minix/drivers/fxp/fxp.c

2205 lines
51 KiB
C
Raw Normal View History

/*
* fxp.c
*
* This file contains an ethernet device driver for Intel 82557, 82558,
* 82559, 82550, and 82562 fast ethernet controllers.
*
* Created: Nov 2004 by Philip Homburg <philip@f-mnx.phicoh.com>
*/
#include <minix/drivers.h>
Driver refactory for live update and crash recovery. SYSLIB CHANGES: - DS calls to publish / retrieve labels consider endpoints instead of u32_t. VFS CHANGES: - mapdriver() only adds an entry in the dmap table in VFS. - dev_up() is only executed upon reception of a driver up event. INET CHANGES: - INET no longer searches for existing drivers instances at startup. - A newtwork driver is (re)initialized upon reception of a driver up event. - Networking startup is now race-free by design. No need to waste 5 seconds at startup any more. DRIVER CHANGES: - Every driver publishes driver up events when starting for the first time or in case of restart when recovery actions must be taken in the upper layers. - Driver up events are published by drivers through DS. - For regular drivers, VFS is normally the only subscriber, but not necessarily. For instance, when the filter driver is in use, it must subscribe to driver up events to initiate recovery. - For network drivers, inet is the only subscriber for now. - Every VFS driver is statically linked with libdriver, every network driver is statically linked with libnetdriver. DRIVER LIBRARIES CHANGES: - Libdriver is extended to provide generic receive() and ds_publish() interfaces for VFS drivers. - driver_receive() is a wrapper for sef_receive() also used in driver_task() to discard spurious messages that were meant to be delivered to a previous version of the driver. - driver_receive_mq() is the same as driver_receive() but integrates support for queued messages. - driver_announce() publishes a driver up event for VFS drivers and marks the driver as initialized and expecting a DEV_OPEN message. - Libnetdriver is introduced to provide similar receive() and ds_publish() interfaces for network drivers (netdriver_announce() and netdriver_receive()). - Network drivers all support live update with no state transfer now. KERNEL CHANGES: - Added kernel call statectl for state management. Used by driver_announce() to unblock eventual callers sendrecing to the driver.
2010-04-08 15:41:35 +02:00
#include <minix/netdriver.h>
#include <stdlib.h>
#include <net/hton.h>
#include <net/gen/ether.h>
#include <net/gen/eth_io.h>
#include <machine/pci.h>
#include <minix/ds.h>
#include <minix/endpoint.h>
#include <minix/timers.h>
#define debug 0
#define RAND_UPDATE /**/
#define printW() ((void)0)
#include "assert.h"
#include "fxp.h"
#include "mii.h"
/* Number of receive buffers */
#define N_RX_BUF 40
/* Number of transmit buffers */
#define N_TX_BUF 4
/* I/O vectors are handled IOVEC_NR entries at a time. */
#define IOVEC_NR 16
/* Configuration */
#define FXP_ENVVAR "FXPETH"
typedef int irq_hook_t;
/* ignore interrupt for the moment */
#define interrupt(x) do { } while(0)
2012-03-25 20:25:53 +02:00
static union tmpbuf
{
char pad[4096];
struct cbl_conf cc;
struct ias ias;
} *tmpbufp;
typedef struct fxp
{
port_t fxp_base_port;
int fxp_mode;
int fxp_got_int;
int fxp_send_int;
int fxp_flags;
int fxp_client;
int fxp_features; /* Needed? */
int fxp_irq;
int fxp_type; /* What kind of hardware */
int fxp_ms_regs; /* Master/slave registers */
int fxp_ee_addrlen; /* #EEPROM address bits */
int fxp_tx_alive;
int fxp_need_reset;
/* Rx */
vir_bytes fxp_read_s;
int fxp_rx_nbuf;
int fxp_rx_bufsize;
struct rfd *fxp_rx_buf;
phys_bytes fxp_rx_busaddr;
int fxp_rx_head;
int fxp_rx_need_restart;
int fxp_need_conf; /* Re-configure after draining send
* queue
*/
/* Tx */
int fxp_tx_nbuf;
int fxp_tx_bufsize;
struct tx *fxp_tx_buf;
phys_bytes fxp_tx_busaddr;
int fxp_tx_idle;
int fxp_tx_head;
int fxp_tx_tail;
int fxp_tx_threshold;
/* Link status */
int fxp_report_link;
int fxp_link_up;
int fxp_mii_busy;
u16_t fxp_mii_scr;
/* PCI related */
int fxp_seen; /* TRUE iff device available */
/* 'large' items */
irq_hook_t fxp_hook;
ether_addr_t fxp_address;
message fxp_rx_mess;
message fxp_tx_mess;
struct sc fxp_stat;
u8_t fxp_conf_bytes[CC_BYTES_NR];
char fxp_name[sizeof("fxp#n")];
iovec_t fxp_iovec[IOVEC_NR];
2006-07-10 14:43:38 +02:00
iovec_s_t fxp_iovec_s[IOVEC_NR];
}
fxp_t;
/* fxp_mode */
#define FM_DISABLED 0x0
#define FM_ENABLED 0x1
/* fxp_flags */
#define FF_EMPTY 0x000
#define FF_PACK_SENT 0x001
#define FF_PACK_RECV 0x002
#define FF_SEND_AVAIL 0x004
#define FF_READING 0x010
#define FF_PROMISC 0x040
#define FF_MULTI 0x080
#define FF_BROAD 0x100
#define FF_ENABLED 0x200
/* fxp_features */
#define FFE_NONE 0x0
/* fxp_type */
#define FT_UNKNOWN 0x0
#define FT_82557 0x1
#define FT_82558A 0x2
#define FT_82559 0x4
2009-08-26 23:01:46 +02:00
#define FT_82801 0x8
2012-03-25 20:25:53 +02:00
static int fxp_instance;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
2012-03-25 20:25:53 +02:00
static fxp_t *fxp_state;
static minix_timer_t fxp_watchdog;
2012-03-25 20:25:53 +02:00
static u32_t system_hz;
#define fxp_inb(port, offset) (do_inb((port) + (offset)))
#define fxp_inl(port, offset) (do_inl((port) + (offset)))
#define fxp_outb(port, offset, value) (do_outb((port) + (offset), (value)))
#define fxp_outl(port, offset, value) (do_outl((port) + (offset), (value)))
static void fxp_init(message *mp);
static void fxp_pci_conf(void);
static int fxp_probe(fxp_t *fp, int skip);
static void fxp_conf_hw(fxp_t *fp);
static void fxp_init_hw(fxp_t *fp);
static void fxp_init_buf(fxp_t *fp);
static void fxp_reset_hw(fxp_t *fp);
static void fxp_confaddr(fxp_t *fp);
static void fxp_rec_mode(fxp_t *fp);
static void fxp_writev_s(const message *mp, int from_int);
static void fxp_readv_s(message *mp, int from_int);
static void fxp_do_conf(fxp_t *fp);
static void fxp_cu_ptr_cmd(fxp_t *fp, int cmd, phys_bytes bus_addr, int
check_idle);
static void fxp_ru_ptr_cmd(fxp_t *fp, int cmd, phys_bytes bus_addr, int
check_idle);
static void fxp_restart_ru(fxp_t *fp);
static void fxp_getstat_s(message *mp);
static void fxp_handler(fxp_t *fp);
static void fxp_check_ints(fxp_t *fp);
static void fxp_watchdog_f(minix_timer_t *tp);
static int fxp_link_changed(fxp_t *fp);
static void fxp_report_link(fxp_t *fp);
static void reply(fxp_t *fp);
static void mess_reply(message *req, message *reply);
static u16_t eeprom_read(fxp_t *fp, int reg);
static void eeprom_addrsize(fxp_t *fp);
static u16_t mii_read(fxp_t *fp, int reg);
static u8_t do_inb(port_t port);
static u32_t do_inl(port_t port);
static void do_outb(port_t port, u8_t v);
static void do_outl(port_t port, u32_t v);
static void tell_dev(vir_bytes start, size_t size, int pci_bus, int
pci_dev, int pci_func);
2012-03-25 20:25:53 +02:00
static void handle_hw_intr(void)
{
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
int r;
fxp_t *fp;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
fp= fxp_state;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
if (fp->fxp_mode != FM_ENABLED)
return;
fxp_handler(fp);
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
r= sys_irqenable(&fp->fxp_hook);
if (r != OK) {
panic("unable enable interrupts: %d", r);
}
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
if (!fp->fxp_got_int)
return;
fp->fxp_got_int= 0;
assert(fp->fxp_flags & FF_ENABLED);
fxp_check_ints(fp);
}
Basic System Event Framework (SEF) with ping and live update. SYSLIB CHANGES: - SEF must be used by every system process and is thereby part of the system library. - The framework provides a receive() interface (sef_receive) for system processes to automatically catch known system even messages and process them. - SEF provides a default behavior for each type of system event, but allows system processes to register callbacks to override the default behavior. - Custom (local to the process) or predefined (provided by SEF) callback implementations can be registered to SEF. - SEF currently includes support for 2 types of system events: 1. SEF Ping. The event occurs every time RS sends a ping to figure out whether a system process is still alive. The default callback implementation provided by SEF is to notify RS back to let it know the process is alive and kicking. 2. SEF Live update. The event occurs every time RS sends a prepare to update message to let a system process know an update is available and to prepare for it. The live update support is very basic for now. SEF only deals with verifying if the prepare state can be supported by the process, dumping the state for debugging purposes, and providing an event-driven programming model to the process to react to state changes check-in when ready to update. - SEF should be extended in the future to integrate support for more types of system events. Ideally, all the cross-cutting concerns should be integrated into SEF to avoid duplicating code and ease extensibility. Examples include: * PM notify messages primarily used at shutdown. * SYSTEM notify messages primarily used for signals. * CLOCK notify messages used for system alarms. * Debug messages. IS could still be in charge of fkey handling but would forward the debug message to the target process (e.g. PM, if the user requested debug information about PM). SEF would then catch the message and do nothing unless the process has registered an appropriate callback to deal with the event. This simplifies the programming model to print debug information, avoids duplicating code, and reduces the effort to print debug information. SYSTEM PROCESSES CHANGES: - Every system process registers SEF callbacks it needs to override the default system behavior and calls sef_startup() right after being started. - sef_startup() does almost nothing now, but will be extended in the future to support callbacks of its own to let RS control and synchronize with every system process at initialization time. - Every system process calls sef_receive() now rather than receive() directly, to let SEF handle predefined system events. RS CHANGES: - RS supports a basic single-component live update protocol now, as follows: * When an update command is issued (via "service update *"), RS notifies the target system process to prepare for a specific update state. * If the process doesn't respond back in time, the update is aborted. * When the process responds back, RS kills it and marks it for refreshing. * The process is then automatically restarted as for a buggy process and can start running again. * Live update is currently prototyped as a controlled failure.
2009-12-21 15:12:21 +01:00
/* SEF functions and variables. */
2012-03-25 20:25:53 +02:00
static void sef_local_startup(void);
static int sef_cb_init_fresh(int type, sef_init_info_t *info);
static void sef_cb_signal_handler(int signo);
Basic System Event Framework (SEF) with ping and live update. SYSLIB CHANGES: - SEF must be used by every system process and is thereby part of the system library. - The framework provides a receive() interface (sef_receive) for system processes to automatically catch known system even messages and process them. - SEF provides a default behavior for each type of system event, but allows system processes to register callbacks to override the default behavior. - Custom (local to the process) or predefined (provided by SEF) callback implementations can be registered to SEF. - SEF currently includes support for 2 types of system events: 1. SEF Ping. The event occurs every time RS sends a ping to figure out whether a system process is still alive. The default callback implementation provided by SEF is to notify RS back to let it know the process is alive and kicking. 2. SEF Live update. The event occurs every time RS sends a prepare to update message to let a system process know an update is available and to prepare for it. The live update support is very basic for now. SEF only deals with verifying if the prepare state can be supported by the process, dumping the state for debugging purposes, and providing an event-driven programming model to the process to react to state changes check-in when ready to update. - SEF should be extended in the future to integrate support for more types of system events. Ideally, all the cross-cutting concerns should be integrated into SEF to avoid duplicating code and ease extensibility. Examples include: * PM notify messages primarily used at shutdown. * SYSTEM notify messages primarily used for signals. * CLOCK notify messages used for system alarms. * Debug messages. IS could still be in charge of fkey handling but would forward the debug message to the target process (e.g. PM, if the user requested debug information about PM). SEF would then catch the message and do nothing unless the process has registered an appropriate callback to deal with the event. This simplifies the programming model to print debug information, avoids duplicating code, and reduces the effort to print debug information. SYSTEM PROCESSES CHANGES: - Every system process registers SEF callbacks it needs to override the default system behavior and calls sef_startup() right after being started. - sef_startup() does almost nothing now, but will be extended in the future to support callbacks of its own to let RS control and synchronize with every system process at initialization time. - Every system process calls sef_receive() now rather than receive() directly, to let SEF handle predefined system events. RS CHANGES: - RS supports a basic single-component live update protocol now, as follows: * When an update command is issued (via "service update *"), RS notifies the target system process to prepare for a specific update state. * If the process doesn't respond back in time, the update is aborted. * When the process responds back, RS kills it and marks it for refreshing. * The process is then automatically restarted as for a buggy process and can start running again. * Live update is currently prototyped as a controlled failure.
2009-12-21 15:12:21 +01:00
/*===========================================================================*
* main *
*===========================================================================*/
int main(int argc, char *argv[])
{
message m;
Driver refactory for live update and crash recovery. SYSLIB CHANGES: - DS calls to publish / retrieve labels consider endpoints instead of u32_t. VFS CHANGES: - mapdriver() only adds an entry in the dmap table in VFS. - dev_up() is only executed upon reception of a driver up event. INET CHANGES: - INET no longer searches for existing drivers instances at startup. - A newtwork driver is (re)initialized upon reception of a driver up event. - Networking startup is now race-free by design. No need to waste 5 seconds at startup any more. DRIVER CHANGES: - Every driver publishes driver up events when starting for the first time or in case of restart when recovery actions must be taken in the upper layers. - Driver up events are published by drivers through DS. - For regular drivers, VFS is normally the only subscriber, but not necessarily. For instance, when the filter driver is in use, it must subscribe to driver up events to initiate recovery. - For network drivers, inet is the only subscriber for now. - Every VFS driver is statically linked with libdriver, every network driver is statically linked with libnetdriver. DRIVER LIBRARIES CHANGES: - Libdriver is extended to provide generic receive() and ds_publish() interfaces for VFS drivers. - driver_receive() is a wrapper for sef_receive() also used in driver_task() to discard spurious messages that were meant to be delivered to a previous version of the driver. - driver_receive_mq() is the same as driver_receive() but integrates support for queued messages. - driver_announce() publishes a driver up event for VFS drivers and marks the driver as initialized and expecting a DEV_OPEN message. - Libnetdriver is introduced to provide similar receive() and ds_publish() interfaces for network drivers (netdriver_announce() and netdriver_receive()). - Network drivers all support live update with no state transfer now. KERNEL CHANGES: - Added kernel call statectl for state management. Used by driver_announce() to unblock eventual callers sendrecing to the driver.
2010-04-08 15:41:35 +02:00
int ipc_status;
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
int r;
Basic System Event Framework (SEF) with ping and live update. SYSLIB CHANGES: - SEF must be used by every system process and is thereby part of the system library. - The framework provides a receive() interface (sef_receive) for system processes to automatically catch known system even messages and process them. - SEF provides a default behavior for each type of system event, but allows system processes to register callbacks to override the default behavior. - Custom (local to the process) or predefined (provided by SEF) callback implementations can be registered to SEF. - SEF currently includes support for 2 types of system events: 1. SEF Ping. The event occurs every time RS sends a ping to figure out whether a system process is still alive. The default callback implementation provided by SEF is to notify RS back to let it know the process is alive and kicking. 2. SEF Live update. The event occurs every time RS sends a prepare to update message to let a system process know an update is available and to prepare for it. The live update support is very basic for now. SEF only deals with verifying if the prepare state can be supported by the process, dumping the state for debugging purposes, and providing an event-driven programming model to the process to react to state changes check-in when ready to update. - SEF should be extended in the future to integrate support for more types of system events. Ideally, all the cross-cutting concerns should be integrated into SEF to avoid duplicating code and ease extensibility. Examples include: * PM notify messages primarily used at shutdown. * SYSTEM notify messages primarily used for signals. * CLOCK notify messages used for system alarms. * Debug messages. IS could still be in charge of fkey handling but would forward the debug message to the target process (e.g. PM, if the user requested debug information about PM). SEF would then catch the message and do nothing unless the process has registered an appropriate callback to deal with the event. This simplifies the programming model to print debug information, avoids duplicating code, and reduces the effort to print debug information. SYSTEM PROCESSES CHANGES: - Every system process registers SEF callbacks it needs to override the default system behavior and calls sef_startup() right after being started. - sef_startup() does almost nothing now, but will be extended in the future to support callbacks of its own to let RS control and synchronize with every system process at initialization time. - Every system process calls sef_receive() now rather than receive() directly, to let SEF handle predefined system events. RS CHANGES: - RS supports a basic single-component live update protocol now, as follows: * When an update command is issued (via "service update *"), RS notifies the target system process to prepare for a specific update state. * If the process doesn't respond back in time, the update is aborted. * When the process responds back, RS kills it and marks it for refreshing. * The process is then automatically restarted as for a buggy process and can start running again. * Live update is currently prototyped as a controlled failure.
2009-12-21 15:12:21 +01:00
/* SEF local startup. */
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
env_setargs(argc, argv);
Basic System Event Framework (SEF) with ping and live update. SYSLIB CHANGES: - SEF must be used by every system process and is thereby part of the system library. - The framework provides a receive() interface (sef_receive) for system processes to automatically catch known system even messages and process them. - SEF provides a default behavior for each type of system event, but allows system processes to register callbacks to override the default behavior. - Custom (local to the process) or predefined (provided by SEF) callback implementations can be registered to SEF. - SEF currently includes support for 2 types of system events: 1. SEF Ping. The event occurs every time RS sends a ping to figure out whether a system process is still alive. The default callback implementation provided by SEF is to notify RS back to let it know the process is alive and kicking. 2. SEF Live update. The event occurs every time RS sends a prepare to update message to let a system process know an update is available and to prepare for it. The live update support is very basic for now. SEF only deals with verifying if the prepare state can be supported by the process, dumping the state for debugging purposes, and providing an event-driven programming model to the process to react to state changes check-in when ready to update. - SEF should be extended in the future to integrate support for more types of system events. Ideally, all the cross-cutting concerns should be integrated into SEF to avoid duplicating code and ease extensibility. Examples include: * PM notify messages primarily used at shutdown. * SYSTEM notify messages primarily used for signals. * CLOCK notify messages used for system alarms. * Debug messages. IS could still be in charge of fkey handling but would forward the debug message to the target process (e.g. PM, if the user requested debug information about PM). SEF would then catch the message and do nothing unless the process has registered an appropriate callback to deal with the event. This simplifies the programming model to print debug information, avoids duplicating code, and reduces the effort to print debug information. SYSTEM PROCESSES CHANGES: - Every system process registers SEF callbacks it needs to override the default system behavior and calls sef_startup() right after being started. - sef_startup() does almost nothing now, but will be extended in the future to support callbacks of its own to let RS control and synchronize with every system process at initialization time. - Every system process calls sef_receive() now rather than receive() directly, to let SEF handle predefined system events. RS CHANGES: - RS supports a basic single-component live update protocol now, as follows: * When an update command is issued (via "service update *"), RS notifies the target system process to prepare for a specific update state. * If the process doesn't respond back in time, the update is aborted. * When the process responds back, RS kills it and marks it for refreshing. * The process is then automatically restarted as for a buggy process and can start running again. * Live update is currently prototyped as a controlled failure.
2009-12-21 15:12:21 +01:00
sef_local_startup();
while (TRUE)
{
Driver refactory for live update and crash recovery. SYSLIB CHANGES: - DS calls to publish / retrieve labels consider endpoints instead of u32_t. VFS CHANGES: - mapdriver() only adds an entry in the dmap table in VFS. - dev_up() is only executed upon reception of a driver up event. INET CHANGES: - INET no longer searches for existing drivers instances at startup. - A newtwork driver is (re)initialized upon reception of a driver up event. - Networking startup is now race-free by design. No need to waste 5 seconds at startup any more. DRIVER CHANGES: - Every driver publishes driver up events when starting for the first time or in case of restart when recovery actions must be taken in the upper layers. - Driver up events are published by drivers through DS. - For regular drivers, VFS is normally the only subscriber, but not necessarily. For instance, when the filter driver is in use, it must subscribe to driver up events to initiate recovery. - For network drivers, inet is the only subscriber for now. - Every VFS driver is statically linked with libdriver, every network driver is statically linked with libnetdriver. DRIVER LIBRARIES CHANGES: - Libdriver is extended to provide generic receive() and ds_publish() interfaces for VFS drivers. - driver_receive() is a wrapper for sef_receive() also used in driver_task() to discard spurious messages that were meant to be delivered to a previous version of the driver. - driver_receive_mq() is the same as driver_receive() but integrates support for queued messages. - driver_announce() publishes a driver up event for VFS drivers and marks the driver as initialized and expecting a DEV_OPEN message. - Libnetdriver is introduced to provide similar receive() and ds_publish() interfaces for network drivers (netdriver_announce() and netdriver_receive()). - Network drivers all support live update with no state transfer now. KERNEL CHANGES: - Added kernel call statectl for state management. Used by driver_announce() to unblock eventual callers sendrecing to the driver.
2010-04-08 15:41:35 +02:00
if ((r= netdriver_receive(ANY, &m, &ipc_status)) != OK)
panic("netdriver_receive failed: %d", r);
Driver refactory for live update and crash recovery. SYSLIB CHANGES: - DS calls to publish / retrieve labels consider endpoints instead of u32_t. VFS CHANGES: - mapdriver() only adds an entry in the dmap table in VFS. - dev_up() is only executed upon reception of a driver up event. INET CHANGES: - INET no longer searches for existing drivers instances at startup. - A newtwork driver is (re)initialized upon reception of a driver up event. - Networking startup is now race-free by design. No need to waste 5 seconds at startup any more. DRIVER CHANGES: - Every driver publishes driver up events when starting for the first time or in case of restart when recovery actions must be taken in the upper layers. - Driver up events are published by drivers through DS. - For regular drivers, VFS is normally the only subscriber, but not necessarily. For instance, when the filter driver is in use, it must subscribe to driver up events to initiate recovery. - For network drivers, inet is the only subscriber for now. - Every VFS driver is statically linked with libdriver, every network driver is statically linked with libnetdriver. DRIVER LIBRARIES CHANGES: - Libdriver is extended to provide generic receive() and ds_publish() interfaces for VFS drivers. - driver_receive() is a wrapper for sef_receive() also used in driver_task() to discard spurious messages that were meant to be delivered to a previous version of the driver. - driver_receive_mq() is the same as driver_receive() but integrates support for queued messages. - driver_announce() publishes a driver up event for VFS drivers and marks the driver as initialized and expecting a DEV_OPEN message. - Libnetdriver is introduced to provide similar receive() and ds_publish() interfaces for network drivers (netdriver_announce() and netdriver_receive()). - Network drivers all support live update with no state transfer now. KERNEL CHANGES: - Added kernel call statectl for state management. Used by driver_announce() to unblock eventual callers sendrecing to the driver.
2010-04-08 15:41:35 +02:00
if (is_ipc_notify(ipc_status)) {
switch (_ENDPOINT_P(m.m_source)) {
case HARDWARE:
handle_hw_intr();
break;
case CLOCK:
expire_timers(m.m_notify.timestamp);
break;
default:
panic(" illegal notify from: %d", m.m_source);
}
/* get new message */
continue;
}
switch (m.m_type)
{
2006-07-10 14:43:38 +02:00
case DL_WRITEV_S: fxp_writev_s(&m, FALSE); break;
case DL_READV_S: fxp_readv_s(&m, FALSE); break;
case DL_CONF: fxp_init(&m); break;
case DL_GETSTAT_S: fxp_getstat_s(&m); break;
default:
panic(" illegal message: %d", m.m_type);
}
}
}
Basic System Event Framework (SEF) with ping and live update. SYSLIB CHANGES: - SEF must be used by every system process and is thereby part of the system library. - The framework provides a receive() interface (sef_receive) for system processes to automatically catch known system even messages and process them. - SEF provides a default behavior for each type of system event, but allows system processes to register callbacks to override the default behavior. - Custom (local to the process) or predefined (provided by SEF) callback implementations can be registered to SEF. - SEF currently includes support for 2 types of system events: 1. SEF Ping. The event occurs every time RS sends a ping to figure out whether a system process is still alive. The default callback implementation provided by SEF is to notify RS back to let it know the process is alive and kicking. 2. SEF Live update. The event occurs every time RS sends a prepare to update message to let a system process know an update is available and to prepare for it. The live update support is very basic for now. SEF only deals with verifying if the prepare state can be supported by the process, dumping the state for debugging purposes, and providing an event-driven programming model to the process to react to state changes check-in when ready to update. - SEF should be extended in the future to integrate support for more types of system events. Ideally, all the cross-cutting concerns should be integrated into SEF to avoid duplicating code and ease extensibility. Examples include: * PM notify messages primarily used at shutdown. * SYSTEM notify messages primarily used for signals. * CLOCK notify messages used for system alarms. * Debug messages. IS could still be in charge of fkey handling but would forward the debug message to the target process (e.g. PM, if the user requested debug information about PM). SEF would then catch the message and do nothing unless the process has registered an appropriate callback to deal with the event. This simplifies the programming model to print debug information, avoids duplicating code, and reduces the effort to print debug information. SYSTEM PROCESSES CHANGES: - Every system process registers SEF callbacks it needs to override the default system behavior and calls sef_startup() right after being started. - sef_startup() does almost nothing now, but will be extended in the future to support callbacks of its own to let RS control and synchronize with every system process at initialization time. - Every system process calls sef_receive() now rather than receive() directly, to let SEF handle predefined system events. RS CHANGES: - RS supports a basic single-component live update protocol now, as follows: * When an update command is issued (via "service update *"), RS notifies the target system process to prepare for a specific update state. * If the process doesn't respond back in time, the update is aborted. * When the process responds back, RS kills it and marks it for refreshing. * The process is then automatically restarted as for a buggy process and can start running again. * Live update is currently prototyped as a controlled failure.
2009-12-21 15:12:21 +01:00
/*===========================================================================*
* sef_local_startup *
*===========================================================================*/
2012-03-25 20:25:53 +02:00
static void sef_local_startup()
Basic System Event Framework (SEF) with ping and live update. SYSLIB CHANGES: - SEF must be used by every system process and is thereby part of the system library. - The framework provides a receive() interface (sef_receive) for system processes to automatically catch known system even messages and process them. - SEF provides a default behavior for each type of system event, but allows system processes to register callbacks to override the default behavior. - Custom (local to the process) or predefined (provided by SEF) callback implementations can be registered to SEF. - SEF currently includes support for 2 types of system events: 1. SEF Ping. The event occurs every time RS sends a ping to figure out whether a system process is still alive. The default callback implementation provided by SEF is to notify RS back to let it know the process is alive and kicking. 2. SEF Live update. The event occurs every time RS sends a prepare to update message to let a system process know an update is available and to prepare for it. The live update support is very basic for now. SEF only deals with verifying if the prepare state can be supported by the process, dumping the state for debugging purposes, and providing an event-driven programming model to the process to react to state changes check-in when ready to update. - SEF should be extended in the future to integrate support for more types of system events. Ideally, all the cross-cutting concerns should be integrated into SEF to avoid duplicating code and ease extensibility. Examples include: * PM notify messages primarily used at shutdown. * SYSTEM notify messages primarily used for signals. * CLOCK notify messages used for system alarms. * Debug messages. IS could still be in charge of fkey handling but would forward the debug message to the target process (e.g. PM, if the user requested debug information about PM). SEF would then catch the message and do nothing unless the process has registered an appropriate callback to deal with the event. This simplifies the programming model to print debug information, avoids duplicating code, and reduces the effort to print debug information. SYSTEM PROCESSES CHANGES: - Every system process registers SEF callbacks it needs to override the default system behavior and calls sef_startup() right after being started. - sef_startup() does almost nothing now, but will be extended in the future to support callbacks of its own to let RS control and synchronize with every system process at initialization time. - Every system process calls sef_receive() now rather than receive() directly, to let SEF handle predefined system events. RS CHANGES: - RS supports a basic single-component live update protocol now, as follows: * When an update command is issued (via "service update *"), RS notifies the target system process to prepare for a specific update state. * If the process doesn't respond back in time, the update is aborted. * When the process responds back, RS kills it and marks it for refreshing. * The process is then automatically restarted as for a buggy process and can start running again. * Live update is currently prototyped as a controlled failure.
2009-12-21 15:12:21 +01:00
{
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
/* Register init callbacks. */
sef_setcb_init_fresh(sef_cb_init_fresh);
Driver refactory for live update and crash recovery. SYSLIB CHANGES: - DS calls to publish / retrieve labels consider endpoints instead of u32_t. VFS CHANGES: - mapdriver() only adds an entry in the dmap table in VFS. - dev_up() is only executed upon reception of a driver up event. INET CHANGES: - INET no longer searches for existing drivers instances at startup. - A newtwork driver is (re)initialized upon reception of a driver up event. - Networking startup is now race-free by design. No need to waste 5 seconds at startup any more. DRIVER CHANGES: - Every driver publishes driver up events when starting for the first time or in case of restart when recovery actions must be taken in the upper layers. - Driver up events are published by drivers through DS. - For regular drivers, VFS is normally the only subscriber, but not necessarily. For instance, when the filter driver is in use, it must subscribe to driver up events to initiate recovery. - For network drivers, inet is the only subscriber for now. - Every VFS driver is statically linked with libdriver, every network driver is statically linked with libnetdriver. DRIVER LIBRARIES CHANGES: - Libdriver is extended to provide generic receive() and ds_publish() interfaces for VFS drivers. - driver_receive() is a wrapper for sef_receive() also used in driver_task() to discard spurious messages that were meant to be delivered to a previous version of the driver. - driver_receive_mq() is the same as driver_receive() but integrates support for queued messages. - driver_announce() publishes a driver up event for VFS drivers and marks the driver as initialized and expecting a DEV_OPEN message. - Libnetdriver is introduced to provide similar receive() and ds_publish() interfaces for network drivers (netdriver_announce() and netdriver_receive()). - Network drivers all support live update with no state transfer now. KERNEL CHANGES: - Added kernel call statectl for state management. Used by driver_announce() to unblock eventual callers sendrecing to the driver.
2010-04-08 15:41:35 +02:00
sef_setcb_init_lu(sef_cb_init_fresh);
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
sef_setcb_init_restart(sef_cb_init_fresh);
Driver refactory for live update and crash recovery. SYSLIB CHANGES: - DS calls to publish / retrieve labels consider endpoints instead of u32_t. VFS CHANGES: - mapdriver() only adds an entry in the dmap table in VFS. - dev_up() is only executed upon reception of a driver up event. INET CHANGES: - INET no longer searches for existing drivers instances at startup. - A newtwork driver is (re)initialized upon reception of a driver up event. - Networking startup is now race-free by design. No need to waste 5 seconds at startup any more. DRIVER CHANGES: - Every driver publishes driver up events when starting for the first time or in case of restart when recovery actions must be taken in the upper layers. - Driver up events are published by drivers through DS. - For regular drivers, VFS is normally the only subscriber, but not necessarily. For instance, when the filter driver is in use, it must subscribe to driver up events to initiate recovery. - For network drivers, inet is the only subscriber for now. - Every VFS driver is statically linked with libdriver, every network driver is statically linked with libnetdriver. DRIVER LIBRARIES CHANGES: - Libdriver is extended to provide generic receive() and ds_publish() interfaces for VFS drivers. - driver_receive() is a wrapper for sef_receive() also used in driver_task() to discard spurious messages that were meant to be delivered to a previous version of the driver. - driver_receive_mq() is the same as driver_receive() but integrates support for queued messages. - driver_announce() publishes a driver up event for VFS drivers and marks the driver as initialized and expecting a DEV_OPEN message. - Libnetdriver is introduced to provide similar receive() and ds_publish() interfaces for network drivers (netdriver_announce() and netdriver_receive()). - Network drivers all support live update with no state transfer now. KERNEL CHANGES: - Added kernel call statectl for state management. Used by driver_announce() to unblock eventual callers sendrecing to the driver.
2010-04-08 15:41:35 +02:00
/* Register live update callbacks. */
sef_setcb_lu_prepare(sef_cb_lu_prepare_always_ready);
sef_setcb_lu_state_isvalid(sef_cb_lu_state_isvalid_workfree);
Basic System Event Framework (SEF) with ping and live update. SYSLIB CHANGES: - SEF must be used by every system process and is thereby part of the system library. - The framework provides a receive() interface (sef_receive) for system processes to automatically catch known system even messages and process them. - SEF provides a default behavior for each type of system event, but allows system processes to register callbacks to override the default behavior. - Custom (local to the process) or predefined (provided by SEF) callback implementations can be registered to SEF. - SEF currently includes support for 2 types of system events: 1. SEF Ping. The event occurs every time RS sends a ping to figure out whether a system process is still alive. The default callback implementation provided by SEF is to notify RS back to let it know the process is alive and kicking. 2. SEF Live update. The event occurs every time RS sends a prepare to update message to let a system process know an update is available and to prepare for it. The live update support is very basic for now. SEF only deals with verifying if the prepare state can be supported by the process, dumping the state for debugging purposes, and providing an event-driven programming model to the process to react to state changes check-in when ready to update. - SEF should be extended in the future to integrate support for more types of system events. Ideally, all the cross-cutting concerns should be integrated into SEF to avoid duplicating code and ease extensibility. Examples include: * PM notify messages primarily used at shutdown. * SYSTEM notify messages primarily used for signals. * CLOCK notify messages used for system alarms. * Debug messages. IS could still be in charge of fkey handling but would forward the debug message to the target process (e.g. PM, if the user requested debug information about PM). SEF would then catch the message and do nothing unless the process has registered an appropriate callback to deal with the event. This simplifies the programming model to print debug information, avoids duplicating code, and reduces the effort to print debug information. SYSTEM PROCESSES CHANGES: - Every system process registers SEF callbacks it needs to override the default system behavior and calls sef_startup() right after being started. - sef_startup() does almost nothing now, but will be extended in the future to support callbacks of its own to let RS control and synchronize with every system process at initialization time. - Every system process calls sef_receive() now rather than receive() directly, to let SEF handle predefined system events. RS CHANGES: - RS supports a basic single-component live update protocol now, as follows: * When an update command is issued (via "service update *"), RS notifies the target system process to prepare for a specific update state. * If the process doesn't respond back in time, the update is aborted. * When the process responds back, RS kills it and marks it for refreshing. * The process is then automatically restarted as for a buggy process and can start running again. * Live update is currently prototyped as a controlled failure.
2009-12-21 15:12:21 +01:00
New RS and new signal handling for system processes. UPDATING INFO: 20100317: /usr/src/etc/system.conf updated to ignore default kernel calls: copy it (or merge it) to /etc/system.conf. The hello driver (/dev/hello) added to the distribution: # cd /usr/src/commands/scripts && make clean install # cd /dev && MAKEDEV hello KERNEL CHANGES: - Generic signal handling support. The kernel no longer assumes PM as a signal manager for every process. The signal manager of a given process can now be specified in its privilege slot. When a signal has to be delivered, the kernel performs the lookup and forwards the signal to the appropriate signal manager. PM is the default signal manager for user processes, RS is the default signal manager for system processes. To enable ptrace()ing for system processes, it is sufficient to change the default signal manager to PM. This will temporarily disable crash recovery, though. - sys_exit() is now split into sys_exit() (i.e. exit() for system processes, which generates a self-termination signal), and sys_clear() (i.e. used by PM to ask the kernel to clear a process slot when a process exits). - Added a new kernel call (i.e. sys_update()) to swap two process slots and implement live update. PM CHANGES: - Posix signal handling is no longer allowed for system processes. System signals are split into two fixed categories: termination and non-termination signals. When a non-termination signaled is processed, PM transforms the signal into an IPC message and delivers the message to the system process. When a termination signal is processed, PM terminates the process. - PM no longer assumes itself as the signal manager for system processes. It now makes sure that every system signal goes through the kernel before being actually processes. The kernel will then dispatch the signal to the appropriate signal manager which may or may not be PM. SYSLIB CHANGES: - Simplified SEF init and LU callbacks. - Added additional predefined SEF callbacks to debug crash recovery and live update. - Fixed a temporary ack in the SEF init protocol. SEF init reply is now completely synchronous. - Added SEF signal event type to provide a uniform interface for system processes to deal with signals. A sef_cb_signal_handler() callback is available for system processes to handle every received signal. A sef_cb_signal_manager() callback is used by signal managers to process system signals on behalf of the kernel. - Fixed a few bugs with memory mapping and DS. VM CHANGES: - Page faults and memory requests coming from the kernel are now implemented using signals. - Added a new VM call to swap two process slots and implement live update. - The call is used by RS at update time and in turn invokes the kernel call sys_update(). RS CHANGES: - RS has been reworked with a better functional decomposition. - Better kernel call masks. com.h now defines the set of very basic kernel calls every system service is allowed to use. This makes system.conf simpler and easier to maintain. In addition, this guarantees a higher level of isolation for system libraries that use one or more kernel calls internally (e.g. printf). - RS is the default signal manager for system processes. By default, RS intercepts every signal delivered to every system process. This makes crash recovery possible before bringing PM and friends in the loop. - RS now supports fast rollback when something goes wrong while initializing the new version during a live update. - Live update is now implemented by keeping the two versions side-by-side and swapping the process slots when the old version is ready to update. - Crash recovery is now implemented by keeping the two versions side-by-side and cleaning up the old version only when the recovery process is complete. DS CHANGES: - Fixed a bug when the process doing ds_publish() or ds_delete() is not known by DS. - Fixed the completely broken support for strings. String publishing is now implemented in the system library and simply wraps publishing of memory ranges. Ideally, we should adopt a similar approach for other data types as well. - Test suite fixed. DRIVER CHANGES: - The hello driver has been added to the Minix distribution to demonstrate basic live update and crash recovery functionalities. - Other drivers have been adapted to conform the new SEF interface.
2010-03-17 02:15:29 +01:00
/* Register signal callbacks. */
sef_setcb_signal_handler(sef_cb_signal_handler);
Basic System Event Framework (SEF) with ping and live update. SYSLIB CHANGES: - SEF must be used by every system process and is thereby part of the system library. - The framework provides a receive() interface (sef_receive) for system processes to automatically catch known system even messages and process them. - SEF provides a default behavior for each type of system event, but allows system processes to register callbacks to override the default behavior. - Custom (local to the process) or predefined (provided by SEF) callback implementations can be registered to SEF. - SEF currently includes support for 2 types of system events: 1. SEF Ping. The event occurs every time RS sends a ping to figure out whether a system process is still alive. The default callback implementation provided by SEF is to notify RS back to let it know the process is alive and kicking. 2. SEF Live update. The event occurs every time RS sends a prepare to update message to let a system process know an update is available and to prepare for it. The live update support is very basic for now. SEF only deals with verifying if the prepare state can be supported by the process, dumping the state for debugging purposes, and providing an event-driven programming model to the process to react to state changes check-in when ready to update. - SEF should be extended in the future to integrate support for more types of system events. Ideally, all the cross-cutting concerns should be integrated into SEF to avoid duplicating code and ease extensibility. Examples include: * PM notify messages primarily used at shutdown. * SYSTEM notify messages primarily used for signals. * CLOCK notify messages used for system alarms. * Debug messages. IS could still be in charge of fkey handling but would forward the debug message to the target process (e.g. PM, if the user requested debug information about PM). SEF would then catch the message and do nothing unless the process has registered an appropriate callback to deal with the event. This simplifies the programming model to print debug information, avoids duplicating code, and reduces the effort to print debug information. SYSTEM PROCESSES CHANGES: - Every system process registers SEF callbacks it needs to override the default system behavior and calls sef_startup() right after being started. - sef_startup() does almost nothing now, but will be extended in the future to support callbacks of its own to let RS control and synchronize with every system process at initialization time. - Every system process calls sef_receive() now rather than receive() directly, to let SEF handle predefined system events. RS CHANGES: - RS supports a basic single-component live update protocol now, as follows: * When an update command is issued (via "service update *"), RS notifies the target system process to prepare for a specific update state. * If the process doesn't respond back in time, the update is aborted. * When the process responds back, RS kills it and marks it for refreshing. * The process is then automatically restarted as for a buggy process and can start running again. * Live update is currently prototyped as a controlled failure.
2009-12-21 15:12:21 +01:00
/* Let SEF perform startup. */
sef_startup();
}
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
/*===========================================================================*
* sef_cb_init_fresh *
*===========================================================================*/
2012-03-25 20:25:53 +02:00
static int sef_cb_init_fresh(int UNUSED(type), sef_init_info_t *UNUSED(info))
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
{
/* Initialize the fxp driver. */
long v;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
int r;
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
vir_bytes ft;
system_hz = sys_hz();
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
v = 0;
(void) env_parse("instance", "d", 0, &v, 0, 255);
fxp_instance = (int) v;
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
ft = sizeof(*fxp_state);
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
if(!(fxp_state = alloc_contig(ft, 0, NULL)))
panic("couldn't allocate table: %d", ENOMEM);
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
memset(fxp_state, 0, ft);
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
if((r=tsc_calibrate()) != OK)
panic("tsc_calibrate failed: %d", r);
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
Driver refactory for live update and crash recovery. SYSLIB CHANGES: - DS calls to publish / retrieve labels consider endpoints instead of u32_t. VFS CHANGES: - mapdriver() only adds an entry in the dmap table in VFS. - dev_up() is only executed upon reception of a driver up event. INET CHANGES: - INET no longer searches for existing drivers instances at startup. - A newtwork driver is (re)initialized upon reception of a driver up event. - Networking startup is now race-free by design. No need to waste 5 seconds at startup any more. DRIVER CHANGES: - Every driver publishes driver up events when starting for the first time or in case of restart when recovery actions must be taken in the upper layers. - Driver up events are published by drivers through DS. - For regular drivers, VFS is normally the only subscriber, but not necessarily. For instance, when the filter driver is in use, it must subscribe to driver up events to initiate recovery. - For network drivers, inet is the only subscriber for now. - Every VFS driver is statically linked with libdriver, every network driver is statically linked with libnetdriver. DRIVER LIBRARIES CHANGES: - Libdriver is extended to provide generic receive() and ds_publish() interfaces for VFS drivers. - driver_receive() is a wrapper for sef_receive() also used in driver_task() to discard spurious messages that were meant to be delivered to a previous version of the driver. - driver_receive_mq() is the same as driver_receive() but integrates support for queued messages. - driver_announce() publishes a driver up event for VFS drivers and marks the driver as initialized and expecting a DEV_OPEN message. - Libnetdriver is introduced to provide similar receive() and ds_publish() interfaces for network drivers (netdriver_announce() and netdriver_receive()). - Network drivers all support live update with no state transfer now. KERNEL CHANGES: - Added kernel call statectl for state management. Used by driver_announce() to unblock eventual callers sendrecing to the driver.
2010-04-08 15:41:35 +02:00
/* Announce we are up! */
netdriver_announce();
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
return(OK);
}
New RS and new signal handling for system processes. UPDATING INFO: 20100317: /usr/src/etc/system.conf updated to ignore default kernel calls: copy it (or merge it) to /etc/system.conf. The hello driver (/dev/hello) added to the distribution: # cd /usr/src/commands/scripts && make clean install # cd /dev && MAKEDEV hello KERNEL CHANGES: - Generic signal handling support. The kernel no longer assumes PM as a signal manager for every process. The signal manager of a given process can now be specified in its privilege slot. When a signal has to be delivered, the kernel performs the lookup and forwards the signal to the appropriate signal manager. PM is the default signal manager for user processes, RS is the default signal manager for system processes. To enable ptrace()ing for system processes, it is sufficient to change the default signal manager to PM. This will temporarily disable crash recovery, though. - sys_exit() is now split into sys_exit() (i.e. exit() for system processes, which generates a self-termination signal), and sys_clear() (i.e. used by PM to ask the kernel to clear a process slot when a process exits). - Added a new kernel call (i.e. sys_update()) to swap two process slots and implement live update. PM CHANGES: - Posix signal handling is no longer allowed for system processes. System signals are split into two fixed categories: termination and non-termination signals. When a non-termination signaled is processed, PM transforms the signal into an IPC message and delivers the message to the system process. When a termination signal is processed, PM terminates the process. - PM no longer assumes itself as the signal manager for system processes. It now makes sure that every system signal goes through the kernel before being actually processes. The kernel will then dispatch the signal to the appropriate signal manager which may or may not be PM. SYSLIB CHANGES: - Simplified SEF init and LU callbacks. - Added additional predefined SEF callbacks to debug crash recovery and live update. - Fixed a temporary ack in the SEF init protocol. SEF init reply is now completely synchronous. - Added SEF signal event type to provide a uniform interface for system processes to deal with signals. A sef_cb_signal_handler() callback is available for system processes to handle every received signal. A sef_cb_signal_manager() callback is used by signal managers to process system signals on behalf of the kernel. - Fixed a few bugs with memory mapping and DS. VM CHANGES: - Page faults and memory requests coming from the kernel are now implemented using signals. - Added a new VM call to swap two process slots and implement live update. - The call is used by RS at update time and in turn invokes the kernel call sys_update(). RS CHANGES: - RS has been reworked with a better functional decomposition. - Better kernel call masks. com.h now defines the set of very basic kernel calls every system service is allowed to use. This makes system.conf simpler and easier to maintain. In addition, this guarantees a higher level of isolation for system libraries that use one or more kernel calls internally (e.g. printf). - RS is the default signal manager for system processes. By default, RS intercepts every signal delivered to every system process. This makes crash recovery possible before bringing PM and friends in the loop. - RS now supports fast rollback when something goes wrong while initializing the new version during a live update. - Live update is now implemented by keeping the two versions side-by-side and swapping the process slots when the old version is ready to update. - Crash recovery is now implemented by keeping the two versions side-by-side and cleaning up the old version only when the recovery process is complete. DS CHANGES: - Fixed a bug when the process doing ds_publish() or ds_delete() is not known by DS. - Fixed the completely broken support for strings. String publishing is now implemented in the system library and simply wraps publishing of memory ranges. Ideally, we should adopt a similar approach for other data types as well. - Test suite fixed. DRIVER CHANGES: - The hello driver has been added to the Minix distribution to demonstrate basic live update and crash recovery functionalities. - Other drivers have been adapted to conform the new SEF interface.
2010-03-17 02:15:29 +01:00
/*===========================================================================*
* sef_cb_signal_handler *
*===========================================================================*/
2012-03-25 20:25:53 +02:00
static void sef_cb_signal_handler(int signo)
New RS and new signal handling for system processes. UPDATING INFO: 20100317: /usr/src/etc/system.conf updated to ignore default kernel calls: copy it (or merge it) to /etc/system.conf. The hello driver (/dev/hello) added to the distribution: # cd /usr/src/commands/scripts && make clean install # cd /dev && MAKEDEV hello KERNEL CHANGES: - Generic signal handling support. The kernel no longer assumes PM as a signal manager for every process. The signal manager of a given process can now be specified in its privilege slot. When a signal has to be delivered, the kernel performs the lookup and forwards the signal to the appropriate signal manager. PM is the default signal manager for user processes, RS is the default signal manager for system processes. To enable ptrace()ing for system processes, it is sufficient to change the default signal manager to PM. This will temporarily disable crash recovery, though. - sys_exit() is now split into sys_exit() (i.e. exit() for system processes, which generates a self-termination signal), and sys_clear() (i.e. used by PM to ask the kernel to clear a process slot when a process exits). - Added a new kernel call (i.e. sys_update()) to swap two process slots and implement live update. PM CHANGES: - Posix signal handling is no longer allowed for system processes. System signals are split into two fixed categories: termination and non-termination signals. When a non-termination signaled is processed, PM transforms the signal into an IPC message and delivers the message to the system process. When a termination signal is processed, PM terminates the process. - PM no longer assumes itself as the signal manager for system processes. It now makes sure that every system signal goes through the kernel before being actually processes. The kernel will then dispatch the signal to the appropriate signal manager which may or may not be PM. SYSLIB CHANGES: - Simplified SEF init and LU callbacks. - Added additional predefined SEF callbacks to debug crash recovery and live update. - Fixed a temporary ack in the SEF init protocol. SEF init reply is now completely synchronous. - Added SEF signal event type to provide a uniform interface for system processes to deal with signals. A sef_cb_signal_handler() callback is available for system processes to handle every received signal. A sef_cb_signal_manager() callback is used by signal managers to process system signals on behalf of the kernel. - Fixed a few bugs with memory mapping and DS. VM CHANGES: - Page faults and memory requests coming from the kernel are now implemented using signals. - Added a new VM call to swap two process slots and implement live update. - The call is used by RS at update time and in turn invokes the kernel call sys_update(). RS CHANGES: - RS has been reworked with a better functional decomposition. - Better kernel call masks. com.h now defines the set of very basic kernel calls every system service is allowed to use. This makes system.conf simpler and easier to maintain. In addition, this guarantees a higher level of isolation for system libraries that use one or more kernel calls internally (e.g. printf). - RS is the default signal manager for system processes. By default, RS intercepts every signal delivered to every system process. This makes crash recovery possible before bringing PM and friends in the loop. - RS now supports fast rollback when something goes wrong while initializing the new version during a live update. - Live update is now implemented by keeping the two versions side-by-side and swapping the process slots when the old version is ready to update. - Crash recovery is now implemented by keeping the two versions side-by-side and cleaning up the old version only when the recovery process is complete. DS CHANGES: - Fixed a bug when the process doing ds_publish() or ds_delete() is not known by DS. - Fixed the completely broken support for strings. String publishing is now implemented in the system library and simply wraps publishing of memory ranges. Ideally, we should adopt a similar approach for other data types as well. - Test suite fixed. DRIVER CHANGES: - The hello driver has been added to the Minix distribution to demonstrate basic live update and crash recovery functionalities. - Other drivers have been adapted to conform the new SEF interface.
2010-03-17 02:15:29 +01:00
{
port_t port;
fxp_t *fp;
/* Only check for termination signal, ignore anything else. */
if (signo != SIGTERM) return;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
fp= fxp_state;
if (fp->fxp_mode == FM_ENABLED && (fp->fxp_flags & FF_ENABLED)) {
New RS and new signal handling for system processes. UPDATING INFO: 20100317: /usr/src/etc/system.conf updated to ignore default kernel calls: copy it (or merge it) to /etc/system.conf. The hello driver (/dev/hello) added to the distribution: # cd /usr/src/commands/scripts && make clean install # cd /dev && MAKEDEV hello KERNEL CHANGES: - Generic signal handling support. The kernel no longer assumes PM as a signal manager for every process. The signal manager of a given process can now be specified in its privilege slot. When a signal has to be delivered, the kernel performs the lookup and forwards the signal to the appropriate signal manager. PM is the default signal manager for user processes, RS is the default signal manager for system processes. To enable ptrace()ing for system processes, it is sufficient to change the default signal manager to PM. This will temporarily disable crash recovery, though. - sys_exit() is now split into sys_exit() (i.e. exit() for system processes, which generates a self-termination signal), and sys_clear() (i.e. used by PM to ask the kernel to clear a process slot when a process exits). - Added a new kernel call (i.e. sys_update()) to swap two process slots and implement live update. PM CHANGES: - Posix signal handling is no longer allowed for system processes. System signals are split into two fixed categories: termination and non-termination signals. When a non-termination signaled is processed, PM transforms the signal into an IPC message and delivers the message to the system process. When a termination signal is processed, PM terminates the process. - PM no longer assumes itself as the signal manager for system processes. It now makes sure that every system signal goes through the kernel before being actually processes. The kernel will then dispatch the signal to the appropriate signal manager which may or may not be PM. SYSLIB CHANGES: - Simplified SEF init and LU callbacks. - Added additional predefined SEF callbacks to debug crash recovery and live update. - Fixed a temporary ack in the SEF init protocol. SEF init reply is now completely synchronous. - Added SEF signal event type to provide a uniform interface for system processes to deal with signals. A sef_cb_signal_handler() callback is available for system processes to handle every received signal. A sef_cb_signal_manager() callback is used by signal managers to process system signals on behalf of the kernel. - Fixed a few bugs with memory mapping and DS. VM CHANGES: - Page faults and memory requests coming from the kernel are now implemented using signals. - Added a new VM call to swap two process slots and implement live update. - The call is used by RS at update time and in turn invokes the kernel call sys_update(). RS CHANGES: - RS has been reworked with a better functional decomposition. - Better kernel call masks. com.h now defines the set of very basic kernel calls every system service is allowed to use. This makes system.conf simpler and easier to maintain. In addition, this guarantees a higher level of isolation for system libraries that use one or more kernel calls internally (e.g. printf). - RS is the default signal manager for system processes. By default, RS intercepts every signal delivered to every system process. This makes crash recovery possible before bringing PM and friends in the loop. - RS now supports fast rollback when something goes wrong while initializing the new version during a live update. - Live update is now implemented by keeping the two versions side-by-side and swapping the process slots when the old version is ready to update. - Crash recovery is now implemented by keeping the two versions side-by-side and cleaning up the old version only when the recovery process is complete. DS CHANGES: - Fixed a bug when the process doing ds_publish() or ds_delete() is not known by DS. - Fixed the completely broken support for strings. String publishing is now implemented in the system library and simply wraps publishing of memory ranges. Ideally, we should adopt a similar approach for other data types as well. - Test suite fixed. DRIVER CHANGES: - The hello driver has been added to the Minix distribution to demonstrate basic live update and crash recovery functionalities. - Other drivers have been adapted to conform the new SEF interface.
2010-03-17 02:15:29 +01:00
port= fp->fxp_base_port;
/* Reset device */
if (debug)
printf("%s: resetting device\n", fp->fxp_name);
fxp_outl(port, CSR_PORT, CP_CMD_SOFT_RESET);
}
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
New RS and new signal handling for system processes. UPDATING INFO: 20100317: /usr/src/etc/system.conf updated to ignore default kernel calls: copy it (or merge it) to /etc/system.conf. The hello driver (/dev/hello) added to the distribution: # cd /usr/src/commands/scripts && make clean install # cd /dev && MAKEDEV hello KERNEL CHANGES: - Generic signal handling support. The kernel no longer assumes PM as a signal manager for every process. The signal manager of a given process can now be specified in its privilege slot. When a signal has to be delivered, the kernel performs the lookup and forwards the signal to the appropriate signal manager. PM is the default signal manager for user processes, RS is the default signal manager for system processes. To enable ptrace()ing for system processes, it is sufficient to change the default signal manager to PM. This will temporarily disable crash recovery, though. - sys_exit() is now split into sys_exit() (i.e. exit() for system processes, which generates a self-termination signal), and sys_clear() (i.e. used by PM to ask the kernel to clear a process slot when a process exits). - Added a new kernel call (i.e. sys_update()) to swap two process slots and implement live update. PM CHANGES: - Posix signal handling is no longer allowed for system processes. System signals are split into two fixed categories: termination and non-termination signals. When a non-termination signaled is processed, PM transforms the signal into an IPC message and delivers the message to the system process. When a termination signal is processed, PM terminates the process. - PM no longer assumes itself as the signal manager for system processes. It now makes sure that every system signal goes through the kernel before being actually processes. The kernel will then dispatch the signal to the appropriate signal manager which may or may not be PM. SYSLIB CHANGES: - Simplified SEF init and LU callbacks. - Added additional predefined SEF callbacks to debug crash recovery and live update. - Fixed a temporary ack in the SEF init protocol. SEF init reply is now completely synchronous. - Added SEF signal event type to provide a uniform interface for system processes to deal with signals. A sef_cb_signal_handler() callback is available for system processes to handle every received signal. A sef_cb_signal_manager() callback is used by signal managers to process system signals on behalf of the kernel. - Fixed a few bugs with memory mapping and DS. VM CHANGES: - Page faults and memory requests coming from the kernel are now implemented using signals. - Added a new VM call to swap two process slots and implement live update. - The call is used by RS at update time and in turn invokes the kernel call sys_update(). RS CHANGES: - RS has been reworked with a better functional decomposition. - Better kernel call masks. com.h now defines the set of very basic kernel calls every system service is allowed to use. This makes system.conf simpler and easier to maintain. In addition, this guarantees a higher level of isolation for system libraries that use one or more kernel calls internally (e.g. printf). - RS is the default signal manager for system processes. By default, RS intercepts every signal delivered to every system process. This makes crash recovery possible before bringing PM and friends in the loop. - RS now supports fast rollback when something goes wrong while initializing the new version during a live update. - Live update is now implemented by keeping the two versions side-by-side and swapping the process slots when the old version is ready to update. - Crash recovery is now implemented by keeping the two versions side-by-side and cleaning up the old version only when the recovery process is complete. DS CHANGES: - Fixed a bug when the process doing ds_publish() or ds_delete() is not known by DS. - Fixed the completely broken support for strings. String publishing is now implemented in the system library and simply wraps publishing of memory ranges. Ideally, we should adopt a similar approach for other data types as well. - Test suite fixed. DRIVER CHANGES: - The hello driver has been added to the Minix distribution to demonstrate basic live update and crash recovery functionalities. - Other drivers have been adapted to conform the new SEF interface.
2010-03-17 02:15:29 +01:00
exit(0);
}
/*===========================================================================*
* fxp_init *
*===========================================================================*/
static void fxp_init(mp)
message *mp;
{
static int first_time= 1;
fxp_t *fp;
message reply_mess;
if (first_time)
{
first_time= 0;
fxp_pci_conf(); /* Configure PCI devices. */
2010-07-09 14:58:18 +02:00
init_timer(&fxp_watchdog);
set_timer(&fxp_watchdog, system_hz, fxp_watchdog_f, 0);
}
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
fp= fxp_state;
if (fp->fxp_mode == FM_DISABLED)
{
/* This is the default, try to (re)locate the device. */
fxp_conf_hw(fp);
if (fp->fxp_mode == FM_DISABLED)
{
/* Probe failed, or the device is configured off. */
2006-07-10 14:43:38 +02:00
reply_mess.m_type= DL_CONF_REPLY;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
reply_mess.DL_STAT= ENXIO;
mess_reply(mp, &reply_mess);
return;
}
if (fp->fxp_mode == FM_ENABLED)
fxp_init_hw(fp);
fxp_report_link(fp);
}
assert(fp->fxp_mode == FM_ENABLED);
assert(fp->fxp_flags & FF_ENABLED);
fp->fxp_flags &= ~(FF_PROMISC | FF_MULTI | FF_BROAD);
if (mp->DL_MODE & DL_PROMISC_REQ)
fp->fxp_flags |= FF_PROMISC;
if (mp->DL_MODE & DL_MULTI_REQ)
fp->fxp_flags |= FF_MULTI;
if (mp->DL_MODE & DL_BROAD_REQ)
fp->fxp_flags |= FF_BROAD;
fxp_rec_mode(fp);
2006-07-10 14:43:38 +02:00
reply_mess.m_type = DL_CONF_REPLY;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
reply_mess.DL_STAT = OK;
*(ether_addr_t *) reply_mess.DL_HWADDR = fp->fxp_address;
mess_reply(mp, &reply_mess);
}
/*===========================================================================*
* fxp_pci_conf *
*===========================================================================*/
static void fxp_pci_conf()
{
fxp_t *fp;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
fp= fxp_state;
strlcpy(fp->fxp_name, "fxp#0", sizeof(fp->fxp_name));
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
fp->fxp_name[4] += fxp_instance;
fp->fxp_seen= FALSE;
fp->fxp_features= FFE_NONE;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
pci_init();
if (fxp_probe(fp, fxp_instance))
fp->fxp_seen= TRUE;
}
/*===========================================================================*
* fxp_probe *
*===========================================================================*/
static int fxp_probe(fxp_t *fp, int skip)
{
int r, devind;
u16_t vid, did;
u32_t bar;
u8_t ilr, rev;
char *str;
#if VERBOSE
char *dname;
#endif
r= pci_first_dev(&devind, &vid, &did);
if (r == 0)
return FALSE;
while (skip--)
{
r= pci_next_dev(&devind, &vid, &did);
if (!r)
return FALSE;
}
#if VERBOSE
dname= pci_dev_name(vid, did);
if (!dname)
dname= "unknown device";
printf("%s: %s (%04x/%04x) at %s\n",
fp->fxp_name, dname, vid, did, pci_slot_name(devind));
#endif
pci_reserve(devind);
bar= pci_attr_r32(devind, PCI_BAR_2) & 0xffffffe0;
if (bar < 0x400) {
panic("fxp_probe: base address is not properly configured");
}
fp->fxp_base_port= bar;
ilr= pci_attr_r8(devind, PCI_ILR);
fp->fxp_irq= ilr;
if (debug)
{
printf("%s: using I/O address 0x%lx, IRQ %d\n",
fp->fxp_name, (unsigned long)bar, ilr);
}
rev= pci_attr_r8(devind, PCI_REV);
str= NULL;
fp->fxp_type= FT_UNKNOWN;
switch(rev)
{
case FXP_REV_82557A: str= "82557A"; /* 0x01 */
fp->fxp_type= FT_82557;
break;
case FXP_REV_82557B: str= "82557B"; break; /* 0x02 */
case FXP_REV_82557C: str= "82557C"; break; /* 0x03 */
case FXP_REV_82558A: str= "82558A"; /* 0x04 */
fp->fxp_type= FT_82558A;
break;
case FXP_REV_82558B: str= "82558B"; /* 0x05 */
fp->fxp_type= FT_82559;
break;
case FXP_REV_82559A: str= "82559A"; break; /* 0x06 */
case FXP_REV_82559B: str= "82559B"; break; /* 0x07 */
case FXP_REV_82559C: str= "82559C"; /* 0x08 */
fp->fxp_type= FT_82559;
break;
2005-12-02 15:45:10 +01:00
case FXP_REV_82559ERA: str= "82559ER-A"; /* 0x09 */
fp->fxp_type= FT_82559;
break;
case FXP_REV_82550_1: str= "82550(1)"; /* 0x0C */
fp->fxp_type= FT_82559;
break;
case FXP_REV_82550_2: str= "82550(2)"; /* 0x0D */
fp->fxp_type= FT_82559;
break;
case FXP_REV_82550_3: str= "82550(3)"; /* 0x0E */
fp->fxp_type= FT_82559;
break;
case FXP_REV_82551_1: str= "82551(1)"; /* 0x0F */
fp->fxp_type= FT_82559;
break;
case FXP_REV_82551_2: str= "82551(2)"; /* 0x10 */
fp->fxp_type= FT_82559;
break;
case FXP_REV_82801CAM: str= "82801CAM"; /* 0x42 */
fp->fxp_type= FT_82801;
break;
2009-08-26 23:01:46 +02:00
case FXP_REV_82801DB: str= "82801DB"; /* 0x81 */
fp->fxp_type= FT_82801;
break;
case FXP_REV_82550_4: str= "82550(4)"; /* 0x83 */
fp->fxp_type= FT_82559;
break;
}
#if VERBOSE
if (str)
printf("%s: device revision: %s\n", fp->fxp_name, str);
else
printf("%s: unknown revision: 0x%x\n", fp->fxp_name, rev);
#endif
if (fp->fxp_type == FT_UNKNOWN)
{
printf("fxp_probe: device is not supported by this driver\n");
return FALSE;
}
return TRUE;
}
/*===========================================================================*
* fxp_conf_hw *
*===========================================================================*/
static void fxp_conf_hw(fxp_t *fp)
{
#if VERBOSE
int i;
#endif
fp->fxp_mode= FM_DISABLED; /* Superfluous */
if (!fp->fxp_seen)
return;
/* PCI device is present */
fp->fxp_mode= FM_ENABLED;
fp->fxp_flags= FF_EMPTY;
fp->fxp_got_int= 0;
fp->fxp_send_int= 0;
fp->fxp_ee_addrlen= 0; /* Unknown */
fp->fxp_need_reset= 0;
fp->fxp_report_link= 0;
fp->fxp_link_up= -1; /* Unknown */
fp->fxp_mii_busy= 0;
fp->fxp_read_s= 0;
fp->fxp_rx_need_restart= 0;
fp->fxp_need_conf= 0;
fp->fxp_tx_head= 0;
fp->fxp_tx_tail= 0;
fp->fxp_tx_alive= 0;
fp->fxp_tx_threshold= TXTT_MIN;
/* Try to come up with a sensible configuration for the current
* device. Unfortunately every device is different, defaults are
* not always zero, and some fields are re-used with a completely
* different interpretation. We start out with a sensible default
* for all devices and then add device specific changes.
*/
fp->fxp_conf_bytes[0]= CC_BYTES_NR;
fp->fxp_conf_bytes[1]= CTL_DEFAULT | CRL_DEFAULT;
fp->fxp_conf_bytes[2]= CAI_DEFAULT;
fp->fxp_conf_bytes[3]= 0;
fp->fxp_conf_bytes[4]= 0;
fp->fxp_conf_bytes[5]= 0;
fp->fxp_conf_bytes[6]= CCB6_ESC | CCB6_ETCB | CCB6_RES;
fp->fxp_conf_bytes[7]= CUR_1;
fp->fxp_conf_bytes[8]= CCB8_503_MII;
fp->fxp_conf_bytes[9]= 0;
fp->fxp_conf_bytes[10]= CLB_NORMAL | CPAL_DEFAULT | CCB10_NSAI |
CCB10_RES1;
fp->fxp_conf_bytes[11]= 0;
fp->fxp_conf_bytes[12]= CIS_DEFAULT;
fp->fxp_conf_bytes[13]= CCB13_DEFAULT;
fp->fxp_conf_bytes[14]= CCB14_DEFAULT;
fp->fxp_conf_bytes[15]= CCB15_RES1 | CCB15_RES2;
fp->fxp_conf_bytes[16]= CCB16_DEFAULT;
fp->fxp_conf_bytes[17]= CCB17_DEFAULT;
fp->fxp_conf_bytes[18]= CCB18_RES1 | CCB18_PFCT | CCB18_PE;
fp->fxp_conf_bytes[19]= CCB19_FDPE;
fp->fxp_conf_bytes[20]= CCB20_PFCL | CCB20_RES1;
fp->fxp_conf_bytes[21]= CCB21_RES21;
#if VERBOSE
for (i= 0; i<CC_BYTES_NR; i++)
printf("%d: %0x, ", i, fp->fxp_conf_bytes[i]);
printf("\n");
#endif
switch(fp->fxp_type)
{
case FT_82557:
break;
case FT_82558A:
case FT_82559:
2009-08-26 23:01:46 +02:00
case FT_82801:
fp->fxp_conf_bytes[18] |= CCB18_LROK;
2009-08-26 23:01:46 +02:00
if (fp->fxp_type == FT_82801)
{
fp->fxp_conf_bytes[6] = 0xba; /* ctrl 1 */
fp->fxp_conf_bytes[15] = 0x48; /* promiscuous */
fp->fxp_conf_bytes[21] = 0x05; /* mc_all */
}
break;
default:
panic("fxp_conf_hw: bad device type: %d", fp->fxp_type);
}
/* Assume an 82555 (compatible) PHY. The should be changed for
* 82557 NICs with different PHYs
*/
fp->fxp_ms_regs = 0; /* No master/slave registers. */
#if VERBOSE
for (i= 0; i<CC_BYTES_NR; i++)
printf("%d: %0x, ", i, fp->fxp_conf_bytes[i]);
printf("\n");
#endif
}
/*===========================================================================*
* fxp_init_hw *
*===========================================================================*/
static void fxp_init_hw(fp)
fxp_t *fp;
{
int i, r, isr;
port_t port;
2012-03-05 00:11:41 +01:00
phys_bytes bus_addr;
port= fp->fxp_base_port;
fxp_init_buf(fp);
fp->fxp_flags = FF_EMPTY;
fp->fxp_flags |= FF_ENABLED;
/* Set the interrupt handler and policy. Do not automatically
* reenable interrupts. Return the IRQ line number on interrupts.
*/
fp->fxp_hook = fp->fxp_irq;
r= sys_irqsetpolicy(fp->fxp_irq, 0, &fp->fxp_hook);
if (r != OK)
panic("sys_irqsetpolicy failed: %d", r);
fxp_reset_hw(fp);
r= sys_irqenable(&fp->fxp_hook);
if (r != OK)
panic("sys_irqenable failed: %d", r);
/* Reset PHY? */
fxp_do_conf(fp);
/* Set pointer to statistical counters */
r= sys_umap(SELF, VM_D, (vir_bytes)&fp->fxp_stat, sizeof(fp->fxp_stat),
&bus_addr);
if (r != OK)
panic("sys_umap failed: %d", r);
fxp_cu_ptr_cmd(fp, SC_CU_LOAD_DCA, bus_addr, TRUE /* check idle */);
/* Ack previous interrupts */
isr= fxp_inb(port, SCB_INT_STAT);
fxp_outb(port, SCB_INT_STAT, isr);
/* Enable interrupts */
fxp_outb(port, SCB_INT_MASK, 0);
fxp_ru_ptr_cmd(fp, SC_RU_START, fp->fxp_rx_busaddr,
TRUE /* check idle */);
fxp_confaddr(fp);
if (debug)
{
printf("%s: Ethernet address ", fp->fxp_name);
for (i= 0; i < 6; i++)
{
printf("%x%c", fp->fxp_address.ea_addr[i],
i < 5 ? ':' : '\n');
}
}
}
/*===========================================================================*
* fxp_init_buf *
*===========================================================================*/
static void fxp_init_buf(fp)
fxp_t *fp;
{
size_t rx_totbufsize, tx_totbufsize, tot_bufsize, alloc_bufsize;
char *alloc_buf;
2012-03-05 00:11:41 +01:00
phys_bytes buf, bus_addr;
int i, r;
struct rfd *rfdp;
struct tx *txp;
phys_bytes ph;
fp->fxp_rx_nbuf= N_RX_BUF;
rx_totbufsize= fp->fxp_rx_nbuf * sizeof(struct rfd);
fp->fxp_rx_bufsize= rx_totbufsize;
fp->fxp_tx_nbuf= N_TX_BUF;
tx_totbufsize= fp->fxp_tx_nbuf * sizeof(struct tx);
fp->fxp_tx_bufsize= tx_totbufsize;
tot_bufsize= sizeof(*tmpbufp) + tx_totbufsize + rx_totbufsize;
if (tot_bufsize % 4096)
tot_bufsize += 4096 - (tot_bufsize % 4096);
alloc_bufsize= tot_bufsize;
alloc_buf= alloc_contig(alloc_bufsize, AC_ALIGN4K, &ph);
if (alloc_buf == NULL) {
panic("fxp_init_buf: unable to alloc_contig size: %d", alloc_bufsize);
}
buf= (phys_bytes)alloc_buf;
tell_dev((vir_bytes)buf, tot_bufsize, 0, 0, 0);
tmpbufp= (union tmpbuf *)buf;
fp->fxp_rx_buf= (struct rfd *)&tmpbufp[1];
r= sys_umap(SELF, VM_D, (vir_bytes)fp->fxp_rx_buf, rx_totbufsize,
2012-03-05 00:11:41 +01:00
&bus_addr);
if (r != OK)
panic("sys_umap failed: %d", r);
2012-03-05 00:11:41 +01:00
fp->fxp_rx_busaddr= bus_addr;
#if 0
printf("fxp_init_buf: got phys 0x%x for vir 0x%x\n",
fp->fxp_rx_busaddr, fp->fxp_rx_buf);
#endif
for (i= 0, rfdp= fp->fxp_rx_buf; i<fp->fxp_rx_nbuf; i++, rfdp++)
{
rfdp->rfd_status= 0;
rfdp->rfd_command= 0;
if (i != fp->fxp_rx_nbuf-1)
{
r= sys_umap(SELF, VM_D, (vir_bytes)&rfdp[1],
2012-03-05 00:11:41 +01:00
sizeof(rfdp[1]), &bus_addr);
if (r != OK)
panic("sys_umap failed: %d", r);
2012-03-05 00:11:41 +01:00
rfdp->rfd_linkaddr= bus_addr;
}
else
{
rfdp->rfd_linkaddr= fp->fxp_rx_busaddr;
rfdp->rfd_command |= RFDC_EL;
}
rfdp->rfd_reserved= 0;
rfdp->rfd_res= 0;
rfdp->rfd_size= sizeof(rfdp->rfd_buf);
}
fp->fxp_rx_head= 0;
fp->fxp_tx_buf= (struct tx *)((char *)fp->fxp_rx_buf+rx_totbufsize);
r= sys_umap(SELF, VM_D, (vir_bytes)fp->fxp_tx_buf,
(phys_bytes)tx_totbufsize, &fp->fxp_tx_busaddr);
if (r != OK)
panic("sys_umap failed: %d", r);
for (i= 0, txp= fp->fxp_tx_buf; i<fp->fxp_tx_nbuf; i++, txp++)
{
txp->tx_status= 0;
txp->tx_command= TXC_EL | CBL_NOP; /* Just in case */
if (i != fp->fxp_tx_nbuf-1)
{
r= sys_umap(SELF, VM_D, (vir_bytes)&txp[1],
2012-03-05 00:11:41 +01:00
(phys_bytes)sizeof(txp[1]), &bus_addr);
if (r != OK)
panic("sys_umap failed: %d", r);
2012-03-05 00:11:41 +01:00
txp->tx_linkaddr= bus_addr;
}
else
{
txp->tx_linkaddr= fp->fxp_tx_busaddr;
}
txp->tx_tbda= TX_TBDA_NIL;
txp->tx_size= 0;
txp->tx_tthresh= fp->fxp_tx_threshold;
txp->tx_ntbd= 0;
}
fp->fxp_tx_idle= 1;
}
/*===========================================================================*
* fxp_reset_hw *
*===========================================================================*/
static void fxp_reset_hw(fp)
fxp_t *fp;
{
/* Inline the function in init? */
port_t port;
port= fp->fxp_base_port;
/* Reset device */
fxp_outl(port, CSR_PORT, CP_CMD_SOFT_RESET);
tickdelay(micros_to_ticks(CSR_PORT_RESET_DELAY));
/* Disable interrupts */
fxp_outb(port, SCB_INT_MASK, SIM_M);
/* Set CU base to zero */
fxp_cu_ptr_cmd(fp, SC_CU_LOAD_BASE, 0, TRUE /* check idle */);
/* Set RU base to zero */
fxp_ru_ptr_cmd(fp, SC_RU_LOAD_BASE, 0, TRUE /* check idle */);
}
/*===========================================================================*
* fxp_confaddr *
*===========================================================================*/
static void fxp_confaddr(fxp_t *fp)
{
static char eakey[]= FXP_ENVVAR "#_EA";
static char eafmt[]= "x:x:x:x:x:x";
int i, r;
2012-03-05 00:11:41 +01:00
phys_bytes bus_addr;
long v;
/* User defined ethernet address? */
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
eakey[sizeof(FXP_ENVVAR)-1]= '0' + fxp_instance;
for (i= 0; i < 6; i++)
{
if (env_parse(eakey, eafmt, i, &v, 0x00L, 0xFFL) != EP_SET)
break;
fp->fxp_address.ea_addr[i]= v;
}
if (i != 0 && i != 6) env_panic(eakey); /* It's all or nothing */
if (i == 0)
{
/* Get ethernet address from EEPROM */
for (i= 0; i<3; i++)
{
v= eeprom_read(fp, i);
fp->fxp_address.ea_addr[i*2]= (v & 0xff);
fp->fxp_address.ea_addr[i*2+1]= ((v >> 8) & 0xff);
}
}
/* Tell NIC about ethernet address */
tmpbufp->ias.ias_status= 0;
tmpbufp->ias.ias_command= CBL_C_EL | CBL_AIS;
tmpbufp->ias.ias_linkaddr= 0;
memcpy(tmpbufp->ias.ias_ethaddr, fp->fxp_address.ea_addr,
sizeof(tmpbufp->ias.ias_ethaddr));
r= sys_umap(SELF, VM_D, (vir_bytes)&tmpbufp->ias,
(phys_bytes)sizeof(tmpbufp->ias), &bus_addr);
if (r != OK)
panic("sys_umap failed: %d", r);
fxp_cu_ptr_cmd(fp, SC_CU_START, bus_addr, TRUE /* check idle */);
/* Wait for CU command to complete */
SPIN_UNTIL(tmpbufp->ias.ias_status & CBL_F_C, 1000);
if (!(tmpbufp->ias.ias_status & CBL_F_C))
panic("fxp_confaddr: CU command failed to complete");
if (!(tmpbufp->ias.ias_status & CBL_F_OK))
panic("fxp_confaddr: CU command failed");
#if VERBOSE
printf("%s: hardware ethernet address: ", fp->fxp_name);
for (i= 0; i<6; i++)
{
printf("%02x%s", fp->fxp_address.ea_addr[i],
i < 5 ? ":" : "");
}
printf("\n");
#endif
}
/*===========================================================================*
* fxp_rec_mode *
*===========================================================================*/
static void fxp_rec_mode(fp)
fxp_t *fp;
{
fp->fxp_conf_bytes[0]= CC_BYTES_NR; /* Just to be sure */
fp->fxp_conf_bytes[15] &= ~(CCB15_BD|CCB15_PM);
fp->fxp_conf_bytes[21] &= ~CCB21_MA;
if (fp->fxp_flags & FF_PROMISC)
fp->fxp_conf_bytes[15] |= CCB15_PM;
if (fp->fxp_flags & FF_MULTI)
fp->fxp_conf_bytes[21] |= CCB21_MA;
if (!(fp->fxp_flags & (FF_BROAD|FF_MULTI|FF_PROMISC)))
fp->fxp_conf_bytes[15] |= CCB15_BD;
/* Queue request if not idle */
if (fp->fxp_tx_idle)
{
fxp_do_conf(fp);
}
else
{
printf("fxp_rec_mode: setting fxp_need_conf\n");
fp->fxp_need_conf= TRUE;
}
}
2006-07-10 14:43:38 +02:00
/*===========================================================================*
* fxp_writev_s *
*===========================================================================*/
static void fxp_writev_s(const message *mp, int from_int)
2006-07-10 14:43:38 +02:00
{
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
endpoint_t iov_endpt;
2006-07-10 14:43:38 +02:00
cp_grant_id_t iov_grant;
vir_bytes iov_offset;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
int i, j, n, o, r, s, count, size, prev_head;
int fxp_tx_nbuf, fxp_tx_head;
2006-07-10 14:43:38 +02:00
u16_t tx_command;
fxp_t *fp;
iovec_s_t *iovp;
struct tx *txp, *prev_txp;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
fp= fxp_state;
2006-07-10 14:43:38 +02:00
count = mp->DL_COUNT;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
fp->fxp_client= mp->m_source;
2006-07-10 14:43:38 +02:00
assert(fp->fxp_mode == FM_ENABLED);
assert(fp->fxp_flags & FF_ENABLED);
if (from_int)
{
assert(fp->fxp_flags & FF_SEND_AVAIL);
fp->fxp_flags &= ~FF_SEND_AVAIL;
fp->fxp_tx_alive= TRUE;
}
if (fp->fxp_tx_idle)
{
txp= fp->fxp_tx_buf;
fxp_tx_head= 0; /* lint */
prev_txp= NULL; /* lint */
}
else
{
fxp_tx_nbuf= fp->fxp_tx_nbuf;
prev_head= fp->fxp_tx_head;
fxp_tx_head= prev_head+1;
if (fxp_tx_head == fxp_tx_nbuf)
fxp_tx_head= 0;
assert(fxp_tx_head < fxp_tx_nbuf);
if (fxp_tx_head == fp->fxp_tx_tail)
{
/* Send queue is full */
assert(!(fp->fxp_flags & FF_SEND_AVAIL));
fp->fxp_flags |= FF_SEND_AVAIL;
goto suspend;
}
prev_txp= &fp->fxp_tx_buf[prev_head];
txp= &fp->fxp_tx_buf[fxp_tx_head];
}
assert(!(fp->fxp_flags & FF_SEND_AVAIL));
assert(!(fp->fxp_flags & FF_PACK_SENT));
iov_endpt= mp->m_source;
2006-07-10 14:43:38 +02:00
iov_grant= mp->DL_GRANT;
size= 0;
o= 0;
iov_offset= 0;
for (i= 0; i<count; i += IOVEC_NR,
iov_offset += IOVEC_NR * sizeof(fp->fxp_iovec_s[0]))
{
n= IOVEC_NR;
if (i+n > count)
n= count-i;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
r= sys_safecopyfrom(iov_endpt, iov_grant, iov_offset,
2006-07-10 14:43:38 +02:00
(vir_bytes)fp->fxp_iovec_s,
n * sizeof(fp->fxp_iovec_s[0]));
2006-07-10 14:43:38 +02:00
if (r != OK)
panic("fxp_writev: sys_safecopyfrom failed: %d", r);
2006-07-10 14:43:38 +02:00
for (j= 0, iovp= fp->fxp_iovec_s; j<n; j++, iovp++)
{
s= iovp->iov_size;
if (size + s > ETH_MAX_PACK_SIZE_TAGGED) {
panic("fxp_writev: invalid packet size: %d", size + s);
2006-07-10 14:43:38 +02:00
}
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
r= sys_safecopyfrom(iov_endpt, iovp->iov_grant,
0, (vir_bytes)(txp->tx_buf+o), s);
if (r != OK) {
panic("fxp_writev_s: sys_safecopyfrom failed: %d", r);
2006-07-10 14:43:38 +02:00
}
size += s;
o += s;
}
}
if (size < ETH_MIN_PACK_SIZE)
panic("fxp_writev: invalid packet size: %d", size);
2006-07-10 14:43:38 +02:00
txp->tx_status= 0;
txp->tx_command= TXC_EL | CBL_XMIT;
txp->tx_tbda= TX_TBDA_NIL;
txp->tx_size= TXSZ_EOF | size;
txp->tx_tthresh= fp->fxp_tx_threshold;
txp->tx_ntbd= 0;
if (fp->fxp_tx_idle)
{
fp->fxp_tx_idle= 0;
fp->fxp_tx_head= fp->fxp_tx_tail= 0;
fxp_cu_ptr_cmd(fp, SC_CU_START, fp->fxp_tx_busaddr,
TRUE /* check idle */);
}
else
{
/* Link new request in transmit list */
tx_command= prev_txp->tx_command;
assert(tx_command == (TXC_EL | CBL_XMIT));
prev_txp->tx_command= CBL_XMIT;
fp->fxp_tx_head= fxp_tx_head;
}
fp->fxp_flags |= FF_PACK_SENT;
/* If the interrupt handler called, don't send a reply. The reply
* will be sent after all interrupts are handled.
*/
if (from_int)
return;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
reply(fp);
2006-07-10 14:43:38 +02:00
return;
suspend:
if (from_int)
panic("fxp: should not be sending");
2006-07-10 14:43:38 +02:00
fp->fxp_tx_mess= *mp;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
reply(fp);
}
2006-07-10 14:43:38 +02:00
/*===========================================================================*
* fxp_readv_s *
*===========================================================================*/
static void fxp_readv_s(mp, from_int)
message *mp;
int from_int;
{
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
int i, j, n, o, r, s, count, size, fxp_rx_head, fxp_rx_nbuf;
endpoint_t iov_endpt;
2006-07-10 14:43:38 +02:00
cp_grant_id_t iov_grant;
port_t port;
unsigned packlen;
vir_bytes iov_offset;
u16_t rfd_status;
u16_t rfd_res;
u8_t scb_status;
fxp_t *fp;
iovec_s_t *iovp;
struct rfd *rfdp, *prev_rfdp;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
fp= fxp_state;
2006-07-10 14:43:38 +02:00
count = mp->DL_COUNT;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
fp->fxp_client= mp->m_source;
2006-07-10 14:43:38 +02:00
assert(fp->fxp_mode == FM_ENABLED);
assert(fp->fxp_flags & FF_ENABLED);
port= fp->fxp_base_port;
fxp_rx_head= fp->fxp_rx_head;
rfdp= &fp->fxp_rx_buf[fxp_rx_head];
rfd_status= rfdp->rfd_status;
if (!(rfd_status & RFDS_C))
{
/* Receive buffer is empty, suspend */
goto suspend;
}
if (!(rfd_status & RFDS_OK))
2006-07-10 14:43:38 +02:00
{
/* Not OK? What happened? */
assert(0);
}
else
{
assert(!(rfd_status & (RFDS_CRCERR | RFDS_ALIGNERR |
RFDS_OUTOFBUF | RFDS_DMAOVR | RFDS_TOOSHORT |
RFDS_RXERR)));
}
rfd_res= rfdp->rfd_res;
assert(rfd_res & RFDR_EOF);
assert(rfd_res & RFDR_F);
packlen= rfd_res & RFDSZ_SIZE;
iov_endpt = mp->m_source;
2006-07-10 14:43:38 +02:00
iov_grant = mp->DL_GRANT;
size= 0;
o= 0;
iov_offset= 0;
for (i= 0; i<count; i += IOVEC_NR,
iov_offset += IOVEC_NR * sizeof(fp->fxp_iovec_s[0]))
{
n= IOVEC_NR;
if (i+n > count)
n= count-i;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
r= sys_safecopyfrom(iov_endpt, iov_grant, iov_offset,
2006-07-10 14:43:38 +02:00
(vir_bytes)fp->fxp_iovec_s,
n * sizeof(fp->fxp_iovec_s[0]));
2006-07-10 14:43:38 +02:00
if (r != OK)
panic("fxp_readv_s: sys_safecopyfrom failed: %d", r);
2006-07-10 14:43:38 +02:00
for (j= 0, iovp= fp->fxp_iovec_s; j<n; j++, iovp++)
{
s= iovp->iov_size;
if (size + s > packlen)
{
assert(packlen > size);
s= packlen-size;
}
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
r= sys_safecopyto(iov_endpt, iovp->iov_grant,
0, (vir_bytes)(rfdp->rfd_buf+o), s);
2006-07-10 14:43:38 +02:00
if (r != OK)
{
panic("fxp_readv: sys_safecopyto failed: %d", r);
2006-07-10 14:43:38 +02:00
}
size += s;
if (size == packlen)
break;
o += s;
}
if (size == packlen)
break;
}
if (size < packlen)
{
assert(0);
}
fp->fxp_read_s= packlen;
fp->fxp_flags= (fp->fxp_flags & ~FF_READING) | FF_PACK_RECV;
/* Re-init the current buffer */
rfdp->rfd_status= 0;
rfdp->rfd_command= RFDC_EL;
rfdp->rfd_reserved= 0;
rfdp->rfd_res= 0;
rfdp->rfd_size= sizeof(rfdp->rfd_buf);
fxp_rx_nbuf= fp->fxp_rx_nbuf;
if (fxp_rx_head == 0)
{
prev_rfdp= &fp->fxp_rx_buf[fxp_rx_nbuf-1];
}
else
prev_rfdp= &rfdp[-1];
assert(prev_rfdp->rfd_command & RFDC_EL);
prev_rfdp->rfd_command &= ~RFDC_EL;
fxp_rx_head++;
if (fxp_rx_head == fxp_rx_nbuf)
fxp_rx_head= 0;
assert(fxp_rx_head < fxp_rx_nbuf);
fp->fxp_rx_head= fxp_rx_head;
if (!from_int)
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
reply(fp);
2006-07-10 14:43:38 +02:00
return;
suspend:
if (fp->fxp_rx_need_restart)
{
fp->fxp_rx_need_restart= 0;
/* Check the status of the RU */
scb_status= fxp_inb(port, SCB_STATUS);
if ((scb_status & SS_RUS_MASK) != SS_RU_NORES)
{
/* Race condition? */
printf("fxp_readv: restart race: 0x%x\n",
scb_status);
assert((scb_status & SS_RUS_MASK) == SS_RU_READY);
}
else
{
fxp_restart_ru(fp);
}
}
if (from_int)
{
assert(fp->fxp_flags & FF_READING);
/* No need to store any state */
return;
}
fp->fxp_rx_mess= *mp;
assert(!(fp->fxp_flags & FF_READING));
fp->fxp_flags |= FF_READING;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
reply(fp);
2006-07-10 14:43:38 +02:00
}
/*===========================================================================*
* fxp_do_conf *
*===========================================================================*/
static void fxp_do_conf(fp)
fxp_t *fp;
{
int r;
2012-03-05 00:11:41 +01:00
phys_bytes bus_addr;
/* Configure device */
tmpbufp->cc.cc_status= 0;
tmpbufp->cc.cc_command= CBL_C_EL | CBL_CONF;
tmpbufp->cc.cc_linkaddr= 0;
memcpy(tmpbufp->cc.cc_bytes, fp->fxp_conf_bytes,
sizeof(tmpbufp->cc.cc_bytes));
r= sys_umap(SELF, VM_D, (vir_bytes)&tmpbufp->cc,
(phys_bytes)sizeof(tmpbufp->cc), &bus_addr);
if (r != OK)
panic("sys_umap failed: %d", r);
fxp_cu_ptr_cmd(fp, SC_CU_START, bus_addr, TRUE /* check idle */);
/* Wait for CU command to complete */
SPIN_UNTIL(tmpbufp->cc.cc_status & CBL_F_C, 100000);
if (!(tmpbufp->cc.cc_status & CBL_F_C))
panic("fxp_do_conf: CU command failed to complete");
if (!(tmpbufp->cc.cc_status & CBL_F_OK))
panic("fxp_do_conf: CU command failed");
}
/*===========================================================================*
* fxp_cu_ptr_cmd *
*===========================================================================*/
static void fxp_cu_ptr_cmd(fp, cmd, bus_addr, check_idle)
fxp_t *fp;
int cmd;
phys_bytes bus_addr;
int check_idle;
{
spin_t spin;
port_t port;
u8_t scb_cmd;
port= fp->fxp_base_port;
if (check_idle)
{
/* Consistency check. Make sure that CU is idle */
if ((fxp_inb(port, SCB_STATUS) & SS_CUS_MASK) != SS_CU_IDLE)
panic("fxp_cu_ptr_cmd: CU is not idle");
}
fxp_outl(port, SCB_POINTER, bus_addr);
fxp_outb(port, SCB_CMD, cmd);
/* What is a reasonable time-out? There is nothing in the
* documentation. 1 ms should be enough. We use 100 ms.
*/
spin_init(&spin, 100000);
do {
/* Wait for CU command to be accepted */
scb_cmd= fxp_inb(port, SCB_CMD);
if ((scb_cmd & SC_CUC_MASK) == SC_CU_NOP)
break;
} while (spin_check(&spin));
if ((scb_cmd & SC_CUC_MASK) != SC_CU_NOP)
panic("fxp_cu_ptr_cmd: CU does not accept command");
}
/*===========================================================================*
* fxp_ru_ptr_cmd *
*===========================================================================*/
static void fxp_ru_ptr_cmd(fp, cmd, bus_addr, check_idle)
fxp_t *fp;
int cmd;
phys_bytes bus_addr;
int check_idle;
{
spin_t spin;
port_t port;
u8_t scb_cmd;
port= fp->fxp_base_port;
if (check_idle)
{
/* Consistency check, make sure that RU is idle */
if ((fxp_inb(port, SCB_STATUS) & SS_RUS_MASK) != SS_RU_IDLE)
panic("fxp_ru_ptr_cmd: RU is not idle");
}
fxp_outl(port, SCB_POINTER, bus_addr);
fxp_outb(port, SCB_CMD, cmd);
spin_init(&spin, 1000);
do {
/* Wait for RU command to be accepted */
scb_cmd= fxp_inb(port, SCB_CMD);
if ((scb_cmd & SC_RUC_MASK) == SC_RU_NOP)
break;
} while (spin_check(&spin));
if ((scb_cmd & SC_RUC_MASK) != SC_RU_NOP)
panic("fxp_ru_ptr_cmd: RU does not accept command");
}
/*===========================================================================*
* fxp_restart_ru *
*===========================================================================*/
static void fxp_restart_ru(fp)
fxp_t *fp;
{
int i, fxp_rx_nbuf;
port_t port;
struct rfd *rfdp;
port= fp->fxp_base_port;
fxp_rx_nbuf= fp->fxp_rx_nbuf;
for (i= 0, rfdp= fp->fxp_rx_buf; i<fxp_rx_nbuf; i++, rfdp++)
{
rfdp->rfd_status= 0;
rfdp->rfd_command= 0;
if (i == fp->fxp_rx_nbuf-1)
rfdp->rfd_command= RFDC_EL;
rfdp->rfd_reserved= 0;
rfdp->rfd_res= 0;
rfdp->rfd_size= sizeof(rfdp->rfd_buf);
}
fp->fxp_rx_head= 0;
/* Make sure that RU is in the 'No resources' state */
if ((fxp_inb(port, SCB_STATUS) & SS_RUS_MASK) != SS_RU_NORES)
panic("fxp_restart_ru: RU is in an unexpected state");
fxp_ru_ptr_cmd(fp, SC_RU_START, fp->fxp_rx_busaddr,
FALSE /* do not check idle */);
}
2006-07-10 14:43:38 +02:00
/*===========================================================================*
* fxp_getstat_s *
*===========================================================================*/
static void fxp_getstat_s(message *mp)
2006-07-10 14:43:38 +02:00
{
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
int r;
2006-07-10 14:43:38 +02:00
fxp_t *fp;
u32_t *p;
eth_stat_t stats;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
fp= fxp_state;
2006-07-10 14:43:38 +02:00
assert(fp->fxp_mode == FM_ENABLED);
assert(fp->fxp_flags & FF_ENABLED);
p= &fp->fxp_stat.sc_tx_fcp;
*p= 0;
/* The dump commmand doesn't take a pointer. Setting a pointer
* doesn't hurt though.
*/
fxp_cu_ptr_cmd(fp, SC_CU_DUMP_SC, 0, FALSE /* do not check idle */);
/* Wait for CU command to complete */
SPIN_UNTIL(*p != 0, 1000);
2006-07-10 14:43:38 +02:00
if (*p == 0)
panic("fxp_getstat: CU command failed to complete");
2006-07-10 14:43:38 +02:00
if (*p != SCM_DSC)
panic("fxp_getstat: bad magic");
2006-07-10 14:43:38 +02:00
stats.ets_recvErr=
fp->fxp_stat.sc_rx_crc +
fp->fxp_stat.sc_rx_align +
fp->fxp_stat.sc_rx_resource +
fp->fxp_stat.sc_rx_overrun +
fp->fxp_stat.sc_rx_cd +
fp->fxp_stat.sc_rx_short;
stats.ets_sendErr=
fp->fxp_stat.sc_tx_maxcol +
fp->fxp_stat.sc_tx_latecol +
fp->fxp_stat.sc_tx_crs;
stats.ets_OVW= fp->fxp_stat.sc_rx_overrun;
stats.ets_CRCerr= fp->fxp_stat.sc_rx_crc;
stats.ets_frameAll= fp->fxp_stat.sc_rx_align;
stats.ets_missedP= fp->fxp_stat.sc_rx_resource;
stats.ets_packetR= fp->fxp_stat.sc_rx_good;
stats.ets_packetT= fp->fxp_stat.sc_tx_good;
stats.ets_transDef= fp->fxp_stat.sc_tx_defered;
stats.ets_collision= fp->fxp_stat.sc_tx_totcol;
stats.ets_transAb= fp->fxp_stat.sc_tx_maxcol;
stats.ets_carrSense= fp->fxp_stat.sc_tx_crs;
stats.ets_fifoUnder= fp->fxp_stat.sc_tx_underrun;
stats.ets_fifoOver= fp->fxp_stat.sc_rx_overrun;
stats.ets_CDheartbeat= 0;
stats.ets_OWC= fp->fxp_stat.sc_tx_latecol;
r= sys_safecopyto(mp->m_source, mp->DL_GRANT, 0, (vir_bytes)&stats,
sizeof(stats));
2006-07-10 14:43:38 +02:00
if (r != OK)
panic("fxp_getstat_s: sys_safecopyto failed: %d", r);
mp->m_type= DL_STAT_REPLY;
r= ipc_send(mp->m_source, mp);
if (r != OK)
panic("fxp_getstat_s: ipc_send failed: %d", r);
}
/*===========================================================================*
* fxp_handler *
*===========================================================================*/
static void fxp_handler(fxp_t *fp)
{
int port;
u16_t isr;
RAND_UPDATE
port= fp->fxp_base_port;
/* Ack interrupt */
isr= fxp_inb(port, SCB_INT_STAT);
fxp_outb(port, SCB_INT_STAT, isr);
if (isr & SIS_FR)
{
isr &= ~SIS_FR;
if (!fp->fxp_got_int && (fp->fxp_flags & FF_READING))
{
fp->fxp_got_int= TRUE;
interrupt(fxp_tasknr);
}
}
if (isr & SIS_CNA)
{
isr &= ~SIS_CNA;
if (!fp->fxp_tx_idle)
{
fp->fxp_send_int= TRUE;
if (!fp->fxp_got_int)
{
fp->fxp_got_int= TRUE;
interrupt(fxp_tasknr);
}
}
}
if (isr & SIS_RNR)
{
isr &= ~SIS_RNR;
/* Assume that receive buffer is full of packets. fxp_readv
* will restart the RU.
*/
fp->fxp_rx_need_restart= 1;
}
if (isr)
{
printf("fxp_handler: unhandled interrupt: isr = 0x%02x\n",
isr);
}
}
/*===========================================================================*
* fxp_check_ints *
*===========================================================================*/
static void fxp_check_ints(fxp_t *fp)
{
int n, fxp_flags, prev_tail;
int fxp_tx_tail, fxp_tx_nbuf, fxp_tx_threshold;
port_t port;
u32_t busaddr;
u16_t tx_status;
u8_t scb_status;
struct tx *txp;
fxp_flags= fp->fxp_flags;
if (fxp_flags & FF_READING)
{
if (!(fp->fxp_rx_buf[fp->fxp_rx_head].rfd_status & RFDS_C))
; /* Nothing */
else
{
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
fxp_readv_s(&fp->fxp_rx_mess, TRUE /* from int */);
}
}
if (fp->fxp_tx_idle)
; /* Nothing to do */
else if (fp->fxp_send_int)
{
fp->fxp_send_int= FALSE;
fxp_tx_tail= fp->fxp_tx_tail;
fxp_tx_nbuf= fp->fxp_tx_nbuf;
n= 0;
for (;;)
{
txp= &fp->fxp_tx_buf[fxp_tx_tail];
tx_status= txp->tx_status;
if (!(tx_status & TXS_C))
break;
n++;
assert(tx_status & TXS_OK);
if (tx_status & TXS_U)
{
fxp_tx_threshold= fp->fxp_tx_threshold;
if (fxp_tx_threshold < TXTT_MAX)
{
fxp_tx_threshold++;
fp->fxp_tx_threshold= fxp_tx_threshold;
}
printf(
"fxp_check_ints: fxp_tx_threshold = 0x%x\n",
fxp_tx_threshold);
}
if (txp->tx_command & TXC_EL)
{
fp->fxp_tx_idle= 1;
break;
}
fxp_tx_tail++;
if (fxp_tx_tail == fxp_tx_nbuf)
fxp_tx_tail= 0;
assert(fxp_tx_tail < fxp_tx_nbuf);
}
if (fp->fxp_need_conf)
{
/* Check the status of the CU */
port= fp->fxp_base_port;
scb_status= fxp_inb(port, SCB_STATUS);
if ((scb_status & SS_CUS_MASK) != SS_CU_IDLE)
{
/* Nothing to do */
printf("scb_status = 0x%x\n", scb_status);
}
else
{
printf("fxp_check_ints: fxp_need_conf\n");
fp->fxp_need_conf= FALSE;
fxp_do_conf(fp);
}
}
if (n)
{
if (!fp->fxp_tx_idle)
{
fp->fxp_tx_tail= fxp_tx_tail;
/* Check the status of the CU */
port= fp->fxp_base_port;
scb_status= fxp_inb(port, SCB_STATUS);
if ((scb_status & SS_CUS_MASK) != SS_CU_IDLE)
{
/* Nothing to do */
printf("scb_status = 0x%x\n",
scb_status);
}
else
{
if (fxp_tx_tail == 0)
prev_tail= fxp_tx_nbuf-1;
else
prev_tail= fxp_tx_tail-1;
busaddr= fp->fxp_tx_buf[prev_tail].
tx_linkaddr;
fxp_cu_ptr_cmd(fp, SC_CU_START,
busaddr, 1 /* check idle */);
}
}
if (fp->fxp_flags & FF_SEND_AVAIL)
{
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
fxp_writev_s(&fp->fxp_tx_mess,
TRUE /* from int */);
}
}
}
if (fp->fxp_report_link)
fxp_report_link(fp);
if (fp->fxp_flags & (FF_PACK_SENT | FF_PACK_RECV))
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
reply(fp);
}
/*===========================================================================*
* fxp_watchdog_f *
*===========================================================================*/
static void fxp_watchdog_f(tp)
minix_timer_t *tp;
{
fxp_t *fp;
2010-07-09 14:58:18 +02:00
set_timer(&fxp_watchdog, system_hz, fxp_watchdog_f, 0);
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
fp= fxp_state;
if (fp->fxp_mode != FM_ENABLED)
return;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
/* Handle race condition, MII interface might be busy */
if(!fp->fxp_mii_busy)
{
/* Check the link status. */
if (fxp_link_changed(fp))
{
2005-08-04 10:07:29 +02:00
#if VERBOSE
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
printf("fxp_watchdog_f: link changed\n");
2005-08-04 10:07:29 +02:00
#endif
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
fp->fxp_report_link= TRUE;
fp->fxp_got_int= TRUE;
interrupt(fxp_tasknr);
}
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
}
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
if (!(fp->fxp_flags & FF_SEND_AVAIL))
{
/* Assume that an idle system is alive */
fp->fxp_tx_alive= TRUE;
return;
}
if (fp->fxp_tx_alive)
{
fp->fxp_tx_alive= FALSE;
return;
}
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
fp->fxp_need_reset= TRUE;
fp->fxp_got_int= TRUE;
interrupt(fxp_tasknr);
}
/*===========================================================================*
* fxp_link_changed *
*===========================================================================*/
static int fxp_link_changed(fp)
fxp_t *fp;
{
u16_t scr;
scr= mii_read(fp, MII_SCR);
scr &= ~(MII_SCR_RES|MII_SCR_RES_1);
return (fp->fxp_mii_scr != scr);
}
/*===========================================================================*
* fxp_report_link *
*===========================================================================*/
static void fxp_report_link(fxp_t *fp)
{
u16_t mii_ctrl, mii_status, mii_id1, mii_id2,
mii_ana, mii_anlpa, mii_ane, mii_extstat,
mii_ms_ctrl, mii_ms_status, scr;
u32_t oui;
int model, rev;
int f, link_up;
fp->fxp_report_link= FALSE;
scr= mii_read(fp, MII_SCR);
scr &= ~(MII_SCR_RES|MII_SCR_RES_1);
fp->fxp_mii_scr= scr;
mii_ctrl= mii_read(fp, MII_CTRL);
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
mii_read(fp, MII_STATUS); /* The status reg is latched, read twice */
mii_status= mii_read(fp, MII_STATUS);
mii_id1= mii_read(fp, MII_PHYID_H);
mii_id2= mii_read(fp, MII_PHYID_L);
mii_ana= mii_read(fp, MII_ANA);
mii_anlpa= mii_read(fp, MII_ANLPA);
mii_ane= mii_read(fp, MII_ANE);
if (mii_status & MII_STATUS_EXT_STAT)
mii_extstat= mii_read(fp, MII_EXT_STATUS);
else
mii_extstat= 0;
if (fp->fxp_ms_regs)
{
mii_ms_ctrl= mii_read(fp, MII_MS_CTRL);
mii_ms_status= mii_read(fp, MII_MS_STATUS);
}
else
{
mii_ms_ctrl= 0;
mii_ms_status= 0;
}
/* How do we know about the link status? */
link_up= !!(mii_status & MII_STATUS_LS);
fp->fxp_link_up= link_up;
if (!link_up)
{
#if VERBOSE
printf("%s: link down\n", fp->fxp_name);
#endif
return;
}
oui= (mii_id1 << MII_PH_OUI_H_C_SHIFT) |
((mii_id2 & MII_PL_OUI_L_MASK) >> MII_PL_OUI_L_SHIFT);
model= ((mii_id2 & MII_PL_MODEL_MASK) >> MII_PL_MODEL_SHIFT);
rev= (mii_id2 & MII_PL_REV_MASK);
#if VERBOSE
printf("OUI 0x%06lx, Model 0x%02x, Revision 0x%x\n", oui, model, rev);
#endif
if (mii_ctrl & (MII_CTRL_LB|MII_CTRL_PD|MII_CTRL_ISO))
{
printf("%s: PHY: ", fp->fxp_name);
f= 1;
if (mii_ctrl & MII_CTRL_LB)
{
printf("loopback mode");
f= 0;
}
if (mii_ctrl & MII_CTRL_PD)
{
if (!f) printf(", ");
f= 0;
printf("powered down");
}
if (mii_ctrl & MII_CTRL_ISO)
{
if (!f) printf(", ");
f= 0;
printf("isolated");
}
printf("\n");
return;
}
if (!(mii_ctrl & MII_CTRL_ANE))
{
printf("%s: manual config: ", fp->fxp_name);
switch(mii_ctrl & (MII_CTRL_SP_LSB|MII_CTRL_SP_MSB))
{
case MII_CTRL_SP_10: printf("10 Mbps"); break;
case MII_CTRL_SP_100: printf("100 Mbps"); break;
case MII_CTRL_SP_1000: printf("1000 Mbps"); break;
case MII_CTRL_SP_RES: printf("reserved speed"); break;
}
if (mii_ctrl & MII_CTRL_DM)
printf(", full duplex");
else
printf(", half duplex");
printf("\n");
return;
}
if (!debug) goto resspeed;
printf("%s: ", fp->fxp_name);
mii_print_stat_speed(mii_status, mii_extstat);
printf("\n");
if (!(mii_status & MII_STATUS_ANC))
printf("%s: auto-negotiation not complete\n", fp->fxp_name);
if (mii_status & MII_STATUS_RF)
printf("%s: remote fault detected\n", fp->fxp_name);
if (!(mii_status & MII_STATUS_ANA))
{
printf("%s: local PHY has no auto-negotiation ability\n",
fp->fxp_name);
}
if (!(mii_status & MII_STATUS_LS))
printf("%s: link down\n", fp->fxp_name);
if (mii_status & MII_STATUS_JD)
printf("%s: jabber condition detected\n", fp->fxp_name);
if (!(mii_status & MII_STATUS_EC))
{
printf("%s: no extended register set\n", fp->fxp_name);
goto resspeed;
}
if (!(mii_status & MII_STATUS_ANC))
goto resspeed;
printf("%s: local cap.: ", fp->fxp_name);
if (mii_ms_ctrl & (MII_MSC_1000T_FD | MII_MSC_1000T_HD))
{
printf("1000 Mbps: T-");
switch(mii_ms_ctrl & (MII_MSC_1000T_FD | MII_MSC_1000T_HD))
{
case MII_MSC_1000T_FD: printf("FD"); break;
case MII_MSC_1000T_HD: printf("HD"); break;
default: printf("FD/HD"); break;
}
if (mii_ana)
printf(", ");
}
mii_print_techab(mii_ana);
printf("\n");
if (mii_ane & MII_ANE_PDF)
printf("%s: parallel detection fault\n", fp->fxp_name);
if (!(mii_ane & MII_ANE_LPANA))
{
printf("%s: link-partner does not support auto-negotiation\n",
fp->fxp_name);
goto resspeed;
}
printf("%s: remote cap.: ", fp->fxp_name);
if (mii_ms_ctrl & (MII_MSC_1000T_FD | MII_MSC_1000T_HD))
if (mii_ms_status & (MII_MSS_LP1000T_FD | MII_MSS_LP1000T_HD))
{
printf("1000 Mbps: T-");
switch(mii_ms_status &
(MII_MSS_LP1000T_FD | MII_MSS_LP1000T_HD))
{
case MII_MSS_LP1000T_FD: printf("FD"); break;
case MII_MSS_LP1000T_HD: printf("HD"); break;
default: printf("FD/HD"); break;
}
if (mii_anlpa)
printf(", ");
}
mii_print_techab(mii_anlpa);
printf("\n");
if (fp->fxp_ms_regs)
{
printf("%s: ", fp->fxp_name);
if (mii_ms_ctrl & MII_MSC_MS_MANUAL)
{
printf("manual %s",
(mii_ms_ctrl & MII_MSC_MS_VAL) ?
"MASTER" : "SLAVE");
}
else
{
printf("%s device",
(mii_ms_ctrl & MII_MSC_MULTIPORT) ?
"multiport" : "single-port");
}
if (mii_ms_ctrl & MII_MSC_RES)
printf(" reserved<0x%x>", mii_ms_ctrl & MII_MSC_RES);
printf(": ");
if (mii_ms_status & MII_MSS_FAULT)
printf("M/S config fault");
else if (mii_ms_status & MII_MSS_MASTER)
printf("MASTER");
else
printf("SLAVE");
printf("\n");
}
if (mii_ms_status & (MII_MSS_LP1000T_FD|MII_MSS_LP1000T_HD))
{
if (!(mii_ms_status & MII_MSS_LOCREC))
{
printf("%s: local receiver not OK\n",
fp->fxp_name);
}
if (!(mii_ms_status & MII_MSS_REMREC))
{
printf("%s: remote receiver not OK\n",
fp->fxp_name);
}
}
if (mii_ms_status & (MII_MSS_RES|MII_MSS_IDLE_ERR))
{
printf("%s", fp->fxp_name);
if (mii_ms_status & MII_MSS_RES)
printf(" reserved<0x%x>", mii_ms_status & MII_MSS_RES);
if (mii_ms_status & MII_MSS_IDLE_ERR)
{
printf(" idle error %d",
mii_ms_status & MII_MSS_IDLE_ERR);
}
printf("\n");
}
resspeed:
2005-08-04 10:07:29 +02:00
#if VERBOSE
printf("%s: link up, %d Mbps, %s duplex\n",
fp->fxp_name, (scr & MII_SCR_100) ? 100 : 10,
(scr & MII_SCR_FD) ? "full" : "half");
2005-08-04 10:07:29 +02:00
#endif
;
}
/*===========================================================================*
* reply *
*===========================================================================*/
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
static void reply(fp)
fxp_t *fp;
{
message reply;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
int flags;
int r;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
flags = DL_NOFLAGS;
if (fp->fxp_flags & FF_PACK_SENT)
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
flags |= DL_PACK_SEND;
if (fp->fxp_flags & FF_PACK_RECV)
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
flags |= DL_PACK_RECV;
reply.m_type = DL_TASK_REPLY;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
reply.DL_FLAGS = flags;
reply.DL_COUNT = fp->fxp_read_s;
r= ipc_send(fp->fxp_client, &reply);
if (r < 0)
panic("fxp: ipc_send failed: %d", r);
fp->fxp_read_s = 0;
fp->fxp_flags &= ~(FF_PACK_SENT | FF_PACK_RECV);
}
/*===========================================================================*
* mess_reply *
*===========================================================================*/
static void mess_reply(req, reply_mess)
message *req;
message *reply_mess;
{
if (ipc_send(req->m_source, reply_mess) != OK)
panic("fxp: unable to mess_reply");
}
/*===========================================================================*
* eeprom_read *
*===========================================================================*/
2012-03-25 20:25:53 +02:00
static u16_t eeprom_read(fp, reg)
fxp_t *fp;
int reg;
{
port_t port;
u16_t v;
int b, i, alen;
alen= fp->fxp_ee_addrlen;
if (!alen)
{
eeprom_addrsize(fp);
alen= fp->fxp_ee_addrlen;
assert(alen == 6 || alen == 8);
}
port= fp->fxp_base_port;
fxp_outb(port, CSR_EEPROM, CE_EECS); /* Enable EEPROM */
v= EEPROM_READ_PREFIX;
for (i= EEPROM_PREFIX_LEN-1; i >= 0; i--)
{
b= ((v & (1 << i)) ? CE_EEDI : 0);
fxp_outb(port, CSR_EEPROM, CE_EECS | b); /* bit */
fxp_outb(port, CSR_EEPROM, CE_EECS | b | CE_EESK); /* Clock */
micro_delay(EESK_PERIOD/2+1);
fxp_outb(port, CSR_EEPROM, CE_EECS | b);
micro_delay(EESK_PERIOD/2+1);
}
v= reg;
for (i= alen-1; i >= 0; i--)
{
b= ((v & (1 << i)) ? CE_EEDI : 0);
fxp_outb(port, CSR_EEPROM, CE_EECS | b); /* bit */
fxp_outb(port, CSR_EEPROM, CE_EECS | b | CE_EESK); /* Clock */
micro_delay(EESK_PERIOD/2+1);
fxp_outb(port, CSR_EEPROM, CE_EECS | b);
micro_delay(EESK_PERIOD/2+1);
}
v= 0;
for (i= 0; i<16; i++)
{
fxp_outb(port, CSR_EEPROM, CE_EECS | CE_EESK); /* Clock */
micro_delay(EESK_PERIOD/2+1);
b= !!(fxp_inb(port, CSR_EEPROM) & CE_EEDO);
v= (v << 1) | b;
fxp_outb(port, CSR_EEPROM, CE_EECS );
micro_delay(EESK_PERIOD/2+1);
}
fxp_outb(port, CSR_EEPROM, 0); /* Disable EEPROM */
micro_delay(EECS_DELAY);
return v;
}
/*===========================================================================*
* eeprom_addrsize *
*===========================================================================*/
2012-03-25 20:25:53 +02:00
static void eeprom_addrsize(fp)
fxp_t *fp;
{
port_t port;
u16_t v;
int b, i;
port= fp->fxp_base_port;
/* Try to find out the size of the EEPROM */
fxp_outb(port, CSR_EEPROM, CE_EECS); /* Enable EEPROM */
v= EEPROM_READ_PREFIX;
for (i= EEPROM_PREFIX_LEN-1; i >= 0; i--)
{
b= ((v & (1 << i)) ? CE_EEDI : 0);
fxp_outb(port, CSR_EEPROM, CE_EECS | b); /* bit */
fxp_outb(port, CSR_EEPROM, CE_EECS | b | CE_EESK); /* Clock */
micro_delay(EESK_PERIOD/2+1);
fxp_outb(port, CSR_EEPROM, CE_EECS | b);
micro_delay(EESK_PERIOD/2+1);
}
for (i= 0; i<32; i++)
{
b= 0;
fxp_outb(port, CSR_EEPROM, CE_EECS | b); /* bit */
fxp_outb(port, CSR_EEPROM, CE_EECS | b | CE_EESK); /* Clock */
micro_delay(EESK_PERIOD/2+1);
fxp_outb(port, CSR_EEPROM, CE_EECS | b);
micro_delay(EESK_PERIOD/2+1);
v= fxp_inb(port, CSR_EEPROM);
if (!(v & CE_EEDO))
break;
}
if (i >= 32)
panic("eeprom_addrsize: failed");
fp->fxp_ee_addrlen= i+1;
/* Discard 16 data bits */
for (i= 0; i<16; i++)
{
fxp_outb(port, CSR_EEPROM, CE_EECS | CE_EESK); /* Clock */
micro_delay(EESK_PERIOD/2+1);
fxp_outb(port, CSR_EEPROM, CE_EECS );
micro_delay(EESK_PERIOD/2+1);
}
fxp_outb(port, CSR_EEPROM, 0); /* Disable EEPROM */
micro_delay(EECS_DELAY);
#if VERBOSE
printf("%s EEPROM address length: %d\n",
fp->fxp_name, fp->fxp_ee_addrlen);
#endif
}
/*===========================================================================*
* mii_read *
*===========================================================================*/
2012-03-25 20:25:53 +02:00
static u16_t mii_read(fp, reg)
fxp_t *fp;
int reg;
{
spin_t spin;
port_t port;
u32_t v;
port= fp->fxp_base_port;
assert(!fp->fxp_mii_busy);
fp->fxp_mii_busy++;
if (!(fxp_inl(port, CSR_MDI_CTL) & CM_READY))
panic("mii_read: MDI not ready");
fxp_outl(port, CSR_MDI_CTL, CM_READ | (1 << CM_PHYADDR_SHIFT) |
(reg << CM_REG_SHIFT));
spin_init(&spin, 100000);
do {
v= fxp_inl(port, CSR_MDI_CTL);
if (v & CM_READY)
break;
} while (spin_check(&spin));
if (!(v & CM_READY))
panic("mii_read: MDI not ready after command");
fp->fxp_mii_busy--;
assert(!fp->fxp_mii_busy);
return v & CM_DATA_MASK;
}
static u8_t do_inb(port_t port)
{
int r;
u32_t value;
r= sys_inb(port, &value);
if (r != OK)
panic("sys_inb failed: %d", r);
return value;
}
static u32_t do_inl(port_t port)
{
int r;
u32_t value;
r= sys_inl(port, &value);
if (r != OK)
panic("sys_inl failed: %d", r);
return value;
}
static void do_outb(port_t port, u8_t value)
{
int r;
r= sys_outb(port, value);
if (r != OK)
panic("sys_outb failed: %d", r);
}
static void do_outl(port_t port, u32_t value)
{
int r;
r= sys_outl(port, value);
if (r != OK)
panic("sys_outl failed: %d", r);
}
2012-03-25 20:25:53 +02:00
static void tell_dev(buf, size, pci_bus, pci_dev, pci_func)
vir_bytes buf;
size_t size;
int pci_bus;
int pci_dev;
int pci_func;
{
int r;
endpoint_t dev_e;
message m;
Driver refactory for live update and crash recovery. SYSLIB CHANGES: - DS calls to publish / retrieve labels consider endpoints instead of u32_t. VFS CHANGES: - mapdriver() only adds an entry in the dmap table in VFS. - dev_up() is only executed upon reception of a driver up event. INET CHANGES: - INET no longer searches for existing drivers instances at startup. - A newtwork driver is (re)initialized upon reception of a driver up event. - Networking startup is now race-free by design. No need to waste 5 seconds at startup any more. DRIVER CHANGES: - Every driver publishes driver up events when starting for the first time or in case of restart when recovery actions must be taken in the upper layers. - Driver up events are published by drivers through DS. - For regular drivers, VFS is normally the only subscriber, but not necessarily. For instance, when the filter driver is in use, it must subscribe to driver up events to initiate recovery. - For network drivers, inet is the only subscriber for now. - Every VFS driver is statically linked with libdriver, every network driver is statically linked with libnetdriver. DRIVER LIBRARIES CHANGES: - Libdriver is extended to provide generic receive() and ds_publish() interfaces for VFS drivers. - driver_receive() is a wrapper for sef_receive() also used in driver_task() to discard spurious messages that were meant to be delivered to a previous version of the driver. - driver_receive_mq() is the same as driver_receive() but integrates support for queued messages. - driver_announce() publishes a driver up event for VFS drivers and marks the driver as initialized and expecting a DEV_OPEN message. - Libnetdriver is introduced to provide similar receive() and ds_publish() interfaces for network drivers (netdriver_announce() and netdriver_receive()). - Network drivers all support live update with no state transfer now. KERNEL CHANGES: - Added kernel call statectl for state management. Used by driver_announce() to unblock eventual callers sendrecing to the driver.
2010-04-08 15:41:35 +02:00
r= ds_retrieve_label_endpt("amddev", &dev_e);
if (r != OK)
{
#if 0
printf(
Driver refactory for live update and crash recovery. SYSLIB CHANGES: - DS calls to publish / retrieve labels consider endpoints instead of u32_t. VFS CHANGES: - mapdriver() only adds an entry in the dmap table in VFS. - dev_up() is only executed upon reception of a driver up event. INET CHANGES: - INET no longer searches for existing drivers instances at startup. - A newtwork driver is (re)initialized upon reception of a driver up event. - Networking startup is now race-free by design. No need to waste 5 seconds at startup any more. DRIVER CHANGES: - Every driver publishes driver up events when starting for the first time or in case of restart when recovery actions must be taken in the upper layers. - Driver up events are published by drivers through DS. - For regular drivers, VFS is normally the only subscriber, but not necessarily. For instance, when the filter driver is in use, it must subscribe to driver up events to initiate recovery. - For network drivers, inet is the only subscriber for now. - Every VFS driver is statically linked with libdriver, every network driver is statically linked with libnetdriver. DRIVER LIBRARIES CHANGES: - Libdriver is extended to provide generic receive() and ds_publish() interfaces for VFS drivers. - driver_receive() is a wrapper for sef_receive() also used in driver_task() to discard spurious messages that were meant to be delivered to a previous version of the driver. - driver_receive_mq() is the same as driver_receive() but integrates support for queued messages. - driver_announce() publishes a driver up event for VFS drivers and marks the driver as initialized and expecting a DEV_OPEN message. - Libnetdriver is introduced to provide similar receive() and ds_publish() interfaces for network drivers (netdriver_announce() and netdriver_receive()). - Network drivers all support live update with no state transfer now. KERNEL CHANGES: - Added kernel call statectl for state management. Used by driver_announce() to unblock eventual callers sendrecing to the driver.
2010-04-08 15:41:35 +02:00
"fxp`tell_dev: ds_retrieve_label_endpt failed for 'amddev': %d\n",
r);
#endif
return;
}
m.m_type= IOMMU_MAP;
m.m2_i1= pci_bus;
m.m2_i2= pci_dev;
m.m2_i3= pci_func;
m.m2_l1= buf;
m.m2_l2= size;
r= ipc_sendrec(dev_e, &m);
if (r != OK)
{
printf("fxp`tell_dev: ipc_sendrec to %d failed: %d\n",
dev_e, r);
return;
}
if (m.m_type != OK)
{
printf("fxp`tell_dev: dma map request failed: %d\n",
m.m_type);
return;
}
}
/*
* $PchId: fxp.c,v 1.4 2005/01/31 22:10:37 philip Exp $
*/