minix/drivers/dec21140A/dec21140A.c

909 lines
25 KiB
C
Raw Normal View History

/*
* dec21041.c
*
* This file contains an ethernet device driver for DEC 21140A
* fast ethernet controllers as emulated by VirtualPC 2007. It is not
* intended to support the real card, as much more error checking
* and testing would be needed. It supports both bridged and NAT mode.
*
* Created: Mar 2008 by Nicolas Tittley <first.last@ google's mail>
*/
#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 <assert.h>
#include <machine/pci.h>
#include <minix/syslib.h>
#include <minix/endpoint.h>
#include <minix/com.h>
#include <minix/sef.h>
#include <minix/ds.h>
#include <net/gen/ether.h>
#include <net/gen/eth_io.h>
#include <stdlib.h>
#include "dec21140A.h"
_PROTOTYPE( PRIVATE u32_t io_inl, (u16_t); );
_PROTOTYPE( PRIVATE void io_outl, (u16_t, u32_t); );
_PROTOTYPE( PRIVATE void do_conf, (const message *); );
_PROTOTYPE( PRIVATE void do_get_stat_s, (message *); );
_PROTOTYPE( PRIVATE void do_interrupt, (const dpeth_t *); );
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
_PROTOTYPE( PRIVATE void do_reply, (dpeth_t *); );
_PROTOTYPE( PRIVATE void do_vread_s, (const message *, int); );
_PROTOTYPE( PRIVATE void do_watchdog, (void *); );
_PROTOTYPE( PRIVATE void de_update_conf, (dpeth_t *); );
_PROTOTYPE( PRIVATE int de_probe, (dpeth_t *, int skip); );
_PROTOTYPE( PRIVATE void de_conf_addr, (dpeth_t *); );
_PROTOTYPE( PRIVATE void de_first_init, (dpeth_t *); );
_PROTOTYPE( PRIVATE void de_reset, (const dpeth_t *); );
_PROTOTYPE( PRIVATE void de_hw_conf, (const dpeth_t *); );
_PROTOTYPE( PRIVATE void de_start, (const dpeth_t *); );
_PROTOTYPE( PRIVATE void de_setup_frame, (const dpeth_t *); );
2010-04-07 09:50:44 +02:00
_PROTOTYPE( PRIVATE u16_t de_read_rom, (const dpeth_t *, u8_t, u8_t); );
_PROTOTYPE( PRIVATE int de_calc_iov_size, (iovec_dat_s_t *); );
_PROTOTYPE( PRIVATE void de_next_iov, (iovec_dat_s_t *); );
_PROTOTYPE( PRIVATE void do_vwrite_s, (const message *, int); );
_PROTOTYPE( PRIVATE void de_get_userdata_s, (int, cp_grant_id_t,
vir_bytes, int, void *); );
/* Error messages */
static char str_CopyErrMsg[] = "unable to read/write user data";
static char str_SendErrMsg[] = "send failed";
static char str_SizeErrMsg[] = "illegal packet size";
static char str_UmapErrMsg[] = "Unable to sys_umap";
static char str_BusyErrMsg[] = "Send/Recv failed: busy";
static char str_StatErrMsg[] = "Unable to send stats";
static char str_AlignErrMsg[] = "Bad align of buffer/descriptor";
static char str_DevName[] = "dec21140A:eth#?";
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
PRIVATE dpeth_t de_state;
PRIVATE int de_instance;
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 functions and variables. */
FORWARD _PROTOTYPE( void sef_local_startup, (void) );
FORWARD _PROTOTYPE( int sef_cb_init_fresh, (int type, sef_init_info_t *info) );
EXTERN char **env_argv;
/*===========================================================================*
* main *
*===========================================================================*/
int main(int argc, char *argv[])
{
dpeth_t *dep;
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;
int r;
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 local startup. */
env_setargs(argc, argv);
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_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 CLOCK:
do_watchdog(&m);
break;
case HARDWARE:
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
dep = &de_state;
if (dep->de_mode == DEM_ENABLED) {
do_interrupt(dep);
if (dep->de_flags & (DEF_ACK_SEND | DEF_ACK_RECV))
do_reply(dep);
sys_irqenable(&dep->de_hook);
}
break;
default:
printf("ignoring notify from %d\n", m.m_source);
break;
}
continue;
}
switch (m.m_type)
{
case DL_WRITEV_S: do_vwrite_s(&m, FALSE); break;
case DL_READV_S: do_vread_s(&m, FALSE); break;
case DL_CONF: do_conf(&m); break;
case DL_GETSTAT_S: do_get_stat_s(&m); break;
default:
printf("message 0x%lx; %d from %d\n",
m.m_type, m.m_type-DL_RQ_BASE, m.m_source);
panic("illegal message: %d", m.m_type);
}
}
}
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_local_startup *
*===========================================================================*/
PRIVATE void sef_local_startup()
{
/* 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);
sef_setcb_init_restart(sef_cb_init_fresh);
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
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);
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_term);
/* Let SEF perform startup. */
sef_startup();
}
/*===========================================================================*
* sef_cb_init_fresh *
*===========================================================================*/
PRIVATE int sef_cb_init_fresh(int type, sef_init_info_t *UNUSED(info))
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
{
/* Initialize the DEC 21140A driver. */
int fkeys, sfkeys;
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
long v;
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
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);
de_instance = (int) v;
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
/* Request function key for debug dumps */
fkeys = sfkeys = 0;
bit_set(sfkeys, DE_FKEY);
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
if ((fkey_map(&fkeys, &sfkeys)) != OK)
printf("%s: error using Shift+F%d key(%d)\n", str_DevName, DE_FKEY, errno);
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();
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
return OK;
}
PRIVATE void do_get_stat_s(message * 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
int rc;
dpeth_t *dep;
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
dep = &de_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 ((rc = sys_safecopyto(mp->DL_ENDPT, mp->DL_GRANT, 0UL,
(vir_bytes)&dep->de_stat,
sizeof(dep->de_stat), 0)) != OK)
panic(str_CopyErrMsg, rc);
mp->m_type = DL_STAT_REPLY;
rc = send(mp->m_source, mp);
if( rc != OK )
panic(str_StatErrMsg, rc);
return;
}
PRIVATE void do_conf(const message * 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
int r;
dpeth_t *dep;
message reply_mess;
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
dep = &de_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
strncpy(dep->de_name, str_DevName, strlen(str_DevName));
dep->de_name[strlen(dep->de_name)-1] = '0' + de_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
if (dep->de_mode == DEM_DISABLED) {
de_update_conf(dep);
pci_init();
if (dep->de_mode == DEM_ENABLED && !de_probe(dep, de_instance)) {
printf("%s: warning no ethernet card found at 0x%04X\n",
dep->de_name, dep->de_base_port);
dep->de_mode = DEM_DISABLED;
}
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
r = OK;
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
/* 'de_mode' may change if probe routines fail, test again */
switch (dep->de_mode) {
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
case DEM_DISABLED:
r = ENXIO; /* Device is OFF or hardware probe failed */
break;
case DEM_ENABLED:
if (dep->de_flags == DEF_EMPTY) {
de_first_init(dep);
dep->de_flags |= DEF_ENABLED;
de_reset(dep);
de_hw_conf(dep);
de_setup_frame(dep);
de_start(dep);
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
/* TODO CHECK PROMISC AND MULTI */
dep->de_flags &= NOT(DEF_PROMISC | DEF_MULTI | DEF_BROAD);
if (mp->DL_MODE & DL_PROMISC_REQ)
dep->de_flags |= DEF_PROMISC | DEF_MULTI | DEF_BROAD;
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 (mp->DL_MODE & DL_MULTI_REQ) dep->de_flags |= DEF_MULTI;
if (mp->DL_MODE & DL_BROAD_REQ) dep->de_flags |= DEF_BROAD;
break;
case DEM_SINK:
DEBUG(printf("%s running in sink mode\n", str_DevName));
memset(dep->de_address.ea_addr, 0, sizeof(ether_addr_t));
de_conf_addr(dep);
break;
default: break;
}
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 = r;
if(r == OK){
*(ether_addr_t *) reply_mess.DL_HWADDR = dep->de_address;
}
if (send(mp->m_source, &reply_mess) != OK)
panic(str_SendErrMsg, mp->m_source);
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
PRIVATE void do_reply(dpeth_t * dep)
{
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 r, flags = DL_NOFLAGS;
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 (dep->de_flags & DEF_ACK_SEND) flags |= DL_PACK_SEND;
if (dep->de_flags & DEF_ACK_RECV) 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_STAT = flags;
reply.DL_COUNT = dep->de_read_s;
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 = send(dep->de_client, &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
if(r < 0)
panic(str_SendErrMsg, r);
dep->de_read_s = 0;
dep->de_flags &= NOT(DEF_ACK_SEND | DEF_ACK_RECV);
return;
}
PRIVATE void do_watchdog(void *UNUSED(message))
{
/* nothing here yet */
return;
}
PRIVATE int de_probe(dpeth_t *dep, int skip){
int i, r, devind;
u16_t vid, did, temp16;
DEBUG(printf("PROBING..."));
r= pci_first_dev(&devind, &vid, &did);
if (r == 0)
return FALSE;
for(;;)
{
if ( DEC21140A_VID == vid &&
DEC21140A_DID == did)
{
if (!skip)
break;
skip--;
}
r= pci_next_dev(&devind, &vid, &did);
if (!r)
return FALSE;
}
pci_reserve(devind);
dep->de_base_port = pci_attr_r32(devind, PCI_BAR) & 0xffffffe0;
dep->de_irq = pci_attr_r8(devind, PCI_ILR);
if (dep->de_base_port < DE_MIN_BASE_ADDR)
panic("de_probe: base address invalid: %d", dep->de_base_port);
DEBUG(printf("%s: using I/O address 0x%lx, IRQ %d\n",
dep->de_name, (unsigned long)dep->de_base_port,
dep->de_irq));
dep->de_type = pci_attr_r8(devind, PCI_REV);
/* device validation. We support only the DEC21140A */
if(dep->de_type != DEC_21140A){
dep->de_type = DE_TYPE_UNKNOWN;
printf("%s: unsupported device\n", str_DevName);
return FALSE;
}
de_reset(dep);
DEBUG(printf("Reading SROM...\n"));
for(i=0;i<(1<<SROM_BITWIDTH)-1;i++){
temp16 = de_read_rom(dep, i, SROM_BITWIDTH);
dep->srom[i*2] = temp16 & 0xFF;
dep->srom[i*2+1] = temp16 >> 8;
}
/* TODO: validate SROM content */
/* acquire MAC addr */
DEBUG(printf("Using MAC addr= "));
for(i=0;i<6;i++){
dep->de_address.ea_addr[i] = dep->srom[i+DE_SROM_EA_OFFSET];
DEBUG(printf("%02X%c",dep->de_address.ea_addr[i],i!=5?'-':'\n'));
}
DEBUG(printf("probe success\n"));
return TRUE;
}
PRIVATE u16_t de_read_rom(const dpeth_t *dep, u8_t addr, u8_t nbAddrBits){
u16_t retVal = 0;
int i;
u32_t csr = 0;
u32_t csr2 = 0; /* csr2 is used to hold constant values that are
setup in the init phase, it makes this a little
more readable, the following macro is also just
to clear up the code a little.*/
#define EMIT do { io_outl(CSR_ADDR(dep, CSR9), csr | csr2); io_outl(CSR_ADDR(dep, CSR1), 0);} while(0)
/* init */
csr = 0; EMIT;
csr = CSR9_SR; EMIT;
csr = CSR9_SR | CSR9_RD; EMIT;
csr2 = CSR9_SR | CSR9_RD;
csr = 0; EMIT;
csr2 |= CSR9_CS;
csr = 0; EMIT;
csr = CSR9_SRC; EMIT;
csr = 0; EMIT;
/* cmd 110 - Read */
csr = CSR9_DI; EMIT;
csr = CSR9_DI | CSR9_SRC; EMIT;
csr = CSR9_DI; EMIT;
csr = CSR9_DI | CSR9_SRC; EMIT;
csr = CSR9_DI; EMIT;
csr = 0; EMIT;
csr = CSR9_SRC; EMIT;
csr = 0; EMIT;
/* addr to read */
for(i=nbAddrBits;i!=0;i--){
csr = (addr&(1<<(i-1))) != 0 ? CSR9_DI : 0; EMIT;
csr ^= CSR9_SRC; EMIT;
csr ^= CSR9_SRC; EMIT;
}
/* actual read */
retVal=0;
for(i=0;i<16;i++){
retVal <<= 1;
csr = CSR9_SRC; EMIT;
retVal |= (io_inl(CSR_ADDR(dep, CSR9)) & CSR9_DO) == 0 ? 0 : 1;
csr = 0; EMIT;
}
/* clean up */
csr = 0; EMIT;
#undef EMIT
return retVal;
}
static void de_update_conf(dpeth_t * dep)
{
static char dpc_fmt[] = "x:d:x";
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
char ec_key[16];
long val;
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
strcpy(ec_key, "DEETH0");
ec_key[5] += de_instance;
dep->de_mode = DEM_ENABLED;
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
switch (env_parse(ec_key, dpc_fmt, 0, &val, 0x000L, 0x3FFL)) {
case EP_OFF: dep->de_mode = DEM_DISABLED; break;
case EP_ON: dep->de_mode = DEM_SINK; break;
}
dep->de_base_port = 0;
return;
}
PRIVATE void do_vread_s(const message * mp, int from_int)
{
char *buffer;
u32_t size;
int r, ix = 0;
vir_bytes bytes;
dpeth_t *dep = NULL;
de_loc_descr_t *descr = NULL;
iovec_dat_s_t *iovp = NULL;
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
dep = &de_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
dep->de_client = mp->m_source;
if (dep->de_mode == DEM_ENABLED) {
descr = &dep->descr[DESCR_RECV][dep->cur_descr[DESCR_RECV]];
/* check if packet is in the current descr and only there */
if( !( !(descr->descr->des[DES0] & DES0_OWN) &&
(descr->descr->des[DES0] & DES0_FS) &&
(descr->descr->des[DES0] & DES0_LS) ))
goto suspend;
/*TODO: multi-descr msgs...*/
/* We only support packets contained in a single descriptor.
Setting the descriptor buffer size to less then
ETH_MAX_PACK_SIZE will result in multi-descriptor
packets that we won't be able to handle
*/
assert(!(descr->descr->des[DES0]&DES0_OWN));
assert(descr->descr->des[DES0]&DES0_FS);
assert(descr->descr->des[DES0]&DES0_LS);
/* Check for abnormal messages. We assert here
because this driver is for a virtualized
envrionment where we will not get bad packets
*/
assert(!(descr->descr->des[DES0]&DES0_ES));
assert(!(descr->descr->des[DES0]&DES0_RE));
/* Setup the iovec entry to allow copying into
client layer
*/
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
dep->de_read_iovec.iod_proc_nr = mp->DL_ENDPT;
de_get_userdata_s(mp->DL_ENDPT, (cp_grant_id_t) mp->DL_GRANT, 0,
mp->DL_COUNT, dep->de_read_iovec.iod_iovec);
dep->de_read_iovec.iod_iovec_s = mp->DL_COUNT;
dep->de_read_iovec.iod_grant = (cp_grant_id_t) mp->DL_GRANT;
dep->de_read_iovec.iod_iovec_offset = 0;
size = de_calc_iov_size(&dep->de_read_iovec);
if (size < ETH_MAX_PACK_SIZE)
panic(str_SizeErrMsg, size);
/* Copy buffer to user area and clear ownage */
size = (descr->descr->des[DES0]&DES0_FL)>>DES0_FL_SHIFT;
/*TODO: Complain to MS */
/*HACK: VPC2007 returns packet of invalid size. Ethernet standard
specify 46 bytes as the minimum for valid payload. However, this is
artificial in so far as for certain packet types, notably ARP, less
then 46 bytes are needed to contain the full information. In a non
virtualized environment the 46 bytes rule is enforced in order to give
guarantee in the collison detection scheme. Of course, this being a
driver for a VPC2007, we won't have collisions and I can only suppose
MS decided to cut packet size to true minimum, regardless of the
46 bytes payload standard. Note that this seems to not happen in
bridged mode. Note also, that the card does not return runt or
incomplete frames to us, so this hack is safe
*/
if(size<60){
bzero(&descr->buf1[size], 60-size);
size=60;
}
/* End ugly hack */
iovp = &dep->de_read_iovec;
buffer = descr->buf1;
dep->bytes_rx += size;
dep->de_stat.ets_packetR++;
dep->de_read_s = size;
do {
bytes = iovp->iod_iovec[ix].iov_size; /* Size of buffer */
if (bytes >= size)
bytes = size;
r= sys_safecopyto(iovp->iod_proc_nr, iovp->iod_iovec[ix].iov_grant, 0,
(vir_bytes)buffer, bytes, D);
if (r != OK)
panic(str_CopyErrMsg, r);
buffer += bytes;
if (++ix >= IOVEC_NR) { /* Next buffer of IO vector */
de_next_iov(iovp);
ix = 0;
}
} while ((size -= bytes) > 0);
descr->descr->des[DES0]=DES0_OWN;
dep->cur_descr[DESCR_RECV]++;
if(dep->cur_descr[DESCR_RECV] >= DE_NB_RECV_DESCR)
dep->cur_descr[DESCR_RECV] = 0;
DEBUG(printf("Read returned size = %d\n", size));
/* Reply information */
dep->de_flags |= DEF_ACK_RECV;
dep->de_flags &= NOT(DEF_READING);
}
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
do_reply(dep);
}
return;
suspend:
if(from_int){
assert(dep->de_flags & DEF_READING);
return;
}
assert(!(dep->de_flags & DEF_READING));
dep->rx_return_msg = *mp;
dep->de_flags |= DEF_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
do_reply(dep);
return;
}
PRIVATE void de_conf_addr(dpeth_t * dep)
{
static char ea_fmt[] = "x:x:x:x:x:x";
char ea_key[16];
int ix;
long val;
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
strcpy(ea_key, "DEETH0_EA");
ea_key[5] += de_instance;
for (ix = 0; ix < SA_ADDR_LEN; ix++) {
val = dep->de_address.ea_addr[ix];
if (env_parse(ea_key, ea_fmt, ix, &val, 0x00L, 0xFFL) != EP_SET)
break;
dep->de_address.ea_addr[ix] = val;
}
if (ix != 0 && ix != SA_ADDR_LEN)
env_parse(ea_key, "?", 0, &val, 0L, 0L);
return;
}
PRIVATE void de_first_init(dpeth_t *dep)
{
int i,j,r;
vir_bytes descr_vir = dep->sendrecv_descr_buf;
vir_bytes buffer_vir = dep->sendrecv_buf;
de_loc_descr_t *loc_descr;
u32_t temp;
for(i=0;i<2;i++){
loc_descr = &dep->descr[i][0];
for(j=0; j < (i==DESCR_RECV ? DE_NB_RECV_DESCR : DE_NB_SEND_DESCR); j++){
/* assign buffer space for descriptor */
loc_descr->descr = descr_vir;
descr_vir += sizeof(de_descr_t);
/* assign space for buffer */
loc_descr->buf1 = buffer_vir;
buffer_vir += (i==DESCR_RECV ? DE_RECV_BUF_SIZE : DE_SEND_BUF_SIZE);
loc_descr->buf2 = 0;
loc_descr++;
}
}
/* Now that we have buffer space and descriptors, we need to
obtain their physical address to pass to the hardware
*/
for(i=0;i<2;i++){
loc_descr = &dep->descr[i][0];
temp = (i==DESCR_RECV ? DE_RECV_BUF_SIZE : DE_SEND_BUF_SIZE);
for(j=0; j < (i==DESCR_RECV ? DE_NB_RECV_DESCR : DE_NB_SEND_DESCR); j++){
/* translate buffers physical address */
r = sys_umap(SELF, VM_D, loc_descr->buf1, temp,
&(loc_descr->descr->des[DES_BUF1]));
if(r != OK) panic("umap failed: %d", r);
loc_descr->descr->des[DES_BUF2] = 0;
memset(&loc_descr->descr->des[DES0],0,sizeof(u32_t));
loc_descr->descr->des[DES1] = temp;
if(j==( (i==DESCR_RECV?DE_NB_RECV_DESCR:DE_NB_SEND_DESCR)-1))
loc_descr->descr->des[DES1] |= DES1_ER;
if(i==DESCR_RECV)
loc_descr->descr->des[DES0] |= DES0_OWN;
loc_descr++;
}
}
/* record physical location of two first descriptor */
r = sys_umap(SELF, VM_D, dep->descr[DESCR_RECV][0].descr,
sizeof(de_descr_t), &dep->sendrecv_descr_phys_addr[DESCR_RECV]);
if(r != OK) panic(str_UmapErrMsg, r);
r = sys_umap(SELF, VM_D, dep->descr[DESCR_TRAN][0].descr,
sizeof(de_descr_t), &dep->sendrecv_descr_phys_addr[DESCR_TRAN]);
if(r != OK) panic(str_UmapErrMsg, r);
DEBUG(printf("Descr: head tran=[%08X] head recv=[%08X]\n",
dep->sendrecv_descr_phys_addr[DESCR_TRAN],
dep->sendrecv_descr_phys_addr[DESCR_RECV]));
/* check alignment just to be extra safe */
for(i=0;i<2;i++){
loc_descr = &dep->descr[i][0];
for(j=0;j< (i==DESCR_RECV?DE_NB_RECV_DESCR:DE_NB_SEND_DESCR);j++){
r = sys_umap(SELF, VM_D, &(loc_descr->descr), sizeof(de_descr_t),
&temp);
if(r != OK)
panic(str_UmapErrMsg, r);
if( ((loc_descr->descr->des[DES_BUF1] & 0x3) != 0) ||
((loc_descr->descr->des[DES_BUF2] & 0x3) != 0) ||
((temp&0x3)!=0) )
panic(str_AlignErrMsg, temp);
loc_descr++;
}
}
/* Init default values */
dep->cur_descr[DESCR_TRAN]=1;
dep->cur_descr[DESCR_RECV]=0;
dep->bytes_rx = 0;
dep->bytes_tx = 0;
/* Set the interrupt handler policy. Request interrupts not to be reenabled
* automatically. Return the IRQ line number when an interrupt occurs.
*/
dep->de_hook = dep->de_irq;
sys_irqsetpolicy(dep->de_irq, 0, &dep->de_hook);
sys_irqenable(&dep->de_hook);
}
PRIVATE void do_interrupt(const dpeth_t *dep){
u32_t val;
val = io_inl(CSR_ADDR(dep, CSR5));
if(val & CSR5_AIS){
panic("Abnormal Int CSR5=: %d", val);
}
if( (dep->de_flags & DEF_READING) && (val & CSR5_RI) ){
do_vread_s(&dep->rx_return_msg, TRUE);
}
if( (dep->de_flags & DEF_SENDING) && (val & CSR5_TI) ){
do_vwrite_s(&dep->tx_return_msg, TRUE);
}
/* ack and reset interrupts */
io_outl(CSR_ADDR(dep, CSR5), 0xFFFFFFFF);
return;
}
PRIVATE void de_reset(const dpeth_t *dep){
io_outl(CSR_ADDR(dep, CSR0), CSR0_SWR);
}
PRIVATE void de_hw_conf(const dpeth_t *dep){
u32_t val;
/* CSR0 - global host bus prop */
val = CSR0_BAR | CSR0_CAL_8;
io_outl(CSR_ADDR(dep, CSR0), val);
/* CSR3 - Receive list BAR */
val = dep->sendrecv_descr_phys_addr[DESCR_RECV];
io_outl(CSR_ADDR(dep, CSR3), val);
/* CSR4 - Transmit list BAR */
val = dep->sendrecv_descr_phys_addr[DESCR_TRAN];
io_outl(CSR_ADDR(dep, CSR4), val);
/* CSR7 - interrupt mask */
val = CSR7_TI | CSR7_RI | CSR7_AI;
io_outl(CSR_ADDR(dep, CSR7), val);
/* CSR6 - operating mode register */
val = CSR6_MBO | CSR6_PS | CSR6_FD | CSR6_HBD |
CSR6_PCS | CSR6_SCR | CSR6_TR_00;
io_outl(CSR_ADDR(dep, CSR6), val);
}
PRIVATE void de_start(const dpeth_t *dep){
u32_t val;
val = io_inl(CSR_ADDR(dep, CSR6)) | CSR6_ST | CSR6_SR;
io_outl(CSR_ADDR(dep, CSR6), val);
}
PRIVATE void de_setup_frame(const dpeth_t *dep){
int i;
u32_t val;
/* this is not perfect... we assume pass all multicast and only
filter non-multicast frames */
dep->descr[DESCR_TRAN][0].buf1[0] = 0xFF;
dep->descr[DESCR_TRAN][0].buf1[1] = 0xFF;
dep->descr[DESCR_TRAN][0].buf1[4] = 0xFF;
dep->descr[DESCR_TRAN][0].buf1[5] = 0xFF;
dep->descr[DESCR_TRAN][0].buf1[8] = 0xFF;
dep->descr[DESCR_TRAN][0].buf1[9] = 0xFF;
for(i=1;i<16;i++){
memset(&(dep->descr[DESCR_TRAN][0].buf1[12*i]), 0, 12);
dep->descr[DESCR_TRAN][0].buf1[12*i+0] = dep->de_address.ea_addr[0];
dep->descr[DESCR_TRAN][0].buf1[12*i+1] = dep->de_address.ea_addr[1];
dep->descr[DESCR_TRAN][0].buf1[12*i+4] = dep->de_address.ea_addr[2];
dep->descr[DESCR_TRAN][0].buf1[12*i+5] = dep->de_address.ea_addr[3];
dep->descr[DESCR_TRAN][0].buf1[12*i+8] = dep->de_address.ea_addr[4];
dep->descr[DESCR_TRAN][0].buf1[12*i+9] = dep->de_address.ea_addr[5];
}
dep->descr[DESCR_TRAN][0].descr->des[DES0] = DES0_OWN;
dep->descr[DESCR_TRAN][0].descr->des[DES1] = DES1_SET |
DE_SETUP_FRAME_SIZE | DES1_IC;
/* start transmit process to process setup frame */
val = io_inl(CSR_ADDR(dep, CSR6)) | CSR6_ST;
io_outl(CSR_ADDR(dep, CSR6), val);
io_outl(CSR_ADDR(dep, CSR1), 0xFFFFFFFF);
return;
}
PRIVATE int de_calc_iov_size(iovec_dat_s_t * iovp){
int size, ix;
size = ix = 0;
do{
size += iovp->iod_iovec[ix].iov_size;
if (++ix >= IOVEC_NR) {
de_next_iov(iovp);
ix = 0;
}
} while (ix < iovp->iod_iovec_s);
return size;
}
PRIVATE void de_get_userdata_s(int user_proc, cp_grant_id_t grant,
vir_bytes offset, int count, void *loc_addr){
int rc;
vir_bytes len;
len = (count > IOVEC_NR ? IOVEC_NR : count) * sizeof(iovec_t);
rc = sys_safecopyfrom(user_proc, grant, 0, (vir_bytes)loc_addr, len, D);
if (rc != OK)
panic(str_CopyErrMsg, rc);
return;
}
PRIVATE void de_next_iov(iovec_dat_s_t * iovp){
iovp->iod_iovec_s -= IOVEC_NR;
iovp->iod_iovec_offset += IOVEC_NR * sizeof(iovec_t);
de_get_userdata_s(iovp->iod_proc_nr, iovp->iod_grant, iovp->iod_iovec_offset,
iovp->iod_iovec_s, iovp->iod_iovec);
return;
}
PRIVATE void do_vwrite_s(const message * mp, int from_int){
static u8_t setupDone = 0;
int size, r, bytes, ix, totalsize;
dpeth_t *dep;
iovec_dat_s_t *iovp = NULL;
de_loc_descr_t *descr = NULL;
char *buffer = NULL;
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
dep = &de_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
dep->de_client = mp->m_source;
if (dep->de_mode == DEM_ENABLED) {
if (!from_int && (dep->de_flags & DEF_SENDING))
panic(str_BusyErrMsg);
descr = &dep->descr[DESCR_TRAN][dep->cur_descr[DESCR_TRAN]];
if(( descr->descr->des[DES0] & DES0_OWN)!=0)
goto suspend;
if(!setupDone && (dep->cur_descr[DESCR_TRAN] == 0) ){
dep->descr[DESCR_TRAN][0].descr->des[DES0] = 0;
setupDone=1;
}
buffer = descr->buf1;
iovp = &dep->de_write_iovec;
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
iovp->iod_proc_nr = mp->DL_ENDPT;
de_get_userdata_s(mp->DL_ENDPT, mp->DL_GRANT, 0,
mp->DL_COUNT, iovp->iod_iovec);
iovp->iod_iovec_s = mp->DL_COUNT;
iovp->iod_grant = (cp_grant_id_t) mp->DL_GRANT;
iovp->iod_iovec_offset = 0;
totalsize = size = de_calc_iov_size(iovp);
if (size < ETH_MIN_PACK_SIZE || size > ETH_MAX_PACK_SIZE)
panic(str_SizeErrMsg, size);
dep->bytes_tx += size;
dep->de_stat.ets_packetT++;
ix=0;
do {
bytes = iovp->iod_iovec[ix].iov_size;
if (bytes >= size)
bytes = size;
r= sys_safecopyfrom(iovp->iod_proc_nr, iovp->iod_iovec[ix].iov_grant,
0, (vir_bytes)buffer, bytes, D);
if (r != OK)
panic(str_CopyErrMsg, r);
buffer += bytes;
if (++ix >= IOVEC_NR) {
de_next_iov(iovp);
ix = 0;
}
} while ((size -= bytes) > 0);
descr->descr->des[DES1] = (descr->descr->des[DES1]&DES1_ER) |
DES1_FS | DES1_LS | DES1_IC | totalsize;
descr->descr->des[DES0] = DES0_OWN;
dep->cur_descr[DESCR_TRAN]++;
if(dep->cur_descr[DESCR_TRAN] >= DE_NB_SEND_DESCR)
dep->cur_descr[DESCR_TRAN] = 0;
io_outl(CSR_ADDR(dep, CSR1), 0xFFFFFFFF);
}
dep->de_flags |= DEF_ACK_SEND;
if(from_int){
dep->de_flags &= NOT(DEF_SENDING);
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
do_reply(dep);
return;
suspend:
if(from_int)
panic("should not happen: %d", 0);
dep->de_stat.ets_transDef++;
dep->de_flags |= DEF_SENDING;
dep->de_stat.ets_transDef++;
dep->tx_return_msg = *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
do_reply(dep);
}
PRIVATE void warning(const char *type, int err){
printf("Warning: %s sys_%s failed (%d)\n", str_DevName, type, err);
return;
}
PRIVATE u32_t io_inl(u16_t port){
u32_t value;
int rc;
if ((rc = sys_inl(port, &value)) != OK) warning("inl", rc);
return value;
}
PRIVATE void io_outl(u16_t port, u32_t value){
int rc;
if ((rc = sys_outl(port, value)) != OK) warning("outl", rc);
return;
}