minix/drivers/atl2/atl2.c

1298 lines
35 KiB
C
Raw Normal View History

/* Attansic/Atheros L2 FastEthernet driver, by D.C. van Moolenbroek */
/* No documentation is available for this card. The FreeBSD driver is based
* heavily on the official Linux driver; this driver is based heavily on both.
*/
#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 <sys/mman.h>
#include <minix/ds.h>
#include <minix/vm.h>
#include <machine/pci.h>
#include <net/gen/ether.h>
#include <net/gen/eth_io.h>
#include <assert.h>
#include "atl2.h"
#define VERBOSE 0 /* Verbose debugging output */
#define ATL2_FKEY 1 /* Register Shift+F11 for dumping statistics */
#if VERBOSE
#define ATL2_DEBUG(x) printf x
#else
#define ATL2_DEBUG(x)
#endif
typedef struct {
u32_t hdr;
u32_t vtag;
u8_t data[ATL2_RXD_SIZE - sizeof(u32_t) * 2];
} rxd_t;
PRIVATE struct {
int devind; /* PCI device index */
int irq; /* IRQ number */
int hook_id; /* IRQ hook ID */
int mode; /* datalink mode */
char *base; /* base address of memory-mapped registers */
2010-07-01 11:10:16 +02:00
u32_t size; /* size of memory-mapped area */
u32_t hwaddr[2]; /* MAC address, in register representation */
u8_t *txd_base; /* local address of TxD ring buffer base */
u32_t *txs_base; /* local address of TxS ring buffer base */
u8_t *rxd_base_u; /* unaligned base address of RxD ring buffer */
rxd_t *rxd_base; /* local address of RxD ring buffer base */
int rxd_align; /* alignment offset of RxD ring buffer */
vir_bytes txd_phys; /* physical address of TxD ring buffer */
vir_bytes txs_phys; /* physical address of TxS ring buffer */
vir_bytes rxd_phys; /* physical address of RxD ring buffer */
int txd_tail; /* tail index into TxD, in bytes */
int txd_num; /* head-tail offset into TxD, in bytes */
int txs_tail; /* tail index into TxS, in elements */
int txs_num; /* head-tail offset into TxS, in elements */
int rxd_tail; /* tail index into RxD, in elements */
int flags; /* state flags (ATL2_FLAG_) */
message read_msg; /* suspended read request (READ_PEND) */
message write_msg; /* suspended write request (WRITE_PEND) */
endpoint_t task_endpt; /* requester endpoint (PACK_RCVD|PACK_SENT) */
size_t recv_count; /* packet size (PACK_RCVD) */
eth_stat_t stat; /* statistics */
} state;
#define ATL2_FLAG_RX_AVAIL 0x01 /* packet available for receipt */
#define ATL2_FLAG_READ_PEND 0x02 /* read request pending */
#define ATL2_FLAG_WRITE_PEND 0x04 /* write request pending */
#define ATL2_FLAG_PACK_RCVD 0x08 /* packet received */
#define ATL2_FLAG_PACK_SENT 0x10 /* packet transmitted */
#define ATL2_READ_U8(off) (* (u8_t *) (state.base + (off)))
#define ATL2_READ_U16(off) (* (u16_t *) (state.base + (off)))
#define ATL2_READ_U32(off) (* (u32_t *) (state.base + (off)))
#define ATL2_WRITE_U8(off, val) * (u8_t *) (state.base + (off)) = (val);
#define ATL2_WRITE_U16(off, val) * (u16_t *) (state.base + (off)) = (val);
#define ATL2_WRITE_U32(off, val) * (u32_t *) (state.base + (off)) = (val);
#define ATL2_ALIGN_32(n) (((n) + 3) & ~3)
PRIVATE iovec_s_t iovec[NR_IOREQS];
PRIVATE struct {
u16_t vid;
u16_t did;
} pcitab[] = {
{ 0x1969, 0x2048 }, /* Attansic Technology Corp, L2 FastEthernet */
{ 0x0000, 0x0000 }
};
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 int instance;
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
/*===========================================================================*
* atl2_read_vpd *
*===========================================================================*/
PRIVATE int atl2_read_vpd(int index, u32_t *res)
{
/* Read a value from the VPD register area.
*/
u32_t off, val;
int i;
ATL2_WRITE_U32(ATL2_VPD_DATA_REG, 0);
off = ATL2_VPD_REGBASE + index * sizeof(u32_t);
ATL2_WRITE_U32(ATL2_VPD_CAP_REG,
(off << ATL2_VPD_CAP_ADDR_SHIFT) & ATL2_VPD_CAP_ADDR_MASK);
for (i = 0; i < ATL2_VPD_NTRIES; i++) {
micro_delay(ATL2_VPD_DELAY);
val = ATL2_READ_U32(ATL2_VPD_CAP_REG);
if (val & ATL2_VPD_CAP_DONE)
break;
}
if (i == ATL2_VPD_NTRIES) {
printf("ATL2: timeout reading EEPROM register %d\n", index);
return FALSE;
}
*res = ATL2_READ_U32(ATL2_VPD_DATA_REG);
return TRUE;
}
/*===========================================================================*
* atl2_get_vpd_hwaddr *
*===========================================================================*/
PRIVATE int atl2_get_vpd_hwaddr(void)
{
/* Read the MAC address from the EEPROM, using the Vital Product Data
* register interface.
*/
u32_t key, val;
int i, n, found[2];
/* No idea, copied from FreeBSD which copied it from Linux. */
val = ATL2_READ_U32(ATL2_SPICTL_REG);
if (val & ATL2_SPICTL_VPD_EN) {
val &= ~ATL2_SPICTL_VPD_EN;
ATL2_WRITE_U32(ATL2_SPICTL_REG, val);
}
/* Is VPD supported? */
#ifdef PCI_CAP_VPD /* FIXME: just a guess at the future name */
if (!pci_find_cap(state.devind, PCI_CAP_VPD, &n))
return FALSE;
#endif
/* Read out the set of key/value pairs. Look for the two parts that
* make up the MAC address.
*/
found[0] = found[1] = FALSE;
for (i = 0; i < ATL2_VPD_NREGS; i += 2) {
if (!atl2_read_vpd(i, &key))
break;
if ((key & ATL2_VPD_SIG_MASK) != ATL2_VPD_SIG)
break;
key >>= ATL2_VPD_REG_SHIFT;
if (key != ATL2_HWADDR0_REG && key != ATL2_HWADDR1_REG)
continue;
if (!atl2_read_vpd(i + 1, &val))
break;
n = (key == ATL2_HWADDR1_REG);
state.hwaddr[n] = val;
found[n] = TRUE;
if (found[1 - n]) break;
}
return found[0] && found[1];
}
/*===========================================================================*
* atl2_get_hwaddr *
*===========================================================================*/
PRIVATE void atl2_get_hwaddr(void)
{
/* Get the MAC address of the card. First try the EEPROM; if that
* fails, just use whatever the card was already set to.
*/
if (!atl2_get_vpd_hwaddr()) {
printf("ATL2: unable to read from VPD\n");
state.hwaddr[0] = ATL2_READ_U32(ATL2_HWADDR0_REG);
state.hwaddr[1] = ATL2_READ_U32(ATL2_HWADDR1_REG) & 0xffff;
}
ATL2_DEBUG(("ATL2: MAC address %04lx%08lx\n",
state.hwaddr[1], state.hwaddr[0]));
}
/*===========================================================================*
* atl2_read_mdio *
*===========================================================================*/
PRIVATE int atl2_read_mdio(int addr, u16_t *res)
{
/* Read a MII PHY register using MDIO.
*/
u32_t rval;
int i;
rval = ((addr << ATL2_MDIO_ADDR_SHIFT) & ATL2_MDIO_ADDR_MASK) |
ATL2_MDIO_START | ATL2_MDIO_READ | ATL2_MDIO_SUP_PREAMBLE |
ATL2_MDIO_CLK_25_4;
ATL2_WRITE_U32(ATL2_MDIO_REG, rval);
for (i = 0; i < ATL2_MDIO_NTRIES; i++) {
micro_delay(ATL2_MDIO_DELAY);
rval = ATL2_READ_U32(ATL2_MDIO_REG);
if (!(rval & (ATL2_MDIO_START | ATL2_MDIO_BUSY)))
break;
}
if (i == ATL2_MDIO_NTRIES) return FALSE;
*res = (u16_t) (rval & ATL2_MDIO_DATA_MASK);
return TRUE;
}
/*===========================================================================*
* atl2_alloc_dma *
*===========================================================================*/
PRIVATE int atl2_alloc_dma(void)
{
/* Allocate DMA ring buffers.
*/
state.txd_base = alloc_contig(ATL2_TXD_BUFSIZE,
AC_ALIGN4K, &state.txd_phys);
state.txs_base = alloc_contig(ATL2_TXS_COUNT * sizeof(u32_t),
AC_ALIGN4K, &state.txs_phys);
/* The data buffer in each RxD descriptor must be 128-byte aligned.
* The two Tx buffers merely require a 4-byte start alignment.
*/
state.rxd_align = 128 - offsetof(rxd_t, data);
state.rxd_base_u =
alloc_contig(state.rxd_align + ATL2_RXD_COUNT * ATL2_RXD_SIZE,
AC_ALIGN4K, &state.rxd_phys);
/* Unlike mmap, alloc_contig returns NULL on failure. */
if (!state.txd_base || !state.txs_base || !state.rxd_base_u)
return ENOMEM;
state.rxd_base = (rxd_t *) (state.rxd_base_u + state.rxd_align);
state.rxd_phys += state.rxd_align;
/* Zero out just in case. */
memset(state.txd_base, 0, ATL2_TXD_BUFSIZE);
memset(state.txs_base, 0, ATL2_TXS_COUNT * sizeof(u32_t));
memset(state.rxd_base, 0, ATL2_RXD_COUNT * ATL2_RXD_SIZE);
return OK;
}
/*===========================================================================*
* atl2_stop *
*===========================================================================*/
PRIVATE int atl2_stop(void)
{
/* Stop the device.
*/
u32_t val;
int i;
/* Clear and disable interrupts. */
ATL2_WRITE_U32(ATL2_IMR_REG, 0);
ATL2_WRITE_U32(ATL2_ISR_REG, 0xffffffff);
/* Stop Rx/Tx MACs. */
val = ATL2_READ_U32(ATL2_MAC_REG);
if (val & (ATL2_MAC_RX_EN | ATL2_MAC_TX_EN)) {
val &= ~(ATL2_MAC_RX_EN | ATL2_MAC_TX_EN);
ATL2_WRITE_U32(ATL2_MAC_REG, val);
}
ATL2_WRITE_U8(ATL2_DMAWRITE_REG, 0);
ATL2_WRITE_U8(ATL2_DMAREAD_REG, 0);
/* Wait until everything is idle. */
for (i = 0; i < ATL2_IDLE_NTRIES; i++) {
if (ATL2_READ_U32(ATL2_IDLE_REG) == 0)
break;
micro_delay(ATL2_IDLE_DELAY);
}
/* The caller will generally ignore this return value. */
return (i < ATL2_IDLE_NTRIES);
}
/*===========================================================================*
* atl2_reset *
*===========================================================================*/
PRIVATE int atl2_reset(void)
{
/* Reset the device to a known good state.
*/
u32_t val;
int i;
/* Issue a soft reset, and wait for the device to respond. */
ATL2_WRITE_U32(ATL2_MASTER_REG, ATL2_MASTER_SOFT_RESET);
for (i = 0; i < ATL2_RESET_NTRIES; i++) {
val = ATL2_READ_U32(ATL2_MASTER_REG);
if (!(val & ATL2_MASTER_SOFT_RESET))
break;
micro_delay(ATL2_RESET_DELAY);
}
if (i == ATL2_RESET_NTRIES)
return FALSE;
/* Wait until everything is idle. */
for (i = 0; i < ATL2_IDLE_NTRIES; i++) {
if (ATL2_READ_U32(ATL2_IDLE_REG) == 0)
break;
micro_delay(ATL2_IDLE_DELAY);
}
return (i < ATL2_IDLE_NTRIES);
}
/*===========================================================================*
* atl2_set_mode *
*===========================================================================*/
PRIVATE void atl2_set_mode(void)
{
/* Reconfigure the device's promiscuity, multicast, and broadcast mode
* settings.
*/
u32_t val;
val = ATL2_READ_U32(ATL2_MAC_REG);
val &= ~(ATL2_MAC_PROMISC_EN | ATL2_MAC_MCAST_EN | ATL2_MAC_BCAST_EN);
if (state.mode & DL_PROMISC_REQ)
val |= ATL2_MAC_PROMISC_EN;
if (state.mode & DL_MULTI_REQ)
val |= ATL2_MAC_MCAST_EN;
if (state.mode & DL_BROAD_REQ)
val |= ATL2_MAC_BCAST_EN;
ATL2_WRITE_U32(ATL2_MAC_REG, val);
}
/*===========================================================================*
* atl2_setup *
*===========================================================================*/
PRIVATE int atl2_setup(void)
{
/* Set up the device for normal operation.
*/
u32_t val;
atl2_stop();
if (!atl2_reset())
return FALSE;
/* Initialize PCIE module. Magic. */
ATL2_WRITE_U32(ATL2_LTSSM_TESTMODE_REG, ATL2_LTSSM_TESTMODE_DEFAULT);
ATL2_WRITE_U32(ATL2_DLL_TX_CTRL_REG, ATL2_DLL_TX_CTRL_DEFAULT);
/* Enable PHY. */
ATL2_WRITE_U32(ATL2_PHY_ENABLE_REG, ATL2_PHY_ENABLE);
micro_delay(1000);
/* Clear and disable interrupts. */
ATL2_WRITE_U32(ATL2_ISR_REG, 0xffffffff);
/* Set the MAC address. */
ATL2_WRITE_U32(ATL2_HWADDR0_REG, state.hwaddr[0]);
ATL2_WRITE_U32(ATL2_HWADDR1_REG, state.hwaddr[1]);
/* Initialize ring buffer addresses and sizes. */
ATL2_WRITE_U32(ATL2_DESC_ADDR_HI_REG, 0); /* no 64 bit */
ATL2_WRITE_U32(ATL2_TXD_ADDR_LO_REG, state.txd_phys);
ATL2_WRITE_U32(ATL2_TXS_ADDR_LO_REG, state.txs_phys);
ATL2_WRITE_U32(ATL2_RXD_ADDR_LO_REG, state.rxd_phys);
ATL2_WRITE_U16(ATL2_RXD_COUNT_REG, ATL2_RXD_COUNT);
ATL2_WRITE_U16(ATL2_TXD_BUFSIZE_REG, ATL2_TXD_BUFSIZE / sizeof(u32_t));
ATL2_WRITE_U16(ATL2_TXS_COUNT_REG, ATL2_TXS_COUNT);
/* A whole lot of other initialization copied from Linux/FreeBSD. */
ATL2_WRITE_U32(ATL2_IFG_REG, ATL2_IFG_DEFAULT);
ATL2_WRITE_U32(ATL2_HDPX_REG, ATL2_HDPX_DEFAULT);
ATL2_WRITE_U16(ATL2_IMT_REG, ATL2_IMT_DEFAULT);
val = ATL2_READ_U32(ATL2_MASTER_REG);
ATL2_WRITE_U32(ATL2_MASTER_REG, val | ATL2_MASTER_IMT_EN);
ATL2_WRITE_U16(ATL2_ICT_REG, ATL2_ICT_DEFAULT);
ATL2_WRITE_U32(ATL2_CUT_THRESH_REG, ATL2_CUT_THRESH_DEFAULT);
ATL2_WRITE_U16(ATL2_FLOW_THRESH_HI_REG, (ATL2_RXD_COUNT / 8) * 7);
ATL2_WRITE_U16(ATL2_FLOW_THRESH_LO_REG, ATL2_RXD_COUNT / 12);
/* Set MTU. */
ATL2_WRITE_U16(ATL2_MTU_REG, ATL2_MTU_DEFAULT);
/* Reset descriptors, and enable DMA. */
state.txd_tail = state.txs_tail = state.rxd_tail = 0;
state.txd_num = state.txs_num = 0;
state.flags &= ~ATL2_FLAG_RX_AVAIL;
ATL2_WRITE_U16(ATL2_TXD_IDX_REG, 0);
ATL2_WRITE_U16(ATL2_RXD_IDX_REG, 0);
ATL2_WRITE_U8(ATL2_DMAREAD_REG, ATL2_DMAREAD_EN);
ATL2_WRITE_U8(ATL2_DMAWRITE_REG, ATL2_DMAWRITE_EN);
/* Did everything go alright? */
val = ATL2_READ_U32(ATL2_ISR_REG);
if (val & ATL2_ISR_PHY_LINKDOWN) {
printf("ATL2: initialization failed\n");
return FALSE;
}
/* Clear interrupt status. */
ATL2_WRITE_U32(ATL2_ISR_REG, 0x3fffffff);
ATL2_WRITE_U32(ATL2_ISR_REG, 0);
/* Enable interrupts. */
ATL2_WRITE_U32(ATL2_IMR_REG, ATL2_IMR_DEFAULT);
/* Configure MAC. */
ATL2_WRITE_U32(ATL2_MAC_REG, ATL2_MAC_DEFAULT);
/* Inet does not tell us about the multicast addresses that it is
* interested in, so we have to simply accept all multicast packets.
*/
ATL2_WRITE_U32(ATL2_MHT0_REG, 0xffffffff);
ATL2_WRITE_U32(ATL2_MHT1_REG, 0xffffffff);
atl2_set_mode();
/* Enable Tx/Rx. */
val = ATL2_READ_U32(ATL2_MAC_REG);
ATL2_WRITE_U32(ATL2_MAC_REG, val | ATL2_MAC_TX_EN | ATL2_MAC_RX_EN);
return TRUE;
}
/*===========================================================================*
* atl2_probe *
*===========================================================================*/
PRIVATE int atl2_probe(int instance)
{
/* Find a matching PCI device.
*/
u16_t vid, did;
char *dname;
int i, r, devind, skip;
pci_init();
r = pci_first_dev(&devind, &vid, &did);
if (r <= 0)
return -1;
skip = 0;
for (;;) {
for (i = 0; pcitab[i].vid != 0; i++)
if (pcitab[i].vid == vid && pcitab[i].did == did)
break;
if (pcitab[i].vid != 0) {
if (skip == instance) break;
skip++;
}
r = pci_next_dev(&devind, &vid, &did);
if (r <= 0)
return -1;
}
dname = pci_dev_name(vid, did);
ATL2_DEBUG(("ATL2: found %s (%x/%x) at %s\n",
dname ? dname : "<unknown>", vid, did,
pci_slot_name(devind)));
pci_reserve(devind);
return devind;
}
/*===========================================================================*
* atl2_init *
*===========================================================================*/
PRIVATE void atl2_init(int devind)
{
/* Initialize the device.
*/
u32_t bar;
2010-07-01 11:10:16 +02:00
int r, flag;
/* Initialize global state. */
state.devind = devind;
state.mode = DL_NOMODE;
state.flags = 0;
state.recv_count = 0;
memset(&state.stat, 0, sizeof(state.stat));
2010-07-01 11:10:16 +02:00
if ((r = pci_get_bar(devind, PCI_BAR, &bar, &state.size, &flag)) != OK)
panic("unable to retrieve bar: %d", r);
2010-07-01 11:10:16 +02:00
if (state.size < ATL2_MIN_MMAP_SIZE || flag)
panic("invalid register bar");
state.base = vm_map_phys(SELF, (void *) bar, state.size);
if (state.base == MAP_FAILED)
panic("unable to map in registers");
if ((r = atl2_alloc_dma()) != OK)
panic("unable to allocate DMA buffers: %d", r);
state.irq = pci_attr_r8(devind, PCI_ILR);
if ((r = sys_irqsetpolicy(state.irq, 0, &state.hook_id)) != OK)
panic("unable to register IRQ: %d", r);
if (!atl2_reset())
panic("unable to reset hardware");
if ((r = sys_irqenable(&state.hook_id)) != OK)
panic("unable to enable IRQ: %d", r);
atl2_get_hwaddr();
atl2_setup();
}
/*===========================================================================*
* atl2_tx_stat *
*===========================================================================*/
PRIVATE void atl2_tx_stat(u32_t stat)
{
/* Update statistics for packet transmission.
*/
if (stat & ATL2_TXS_SUCCESS)
state.stat.ets_packetT++;
else
state.stat.ets_recvErr++;
if (stat & ATL2_TXS_DEFER)
state.stat.ets_transDef++;
if (stat & (ATL2_TXS_EXCDEFER | ATL2_TXS_ABORTCOL))
state.stat.ets_transAb++;
if (stat & ATL2_TXS_SINGLECOL)
state.stat.ets_collision++;
if (stat & ATL2_TXS_MULTICOL)
state.stat.ets_collision++;
if (stat & ATL2_TXS_LATECOL)
state.stat.ets_OWC++;
if (stat & ATL2_TXS_UNDERRUN)
state.stat.ets_fifoUnder++;
}
/*===========================================================================*
* atl2_rx_stat *
*===========================================================================*/
PRIVATE void atl2_rx_stat(u32_t stat)
{
/* Update statistics for packet receipt.
*/
if (stat & ATL2_RXD_SUCCESS)
state.stat.ets_packetR++;
else
state.stat.ets_recvErr++;
if (stat & ATL2_RXD_CRCERR)
state.stat.ets_CRCerr++;
if (stat & ATL2_RXD_FRAG)
state.stat.ets_collision++;
if (stat & ATL2_RXD_TRUNC)
state.stat.ets_fifoOver++;
if (stat & ATL2_RXD_ALIGN)
state.stat.ets_frameAll++;
}
/*===========================================================================*
* atl2_tx_advance *
*===========================================================================*/
PRIVATE int atl2_tx_advance(void)
{
/* Advance the TxD/TxS tails by as many sent packets as found.
*/
u32_t stat, size, dsize;
int advanced;
advanced = FALSE;
while (state.txs_num > 0) {
/* Has the tail packet been processed by the driver? */
stat = state.txs_base[state.txs_tail];
if (!(stat & ATL2_TXS_UPDATE))
break;
/* The packet size from the status must match the packet size
* we put in. If they don't, there's not much we can do..
*/
size = stat & ATL2_TXS_SIZE_MASK;
assert(state.txd_tail <= ATL2_TXD_BUFSIZE - sizeof(u32_t));
dsize = * (u32_t *) (state.txd_base + state.txd_tail);
if (size != dsize)
printf("ATL2: TxD/TxS size mismatch (%lx vs %lx)\n",
size, dsize);
/* Advance tails accordingly. */
size = sizeof(u32_t) + ATL2_ALIGN_32(dsize);
assert(state.txd_num >= size);
state.txd_tail = (state.txd_tail + size) % ATL2_TXD_BUFSIZE;
state.txd_num -= size;
state.txs_tail = (state.txs_tail + 1) % ATL2_TXS_COUNT;
state.txs_num--;
if (stat & ATL2_TXS_SUCCESS)
ATL2_DEBUG(("ATL2: successfully sent packet\n"));
else
ATL2_DEBUG(("ATL2: failed to send packet\n"));
/* Update statistics. */
atl2_tx_stat(stat);
advanced = TRUE;
}
return advanced;
}
/*===========================================================================*
* atl2_rx_advance *
*===========================================================================*/
PRIVATE void atl2_rx_advance(int next)
{
/* Advance the RxD tail by as many failed receipts as possible, and
* see if there is an actual packet left to receive. If 'next' is set,
* the packet at the current tail has been processed.
*/
int update_tail;
rxd_t *rxd;
u32_t hdr, size;
update_tail = FALSE;
if (next) {
state.rxd_tail = (state.rxd_tail + 1) % ATL2_RXD_COUNT;
update_tail = TRUE;
ATL2_DEBUG(("ATL2: successfully received packet\n"));
state.flags &= ~ATL2_FLAG_RX_AVAIL;
}
assert(!(state.flags & ATL2_FLAG_RX_AVAIL));
for (;;) {
/* Check the RxD tail for updates. */
rxd = &state.rxd_base[state.rxd_tail];
hdr = rxd->hdr;
if (!(hdr & ATL2_RXD_UPDATE))
break;
rxd->hdr = hdr & ~(ATL2_RXD_UPDATE);
/* Update statistics. */
atl2_rx_stat(hdr);
/* Stop at the first successful receipt. The packet will be
* picked up by Inet later.
*/
size = hdr & ATL2_RXD_SIZE_MASK;
if ((hdr & ATL2_RXD_SUCCESS) && size >= ETH_MIN_PACK_SIZE) {
ATL2_DEBUG(("ATL2: packet available, size %ld\n",
size));
state.flags |= ATL2_FLAG_RX_AVAIL;
break;
}
ATL2_DEBUG(("ATL2: packet receipt failed\n"));
/* Advance tail. */
state.rxd_tail = (state.rxd_tail + 1) % ATL2_RXD_COUNT;
update_tail = TRUE;
}
/* If new RxD descriptors are now up for reuse, tell the device. */
if (update_tail)
ATL2_WRITE_U32(ATL2_RXD_IDX_REG, state.rxd_tail);
}
/*===========================================================================*
* atl2_reply *
*===========================================================================*/
PRIVATE void atl2_reply(void)
{
/* Send a task reply to Inet.
*/
message m;
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;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
flags = DL_NOFLAGS;
if (state.flags & ATL2_FLAG_PACK_SENT)
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
flags |= DL_PACK_SEND;
if (state.flags & ATL2_FLAG_PACK_RCVD)
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
flags |= DL_PACK_RECV;
m.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
m.DL_FLAGS = flags;
m.DL_COUNT = state.recv_count;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
ATL2_DEBUG(("ATL2: sending reply, flags %x count %d\n", flags,
m.DL_COUNT));
if ((r = send(state.task_endpt, &m)) != OK)
panic("unable to reply: %d", r);
state.flags &= ~(ATL2_FLAG_PACK_SENT | ATL2_FLAG_PACK_RCVD);
state.recv_count = 0;
}
/*===========================================================================*
* atl2_readv *
*===========================================================================*/
PRIVATE void atl2_readv(const message *m, int from_int)
{
/* Read packet data.
*/
rxd_t *rxd;
iovec_s_t *iovp;
size_t count, off, left, size;
u8_t *pos;
int i, j, r, batch;
/* We can deal with only one read request from Inet at a time. */
assert(from_int || !(state.flags & ATL2_FLAG_READ_PEND));
state.task_endpt = m->m_source;
/* Are there any packets available at all? */
if (!(state.flags & ATL2_FLAG_RX_AVAIL))
goto suspend;
/* Get the first available packet's size. Cut off the CRC. */
rxd = &state.rxd_base[state.rxd_tail];
count = rxd->hdr & ATL2_RXD_SIZE_MASK;
count -= ETH_CRC_SIZE;
ATL2_DEBUG(("ATL2: readv: found packet with length %d\n", count));
/* Copy out the packet. */
off = 0;
left = count;
pos = rxd->data;
for (i = 0; i < m->DL_COUNT && left > 0; i += batch) {
/* Copy in the next batch. */
batch = MIN(m->DL_COUNT - i, NR_IOREQS);
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
r = sys_safecopyfrom(m->DL_ENDPT, m->DL_GRANT, off,
(vir_bytes) iovec, batch * sizeof(iovec[0]), D);
if (r != OK)
panic("vector copy failed: %d", r);
/* Copy out each element in the batch, until we run out. */
for (j = 0, iovp = iovec; j < batch && left > 0; j++, iovp++) {
size = MIN(iovp->iov_size, left);
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
r = sys_safecopyto(m->DL_ENDPT, iovp->iov_grant, 0,
(vir_bytes) pos, size, D);
if (r != OK)
panic("safe copy failed: %d", r);
pos += size;
left -= size;
}
off += batch * sizeof(iovec[0]);
}
/* Not sure what to do here. Inet shouldn't mess this up anyway. */
if (left > 0) {
printf("ATL2: truncated packet of %d bytes by %d bytes\n",
count, left);
count -= left;
}
/* We are done with this packet. Move on to the next. */
atl2_rx_advance(TRUE /*next*/);
/* We have now successfully received a packet. */
state.flags &= ~ATL2_FLAG_READ_PEND;
state.flags |= ATL2_FLAG_PACK_RCVD;
state.recv_count = count;
/* If called from the interrupt handler, the caller will reply. */
if (!from_int)
atl2_reply();
return;
suspend:
/* No packets are available at this time. If we were not already
* trying to resume receipt, save the read request for later, and tell
* Inet that the request has been suspended.
*/
if (from_int)
return;
state.flags |= ATL2_FLAG_READ_PEND;
state.read_msg = *m;
atl2_reply();
}
/*===========================================================================*
* atl2_writev *
*===========================================================================*/
PRIVATE void atl2_writev(const message *m, int from_int)
{
/* Write packet data.
*/
iovec_s_t *iovp;
size_t off, count, left, pos, skip;
vir_bytes size;
u8_t *sizep;
int i, j, r, batch, maxnum;
/* We can deal with only one write request from Inet at a time. */
assert(from_int || !(state.flags & ATL2_FLAG_WRITE_PEND));
state.task_endpt = m->m_source;
/* If we are already certain that the packet won't fit, bail out.
* Keep at least some space between TxD head and tail, as it is not
* clear whether the device deals well with the case that they collide.
*/
if (state.txs_num >= ATL2_TXS_COUNT)
goto suspend;
maxnum = ATL2_TXD_BUFSIZE - ETH_MIN_PACK_SIZE - sizeof(u32_t);
if (state.txd_num >= maxnum)
goto suspend;
/* Optimistically try to copy in the data; suspend if it turns out
* that it does not fit.
*/
off = 0;
count = 0;
left = state.txd_num - sizeof(u32_t);
pos = (state.txd_tail + state.txd_num +
sizeof(u32_t)) % ATL2_TXD_BUFSIZE;
for (i = 0; i < m->DL_COUNT; i += batch) {
/* Copy in the next batch. */
batch = MIN(m->DL_COUNT - i, NR_IOREQS);
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
r = sys_safecopyfrom(m->DL_ENDPT, m->DL_GRANT, off,
(vir_bytes) iovec, batch * sizeof(iovec[0]), D);
if (r != OK)
panic("vector copy failed: %d", r);
/* Copy in each element in the batch. */
for (j = 0, iovp = iovec; j < batch; j++, iovp++) {
size = iovp->iov_size;
if (size > left)
goto suspend;
skip = 0;
if (size > ATL2_TXD_BUFSIZE - pos) {
skip = ATL2_TXD_BUFSIZE - pos;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
r = sys_safecopyfrom(m->DL_ENDPT,
iovp->iov_grant, 0,
(vir_bytes) (state.txd_base + pos),
skip, D);
if (r != OK)
panic("safe copy failed: %d", r);
pos = 0;
}
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
r = sys_safecopyfrom(m->DL_ENDPT, iovp->iov_grant,
skip, (vir_bytes) (state.txd_base + pos),
size - skip, D);
if (r != OK)
panic("safe copy failed: %d", r);
pos = (pos + size - skip) % ATL2_TXD_BUFSIZE;
left -= size;
count += size;
}
off += batch * sizeof(iovec[0]);
}
assert(count <= ETH_MAX_PACK_SIZE_TAGGED);
/* Write the length to the DWORD right before the packet. */
sizep = state.txd_base +
(state.txd_tail + state.txd_num) % ATL2_TXD_BUFSIZE;
* (u32_t *) sizep = count;
/* Update the TxD head. */
state.txd_num += sizeof(u32_t) + ATL2_ALIGN_32(count);
pos = ATL2_ALIGN_32(pos) % ATL2_TXD_BUFSIZE;
assert(pos == (state.txd_tail + state.txd_num) % ATL2_TXD_BUFSIZE);
/* Initialize and update the TxS head. */
state.txs_base[(state.txs_tail + state.txs_num) % ATL2_TXS_COUNT] = 0;
state.txs_num++;
/* Tell the device about our new position. */
ATL2_WRITE_U32(ATL2_TXD_IDX_REG, pos / sizeof(u32_t));
/* We have now successfully set up the transmission of a packet. */
state.flags &= ~ATL2_FLAG_WRITE_PEND;
state.flags |= ATL2_FLAG_PACK_SENT;
/* If called from the interrupt handler, the caller will reply. */
if (!from_int)
atl2_reply();
return;
suspend:
/* We cannot transmit the packet at this time. If we were not already
* trying to resume transmission, save the write request for later,
* and tell Inet that the request has been suspended.
*/
if (from_int)
return;
state.flags |= ATL2_FLAG_WRITE_PEND;
state.write_msg = *m;
atl2_reply();
}
/*===========================================================================*
* atl2_intr *
*===========================================================================*/
PRIVATE void atl2_intr(const message *m)
{
/* Interrupt received.
*/
u32_t val;
int r, try_write, try_read;
/* Clear and disable interrupts. */
val = ATL2_READ_U32(ATL2_ISR_REG);
ATL2_WRITE_U32(ATL2_ISR_REG, val | ATL2_ISR_DISABLE);
ATL2_DEBUG(("ATL2: interrupt (0x%08lx)\n", val));
/* If an error occurred, reset the card. */
if (val & (ATL2_ISR_DMAR_TIMEOUT | ATL2_ISR_DMAW_TIMEOUT |
ATL2_ISR_PHY_LINKDOWN)) {
atl2_setup();
}
try_write = try_read = FALSE;
/* Process sent data, and possibly send pending data. */
if (val & ATL2_ISR_TX_EVENT) {
if (atl2_tx_advance())
try_write = (state.flags & ATL2_FLAG_WRITE_PEND);
}
/* Receive new data, and possible satisfy a pending receive request. */
if (val & ATL2_ISR_RX_EVENT) {
if (!(state.flags & ATL2_FLAG_RX_AVAIL)) {
atl2_rx_advance(FALSE /*next*/);
try_read = (state.flags & ATL2_FLAG_READ_PEND);
}
}
/* Reenable interrupts. */
ATL2_WRITE_U32(ATL2_ISR_REG, 0);
if ((r = sys_irqenable(&state.hook_id)) != OK)
panic("unable to enable IRQ: %d", r);
/* Attempt to satisfy pending write and read requests. */
if (try_write)
atl2_writev(&state.write_msg, TRUE /*from_int*/);
if (try_read)
atl2_readv(&state.read_msg, TRUE /*from_int*/);
if (state.flags & (ATL2_FLAG_PACK_SENT | ATL2_FLAG_PACK_RCVD))
atl2_reply();
}
/*===========================================================================*
* atl2_conf *
*===========================================================================*/
PRIVATE void atl2_conf(message *m)
{
/* Configure the mode of the card.
*/
ether_addr_t *addr;
int r;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
state.mode = m->DL_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
atl2_set_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
addr = (ether_addr_t *) m->DL_HWADDR;
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
addr->ea_addr[0] = state.hwaddr[1] >> 8;
addr->ea_addr[1] = state.hwaddr[1] & 0xff;
addr->ea_addr[2] = state.hwaddr[0] >> 24;
addr->ea_addr[3] = (state.hwaddr[0] >> 16) & 0xff;
addr->ea_addr[4] = (state.hwaddr[0] >> 8) & 0xff;
addr->ea_addr[5] = state.hwaddr[0] & 0xff;
m->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
m->DL_STAT = OK;
if ((r = send(m->m_source, m)) != OK)
printf("ATL2: unable to send reply (%d)\n", r);
}
/*===========================================================================*
* atl2_getstat *
*===========================================================================*/
PRIVATE void atl2_getstat(message *m)
{
/* Copy out statistics.
*/
int r;
This patch switches the MINIX3 ethernet driver stack from a port-based model to an instance-based model. Each ethernet driver instance is now responsible for exactly one network interface card. The port field in /etc/inet.conf now acts as an instance field instead. This patch also updates the data link protocol. This update: - eliminates the concept of ports entirely; - eliminates DL_GETNAME entirely; - standardizes on using m_source for IPC and DL_ENDPT for safecopies; - removes error codes from TASK/STAT replies, as they were unused; - removes a number of other old or unused fields; - names and renames a few other fields. All ethernet drivers have been changed to: - conform to the new protocol, and exactly that; - take on an instance number based on a given "instance" argument; - skip that number of PCI devices in probe iterations; - use config tables and environment variables based on that number; - no longer be limited to a predefined maximum of cards in any way; - get rid of any leftover non-safecopy support and other ancient junk; - have a correct banner protocol figure, or none at all. Other changes: * Inet.conf is now taken to be line-based, and supports #-comments. No existing installations are expected to be affected by this. * A new, select-based asynchio library replaces the old one. Kindly contributed by Kees J. Bot. * Inet now supports use of select() on IP devices. Combined, the last two changes together speed up dhcpd considerably in the presence of multiple interfaces. * A small bug has been fixed in nonamed.
2010-05-18 00:22:53 +02:00
sys_safecopyto(m->DL_ENDPT, m->DL_GRANT, 0,
(vir_bytes) &state.stat, sizeof(state.stat), D);
m->m_type = DL_STAT_REPLY;
if ((r = send(m->m_source, m)) != OK)
printf("ATL2: unable to send reply (%d)\n", r);
}
/*===========================================================================*
* atl2_dump_link *
*===========================================================================*/
PRIVATE void atl2_dump_link(void)
{
/* Dump link status.
*/
u16_t val;
int link_up;
/* The link status bit is latched. Read the status register twice. */
atl2_read_mdio(ATL2_MII_BMSR, &val);
if (!atl2_read_mdio(ATL2_MII_BMSR, &val)) return;
link_up = val & ATL2_MII_BMSR_LSTATUS;
printf("link status: %4s\t", link_up ? "up" : "down");
if (!link_up) return;
if (!atl2_read_mdio(ATL2_MII_PSSR, &val)) return;
if (!(val & ATL2_MII_PSSR_RESOLVED)) {
printf("(not resolved)\n");
return;
}
switch (val & ATL2_MII_PSSR_SPEED) {
case ATL2_MII_PSSR_10: printf("(10Mbps "); break;
case ATL2_MII_PSSR_100: printf("(100Mbps "); break;
case ATL2_MII_PSSR_1000: printf("(1000Mbps "); break;
default: printf("(unknown, ");
}
printf("%s duplex)", (val & ATL2_MII_PSSR_DUPLEX) ? "full" : "half");
}
/*===========================================================================*
* atl2_dump *
*===========================================================================*/
PRIVATE void atl2_dump(void)
{
/* Dump statistics.
*/
printf("\n");
printf("Attansic L2 statistics:\n");
printf("recvErr: %8ld\t", state.stat.ets_recvErr);
printf("sendErr: %8ld\t", state.stat.ets_sendErr);
printf("OVW: %8ld\n", state.stat.ets_OVW);
printf("CRCerr: %8ld\t", state.stat.ets_CRCerr);
printf("frameAll: %8ld\t", state.stat.ets_frameAll);
printf("missedP: %8ld\n", state.stat.ets_missedP);
printf("packetR: %8ld\t", state.stat.ets_packetR);
printf("packetT: %8ld\t", state.stat.ets_packetT);
printf("transDef: %8ld\n", state.stat.ets_transDef);
printf("collision: %8ld\t", state.stat.ets_collision);
printf("transAb: %8ld\t", state.stat.ets_transAb);
printf("carrSense: %8ld\n", state.stat.ets_carrSense);
printf("fifoUnder: %8ld\t", state.stat.ets_fifoUnder);
printf("fifoOver: %8ld\t", state.stat.ets_fifoOver);
printf("CDheartbeat: %8ld\n", state.stat.ets_CDheartbeat);
printf("OWC: %8ld\t", state.stat.ets_OWC);
printf("TxD tail: %8d\t", state.txd_tail);
printf("TxD count: %8d\n", state.txd_num);
printf("RxD tail: %8d\t", state.rxd_tail);
printf("TxS tail: %8d\t", state.txs_tail);
printf("TxS count: %8d\n", state.txs_num);
printf("flags: 0x%04x\t", state.flags);
atl2_dump_link();
printf("\n");
}
/*===========================================================================*
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
* sef_cb_init_fresh *
*===========================================================================*/
PRIVATE int sef_cb_init_fresh(int type, sef_init_info_t *UNUSED(info))
{
2010-01-13 11:52:47 +01:00
/* Initialize the atl2 driver.
*/
int r, devind;
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;
#if ATL2_FKEY
int fkeys, sfkeys;
#endif
/* How many matching devices should we skip? */
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);
instance = (int) v;
/* Try to find a recognized device. */
devind = atl2_probe(instance);
if (devind < 0)
panic("no matching device found");
/* Initialize the device. */
atl2_init(devind);
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();
#if ATL2_FKEY
/* Register debug dump function key. */
fkeys = sfkeys = 0;
bit_set(sfkeys, 11);
if ((r = fkey_map(&fkeys, &sfkeys)) != OK)
printf("ATL2: warning, could not map Shift+F11 key (%d)\n", r);
#endif
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
return(OK);
}
New RS and new signal handling for system processes. UPDATING INFO: 20100317: /usr/src/etc/system.conf updated to ignore default kernel calls: copy it (or merge it) to /etc/system.conf. The hello driver (/dev/hello) added to the distribution: # cd /usr/src/commands/scripts && make clean install # cd /dev && MAKEDEV hello KERNEL CHANGES: - Generic signal handling support. The kernel no longer assumes PM as a signal manager for every process. The signal manager of a given process can now be specified in its privilege slot. When a signal has to be delivered, the kernel performs the lookup and forwards the signal to the appropriate signal manager. PM is the default signal manager for user processes, RS is the default signal manager for system processes. To enable ptrace()ing for system processes, it is sufficient to change the default signal manager to PM. This will temporarily disable crash recovery, though. - sys_exit() is now split into sys_exit() (i.e. exit() for system processes, which generates a self-termination signal), and sys_clear() (i.e. used by PM to ask the kernel to clear a process slot when a process exits). - Added a new kernel call (i.e. sys_update()) to swap two process slots and implement live update. PM CHANGES: - Posix signal handling is no longer allowed for system processes. System signals are split into two fixed categories: termination and non-termination signals. When a non-termination signaled is processed, PM transforms the signal into an IPC message and delivers the message to the system process. When a termination signal is processed, PM terminates the process. - PM no longer assumes itself as the signal manager for system processes. It now makes sure that every system signal goes through the kernel before being actually processes. The kernel will then dispatch the signal to the appropriate signal manager which may or may not be PM. SYSLIB CHANGES: - Simplified SEF init and LU callbacks. - Added additional predefined SEF callbacks to debug crash recovery and live update. - Fixed a temporary ack in the SEF init protocol. SEF init reply is now completely synchronous. - Added SEF signal event type to provide a uniform interface for system processes to deal with signals. A sef_cb_signal_handler() callback is available for system processes to handle every received signal. A sef_cb_signal_manager() callback is used by signal managers to process system signals on behalf of the kernel. - Fixed a few bugs with memory mapping and DS. VM CHANGES: - Page faults and memory requests coming from the kernel are now implemented using signals. - Added a new VM call to swap two process slots and implement live update. - The call is used by RS at update time and in turn invokes the kernel call sys_update(). RS CHANGES: - RS has been reworked with a better functional decomposition. - Better kernel call masks. com.h now defines the set of very basic kernel calls every system service is allowed to use. This makes system.conf simpler and easier to maintain. In addition, this guarantees a higher level of isolation for system libraries that use one or more kernel calls internally (e.g. printf). - RS is the default signal manager for system processes. By default, RS intercepts every signal delivered to every system process. This makes crash recovery possible before bringing PM and friends in the loop. - RS now supports fast rollback when something goes wrong while initializing the new version during a live update. - Live update is now implemented by keeping the two versions side-by-side and swapping the process slots when the old version is ready to update. - Crash recovery is now implemented by keeping the two versions side-by-side and cleaning up the old version only when the recovery process is complete. DS CHANGES: - Fixed a bug when the process doing ds_publish() or ds_delete() is not known by DS. - Fixed the completely broken support for strings. String publishing is now implemented in the system library and simply wraps publishing of memory ranges. Ideally, we should adopt a similar approach for other data types as well. - Test suite fixed. DRIVER CHANGES: - The hello driver has been added to the Minix distribution to demonstrate basic live update and crash recovery functionalities. - Other drivers have been adapted to conform the new SEF interface.
2010-03-17 02:15:29 +01:00
/*===========================================================================*
* sef_cb_signal_handler *
*===========================================================================*/
PRIVATE void sef_cb_signal_handler(int signo)
{
/* In case of a termination signal, shut down this driver.
* Stop the device, and deallocate resources as proof of concept.
*/
int r;
/* Only check for termination signal, ignore anything else. */
if (signo != SIGTERM) return;
atl2_stop();
if ((r = sys_irqrmpolicy(&state.hook_id)) != OK)
panic("unable to deregister IRQ: %d", r);
free_contig(state.txd_base, ATL2_TXD_BUFSIZE);
free_contig(state.txs_base, ATL2_TXS_COUNT * sizeof(u32_t));
free_contig(state.rxd_base_u,
state.rxd_align + ATL2_RXD_COUNT * ATL2_RXD_SIZE);
2010-07-01 11:10:16 +02:00
vm_unmap_phys(SELF, state.base, state.size);
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
/* We cannot free the PCI device at this time. */
exit(0);
}
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
/*===========================================================================*
* sef_local_startup *
*===========================================================================*/
PRIVATE void sef_local_startup(void)
{
/* Register init callbacks. */
sef_setcb_init_fresh(sef_cb_init_fresh);
Driver refactory for live update and crash recovery. SYSLIB CHANGES: - DS calls to publish / retrieve labels consider endpoints instead of u32_t. VFS CHANGES: - mapdriver() only adds an entry in the dmap table in VFS. - dev_up() is only executed upon reception of a driver up event. INET CHANGES: - INET no longer searches for existing drivers instances at startup. - A newtwork driver is (re)initialized upon reception of a driver up event. - Networking startup is now race-free by design. No need to waste 5 seconds at startup any more. DRIVER CHANGES: - Every driver publishes driver up events when starting for the first time or in case of restart when recovery actions must be taken in the upper layers. - Driver up events are published by drivers through DS. - For regular drivers, VFS is normally the only subscriber, but not necessarily. For instance, when the filter driver is in use, it must subscribe to driver up events to initiate recovery. - For network drivers, inet is the only subscriber for now. - Every VFS driver is statically linked with libdriver, every network driver is statically linked with libnetdriver. DRIVER LIBRARIES CHANGES: - Libdriver is extended to provide generic receive() and ds_publish() interfaces for VFS drivers. - driver_receive() is a wrapper for sef_receive() also used in driver_task() to discard spurious messages that were meant to be delivered to a previous version of the driver. - driver_receive_mq() is the same as driver_receive() but integrates support for queued messages. - driver_announce() publishes a driver up event for VFS drivers and marks the driver as initialized and expecting a DEV_OPEN message. - Libnetdriver is introduced to provide similar receive() and ds_publish() interfaces for network drivers (netdriver_announce() and netdriver_receive()). - Network drivers all support live update with no state transfer now. KERNEL CHANGES: - Added kernel call statectl for state management. Used by driver_announce() to unblock eventual callers sendrecing to the driver.
2010-04-08 15:41:35 +02:00
sef_setcb_init_lu(sef_cb_init_fresh);
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
sef_setcb_init_restart(sef_cb_init_fresh);
Driver refactory for live update and crash recovery. SYSLIB CHANGES: - DS calls to publish / retrieve labels consider endpoints instead of u32_t. VFS CHANGES: - mapdriver() only adds an entry in the dmap table in VFS. - dev_up() is only executed upon reception of a driver up event. INET CHANGES: - INET no longer searches for existing drivers instances at startup. - A newtwork driver is (re)initialized upon reception of a driver up event. - Networking startup is now race-free by design. No need to waste 5 seconds at startup any more. DRIVER CHANGES: - Every driver publishes driver up events when starting for the first time or in case of restart when recovery actions must be taken in the upper layers. - Driver up events are published by drivers through DS. - For regular drivers, VFS is normally the only subscriber, but not necessarily. For instance, when the filter driver is in use, it must subscribe to driver up events to initiate recovery. - For network drivers, inet is the only subscriber for now. - Every VFS driver is statically linked with libdriver, every network driver is statically linked with libnetdriver. DRIVER LIBRARIES CHANGES: - Libdriver is extended to provide generic receive() and ds_publish() interfaces for VFS drivers. - driver_receive() is a wrapper for sef_receive() also used in driver_task() to discard spurious messages that were meant to be delivered to a previous version of the driver. - driver_receive_mq() is the same as driver_receive() but integrates support for queued messages. - driver_announce() publishes a driver up event for VFS drivers and marks the driver as initialized and expecting a DEV_OPEN message. - Libnetdriver is introduced to provide similar receive() and ds_publish() interfaces for network drivers (netdriver_announce() and netdriver_receive()). - Network drivers all support live update with no state transfer now. KERNEL CHANGES: - Added kernel call statectl for state management. Used by driver_announce() to unblock eventual callers sendrecing to the driver.
2010-04-08 15:41:35 +02:00
/* Register live update callbacks. */
sef_setcb_lu_prepare(sef_cb_lu_prepare_always_ready);
sef_setcb_lu_state_isvalid(sef_cb_lu_state_isvalid_workfree);
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
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);
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
/* Let SEF perform startup. */
sef_startup();
}
/*===========================================================================*
* main *
*===========================================================================*/
int main(int argc, char **argv)
{
/* Driver task.
*/
message m;
Driver refactory for live update and crash recovery. SYSLIB CHANGES: - DS calls to publish / retrieve labels consider endpoints instead of u32_t. VFS CHANGES: - mapdriver() only adds an entry in the dmap table in VFS. - dev_up() is only executed upon reception of a driver up event. INET CHANGES: - INET no longer searches for existing drivers instances at startup. - A newtwork driver is (re)initialized upon reception of a driver up event. - Networking startup is now race-free by design. No need to waste 5 seconds at startup any more. DRIVER CHANGES: - Every driver publishes driver up events when starting for the first time or in case of restart when recovery actions must be taken in the upper layers. - Driver up events are published by drivers through DS. - For regular drivers, VFS is normally the only subscriber, but not necessarily. For instance, when the filter driver is in use, it must subscribe to driver up events to initiate recovery. - For network drivers, inet is the only subscriber for now. - Every VFS driver is statically linked with libdriver, every network driver is statically linked with libnetdriver. DRIVER LIBRARIES CHANGES: - Libdriver is extended to provide generic receive() and ds_publish() interfaces for VFS drivers. - driver_receive() is a wrapper for sef_receive() also used in driver_task() to discard spurious messages that were meant to be delivered to a previous version of the driver. - driver_receive_mq() is the same as driver_receive() but integrates support for queued messages. - driver_announce() publishes a driver up event for VFS drivers and marks the driver as initialized and expecting a DEV_OPEN message. - Libnetdriver is introduced to provide similar receive() and ds_publish() interfaces for network drivers (netdriver_announce() and netdriver_receive()). - Network drivers all support live update with no state transfer now. KERNEL CHANGES: - Added kernel call statectl for state management. Used by driver_announce() to unblock eventual callers sendrecing to the driver.
2010-04-08 15:41:35 +02:00
int ipc_status;
Initialization protocol for system services. SYSLIB CHANGES: - SEF framework now supports a new SEF Init request type from RS. 3 different callbacks are available (init_fresh, init_lu, init_restart) to specify initialization code when a service starts fresh, starts after a live update, or restarts. SYSTEM SERVICE CHANGES: - Initialization code for system services is now enclosed in a callback SEF will automatically call at init time. The return code of the callback will tell RS whether the initialization completed successfully. - Each init callback can access information passed by RS to initialize. As of now, each system service has access to the public entries of RS's system process table to gather all the information required to initialize. This design eliminates many existing or potential races at boot time and provides a uniform initialization interface to system services. The same interface will be reused for the upcoming publish/subscribe model to handle dynamic registration / deregistration of system services. VM CHANGES: - Uniform privilege management for all system services. Every service uses the same call mask format. For boot services, VM copies the call mask from init data. For dynamic services, VM still receives the call mask via rs_set_priv call that will be soon replaced by the upcoming publish/subscribe model. RS CHANGES: - The system process table has been reorganized and split into private entries and public entries. Only the latter ones are exposed to system services. - VM call masks are now entirely configured in rs/table.c - RS has now its own slot in the system process table. Only kernel tasks and user processes not included in the boot image are now left out from the system process table. - RS implements the initialization protocol for system services. - For services in the boot image, RS blocks till initialization is complete and panics when failure is reported back. Services are initialized in their order of appearance in the boot image priv table and RS blocks to implements synchronous initialization for every system service having the flag SF_SYNCH_BOOT set. - For services started dynamically, the initialization protocol is implemented as though it were the first ping for the service. In this case, if the system service fails to report back (or reports failure), RS brings the service down rather than trying to restart it.
2010-01-08 02:20:42 +01:00
int r;
/* Initialize SEF. */
env_setargs(argc, argv);
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 (m.m_source) {
case HARDWARE: /* interrupt */
atl2_intr(&m);
break;
case TTY_PROC_NR: /* function key */
atl2_dump();
break;
default:
printf("ATL2: illegal notify from %d\n",
m.m_source);
}
continue;
}
/* Process requests from Inet. */
switch (m.m_type) {
case DL_CONF: atl2_conf(&m); break;
case DL_GETSTAT_S: atl2_getstat(&m); break;
case DL_WRITEV_S: atl2_writev(&m, FALSE); break;
case DL_READV_S: atl2_readv(&m, FALSE); break;
default:
printf("ATL2: illegal message %d from %d\n",
m.m_type, m.m_source);
}
}
}