view packages/net/tcpip/current/src/ecos/support.c @ 112:02ea4320376c ecos-sw-2000-07-21

Merge from eCos master repository on 2000-07-21-21:01:39-BST
author jlarmour
date Fri, 21 Jul 2000 21:34:11 +0000
parents 84e4bde58b26
children 6ed91473a1cd
line wrap: on
line source

//==========================================================================
//
//      ecos/support.c
//
//      eCos wrapper and support functions
//
//==========================================================================
//####COPYRIGHTBEGIN####
//                                                                          
// -------------------------------------------                              
// The contents of this file are subject to the Red Hat eCos Public License 
// Version 1.1 (the "License"); you may not use this file except in         
// compliance with the License.  You may obtain a copy of the License at    
// http://www.redhat.com/                                                   
//                                                                          
// Software distributed under the License is distributed on an "AS IS"      
// basis, WITHOUT WARRANTY OF ANY KIND, either express or implied.  See the 
// License for the specific language governing rights and limitations under 
// the License.                                                             
//                                                                          
// The Original Code is eCos - Embedded Configurable Operating System,      
// released September 30, 1998.                                             
//                                                                          
// The Initial Developer of the Original Code is Red Hat.                   
// Portions created by Red Hat are                                          
// Copyright (C) 1998, 1999, 2000 Red Hat, Inc.                             
// All Rights Reserved.                                                     
// -------------------------------------------                              
//                                                                          
//####COPYRIGHTEND####
//####BSDCOPYRIGHTBEGIN####
//
// -------------------------------------------
//
// Portions of this software may have been derived from OpenBSD or other sources,
// and are covered by the appropriate copyright disclaimers included herein.
//
// -------------------------------------------
//
//####BSDCOPYRIGHTEND####
//==========================================================================
//#####DESCRIPTIONBEGIN####
//
// Author(s):    gthomas
// Contributors: gthomas
// Date:         2000-01-10
// Purpose:      
// Description:  
//              
//
//####DESCRIPTIONEND####
//
//==========================================================================


// Support routines, etc., used by network code

#include <sys/param.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/kernel.h>
#include <sys/domain.h>
#include <sys/protosw.h>
#include <net/netisr.h>

#include <machine/cpu.h>

#include <pkgconf/net.h>

#include <cyg/infra/diag.h>
#include <cyg/hal/hal_intr.h>
#include <cyg/kernel/kapi.h>

#include <cyg/infra/cyg_ass.h>

#include <netdev.h>

// Used for system-wide "ticks per second"
int hz = 100;
int tick = 10000;  // usec per "tick"

volatile struct timeval mono_time;

// Low-level network debugging
int net_debug = 0;

#define STACK_SIZE CYGNUM_HAL_STACK_SIZE_TYPICAL
static char netint_stack[STACK_SIZE];
static cyg_thread netint_thread_data;
static cyg_handle_t netint_thread_handle;

cyg_flag_t netint_flags;  
#define NETISR_ANY 0xFFFFFFFF  // Any possible bit...

extern void cyg_test_exit(void);  // TEMP
void
cyg_panic(const char *msg, ...)
{
    cyg_uint32 old_ints;
    HAL_DISABLE_INTERRUPTS(old_ints);
    diag_printf("PANIC: %s\n", msg);
    cyg_test_exit();  // FIXME
}

//---------------------------- splx() emulation ------------------------------
//
// This variable (and the associated bit patterns) is used to keep track
// of the "splx()" level.  This is an artifact of the original stack, based
// on the BSD interrupt world (interrupts and processing could be masked
// based on a level value, supported by hardware).  This is not very real-time,
// so the emulation uses proper eCos tools and techniques to accomplish the
// same result.  The key here is in the analysis of the various "levels", why
// they are used, etc.  
//
static cyg_uint32 spl_state;
#define SPL_STATE_IMP     0x01
#define SPL_STATE_NET     0x02
#define SPL_STATE_CLOCK   0x04
#define SPL_STATE_SOFTNET 0x08

static cyg_mutex_t softnet_mutex;
static volatile cyg_handle_t softnet_thread;

#define SPLINIT() CYG_MACRO_START               \
    cyg_mutex_init( &softnet_mutex );           \
    softnet_thread = 0;                         \
CYG_MACRO_END

//
// This function is called in order to protect internal data structures
// short-term, primarily so that interrupt processing does not interfere
// with them.
//
// Simply protecting against interrupts (DSRs) should suffice.
//
cyg_uint32
#ifdef CYGIMPL_TRACE_SPLX   
cyg_splimp(const char *file, const int line)
#else
cyg_splimp(void)
#endif
{
    cyg_uint32 old_ints;
    cyg_scheduler_lock();
#ifdef CYGIMPL_TRACE_SPLX   
    do_sched_event(__FUNCTION__, file, line, cyg_scheduler_read_lock());
#endif
    old_ints = spl_state;
    spl_state |= SPL_STATE_IMP;
    if (old_ints & SPL_STATE_IMP) {
        // Already at this state/level, no need to retake scheduler lock
        cyg_scheduler_unlock();
    }
    return old_ints;
}

//
// This function is called in order to ensure that a timestamp is valid
// i.e. no time passes while the stamp is being taken (since it is a 
// potentially non-idempotent data structure).
//
// Simply protecting against interrupts (DSRs) should suffice.
//
cyg_uint32
#ifdef CYGIMPL_TRACE_SPLX   
cyg_splclock(const char *file, const int line)
#else
cyg_splclock(void)
#endif
{
    cyg_uint32 old_ints;
    cyg_scheduler_lock();
#ifdef CYGIMPL_TRACE_SPLX   
    do_sched_event(__FUNCTION__, file, line, cyg_scheduler_read_lock());
#endif
    old_ints = spl_state;
    spl_state |= SPL_STATE_CLOCK;
    if (old_ints & SPL_STATE_CLOCK) {
        // Already at this state/level, no need to retake scheduler lock
        cyg_scheduler_unlock();
    }
    return old_ints;
}

cyg_uint32
#ifdef CYGIMPL_TRACE_SPLX   
cyg_splnet(const char *file, const int line)
#else
cyg_splnet(void)
#endif
{
    cyg_uint32 old_ints;
    cyg_scheduler_lock();
#ifdef CYGIMPL_TRACE_SPLX   
    do_sched_event(__FUNCTION__, file, line, cyg_scheduler_read_lock());
#endif
    old_ints = spl_state;
    spl_state |= SPL_STATE_NET;
    if (old_ints & SPL_STATE_NET) {
        // Already at this state/level, no need to retake scheduler lock
        cyg_scheduler_unlock();
    }
    return old_ints;
}

cyg_uint32
#ifdef CYGIMPL_TRACE_SPLX   
cyg_splhigh(const char *file, const int line)
#else
cyg_splhigh(void)
#endif
{
    // splhigh did SPLSOFTNET in the contrib, so this is the same
    return cyg_splsoftnet(
#ifdef CYGIMPL_TRACE_SPLX   
        file, line
#endif
        );
}

//
// Prevent all other stack processing, including interrupts (DSRs), etc.
//
cyg_uint32
#ifdef CYGIMPL_TRACE_SPLX   
cyg_splsoftnet(const char *file, const int line)
#else
cyg_splsoftnet(void)
#endif
{
    cyg_uint32 old_ints;
    cyg_scheduler_lock();
#ifdef CYGIMPL_TRACE_SPLX   
    do_sched_event(__FUNCTION__, file, line, cyg_scheduler_read_lock());
#endif
    if (spl_state & SPL_STATE_SOFTNET) {
        if (softnet_thread == cyg_thread_self()) {
            // Do nothing
            old_ints = spl_state;
            spl_state |= SPL_STATE_SOFTNET;
            cyg_scheduler_unlock();
            return old_ints;
        }
    }
    cyg_scheduler_unlock();
    cyg_mutex_lock(&softnet_mutex);
    cyg_scheduler_lock();
    softnet_thread = cyg_thread_self();
    old_ints = spl_state;
    spl_state |= SPL_STATE_SOFTNET;
    return old_ints;
}

//
// Return to a previous interrupt state/level.
//
void
#ifdef CYGIMPL_TRACE_SPLX   
cyg_splx(cyg_uint32 old_state, const char *file, const int line)
#else
cyg_splx(cyg_uint32 old_state)
#endif
{
    cyg_uint32 new_state = spl_state;
    cyg_scheduler_lock();  // Extra security while messing about
#ifdef CYGIMPL_TRACE_SPLX   
    do_sched_event(__FUNCTION__, file, line, cyg_scheduler_read_lock());
#endif
    if ((spl_state & SPL_STATE_SOFTNET) && !(old_state & SPL_STATE_SOFTNET)) {
        new_state &= ~SPL_STATE_SOFTNET;
        softnet_thread = 0;
        cyg_mutex_unlock(&softnet_mutex);
        cyg_scheduler_unlock();
    }
    if ((spl_state & SPL_STATE_NET) && !(old_state & SPL_STATE_NET)) {
        new_state &= ~SPL_STATE_NET;
        cyg_scheduler_unlock();
    }
    if ((spl_state & SPL_STATE_CLOCK) && !(old_state & SPL_STATE_CLOCK)) {
        new_state &= ~SPL_STATE_CLOCK;
        cyg_scheduler_unlock();
    }
    if ((spl_state & SPL_STATE_IMP) && !(old_state & SPL_STATE_IMP)) {
        new_state &= ~SPL_STATE_IMP;
        cyg_scheduler_unlock();
    }
    spl_state = new_state;
    cyg_scheduler_unlock();
}
//---------------------------- splx() emulation ------------------------------

void
setsoftnet(void)
{
    diag_printf("setsoftnet\n");
    schednetisr(NETISR_SOFTNET);
}

// Round a number 'n' up to a multiple of 'm'
#define round(n,m) ((((n)+((m)-1))/(m))*(m))

#define NET_MEMPOOL_SIZE  round(CYGPKG_NET_MEM_USAGE/4,MSIZE)
#define NET_MBUFS_SIZE    round(CYGPKG_NET_MEM_USAGE/4,MSIZE)
#define NET_CLUSTERS_SIZE round(CYGPKG_NET_MEM_USAGE/2,MCLBYTES)

static unsigned char net_mempool_area[NET_MEMPOOL_SIZE];
static cyg_mempool_var net_mem_pool;
static cyg_handle_t    net_mem;
static unsigned char net_mbufs_area[NET_MBUFS_SIZE];
static cyg_mempool_fix net_mbufs_pool;
static cyg_handle_t    net_mbufs;
static unsigned char net_clusters_area[NET_CLUSTERS_SIZE];
static cyg_mempool_fix net_clusters_pool;
static cyg_handle_t    net_clusters;
static char            net_clusters_refcnt[(NET_CLUSTERS_SIZE/MCLBYTES)+1];

static struct net_stats  stats_malloc, stats_free, 
    stats_memcpy, stats_memset,
    stats_mbuf_alloc, stats_mbuf_free, stats_cluster_alloc;
extern struct net_stats stats_in_cksum;

// Display a number of ticks as microseconds
// Note: for improved calculation significance, values are kept in ticks*1000
static long rtc_resolution[] = CYGNUM_KERNEL_COUNTERS_RTC_RESOLUTION;
static long ns_per_system_clock;

static void
show_ticks_in_us(cyg_uint32 ticks)
{
    long long ns;
    ns_per_system_clock = 1000000/rtc_resolution[1];
    ns = (ns_per_system_clock * ((long long)ticks * 1000)) / 
        CYGNUM_KERNEL_COUNTERS_RTC_PERIOD;
    ns += 5;  // for rounding to .01us
    diag_printf("%7d.%02d", (int)(ns/1000), (int)((ns%1000)/10));
}

void
show_net_stats(struct net_stats *stats, const char *title)
{
    int ave;
    ave = stats->total_time / stats->count;
    diag_printf("%s:\n", title);
    diag_printf("  count: %6d", stats->count);
    diag_printf(", min: ");
    show_ticks_in_us(stats->min_time);
    diag_printf(", max: ");
    show_ticks_in_us(stats->max_time);
    diag_printf(", total: ");
    show_ticks_in_us(stats->total_time);
    diag_printf(", ave: ");
    show_ticks_in_us(ave);
    diag_printf("\n");
    // Reset stats
    memset(stats, 0, sizeof(*stats));
}

void
show_net_times(void)
{
    show_net_stats(&stats_malloc,        "Net malloc");
    show_net_stats(&stats_free,          "Net free");
    show_net_stats(&stats_mbuf_alloc,    "Mbuf alloc");
    show_net_stats(&stats_mbuf_free,     "Mbuf free");
    show_net_stats(&stats_cluster_alloc, "Cluster alloc");
    show_net_stats(&stats_in_cksum,      "Checksum");
    show_net_stats(&stats_memcpy,        "Net memcpy");
    show_net_stats(&stats_memset,        "Net memset");
}

void *
cyg_net_malloc(u_long size, int type, int flags)
{
    void *res;
    START_STATS();
    if (flags & M_NOWAIT) {
        res = cyg_mempool_var_try_alloc(net_mem, size);
    } else {
        res = cyg_mempool_var_alloc(net_mem, size);
    }
    FINISH_STATS(stats_malloc);
    return (res);
}

void 
cyg_net_free(caddr_t addr, int type)
{
    START_STATS();
    cyg_mempool_var_free(net_mem, addr);
    FINISH_STATS(stats_free);
}

void *
cyg_net_mbuf_alloc(int type, int flags)
{
    void *res;    
    START_STATS();
    mbstat.m_mbufs++;
    if (flags & M_NOWAIT) {
        res = cyg_mempool_fix_try_alloc(net_mbufs);
    } else {
        res = cyg_mempool_fix_alloc(net_mbufs);
    }
    FINISH_STATS(stats_mbuf_alloc);
    // Check that this nastiness works OK
    CYG_ASSERT( dtom(res) == res, "dtom failed, base of mbuf" );
    CYG_ASSERT( dtom((char *)res + MSIZE/2) == res, "dtom failed, mid mbuf" );
    return (res);
}

void 
cyg_net_mbuf_free(caddr_t addr, int type)
{
    START_STATS();
    mbstat.m_mbufs--;
    cyg_mempool_fix_free(net_mbufs, addr);
    FINISH_STATS(stats_mbuf_free);
}

void *
cyg_net_cluster_alloc(void)
{
    void *res;
    START_STATS();
    res = cyg_mempool_fix_try_alloc(net_clusters);
    FINISH_STATS(stats_cluster_alloc);
    return res;
}

static void
cyg_kmem_init(void)
{
    unsigned char *p;
    diag_printf("Network stack using %d bytes for misc space\n", NET_MEMPOOL_SIZE);
    diag_printf("                    %d bytes for mbufs\n", NET_MBUFS_SIZE);
    diag_printf("                    %d bytes for mbuf clusters\n", NET_CLUSTERS_SIZE);
    cyg_mempool_var_create(&net_mempool_area,
                           NET_MEMPOOL_SIZE,
                           &net_mem,
                           &net_mem_pool);
    // Align the mbufs on MSIZE boudaries so that dtom() can work.
    p = (unsigned char *)(((long)(&net_mbufs_area) + MSIZE - 1) & ~(MSIZE-1));
    cyg_mempool_fix_create(p,
                           ((&(net_mbufs_area[NET_MBUFS_SIZE])) - p) & ~(MSIZE-1),
                           MSIZE,
                           &net_mbufs,
                           &net_mbufs_pool);
    cyg_mempool_fix_create(&net_clusters_area,
                           NET_CLUSTERS_SIZE,
                           MCLBYTES,
                           &net_clusters,
                           &net_clusters_pool);
    mbutl = (struct mbuf *)&net_clusters_area;
    mclrefcnt = net_clusters_refcnt;
}

void cyg_kmem_print_stats( void )
{
    cyg_mempool_info info;

    diag_printf( "Network stack mbuf stats:\n" );
    diag_printf( "   mbufs %d, clusters %d, free clusters %d\n",
                 mbstat.m_mbufs,	/* mbufs obtained from page pool */
                 mbstat.m_clusters,	/* clusters obtained from page pool */
                 /* mbstat.m_spare, */	/* spare field */
                 mbstat.m_clfree	/* free clusters */
        );
    diag_printf( "   Failed to get %d times\n"
                 "   Waited to get %d times\n"
                 "   Drained queues to get %d times\n",
                 mbstat.m_drops,	/* times failed to find space */
                 mbstat.m_wait, 	/* times waited for space */
                 mbstat.m_drain         /* times drained protocols for space */
                 /* mbstat.m_mtypes[256]; type specific mbuf allocations */
        );

    cyg_mempool_var_get_info( net_mem, &info );
    diag_printf( "Misc mpool: total %7d, free %7d, max free block %d\n",
                 info.totalmem,
                 info.freemem,
                 info.maxfree
        );

    cyg_mempool_fix_get_info( net_mbufs, &info );
    diag_printf( "Mbufs pool: total %7d, free %7d, blocksize %4d\n",
                 info.totalmem,
                 info.freemem,
                 info.blocksize
        );


    cyg_mempool_fix_get_info( net_clusters, &info );
    diag_printf( "Clust pool: total %7d, free %7d, blocksize %4d\n",
                 info.totalmem,
                 info.freemem,
                 info.blocksize
        );
}

int
cyg_mtocl(u_long x)
{
    int res;
    res = (((u_long)(x) - (u_long)mbutl) >> MCLSHIFT);
    return res;
}

struct mbuf *
cyg_cltom(u_long x)
{
    struct mbuf *res;
    res = (struct mbuf *)((caddr_t)((u_long)mbutl + ((u_long)(x) << MCLSHIFT)));
    return res;
}

void 
net_memcpy(void *d, void *s, int n)
{
    START_STATS();
    memcpy(d, s, n);
    FINISH_STATS(stats_memcpy);
}

void 
net_memset(void *s, int v, int n)
{
    START_STATS();
    memset(s, v, n);
    FINISH_STATS(stats_memset);
}

// Rather than bring in the whole BSD 'random' code...
int
arc4random(void)
{
    cyg_uint32 res;
    static unsigned long seed = 0xDEADB00B;
    HAL_CLOCK_READ(&res);  // Not so bad...
    seed = ((seed & 0x07F00FF) << 7) ^
        (seed & 0xF80FF00) ^ 
        (res << 13);
    return (int)seed;
}

void 
get_random_bytes(void *buf, size_t len)
{
    unsigned long ranbuf, *lp;
    lp = (unsigned long *)buf;
    while (len > 0) {
        ranbuf = arc4random();
        *lp++ = ranbuf;
        len -= sizeof(ranbuf);
    }
}

void 
microtime(struct timeval *tp)
{
    panic("microtime");
}

void
get_mono_time(void)
{
    panic("get_mono_time");
}

void 
csignal(pid_t pgid, int signum, uid_t uid, uid_t euid)
{
    panic("csignal");
}

int
bcmp(const void *_p1, const void *_p2, size_t len)
{
    int res = 0;
    unsigned char *p1 = (unsigned char *)_p1;
    unsigned char *p2 = (unsigned char *)_p2;
    while (len-- > 0) {
        res = *p1++ - *p2++;
        if (res) break;
    }
    return res;
}

int
copyout(const void *s, void *d, size_t len)
{
    memcpy(d, s, len);
    return 0;
}

int
copyin(const void *s, void *d, size_t len)
{
    memcpy(d, s, len);
    return 0;
}

void
ovbcopy(const void *s, void *d, size_t len)
{
    memcpy(d, s, len);
}

//------------------ tsleep() and wakeup() emulation ---------------------------
//
// Structure used to keep track of 'tsleep' style events
//
struct wakeup_event {
    void *chan;
    cyg_sem_t sem;
};
static struct wakeup_event wakeup_list[CYGPKG_NET_NUM_WAKEUP_EVENTS];

//
// Signal an event
void
cyg_wakeup(void *chan)
{
    int i;
    struct wakeup_event *ev;
    cyg_scheduler_lock();  // Ensure scan is safe
    for (i = 0, ev = wakeup_list;  i < CYGPKG_NET_NUM_WAKEUP_EVENTS;  i++, ev++) {
        if (ev->chan == chan) {
            cyg_semaphore_post(&ev->sem);
            ev->chan = 0;
        }
    }
    cyg_scheduler_unlock();
}

//
// Wait for an event with timeout
//   tsleep(event, priority, state, timeout)
//     event - the thing to wait for
//     priority - unused
//     state    - a descriptive message
//     timeout  - max time (in ticks) to wait
//   returns:
//     0         - event was "signalled"
//     ETIMEDOUT - timeout occurred
//
int       
cyg_tsleep(void *chan, int pri, char *wmesg, int timo)
{
    int i, res = 0;
    struct wakeup_event *ev;    
    cyg_tick_count_t sleep_time;
    int olock; // current state of scheduler lock - so it can be replaced
    cyg_handle_t self = cyg_thread_self();

    cyg_scheduler_lock(); // ...around.
    olock = cyg_scheduler_read_lock();
    if ( olock > 1 )
        cyg_scheduler_unlock();

    for (i = 0, ev = wakeup_list;  i < CYGPKG_NET_NUM_WAKEUP_EVENTS;  i++, ev++) {
        if (ev->chan == 0) {
            ev->chan = chan;
            break;
        }
    }
    if (i == CYGPKG_NET_NUM_WAKEUP_EVENTS) {
        panic("no sleep slots");
    }
    if ( 1 != cyg_scheduler_read_lock()) {
        panic("Tsleep - called with scheduler locked\n");
    }

    // Then we must release the 'softnet' mutex when we wait - if we have it
    if ( self == softnet_thread ) {
        softnet_thread = 0;
        spl_state &= ~SPL_STATE_SOFTNET;
        cyg_mutex_unlock( &softnet_mutex );
    } else {
        self = 0; // Flag no need to reclaim
    }

    // This part actually does the wait:
    cyg_scheduler_unlock();
    if (timo) {
        sleep_time = cyg_current_time() + timo;
        if (!cyg_semaphore_timed_wait(&ev->sem, sleep_time)) {
            res = ETIMEDOUT;
            ev->chan = 0;  // Free slot
        }
    } else {
        cyg_semaphore_wait(&ev->sem);
    }

    if ( self ) { // return to previous state
        cyg_mutex_lock( &softnet_mutex ); // this might wait
        cyg_scheduler_lock();
        softnet_thread = self; // got it now...
        spl_state |= SPL_STATE_SOFTNET;
        cyg_scheduler_unlock();
    }
    if ( olock > 1 )
        cyg_scheduler_lock();

    return res;
}

// Called to initialize structures used by timeout functions
static void
cyg_timeout_init(void)
{
    int i;
    struct wakeup_event *ev;
    // Create list of "wakeup event" semaphores
    for (i = 0, ev = wakeup_list;  i < CYGPKG_NET_NUM_WAKEUP_EVENTS;  i++, ev++) {
        ev->chan = 0;
        cyg_semaphore_init(&ev->sem, 0);
    }
}
//------------------ tsleep() and wakeup() emulation ---------------------------

//
// Network software interrupt handler
//   This function is run as a separate thread to allow
// processing of network events (mostly incoming packets)
// at "user level" instead of at interrupt time.
//
static void
cyg_netint(cyg_addrword_t param)
{
    cyg_flag_value_t curisr;
    int s;
    while (true) {
        curisr = cyg_flag_wait(&netint_flags, NETISR_ANY, 
                               CYG_FLAG_WAITMODE_OR|CYG_FLAG_WAITMODE_CLR);
        s = splsoftnet();      // Prevent any overlapping "stack" processing
#ifdef INET
        if (curisr & (1 << NETISR_ARP)) {
            // Pending ARP requests
            arpintr();
        }
        if (curisr & (1 << NETISR_IP)) {
            // Pending IPv4 input
            ipintr();
        }
#endif
#ifdef INET6
        if (curisr & (1 << NETISR_IPV6)) {
            // Pending IPv6 input
            ip6intr();
        }
#endif
#if NBRIDGE > 0
        if (curisr & (1 << NETISR_BRIDGE)) {
            // Pending bridge input
            bridgeintr();
        }
#endif
        splx(s);
    }
}

//
// Network initialization
//   This function is called during system initialization to setup the whole
// networking environment.
//
extern void cyg_do_net_init(void);  // Linker magic to execute this function as 'init'
extern void ifinit(void);
extern void loopattach(int);
extern void bridgeattach(int);

void
cyg_net_init(void)
{
    static int _init = false;
    cyg_netdevtab_entry_t *t;

#ifdef CYGIMPL_TRACE_SPLX   
    show_sched_events();
#endif
    if (_init) return;
    cyg_do_net_init();  // Just forces the linking in of the initializer/constructor
    // Initialize interrupt "flags"
    cyg_flag_init(&netint_flags);
    // Anthing else needed?
#ifdef SPLINIT
    SPLINIT();
#endif
    // Create network background thread
    cyg_thread_create(CYGPKG_NET_THREAD_PRIORITY, // Priority
                      cyg_netint,               // entry
                      0,                        // entry parameter
                      "Network support",        // Name
                      &netint_stack[0],         // Stack
                      STACK_SIZE,               // Size
                      &netint_thread_handle,    // Handle
                      &netint_thread_data       // Thread data structure
            );
    cyg_thread_resume(netint_thread_handle);    // Start it
    // Initialize timeout support
    cyg_timeout_init();
    // Initialize network memory system
    cyg_kmem_init();
    mbinit();
    // Initialize all network devices
    for (t = &__NETDEVTAB__[0]; t != &__NETDEVTAB_END__; t++) {
        diag_printf("Init device '%s'\n", t->name);
        if (t->init(t)) {
            t->status = CYG_NETDEVTAB_STATUS_AVAIL;
        } else {
            // What to do if device init fails?
            t->status = 0;  // Device not [currently] available
        }
    }
    // And attack the loopback interface
#ifdef CYGPKG_NET_NLOOP
#if 0 < CYGPKG_NET_NLOOP
    loopattach(0);
#endif
#endif
#if NBRIDGE > 0
    bridgeattach(0);
#endif
    // Start up the network processing
    ifinit();
    domaininit();
    // Done
    _init = true;
}

#ifdef CYGIMPL_TRACE_SPLX   
#undef cyg_scheduler_lock
#undef cyg_scheduler_safe_lock
#undef cyg_scheduler_unlock

#define MAX_SCHED_EVENTS 256
static struct _sched_event {
    char *fun, *file;
    int line, lock;
} sched_event[MAX_SCHED_EVENTS];
static int next_sched_event = 0;
static int total_sched_events = 0;

static void
do_sched_event(char *fun, char *file, int line, int lock)
{
    struct _sched_event *se = &sched_event[next_sched_event];
    if (++next_sched_event == MAX_SCHED_EVENTS) {
        next_sched_event = 0;
    }
    se->fun = fun;
    se->file = file;
    se->line = line;
    se->lock = lock;
    total_sched_events++;
}

static void
show_sched_events(void)
{
    int i;
    struct _sched_event *se;
    if (total_sched_events < MAX_SCHED_EVENTS) {
        i = 0;
    } else {
        i = next_sched_event + 1;
        if (i == MAX_SCHED_EVENTS) i = 0;
    }
    diag_printf("%d total scheduler events\n", total_sched_events);
    while (i != next_sched_event) {
        se = &sched_event[i];
        diag_printf("%s - lock: %d, called from %s.%d\n", se->fun, se->lock, se->file, se->line);
        if (++i == MAX_SCHED_EVENTS) i = 0;
    }
}

void
_cyg_scheduler_lock(char *file, int line)
{
    cyg_scheduler_lock();
    do_sched_event(__FUNCTION__, file, line, cyg_scheduler_read_lock());
}

void
_cyg_scheduler_safe_lock(char *file, int line)
{
    cyg_scheduler_safe_lock();
    do_sched_event(__FUNCTION__, file, line, cyg_scheduler_read_lock());
}

void
_cyg_scheduler_unlock(char *file, int line)
{
    cyg_scheduler_unlock();
    do_sched_event(__FUNCTION__, file, line, cyg_scheduler_read_lock());
}
#endif // CYGIMPL_TRACE_SPLX