source: trunk/kernel/kern/process.c @ 657

/*
 * process.c - process related functions definition.
 *
 * Authors  Ghassan Almaless (2008,2009,2010,2011,2012)
 *          Mohamed Lamine Karaoui (2015)
 *          Alain Greiner (2016,2017,2018,2019,2020)
 *
 * Copyright (c) UPMC Sorbonne Universites
 *
 * This file is part of ALMOS-MKH.
 *
 * ALMOS-MKH is free software; you can redistribute it and/or modify it
 * under the terms of the GNU General Public License as published by
 * the Free Software Foundation; version 2.0 of the License.
 *
 * ALMOS-MKH is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with ALMOS-MKH; if not, write to the Free Software Foundation,
 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
 */

#include <kernel_config.h>
#include <hal_kernel_types.h>
#include <hal_remote.h>
#include <hal_uspace.h>
#include <hal_irqmask.h>
#include <hal_vmm.h>
#include <errno.h>
#include <printk.h>
#include <memcpy.h>
#include <bits.h>
#include <kmem.h>
#include <page.h>
#include <vmm.h>
#include <vfs.h>
#include <core.h>
#include <thread.h>
#include <chdev.h>
#include <list.h>
#include <string.h>
#include <scheduler.h>
#include <busylock.h>
#include <queuelock.h>
#include <remote_queuelock.h>
#include <rwlock.h>
#include <remote_rwlock.h>
#include <dqdt.h>
#include <cluster.h>
#include <ppm.h>
#include <boot_info.h>
#include <process.h>
#include <elf.h>
#include <syscalls.h>
#include <shared_syscalls.h>

//////////////////////////////////////////////////////////////////////////////////////////
// Extern global variables
//////////////////////////////////////////////////////////////////////////////////////////

extern process_t           process_zero;     // allocated in kernel_init.c
extern chdev_directory_t   chdev_dir;        // allocated in kernel_init.c

//////////////////////////////////////////////////////////////////////////////////////////
// Process initialisation related functions
//////////////////////////////////////////////////////////////////////////////////////////

/////////////////////////////////
process_t * process_alloc( void )
{
        kmem_req_t req;

    req.type  = KMEM_KCM;
        req.order = bits_log2( sizeof(process_t) );
        req.flags = AF_KERNEL;

    return kmem_alloc( &req );
}

////////////////////////////////////////
void process_free( process_t * process )
{
    kmem_req_t  req;

        req.type = KMEM_KCM;
        req.ptr  = process;
        kmem_free( &req );
}

////////////////////////////////////////////////////
error_t process_reference_init( process_t * process,
                                pid_t       pid,
                                xptr_t      parent_xp )
{
    error_t     error;
    xptr_t      process_xp;
    cxy_t       parent_cxy;
    process_t * parent_ptr;
    xptr_t      stdin_xp;
    xptr_t      stdout_xp;
    xptr_t      stderr_xp;
    uint32_t    stdin_id;
    uint32_t    stdout_id;
    uint32_t    stderr_id;
    uint32_t    txt_id;
    char        rx_path[40];
    char        tx_path[40];
    xptr_t      file_xp;
    xptr_t      chdev_xp;
    chdev_t   * chdev_ptr;
    cxy_t       chdev_cxy;
    pid_t       parent_pid;
    vmm_t     * vmm;

    // build extended pointer on this reference process
    process_xp = XPTR( local_cxy , process );

    // get pointer on process vmm
    vmm = &process->vmm;

    // get parent process cluster and local pointer
    parent_cxy = GET_CXY( parent_xp );
    parent_ptr = GET_PTR( parent_xp );

    // get parent_pid
    parent_pid = hal_remote_l32( XPTR( parent_cxy , &parent_ptr->pid ) );

#if DEBUG_PROCESS_REFERENCE_INIT
thread_t * this = CURRENT_THREAD;
uint32_t cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_REFERENCE_INIT < cycle )
printk("\n[%s] thread[%x,%x] enter to initialize process %x / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, pid, cycle );
#endif

    // initialize pid, ref_xp, parent_xp, owner_xp, term_state fields
        process->pid        = pid;
    process->ref_xp     = XPTR( local_cxy , process );
    process->owner_xp   = XPTR( local_cxy , process );
    process->parent_xp  = parent_xp;
    process->term_state = 0;

    // initialize VFS root inode and CWD inode
    process->vfs_root_xp = hal_remote_l64( XPTR( parent_cxy, &parent_ptr->vfs_root_xp ) );
    process->cwd_xp      = hal_remote_l64( XPTR( parent_cxy, &parent_ptr->cwd_xp ) );

    // initialize VSL as empty
    vmm->vsegs_nr = 0;
        xlist_root_init( XPTR( local_cxy , &vmm->vsegs_root ) );

    // create an empty GPT as required by the architecture
    error = hal_gpt_create( &vmm->gpt );
    if( error ) 
    {
        printk("\n[ERROR] in %s : cannot create empty GPT\n", __FUNCTION__ );
        return -1;
    }

#if (DEBUG_PROCESS_REFERENCE_INIT & 1)
if( DEBUG_PROCESS_REFERENCE_INIT < cycle )
printk("\n[%s] thread[%x,%x] created empty GPT for process %x\n",
__FUNCTION__, parent_pid, this->trdid, pid );
#endif

    // initialize VSL lock
        remote_rwlock_init( XPTR( local_cxy , &vmm->vsl_lock ) , LOCK_VMM_VSL );

    // register kernel vsegs in user process VMM as required by the architecture
    error = hal_vmm_kernel_update( process );
    if( error ) 
    {
        printk("\n[ERROR] in %s : cannot register kernel vsegs in VMM\n", __FUNCTION__ );
        return -1;
    }

#if (DEBUG_PROCESS_REFERENCE_INIT & 1)
if( DEBUG_PROCESS_REFERENCE_INIT < cycle )
printk("\n[%s] thread[%x,%x] registered kernel vsegs in VSL for process %x\n",
__FUNCTION__, parent_pid, this->trdid, pid );
#endif

    // create "args" and "envs" vsegs
    // create "stacks" and "mmap" vsegs allocators 
    // initialize locks protecting GPT and VSL
    error = vmm_user_init( process );
    if( error ) 
    {
        printk("\n[ERROR] in %s : cannot register user vsegs in VMM\n", __FUNCTION__ );
        return -1;
    }
 
#if (DEBUG_PROCESS_REFERENCE_INIT & 1)
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_REFERENCE_INIT < cycle )
printk("\n[%s] thread[%x,%x] initialized vmm for process %x\n", 
__FUNCTION__, parent_pid, this->trdid, pid );
#endif

    // initialize fd_array as empty
    process_fd_init( process );

    // define the stdin/stdout/stderr pseudo files <=> select a TXT terminal.
    if( (pid == 1) || (parent_pid  == 1) )      // INIT or KSH  process
    {
        // select a TXT channel
        if( pid == 1 )  txt_id = 0;                     // INIT
        else            txt_id = process_txt_alloc();   // KSH

        // attach process to TXT
        process_txt_attach( process , txt_id ); 

#if (DEBUG_PROCESS_REFERENCE_INIT & 1)
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_REFERENCE_INIT < cycle )
printk("\n[%s] thread[%x,%x] / process %x attached to TXT%d / cycle %d\n", 
__FUNCTION__, parent_pid, this->trdid, pid, txt_id, cycle );
#endif
        // build path to TXT_RX[i] and TXT_TX[i] chdevs
        snprintf( rx_path , 40 , "/dev/external/txt%d_rx", txt_id );
        snprintf( tx_path , 40 , "/dev/external/txt%d_tx", txt_id );

        // create stdin pseudo file          
        error = vfs_open(  process->vfs_root_xp,
                           rx_path,
                           process_xp,
                           O_RDONLY, 
                           0,                // FIXME chmod
                           &stdin_xp, 
                           &stdin_id );
        if( error )
        {
            printk("\n[ERROR] in %s : cannot open stdout pseudo-file\n", __FUNCTION__ );
            return -1;
        }

assert( (stdin_id == 0) , "stdin index must be 0" );

#if (DEBUG_PROCESS_REFERENCE_INIT & 1)
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_REFERENCE_INIT < cycle )
printk("\n[%s] thread[%x,%x] / stdin open for process %x / cycle %d\n", 
__FUNCTION__, parent_pid, this->trdid, pid, cycle );
#endif

        // create stdout pseudo file          
        error = vfs_open(  process->vfs_root_xp,
                           tx_path,
                           process_xp,
                           O_WRONLY, 
                           0,                // FIXME chmod
                           &stdout_xp, 
                           &stdout_id );
        if( error )
        {
            printk("\n[ERROR] in %s : cannot open stdout pseudo-file\n", __FUNCTION__ );
            return -1;
        }

assert( (stdout_id == 1) , "stdout index must be 1" );

#if (DEBUG_PROCESS_REFERENCE_INIT & 1)
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_REFERENCE_INIT < cycle )
printk("\n[%s] thread[%x,%x] / stdout open for process %x / cycle %d\n", 
__FUNCTION__, parent_pid, this->trdid, pid, cycle );
#endif

        // create stderr pseudo file          
        error = vfs_open(  process->vfs_root_xp,
                           tx_path,
                           process_xp,
                           O_WRONLY, 
                           0,                // FIXME chmod
                           &stderr_xp, 
                           &stderr_id );
        if( error )
        {
            printk("\n[ERROR] in %s : cannot open stderr pseudo-file\n", __FUNCTION__ );
            return -1;
        }

assert( (stderr_id == 2) , "stderr index must be 2" );

#if (DEBUG_PROCESS_REFERENCE_INIT & 1)
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_REFERENCE_INIT < cycle )
printk("\n[%s] thread[%x,%x] / stderr open for process %x / cycle %d\n", 
__FUNCTION__, parent_pid, this->trdid, pid, cycle );
#endif

    }
    else                                            // normal user process 
    {
        // get extended pointer on stdin pseudo file in parent process
        file_xp = (xptr_t)hal_remote_l64( XPTR( parent_cxy,
                                                &parent_ptr->fd_array.array[0] ) );

        // get extended pointer on parent process TXT chdev
        chdev_xp = chdev_from_file( file_xp );
 
        // get cluster and local pointer on chdev
        chdev_cxy = GET_CXY( chdev_xp );
        chdev_ptr = GET_PTR( chdev_xp );
 
        // get parent process TXT terminal index
        txt_id = hal_remote_l32( XPTR( chdev_cxy , &chdev_ptr->channel ) );

        // attach child process to parent process TXT terminal
        process_txt_attach( process , txt_id ); 

        // copy all open files from parent process fd_array to this process
        process_fd_remote_copy( XPTR( local_cxy , &process->fd_array ),
                                XPTR( parent_cxy , &parent_ptr->fd_array ) );
    }

    // initialize lock protecting CWD changes
    remote_busylock_init( XPTR( local_cxy , 
                                &process->cwd_lock ), LOCK_PROCESS_CWD );

#if (DEBUG_PROCESS_REFERENCE_INIT & 1)
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_REFERENCE_INIT < cycle )
printk("\n[%s] thread[%x,%x] / set fd_array for process %x / cycle %d\n", 
__FUNCTION__, parent_pid, this->trdid, pid , cycle );
#endif

    // reset children list root 
    xlist_root_init( XPTR( local_cxy , &process->children_root ) );
    process->children_nr     = 0;
    remote_queuelock_init( XPTR( local_cxy,
                                 &process->children_lock ), LOCK_PROCESS_CHILDREN );

    // reset semaphore / mutex / barrier / condvar list roots and lock
    xlist_root_init( XPTR( local_cxy , &process->sem_root ) );
    xlist_root_init( XPTR( local_cxy , &process->mutex_root ) );
    xlist_root_init( XPTR( local_cxy , &process->barrier_root ) );
    xlist_root_init( XPTR( local_cxy , &process->condvar_root ) );
    remote_queuelock_init( XPTR( local_cxy , 
                                 &process->sync_lock ), LOCK_PROCESS_USERSYNC );

    // reset open directories root and lock 
    xlist_root_init( XPTR( local_cxy , &process->dir_root ) );
    remote_queuelock_init( XPTR( local_cxy , 
                                 &process->dir_lock ), LOCK_PROCESS_DIR );

    // register new process in the local cluster manager pref_tbl[]
    lpid_t lpid = LPID_FROM_PID( pid );
    LOCAL_CLUSTER->pmgr.pref_tbl[lpid] = XPTR( local_cxy , process );

    // register new process descriptor in local cluster manager local_list
    cluster_process_local_link( process );

    // register new process descriptor in local cluster manager copies_list 
    cluster_process_copies_link( process );

    // initialize th_tbl[] array and associated threads
    uint32_t i;

    for( i = 0 ; i < CONFIG_THREADS_MAX_PER_CLUSTER ; i++ )
        {
        process->th_tbl[i] = NULL;
    }
    process->th_nr  = 0;
    rwlock_init( &process->th_lock , LOCK_PROCESS_THTBL );

        hal_fence();

#if (DEBUG_PROCESS_REFERENCE_INIT & 1)
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_REFERENCE_INIT < cycle )
printk("\n[%s] thread[%x,%x] exit for process %x / cycle %d\n", 
__FUNCTION__, parent_pid, this->trdid, pid, cycle );
#endif

#if (DEBUG_PROCESS_REFERENCE_INIT & 1)
hal_vmm_display( parent_xp , false );
hal_vmm_display( XPTR( local_cxy , process ) , false );
#endif

    return 0;

}  // process_reference_init()

/////////////////////////////////////////////////////
error_t process_copy_init( process_t * local_process,
                           xptr_t      reference_process_xp )
{
    error_t   error;
    vmm_t   * vmm;

    // get reference process cluster and local pointer
    cxy_t       ref_cxy = GET_CXY( reference_process_xp );
    process_t * ref_ptr = GET_PTR( reference_process_xp );

    // get pointer on process vmm
    vmm = &local_process->vmm;

    // initialize PID, REF_XP, PARENT_XP, and STATE
    local_process->pid        = hal_remote_l32(  XPTR( ref_cxy , &ref_ptr->pid ) );
    local_process->parent_xp  = hal_remote_l64( XPTR( ref_cxy , &ref_ptr->parent_xp ) );
    local_process->ref_xp     = reference_process_xp;
    local_process->owner_xp   = reference_process_xp;
    local_process->term_state = 0;

#if DEBUG_PROCESS_COPY_INIT
thread_t * this = CURRENT_THREAD;  
uint32_t cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_COPY_INIT < cycle )
printk("\n[%s] thread[%x,%x] enter for process %x / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, local_process->pid, cycle );
#endif

// check user process
assert( (local_process->pid != 0), "LPID cannot be 0" );

    // initialize VSL as empty
    vmm->vsegs_nr = 0;
        xlist_root_init( XPTR( local_cxy , &vmm->vsegs_root ) );

    // create an empty GPT as required by the architecture
    error = hal_gpt_create( &vmm->gpt );
    if( error ) 
    {
        printk("\n[ERROR] in %s : cannot create empty GPT\n", __FUNCTION__ );
        return -1;
    }

    // initialize GPT and VSL locks
        remote_rwlock_init( XPTR( local_cxy , &vmm->vsl_lock ) , LOCK_VMM_VSL );

    // register kernel vsegs in VMM as required by the architecture
    error = hal_vmm_kernel_update( local_process );
    if( error ) 
    {
        printk("\n[ERROR] in %s : cannot register kernel vsegs in VMM\n", __FUNCTION__ );
        return -1;
    }

    // create "args" and "envs" vsegs
    // create "stacks" and "mmap" vsegs allocators 
    // initialize locks protecting GPT and VSL
    error = vmm_user_init( local_process );
    if( error ) 
    {
        printk("\n[ERROR] in %s : cannot register user vsegs in VMM\n", __FUNCTION__ );
        return -1;
    }
 
#if (DEBUG_PROCESS_COPY_INIT & 1)
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_COPY_INIT < cycle )
printk("\n[%s] thread[%x,%x] initialized vmm for process %x / cycle %d\n", 
__FUNCTION__, parent_pid, this->trdid, pid, cycle );
#endif

    // set process file descriptors array
        process_fd_init( local_process );

    // set vfs_root_xp / vfs_bin_xp / cwd_xp fields
    local_process->vfs_root_xp = hal_remote_l64( XPTR( ref_cxy , &ref_ptr->vfs_root_xp ) );
    local_process->vfs_bin_xp  = hal_remote_l64( XPTR( ref_cxy , &ref_ptr->vfs_bin_xp ) );
    local_process->cwd_xp      = XPTR_NULL;

    // reset children list root (not used in a process descriptor copy)
    xlist_root_init( XPTR( local_cxy , &local_process->children_root ) );
    local_process->children_nr   = 0;
    remote_queuelock_init( XPTR( local_cxy , &local_process->children_lock ),
                           LOCK_PROCESS_CHILDREN );

    // reset children_list (not used in a process descriptor copy)
    xlist_entry_init( XPTR( local_cxy , &local_process->children_list ) );

    // reset semaphores list root (not used in a process descriptor copy)
    xlist_root_init( XPTR( local_cxy , &local_process->sem_root ) );
    xlist_root_init( XPTR( local_cxy , &local_process->mutex_root ) );
    xlist_root_init( XPTR( local_cxy , &local_process->barrier_root ) );
    xlist_root_init( XPTR( local_cxy , &local_process->condvar_root ) );

    // initialize th_tbl[] array and associated fields
    uint32_t i;
    for( i = 0 ; i < CONFIG_THREADS_MAX_PER_CLUSTER ; i++ )
        {
        local_process->th_tbl[i] = NULL;
    }
    local_process->th_nr  = 0;
    rwlock_init( &local_process->th_lock , LOCK_PROCESS_THTBL );

    // register new process descriptor in local cluster manager local_list
    cluster_process_local_link( local_process );

    // register new process descriptor in owner cluster manager copies_list
    cluster_process_copies_link( local_process );

        hal_fence();

#if DEBUG_PROCESS_COPY_INIT 
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_COPY_INIT < cycle )
printk("\n[%s] thread[%x,%x] exit for process %x / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, local_process->pid, cycle );
#endif

    return 0;

} // end process_copy_init()

///////////////////////////////////////////
void process_destroy( process_t * process )
{
    xptr_t      parent_xp;
    process_t * parent_ptr;
    cxy_t       parent_cxy;
    xptr_t      children_lock_xp;
    xptr_t      children_nr_xp;

    pid_t       pid = process->pid;

// check no more threads
assert( (process->th_nr == 0),
"process %x in cluster %x contains threads", pid , local_cxy );

#if DEBUG_PROCESS_DESTROY
thread_t * this = CURRENT_THREAD;
uint32_t cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_DESTROY < cycle )
printk("\n[%s] thread[%x,%x] enter for process %x in cluster %x / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, pid, local_cxy, cycle );
#endif

    // Destroy VMM
    vmm_destroy( process );

#if (DEBUG_PROCESS_DESTROY & 1)
if( DEBUG_PROCESS_DESTROY < cycle )
printk("\n[%s] thread[%x,%x] destroyed VMM for process %x in cluster %x\n",
__FUNCTION__, this->process->pid, this->trdid, pid, local_cxy );
#endif

    // remove process from local_list in local cluster manager
    cluster_process_local_unlink( process );

#if (DEBUG_PROCESS_DESTROY & 1)
if( DEBUG_PROCESS_DESTROY < cycle )
printk("\n[%s] thread[%x,%x] removed process %x in cluster %x from local list\n",
__FUNCTION__, this->process->pid, this->trdid, pid, local_cxy );
#endif

    // remove process from copies_list in owner cluster manager
    cluster_process_copies_unlink( process );

#if (DEBUG_PROCESS_DESTROY & 1)
if( DEBUG_PROCESS_DESTROY < cycle )
printk("\n[%s] thread[%x,%x] removed process %x in cluster %x from copies list\n",
__FUNCTION__, this->process->pid, this->trdid, pid, local_cxy );
#endif

    // when target process cluster is the owner cluster
    // - remove process from TXT list and transfer ownership
    // - remove process from children_list 
    // - release PID
    if( CXY_FROM_PID( pid ) == local_cxy )
    {
        process_txt_detach( XPTR( local_cxy , process ) );

#if (DEBUG_PROCESS_DESTROY & 1)
if( DEBUG_PROCESS_DESTROY < cycle )
printk("\n[%s] thread[%x,%x] removed process %x from TXT list\n",
__FUNCTION__, this->process->pid, this->trdid, pid );
#endif

        // get pointers on parent process
        parent_xp  = process->parent_xp;
        parent_cxy = GET_CXY( parent_xp );
        parent_ptr = GET_PTR( parent_xp );

        // get extended pointer on children_lock in parent process
        children_lock_xp = XPTR( parent_cxy , &parent_ptr->children_lock );
        children_nr_xp   = XPTR( parent_cxy , &parent_ptr->children_nr );

        // remove process from children_list
        remote_queuelock_acquire( children_lock_xp );
        xlist_unlink( XPTR( local_cxy , &process->children_list ) );
            hal_remote_atomic_add( children_nr_xp , -1 );
        remote_queuelock_release( children_lock_xp );

#if (DEBUG_PROCESS_DESTROY & 1)
if( DEBUG_PROCESS_DESTROY < cycle )
printk("\n[%s] thread[%x,%x] removed process %x from parent process children list\n",
__FUNCTION__, this->process->pid, this->trdid, pid );
#endif

        // release the process PID to cluster manager
        cluster_pid_release( pid );

#if (DEBUG_PROCESS_DESTROY & 1)
if( DEBUG_PROCESS_DESTROY < cycle )
printk("\n[%s] thread[%x,%x] released process PID %x to pmgr in cluster %x\n",
__FUNCTION__, this->process->pid, this->trdid, pid, local_cxy );
#endif

    }

    // FIXME decrement the refcount on file pointer for vfs_bin_xp [AG]

    // FIXME close all open files [AG]

    // FIXME synchronize dirty files [AG]

    // release memory allocated to process descriptor
    process_free( process );

#if DEBUG_PROCESS_DESTROY
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_DESTROY < cycle )
printk("\n[%s] thread[%x,%x] exit / process %x in cluster %x / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, pid, local_cxy, cycle );
#endif

}  // end process_destroy()

///////////////////////////////////////////////////////////////////
const char * process_action_str( process_sigactions_t action_type )
{
    switch ( action_type )
    {
        case BLOCK_ALL_THREADS:   return "BLOCK";
        case UNBLOCK_ALL_THREADS: return "UNBLOCK";
        case DELETE_ALL_THREADS:  return "DELETE";
        default:                  return "undefined";
    }
}

////////////////////////////////////////
void process_sigaction( pid_t       pid,
                        uint32_t    type )
{
    cxy_t              owner_cxy;         // owner cluster identifier
    lpid_t             lpid;              // process index in owner cluster
    cluster_t        * cluster;           // pointer on cluster manager
    xptr_t             root_xp;           // extended pointer on root of copies
    xptr_t             lock_xp;           // extended pointer on lock protecting copies
    xptr_t             iter_xp;           // iterator on copies list
    xptr_t             process_xp;        // extended pointer on process copy
    cxy_t              process_cxy;       // process copy cluster identifier
    process_t        * process_ptr;       // local pointer on process copy
    reg_t              save_sr;           // for critical section
    thread_t         * client;            // pointer on client thread
    xptr_t             client_xp;         // extended pointer on client thread
    process_t        * local;             // pointer on process copy in local cluster
    uint32_t           remote_nr;         // number of remote process copies
    rpc_desc_t         rpc;               // shared RPC descriptor
    uint32_t           responses;         // shared RPC responses counter

    client    = CURRENT_THREAD;
    client_xp = XPTR( local_cxy , client );
    local     = NULL;
    remote_nr = 0;

    // check calling thread can yield
    thread_assert_can_yield( client , __FUNCTION__ );

#if DEBUG_PROCESS_SIGACTION
uint32_t cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_SIGACTION < cycle )
printk("\n[%s] thread[%x,%x] enter to %s process %x / cycle %d\n",
__FUNCTION__ , client->process->pid, client->trdid,
process_action_str( type ) , pid , cycle );
#endif

    // get pointer on local cluster manager
    cluster = LOCAL_CLUSTER;

    // get owner cluster identifier and process lpid
    owner_cxy = CXY_FROM_PID( pid );
    lpid      = LPID_FROM_PID( pid );

    // get root of list of copies and lock from owner cluster
    root_xp   = XPTR( owner_cxy , &cluster->pmgr.copies_root[lpid] );
    lock_xp   = XPTR( owner_cxy , &cluster->pmgr.copies_lock[lpid] );

// check action type
assert( ((type == DELETE_ALL_THREADS ) ||
         (type == BLOCK_ALL_THREADS )  ||
         (type == UNBLOCK_ALL_THREADS )), "illegal action type" );
             
    // This client thread send parallel RPCs to all remote clusters containing
    // target process copies, wait all responses, and then handles directly 
    // the threads in local cluster, when required.
    // The client thread allocates a - shared - RPC descriptor in the stack,
    // because all parallel, non-blocking, server threads use the same input 
    // arguments, and use the shared RPC response field

    // mask IRQs
    hal_disable_irq( &save_sr);

    // client thread blocks itself
    thread_block( client_xp , THREAD_BLOCKED_RPC );

    // initialize RPC responses counter
    responses = 0;

    // initialize shared RPC descriptor
    // can be shared, because no out arguments
    rpc.rsp       = &responses;
    rpc.blocking  = false;
    rpc.index     = RPC_PROCESS_SIGACTION;
    rpc.thread    = client;
    rpc.lid       = client->core->lid;
    rpc.args[0]   = pid;
    rpc.args[1]   = type;

    // take the lock protecting process copies
    remote_queuelock_acquire( lock_xp );

    // scan list of process copies
    XLIST_FOREACH( root_xp , iter_xp )
    {
        // get extended pointers and cluster on process
        process_xp  = XLIST_ELEMENT( iter_xp , process_t , copies_list );
        process_cxy = GET_CXY( process_xp );
        process_ptr = GET_PTR( process_xp );

        if( process_cxy == local_cxy )    // process copy is local
        { 
            local = process_ptr;
        }
        else                              // process copy is remote
        {
            // update number of remote process copies
            remote_nr++;

            // atomically increment RPC responses counter
            hal_atomic_add( &responses , 1 );

#if DEBUG_PROCESS_SIGACTION 
if( DEBUG_PROCESS_SIGACTION < cycle )
printk("\n[%s] thread[%x,%x] send RPC to cluster %x for process %x\n",
__FUNCTION__, client->process->pid, client->trdid, process_cxy, pid );
#endif
            // call RPC in target cluster
            rpc_send( process_cxy , &rpc );
        }
    }  // end list of copies

    // release the lock protecting process copies
    remote_queuelock_release( lock_xp );

    // restore IRQs
    hal_restore_irq( save_sr);

    // - if there is remote process copies, the client thread deschedules,
    //   (it will be unblocked by the last RPC server thread). 
    // - if there is no remote copies, the client thread unblock itself.
    if( remote_nr )
    {
        sched_yield("blocked on rpc_process_sigaction");
    } 
    else
    {
        thread_unblock( client_xp , THREAD_BLOCKED_RPC );
    }

    // handle the local process copy if required
    if( local != NULL )
    {

#if DEBUG_PROCESS_SIGACTION 
if( DEBUG_PROCESS_SIGACTION < cycle )
printk("\n[%s] thread[%x,%x] handles local process %x in cluster %x\n",
__FUNCTION__, client->process->pid, client->trdid, pid , local_cxy );
#endif
        if     (type == DELETE_ALL_THREADS  ) process_delete_threads ( local , client_xp ); 
        else if(type == BLOCK_ALL_THREADS   ) process_block_threads  ( local ); 
        else if(type == UNBLOCK_ALL_THREADS ) process_unblock_threads( local );
    }

#if DEBUG_PROCESS_SIGACTION
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_SIGACTION < cycle )
printk("\n[%s] thread[%x,%x] exit after %s process %x / cycle %d\n",
__FUNCTION__, client->process->pid, client->trdid,
process_action_str( type ), pid, cycle );
#endif

}  // end process_sigaction()

/////////////////////////////////////////////////
void process_block_threads( process_t * process )
{
    thread_t          * target;         // pointer on target thread 
    thread_t          * this;           // pointer on calling thread 
    uint32_t            ltid;           // index in process th_tbl[]
    uint32_t            count;          // requests counter
    volatile uint32_t   ack_count;      // acknowledges counter

    // get calling thread pointer
    this = CURRENT_THREAD;

#if DEBUG_PROCESS_SIGACTION
pid_t pid = process->pid;
uint32_t cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_SIGACTION < cycle )
printk("\n[%s] thread[%x,%x] enter for process %x in cluster %x / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, pid, local_cxy , cycle );
#endif

// check target process is an user process
assert( (LPID_FROM_PID( process->pid ) != 0 ),
"process %x is not an user process\n", process->pid );

    // get lock protecting process th_tbl[]
    rwlock_rd_acquire( &process->th_lock );

    // loop on target process local threads 
    // we use both "ltid" and "count" because it can exist "holes" in th_tbl
    // - if the calling thread and the target thread are not running on the same
    //   core, we ask the target scheduler to acknowlege the blocking
    //   to be sure that the target thread is not running.
    // - if the calling thread and the target thread are running on the same core,
    //   we don't need confirmation from scheduler.
            
    for( ltid = 0 , count = 0 , ack_count = 0 ; count < process->th_nr ; ltid++ )
    {
        target = process->th_tbl[ltid];

        if( target != NULL )                                 // thread exist
        {
            count++;

            // set the global blocked bit in target thread descriptor.
            thread_block( XPTR( local_cxy , target ) , THREAD_BLOCKED_GLOBAL );
  
            if( this->core->lid != target->core->lid )
            {
                // increment responses counter 
                hal_atomic_add( (void*)&ack_count , 1 );

                // set FLAG_REQ_ACK and &ack_rsp_count in target descriptor
                thread_set_req_ack( target , (uint32_t *)&ack_count );

                // force scheduling on target thread
                dev_pic_send_ipi( local_cxy , target->core->lid );
            }
        }
    }

    // release lock protecting process th_tbl[]
    rwlock_rd_release( &process->th_lock );

    // wait other threads acknowledges  TODO this could be improved...
    while( 1 )
    {
        // exit when all scheduler acknowledges received
        if ( ack_count == 0 ) break;
    
        // wait 1000 cycles before retry
        hal_fixed_delay( 1000 );
    }

#if DEBUG_PROCESS_SIGACTION
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_SIGACTION < cycle )
printk("\n[%s] thread[%x,%x] exit for process %x in cluster %x / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, pid, local_cxy , cycle );
#endif

}  // end process_block_threads()

/////////////////////////////////////////////////
void process_delete_threads( process_t * process,
                             xptr_t      client_xp )
{
    thread_t          * target;        // local pointer on target thread 
    xptr_t              target_xp;     // extended pointer on target thread
    cxy_t               owner_cxy;     // owner process cluster
    uint32_t            ltid;          // index in process th_tbl
    uint32_t            count;         // threads counter

    // get calling thread pointer

    // get target process owner cluster
    owner_cxy = CXY_FROM_PID( process->pid );

#if DEBUG_PROCESS_SIGACTION
thread_t * this  = CURRENT_THREAD;
uint32_t   cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_SIGACTION < cycle )
printk("\n[%s] thread[%x,%x] enter for process %x n cluster %x / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, process->pid, local_cxy, cycle );
#endif

// check target process is an user process
assert( (LPID_FROM_PID( process->pid ) != 0),
"process %x is not an user process\n", process->pid );

    // get lock protecting process th_tbl[]
    rwlock_wr_acquire( &process->th_lock );

    // loop on target process local threads                       
    // we use both "ltid" and "count" because it can exist "holes" in th_tbl
    for( ltid = 0 , count = 0  ; count < process->th_nr ; ltid++ )
    {
        target = process->th_tbl[ltid];

        if( target != NULL )    // valid thread  
        {
            count++;
            target_xp = XPTR( local_cxy , target );

            // main thread and client thread should not be deleted
            if( ((ltid != 0) || (owner_cxy != local_cxy)) &&         // not main thread
                (client_xp) != target_xp )                           // not client thread
            {
                // mark target thread for delete and block it
                thread_delete( target_xp , true );                   // forced 
            }
        }
    }

    // release lock protecting process th_tbl[]
    rwlock_wr_release( &process->th_lock );

#if DEBUG_PROCESS_SIGACTION
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_SIGACTION < cycle )
printk("\n[%s] thread[%x,%x] exit for process %x in cluster %x / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, process->pid, local_cxy , cycle );
#endif

}  // end process_delete_threads()

///////////////////////////////////////////////////
void process_unblock_threads( process_t * process )
{
    thread_t          * target;        // pointer on target thead 
    uint32_t            ltid;          // index in process th_tbl
    uint32_t            count;         // requests counter

#if DEBUG_PROCESS_SIGACTION
thread_t * this  = CURRENT_THREAD;
pid_t      pid   = process->pid;
uint32_t   cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_SIGACTION < cycle )
printk("\n[%s] thread[%x,%x] enter for process %x in cluster %x / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, pid, local_cxy , cycle );
#endif

// check target process is an user process
assert( ( LPID_FROM_PID( process->pid ) != 0 ),
"process %x is not an user process\n", process->pid );

    // get lock protecting process th_tbl[]
    rwlock_rd_acquire( &process->th_lock );

    // loop on process threads to unblock all threads
    // we use both "ltid" and "count" because it can exist "holes" in th_tbl
    for( ltid = 0 , count = 0 ; count < process->th_nr ; ltid++ )
    {
        target = process->th_tbl[ltid];

        if( target != NULL )             // thread found
        {
            count++;

            // reset the global blocked bit in target thread descriptor.
            thread_unblock( XPTR( local_cxy , target ) , THREAD_BLOCKED_GLOBAL );
        }
    }

    // release lock protecting process th_tbl[]
    rwlock_rd_release( &process->th_lock );

#if DEBUG_PROCESS_SIGACTION
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_SIGACTION < cycle )
printk("\n[%s] thread[%x,%x] exit for process %x in cluster %x / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, pid, local_cxy, cycle );
#endif

}  // end process_unblock_threads()

///////////////////////////////////////////////
process_t * process_get_local_copy( pid_t pid )
{
    error_t        error;
    process_t    * process_ptr;   // local pointer on process
    xptr_t         process_xp;    // extended pointer on process

    cluster_t * cluster = LOCAL_CLUSTER;

#if DEBUG_PROCESS_GET_LOCAL_COPY
thread_t * this = CURRENT_THREAD;
uint32_t cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_GET_LOCAL_COPY < cycle )
printk("\n[%s] thread[%x,%x] enter for process %x in cluster %x / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, pid, local_cxy, cycle );
#endif

    // get lock protecting local list of processes
    remote_queuelock_acquire( XPTR( local_cxy , &cluster->pmgr.local_lock ) );

    // scan the local list of process descriptors to find the process
    xptr_t  iter;
    bool_t  found = false;
    XLIST_FOREACH( XPTR( local_cxy , &cluster->pmgr.local_root ) , iter )
    {
        process_xp  = XLIST_ELEMENT( iter , process_t , local_list );
        process_ptr = GET_PTR( process_xp );
        if( process_ptr->pid == pid )
        {
            found = true;
            break;
        }
    }

    // release lock protecting local list of processes
    remote_queuelock_release( XPTR( local_cxy , &cluster->pmgr.local_lock ) );

    // allocate memory for a new local process descriptor
    // and initialise it from reference cluster if not found
    if( !found )
    {
        // get extended pointer on reference process descriptor
        xptr_t ref_xp = cluster_get_reference_process_from_pid( pid );

        assert( (ref_xp != XPTR_NULL) , "illegal pid\n" );

        // allocate memory for local process descriptor
        process_ptr = process_alloc();

        if( process_ptr == NULL )  return NULL;

        // initialize local process descriptor copy
        error = process_copy_init( process_ptr , ref_xp );

        if( error ) return NULL;
    }

#if DEBUG_PROCESS_GET_LOCAL_COPY
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_GET_LOCAL_COPY < cycle )
printk("\n[%s] thread[%x,%x] exit in cluster %x / process %x / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, local_cxy, process_ptr, cycle );
#endif

    return process_ptr;

}  // end process_get_local_copy()

////////////////////////////////////////////
pid_t process_get_ppid( xptr_t  process_xp )
{
    cxy_t       process_cxy;
    process_t * process_ptr;
    xptr_t      parent_xp;
    cxy_t       parent_cxy;
    process_t * parent_ptr;

    // get process cluster and local pointer
    process_cxy = GET_CXY( process_xp );
    process_ptr = GET_PTR( process_xp );

    // get pointers on parent process
    parent_xp  = (xptr_t)hal_remote_l64( XPTR( process_cxy , &process_ptr->parent_xp ) );
    parent_cxy = GET_CXY( parent_xp );
    parent_ptr = GET_PTR( parent_xp );

    return hal_remote_l32( XPTR( parent_cxy , &parent_ptr->pid ) );
}

//////////////////////////////////////////////////////////////////////////////////////////
// File descriptor array related functions
//////////////////////////////////////////////////////////////////////////////////////////

///////////////////////////////////////////
void process_fd_init( process_t * process )
{
    uint32_t fd;

    // initialize lock
    remote_queuelock_init( XPTR( local_cxy , &process->fd_array.lock ), LOCK_PROCESS_FDARRAY );

    // initialize number of open files
    process->fd_array.current = 0;

    // initialize array
    for ( fd = 0 ; fd < CONFIG_PROCESS_FILE_MAX_NR ; fd++ )
    {
        process->fd_array.array[fd] = XPTR_NULL;
    }
}

////////////////////////////////////////////////////
error_t process_fd_register( xptr_t      process_xp,
                             xptr_t      file_xp,
                             uint32_t  * fdid )
{
    bool_t    found;
    uint32_t  id;
    xptr_t    xp;

    // get target process cluster and local pointer
    process_t * process_ptr = GET_PTR( process_xp );
    cxy_t       process_cxy = GET_CXY( process_xp );

// check target process is reference process
assert( (process_xp == hal_remote_l64( XPTR( process_cxy , &process_ptr->ref_xp ) ) ),
"client process must be reference process\n" );

#if DEBUG_PROCESS_FD_REGISTER
thread_t * this  = CURRENT_THREAD;
uint32_t   cycle = (uint32_t)hal_get_cycles();
pid_t      pid   = hal_remote_l32( XPTR( process_cxy , &process_ptr->pid) );
if( DEBUG_PROCESS_FD_REGISTER < cycle )
printk("\n[%s] thread[%x,%x] enter for process %x / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, pid, cycle );
#endif

    // build extended pointer on lock protecting reference fd_array
    xptr_t lock_xp = XPTR( process_cxy , &process_ptr->fd_array.lock );

    // take lock protecting reference fd_array
        remote_queuelock_acquire( lock_xp );

    found   = false;

    for ( id = 0; id < CONFIG_PROCESS_FILE_MAX_NR ; id++ )
    {
        xp = hal_remote_l64( XPTR( process_cxy , &process_ptr->fd_array.array[id] ) );
        if ( xp == XPTR_NULL )
        {
            // update reference fd_array
            hal_remote_s64( XPTR( process_cxy , &process_ptr->fd_array.array[id] ) , file_xp );
                hal_remote_atomic_add( XPTR( process_cxy , &process_ptr->fd_array.current ) , 1 );

            // exit
                        *fdid = id;
            found = true;
            break;
        }
    }

    // release lock protecting fd_array
        remote_queuelock_release( lock_xp );

#if DEBUG_PROCESS_FD_REGISTER
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_FD_REGISTER < cycle )
printk("\n[%s] thread[%x,%x] exit for process %x / fdid %d / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, pid, id, cycle );
#endif

    if ( !found ) return -1;
    else          return 0;

}  // end process_fd_register()

/////////////////////////////////////////////
void process_fd_remove( xptr_t    process_xp,
                        uint32_t  fdid )
{
    pid_t       pid;           // target process PID 
    lpid_t      lpid;          // target process LPID
    xptr_t      iter_xp;       // iterator for list of process copies
    xptr_t      copy_xp;       // extended pointer on process copy
    process_t * copy_ptr;      // local pointer on process copy  
    cxy_t       copy_cxy;      // process copy cluster identifier

// check process_xp argument
assert( (process_xp != XPTR_NULL), "process_xp argument cannot be XPTR_NULL");

    // get target process cluster and local pointer
    process_t * process_ptr = GET_PTR( process_xp );
    cxy_t       process_cxy = GET_CXY( process_xp );

    // get target process pid and lpid
    pid  = hal_remote_l32( XPTR( process_cxy , &process_ptr->pid) );
    lpid = LPID_FROM_PID( pid );

    // get process descriptor in owner cluster and in reference cluster
    xptr_t  owner_xp = hal_remote_l64( XPTR( process_cxy , &process_ptr->owner_xp ));
    xptr_t  ref_xp   = hal_remote_l64( XPTR( process_cxy , &process_ptr->ref_xp   ));

// check target process in in owner cluster 
assert( (process_xp == owner_xp), "target process must be in owner process\n" );

#if DEBUG_PROCESS_FD_REMOVE
uint32_t    cycle = (uint32_t)hal_get_cycles();
thread_t  * this  = CURRENT_THREAD;
if( DEBUG_PROCESS_FD_REMOVE < cycle )
printk("\n[%s] thread[%x,%x] enter for fdid %d in process %x / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, fdid, pid, cycle );
#endif

    // build extended pointers on list_of_copies root and lock (in owner cluster)
    xptr_t copies_root_xp = XPTR( process_cxy , &LOCAL_CLUSTER->pmgr.copies_root[lpid] );
    xptr_t copies_lock_xp = XPTR( process_cxy , &LOCAL_CLUSTER->pmgr.copies_lock[lpid] );
 
    // get reference process cluster and local pointer
    process_t * ref_ptr = GET_PTR( ref_xp );
    cxy_t       ref_cxy = GET_CXY( ref_xp );

    // build extended pointer on lock protecting reference fd_array
    xptr_t fd_lock_xp = XPTR( ref_cxy , &ref_ptr->fd_array.lock );

    // take lock protecting reference fd_array
        remote_queuelock_acquire( fd_lock_xp );

    // take the lock protecting the list of copies
    remote_queuelock_acquire( copies_lock_xp );

    // loop on list of process copies
    XLIST_FOREACH( copies_root_xp , iter_xp )
    {
        // get pointers on process copy
        copy_xp  = XLIST_ELEMENT( iter_xp , process_t , copies_list );
        copy_ptr = GET_PTR( copy_xp );
        copy_cxy = GET_CXY( copy_xp );

        // release the fd_array entry in process copy
        hal_remote_s64( XPTR( copy_cxy , &copy_ptr->fd_array.array[fdid] ), XPTR_NULL );
    }

    // release the lock protecting reference fd_array
        remote_queuelock_release( fd_lock_xp );

    // release the lock protecting the list of copies
    remote_queuelock_release( copies_lock_xp );

#if DEBUG_PROCESS_FD_REMOVE
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_FD_REMOVE < cycle )
printk("\n[%s] thread[%x,%x] exit for fdid %d in process %x / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, fdid, pid, cycle );
#endif

}  // end process_fd_remove()

////////////////////////////////////////////////
xptr_t process_fd_get_xptr( process_t * process,
                            uint32_t    fdid )
{
    xptr_t  file_xp;
    xptr_t  lock_xp;

    // access local copy of process descriptor
    file_xp = process->fd_array.array[fdid];

    if( file_xp == XPTR_NULL )
    {
        // get reference process cluster and local pointer
        xptr_t      ref_xp  = process->ref_xp;
        cxy_t       ref_cxy = GET_CXY( ref_xp );
        process_t * ref_ptr = GET_PTR( ref_xp );

        // build extended pointer on lock protecting reference fd_array
        lock_xp = XPTR( ref_cxy , &ref_ptr->fd_array.lock );

        // take lock protecting reference fd_array
            remote_queuelock_acquire( lock_xp );

        // access reference process descriptor
        file_xp = hal_remote_l64( XPTR( ref_cxy , &ref_ptr->fd_array.array[fdid] ) );

        // update local fd_array if found
        if( file_xp != XPTR_NULL )  process->fd_array.array[fdid] = file_xp;
        
        // release lock protecting reference fd_array
            remote_queuelock_release( lock_xp );
    }

    return file_xp;

}  // end process_fd_get_xptr()

///////////////////////////////////////////
void process_fd_remote_copy( xptr_t dst_xp,
                             xptr_t src_xp )
{
    uint32_t fd;
    xptr_t   entry;

    // get cluster and local pointer for src fd_array
    cxy_t        src_cxy = GET_CXY( src_xp );
    fd_array_t * src_ptr = GET_PTR( src_xp );

    // get cluster and local pointer for dst fd_array
    cxy_t        dst_cxy = GET_CXY( dst_xp );
    fd_array_t * dst_ptr = GET_PTR( dst_xp );

    // get the remote lock protecting the src fd_array
        remote_queuelock_acquire( XPTR( src_cxy , &src_ptr->lock ) );

    // loop on all fd_array entries
    for( fd = 0 ; fd < CONFIG_PROCESS_FILE_MAX_NR ; fd++ )
        {
                entry = (xptr_t)hal_remote_l64( XPTR( src_cxy , &src_ptr->array[fd] ) );

                if( entry != XPTR_NULL )
                {
            // increment file descriptor refcount
            vfs_file_count_up( entry );

                        // copy entry in destination process fd_array
                        hal_remote_s64( XPTR( dst_cxy , &dst_ptr->array[fd] ) , entry );
                }
        }

    // release lock on source process fd_array
        remote_queuelock_release( XPTR( src_cxy , &src_ptr->lock ) );

}  // end process_fd_remote_copy()


////////////////////////////////////
bool_t process_fd_array_full( void )
{
    // get extended pointer on reference process
    xptr_t ref_xp = CURRENT_THREAD->process->ref_xp;

    // get reference process cluster and local pointer
    process_t * ref_ptr = GET_PTR( ref_xp );
    cxy_t       ref_cxy = GET_CXY( ref_xp );

    // get number of open file descriptors from reference fd_array
    uint32_t current = hal_remote_l32( XPTR( ref_cxy , &ref_ptr->fd_array.current ) );

        return ( current >= CONFIG_PROCESS_FILE_MAX_NR );
}


////////////////////////////////////////////////////////////////////////////////////
//  Thread related functions
////////////////////////////////////////////////////////////////////////////////////

/////////////////////////////////////////////////////
error_t process_register_thread( process_t * process,
                                 thread_t  * thread,
                                 trdid_t   * trdid )
{
    ltid_t         ltid;
    bool_t         found = false;
 
// check arguments
assert( (process != NULL) , "process argument is NULL" );
assert( (thread != NULL) , "thread argument is NULL" );

    // get the lock protecting th_tbl for all threads
    // but the idle thread executing kernel_init (cannot yield)
    if( thread->type != THREAD_IDLE ) rwlock_wr_acquire( &process->th_lock );

    // scan th_tbl
    for( ltid = 0 ; ltid < CONFIG_THREADS_MAX_PER_CLUSTER ; ltid++ )
    {
        if( process->th_tbl[ltid] == NULL )
        {
            found = true;
            break;
        }
    }

    if( found )
    {
        // register thread in th_tbl[]
        process->th_tbl[ltid] = thread;
        process->th_nr++;

        // returns trdid
        *trdid = TRDID( local_cxy , ltid );
    }

    // release the lock protecting th_tbl
    if( thread->type != THREAD_IDLE ) rwlock_wr_release( &process->th_lock );

    return (found) ? 0 : 0xFFFFFFFF;

}  // end process_register_thread()

///////////////////////////////////////////////////
uint32_t process_remove_thread( thread_t * thread )
{
    uint32_t count;  // number of threads in local process descriptor

// check thread
assert( (thread != NULL) , "thread argument is NULL" );

    process_t * process = thread->process;

    // get thread local index
    ltid_t  ltid = LTID_FROM_TRDID( thread->trdid );
    
    // get the lock protecting th_tbl[]
    rwlock_wr_acquire( &process->th_lock );

    // get number of threads
    count = process->th_nr;

// check th_nr value
assert( (count > 0) , "process th_nr cannot be 0" );

    // remove thread from th_tbl[]
    process->th_tbl[ltid] = NULL;
    process->th_nr = count-1;

    // release lock protecting th_tbl
    rwlock_wr_release( &process->th_lock );

    return count;

}  // end process_remove_thread()

/////////////////////////////////////////////////////////
error_t process_make_fork( xptr_t      parent_process_xp,
                           xptr_t      parent_thread_xp,
                           pid_t     * child_pid,
                           thread_t ** child_thread )
{
    process_t * process;         // local pointer on child process descriptor
    thread_t  * thread;          // local pointer on child thread descriptor
    pid_t       new_pid;         // process identifier for child process
    pid_t       parent_pid;      // process identifier for parent process
    xptr_t      ref_xp;          // extended pointer on reference process
    xptr_t      vfs_bin_xp;      // extended pointer on .elf file
    error_t     error;

    // get cluster and local pointer for parent process
    cxy_t       parent_process_cxy = GET_CXY( parent_process_xp );
    process_t * parent_process_ptr = GET_PTR( parent_process_xp );

    // get parent process PID and extended pointer on .elf file
    parent_pid = hal_remote_l32 (XPTR( parent_process_cxy , &parent_process_ptr->pid));
    vfs_bin_xp = hal_remote_l64(XPTR( parent_process_cxy , &parent_process_ptr->vfs_bin_xp));

    // get extended pointer on reference process
    ref_xp = hal_remote_l64( XPTR( parent_process_cxy , &parent_process_ptr->ref_xp ) );

// check parent process is the reference process
assert( (parent_process_xp == ref_xp ) ,
"parent process must be the reference process" );

#if DEBUG_PROCESS_MAKE_FORK
uint32_t   cycle;
thread_t * this  = CURRENT_THREAD;
trdid_t    trdid = this->trdid;
pid_t      pid   = this->process->pid;
#endif

#if( DEBUG_PROCESS_MAKE_FORK & 1 )
cycle   = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_MAKE_FORK < cycle )
printk("\n[%s] thread[%x,%x] enter / cluster %x / cycle %d\n",
__FUNCTION__, pid, trdid, local_cxy, cycle );
#endif

    // allocate a process descriptor
    process = process_alloc();

    if( process == NULL )
    {
        printk("\n[ERROR] in %s : cannot get process in cluster %x\n", 
        __FUNCTION__, local_cxy ); 
        return -1;
    }

    // allocate a child PID from local cluster
    error = cluster_pid_alloc( process , &new_pid );
    if( error ) 
    {
        printk("\n[ERROR] in %s : cannot get PID in cluster %x\n", 
        __FUNCTION__, local_cxy ); 
        process_free( process );
        return -1;
    }

#if( DEBUG_PROCESS_MAKE_FORK & 1 )
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_MAKE_FORK < cycle )
printk("\n[%s] thread[%x,%x] allocated child_process %x / cycle %d\n",
__FUNCTION__, pid, trdid, new_pid, cycle );
#endif

    // initializes child process descriptor from parent process descriptor 
    error = process_reference_init( process,
                                    new_pid,
                                    parent_process_xp );
    if( error ) 
    {
        printk("\n[ERROR] in %s : cannot initialize child process in cluster %x\n", 
        __FUNCTION__, local_cxy ); 
        process_free( process );
        return -1;
    }

#if( DEBUG_PROCESS_MAKE_FORK & 1 )
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_MAKE_FORK < cycle )
printk("\n[%s] thread[%x,%x] initialized child_process %x / cycle %d\n",
__FUNCTION__, pid, trdid, new_pid, cycle );
#endif

    // copy VMM from parent descriptor to child descriptor
    error = vmm_fork_copy( process,
                           parent_process_xp );
    if( error )
    {
        printk("\n[ERROR] in %s : cannot copy VMM in cluster %x\n", 
        __FUNCTION__, local_cxy ); 
        process_free( process );
        cluster_pid_release( new_pid );
        return -1;
    }

#if( DEBUG_PROCESS_MAKE_FORK & 1 )
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_MAKE_FORK < cycle )
printk("\n[%s] thread[%x,%x] copied VMM from parent to child / cycle %d\n",
__FUNCTION__, pid, trdid, cycle );
hal_vmm_display( XPTR( local_cxy , process ) , true );
#endif

    // if parent_process is INIT, or if parent_process is the TXT owner,
    // the child_process becomes the owner of its TXT terminal
    if( (parent_pid == 1) || process_txt_is_owner( parent_process_xp ) )
    {
        process_txt_set_ownership( XPTR( local_cxy , process ) );

#if( DEBUG_PROCESS_MAKE_FORK & 1 )
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_MAKE_FORK < cycle )
printk("\n[%s] thread[%x,%x] / child_process %x takes TXT ownership / cycle %d\n",
__FUNCTION__ , pid, trdid, new_pid, cycle );
#endif

    }

    // update extended pointer on .elf file
    process->vfs_bin_xp = vfs_bin_xp;

    // create child thread descriptor from parent thread descriptor
    error = thread_user_fork( parent_thread_xp,
                              process,
                              &thread );
    if( error )
    {
        printk("\n[ERROR] in %s : cannot create thread in cluster %x\n",
        __FUNCTION__, local_cxy ); 
        process_free( process );
        cluster_pid_release( new_pid );
        return -1;
    }

// check main thread LTID
assert( (LTID_FROM_TRDID(thread->trdid) == 0) ,
"main thread must have LTID == 0" );

#if( DEBUG_PROCESS_MAKE_FORK & 1 )
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_MAKE_FORK < cycle )
printk("\n[%s] thread[%x,%x] created main thread %x / cycle %d\n", 
__FUNCTION__, pid, trdid, thread, cycle );
#endif

    // set COW flag in DATA, ANON, REMOTE vsegs in parent process VMM
    // this includes all parent process copies in all clusters
    if( parent_process_cxy == local_cxy )   // reference is local
    {
        vmm_set_cow( parent_process_ptr );
    }
    else                                    // reference is remote
    {
        rpc_vmm_set_cow_client( parent_process_cxy,
                                parent_process_ptr );
    }

    // set COW flag in DATA, ANON, REMOTE vsegs for child process VMM
    vmm_set_cow( process );
 
#if( DEBUG_PROCESS_MAKE_FORK & 1 )
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_MAKE_FORK < cycle )
printk("\n[%s] thread[%x,%x] set COW in DATA / ANON / REMOTE for parent and child / cycle %d\n",
__FUNCTION__, pid, trdid, cycle );
#endif

    // get extended pointers on parent children_root, children_lock and children_nr
    xptr_t children_root_xp = XPTR( parent_process_cxy , &parent_process_ptr->children_root );
    xptr_t children_lock_xp = XPTR( parent_process_cxy , &parent_process_ptr->children_lock );
    xptr_t children_nr_xp   = XPTR( parent_process_cxy , &parent_process_ptr->children_nr   );

    // register process in parent children list
    remote_queuelock_acquire( children_lock_xp );
        xlist_add_last( children_root_xp , XPTR( local_cxy , &process->children_list ) );
        hal_remote_atomic_add( children_nr_xp , 1 );
    remote_queuelock_release( children_lock_xp );

    // return success
    *child_thread = thread;
    *child_pid    = new_pid;

#if DEBUG_PROCESS_MAKE_FORK
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_MAKE_FORK < cycle )
printk("\n[%s] thread[%x,%x] exit / created process %x / cycle %d\n",
__FUNCTION__, pid, trdid, new_pid, cycle );
#endif

    return 0;

}   // end process_make_fork()

/////////////////////////////////////////////////////
error_t process_make_exec( exec_info_t  * exec_info )
{
    thread_t       * thread;                  // local pointer on this thread
    process_t      * process;                 // local pointer on this process
    pid_t            pid;                     // this process identifier
    xptr_t           ref_xp;                  // reference process for this process
        error_t          error;                   // value returned by called functions
    char           * path;                    // path to .elf file
    xptr_t           file_xp;                 // extended pointer on .elf file descriptor
    uint32_t         file_id;                 // file index in fd_array
    uint32_t         args_nr;                 // number of main thread arguments
    char          ** args_pointers;           // array of pointers on main thread arguments

    // get calling thread, process, pid and ref_xp 
    thread  = CURRENT_THREAD;
    process = thread->process;
    pid     = process->pid;
    ref_xp  = process->ref_xp;

        // get relevant infos from exec_info
        path          = exec_info->path;
    args_nr       = exec_info->args_nr;
    args_pointers = exec_info->args_pointers;

#if DEBUG_PROCESS_MAKE_EXEC
uint32_t cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_MAKE_EXEC < cycle )
printk("\n[%s] thread[%x,%x] enters for %s / cycle %d\n",
__FUNCTION__, pid, thread->trdid, path, cycle );
#endif

    // open the file identified by <path>
    file_xp = XPTR_NULL;
    file_id = 0xFFFFFFFF;
        error   = vfs_open( process->vfs_root_xp,
                            path,
                        ref_xp,
                            O_RDONLY,
                            0,
                            &file_xp,
                            &file_id );
        if( error )
        {
                printk("\n[ERROR] in %s : failed to open file <%s>\n", __FUNCTION__ , path );
                return -1;
        }

#if (DEBUG_PROCESS_MAKE_EXEC & 1)
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_MAKE_EXEC < cycle )
printk("\n[%s] thread[%x,%x] opened file <%s> / cycle %d\n",
__FUNCTION__, pid, thread->trdid, path, cycle );
#endif

    // delete all threads other than this main thread in all clusters
    process_sigaction( pid , DELETE_ALL_THREADS );

#if (DEBUG_PROCESS_MAKE_EXEC & 1)
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_MAKE_EXEC < cycle )
printk("\n[%s] thread[%x,%x] deleted existing threads / cycle %d\n",
__FUNCTION__, pid, thread->trdid, cycle );
#endif

    // reset calling process VMM
    vmm_user_reset( process );

#if( DEBUG_PROCESS_MAKE_EXEC & 1 )
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_MAKE_EXEC < cycle )
printk("\n[%s] thread[%x,%x] completed VMM reset / cycle %d\n",
__FUNCTION__, pid, thread->trdid, cycle );
#endif

    // re-initialize the VMM (args/envs vsegs registration)
    error = vmm_user_init( process );
    if( error )
    {
        printk("\n[ERROR] in %s : cannot initialise VMM for %s\n", __FUNCTION__ , path );
        vfs_close( file_xp , file_id );
        // FIXME restore old process VMM [AG]
        return -1;
    }
    
#if( DEBUG_PROCESS_MAKE_EXEC & 1 )
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_MAKE_EXEC < cycle )
printk("\n[%s] thread[%x,%x] registered args/envs vsegs / cycle %d\n",
__FUNCTION__, pid, thread->trdid, cycle );
#endif

    // register code & data vsegs as well as entry-point in process VMM,
    // and register extended pointer on .elf file in process descriptor
        error = elf_load_process( file_xp , process );
    if( error )
        {
                printk("\n[ERROR] in %s : failed to access <%s>\n", __FUNCTION__ , path );
        vfs_close( file_xp , file_id );
        // FIXME restore old process VMM [AG]
        return -1;
        }

#if( DEBUG_PROCESS_MAKE_EXEC & 1 )
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_MAKE_EXEC < cycle )
printk("\n[%s] thread[%x,%x] registered code/data vsegs / cycle %d\n",
__FUNCTION__, pid, thread->trdid, cycle );
#endif

    // update the existing main thread descriptor... and jump to user code 
    error = thread_user_exec( (void *)process->vmm.entry_point,
                              args_nr,
                              args_pointers );
    if( error )
    {
        printk("\n[ERROR] in %s : cannot update main thread for %s\n", __FUNCTION__ , path );
        vfs_close( file_xp , file_id );
        // FIXME restore old process VMM 
        return -1;
    }

    assert( false, "we should not execute this code");
 
        return 0;

}  // end process_make_exec()


////////////////////////////////////////////////
void process_zero_create( process_t   * process,
                          boot_info_t * info )
{
    error_t error;
    pid_t   pid;

#if DEBUG_PROCESS_ZERO_CREATE
uint32_t cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_ZERO_CREATE < cycle )
printk("\n[%s] enter / cluster %x / cycle %d\n",
__FUNCTION__, local_cxy, cycle );
#endif

    // get pointer on VMM
    vmm_t * vmm = &process->vmm;

    // get PID from local cluster manager for this kernel process
    error = cluster_pid_alloc( process , &pid );

    if( error || (LPID_FROM_PID( pid ) != 0) )
    {
        printk("\n[PANIC] in %s : cannot get valid PID in cluster %x / PID = %x\n",
        __FUNCTION__ , local_cxy, pid );
        hal_core_sleep();
    }

#if (DEBUG_PROCESS_ZERO_CREATE & 1)
if( DEBUG_PROCESS_ZERO_CREATE < cycle )
printk("\n[%s] allocated pid %x in cluster %x\n", __FUNCTION__, pid, local_cxy );
#endif

    // initialize PID, REF_XP, PARENT_XP, and STATE
    // the kernel process_zero is its own parent_process,
    // reference_process, and owner_process, and cannot be killed...
    process->pid        = pid;
    process->ref_xp     = XPTR( local_cxy , process );
    process->owner_xp   = XPTR( local_cxy , process );
    process->parent_xp  = XPTR( local_cxy , process );
    process->term_state = 0;

    // initialize VSL as empty
    vmm->vsegs_nr = 0;
        xlist_root_init( XPTR( local_cxy , &vmm->vsegs_root ) );

#if (DEBUG_PROCESS_ZERO_CREATE & 1)
if( DEBUG_PROCESS_ZERO_CREATE < cycle )
printk("\n[%s] initialized VSL empty in cluster %x\n", __FUNCTION__, local_cxy );
#endif

    // initialize GPT as empty
    error = hal_gpt_create( &vmm->gpt );

    if( error ) 
    {
        printk("\n[PANIC] in %s : cannot create empty GPT\n", __FUNCTION__ );
        hal_core_sleep();
    }

#if (DEBUG_PROCESS_ZERO_CREATE & 1)
if( DEBUG_PROCESS_ZERO_CREATE < cycle )
printk("\n[%s] initialized GPT empty in cluster %x\n", __FUNCTION__, local_cxy );
#endif

    // initialize VSL and GPT locks
    remote_rwlock_init( XPTR( local_cxy , &vmm->vsl_lock ) , LOCK_VMM_VSL );
    
    // create kernel vsegs in GPT and VSL, as required by the hardware architecture
    error = hal_vmm_kernel_init( info );

    if( error ) 
    {
        printk("\n[PANIC] in %s : cannot create kernel vsegs in VMM\n", __FUNCTION__ );
        hal_core_sleep();
    }

#if (DEBUG_PROCESS_ZERO_CREATE & 1)
if( DEBUG_PROCESS_ZERO_CREATE < cycle )
printk("\n[%s] initialized hal specific VMM in cluster%x\n", __FUNCTION__, local_cxy );
#endif

    // reset th_tbl[] array and associated fields
    uint32_t i;
    for( i = 0 ; i < CONFIG_THREADS_MAX_PER_CLUSTER ; i++ )
        {
        process->th_tbl[i] = NULL;
    }
    process->th_nr  = 0;
    rwlock_init( &process->th_lock , LOCK_PROCESS_THTBL );

#if (DEBUG_PROCESS_ZERO_CREATE & 1)
if( DEBUG_PROCESS_ZERO_CREATE < cycle )
printk("\n[%s] initialized th_tbl[] in cluster%x\n", __FUNCTION__, local_cxy );
#endif

    // reset children list as empty
    xlist_root_init( XPTR( local_cxy , &process->children_root ) );
    process->children_nr = 0;
    remote_queuelock_init( XPTR( local_cxy , &process->children_lock ),
                           LOCK_PROCESS_CHILDREN );

#if (DEBUG_PROCESS_ZERO_CREATE & 1)
if( DEBUG_PROCESS_ZERO_CREATE < cycle )
printk("\n[%s] initialized children list in cluster%x\n", __FUNCTION__, local_cxy );
#endif

    // register kernel process in cluster manager local_list
    cluster_process_local_link( process );
    
        hal_fence();

#if DEBUG_PROCESS_ZERO_CREATE
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_ZERO_CREATE < cycle )
printk("\n[%s] exit / cluster %x / cycle %d\n",
__FUNCTION__, local_cxy, cycle );
#endif

}  // end process_zero_create()

////////////////////////////////
void process_init_create( void )
{
    process_t      * process;       // local pointer on process descriptor
    pid_t            pid;           // process_init identifier
    thread_t       * thread;        // local pointer on main thread
    pthread_attr_t   attr;          // main thread attributes
    lid_t            lid;           // selected core local index for main thread
    xptr_t           file_xp;       // extended pointer on .elf file descriptor
    uint32_t         file_id;       // file index in fd_array
    error_t          error;

#if DEBUG_PROCESS_INIT_CREATE
thread_t * this = CURRENT_THREAD;
uint32_t cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_INIT_CREATE < cycle )
printk("\n[%s] thread[%x,%x] enter / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, cycle );
#endif

    // allocates memory for process descriptor from local cluster
        process = process_alloc(); 
    if( process == NULL )
    {
        printk("\n[PANIC] in %s : cannot allocate process\n", __FUNCTION__ );
        hal_core_sleep();
    }

    // set the CWD and VFS_ROOT fields in process descriptor
    process->cwd_xp      = process_zero.vfs_root_xp;
    process->vfs_root_xp = process_zero.vfs_root_xp;

    // get PID from local cluster
    error = cluster_pid_alloc( process , &pid );
    if( error ) 
    {
        printk("\n[PANIC] in %s : cannot allocate PID\n", __FUNCTION__ );
        hal_core_sleep();
    }
    if( pid != 1 ) 
    {
        printk("\n[PANIC] in %s : process PID must be 0x1\n", __FUNCTION__ );
        hal_core_sleep();
    }

    // initialize process descriptor / parent is local process_zero
    error = process_reference_init( process,
                                    pid,
                                    XPTR( local_cxy , &process_zero ) );  
    if( error )
    {
        printk("\n[PANIC] in %s : cannot initialize process\n", __FUNCTION__ );
        hal_core_sleep();
    }

#if(DEBUG_PROCESS_INIT_CREATE & 1)
if( DEBUG_PROCESS_INIT_CREATE < cycle )
printk("\n[%s] thread[%x,%x] initialized process descriptor\n",
__FUNCTION__, this->process->pid, this->trdid );
#endif

    // open the file identified by CONFIG_PROCESS_INIT_PATH
    file_xp = XPTR_NULL;
    file_id = -1;
        error   = vfs_open( process->vfs_root_xp,
                            CONFIG_PROCESS_INIT_PATH,
                        XPTR( local_cxy , process ),
                            O_RDONLY,
                            0,
                            &file_xp,
                            &file_id );
    if( error )
    {
        printk("\n[PANIC] in %s : cannot open file <%s>\n",
         __FUNCTION__, CONFIG_PROCESS_INIT_PATH  );
        hal_core_sleep();
    }

#if(DEBUG_PROCESS_INIT_CREATE & 1)
if( DEBUG_PROCESS_INIT_CREATE < cycle )
printk("\n[%s] thread[%x,%x] open .elf file decriptor\n",
__FUNCTION__, this->process->pid, this->trdid );
#endif

    // register "code" and "data" vsegs as well as entry-point
    // in process VMM, using information contained in the elf file.
        error = elf_load_process( file_xp , process );

    if( error ) 
    {
        printk("\n[PANIC] in %s : cannot access file <%s>\n",
         __FUNCTION__, CONFIG_PROCESS_INIT_PATH  );
        hal_core_sleep();
    }


#if(DEBUG_PROCESS_INIT_CREATE & 1)
if( DEBUG_PROCESS_INIT_CREATE < cycle )
printk("\n[%s] thread[%x,%x] registered code/data vsegs in VMM\n",
__FUNCTION__, this->process->pid, this->trdid );
#endif

#if (DEBUG_PROCESS_INIT_CREATE & 1)
hal_vmm_display( XPTR( local_cxy , process ) , true );
#endif

    // get extended pointers on process_zero children_root, children_lock
    xptr_t children_root_xp = XPTR( local_cxy , &process_zero.children_root );
    xptr_t children_lock_xp = XPTR( local_cxy , &process_zero.children_lock );

    // take lock protecting kernel process children list
    remote_queuelock_acquire( children_lock_xp );

    // register process INIT in parent local process_zero 
        xlist_add_last( children_root_xp , XPTR( local_cxy , &process->children_list ) );
        hal_atomic_add( &process_zero.children_nr , 1 );

    // release lock protecting kernel process children list
    remote_queuelock_release( children_lock_xp );

#if(DEBUG_PROCESS_INIT_CREATE & 1)
if( DEBUG_PROCESS_INIT_CREATE < cycle )
printk("\n[%s] thread[%x,%x] registered init process in parent\n",
__FUNCTION__, this->process->pid, this->trdid );
#endif

    // select a core in local cluster to execute the main thread
    lid  = cluster_select_local_core( local_cxy );

    // initialize pthread attributes for main thread
    attr.attributes = PT_ATTR_DETACH | PT_ATTR_CLUSTER_DEFINED | PT_ATTR_CORE_DEFINED;
    attr.cxy        = local_cxy;
    attr.lid        = lid;

    // create and initialize thread descriptor
        error = thread_user_create( pid,
                                (void *)process->vmm.entry_point,
                                NULL,
                                &attr,
                                &thread );

    if( error )
    {
        printk("\n[PANIC] in %s : cannot create main thread\n", __FUNCTION__  );
        hal_core_sleep();
    }
    if( thread->trdid != 0 )
    {
        printk("\n[PANIC] in %s : bad main thread trdid\n", __FUNCTION__  );
        hal_core_sleep();
    }

#if(DEBUG_PROCESS_INIT_CREATE & 1)
if( DEBUG_PROCESS_INIT_CREATE < cycle )
printk("\n[%s] thread[%x,%x] created main thread\n",
__FUNCTION__, this->process->pid, this->trdid );
#endif

    // activate thread
        thread_unblock( XPTR( local_cxy , thread ) , THREAD_BLOCKED_GLOBAL );

    hal_fence();

#if DEBUG_PROCESS_INIT_CREATE
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_INIT_CREATE < cycle )
printk("\n[%s] thread[%x,%x] exit / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, cycle );
#endif

}  // end process_init_create()

/////////////////////////////////////////
void process_display( xptr_t process_xp )
{
    process_t   * process_ptr;
    cxy_t         process_cxy;

    xptr_t        parent_xp;       // extended pointer on parent process
    process_t   * parent_ptr;
    cxy_t         parent_cxy;

    xptr_t        owner_xp;        // extended pointer on owner process
    process_t   * owner_ptr;
    cxy_t         owner_cxy;

    pid_t         pid;
    pid_t         ppid;
    lpid_t        lpid;
    uint32_t      state;
    uint32_t      th_nr;

    xptr_t        txt_file_xp;     // extended pointer on TXT_RX file descriptor
    xptr_t        txt_chdev_xp;    // extended pointer on TXT_RX chdev
    chdev_t     * txt_chdev_ptr;
    cxy_t         txt_chdev_cxy;
    xptr_t        txt_owner_xp;    // extended pointer on TXT owner process 

    xptr_t        elf_file_xp;     // extended pointer on .elf file 
    cxy_t         elf_file_cxy;
    vfs_file_t  * elf_file_ptr;
    vfs_inode_t * elf_inode_ptr;   // local pointer on .elf inode

    char          txt_name[CONFIG_VFS_MAX_NAME_LENGTH];
    char          elf_name[CONFIG_VFS_MAX_NAME_LENGTH];

    // get cluster and local pointer on process
    process_ptr = GET_PTR( process_xp );
    process_cxy = GET_CXY( process_xp );

    // get process PID, LPID, and state
    pid   = hal_remote_l32( XPTR( process_cxy , &process_ptr->pid ) );
    lpid  = LPID_FROM_PID( pid );
    state = hal_remote_l32( XPTR( process_cxy , &process_ptr->term_state ) );

    // get process PPID
    parent_xp  = hal_remote_l64( XPTR( process_cxy , &process_ptr->parent_xp ) );
    parent_cxy = GET_CXY( parent_xp );
    parent_ptr = GET_PTR( parent_xp );
    ppid       = hal_remote_l32( XPTR( parent_cxy , &parent_ptr->pid ) );

    // get number of threads
    th_nr      = hal_remote_l32( XPTR( process_cxy , &process_ptr->th_nr ) );

    // get pointers on owner process descriptor
    owner_xp  = hal_remote_l64( XPTR( process_cxy , &process_ptr->owner_xp ) );
    owner_cxy = GET_CXY( owner_xp );
    owner_ptr = GET_PTR( owner_xp );

    // get process TXT name and .elf name
    if( lpid )                                   // user process
    {

        // get extended pointer on file descriptor associated to TXT_RX 
        txt_file_xp = hal_remote_l64( XPTR( owner_cxy , &owner_ptr->fd_array.array[0] ) );

        assert( (txt_file_xp != XPTR_NULL) ,
        "process must be attached to one TXT terminal" ); 

        // get TXT_RX chdev pointers
        txt_chdev_xp  = chdev_from_file( txt_file_xp );
        txt_chdev_cxy = GET_CXY( txt_chdev_xp );
        txt_chdev_ptr = GET_PTR( txt_chdev_xp );

        // get TXT_RX name and ownership
        hal_remote_strcpy( XPTR( local_cxy , txt_name ) ,
                           XPTR( txt_chdev_cxy , txt_chdev_ptr->name ) );
    
        txt_owner_xp = (xptr_t)hal_remote_l64( XPTR( txt_chdev_cxy, 
                                                     &txt_chdev_ptr->ext.txt.owner_xp ) );

        // get process .elf name
        elf_file_xp   = hal_remote_l64( XPTR( process_cxy , &process_ptr->vfs_bin_xp ) );
        elf_file_cxy  = GET_CXY( elf_file_xp );
        elf_file_ptr  = GET_PTR( elf_file_xp );
        elf_inode_ptr = hal_remote_lpt( XPTR( elf_file_cxy , &elf_file_ptr->inode ) );
        vfs_inode_get_name( XPTR( elf_file_cxy , elf_inode_ptr ) , elf_name );
    }
    else                                         // kernel process_zero
    {
        // TXT name and .elf name are not registered in kernel process_zero
        strcpy( txt_name , "txt0_rx" );
        txt_owner_xp = process_xp; 
        strcpy( elf_name , "kernel.elf" );
    }

    // display process info
    if( txt_owner_xp == process_xp )
    {
        nolock_printk("PID %X | %s (FG) | %X | PPID %X | TS %X | %d | %s\n", 
        pid, txt_name, process_ptr, ppid, state, th_nr, elf_name );
    }
    else
    {
        nolock_printk("PID %X | %s (BG) | %X | PPID %X | TS %X | %d | %s\n", 
        pid, txt_name, process_ptr, ppid, state, th_nr, elf_name );
    }
}  // end process_display() 


////////////////////////////////////////////////////////////////////////////////////////
//     Terminals related functions
////////////////////////////////////////////////////////////////////////////////////////

//////////////////////////////////
uint32_t process_txt_alloc( void )
{
    uint32_t  index;       // TXT terminal index
    xptr_t    chdev_xp;    // extended pointer on TXT_RX chdev
    chdev_t * chdev_ptr;   // local pointer on TXT_RX chdev
    cxy_t     chdev_cxy;   // TXT_RX chdev cluster
    xptr_t    root_xp;     // extended pointer on owner field in chdev

    // scan the user TXT_RX chdevs (TXT0 is reserved for kernel)
    for( index = 1 ; index < LOCAL_CLUSTER->nb_txt_channels ; index ++ )
    {
        // get pointers on TXT_RX[index]
        chdev_xp  = chdev_dir.txt_rx[index];
        chdev_cxy = GET_CXY( chdev_xp );
        chdev_ptr = GET_PTR( chdev_xp );

        // get extended pointer on root of attached process
        root_xp = XPTR( chdev_cxy , &chdev_ptr->ext.txt.root );

        // return free TXT index if found
        if( xlist_is_empty( root_xp ) ) return index;  
    }

    assert( false , "no free TXT terminal found" );

    return -1;

} // end process_txt_alloc()

/////////////////////////////////////////////
void process_txt_attach( process_t * process,
                         uint32_t    txt_id )
{
    xptr_t      chdev_xp;     // extended pointer on TXT_RX chdev
    cxy_t       chdev_cxy;    // TXT_RX chdev cluster
    chdev_t *   chdev_ptr;    // local pointer on TXT_RX chdev
    xptr_t      root_xp;      // extended pointer on list root in chdev
    xptr_t      lock_xp;      // extended pointer on list lock in chdev

// check process is in owner cluster
assert( (CXY_FROM_PID( process->pid ) == local_cxy) ,
"process descriptor not in owner cluster" );

// check terminal index
assert( (txt_id < LOCAL_CLUSTER->nb_txt_channels) ,
"illegal TXT terminal index" );

    // get pointers on TXT_RX[txt_id] chdev
    chdev_xp  = chdev_dir.txt_rx[txt_id];
    chdev_cxy = GET_CXY( chdev_xp );
    chdev_ptr = GET_PTR( chdev_xp );

    // get extended pointer on root & lock of attached process list
    root_xp = XPTR( chdev_cxy , &chdev_ptr->ext.txt.root );
    lock_xp = XPTR( chdev_cxy , &chdev_ptr->ext.txt.lock );

    // get lock protecting list of processes attached to TXT
    remote_busylock_acquire( lock_xp );

    // insert process in attached process list
    xlist_add_last( root_xp , XPTR( local_cxy , &process->txt_list ) );

    // release lock protecting list of processes attached to TXT
    remote_busylock_release( lock_xp );

#if DEBUG_PROCESS_TXT
thread_t * this = CURRENT_THREAD;
uint32_t cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_TXT < cycle )
printk("\n[%s] thread[%x,%x] attached process %x to TXT %d / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, process->pid, txt_id , cycle );
#endif

} // end process_txt_attach()

/////////////////////////////////////////////
void process_txt_detach( xptr_t  process_xp )
{
    process_t * process_ptr;  // local pointer on process in owner cluster
    cxy_t       process_cxy;  // process owner cluster
    pid_t       process_pid;  // process identifier
    xptr_t      file_xp;      // extended pointer on stdin file
    xptr_t      chdev_xp;     // extended pointer on TXT_RX chdev
    cxy_t       chdev_cxy;    // TXT_RX chdev cluster
    chdev_t *   chdev_ptr;    // local pointer on TXT_RX chdev
    xptr_t      lock_xp;      // extended pointer on list lock in chdev

    // get process cluster, local pointer, and PID
    process_cxy = GET_CXY( process_xp );
    process_ptr = GET_PTR( process_xp );
    process_pid = hal_remote_l32( XPTR( process_cxy , &process_ptr->pid ) );

// check process descriptor in owner cluster
assert( (CXY_FROM_PID( process_pid ) == process_cxy ) ,
"process descriptor not in owner cluster" );

    // release TXT ownership (does nothing if not TXT owner)
    process_txt_transfer_ownership( process_xp );

    // get extended pointer on process stdin pseudo file
    file_xp = (xptr_t)hal_remote_l64( XPTR( process_cxy , &process_ptr->fd_array.array[0] ) );

    // get pointers on TXT_RX chdev
    chdev_xp  = chdev_from_file( file_xp );
    chdev_cxy = GET_CXY( chdev_xp );
    chdev_ptr = (chdev_t *)GET_PTR( chdev_xp );

    // get extended pointer on lock protecting attached process list
    lock_xp = XPTR( chdev_cxy , &chdev_ptr->ext.txt.lock );

    // get lock protecting list of processes attached to TXT
    remote_busylock_acquire( lock_xp );

    // unlink process from attached process list
    xlist_unlink( XPTR( process_cxy , &process_ptr->txt_list ) );

    // release lock protecting list of processes attached to TXT
    remote_busylock_release( lock_xp );

#if DEBUG_PROCESS_TXT
thread_t * this = CURRENT_THREAD;
uint32_t cycle  = (uint32_t)hal_get_cycles();
uint32_t txt_id = hal_remote_l32( XPTR( chdev_cxy , &chdev_ptr->channel ) );
if( DEBUG_PROCESS_TXT < cycle )
printk("\n[%s] thread[%x,%x] detached process %x from TXT%d / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, process_pid, txt_id, cycle );
#endif

} // end process_txt_detach()

///////////////////////////////////////////////////
void process_txt_set_ownership( xptr_t process_xp )
{
    process_t * process_ptr;
    cxy_t       process_cxy;
    pid_t       process_pid;
    xptr_t      file_xp;
    xptr_t      txt_xp;     
    chdev_t   * txt_ptr;
    cxy_t       txt_cxy;

    // get pointers on process in owner cluster
    process_cxy = GET_CXY( process_xp );
    process_ptr = GET_PTR( process_xp );
    process_pid = hal_remote_l32( XPTR( process_cxy , &process_ptr->pid ) );

    // check owner cluster
    assert( (process_cxy == CXY_FROM_PID( process_pid )) ,
    "process descriptor not in owner cluster" );

    // get extended pointer on stdin pseudo file
    file_xp = hal_remote_l64( XPTR( process_cxy , &process_ptr->fd_array.array[0] ) );

    // get pointers on TXT chdev
    txt_xp  = chdev_from_file( file_xp );
    txt_cxy = GET_CXY( txt_xp );
    txt_ptr = GET_PTR( txt_xp );

    // set owner field in TXT chdev
    hal_remote_s64( XPTR( txt_cxy , &txt_ptr->ext.txt.owner_xp ) , process_xp );

#if DEBUG_PROCESS_TXT
thread_t * this = CURRENT_THREAD;
uint32_t cycle  = (uint32_t)hal_get_cycles();
uint32_t txt_id = hal_remote_l32( XPTR( txt_cxy , &txt_ptr->channel ) );
if( DEBUG_PROCESS_TXT < cycle )
printk("\n[%s] thread[%x,%x] give TXT%d ownership to process %x / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, txt_id, process_pid, cycle );
#endif

}  // end process_txt_set ownership()

////////////////////////////////////////////////////////
void process_txt_transfer_ownership( xptr_t process_xp )
{
    process_t * process_ptr;     // local pointer on process releasing ownership
    cxy_t       process_cxy;     // process cluster
    pid_t       process_pid;     // process identifier
    xptr_t      file_xp;         // extended pointer on TXT_RX pseudo file
    xptr_t      txt_xp;          // extended pointer on TXT_RX chdev
    chdev_t   * txt_ptr;         // local pointer on TXT_RX chdev
    cxy_t       txt_cxy;         // cluster of TXT_RX chdev
    uint32_t    txt_id;          // TXT_RX channel 
    xptr_t      owner_xp;        // extended pointer on current TXT_RX owner
    xptr_t      root_xp;         // extended pointer on root of attached process list
    xptr_t      lock_xp;         // extended pointer on lock protecting attached process list
    xptr_t      iter_xp;         // iterator for xlist
    xptr_t      current_xp;      // extended pointer on current process
    bool_t      found;

#if DEBUG_PROCESS_TXT
thread_t * this  = CURRENT_THREAD;
uint32_t   cycle;
#endif

    // get pointers on target process 
    process_cxy = GET_CXY( process_xp );
    process_ptr = GET_PTR( process_xp );
    process_pid = hal_remote_l32( XPTR( process_cxy , &process_ptr->pid ) );

// check owner cluster
assert( (process_cxy == CXY_FROM_PID( process_pid )) ,
"process descriptor not in owner cluster" );

    // get extended pointer on stdin pseudo file
    file_xp = hal_remote_l64( XPTR( process_cxy , &process_ptr->fd_array.array[0] ) );

    // get pointers on TXT chdev
    txt_xp  = chdev_from_file( file_xp );
    txt_cxy = GET_CXY( txt_xp );
    txt_ptr = GET_PTR( txt_xp );

    // get relevant infos from chdev descriptor
    owner_xp = hal_remote_l64( XPTR( txt_cxy , &txt_ptr->ext.txt.owner_xp ) );
    txt_id   = hal_remote_l32( XPTR( txt_cxy , &txt_ptr->channel ) );

    // transfer ownership only if target process is the TXT owner
    if( (owner_xp == process_xp) && (txt_id > 0) )  
    {
        // get extended pointers on root and lock of attached processes list
        root_xp = XPTR( txt_cxy , &txt_ptr->ext.txt.root );
        lock_xp = XPTR( txt_cxy , &txt_ptr->ext.txt.lock );

        if( process_get_ppid( process_xp ) != 1 )       // target process is not KSH
        {
            // get lock
            remote_busylock_acquire( lock_xp );

            // scan attached process list to find KSH process
            found = false;
            for( iter_xp = hal_remote_l64( root_xp ) ;
                 (iter_xp != root_xp) && (found == false) ;
                 iter_xp = hal_remote_l64( iter_xp ) )
            {
                current_xp = XLIST_ELEMENT( iter_xp , process_t , txt_list );

                if( process_get_ppid( current_xp ) == 1 )  // current is KSH 
                {
                    // set owner field in TXT chdev 
                    hal_remote_s64( XPTR( txt_cxy , &txt_ptr->ext.txt.owner_xp ) , current_xp );

#if DEBUG_PROCESS_TXT
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_TXT < cycle )
printk("\n[%s] thread[%x,%x] transfered TXT%d ownership to KSH / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, txt_id, cycle );
#endif
                    found = true;
                }
            }

            // release lock
            remote_busylock_release( lock_xp );

// It must exist a KSH process for each user TXT channel
assert( (found == true), "KSH process not found for TXT%d", txt_id );

        }
        else                                           // target process is KSH
        {
            // get lock
            remote_busylock_acquire( lock_xp );

            // scan attached process list to find another process 
            found = false;
            for( iter_xp = hal_remote_l64( root_xp ) ;
                 (iter_xp != root_xp) && (found == false) ;
                 iter_xp = hal_remote_l64( iter_xp ) )
            {
                current_xp  = XLIST_ELEMENT( iter_xp , process_t , txt_list );

                if( current_xp != process_xp )            // current is not KSH 
                {
                    // set owner field in TXT chdev 
                    hal_remote_s64( XPTR( txt_cxy , &txt_ptr->ext.txt.owner_xp ) , current_xp );

#if DEBUG_PROCESS_TXT
cycle  = (uint32_t)hal_get_cycles();
cxy_t       current_cxy = GET_CXY( current_xp );
process_t * current_ptr = GET_PTR( current_xp );
uint32_t    new_pid     = hal_remote_l32( XPTR( current_cxy , &current_ptr->pid ) );
if( DEBUG_PROCESS_TXT < cycle )
printk("\n[%s] thread[%x,%x] transfered TXT%d ownership to process %x / cycle %d\n",
__FUNCTION__,this->process->pid, this->trdid, txt_id, new_pid, cycle );
#endif
                    found = true;
                }
            }

            // release lock
            remote_busylock_release( lock_xp );

            // no more owner for TXT if no other process found
            if( found == false )
            {
                // set owner field in TXT chdev 
                hal_remote_s64( XPTR( txt_cxy , &txt_ptr->ext.txt.owner_xp ) , XPTR_NULL );

#if DEBUG_PROCESS_TXT
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_TXT < cycle )
printk("\n[%s] thread[%x,%x] released TXT%d (no attached process) / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, txt_id, cycle );
#endif
            }
        }
    }
    else
    {

#if DEBUG_PROCESS_TXT
cycle = (uint32_t)hal_get_cycles();
if( DEBUG_PROCESS_TXT < cycle )
printk("\n[%s] thread[%x,%x] does nothing for process %x (not TXT owner) / cycle %d\n",
__FUNCTION__, this->process->pid, this->trdid, process_pid, cycle );
#endif

    }

}  // end process_txt_transfer_ownership()


////////////////////////////////////////////////
bool_t process_txt_is_owner( xptr_t process_xp )
{
    // get local pointer and cluster of process in owner cluster
    cxy_t       process_cxy = GET_CXY( process_xp );
    process_t * process_ptr = GET_PTR( process_xp );

// check calling thread execute in target process owner cluster
pid_t process_pid = hal_remote_l32( XPTR( process_cxy , &process_ptr->pid ) );
assert( (process_cxy == CXY_FROM_PID( process_pid )) ,
"process descriptor not in owner cluster" );

    // get extended pointer on stdin pseudo file
    xptr_t file_xp = hal_remote_l64( XPTR( process_cxy , &process_ptr->fd_array.array[0] ) );

    // get pointers on TXT chdev
    xptr_t    txt_xp  = chdev_from_file( file_xp );
    cxy_t     txt_cxy = GET_CXY( txt_xp );
    chdev_t * txt_ptr = GET_PTR( txt_xp );

    // get extended pointer on TXT_RX owner process 
    xptr_t owner_xp = hal_remote_l64( XPTR( txt_cxy , &txt_ptr->ext.txt.owner_xp ) );

    return (process_xp == owner_xp);

}   // end process_txt_is_owner()

////////////////////////////////////////////////     
xptr_t process_txt_get_owner( uint32_t channel )
{
    xptr_t      txt_rx_xp  = chdev_dir.txt_rx[channel];
    cxy_t       txt_rx_cxy = GET_CXY( txt_rx_xp );
    chdev_t *   txt_rx_ptr = GET_PTR( txt_rx_xp );

    return (xptr_t)hal_remote_l64( XPTR( txt_rx_cxy , &txt_rx_ptr->ext.txt.owner_xp ) );

}  // end process_txt_get_owner()

///////////////////////////////////////////
void process_txt_display( uint32_t txt_id )
{
    xptr_t      chdev_xp;
    cxy_t       chdev_cxy;
    chdev_t   * chdev_ptr;
    xptr_t      root_xp;
    xptr_t      lock_xp;
    xptr_t      current_xp;
    xptr_t      iter_xp;
    cxy_t       txt0_cxy;
    chdev_t   * txt0_ptr;
    xptr_t      txt0_xp;
    xptr_t      txt0_lock_xp;
    
    assert( (txt_id < LOCAL_CLUSTER->nb_txt_channels) ,
    "illegal TXT terminal index" );

    // get pointers on TXT0 chdev
    txt0_xp  = chdev_dir.txt_tx[0];
    txt0_cxy = GET_CXY( txt0_xp );
    txt0_ptr = GET_PTR( txt0_xp );

    // get extended pointer on TXT0 lock
    txt0_lock_xp = XPTR( txt0_cxy , &txt0_ptr->wait_lock );

    // get pointers on TXT_RX[txt_id] chdev
    chdev_xp  = chdev_dir.txt_rx[txt_id];
    chdev_cxy = GET_CXY( chdev_xp );
    chdev_ptr = GET_PTR( chdev_xp );

    // get extended pointer on root & lock of attached process list
    root_xp = XPTR( chdev_cxy , &chdev_ptr->ext.txt.root );
    lock_xp = XPTR( chdev_cxy , &chdev_ptr->ext.txt.lock );

    // get lock on attached process list
    remote_busylock_acquire( lock_xp );

    // get TXT0 lock in busy waiting mode
    remote_busylock_acquire( txt0_lock_xp );

    // display header
    nolock_printk("\n***** processes attached to TXT_%d / cycle %d\n",
    txt_id , (uint32_t)hal_get_cycles() );

    // scan attached process list 
    XLIST_FOREACH( root_xp , iter_xp )
    {
        current_xp  = XLIST_ELEMENT( iter_xp , process_t , txt_list );
        process_display( current_xp );
    }

    // release TXT0 lock in busy waiting mode
    remote_busylock_release( txt0_lock_xp );

    // release lock on attached process list
    remote_busylock_release( lock_xp );

}  // end process_txt_display