Non-stop mode support.

[deliverable/binutils-gdb.git] / gdb / gdbserver / linux-low.c

/* Low level interface to ptrace, for the remote server for GDB.
   Copyright (C) 1995, 1996, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005,
   2006, 2007, 2008, 2009 Free Software Foundation, Inc.

   This file is part of GDB.

   This program 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; either version 3 of the License, or
   (at your option) any later version.

   This program 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 this program.  If not, see <http://www.gnu.org/licenses/>.  */

#include "server.h"
#include "linux-low.h"

#include <sys/wait.h>
#include <stdio.h>
#include <sys/param.h>
#include <sys/ptrace.h>
#include <signal.h>
#include <sys/ioctl.h>
#include <fcntl.h>
#include <string.h>
#include <stdlib.h>
#include <unistd.h>
#include <errno.h>
#include <sys/syscall.h>
#include <sched.h>
#include <ctype.h>
#include <pwd.h>
#include <sys/types.h>
#include <dirent.h>

#ifndef PTRACE_GETSIGINFO
# define PTRACE_GETSIGINFO 0x4202
# define PTRACE_SETSIGINFO 0x4203
#endif

#ifndef O_LARGEFILE
#define O_LARGEFILE 0
#endif

/* If the system headers did not provide the constants, hard-code the normal
   values.  */
#ifndef PTRACE_EVENT_FORK

#define PTRACE_SETOPTIONS       0x4200
#define PTRACE_GETEVENTMSG      0x4201

/* options set using PTRACE_SETOPTIONS */
#define PTRACE_O_TRACESYSGOOD   0x00000001
#define PTRACE_O_TRACEFORK      0x00000002
#define PTRACE_O_TRACEVFORK     0x00000004
#define PTRACE_O_TRACECLONE     0x00000008
#define PTRACE_O_TRACEEXEC      0x00000010
#define PTRACE_O_TRACEVFORKDONE 0x00000020
#define PTRACE_O_TRACEEXIT      0x00000040

/* Wait extended result codes for the above trace options.  */
#define PTRACE_EVENT_FORK       1
#define PTRACE_EVENT_VFORK      2
#define PTRACE_EVENT_CLONE      3
#define PTRACE_EVENT_EXEC       4
#define PTRACE_EVENT_VFORK_DONE 5
#define PTRACE_EVENT_EXIT       6

#endif /* PTRACE_EVENT_FORK */

/* We can't always assume that this flag is available, but all systems
   with the ptrace event handlers also have __WALL, so it's safe to use
   in some contexts.  */
#ifndef __WALL
#define __WALL          0x40000000 /* Wait for any child.  */
#endif

#ifdef __UCLIBC__
#if !(defined(__UCLIBC_HAS_MMU__) || defined(__ARCH_HAS_MMU__))
#define HAS_NOMMU
#endif
#endif

/* ``all_threads'' is keyed by the LWP ID, which we use as the GDB protocol
   representation of the thread ID.

   ``all_lwps'' is keyed by the process ID - which on Linux is (presently)
   the same as the LWP ID.  */

struct inferior_list all_lwps;

/* A list of all unknown processes which receive stop signals.  Some other
   process will presumably claim each of these as forked children
   momentarily.  */

struct inferior_list stopped_pids;

/* FIXME this is a bit of a hack, and could be removed.  */
int stopping_threads;

/* FIXME make into a target method?  */
int using_threads = 1;
static int thread_db_active;

static int must_set_ptrace_flags;

/* This flag is true iff we've just created or attached to a new inferior
   but it has not stopped yet.  As soon as it does, we need to call the
   low target's arch_setup callback.  */
static int new_inferior;

static void linux_resume_one_lwp (struct inferior_list_entry *entry,
                                  int step, int signal, siginfo_t *info);
static void linux_resume (struct thread_resume *resume_info, size_t n);
static void stop_all_lwps (void);
static int linux_wait_for_event (int pid, int *wstat, int options);
static int check_removed_breakpoint (struct lwp_info *event_child);
static void *add_lwp (unsigned long pid);
static int my_waitpid (int pid, int *status, int flags);
static int linux_stopped_by_watchpoint (void);

struct pending_signals
{
  int signal;
  siginfo_t info;
  struct pending_signals *prev;
};

#define PTRACE_ARG3_TYPE long
#define PTRACE_XFER_TYPE long

#ifdef HAVE_LINUX_REGSETS
static char *disabled_regsets;
static int num_regsets;
#endif

/* FIXME: Delete eventually.  */
#define inferior_pid (lwpid_of (get_thread_lwp (current_inferior)))

/* The read/write ends of the pipe registered as waitable file in the
   event loop.  */
static int linux_event_pipe[2] = { -1, -1 };

/* True if we're currently in async mode.  */
#define target_is_async_p() (linux_event_pipe[0] != -1)

static void send_sigstop (struct inferior_list_entry *entry);
static void wait_for_sigstop (struct inferior_list_entry *entry);

static void
delete_lwp (struct lwp_info *lwp)
{
  remove_thread (get_lwp_thread (lwp));
  remove_inferior (&all_lwps, &lwp->head);
  free (lwp);
}

/* Handle a GNU/Linux extended wait response.  If we see a clone
   event, we need to add the new LWP to our list (and not report the
   trap to higher layers).  */

static void
handle_extended_wait (struct lwp_info *event_child, int wstat)
{
  int event = wstat >> 16;
  struct lwp_info *new_lwp;

  if (event == PTRACE_EVENT_CLONE)
    {
      unsigned long new_pid;
      int ret, status = W_STOPCODE (SIGSTOP);

      ptrace (PTRACE_GETEVENTMSG, lwpid_of (event_child), 0, &new_pid);

      /* If we haven't already seen the new PID stop, wait for it now.  */
      if (! pull_pid_from_list (&stopped_pids, new_pid))
        {
          /* The new child has a pending SIGSTOP.  We can't affect it until it
             hits the SIGSTOP, but we're already attached.  */

          ret = my_waitpid (new_pid, &status, __WALL);

          if (ret == -1)
            perror_with_name ("waiting for new child");
          else if (ret != new_pid)
            warning ("wait returned unexpected PID %d", ret);
          else if (!WIFSTOPPED (status))
            warning ("wait returned unexpected status 0x%x", status);
        }

      ptrace (PTRACE_SETOPTIONS, new_pid, 0, PTRACE_O_TRACECLONE);

      new_lwp = (struct lwp_info *) add_lwp (new_pid);
      add_thread (new_pid, new_lwp, new_pid);
      new_thread_notify (thread_id_to_gdb_id (lwpid_of (new_lwp)));

      /* Normally we will get the pending SIGSTOP.  But in some cases
         we might get another signal delivered to the group first.
         If we do get another signal, be sure not to lose it.  */
      if (WSTOPSIG (status) == SIGSTOP)
        {
          if (stopping_threads)
            new_lwp->stopped = 1;
          else
            ptrace (PTRACE_CONT, new_pid, 0, 0);
        }
      else
        {
          new_lwp->stop_expected = 1;
          if (stopping_threads)
            {
              new_lwp->stopped = 1;
              new_lwp->status_pending_p = 1;
              new_lwp->status_pending = status;
            }
          else
            /* Pass the signal on.  This is what GDB does - except
               shouldn't we really report it instead?  */
            ptrace (PTRACE_CONT, new_pid, 0, WSTOPSIG (status));
        }

      /* Always resume the current thread.  If we are stopping
         threads, it will have a pending SIGSTOP; we may as well
         collect it now.  */
      linux_resume_one_lwp (&event_child->head,
                            event_child->stepping, 0, NULL);
    }
}

/* This function should only be called if the process got a SIGTRAP.
   The SIGTRAP could mean several things.

   On i386, where decr_pc_after_break is non-zero:
   If we were single-stepping this process using PTRACE_SINGLESTEP,
   we will get only the one SIGTRAP (even if the instruction we
   stepped over was a breakpoint).  The value of $eip will be the
   next instruction.
   If we continue the process using PTRACE_CONT, we will get a
   SIGTRAP when we hit a breakpoint.  The value of $eip will be
   the instruction after the breakpoint (i.e. needs to be
   decremented).  If we report the SIGTRAP to GDB, we must also
   report the undecremented PC.  If we cancel the SIGTRAP, we
   must resume at the decremented PC.

   (Presumably, not yet tested) On a non-decr_pc_after_break machine
   with hardware or kernel single-step:
   If we single-step over a breakpoint instruction, our PC will
   point at the following instruction.  If we continue and hit a
   breakpoint instruction, our PC will point at the breakpoint
   instruction.  */

static CORE_ADDR
get_stop_pc (void)
{
  CORE_ADDR stop_pc = (*the_low_target.get_pc) ();

  if (get_thread_lwp (current_inferior)->stepping)
    return stop_pc;
  else
    return stop_pc - the_low_target.decr_pc_after_break;
}

static void *
add_lwp (unsigned long pid)
{
  struct lwp_info *lwp;

  lwp = (struct lwp_info *) xmalloc (sizeof (*lwp));
  memset (lwp, 0, sizeof (*lwp));

  lwp->head.id = pid;

  add_inferior_to_list (&all_lwps, &lwp->head);

  return lwp;
}

/* Start an inferior process and returns its pid.
   ALLARGS is a vector of program-name and args. */

static int
linux_create_inferior (char *program, char **allargs)
{
  void *new_lwp;
  int pid;

#if defined(__UCLIBC__) && defined(HAS_NOMMU)
  pid = vfork ();
#else
  pid = fork ();
#endif
  if (pid < 0)
    perror_with_name ("fork");

  if (pid == 0)
    {
      ptrace (PTRACE_TRACEME, 0, 0, 0);

      signal (__SIGRTMIN + 1, SIG_DFL);

      setpgid (0, 0);

      execv (program, allargs);
      if (errno == ENOENT)
        execvp (program, allargs);

      fprintf (stderr, "Cannot exec %s: %s.\n", program,
               strerror (errno));
      fflush (stderr);
      _exit (0177);
    }

  new_lwp = add_lwp (pid);
  add_thread (pid, new_lwp, pid);
  must_set_ptrace_flags = 1;
  new_inferior = 1;

  return pid;
}

/* Attach to an inferior process.  */

void
linux_attach_lwp (unsigned long pid)
{
  struct lwp_info *new_lwp;

  if (ptrace (PTRACE_ATTACH, pid, 0, 0) != 0)
    {
      if (all_threads.head != NULL)
        {
          /* If we fail to attach to an LWP, just warn.  */
          fprintf (stderr, "Cannot attach to lwp %ld: %s (%d)\n", pid,
                   strerror (errno), errno);
          fflush (stderr);
          return;
        }
      else
        /* If we fail to attach to a process, report an error.  */
        error ("Cannot attach to lwp %ld: %s (%d)\n", pid,
               strerror (errno), errno);
    }

  /* FIXME: This intermittently fails.
     We need to wait for SIGSTOP first.  */
  ptrace (PTRACE_SETOPTIONS, pid, 0, PTRACE_O_TRACECLONE);

  new_lwp = (struct lwp_info *) add_lwp (pid);
  add_thread (pid, new_lwp, pid);
  new_thread_notify (thread_id_to_gdb_id (lwpid_of (new_lwp)));

  /* The next time we wait for this LWP we'll see a SIGSTOP as PTRACE_ATTACH
     brings it to a halt.

     There are several cases to consider here:

     1) gdbserver has already attached to the process and is being notified
        of a new thread that is being created.
        In this case we should ignore that SIGSTOP and resume the process.
        This is handled below by setting stop_expected = 1.

     2) This is the first thread (the process thread), and we're attaching
        to it via attach_inferior.
        In this case we want the process thread to stop.
        This is handled by having linux_attach clear stop_expected after
        we return.
        ??? If the process already has several threads we leave the other
        threads running.

     3) GDB is connecting to gdbserver and is requesting an enumeration of all
        existing threads.
        In this case we want the thread to stop.
        FIXME: This case is currently not properly handled.
        We should wait for the SIGSTOP but don't.  Things work apparently
        because enough time passes between when we ptrace (ATTACH) and when
        gdb makes the next ptrace call on the thread.

     On the other hand, if we are currently trying to stop all threads, we
     should treat the new thread as if we had sent it a SIGSTOP.  This works
     because we are guaranteed that the add_lwp call above added us to the
     end of the list, and so the new thread has not yet reached
     wait_for_sigstop (but will).  */
  if (! stopping_threads)
    new_lwp->stop_expected = 1;
}

int
linux_attach (unsigned long pid)
{
  struct lwp_info *lwp;

  linux_attach_lwp (pid);

  if (!non_stop)
    {
      /* Don't ignore the initial SIGSTOP if we just attached to this
         process.  It will be collected by wait shortly.  */
      lwp = (struct lwp_info *) find_inferior_id (&all_lwps, pid);
      lwp->stop_expected = 0;
    }

  new_inferior = 1;

  return 0;
}

/* Kill the inferior process.  Make us have no inferior.  */

static void
linux_kill_one_lwp (struct inferior_list_entry *entry)
{
  struct thread_info *thread = (struct thread_info *) entry;
  struct lwp_info *lwp = get_thread_lwp (thread);
  int pid;
  int wstat;

  /* We avoid killing the first thread here, because of a Linux kernel (at
     least 2.6.0-test7 through 2.6.8-rc4) bug; if we kill the parent before
     the children get a chance to be reaped, it will remain a zombie
     forever.  */
  if (entry == all_threads.head)
    return;

  /* If we're killing a running inferior, make sure it is stopped
     first, as PTRACE_KILL will not work otherwise.  */
  if (!lwp->stopped)
    send_sigstop (&lwp->head);

  do
    {
      ptrace (PTRACE_KILL, lwpid_of (lwp), 0, 0);

      /* Make sure it died.  The loop is most likely unnecessary.  */
      pid = linux_wait_for_event (lwpid_of (lwp), &wstat, __WALL);
    } while (pid > 0 && WIFSTOPPED (wstat));
}

static void
linux_kill (void)
{
  struct thread_info *thread = (struct thread_info *) all_threads.head;
  struct lwp_info *lwp;
  int wstat;
  int pid;

  if (thread == NULL)
    return;

  for_each_inferior (&all_threads, linux_kill_one_lwp);

  /* See the comment in linux_kill_one_lwp.  We did not kill the first
     thread in the list, so do so now.  */
  lwp = get_thread_lwp (thread);

  if (debug_threads)
    fprintf (stderr, "lk_1: killing lwp %ld\n", lwpid_of (lwp));

  /* If we're killing a running inferior, make sure it is stopped
     first, as PTRACE_KILL will not work otherwise.  */
  if (!lwp->stopped)
    send_sigstop (&lwp->head);

  do
    {
      ptrace (PTRACE_KILL, lwpid_of (lwp), 0, 0);

      /* Make sure it died.  The loop is most likely unnecessary.  */
      pid = linux_wait_for_event (lwpid_of (lwp), &wstat, __WALL);
    } while (pid > 0 && WIFSTOPPED (wstat));

  delete_lwp (lwp);
  clear_inferiors ();
}

static void
linux_detach_one_lwp (struct inferior_list_entry *entry)
{
  struct thread_info *thread = (struct thread_info *) entry;
  struct lwp_info *lwp = get_thread_lwp (thread);

  /* If we're detaching from a running inferior, make sure it is
     stopped first, as PTRACE_DETACH will not work otherwise.  */
  if (!lwp->stopped)
    {
      int pid = lwpid_of (lwp);

      stopping_threads = 1;
      send_sigstop (&lwp->head);

      /* If this detects a new thread through a clone event, the new
         thread is appended to the end of the lwp list, so we'll
         eventually detach from it.  */
      wait_for_sigstop (&lwp->head);
      stopping_threads = 0;

      /* If LWP exits while we're trying to stop it, there's nothing
         left to do.  */
      lwp = (struct lwp_info *) find_inferior_id (&all_lwps, pid);
      if (lwp == NULL)
        return;
    }

  /* Make sure the process isn't stopped at a breakpoint that's
     no longer there.  */
  check_removed_breakpoint (lwp);

  /* If this process is stopped but is expecting a SIGSTOP, then make
     sure we take care of that now.  This isn't absolutely guaranteed
     to collect the SIGSTOP, but is fairly likely to.  */
  if (lwp->stop_expected)
    {
      int wstat;
      /* Clear stop_expected, so that the SIGSTOP will be reported.  */
      lwp->stop_expected = 0;
      if (lwp->stopped)
        linux_resume_one_lwp (&lwp->head, 0, 0, NULL);
      linux_wait_for_event (lwpid_of (lwp), &wstat, __WALL);
    }

  /* Flush any pending changes to the process's registers.  */
  regcache_invalidate_one ((struct inferior_list_entry *)
                           get_lwp_thread (lwp));

  /* Finally, let it resume.  */
  ptrace (PTRACE_DETACH, lwpid_of (lwp), 0, 0);

  delete_lwp (lwp);
}

static int
linux_detach (void)
{
  delete_all_breakpoints ();
  for_each_inferior (&all_threads, linux_detach_one_lwp);
  clear_inferiors ();
  return 0;
}

static void
linux_join (void)
{
  int status, ret;
  struct thread_info *thread;
  struct lwp_info *lwp;

  thread = (struct thread_info *) all_threads.head;
  lwp = get_thread_lwp (thread);

  do {
    ret = my_waitpid (lwpid_of (lwp), &status, 0);
    if (WIFEXITED (status) || WIFSIGNALED (status))
      break;
  } while (ret != -1 || errno != ECHILD);
}

/* Return nonzero if the given thread is still alive.  */
static int
linux_thread_alive (unsigned long lwpid)
{
  if (find_inferior_id (&all_threads, lwpid) != NULL)
    return 1;
  else
    return 0;
}

/* Return nonzero if this process stopped at a breakpoint which
   no longer appears to be inserted.  Also adjust the PC
   appropriately to resume where the breakpoint used to be.  */
static int
check_removed_breakpoint (struct lwp_info *event_child)
{
  CORE_ADDR stop_pc;
  struct thread_info *saved_inferior;

  if (event_child->pending_is_breakpoint == 0)
    return 0;

  if (debug_threads)
    fprintf (stderr, "Checking for breakpoint in lwp %ld.\n",
             lwpid_of (event_child));

  saved_inferior = current_inferior;
  current_inferior = get_lwp_thread (event_child);

  stop_pc = get_stop_pc ();

  /* If the PC has changed since we stopped, then we shouldn't do
     anything.  This happens if, for instance, GDB handled the
     decr_pc_after_break subtraction itself.  */
  if (stop_pc != event_child->pending_stop_pc)
    {
      if (debug_threads)
        fprintf (stderr, "Ignoring, PC was changed.  Old PC was 0x%08llx\n",
                 event_child->pending_stop_pc);

      event_child->pending_is_breakpoint = 0;
      current_inferior = saved_inferior;
      return 0;
    }

  /* If the breakpoint is still there, we will report hitting it.  */
  if ((*the_low_target.breakpoint_at) (stop_pc))
    {
      if (debug_threads)
        fprintf (stderr, "Ignoring, breakpoint is still present.\n");
      current_inferior = saved_inferior;
      return 0;
    }

  if (debug_threads)
    fprintf (stderr, "Removed breakpoint.\n");

  /* For decr_pc_after_break targets, here is where we perform the
     decrement.  We go immediately from this function to resuming,
     and can not safely call get_stop_pc () again.  */
  if (the_low_target.set_pc != NULL)
    (*the_low_target.set_pc) (stop_pc);

  /* We consumed the pending SIGTRAP.  */
  event_child->pending_is_breakpoint = 0;
  event_child->status_pending_p = 0;
  event_child->status_pending = 0;

  current_inferior = saved_inferior;
  return 1;
}

/* Return 1 if this lwp has an interesting status pending.  This
   function may silently resume an inferior lwp.  */
static int
status_pending_p (struct inferior_list_entry *entry, void *dummy)
{
  struct lwp_info *lwp = (struct lwp_info *) entry;

  if (lwp->status_pending_p && !lwp->suspended)
    if (check_removed_breakpoint (lwp))
      {
        /* This thread was stopped at a breakpoint, and the breakpoint
           is now gone.  We were told to continue (or step...) all threads,
           so GDB isn't trying to single-step past this breakpoint.
           So instead of reporting the old SIGTRAP, pretend we got to
           the breakpoint just after it was removed instead of just
           before; resume the process.  */
        linux_resume_one_lwp (&lwp->head, 0, 0, NULL);
        return 0;
      }

  return (lwp->status_pending_p && !lwp->suspended);
}

static struct lwp_info *
linux_wait_for_lwp (int pid, int *wstatp, int options)
{
  int ret;
  int to_wait_for = pid;
  struct lwp_info *child = NULL;

  if (debug_threads)
    fprintf (stderr, "linux_wait_for_lwp: %d\n", pid);

  options |= __WALL;

retry:

  ret = my_waitpid (to_wait_for, wstatp, options);
  if (ret == 0 || (ret == -1 && errno == ECHILD && (options & WNOHANG)))
    return NULL;
  else if (ret == -1)
    perror_with_name ("waitpid");

  if (debug_threads
      && (!WIFSTOPPED (*wstatp)
          || (WSTOPSIG (*wstatp) != 32
              && WSTOPSIG (*wstatp) != 33)))
    fprintf (stderr, "Got an event from %d (%x)\n", ret, *wstatp);

  child = (struct lwp_info *) find_inferior_id (&all_lwps, ret);

  /* If we didn't find a process, one of two things presumably happened:
     - A process we started and then detached from has exited.  Ignore it.
     - A process we are controlling has forked and the new child's stop
     was reported to us by the kernel.  Save its PID.  */
  if (child == NULL && WIFSTOPPED (*wstatp))
    {
      add_pid_to_list (&stopped_pids, ret);
      goto retry;
    }
  else if (child == NULL)
    goto retry;

  child->stopped = 1;
  child->pending_is_breakpoint = 0;

  child->last_status = *wstatp;

  /* Architecture-specific setup after inferior is running.
     This needs to happen after we have attached to the inferior
     and it is stopped for the first time, but before we access
     any inferior registers.  */
  if (new_inferior)
    {
      the_low_target.arch_setup ();
#ifdef HAVE_LINUX_REGSETS
      memset (disabled_regsets, 0, num_regsets);
#endif
      new_inferior = 0;
    }

  if (debug_threads
      && WIFSTOPPED (*wstatp))
    {
      struct thread_info *saved_inferior = current_inferior;
      current_inferior = (struct thread_info *)
        find_inferior_id (&all_threads, lwpid_of (child));
      /* For testing only; i386_stop_pc prints out a diagnostic.  */
      if (the_low_target.get_pc != NULL)
        get_stop_pc ();
      current_inferior = saved_inferior;
    }

  return child;
}

/* Wait for an event from child PID.  If PID is -1, wait for any
   child.  Store the stop status through the status pointer WSTAT.
   OPTIONS is passed to the waitpid call.  Return 0 if no child stop
   event was found and OPTIONS contains WNOHANG.  Return the PID of
   the stopped child otherwise.  */

static int
linux_wait_for_event (int pid, int *wstat, int options)
{
  CORE_ADDR stop_pc;
  struct lwp_info *event_child = NULL;
  int bp_status;
  struct lwp_info *requested_child = NULL;

  /* Check for a process with a pending status.  */
  /* It is possible that the user changed the pending task's registers since
     it stopped.  We correctly handle the change of PC if we hit a breakpoint
     (in check_removed_breakpoint); signals should be reported anyway.  */

  if (pid == -1)
    {
      event_child = (struct lwp_info *)
        find_inferior (&all_lwps, status_pending_p, NULL);
      if (debug_threads && event_child)
        fprintf (stderr, "Got a pending child %ld\n", lwpid_of (event_child));
    }
  else
    {
      requested_child = (struct lwp_info *)
        find_inferior_id (&all_lwps, pid);
      if (requested_child->status_pending_p
          && !check_removed_breakpoint (requested_child))
        event_child = requested_child;
    }

  if (event_child != NULL)
    {
      if (debug_threads)
        fprintf (stderr, "Got an event from pending child %ld (%04x)\n",
                 lwpid_of (event_child), event_child->status_pending);
      *wstat = event_child->status_pending;
      event_child->status_pending_p = 0;
      event_child->status_pending = 0;
      current_inferior = get_lwp_thread (event_child);
      return lwpid_of (event_child);
    }

  /* We only enter this loop if no process has a pending wait status.  Thus
     any action taken in response to a wait status inside this loop is
     responding as soon as we detect the status, not after any pending
     events.  */
  while (1)
    {
      event_child = linux_wait_for_lwp (pid, wstat, options);

      if ((options & WNOHANG) && event_child == NULL)
        return 0;

      if (event_child == NULL)
        error ("event from unknown child");

      current_inferior = get_lwp_thread (event_child);

      /* Check for thread exit.  */
      if (! WIFSTOPPED (*wstat))
        {
          int lwpid = lwpid_of (event_child);
          if (debug_threads)
            fprintf (stderr, "LWP %d exiting\n", lwpid);

          /* If the last thread is exiting, just return.  */
          if (all_threads.head == all_threads.tail)
            {
              if (debug_threads)
                fprintf (stderr, "LWP %d is last lwp of process\n", lwpid);
              return lwpid_of (event_child);
            }

          dead_thread_notify (thread_id_to_gdb_id (lwpid_of (event_child)));
          delete_lwp (event_child);

          if (!non_stop)
            {
              current_inferior = (struct thread_info *) all_threads.head;
              if (debug_threads)
                fprintf (stderr, "Current inferior is now %ld\n",
                         lwpid_of (get_thread_lwp (current_inferior)));
            }
          else
            {
              current_inferior = NULL;
              if (debug_threads)
                fprintf (stderr, "Current inferior is now <NULL>\n");
            }

          /* If we were waiting for this particular child to do something...
             well, it did something.  */
          if (requested_child != NULL)
            return lwpid;

          /* Wait for a more interesting event.  */
          continue;
        }

      if (WIFSTOPPED (*wstat)
          && WSTOPSIG (*wstat) == SIGSTOP
          && event_child->stop_expected)
        {
          if (debug_threads)
            fprintf (stderr, "Expected stop.\n");
          event_child->stop_expected = 0;
          linux_resume_one_lwp (&event_child->head,
                                event_child->stepping, 0, NULL);
          continue;
        }

      if (WIFSTOPPED (*wstat) && WSTOPSIG (*wstat) == SIGTRAP
          && *wstat >> 16 != 0)
        {
          handle_extended_wait (event_child, *wstat);
          continue;
        }

      /* If GDB is not interested in this signal, don't stop other
         threads, and don't report it to GDB.  Just resume the
         inferior right away.  We do this for threading-related
         signals as well as any that GDB specifically requested we
         ignore.  But never ignore SIGSTOP if we sent it ourselves,
         and do not ignore signals when stepping - they may require
         special handling to skip the signal handler.  */
      /* FIXME drow/2002-06-09: Get signal numbers from the inferior's
         thread library?  */
      if (WIFSTOPPED (*wstat)
          && !event_child->stepping
          && (
#ifdef USE_THREAD_DB
              (thread_db_active
               && (WSTOPSIG (*wstat) == __SIGRTMIN
                   || WSTOPSIG (*wstat) == __SIGRTMIN + 1))
              ||
#endif
              (pass_signals[target_signal_from_host (WSTOPSIG (*wstat))]
               && (WSTOPSIG (*wstat) != SIGSTOP || !stopping_threads))))
        {
          siginfo_t info, *info_p;

          if (debug_threads)
            fprintf (stderr, "Ignored signal %d for LWP %ld.\n",
                     WSTOPSIG (*wstat), lwpid_of (event_child));

          if (ptrace (PTRACE_GETSIGINFO, lwpid_of (event_child), 0, &info) == 0)
            info_p = &info;
          else
            info_p = NULL;
          linux_resume_one_lwp (&event_child->head,
                                event_child->stepping,
                                WSTOPSIG (*wstat), info_p);
          continue;
        }

      /* If this event was not handled above, and is not a SIGTRAP, report
         it.  */
      if (!WIFSTOPPED (*wstat) || WSTOPSIG (*wstat) != SIGTRAP)
        return lwpid_of (event_child);

      /* If this target does not support breakpoints, we simply report the
         SIGTRAP; it's of no concern to us.  */
      if (the_low_target.get_pc == NULL)
        return lwpid_of (event_child);

      stop_pc = get_stop_pc ();

      /* bp_reinsert will only be set if we were single-stepping.
         Notice that we will resume the process after hitting
         a gdbserver breakpoint; single-stepping to/over one
         is not supported (yet).  */
      if (event_child->bp_reinsert != 0)
        {
          if (debug_threads)
            fprintf (stderr, "Reinserted breakpoint.\n");
          reinsert_breakpoint (event_child->bp_reinsert);
          event_child->bp_reinsert = 0;

          /* Clear the single-stepping flag and SIGTRAP as we resume.  */
          linux_resume_one_lwp (&event_child->head, 0, 0, NULL);
          continue;
        }

      bp_status = check_breakpoints (stop_pc);

      if (bp_status != 0)
        {
          if (debug_threads)
            fprintf (stderr, "Hit a gdbserver breakpoint.\n");

          /* We hit one of our own breakpoints.  We mark it as a pending
             breakpoint, so that check_removed_breakpoint () will do the PC
             adjustment for us at the appropriate time.  */
          event_child->pending_is_breakpoint = 1;
          event_child->pending_stop_pc = stop_pc;

          /* We may need to put the breakpoint back.  We continue in the event
             loop instead of simply replacing the breakpoint right away,
             in order to not lose signals sent to the thread that hit the
             breakpoint.  Unfortunately this increases the window where another
             thread could sneak past the removed breakpoint.  For the current
             use of server-side breakpoints (thread creation) this is
             acceptable; but it needs to be considered before this breakpoint
             mechanism can be used in more general ways.  For some breakpoints
             it may be necessary to stop all other threads, but that should
             be avoided where possible.

             If breakpoint_reinsert_addr is NULL, that means that we can
             use PTRACE_SINGLESTEP on this platform.  Uninsert the breakpoint,
             mark it for reinsertion, and single-step.

             Otherwise, call the target function to figure out where we need
             our temporary breakpoint, create it, and continue executing this
             process.  */

          /* NOTE: we're lifting breakpoints in non-stop mode.  This
             is currently only used for thread event breakpoints, so
             it isn't that bad as long as we have PTRACE_EVENT_CLONE
             events.  */
          if (bp_status == 2)
            /* No need to reinsert.  */
            linux_resume_one_lwp (&event_child->head, 0, 0, NULL);
          else if (the_low_target.breakpoint_reinsert_addr == NULL)
            {
              event_child->bp_reinsert = stop_pc;
              uninsert_breakpoint (stop_pc);
              linux_resume_one_lwp (&event_child->head, 1, 0, NULL);
            }
          else
            {
              reinsert_breakpoint_by_bp
                (stop_pc, (*the_low_target.breakpoint_reinsert_addr) ());
              linux_resume_one_lwp (&event_child->head, 0, 0, NULL);
            }

          continue;
        }

      if (debug_threads)
        fprintf (stderr, "Hit a non-gdbserver breakpoint.\n");

      /* If we were single-stepping, we definitely want to report the
         SIGTRAP.  Although the single-step operation has completed,
         do not clear clear the stepping flag yet; we need to check it
         in wait_for_sigstop.  */
      if (event_child->stepping)
        return lwpid_of (event_child);

      /* A SIGTRAP that we can't explain.  It may have been a breakpoint.
         Check if it is a breakpoint, and if so mark the process information
         accordingly.  This will handle both the necessary fiddling with the
         PC on decr_pc_after_break targets and suppressing extra threads
         hitting a breakpoint if two hit it at once and then GDB removes it
         after the first is reported.  Arguably it would be better to report
         multiple threads hitting breakpoints simultaneously, but the current
         remote protocol does not allow this.  */
      if ((*the_low_target.breakpoint_at) (stop_pc))
        {
          event_child->pending_is_breakpoint = 1;
          event_child->pending_stop_pc = stop_pc;
        }

      return lwpid_of (event_child);
    }

  /* NOTREACHED */
  return 0;
}

/* Wait for process, returns status.  */

static unsigned long
linux_wait_1 (struct target_waitstatus *ourstatus, int target_options)
{
  int w;
  struct thread_info *thread = NULL;
  struct lwp_info *lwp = NULL;
  int options;
  int wait_pid = -1;
  int pid;

  /* Translate generic target options into linux options.  */
  options = __WALL;
  if (target_options & TARGET_WNOHANG)
    options |= WNOHANG;

retry:
  ourstatus->kind = TARGET_WAITKIND_IGNORE;

  /* If we were only supposed to resume one thread, only wait for
     that thread - if it's still alive.  If it died, however - which
     can happen if we're coming from the thread death case below -
     then we need to make sure we restart the other threads.  We could
     pick a thread at random or restart all; restarting all is less
     arbitrary.  */
  if (!non_stop && cont_thread != 0 && cont_thread != -1)
    {
      thread = (struct thread_info *) find_inferior_id (&all_threads,
                                                        cont_thread);

      /* No stepping, no signal - unless one is pending already, of course.  */
      if (thread == NULL)
        {
          struct thread_resume resume_info;
          resume_info.thread = -1;
          resume_info.kind = resume_continue;
          resume_info.sig = 0;
          linux_resume (&resume_info, 1);
        }
      else
        wait_pid = cont_thread;
    }

  pid = linux_wait_for_event (wait_pid, &w, options);
  if (pid == 0) /* only if TARGET_WNOHANG */
    return pid;

  lwp = get_thread_lwp (current_inferior);

  if (must_set_ptrace_flags)
    {
      ptrace (PTRACE_SETOPTIONS, lwpid_of (lwp), 0, PTRACE_O_TRACECLONE);
      must_set_ptrace_flags = 0;
    }
  /* If we are waiting for a particular child, and it exited,
     linux_wait_for_event will return its exit status.  Similarly if
     the last child exited.  If this is not the last child, however,
     do not report it as exited until there is a 'thread exited' response
     available in the remote protocol.  Instead, just wait for another event.
     This should be safe, because if the thread crashed we will already
     have reported the termination signal to GDB; that should stop any
     in-progress stepping operations, etc.

     Report the exit status of the last thread to exit.  This matches
     LinuxThreads' behavior.  */

  if (all_threads.head == all_threads.tail)
    {
      if (WIFEXITED (w) || WIFSIGNALED (w))
        {
          int pid;

          pid = pid_of (lwp);

          delete_lwp (lwp);
          clear_inferiors ();

          current_inferior = NULL;

          if (WIFEXITED (w))
            {
              ourstatus->kind = TARGET_WAITKIND_EXITED;
              ourstatus->value.integer = WEXITSTATUS (w);

              if (debug_threads)
                fprintf (stderr, "\nChild exited with retcode = %x \n", WEXITSTATUS (w));
            }
          else
            {
              ourstatus->kind = TARGET_WAITKIND_SIGNALLED;
              ourstatus->value.sig = target_signal_from_host (WTERMSIG (w));

              if (debug_threads)
                fprintf (stderr, "\nChild terminated with signal = %x \n", WTERMSIG (w));

            }

          return pid;
        }
    }
  else
    {
      if (!WIFSTOPPED (w))
        goto retry;
    }

  /* In all-stop, stop all threads.  Be careful to only do this if
     we're about to report an event to GDB.  */
  if (!non_stop)
    stop_all_lwps ();

  ourstatus->kind = TARGET_WAITKIND_STOPPED;

  if (lwp->suspended && WSTOPSIG (w) == SIGSTOP)
    {
      /* A thread that has been requested to stop by GDB with vCont;t,
         and it stopped cleanly, so report as SIG0.  The use of
         SIGSTOP is an implementation detail.  */
      ourstatus->value.sig = TARGET_SIGNAL_0;
    }
  else if (lwp->suspended && WSTOPSIG (w) != SIGSTOP)
    {
      /* A thread that has been requested to stop by GDB with vCont;t,
         but, it stopped for other reasons.  Set stop_expected so the
         pending SIGSTOP is ignored and the LWP is resumed.  */
      lwp->stop_expected = 1;
      ourstatus->value.sig = target_signal_from_host (WSTOPSIG (w));
    }
  else
    {
      ourstatus->value.sig = target_signal_from_host (WSTOPSIG (w));
    }

  if (debug_threads)
    fprintf (stderr, "linux_wait ret = %ld, %d, %d\n",
             lwpid_of (lwp),
             ourstatus->kind,
             ourstatus->value.sig);

  return lwpid_of (lwp);
}

/* Get rid of any pending event in the pipe.  */
static void
async_file_flush (void)
{
  int ret;
  char buf;

  do
    ret = read (linux_event_pipe[0], &buf, 1);
  while (ret >= 0 || (ret == -1 && errno == EINTR));
}

/* Put something in the pipe, so the event loop wakes up.  */
static void
async_file_mark (void)
{
  int ret;

  async_file_flush ();

  do
    ret = write (linux_event_pipe[1], "+", 1);
  while (ret == 0 || (ret == -1 && errno == EINTR));

  /* Ignore EAGAIN.  If the pipe is full, the event loop will already
     be awakened anyway.  */
}

static unsigned long
linux_wait (struct target_waitstatus *ourstatus, int target_options)
{
  unsigned long event_ptid;

  if (debug_threads)
    fprintf (stderr, "linux_wait\n");

  /* Flush the async file first.  */
  if (target_is_async_p ())
    async_file_flush ();

  event_ptid = linux_wait_1 (ourstatus, target_options);

  /* If at least one stop was reported, there may be more.  A single
     SIGCHLD can signal more than one child stop.  */
  if (target_is_async_p ()
      && (target_options & TARGET_WNOHANG) != 0
      && event_ptid != 0)
    async_file_mark ();

  return event_ptid;
}

/* Send a signal to an LWP.  For LinuxThreads, kill is enough; however, if
   thread groups are in use, we need to use tkill.  */

static int
kill_lwp (unsigned long lwpid, int signo)
{
  static int tkill_failed;

  errno = 0;

#ifdef SYS_tkill
  if (!tkill_failed)
    {
      int ret = syscall (SYS_tkill, lwpid, signo);
      if (errno != ENOSYS)
        return ret;
      errno = 0;
      tkill_failed = 1;
    }
#endif

  return kill (lwpid, signo);
}

static void
send_sigstop (struct inferior_list_entry *entry)
{
  struct lwp_info *lwp = (struct lwp_info *) entry;
  int pid;

  if (lwp->stopped)
    return;

  pid = lwpid_of (lwp);

  /* If we already have a pending stop signal for this process, don't
     send another.  */
  if (lwp->stop_expected)
    {
      if (debug_threads)
        fprintf (stderr, "Have pending sigstop for lwp %d\n", pid);

      /* We clear the stop_expected flag so that wait_for_sigstop
         will receive the SIGSTOP event (instead of silently resuming and
         waiting again).  It'll be reset below.  */
      lwp->stop_expected = 0;
      return;
    }

  if (debug_threads)
    fprintf (stderr, "Sending sigstop to lwp %d\n", pid);

  kill_lwp (pid, SIGSTOP);
}

static void
wait_for_sigstop (struct inferior_list_entry *entry)
{
  struct lwp_info *lwp = (struct lwp_info *) entry;
  struct thread_info *saved_inferior;
  int wstat;
  unsigned long saved_tid;
  unsigned long ptid;

  if (lwp->stopped)
    return;

  saved_inferior = current_inferior;
  if (saved_inferior != NULL)
    saved_tid = ((struct inferior_list_entry *) saved_inferior)->id;
  else
    saved_tid = 0; /* avoid bogus unused warning */

  ptid = lwpid_of (lwp);

  linux_wait_for_event (ptid, &wstat, __WALL);

  /* If we stopped with a non-SIGSTOP signal, save it for later
     and record the pending SIGSTOP.  If the process exited, just
     return.  */
  if (WIFSTOPPED (wstat)
      && WSTOPSIG (wstat) != SIGSTOP)
    {
      if (debug_threads)
        fprintf (stderr, "LWP %ld stopped with non-sigstop status %06x\n",
                 lwpid_of (lwp), wstat);

      /* Do not leave a pending single-step finish to be reported to
         the client.  The client will give us a new action for this
         thread, possibly a continue request --- otherwise, the client
         would consider this pending SIGTRAP reported later a spurious
         signal.  */
      if (WSTOPSIG (wstat) == SIGTRAP
          && lwp->stepping
          && !linux_stopped_by_watchpoint ())
        {
          if (debug_threads)
            fprintf (stderr, "  single-step SIGTRAP ignored\n");
        }
      else
        {
          lwp->status_pending_p = 1;
          lwp->status_pending = wstat;
        }
      lwp->stop_expected = 1;
    }

  if (saved_inferior == NULL || linux_thread_alive (saved_tid))
    current_inferior = saved_inferior;
  else
    {
      if (debug_threads)
        fprintf (stderr, "Previously current thread died.\n");

      if (non_stop)
        {
          /* We can't change the current inferior behind GDB's back,
             otherwise, a subsequent command may apply to the wrong
             process.  */
          current_inferior = NULL;
        }
      else
        {
          /* Set a valid thread as current.  */
          set_desired_inferior (0);
        }
    }
}

static void
stop_all_lwps (void)
{
  stopping_threads = 1;
  for_each_inferior (&all_lwps, send_sigstop);
  for_each_inferior (&all_lwps, wait_for_sigstop);
  stopping_threads = 0;
}

/* Resume execution of the inferior process.
   If STEP is nonzero, single-step it.
   If SIGNAL is nonzero, give it that signal.  */

static void
linux_resume_one_lwp (struct inferior_list_entry *entry,
                      int step, int signal, siginfo_t *info)
{
  struct lwp_info *lwp = (struct lwp_info *) entry;
  struct thread_info *saved_inferior;

  if (lwp->stopped == 0)
    return;

  /* If we have pending signals or status, and a new signal, enqueue the
     signal.  Also enqueue the signal if we are waiting to reinsert a
     breakpoint; it will be picked up again below.  */
  if (signal != 0
      && (lwp->status_pending_p || lwp->pending_signals != NULL
          || lwp->bp_reinsert != 0))
    {
      struct pending_signals *p_sig;
      p_sig = xmalloc (sizeof (*p_sig));
      p_sig->prev = lwp->pending_signals;
      p_sig->signal = signal;
      if (info == NULL)
        memset (&p_sig->info, 0, sizeof (siginfo_t));
      else
        memcpy (&p_sig->info, info, sizeof (siginfo_t));
      lwp->pending_signals = p_sig;
    }

  if (lwp->status_pending_p && !check_removed_breakpoint (lwp))
    return;

  saved_inferior = current_inferior;
  current_inferior = get_lwp_thread (lwp);

  if (debug_threads)
    fprintf (stderr, "Resuming lwp %ld (%s, signal %d, stop %s)\n",
             lwpid_of (lwp), step ? "step" : "continue", signal,
             lwp->stop_expected ? "expected" : "not expected");

  /* This bit needs some thinking about.  If we get a signal that
     we must report while a single-step reinsert is still pending,
     we often end up resuming the thread.  It might be better to
     (ew) allow a stack of pending events; then we could be sure that
     the reinsert happened right away and not lose any signals.

     Making this stack would also shrink the window in which breakpoints are
     uninserted (see comment in linux_wait_for_lwp) but not enough for
     complete correctness, so it won't solve that problem.  It may be
     worthwhile just to solve this one, however.  */
  if (lwp->bp_reinsert != 0)
    {
      if (debug_threads)
        fprintf (stderr, "  pending reinsert at %08lx", (long)lwp->bp_reinsert);
      if (step == 0)
        fprintf (stderr, "BAD - reinserting but not stepping.\n");
      step = 1;

      /* Postpone any pending signal.  It was enqueued above.  */
      signal = 0;
    }

  check_removed_breakpoint (lwp);

  if (debug_threads && the_low_target.get_pc != NULL)
    {
      fprintf (stderr, "  ");
      (*the_low_target.get_pc) ();
    }

  /* If we have pending signals, consume one unless we are trying to reinsert
     a breakpoint.  */
  if (lwp->pending_signals != NULL && lwp->bp_reinsert == 0)
    {
      struct pending_signals **p_sig;

      p_sig = &lwp->pending_signals;
      while ((*p_sig)->prev != NULL)
        p_sig = &(*p_sig)->prev;

      signal = (*p_sig)->signal;
      if ((*p_sig)->info.si_signo != 0)
        ptrace (PTRACE_SETSIGINFO, lwpid_of (lwp), 0, &(*p_sig)->info);

      free (*p_sig);
      *p_sig = NULL;
    }

  regcache_invalidate_one ((struct inferior_list_entry *)
                           get_lwp_thread (lwp));
  errno = 0;
  lwp->stopped = 0;
  lwp->stepping = step;
  ptrace (step ? PTRACE_SINGLESTEP : PTRACE_CONT, lwpid_of (lwp), 0, signal);

  current_inferior = saved_inferior;
  if (errno)
    {
      /* ESRCH from ptrace either means that the thread was already
         running (an error) or that it is gone (a race condition).  If
         it's gone, we will get a notification the next time we wait,
         so we can ignore the error.  We could differentiate these
         two, but it's tricky without waiting; the thread still exists
         as a zombie, so sending it signal 0 would succeed.  So just
         ignore ESRCH.  */
      if (errno == ESRCH)
        return;

      perror_with_name ("ptrace");
    }
}

struct thread_resume_array
{
  struct thread_resume *resume;
  size_t n;
};

/* This function is called once per thread.  We look up the thread
   in RESUME_PTR, and mark the thread with a pointer to the appropriate
   resume request.

   This algorithm is O(threads * resume elements), but resume elements
   is small (and will remain small at least until GDB supports thread
   suspension).  */
static int
linux_set_resume_request (struct inferior_list_entry *entry, void *arg)
{
  struct lwp_info *lwp;
  struct thread_info *thread;
  int ndx;
  struct thread_resume_array *r;

  thread = (struct thread_info *) entry;
  lwp = get_thread_lwp (thread);
  r = arg;

  for (ndx = 0; ndx < r->n; ndx++)
    if (r->resume[ndx].thread == -1 || r->resume[ndx].thread == entry->id)
      {
        lwp->resume = &r->resume[ndx];
        return 0;
      }

  /* No resume action for this thread.  */
  lwp->resume = NULL;

  return 0;
}


/* Set *FLAG_P if this lwp has an interesting status pending.  */
static int
resume_status_pending_p (struct inferior_list_entry *entry, void *flag_p)
{
  struct lwp_info *lwp = (struct lwp_info *) entry;

  /* LWPs which will not be resumed are not interesting, because
     we might not wait for them next time through linux_wait.  */
  if (lwp->resume == NULL)
    return 0;

  /* If this thread has a removed breakpoint, we won't have any
     events to report later, so check now.  check_removed_breakpoint
     may clear status_pending_p.  We avoid calling check_removed_breakpoint
     for any thread that we are not otherwise going to resume - this
     lets us preserve stopped status when two threads hit a breakpoint.
     GDB removes the breakpoint to single-step a particular thread
     past it, then re-inserts it and resumes all threads.  We want
     to report the second thread without resuming it in the interim.  */
  if (lwp->status_pending_p)
    check_removed_breakpoint (lwp);

  if (lwp->status_pending_p)
    * (int *) flag_p = 1;

  return 0;
}

/* This function is called once per thread.  We check the thread's resume
   request, which will tell us whether to resume, step, or leave the thread
   stopped; and what signal, if any, it should be sent.

   For threads which we aren't explicitly told otherwise, we preserve
   the stepping flag; this is used for stepping over gdbserver-placed
   breakpoints.

   If pending_flags was set in any thread, we queue any needed
   signals, since we won't actually resume.  We already have a pending
   event to report, so we don't need to preserve any step requests;
   they should be re-issued if necessary.  */

static int
linux_resume_one_thread (struct inferior_list_entry *entry, void *arg)
{
  struct lwp_info *lwp;
  struct thread_info *thread;
  int step;
  int pending_flag = * (int *) arg;

  thread = (struct thread_info *) entry;
  lwp = get_thread_lwp (thread);

  if (lwp->resume == NULL)
    return 0;

  if (lwp->resume->kind == resume_stop)
    {
      if (debug_threads)
        fprintf (stderr, "suspending LWP %ld\n", lwpid_of (lwp));

      if (!lwp->stopped)
        {
          if (debug_threads)
            fprintf (stderr, "running -> suspending %ld\n", lwpid_of (lwp));

          lwp->suspended = 1;
          send_sigstop (&lwp->head);
        }
      else
        {
          if (debug_threads)
            {
              if (lwp->suspended)
                fprintf (stderr, "already stopped/suspended LWP %ld\n",
                         lwpid_of (lwp));
              else
                fprintf (stderr, "already stopped/not suspended LWP %ld\n",
                         lwpid_of (lwp));
            }

          /* Make sure we leave the LWP suspended, so we don't try to
             resume it without GDB telling us to.  FIXME: The LWP may
             have been stopped in an internal event that was not meant
             to be notified back to GDB (e.g., gdbserver breakpoint),
             so we should be reporting a stop event in that case
             too.  */
          lwp->suspended = 1;
        }

      /* For stop requests, we're done.  */
      lwp->resume = NULL;
      return 0;
    }
  else
    lwp->suspended = 0;

  /* If this thread which is about to be resumed has a pending status,
     then don't resume any threads - we can just report the pending
     status.  Make sure to queue any signals that would otherwise be
     sent.  In all-stop mode, we do this decision based on if *any*
     thread has a pending status.  */
  if (non_stop)
    resume_status_pending_p (&lwp->head, &pending_flag);

  if (!pending_flag)
    {
      if (debug_threads)
        fprintf (stderr, "resuming LWP %ld\n", lwpid_of (lwp));

      if (lwp->resume->thread == -1
          && lwp->stepping
          && lwp->pending_is_breakpoint)
        step = 1;
      else
        step = (lwp->resume->kind == resume_step);

      linux_resume_one_lwp (&lwp->head, step, lwp->resume->sig, NULL);
    }
  else
    {
      if (debug_threads)
        fprintf (stderr, "leaving LWP %ld stopped\n", lwpid_of (lwp));

      /* If we have a new signal, enqueue the signal.  */
      if (lwp->resume->sig != 0)
        {
          struct pending_signals *p_sig;
          p_sig = xmalloc (sizeof (*p_sig));
          p_sig->prev = lwp->pending_signals;
          p_sig->signal = lwp->resume->sig;
          memset (&p_sig->info, 0, sizeof (siginfo_t));

          /* If this is the same signal we were previously stopped by,
             make sure to queue its siginfo.  We can ignore the return
             value of ptrace; if it fails, we'll skip
             PTRACE_SETSIGINFO.  */
          if (WIFSTOPPED (lwp->last_status)
              && WSTOPSIG (lwp->last_status) == lwp->resume->sig)
            ptrace (PTRACE_GETSIGINFO, lwpid_of (lwp), 0, &p_sig->info);

          lwp->pending_signals = p_sig;
        }
    }

  lwp->resume = NULL;
  return 0;
}

static void
linux_resume (struct thread_resume *resume_info, size_t n)
{
  int pending_flag;
  struct thread_resume_array array = { resume_info, n };

  find_inferior (&all_threads, linux_set_resume_request, &array);

  /* If there is a thread which would otherwise be resumed, which
     has a pending status, then don't resume any threads - we can just
     report the pending status.  Make sure to queue any signals
     that would otherwise be sent.  In non-stop mode, we'll apply this
     logic to each thread individually.  */
  pending_flag = 0;
  if (!non_stop)
    find_inferior (&all_lwps, resume_status_pending_p, &pending_flag);

  if (debug_threads)
    {
      if (pending_flag)
        fprintf (stderr, "Not resuming, pending status\n");
      else
        fprintf (stderr, "Resuming, no pending status\n");
    }

  find_inferior (&all_threads, linux_resume_one_thread, &pending_flag);
}

#ifdef HAVE_LINUX_USRREGS

int
register_addr (int regnum)
{
  int addr;

  if (regnum < 0 || regnum >= the_low_target.num_regs)
    error ("Invalid register number %d.", regnum);

  addr = the_low_target.regmap[regnum];

  return addr;
}

/* Fetch one register.  */
static void
fetch_register (int regno)
{
  CORE_ADDR regaddr;
  int i, size;
  char *buf;

  if (regno >= the_low_target.num_regs)
    return;
  if ((*the_low_target.cannot_fetch_register) (regno))
    return;

  regaddr = register_addr (regno);
  if (regaddr == -1)
    return;
  size = ((register_size (regno) + sizeof (PTRACE_XFER_TYPE) - 1)
          & - sizeof (PTRACE_XFER_TYPE));
  buf = alloca (size);
  for (i = 0; i < size; i += sizeof (PTRACE_XFER_TYPE))
    {
      errno = 0;
      *(PTRACE_XFER_TYPE *) (buf + i) =
        ptrace (PTRACE_PEEKUSER, inferior_pid, (PTRACE_ARG3_TYPE) regaddr, 0);
      regaddr += sizeof (PTRACE_XFER_TYPE);
      if (errno != 0)
        {
          /* Warning, not error, in case we are attached; sometimes the
             kernel doesn't let us at the registers.  */
          char *err = strerror (errno);
          char *msg = alloca (strlen (err) + 128);
          sprintf (msg, "reading register %d: %s", regno, err);
          error (msg);
          goto error_exit;
        }
    }

  if (the_low_target.supply_ptrace_register)
    the_low_target.supply_ptrace_register (regno, buf);
  else
    supply_register (regno, buf);

error_exit:;
}

/* Fetch all registers, or just one, from the child process.  */
static void
usr_fetch_inferior_registers (int regno)
{
  if (regno == -1 || regno == 0)
    for (regno = 0; regno < the_low_target.num_regs; regno++)
      fetch_register (regno);
  else
    fetch_register (regno);
}

/* Store our register values back into the inferior.
   If REGNO is -1, do this for all registers.
   Otherwise, REGNO specifies which register (so we can save time).  */
static void
usr_store_inferior_registers (int regno)
{
  CORE_ADDR regaddr;
  int i, size;
  char *buf;

  if (regno >= 0)
    {
      if (regno >= the_low_target.num_regs)
        return;

      if ((*the_low_target.cannot_store_register) (regno) == 1)
        return;

      regaddr = register_addr (regno);
      if (regaddr == -1)
        return;
      errno = 0;
      size = (register_size (regno) + sizeof (PTRACE_XFER_TYPE) - 1)
             & - sizeof (PTRACE_XFER_TYPE);
      buf = alloca (size);
      memset (buf, 0, size);

      if (the_low_target.collect_ptrace_register)
        the_low_target.collect_ptrace_register (regno, buf);
      else
        collect_register (regno, buf);

      for (i = 0; i < size; i += sizeof (PTRACE_XFER_TYPE))
        {
          errno = 0;
          ptrace (PTRACE_POKEUSER, inferior_pid, (PTRACE_ARG3_TYPE) regaddr,
                  *(PTRACE_XFER_TYPE *) (buf + i));
          if (errno != 0)
            {
              /* At this point, ESRCH should mean the process is
                 already gone, in which case we simply ignore attempts
                 to change its registers.  See also the related
                 comment in linux_resume_one_lwp.  */
              if (errno == ESRCH)
                return;

              if ((*the_low_target.cannot_store_register) (regno) == 0)
                {
                  char *err = strerror (errno);
                  char *msg = alloca (strlen (err) + 128);
                  sprintf (msg, "writing register %d: %s",
                           regno, err);
                  error (msg);
                  return;
                }
            }
          regaddr += sizeof (PTRACE_XFER_TYPE);
        }
    }
  else
    for (regno = 0; regno < the_low_target.num_regs; regno++)
      usr_store_inferior_registers (regno);
}
#endif /* HAVE_LINUX_USRREGS */



#ifdef HAVE_LINUX_REGSETS

static int
regsets_fetch_inferior_registers ()
{
  struct regset_info *regset;
  int saw_general_regs = 0;

  regset = target_regsets;

  while (regset->size >= 0)
    {
      void *buf;
      int res;

      if (regset->size == 0 || disabled_regsets[regset - target_regsets])
        {
          regset ++;
          continue;
        }

      buf = xmalloc (regset->size);
#ifndef __sparc__
      res = ptrace (regset->get_request, inferior_pid, 0, buf);
#else
      res = ptrace (regset->get_request, inferior_pid, buf, 0);
#endif
      if (res < 0)
        {
          if (errno == EIO)
            {
              /* If we get EIO on a regset, do not try it again for
                 this process.  */
              disabled_regsets[regset - target_regsets] = 1;
              continue;
            }
          else
            {
              char s[256];
              sprintf (s, "ptrace(regsets_fetch_inferior_registers) PID=%ld",
                       inferior_pid);
              perror (s);
            }
        }
      else if (regset->type == GENERAL_REGS)
        saw_general_regs = 1;
      regset->store_function (buf);
      regset ++;
    }
  if (saw_general_regs)
    return 0;
  else
    return 1;
}

static int
regsets_store_inferior_registers ()
{
  struct regset_info *regset;
  int saw_general_regs = 0;

  regset = target_regsets;

  while (regset->size >= 0)
    {
      void *buf;
      int res;

      if (regset->size == 0 || disabled_regsets[regset - target_regsets])
        {
          regset ++;
          continue;
        }

      buf = xmalloc (regset->size);

      /* First fill the buffer with the current register set contents,
         in case there are any items in the kernel's regset that are
         not in gdbserver's regcache.  */
#ifndef __sparc__
      res = ptrace (regset->get_request, inferior_pid, 0, buf);
#else
      res = ptrace (regset->get_request, inferior_pid, buf, 0);
#endif

      if (res == 0)
        {
          /* Then overlay our cached registers on that.  */
          regset->fill_function (buf);

          /* Only now do we write the register set.  */
#ifndef __sparc__
          res = ptrace (regset->set_request, inferior_pid, 0, buf);
#else
          res = ptrace (regset->set_request, inferior_pid, buf, 0);
#endif
        }

      if (res < 0)
        {
          if (errno == EIO)
            {
              /* If we get EIO on a regset, do not try it again for
                 this process.  */
              disabled_regsets[regset - target_regsets] = 1;
              continue;
            }
          else if (errno == ESRCH)
            {
              /* At this point, ESRCH should mean the process is
                 already gone, in which case we simply ignore attempts
                 to change its registers.  See also the related
                 comment in linux_resume_one_lwp.  */
              return 0;
            }
          else
            {
              perror ("Warning: ptrace(regsets_store_inferior_registers)");
            }
        }
      else if (regset->type == GENERAL_REGS)
        saw_general_regs = 1;
      regset ++;
      free (buf);
    }
  if (saw_general_regs)
    return 0;
  else
    return 1;
  return 0;
}

#endif /* HAVE_LINUX_REGSETS */


void
linux_fetch_registers (int regno)
{
#ifdef HAVE_LINUX_REGSETS
  if (regsets_fetch_inferior_registers () == 0)
    return;
#endif
#ifdef HAVE_LINUX_USRREGS
  usr_fetch_inferior_registers (regno);
#endif
}

void
linux_store_registers (int regno)
{
#ifdef HAVE_LINUX_REGSETS
  if (regsets_store_inferior_registers () == 0)
    return;
#endif
#ifdef HAVE_LINUX_USRREGS
  usr_store_inferior_registers (regno);
#endif
}


/* Copy LEN bytes from inferior's memory starting at MEMADDR
   to debugger memory starting at MYADDR.  */

static int
linux_read_memory (CORE_ADDR memaddr, unsigned char *myaddr, int len)
{
  register int i;
  /* Round starting address down to longword boundary.  */
  register CORE_ADDR addr = memaddr & -(CORE_ADDR) sizeof (PTRACE_XFER_TYPE);
  /* Round ending address up; get number of longwords that makes.  */
  register int count
    = (((memaddr + len) - addr) + sizeof (PTRACE_XFER_TYPE) - 1)
      / sizeof (PTRACE_XFER_TYPE);
  /* Allocate buffer of that many longwords.  */
  register PTRACE_XFER_TYPE *buffer
    = (PTRACE_XFER_TYPE *) alloca (count * sizeof (PTRACE_XFER_TYPE));
  int fd;
  char filename[64];

  /* Try using /proc.  Don't bother for one word.  */
  if (len >= 3 * sizeof (long))
    {
      /* We could keep this file open and cache it - possibly one per
         thread.  That requires some juggling, but is even faster.  */
      sprintf (filename, "/proc/%ld/mem", inferior_pid);
      fd = open (filename, O_RDONLY | O_LARGEFILE);
      if (fd == -1)
        goto no_proc;

      /* If pread64 is available, use it.  It's faster if the kernel
         supports it (only one syscall), and it's 64-bit safe even on
         32-bit platforms (for instance, SPARC debugging a SPARC64
         application).  */
#ifdef HAVE_PREAD64
      if (pread64 (fd, myaddr, len, memaddr) != len)
#else
      if (lseek (fd, memaddr, SEEK_SET) == -1 || read (fd, memaddr, len) != len)
#endif
        {
          close (fd);
          goto no_proc;
        }

      close (fd);
      return 0;
    }

 no_proc:
  /* Read all the longwords */
  for (i = 0; i < count; i++, addr += sizeof (PTRACE_XFER_TYPE))
    {
      errno = 0;
      buffer[i] = ptrace (PTRACE_PEEKTEXT, inferior_pid,
                          (PTRACE_ARG3_TYPE) addr, 0);
      if (errno)
        return errno;
    }

  /* Copy appropriate bytes out of the buffer.  */
  memcpy (myaddr,
          (char *) buffer + (memaddr & (sizeof (PTRACE_XFER_TYPE) - 1)),
          len);

  return 0;
}

/* Copy LEN bytes of data from debugger memory at MYADDR
   to inferior's memory at MEMADDR.
   On failure (cannot write the inferior)
   returns the value of errno.  */

static int
linux_write_memory (CORE_ADDR memaddr, const unsigned char *myaddr, int len)
{
  register int i;
  /* Round starting address down to longword boundary.  */
  register CORE_ADDR addr = memaddr & -(CORE_ADDR) sizeof (PTRACE_XFER_TYPE);
  /* Round ending address up; get number of longwords that makes.  */
  register int count
  = (((memaddr + len) - addr) + sizeof (PTRACE_XFER_TYPE) - 1) / sizeof (PTRACE_XFER_TYPE);
  /* Allocate buffer of that many longwords.  */
  register PTRACE_XFER_TYPE *buffer = (PTRACE_XFER_TYPE *) alloca (count * sizeof (PTRACE_XFER_TYPE));

  if (debug_threads)
    {
      fprintf (stderr, "Writing %02x to %08lx\n", (unsigned)myaddr[0], (long)memaddr);
    }

  /* Fill start and end extra bytes of buffer with existing memory data.  */

  buffer[0] = ptrace (PTRACE_PEEKTEXT, inferior_pid,
                      (PTRACE_ARG3_TYPE) addr, 0);

  if (count > 1)
    {
      buffer[count - 1]
        = ptrace (PTRACE_PEEKTEXT, inferior_pid,
                  (PTRACE_ARG3_TYPE) (addr + (count - 1)
                                      * sizeof (PTRACE_XFER_TYPE)),
                  0);
    }

  /* Copy data to be written over corresponding part of buffer */

  memcpy ((char *) buffer + (memaddr & (sizeof (PTRACE_XFER_TYPE) - 1)), myaddr, len);

  /* Write the entire buffer.  */

  for (i = 0; i < count; i++, addr += sizeof (PTRACE_XFER_TYPE))
    {
      errno = 0;
      ptrace (PTRACE_POKETEXT, inferior_pid, (PTRACE_ARG3_TYPE) addr, buffer[i]);
      if (errno)
        return errno;
    }

  return 0;
}

static int linux_supports_tracefork_flag;

/* Helper functions for linux_test_for_tracefork, called via clone ().  */

static int
linux_tracefork_grandchild (void *arg)
{
  _exit (0);
}

#define STACK_SIZE 4096

static int
linux_tracefork_child (void *arg)
{
  ptrace (PTRACE_TRACEME, 0, 0, 0);
  kill (getpid (), SIGSTOP);
#ifdef __ia64__
  __clone2 (linux_tracefork_grandchild, arg, STACK_SIZE,
            CLONE_VM | SIGCHLD, NULL);
#else
  clone (linux_tracefork_grandchild, arg + STACK_SIZE,
         CLONE_VM | SIGCHLD, NULL);
#endif
  _exit (0);
}

/* Wrapper function for waitpid which handles EINTR, and emulates
   __WALL for systems where that is not available.  */

static int
my_waitpid (int pid, int *status, int flags)
{
  int ret, out_errno;

  if (debug_threads)
    fprintf (stderr, "my_waitpid (%d, 0x%x)\n", pid, flags);

  if (flags & __WALL)
    {
      sigset_t block_mask, org_mask, wake_mask;
      int wnohang;

      wnohang = (flags & WNOHANG) != 0;
      flags &= ~(__WALL | __WCLONE);
      flags |= WNOHANG;

      /* Block all signals while here.  This avoids knowing about
         LinuxThread's signals.  */
      sigfillset (&block_mask);
      sigprocmask (SIG_BLOCK, &block_mask, &org_mask);

      /* ... except during the sigsuspend below.  */
      sigemptyset (&wake_mask);

      while (1)
        {
          /* Since all signals are blocked, there's no need to check
             for EINTR here.  */
          ret = waitpid (pid, status, flags);
          out_errno = errno;

          if (ret == -1 && out_errno != ECHILD)
            break;
          else if (ret > 0)
            break;

          if (flags & __WCLONE)
            {
              /* We've tried both flavors now.  If WNOHANG is set,
                 there's nothing else to do, just bail out.  */
              if (wnohang)
                break;

              if (debug_threads)
                fprintf (stderr, "blocking\n");

              /* Block waiting for signals.  */
              sigsuspend (&wake_mask);
            }

          flags ^= __WCLONE;
        }

      sigprocmask (SIG_SETMASK, &org_mask, NULL);
    }
  else
    {
      do
        ret = waitpid (pid, status, flags);
      while (ret == -1 && errno == EINTR);
      out_errno = errno;
    }

  if (debug_threads)
    fprintf (stderr, "my_waitpid (%d, 0x%x): status(%x), %d\n",
             pid, flags, status ? *status : -1, ret);

  errno = out_errno;
  return ret;
}

/* Determine if PTRACE_O_TRACEFORK can be used to follow fork events.  Make
   sure that we can enable the option, and that it had the desired
   effect.  */

static void
linux_test_for_tracefork (void)
{
  int child_pid, ret, status;
  long second_pid;
  char *stack = xmalloc (STACK_SIZE * 4);

  linux_supports_tracefork_flag = 0;

  /* Use CLONE_VM instead of fork, to support uClinux (no MMU).  */
#ifdef __ia64__
  child_pid = __clone2 (linux_tracefork_child, stack, STACK_SIZE,
                        CLONE_VM | SIGCHLD, stack + STACK_SIZE * 2);
#else
  child_pid = clone (linux_tracefork_child, stack + STACK_SIZE,
                     CLONE_VM | SIGCHLD, stack + STACK_SIZE * 2);
#endif
  if (child_pid == -1)
    perror_with_name ("clone");

  ret = my_waitpid (child_pid, &status, 0);
  if (ret == -1)
    perror_with_name ("waitpid");
  else if (ret != child_pid)
    error ("linux_test_for_tracefork: waitpid: unexpected result %d.", ret);
  if (! WIFSTOPPED (status))
    error ("linux_test_for_tracefork: waitpid: unexpected status %d.", status);

  ret = ptrace (PTRACE_SETOPTIONS, child_pid, 0, PTRACE_O_TRACEFORK);
  if (ret != 0)
    {
      ret = ptrace (PTRACE_KILL, child_pid, 0, 0);
      if (ret != 0)
        {
          warning ("linux_test_for_tracefork: failed to kill child");
          return;
        }

      ret = my_waitpid (child_pid, &status, 0);
      if (ret != child_pid)
        warning ("linux_test_for_tracefork: failed to wait for killed child");
      else if (!WIFSIGNALED (status))
        warning ("linux_test_for_tracefork: unexpected wait status 0x%x from "
                 "killed child", status);

      return;
    }

  ret = ptrace (PTRACE_CONT, child_pid, 0, 0);
  if (ret != 0)
    warning ("linux_test_for_tracefork: failed to resume child");

  ret = my_waitpid (child_pid, &status, 0);

  if (ret == child_pid && WIFSTOPPED (status)
      && status >> 16 == PTRACE_EVENT_FORK)
    {
      second_pid = 0;
      ret = ptrace (PTRACE_GETEVENTMSG, child_pid, 0, &second_pid);
      if (ret == 0 && second_pid != 0)
        {
          int second_status;

          linux_supports_tracefork_flag = 1;
          my_waitpid (second_pid, &second_status, 0);
          ret = ptrace (PTRACE_KILL, second_pid, 0, 0);
          if (ret != 0)
            warning ("linux_test_for_tracefork: failed to kill second child");
          my_waitpid (second_pid, &status, 0);
        }
    }
  else
    warning ("linux_test_for_tracefork: unexpected result from waitpid "
             "(%d, status 0x%x)", ret, status);

  do
    {
      ret = ptrace (PTRACE_KILL, child_pid, 0, 0);
      if (ret != 0)
        warning ("linux_test_for_tracefork: failed to kill child");
      my_waitpid (child_pid, &status, 0);
    }
  while (WIFSTOPPED (status));

  free (stack);
}


static void
linux_look_up_symbols (void)
{
#ifdef USE_THREAD_DB
  if (thread_db_active)
    return;

  thread_db_active = thread_db_init (!linux_supports_tracefork_flag);
#endif
}

static void
linux_request_interrupt (void)
{
  extern unsigned long signal_pid;

  if (cont_thread != 0 && cont_thread != -1)
    {
      struct lwp_info *lwp;
      int lwpid;

      lwp = get_thread_lwp (current_inferior);
      lwpid = lwpid_of (lwp);
      kill_lwp (lwpid, SIGINT);
    }
  else
    kill_lwp (signal_pid, SIGINT);
}

/* Copy LEN bytes from inferior's auxiliary vector starting at OFFSET
   to debugger memory starting at MYADDR.  */

static int
linux_read_auxv (CORE_ADDR offset, unsigned char *myaddr, unsigned int len)
{
  char filename[PATH_MAX];
  int fd, n;

  snprintf (filename, sizeof filename, "/proc/%ld/auxv", inferior_pid);

  fd = open (filename, O_RDONLY);
  if (fd < 0)
    return -1;

  if (offset != (CORE_ADDR) 0
      && lseek (fd, (off_t) offset, SEEK_SET) != (off_t) offset)
    n = -1;
  else
    n = read (fd, myaddr, len);

  close (fd);

  return n;
}

/* These watchpoint related wrapper functions simply pass on the function call
   if the target has registered a corresponding function.  */

static int
linux_insert_watchpoint (char type, CORE_ADDR addr, int len)
{
  if (the_low_target.insert_watchpoint != NULL)
    return the_low_target.insert_watchpoint (type, addr, len);
  else
    /* Unsupported (see target.h).  */
    return 1;
}

static int
linux_remove_watchpoint (char type, CORE_ADDR addr, int len)
{
  if (the_low_target.remove_watchpoint != NULL)
    return the_low_target.remove_watchpoint (type, addr, len);
  else
    /* Unsupported (see target.h).  */
    return 1;
}

static int
linux_stopped_by_watchpoint (void)
{
  if (the_low_target.stopped_by_watchpoint != NULL)
    return the_low_target.stopped_by_watchpoint ();
  else
    return 0;
}

static CORE_ADDR
linux_stopped_data_address (void)
{
  if (the_low_target.stopped_data_address != NULL)
    return the_low_target.stopped_data_address ();
  else
    return 0;
}

#if defined(__UCLIBC__) && defined(HAS_NOMMU)
#if defined(__mcoldfire__)
/* These should really be defined in the kernel's ptrace.h header.  */
#define PT_TEXT_ADDR 49*4
#define PT_DATA_ADDR 50*4
#define PT_TEXT_END_ADDR  51*4
#endif

/* Under uClinux, programs are loaded at non-zero offsets, which we need
   to tell gdb about.  */

static int
linux_read_offsets (CORE_ADDR *text_p, CORE_ADDR *data_p)
{
#if defined(PT_TEXT_ADDR) && defined(PT_DATA_ADDR) && defined(PT_TEXT_END_ADDR)
  unsigned long text, text_end, data;
  int pid = lwpid_of (get_thread_lwp (current_inferior));

  errno = 0;

  text = ptrace (PTRACE_PEEKUSER, pid, (long)PT_TEXT_ADDR, 0);
  text_end = ptrace (PTRACE_PEEKUSER, pid, (long)PT_TEXT_END_ADDR, 0);
  data = ptrace (PTRACE_PEEKUSER, pid, (long)PT_DATA_ADDR, 0);

  if (errno == 0)
    {
      /* Both text and data offsets produced at compile-time (and so
         used by gdb) are relative to the beginning of the program,
         with the data segment immediately following the text segment.
         However, the actual runtime layout in memory may put the data
         somewhere else, so when we send gdb a data base-address, we
         use the real data base address and subtract the compile-time
         data base-address from it (which is just the length of the
         text segment).  BSS immediately follows data in both
         cases.  */
      *text_p = text;
      *data_p = data - (text_end - text);

      return 1;
    }
#endif
 return 0;
}
#endif

static int
linux_qxfer_osdata (const char *annex,
                    unsigned char *readbuf, unsigned const char *writebuf,
                    CORE_ADDR offset, int len)
{
  /* We make the process list snapshot when the object starts to be
     read.  */
  static const char *buf;
  static long len_avail = -1;
  static struct buffer buffer;

  DIR *dirp;

  if (strcmp (annex, "processes") != 0)
    return 0;

  if (!readbuf || writebuf)
    return 0;

  if (offset == 0)
    {
      if (len_avail != -1 && len_avail != 0)
       buffer_free (&buffer);
      len_avail = 0;
      buf = NULL;
      buffer_init (&buffer);
      buffer_grow_str (&buffer, "<osdata type=\"processes\">");

      dirp = opendir ("/proc");
      if (dirp)
       {
         struct dirent *dp;
         while ((dp = readdir (dirp)) != NULL)
           {
             struct stat statbuf;
             char procentry[sizeof ("/proc/4294967295")];

             if (!isdigit (dp->d_name[0])
                 || strlen (dp->d_name) > sizeof ("4294967295") - 1)
               continue;

             sprintf (procentry, "/proc/%s", dp->d_name);
             if (stat (procentry, &statbuf) == 0
                 && S_ISDIR (statbuf.st_mode))
               {
                 char pathname[128];
                 FILE *f;
                 char cmd[MAXPATHLEN + 1];
                 struct passwd *entry;

                 sprintf (pathname, "/proc/%s/cmdline", dp->d_name);
                 entry = getpwuid (statbuf.st_uid);

                 if ((f = fopen (pathname, "r")) != NULL)
                   {
                     size_t len = fread (cmd, 1, sizeof (cmd) - 1, f);
                     if (len > 0)
                       {
                         int i;
                         for (i = 0; i < len; i++)
                           if (cmd[i] == '\0')
                             cmd[i] = ' ';
                         cmd[len] = '\0';

                         buffer_xml_printf (
                           &buffer,
                           "<item>"
                           "<column name=\"pid\">%s</column>"
                           "<column name=\"user\">%s</column>"
                           "<column name=\"command\">%s</column>"
                           "</item>",
                           dp->d_name,
                           entry ? entry->pw_name : "?",
                           cmd);
                       }
                     fclose (f);
                   }
               }
           }

         closedir (dirp);
       }
      buffer_grow_str0 (&buffer, "</osdata>\n");
      buf = buffer_finish (&buffer);
      len_avail = strlen (buf);
    }

  if (offset >= len_avail)
    {
      /* Done.  Get rid of the data.  */
      buffer_free (&buffer);
      buf = NULL;
      len_avail = 0;
      return 0;
    }

  if (len > len_avail - offset)
    len = len_avail - offset;
  memcpy (readbuf, buf + offset, len);

  return len;
}

static int
linux_xfer_siginfo (const char *annex, unsigned char *readbuf,
                    unsigned const char *writebuf, CORE_ADDR offset, int len)
{
  struct siginfo siginfo;
  long pid = -1;

  if (current_inferior == NULL)
    return -1;

  pid = lwpid_of (get_thread_lwp (current_inferior));

  if (debug_threads)
    fprintf (stderr, "%s siginfo for lwp %ld.\n",
             readbuf != NULL ? "Reading" : "Writing",
             pid);

  if (offset > sizeof (siginfo))
    return -1;

  if (ptrace (PTRACE_GETSIGINFO, pid, 0, &siginfo) != 0)
    return -1;

  if (offset + len > sizeof (siginfo))
    len = sizeof (siginfo) - offset;

  if (readbuf != NULL)
    memcpy (readbuf, (char *) &siginfo + offset, len);
  else
    {
      memcpy ((char *) &siginfo + offset, writebuf, len);
      if (ptrace (PTRACE_SETSIGINFO, pid, 0, &siginfo) != 0)
        return -1;
    }

  return len;
}

/* SIGCHLD handler that serves two purposes: In non-stop/async mode,
   so we notice when children change state; as the handler for the
   sigsuspend in my_waitpid.  */

static void
sigchld_handler (int signo)
{
  int old_errno = errno;

  if (debug_threads)
    /* fprintf is not async-signal-safe, so call write directly.  */
    write (2, "sigchld_handler\n", sizeof ("sigchld_handler\n") - 1);

  if (target_is_async_p ())
    async_file_mark (); /* trigger a linux_wait */

  errno = old_errno;
}

static int
linux_supports_non_stop (void)
{
  return 1;
}

static int
linux_async (int enable)
{
  int previous = (linux_event_pipe[0] != -1);

  if (previous != enable)
    {
      sigset_t mask;
      sigemptyset (&mask);
      sigaddset (&mask, SIGCHLD);

      sigprocmask (SIG_BLOCK, &mask, NULL);

      if (enable)
        {
          if (pipe (linux_event_pipe) == -1)
            fatal ("creating event pipe failed.");

          fcntl (linux_event_pipe[0], F_SETFL, O_NONBLOCK);
          fcntl (linux_event_pipe[1], F_SETFL, O_NONBLOCK);

          /* Register the event loop handler.  */
          add_file_handler (linux_event_pipe[0],
                            handle_target_event, NULL);

          /* Always trigger a linux_wait.  */
          async_file_mark ();
        }
      else
        {
          delete_file_handler (linux_event_pipe[0]);

          close (linux_event_pipe[0]);
          close (linux_event_pipe[1]);
          linux_event_pipe[0] = -1;
          linux_event_pipe[1] = -1;
        }

      sigprocmask (SIG_UNBLOCK, &mask, NULL);
    }

  return previous;
}

static int
linux_start_non_stop (int nonstop)
{
  /* Register or unregister from event-loop accordingly.  */
  linux_async (nonstop);
  return 0;
}

static struct target_ops linux_target_ops = {
  linux_create_inferior,
  linux_attach,
  linux_kill,
  linux_detach,
  linux_join,
  linux_thread_alive,
  linux_resume,
  linux_wait,
  linux_fetch_registers,
  linux_store_registers,
  linux_read_memory,
  linux_write_memory,
  linux_look_up_symbols,
  linux_request_interrupt,
  linux_read_auxv,
  linux_insert_watchpoint,
  linux_remove_watchpoint,
  linux_stopped_by_watchpoint,
  linux_stopped_data_address,
#if defined(__UCLIBC__) && defined(HAS_NOMMU)
  linux_read_offsets,
#else
  NULL,
#endif
#ifdef USE_THREAD_DB
  thread_db_get_tls_address,
#else
  NULL,
#endif
  NULL,
  hostio_last_error_from_errno,
  linux_qxfer_osdata,
  linux_xfer_siginfo,
  linux_supports_non_stop,
  linux_async,
  linux_start_non_stop,
};

static void
linux_init_signals ()
{
  /* FIXME drow/2002-06-09: As above, we should check with LinuxThreads
     to find what the cancel signal actually is.  */
  signal (__SIGRTMIN+1, SIG_IGN);
}

void
initialize_low (void)
{
  struct sigaction sigchld_action;
  memset (&sigchld_action, 0, sizeof (sigchld_action));
  thread_db_active = 0;
  set_target_ops (&linux_target_ops);
  set_breakpoint_data (the_low_target.breakpoint,
                       the_low_target.breakpoint_len);
  linux_init_signals ();
  linux_test_for_tracefork ();
#ifdef HAVE_LINUX_REGSETS
  for (num_regsets = 0; target_regsets[num_regsets].size >= 0; num_regsets++)
    ;
  disabled_regsets = xmalloc (num_regsets);
#endif

  sigchld_action.sa_handler = sigchld_handler;
  sigemptyset (&sigchld_action.sa_mask);
  sigchld_action.sa_flags = SA_RESTART;
  sigaction (SIGCHLD, &sigchld_action, NULL);
}