gdb/infrun.c

gdb/infrun.c - gdb

Source code

/* Target-struct-independent code to start (run) and stop an inferior

   process.



   Copyright (C) 1986-2015 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 "defs.h"

#include "infrun.h"

#include <ctype.h>

#include "symtab.h"

#include "frame.h"

#include "inferior.h"

#include "breakpoint.h"

#include "gdb_wait.h"

#include "gdbcore.h"

#include "gdbcmd.h"

#include "cli/cli-script.h"

#include "target.h"

#include "gdbthread.h"

#include "annotate.h"

#include "symfile.h"

#include "top.h"

#include <signal.h>

#include "inf-loop.h"

#include "regcache.h"

#include "value.h"

#include "observer.h"

#include "language.h"

#include "solib.h"

#include "main.h"

#include "dictionary.h"

#include "block.h"

#include "mi/mi-common.h"

#include "event-top.h"

#include "record.h"

#include "record-full.h"

#include "inline-frame.h"

#include "jit.h"

#include "tracepoint.h"

#include "continuations.h"

#include "interps.h"

#include "skip.h"

#include "probe.h"

#include "objfiles.h"

#include "completer.h"

#include "target-descriptions.h"

#include "target-dcache.h"

#include "terminal.h"



/* Prototypes for local functions */



static void signals_info (char *, int);



static void handle_command (char *, int);



static void sig_print_info (enum gdb_signal);



static void sig_print_header (void);



static void resume_cleanups (void *);



static int hook_stop_stub (void *);



static int restore_selected_frame (void *);



static int follow_fork (void);



static int follow_fork_inferior (int follow_child, int detach_fork);



static void follow_inferior_reset_breakpoints (void);



static void set_schedlock_func (char *args, int from_tty,

                                struct cmd_list_element *c);



static int currently_stepping (struct thread_info *tp);



static void xdb_handle_command (char *args, int from_tty);



void _initialize_infrun (void);



void nullify_last_target_wait_ptid (void);



static void insert_hp_step_resume_breakpoint_at_frame (struct frame_info *);



static void insert_step_resume_breakpoint_at_caller (struct frame_info *);



static void insert_longjmp_resume_breakpoint (struct gdbarch *, CORE_ADDR);



/* When set, stop the 'step' command if we enter a function which has

   no line number information.  The normal behavior is that we step

   over such function.  */

‌int step_stop_if_no_debug = 0;

static void

‌show_step_stop_if_no_debug (struct ui_file *file, int from_tty,

                            struct cmd_list_element *c, const char *value)

{

  fprintf_filtered (file, _("Mode of the step operation is %s.\n"), value);

}



/* In asynchronous mode, but simulating synchronous execution.  */



‌int sync_execution = 0;



/* proceed and normal_stop use this to notify the user when the

   inferior stopped in a different thread than it had been running

   in.  */



‌static ptid_t previous_inferior_ptid;



/* If set (default for legacy reasons), when following a fork, GDB

   will detach from one of the fork branches, child or parent.

   Exactly which branch is detached depends on 'set follow-fork-mode'

   setting.  */



‌static int detach_fork = 1;



‌int debug_displaced = 0;

static void

‌show_debug_displaced (struct ui_file *file, int from_tty,

                      struct cmd_list_element *c, const char *value)

{

  fprintf_filtered (file, _("Displace stepping debugging is %s.\n"), value);

}



‌unsigned int debug_infrun = 0;

static void

‌show_debug_infrun (struct ui_file *file, int from_tty,

                   struct cmd_list_element *c, const char *value)

{

  fprintf_filtered (file, _("Inferior debugging is %s.\n"), value);

}





/* Support for disabling address space randomization.  */



‌int disable_randomization = 1;



static void

‌show_disable_randomization (struct ui_file *file, int from_tty,

                            struct cmd_list_element *c, const char *value)

{

  if (target_supports_disable_randomization ())

    fprintf_filtered (file,

                      _("Disabling randomization of debuggee's "

                        "virtual address space is %s.\n"),

                      value);

  else

    fputs_filtered (_("Disabling randomization of debuggee's "

                      "virtual address space is unsupported on\n"

                      "this platform.\n"), file);

}



static void

‌set_disable_randomization (char *args, int from_tty,

                           struct cmd_list_element *c)

{

  if (!target_supports_disable_randomization ())

    error (_("Disabling randomization of debuggee's "

             "virtual address space is unsupported on\n"

             "this platform."));

}



/* User interface for non-stop mode.  */



‌int non_stop = 0;

‌static int non_stop_1 = 0;



static void

‌set_non_stop (char *args, int from_tty,

              struct cmd_list_element *c)

{

  if (target_has_execution)

    {

      non_stop_1 = non_stop;

      error (_("Cannot change this setting while the inferior is running."));

    }



  non_stop = non_stop_1;

}



static void

‌show_non_stop (struct ui_file *file, int from_tty,

               struct cmd_list_element *c, const char *value)

{

  fprintf_filtered (file,

                    _("Controlling the inferior in non-stop mode is %s.\n"),

                    value);

}



/* "Observer mode" is somewhat like a more extreme version of

   non-stop, in which all GDB operations that might affect the

   target's execution have been disabled.  */



‌int observer_mode = 0;

‌static int observer_mode_1 = 0;



static void

‌set_observer_mode (char *args, int from_tty,

                   struct cmd_list_element *c)

{

  if (target_has_execution)

    {

      observer_mode_1 = observer_mode;

      error (_("Cannot change this setting while the inferior is running."));

    }



  observer_mode = observer_mode_1;



  may_write_registers = !observer_mode;

  may_write_memory = !observer_mode;

  may_insert_breakpoints = !observer_mode;

  may_insert_tracepoints = !observer_mode;

  /* We can insert fast tracepoints in or out of observer mode,

     but enable them if we're going into this mode.  */

  if (observer_mode)

    may_insert_fast_tracepoints = 1;

  may_stop = !observer_mode;

  update_target_permissions ();



  /* Going *into* observer mode we must force non-stop, then

     going out we leave it that way.  */

  if (observer_mode)

    {

      pagination_enabled = 0;

      non_stop = non_stop_1 = 1;

    }



  if (from_tty)

    printf_filtered (_("Observer mode is now %s.\n"),

                     (observer_mode ? "on" : "off"));

}



static void

‌show_observer_mode (struct ui_file *file, int from_tty,

                    struct cmd_list_element *c, const char *value)

{

  fprintf_filtered (file, _("Observer mode is %s.\n"), value);

}



/* This updates the value of observer mode based on changes in

   permissions.  Note that we are deliberately ignoring the values of

   may-write-registers and may-write-memory, since the user may have

   reason to enable these during a session, for instance to turn on a

   debugging-related global.  */



void

‌update_observer_mode (void)

{

  int newval;



  newval = (!may_insert_breakpoints

            && !may_insert_tracepoints

            && may_insert_fast_tracepoints

            && !may_stop

            && non_stop);



  /* Let the user know if things change.  */

  if (newval != observer_mode)

    printf_filtered (_("Observer mode is now %s.\n"),

                     (newval ? "on" : "off"));



  observer_mode = observer_mode_1 = newval;

}



/* Tables of how to react to signals; the user sets them.  */



‌static unsigned char *signal_stop;

‌static unsigned char *signal_print;

‌static unsigned char *signal_program;



/* Table of signals that are registered with "catch signal".  A

   non-zero entry indicates that the signal is caught by some "catch

   signal" command.  This has size GDB_SIGNAL_LAST, to accommodate all

   signals.  */

‌static unsigned char *signal_catch;



/* Table of signals that the target may silently handle.

   This is automatically determined from the flags above,

   and simply cached here.  */

‌static unsigned char *signal_pass;



‌#define SET_SIGS(nsigs,sigs,flags) \

  do { \

    int signum = (nsigs); \

    while (signum-- > 0) \

      if ((sigs)[signum]) \

        (flags)[signum] = 1; \

  } while (0)



‌#define UNSET_SIGS(nsigs,sigs,flags) \

  do { \

    int signum = (nsigs); \

    while (signum-- > 0) \

      if ((sigs)[signum]) \

        (flags)[signum] = 0; \

  } while (0)



/* Update the target's copy of SIGNAL_PROGRAM.  The sole purpose of

   this function is to avoid exporting `signal_program'.  */



void

‌update_signals_program_target (void)

{

  target_program_signals ((int) GDB_SIGNAL_LAST, signal_program);

}



/* Value to pass to target_resume() to cause all threads to resume.  */



‌#define RESUME_ALL minus_one_ptid



/* Command list pointer for the "stop" placeholder.  */



‌static struct cmd_list_element *stop_command;



/* Function inferior was in as of last step command.  */



‌static struct symbol *step_start_function;



/* Nonzero if we want to give control to the user when we're notified

   of shared library events by the dynamic linker.  */

‌int stop_on_solib_events;



/* Enable or disable optional shared library event breakpoints

   as appropriate when the above flag is changed.  */



static void

‌set_stop_on_solib_events (char *args, int from_tty, struct cmd_list_element *c)

{

  update_solib_breakpoints ();

}



static void

‌show_stop_on_solib_events (struct ui_file *file, int from_tty,

                           struct cmd_list_element *c, const char *value)

{

  fprintf_filtered (file, _("Stopping for shared library events is %s.\n"),

                    value);

}



/* Nonzero means expecting a trace trap

   and should stop the inferior and return silently when it happens.  */



‌int stop_after_trap;



/* Save register contents here when executing a "finish" command or are

   about to pop a stack dummy frame, if-and-only-if proceed_to_finish is set.

   Thus this contains the return value from the called function (assuming

   values are returned in a register).  */



‌struct regcache *stop_registers;



/* Nonzero after stop if current stack frame should be printed.  */



‌static int stop_print_frame;



/* This is a cached copy of the pid/waitstatus of the last event

   returned by target_wait()/deprecated_target_wait_hook().  This

   information is returned by get_last_target_status().  */

‌static ptid_t target_last_wait_ptid;

‌static struct target_waitstatus target_last_waitstatus;



static void context_switch (ptid_t ptid);



void init_thread_stepping_state (struct thread_info *tss);



‌static const char follow_fork_mode_child[] = "child";

‌static const char follow_fork_mode_parent[] = "parent";



‌static const char *const follow_fork_mode_kind_names[] = {

  follow_fork_mode_child,

  follow_fork_mode_parent,

  NULL

};



‌static const char *follow_fork_mode_string = follow_fork_mode_parent;

static void

‌show_follow_fork_mode_string (struct ui_file *file, int from_tty,

                              struct cmd_list_element *c, const char *value)

{

  fprintf_filtered (file,

                    _("Debugger response to a program "

                      "call of fork or vfork is \"%s\".\n"),

                    value);

}





/* Handle changes to the inferior list based on the type of fork,

   which process is being followed, and whether the other process

   should be detached.  On entry inferior_ptid must be the ptid of

   the fork parent.  At return inferior_ptid is the ptid of the

   followed inferior.  */



static int

‌follow_fork_inferior (int follow_child, int detach_fork)

{

  int has_vforked;

  int parent_pid, child_pid;



  has_vforked = (inferior_thread ()->pending_follow.kind

                 == TARGET_WAITKIND_VFORKED);

  parent_pid = ptid_get_lwp (inferior_ptid);

  if (parent_pid == 0)

    parent_pid = ptid_get_pid (inferior_ptid);

  child_pid

    = ptid_get_pid (inferior_thread ()->pending_follow.value.related_pid);



  if (has_vforked

      && !non_stop /* Non-stop always resumes both branches.  */

      && (!target_is_async_p () || sync_execution)

      && !(follow_child || detach_fork || sched_multi))

    {

      /* The parent stays blocked inside the vfork syscall until the

         child execs or exits.  If we don't let the child run, then

         the parent stays blocked.  If we're telling the parent to run

         in the foreground, the user will not be able to ctrl-c to get

         back the terminal, effectively hanging the debug session.  */

      fprintf_filtered (gdb_stderr, _("\

Can not resume the parent process over vfork in the foreground while\n\

holding the child stopped.  Try \"set detach-on-fork\" or \

\"set schedule-multiple\".\n"));

      /* FIXME output string > 80 columns.  */

      return 1;

    }



  if (!follow_child)

    {

      /* Detach new forked process?  */

      if (detach_fork)

        {

          struct cleanup *old_chain;



          /* Before detaching from the child, remove all breakpoints

             from it.  If we forked, then this has already been taken

             care of by infrun.c.  If we vforked however, any

             breakpoint inserted in the parent is visible in the

             child, even those added while stopped in a vfork

             catchpoint.  This will remove the breakpoints from the

             parent also, but they'll be reinserted below.  */

          if (has_vforked)

            {

              /* Keep breakpoints list in sync.  */

              remove_breakpoints_pid (ptid_get_pid (inferior_ptid));

            }



          if (info_verbose || debug_infrun)

            {

              target_terminal_ours_for_output ();

              fprintf_filtered (gdb_stdlog,

                                _("Detaching after %s from "

                                  "child process %d.\n"),

                                has_vforked ? "vfork" : "fork",

                                child_pid);

            }

        }

      else

        {

          struct inferior *parent_inf, *child_inf;

          struct cleanup *old_chain;



          /* Add process to GDB's tables.  */

          child_inf = add_inferior (child_pid);



          parent_inf = current_inferior ();

          child_inf->attach_flag = parent_inf->attach_flag;

          copy_terminal_info (child_inf, parent_inf);

          child_inf->gdbarch = parent_inf->gdbarch;

          copy_inferior_target_desc_info (child_inf, parent_inf);



          old_chain = save_inferior_ptid ();

          save_current_program_space ();



          inferior_ptid = ptid_build (child_pid, child_pid, 0);

          add_thread (inferior_ptid);

          child_inf->symfile_flags = SYMFILE_NO_READ;



          /* If this is a vfork child, then the address-space is

             shared with the parent.  */

          if (has_vforked)

            {

              child_inf->pspace = parent_inf->pspace;

              child_inf->aspace = parent_inf->aspace;



              /* The parent will be frozen until the child is done

                 with the shared region.  Keep track of the

                 parent.  */

              child_inf->vfork_parent = parent_inf;

              child_inf->pending_detach = 0;

              parent_inf->vfork_child = child_inf;

              parent_inf->pending_detach = 0;

            }

          else

            {

              child_inf->aspace = new_address_space ();

              child_inf->pspace = add_program_space (child_inf->aspace);

              child_inf->removable = 1;

              set_current_program_space (child_inf->pspace);

              clone_program_space (child_inf->pspace, parent_inf->pspace);



              /* Let the shared library layer (e.g., solib-svr4) learn

                 about this new process, relocate the cloned exec, pull

                 in shared libraries, and install the solib event

                 breakpoint.  If a "cloned-VM" event was propagated

                 better throughout the core, this wouldn't be

                 required.  */

              solib_create_inferior_hook (0);

            }



          do_cleanups (old_chain);

        }



      if (has_vforked)

        {

          struct inferior *parent_inf;



          parent_inf = current_inferior ();



          /* If we detached from the child, then we have to be careful

             to not insert breakpoints in the parent until the child

             is done with the shared memory region.  However, if we're

             staying attached to the child, then we can and should

             insert breakpoints, so that we can debug it.  A

             subsequent child exec or exit is enough to know when does

             the child stops using the parent's address space.  */

          parent_inf->waiting_for_vfork_done = detach_fork;

          parent_inf->pspace->breakpoints_not_allowed = detach_fork;

        }

    }

  else

    {

      /* Follow the child.  */

      struct inferior *parent_inf, *child_inf;

      struct program_space *parent_pspace;



      if (info_verbose || debug_infrun)

        {

          target_terminal_ours_for_output ();

          fprintf_filtered (gdb_stdlog,

                            _("Attaching after process %d "

                              "%s to child process %d.\n"),

                            parent_pid,

                            has_vforked ? "vfork" : "fork",

                            child_pid);

        }



      /* Add the new inferior first, so that the target_detach below

         doesn't unpush the target.  */



      child_inf = add_inferior (child_pid);



      parent_inf = current_inferior ();

      child_inf->attach_flag = parent_inf->attach_flag;

      copy_terminal_info (child_inf, parent_inf);

      child_inf->gdbarch = parent_inf->gdbarch;

      copy_inferior_target_desc_info (child_inf, parent_inf);



      parent_pspace = parent_inf->pspace;



      /* If we're vforking, we want to hold on to the parent until the

         child exits or execs.  At child exec or exit time we can

         remove the old breakpoints from the parent and detach or

         resume debugging it.  Otherwise, detach the parent now; we'll

         want to reuse it's program/address spaces, but we can't set

         them to the child before removing breakpoints from the

         parent, otherwise, the breakpoints module could decide to

         remove breakpoints from the wrong process (since they'd be

         assigned to the same address space).  */



      if (has_vforked)

        {

          gdb_assert (child_inf->vfork_parent == NULL);

          gdb_assert (parent_inf->vfork_child == NULL);

          child_inf->vfork_parent = parent_inf;

          child_inf->pending_detach = 0;

          parent_inf->vfork_child = child_inf;

          parent_inf->pending_detach = detach_fork;

          parent_inf->waiting_for_vfork_done = 0;

        }

      else if (detach_fork)

        {

          if (info_verbose || debug_infrun)

            {

              target_terminal_ours_for_output ();

              fprintf_filtered (gdb_stdlog,

                                _("Detaching after fork from "

                                  "child process %d.\n"),

                                child_pid);

            }



          target_detach (NULL, 0);

        }



      /* Note that the detach above makes PARENT_INF dangling.  */



      /* Add the child thread to the appropriate lists, and switch to

         this new thread, before cloning the program space, and

         informing the solib layer about this new process.  */



      inferior_ptid = ptid_build (child_pid, child_pid, 0);

      add_thread (inferior_ptid);



      /* If this is a vfork child, then the address-space is shared

         with the parent.  If we detached from the parent, then we can

         reuse the parent's program/address spaces.  */

      if (has_vforked || detach_fork)

        {

          child_inf->pspace = parent_pspace;

          child_inf->aspace = child_inf->pspace->aspace;

        }

      else

        {

          child_inf->aspace = new_address_space ();

          child_inf->pspace = add_program_space (child_inf->aspace);

          child_inf->removable = 1;

          child_inf->symfile_flags = SYMFILE_NO_READ;

          set_current_program_space (child_inf->pspace);

          clone_program_space (child_inf->pspace, parent_pspace);



          /* Let the shared library layer (e.g., solib-svr4) learn

             about this new process, relocate the cloned exec, pull in

             shared libraries, and install the solib event breakpoint.

             If a "cloned-VM" event was propagated better throughout

             the core, this wouldn't be required.  */

          solib_create_inferior_hook (0);

        }

    }



  return target_follow_fork (follow_child, detach_fork);

}



/* Tell the target to follow the fork we're stopped at.  Returns true

   if the inferior should be resumed; false, if the target for some

   reason decided it's best not to resume.  */



static int

‌follow_fork (void)

{

  int follow_child = (follow_fork_mode_string == follow_fork_mode_child);

  int should_resume = 1;

  struct thread_info *tp;



  /* Copy user stepping state to the new inferior thread.  FIXME: the

     followed fork child thread should have a copy of most of the

     parent thread structure's run control related fields, not just these.

     Initialized to avoid "may be used uninitialized" warnings from gcc.  */

  struct breakpoint *step_resume_breakpoint = NULL;

  struct breakpoint *exception_resume_breakpoint = NULL;

  CORE_ADDR step_range_start = 0;

  CORE_ADDR step_range_end = 0;

  struct frame_id step_frame_id = { 0 };

  struct interp *command_interp = NULL;



  if (!non_stop)

    {

      ptid_t wait_ptid;

      struct target_waitstatus wait_status;



      /* Get the last target status returned by target_wait().  */

      get_last_target_status (&wait_ptid, &wait_status);



      /* If not stopped at a fork event, then there's nothing else to

         do.  */

      if (wait_status.kind != TARGET_WAITKIND_FORKED

          && wait_status.kind != TARGET_WAITKIND_VFORKED)

        return 1;



      /* Check if we switched over from WAIT_PTID, since the event was

         reported.  */

      if (!ptid_equal (wait_ptid, minus_one_ptid)

          && !ptid_equal (inferior_ptid, wait_ptid))

        {

          /* We did.  Switch back to WAIT_PTID thread, to tell the

             target to follow it (in either direction).  We'll

             afterwards refuse to resume, and inform the user what

             happened.  */

          switch_to_thread (wait_ptid);

          should_resume = 0;

        }

    }



  tp = inferior_thread ();



  /* If there were any forks/vforks that were caught and are now to be

     followed, then do so now.  */

  switch (tp->pending_follow.kind)

    {

    case TARGET_WAITKIND_FORKED:

    case TARGET_WAITKIND_VFORKED:

      {

        ptid_t parent, child;



        /* If the user did a next/step, etc, over a fork call,

           preserve the stepping state in the fork child.  */

        if (follow_child && should_resume)

          {

            step_resume_breakpoint = clone_momentary_breakpoint

                                         (tp->control.step_resume_breakpoint);

            step_range_start = tp->control.step_range_start;

            step_range_end = tp->control.step_range_end;

            step_frame_id = tp->control.step_frame_id;

            exception_resume_breakpoint

              = clone_momentary_breakpoint (tp->control.exception_resume_breakpoint);

            command_interp = tp->control.command_interp;



            /* For now, delete the parent's sr breakpoint, otherwise,

               parent/child sr breakpoints are considered duplicates,

               and the child version will not be installed.  Remove

               this when the breakpoints module becomes aware of

               inferiors and address spaces.  */

            delete_step_resume_breakpoint (tp);

            tp->control.step_range_start = 0;

            tp->control.step_range_end = 0;

            tp->control.step_frame_id = null_frame_id;

            delete_exception_resume_breakpoint (tp);

            tp->control.command_interp = NULL;

          }



        parent = inferior_ptid;

        child = tp->pending_follow.value.related_pid;



        /* Set up inferior(s) as specified by the caller, and tell the

           target to do whatever is necessary to follow either parent

           or child.  */

        if (follow_fork_inferior (follow_child, detach_fork))

          {

            /* Target refused to follow, or there's some other reason

               we shouldn't resume.  */

            should_resume = 0;

          }

        else

          {

            /* This pending follow fork event is now handled, one way

               or another.  The previous selected thread may be gone

               from the lists by now, but if it is still around, need

               to clear the pending follow request.  */

            tp = find_thread_ptid (parent);

            if (tp)

              tp->pending_follow.kind = TARGET_WAITKIND_SPURIOUS;



            /* This makes sure we don't try to apply the "Switched

               over from WAIT_PID" logic above.  */

            nullify_last_target_wait_ptid ();



            /* If we followed the child, switch to it...  */

            if (follow_child)

              {

                switch_to_thread (child);



                /* ... and preserve the stepping state, in case the

                   user was stepping over the fork call.  */

                if (should_resume)

                  {

                    tp = inferior_thread ();

                    tp->control.step_resume_breakpoint

                      = step_resume_breakpoint;

                    tp->control.step_range_start = step_range_start;

                    tp->control.step_range_end = step_range_end;

                    tp->control.step_frame_id = step_frame_id;

                    tp->control.exception_resume_breakpoint

                      = exception_resume_breakpoint;

                    tp->control.command_interp = command_interp;

                  }

                else

                  {

                    /* If we get here, it was because we're trying to

                       resume from a fork catchpoint, but, the user

                       has switched threads away from the thread that

                       forked.  In that case, the resume command

                       issued is most likely not applicable to the

                       child, so just warn, and refuse to resume.  */

                    warning (_("Not resuming: switched threads "

                               "before following fork child.\n"));

                  }



                /* Reset breakpoints in the child as appropriate.  */

                follow_inferior_reset_breakpoints ();

              }

            else

              switch_to_thread (parent);

          }

      }

      break;

    case TARGET_WAITKIND_SPURIOUS:

      /* Nothing to follow.  */

      break;

    default:

      internal_error (__FILE__, __LINE__,

                      "Unexpected pending_follow.kind %d\n",

                      tp->pending_follow.kind);

      break;

    }



  return should_resume;

}



static void

‌follow_inferior_reset_breakpoints (void)

{

  struct thread_info *tp = inferior_thread ();



  /* Was there a step_resume breakpoint?  (There was if the user

     did a "next" at the fork() call.)  If so, explicitly reset its

     thread number.  Cloned step_resume breakpoints are disabled on

     creation, so enable it here now that it is associated with the

     correct thread.



     step_resumes are a form of bp that are made to be per-thread.

     Since we created the step_resume bp when the parent process

     was being debugged, and now are switching to the child process,

     from the breakpoint package's viewpoint, that's a switch of

     "threads".  We must update the bp's notion of which thread

     it is for, or it'll be ignored when it triggers.  */



  if (tp->control.step_resume_breakpoint)

    {

      breakpoint_re_set_thread (tp->control.step_resume_breakpoint);

      tp->control.step_resume_breakpoint->loc->enabled = 1;

    }



  /* Treat exception_resume breakpoints like step_resume breakpoints.  */

  if (tp->control.exception_resume_breakpoint)

    {

      breakpoint_re_set_thread (tp->control.exception_resume_breakpoint);

      tp->control.exception_resume_breakpoint->loc->enabled = 1;

    }



  /* Reinsert all breakpoints in the child.  The user may have set

     breakpoints after catching the fork, in which case those

     were never set in the child, but only in the parent.  This makes

     sure the inserted breakpoints match the breakpoint list.  */



  breakpoint_re_set ();

  insert_breakpoints ();

}



/* The child has exited or execed: resume threads of the parent the

   user wanted to be executing.  */



static int

‌proceed_after_vfork_done (struct thread_info *thread,

                          void *arg)

{

  int pid = * (int *) arg;



  if (ptid_get_pid (thread->ptid) == pid

      && is_running (thread->ptid)

      && !is_executing (thread->ptid)

      && !thread->stop_requested

      && thread->suspend.stop_signal == GDB_SIGNAL_0)

    {

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog,

                            "infrun: resuming vfork parent thread %s\n",

                            target_pid_to_str (thread->ptid));



      switch_to_thread (thread->ptid);

      clear_proceed_status (0);

      proceed ((CORE_ADDR) -1, GDB_SIGNAL_DEFAULT, 0);

    }



  return 0;

}



/* Called whenever we notice an exec or exit event, to handle

   detaching or resuming a vfork parent.  */



static void

‌handle_vfork_child_exec_or_exit (int exec)

{

  struct inferior *inf = current_inferior ();



  if (inf->vfork_parent)

    {

      int resume_parent = -1;



      /* This exec or exit marks the end of the shared memory region

         between the parent and the child.  If the user wanted to

         detach from the parent, now is the time.  */



      if (inf->vfork_parent->pending_detach)

        {

          struct thread_info *tp;

          struct cleanup *old_chain;

          struct program_space *pspace;

          struct address_space *aspace;



          /* follow-fork child, detach-on-fork on.  */



          inf->vfork_parent->pending_detach = 0;



          if (!exec)

            {

              /* If we're handling a child exit, then inferior_ptid

                 points at the inferior's pid, not to a thread.  */

              old_chain = save_inferior_ptid ();

              save_current_program_space ();

              save_current_inferior ();

            }

          else

            old_chain = save_current_space_and_thread ();



          /* We're letting loose of the parent.  */

          tp = any_live_thread_of_process (inf->vfork_parent->pid);

          switch_to_thread (tp->ptid);



          /* We're about to detach from the parent, which implicitly

             removes breakpoints from its address space.  There's a

             catch here: we want to reuse the spaces for the child,

             but, parent/child are still sharing the pspace at this

             point, although the exec in reality makes the kernel give

             the child a fresh set of new pages.  The problem here is

             that the breakpoints module being unaware of this, would

             likely chose the child process to write to the parent

             address space.  Swapping the child temporarily away from

             the spaces has the desired effect.  Yes, this is "sort

             of" a hack.  */



          pspace = inf->pspace;

          aspace = inf->aspace;

          inf->aspace = NULL;

          inf->pspace = NULL;



          if (debug_infrun || info_verbose)

            {

              target_terminal_ours_for_output ();



              if (exec)

                {

                  fprintf_filtered (gdb_stdlog,

                                    _("Detaching vfork parent process "

                                      "%d after child exec.\n"),

                                    inf->vfork_parent->pid);

                }

              else

                {

                  fprintf_filtered (gdb_stdlog,

                                    _("Detaching vfork parent process "

                                      "%d after child exit.\n"),

                                    inf->vfork_parent->pid);

                }

            }



          target_detach (NULL, 0);



          /* Put it back.  */

          inf->pspace = pspace;

          inf->aspace = aspace;



          do_cleanups (old_chain);

        }

      else if (exec)

        {

          /* We're staying attached to the parent, so, really give the

             child a new address space.  */

          inf->pspace = add_program_space (maybe_new_address_space ());

          inf->aspace = inf->pspace->aspace;

          inf->removable = 1;

          set_current_program_space (inf->pspace);



          resume_parent = inf->vfork_parent->pid;



          /* Break the bonds.  */

          inf->vfork_parent->vfork_child = NULL;

        }

      else

        {

          struct cleanup *old_chain;

          struct program_space *pspace;



          /* If this is a vfork child exiting, then the pspace and

             aspaces were shared with the parent.  Since we're

             reporting the process exit, we'll be mourning all that is

             found in the address space, and switching to null_ptid,

             preparing to start a new inferior.  But, since we don't

             want to clobber the parent's address/program spaces, we

             go ahead and create a new one for this exiting

             inferior.  */



          /* Switch to null_ptid, so that clone_program_space doesn't want

             to read the selected frame of a dead process.  */

          old_chain = save_inferior_ptid ();

          inferior_ptid = null_ptid;



          /* This inferior is dead, so avoid giving the breakpoints

             module the option to write through to it (cloning a

             program space resets breakpoints).  */

          inf->aspace = NULL;

          inf->pspace = NULL;

          pspace = add_program_space (maybe_new_address_space ());

          set_current_program_space (pspace);

          inf->removable = 1;

          inf->symfile_flags = SYMFILE_NO_READ;

          clone_program_space (pspace, inf->vfork_parent->pspace);

          inf->pspace = pspace;

          inf->aspace = pspace->aspace;



          /* Put back inferior_ptid.  We'll continue mourning this

             inferior.  */

          do_cleanups (old_chain);



          resume_parent = inf->vfork_parent->pid;

          /* Break the bonds.  */

          inf->vfork_parent->vfork_child = NULL;

        }



      inf->vfork_parent = NULL;



      gdb_assert (current_program_space == inf->pspace);



      if (non_stop && resume_parent != -1)

        {

          /* If the user wanted the parent to be running, let it go

             free now.  */

          struct cleanup *old_chain = make_cleanup_restore_current_thread ();



          if (debug_infrun)

            fprintf_unfiltered (gdb_stdlog,

                                "infrun: resuming vfork parent process %d\n",

                                resume_parent);



          iterate_over_threads (proceed_after_vfork_done, &resume_parent);



          do_cleanups (old_chain);

        }

    }

}



/* Enum strings for "set|show follow-exec-mode".  */



‌static const char follow_exec_mode_new[] = "new";

‌static const char follow_exec_mode_same[] = "same";

‌static const char *const follow_exec_mode_names[] =

{

  follow_exec_mode_new,

  follow_exec_mode_same,

  NULL,

};



‌static const char *follow_exec_mode_string = follow_exec_mode_same;

static void

‌show_follow_exec_mode_string (struct ui_file *file, int from_tty,

                              struct cmd_list_element *c, const char *value)

{

  fprintf_filtered (file, _("Follow exec mode is \"%s\".\n"),  value);

}



/* EXECD_PATHNAME is assumed to be non-NULL.  */



static void

‌follow_exec (ptid_t pid, char *execd_pathname)

{

  struct thread_info *th = inferior_thread ();

  struct inferior *inf = current_inferior ();



  /* This is an exec event that we actually wish to pay attention to.

     Refresh our symbol table to the newly exec'd program, remove any

     momentary bp's, etc.



     If there are breakpoints, they aren't really inserted now,

     since the exec() transformed our inferior into a fresh set

     of instructions.



     We want to preserve symbolic breakpoints on the list, since

     we have hopes that they can be reset after the new a.out's

     symbol table is read.



     However, any "raw" breakpoints must be removed from the list

     (e.g., the solib bp's), since their address is probably invalid

     now.



     And, we DON'T want to call delete_breakpoints() here, since

     that may write the bp's "shadow contents" (the instruction

     value that was overwritten witha TRAP instruction).  Since

     we now have a new a.out, those shadow contents aren't valid.  */



  mark_breakpoints_out ();



  update_breakpoints_after_exec ();



  /* If there was one, it's gone now.  We cannot truly step-to-next

     statement through an exec().  */

  th->control.step_resume_breakpoint = NULL;

  th->control.exception_resume_breakpoint = NULL;

  th->control.single_step_breakpoints = NULL;

  th->control.step_range_start = 0;

  th->control.step_range_end = 0;



  /* The target reports the exec event to the main thread, even if

     some other thread does the exec, and even if the main thread was

     already stopped --- if debugging in non-stop mode, it's possible

     the user had the main thread held stopped in the previous image

     --- release it now.  This is the same behavior as step-over-exec

     with scheduler-locking on in all-stop mode.  */

  th->stop_requested = 0;



  /* What is this a.out's name?  */

  printf_unfiltered (_("%s is executing new program: %s\n"),

                     target_pid_to_str (inferior_ptid),

                     execd_pathname);



  /* We've followed the inferior through an exec.  Therefore, the

     inferior has essentially been killed & reborn.  */



  gdb_flush (gdb_stdout);



  breakpoint_init_inferior (inf_execd);



  if (gdb_sysroot && *gdb_sysroot)

    {

      char *name = alloca (strlen (gdb_sysroot)

                            + strlen (execd_pathname)

                            + 1);



      strcpy (name, gdb_sysroot);

      strcat (name, execd_pathname);

      execd_pathname = name;

    }



  /* Reset the shared library package.  This ensures that we get a

     shlib event when the child reaches "_start", at which point the

     dld will have had a chance to initialize the child.  */

  /* Also, loading a symbol file below may trigger symbol lookups, and

     we don't want those to be satisfied by the libraries of the

     previous incarnation of this process.  */

  no_shared_libraries (NULL, 0);



  if (follow_exec_mode_string == follow_exec_mode_new)

    {

      struct program_space *pspace;



      /* The user wants to keep the old inferior and program spaces

         around.  Create a new fresh one, and switch to it.  */



      inf = add_inferior (current_inferior ()->pid);

      pspace = add_program_space (maybe_new_address_space ());

      inf->pspace = pspace;

      inf->aspace = pspace->aspace;



      exit_inferior_num_silent (current_inferior ()->num);



      set_current_inferior (inf);

      set_current_program_space (pspace);

    }

  else

    {

      /* The old description may no longer be fit for the new image.

         E.g, a 64-bit process exec'ed a 32-bit process.  Clear the

         old description; we'll read a new one below.  No need to do

         this on "follow-exec-mode new", as the old inferior stays

         around (its description is later cleared/refetched on

         restart).  */

      target_clear_description ();

    }



  gdb_assert (current_program_space == inf->pspace);



  /* That a.out is now the one to use.  */

  exec_file_attach (execd_pathname, 0);



  /* SYMFILE_DEFER_BP_RESET is used as the proper displacement for PIE

     (Position Independent Executable) main symbol file will get applied by

     solib_create_inferior_hook below.  breakpoint_re_set would fail to insert

     the breakpoints with the zero displacement.  */



  symbol_file_add (execd_pathname,

                   (inf->symfile_flags

                    | SYMFILE_MAINLINE | SYMFILE_DEFER_BP_RESET),

                   NULL, 0);



  if ((inf->symfile_flags & SYMFILE_NO_READ) == 0)

    set_initial_language ();



  /* If the target can specify a description, read it.  Must do this

     after flipping to the new executable (because the target supplied

     description must be compatible with the executable's

     architecture, and the old executable may e.g., be 32-bit, while

     the new one 64-bit), and before anything involving memory or

     registers.  */

  target_find_description ();



  solib_create_inferior_hook (0);



  jit_inferior_created_hook ();



  breakpoint_re_set ();



  /* Reinsert all breakpoints.  (Those which were symbolic have

     been reset to the proper address in the new a.out, thanks

     to symbol_file_command...).  */

  insert_breakpoints ();



  /* The next resume of this inferior should bring it to the shlib

     startup breakpoints.  (If the user had also set bp's on

     "main" from the old (parent) process, then they'll auto-

     matically get reset there in the new process.).  */

}



/* Info about an instruction that is being stepped over.  */



‌struct step_over_info

{

  /* If we're stepping past a breakpoint, this is the address space

     and address of the instruction the breakpoint is set at.  We'll

     skip inserting all breakpoints here.  Valid iff ASPACE is

     non-NULL.  */

  struct address_space *aspace;

  CORE_ADDR address;



  /* The instruction being stepped over triggers a nonsteppable

     watchpoint.  If true, we'll skip inserting watchpoints.  */

  int nonsteppable_watchpoint_p;

};



/* The step-over info of the location that is being stepped over.



   Note that with async/breakpoint always-inserted mode, a user might

   set a new breakpoint/watchpoint/etc. exactly while a breakpoint is

   being stepped over.  As setting a new breakpoint inserts all

   breakpoints, we need to make sure the breakpoint being stepped over

   isn't inserted then.  We do that by only clearing the step-over

   info when the step-over is actually finished (or aborted).



   Presently GDB can only step over one breakpoint at any given time.

   Given threads that can't run code in the same address space as the

   breakpoint's can't really miss the breakpoint, GDB could be taught

   to step-over at most one breakpoint per address space (so this info

   could move to the address space object if/when GDB is extended).

   The set of breakpoints being stepped over will normally be much

   smaller than the set of all breakpoints, so a flag in the

   breakpoint location structure would be wasteful.  A separate list

   also saves complexity and run-time, as otherwise we'd have to go

   through all breakpoint locations clearing their flag whenever we

   start a new sequence.  Similar considerations weigh against storing

   this info in the thread object.  Plus, not all step overs actually

   have breakpoint locations -- e.g., stepping past a single-step

   breakpoint, or stepping to complete a non-continuable

   watchpoint.  */

‌static struct step_over_info step_over_info;



/* Record the address of the breakpoint/instruction we're currently

   stepping over.  */



static void

‌set_step_over_info (struct address_space *aspace, CORE_ADDR address,

                    int nonsteppable_watchpoint_p)

{

  step_over_info.aspace = aspace;

  step_over_info.address = address;

  step_over_info.nonsteppable_watchpoint_p = nonsteppable_watchpoint_p;

}



/* Called when we're not longer stepping over a breakpoint / an

   instruction, so all breakpoints are free to be (re)inserted.  */



static void

‌clear_step_over_info (void)

{

  step_over_info.aspace = NULL;

  step_over_info.address = 0;

  step_over_info.nonsteppable_watchpoint_p = 0;

}



/* See infrun.h.  */



int

‌stepping_past_instruction_at (struct address_space *aspace,

                              CORE_ADDR address)

{

  return (step_over_info.aspace != NULL

          && breakpoint_address_match (aspace, address,

                                       step_over_info.aspace,

                                       step_over_info.address));

}



/* See infrun.h.  */



int

‌stepping_past_nonsteppable_watchpoint (void)

{

  return step_over_info.nonsteppable_watchpoint_p;

}



/* Returns true if step-over info is valid.  */



static int

‌step_over_info_valid_p (void)

{

  return (step_over_info.aspace != NULL

          || stepping_past_nonsteppable_watchpoint ());

}





/* Displaced stepping.  */



/* In non-stop debugging mode, we must take special care to manage

   breakpoints properly; in particular, the traditional strategy for

   stepping a thread past a breakpoint it has hit is unsuitable.

   'Displaced stepping' is a tactic for stepping one thread past a

   breakpoint it has hit while ensuring that other threads running

   concurrently will hit the breakpoint as they should.



   The traditional way to step a thread T off a breakpoint in a

   multi-threaded program in all-stop mode is as follows:



   a0) Initially, all threads are stopped, and breakpoints are not

       inserted.

   a1) We single-step T, leaving breakpoints uninserted.

   a2) We insert breakpoints, and resume all threads.



   In non-stop debugging, however, this strategy is unsuitable: we

   don't want to have to stop all threads in the system in order to

   continue or step T past a breakpoint.  Instead, we use displaced

   stepping:



   n0) Initially, T is stopped, other threads are running, and

       breakpoints are inserted.

   n1) We copy the instruction "under" the breakpoint to a separate

       location, outside the main code stream, making any adjustments

       to the instruction, register, and memory state as directed by

       T's architecture.

   n2) We single-step T over the instruction at its new location.

   n3) We adjust the resulting register and memory state as directed

       by T's architecture.  This includes resetting T's PC to point

       back into the main instruction stream.

   n4) We resume T.



   This approach depends on the following gdbarch methods:



   - gdbarch_max_insn_length and gdbarch_displaced_step_location

     indicate where to copy the instruction, and how much space must

     be reserved there.  We use these in step n1.



   - gdbarch_displaced_step_copy_insn copies a instruction to a new

     address, and makes any necessary adjustments to the instruction,

     register contents, and memory.  We use this in step n1.



   - gdbarch_displaced_step_fixup adjusts registers and memory after

     we have successfuly single-stepped the instruction, to yield the

     same effect the instruction would have had if we had executed it

     at its original address.  We use this in step n3.



   - gdbarch_displaced_step_free_closure provides cleanup.



   The gdbarch_displaced_step_copy_insn and

   gdbarch_displaced_step_fixup functions must be written so that

   copying an instruction with gdbarch_displaced_step_copy_insn,

   single-stepping across the copied instruction, and then applying

   gdbarch_displaced_insn_fixup should have the same effects on the

   thread's memory and registers as stepping the instruction in place

   would have.  Exactly which responsibilities fall to the copy and

   which fall to the fixup is up to the author of those functions.



   See the comments in gdbarch.sh for details.



   Note that displaced stepping and software single-step cannot

   currently be used in combination, although with some care I think

   they could be made to.  Software single-step works by placing

   breakpoints on all possible subsequent instructions; if the

   displaced instruction is a PC-relative jump, those breakpoints

   could fall in very strange places --- on pages that aren't

   executable, or at addresses that are not proper instruction

   boundaries.  (We do generally let other threads run while we wait

   to hit the software single-step breakpoint, and they might

   encounter such a corrupted instruction.)  One way to work around

   this would be to have gdbarch_displaced_step_copy_insn fully

   simulate the effect of PC-relative instructions (and return NULL)

   on architectures that use software single-stepping.



   In non-stop mode, we can have independent and simultaneous step

   requests, so more than one thread may need to simultaneously step

   over a breakpoint.  The current implementation assumes there is

   only one scratch space per process.  In this case, we have to

   serialize access to the scratch space.  If thread A wants to step

   over a breakpoint, but we are currently waiting for some other

   thread to complete a displaced step, we leave thread A stopped and

   place it in the displaced_step_request_queue.  Whenever a displaced

   step finishes, we pick the next thread in the queue and start a new

   displaced step operation on it.  See displaced_step_prepare and

   displaced_step_fixup for details.  */



‌struct displaced_step_request

{

  ptid_t ptid;

  struct displaced_step_request *next;

};



/* Per-inferior displaced stepping state.  */

‌struct displaced_step_inferior_state

{

  /* Pointer to next in linked list.  */

  struct displaced_step_inferior_state *next;



  /* The process this displaced step state refers to.  */

  int pid;



  /* A queue of pending displaced stepping requests.  One entry per

     thread that needs to do a displaced step.  */

  struct displaced_step_request *step_request_queue;



  /* If this is not null_ptid, this is the thread carrying out a

     displaced single-step in process PID.  This thread's state will

     require fixing up once it has completed its step.  */

  ptid_t step_ptid;



  /* The architecture the thread had when we stepped it.  */

  struct gdbarch *step_gdbarch;



  /* The closure provided gdbarch_displaced_step_copy_insn, to be used

     for post-step cleanup.  */

  struct displaced_step_closure *step_closure;



  /* The address of the original instruction, and the copy we

     made.  */

  CORE_ADDR step_original, step_copy;



  /* Saved contents of copy area.  */

  gdb_byte *step_saved_copy;

};



/* The list of states of processes involved in displaced stepping

   presently.  */

‌static struct displaced_step_inferior_state *displaced_step_inferior_states;



/* Get the displaced stepping state of process PID.  */



static struct displaced_step_inferior_state *

‌get_displaced_stepping_state (int pid)

{

  struct displaced_step_inferior_state *state;



  for (state = displaced_step_inferior_states;

       state != NULL;

       state = state->next)

    if (state->pid == pid)

      return state;



  return NULL;

}



/* Add a new displaced stepping state for process PID to the displaced

   stepping state list, or return a pointer to an already existing

   entry, if it already exists.  Never returns NULL.  */



static struct displaced_step_inferior_state *

‌add_displaced_stepping_state (int pid)

{

  struct displaced_step_inferior_state *state;



  for (state = displaced_step_inferior_states;

       state != NULL;

       state = state->next)

    if (state->pid == pid)

      return state;



  state = xcalloc (1, sizeof (*state));

  state->pid = pid;

  state->next = displaced_step_inferior_states;

  displaced_step_inferior_states = state;



  return state;

}



/* If inferior is in displaced stepping, and ADDR equals to starting address

   of copy area, return corresponding displaced_step_closure.  Otherwise,

   return NULL.  */



struct displaced_step_closure*

‌get_displaced_step_closure_by_addr (CORE_ADDR addr)

{

  struct displaced_step_inferior_state *displaced

    = get_displaced_stepping_state (ptid_get_pid (inferior_ptid));



  /* If checking the mode of displaced instruction in copy area.  */

  if (displaced && !ptid_equal (displaced->step_ptid, null_ptid)

     && (displaced->step_copy == addr))

    return displaced->step_closure;



  return NULL;

}



/* Remove the displaced stepping state of process PID.  */



static void

‌remove_displaced_stepping_state (int pid)

{

  struct displaced_step_inferior_state *it, **prev_next_p;



  gdb_assert (pid != 0);



  it = displaced_step_inferior_states;

  prev_next_p = &displaced_step_inferior_states;

  while (it)

    {

      if (it->pid == pid)

        {

          *prev_next_p = it->next;

          xfree (it);

          return;

        }



      prev_next_p = &it->next;

      it = *prev_next_p;

    }

}



static void

‌infrun_inferior_exit (struct inferior *inf)

{

  remove_displaced_stepping_state (inf->pid);

}



/* If ON, and the architecture supports it, GDB will use displaced

   stepping to step over breakpoints.  If OFF, or if the architecture

   doesn't support it, GDB will instead use the traditional

   hold-and-step approach.  If AUTO (which is the default), GDB will

   decide which technique to use to step over breakpoints depending on

   which of all-stop or non-stop mode is active --- displaced stepping

   in non-stop mode; hold-and-step in all-stop mode.  */



‌static enum auto_boolean can_use_displaced_stepping = AUTO_BOOLEAN_AUTO;



static void

‌show_can_use_displaced_stepping (struct ui_file *file, int from_tty,

                                 struct cmd_list_element *c,

                                 const char *value)

{

  if (can_use_displaced_stepping == AUTO_BOOLEAN_AUTO)

    fprintf_filtered (file,

                      _("Debugger's willingness to use displaced stepping "

                        "to step over breakpoints is %s (currently %s).\n"),

                      value, non_stop ? "on" : "off");

  else

    fprintf_filtered (file,

                      _("Debugger's willingness to use displaced stepping "

                        "to step over breakpoints is %s.\n"), value);

}



/* Return non-zero if displaced stepping can/should be used to step

   over breakpoints.  */



static int

‌use_displaced_stepping (struct gdbarch *gdbarch)

{

  return (((can_use_displaced_stepping == AUTO_BOOLEAN_AUTO && non_stop)

           || can_use_displaced_stepping == AUTO_BOOLEAN_TRUE)

          && gdbarch_displaced_step_copy_insn_p (gdbarch)

          && find_record_target () == NULL);

}



/* Clean out any stray displaced stepping state.  */

static void

‌displaced_step_clear (struct displaced_step_inferior_state *displaced)

{

  /* Indicate that there is no cleanup pending.  */

  displaced->step_ptid = null_ptid;



  if (displaced->step_closure)

    {

      gdbarch_displaced_step_free_closure (displaced->step_gdbarch,

                                           displaced->step_closure);

      displaced->step_closure = NULL;

    }

}



static void

‌displaced_step_clear_cleanup (void *arg)

{

  struct displaced_step_inferior_state *state = arg;



  displaced_step_clear (state);

}



/* Dump LEN bytes at BUF in hex to FILE, followed by a newline.  */

void

‌displaced_step_dump_bytes (struct ui_file *file,

                           const gdb_byte *buf,

                           size_t len)

{

  int i;



  for (i = 0; i < len; i++)

    fprintf_unfiltered (file, "%02x ", buf[i]);

  fputs_unfiltered ("\n", file);

}



/* Prepare to single-step, using displaced stepping.



   Note that we cannot use displaced stepping when we have a signal to

   deliver.  If we have a signal to deliver and an instruction to step

   over, then after the step, there will be no indication from the

   target whether the thread entered a signal handler or ignored the

   signal and stepped over the instruction successfully --- both cases

   result in a simple SIGTRAP.  In the first case we mustn't do a

   fixup, and in the second case we must --- but we can't tell which.

   Comments in the code for 'random signals' in handle_inferior_event

   explain how we handle this case instead.



   Returns 1 if preparing was successful -- this thread is going to be

   stepped now; or 0 if displaced stepping this thread got queued.  */

static int

‌displaced_step_prepare (ptid_t ptid)

{

  struct cleanup *old_cleanups, *ignore_cleanups;

  struct thread_info *tp = find_thread_ptid (ptid);

  struct regcache *regcache = get_thread_regcache (ptid);

  struct gdbarch *gdbarch = get_regcache_arch (regcache);

  CORE_ADDR original, copy;

  ULONGEST len;

  struct displaced_step_closure *closure;

  struct displaced_step_inferior_state *displaced;

  int status;



  /* We should never reach this function if the architecture does not

     support displaced stepping.  */

  gdb_assert (gdbarch_displaced_step_copy_insn_p (gdbarch));



  /* Disable range stepping while executing in the scratch pad.  We

     want a single-step even if executing the displaced instruction in

     the scratch buffer lands within the stepping range (e.g., a

     jump/branch).  */

  tp->control.may_range_step = 0;



  /* We have to displaced step one thread at a time, as we only have

     access to a single scratch space per inferior.  */



  displaced = add_displaced_stepping_state (ptid_get_pid (ptid));



  if (!ptid_equal (displaced->step_ptid, null_ptid))

    {

      /* Already waiting for a displaced step to finish.  Defer this

         request and place in queue.  */

      struct displaced_step_request *req, *new_req;



      if (debug_displaced)

        fprintf_unfiltered (gdb_stdlog,

                            "displaced: defering step of %s\n",

                            target_pid_to_str (ptid));



      new_req = xmalloc (sizeof (*new_req));

      new_req->ptid = ptid;

      new_req->next = NULL;



      if (displaced->step_request_queue)

        {

          for (req = displaced->step_request_queue;

               req && req->next;

               req = req->next)

            ;

          req->next = new_req;

        }

      else

        displaced->step_request_queue = new_req;



      return 0;

    }

  else

    {

      if (debug_displaced)

        fprintf_unfiltered (gdb_stdlog,

                            "displaced: stepping %s now\n",

                            target_pid_to_str (ptid));

    }



  displaced_step_clear (displaced);



  old_cleanups = save_inferior_ptid ();

  inferior_ptid = ptid;



  original = regcache_read_pc (regcache);



  copy = gdbarch_displaced_step_location (gdbarch);

  len = gdbarch_max_insn_length (gdbarch);



  /* Save the original contents of the copy area.  */

  displaced->step_saved_copy = xmalloc (len);

  ignore_cleanups = make_cleanup (free_current_contents,

                                  &displaced->step_saved_copy);

  status = target_read_memory (copy, displaced->step_saved_copy, len);

  if (status != 0)

    throw_error (MEMORY_ERROR,

                 _("Error accessing memory address %s (%s) for "

                   "displaced-stepping scratch space."),

                 paddress (gdbarch, copy), safe_strerror (status));

  if (debug_displaced)

    {

      fprintf_unfiltered (gdb_stdlog, "displaced: saved %s: ",

                          paddress (gdbarch, copy));

      displaced_step_dump_bytes (gdb_stdlog,

                                 displaced->step_saved_copy,

                                 len);

    };



  closure = gdbarch_displaced_step_copy_insn (gdbarch,

                                              original, copy, regcache);



  /* We don't support the fully-simulated case at present.  */

  gdb_assert (closure);



  /* Save the information we need to fix things up if the step

     succeeds.  */

  displaced->step_ptid = ptid;

  displaced->step_gdbarch = gdbarch;

  displaced->step_closure = closure;

  displaced->step_original = original;

  displaced->step_copy = copy;



  make_cleanup (displaced_step_clear_cleanup, displaced);



  /* Resume execution at the copy.  */

  regcache_write_pc (regcache, copy);



  discard_cleanups (ignore_cleanups);



  do_cleanups (old_cleanups);



  if (debug_displaced)

    fprintf_unfiltered (gdb_stdlog, "displaced: displaced pc to %s\n",

                        paddress (gdbarch, copy));



  return 1;

}



static void

‌write_memory_ptid (ptid_t ptid, CORE_ADDR memaddr,

                   const gdb_byte *myaddr, int len)

{

  struct cleanup *ptid_cleanup = save_inferior_ptid ();



  inferior_ptid = ptid;

  write_memory (memaddr, myaddr, len);

  do_cleanups (ptid_cleanup);

}



/* Restore the contents of the copy area for thread PTID.  */



static void

‌displaced_step_restore (struct displaced_step_inferior_state *displaced,

                        ptid_t ptid)

{

  ULONGEST len = gdbarch_max_insn_length (displaced->step_gdbarch);



  write_memory_ptid (ptid, displaced->step_copy,

                     displaced->step_saved_copy, len);

  if (debug_displaced)

    fprintf_unfiltered (gdb_stdlog, "displaced: restored %s %s\n",

                        target_pid_to_str (ptid),

                        paddress (displaced->step_gdbarch,

                                  displaced->step_copy));

}



static void

‌displaced_step_fixup (ptid_t event_ptid, enum gdb_signal signal)

{

  struct cleanup *old_cleanups;

  struct displaced_step_inferior_state *displaced

    = get_displaced_stepping_state (ptid_get_pid (event_ptid));



  /* Was any thread of this process doing a displaced step?  */

  if (displaced == NULL)

    return;



  /* Was this event for the pid we displaced?  */

  if (ptid_equal (displaced->step_ptid, null_ptid)

      || ! ptid_equal (displaced->step_ptid, event_ptid))

    return;



  old_cleanups = make_cleanup (displaced_step_clear_cleanup, displaced);



  displaced_step_restore (displaced, displaced->step_ptid);



  /* Did the instruction complete successfully?  */

  if (signal == GDB_SIGNAL_TRAP)

    {

      /* Fix up the resulting state.  */

      gdbarch_displaced_step_fixup (displaced->step_gdbarch,

                                    displaced->step_closure,

                                    displaced->step_original,

                                    displaced->step_copy,

                                    get_thread_regcache (displaced->step_ptid));

    }

  else

    {

      /* Since the instruction didn't complete, all we can do is

         relocate the PC.  */

      struct regcache *regcache = get_thread_regcache (event_ptid);

      CORE_ADDR pc = regcache_read_pc (regcache);



      pc = displaced->step_original + (pc - displaced->step_copy);

      regcache_write_pc (regcache, pc);

    }



  do_cleanups (old_cleanups);



  displaced->step_ptid = null_ptid;



  /* Are there any pending displaced stepping requests?  If so, run

     one now.  Leave the state object around, since we're likely to

     need it again soon.  */

  while (displaced->step_request_queue)

    {

      struct displaced_step_request *head;

      ptid_t ptid;

      struct regcache *regcache;

      struct gdbarch *gdbarch;

      CORE_ADDR actual_pc;

      struct address_space *aspace;



      head = displaced->step_request_queue;

      ptid = head->ptid;

      displaced->step_request_queue = head->next;

      xfree (head);



      context_switch (ptid);



      regcache = get_thread_regcache (ptid);

      actual_pc = regcache_read_pc (regcache);

      aspace = get_regcache_aspace (regcache);



      if (breakpoint_here_p (aspace, actual_pc))

        {

          if (debug_displaced)

            fprintf_unfiltered (gdb_stdlog,

                                "displaced: stepping queued %s now\n",

                                target_pid_to_str (ptid));



          displaced_step_prepare (ptid);



          gdbarch = get_regcache_arch (regcache);



          if (debug_displaced)

            {

              CORE_ADDR actual_pc = regcache_read_pc (regcache);

              gdb_byte buf[4];



              fprintf_unfiltered (gdb_stdlog, "displaced: run %s: ",

                                  paddress (gdbarch, actual_pc));

              read_memory (actual_pc, buf, sizeof (buf));

              displaced_step_dump_bytes (gdb_stdlog, buf, sizeof (buf));

            }



          if (gdbarch_displaced_step_hw_singlestep (gdbarch,

                                                    displaced->step_closure))

            target_resume (ptid, 1, GDB_SIGNAL_0);

          else

            target_resume (ptid, 0, GDB_SIGNAL_0);



          /* Done, we're stepping a thread.  */

          break;

        }

      else

        {

          int step;

          struct thread_info *tp = inferior_thread ();



          /* The breakpoint we were sitting under has since been

             removed.  */

          tp->control.trap_expected = 0;



          /* Go back to what we were trying to do.  */

          step = currently_stepping (tp);



          if (debug_displaced)

            fprintf_unfiltered (gdb_stdlog,

                                "displaced: breakpoint is gone: %s, step(%d)\n",

                                target_pid_to_str (tp->ptid), step);



          target_resume (ptid, step, GDB_SIGNAL_0);

          tp->suspend.stop_signal = GDB_SIGNAL_0;



          /* This request was discarded.  See if there's any other

             thread waiting for its turn.  */

        }

    }

}



/* Update global variables holding ptids to hold NEW_PTID if they were

   holding OLD_PTID.  */

static void

‌infrun_thread_ptid_changed (ptid_t old_ptid, ptid_t new_ptid)

{

  struct displaced_step_request *it;

  struct displaced_step_inferior_state *displaced;



  if (ptid_equal (inferior_ptid, old_ptid))

    inferior_ptid = new_ptid;



  for (displaced = displaced_step_inferior_states;

       displaced;

       displaced = displaced->next)

    {

      if (ptid_equal (displaced->step_ptid, old_ptid))

        displaced->step_ptid = new_ptid;



      for (it = displaced->step_request_queue; it; it = it->next)

        if (ptid_equal (it->ptid, old_ptid))

          it->ptid = new_ptid;

    }

}





/* Resuming.  */



/* Things to clean up if we QUIT out of resume ().  */

static void

‌resume_cleanups (void *ignore)

{

  if (!ptid_equal (inferior_ptid, null_ptid))

    delete_single_step_breakpoints (inferior_thread ());



  normal_stop ();

}



‌static const char schedlock_off[] = "off";

‌static const char schedlock_on[] = "on";

‌static const char schedlock_step[] = "step";

‌static const char *const scheduler_enums[] = {

  schedlock_off,

  schedlock_on,

  schedlock_step,

  NULL

};

‌static const char *scheduler_mode = schedlock_off;

static void

‌show_scheduler_mode (struct ui_file *file, int from_tty,

                     struct cmd_list_element *c, const char *value)

{

  fprintf_filtered (file,

                    _("Mode for locking scheduler "

                      "during execution is \"%s\".\n"),

                    value);

}



static void

‌set_schedlock_func (char *args, int from_tty, struct cmd_list_element *c)

{

  if (!target_can_lock_scheduler)

    {

      scheduler_mode = schedlock_off;

      error (_("Target '%s' cannot support this command."), target_shortname);

    }

}



/* True if execution commands resume all threads of all processes by

   default; otherwise, resume only threads of the current inferior

   process.  */

‌int sched_multi = 0;



/* Try to setup for software single stepping over the specified location.

   Return 1 if target_resume() should use hardware single step.



   GDBARCH the current gdbarch.

   PC the location to step over.  */



static int

‌maybe_software_singlestep (struct gdbarch *gdbarch, CORE_ADDR pc)

{

  int hw_step = 1;



  if (execution_direction == EXEC_FORWARD

      && gdbarch_software_single_step_p (gdbarch)

      && gdbarch_software_single_step (gdbarch, get_current_frame ()))

    {

      hw_step = 0;

    }

  return hw_step;

}



ptid_t

‌user_visible_resume_ptid (int step)

{

  /* By default, resume all threads of all processes.  */

  ptid_t resume_ptid = RESUME_ALL;



  /* Maybe resume only all threads of the current process.  */

  if (!sched_multi && target_supports_multi_process ())

    {

      resume_ptid = pid_to_ptid (ptid_get_pid (inferior_ptid));

    }



  /* Maybe resume a single thread after all.  */

  if (non_stop)

    {

      /* With non-stop mode on, threads are always handled

         individually.  */

      resume_ptid = inferior_ptid;

    }

  else if ((scheduler_mode == schedlock_on)

           || (scheduler_mode == schedlock_step && step))

    {

      /* User-settable 'scheduler' mode requires solo thread resume.  */

      resume_ptid = inferior_ptid;

    }



  /* We may actually resume fewer threads at first, e.g., if a thread

     is stopped at a breakpoint that needs stepping-off, but that

     should not be visible to the user/frontend, and neither should

     the frontend/user be allowed to proceed any of the threads that

     happen to be stopped for internal run control handling, if a

     previous command wanted them resumed.  */

  return resume_ptid;

}



/* Resume the inferior, but allow a QUIT.  This is useful if the user

   wants to interrupt some lengthy single-stepping operation

   (for child processes, the SIGINT goes to the inferior, and so

   we get a SIGINT random_signal, but for remote debugging and perhaps

   other targets, that's not true).



   STEP nonzero if we should step (zero to continue instead).

   SIG is the signal to give the inferior (zero for none).  */

void

‌resume (int step, enum gdb_signal sig)

{

  struct cleanup *old_cleanups = make_cleanup (resume_cleanups, 0);

  struct regcache *regcache = get_current_regcache ();

  struct gdbarch *gdbarch = get_regcache_arch (regcache);

  struct thread_info *tp = inferior_thread ();

  CORE_ADDR pc = regcache_read_pc (regcache);

  struct address_space *aspace = get_regcache_aspace (regcache);

  ptid_t resume_ptid;

  /* From here on, this represents the caller's step vs continue

     request, while STEP represents what we'll actually request the

     target to do.  STEP can decay from a step to a continue, if e.g.,

     we need to implement single-stepping with breakpoints (software

     single-step).  When deciding whether "set scheduler-locking step"

     applies, it's the callers intention that counts.  */

  const int entry_step = step;



  tp->stepped_breakpoint = 0;



  QUIT;



  if (current_inferior ()->waiting_for_vfork_done)

    {

      /* Don't try to single-step a vfork parent that is waiting for

         the child to get out of the shared memory region (by exec'ing

         or exiting).  This is particularly important on software

         single-step archs, as the child process would trip on the

         software single step breakpoint inserted for the parent

         process.  Since the parent will not actually execute any

         instruction until the child is out of the shared region (such

         are vfork's semantics), it is safe to simply continue it.

         Eventually, we'll see a TARGET_WAITKIND_VFORK_DONE event for

         the parent, and tell it to `keep_going', which automatically

         re-sets it stepping.  */

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog,

                            "infrun: resume : clear step\n");

      step = 0;

    }



  if (debug_infrun)

    fprintf_unfiltered (gdb_stdlog,

                        "infrun: resume (step=%d, signal=%s), "

                        "trap_expected=%d, current thread [%s] at %s\n",

                        step, gdb_signal_to_symbol_string (sig),

                        tp->control.trap_expected,

                        target_pid_to_str (inferior_ptid),

                        paddress (gdbarch, pc));



  /* Normally, by the time we reach `resume', the breakpoints are either

     removed or inserted, as appropriate.  The exception is if we're sitting

     at a permanent breakpoint; we need to step over it, but permanent

     breakpoints can't be removed.  So we have to test for it here.  */

  if (breakpoint_here_p (aspace, pc) == permanent_breakpoint_here)

    {

      if (sig != GDB_SIGNAL_0)

        {

          /* We have a signal to pass to the inferior.  The resume

             may, or may not take us to the signal handler.  If this

             is a step, we'll need to stop in the signal handler, if

             there's one, (if the target supports stepping into

             handlers), or in the next mainline instruction, if

             there's no handler.  If this is a continue, we need to be

             sure to run the handler with all breakpoints inserted.

             In all cases, set a breakpoint at the current address

             (where the handler returns to), and once that breakpoint

             is hit, resume skipping the permanent breakpoint.  If

             that breakpoint isn't hit, then we've stepped into the

             signal handler (or hit some other event).  We'll delete

             the step-resume breakpoint then.  */



          if (debug_infrun)

            fprintf_unfiltered (gdb_stdlog,

                                "infrun: resume: skipping permanent breakpoint, "

                                "deliver signal first\n");



          clear_step_over_info ();

          tp->control.trap_expected = 0;



          if (tp->control.step_resume_breakpoint == NULL)

            {

              /* Set a "high-priority" step-resume, as we don't want

                 user breakpoints at PC to trigger (again) when this

                 hits.  */

              insert_hp_step_resume_breakpoint_at_frame (get_current_frame ());

              gdb_assert (tp->control.step_resume_breakpoint->loc->permanent);



              tp->step_after_step_resume_breakpoint = step;

            }



          insert_breakpoints ();

        }

      else

        {

          /* There's no signal to pass, we can go ahead and skip the

             permanent breakpoint manually.  */

          if (debug_infrun)

            fprintf_unfiltered (gdb_stdlog,

                                "infrun: resume: skipping permanent breakpoint\n");

          gdbarch_skip_permanent_breakpoint (gdbarch, regcache);

          /* Update pc to reflect the new address from which we will

             execute instructions.  */

          pc = regcache_read_pc (regcache);



          if (step)

            {

              /* We've already advanced the PC, so the stepping part

                 is done.  Now we need to arrange for a trap to be

                 reported to handle_inferior_event.  Set a breakpoint

                 at the current PC, and run to it.  Don't update

                 prev_pc, because if we end in

                 switch_back_to_stepping, we want the "expected thread

                 advanced also" branch to be taken.  IOW, we don't

                 want this thread to step further from PC

                 (overstep).  */

              insert_single_step_breakpoint (gdbarch, aspace, pc);

              insert_breakpoints ();



              tp->suspend.stop_signal = GDB_SIGNAL_0;

              /* We're continuing with all breakpoints inserted.  It's

                 safe to let the target bypass signals.  */

              target_pass_signals ((int) GDB_SIGNAL_LAST, signal_pass);

              /* ... and safe to let other threads run, according to

                 schedlock.  */

              resume_ptid = user_visible_resume_ptid (entry_step);

              target_resume (resume_ptid, 0, GDB_SIGNAL_0);

              discard_cleanups (old_cleanups);

              return;

            }

        }

    }



  /* If we have a breakpoint to step over, make sure to do a single

     step only.  Same if we have software watchpoints.  */

  if (tp->control.trap_expected || bpstat_should_step ())

    tp->control.may_range_step = 0;



  /* If enabled, step over breakpoints by executing a copy of the

     instruction at a different address.



     We can't use displaced stepping when we have a signal to deliver;

     the comments for displaced_step_prepare explain why.  The

     comments in the handle_inferior event for dealing with 'random

     signals' explain what we do instead.



     We can't use displaced stepping when we are waiting for vfork_done

     event, displaced stepping breaks the vfork child similarly as single

     step software breakpoint.  */

  if (use_displaced_stepping (gdbarch)

      && tp->control.trap_expected

      && sig == GDB_SIGNAL_0

      && !current_inferior ()->waiting_for_vfork_done)

    {

      struct displaced_step_inferior_state *displaced;



      if (!displaced_step_prepare (inferior_ptid))

        {

          /* Got placed in displaced stepping queue.  Will be resumed

             later when all the currently queued displaced stepping

             requests finish.  The thread is not executing at this

             point, and the call to set_executing will be made later.

             But we need to call set_running here, since from the

             user/frontend's point of view, threads were set running.

             Unless we're calling an inferior function, as in that

             case we pretend the inferior doesn't run at all.  */

          if (!tp->control.in_infcall)

            set_running (user_visible_resume_ptid (entry_step), 1);

          discard_cleanups (old_cleanups);

          return;

        }



      /* Update pc to reflect the new address from which we will execute

         instructions due to displaced stepping.  */

      pc = regcache_read_pc (get_thread_regcache (inferior_ptid));



      displaced = get_displaced_stepping_state (ptid_get_pid (inferior_ptid));

      step = gdbarch_displaced_step_hw_singlestep (gdbarch,

                                                   displaced->step_closure);

    }



  /* Do we need to do it the hard way, w/temp breakpoints?  */

  else if (step)

    step = maybe_software_singlestep (gdbarch, pc);



  /* Currently, our software single-step implementation leads to different

     results than hardware single-stepping in one situation: when stepping

     into delivering a signal which has an associated signal handler,

     hardware single-step will stop at the first instruction of the handler,

     while software single-step will simply skip execution of the handler.



     For now, this difference in behavior is accepted since there is no

     easy way to actually implement single-stepping into a signal handler

     without kernel support.



     However, there is one scenario where this difference leads to follow-on

     problems: if we're stepping off a breakpoint by removing all breakpoints

     and then single-stepping.  In this case, the software single-step

     behavior means that even if there is a *breakpoint* in the signal

     handler, GDB still would not stop.



     Fortunately, we can at least fix this particular issue.  We detect

     here the case where we are about to deliver a signal while software

     single-stepping with breakpoints removed.  In this situation, we

     revert the decisions to remove all breakpoints and insert single-

     step breakpoints, and instead we install a step-resume breakpoint

     at the current address, deliver the signal without stepping, and

     once we arrive back at the step-resume breakpoint, actually step

     over the breakpoint we originally wanted to step over.  */

  if (thread_has_single_step_breakpoints_set (tp)

      && sig != GDB_SIGNAL_0

      && step_over_info_valid_p ())

    {

      /* If we have nested signals or a pending signal is delivered

         immediately after a handler returns, might might already have

         a step-resume breakpoint set on the earlier handler.  We cannot

         set another step-resume breakpoint; just continue on until the

         original breakpoint is hit.  */

      if (tp->control.step_resume_breakpoint == NULL)

        {

          insert_hp_step_resume_breakpoint_at_frame (get_current_frame ());

          tp->step_after_step_resume_breakpoint = 1;

        }



      delete_single_step_breakpoints (tp);



      clear_step_over_info ();

      tp->control.trap_expected = 0;



      insert_breakpoints ();

    }



  /* If STEP is set, it's a request to use hardware stepping

     facilities.  But in that case, we should never

     use singlestep breakpoint.  */

  gdb_assert (!(thread_has_single_step_breakpoints_set (tp) && step));



  /* Decide the set of threads to ask the target to resume.  Start

     by assuming everything will be resumed, than narrow the set

     by applying increasingly restricting conditions.  */

  resume_ptid = user_visible_resume_ptid (entry_step);



  /* Even if RESUME_PTID is a wildcard, and we end up resuming less

     (e.g., we might need to step over a breakpoint), from the

     user/frontend's point of view, all threads in RESUME_PTID are now

     running.  Unless we're calling an inferior function, as in that

     case pretend we inferior doesn't run at all.  */

  if (!tp->control.in_infcall)

    set_running (resume_ptid, 1);



  /* Maybe resume a single thread after all.  */

  if ((step || thread_has_single_step_breakpoints_set (tp))

      && tp->control.trap_expected)

    {

      /* We're allowing a thread to run past a breakpoint it has

         hit, by single-stepping the thread with the breakpoint

         removed.  In which case, we need to single-step only this

         thread, and keep others stopped, as they can miss this

         breakpoint if allowed to run.  */

      resume_ptid = inferior_ptid;

    }



  if (execution_direction != EXEC_REVERSE

      && step && breakpoint_inserted_here_p (aspace, pc))

    {

      /* The only case we currently need to step a breakpoint

         instruction is when we have a signal to deliver.  See

         handle_signal_stop where we handle random signals that could

         take out us out of the stepping range.  Normally, in that

         case we end up continuing (instead of stepping) over the

         signal handler with a breakpoint at PC, but there are cases

         where we should _always_ single-step, even if we have a

         step-resume breakpoint, like when a software watchpoint is

         set.  Assuming single-stepping and delivering a signal at the

         same time would takes us to the signal handler, then we could

         have removed the breakpoint at PC to step over it.  However,

         some hardware step targets (like e.g., Mac OS) can't step

         into signal handlers, and for those, we need to leave the

         breakpoint at PC inserted, as otherwise if the handler

         recurses and executes PC again, it'll miss the breakpoint.

         So we leave the breakpoint inserted anyway, but we need to

         record that we tried to step a breakpoint instruction, so

         that adjust_pc_after_break doesn't end up confused.  */

      gdb_assert (sig != GDB_SIGNAL_0);



      tp->stepped_breakpoint = 1;



      /* Most targets can step a breakpoint instruction, thus

         executing it normally.  But if this one cannot, just

         continue and we will hit it anyway.  */

      if (gdbarch_cannot_step_breakpoint (gdbarch))

        step = 0;

    }



  if (debug_displaced

      && use_displaced_stepping (gdbarch)

      && tp->control.trap_expected)

    {

      struct regcache *resume_regcache = get_thread_regcache (resume_ptid);

      struct gdbarch *resume_gdbarch = get_regcache_arch (resume_regcache);

      CORE_ADDR actual_pc = regcache_read_pc (resume_regcache);

      gdb_byte buf[4];



      fprintf_unfiltered (gdb_stdlog, "displaced: run %s: ",

                          paddress (resume_gdbarch, actual_pc));

      read_memory (actual_pc, buf, sizeof (buf));

      displaced_step_dump_bytes (gdb_stdlog, buf, sizeof (buf));

    }



  if (tp->control.may_range_step)

    {

      /* If we're resuming a thread with the PC out of the step

         range, then we're doing some nested/finer run control

         operation, like stepping the thread out of the dynamic

         linker or the displaced stepping scratch pad.  We

         shouldn't have allowed a range step then.  */

      gdb_assert (pc_in_thread_step_range (pc, tp));

    }



  /* Install inferior's terminal modes.  */

  target_terminal_inferior ();



  /* Avoid confusing the next resume, if the next stop/resume

     happens to apply to another thread.  */

  tp->suspend.stop_signal = GDB_SIGNAL_0;



  /* Advise target which signals may be handled silently.  If we have

     removed breakpoints because we are stepping over one (in any

     thread), we need to receive all signals to avoid accidentally

     skipping a breakpoint during execution of a signal handler.  */

  if (step_over_info_valid_p ())

    target_pass_signals (0, NULL);

  else

    target_pass_signals ((int) GDB_SIGNAL_LAST, signal_pass);



  target_resume (resume_ptid, step, sig);



  discard_cleanups (old_cleanups);

}



/* Proceeding.  */



/* Clear out all variables saying what to do when inferior is continued.

   First do this, then set the ones you want, then call `proceed'.  */



static void

‌clear_proceed_status_thread (struct thread_info *tp)

{

  if (debug_infrun)

    fprintf_unfiltered (gdb_stdlog,

                        "infrun: clear_proceed_status_thread (%s)\n",

                        target_pid_to_str (tp->ptid));



  /* If this signal should not be seen by program, give it zero.

     Used for debugging signals.  */

  if (!signal_pass_state (tp->suspend.stop_signal))

    tp->suspend.stop_signal = GDB_SIGNAL_0;



  tp->control.trap_expected = 0;

  tp->control.step_range_start = 0;

  tp->control.step_range_end = 0;

  tp->control.may_range_step = 0;

  tp->control.step_frame_id = null_frame_id;

  tp->control.step_stack_frame_id = null_frame_id;

  tp->control.step_over_calls = STEP_OVER_UNDEBUGGABLE;

  tp->stop_requested = 0;



  tp->control.stop_step = 0;



  tp->control.proceed_to_finish = 0;



  tp->control.command_interp = NULL;



  /* Discard any remaining commands or status from previous stop.  */

  bpstat_clear (&tp->control.stop_bpstat);

}



void

‌clear_proceed_status (int step)

{

  if (!non_stop)

    {

      struct thread_info *tp;

      ptid_t resume_ptid;



      resume_ptid = user_visible_resume_ptid (step);



      /* In all-stop mode, delete the per-thread status of all threads

         we're about to resume, implicitly and explicitly.  */

      ALL_NON_EXITED_THREADS (tp)

        {

          if (!ptid_match (tp->ptid, resume_ptid))

            continue;

          clear_proceed_status_thread (tp);

        }

    }



  if (!ptid_equal (inferior_ptid, null_ptid))

    {

      struct inferior *inferior;



      if (non_stop)

        {

          /* If in non-stop mode, only delete the per-thread status of

             the current thread.  */

          clear_proceed_status_thread (inferior_thread ());

        }



      inferior = current_inferior ();

      inferior->control.stop_soon = NO_STOP_QUIETLY;

    }



  stop_after_trap = 0;



  clear_step_over_info ();



  observer_notify_about_to_proceed ();



  if (stop_registers)

    {

      regcache_xfree (stop_registers);

      stop_registers = NULL;

    }

}



/* Returns true if TP is still stopped at a breakpoint that needs

   stepping-over in order to make progress.  If the breakpoint is gone

   meanwhile, we can skip the whole step-over dance.  */



static int

‌thread_still_needs_step_over (struct thread_info *tp)

{

  if (tp->stepping_over_breakpoint)

    {

      struct regcache *regcache = get_thread_regcache (tp->ptid);



      if (breakpoint_here_p (get_regcache_aspace (regcache),

                             regcache_read_pc (regcache))

          == ordinary_breakpoint_here)

        return 1;



      tp->stepping_over_breakpoint = 0;

    }



  return 0;

}



/* Returns true if scheduler locking applies.  STEP indicates whether

   we're about to do a step/next-like command to a thread.  */



static int

‌schedlock_applies (int step)

{

  return (scheduler_mode == schedlock_on

          || (scheduler_mode == schedlock_step

              && step));

}



/* Look a thread other than EXCEPT that has previously reported a

   breakpoint event, and thus needs a step-over in order to make

   progress.  Returns NULL is none is found.  STEP indicates whether

   we're about to step the current thread, in order to decide whether

   "set scheduler-locking step" applies.  */



static struct thread_info *

‌find_thread_needs_step_over (int step, struct thread_info *except)

{

  struct thread_info *tp, *current;



  /* With non-stop mode on, threads are always handled individually.  */

  gdb_assert (! non_stop);



  current = inferior_thread ();



  /* If scheduler locking applies, we can avoid iterating over all

     threads.  */

  if (schedlock_applies (step))

    {

      if (except != current

          && thread_still_needs_step_over (current))

        return current;



      return NULL;

    }



  ALL_NON_EXITED_THREADS (tp)

    {

      /* Ignore the EXCEPT thread.  */

      if (tp == except)

        continue;

      /* Ignore threads of processes we're not resuming.  */

      if (!sched_multi

          && ptid_get_pid (tp->ptid) != ptid_get_pid (inferior_ptid))

        continue;



      if (thread_still_needs_step_over (tp))

        return tp;

    }



  return NULL;

}



/* Basic routine for continuing the program in various fashions.



   ADDR is the address to resume at, or -1 for resume where stopped.

   SIGGNAL is the signal to give it, or 0 for none,

   or -1 for act according to how it stopped.

   STEP is nonzero if should trap after one instruction.

   -1 means return after that and print nothing.

   You should probably set various step_... variables

   before calling here, if you are stepping.



   You should call clear_proceed_status before calling proceed.  */



void

‌proceed (CORE_ADDR addr, enum gdb_signal siggnal, int step)

{

  struct regcache *regcache;

  struct gdbarch *gdbarch;

  struct thread_info *tp;

  CORE_ADDR pc;

  struct address_space *aspace;



  /* If we're stopped at a fork/vfork, follow the branch set by the

     "set follow-fork-mode" command; otherwise, we'll just proceed

     resuming the current thread.  */

  if (!follow_fork ())

    {

      /* The target for some reason decided not to resume.  */

      normal_stop ();

      if (target_can_async_p ())

        inferior_event_handler (INF_EXEC_COMPLETE, NULL);

      return;

    }



  /* We'll update this if & when we switch to a new thread.  */

  previous_inferior_ptid = inferior_ptid;



  regcache = get_current_regcache ();

  gdbarch = get_regcache_arch (regcache);

  aspace = get_regcache_aspace (regcache);

  pc = regcache_read_pc (regcache);

  tp = inferior_thread ();



  if (step > 0)

    step_start_function = find_pc_function (pc);

  if (step < 0)

    stop_after_trap = 1;



  /* Fill in with reasonable starting values.  */

  init_thread_stepping_state (tp);



  if (addr == (CORE_ADDR) -1)

    {

      if (pc == stop_pc

          && breakpoint_here_p (aspace, pc) == ordinary_breakpoint_here

          && execution_direction != EXEC_REVERSE)

        /* There is a breakpoint at the address we will resume at,

           step one instruction before inserting breakpoints so that

           we do not stop right away (and report a second hit at this

           breakpoint).



           Note, we don't do this in reverse, because we won't

           actually be executing the breakpoint insn anyway.

           We'll be (un-)executing the previous instruction.  */

        tp->stepping_over_breakpoint = 1;

      else if (gdbarch_single_step_through_delay_p (gdbarch)

               && gdbarch_single_step_through_delay (gdbarch,

                                                     get_current_frame ()))

        /* We stepped onto an instruction that needs to be stepped

           again before re-inserting the breakpoint, do so.  */

        tp->stepping_over_breakpoint = 1;

    }

  else

    {

      regcache_write_pc (regcache, addr);

    }



  if (siggnal != GDB_SIGNAL_DEFAULT)

    tp->suspend.stop_signal = siggnal;



  /* Record the interpreter that issued the execution command that

     caused this thread to resume.  If the top level interpreter is

     MI/async, and the execution command was a CLI command

     (next/step/etc.), we'll want to print stop event output to the MI

     console channel (the stepped-to line, etc.), as if the user

     entered the execution command on a real GDB console.  */

  inferior_thread ()->control.command_interp = command_interp ();



  if (debug_infrun)

    fprintf_unfiltered (gdb_stdlog,

                        "infrun: proceed (addr=%s, signal=%s, step=%d)\n",

                        paddress (gdbarch, addr),

                        gdb_signal_to_symbol_string (siggnal), step);



  if (non_stop)

    /* In non-stop, each thread is handled individually.  The context

       must already be set to the right thread here.  */

    ;

  else

    {

      struct thread_info *step_over;



      /* In a multi-threaded task we may select another thread and

         then continue or step.



         But if the old thread was stopped at a breakpoint, it will

         immediately cause another breakpoint stop without any

         execution (i.e. it will report a breakpoint hit incorrectly).

         So we must step over it first.



         Look for a thread other than the current (TP) that reported a

         breakpoint hit and hasn't been resumed yet since.  */

      step_over = find_thread_needs_step_over (step, tp);

      if (step_over != NULL)

        {

          if (debug_infrun)

            fprintf_unfiltered (gdb_stdlog,

                                "infrun: need to step-over [%s] first\n",

                                target_pid_to_str (step_over->ptid));



          /* Store the prev_pc for the stepping thread too, needed by

             switch_back_to_stepping thread.  */

          tp->prev_pc = regcache_read_pc (get_current_regcache ());

          switch_to_thread (step_over->ptid);

          tp = step_over;

        }

    }



  /* If we need to step over a breakpoint, and we're not using

     displaced stepping to do so, insert all breakpoints (watchpoints,

     etc.) but the one we're stepping over, step one instruction, and

     then re-insert the breakpoint when that step is finished.  */

  if (tp->stepping_over_breakpoint && !use_displaced_stepping (gdbarch))

    {

      struct regcache *regcache = get_current_regcache ();



      set_step_over_info (get_regcache_aspace (regcache),

                          regcache_read_pc (regcache), 0);

    }

  else

    clear_step_over_info ();



  insert_breakpoints ();



  tp->control.trap_expected = tp->stepping_over_breakpoint;



  annotate_starting ();



  /* Make sure that output from GDB appears before output from the

     inferior.  */

  gdb_flush (gdb_stdout);



  /* Refresh prev_pc value just prior to resuming.  This used to be

     done in stop_waiting, however, setting prev_pc there did not handle

     scenarios such as inferior function calls or returning from

     a function via the return command.  In those cases, the prev_pc

     value was not set properly for subsequent commands.  The prev_pc value

     is used to initialize the starting line number in the ecs.  With an

     invalid value, the gdb next command ends up stopping at the position

     represented by the next line table entry past our start position.

     On platforms that generate one line table entry per line, this

     is not a problem.  However, on the ia64, the compiler generates

     extraneous line table entries that do not increase the line number.

     When we issue the gdb next command on the ia64 after an inferior call

     or a return command, we often end up a few instructions forward, still

     within the original line we started.



     An attempt was made to refresh the prev_pc at the same time the

     execution_control_state is initialized (for instance, just before

     waiting for an inferior event).  But this approach did not work

     because of platforms that use ptrace, where the pc register cannot

     be read unless the inferior is stopped.  At that point, we are not

     guaranteed the inferior is stopped and so the regcache_read_pc() call

     can fail.  Setting the prev_pc value here ensures the value is updated

     correctly when the inferior is stopped.  */

  tp->prev_pc = regcache_read_pc (get_current_regcache ());



  /* Resume inferior.  */

  resume (tp->control.trap_expected || step || bpstat_should_step (),

          tp->suspend.stop_signal);



  /* Wait for it to stop (if not standalone)

     and in any case decode why it stopped, and act accordingly.  */

  /* Do this only if we are not using the event loop, or if the target

     does not support asynchronous execution.  */

  if (!target_can_async_p ())

    {

      wait_for_inferior ();

      normal_stop ();

    }

}





/* Start remote-debugging of a machine over a serial link.  */



void

‌start_remote (int from_tty)

{

  struct inferior *inferior;



  inferior = current_inferior ();

  inferior->control.stop_soon = STOP_QUIETLY_REMOTE;



  /* Always go on waiting for the target, regardless of the mode.  */

  /* FIXME: cagney/1999-09-23: At present it isn't possible to

     indicate to wait_for_inferior that a target should timeout if

     nothing is returned (instead of just blocking).  Because of this,

     targets expecting an immediate response need to, internally, set

     things up so that the target_wait() is forced to eventually

     timeout.  */

  /* FIXME: cagney/1999-09-24: It isn't possible for target_open() to

     differentiate to its caller what the state of the target is after

     the initial open has been performed.  Here we're assuming that

     the target has stopped.  It should be possible to eventually have

     target_open() return to the caller an indication that the target

     is currently running and GDB state should be set to the same as

     for an async run.  */

  wait_for_inferior ();



  /* Now that the inferior has stopped, do any bookkeeping like

     loading shared libraries.  We want to do this before normal_stop,

     so that the displayed frame is up to date.  */

  post_create_inferior (&current_target, from_tty);



  normal_stop ();

}



/* Initialize static vars when a new inferior begins.  */



void

‌init_wait_for_inferior (void)

{

  /* These are meaningless until the first time through wait_for_inferior.  */



  breakpoint_init_inferior (inf_starting);



  clear_proceed_status (0);



  target_last_wait_ptid = minus_one_ptid;



  previous_inferior_ptid = inferior_ptid;



  /* Discard any skipped inlined frames.  */

  clear_inline_frame_state (minus_one_ptid);

}





/* Data to be passed around while handling an event.  This data is

   discarded between events.  */

‌struct execution_control_state

{

  ptid_t ptid;

  /* The thread that got the event, if this was a thread event; NULL

     otherwise.  */

  struct thread_info *event_thread;



  struct target_waitstatus ws;

  int stop_func_filled_in;

  CORE_ADDR stop_func_start;

  CORE_ADDR stop_func_end;

  const char *stop_func_name;

  int wait_some_more;



  /* True if the event thread hit the single-step breakpoint of

     another thread.  Thus the event doesn't cause a stop, the thread

     needs to be single-stepped past the single-step breakpoint before

     we can switch back to the original stepping thread.  */

  int hit_singlestep_breakpoint;

};



static void handle_inferior_event (struct execution_control_state *ecs);



static void handle_step_into_function (struct gdbarch *gdbarch,

                                       struct execution_control_state *ecs);

static void handle_step_into_function_backward (struct gdbarch *gdbarch,

                                                struct execution_control_state *ecs);

static void handle_signal_stop (struct execution_control_state *ecs);

static void check_exception_resume (struct execution_control_state *,

                                    struct frame_info *);



static void end_stepping_range (struct execution_control_state *ecs);

static void stop_waiting (struct execution_control_state *ecs);

static void prepare_to_wait (struct execution_control_state *ecs);

static void keep_going (struct execution_control_state *ecs);

static void process_event_stop_test (struct execution_control_state *ecs);

static int switch_back_to_stepped_thread (struct execution_control_state *ecs);



/* Callback for iterate over threads.  If the thread is stopped, but

   the user/frontend doesn't know about that yet, go through

   normal_stop, as if the thread had just stopped now.  ARG points at

   a ptid.  If PTID is MINUS_ONE_PTID, applies to all threads.  If

   ptid_is_pid(PTID) is true, applies to all threads of the process

   pointed at by PTID.  Otherwise, apply only to the thread pointed by

   PTID.  */



static int

‌infrun_thread_stop_requested_callback (struct thread_info *info, void *arg)

{

  ptid_t ptid = * (ptid_t *) arg;



  if ((ptid_equal (info->ptid, ptid)

       || ptid_equal (minus_one_ptid, ptid)

       || (ptid_is_pid (ptid)

           && ptid_get_pid (ptid) == ptid_get_pid (info->ptid)))

      && is_running (info->ptid)

      && !is_executing (info->ptid))

    {

      struct cleanup *old_chain;

      struct execution_control_state ecss;

      struct execution_control_state *ecs = &ecss;



      memset (ecs, 0, sizeof (*ecs));



      old_chain = make_cleanup_restore_current_thread ();



      overlay_cache_invalid = 1;

      /* Flush target cache before starting to handle each event.

         Target was running and cache could be stale.  This is just a

         heuristic.  Running threads may modify target memory, but we

         don't get any event.  */

      target_dcache_invalidate ();



      /* Go through handle_inferior_event/normal_stop, so we always

         have consistent output as if the stop event had been

         reported.  */

      ecs->ptid = info->ptid;

      ecs->event_thread = find_thread_ptid (info->ptid);

      ecs->ws.kind = TARGET_WAITKIND_STOPPED;

      ecs->ws.value.sig = GDB_SIGNAL_0;



      handle_inferior_event (ecs);



      if (!ecs->wait_some_more)

        {

          struct thread_info *tp;



          normal_stop ();



          /* Finish off the continuations.  */

          tp = inferior_thread ();

          do_all_intermediate_continuations_thread (tp, 1);

          do_all_continuations_thread (tp, 1);

        }



      do_cleanups (old_chain);

    }



  return 0;

}



/* This function is attached as a "thread_stop_requested" observer.

   Cleanup local state that assumed the PTID was to be resumed, and

   report the stop to the frontend.  */



static void

‌infrun_thread_stop_requested (ptid_t ptid)

{

  struct displaced_step_inferior_state *displaced;



  /* PTID was requested to stop.  Remove it from the displaced

     stepping queue, so we don't try to resume it automatically.  */



  for (displaced = displaced_step_inferior_states;

       displaced;

       displaced = displaced->next)

    {

      struct displaced_step_request *it, **prev_next_p;



      it = displaced->step_request_queue;

      prev_next_p = &displaced->step_request_queue;

      while (it)

        {

          if (ptid_match (it->ptid, ptid))

            {

              *prev_next_p = it->next;

              it->next = NULL;

              xfree (it);

            }

          else

            {

              prev_next_p = &it->next;

            }



          it = *prev_next_p;

        }

    }



  iterate_over_threads (infrun_thread_stop_requested_callback, &ptid);

}



static void

‌infrun_thread_thread_exit (struct thread_info *tp, int silent)

{

  if (ptid_equal (target_last_wait_ptid, tp->ptid))

    nullify_last_target_wait_ptid ();

}



/* Delete the step resume, single-step and longjmp/exception resume

   breakpoints of TP.  */



static void

‌delete_thread_infrun_breakpoints (struct thread_info *tp)

{

  delete_step_resume_breakpoint (tp);

  delete_exception_resume_breakpoint (tp);

  delete_single_step_breakpoints (tp);

}



/* If the target still has execution, call FUNC for each thread that

   just stopped.  In all-stop, that's all the non-exited threads; in

   non-stop, that's the current thread, only.  */



‌typedef void (*for_each_just_stopped_thread_callback_func)

  (struct thread_info *tp);



static void

‌for_each_just_stopped_thread (for_each_just_stopped_thread_callback_func func)

{

  if (!target_has_execution || ptid_equal (inferior_ptid, null_ptid))

    return;



  if (non_stop)

    {

      /* If in non-stop mode, only the current thread stopped.  */

      func (inferior_thread ());

    }

  else

    {

      struct thread_info *tp;



      /* In all-stop mode, all threads have stopped.  */

      ALL_NON_EXITED_THREADS (tp)

        {

          func (tp);

        }

    }

}



/* Delete the step resume and longjmp/exception resume breakpoints of

   the threads that just stopped.  */



static void

‌delete_just_stopped_threads_infrun_breakpoints (void)

{

  for_each_just_stopped_thread (delete_thread_infrun_breakpoints);

}



/* Delete the single-step breakpoints of the threads that just

   stopped.  */



static void

‌delete_just_stopped_threads_single_step_breakpoints (void)

{

  for_each_just_stopped_thread (delete_single_step_breakpoints);

}



/* A cleanup wrapper.  */



static void

‌delete_just_stopped_threads_infrun_breakpoints_cleanup (void *arg)

{

  delete_just_stopped_threads_infrun_breakpoints ();

}



/* Pretty print the results of target_wait, for debugging purposes.  */



static void

‌print_target_wait_results (ptid_t waiton_ptid, ptid_t result_ptid,

                           const struct target_waitstatus *ws)

{

  char *status_string = target_waitstatus_to_string (ws);

  struct ui_file *tmp_stream = mem_fileopen ();

  char *text;



  /* The text is split over several lines because it was getting too long.

     Call fprintf_unfiltered (gdb_stdlog) once so that the text is still

     output as a unit; we want only one timestamp printed if debug_timestamp

     is set.  */



  fprintf_unfiltered (tmp_stream,

                      "infrun: target_wait (%d", ptid_get_pid (waiton_ptid));

  if (ptid_get_pid (waiton_ptid) != -1)

    fprintf_unfiltered (tmp_stream,

                        " [%s]", target_pid_to_str (waiton_ptid));

  fprintf_unfiltered (tmp_stream, ", status) =\n");

  fprintf_unfiltered (tmp_stream,

                      "infrun:   %d [%s],\n",

                      ptid_get_pid (result_ptid),

                      target_pid_to_str (result_ptid));

  fprintf_unfiltered (tmp_stream,

                      "infrun:   %s\n",

                      status_string);



  text = ui_file_xstrdup (tmp_stream, NULL);



  /* This uses %s in part to handle %'s in the text, but also to avoid

     a gcc error: the format attribute requires a string literal.  */

  fprintf_unfiltered (gdb_stdlog, "%s", text);



  xfree (status_string);

  xfree (text);

  ui_file_delete (tmp_stream);

}



/* Prepare and stabilize the inferior for detaching it.  E.g.,

   detaching while a thread is displaced stepping is a recipe for

   crashing it, as nothing would readjust the PC out of the scratch

   pad.  */



void

‌prepare_for_detach (void)

{

  struct inferior *inf = current_inferior ();

  ptid_t pid_ptid = pid_to_ptid (inf->pid);

  struct cleanup *old_chain_1;

  struct displaced_step_inferior_state *displaced;



  displaced = get_displaced_stepping_state (inf->pid);



  /* Is any thread of this process displaced stepping?  If not,

     there's nothing else to do.  */

  if (displaced == NULL || ptid_equal (displaced->step_ptid, null_ptid))

    return;



  if (debug_infrun)

    fprintf_unfiltered (gdb_stdlog,

                        "displaced-stepping in-process while detaching");



  old_chain_1 = make_cleanup_restore_integer (&inf->detaching);

  inf->detaching = 1;



  while (!ptid_equal (displaced->step_ptid, null_ptid))

    {

      struct cleanup *old_chain_2;

      struct execution_control_state ecss;

      struct execution_control_state *ecs;



      ecs = &ecss;

      memset (ecs, 0, sizeof (*ecs));



      overlay_cache_invalid = 1;

      /* Flush target cache before starting to handle each event.

         Target was running and cache could be stale.  This is just a

         heuristic.  Running threads may modify target memory, but we

         don't get any event.  */

      target_dcache_invalidate ();



      if (deprecated_target_wait_hook)

        ecs->ptid = deprecated_target_wait_hook (pid_ptid, &ecs->ws, 0);

      else

        ecs->ptid = target_wait (pid_ptid, &ecs->ws, 0);



      if (debug_infrun)

        print_target_wait_results (pid_ptid, ecs->ptid, &ecs->ws);



      /* If an error happens while handling the event, propagate GDB's

         knowledge of the executing state to the frontend/user running

         state.  */

      old_chain_2 = make_cleanup (finish_thread_state_cleanup,

                                  &minus_one_ptid);



      /* Now figure out what to do with the result of the result.  */

      handle_inferior_event (ecs);



      /* No error, don't finish the state yet.  */

      discard_cleanups (old_chain_2);



      /* Breakpoints and watchpoints are not installed on the target

         at this point, and signals are passed directly to the

         inferior, so this must mean the process is gone.  */

      if (!ecs->wait_some_more)

        {

          discard_cleanups (old_chain_1);

          error (_("Program exited while detaching"));

        }

    }



  discard_cleanups (old_chain_1);

}



/* Wait for control to return from inferior to debugger.



   If inferior gets a signal, we may decide to start it up again

   instead of returning.  That is why there is a loop in this function.

   When this function actually returns it means the inferior

   should be left stopped and GDB should read more commands.  */



void

‌wait_for_inferior (void)

{

  struct cleanup *old_cleanups;



  if (debug_infrun)

    fprintf_unfiltered

      (gdb_stdlog, "infrun: wait_for_inferior ()\n");



  old_cleanups

    = make_cleanup (delete_just_stopped_threads_infrun_breakpoints_cleanup,

                    NULL);



  while (1)

    {

      struct execution_control_state ecss;

      struct execution_control_state *ecs = &ecss;

      struct cleanup *old_chain;

      ptid_t waiton_ptid = minus_one_ptid;



      memset (ecs, 0, sizeof (*ecs));



      overlay_cache_invalid = 1;



      /* Flush target cache before starting to handle each event.

         Target was running and cache could be stale.  This is just a

         heuristic.  Running threads may modify target memory, but we

         don't get any event.  */

      target_dcache_invalidate ();



      if (deprecated_target_wait_hook)

        ecs->ptid = deprecated_target_wait_hook (waiton_ptid, &ecs->ws, 0);

      else

        ecs->ptid = target_wait (waiton_ptid, &ecs->ws, 0);



      if (debug_infrun)

        print_target_wait_results (waiton_ptid, ecs->ptid, &ecs->ws);



      /* If an error happens while handling the event, propagate GDB's

         knowledge of the executing state to the frontend/user running

         state.  */

      old_chain = make_cleanup (finish_thread_state_cleanup, &minus_one_ptid);



      /* Now figure out what to do with the result of the result.  */

      handle_inferior_event (ecs);



      /* No error, don't finish the state yet.  */

      discard_cleanups (old_chain);



      if (!ecs->wait_some_more)

        break;

    }



  do_cleanups (old_cleanups);

}



/* Cleanup that reinstalls the readline callback handler, if the

   target is running in the background.  If while handling the target

   event something triggered a secondary prompt, like e.g., a

   pagination prompt, we'll have removed the callback handler (see

   gdb_readline_wrapper_line).  Need to do this as we go back to the

   event loop, ready to process further input.  Note this has no

   effect if the handler hasn't actually been removed, because calling

   rl_callback_handler_install resets the line buffer, thus losing

   input.  */



static void

‌reinstall_readline_callback_handler_cleanup (void *arg)

{

  if (async_command_editing_p && !sync_execution)

    gdb_rl_callback_handler_reinstall ();

}



/* Asynchronous version of wait_for_inferior.  It is called by the

   event loop whenever a change of state is detected on the file

   descriptor corresponding to the target.  It can be called more than

   once to complete a single execution command.  In such cases we need

   to keep the state in a global variable ECSS.  If it is the last time

   that this function is called for a single execution command, then

   report to the user that the inferior has stopped, and do the

   necessary cleanups.  */



void

‌fetch_inferior_event (void *client_data)

{

  struct execution_control_state ecss;

  struct execution_control_state *ecs = &ecss;

  struct cleanup *old_chain = make_cleanup (null_cleanup, NULL);

  struct cleanup *ts_old_chain;

  int was_sync = sync_execution;

  int cmd_done = 0;

  ptid_t waiton_ptid = minus_one_ptid;



  memset (ecs, 0, sizeof (*ecs));



  /* End up with readline processing input, if necessary.  */

  make_cleanup (reinstall_readline_callback_handler_cleanup, NULL);



  /* We're handling a live event, so make sure we're doing live

     debugging.  If we're looking at traceframes while the target is

     running, we're going to need to get back to that mode after

     handling the event.  */

  if (non_stop)

    {

      make_cleanup_restore_current_traceframe ();

      set_current_traceframe (-1);

    }



  if (non_stop)

    /* In non-stop mode, the user/frontend should not notice a thread

       switch due to internal events.  Make sure we reverse to the

       user selected thread and frame after handling the event and

       running any breakpoint commands.  */

    make_cleanup_restore_current_thread ();



  overlay_cache_invalid = 1;

  /* Flush target cache before starting to handle each event.  Target

     was running and cache could be stale.  This is just a heuristic.

     Running threads may modify target memory, but we don't get any

     event.  */

  target_dcache_invalidate ();



  make_cleanup_restore_integer (&execution_direction);

  execution_direction = target_execution_direction ();



  if (deprecated_target_wait_hook)

    ecs->ptid =

      deprecated_target_wait_hook (waiton_ptid, &ecs->ws, TARGET_WNOHANG);

  else

    ecs->ptid = target_wait (waiton_ptid, &ecs->ws, TARGET_WNOHANG);



  if (debug_infrun)

    print_target_wait_results (waiton_ptid, ecs->ptid, &ecs->ws);



  /* If an error happens while handling the event, propagate GDB's

     knowledge of the executing state to the frontend/user running

     state.  */

  if (!non_stop)

    ts_old_chain = make_cleanup (finish_thread_state_cleanup, &minus_one_ptid);

  else

    ts_old_chain = make_cleanup (finish_thread_state_cleanup, &ecs->ptid);



  /* Get executed before make_cleanup_restore_current_thread above to apply

     still for the thread which has thrown the exception.  */

  make_bpstat_clear_actions_cleanup ();



  make_cleanup (delete_just_stopped_threads_infrun_breakpoints_cleanup, NULL);



  /* Now figure out what to do with the result of the result.  */

  handle_inferior_event (ecs);



  if (!ecs->wait_some_more)

    {

      struct inferior *inf = find_inferior_ptid (ecs->ptid);



      delete_just_stopped_threads_infrun_breakpoints ();



      /* We may not find an inferior if this was a process exit.  */

      if (inf == NULL || inf->control.stop_soon == NO_STOP_QUIETLY)

        normal_stop ();



      if (target_has_execution

          && ecs->ws.kind != TARGET_WAITKIND_NO_RESUMED

          && ecs->ws.kind != TARGET_WAITKIND_EXITED

          && ecs->ws.kind != TARGET_WAITKIND_SIGNALLED

          && ecs->event_thread->step_multi

          && ecs->event_thread->control.stop_step)

        inferior_event_handler (INF_EXEC_CONTINUE, NULL);

      else

        {

          inferior_event_handler (INF_EXEC_COMPLETE, NULL);

          cmd_done = 1;

        }

    }



  /* No error, don't finish the thread states yet.  */

  discard_cleanups (ts_old_chain);



  /* Revert thread and frame.  */

  do_cleanups (old_chain);



  /* If the inferior was in sync execution mode, and now isn't,

     restore the prompt (a synchronous execution command has finished,

     and we're ready for input).  */

  if (interpreter_async && was_sync && !sync_execution)

    observer_notify_sync_execution_done ();



  if (cmd_done

      && !was_sync

      && exec_done_display_p

      && (ptid_equal (inferior_ptid, null_ptid)

          || !is_running (inferior_ptid)))

    printf_unfiltered (_("completed.\n"));

}



/* Record the frame and location we're currently stepping through.  */

void

‌set_step_info (struct frame_info *frame, struct symtab_and_line sal)

{

  struct thread_info *tp = inferior_thread ();



  tp->control.step_frame_id = get_frame_id (frame);

  tp->control.step_stack_frame_id = get_stack_frame_id (frame);



  tp->current_symtab = sal.symtab;

  tp->current_line = sal.line;

}



/* Clear context switchable stepping state.  */



void

‌init_thread_stepping_state (struct thread_info *tss)

{

  tss->stepped_breakpoint = 0;

  tss->stepping_over_breakpoint = 0;

  tss->stepping_over_watchpoint = 0;

  tss->step_after_step_resume_breakpoint = 0;

}



/* Set the cached copy of the last ptid/waitstatus.  */



static void

‌set_last_target_status (ptid_t ptid, struct target_waitstatus status)

{

  target_last_wait_ptid = ptid;

  target_last_waitstatus = status;

}



/* Return the cached copy of the last pid/waitstatus returned by

   target_wait()/deprecated_target_wait_hook().  The data is actually

   cached by handle_inferior_event(), which gets called immediately

   after target_wait()/deprecated_target_wait_hook().  */



void

‌get_last_target_status (ptid_t *ptidp, struct target_waitstatus *status)

{

  *ptidp = target_last_wait_ptid;

  *status = target_last_waitstatus;

}



void

‌nullify_last_target_wait_ptid (void)

{

  target_last_wait_ptid = minus_one_ptid;

}



/* Switch thread contexts.  */



static void

‌context_switch (ptid_t ptid)

{

  if (debug_infrun && !ptid_equal (ptid, inferior_ptid))

    {

      fprintf_unfiltered (gdb_stdlog, "infrun: Switching context from %s ",

                          target_pid_to_str (inferior_ptid));

      fprintf_unfiltered (gdb_stdlog, "to %s\n",

                          target_pid_to_str (ptid));

    }



  switch_to_thread (ptid);

}



static void

‌adjust_pc_after_break (struct execution_control_state *ecs)

{

  struct regcache *regcache;

  struct gdbarch *gdbarch;

  struct address_space *aspace;

  CORE_ADDR breakpoint_pc, decr_pc;



  /* If we've hit a breakpoint, we'll normally be stopped with SIGTRAP.  If

     we aren't, just return.



     We assume that waitkinds other than TARGET_WAITKIND_STOPPED are not

     affected by gdbarch_decr_pc_after_break.  Other waitkinds which are

     implemented by software breakpoints should be handled through the normal

     breakpoint layer.



     NOTE drow/2004-01-31: On some targets, breakpoints may generate

     different signals (SIGILL or SIGEMT for instance), but it is less

     clear where the PC is pointing afterwards.  It may not match

     gdbarch_decr_pc_after_break.  I don't know any specific target that

     generates these signals at breakpoints (the code has been in GDB since at

     least 1992) so I can not guess how to handle them here.



     In earlier versions of GDB, a target with

     gdbarch_have_nonsteppable_watchpoint would have the PC after hitting a

     watchpoint affected by gdbarch_decr_pc_after_break.  I haven't found any

     target with both of these set in GDB history, and it seems unlikely to be

     correct, so gdbarch_have_nonsteppable_watchpoint is not checked here.  */



  if (ecs->ws.kind != TARGET_WAITKIND_STOPPED)

    return;



  if (ecs->ws.value.sig != GDB_SIGNAL_TRAP)

    return;



  /* In reverse execution, when a breakpoint is hit, the instruction

     under it has already been de-executed.  The reported PC always

     points at the breakpoint address, so adjusting it further would

     be wrong.  E.g., consider this case on a decr_pc_after_break == 1

     architecture:



       B1         0x08000000 :   INSN1

       B2         0x08000001 :   INSN2

                  0x08000002 :   INSN3

            PC -> 0x08000003 :   INSN4



     Say you're stopped at 0x08000003 as above.  Reverse continuing

     from that point should hit B2 as below.  Reading the PC when the

     SIGTRAP is reported should read 0x08000001 and INSN2 should have

     been de-executed already.



       B1         0x08000000 :   INSN1

       B2   PC -> 0x08000001 :   INSN2

                  0x08000002 :   INSN3

                  0x08000003 :   INSN4



     We can't apply the same logic as for forward execution, because

     we would wrongly adjust the PC to 0x08000000, since there's a

     breakpoint at PC - 1.  We'd then report a hit on B1, although

     INSN1 hadn't been de-executed yet.  Doing nothing is the correct

     behaviour.  */

  if (execution_direction == EXEC_REVERSE)

    return;



  /* If this target does not decrement the PC after breakpoints, then

     we have nothing to do.  */

  regcache = get_thread_regcache (ecs->ptid);

  gdbarch = get_regcache_arch (regcache);



  decr_pc = target_decr_pc_after_break (gdbarch);

  if (decr_pc == 0)

    return;



  aspace = get_regcache_aspace (regcache);



  /* Find the location where (if we've hit a breakpoint) the

     breakpoint would be.  */

  breakpoint_pc = regcache_read_pc (regcache) - decr_pc;



  /* Check whether there actually is a software breakpoint inserted at

     that location.



     If in non-stop mode, a race condition is possible where we've

     removed a breakpoint, but stop events for that breakpoint were

     already queued and arrive later.  To suppress those spurious

     SIGTRAPs, we keep a list of such breakpoint locations for a bit,

     and retire them after a number of stop events are reported.  */

  if (software_breakpoint_inserted_here_p (aspace, breakpoint_pc)

      || (non_stop && moribund_breakpoint_here_p (aspace, breakpoint_pc)))

    {

      struct cleanup *old_cleanups = make_cleanup (null_cleanup, NULL);



      if (record_full_is_used ())

        record_full_gdb_operation_disable_set ();



      /* When using hardware single-step, a SIGTRAP is reported for both

         a completed single-step and a software breakpoint.  Need to

         differentiate between the two, as the latter needs adjusting

         but the former does not.



         The SIGTRAP can be due to a completed hardware single-step only if

          - we didn't insert software single-step breakpoints

          - the thread to be examined is still the current thread

          - this thread is currently being stepped



         If any of these events did not occur, we must have stopped due

         to hitting a software breakpoint, and have to back up to the

         breakpoint address.



         As a special case, we could have hardware single-stepped a

         software breakpoint.  In this case (prev_pc == breakpoint_pc),

         we also need to back up to the breakpoint address.  */



      if (thread_has_single_step_breakpoints_set (ecs->event_thread)

          || !ptid_equal (ecs->ptid, inferior_ptid)

          || !currently_stepping (ecs->event_thread)

          || (ecs->event_thread->stepped_breakpoint

              && ecs->event_thread->prev_pc == breakpoint_pc))

        regcache_write_pc (regcache, breakpoint_pc);



      do_cleanups (old_cleanups);

    }

}



static int

‌stepped_in_from (struct frame_info *frame, struct frame_id step_frame_id)

{

  for (frame = get_prev_frame (frame);

       frame != NULL;

       frame = get_prev_frame (frame))

    {

      if (frame_id_eq (get_frame_id (frame), step_frame_id))

        return 1;

      if (get_frame_type (frame) != INLINE_FRAME)

        break;

    }



  return 0;

}



/* Auxiliary function that handles syscall entry/return events.

   It returns 1 if the inferior should keep going (and GDB

   should ignore the event), or 0 if the event deserves to be

   processed.  */



static int

‌handle_syscall_event (struct execution_control_state *ecs)

{

  struct regcache *regcache;

  int syscall_number;



  if (!ptid_equal (ecs->ptid, inferior_ptid))

    context_switch (ecs->ptid);



  regcache = get_thread_regcache (ecs->ptid);

  syscall_number = ecs->ws.value.syscall_number;

  stop_pc = regcache_read_pc (regcache);



  if (catch_syscall_enabled () > 0

      && catching_syscall_number (syscall_number) > 0)

    {

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog, "infrun: syscall number = '%d'\n",

                            syscall_number);



      ecs->event_thread->control.stop_bpstat

        = bpstat_stop_status (get_regcache_aspace (regcache),

                              stop_pc, ecs->ptid, &ecs->ws);



      if (bpstat_causes_stop (ecs->event_thread->control.stop_bpstat))

        {

          /* Catchpoint hit.  */

          return 0;

        }

    }



  /* If no catchpoint triggered for this, then keep going.  */

  keep_going (ecs);

  return 1;

}



/* Lazily fill in the execution_control_state's stop_func_* fields.  */



static void

‌fill_in_stop_func (struct gdbarch *gdbarch,

                   struct execution_control_state *ecs)

{

  if (!ecs->stop_func_filled_in)

    {

      /* Don't care about return value; stop_func_start and stop_func_name

         will both be 0 if it doesn't work.  */

      find_pc_partial_function (stop_pc, &ecs->stop_func_name,

                                &ecs->stop_func_start, &ecs->stop_func_end);

      ecs->stop_func_start

        += gdbarch_deprecated_function_start_offset (gdbarch);



      if (gdbarch_skip_entrypoint_p (gdbarch))

        ecs->stop_func_start = gdbarch_skip_entrypoint (gdbarch,

                                                        ecs->stop_func_start);



      ecs->stop_func_filled_in = 1;

    }

}





/* Return the STOP_SOON field of the inferior pointed at by PTID.  */



static enum stop_kind

‌get_inferior_stop_soon (ptid_t ptid)

{

  struct inferior *inf = find_inferior_ptid (ptid);



  gdb_assert (inf != NULL);

  return inf->control.stop_soon;

}



/* Given an execution control state that has been freshly filled in by

   an event from the inferior, figure out what it means and take

   appropriate action.



   The alternatives are:



   1) stop_waiting and return; to really stop and return to the

   debugger.



   2) keep_going and return; to wait for the next event (set

   ecs->event_thread->stepping_over_breakpoint to 1 to single step

   once).  */



static void

‌handle_inferior_event (struct execution_control_state *ecs)

{

  enum stop_kind stop_soon;



  if (ecs->ws.kind == TARGET_WAITKIND_IGNORE)

    {

      /* We had an event in the inferior, but we are not interested in

         handling it at this level.  The lower layers have already

         done what needs to be done, if anything.



         One of the possible circumstances for this is when the

         inferior produces output for the console.  The inferior has

         not stopped, and we are ignoring the event.  Another possible

         circumstance is any event which the lower level knows will be

         reported multiple times without an intervening resume.  */

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog, "infrun: TARGET_WAITKIND_IGNORE\n");

      prepare_to_wait (ecs);

      return;

    }



  if (ecs->ws.kind == TARGET_WAITKIND_NO_RESUMED

      && target_can_async_p () && !sync_execution)

    {

      /* There were no unwaited-for children left in the target, but,

         we're not synchronously waiting for events either.  Just

         ignore.  Otherwise, if we were running a synchronous

         execution command, we need to cancel it and give the user

         back the terminal.  */

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog,

                            "infrun: TARGET_WAITKIND_NO_RESUMED (ignoring)\n");

      prepare_to_wait (ecs);

      return;

    }



  /* Cache the last pid/waitstatus.  */

  set_last_target_status (ecs->ptid, ecs->ws);



  /* Always clear state belonging to the previous time we stopped.  */

  stop_stack_dummy = STOP_NONE;



  if (ecs->ws.kind == TARGET_WAITKIND_NO_RESUMED)

    {

      /* No unwaited-for children left.  IOW, all resumed children

         have exited.  */

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog, "infrun: TARGET_WAITKIND_NO_RESUMED\n");



      stop_print_frame = 0;

      stop_waiting (ecs);

      return;

    }



  if (ecs->ws.kind != TARGET_WAITKIND_EXITED

      && ecs->ws.kind != TARGET_WAITKIND_SIGNALLED)

    {

      ecs->event_thread = find_thread_ptid (ecs->ptid);

      /* If it's a new thread, add it to the thread database.  */

      if (ecs->event_thread == NULL)

        ecs->event_thread = add_thread (ecs->ptid);



      /* Disable range stepping.  If the next step request could use a

         range, this will be end up re-enabled then.  */

      ecs->event_thread->control.may_range_step = 0;

    }



  /* Dependent on valid ECS->EVENT_THREAD.  */

  adjust_pc_after_break (ecs);



  /* Dependent on the current PC value modified by adjust_pc_after_break.  */

  reinit_frame_cache ();



  breakpoint_retire_moribund ();



  /* First, distinguish signals caused by the debugger from signals

     that have to do with the program's own actions.  Note that

     breakpoint insns may cause SIGTRAP or SIGILL or SIGEMT, depending

     on the operating system version.  Here we detect when a SIGILL or

     SIGEMT is really a breakpoint and change it to SIGTRAP.  We do

     something similar for SIGSEGV, since a SIGSEGV will be generated

     when we're trying to execute a breakpoint instruction on a

     non-executable stack.  This happens for call dummy breakpoints

     for architectures like SPARC that place call dummies on the

     stack.  */

  if (ecs->ws.kind == TARGET_WAITKIND_STOPPED

      && (ecs->ws.value.sig == GDB_SIGNAL_ILL

          || ecs->ws.value.sig == GDB_SIGNAL_SEGV

          || ecs->ws.value.sig == GDB_SIGNAL_EMT))

    {

      struct regcache *regcache = get_thread_regcache (ecs->ptid);



      if (breakpoint_inserted_here_p (get_regcache_aspace (regcache),

                                      regcache_read_pc (regcache)))

        {

          if (debug_infrun)

            fprintf_unfiltered (gdb_stdlog,

                                "infrun: Treating signal as SIGTRAP\n");

          ecs->ws.value.sig = GDB_SIGNAL_TRAP;

        }

    }



  /* Mark the non-executing threads accordingly.  In all-stop, all

     threads of all processes are stopped when we get any event

     reported.  In non-stop mode, only the event thread stops.  If

     we're handling a process exit in non-stop mode, there's nothing

     to do, as threads of the dead process are gone, and threads of

     any other process were left running.  */

  if (!non_stop)

    set_executing (minus_one_ptid, 0);

  else if (ecs->ws.kind != TARGET_WAITKIND_SIGNALLED

           && ecs->ws.kind != TARGET_WAITKIND_EXITED)

    set_executing (ecs->ptid, 0);



  switch (ecs->ws.kind)

    {

    case TARGET_WAITKIND_LOADED:

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog, "infrun: TARGET_WAITKIND_LOADED\n");

      if (!ptid_equal (ecs->ptid, inferior_ptid))

        context_switch (ecs->ptid);

      /* Ignore gracefully during startup of the inferior, as it might

         be the shell which has just loaded some objects, otherwise

         add the symbols for the newly loaded objects.  Also ignore at

         the beginning of an attach or remote session; we will query

         the full list of libraries once the connection is

         established.  */



      stop_soon = get_inferior_stop_soon (ecs->ptid);

      if (stop_soon == NO_STOP_QUIETLY)

        {

          struct regcache *regcache;



          regcache = get_thread_regcache (ecs->ptid);



          handle_solib_event ();



          ecs->event_thread->control.stop_bpstat

            = bpstat_stop_status (get_regcache_aspace (regcache),

                                  stop_pc, ecs->ptid, &ecs->ws);



          if (bpstat_causes_stop (ecs->event_thread->control.stop_bpstat))

            {

              /* A catchpoint triggered.  */

              process_event_stop_test (ecs);

              return;

            }



          /* If requested, stop when the dynamic linker notifies

             gdb of events.  This allows the user to get control

             and place breakpoints in initializer routines for

             dynamically loaded objects (among other things).  */

          ecs->event_thread->suspend.stop_signal = GDB_SIGNAL_0;

          if (stop_on_solib_events)

            {

              /* Make sure we print "Stopped due to solib-event" in

                 normal_stop.  */

              stop_print_frame = 1;



              stop_waiting (ecs);

              return;

            }

        }



      /* If we are skipping through a shell, or through shared library

         loading that we aren't interested in, resume the program.  If

         we're running the program normally, also resume.  */

      if (stop_soon == STOP_QUIETLY || stop_soon == NO_STOP_QUIETLY)

        {

          /* Loading of shared libraries might have changed breakpoint

             addresses.  Make sure new breakpoints are inserted.  */

          if (stop_soon == NO_STOP_QUIETLY)

            insert_breakpoints ();

          resume (0, GDB_SIGNAL_0);

          prepare_to_wait (ecs);

          return;

        }



      /* But stop if we're attaching or setting up a remote

         connection.  */

      if (stop_soon == STOP_QUIETLY_NO_SIGSTOP

          || stop_soon == STOP_QUIETLY_REMOTE)

        {

          if (debug_infrun)

            fprintf_unfiltered (gdb_stdlog, "infrun: quietly stopped\n");

          stop_waiting (ecs);

          return;

        }



      internal_error (__FILE__, __LINE__,

                      _("unhandled stop_soon: %d"), (int) stop_soon);



    case TARGET_WAITKIND_SPURIOUS:

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog, "infrun: TARGET_WAITKIND_SPURIOUS\n");

      if (!ptid_equal (ecs->ptid, inferior_ptid))

        context_switch (ecs->ptid);

      resume (0, GDB_SIGNAL_0);

      prepare_to_wait (ecs);

      return;



    case TARGET_WAITKIND_EXITED:

    case TARGET_WAITKIND_SIGNALLED:

      if (debug_infrun)

        {

          if (ecs->ws.kind == TARGET_WAITKIND_EXITED)

            fprintf_unfiltered (gdb_stdlog,

                                "infrun: TARGET_WAITKIND_EXITED\n");

          else

            fprintf_unfiltered (gdb_stdlog,

                                "infrun: TARGET_WAITKIND_SIGNALLED\n");

        }



      inferior_ptid = ecs->ptid;

      set_current_inferior (find_inferior_ptid (ecs->ptid));

      set_current_program_space (current_inferior ()->pspace);

      handle_vfork_child_exec_or_exit (0);

      target_terminal_ours ();        /* Must do this before mourn anyway.  */



      /* Clearing any previous state of convenience variables.  */

      clear_exit_convenience_vars ();



      if (ecs->ws.kind == TARGET_WAITKIND_EXITED)

        {

          /* Record the exit code in the convenience variable $_exitcode, so

             that the user can inspect this again later.  */

          set_internalvar_integer (lookup_internalvar ("_exitcode"),

                                   (LONGEST) ecs->ws.value.integer);



          /* Also record this in the inferior itself.  */

          current_inferior ()->has_exit_code = 1;

          current_inferior ()->exit_code = (LONGEST) ecs->ws.value.integer;



          /* Support the --return-child-result option.  */

          return_child_result_value = ecs->ws.value.integer;



          observer_notify_exited (ecs->ws.value.integer);

        }

      else

        {

          struct regcache *regcache = get_thread_regcache (ecs->ptid);

          struct gdbarch *gdbarch = get_regcache_arch (regcache);



          if (gdbarch_gdb_signal_to_target_p (gdbarch))

            {

              /* Set the value of the internal variable $_exitsignal,

                 which holds the signal uncaught by the inferior.  */

              set_internalvar_integer (lookup_internalvar ("_exitsignal"),

                                       gdbarch_gdb_signal_to_target (gdbarch,

                                                          ecs->ws.value.sig));

            }

          else

            {

              /* We don't have access to the target's method used for

                 converting between signal numbers (GDB's internal

                 representation <-> target's representation).

                 Therefore, we cannot do a good job at displaying this

                 information to the user.  It's better to just warn

                 her about it (if infrun debugging is enabled), and

                 give up.  */

              if (debug_infrun)

                fprintf_filtered (gdb_stdlog, _("\

Cannot fill $_exitsignal with the correct signal number.\n"));

            }



          observer_notify_signal_exited (ecs->ws.value.sig);

        }



      gdb_flush (gdb_stdout);

      target_mourn_inferior ();

      stop_print_frame = 0;

      stop_waiting (ecs);

      return;



      /* The following are the only cases in which we keep going;

         the above cases end in a continue or goto.  */

    case TARGET_WAITKIND_FORKED:

    case TARGET_WAITKIND_VFORKED:

      if (debug_infrun)

        {

          if (ecs->ws.kind == TARGET_WAITKIND_FORKED)

            fprintf_unfiltered (gdb_stdlog, "infrun: TARGET_WAITKIND_FORKED\n");

          else

            fprintf_unfiltered (gdb_stdlog, "infrun: TARGET_WAITKIND_VFORKED\n");

        }



      /* Check whether the inferior is displaced stepping.  */

      {

        struct regcache *regcache = get_thread_regcache (ecs->ptid);

        struct gdbarch *gdbarch = get_regcache_arch (regcache);

        struct displaced_step_inferior_state *displaced

          = get_displaced_stepping_state (ptid_get_pid (ecs->ptid));



        /* If checking displaced stepping is supported, and thread

           ecs->ptid is displaced stepping.  */

        if (displaced && ptid_equal (displaced->step_ptid, ecs->ptid))

          {

            struct inferior *parent_inf

              = find_inferior_ptid (ecs->ptid);

            struct regcache *child_regcache;

            CORE_ADDR parent_pc;



            /* GDB has got TARGET_WAITKIND_FORKED or TARGET_WAITKIND_VFORKED,

               indicating that the displaced stepping of syscall instruction

               has been done.  Perform cleanup for parent process here.  Note

               that this operation also cleans up the child process for vfork,

               because their pages are shared.  */

            displaced_step_fixup (ecs->ptid, GDB_SIGNAL_TRAP);



            if (ecs->ws.kind == TARGET_WAITKIND_FORKED)

              {

                /* Restore scratch pad for child process.  */

                displaced_step_restore (displaced, ecs->ws.value.related_pid);

              }



            /* Since the vfork/fork syscall instruction was executed in the scratchpad,

               the child's PC is also within the scratchpad.  Set the child's PC

               to the parent's PC value, which has already been fixed up.

               FIXME: we use the parent's aspace here, although we're touching

               the child, because the child hasn't been added to the inferior

               list yet at this point.  */



            child_regcache

              = get_thread_arch_aspace_regcache (ecs->ws.value.related_pid,

                                                 gdbarch,

                                                 parent_inf->aspace);

            /* Read PC value of parent process.  */

            parent_pc = regcache_read_pc (regcache);



            if (debug_displaced)

              fprintf_unfiltered (gdb_stdlog,

                                  "displaced: write child pc from %s to %s\n",

                                  paddress (gdbarch,

                                            regcache_read_pc (child_regcache)),

                                  paddress (gdbarch, parent_pc));



            regcache_write_pc (child_regcache, parent_pc);

          }

      }



      if (!ptid_equal (ecs->ptid, inferior_ptid))

        context_switch (ecs->ptid);



      /* Immediately detach breakpoints from the child before there's

         any chance of letting the user delete breakpoints from the

         breakpoint lists.  If we don't do this early, it's easy to

         leave left over traps in the child, vis: "break foo; catch

         fork; c; <fork>; del; c; <child calls foo>".  We only follow

         the fork on the last `continue', and by that time the

         breakpoint at "foo" is long gone from the breakpoint table.

         If we vforked, then we don't need to unpatch here, since both

         parent and child are sharing the same memory pages; we'll

         need to unpatch at follow/detach time instead to be certain

         that new breakpoints added between catchpoint hit time and

         vfork follow are detached.  */

      if (ecs->ws.kind != TARGET_WAITKIND_VFORKED)

        {

          /* This won't actually modify the breakpoint list, but will

             physically remove the breakpoints from the child.  */

          detach_breakpoints (ecs->ws.value.related_pid);

        }



      delete_just_stopped_threads_single_step_breakpoints ();



      /* In case the event is caught by a catchpoint, remember that

         the event is to be followed at the next resume of the thread,

         and not immediately.  */

      ecs->event_thread->pending_follow = ecs->ws;



      stop_pc = regcache_read_pc (get_thread_regcache (ecs->ptid));



      ecs->event_thread->control.stop_bpstat

        = bpstat_stop_status (get_regcache_aspace (get_current_regcache ()),

                              stop_pc, ecs->ptid, &ecs->ws);



      /* If no catchpoint triggered for this, then keep going.  Note

         that we're interested in knowing the bpstat actually causes a

         stop, not just if it may explain the signal.  Software

         watchpoints, for example, always appear in the bpstat.  */

      if (!bpstat_causes_stop (ecs->event_thread->control.stop_bpstat))

        {

          ptid_t parent;

          ptid_t child;

          int should_resume;

          int follow_child

            = (follow_fork_mode_string == follow_fork_mode_child);



          ecs->event_thread->suspend.stop_signal = GDB_SIGNAL_0;



          should_resume = follow_fork ();



          parent = ecs->ptid;

          child = ecs->ws.value.related_pid;



          /* In non-stop mode, also resume the other branch.  */

          if (non_stop && !detach_fork)

            {

              if (follow_child)

                switch_to_thread (parent);

              else

                switch_to_thread (child);



              ecs->event_thread = inferior_thread ();

              ecs->ptid = inferior_ptid;

              keep_going (ecs);

            }



          if (follow_child)

            switch_to_thread (child);

          else

            switch_to_thread (parent);



          ecs->event_thread = inferior_thread ();

          ecs->ptid = inferior_ptid;



          if (should_resume)

            keep_going (ecs);

          else

            stop_waiting (ecs);

          return;

        }

      process_event_stop_test (ecs);

      return;



    case TARGET_WAITKIND_VFORK_DONE:

      /* Done with the shared memory region.  Re-insert breakpoints in

         the parent, and keep going.  */



      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog,

                            "infrun: TARGET_WAITKIND_VFORK_DONE\n");



      if (!ptid_equal (ecs->ptid, inferior_ptid))

        context_switch (ecs->ptid);



      current_inferior ()->waiting_for_vfork_done = 0;

      current_inferior ()->pspace->breakpoints_not_allowed = 0;

      /* This also takes care of reinserting breakpoints in the

         previously locked inferior.  */

      keep_going (ecs);

      return;



    case TARGET_WAITKIND_EXECD:

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog, "infrun: TARGET_WAITKIND_EXECD\n");



      if (!ptid_equal (ecs->ptid, inferior_ptid))

        context_switch (ecs->ptid);



      stop_pc = regcache_read_pc (get_thread_regcache (ecs->ptid));



      /* Do whatever is necessary to the parent branch of the vfork.  */

      handle_vfork_child_exec_or_exit (1);



      /* This causes the eventpoints and symbol table to be reset.

         Must do this now, before trying to determine whether to

         stop.  */

      follow_exec (inferior_ptid, ecs->ws.value.execd_pathname);



      ecs->event_thread->control.stop_bpstat

        = bpstat_stop_status (get_regcache_aspace (get_current_regcache ()),

                              stop_pc, ecs->ptid, &ecs->ws);



      /* Note that this may be referenced from inside

         bpstat_stop_status above, through inferior_has_execd.  */

      xfree (ecs->ws.value.execd_pathname);

      ecs->ws.value.execd_pathname = NULL;



      /* If no catchpoint triggered for this, then keep going.  */

      if (!bpstat_causes_stop (ecs->event_thread->control.stop_bpstat))

        {

          ecs->event_thread->suspend.stop_signal = GDB_SIGNAL_0;

          keep_going (ecs);

          return;

        }

      process_event_stop_test (ecs);

      return;



      /* Be careful not to try to gather much state about a thread

         that's in a syscall.  It's frequently a losing proposition.  */

    case TARGET_WAITKIND_SYSCALL_ENTRY:

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog,

                            "infrun: TARGET_WAITKIND_SYSCALL_ENTRY\n");

      /* Getting the current syscall number.  */

      if (handle_syscall_event (ecs) == 0)

        process_event_stop_test (ecs);

      return;



      /* Before examining the threads further, step this thread to

         get it entirely out of the syscall.  (We get notice of the

         event when the thread is just on the verge of exiting a

         syscall.  Stepping one instruction seems to get it back

         into user code.)  */

    case TARGET_WAITKIND_SYSCALL_RETURN:

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog,

                            "infrun: TARGET_WAITKIND_SYSCALL_RETURN\n");

      if (handle_syscall_event (ecs) == 0)

        process_event_stop_test (ecs);

      return;



    case TARGET_WAITKIND_STOPPED:

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog, "infrun: TARGET_WAITKIND_STOPPED\n");

      ecs->event_thread->suspend.stop_signal = ecs->ws.value.sig;

      handle_signal_stop (ecs);

      return;



    case TARGET_WAITKIND_NO_HISTORY:

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog, "infrun: TARGET_WAITKIND_NO_HISTORY\n");

      /* Reverse execution: target ran out of history info.  */



      delete_just_stopped_threads_single_step_breakpoints ();

      stop_pc = regcache_read_pc (get_thread_regcache (ecs->ptid));

      observer_notify_no_history ();

      stop_waiting (ecs);

      return;

    }

}



/* Come here when the program has stopped with a signal.  */



static void

‌handle_signal_stop (struct execution_control_state *ecs)

{

  struct frame_info *frame;

  struct gdbarch *gdbarch;

  int stopped_by_watchpoint;

  enum stop_kind stop_soon;

  int random_signal;



  gdb_assert (ecs->ws.kind == TARGET_WAITKIND_STOPPED);



  /* Do we need to clean up the state of a thread that has

     completed a displaced single-step?  (Doing so usually affects

     the PC, so do it here, before we set stop_pc.)  */

  displaced_step_fixup (ecs->ptid,

                        ecs->event_thread->suspend.stop_signal);



  /* If we either finished a single-step or hit a breakpoint, but

     the user wanted this thread to be stopped, pretend we got a

     SIG0 (generic unsignaled stop).  */

  if (ecs->event_thread->stop_requested

      && ecs->event_thread->suspend.stop_signal == GDB_SIGNAL_TRAP)

    ecs->event_thread->suspend.stop_signal = GDB_SIGNAL_0;



  stop_pc = regcache_read_pc (get_thread_regcache (ecs->ptid));



  if (debug_infrun)

    {

      struct regcache *regcache = get_thread_regcache (ecs->ptid);

      struct gdbarch *gdbarch = get_regcache_arch (regcache);

      struct cleanup *old_chain = save_inferior_ptid ();



      inferior_ptid = ecs->ptid;



      fprintf_unfiltered (gdb_stdlog, "infrun: stop_pc = %s\n",

                          paddress (gdbarch, stop_pc));

      if (target_stopped_by_watchpoint ())

        {

          CORE_ADDR addr;



          fprintf_unfiltered (gdb_stdlog, "infrun: stopped by watchpoint\n");



          if (target_stopped_data_address (&current_target, &addr))

            fprintf_unfiltered (gdb_stdlog,

                                "infrun: stopped data address = %s\n",

                                paddress (gdbarch, addr));

          else

            fprintf_unfiltered (gdb_stdlog,

                                "infrun: (no data address available)\n");

        }



      do_cleanups (old_chain);

    }



  /* This is originated from start_remote(), start_inferior() and

     shared libraries hook functions.  */

  stop_soon = get_inferior_stop_soon (ecs->ptid);

  if (stop_soon == STOP_QUIETLY || stop_soon == STOP_QUIETLY_REMOTE)

    {

      if (!ptid_equal (ecs->ptid, inferior_ptid))

        context_switch (ecs->ptid);

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog, "infrun: quietly stopped\n");

      stop_print_frame = 1;

      stop_waiting (ecs);

      return;

    }



  if (ecs->event_thread->suspend.stop_signal == GDB_SIGNAL_TRAP

      && stop_after_trap)

    {

      if (!ptid_equal (ecs->ptid, inferior_ptid))

        context_switch (ecs->ptid);

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog, "infrun: stopped\n");

      stop_print_frame = 0;

      stop_waiting (ecs);

      return;

    }



  /* This originates from attach_command().  We need to overwrite

     the stop_signal here, because some kernels don't ignore a

     SIGSTOP in a subsequent ptrace(PTRACE_CONT,SIGSTOP) call.

     See more comments in inferior.h.  On the other hand, if we

     get a non-SIGSTOP, report it to the user - assume the backend

     will handle the SIGSTOP if it should show up later.



     Also consider that the attach is complete when we see a

     SIGTRAP.  Some systems (e.g. Windows), and stubs supporting

     target extended-remote report it instead of a SIGSTOP

     (e.g. gdbserver).  We already rely on SIGTRAP being our

     signal, so this is no exception.



     Also consider that the attach is complete when we see a

     GDB_SIGNAL_0.  In non-stop mode, GDB will explicitly tell

     the target to stop all threads of the inferior, in case the

     low level attach operation doesn't stop them implicitly.  If

     they weren't stopped implicitly, then the stub will report a

     GDB_SIGNAL_0, meaning: stopped for no particular reason

     other than GDB's request.  */

  if (stop_soon == STOP_QUIETLY_NO_SIGSTOP

      && (ecs->event_thread->suspend.stop_signal == GDB_SIGNAL_STOP

          || ecs->event_thread->suspend.stop_signal == GDB_SIGNAL_TRAP

          || ecs->event_thread->suspend.stop_signal == GDB_SIGNAL_0))

    {

      stop_print_frame = 1;

      stop_waiting (ecs);

      ecs->event_thread->suspend.stop_signal = GDB_SIGNAL_0;

      return;

    }



  /* See if something interesting happened to the non-current thread.  If

     so, then switch to that thread.  */

  if (!ptid_equal (ecs->ptid, inferior_ptid))

    {

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog, "infrun: context switch\n");



      context_switch (ecs->ptid);



      if (deprecated_context_hook)

        deprecated_context_hook (pid_to_thread_id (ecs->ptid));

    }



  /* At this point, get hold of the now-current thread's frame.  */

  frame = get_current_frame ();

  gdbarch = get_frame_arch (frame);



  /* Pull the single step breakpoints out of the target.  */

  if (ecs->event_thread->suspend.stop_signal == GDB_SIGNAL_TRAP)

    {

      struct regcache *regcache;

      struct address_space *aspace;

      CORE_ADDR pc;



      regcache = get_thread_regcache (ecs->ptid);

      aspace = get_regcache_aspace (regcache);

      pc = regcache_read_pc (regcache);



      /* However, before doing so, if this single-step breakpoint was

         actually for another thread, set this thread up for moving

         past it.  */

      if (!thread_has_single_step_breakpoint_here (ecs->event_thread,

                                                   aspace, pc))

        {

          if (single_step_breakpoint_inserted_here_p (aspace, pc))

            {

              if (debug_infrun)

                {

                  fprintf_unfiltered (gdb_stdlog,

                                      "infrun: [%s] hit another thread's "

                                      "single-step breakpoint\n",

                                      target_pid_to_str (ecs->ptid));

                }

              ecs->hit_singlestep_breakpoint = 1;

            }

        }

      else

        {

          if (debug_infrun)

            {

              fprintf_unfiltered (gdb_stdlog,

                                  "infrun: [%s] hit its "

                                  "single-step breakpoint\n",

                                  target_pid_to_str (ecs->ptid));

            }

        }

    }

  delete_just_stopped_threads_single_step_breakpoints ();



  if (ecs->event_thread->suspend.stop_signal == GDB_SIGNAL_TRAP

      && ecs->event_thread->control.trap_expected

      && ecs->event_thread->stepping_over_watchpoint)

    stopped_by_watchpoint = 0;

  else

    stopped_by_watchpoint = watchpoints_triggered (&ecs->ws);



  /* If necessary, step over this watchpoint.  We'll be back to display

     it in a moment.  */

  if (stopped_by_watchpoint

      && (target_have_steppable_watchpoint

          || gdbarch_have_nonsteppable_watchpoint (gdbarch)))

    {

      /* At this point, we are stopped at an instruction which has

         attempted to write to a piece of memory under control of

         a watchpoint.  The instruction hasn't actually executed

         yet.  If we were to evaluate the watchpoint expression

         now, we would get the old value, and therefore no change

         would seem to have occurred.



         In order to make watchpoints work `right', we really need

         to complete the memory write, and then evaluate the

         watchpoint expression.  We do this by single-stepping the

         target.



         It may not be necessary to disable the watchpoint to step over

         it.  For example, the PA can (with some kernel cooperation)

         single step over a watchpoint without disabling the watchpoint.



         It is far more common to need to disable a watchpoint to step

         the inferior over it.  If we have non-steppable watchpoints,

         we must disable the current watchpoint; it's simplest to

         disable all watchpoints.



         Any breakpoint at PC must also be stepped over -- if there's

         one, it will have already triggered before the watchpoint

         triggered, and we either already reported it to the user, or

         it didn't cause a stop and we called keep_going.  In either

         case, if there was a breakpoint at PC, we must be trying to

         step past it.  */

      ecs->event_thread->stepping_over_watchpoint = 1;

      keep_going (ecs);

      return;

    }



  ecs->event_thread->stepping_over_breakpoint = 0;

  ecs->event_thread->stepping_over_watchpoint = 0;

  bpstat_clear (&ecs->event_thread->control.stop_bpstat);

  ecs->event_thread->control.stop_step = 0;

  stop_print_frame = 1;

  stopped_by_random_signal = 0;



  /* Hide inlined functions starting here, unless we just performed stepi or

     nexti.  After stepi and nexti, always show the innermost frame (not any

     inline function call sites).  */

  if (ecs->event_thread->control.step_range_end != 1)

    {

      struct address_space *aspace =

        get_regcache_aspace (get_thread_regcache (ecs->ptid));



      /* skip_inline_frames is expensive, so we avoid it if we can

         determine that the address is one where functions cannot have

         been inlined.  This improves performance with inferiors that

         load a lot of shared libraries, because the solib event

         breakpoint is defined as the address of a function (i.e. not

         inline).  Note that we have to check the previous PC as well

         as the current one to catch cases when we have just

         single-stepped off a breakpoint prior to reinstating it.

         Note that we're assuming that the code we single-step to is

         not inline, but that's not definitive: there's nothing

         preventing the event breakpoint function from containing

         inlined code, and the single-step ending up there.  If the

         user had set a breakpoint on that inlined code, the missing

         skip_inline_frames call would break things.  Fortunately

         that's an extremely unlikely scenario.  */

      if (!pc_at_non_inline_function (aspace, stop_pc, &ecs->ws)

          && !(ecs->event_thread->suspend.stop_signal == GDB_SIGNAL_TRAP

               && ecs->event_thread->control.trap_expected

               && pc_at_non_inline_function (aspace,

                                             ecs->event_thread->prev_pc,

                                             &ecs->ws)))

        {

          skip_inline_frames (ecs->ptid);



          /* Re-fetch current thread's frame in case that invalidated

             the frame cache.  */

          frame = get_current_frame ();

          gdbarch = get_frame_arch (frame);

        }

    }



  if (ecs->event_thread->suspend.stop_signal == GDB_SIGNAL_TRAP

      && ecs->event_thread->control.trap_expected

      && gdbarch_single_step_through_delay_p (gdbarch)

      && currently_stepping (ecs->event_thread))

    {

      /* We're trying to step off a breakpoint.  Turns out that we're

         also on an instruction that needs to be stepped multiple

         times before it's been fully executing.  E.g., architectures

         with a delay slot.  It needs to be stepped twice, once for

         the instruction and once for the delay slot.  */

      int step_through_delay

        = gdbarch_single_step_through_delay (gdbarch, frame);



      if (debug_infrun && step_through_delay)

        fprintf_unfiltered (gdb_stdlog, "infrun: step through delay\n");

      if (ecs->event_thread->control.step_range_end == 0

          && step_through_delay)

        {

          /* The user issued a continue when stopped at a breakpoint.

             Set up for another trap and get out of here.  */

         ecs->event_thread->stepping_over_breakpoint = 1;

         keep_going (ecs);

         return;

        }

      else if (step_through_delay)

        {

          /* The user issued a step when stopped at a breakpoint.

             Maybe we should stop, maybe we should not - the delay

             slot *might* correspond to a line of source.  In any

             case, don't decide that here, just set

             ecs->stepping_over_breakpoint, making sure we

             single-step again before breakpoints are re-inserted.  */

          ecs->event_thread->stepping_over_breakpoint = 1;

        }

    }



  /* See if there is a breakpoint/watchpoint/catchpoint/etc. that

     handles this event.  */

  ecs->event_thread->control.stop_bpstat

    = bpstat_stop_status (get_regcache_aspace (get_current_regcache ()),

                          stop_pc, ecs->ptid, &ecs->ws);



  /* Following in case break condition called a

     function.  */

  stop_print_frame = 1;



  /* This is where we handle "moribund" watchpoints.  Unlike

     software breakpoints traps, hardware watchpoint traps are

     always distinguishable from random traps.  If no high-level

     watchpoint is associated with the reported stop data address

     anymore, then the bpstat does not explain the signal ---

     simply make sure to ignore it if `stopped_by_watchpoint' is

     set.  */



  if (debug_infrun

      && ecs->event_thread->suspend.stop_signal == GDB_SIGNAL_TRAP

      && !bpstat_explains_signal (ecs->event_thread->control.stop_bpstat,

                                  GDB_SIGNAL_TRAP)

      && stopped_by_watchpoint)

    fprintf_unfiltered (gdb_stdlog,

                        "infrun: no user watchpoint explains "

                        "watchpoint SIGTRAP, ignoring\n");



  /* NOTE: cagney/2003-03-29: These checks for a random signal

     at one stage in the past included checks for an inferior

     function call's call dummy's return breakpoint.  The original

     comment, that went with the test, read:



     ``End of a stack dummy.  Some systems (e.g. Sony news) give

     another signal besides SIGTRAP, so check here as well as

     above.''



     If someone ever tries to get call dummys on a

     non-executable stack to work (where the target would stop

     with something like a SIGSEGV), then those tests might need

     to be re-instated.  Given, however, that the tests were only

     enabled when momentary breakpoints were not being used, I

     suspect that it won't be the case.



     NOTE: kettenis/2004-02-05: Indeed such checks don't seem to

     be necessary for call dummies on a non-executable stack on

     SPARC.  */



  /* See if the breakpoints module can explain the signal.  */

  random_signal

    = !bpstat_explains_signal (ecs->event_thread->control.stop_bpstat,

                               ecs->event_thread->suspend.stop_signal);



  /* If not, perhaps stepping/nexting can.  */

  if (random_signal)

    random_signal = !(ecs->event_thread->suspend.stop_signal == GDB_SIGNAL_TRAP

                      && currently_stepping (ecs->event_thread));



  /* Perhaps the thread hit a single-step breakpoint of _another_

     thread.  Single-step breakpoints are transparent to the

     breakpoints module.  */

  if (random_signal)

    random_signal = !ecs->hit_singlestep_breakpoint;



  /* No?  Perhaps we got a moribund watchpoint.  */

  if (random_signal)

    random_signal = !stopped_by_watchpoint;



  /* For the program's own signals, act according to

     the signal handling tables.  */



  if (random_signal)

    {

      /* Signal not for debugging purposes.  */

      struct inferior *inf = find_inferior_ptid (ecs->ptid);

      enum gdb_signal stop_signal = ecs->event_thread->suspend.stop_signal;



      if (debug_infrun)

         fprintf_unfiltered (gdb_stdlog, "infrun: random signal (%s)\n",

                             gdb_signal_to_symbol_string (stop_signal));



      stopped_by_random_signal = 1;



      /* Always stop on signals if we're either just gaining control

         of the program, or the user explicitly requested this thread

         to remain stopped.  */

      if (stop_soon != NO_STOP_QUIETLY

          || ecs->event_thread->stop_requested

          || (!inf->detaching

              && signal_stop_state (ecs->event_thread->suspend.stop_signal)))

        {

          stop_waiting (ecs);

          return;

        }



      /* Notify observers the signal has "handle print" set.  Note we

         returned early above if stopping; normal_stop handles the

         printing in that case.  */

      if (signal_print[ecs->event_thread->suspend.stop_signal])

        {

          /* The signal table tells us to print about this signal.  */

          target_terminal_ours_for_output ();

          observer_notify_signal_received (ecs->event_thread->suspend.stop_signal);

          target_terminal_inferior ();

        }



      /* Clear the signal if it should not be passed.  */

      if (signal_program[ecs->event_thread->suspend.stop_signal] == 0)

        ecs->event_thread->suspend.stop_signal = GDB_SIGNAL_0;



      if (ecs->event_thread->prev_pc == stop_pc

          && ecs->event_thread->control.trap_expected

          && ecs->event_thread->control.step_resume_breakpoint == NULL)

        {

          /* We were just starting a new sequence, attempting to

             single-step off of a breakpoint and expecting a SIGTRAP.

             Instead this signal arrives.  This signal will take us out

             of the stepping range so GDB needs to remember to, when

             the signal handler returns, resume stepping off that

             breakpoint.  */

          /* To simplify things, "continue" is forced to use the same

             code paths as single-step - set a breakpoint at the

             signal return address and then, once hit, step off that

             breakpoint.  */

          if (debug_infrun)

            fprintf_unfiltered (gdb_stdlog,

                                "infrun: signal arrived while stepping over "

                                "breakpoint\n");



          insert_hp_step_resume_breakpoint_at_frame (frame);

          ecs->event_thread->step_after_step_resume_breakpoint = 1;

          /* Reset trap_expected to ensure breakpoints are re-inserted.  */

          ecs->event_thread->control.trap_expected = 0;



          /* If we were nexting/stepping some other thread, switch to

             it, so that we don't continue it, losing control.  */

          if (!switch_back_to_stepped_thread (ecs))

            keep_going (ecs);

          return;

        }



      if (ecs->event_thread->suspend.stop_signal != GDB_SIGNAL_0

          && (pc_in_thread_step_range (stop_pc, ecs->event_thread)

              || ecs->event_thread->control.step_range_end == 1)

          && frame_id_eq (get_stack_frame_id (frame),

                          ecs->event_thread->control.step_stack_frame_id)

          && ecs->event_thread->control.step_resume_breakpoint == NULL)

        {

          /* The inferior is about to take a signal that will take it

             out of the single step range.  Set a breakpoint at the

             current PC (which is presumably where the signal handler

             will eventually return) and then allow the inferior to

             run free.



             Note that this is only needed for a signal delivered

             while in the single-step range.  Nested signals aren't a

             problem as they eventually all return.  */

          if (debug_infrun)

            fprintf_unfiltered (gdb_stdlog,

                                "infrun: signal may take us out of "

                                "single-step range\n");



          insert_hp_step_resume_breakpoint_at_frame (frame);

          ecs->event_thread->step_after_step_resume_breakpoint = 1;

          /* Reset trap_expected to ensure breakpoints are re-inserted.  */

          ecs->event_thread->control.trap_expected = 0;

          keep_going (ecs);

          return;

        }



      /* Note: step_resume_breakpoint may be non-NULL.  This occures

         when either there's a nested signal, or when there's a

         pending signal enabled just as the signal handler returns

         (leaving the inferior at the step-resume-breakpoint without

         actually executing it).  Either way continue until the

         breakpoint is really hit.  */



      if (!switch_back_to_stepped_thread (ecs))

        {

          if (debug_infrun)

            fprintf_unfiltered (gdb_stdlog,

                                "infrun: random signal, keep going\n");



          keep_going (ecs);

        }

      return;

    }



  process_event_stop_test (ecs);

}



/* Come here when we've got some debug event / signal we can explain

   (IOW, not a random signal), and test whether it should cause a

   stop, or whether we should resume the inferior (transparently).

   E.g., could be a breakpoint whose condition evaluates false; we

   could be still stepping within the line; etc.  */



static void

‌process_event_stop_test (struct execution_control_state *ecs)

{

  struct symtab_and_line stop_pc_sal;

  struct frame_info *frame;

  struct gdbarch *gdbarch;

  CORE_ADDR jmp_buf_pc;

  struct bpstat_what what;



  /* Handle cases caused by hitting a breakpoint.  */



  frame = get_current_frame ();

  gdbarch = get_frame_arch (frame);



  what = bpstat_what (ecs->event_thread->control.stop_bpstat);



  if (what.call_dummy)

    {

      stop_stack_dummy = what.call_dummy;

    }



  /* If we hit an internal event that triggers symbol changes, the

     current frame will be invalidated within bpstat_what (e.g., if we

     hit an internal solib event).  Re-fetch it.  */

  frame = get_current_frame ();

  gdbarch = get_frame_arch (frame);



  switch (what.main_action)

    {

    case BPSTAT_WHAT_SET_LONGJMP_RESUME:

      /* If we hit the breakpoint at longjmp while stepping, we

         install a momentary breakpoint at the target of the

         jmp_buf.  */



      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog,

                            "infrun: BPSTAT_WHAT_SET_LONGJMP_RESUME\n");



      ecs->event_thread->stepping_over_breakpoint = 1;



      if (what.is_longjmp)

        {

          struct value *arg_value;



          /* If we set the longjmp breakpoint via a SystemTap probe,

             then use it to extract the arguments.  The destination PC

             is the third argument to the probe.  */

          arg_value = probe_safe_evaluate_at_pc (frame, 2);

          if (arg_value)

            {

              jmp_buf_pc = value_as_address (arg_value);

              jmp_buf_pc = gdbarch_addr_bits_remove (gdbarch, jmp_buf_pc);

            }

          else if (!gdbarch_get_longjmp_target_p (gdbarch)

                   || !gdbarch_get_longjmp_target (gdbarch,

                                                   frame, &jmp_buf_pc))

            {

              if (debug_infrun)

                fprintf_unfiltered (gdb_stdlog,

                                    "infrun: BPSTAT_WHAT_SET_LONGJMP_RESUME "

                                    "(!gdbarch_get_longjmp_target)\n");

              keep_going (ecs);

              return;

            }



          /* Insert a breakpoint at resume address.  */

          insert_longjmp_resume_breakpoint (gdbarch, jmp_buf_pc);

        }

      else

        check_exception_resume (ecs, frame);

      keep_going (ecs);

      return;



    case BPSTAT_WHAT_CLEAR_LONGJMP_RESUME:

      {

        struct frame_info *init_frame;



        /* There are several cases to consider.



           1. The initiating frame no longer exists.  In this case we

           must stop, because the exception or longjmp has gone too

           far.



           2. The initiating frame exists, and is the same as the

           current frame.  We stop, because the exception or longjmp

           has been caught.



           3. The initiating frame exists and is different from the

           current frame.  This means the exception or longjmp has

           been caught beneath the initiating frame, so keep going.



           4. longjmp breakpoint has been placed just to protect

           against stale dummy frames and user is not interested in

           stopping around longjmps.  */



        if (debug_infrun)

          fprintf_unfiltered (gdb_stdlog,

                              "infrun: BPSTAT_WHAT_CLEAR_LONGJMP_RESUME\n");



        gdb_assert (ecs->event_thread->control.exception_resume_breakpoint

                    != NULL);

        delete_exception_resume_breakpoint (ecs->event_thread);



        if (what.is_longjmp)

          {

            check_longjmp_breakpoint_for_call_dummy (ecs->event_thread);



            if (!frame_id_p (ecs->event_thread->initiating_frame))

              {

                /* Case 4.  */

                keep_going (ecs);

                return;

              }

          }



        init_frame = frame_find_by_id (ecs->event_thread->initiating_frame);



        if (init_frame)

          {

            struct frame_id current_id

              = get_frame_id (get_current_frame ());

            if (frame_id_eq (current_id,

                             ecs->event_thread->initiating_frame))

              {

                /* Case 2.  Fall through.  */

              }

            else

              {

                /* Case 3.  */

                keep_going (ecs);

                return;

              }

          }



        /* For Cases 1 and 2, remove the step-resume breakpoint, if it

           exists.  */

        delete_step_resume_breakpoint (ecs->event_thread);



        end_stepping_range (ecs);

      }

      return;



    case BPSTAT_WHAT_SINGLE:

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog, "infrun: BPSTAT_WHAT_SINGLE\n");

      ecs->event_thread->stepping_over_breakpoint = 1;

      /* Still need to check other stuff, at least the case where we

         are stepping and step out of the right range.  */

      break;



    case BPSTAT_WHAT_STEP_RESUME:

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog, "infrun: BPSTAT_WHAT_STEP_RESUME\n");



      delete_step_resume_breakpoint (ecs->event_thread);

      if (ecs->event_thread->control.proceed_to_finish

          && execution_direction == EXEC_REVERSE)

        {

          struct thread_info *tp = ecs->event_thread;



          /* We are finishing a function in reverse, and just hit the

             step-resume breakpoint at the start address of the

             function, and we're almost there -- just need to back up

             by one more single-step, which should take us back to the

             function call.  */

          tp->control.step_range_start = tp->control.step_range_end = 1;

          keep_going (ecs);

          return;

        }

      fill_in_stop_func (gdbarch, ecs);

      if (stop_pc == ecs->stop_func_start

          && execution_direction == EXEC_REVERSE)

        {

          /* We are stepping over a function call in reverse, and just

             hit the step-resume breakpoint at the start address of

             the function.  Go back to single-stepping, which should

             take us back to the function call.  */

          ecs->event_thread->stepping_over_breakpoint = 1;

          keep_going (ecs);

          return;

        }

      break;



    case BPSTAT_WHAT_STOP_NOISY:

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog, "infrun: BPSTAT_WHAT_STOP_NOISY\n");

      stop_print_frame = 1;



      /* Assume the thread stopped for a breapoint.  We'll still check

         whether a/the breakpoint is there when the thread is next

         resumed.  */

      ecs->event_thread->stepping_over_breakpoint = 1;



      stop_waiting (ecs);

      return;



    case BPSTAT_WHAT_STOP_SILENT:

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog, "infrun: BPSTAT_WHAT_STOP_SILENT\n");

      stop_print_frame = 0;



      /* Assume the thread stopped for a breapoint.  We'll still check

         whether a/the breakpoint is there when the thread is next

         resumed.  */

      ecs->event_thread->stepping_over_breakpoint = 1;

      stop_waiting (ecs);

      return;



    case BPSTAT_WHAT_HP_STEP_RESUME:

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog, "infrun: BPSTAT_WHAT_HP_STEP_RESUME\n");



      delete_step_resume_breakpoint (ecs->event_thread);

      if (ecs->event_thread->step_after_step_resume_breakpoint)

        {

          /* Back when the step-resume breakpoint was inserted, we

             were trying to single-step off a breakpoint.  Go back to

             doing that.  */

          ecs->event_thread->step_after_step_resume_breakpoint = 0;

          ecs->event_thread->stepping_over_breakpoint = 1;

          keep_going (ecs);

          return;

        }

      break;



    case BPSTAT_WHAT_KEEP_CHECKING:

      break;

    }



  /* If we stepped a permanent breakpoint and we had a high priority

     step-resume breakpoint for the address we stepped, but we didn't

     hit it, then we must have stepped into the signal handler.  The

     step-resume was only necessary to catch the case of _not_

     stepping into the handler, so delete it, and fall through to

     checking whether the step finished.  */

  if (ecs->event_thread->stepped_breakpoint)

    {

      struct breakpoint *sr_bp

        = ecs->event_thread->control.step_resume_breakpoint;



      if (sr_bp->loc->permanent

          && sr_bp->type == bp_hp_step_resume

          && sr_bp->loc->address == ecs->event_thread->prev_pc)

        {

          if (debug_infrun)

            fprintf_unfiltered (gdb_stdlog,

                                "infrun: stepped permanent breakpoint, stopped in "

                                "handler\n");

          delete_step_resume_breakpoint (ecs->event_thread);

          ecs->event_thread->step_after_step_resume_breakpoint = 0;

        }

    }



  /* We come here if we hit a breakpoint but should not stop for it.

     Possibly we also were stepping and should stop for that.  So fall

     through and test for stepping.  But, if not stepping, do not

     stop.  */



  /* In all-stop mode, if we're currently stepping but have stopped in

     some other thread, we need to switch back to the stepped thread.  */

  if (switch_back_to_stepped_thread (ecs))

    return;



  if (ecs->event_thread->control.step_resume_breakpoint)

    {

      if (debug_infrun)

         fprintf_unfiltered (gdb_stdlog,

                             "infrun: step-resume breakpoint is inserted\n");



      /* Having a step-resume breakpoint overrides anything

         else having to do with stepping commands until

         that breakpoint is reached.  */

      keep_going (ecs);

      return;

    }



  if (ecs->event_thread->control.step_range_end == 0)

    {

      if (debug_infrun)

         fprintf_unfiltered (gdb_stdlog, "infrun: no stepping, continue\n");

      /* Likewise if we aren't even stepping.  */

      keep_going (ecs);

      return;

    }



  /* Re-fetch current thread's frame in case the code above caused

     the frame cache to be re-initialized, making our FRAME variable

     a dangling pointer.  */

  frame = get_current_frame ();

  gdbarch = get_frame_arch (frame);

  fill_in_stop_func (gdbarch, ecs);



  /* If stepping through a line, keep going if still within it.



     Note that step_range_end is the address of the first instruction

     beyond the step range, and NOT the address of the last instruction

     within it!



     Note also that during reverse execution, we may be stepping

     through a function epilogue and therefore must detect when

     the current-frame changes in the middle of a line.  */



  if (pc_in_thread_step_range (stop_pc, ecs->event_thread)

      && (execution_direction != EXEC_REVERSE

          || frame_id_eq (get_frame_id (frame),

                          ecs->event_thread->control.step_frame_id)))

    {

      if (debug_infrun)

        fprintf_unfiltered

          (gdb_stdlog, "infrun: stepping inside range [%s-%s]\n",

           paddress (gdbarch, ecs->event_thread->control.step_range_start),

           paddress (gdbarch, ecs->event_thread->control.step_range_end));



      /* Tentatively re-enable range stepping; `resume' disables it if

         necessary (e.g., if we're stepping over a breakpoint or we

         have software watchpoints).  */

      ecs->event_thread->control.may_range_step = 1;



      /* When stepping backward, stop at beginning of line range

         (unless it's the function entry point, in which case

         keep going back to the call point).  */

      if (stop_pc == ecs->event_thread->control.step_range_start

          && stop_pc != ecs->stop_func_start

          && execution_direction == EXEC_REVERSE)

        end_stepping_range (ecs);

      else

        keep_going (ecs);



      return;

    }



  /* We stepped out of the stepping range.  */



  /* If we are stepping at the source level and entered the runtime

     loader dynamic symbol resolution code...



     EXEC_FORWARD: we keep on single stepping until we exit the run

     time loader code and reach the callee's address.



     EXEC_REVERSE: we've already executed the callee (backward), and

     the runtime loader code is handled just like any other

     undebuggable function call.  Now we need only keep stepping

     backward through the trampoline code, and that's handled further

     down, so there is nothing for us to do here.  */



  if (execution_direction != EXEC_REVERSE

      && ecs->event_thread->control.step_over_calls == STEP_OVER_UNDEBUGGABLE

      && in_solib_dynsym_resolve_code (stop_pc))

    {

      CORE_ADDR pc_after_resolver =

        gdbarch_skip_solib_resolver (gdbarch, stop_pc);



      if (debug_infrun)

         fprintf_unfiltered (gdb_stdlog,

                             "infrun: stepped into dynsym resolve code\n");



      if (pc_after_resolver)

        {

          /* Set up a step-resume breakpoint at the address

             indicated by SKIP_SOLIB_RESOLVER.  */

          struct symtab_and_line sr_sal;



          init_sal (&sr_sal);

          sr_sal.pc = pc_after_resolver;

          sr_sal.pspace = get_frame_program_space (frame);



          insert_step_resume_breakpoint_at_sal (gdbarch,

                                                sr_sal, null_frame_id);

        }



      keep_going (ecs);

      return;

    }



  if (ecs->event_thread->control.step_range_end != 1

      && (ecs->event_thread->control.step_over_calls == STEP_OVER_UNDEBUGGABLE

          || ecs->event_thread->control.step_over_calls == STEP_OVER_ALL)

      && get_frame_type (frame) == SIGTRAMP_FRAME)

    {

      if (debug_infrun)

         fprintf_unfiltered (gdb_stdlog,

                             "infrun: stepped into signal trampoline\n");

      /* The inferior, while doing a "step" or "next", has ended up in

         a signal trampoline (either by a signal being delivered or by

         the signal handler returning).  Just single-step until the

         inferior leaves the trampoline (either by calling the handler

         or returning).  */

      keep_going (ecs);

      return;

    }



  /* If we're in the return path from a shared library trampoline,

     we want to proceed through the trampoline when stepping.  */

  /* macro/2012-04-25: This needs to come before the subroutine

     call check below as on some targets return trampolines look

     like subroutine calls (MIPS16 return thunks).  */

  if (gdbarch_in_solib_return_trampoline (gdbarch,

                                          stop_pc, ecs->stop_func_name)

      && ecs->event_thread->control.step_over_calls != STEP_OVER_NONE)

    {

      /* Determine where this trampoline returns.  */

      CORE_ADDR real_stop_pc;



      real_stop_pc = gdbarch_skip_trampoline_code (gdbarch, frame, stop_pc);



      if (debug_infrun)

         fprintf_unfiltered (gdb_stdlog,

                             "infrun: stepped into solib return tramp\n");



      /* Only proceed through if we know where it's going.  */

      if (real_stop_pc)

        {

          /* And put the step-breakpoint there and go until there.  */

          struct symtab_and_line sr_sal;



          init_sal (&sr_sal);        /* initialize to zeroes */

          sr_sal.pc = real_stop_pc;

          sr_sal.section = find_pc_overlay (sr_sal.pc);

          sr_sal.pspace = get_frame_program_space (frame);



          /* Do not specify what the fp should be when we stop since

             on some machines the prologue is where the new fp value

             is established.  */

          insert_step_resume_breakpoint_at_sal (gdbarch,

                                                sr_sal, null_frame_id);



          /* Restart without fiddling with the step ranges or

             other state.  */

          keep_going (ecs);

          return;

        }

    }



  /* Check for subroutine calls.  The check for the current frame

     equalling the step ID is not necessary - the check of the

     previous frame's ID is sufficient - but it is a common case and

     cheaper than checking the previous frame's ID.



     NOTE: frame_id_eq will never report two invalid frame IDs as

     being equal, so to get into this block, both the current and

     previous frame must have valid frame IDs.  */

  /* The outer_frame_id check is a heuristic to detect stepping

     through startup code.  If we step over an instruction which

     sets the stack pointer from an invalid value to a valid value,

     we may detect that as a subroutine call from the mythical

     "outermost" function.  This could be fixed by marking

     outermost frames as !stack_p,code_p,special_p.  Then the

     initial outermost frame, before sp was valid, would

     have code_addr == &_start.  See the comment in frame_id_eq

     for more.  */

  if (!frame_id_eq (get_stack_frame_id (frame),

                    ecs->event_thread->control.step_stack_frame_id)

      && (frame_id_eq (frame_unwind_caller_id (get_current_frame ()),

                       ecs->event_thread->control.step_stack_frame_id)

          && (!frame_id_eq (ecs->event_thread->control.step_stack_frame_id,

                            outer_frame_id)

              || step_start_function != find_pc_function (stop_pc))))

    {

      CORE_ADDR real_stop_pc;



      if (debug_infrun)

         fprintf_unfiltered (gdb_stdlog, "infrun: stepped into subroutine\n");



      if (ecs->event_thread->control.step_over_calls == STEP_OVER_NONE)

        {

          /* I presume that step_over_calls is only 0 when we're

             supposed to be stepping at the assembly language level

             ("stepi").  Just stop.  */

          /* And this works the same backward as frontward.  MVS */

          end_stepping_range (ecs);

          return;

        }



      /* Reverse stepping through solib trampolines.  */



      if (execution_direction == EXEC_REVERSE

          && ecs->event_thread->control.step_over_calls != STEP_OVER_NONE

          && (gdbarch_skip_trampoline_code (gdbarch, frame, stop_pc)

              || (ecs->stop_func_start == 0

                  && in_solib_dynsym_resolve_code (stop_pc))))

        {

          /* Any solib trampoline code can be handled in reverse

             by simply continuing to single-step.  We have already

             executed the solib function (backwards), and a few

             steps will take us back through the trampoline to the

             caller.  */

          keep_going (ecs);

          return;

        }



      if (ecs->event_thread->control.step_over_calls == STEP_OVER_ALL)

        {

          /* We're doing a "next".



             Normal (forward) execution: set a breakpoint at the

             callee's return address (the address at which the caller

             will resume).



             Reverse (backward) execution.  set the step-resume

             breakpoint at the start of the function that we just

             stepped into (backwards), and continue to there.  When we

             get there, we'll need to single-step back to the caller.  */



          if (execution_direction == EXEC_REVERSE)

            {

              /* If we're already at the start of the function, we've either

                 just stepped backward into a single instruction function,

                 or stepped back out of a signal handler to the first instruction

                 of the function.  Just keep going, which will single-step back

                 to the caller.  */

              if (ecs->stop_func_start != stop_pc && ecs->stop_func_start != 0)

                {

                  struct symtab_and_line sr_sal;



                  /* Normal function call return (static or dynamic).  */

                  init_sal (&sr_sal);

                  sr_sal.pc = ecs->stop_func_start;

                  sr_sal.pspace = get_frame_program_space (frame);

                  insert_step_resume_breakpoint_at_sal (gdbarch,

                                                        sr_sal, null_frame_id);

                }

            }

          else

            insert_step_resume_breakpoint_at_caller (frame);



          keep_going (ecs);

          return;

        }



      /* If we are in a function call trampoline (a stub between the

         calling routine and the real function), locate the real

         function.  That's what tells us (a) whether we want to step

         into it at all, and (b) what prologue we want to run to the

         end of, if we do step into it.  */

      real_stop_pc = skip_language_trampoline (frame, stop_pc);

      if (real_stop_pc == 0)

        real_stop_pc = gdbarch_skip_trampoline_code (gdbarch, frame, stop_pc);

      if (real_stop_pc != 0)

        ecs->stop_func_start = real_stop_pc;



      if (real_stop_pc != 0 && in_solib_dynsym_resolve_code (real_stop_pc))

        {

          struct symtab_and_line sr_sal;



          init_sal (&sr_sal);

          sr_sal.pc = ecs->stop_func_start;

          sr_sal.pspace = get_frame_program_space (frame);



          insert_step_resume_breakpoint_at_sal (gdbarch,

                                                sr_sal, null_frame_id);

          keep_going (ecs);

          return;

        }



      /* If we have line number information for the function we are

         thinking of stepping into and the function isn't on the skip

         list, step into it.



         If there are several symtabs at that PC (e.g. with include

         files), just want to know whether *any* of them have line

         numbers.  find_pc_line handles this.  */

      {

        struct symtab_and_line tmp_sal;



        tmp_sal = find_pc_line (ecs->stop_func_start, 0);

        if (tmp_sal.line != 0

            && !function_name_is_marked_for_skip (ecs->stop_func_name,

                                                  &tmp_sal))

          {

            if (execution_direction == EXEC_REVERSE)

              handle_step_into_function_backward (gdbarch, ecs);

            else

              handle_step_into_function (gdbarch, ecs);

            return;

          }

      }



      /* If we have no line number and the step-stop-if-no-debug is

         set, we stop the step so that the user has a chance to switch

         in assembly mode.  */

      if (ecs->event_thread->control.step_over_calls == STEP_OVER_UNDEBUGGABLE

          && step_stop_if_no_debug)

        {

          end_stepping_range (ecs);

          return;

        }



      if (execution_direction == EXEC_REVERSE)

        {

          /* If we're already at the start of the function, we've either just

             stepped backward into a single instruction function without line

             number info, or stepped back out of a signal handler to the first

             instruction of the function without line number info.  Just keep

             going, which will single-step back to the caller.  */

          if (ecs->stop_func_start != stop_pc)

            {

              /* Set a breakpoint at callee's start address.

                 From there we can step once and be back in the caller.  */

              struct symtab_and_line sr_sal;



              init_sal (&sr_sal);

              sr_sal.pc = ecs->stop_func_start;

              sr_sal.pspace = get_frame_program_space (frame);

              insert_step_resume_breakpoint_at_sal (gdbarch,

                                                    sr_sal, null_frame_id);

            }

        }

      else

        /* Set a breakpoint at callee's return address (the address

           at which the caller will resume).  */

        insert_step_resume_breakpoint_at_caller (frame);



      keep_going (ecs);

      return;

    }



  /* Reverse stepping through solib trampolines.  */



  if (execution_direction == EXEC_REVERSE

      && ecs->event_thread->control.step_over_calls != STEP_OVER_NONE)

    {

      if (gdbarch_skip_trampoline_code (gdbarch, frame, stop_pc)

          || (ecs->stop_func_start == 0

              && in_solib_dynsym_resolve_code (stop_pc)))

        {

          /* Any solib trampoline code can be handled in reverse

             by simply continuing to single-step.  We have already

             executed the solib function (backwards), and a few

             steps will take us back through the trampoline to the

             caller.  */

          keep_going (ecs);

          return;

        }

      else if (in_solib_dynsym_resolve_code (stop_pc))

        {

          /* Stepped backward into the solib dynsym resolver.

             Set a breakpoint at its start and continue, then

             one more step will take us out.  */

          struct symtab_and_line sr_sal;



          init_sal (&sr_sal);

          sr_sal.pc = ecs->stop_func_start;

          sr_sal.pspace = get_frame_program_space (frame);

          insert_step_resume_breakpoint_at_sal (gdbarch,

                                                sr_sal, null_frame_id);

          keep_going (ecs);

          return;

        }

    }



  stop_pc_sal = find_pc_line (stop_pc, 0);



  /* NOTE: tausq/2004-05-24: This if block used to be done before all

     the trampoline processing logic, however, there are some trampolines

     that have no names, so we should do trampoline handling first.  */

  if (ecs->event_thread->control.step_over_calls == STEP_OVER_UNDEBUGGABLE

      && ecs->stop_func_name == NULL

      && stop_pc_sal.line == 0)

    {

      if (debug_infrun)

         fprintf_unfiltered (gdb_stdlog,

                             "infrun: stepped into undebuggable function\n");



      /* The inferior just stepped into, or returned to, an

         undebuggable function (where there is no debugging information

         and no line number corresponding to the address where the

         inferior stopped).  Since we want to skip this kind of code,

         we keep going until the inferior returns from this

         function - unless the user has asked us not to (via

         set step-mode) or we no longer know how to get back

         to the call site.  */

      if (step_stop_if_no_debug

          || !frame_id_p (frame_unwind_caller_id (frame)))

        {

          /* If we have no line number and the step-stop-if-no-debug

             is set, we stop the step so that the user has a chance to

             switch in assembly mode.  */

          end_stepping_range (ecs);

          return;

        }

      else

        {

          /* Set a breakpoint at callee's return address (the address

             at which the caller will resume).  */

          insert_step_resume_breakpoint_at_caller (frame);

          keep_going (ecs);

          return;

        }

    }



  if (ecs->event_thread->control.step_range_end == 1)

    {

      /* It is stepi or nexti.  We always want to stop stepping after

         one instruction.  */

      if (debug_infrun)

         fprintf_unfiltered (gdb_stdlog, "infrun: stepi/nexti\n");

      end_stepping_range (ecs);

      return;

    }



  if (stop_pc_sal.line == 0)

    {

      /* We have no line number information.  That means to stop

         stepping (does this always happen right after one instruction,

         when we do "s" in a function with no line numbers,

         or can this happen as a result of a return or longjmp?).  */

      if (debug_infrun)

         fprintf_unfiltered (gdb_stdlog, "infrun: no line number info\n");

      end_stepping_range (ecs);

      return;

    }



  /* Look for "calls" to inlined functions, part one.  If the inline

     frame machinery detected some skipped call sites, we have entered

     a new inline function.  */



  if (frame_id_eq (get_frame_id (get_current_frame ()),

                   ecs->event_thread->control.step_frame_id)

      && inline_skipped_frames (ecs->ptid))

    {

      struct symtab_and_line call_sal;



      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog,

                            "infrun: stepped into inlined function\n");



      find_frame_sal (get_current_frame (), &call_sal);



      if (ecs->event_thread->control.step_over_calls != STEP_OVER_ALL)

        {

          /* For "step", we're going to stop.  But if the call site

             for this inlined function is on the same source line as

             we were previously stepping, go down into the function

             first.  Otherwise stop at the call site.  */



          if (call_sal.line == ecs->event_thread->current_line

              && call_sal.symtab == ecs->event_thread->current_symtab)

            step_into_inline_frame (ecs->ptid);



          end_stepping_range (ecs);

          return;

        }

      else

        {

          /* For "next", we should stop at the call site if it is on a

             different source line.  Otherwise continue through the

             inlined function.  */

          if (call_sal.line == ecs->event_thread->current_line

              && call_sal.symtab == ecs->event_thread->current_symtab)

            keep_going (ecs);

          else

            end_stepping_range (ecs);

          return;

        }

    }



  /* Look for "calls" to inlined functions, part two.  If we are still

     in the same real function we were stepping through, but we have

     to go further up to find the exact frame ID, we are stepping

     through a more inlined call beyond its call site.  */



  if (get_frame_type (get_current_frame ()) == INLINE_FRAME

      && !frame_id_eq (get_frame_id (get_current_frame ()),

                       ecs->event_thread->control.step_frame_id)

      && stepped_in_from (get_current_frame (),

                          ecs->event_thread->control.step_frame_id))

    {

      if (debug_infrun)

        fprintf_unfiltered (gdb_stdlog,

                            "infrun: stepping through inlined function\n");



      if (ecs->event_thread->control.step_over_calls == STEP_OVER_ALL)

        keep_going (ecs);

      else

        end_stepping_range (ecs);

      return;

    }



  if ((stop_pc == stop_pc_sal.pc)

      && (ecs->event_thread->current_line != stop_pc_sal.line

           || ecs->event_thread->current_symtab != stop_pc_sal.symtab))

    {

      /* We are at the start of a different line.  So stop.  Note that

         we don't stop if we step into the middle of a different line.

         That is said to make things like for (;;) statements work

         better.  */

      if (debug_infrun)

         fprintf_unfiltered (gdb_stdlog,

                             "infrun: stepped to a different line\n");

      end_stepping_range (ecs);

      return;

    }



  /* We aren't done stepping.



     Optimize by setting the stepping range to the line.

     (We might not be in the original line, but if we entered a

     new line in mid-statement, we continue stepping.  This makes

     things like for(;;) statements work better.)  */



  ecs->event_thread->control.step_range_start = stop_pc_sal.pc;

  ecs->event_thread->control.step_range_end = stop_pc_sal.end;

  ecs->event_thread->control.may_range_step = 1;

  set_step_info (frame, stop_pc_sal);



  if (debug_infrun)

     fprintf_unfiltered (gdb_stdlog, "infrun: keep going\n");

  keep_going (ecs);

}



/* In all-stop mode, if we're currently stepping but have stopped in

   some other thread, we may need to switch back to the stepped

   thread.  Returns true we set the inferior running, false if we left

   it stopped (and the event needs further processing).  */



static int

‌switch_back_to_stepped_thread (struct execution_control_state *ecs)

{

  if (!non_stop)

    {

      struct thread_info *tp;

      struct thread_info *stepping_thread;

      struct thread_info *step_over;



      /* If any thread is blocked on some internal breakpoint, and we

         simply need to step over that breakpoint to get it going

         again, do that first.  */



      /* However, if we see an event for the stepping thread, then we

         know all other threads have been moved past their breakpoints

         already.  Let the caller check whether the step is finished,

         etc., before deciding to move it past a breakpoint.  */

      if (ecs->event_thread->control.step_range_end != 0)

        return 0;



      /* Check if the current thread is blocked on an incomplete

         step-over, interrupted by a random signal.  */

      if (ecs->event_thread->control.trap_expected

          && ecs->event_thread->suspend.stop_signal != GDB_SIGNAL_TRAP)

        {

          if (debug_infrun)

            {

              fprintf_unfiltered (gdb_stdlog,

                                  "infrun: need to finish step-over of [%s]\n",

                                  target_pid_to_str (ecs->event_thread->ptid));

            }

          keep_going (ecs);

          return 1;

        }



      /* Check if the current thread is blocked by a single-step

         breakpoint of another thread.  */

      if (ecs->hit_singlestep_breakpoint)

       {

         if (debug_infrun)

           {

             fprintf_unfiltered (gdb_stdlog,

                                 "infrun: need to step [%s] over single-step "

                                 "breakpoint\n",

                                 target_pid_to_str (ecs->ptid));

           }

         keep_going (ecs);

         return 1;

       }



      /* Otherwise, we no longer expect a trap in the current thread.

         Clear the trap_expected flag before switching back -- this is

         what keep_going does as well, if we call it.  */

      ecs->event_thread->control.trap_expected = 0;



      /* Likewise, clear the signal if it should not be passed.  */

      if (!signal_program[ecs->event_thread->suspend.stop_signal])

        ecs->event_thread->suspend.stop_signal = GDB_SIGNAL_0;



      /* If scheduler locking applies even if not stepping, there's no

         need to walk over threads.  Above we've checked whether the

         current thread is stepping.  If some other thread not the

         event thread is stepping, then it must be that scheduler

         locking is not in effect.  */

      if (schedlock_applies (0))

        return 0;



      /* Look for the stepping/nexting thread, and check if any other

         thread other than the stepping thread needs to start a

         step-over.  Do all step-overs before actually proceeding with

         step/next/etc.  */

      stepping_thread = NULL;

      step_over = NULL;

      ALL_NON_EXITED_THREADS (tp)

        {

          /* Ignore threads of processes we're not resuming.  */

          if (!sched_multi

              && ptid_get_pid (tp->ptid) != ptid_get_pid (inferior_ptid))

            continue;



          /* When stepping over a breakpoint, we lock all threads

             except the one that needs to move past the breakpoint.

             If a non-event thread has this set, the "incomplete

             step-over" check above should have caught it earlier.  */

          gdb_assert (!tp->control.trap_expected);



          /* Did we find the stepping thread?  */

          if (tp->control.step_range_end)

            {

              /* Yep.  There should only one though.  */

              gdb_assert (stepping_thread == NULL);



              /* The event thread is handled at the top, before we

                 enter this loop.  */

              gdb_assert (tp != ecs->event_thread);



              /* If some thread other than the event thread is

                 stepping, then scheduler locking can't be in effect,

                 otherwise we wouldn't have resumed the current event

                 thread in the first place.  */

              gdb_assert (!schedlock_applies (currently_stepping (tp)));



              stepping_thread = tp;

            }

          else if (thread_still_needs_step_over (tp))

            {

              step_over = tp;



              /* At the top we've returned early if the event thread

                 is stepping.  If some other thread not the event

                 thread is stepping, then scheduler locking can't be

                 in effect, and we can resume this thread.  No need to

                 keep looking for the stepping thread then.  */

              break;

            }

        }



      if (step_over != NULL)

        {

          tp = step_over;

          if (debug_infrun)

            {

              fprintf_unfiltered (gdb_stdlog,

                                  "infrun: need to step-over [%s]\n",

                                  target_pid_to_str (tp->ptid));

            }



          /* Only the stepping thread should have this set.  */

          gdb_assert (tp->control.step_range_end == 0);



          ecs->ptid = tp->ptid;

          ecs->event_thread = tp;

          switch_to_thread (ecs->ptid);

          keep_going (ecs);

          return 1;

        }



      if (stepping_thread != NULL)

        {

          struct frame_info *frame;

          struct gdbarch *gdbarch;



          tp = stepping_thread;



          /* If the stepping thread exited, then don't try to switch

             back and resume it, which could fail in several different

             ways depending on the target.  Instead, just keep going.



             We can find a stepping dead thread in the thread list in

             two cases:



             - The target supports thread exit events, and when the

             target tries to delete the thread from the thread list,

             inferior_ptid pointed at the exiting thread.  In such

             case, calling delete_thread does not really remove the

             thread from the list; instead, the thread is left listed,

             with 'exited' state.



             - The target's debug interface does not support thread

             exit events, and so we have no idea whatsoever if the

             previously stepping thread is still alive.  For that

             reason, we need to synchronously query the target

             now.  */

          if (is_exited (tp->ptid)

              || !target_thread_alive (tp->ptid))

            {

              if (debug_infrun)

                fprintf_unfiltered (gdb_stdlog,

                                    "infrun: not switching back to "

                                    "stepped thread, it has vanished\n");



              delete_thread (tp->ptid);

              keep_going (ecs);

              return 1;

            }



          if (debug_infrun)

            fprintf_unfiltered (gdb_stdlog,

                                "infrun: switching back to stepped thread\n");



          ecs->event_thread = tp;

          ecs->ptid = tp->ptid;

          context_switch (ecs->ptid);



          stop_pc = regcache_read_pc (get_thread_regcache (ecs->ptid));

          frame = get_current_frame ();

          gdbarch = get_frame_arch (frame);



          /* If the PC of the thread we were trying to single-step has

             changed, then that thread has trapped or been signaled,

             but the event has not been reported to GDB yet.  Re-poll

             the target looking for this particular thread's event

             (i.e. temporarily enable schedlock) by:



               - setting a break at the current PC

               - resuming that particular thread, only (by setting

                 trap expected)



             This prevents us continuously moving the single-step

             breakpoint forward, one instruction at a time,

             overstepping.  */



          if (stop_pc != tp->prev_pc)

            {

              if (debug_infrun)

                fprintf_unfiltered (gdb_stdlog,

                                    "infrun: expected thread advanced also\n");



              /* Clear the info of the previous step-over, as it's no

                 longer valid.  It's what keep_going would do too, if

                 we called it.  Must do this before trying to insert

                 the sss breakpoint, otherwise if we were previously

                 trying to step over this exact address in another

                 thread, the breakpoint ends up not installed.  */

              clear_step_over_info ();



              insert_single_step_breakpoint (get_frame_arch (frame),

                                             get_frame_address_space (frame),

                                             stop_pc);

              ecs->event_thread->control.trap_expected = 1;



              resume (0, GDB_SIGNAL_0);

              prepare_to_wait (ecs);

            }

          else

            {

              if (debug_infrun)

                fprintf_unfiltered (gdb_stdlog,

                                    "infrun: expected thread still "

                                    "hasn't advanced\n");

              keep_going (ecs);

            }



          return 1;

        }

    }

  return 0;

}



/* Is thread TP in the middle of single-stepping?  */



static int

‌currently_stepping (struct thread_info *tp)

{

  return ((tp->control.step_range_end

           && tp->control.step_resume_breakpoint == NULL)

          || tp->control.trap_expected

          || tp->stepped_breakpoint

          || bpstat_should_step ());

}



/* Inferior has stepped into a subroutine call with source code that

   we should not step over.  Do step to the first line of code in

   it.  */



static void

‌handle_step_into_function (struct gdbarch *gdbarch,

                           struct execution_control_state *ecs)

{

  struct compunit_symtab *cust;

  struct symtab_and_line stop_func_sal, sr_sal;



  fill_in_stop_func (gdbarch, ecs);



  cust = find_pc_compunit_symtab (stop_pc);

  if (cust != NULL && compunit_language (cust) != language_asm)

    ecs->stop_func_start = gdbarch_skip_prologue (gdbarch,

                                                  ecs->stop_func_start);



  stop_func_sal = find_pc_line (ecs->stop_func_start, 0);

  /* Use the step_resume_break to step until the end of the prologue,

     even if that involves jumps (as it seems to on the vax under

     4.2).  */

  /* If the prologue ends in the middle of a source line, continue to

     the end of that source line (if it is still within the function).

     Otherwise, just go to end of prologue.  */

  if (stop_func_sal.end

      && stop_func_sal.pc != ecs->stop_func_start

      && stop_func_sal.end < ecs->stop_func_end)

    ecs->stop_func_start = stop_func_sal.end;



  /* Architectures which require breakpoint adjustment might not be able

     to place a breakpoint at the computed address.  If so, the test

     ``ecs->stop_func_start == stop_pc'' will never succeed.  Adjust

     ecs->stop_func_start to an address at which a breakpoint may be

     legitimately placed.



     Note:  kevinb/2004-01-19:  On FR-V, if this adjustment is not

     made, GDB will enter an infinite loop when stepping through

     optimized code consisting of VLIW instructions which contain

     subinstructions corresponding to different source lines.  On

     FR-V, it's not permitted to place a breakpoint on any but the

     first subinstruction of a VLIW instruction.  When a breakpoint is

     set, GDB will adjust the breakpoint address to the beginning of

     the VLIW instruction.  Thus, we need to make the corresponding

     adjustment here when computing the stop address.  */



  if (gdbarch_adjust_breakpoint_address_p (gdbarch))

    {

      ecs->stop_func_start

        = gdbarch_adjust_breakpoint_address (gdbarch,

                                             ecs->stop_func_start);

    }



  if (ecs->stop_func_start == stop_pc)

    {

      /* We are already there: stop now.  */

      end_stepping_range (ecs);

      return;

    }

  else

    {

      /* Put the step-breakpoint there and go until there.  */

      init_sal (&sr_sal);        /* initialize to zeroes */

      sr_sal.pc = ecs->stop_func_start;

      sr_sal.section = find_pc_overlay (ecs->stop_func_start);

      sr_sal.pspace = get_frame_program_space (get_current_frame ());



      /* Do not specify what the fp should be when we stop since on

         some machines the prologue is where the new fp value is

         established.  */

      insert_step_resume_breakpoint_at_sal (gdbarch, sr_sal, null_frame_id);



      /* And make sure stepping stops right away then.  */

      ecs->event_thread->control.step_range_end

        = ecs->event_thread->control.step_range_start;

    }

  keep_going (ecs);

}



/* Inferior has stepped backward into a subroutine call with source

   code that we should not step over.  Do step to the beginning of the

   last line of code in it.  */



static void

‌handle_step_into_function_backward (struct gdbarch *gdbarch,

                                    struct execution_control_state *ecs)

{

  struct compunit_symtab *cust;

  struct symtab_and_line stop_func_sal;



  fill_in_stop_func (gdbarch, ecs);



  cust = find_pc_compunit_symtab (stop_pc);

  if (cust != NULL && compunit_language (cust) != language_asm)

    ecs->stop_func_start = gdbarch_skip_prologue (gdbarch,

                                                  ecs->stop_func_start);



  stop_func_sal = find_pc_line (stop_pc, 0);



  /* OK, we're just going to keep stepping here.  */

  if (stop_func_sal.pc == stop_pc)

    {

      /* We're there already.  Just stop stepping now.  */

      end_stepping_range (ecs);

    }

  else

    {

      /* Else just reset the step range and keep going.

         No step-resume breakpoint, they don't work for

         epilogues, which can have multiple entry paths.  */

      ecs->event_thread->control.step_range_start = stop_func_sal.pc;

      ecs->event_thread->control.step_range_end = stop_func_sal.end;

      keep_going (ecs);

    }

  return;

}



/* Insert a "step-resume breakpoint" at SR_SAL with frame ID SR_ID.

   This is used to both functions and to skip over code.  */



static void

‌insert_step_resume_breakpoint_at_sal_1 (struct gdbarch *gdbarch,

                                        struct symtab_and_line sr_sal,

                                        struct frame_id sr_id,

                                        enum bptype sr_type)

{

  /* There should never be more than one step-resume or longjmp-resume

     breakpoint per thread, so we should never be setting a new

     step_resume_breakpoint when one is already active.  */

  gdb_assert (inferior_thread ()->control.step_resume_breakpoint == NULL);

  gdb_assert (sr_type == bp_step_resume || sr_type == bp_hp_step_resume);



  if (debug_infrun)

    fprintf_unfiltered (gdb_stdlog,

                        "infrun: inserting step-resume breakpoint at %s\n",

                        paddress (gdbarch, sr_sal.pc));



  inferior_thread ()->control.step_resume_breakpoint

    = set_momentary_breakpoint (gdbarch, sr_sal, sr_id, sr_type);

}



void

‌insert_step_resume_breakpoint_at_sal (struct gdbarch *gdbarch,

                                      struct symtab_and_line sr_sal,

                                      struct frame_id sr_id)

{

  insert_step_resume_breakpoint_at_sal_1 (gdbarch,

                                          sr_sal, sr_id,

                                          bp_step_resume);

}



/* Insert a "high-priority step-resume breakpoint" at RETURN_FRAME.pc.

   This is used to skip a potential signal handler.



   This is called with the interrupted function's frame.  The signal

   handler, when it returns, will resume the interrupted function at

   RETURN_FRAME.pc.  */



static void

‌insert_hp_step_resume_breakpoint_at_frame (struct frame_info *return_frame)

{

  struct symtab_and_line sr_sal;

  struct gdbarch *gdbarch;



  gdb_assert (return_frame != NULL);

  init_sal (&sr_sal);                /* initialize to zeros */



  gdbarch = get_frame_arch (return_frame);

  sr_sal.pc = gdbarch_addr_bits_remove (gdbarch, get_frame_pc (return_frame));

  sr_sal.section = find_pc_overlay (sr_sal.pc);

  sr_sal.pspace = get_frame_program_space (return_frame);



  insert_step_resume_breakpoint_at_sal_1 (gdbarch, sr_sal,

                                          get_stack_frame_id (return_frame),

                                          bp_hp_step_resume);

}



/* Insert a "step-resume breakpoint" at the previous frame's PC.  This

   is used to skip a function after stepping into it (for "next" or if

   the called function has no debugging information).



   The current function has almost always been reached by single

   stepping a call or return instruction.  NEXT_FRAME belongs to the

   current function, and the breakpoint will be set at the caller's

   resume address.



   This is a separate function rather than reusing

   insert_hp_step_resume_breakpoint_at_frame in order to avoid

   get_prev_frame, which may stop prematurely (see the implementation

   of frame_unwind_caller_id for an example).  */



static void

‌insert_step_resume_breakpoint_at_caller (struct frame_info *next_frame)

{

  struct symtab_and_line sr_sal;

  struct gdbarch *gdbarch;



  /* We shouldn't have gotten here if we don't know where the call site

     is.  */

  gdb_assert (frame_id_p (frame_unwind_caller_id (next_frame)));



  init_sal (&sr_sal);                /* initialize to zeros */



  gdbarch = frame_unwind_caller_arch (next_frame);

  sr_sal.pc = gdbarch_addr_bits_remove (gdbarch,

                                        frame_unwind_caller_pc (next_frame));

  sr_sal.section = find_pc_overlay (sr_sal.pc);

  sr_sal.pspace = frame_unwind_program_space (next_frame);



  insert_step_resume_breakpoint_at_sal (gdbarch, sr_sal,

                                        frame_unwind_caller_id (next_frame));

}



/* Insert a "longjmp-resume" breakpoint at PC.  This is used to set a

   new breakpoint at the target of a jmp_buf.  The handling of

   longjmp-resume uses the same mechanisms used for handling

   "step-resume" breakpoints.  */



static void

‌insert_longjmp_resume_breakpoint (struct gdbarch *gdbarch, CORE_ADDR pc)

{

  /* There should never be more than one longjmp-resume breakpoint per

     thread, so we should never be setting a new

     longjmp_resume_breakpoint when one is already active.  */

  gdb_assert (inferior_thread ()->control.exception_resume_breakpoint == NULL);



  if (debug_infrun)

    fprintf_unfiltered (gdb_stdlog,

                        "infrun: inserting longjmp-resume breakpoint at %s\n",

                        paddress (gdbarch, pc));



  inferior_thread ()->control.exception_resume_breakpoint =

    set_momentary_breakpoint_at_pc (gdbarch, pc, bp_longjmp_resume);

}



/* Insert an exception resume breakpoint.  TP is the thread throwing

   the exception.  The block B is the block of the unwinder debug hook

   function.  FRAME is the frame corresponding to the call to this

   function.  SYM is the symbol of the function argument holding the

   target PC of the exception.  */



static void

‌insert_exception_resume_breakpoint (struct thread_info *tp,

                                    const struct block *b,

                                    struct frame_info *frame,

                                    struct symbol *sym)

{

  volatile struct gdb_exception e;



  /* We want to ignore errors here.  */

  TRY_CATCH (e, RETURN_MASK_ERROR)

    {

      struct symbol *vsym;

      struct value *value;

      CORE_ADDR handler;

      struct breakpoint *bp;



      vsym = lookup_symbol (SYMBOL_LINKAGE_NAME (sym), b, VAR_DOMAIN, NULL);

      value = read_var_value (vsym, frame);

      /* If the value was optimized out, revert to the old behavior.  */

      if (! value_optimized_out (value))

        {

          handler = value_as_address (value);



          if (debug_infrun)

            fprintf_unfiltered (gdb_stdlog,

                                "infrun: exception resume at %lx\n",

                                (unsigned long) handler);



          bp = set_momentary_breakpoint_at_pc (get_frame_arch (frame),

                                               handler, bp_exception_resume);



          /* set_momentary_breakpoint_at_pc invalidates FRAME.  */

          frame = NULL;



          bp->thread = tp->num;

          inferior_thread ()->control.exception_resume_breakpoint = bp;

        }

    }

}



/* A helper for check_exception_resume that sets an

   exception-breakpoint based on a SystemTap probe.  */



static void

‌insert_exception_resume_from_probe (struct thread_info *tp,

                                    const struct bound_probe *probe,

                                    struct frame_info *frame)

{

  struct value *arg_value;

  CORE_ADDR handler;

  struct breakpoint *bp;



  arg_value = probe_safe_evaluate_at_pc (frame, 1);

  if (!arg_value)

    return;



  handler = value_as_address (arg_value);



  if (debug_infrun)

    fprintf_unfiltered (gdb_stdlog,

                        "infrun: exception resume at %s\n",

                        paddress (get_objfile_arch (probe->objfile),

                                  handler));



  bp = set_momentary_breakpoint_at_pc (get_frame_arch (frame),

                                       handler, bp_exception_resume);

  bp->thread = tp->num;

  inferior_thread ()->control.exception_resume_breakpoint = bp;

}



/* This is called when an exception has been intercepted.  Check to

   see whether the exception's destination is of interest, and if so,

   set an exception resume breakpoint there.  */



static void

‌check_exception_resume (struct execution_control_state *ecs,

                        struct frame_info *frame)

{

  volatile struct gdb_exception e;

  struct bound_probe probe;

  struct symbol *func;



  /* First see if this exception unwinding breakpoint was set via a

     SystemTap probe point.  If so, the probe has two arguments: the

     CFA and the HANDLER.  We ignore the CFA, extract the handler, and

     set a breakpoint there.  */

  probe = find_probe_by_pc (get_frame_pc (frame));

  if (probe.probe)

    {

      insert_exception_resume_from_probe (ecs->event_thread, &probe, frame);

      return;

    }



  func = get_frame_function (frame);

  if (!func)

    return;



  TRY_CATCH (e, RETURN_MASK_ERROR)

    {

      const struct block *b;

      struct block_iterator iter;

      struct symbol *sym;

      int argno = 0;



      /* The exception breakpoint is a thread-specific breakpoint on

         the unwinder's debug hook, declared as:



         void _Unwind_DebugHook (void *cfa, void *handler);



         The CFA argument indicates the frame to which control is

         about to be transferred.  HANDLER is the destination PC.



         We ignore the CFA and set a temporary breakpoint at HANDLER.

         This is not extremely efficient but it avoids issues in gdb

         with computing the DWARF CFA, and it also works even in weird

         cases such as throwing an exception from inside a signal

         handler.  */



      b = SYMBOL_BLOCK_VALUE (func);

      ALL_BLOCK_SYMBOLS (b, iter, sym)

        {

          if (!SYMBOL_IS_ARGUMENT (sym))

            continue;



          if (argno == 0)

            ++argno;

          else

            {

              insert_exception_resume_breakpoint (ecs->event_thread,

                                                  b, frame, sym);

              break;

            }

        }

    }

}



static void

‌stop_waiting (struct execution_control_state *ecs)

{

  if (debug_infrun)

    fprintf_unfiltered (gdb_stdlog, "infrun: stop_waiting\n");



  clear_step_over_info ();



  /* Let callers know we don't want to wait for the inferior anymore.  */

  ecs->wait_some_more = 0;

}



/* Called when we should continue running the inferior, because the

   current event doesn't cause a user visible stop.  This does the

   resuming part; waiting for the next event is done elsewhere.  */



static void

‌keep_going (struct execution_control_state *ecs)

{

  /* Make sure normal_stop is called if we get a QUIT handled before

     reaching resume.  */

  struct cleanup *old_cleanups = make_cleanup (resume_cleanups, 0);



  /* Save the pc before execution, to compare with pc after stop.  */

  ecs->event_thread->prev_pc

    = regcache_read_pc (get_thread_regcache (ecs->ptid));



  if (ecs->event_thread->control.trap_expected

      && ecs->event_thread->suspend.stop_signal != GDB_SIGNAL_TRAP)

    {

      /* We haven't yet gotten our trap, and either: intercepted a

         non-signal event (e.g., a fork); or took a signal which we

         are supposed to pass through to the inferior.  Simply

         continue.  */

      discard_cleanups (old_cleanups);

      resume (currently_stepping (ecs->event_thread),

              ecs->event_thread->suspend.stop_signal);

    }

  else

    {

      volatile struct gdb_exception e;

      struct regcache *regcache = get_current_regcache ();

      int remove_bp;

      int remove_wps;



      /* Either the trap was not expected, but we are continuing

         anyway (if we got a signal, the user asked it be passed to

         the child)

         -- or --

         We got our expected trap, but decided we should resume from

         it.



         We're going to run this baby now!



         Note that insert_breakpoints won't try to re-insert

         already inserted breakpoints.  Therefore, we don't

         care if breakpoints were already inserted, or not.  */



      /* If we need to step over a breakpoint, and we're not using

         displaced stepping to do so, insert all breakpoints

         (watchpoints, etc.) but the one we're stepping over, step one

         instruction, and then re-insert the breakpoint when that step

         is finished.  */



      remove_bp = (ecs->hit_singlestep_breakpoint

                   || thread_still_needs_step_over (ecs->event_thread));

      remove_wps = (ecs->event_thread->stepping_over_watchpoint

                    && !target_have_steppable_watchpoint);



      if (remove_bp && !use_displaced_stepping (get_regcache_arch (regcache)))

        {

          set_step_over_info (get_regcache_aspace (regcache),

                              regcache_read_pc (regcache), remove_wps);

        }

      else if (remove_wps)

        set_step_over_info (NULL, 0, remove_wps);

      else

        clear_step_over_info ();



      /* Stop stepping if inserting breakpoints fails.  */

      TRY_CATCH (e, RETURN_MASK_ERROR)

        {

          insert_breakpoints ();

        }

      if (e.reason < 0)

        {

          exception_print (gdb_stderr, e);

          stop_waiting (ecs);

          return;

        }



      ecs->event_thread->control.trap_expected = (remove_bp || remove_wps);



      /* Do not deliver GDB_SIGNAL_TRAP (except when the user

         explicitly specifies that such a signal should be delivered

         to the target program).  Typically, that would occur when a

         user is debugging a target monitor on a simulator: the target

         monitor sets a breakpoint; the simulator encounters this

         breakpoint and halts the simulation handing control to GDB;

         GDB, noting that the stop address doesn't map to any known

         breakpoint, returns control back to the simulator; the

         simulator then delivers the hardware equivalent of a

         GDB_SIGNAL_TRAP to the program being debugged.         */

      if (ecs->event_thread->suspend.stop_signal == GDB_SIGNAL_TRAP

          && !signal_program[ecs->event_thread->suspend.stop_signal])

        ecs->event_thread->suspend.stop_signal = GDB_SIGNAL_0;



      discard_cleanups (old_cleanups);

      resume (currently_stepping (ecs->event_thread),

              ecs->event_thread->suspend.stop_signal);

    }



  prepare_to_wait (ecs);

}



/* This function normally comes after a resume, before

   handle_inferior_event exits.  It takes care of any last bits of

   housekeeping, and sets the all-important wait_some_more flag.  */



static void

‌prepare_to_wait (struct execution_control_state *ecs)

{

  if (debug_infrun)

    fprintf_unfiltered (gdb_stdlog, "infrun: prepare_to_wait\n");



  /* This is the old end of the while loop.  Let everybody know we

     want to wait for the inferior some more and get called again

     soon.  */

  ecs->wait_some_more = 1;

}



/* We are done with the step range of a step/next/si/ni command.

   Called once for each n of a "step n" operation.  */



static void

‌end_stepping_range (struct execution_control_state *ecs)

{

  ecs->event_thread->control.stop_step = 1;

  stop_waiting (ecs);

}



/* Several print_*_reason functions to print why the inferior has stopped.

   We always print something when the inferior exits, or receives a signal.

   The rest of the cases are dealt with later on in normal_stop and

   print_it_typical.  Ideally there should be a call to one of these

   print_*_reason functions functions from handle_inferior_event each time

   stop_waiting is called.



   Note that we don't call these directly, instead we delegate that to

   the interpreters, through observers.  Interpreters then call these

   with whatever uiout is right.  */



void

‌print_end_stepping_range_reason (struct ui_out *uiout)

{

  /* For CLI-like interpreters, print nothing.  */



  if (ui_out_is_mi_like_p (uiout))

    {

      ui_out_field_string (uiout, "reason",

                           async_reason_lookup (EXEC_ASYNC_END_STEPPING_RANGE));

    }

}



void

‌print_signal_exited_reason (struct ui_out *uiout, enum gdb_signal siggnal)

{

  annotate_signalled ();

  if (ui_out_is_mi_like_p (uiout))

    ui_out_field_string

      (uiout, "reason", async_reason_lookup (EXEC_ASYNC_EXITED_SIGNALLED));

  ui_out_text (uiout, "\nProgram terminated with signal ");

  annotate_signal_name ();

  ui_out_field_string (uiout, "signal-name",

                       gdb_signal_to_name (siggnal));

  annotate_signal_name_end ();

  ui_out_text (uiout, ", ");

  annotate_signal_string ();

  ui_out_field_string (uiout, "signal-meaning",

                       gdb_signal_to_string (siggnal));

  annotate_signal_string_end ();

  ui_out_text (uiout, ".\n");

  ui_out_text (uiout, "The program no longer exists.\n");

}



void

‌print_exited_reason (struct ui_out *uiout, int exitstatus)

{

  struct inferior *inf = current_inferior ();

  const char *pidstr = target_pid_to_str (pid_to_ptid (inf->pid));



  annotate_exited (exitstatus);

  if (exitstatus)

    {

      if (ui_out_is_mi_like_p (uiout))

        ui_out_field_string (uiout, "reason",

                             async_reason_lookup (EXEC_ASYNC_EXITED));

      ui_out_text (uiout, "[Inferior ");

      ui_out_text (uiout, plongest (inf->num));

      ui_out_text (uiout, " (");

      ui_out_text (uiout, pidstr);

      ui_out_text (uiout, ") exited with code ");

      ui_out_field_fmt (uiout, "exit-code", "0%o", (unsigned int) exitstatus);

      ui_out_text (uiout, "]\n");

    }

  else

    {

      if (ui_out_is_mi_like_p (uiout))

        ui_out_field_string

          (uiout, "reason", async_reason_lookup (EXEC_ASYNC_EXITED_NORMALLY));

      ui_out_text (uiout, "[Inferior ");

      ui_out_text (uiout, plongest (inf->num));

      ui_out_text (uiout, " (");

      ui_out_text (uiout, pidstr);

      ui_out_text (uiout, ") exited normally]\n");

    }

}



void

‌print_signal_received_reason (struct ui_out *uiout, enum gdb_signal siggnal)

{

  annotate_signal ();



  if (siggnal == GDB_SIGNAL_0 && !ui_out_is_mi_like_p (uiout))

    {

      struct thread_info *t = inferior_thread ();



      ui_out_text (uiout, "\n[");

      ui_out_field_string (uiout, "thread-name",

                           target_pid_to_str (t->ptid));

      ui_out_field_fmt (uiout, "thread-id", "] #%d", t->num);

      ui_out_text (uiout, " stopped");

    }

  else

    {

      ui_out_text (uiout, "\nProgram received signal ");

      annotate_signal_name ();

      if (ui_out_is_mi_like_p (uiout))

        ui_out_field_string

          (uiout, "reason", async_reason_lookup (EXEC_ASYNC_SIGNAL_RECEIVED));

      ui_out_field_string (uiout, "signal-name",

                           gdb_signal_to_name (siggnal));

      annotate_signal_name_end ();

      ui_out_text (uiout, ", ");

      annotate_signal_string ();

      ui_out_field_string (uiout, "signal-meaning",

                           gdb_signal_to_string (siggnal));

      annotate_signal_string_end ();

    }

  ui_out_text (uiout, ".\n");

}



void

‌print_no_history_reason (struct ui_out *uiout)

{

  ui_out_text (uiout, "\nNo more reverse-execution history.\n");

}



/* Print current location without a level number, if we have changed

   functions or hit a breakpoint.  Print source line if we have one.

   bpstat_print contains the logic deciding in detail what to print,

   based on the event(s) that just occurred.  */



void

‌print_stop_event (struct target_waitstatus *ws)

{

  int bpstat_ret;

  int source_flag;

  int do_frame_printing = 1;

  struct thread_info *tp = inferior_thread ();



  bpstat_ret = bpstat_print (tp->control.stop_bpstat, ws->kind);

  switch (bpstat_ret)

    {

    case PRINT_UNKNOWN:

      /* FIXME: cagney/2002-12-01: Given that a frame ID does (or

         should) carry around the function and does (or should) use

         that when doing a frame comparison.  */

      if (tp->control.stop_step

          && frame_id_eq (tp->control.step_frame_id,

                          get_frame_id (get_current_frame ()))

          && step_start_function == find_pc_function (stop_pc))

        {

          /* Finished step, just print source line.  */

          source_flag = SRC_LINE;

        }

      else

        {

          /* Print location and source line.  */

          source_flag = SRC_AND_LOC;

        }

      break;

    case PRINT_SRC_AND_LOC:

      /* Print location and source line.  */

      source_flag = SRC_AND_LOC;

      break;

    case PRINT_SRC_ONLY:

      source_flag = SRC_LINE;

      break;

    case PRINT_NOTHING:

      /* Something bogus.  */

      source_flag = SRC_LINE;

      do_frame_printing = 0;

      break;

    default:

      internal_error (__FILE__, __LINE__, _("Unknown value."));

    }



  /* The behavior of this routine with respect to the source

     flag is:

     SRC_LINE: Print only source line

     LOCATION: Print only location

     SRC_AND_LOC: Print location and source line.  */

  if (do_frame_printing)

    print_stack_frame (get_selected_frame (NULL), 0, source_flag, 1);



  /* Display the auto-display expressions.  */

  do_displays ();

}



/* Here to return control to GDB when the inferior stops for real.

   Print appropriate messages, remove breakpoints, give terminal our modes.



   STOP_PRINT_FRAME nonzero means print the executing frame

   (pc, function, args, file, line number and line text).

   BREAKPOINTS_FAILED nonzero means stop was due to error

   attempting to insert breakpoints.  */



void

‌normal_stop (void)

{

  struct target_waitstatus last;

  ptid_t last_ptid;

  struct cleanup *old_chain = make_cleanup (null_cleanup, NULL);



  get_last_target_status (&last_ptid, &last);



  /* If an exception is thrown from this point on, make sure to

     propagate GDB's knowledge of the executing state to the

     frontend/user running state.  A QUIT is an easy exception to see

     here, so do this before any filtered output.  */

  if (!non_stop)

    make_cleanup (finish_thread_state_cleanup, &minus_one_ptid);

  else if (last.kind != TARGET_WAITKIND_SIGNALLED

           && last.kind != TARGET_WAITKIND_EXITED

           && last.kind != TARGET_WAITKIND_NO_RESUMED)

    make_cleanup (finish_thread_state_cleanup, &inferior_ptid);



  /* As we're presenting a stop, and potentially removing breakpoints,

     update the thread list so we can tell whether there are threads

     running on the target.  With target remote, for example, we can

     only learn about new threads when we explicitly update the thread

     list.  Do this before notifying the interpreters about signal

     stops, end of stepping ranges, etc., so that the "new thread"

     output is emitted before e.g., "Program received signal FOO",

     instead of after.  */

  update_thread_list ();



  if (last.kind == TARGET_WAITKIND_STOPPED && stopped_by_random_signal)

    observer_notify_signal_received (inferior_thread ()->suspend.stop_signal);



  /* As with the notification of thread events, we want to delay

     notifying the user that we've switched thread context until

     the inferior actually stops.



     There's no point in saying anything if the inferior has exited.

     Note that SIGNALLED here means "exited with a signal", not

     "received a signal".



     Also skip saying anything in non-stop mode.  In that mode, as we

     don't want GDB to switch threads behind the user's back, to avoid

     races where the user is typing a command to apply to thread x,

     but GDB switches to thread y before the user finishes entering

     the command, fetch_inferior_event installs a cleanup to restore

     the current thread back to the thread the user had selected right

     after this event is handled, so we're not really switching, only

     informing of a stop.  */

  if (!non_stop

      && !ptid_equal (previous_inferior_ptid, inferior_ptid)

      && target_has_execution

      && last.kind != TARGET_WAITKIND_SIGNALLED

      && last.kind != TARGET_WAITKIND_EXITED

      && last.kind != TARGET_WAITKIND_NO_RESUMED)

    {

      target_terminal_ours_for_output ();

      printf_filtered (_("[Switching to %s]\n"),

                       target_pid_to_str (inferior_ptid));

      annotate_thread_changed ();

      previous_inferior_ptid = inferior_ptid;

    }



  if (last.kind == TARGET_WAITKIND_NO_RESUMED)

    {

      gdb_assert (sync_execution || !target_can_async_p ());



      target_terminal_ours_for_output ();

      printf_filtered (_("No unwaited-for children left.\n"));

    }



  /* Note: this depends on the update_thread_list call above.  */

  if (!breakpoints_should_be_inserted_now () && target_has_execution)

    {

      if (remove_breakpoints ())

        {

          target_terminal_ours_for_output ();

          printf_filtered (_("Cannot remove breakpoints because "

                             "program is no longer writable.\nFurther "

                             "execution is probably impossible.\n"));

        }

    }



  /* If an auto-display called a function and that got a signal,

     delete that auto-display to avoid an infinite recursion.  */



  if (stopped_by_random_signal)

    disable_current_display ();



  /* Notify observers if we finished a "step"-like command, etc.  */

  if (target_has_execution

      && last.kind != TARGET_WAITKIND_SIGNALLED

      && last.kind != TARGET_WAITKIND_EXITED

      && inferior_thread ()->control.stop_step)

    {

      /* But not if in the middle of doing a "step n" operation for

         n > 1 */

      if (inferior_thread ()->step_multi)

        goto done;



      observer_notify_end_stepping_range ();

    }



  target_terminal_ours ();

  async_enable_stdin ();



  /* Set the current source location.  This will also happen if we

     display the frame below, but the current SAL will be incorrect

     during a user hook-stop function.  */

  if (has_stack_frames () && !stop_stack_dummy)

    set_current_sal_from_frame (get_current_frame ());



  /* Let the user/frontend see the threads as stopped, but do nothing

     if the thread was running an infcall.  We may be e.g., evaluating

     a breakpoint condition.  In that case, the thread had state

     THREAD_RUNNING before the infcall, and shall remain set to

     running, all without informing the user/frontend about state

     transition changes.  If this is actually a call command, then the

     thread was originally already stopped, so there's no state to

     finish either.  */

  if (target_has_execution && inferior_thread ()->control.in_infcall)

    discard_cleanups (old_chain);

  else

    do_cleanups (old_chain);



  /* Look up the hook_stop and run it (CLI internally handles problem

     of stop_command's pre-hook not existing).  */

  if (stop_command)

    catch_errors (hook_stop_stub, stop_command,

                  "Error while running hook_stop:\n", RETURN_MASK_ALL);



  if (!has_stack_frames ())

    goto done;



  if (last.kind == TARGET_WAITKIND_SIGNALLED

      || last.kind == TARGET_WAITKIND_EXITED)

    goto done;



  /* Select innermost stack frame - i.e., current frame is frame 0,

     and current location is based on that.

     Don't do this on return from a stack dummy routine,

     or if the program has exited.  */



  if (!stop_stack_dummy)

    {

      select_frame (get_current_frame ());



      /* If --batch-silent is enabled then there's no need to print the current

         source location, and to try risks causing an error message about

         missing source files.  */

      if (stop_print_frame && !batch_silent)

        print_stop_event (&last);

    }



  /* Save the function value return registers, if we care.

     We might be about to restore their previous contents.  */

  if (inferior_thread ()->control.proceed_to_finish

      && execution_direction != EXEC_REVERSE)

    {

      /* This should not be necessary.  */

      if (stop_registers)

        regcache_xfree (stop_registers);



      /* NB: The copy goes through to the target picking up the value of

         all the registers.  */

      stop_registers = regcache_dup (get_current_regcache ());

    }



  if (stop_stack_dummy == STOP_STACK_DUMMY)

    {

      /* Pop the empty frame that contains the stack dummy.

         This also restores inferior state prior to the call

         (struct infcall_suspend_state).  */

      struct frame_info *frame = get_current_frame ();



      gdb_assert (get_frame_type (frame) == DUMMY_FRAME);

      frame_pop (frame);

      /* frame_pop() calls reinit_frame_cache as the last thing it

         does which means there's currently no selected frame.  We

         don't need to re-establish a selected frame if the dummy call

         returns normally, that will be done by

         restore_infcall_control_state.  However, we do have to handle

         the case where the dummy call is returning after being

         stopped (e.g. the dummy call previously hit a breakpoint).

         We can't know which case we have so just always re-establish

         a selected frame here.  */

      select_frame (get_current_frame ());

    }



done:

  annotate_stopped ();



  /* Suppress the stop observer if we're in the middle of:



     - a step n (n > 1), as there still more steps to be done.



     - a "finish" command, as the observer will be called in

       finish_command_continuation, so it can include the inferior

       function's return value.



     - calling an inferior function, as we pretend we inferior didn't

       run at all.  The return value of the call is handled by the

       expression evaluator, through call_function_by_hand.  */



  if (!target_has_execution

      || last.kind == TARGET_WAITKIND_SIGNALLED

      || last.kind == TARGET_WAITKIND_EXITED

      || last.kind == TARGET_WAITKIND_NO_RESUMED

      || (!(inferior_thread ()->step_multi

            && inferior_thread ()->control.stop_step)

          && !(inferior_thread ()->control.stop_bpstat

               && inferior_thread ()->control.proceed_to_finish)

          && !inferior_thread ()->control.in_infcall))

    {

      if (!ptid_equal (inferior_ptid, null_ptid))

        observer_notify_normal_stop (inferior_thread ()->control.stop_bpstat,

                                     stop_print_frame);

      else

        observer_notify_normal_stop (NULL, stop_print_frame);

    }



  if (target_has_execution)

    {

      if (last.kind != TARGET_WAITKIND_SIGNALLED

          && last.kind != TARGET_WAITKIND_EXITED)

        /* Delete the breakpoint we stopped at, if it wants to be deleted.

           Delete any breakpoint that is to be deleted at the next stop.  */

        breakpoint_auto_delete (inferior_thread ()->control.stop_bpstat);

    }



  /* Try to get rid of automatically added inferiors that are no

     longer needed.  Keeping those around slows down things linearly.

     Note that this never removes the current inferior.  */

  prune_inferiors ();

}



static int

‌hook_stop_stub (void *cmd)

{

  execute_cmd_pre_hook ((struct cmd_list_element *) cmd);

  return (0);

}



int

‌signal_stop_state (int signo)

{

  return signal_stop[signo];

}



int

‌signal_print_state (int signo)

{

  return signal_print[signo];

}



int

‌signal_pass_state (int signo)

{

  return signal_program[signo];

}



static void

‌signal_cache_update (int signo)

{

  if (signo == -1)

    {

      for (signo = 0; signo < (int) GDB_SIGNAL_LAST; signo++)

        signal_cache_update (signo);



      return;

    }



  signal_pass[signo] = (signal_stop[signo] == 0

                        && signal_print[signo] == 0

                        && signal_program[signo] == 1

                        && signal_catch[signo] == 0);

}



int

‌signal_stop_update (int signo, int state)

{

  int ret = signal_stop[signo];



  signal_stop[signo] = state;

  signal_cache_update (signo);

  return ret;

}



int

‌signal_print_update (int signo, int state)

{

  int ret = signal_print[signo];



  signal_print[signo] = state;

  signal_cache_update (signo);

  return ret;

}



int

‌signal_pass_update (int signo, int state)

{

  int ret = signal_program[signo];



  signal_program[signo] = state;

  signal_cache_update (signo);

  return ret;

}



/* Update the global 'signal_catch' from INFO and notify the

   target.  */



void

‌signal_catch_update (const unsigned int *info)

{

  int i;



  for (i = 0; i < GDB_SIGNAL_LAST; ++i)

    signal_catch[i] = info[i] > 0;

  signal_cache_update (-1);

  target_pass_signals ((int) GDB_SIGNAL_LAST, signal_pass);

}



static void

‌sig_print_header (void)

{

  printf_filtered (_("Signal        Stop\tPrint\tPass "

                     "to program\tDescription\n"));

}



static void

‌sig_print_info (enum gdb_signal oursig)

{

  const char *name = gdb_signal_to_name (oursig);

  int name_padding = 13 - strlen (name);



  if (name_padding <= 0)

    name_padding = 0;



  printf_filtered ("%s", name);

  printf_filtered ("%*.*s ", name_padding, name_padding, "                 ");

  printf_filtered ("%s\t", signal_stop[oursig] ? "Yes" : "No");

  printf_filtered ("%s\t", signal_print[oursig] ? "Yes" : "No");

  printf_filtered ("%s\t\t", signal_program[oursig] ? "Yes" : "No");

  printf_filtered ("%s\n", gdb_signal_to_string (oursig));

}



/* Specify how various signals in the inferior should be handled.  */



static void

‌handle_command (char *args, int from_tty)

{

  char **argv;

  int digits, wordlen;

  int sigfirst, signum, siglast;

  enum gdb_signal oursig;

  int allsigs;

  int nsigs;

  unsigned char *sigs;

  struct cleanup *old_chain;



  if (args == NULL)

    {

      error_no_arg (_("signal to handle"));

    }



  /* Allocate and zero an array of flags for which signals to handle.  */



  nsigs = (int) GDB_SIGNAL_LAST;

  sigs = (unsigned char *) alloca (nsigs);

  memset (sigs, 0, nsigs);



  /* Break the command line up into args.  */



  argv = gdb_buildargv (args);

  old_chain = make_cleanup_freeargv (argv);



  /* Walk through the args, looking for signal oursigs, signal names, and

     actions.  Signal numbers and signal names may be interspersed with

     actions, with the actions being performed for all signals cumulatively

     specified.  Signal ranges can be specified as <LOW>-<HIGH>.  */



  while (*argv != NULL)

    {

      wordlen = strlen (*argv);

      for (digits = 0; isdigit ((*argv)[digits]); digits++)

        {;

        }

      allsigs = 0;

      sigfirst = siglast = -1;



      if (wordlen >= 1 && !strncmp (*argv, "all", wordlen))

        {

          /* Apply action to all signals except those used by the

             debugger.  Silently skip those.  */

          allsigs = 1;

          sigfirst = 0;

          siglast = nsigs - 1;

        }

      else if (wordlen >= 1 && !strncmp (*argv, "stop", wordlen))

        {

          SET_SIGS (nsigs, sigs, signal_stop);

          SET_SIGS (nsigs, sigs, signal_print);

        }

      else if (wordlen >= 1 && !strncmp (*argv, "ignore", wordlen))

        {

          UNSET_SIGS (nsigs, sigs, signal_program);

        }

      else if (wordlen >= 2 && !strncmp (*argv, "print", wordlen))

        {

          SET_SIGS (nsigs, sigs, signal_print);

        }

      else if (wordlen >= 2 && !strncmp (*argv, "pass", wordlen))

        {

          SET_SIGS (nsigs, sigs, signal_program);

        }

      else if (wordlen >= 3 && !strncmp (*argv, "nostop", wordlen))

        {

          UNSET_SIGS (nsigs, sigs, signal_stop);

        }

      else if (wordlen >= 3 && !strncmp (*argv, "noignore", wordlen))

        {

          SET_SIGS (nsigs, sigs, signal_program);

        }

      else if (wordlen >= 4 && !strncmp (*argv, "noprint", wordlen))

        {

          UNSET_SIGS (nsigs, sigs, signal_print);

          UNSET_SIGS (nsigs, sigs, signal_stop);

        }

      else if (wordlen >= 4 && !strncmp (*argv, "nopass", wordlen))

        {

          UNSET_SIGS (nsigs, sigs, signal_program);

        }

      else if (digits > 0)

        {

          /* It is numeric.  The numeric signal refers to our own

             internal signal numbering from target.h, not to host/target

             signal  number.  This is a feature; users really should be

             using symbolic names anyway, and the common ones like

             SIGHUP, SIGINT, SIGALRM, etc. will work right anyway.  */



          sigfirst = siglast = (int)

            gdb_signal_from_command (atoi (*argv));

          if ((*argv)[digits] == '-')

            {

              siglast = (int)

                gdb_signal_from_command (atoi ((*argv) + digits + 1));

            }

          if (sigfirst > siglast)

            {

              /* Bet he didn't figure we'd think of this case...  */

              signum = sigfirst;

              sigfirst = siglast;

              siglast = signum;

            }

        }

      else

        {

          oursig = gdb_signal_from_name (*argv);

          if (oursig != GDB_SIGNAL_UNKNOWN)

            {

              sigfirst = siglast = (int) oursig;

            }

          else

            {

              /* Not a number and not a recognized flag word => complain.  */

              error (_("Unrecognized or ambiguous flag word: \"%s\"."), *argv);

            }

        }



      /* If any signal numbers or symbol names were found, set flags for

         which signals to apply actions to.  */



      for (signum = sigfirst; signum >= 0 && signum <= siglast; signum++)

        {

          switch ((enum gdb_signal) signum)

            {

            case GDB_SIGNAL_TRAP:

            case GDB_SIGNAL_INT:

              if (!allsigs && !sigs[signum])

                {

                  if (query (_("%s is used by the debugger.\n\

Are you sure you want to change it? "),

                             gdb_signal_to_name ((enum gdb_signal) signum)))

                    {

                      sigs[signum] = 1;

                    }

                  else

                    {

                      printf_unfiltered (_("Not confirmed, unchanged.\n"));

                      gdb_flush (gdb_stdout);

                    }

                }

              break;

            case GDB_SIGNAL_0:

            case GDB_SIGNAL_DEFAULT:

            case GDB_SIGNAL_UNKNOWN:

              /* Make sure that "all" doesn't print these.  */

              break;

            default:

              sigs[signum] = 1;

              break;

            }

        }



      argv++;

    }



  for (signum = 0; signum < nsigs; signum++)

    if (sigs[signum])

      {

        signal_cache_update (-1);

        target_pass_signals ((int) GDB_SIGNAL_LAST, signal_pass);

        target_program_signals ((int) GDB_SIGNAL_LAST, signal_program);



        if (from_tty)

          {

            /* Show the results.  */

            sig_print_header ();

            for (; signum < nsigs; signum++)

              if (sigs[signum])

                sig_print_info (signum);

          }



        break;

      }



  do_cleanups (old_chain);

}



/* Complete the "handle" command.  */



‌static VEC (char_ptr) *

handle_completer (struct cmd_list_element *ignore,

                  const char *text, const char *word)

{

  VEC (char_ptr) *vec_signals, *vec_keywords, *return_val;

  static const char * const keywords[] =

    {

      "all",

      "stop",

      "ignore",

      "print",

      "pass",

      "nostop",

      "noignore",

      "noprint",

      "nopass",

      NULL,

    };



  vec_signals = signal_completer (ignore, text, word);

  vec_keywords = complete_on_enum (keywords, word, word);



  return_val = VEC_merge (char_ptr, vec_signals, vec_keywords);

  VEC_free (char_ptr, vec_signals);

  VEC_free (char_ptr, vec_keywords);

  return return_val;

}



static void

‌xdb_handle_command (char *args, int from_tty)

{

  char **argv;

  struct cleanup *old_chain;



  if (args == NULL)

    error_no_arg (_("xdb command"));



  /* Break the command line up into args.  */



  argv = gdb_buildargv (args);

  old_chain = make_cleanup_freeargv (argv);

  if (argv[1] != (char *) NULL)

    {

      char *argBuf;

      int bufLen;



      bufLen = strlen (argv[0]) + 20;

      argBuf = (char *) xmalloc (bufLen);

      if (argBuf)

        {

          int validFlag = 1;

          enum gdb_signal oursig;



          oursig = gdb_signal_from_name (argv[0]);

          memset (argBuf, 0, bufLen);

          if (strcmp (argv[1], "Q") == 0)

            sprintf (argBuf, "%s %s", argv[0], "noprint");

          else

            {

              if (strcmp (argv[1], "s") == 0)

                {

                  if (!signal_stop[oursig])

                    sprintf (argBuf, "%s %s", argv[0], "stop");

                  else

                    sprintf (argBuf, "%s %s", argv[0], "nostop");

                }

              else if (strcmp (argv[1], "i") == 0)

                {

                  if (!signal_program[oursig])

                    sprintf (argBuf, "%s %s", argv[0], "pass");

                  else

                    sprintf (argBuf, "%s %s", argv[0], "nopass");

                }

              else if (strcmp (argv[1], "r") == 0)

                {

                  if (!signal_print[oursig])

                    sprintf (argBuf, "%s %s", argv[0], "print");

                  else

                    sprintf (argBuf, "%s %s", argv[0], "noprint");

                }

              else

                validFlag = 0;

            }

          if (validFlag)

            handle_command (argBuf, from_tty);

          else

            printf_filtered (_("Invalid signal handling flag.\n"));

          if (argBuf)

            xfree (argBuf);

        }

    }

  do_cleanups (old_chain);

}



enum gdb_signal

‌gdb_signal_from_command (int num)

{

  if (num >= 1 && num <= 15)

    return (enum gdb_signal) num;

  error (_("Only signals 1-15 are valid as numeric signals.\n\

Use \"info signals\" for a list of symbolic signals."));

}



/* Print current contents of the tables set by the handle command.

   It is possible we should just be printing signals actually used

   by the current target (but for things to work right when switching

   targets, all signals should be in the signal tables).  */



static void

‌signals_info (char *signum_exp, int from_tty)

{

  enum gdb_signal oursig;



  sig_print_header ();



  if (signum_exp)

    {

      /* First see if this is a symbol name.  */

      oursig = gdb_signal_from_name (signum_exp);

      if (oursig == GDB_SIGNAL_UNKNOWN)

        {

          /* No, try numeric.  */

          oursig =

            gdb_signal_from_command (parse_and_eval_long (signum_exp));

        }

      sig_print_info (oursig);

      return;

    }



  printf_filtered ("\n");

  /* These ugly casts brought to you by the native VAX compiler.  */

  for (oursig = GDB_SIGNAL_FIRST;

       (int) oursig < (int) GDB_SIGNAL_LAST;

       oursig = (enum gdb_signal) ((int) oursig + 1))

    {

      QUIT;



      if (oursig != GDB_SIGNAL_UNKNOWN

          && oursig != GDB_SIGNAL_DEFAULT && oursig != GDB_SIGNAL_0)

        sig_print_info (oursig);

    }



  printf_filtered (_("\nUse the \"handle\" command "

                     "to change these tables.\n"));

}



/* Check if it makes sense to read $_siginfo from the current thread

   at this point.  If not, throw an error.  */



static void

‌validate_siginfo_access (void)

{

  /* No current inferior, no siginfo.  */

  if (ptid_equal (inferior_ptid, null_ptid))

    error (_("No thread selected."));



  /* Don't try to read from a dead thread.  */

  if (is_exited (inferior_ptid))

    error (_("The current thread has terminated"));



  /* ... or from a spinning thread.  */

  if (is_running (inferior_ptid))

    error (_("Selected thread is running."));

}



/* The $_siginfo convenience variable is a bit special.  We don't know

   for sure the type of the value until we actually have a chance to

   fetch the data.  The type can change depending on gdbarch, so it is

   also dependent on which thread you have selected.



     1. making $_siginfo be an internalvar that creates a new value on

     access.



     2. making the value of $_siginfo be an lval_computed value.  */



/* This function implements the lval_computed support for reading a

   $_siginfo value.  */



static void

‌siginfo_value_read (struct value *v)

{

  LONGEST transferred;



  validate_siginfo_access ();



  transferred =

    target_read (&current_target, TARGET_OBJECT_SIGNAL_INFO,

                 NULL,

                 value_contents_all_raw (v),

                 value_offset (v),

                 TYPE_LENGTH (value_type (v)));



  if (transferred != TYPE_LENGTH (value_type (v)))

    error (_("Unable to read siginfo"));

}



/* This function implements the lval_computed support for writing a

   $_siginfo value.  */



static void

‌siginfo_value_write (struct value *v, struct value *fromval)

{

  LONGEST transferred;



  validate_siginfo_access ();



  transferred = target_write (&current_target,

                              TARGET_OBJECT_SIGNAL_INFO,

                              NULL,

                              value_contents_all_raw (fromval),

                              value_offset (v),

                              TYPE_LENGTH (value_type (fromval)));



  if (transferred != TYPE_LENGTH (value_type (fromval)))

    error (_("Unable to write siginfo"));

}



‌static const struct lval_funcs siginfo_value_funcs =

  {

    siginfo_value_read,

    siginfo_value_write

  };



/* Return a new value with the correct type for the siginfo object of

   the current thread using architecture GDBARCH.  Return a void value

   if there's no object available.  */



static struct value *

‌siginfo_make_value (struct gdbarch *gdbarch, struct internalvar *var,

                    void *ignore)

{

  if (target_has_stack

      && !ptid_equal (inferior_ptid, null_ptid)

      && gdbarch_get_siginfo_type_p (gdbarch))

    {

      struct type *type = gdbarch_get_siginfo_type (gdbarch);



      return allocate_computed_value (type, &siginfo_value_funcs, NULL);

    }



  return allocate_value (builtin_type (gdbarch)->builtin_void);

}





/* infcall_suspend_state contains state about the program itself like its

   registers and any signal it received when it last stopped.

   This state must be restored regardless of how the inferior function call

   ends (either successfully, or after it hits a breakpoint or signal)

   if the program is to properly continue where it left off.  */



‌struct infcall_suspend_state

{

  struct thread_suspend_state thread_suspend;

#if 0 /* Currently unused and empty structures are not valid C.  */

  struct inferior_suspend_state inferior_suspend;

#endif



  /* Other fields:  */

  CORE_ADDR stop_pc;

  struct regcache *registers;



  /* Format of SIGINFO_DATA or NULL if it is not present.  */

  struct gdbarch *siginfo_gdbarch;



  /* The inferior format depends on SIGINFO_GDBARCH and it has a length of

     TYPE_LENGTH (gdbarch_get_siginfo_type ()).  For different gdbarch the

     content would be invalid.  */

  gdb_byte *siginfo_data;

};



struct infcall_suspend_state *

‌save_infcall_suspend_state (void)

{

  struct infcall_suspend_state *inf_state;

  struct thread_info *tp = inferior_thread ();

#if 0

  struct inferior *inf = current_inferior ();

#endif

  struct regcache *regcache = get_current_regcache ();

  struct gdbarch *gdbarch = get_regcache_arch (regcache);

  gdb_byte *siginfo_data = NULL;



  if (gdbarch_get_siginfo_type_p (gdbarch))

    {

      struct type *type = gdbarch_get_siginfo_type (gdbarch);

      size_t len = TYPE_LENGTH (type);

      struct cleanup *back_to;



      siginfo_data = xmalloc (len);

      back_to = make_cleanup (xfree, siginfo_data);



      if (target_read (&current_target, TARGET_OBJECT_SIGNAL_INFO, NULL,

                       siginfo_data, 0, len) == len)

        discard_cleanups (back_to);

      else

        {

          /* Errors ignored.  */

          do_cleanups (back_to);

          siginfo_data = NULL;

        }

    }



  inf_state = XCNEW (struct infcall_suspend_state);



  if (siginfo_data)

    {

      inf_state->siginfo_gdbarch = gdbarch;

      inf_state->siginfo_data = siginfo_data;

    }



  inf_state->thread_suspend = tp->suspend;

#if 0 /* Currently unused and empty structures are not valid C.  */

  inf_state->inferior_suspend = inf->suspend;

#endif



  /* run_inferior_call will not use the signal due to its `proceed' call with

     GDB_SIGNAL_0 anyway.  */

  tp->suspend.stop_signal = GDB_SIGNAL_0;



  inf_state->stop_pc = stop_pc;



  inf_state->registers = regcache_dup (regcache);



  return inf_state;

}



/* Restore inferior session state to INF_STATE.  */



void

‌restore_infcall_suspend_state (struct infcall_suspend_state *inf_state)

{

  struct thread_info *tp = inferior_thread ();

#if 0

  struct inferior *inf = current_inferior ();

#endif

  struct regcache *regcache = get_current_regcache ();

  struct gdbarch *gdbarch = get_regcache_arch (regcache);



  tp->suspend = inf_state->thread_suspend;

#if 0 /* Currently unused and empty structures are not valid C.  */

  inf->suspend = inf_state->inferior_suspend;

#endif



  stop_pc = inf_state->stop_pc;



  if (inf_state->siginfo_gdbarch == gdbarch)

    {

      struct type *type = gdbarch_get_siginfo_type (gdbarch);



      /* Errors ignored.  */

      target_write (&current_target, TARGET_OBJECT_SIGNAL_INFO, NULL,

                    inf_state->siginfo_data, 0, TYPE_LENGTH (type));

    }



  /* The inferior can be gone if the user types "print exit(0)"

     (and perhaps other times).  */

  if (target_has_execution)

    /* NB: The register write goes through to the target.  */

    regcache_cpy (regcache, inf_state->registers);



  discard_infcall_suspend_state (inf_state);

}



static void

‌do_restore_infcall_suspend_state_cleanup (void *state)

{

  restore_infcall_suspend_state (state);

}



struct cleanup *

‌make_cleanup_restore_infcall_suspend_state

  (struct infcall_suspend_state *inf_state)

{

  return make_cleanup (do_restore_infcall_suspend_state_cleanup, inf_state);

}



void

‌discard_infcall_suspend_state (struct infcall_suspend_state *inf_state)

{

  regcache_xfree (inf_state->registers);

  xfree (inf_state->siginfo_data);

  xfree (inf_state);

}



struct regcache *

‌get_infcall_suspend_state_regcache (struct infcall_suspend_state *inf_state)

{

  return inf_state->registers;

}



/* infcall_control_state contains state regarding gdb's control of the

   inferior itself like stepping control.  It also contains session state like

   the user's currently selected frame.  */



‌struct infcall_control_state

{

  struct thread_control_state thread_control;

  struct inferior_control_state inferior_control;



  /* Other fields:  */

  enum stop_stack_kind stop_stack_dummy;

  int stopped_by_random_signal;

  int stop_after_trap;



  /* ID if the selected frame when the inferior function call was made.  */

  struct frame_id selected_frame_id;

};



/* Save all of the information associated with the inferior<==>gdb

   connection.  */



struct infcall_control_state *

‌save_infcall_control_state (void)

{

  struct infcall_control_state *inf_status = xmalloc (sizeof (*inf_status));

  struct thread_info *tp = inferior_thread ();

  struct inferior *inf = current_inferior ();



  inf_status->thread_control = tp->control;

  inf_status->inferior_control = inf->control;



  tp->control.step_resume_breakpoint = NULL;

  tp->control.exception_resume_breakpoint = NULL;



  /* Save original bpstat chain to INF_STATUS; replace it in TP with copy of

     chain.  If caller's caller is walking the chain, they'll be happier if we

     hand them back the original chain when restore_infcall_control_state is

     called.  */

  tp->control.stop_bpstat = bpstat_copy (tp->control.stop_bpstat);



  /* Other fields:  */

  inf_status->stop_stack_dummy = stop_stack_dummy;

  inf_status->stopped_by_random_signal = stopped_by_random_signal;

  inf_status->stop_after_trap = stop_after_trap;



  inf_status->selected_frame_id = get_frame_id (get_selected_frame (NULL));



  return inf_status;

}



static int

‌restore_selected_frame (void *args)

{

  struct frame_id *fid = (struct frame_id *) args;

  struct frame_info *frame;



  frame = frame_find_by_id (*fid);



  /* If inf_status->selected_frame_id is NULL, there was no previously

     selected frame.  */

  if (frame == NULL)

    {

      warning (_("Unable to restore previously selected frame."));

      return 0;

    }



  select_frame (frame);



  return (1);

}



/* Restore inferior session state to INF_STATUS.  */



void

‌restore_infcall_control_state (struct infcall_control_state *inf_status)

{

  struct thread_info *tp = inferior_thread ();

  struct inferior *inf = current_inferior ();



  if (tp->control.step_resume_breakpoint)

    tp->control.step_resume_breakpoint->disposition = disp_del_at_next_stop;



  if (tp->control.exception_resume_breakpoint)

    tp->control.exception_resume_breakpoint->disposition

      = disp_del_at_next_stop;



  /* Handle the bpstat_copy of the chain.  */

  bpstat_clear (&tp->control.stop_bpstat);



  tp->control = inf_status->thread_control;

  inf->control = inf_status->inferior_control;



  /* Other fields:  */

  stop_stack_dummy = inf_status->stop_stack_dummy;

  stopped_by_random_signal = inf_status->stopped_by_random_signal;

  stop_after_trap = inf_status->stop_after_trap;



  if (target_has_stack)

    {

      /* The point of catch_errors is that if the stack is clobbered,

         walking the stack might encounter a garbage pointer and

         error() trying to dereference it.  */

      if (catch_errors

          (restore_selected_frame, &inf_status->selected_frame_id,

           "Unable to restore previously selected frame:\n",

           RETURN_MASK_ERROR) == 0)

        /* Error in restoring the selected frame.  Select the innermost

           frame.  */

        select_frame (get_current_frame ());

    }



  xfree (inf_status);

}



static void

‌do_restore_infcall_control_state_cleanup (void *sts)

{

  restore_infcall_control_state (sts);

}



struct cleanup *

‌make_cleanup_restore_infcall_control_state

  (struct infcall_control_state *inf_status)

{

  return make_cleanup (do_restore_infcall_control_state_cleanup, inf_status);

}



void

‌discard_infcall_control_state (struct infcall_control_state *inf_status)

{

  if (inf_status->thread_control.step_resume_breakpoint)

    inf_status->thread_control.step_resume_breakpoint->disposition

      = disp_del_at_next_stop;



  if (inf_status->thread_control.exception_resume_breakpoint)

    inf_status->thread_control.exception_resume_breakpoint->disposition

      = disp_del_at_next_stop;



  /* See save_infcall_control_state for info on stop_bpstat.  */

  bpstat_clear (&inf_status->thread_control.stop_bpstat);



  xfree (inf_status);

}



/* restore_inferior_ptid() will be used by the cleanup machinery

   to restore the inferior_ptid value saved in a call to

   save_inferior_ptid().  */



static void

‌restore_inferior_ptid (void *arg)

{

  ptid_t *saved_ptid_ptr = arg;



  inferior_ptid = *saved_ptid_ptr;

  xfree (arg);

}



/* Save the value of inferior_ptid so that it may be restored by a

   later call to do_cleanups().  Returns the struct cleanup pointer

   needed for later doing the cleanup.  */



struct cleanup *

‌save_inferior_ptid (void)

{

  ptid_t *saved_ptid_ptr;



  saved_ptid_ptr = xmalloc (sizeof (ptid_t));

  *saved_ptid_ptr = inferior_ptid;

  return make_cleanup (restore_inferior_ptid, saved_ptid_ptr);

}



/* See infrun.h.  */



void

‌clear_exit_convenience_vars (void)

{

  clear_internalvar (lookup_internalvar ("_exitsignal"));

  clear_internalvar (lookup_internalvar ("_exitcode"));

}





/* User interface for reverse debugging:

   Set exec-direction / show exec-direction commands

   (returns error unless target implements to_set_exec_direction method).  */



‌int execution_direction = EXEC_FORWARD;

‌static const char exec_forward[] = "forward";

‌static const char exec_reverse[] = "reverse";

‌static const char *exec_direction = exec_forward;

‌static const char *const exec_direction_names[] = {

  exec_forward,

  exec_reverse,

  NULL

};



static void

‌set_exec_direction_func (char *args, int from_tty,

                         struct cmd_list_element *cmd)

{

  if (target_can_execute_reverse)

    {

      if (!strcmp (exec_direction, exec_forward))

        execution_direction = EXEC_FORWARD;

      else if (!strcmp (exec_direction, exec_reverse))

        execution_direction = EXEC_REVERSE;

    }

  else

    {

      exec_direction = exec_forward;

      error (_("Target does not support this operation."));

    }

}



static void

‌show_exec_direction_func (struct ui_file *out, int from_tty,

                          struct cmd_list_element *cmd, const char *value)

{

  switch (execution_direction) {

  case EXEC_FORWARD:

    fprintf_filtered (out, _("Forward.\n"));

    break;

  case EXEC_REVERSE:

    fprintf_filtered (out, _("Reverse.\n"));

    break;

  default:

    internal_error (__FILE__, __LINE__,

                    _("bogus execution_direction value: %d"),

                    (int) execution_direction);

  }

}



static void

‌show_schedule_multiple (struct ui_file *file, int from_tty,

                        struct cmd_list_element *c, const char *value)

{

  fprintf_filtered (file, _("Resuming the execution of threads "

                            "of all processes is %s.\n"), value);

}



/* Implementation of `siginfo' variable.  */



‌static const struct internalvar_funcs siginfo_funcs =

{

  siginfo_make_value,

  NULL,

  NULL

};



void

‌_initialize_infrun (void)

{

  int i;

  int numsigs;

  struct cmd_list_element *c;



  add_info ("signals", signals_info, _("\

What debugger does when program gets various signals.\n\

Specify a signal as argument to print info on that signal only."));

  add_info_alias ("handle", "signals", 0);



  c = add_com ("handle", class_run, handle_command, _("\

Specify how to handle signals.\n\

Usage: handle SIGNAL [ACTIONS]\n\

Args are signals and actions to apply to those signals.\n\

If no actions are specified, the current settings for the specified signals\n\

will be displayed instead.\n\

\n\

Symbolic signals (e.g. SIGSEGV) are recommended but numeric signals\n\

from 1-15 are allowed for compatibility with old versions of GDB.\n\

Numeric ranges may be specified with the form LOW-HIGH (e.g. 1-5).\n\

The special arg \"all\" is recognized to mean all signals except those\n\

used by the debugger, typically SIGTRAP and SIGINT.\n\

\n\

Recognized actions include \"stop\", \"nostop\", \"print\", \"noprint\",\n\

\"pass\", \"nopass\", \"ignore\", or \"noignore\".\n\

Stop means reenter debugger if this signal happens (implies print).\n\

Print means print a message if this signal happens.\n\

Pass means let program see this signal; otherwise program doesn't know.\n\

Ignore is a synonym for nopass and noignore is a synonym for pass.\n\

Pass and Stop may be combined.\n\

\n\

Multiple signals may be specified.  Signal numbers and signal names\n\

may be interspersed with actions, with the actions being performed for\n\

all signals cumulatively specified."));

  set_cmd_completer (c, handle_completer);



  if (xdb_commands)

    {

      add_com ("lz", class_info, signals_info, _("\

What debugger does when program gets various signals.\n\

Specify a signal as argument to print info on that signal only."));

      add_com ("z", class_run, xdb_handle_command, _("\

Specify how to handle a signal.\n\

Args are signals and actions to apply to those signals.\n\

Symbolic signals (e.g. SIGSEGV) are recommended but numeric signals\n\

from 1-15 are allowed for compatibility with old versions of GDB.\n\

Numeric ranges may be specified with the form LOW-HIGH (e.g. 1-5).\n\

The special arg \"all\" is recognized to mean all signals except those\n\

used by the debugger, typically SIGTRAP and SIGINT.\n\

Recognized actions include \"s\" (toggles between stop and nostop),\n\

\"r\" (toggles between print and noprint), \"i\" (toggles between pass and \

nopass), \"Q\" (noprint)\n\

Stop means reenter debugger if this signal happens (implies print).\n\

Print means print a message if this signal happens.\n\

Pass means let program see this signal; otherwise program doesn't know.\n\

Ignore is a synonym for nopass and noignore is a synonym for pass.\n\

Pass and Stop may be combined."));

    }



  if (!dbx_commands)

    stop_command = add_cmd ("stop", class_obscure,

                            not_just_help_class_command, _("\

There is no `stop' command, but you can set a hook on `stop'.\n\

This allows you to set a list of commands to be run each time execution\n\

of the program stops."), &cmdlist);



  add_setshow_zuinteger_cmd ("infrun", class_maintenance, &debug_infrun, _("\

Set inferior debugging."), _("\

Show inferior debugging."), _("\

When non-zero, inferior specific debugging is enabled."),

                             NULL,

                             show_debug_infrun,

                             &setdebuglist, &showdebuglist);



  add_setshow_boolean_cmd ("displaced", class_maintenance,

                           &debug_displaced, _("\

Set displaced stepping debugging."), _("\

Show displaced stepping debugging."), _("\

When non-zero, displaced stepping specific debugging is enabled."),

                            NULL,

                            show_debug_displaced,

                            &setdebuglist, &showdebuglist);



  add_setshow_boolean_cmd ("non-stop", no_class,

                           &non_stop_1, _("\

Set whether gdb controls the inferior in non-stop mode."), _("\

Show whether gdb controls the inferior in non-stop mode."), _("\

When debugging a multi-threaded program and this setting is\n\

off (the default, also called all-stop mode), when one thread stops\n\

(for a breakpoint, watchpoint, exception, or similar events), GDB stops\n\

all other threads in the program while you interact with the thread of\n\

interest.  When you continue or step a thread, you can allow the other\n\

threads to run, or have them remain stopped, but while you inspect any\n\

thread's state, all threads stop.\n\

\n\

In non-stop mode, when one thread stops, other threads can continue\n\

to run freely.  You'll be able to step each thread independently,\n\

leave it stopped or free to run as needed."),

                           set_non_stop,

                           show_non_stop,

                           &setlist,

                           &showlist);



  numsigs = (int) GDB_SIGNAL_LAST;

  signal_stop = (unsigned char *) xmalloc (sizeof (signal_stop[0]) * numsigs);

  signal_print = (unsigned char *)

    xmalloc (sizeof (signal_print[0]) * numsigs);

  signal_program = (unsigned char *)

    xmalloc (sizeof (signal_program[0]) * numsigs);

  signal_catch = (unsigned char *)

    xmalloc (sizeof (signal_catch[0]) * numsigs);

  signal_pass = (unsigned char *)

    xmalloc (sizeof (signal_pass[0]) * numsigs);

  for (i = 0; i < numsigs; i++)

    {

      signal_stop[i] = 1;

      signal_print[i] = 1;

      signal_program[i] = 1;

      signal_catch[i] = 0;

    }



  /* Signals caused by debugger's own actions

     should not be given to the program afterwards.  */

  signal_program[GDB_SIGNAL_TRAP] = 0;

  signal_program[GDB_SIGNAL_INT] = 0;



  /* Signals that are not errors should not normally enter the debugger.  */

  signal_stop[GDB_SIGNAL_ALRM] = 0;

  signal_print[GDB_SIGNAL_ALRM] = 0;

  signal_stop[GDB_SIGNAL_VTALRM] = 0;

  signal_print[GDB_SIGNAL_VTALRM] = 0;

  signal_stop[GDB_SIGNAL_PROF] = 0;

  signal_print[GDB_SIGNAL_PROF] = 0;

  signal_stop[GDB_SIGNAL_CHLD] = 0;

  signal_print[GDB_SIGNAL_CHLD] = 0;

  signal_stop[GDB_SIGNAL_IO] = 0;

  signal_print[GDB_SIGNAL_IO] = 0;

  signal_stop[GDB_SIGNAL_POLL] = 0;

  signal_print[GDB_SIGNAL_POLL] = 0;

  signal_stop[GDB_SIGNAL_URG] = 0;

  signal_print[GDB_SIGNAL_URG] = 0;

  signal_stop[GDB_SIGNAL_WINCH] = 0;

  signal_print[GDB_SIGNAL_WINCH] = 0;

  signal_stop[GDB_SIGNAL_PRIO] = 0;

  signal_print[GDB_SIGNAL_PRIO] = 0;



  /* These signals are used internally by user-level thread

     implementations.  (See signal(5) on Solaris.)  Like the above

     signals, a healthy program receives and handles them as part of

     its normal operation.  */

  signal_stop[GDB_SIGNAL_LWP] = 0;

  signal_print[GDB_SIGNAL_LWP] = 0;

  signal_stop[GDB_SIGNAL_WAITING] = 0;

  signal_print[GDB_SIGNAL_WAITING] = 0;

  signal_stop[GDB_SIGNAL_CANCEL] = 0;

  signal_print[GDB_SIGNAL_CANCEL] = 0;



  /* Update cached state.  */

  signal_cache_update (-1);



  add_setshow_zinteger_cmd ("stop-on-solib-events", class_support,

                            &stop_on_solib_events, _("\

Set stopping for shared library events."), _("\

Show stopping for shared library events."), _("\

If nonzero, gdb will give control to the user when the dynamic linker\n\

notifies gdb of shared library events.  The most common event of interest\n\

to the user would be loading/unloading of a new library."),

                            set_stop_on_solib_events,

                            show_stop_on_solib_events,

                            &setlist, &showlist);



  add_setshow_enum_cmd ("follow-fork-mode", class_run,

                        follow_fork_mode_kind_names,

                        &follow_fork_mode_string, _("\

Set debugger response to a program call of fork or vfork."), _("\

Show debugger response to a program call of fork or vfork."), _("\

A fork or vfork creates a new process.  follow-fork-mode can be:\n\

  parent  - the original process is debugged after a fork\n\

  child   - the new process is debugged after a fork\n\

The unfollowed process will continue to run.\n\

By default, the debugger will follow the parent process."),

                        NULL,

                        show_follow_fork_mode_string,

                        &setlist, &showlist);



  add_setshow_enum_cmd ("follow-exec-mode", class_run,

                        follow_exec_mode_names,

                        &follow_exec_mode_string, _("\

Set debugger response to a program call of exec."), _("\

Show debugger response to a program call of exec."), _("\

An exec call replaces the program image of a process.\n\

\n\

follow-exec-mode can be:\n\

\n\

  new - the debugger creates a new inferior and rebinds the process\n\

to this new inferior.  The program the process was running before\n\

the exec call can be restarted afterwards by restarting the original\n\

inferior.\n\

\n\

  same - the debugger keeps the process bound to the same inferior.\n\

The new executable image replaces the previous executable loaded in\n\

the inferior.  Restarting the inferior after the exec call restarts\n\

the executable the process was running after the exec call.\n\

\n\

By default, the debugger will use the same inferior."),

                        NULL,

                        show_follow_exec_mode_string,

                        &setlist, &showlist);



  add_setshow_enum_cmd ("scheduler-locking", class_run,

                        scheduler_enums, &scheduler_mode, _("\

Set mode for locking scheduler during execution."), _("\

Show mode for locking scheduler during execution."), _("\

off  == no locking (threads may preempt at any time)\n\

on   == full locking (no thread except the current thread may run)\n\

step == scheduler locked during every single-step operation.\n\

        In this mode, no other thread may run during a step command.\n\

        Other threads may run while stepping over a function call ('next')."),

                        set_schedlock_func,        /* traps on target vector */

                        show_scheduler_mode,

                        &setlist, &showlist);



  add_setshow_boolean_cmd ("schedule-multiple", class_run, &sched_multi, _("\

Set mode for resuming threads of all processes."), _("\

Show mode for resuming threads of all processes."), _("\

When on, execution commands (such as 'continue' or 'next') resume all\n\

threads of all processes.  When off (which is the default), execution\n\

commands only resume the threads of the current process.  The set of\n\

threads that are resumed is further refined by the scheduler-locking\n\

mode (see help set scheduler-locking)."),

                           NULL,

                           show_schedule_multiple,

                           &setlist, &showlist);



  add_setshow_boolean_cmd ("step-mode", class_run, &step_stop_if_no_debug, _("\

Set mode of the step operation."), _("\

Show mode of the step operation."), _("\

When set, doing a step over a function without debug line information\n\

will stop at the first instruction of that function. Otherwise, the\n\

function is skipped and the step command stops at a different source line."),

                           NULL,

                           show_step_stop_if_no_debug,

                           &setlist, &showlist);



  add_setshow_auto_boolean_cmd ("displaced-stepping", class_run,

                                &can_use_displaced_stepping, _("\

Set debugger's willingness to use displaced stepping."), _("\

Show debugger's willingness to use displaced stepping."), _("\

If on, gdb will use displaced stepping to step over breakpoints if it is\n\

supported by the target architecture.  If off, gdb will not use displaced\n\

stepping to step over breakpoints, even if such is supported by the target\n\

architecture.  If auto (which is the default), gdb will use displaced stepping\n\

if the target architecture supports it and non-stop mode is active, but will not\n\

use it in all-stop mode (see help set non-stop)."),

                                NULL,

                                show_can_use_displaced_stepping,

                                &setlist, &showlist);



  add_setshow_enum_cmd ("exec-direction", class_run, exec_direction_names,

                        &exec_direction, _("Set direction of execution.\n\

Options are 'forward' or 'reverse'."),

                        _("Show direction of execution (forward/reverse)."),

                        _("Tells gdb whether to execute forward or backward."),

                        set_exec_direction_func, show_exec_direction_func,

                        &setlist, &showlist);



  /* Set/show detach-on-fork: user-settable mode.  */



  add_setshow_boolean_cmd ("detach-on-fork", class_run, &detach_fork, _("\

Set whether gdb will detach the child of a fork."), _("\

Show whether gdb will detach the child of a fork."), _("\

Tells gdb whether to detach the child of a fork."),

                           NULL, NULL, &setlist, &showlist);



  /* Set/show disable address space randomization mode.  */



  add_setshow_boolean_cmd ("disable-randomization", class_support,

                           &disable_randomization, _("\

Set disabling of debuggee's virtual address space randomization."), _("\

Show disabling of debuggee's virtual address space randomization."), _("\

When this mode is on (which is the default), randomization of the virtual\n\

address space is disabled.  Standalone programs run with the randomization\n\

enabled by default on some platforms."),

                           &set_disable_randomization,

                           &show_disable_randomization,

                           &setlist, &showlist);



  /* ptid initializations */

  inferior_ptid = null_ptid;

  target_last_wait_ptid = minus_one_ptid;



  observer_attach_thread_ptid_changed (infrun_thread_ptid_changed);

  observer_attach_thread_stop_requested (infrun_thread_stop_requested);

  observer_attach_thread_exit (infrun_thread_thread_exit);

  observer_attach_inferior_exit (infrun_inferior_exit);



  /* Explicitly create without lookup, since that tries to create a

     value with a void typed value, and when we get here, gdbarch

     isn't initialized yet.  At this point, we're quite sure there

     isn't another convenience variable of the same name.  */

  create_internalvar_type_lazy ("_siginfo", &siginfo_funcs, NULL);



  add_setshow_boolean_cmd ("observer", no_class,

                           &observer_mode_1, _("\

Set whether gdb controls the inferior in observer mode."), _("\

Show whether gdb controls the inferior in observer mode."), _("\

In observer mode, GDB can get data from the inferior, but not\n\

affect its execution.  Registers and memory may not be changed,\n\

breakpoints may not be set, and the program cannot be interrupted\n\

or signalled."),

                           set_observer_mode,

                           show_observer_mode,

                           &setlist,

                           &showlist);

}
gdb/infrun.c - gdb

Global variables defined

Data types defined

Functions defined

Macros defined

Source code
