binutils-gdb/gdb/gnu-nat.c
Simon Marchi 7831bc9185 gdb: remove get_exec_file
I believe that the get_exec_file function is unnecessary, and the code
can be simplified if we remove it.

Consider for instance when you "run" a program on Linux with native
debugging.

 1. run_command_1 obtains the executable file from
    `current_program_space->exec_filename ()`
 2. it passes it to `run_target->create_inferior()`, which is
    `inf_ptrace_target::create_inferior()` in this case, which then
    passes it to `fork_inferior()`
 3. `fork_inferior()` then has a fallback, where if the passed exec file
    is nullptr, it gets its from `get_exec_file()`.
 4. `get_exec_file()` returns `current_program_space->exec_filename ()`
    - just like the things we started with - or errors out if the
    current program space doesn't have a specified executable.

If there's no exec filename passed in step 1, there's not going to be
any in step 4, so it seems pointless to call `get_exec_file()`, we could
just error out when `exec_file` is nullptr.  But we can't error out
directly in `fork_inferior()`, since the error is GDB-specific, and that
function is shared with GDBserver.

Speaking of GDBserver, all code paths that lead to `fork_inferior()`
provide a non-nullptr exec file.

Therefore, to simplify things:

 - Make `fork_inferior()` assume that the passed exec file is not
   nullptr, don't call `get_exec_file()`
 - Change some targets (darwin-nat, go32-nat, gnu-nat, inf-ptrace,
   nto-procfs, procfs) to error out when the exec file passed to their
   create_inferior method is nullptr.  Some targets are fine with a
   nullptr exec file, so we can't check that in `run_command_1()`.
 - Add the `no_executable_specified_error()` function, which re-uses the
   error message that `get_exec_file()` had.
 - Change some targets (go32-nat, nto-procfs) to not call
   `get_exec_file()`, since it's pointless for the same reason as in the
   example above, if it returns, it's going the be the same value as the
   `exec_file` parameter.  Just rely on `exec_file`.
 - Remove the final use of `get_exec_file()`, in `load_command()`.
 - Remove the `get_exec_file()` implementations in GDB and GDBserver and
   remove the shared declaration.

Change-Id: I601c16498e455f7baa1f111a179da2f6c913baa3
Approved-By: Tom Tromey <tom@tromey.com>
2024-06-07 23:09:03 -04:00

3476 lines
97 KiB
C
Raw Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

/* Interface GDB to the GNU Hurd.
Copyright (C) 1992-2024 Free Software Foundation, Inc.
This file is part of GDB.
Written by Miles Bader <miles@gnu.ai.mit.edu>
Some code and ideas from m3-nat.c by Jukka Virtanen <jtv@hut.fi>
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 this first, to pick up the <mach.h> 'thread_info' diversion. */
#include "gnu-nat.h"
/* Mach/Hurd headers are not yet ready for C++ compilation. */
extern "C"
{
#include <mach.h>
#include <mach_error.h>
#include <mach/exception.h>
#include <mach/message.h>
#include <mach/notify.h>
#include <mach/vm_attributes.h>
#include <mach/vm_param.h>
#include <hurd.h>
#include <hurd/interrupt.h>
#include <hurd/msg.h>
#include <hurd/msg_request.h>
#include <hurd/process.h>
/* Defined in <hurd/process.h>, but we need forward declarations from
<hurd/process_request.h> as well. */
#undef _process_user_
#include <hurd/process_request.h>
#include <hurd/signal.h>
#include <hurd/sigpreempt.h>
#include <portinfo.h>
}
#include <ctype.h>
#include <setjmp.h>
#include <signal.h>
#include <sys/ptrace.h>
#include <elf.h>
#include <link.h>
#include "inferior.h"
#include "symtab.h"
#include "value.h"
#include "language.h"
#include "target.h"
#include "gdbsupport/gdb_wait.h"
#include "gdbarch.h"
#include "cli/cli-cmds.h"
#include "gdbcore.h"
#include "gdbthread.h"
#include "gdbsupport/gdb_obstack.h"
#include "tid-parse.h"
#include "nat/fork-inferior.h"
#include "inf-child.h"
/* MIG stubs are not yet ready for C++ compilation. */
extern "C"
{
#include "exc_request_S.h"
#include "notify_S.h"
#include "process_reply_S.h"
#include "msg_reply_S.h"
#include "exc_request_U.h"
#include "msg_U.h"
#include "gnu-nat-mig.h"
}
struct gnu_nat_target *gnu_target;
static process_t proc_server = MACH_PORT_NULL;
/* If we've sent a proc_wait_request to the proc server, the pid of the
process we asked about. We can only ever have one outstanding. */
int proc_wait_pid = 0;
/* The number of wait requests we've sent, and expect replies from. */
int proc_waits_pending = 0;
bool gnu_debug_flag = false;
/* Forward decls */
static struct inf *make_inf ();
#define inf_debug(_inf, msg, args...) \
do { struct inf *__inf = (_inf); \
debug ("{inf %d %s}: " msg, __inf->pid, \
host_address_to_string (__inf) , ##args); } while (0)
/* Evaluate RPC_EXPR in a scope with the variables MSGPORT and REFPORT bound
to INF's msg port and task port respectively. If it has no msg port,
EIEIO is returned. INF must refer to a running process! */
#define INF_MSGPORT_RPC(inf, rpc_expr) \
HURD_MSGPORT_RPC (proc_getmsgport (proc_server, inf->pid, &msgport), \
(refport = inf->task->port, 0), 0, \
msgport ? (rpc_expr) : EIEIO)
/* Like INF_MSGPORT_RPC, but will also resume the signal thread to ensure
there's someone around to deal with the RPC (and resuspend things
afterwards). This effects INF's threads' resume_sc count. */
#define INF_RESUME_MSGPORT_RPC(inf, rpc_expr) \
(inf_set_threads_resume_sc_for_signal_thread (inf) \
? ({ kern_return_t __e; \
inf_resume (inf); \
__e = INF_MSGPORT_RPC (inf, rpc_expr); \
inf_suspend (inf); \
__e; }) \
: EIEIO)
/* The state passed by an exception message. */
struct exc_state
{
int exception; /* The exception code. */
int code, subcode;
mach_port_t handler; /* The real exception port to handle this. */
mach_port_t reply; /* The reply port from the exception call. */
};
/* The results of the last wait an inf did. */
struct inf_wait
{
struct target_waitstatus status; /* The status returned to gdb. */
struct exc_state exc; /* The exception that caused us to return. */
struct proc *thread; /* The thread in question. */
int suppress; /* Something trivial happened. */
};
/* Further Hurd-specific state of an inferior. */
struct inf
{
/* Fields describing the current inferior. */
struct proc *task; /* The mach task. */
struct proc *threads; /* A linked list of all threads in TASK. */
/* True if THREADS needn't be validated by querying the task. We
assume that we and the task in question are the only ones
frobbing the thread list, so as long as we don't let any code
run, we don't have to worry about THREADS changing. */
int threads_up_to_date;
pid_t pid; /* The real system PID. */
struct inf_wait wait; /* What to return from target_wait. */
/* One thread proc in INF may be in `single-stepping mode'. This
is it. */
struct proc *step_thread;
/* The thread we think is the signal thread. */
struct proc *signal_thread;
mach_port_t event_port; /* Where we receive various msgs. */
/* True if we think at least one thread in the inferior could currently be
running. */
unsigned int running:1;
/* True if the process has stopped (in the proc server sense). Note that
since a proc server `stop' leaves the signal thread running, the inf can
be RUNNING && STOPPED... */
unsigned int stopped:1;
/* True if the inferior has no message port. */
unsigned int nomsg:1;
/* True if the inferior is traced. */
unsigned int traced:1;
/* True if we shouldn't try waiting for the inferior, usually because we
can't for some reason. */
unsigned int no_wait:1;
/* When starting a new inferior, we don't try to validate threads until all
the proper execs have been done, which this flag states we still
expect to happen. */
unsigned int pending_execs:1;
/* Fields describing global state. */
/* The task suspend count used when gdb has control. This is normally 1 to
make things easier for us, but sometimes (like when attaching to vital
system servers) it may be desirable to let the task continue to run
(pausing individual threads as necessary). */
int pause_sc;
/* The task suspend count left when detaching from a task. */
int detach_sc;
/* The initial values used for the run_sc and pause_sc of newly discovered
threads -- see the definition of those fields in struct proc. */
int default_thread_run_sc;
int default_thread_pause_sc;
int default_thread_detach_sc;
/* True if the process should be traced when started/attached. Newly
started processes *must* be traced at first to exec them properly, but
if this is false, tracing is turned off as soon it has done so. */
int want_signals;
/* True if exceptions from the inferior process should be trapped. This
must be on to use breakpoints. */
int want_exceptions;
};
int
__proc_pid (struct proc *proc)
{
return proc->inf->pid;
}
/* Update PROC's real suspend count to match it's desired one. Returns true
if we think PROC is now in a runnable state. */
int
gnu_nat_target::proc_update_sc (struct proc *proc)
{
int running;
int err = 0;
int delta = proc->sc - proc->cur_sc;
if (delta)
proc_debug (proc, "sc: %d --> %d", proc->cur_sc, proc->sc);
if (proc->sc == 0 && proc->state_changed)
/* Since PROC may start running, we must write back any state changes. */
{
gdb_assert (proc_is_thread (proc));
proc_debug (proc, "storing back changed thread state");
err = thread_set_state (proc->port, THREAD_STATE_FLAVOR,
(thread_state_t) &proc->state, THREAD_STATE_SIZE);
if (!err)
proc->state_changed = 0;
}
if (delta > 0)
{
while (delta-- > 0 && !err)
{
if (proc_is_task (proc))
err = task_suspend (proc->port);
else
err = thread_suspend (proc->port);
}
}
else
{
while (delta++ < 0 && !err)
{
if (proc_is_task (proc))
err = task_resume (proc->port);
else
err = thread_resume (proc->port);
}
}
if (!err)
proc->cur_sc = proc->sc;
/* If we got an error, then the task/thread has disappeared. */
running = !err && proc->sc == 0;
proc_debug (proc, "is %s", err ? "dead" : running ? "running" : "suspended");
if (err)
proc_debug (proc, "err = %s", safe_strerror (err));
if (running)
{
proc->aborted = 0;
proc->state_valid = proc->state_changed = 0;
proc->fetched_regs = 0;
}
return running;
}
/* Thread_abort is called on PROC if needed. PROC must be a thread proc.
If PROC is deemed `precious', then nothing is done unless FORCE is true.
In particular, a thread is precious if it's running (in which case forcing
it includes suspending it first), or if it has an exception pending. */
void
gnu_nat_target::proc_abort (struct proc *proc, int force)
{
gdb_assert (proc_is_thread (proc));
if (!proc->aborted)
{
struct inf *inf = proc->inf;
int running = (proc->cur_sc == 0 && inf->task->cur_sc == 0);
if (running && force)
{
proc->sc = 1;
inf_update_suspends (proc->inf);
running = 0;
warning (_("Stopped %s."), proc_string (proc));
}
else if (proc == inf->wait.thread && inf->wait.exc.reply && !force)
/* An exception is pending on PROC, which don't mess with. */
running = 1;
if (!running)
/* We only abort the thread if it's not actually running. */
{
thread_abort (proc->port);
proc_debug (proc, "aborted");
proc->aborted = 1;
}
else
proc_debug (proc, "not aborting");
}
}
/* Make sure that the state field in PROC is up to date, and return a pointer
to it, or 0 if something is wrong. If WILL_MODIFY is true, makes sure
that the thread is stopped and aborted first, and sets the state_changed
field in PROC to true. */
thread_state_t
gnu_nat_target::proc_get_state (struct proc *proc, int will_modify)
{
int was_aborted = proc->aborted;
proc_debug (proc, "updating state info%s",
will_modify ? " (with intention to modify)" : "");
proc_abort (proc, will_modify);
if (!was_aborted && proc->aborted)
/* PROC's state may have changed since we last fetched it. */
proc->state_valid = 0;
if (!proc->state_valid)
{
mach_msg_type_number_t state_size = THREAD_STATE_SIZE;
kern_return_t err =
thread_get_state (proc->port, THREAD_STATE_FLAVOR,
(thread_state_t) &proc->state, &state_size);
proc_debug (proc, "getting thread state");
proc->state_valid = !err;
}
if (proc->state_valid)
{
if (will_modify)
proc->state_changed = 1;
return (thread_state_t) &proc->state;
}
else
return 0;
}
/* Set PORT to PROC's exception port. */
kern_return_t
gnu_nat_target::proc_get_exception_port (struct proc * proc, mach_port_t * port)
{
if (proc_is_task (proc))
return task_get_exception_port (proc->port, port);
else
return thread_get_exception_port (proc->port, port);
}
/* Set PROC's exception port to PORT. */
kern_return_t
gnu_nat_target::proc_set_exception_port (struct proc * proc, mach_port_t port)
{
proc_debug (proc, "setting exception port: %lu", port);
if (proc_is_task (proc))
return task_set_exception_port (proc->port, port);
else
return thread_set_exception_port (proc->port, port);
}
/* Get PROC's exception port, cleaning up a bit if proc has died. */
mach_port_t
gnu_nat_target::_proc_get_exc_port (struct proc *proc)
{
mach_port_t exc_port;
kern_return_t err = proc_get_exception_port (proc, &exc_port);
if (err)
/* PROC must be dead. */
{
if (proc->exc_port)
mach_port_deallocate (mach_task_self (), proc->exc_port);
proc->exc_port = MACH_PORT_NULL;
if (proc->saved_exc_port)
mach_port_deallocate (mach_task_self (), proc->saved_exc_port);
proc->saved_exc_port = MACH_PORT_NULL;
}
return exc_port;
}
/* Replace PROC's exception port with EXC_PORT, unless it's already
been done. Stash away any existing exception port so we can
restore it later. */
void
gnu_nat_target::proc_steal_exc_port (struct proc *proc, mach_port_t exc_port)
{
mach_port_t cur_exc_port = _proc_get_exc_port (proc);
if (cur_exc_port)
{
kern_return_t err = 0;
proc_debug (proc, "inserting exception port: %lu", exc_port);
if (cur_exc_port != exc_port)
/* Put in our exception port. */
err = proc_set_exception_port (proc, exc_port);
if (err || cur_exc_port == proc->exc_port)
/* We previously set the exception port, and it's still set. So we
just keep the old saved port which is what the proc set. */
{
if (cur_exc_port)
mach_port_deallocate (mach_task_self (), cur_exc_port);
}
else
/* Keep a copy of PROC's old exception port so it can be restored. */
{
if (proc->saved_exc_port)
mach_port_deallocate (mach_task_self (), proc->saved_exc_port);
proc->saved_exc_port = cur_exc_port;
}
proc_debug (proc, "saved exception port: %lu", proc->saved_exc_port);
if (!err)
proc->exc_port = exc_port;
else
warning (_("Error setting exception port for %s: %s"),
proc_string (proc), safe_strerror (err));
}
}
/* If we previously replaced PROC's exception port, put back what we
found there at the time, unless *our* exception port has since been
overwritten, in which case who knows what's going on. */
void
gnu_nat_target::proc_restore_exc_port (struct proc *proc)
{
mach_port_t cur_exc_port = _proc_get_exc_port (proc);
if (cur_exc_port)
{
kern_return_t err = 0;
proc_debug (proc, "restoring real exception port");
if (proc->exc_port == cur_exc_port)
/* Our's is still there. */
err = proc_set_exception_port (proc, proc->saved_exc_port);
if (proc->saved_exc_port)
mach_port_deallocate (mach_task_self (), proc->saved_exc_port);
proc->saved_exc_port = MACH_PORT_NULL;
if (!err)
proc->exc_port = MACH_PORT_NULL;
else
warning (_("Error setting exception port for %s: %s"),
proc_string (proc), safe_strerror (err));
}
}
/* Turns hardware tracing in PROC on or off when SET is true or false,
respectively. Returns true on success. */
int
gnu_nat_target::proc_trace (struct proc *proc, int set)
{
thread_state_t state = proc_get_state (proc, 1);
if (!state)
return 0; /* The thread must be dead. */
proc_debug (proc, "tracing %s", set ? "on" : "off");
if (set)
{
/* XXX We don't get the exception unless the thread has its own
exception port???? */
if (proc->exc_port == MACH_PORT_NULL)
proc_steal_exc_port (proc, proc->inf->event_port);
THREAD_STATE_SET_TRACED (state);
}
else
THREAD_STATE_CLEAR_TRACED (state);
return 1;
}
/* A variable from which to assign new TIDs. */
static int next_thread_id = 1;
/* Returns a new proc structure with the given fields. Also adds a
notification for PORT becoming dead to be sent to INF's notify port. */
struct proc *
gnu_nat_target::make_proc (struct inf *inf, mach_port_t port, int tid)
{
kern_return_t err;
mach_port_t prev_port = MACH_PORT_NULL;
struct proc *proc = XNEW (struct proc);
proc->port = port;
proc->tid = tid;
proc->inf = inf;
proc->next = 0;
proc->saved_exc_port = MACH_PORT_NULL;
proc->exc_port = MACH_PORT_NULL;
proc->sc = 0;
proc->cur_sc = 0;
/* Note that these are all the values for threads; the task simply uses the
corresponding field in INF directly. */
proc->run_sc = inf->default_thread_run_sc;
proc->pause_sc = inf->default_thread_pause_sc;
proc->detach_sc = inf->default_thread_detach_sc;
proc->resume_sc = proc->run_sc;
proc->aborted = 0;
proc->dead = 0;
proc->state_valid = 0;
proc->state_changed = 0;
proc_debug (proc, "is new");
/* Get notified when things die. */
err =
mach_port_request_notification (mach_task_self (), port,
MACH_NOTIFY_DEAD_NAME, 1,
inf->event_port,
MACH_MSG_TYPE_MAKE_SEND_ONCE,
&prev_port);
if (err)
warning (_("Couldn't request notification for port %lu: %s"),
port, safe_strerror (err));
else
{
proc_debug (proc, "notifications to: %lu", inf->event_port);
if (prev_port != MACH_PORT_NULL)
mach_port_deallocate (mach_task_self (), prev_port);
}
if (inf->want_exceptions)
{
if (proc_is_task (proc))
/* Make the task exception port point to us. */
proc_steal_exc_port (proc, inf->event_port);
else
/* Just clear thread exception ports -- they default to the
task one. */
proc_steal_exc_port (proc, MACH_PORT_NULL);
}
return proc;
}
/* Frees PROC and any resources it uses, and returns the value of PROC's
next field. */
struct proc *
gnu_nat_target::_proc_free (struct proc *proc)
{
struct inf *inf = proc->inf;
struct proc *next = proc->next;
proc_debug (proc, "freeing...");
if (proc == inf->step_thread)
/* Turn off single stepping. */
inf_set_step_thread (inf, 0);
if (proc == inf->wait.thread)
inf_clear_wait (inf);
if (proc == inf->signal_thread)
inf->signal_thread = 0;
if (proc->port != MACH_PORT_NULL)
{
if (proc->exc_port != MACH_PORT_NULL)
/* Restore the original exception port. */
proc_restore_exc_port (proc);
if (proc->cur_sc != 0)
/* Resume the thread/task. */
{
proc->sc = 0;
proc_update_sc (proc);
}
mach_port_deallocate (mach_task_self (), proc->port);
}
xfree (proc);
return next;
}
static struct inf *
make_inf (void)
{
struct inf *inf = new struct inf;
inf->task = 0;
inf->threads = 0;
inf->threads_up_to_date = 0;
inf->pid = 0;
inf->wait.status.set_spurious ();
inf->wait.thread = 0;
inf->wait.exc.handler = MACH_PORT_NULL;
inf->wait.exc.reply = MACH_PORT_NULL;
inf->step_thread = 0;
inf->signal_thread = 0;
inf->event_port = MACH_PORT_NULL;
inf->running = 0;
inf->stopped = 0;
inf->nomsg = 1;
inf->traced = 0;
inf->no_wait = 0;
inf->pending_execs = 0;
inf->pause_sc = 1;
inf->detach_sc = 0;
inf->default_thread_run_sc = 0;
inf->default_thread_pause_sc = 0;
inf->default_thread_detach_sc = 0;
inf->want_signals = 1; /* By default */
inf->want_exceptions = 1; /* By default */
return inf;
}
/* Clear INF's target wait status. */
void
gnu_nat_target::inf_clear_wait (struct inf *inf)
{
inf_debug (inf, "clearing wait");
inf->wait.status.set_spurious ();
inf->wait.thread = 0;
inf->wait.suppress = 0;
if (inf->wait.exc.handler != MACH_PORT_NULL)
{
mach_port_deallocate (mach_task_self (), inf->wait.exc.handler);
inf->wait.exc.handler = MACH_PORT_NULL;
}
if (inf->wait.exc.reply != MACH_PORT_NULL)
{
mach_port_deallocate (mach_task_self (), inf->wait.exc.reply);
inf->wait.exc.reply = MACH_PORT_NULL;
}
}
void
gnu_nat_target::inf_cleanup (struct inf *inf)
{
inf_debug (inf, "cleanup");
inf_clear_wait (inf);
inf_set_pid (inf, -1);
inf->pid = 0;
inf->running = 0;
inf->stopped = 0;
inf->nomsg = 1;
inf->traced = 0;
inf->no_wait = 0;
inf->pending_execs = 0;
if (inf->event_port)
{
mach_port_destroy (mach_task_self (), inf->event_port);
inf->event_port = MACH_PORT_NULL;
}
}
void
gnu_nat_target::inf_startup (struct inf *inf, int pid)
{
kern_return_t err;
inf_debug (inf, "startup: pid = %d", pid);
inf_cleanup (inf);
/* Make the port on which we receive all events. */
err = mach_port_allocate (mach_task_self (),
MACH_PORT_RIGHT_RECEIVE, &inf->event_port);
if (err)
error (_("Error allocating event port: %s"), safe_strerror (err));
/* Make a send right for it, so we can easily copy it for other people. */
mach_port_insert_right (mach_task_self (), inf->event_port,
inf->event_port, MACH_MSG_TYPE_MAKE_SEND);
inf_set_pid (inf, pid);
}
/* Close current process, if any, and attach INF to process PORT. */
void
gnu_nat_target::inf_set_pid (struct inf *inf, pid_t pid)
{
task_t task_port;
struct proc *task = inf->task;
inf_debug (inf, "setting pid: %d", pid);
if (pid < 0)
task_port = MACH_PORT_NULL;
else
{
kern_return_t err = proc_pid2task (proc_server, pid, &task_port);
if (err)
error (_("Error getting task for pid %d: %s"),
pid, safe_strerror (err));
}
inf_debug (inf, "setting task: %lu", task_port);
if (inf->pause_sc)
task_suspend (task_port);
if (task && task->port != task_port)
{
inf->task = 0;
inf_validate_procs (inf); /* Trash all the threads. */
_proc_free (task); /* And the task. */
}
if (task_port != MACH_PORT_NULL)
{
inf->task = make_proc (inf, task_port, PROC_TID_TASK);
inf->threads_up_to_date = 0;
}
if (inf->task)
{
inf->pid = pid;
if (inf->pause_sc)
/* Reflect task_suspend above. */
inf->task->sc = inf->task->cur_sc = 1;
}
else
inf->pid = -1;
}
/* Validates INF's stopped, nomsg and traced field from the actual
proc server state. Note that the traced field is only updated from
the proc server state if we do not have a message port. If we do
have a message port we'd better look at the tracemask itself. */
void
gnu_nat_target::inf_validate_procinfo (struct inf *inf)
{
char *noise;
mach_msg_type_number_t noise_len = 0;
struct procinfo *pi;
mach_msg_type_number_t pi_len = 0;
int info_flags = 0;
kern_return_t err =
proc_getprocinfo (proc_server, inf->pid, &info_flags,
(procinfo_t *) &pi, &pi_len, &noise, &noise_len);
if (!err)
{
inf->stopped = !!(pi->state & PI_STOPPED);
inf->nomsg = !!(pi->state & PI_NOMSG);
if (inf->nomsg)
inf->traced = !!(pi->state & PI_TRACED);
vm_deallocate (mach_task_self (), (vm_address_t) pi,
pi_len * sizeof (*(procinfo_t) 0));
if (noise_len > 0)
vm_deallocate (mach_task_self (), (vm_address_t) noise, noise_len);
}
}
/* Validates INF's task suspend count. If it's higher than we expect,
verify with the user before `stealing' the extra count. */
void
gnu_nat_target::inf_validate_task_sc (struct inf *inf)
{
char *noise;
mach_msg_type_number_t noise_len = 0;
struct procinfo *pi;
mach_msg_type_number_t pi_len = 0;
int info_flags = PI_FETCH_TASKINFO;
int suspend_count = -1;
kern_return_t err;
retry:
err = proc_getprocinfo (proc_server, inf->pid, &info_flags,
(procinfo_t *) &pi, &pi_len, &noise, &noise_len);
if (err)
{
inf->task->dead = 1; /* oh well */
return;
}
if (inf->task->cur_sc < pi->taskinfo.suspend_count && suspend_count == -1)
{
/* The proc server might have suspended the task while stopping
it. This happens when the task is handling a traced signal.
Refetch the suspend count. The proc server should be
finished stopping the task by now. */
suspend_count = pi->taskinfo.suspend_count;
goto retry;
}
suspend_count = pi->taskinfo.suspend_count;
vm_deallocate (mach_task_self (), (vm_address_t) pi,
pi_len * sizeof (*(procinfo_t) 0));
if (noise_len > 0)
vm_deallocate (mach_task_self (), (vm_address_t) noise, noise_len);
if (inf->task->cur_sc < suspend_count)
{
if (!query (_("Pid %d has an additional task suspend count of %d;"
" clear it? "), inf->pid,
suspend_count - inf->task->cur_sc))
error (_("Additional task suspend count left untouched."));
inf->task->cur_sc = suspend_count;
}
}
/* Turns tracing for INF on or off, depending on ON, unless it already
is. If INF is running, the resume_sc count of INF's threads will
be modified, and the signal thread will briefly be run to change
the trace state. */
void
gnu_nat_target::inf_set_traced (struct inf *inf, int on)
{
if (on == inf->traced)
return;
if (inf->task && !inf->task->dead)
/* Make it take effect immediately. */
{
sigset_t mask = on ? ~(sigset_t) 0 : 0;
kern_return_t err =
INF_RESUME_MSGPORT_RPC (inf, msg_set_init_int (msgport, refport,
INIT_TRACEMASK, mask));
if (err == EIEIO)
{
if (on)
warning (_("Can't modify tracing state for pid %d: %s"),
inf->pid, "No signal thread");
inf->traced = on;
}
else if (err)
warning (_("Can't modify tracing state for pid %d: %s"),
inf->pid, safe_strerror (err));
else
inf->traced = on;
}
else
inf->traced = on;
}
/* Makes all the real suspend count deltas of all the procs in INF
match the desired values. Careful to always do thread/task suspend
counts in the safe order. Returns true if at least one thread is
thought to be running. */
int
gnu_nat_target::inf_update_suspends (struct inf *inf)
{
struct proc *task = inf->task;
/* We don't have to update INF->threads even though we're iterating over it
because we'll change a thread only if it already has an existing proc
entry. */
inf_debug (inf, "updating suspend counts");
if (task)
{
struct proc *thread;
int task_running = (task->sc == 0), thread_running = 0;
if (task->sc > task->cur_sc)
/* The task is becoming _more_ suspended; do before any threads. */
task_running = proc_update_sc (task);
if (inf->pending_execs)
/* When we're waiting for an exec, things may be happening behind our
back, so be conservative. */
thread_running = 1;
/* Do all the thread suspend counts. */
for (thread = inf->threads; thread; thread = thread->next)
thread_running |= proc_update_sc (thread);
if (task->sc != task->cur_sc)
/* We didn't do the task first, because we wanted to wait for the
threads; do it now. */
task_running = proc_update_sc (task);
inf_debug (inf, "%srunning...",
(thread_running && task_running) ? "" : "not ");
inf->running = thread_running && task_running;
/* Once any thread has executed some code, we can't depend on the
threads list any more. */
if (inf->running)
inf->threads_up_to_date = 0;
return inf->running;
}
return 0;
}
/* Converts a GDB pid to a struct proc. */
struct proc *
inf_tid_to_thread (struct inf *inf, int tid)
{
struct proc *thread = inf->threads;
while (thread)
if (thread->tid == tid)
return thread;
else
thread = thread->next;
return 0;
}
/* Converts a thread port to a struct proc. */
static struct proc *
inf_port_to_thread (struct inf *inf, mach_port_t port)
{
struct proc *thread = inf->threads;
while (thread)
if (thread->port == port)
return thread;
else
thread = thread->next;
return 0;
}
/* See gnu-nat.h. */
void
inf_threads (struct inf *inf, inf_threads_ftype *f, void *arg)
{
struct proc *thread;
for (thread = inf->threads; thread; thread = thread->next)
f (thread, arg);
}
/* Make INF's list of threads be consistent with reality of TASK. */
void
gnu_nat_target::inf_validate_procs (struct inf *inf)
{
thread_array_t threads;
mach_msg_type_number_t num_threads, i;
struct proc *task = inf->task;
/* If no threads are currently running, this function will guarantee that
things are up to date. The exception is if there are zero threads --
then it is almost certainly in an odd state, and probably some outside
agent will create threads. */
inf->threads_up_to_date = inf->threads ? !inf->running : 0;
if (task)
{
kern_return_t err = task_threads (task->port, &threads, &num_threads);
inf_debug (inf, "fetching threads");
if (err)
/* TASK must be dead. */
{
task->dead = 1;
task = 0;
}
}
if (!task)
{
num_threads = 0;
inf_debug (inf, "no task");
}
{
/* Make things normally linear. */
mach_msg_type_number_t search_start = 0;
/* Which thread in PROCS corresponds to each task thread, & the task. */
struct proc *matched[num_threads + 1];
/* The last thread in INF->threads, so we can add to the end. */
struct proc *last = 0;
/* The current thread we're considering. */
struct proc *thread = inf->threads;
memset (matched, 0, sizeof (matched));
while (thread)
{
mach_msg_type_number_t left;
for (i = search_start, left = num_threads; left; i++, left--)
{
if (i >= num_threads)
i -= num_threads; /* I wrapped around. */
if (thread->port == threads[i])
/* We already know about this thread. */
{
matched[i] = thread;
last = thread;
thread = thread->next;
search_start++;
break;
}
}
if (!left)
{
proc_debug (thread, "died!");
thread->port = MACH_PORT_NULL;
thread = _proc_free (thread); /* THREAD is dead. */
if (last)
last->next = thread;
else
inf->threads = thread;
}
}
for (i = 0; i < num_threads; i++)
{
if (matched[i])
/* Throw away the duplicate send right. */
mach_port_deallocate (mach_task_self (), threads[i]);
else
/* THREADS[I] is a thread we don't know about yet! */
{
ptid_t ptid;
thread = make_proc (inf, threads[i], next_thread_id++);
if (last)
last->next = thread;
else
inf->threads = thread;
last = thread;
proc_debug (thread, "new thread: %lu", threads[i]);
ptid = ptid_t (inf->pid, thread->tid, 0);
/* Tell GDB's generic thread code. */
if (inferior_ptid == ptid_t (inf->pid))
/* This is the first time we're hearing about thread
ids, after a fork-child. */
thread_change_ptid (this, inferior_ptid, ptid);
else if (inf->pending_execs != 0)
/* This is a shell thread. */
add_thread_silent (this, ptid);
else
add_thread (this, ptid);
}
}
vm_deallocate (mach_task_self (),
(vm_address_t) threads, (num_threads * sizeof (thread_t)));
}
}
/* Makes sure that INF's thread list is synced with the actual process. */
int
inf_update_procs (struct inf *inf)
{
if (!inf->task)
return 0;
if (!inf->threads_up_to_date)
gnu_target->inf_validate_procs (inf);
return !!inf->task;
}
/* Sets the resume_sc of each thread in inf. That of RUN_THREAD is set to 0,
and others are set to their run_sc if RUN_OTHERS is true, and otherwise
their pause_sc. */
void
gnu_nat_target::inf_set_threads_resume_sc (struct inf *inf,
struct proc *run_thread, int run_others)
{
struct proc *thread;
inf_update_procs (inf);
for (thread = inf->threads; thread; thread = thread->next)
if (thread == run_thread)
thread->resume_sc = 0;
else if (run_others)
thread->resume_sc = thread->run_sc;
else
thread->resume_sc = thread->pause_sc;
}
/* Cause INF to continue execution immediately; individual threads may still
be suspended (but their suspend counts will be updated). */
void
gnu_nat_target::inf_resume (struct inf *inf)
{
struct proc *thread;
inf_update_procs (inf);
for (thread = inf->threads; thread; thread = thread->next)
thread->sc = thread->resume_sc;
if (inf->task)
{
if (!inf->pending_execs)
/* Try to make sure our task count is correct -- in the case where
we're waiting for an exec though, things are too volatile, so just
assume things will be reasonable (which they usually will be). */
inf_validate_task_sc (inf);
inf->task->sc = 0;
}
inf_update_suspends (inf);
}
/* Cause INF to stop execution immediately; individual threads may still
be running. */
void
gnu_nat_target::inf_suspend (struct inf *inf)
{
struct proc *thread;
inf_update_procs (inf);
for (thread = inf->threads; thread; thread = thread->next)
thread->sc = thread->pause_sc;
if (inf->task)
inf->task->sc = inf->pause_sc;
inf_update_suspends (inf);
}
/* INF has one thread PROC that is in single-stepping mode. This
function changes it to be PROC, changing any old step_thread to be
a normal one. A PROC of 0 clears any existing value. */
void
gnu_nat_target::inf_set_step_thread (struct inf *inf, struct proc *thread)
{
gdb_assert (!thread || proc_is_thread (thread));
if (thread)
inf_debug (inf, "setting step thread: %d/%d", inf->pid, thread->tid);
else
inf_debug (inf, "clearing step thread");
if (inf->step_thread != thread)
{
if (inf->step_thread && inf->step_thread->port != MACH_PORT_NULL)
if (!proc_trace (inf->step_thread, 0))
return;
if (thread && proc_trace (thread, 1))
inf->step_thread = thread;
else
inf->step_thread = 0;
}
}
/* Set up the thread resume_sc's so that only the signal thread is running
(plus whatever other thread are set to always run). Returns true if we
did so, or false if we can't find a signal thread. */
int
gnu_nat_target::inf_set_threads_resume_sc_for_signal_thread (struct inf *inf)
{
if (inf->signal_thread)
{
inf_set_threads_resume_sc (inf, inf->signal_thread, 0);
return 1;
}
else
return 0;
}
static void
inf_update_signal_thread (struct inf *inf)
{
/* XXX for now we assume that if there's a msgport, the 2nd thread is
the signal thread. */
inf->signal_thread = inf->threads ? inf->threads->next : 0;
}
/* Detachs from INF's inferior task, letting it run once again... */
void
gnu_nat_target::inf_detach (struct inf *inf)
{
struct proc *task = inf->task;
inf_debug (inf, "detaching...");
inf_clear_wait (inf);
inf_set_step_thread (inf, 0);
if (task)
{
struct proc *thread;
inf_validate_procinfo (inf);
inf_set_traced (inf, 0);
if (inf->stopped)
{
if (inf->nomsg)
inf_continue (inf);
else
inf_signal (inf, GDB_SIGNAL_0);
}
proc_restore_exc_port (task);
task->sc = inf->detach_sc;
for (thread = inf->threads; thread; thread = thread->next)
{
proc_restore_exc_port (thread);
thread->sc = thread->detach_sc;
}
inf_update_suspends (inf);
}
inf_cleanup (inf);
}
/* Attaches INF to the process with process id PID, returning it in a
suspended state suitable for debugging. */
void
gnu_nat_target::inf_attach (struct inf *inf, int pid)
{
inf_debug (inf, "attaching: %d", pid);
if (inf->pid)
inf_detach (inf);
inf_startup (inf, pid);
}
/* Makes sure that we've got our exception ports entrenched in the process. */
void
gnu_nat_target::inf_steal_exc_ports (struct inf *inf)
{
struct proc *thread;
inf_debug (inf, "stealing exception ports");
inf_set_step_thread (inf, 0); /* The step thread is special. */
proc_steal_exc_port (inf->task, inf->event_port);
for (thread = inf->threads; thread; thread = thread->next)
proc_steal_exc_port (thread, MACH_PORT_NULL);
}
/* Makes sure the process has its own exception ports. */
void
gnu_nat_target::inf_restore_exc_ports (struct inf *inf)
{
struct proc *thread;
inf_debug (inf, "restoring exception ports");
inf_set_step_thread (inf, 0); /* The step thread is special. */
proc_restore_exc_port (inf->task);
for (thread = inf->threads; thread; thread = thread->next)
proc_restore_exc_port (thread);
}
/* Deliver signal SIG to INF. If INF is stopped, delivering a signal, even
signal 0, will continue it. INF is assumed to be in a paused state, and
the resume_sc's of INF's threads may be affected. */
void
gnu_nat_target::inf_signal (struct inf *inf, enum gdb_signal sig)
{
kern_return_t err = 0;
int host_sig = gdb_signal_to_host (sig);
#define NAME gdb_signal_to_name (sig)
if (host_sig >= _NSIG)
/* A mach exception. Exceptions are encoded in the signal space by
putting them after _NSIG; this assumes they're positive (and not
extremely large)! */
{
struct inf_wait *w = &inf->wait;
if (w->status.kind () == TARGET_WAITKIND_STOPPED
&& w->status.sig () == sig
&& w->thread && !w->thread->aborted)
/* We're passing through the last exception we received. This is
kind of bogus, because exceptions are per-thread whereas gdb
treats signals as per-process. We just forward the exception to
the correct handler, even it's not for the same thread as TID --
i.e., we pretend it's global. */
{
struct exc_state *e = &w->exc;
inf_debug (inf, "passing through exception:"
" task = %lu, thread = %lu, exc = %d"
", code = %d, subcode = %d",
w->thread->port, inf->task->port,
e->exception, e->code, e->subcode);
err =
exception_raise_request (e->handler,
e->reply, MACH_MSG_TYPE_MOVE_SEND_ONCE,
w->thread->port, inf->task->port,
e->exception, e->code, e->subcode);
}
else
error (_("Can't forward spontaneous exception (%s)."), NAME);
}
else
/* A Unix signal. */
if (inf->stopped)
/* The process is stopped and expecting a signal. Just send off a
request and let it get handled when we resume everything. */
{
inf_debug (inf, "sending %s to stopped process", NAME);
err =
INF_MSGPORT_RPC (inf,
msg_sig_post_untraced_request (msgport,
inf->event_port,
MACH_MSG_TYPE_MAKE_SEND_ONCE,
host_sig, 0,
refport));
if (!err)
/* Posting an untraced signal automatically continues it.
We clear this here rather than when we get the reply
because we'd rather assume it's not stopped when it
actually is, than the reverse. */
inf->stopped = 0;
}
else
/* It's not expecting it. We have to let just the signal thread
run, and wait for it to get into a reasonable state before we
can continue the rest of the process. When we finally resume the
process the signal we request will be the very first thing that
happens. */
{
inf_debug (inf, "sending %s to unstopped process"
" (so resuming signal thread)", NAME);
err =
INF_RESUME_MSGPORT_RPC (inf,
msg_sig_post_untraced (msgport, host_sig,
0, refport));
}
if (err == EIEIO)
/* Can't do too much... */
warning (_("Can't deliver signal %s: No signal thread."), NAME);
else if (err)
warning (_("Delivering signal %s: %s"), NAME, safe_strerror (err));
#undef NAME
}
/* Continue INF without delivering a signal. This is meant to be used
when INF does not have a message port. */
void
gnu_nat_target::inf_continue (struct inf *inf)
{
process_t proc;
kern_return_t err = proc_pid2proc (proc_server, inf->pid, &proc);
if (!err)
{
inf_debug (inf, "continuing process");
err = proc_mark_cont (proc);
if (!err)
{
struct proc *thread;
for (thread = inf->threads; thread; thread = thread->next)
thread_resume (thread->port);
inf->stopped = 0;
}
}
if (err)
warning (_("Can't continue process: %s"), safe_strerror (err));
}
/* The inferior used for all gdb target ops. */
struct inf *gnu_current_inf = 0;
/* The inferior being waited for by gnu_wait. Since GDB is decidedly not
multi-threaded, we don't bother to lock this. */
static struct inf *waiting_inf;
/* Wait for something to happen in the inferior, returning what in STATUS. */
ptid_t
gnu_nat_target::wait (ptid_t ptid, struct target_waitstatus *status,
target_wait_flags options)
{
struct msg
{
mach_msg_header_t hdr;
mach_msg_type_t type;
int data[8000];
} msg;
kern_return_t err;
struct proc *thread;
struct inf *inf = gnu_current_inf;
gdb_assert (inf->task);
if (!inf->threads && !inf->pending_execs)
/* No threads! Assume that maybe some outside agency is frobbing our
task, and really look for new threads. If we can't find any, just tell
the user to try again later. */
{
inf_validate_procs (inf);
if (!inf->threads && !inf->task->dead)
error (_("There are no threads; try again later."));
}
waiting_inf = inf;
inf_debug (inf, "waiting for: %s", ptid.to_string ().c_str ());
rewait:
if (proc_wait_pid != inf->pid && !inf->no_wait)
/* Always get information on events from the proc server. */
{
inf_debug (inf, "requesting wait on pid %d", inf->pid);
if (proc_wait_pid)
/* The proc server is single-threaded, and only allows a single
outstanding wait request, so we have to cancel the previous one. */
{
inf_debug (inf, "cancelling previous wait on pid %d", proc_wait_pid);
interrupt_operation (proc_server, 0);
}
err =
proc_wait_request (proc_server, inf->event_port, inf->pid, WUNTRACED);
if (err)
warning (_("wait request failed: %s"), safe_strerror (err));
else
{
inf_debug (inf, "waits pending: %d", proc_waits_pending);
proc_wait_pid = inf->pid;
/* Even if proc_waits_pending was > 0 before, we still won't
get any other replies, because it was either from a
different INF, or a different process attached to INF --
and the event port, which is the wait reply port, changes
when you switch processes. */
proc_waits_pending = 1;
}
}
inf_clear_wait (inf);
/* What can happen? (1) Dead name notification; (2) Exceptions arrive;
(3) wait reply from the proc server. */
inf_debug (inf, "waiting for an event...");
err = mach_msg (&msg.hdr, MACH_RCV_MSG | MACH_RCV_INTERRUPT,
0, sizeof (struct msg), inf->event_port,
MACH_MSG_TIMEOUT_NONE, MACH_PORT_NULL);
/* Re-suspend the task. */
inf_suspend (inf);
if (!inf->task && inf->pending_execs)
/* When doing an exec, it's possible that the old task wasn't reused
(e.g., setuid execs). So if the task seems to have disappeared,
attempt to refetch it, as the pid should still be the same. */
inf_set_pid (inf, inf->pid);
if (err == EMACH_RCV_INTERRUPTED)
inf_debug (inf, "interrupted");
else if (err)
error (_("Couldn't wait for an event: %s"), safe_strerror (err));
else
{
struct
{
mach_msg_header_t hdr;
mach_msg_type_t err_type;
kern_return_t err;
char noise[200];
}
reply;
inf_debug (inf, "event: msgid = %d", msg.hdr.msgh_id);
/* Handle what we got. */
if (!notify_server (&msg.hdr, &reply.hdr)
&& !exc_server (&msg.hdr, &reply.hdr)
&& !process_reply_server (&msg.hdr, &reply.hdr)
&& !msg_reply_server (&msg.hdr, &reply.hdr))
/* Whatever it is, it's something strange. */
error (_("Got a strange event, msg id = %d."), msg.hdr.msgh_id);
if (reply.err)
error (_("Handling event, msgid = %d: %s"),
msg.hdr.msgh_id, safe_strerror (reply.err));
}
if (inf->pending_execs)
/* We're waiting for the inferior to finish execing. */
{
struct inf_wait *w = &inf->wait;
enum target_waitkind kind = w->status.kind ();
if (kind == TARGET_WAITKIND_SPURIOUS)
/* Since gdb is actually counting the number of times the inferior
stops, expecting one stop per exec, we only return major events
while execing. */
{
w->suppress = 1;
inf_debug (inf, "pending_execs, ignoring minor event");
}
else if (kind == TARGET_WAITKIND_STOPPED
&& w->status.sig () == GDB_SIGNAL_TRAP)
/* Ah hah! A SIGTRAP from the inferior while starting up probably
means we've successfully completed an exec! */
{
inf_debug (inf, "one pending exec completed");
}
else if (kind == TARGET_WAITKIND_STOPPED)
/* It's possible that this signal is because of a crashed process
being handled by the hurd crash server; in this case, the process
will have an extra task suspend, which we need to know about.
Since the code in inf_resume that normally checks for this is
disabled while INF->pending_execs, we do the check here instead. */
inf_validate_task_sc (inf);
}
if (inf->wait.suppress)
/* Some totally spurious event happened that we don't consider
worth returning to gdb. Just keep waiting. */
{
inf_debug (inf, "suppressing return, rewaiting...");
inf_resume (inf);
goto rewait;
}
/* Pass back out our results. */
*status = inf->wait.status;
thread = inf->wait.thread;
if (thread)
ptid = ptid_t (inf->pid, thread->tid, 0);
else if (ptid == minus_one_ptid)
thread = inf_tid_to_thread (inf, -1);
else
thread = inf_tid_to_thread (inf, ptid.lwp ());
if (!thread || thread->port == MACH_PORT_NULL)
{
/* TID is dead; try and find a new thread. */
if (inf_update_procs (inf) && inf->threads)
ptid = ptid_t (inf->pid, inf->threads->tid, 0); /* The first
available
thread. */
else
{
/* The process exited. */
ptid = ptid_t (inf->pid);
}
}
if (thread
&& ptid != minus_one_ptid
&& status->kind () != TARGET_WAITKIND_SPURIOUS
&& inf->pause_sc == 0 && thread->pause_sc == 0)
/* If something actually happened to THREAD, make sure we
suspend it. */
{
thread->sc = 1;
inf_update_suspends (inf);
}
inf_debug (inf, "returning ptid = %s, %s",
ptid.to_string ().c_str (),
status->to_string ().c_str ());
return ptid;
}
/* The rpc handler called by exc_server. */
kern_return_t
S_exception_raise_request (mach_port_t port, mach_port_t reply_port,
thread_t thread_port, task_t task_port,
int exception, int code, int subcode)
{
struct inf *inf = waiting_inf;
struct proc *thread = inf_port_to_thread (inf, thread_port);
inf_debug (waiting_inf,
"thread = %lu, task = %lu, exc = %d, code = %d, subcode = %d",
thread_port, task_port, exception, code, subcode);
if (!thread)
/* We don't know about thread? */
{
inf_update_procs (inf);
thread = inf_port_to_thread (inf, thread_port);
if (!thread)
/* Give up, the generating thread is gone. */
return 0;
}
mach_port_deallocate (mach_task_self (), thread_port);
mach_port_deallocate (mach_task_self (), task_port);
if (!thread->aborted)
/* THREAD hasn't been aborted since this exception happened (abortion
clears any exception state), so it must be real. */
{
/* Store away the details; this will destroy any previous info. */
inf->wait.thread = thread;
if (exception == EXC_BREAKPOINT)
/* GDB likes to get SIGTRAP for breakpoints. */
{
inf->wait.status.set_stopped (GDB_SIGNAL_TRAP);
mach_port_deallocate (mach_task_self (), reply_port);
}
else
/* Record the exception so that we can forward it later. */
{
if (thread->exc_port == port)
{
inf_debug (waiting_inf, "Handler is thread exception port <%lu>",
thread->saved_exc_port);
inf->wait.exc.handler = thread->saved_exc_port;
}
else
{
inf_debug (waiting_inf, "Handler is task exception port <%lu>",
inf->task->saved_exc_port);
inf->wait.exc.handler = inf->task->saved_exc_port;
gdb_assert (inf->task->exc_port == port);
}
if (inf->wait.exc.handler != MACH_PORT_NULL)
/* Add a reference to the exception handler. */
mach_port_mod_refs (mach_task_self (),
inf->wait.exc.handler, MACH_PORT_RIGHT_SEND,
1);
inf->wait.exc.exception = exception;
inf->wait.exc.code = code;
inf->wait.exc.subcode = subcode;
inf->wait.exc.reply = reply_port;
/* Exceptions are encoded in the signal space by putting
them after _NSIG; this assumes they're positive (and not
extremely large)! */
inf->wait.status.set_stopped
(gdb_signal_from_host (_NSIG + exception));
}
}
else
/* A suppressed exception, which ignore. */
{
inf->wait.suppress = 1;
mach_port_deallocate (mach_task_self (), reply_port);
}
return 0;
}
/* Fill in INF's wait field after a task has died without giving us more
detailed information. */
static void
inf_task_died_status (struct inf *inf)
{
warning (_("Pid %d died with unknown exit status, using SIGKILL."),
inf->pid);
inf->wait.status.set_signalled (GDB_SIGNAL_KILL);
}
/* Notify server routines. The only real one is dead name notification. */
kern_return_t
do_mach_notify_dead_name (mach_port_t notify, mach_port_t dead_port)
{
struct inf *inf = waiting_inf;
inf_debug (waiting_inf, "port = %lu", dead_port);
if (inf->task && inf->task->port == dead_port)
{
proc_debug (inf->task, "is dead");
inf->task->port = MACH_PORT_NULL;
if (proc_wait_pid == inf->pid)
/* We have a wait outstanding on the process, which will return more
detailed information, so delay until we get that. */
inf->wait.suppress = 1;
else
/* We never waited for the process (maybe it wasn't a child), so just
pretend it got a SIGKILL. */
inf_task_died_status (inf);
}
else
{
struct proc *thread = inf_port_to_thread (inf, dead_port);
if (thread)
{
proc_debug (thread, "is dead");
thread->port = MACH_PORT_NULL;
}
if (inf->task->dead)
/* Since the task is dead, its threads are dying with it. */
inf->wait.suppress = 1;
}
mach_port_deallocate (mach_task_self (), dead_port);
inf->threads_up_to_date = 0; /* Just in case. */
return 0;
}
#define ILL_RPC(fun, ...) \
extern "C" kern_return_t fun (__VA_ARGS__); \
kern_return_t fun (__VA_ARGS__) \
{ \
warning (_("illegal rpc: %s"), #fun); \
return 0; \
}
ILL_RPC (do_mach_notify_no_senders,
mach_port_t notify, mach_port_mscount_t count)
ILL_RPC (do_mach_notify_port_deleted,
mach_port_t notify, mach_port_t name)
ILL_RPC (do_mach_notify_msg_accepted,
mach_port_t notify, mach_port_t name)
ILL_RPC (do_mach_notify_port_destroyed,
mach_port_t notify, mach_port_t name)
ILL_RPC (do_mach_notify_send_once,
mach_port_t notify)
/* Process_reply server routines. We only use process_wait_reply. */
kern_return_t
S_proc_wait_reply (mach_port_t reply, kern_return_t err,
int status, int sigcode, rusage_t rusage, pid_t pid)
{
struct inf *inf = waiting_inf;
inf_debug (inf, "err = %s, pid = %d, status = 0x%x, sigcode = %d",
err ? safe_strerror (err) : "0", pid, status, sigcode);
if (err && proc_wait_pid && (!inf->task || !inf->task->port))
/* Ack. The task has died, but the task-died notification code didn't
tell anyone because it thought a more detailed reply from the
procserver was forthcoming. However, we now learn that won't
happen... So we have to act like the task just died, and this time,
tell the world. */
inf_task_died_status (inf);
if (--proc_waits_pending == 0)
/* PROC_WAIT_PID represents the most recent wait. We will always get
replies in order because the proc server is single threaded. */
proc_wait_pid = 0;
inf_debug (inf, "waits pending now: %d", proc_waits_pending);
if (err)
{
if (err != EINTR)
{
warning (_("Can't wait for pid %d: %s"),
inf->pid, safe_strerror (err));
inf->no_wait = 1;
/* Since we can't see the inferior's signals, don't trap them. */
gnu_target->inf_set_traced (inf, 0);
}
}
else if (pid == inf->pid)
{
inf->wait.status = host_status_to_waitstatus (status);
if (inf->wait.status.kind () == TARGET_WAITKIND_STOPPED)
/* The process has sent us a signal, and stopped itself in a sane
state pending our actions. */
{
inf_debug (inf, "process has stopped itself");
inf->stopped = 1;
}
}
else
inf->wait.suppress = 1; /* Something odd happened. Ignore. */
return 0;
}
ILL_RPC (S_proc_setmsgport_reply,
mach_port_t reply_port, kern_return_t return_code,
mach_port_t oldmsgport)
ILL_RPC (S_proc_getmsgport_reply,
mach_port_t reply_port, kern_return_t return_code,
mach_port_t msgports, mach_msg_type_name_t msgportsPoly)
ILL_RPC (S_proc_pid2task_reply,
mach_port_t reply_port, kern_return_t return_code, mach_port_t task)
ILL_RPC (S_proc_task2pid_reply,
mach_port_t reply_port, kern_return_t return_code, pid_t pid)
ILL_RPC (S_proc_task2proc_reply,
mach_port_t reply_port, kern_return_t return_code,
mach_port_t proc, mach_msg_type_name_t procPoly)
ILL_RPC (S_proc_proc2task_reply,
mach_port_t reply_port, kern_return_t return_code, mach_port_t task)
ILL_RPC (S_proc_pid2proc_reply,
mach_port_t reply_port, kern_return_t return_code,
mach_port_t proc, mach_msg_type_name_t procPoly)
ILL_RPC (S_proc_getprocinfo_reply,
mach_port_t reply_port, kern_return_t return_code,
int flags, const_procinfo_t procinfo, mach_msg_type_number_t procinfoCnt,
const_data_t threadwaits, mach_msg_type_number_t threadwaitsCnt)
ILL_RPC (S_proc_getprocargs_reply,
mach_port_t reply_port, kern_return_t return_code,
const_data_t procargs, mach_msg_type_number_t procargsCnt)
ILL_RPC (S_proc_getprocenv_reply,
mach_port_t reply_port, kern_return_t return_code,
const_data_t procenv, mach_msg_type_number_t procenvCnt)
ILL_RPC (S_proc_getloginid_reply,
mach_port_t reply_port, kern_return_t return_code, pid_t login_id)
ILL_RPC (S_proc_getloginpids_reply,
mach_port_t reply_port, kern_return_t return_code,
const_pidarray_t pids, mach_msg_type_number_t pidsCnt)
ILL_RPC (S_proc_getlogin_reply,
mach_port_t reply_port, kern_return_t return_code, const_string_t logname)
ILL_RPC (S_proc_getsid_reply,
mach_port_t reply_port, kern_return_t return_code, pid_t sid)
ILL_RPC (S_proc_getsessionpgids_reply,
mach_port_t reply_port, kern_return_t return_code,
const_pidarray_t pgidset, mach_msg_type_number_t pgidsetCnt)
ILL_RPC (S_proc_getsessionpids_reply,
mach_port_t reply_port, kern_return_t return_code,
const_pidarray_t pidset, mach_msg_type_number_t pidsetCnt)
ILL_RPC (S_proc_getsidport_reply,
mach_port_t reply_port, kern_return_t return_code,
mach_port_t sessport)
ILL_RPC (S_proc_getpgrp_reply,
mach_port_t reply_port, kern_return_t return_code, pid_t pgrp)
ILL_RPC (S_proc_getpgrppids_reply,
mach_port_t reply_port, kern_return_t return_code,
const_pidarray_t pidset, mach_msg_type_number_t pidsetCnt)
ILL_RPC (S_proc_get_tty_reply,
mach_port_t reply_port, kern_return_t return_code, mach_port_t tty)
ILL_RPC (S_proc_getnports_reply,
mach_port_t reply_port, kern_return_t return_code,
mach_msg_type_number_t nports)
ILL_RPC (S_proc_is_important_reply,
mach_port_t reply_port, kern_return_t return_code,
boolean_t essential)
ILL_RPC (S_proc_get_code_reply,
mach_port_t reply_port, kern_return_t return_code,
vm_address_t start_code, vm_address_t end_code)
/* Msg_reply server routines. We only use msg_sig_post_untraced_reply. */
kern_return_t
S_msg_sig_post_untraced_reply (mach_port_t reply, kern_return_t err)
{
struct inf *inf = waiting_inf;
if (err == EBUSY)
/* EBUSY is what we get when the crash server has grabbed control of the
process and doesn't like what signal we tried to send it. Just act
like the process stopped (using a signal of 0 should mean that the
*next* time the user continues, it will pass signal 0, which the crash
server should like). */
inf->wait.status.set_stopped (GDB_SIGNAL_0);
else if (err)
warning (_("Signal delivery failed: %s"), safe_strerror (err));
if (err)
/* We only get this reply when we've posted a signal to a process which we
thought was stopped, and which we expected to continue after the signal.
Given that the signal has failed for some reason, it's reasonable to
assume it's still stopped. */
inf->stopped = 1;
else
inf->wait.suppress = 1;
return 0;
}
ILL_RPC (S_msg_sig_post_reply,
mach_port_t reply, kern_return_t err)
/* Returns the number of messages queued for the receive right PORT. */
static mach_port_msgcount_t
port_msgs_queued (mach_port_t port)
{
struct mach_port_status status;
kern_return_t err =
mach_port_get_receive_status (mach_task_self (), port, &status);
if (err)
return 0;
else
return status.mps_msgcount;
}
/* Resume execution of the inferior process.
If STEP is nonzero, single-step it.
If SIGNAL is nonzero, give it that signal.
TID STEP:
-1 true Single step the current thread allowing other threads to run.
-1 false Continue the current thread allowing other threads to run.
X true Single step the given thread, don't allow any others to run.
X false Continue the given thread, do not allow any others to run.
(Where X, of course, is anything except -1)
Note that a resume may not `take' if there are pending exceptions/&c
still unprocessed from the last resume we did (any given resume may result
in multiple events returned by wait). */
void
gnu_nat_target::resume (ptid_t ptid, int step, enum gdb_signal sig)
{
struct proc *step_thread = 0;
int resume_all;
struct inf *inf = gnu_current_inf;
inf_debug (inf, "ptid = %s, step = %d, sig = %d",
ptid.to_string ().c_str (), step, sig);
inf_validate_procinfo (inf);
if (sig != GDB_SIGNAL_0 || inf->stopped)
{
if (sig == GDB_SIGNAL_0 && inf->nomsg)
inf_continue (inf);
else
inf_signal (inf, sig);
}
else if (inf->wait.exc.reply != MACH_PORT_NULL)
/* We received an exception to which we have chosen not to forward, so
abort the faulting thread, which will perhaps retake it. */
{
proc_abort (inf->wait.thread, 1);
warning (_("Aborting %s with unforwarded exception %s."),
proc_string (inf->wait.thread),
gdb_signal_to_name (inf->wait.status.sig ()));
}
if (port_msgs_queued (inf->event_port))
/* If there are still messages in our event queue, don't bother resuming
the process, as we're just going to stop it right away anyway. */
return;
inf_update_procs (inf);
/* A specific PTID means `step only this process id'. */
resume_all = ptid == minus_one_ptid;
if (resume_all)
/* Allow all threads to run, except perhaps single-stepping one. */
{
inf_debug (inf, "running all threads; tid = %d",
inferior_ptid.pid ());
ptid = inferior_ptid; /* What to step. */
inf_set_threads_resume_sc (inf, 0, 1);
}
else
/* Just allow a single thread to run. */
{
struct proc *thread = inf_tid_to_thread (inf, ptid.lwp ());
if (!thread)
error (_("Can't run single thread id %s: no such thread!"),
target_pid_to_str (ptid).c_str ());
inf_debug (inf, "running one thread: %s",
ptid.to_string ().c_str ());
inf_set_threads_resume_sc (inf, thread, 0);
}
if (step)
{
step_thread = inf_tid_to_thread (inf, ptid.lwp ());
if (!step_thread)
warning (_("Can't step thread id %s: no such thread."),
target_pid_to_str (ptid).c_str ());
else
inf_debug (inf, "stepping thread: %s",
ptid.to_string ().c_str ());
}
if (step_thread != inf->step_thread)
inf_set_step_thread (inf, step_thread);
inf_debug (inf, "here we go...");
inf_resume (inf);
}
void
gnu_nat_target::kill ()
{
struct proc *task = gnu_current_inf->task;
if (task)
{
proc_debug (task, "terminating...");
task_terminate (task->port);
inf_set_pid (gnu_current_inf, -1);
}
target_mourn_inferior (inferior_ptid);
}
/* Clean up after the inferior dies. */
void
gnu_nat_target::mourn_inferior ()
{
inf_debug (gnu_current_inf, "rip");
inf_detach (gnu_current_inf);
inf_child_target::mourn_inferior ();
}
/* Fork an inferior process, and start debugging it. */
/* Set INFERIOR_PID to the first thread available in the child, if any. */
static int
inf_pick_first_thread (void)
{
if (gnu_current_inf->task && gnu_current_inf->threads)
/* The first thread. */
return gnu_current_inf->threads->tid;
else
/* What may be the next thread. */
return next_thread_id;
}
static struct inf *
cur_inf (void)
{
if (!gnu_current_inf)
gnu_current_inf = make_inf ();
return gnu_current_inf;
}
static void
gnu_ptrace_me (void)
{
/* We're in the child; make this process stop as soon as it execs. */
struct inf *inf = cur_inf ();
inf_debug (inf, "tracing self");
if (ptrace (PTRACE_TRACEME) != 0)
trace_start_error_with_name ("ptrace");
}
void
gnu_nat_target::create_inferior (const char *exec_file,
const std::string &allargs,
char **env,
int from_tty)
{
if (exec_file == nullptr)
no_executable_specified_error ();
struct inf *inf = cur_inf ();
inferior *inferior = current_inferior ();
int pid;
inf_debug (inf, "creating inferior");
if (!inferior->target_is_pushed (this))
inferior->push_target (this);
pid = fork_inferior (exec_file, allargs, env, gnu_ptrace_me,
NULL, NULL, NULL, NULL);
/* We have something that executes now. We'll be running through
the shell at this point (if startup-with-shell is true), but the
pid shouldn't change. */
thread_info *thr = add_thread_silent (this, ptid_t (pid));
switch_to_thread (thr);
/* Attach to the now stopped child, which is actually a shell... */
inf_debug (inf, "attaching to child: %d", pid);
inf_attach (inf, pid);
inf->pending_execs = 1;
inf->nomsg = 1;
inf->traced = 1;
/* Now let the child run again, knowing that it will stop
immediately because of the ptrace. */
inf_resume (inf);
/* We now have thread info. */
thread_change_ptid (this, inferior_ptid,
ptid_t (inf->pid, inf_pick_first_thread (), 0));
gdb_startup_inferior (pid, START_INFERIOR_TRAPS_EXPECTED);
inf->pending_execs = 0;
/* Get rid of the old shell threads. */
prune_threads ();
inf_validate_procinfo (inf);
inf_update_signal_thread (inf);
inf_set_traced (inf, inf->want_signals);
/* Execing the process will have trashed our exception ports; steal them
back (or make sure they're restored if the user wants that). */
if (inf->want_exceptions)
inf_steal_exc_ports (inf);
else
inf_restore_exc_ports (inf);
}
/* Attach to process PID, then initialize for debugging it
and wait for the trace-trap that results from attaching. */
void
gnu_nat_target::attach (const char *args, int from_tty)
{
int pid;
struct inf *inf = cur_inf ();
struct inferior *inferior;
pid = parse_pid_to_attach (args);
if (pid == getpid ()) /* Trying to masturbate? */
error (_("I refuse to debug myself!"));
target_announce_attach (from_tty, pid);
inf_debug (inf, "attaching to pid: %d", pid);
inf_attach (inf, pid);
inferior = current_inferior ();
inferior->push_target (this);
inferior_appeared (inferior, pid);
inferior->attach_flag = true;
inf_update_procs (inf);
thread_info *thr = this->find_thread (ptid_t (pid, inf_pick_first_thread ()));
switch_to_thread (thr);
/* We have to initialize the terminal settings now, since the code
below might try to restore them. */
target_terminal::init ();
/* If the process was stopped before we attached, make it continue the next
time the user does a continue. */
inf_validate_procinfo (inf);
inf_update_signal_thread (inf);
inf_set_traced (inf, inf->want_signals);
#if 0 /* Do we need this? */
renumber_threads (0); /* Give our threads reasonable names. */
#endif
}
/* Take a program previously attached to and detaches it.
The program resumes execution and will no longer stop
on signals, etc. We'd better not have left any breakpoints
in the program or it'll die when it hits one. For this
to work, it may be necessary for the process to have been
previously attached. It *might* work if the program was
started via fork. */
void
gnu_nat_target::detach (inferior *inf, int from_tty)
{
target_announce_detach (from_tty);
inf_detach (gnu_current_inf);
switch_to_no_thread ();
detach_inferior (inf);
maybe_unpush_target ();
}
void
gnu_nat_target::stop (ptid_t ptid)
{
error (_("stop target function not implemented"));
}
bool
gnu_nat_target::thread_alive (ptid_t ptid)
{
inf_update_procs (gnu_current_inf);
return !!inf_tid_to_thread (gnu_current_inf,
ptid.lwp ());
}
/* Read inferior task's LEN bytes from ADDR and copy it to MYADDR in
gdb's address space. Return 0 on failure; number of bytes read
otherwise. */
static int
gnu_read_inferior (task_t task, CORE_ADDR addr, gdb_byte *myaddr, int length)
{
kern_return_t err;
vm_address_t low_address = (vm_address_t) trunc_page (addr);
vm_size_t aligned_length =
(vm_size_t) round_page (addr + length) - low_address;
pointer_t copied;
mach_msg_type_number_t copy_count;
/* Get memory from inferior with page aligned addresses. */
err = vm_read (task, low_address, aligned_length, &copied, &copy_count);
if (err)
return 0;
err = hurd_safe_copyin (myaddr, (void *) (addr - low_address + copied),
length);
if (err)
{
warning (_("Read from inferior faulted: %s"), safe_strerror (err));
length = 0;
}
err = vm_deallocate (mach_task_self (), copied, copy_count);
if (err)
warning (_("gnu_read_inferior vm_deallocate failed: %s"),
safe_strerror (err));
return length;
}
#define CHK_GOTO_OUT(str,ret) \
do if (ret != KERN_SUCCESS) { errstr = #str; goto out; } while(0)
struct vm_region_list
{
struct vm_region_list *next;
vm_prot_t protection;
vm_address_t start;
vm_size_t length;
};
struct obstack region_obstack;
/* Write gdb's LEN bytes from MYADDR and copy it to ADDR in inferior
task's address space. */
static int
gnu_write_inferior (task_t task, CORE_ADDR addr,
const gdb_byte *myaddr, int length)
{
kern_return_t err;
vm_address_t low_address = (vm_address_t) trunc_page (addr);
vm_size_t aligned_length =
(vm_size_t) round_page (addr + length) - low_address;
pointer_t copied;
mach_msg_type_number_t copy_count;
int deallocate = 0;
const char *errstr = "Bug in gnu_write_inferior";
struct vm_region_list *region_element;
struct vm_region_list *region_head = NULL;
/* Get memory from inferior with page aligned addresses. */
err = vm_read (task,
low_address,
aligned_length,
&copied,
&copy_count);
CHK_GOTO_OUT ("gnu_write_inferior vm_read failed", err);
deallocate++;
err = hurd_safe_copyout ((void *) (addr - low_address + copied),
myaddr, length);
CHK_GOTO_OUT ("Write to inferior faulted", err);
obstack_init (&region_obstack);
/* Do writes atomically.
First check for holes and unwritable memory. */
{
vm_size_t remaining_length = aligned_length;
vm_address_t region_address = low_address;
struct vm_region_list *scan;
while (region_address < low_address + aligned_length)
{
vm_prot_t protection;
vm_prot_t max_protection;
vm_inherit_t inheritance;
boolean_t shared;
mach_port_t object_name;
vm_offset_t offset;
vm_size_t region_length = remaining_length;
vm_address_t old_address = region_address;
err = vm_region (task,
&region_address,
&region_length,
&protection,
&max_protection,
&inheritance,
&shared,
&object_name,
&offset);
CHK_GOTO_OUT ("vm_region failed", err);
/* Check for holes in memory. */
if (old_address != region_address)
{
warning (_("No memory at 0x%lx. Nothing written"),
old_address);
err = KERN_SUCCESS;
length = 0;
goto out;
}
if (!(max_protection & VM_PROT_WRITE))
{
warning (_("Memory at address 0x%lx is unwritable. "
"Nothing written"),
old_address);
err = KERN_SUCCESS;
length = 0;
goto out;
}
/* Chain the regions for later use. */
region_element = XOBNEW (&region_obstack, struct vm_region_list);
region_element->protection = protection;
region_element->start = region_address;
region_element->length = region_length;
/* Chain the regions along with protections. */
region_element->next = region_head;
region_head = region_element;
region_address += region_length;
remaining_length = remaining_length - region_length;
}
/* If things fail after this, we give up.
Somebody is messing up inferior_task's mappings. */
/* Enable writes to the chained vm regions. */
for (scan = region_head; scan; scan = scan->next)
{
if (!(scan->protection & VM_PROT_WRITE))
{
err = vm_protect (task,
scan->start,
scan->length,
FALSE,
scan->protection | VM_PROT_WRITE);
CHK_GOTO_OUT ("vm_protect: enable write failed", err);
}
}
err = vm_write (task,
low_address,
copied,
aligned_length);
CHK_GOTO_OUT ("vm_write failed", err);
/* Set up the original region protections, if they were changed. */
for (scan = region_head; scan; scan = scan->next)
{
if (!(scan->protection & VM_PROT_WRITE))
{
err = vm_protect (task,
scan->start,
scan->length,
FALSE,
scan->protection);
CHK_GOTO_OUT ("vm_protect: enable write failed", err);
}
}
}
out:
if (deallocate)
{
obstack_free (&region_obstack, 0);
(void) vm_deallocate (mach_task_self (),
copied,
copy_count);
}
if (err != KERN_SUCCESS)
{
warning (_("%s: %s"), errstr, mach_error_string (err));
return 0;
}
return length;
}
/* Implement the to_xfer_partial target_ops method for
TARGET_OBJECT_MEMORY. */
static enum target_xfer_status
gnu_xfer_memory (gdb_byte *readbuf, const gdb_byte *writebuf,
CORE_ADDR memaddr, ULONGEST len, ULONGEST *xfered_len)
{
task_t task = (gnu_current_inf
? (gnu_current_inf->task
? gnu_current_inf->task->port : 0)
: 0);
int res;
if (task == MACH_PORT_NULL)
return TARGET_XFER_E_IO;
if (writebuf != NULL)
{
inf_debug (gnu_current_inf, "writing %s[%s] <-- %s",
paddress (current_inferior ()->arch (), memaddr), pulongest (len),
host_address_to_string (writebuf));
res = gnu_write_inferior (task, memaddr, writebuf, len);
}
else
{
inf_debug (gnu_current_inf, "reading %s[%s] --> %s",
paddress (current_inferior ()->arch (), memaddr), pulongest (len),
host_address_to_string (readbuf));
res = gnu_read_inferior (task, memaddr, readbuf, len);
}
gdb_assert (res >= 0);
if (res == 0)
return TARGET_XFER_E_IO;
else
{
*xfered_len = (ULONGEST) res;
return TARGET_XFER_OK;
}
}
/* GNU does not have auxv, but we can at least fake the AT_ENTRY entry for PIE
binaries. */
static enum target_xfer_status
gnu_xfer_auxv (gdb_byte *readbuf, const gdb_byte *writebuf,
CORE_ADDR memaddr, ULONGEST len, ULONGEST *xfered_len)
{
task_t task = (gnu_current_inf
? (gnu_current_inf->task
? gnu_current_inf->task->port : 0)
: 0);
process_t proc;
kern_return_t err;
vm_address_t entry;
ElfW(auxv_t) auxv[2];
if (task == MACH_PORT_NULL)
return TARGET_XFER_E_IO;
if (writebuf != NULL)
return TARGET_XFER_E_IO;
if (memaddr == sizeof (auxv))
return TARGET_XFER_EOF;
if (memaddr > sizeof (auxv))
return TARGET_XFER_E_IO;
err = proc_task2proc (proc_server, task, &proc);
if (err != 0)
return TARGET_XFER_E_IO;
/* Get entry from proc server. */
err = proc_get_entry (proc, &entry);
if (err != 0)
return TARGET_XFER_E_IO;
/* Fake auxv entry. */
auxv[0].a_type = AT_ENTRY;
auxv[0].a_un.a_val = entry;
auxv[1].a_type = AT_NULL;
auxv[1].a_un.a_val = 0;
inf_debug (gnu_current_inf, "reading auxv %s[%s] --> %s",
paddress (current_inferior ()->arch (), memaddr), pulongest (len),
host_address_to_string (readbuf));
if (memaddr + len > sizeof (auxv))
len = sizeof (auxv) - memaddr;
memcpy (readbuf, (gdb_byte *) &auxv + memaddr, len);
*xfered_len = len;
return TARGET_XFER_OK;
}
/* Target to_xfer_partial implementation. */
enum target_xfer_status
gnu_nat_target::xfer_partial (enum target_object object,
const char *annex, gdb_byte *readbuf,
const gdb_byte *writebuf, ULONGEST offset,
ULONGEST len, ULONGEST *xfered_len)
{
switch (object)
{
case TARGET_OBJECT_MEMORY:
return gnu_xfer_memory (readbuf, writebuf, offset, len, xfered_len);
case TARGET_OBJECT_AUXV:
return gnu_xfer_auxv (readbuf, writebuf, offset, len, xfered_len);
default:
return TARGET_XFER_E_IO;
}
}
/* Call FUNC on each memory region in the task. */
int
gnu_nat_target::find_memory_regions (find_memory_region_ftype func,
void *data)
{
kern_return_t err;
task_t task;
vm_address_t region_address, last_region_address, last_region_end;
vm_prot_t last_protection;
if (gnu_current_inf == 0 || gnu_current_inf->task == 0)
return 0;
task = gnu_current_inf->task->port;
if (task == MACH_PORT_NULL)
return 0;
region_address = last_region_address = last_region_end = VM_MIN_ADDRESS;
last_protection = VM_PROT_NONE;
while (region_address < VM_MAX_ADDRESS)
{
vm_prot_t protection;
vm_prot_t max_protection;
vm_inherit_t inheritance;
boolean_t shared;
mach_port_t object_name;
vm_offset_t offset;
vm_size_t region_length = VM_MAX_ADDRESS - region_address;
err = vm_region (task,
&region_address,
&region_length,
&protection,
&max_protection,
&inheritance,
&shared,
&object_name,
&offset);
if (err == KERN_NO_SPACE)
break;
if (err != KERN_SUCCESS)
{
warning (_("vm_region failed: %s"), mach_error_string (err));
return -1;
}
if (protection == last_protection && region_address == last_region_end)
/* This region is contiguous with and indistinguishable from
the previous one, so we just extend that one. */
last_region_end = region_address += region_length;
else
{
/* This region is distinct from the last one we saw, so report
that previous one. */
if (last_protection != VM_PROT_NONE)
(*func) (last_region_address,
last_region_end - last_region_address,
last_protection & VM_PROT_READ,
last_protection & VM_PROT_WRITE,
last_protection & VM_PROT_EXECUTE,
1, /* MODIFIED is unknown, pass it as true. */
false, /* No memory tags in the object file. */
data);
last_region_address = region_address;
last_region_end = region_address += region_length;
last_protection = protection;
}
}
/* Report the final region. */
if (last_region_end > last_region_address && last_protection != VM_PROT_NONE)
(*func) (last_region_address, last_region_end - last_region_address,
last_protection & VM_PROT_READ,
last_protection & VM_PROT_WRITE,
last_protection & VM_PROT_EXECUTE,
1, /* MODIFIED is unknown, pass it as true. */
false, /* No memory tags in the object file. */
data);
return 0;
}
/* Return printable description of proc. */
char *
proc_string (struct proc *proc)
{
static char tid_str[80];
if (proc_is_task (proc))
xsnprintf (tid_str, sizeof (tid_str), "process %d", proc->inf->pid);
else
xsnprintf (tid_str, sizeof (tid_str), "Thread %d.%d",
proc->inf->pid, proc->tid);
return tid_str;
}
std::string
gnu_nat_target::pid_to_str (ptid_t ptid)
{
struct inf *inf = gnu_current_inf;
int tid = ptid.lwp ();
struct proc *thread = inf_tid_to_thread (inf, tid);
if (thread)
return proc_string (thread);
else
return string_printf ("bogus thread id %d", tid);
}
/* User task commands. */
static struct cmd_list_element *set_task_cmd_list = 0;
static struct cmd_list_element *show_task_cmd_list = 0;
/* User thread commands. */
/* Commands with a prefix of `set/show thread'. */
extern struct cmd_list_element *thread_cmd_list;
struct cmd_list_element *set_thread_cmd_list = NULL;
struct cmd_list_element *show_thread_cmd_list = NULL;
/* Commands with a prefix of `set/show thread default'. */
struct cmd_list_element *set_thread_default_cmd_list = NULL;
struct cmd_list_element *show_thread_default_cmd_list = NULL;
static int
parse_int_arg (const char *args, const char *cmd_prefix)
{
if (args)
{
char *arg_end;
int val = strtoul (args, &arg_end, 10);
if (*args && *arg_end == '\0')
return val;
}
error (_("Illegal argument for \"%s\" command, should be an integer."),
cmd_prefix);
}
static int
_parse_bool_arg (const char *args, const char *t_val, const char *f_val,
const char *cmd_prefix)
{
if (!args || strcmp (args, t_val) == 0)
return 1;
else if (strcmp (args, f_val) == 0)
return 0;
else
error (_("Illegal argument for \"%s\" command, "
"should be \"%s\" or \"%s\"."),
cmd_prefix, t_val, f_val);
}
#define parse_bool_arg(args, cmd_prefix) \
_parse_bool_arg (args, "on", "off", cmd_prefix)
static void
check_empty (const char *args, const char *cmd_prefix)
{
if (args)
error (_("Garbage after \"%s\" command: `%s'"), cmd_prefix, args);
}
/* Returns the alive thread named by INFERIOR_PID, or signals an error. */
static struct proc *
cur_thread (void)
{
struct inf *inf = cur_inf ();
struct proc *thread = inf_tid_to_thread (inf,
inferior_ptid.lwp ());
if (!thread)
error (_("No current thread."));
return thread;
}
/* Returns the current inferior, but signals an error if it has no task. */
static struct inf *
active_inf (void)
{
struct inf *inf = cur_inf ();
if (!inf->task)
error (_("No current process."));
return inf;
}
static void
set_task_pause_cmd (int arg, int from_tty)
{
struct inf *inf = cur_inf ();
int old_sc = inf->pause_sc;
inf->pause_sc = arg;
if (old_sc == 0 && inf->pause_sc != 0)
/* If the task is currently unsuspended, immediately suspend it,
otherwise wait until the next time it gets control. */
gnu_target->inf_suspend (inf);
}
static void
set_task_pause_cmd (const char *args, int from_tty)
{
set_task_pause_cmd (parse_bool_arg (args, "set task pause"), from_tty);
}
static void
show_task_pause_cmd (const char *args, int from_tty)
{
struct inf *inf = cur_inf ();
check_empty (args, "show task pause");
gdb_printf ("The inferior task %s suspended while gdb has control.\n",
inf->task
? (inf->pause_sc == 0 ? "isn't" : "is")
: (inf->pause_sc == 0 ? "won't be" : "will be"));
}
static void
set_task_detach_sc_cmd (const char *args, int from_tty)
{
cur_inf ()->detach_sc = parse_int_arg (args,
"set task detach-suspend-count");
}
static void
show_task_detach_sc_cmd (const char *args, int from_tty)
{
check_empty (args, "show task detach-suspend-count");
gdb_printf ("The inferior task will be left with a "
"suspend count of %d when detaching.\n",
cur_inf ()->detach_sc);
}
static void
set_thread_default_pause_cmd (const char *args, int from_tty)
{
struct inf *inf = cur_inf ();
inf->default_thread_pause_sc =
parse_bool_arg (args, "set thread default pause") ? 0 : 1;
}
static void
show_thread_default_pause_cmd (const char *args, int from_tty)
{
struct inf *inf = cur_inf ();
int sc = inf->default_thread_pause_sc;
check_empty (args, "show thread default pause");
gdb_printf ("New threads %s suspended while gdb has control%s.\n",
sc ? "are" : "aren't",
!sc && inf->pause_sc ? " (but the task is)" : "");
}
static void
set_thread_default_run_cmd (const char *args, int from_tty)
{
struct inf *inf = cur_inf ();
inf->default_thread_run_sc =
parse_bool_arg (args, "set thread default run") ? 0 : 1;
}
static void
show_thread_default_run_cmd (const char *args, int from_tty)
{
struct inf *inf = cur_inf ();
check_empty (args, "show thread default run");
gdb_printf ("New threads %s allowed to run.\n",
inf->default_thread_run_sc == 0 ? "are" : "aren't");
}
static void
set_thread_default_detach_sc_cmd (const char *args, int from_tty)
{
cur_inf ()->default_thread_detach_sc =
parse_int_arg (args, "set thread default detach-suspend-count");
}
static void
show_thread_default_detach_sc_cmd (const char *args, int from_tty)
{
check_empty (args, "show thread default detach-suspend-count");
gdb_printf ("New threads will get a detach-suspend-count of %d.\n",
cur_inf ()->default_thread_detach_sc);
}
/* Steal a send right called NAME in the inferior task, and make it PROC's
saved exception port. */
void
gnu_nat_target::steal_exc_port (struct proc *proc, mach_port_t name)
{
kern_return_t err;
mach_port_t port;
mach_msg_type_name_t port_type;
if (!proc || !proc->inf->task)
error (_("No inferior task."));
err = mach_port_extract_right (proc->inf->task->port,
name, MACH_MSG_TYPE_COPY_SEND,
&port, &port_type);
if (err)
error (_("Couldn't extract send right %lu from inferior: %s"),
name, safe_strerror (err));
if (proc->saved_exc_port)
/* Get rid of our reference to the old one. */
mach_port_deallocate (mach_task_self (), proc->saved_exc_port);
proc->saved_exc_port = port;
if (!proc->exc_port)
/* If PROC is a thread, we may not have set its exception port
before. We can't use proc_steal_exc_port because it also sets
saved_exc_port. */
{
proc->exc_port = proc->inf->event_port;
err = proc_set_exception_port (proc, proc->exc_port);
error (_("Can't set exception port for %s: %s"),
proc_string (proc), safe_strerror (err));
}
}
static void
set_task_exc_port_cmd (const char *args, int from_tty)
{
struct inf *inf = cur_inf ();
if (!args)
error (_("No argument to \"set task exception-port\" command."));
gnu_target->steal_exc_port (inf->task, parse_and_eval_address (args));
}
static void
set_stopped_cmd (const char *args, int from_tty)
{
cur_inf ()->stopped = _parse_bool_arg (args, "yes", "no", "set stopped");
}
static void
show_stopped_cmd (const char *args, int from_tty)
{
struct inf *inf = active_inf ();
check_empty (args, "show stopped");
gdb_printf ("The inferior process %s stopped.\n",
inf->stopped ? "is" : "isn't");
}
static void
set_sig_thread_cmd (const char *args, int from_tty)
{
struct inf *inf = cur_inf ();
if (!args || (!isdigit (*args) && strcmp (args, "none") != 0))
error (_("Illegal argument to \"set signal-thread\" command.\n"
"Should be a thread ID, or \"none\"."));
if (strcmp (args, "none") == 0)
inf->signal_thread = 0;
else
{
struct thread_info *tp = parse_thread_id (args, NULL);
inf->signal_thread = inf_tid_to_thread (inf, tp->ptid.lwp ());
}
}
static void
show_sig_thread_cmd (const char *args, int from_tty)
{
struct inf *inf = active_inf ();
check_empty (args, "show signal-thread");
if (inf->signal_thread)
gdb_printf ("The signal thread is %s.\n",
proc_string (inf->signal_thread));
else
gdb_printf ("There is no signal thread.\n");
}
static void
set_signals_cmd (int arg, int from_tty)
{
struct inf *inf = cur_inf ();
inf->want_signals = arg;
if (inf->task && inf->want_signals != inf->traced)
/* Make this take effect immediately in a running process. */
gnu_target->inf_set_traced (inf, inf->want_signals);
}
static void
set_signals_cmd (const char *args, int from_tty)
{
set_signals_cmd(parse_bool_arg (args, "set signals"), from_tty);
}
static void
show_signals_cmd (const char *args, int from_tty)
{
struct inf *inf = cur_inf ();
check_empty (args, "show signals");
gdb_printf ("The inferior process's signals %s intercepted.\n",
inf->task
? (inf->traced ? "are" : "aren't")
: (inf->want_signals ? "will be" : "won't be"));
}
static void
set_exceptions_cmd (int arg, int from_tty)
{
struct inf *inf = cur_inf ();
/* Make this take effect immediately in a running process. */
/* XXX */ ;
inf->want_exceptions = arg;
}
static void
set_exceptions_cmd (const char *args, int from_tty)
{
set_exceptions_cmd (parse_bool_arg (args, "set exceptions"), from_tty);
}
static void
show_exceptions_cmd (const char *args, int from_tty)
{
struct inf *inf = cur_inf ();
check_empty (args, "show exceptions");
gdb_printf ("Exceptions in the inferior %s trapped.\n",
inf->task
? (inf->want_exceptions ? "are" : "aren't")
: (inf->want_exceptions ? "will be" : "won't be"));
}
static void
set_task_cmd (const char *args, int from_tty)
{
gdb_printf ("\"set task\" must be followed by the name"
" of a task property.\n");
}
static void
show_task_cmd (const char *args, int from_tty)
{
struct inf *inf = cur_inf ();
check_empty (args, "show task");
show_signals_cmd (0, from_tty);
show_exceptions_cmd (0, from_tty);
show_task_pause_cmd (0, from_tty);
if (inf->pause_sc == 0)
show_thread_default_pause_cmd (0, from_tty);
show_thread_default_run_cmd (0, from_tty);
if (inf->task)
{
show_stopped_cmd (0, from_tty);
show_sig_thread_cmd (0, from_tty);
}
if (inf->detach_sc != 0)
show_task_detach_sc_cmd (0, from_tty);
if (inf->default_thread_detach_sc != 0)
show_thread_default_detach_sc_cmd (0, from_tty);
}
static void
set_noninvasive_cmd (const char *args, int from_tty)
{
/* Invert the sense of the arg for each component. */
int inv_arg = parse_bool_arg (args, "set noninvasive") ? 0 : 1;
set_task_pause_cmd (inv_arg, from_tty);
set_signals_cmd (inv_arg, from_tty);
set_exceptions_cmd (inv_arg, from_tty);
}
static void
info_port_rights (const char *args, mach_port_type_t only)
{
struct inf *inf = active_inf ();
scoped_value_mark vmark;
if (args)
/* Explicit list of port rights. */
{
while (*args)
{
struct value *val = parse_to_comma_and_eval (&args);
long right = value_as_long (val);
error_t err =
print_port_info (right, 0, inf->task->port, PORTINFO_DETAILS,
stdout);
if (err)
error (_("%ld: %s."), right, safe_strerror (err));
}
}
else
/* Print all of them. */
{
error_t err =
print_task_ports_info (inf->task->port, only, PORTINFO_DETAILS,
stdout);
if (err)
error (_("%s."), safe_strerror (err));
}
}
static void
info_send_rights_cmd (const char *args, int from_tty)
{
info_port_rights (args, MACH_PORT_TYPE_SEND);
}
static void
info_recv_rights_cmd (const char *args, int from_tty)
{
info_port_rights (args, MACH_PORT_TYPE_RECEIVE);
}
static void
info_port_sets_cmd (const char *args, int from_tty)
{
info_port_rights (args, MACH_PORT_TYPE_PORT_SET);
}
static void
info_dead_names_cmd (const char *args, int from_tty)
{
info_port_rights (args, MACH_PORT_TYPE_DEAD_NAME);
}
static void
info_port_rights_cmd (const char *args, int from_tty)
{
info_port_rights (args, ~0);
}
static void
add_task_commands (void)
{
add_cmd ("pause", class_run, set_thread_default_pause_cmd, _("\
Set whether the new threads are suspended while gdb has control.\n\
This property normally has no effect because the whole task is\n\
suspended, however, that may be disabled with \"set task pause off\".\n\
The default value is \"off\"."),
&set_thread_default_cmd_list);
add_cmd ("pause", no_class, show_thread_default_pause_cmd, _("\
Show whether new threads are suspended while gdb has control."),
&show_thread_default_cmd_list);
add_cmd ("run", class_run, set_thread_default_run_cmd, _("\
Set whether new threads are allowed to run (once gdb has noticed them)."),
&set_thread_default_cmd_list);
add_cmd ("run", no_class, show_thread_default_run_cmd, _("\
Show whether new threads are allowed to run (once gdb has noticed them)."),
&show_thread_default_cmd_list);
add_cmd ("detach-suspend-count", class_run, set_thread_default_detach_sc_cmd,
_("Set the default detach-suspend-count value for new threads."),
&set_thread_default_cmd_list);
add_cmd ("detach-suspend-count", no_class, show_thread_default_detach_sc_cmd,
_("Show the default detach-suspend-count value for new threads."),
&show_thread_default_cmd_list);
cmd_list_element *set_signals_cmd_
= add_cmd ("signals", class_run, set_signals_cmd, _("\
Set whether the inferior process's signals will be intercepted.\n\
Mach exceptions (such as breakpoint traps) are not affected."),
&setlist);
add_alias_cmd ("sigs", set_signals_cmd_, class_run, 1, &setlist);
cmd_list_element *show_signals_cmd_
= add_cmd ("signals", no_class, show_signals_cmd, _("\
Show whether the inferior process's signals will be intercepted."),
&showlist);
add_alias_cmd ("sigs", show_signals_cmd_, no_class, 1, &showlist);
cmd_list_element *set_signal_thread_cmd_
= add_cmd ("signal-thread", class_run, set_sig_thread_cmd, _("\
Set the thread that gdb thinks is the libc signal thread.\n\
This thread is run when delivering a signal to a non-stopped process."),
&setlist);
add_alias_cmd ("sigthread", set_signal_thread_cmd_, class_run, 1, &setlist);
cmd_list_element *show_signal_thread_cmd_
= add_cmd ("signal-thread", no_class, show_sig_thread_cmd, _("\
Set the thread that gdb thinks is the libc signal thread."),
&showlist);
add_alias_cmd ("sigthread", show_signal_thread_cmd_, no_class, 1, &showlist);
add_cmd ("stopped", class_run, set_stopped_cmd, _("\
Set whether gdb thinks the inferior process is stopped as with SIGSTOP.\n\
Stopped process will be continued by sending them a signal."),
&setlist);
add_cmd ("stopped", no_class, show_stopped_cmd, _("\
Show whether gdb thinks the inferior process is stopped as with SIGSTOP."),
&showlist);
cmd_list_element *set_exceptions_cmd_
= add_cmd ("exceptions", class_run, set_exceptions_cmd, _("\
Set whether exceptions in the inferior process will be trapped.\n\
When exceptions are turned off, neither breakpoints nor single-stepping\n\
will work."), &setlist);
/* Allow `set exc' despite conflict with `set exception-port'. */
add_alias_cmd ("exc", set_exceptions_cmd_, class_run, 1, &setlist);
add_cmd ("exceptions", no_class, show_exceptions_cmd, _("\
Show whether exceptions in the inferior process will be trapped."),
&showlist);
add_prefix_cmd ("task", no_class, set_task_cmd,
_("Command prefix for setting task attributes."),
&set_task_cmd_list, 0, &setlist);
add_prefix_cmd ("task", no_class, show_task_cmd,
_("Command prefix for showing task attributes."),
&show_task_cmd_list, 0, &showlist);
add_cmd ("pause", class_run, set_task_pause_cmd, _("\
Set whether the task is suspended while gdb has control.\n\
A value of \"on\" takes effect immediately, otherwise nothing happens\n\
until the next time the program is continued.\n\
When setting this to \"off\", \"set thread default pause on\" can be\n\
used to pause individual threads by default instead."),
&set_task_cmd_list);
add_cmd ("pause", no_class, show_task_pause_cmd,
_("Show whether the task is suspended while gdb has control."),
&show_task_cmd_list);
add_cmd ("detach-suspend-count", class_run, set_task_detach_sc_cmd,
_("Set the suspend count will leave on the thread when detaching."),
&set_task_cmd_list);
add_cmd ("detach-suspend-count", no_class, show_task_detach_sc_cmd,
_("Show the suspend count will leave "
"on the thread when detaching."),
&show_task_cmd_list);
cmd_list_element *set_task_exception_port_cmd_
= add_cmd ("exception-port", no_class, set_task_exc_port_cmd, _("\
Set the task exception port to which we forward exceptions.\n\
The argument should be the value of the send right in the task."),
&set_task_cmd_list);
add_alias_cmd ("excp", set_task_exception_port_cmd_, no_class, 1,
&set_task_cmd_list);
add_alias_cmd ("exc-port", set_task_exception_port_cmd_, no_class, 1,
&set_task_cmd_list);
/* A convenient way of turning on all options require to noninvasively
debug running tasks. */
add_cmd ("noninvasive", no_class, set_noninvasive_cmd, _("\
Set task options so that we interfere as little as possible.\n\
This is the same as setting `task pause', `exceptions', and\n\
`signals' to the opposite value."),
&setlist);
/* Commands to show information about the task's ports. */
add_info ("send-rights", info_send_rights_cmd,
_("Show information about the task's send rights."));
add_info ("receive-rights", info_recv_rights_cmd,
_("Show information about the task's receive rights."));
cmd_list_element *port_rights_cmd
= add_info ("port-rights", info_port_rights_cmd,
_("Show information about the task's port rights."));
cmd_list_element *port_sets_cmd
= add_info ("port-sets", info_port_sets_cmd,
_("Show information about the task's port sets."));
add_info ("dead-names", info_dead_names_cmd,
_("Show information about the task's dead names."));
add_info_alias ("ports", port_rights_cmd, 1);
add_info_alias ("port", port_rights_cmd, 1);
add_info_alias ("psets", port_sets_cmd, 1);
}
static void
set_thread_pause_cmd (const char *args, int from_tty)
{
struct proc *thread = cur_thread ();
int old_sc = thread->pause_sc;
thread->pause_sc = parse_bool_arg (args, "set thread pause");
if (old_sc == 0 && thread->pause_sc != 0 && thread->inf->pause_sc == 0)
/* If the task is currently unsuspended, immediately suspend it,
otherwise wait until the next time it gets control. */
gnu_target->inf_suspend (thread->inf);
}
static void
show_thread_pause_cmd (const char *args, int from_tty)
{
struct proc *thread = cur_thread ();
int sc = thread->pause_sc;
check_empty (args, "show task pause");
gdb_printf ("Thread %s %s suspended while gdb has control%s.\n",
proc_string (thread),
sc ? "is" : "isn't",
!sc && thread->inf->pause_sc ? " (but the task is)" : "");
}
static void
set_thread_run_cmd (const char *args, int from_tty)
{
struct proc *thread = cur_thread ();
thread->run_sc = parse_bool_arg (args, "set thread run") ? 0 : 1;
}
static void
show_thread_run_cmd (const char *args, int from_tty)
{
struct proc *thread = cur_thread ();
check_empty (args, "show thread run");
gdb_printf ("Thread %s %s allowed to run.",
proc_string (thread),
thread->run_sc == 0 ? "is" : "isn't");
}
static void
set_thread_detach_sc_cmd (const char *args, int from_tty)
{
cur_thread ()->detach_sc = parse_int_arg (args,
"set thread detach-suspend-count");
}
static void
show_thread_detach_sc_cmd (const char *args, int from_tty)
{
struct proc *thread = cur_thread ();
check_empty (args, "show thread detach-suspend-count");
gdb_printf ("Thread %s will be left with a suspend count"
" of %d when detaching.\n",
proc_string (thread),
thread->detach_sc);
}
static void
set_thread_exc_port_cmd (const char *args, int from_tty)
{
struct proc *thread = cur_thread ();
if (!args)
error (_("No argument to \"set thread exception-port\" command."));
gnu_target->steal_exc_port (thread, parse_and_eval_address (args));
}
#if 0
static void
show_thread_cmd (char *args, int from_tty)
{
struct proc *thread = cur_thread ();
check_empty (args, "show thread");
show_thread_run_cmd (0, from_tty);
show_thread_pause_cmd (0, from_tty);
if (thread->detach_sc != 0)
show_thread_detach_sc_cmd (0, from_tty);
}
#endif
static void
thread_takeover_sc_cmd (const char *args, int from_tty)
{
struct proc *thread = cur_thread ();
thread_basic_info_data_t _info;
thread_basic_info_t info = &_info;
mach_msg_type_number_t info_len = THREAD_BASIC_INFO_COUNT;
kern_return_t err
= mach_thread_info (thread->port, THREAD_BASIC_INFO,
(int *) &info, &info_len);
if (err)
error (("%s."), safe_strerror (err));
thread->sc = info->suspend_count;
if (from_tty)
gdb_printf ("Suspend count was %d.\n", thread->sc);
if (info != &_info)
vm_deallocate (mach_task_self (), (vm_address_t) info,
info_len * sizeof (int));
}
static void
add_thread_commands (void)
{
add_setshow_prefix_cmd ("thread", no_class,
_("Command prefix for setting thread properties."),
_("Command prefix for showing thread properties."),
&set_thread_cmd_list,
&show_thread_cmd_list,
&setlist, &showlist);
add_setshow_prefix_cmd ("default", no_class,
_("Command prefix for setting default thread properties."),
_("Command prefix for showing default thread properties."),
&set_thread_default_cmd_list,
&show_thread_default_cmd_list,
&set_thread_cmd_list, &show_thread_cmd_list);
add_cmd ("pause", class_run, set_thread_pause_cmd, _("\
Set whether the current thread is suspended while gdb has control.\n\
A value of \"on\" takes effect immediately, otherwise nothing happens\n\
until the next time the program is continued. This property normally\n\
has no effect because the whole task is suspended, however, that may\n\
be disabled with \"set task pause off\".\n\
The default value is \"off\"."),
&set_thread_cmd_list);
add_cmd ("pause", no_class, show_thread_pause_cmd, _("\
Show whether the current thread is suspended while gdb has control."),
&show_thread_cmd_list);
add_cmd ("run", class_run, set_thread_run_cmd,
_("Set whether the current thread is allowed to run."),
&set_thread_cmd_list);
add_cmd ("run", no_class, show_thread_run_cmd,
_("Show whether the current thread is allowed to run."),
&show_thread_cmd_list);
add_cmd ("detach-suspend-count", class_run, set_thread_detach_sc_cmd, _("\
Set the suspend count will leave on the thread when detaching.\n\
Note that this is relative to suspend count when gdb noticed the thread;\n\
use the `thread takeover-suspend-count' to force it to an absolute value."),
&set_thread_cmd_list);
add_cmd ("detach-suspend-count", no_class, show_thread_detach_sc_cmd, _("\
Show the suspend count will leave on the thread when detaching.\n\
Note that this is relative to suspend count when gdb noticed the thread;\n\
use the `thread takeover-suspend-count' to force it to an absolute value."),
&show_thread_cmd_list);
cmd_list_element *set_thread_exception_port_cmd_
= add_cmd ("exception-port", no_class, set_thread_exc_port_cmd, _("\
Set the thread exception port to which we forward exceptions.\n\
This overrides the task exception port.\n\
The argument should be the value of the send right in the task."),
&set_thread_cmd_list);
add_alias_cmd ("excp", set_thread_exception_port_cmd_, no_class, 1,
&set_thread_cmd_list);
add_alias_cmd ("exc-port", set_thread_exception_port_cmd_, no_class, 1,
&set_thread_cmd_list);
add_cmd ("takeover-suspend-count", no_class, thread_takeover_sc_cmd, _("\
Force the threads absolute suspend-count to be gdb's.\n\
Prior to giving this command, gdb's thread suspend-counts are relative\n\
to the thread's initial suspend-count when gdb notices the threads."),
&thread_cmd_list);
}
void _initialize_gnu_nat ();
void
_initialize_gnu_nat ()
{
proc_server = getproc ();
add_task_commands ();
add_thread_commands ();
add_setshow_boolean_cmd ("gnu-nat", class_maintenance,
&gnu_debug_flag,
_("Set debugging output for the gnu backend."),
_("Show debugging output for the gnu backend."),
NULL,
NULL,
NULL,
&setdebuglist,
&showdebuglist);
}
#ifdef FLUSH_INFERIOR_CACHE
/* When over-writing code on some machines the I-Cache must be flushed
explicitly, because it is not kept coherent by the lazy hardware.
This definitely includes breakpoints, for instance, or else we
end up looping in mysterious Bpt traps. */
void
flush_inferior_icache (CORE_ADDR pc, int amount)
{
vm_machine_attribute_val_t flush = MATTR_VAL_ICACHE_FLUSH;
kern_return_t ret;
ret = vm_machine_attribute (gnu_current_inf->task->port,
pc,
amount,
MATTR_CACHE,
&flush);
if (ret != KERN_SUCCESS)
warning (_("Error flushing inferior's cache : %s"), safe_strerror (ret));
}
#endif /* FLUSH_INFERIOR_CACHE */