Simple test-case,

	int main(void)
	{
		int pid, status;

		pid = fork();
		if (!pid) {
			pause();
			assert(0);
			return 0x23;
		}

		assert(ptrace(PTRACE_ATTACH, pid, 0,0) == 0);
		assert(wait(&status) == pid);
		assert(WIFSTOPPED(status) && WSTOPSIG(status) == SIGSTOP);

		kill(pid, SIGCONT);	// <--- also clears STOP_DEQUEUD

		assert(ptrace(PTRACE_CONT, pid, 0,0) == 0);
		assert(wait(&status) == pid);
		assert(WIFSTOPPED(status) && WSTOPSIG(status) == SIGCONT);

		assert(ptrace(PTRACE_CONT, pid, 0, SIGSTOP) == 0);
		assert(wait(&status) == pid);
		assert(WIFSTOPPED(status) && WSTOPSIG(status) == SIGSTOP);

		kill(pid, SIGKILL);
		return 0;
	}

Without the patch it hangs. After the patch SIGSTOP "injected" by the
tracer is not ignored and stops the tracee.

Note also that if this test-case uses, say, SIGWINCH instead of SIGCONT,
everything works without the patch. This can't be right, and this is
confusing.

The problem is that SIGSTOP (or any other sig_kernel_stop() signal) has
no effect without JOBCTL_STOP_DEQUEUED. This means it is simply ignored
after PTRACE_CONT unless JOBCTL_STOP_DEQUEUED was set "by accident", say
it wasn't cleared after initial SIGSTOP sent by PTRACE_ATTACH.

At first glance we could change ptrace_signal() to add STOP_DEQUEUED
after return from ptrace_stop(), but this is not right in case when the
tracer does not change the reported SIGSTOP and SIGCONT comes in between.
This is even more wrong with PT_SEIZED, SIGCONT adds JOBCTL_TRAP_NOTIFY
which will be "lost" during the TRAP_STOP | TRAP_NOTIFY report.

So lets add STOP_DEQUEUED _before_ we report the signal. It has no effect
unless sig_kernel_stop() == T after the tracer resumes us, and in the
latter case the pending STOP_DEQUEUED means no SIGCONT in between, we
should stop.

Note also that if SIGCONT was sent, PT_SEIZED tracee will correctly
report PTRACE_EVENT_STOP/SIGTRAP and thus the tracer can notice the fact
SIGSTOP was cancelled.

Also, move the current->ptrace check from ptrace_signal() to its caller,
get_signal_to_deliver(), this looks more natural.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Acked-by: NTejun Heo <tj@kernel.org>

The __lock_task_sighand() function calls rcu_read_lock() with interrupts
and preemption enabled, but later calls rcu_read_unlock() with interrupts
disabled.  It is therefore possible that this RCU read-side critical
section will be preempted and later RCU priority boosted, which means that
rcu_read_unlock() will call rt_mutex_unlock() in order to deboost itself, but
with interrupts disabled. This results in lockdep splats, so this commit
nests the RCU read-side critical section within the interrupt-disabled
region of code.  This prevents the RCU read-side critical section from
being preempted, and thus prevents the attempt to deboost with interrupts
disabled.

It is quite possible that a better long-term fix is to make rt_mutex_unlock()
disable irqs when acquiring the rt_mutex structure's ->wait_lock.
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

Kill real_parent_is_ptracer() and update the callers to use
ptrace_reparented(), after the previous patch they do the same.

Remove the unnecessary ->ptrace != 0 check in get_signal_to_deliver(),
if ptrace_reparented() == T then the task must be ptraced.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Acked-by: NTejun Heo <tj@kernel.org>

__ptrace_detach() and do_notify_parent() set task->exit_signal = -1
to mark the task dead. This is no longer needed, nobody checks
exit_signal to detect the EXIT_DEAD task.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Reviewed-by: NTejun Heo <tj@kernel.org>

- change do_notify_parent() to return a boolean, true if the task should
  be reaped because its parent ignores SIGCHLD.

- update the only caller which checks the returned value, exit_notify().

This temporary uglifies exit_notify() even more, will be cleanuped by
the next change.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Acked-by: NTejun Heo <tj@kernel.org>

At this point, tracehooks aren't useful to mainline kernel and mostly
just add an extra layer of obfuscation.  Although they have comments,
without actual in-kernel users, it is difficult to tell what are their
assumptions and they're actually trying to achieve.  To mainline
kernel, they just aren't worth keeping around.

This patch kills the following trivial tracehooks.

* Ones testing whether task is ptraced.  Replace with ->ptrace test.

	tracehook_expect_breakpoints()
	tracehook_consider_ignored_signal()
	tracehook_consider_fatal_signal()

* ptrace_event() wrappers.  Call directly.

	tracehook_report_exec()
	tracehook_report_exit()
	tracehook_report_vfork_done()

* ptrace_release_task() wrapper.  Call directly.

	tracehook_finish_release_task()

* noop

	tracehook_prepare_release_task()
	tracehook_report_death()

This doesn't introduce any behavior change.
Signed-off-by: NTejun Heo <tj@kernel.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: NOleg Nesterov <oleg@redhat.com>

task_ptrace(task) simply dereferences task->ptrace and isn't even used
consistently only adding confusion.  Kill it and directly access
->ptrace instead.

This doesn't introduce any behavior change.
Signed-off-by: NTejun Heo <tj@kernel.org>
Signed-off-by: NOleg Nesterov <oleg@redhat.com>

The previous patch implemented async notification for ptrace but it
only worked while trace is running.  This patch introduces
PTRACE_LISTEN which is suggested by Oleg Nestrov.

It's allowed iff tracee is in STOP trap and puts tracee into
quasi-running state - tracee never really runs but wait(2) and
ptrace(2) consider it to be running.  While ptracer is listening,
tracee is allowed to re-enter STOP to notify an async event.
Listening state is cleared on the first notification.  Ptracer can
also clear it by issuing INTERRUPT - tracee will re-trap into STOP
with listening state cleared.

This allows ptracer to monitor group stop state without running tracee
- use INTERRUPT to put tracee into STOP trap, issue LISTEN and then
wait(2) to wait for the next group stop event.  When it happens,
PTRACE_GETSIGINFO provides information to determine the current state.

Test program follows.

  #define PTRACE_SEIZE		0x4206
  #define PTRACE_INTERRUPT	0x4207
  #define PTRACE_LISTEN		0x4208

  #define PTRACE_SEIZE_DEVEL	0x80000000

  static const struct timespec ts1s = { .tv_sec = 1 };

  int main(int argc, char **argv)
  {
	  pid_t tracee, tracer;
	  int i;

	  tracee = fork();
	  if (!tracee)
		  while (1)
			  pause();

	  tracer = fork();
	  if (!tracer) {
		  siginfo_t si;

		  ptrace(PTRACE_SEIZE, tracee, NULL,
			 (void *)(unsigned long)PTRACE_SEIZE_DEVEL);
		  ptrace(PTRACE_INTERRUPT, tracee, NULL, NULL);
	  repeat:
		  waitid(P_PID, tracee, NULL, WSTOPPED);

		  ptrace(PTRACE_GETSIGINFO, tracee, NULL, &si);
		  if (!si.si_code) {
			  printf("tracer: SIG %d\n", si.si_signo);
			  ptrace(PTRACE_CONT, tracee, NULL,
				 (void *)(unsigned long)si.si_signo);
			  goto repeat;
		  }
		  printf("tracer: stopped=%d signo=%d\n",
			 si.si_signo != SIGTRAP, si.si_signo);
		  if (si.si_signo != SIGTRAP)
			  ptrace(PTRACE_LISTEN, tracee, NULL, NULL);
		  else
			  ptrace(PTRACE_CONT, tracee, NULL, NULL);
		  goto repeat;
	  }

	  for (i = 0; i < 3; i++) {
		  nanosleep(&ts1s, NULL);
		  printf("mother: SIGSTOP\n");
		  kill(tracee, SIGSTOP);
		  nanosleep(&ts1s, NULL);
		  printf("mother: SIGCONT\n");
		  kill(tracee, SIGCONT);
	  }
	  nanosleep(&ts1s, NULL);

	  kill(tracer, SIGKILL);
	  kill(tracee, SIGKILL);
	  return 0;
  }

This is identical to the program to test TRAP_NOTIFY except that
tracee is PTRACE_LISTEN'd instead of PTRACE_CONT'd when group stopped.
This allows ptracer to monitor when group stop ends without running
tracee.

  # ./test-listen
  tracer: stopped=0 signo=5
  mother: SIGSTOP
  tracer: SIG 19
  tracer: stopped=1 signo=19
  mother: SIGCONT
  tracer: stopped=0 signo=5
  tracer: SIG 18
  mother: SIGSTOP
  tracer: SIG 19
  tracer: stopped=1 signo=19
  mother: SIGCONT
  tracer: stopped=0 signo=5
  tracer: SIG 18
  mother: SIGSTOP
  tracer: SIG 19
  tracer: stopped=1 signo=19
  mother: SIGCONT
  tracer: stopped=0 signo=5
  tracer: SIG 18

-v2: Moved JOBCTL_LISTENING check in wait_task_stopped() into
     task_stopped_code() as suggested by Oleg.
Signed-off-by: NTejun Heo <tj@kernel.org>
Cc: Oleg Nesterov <oleg@redhat.com>

Currently there's no way for ptracer to find out whether group stop
finished other than polling with INTERRUPT - GETSIGINFO - CONT
sequence.  This patch implements group stop notification for ptracer
using STOP traps.

When group stop state of a seized tracee changes, JOBCTL_TRAP_NOTIFY
is set, which schedules a STOP trap which is sticky - it isn't cleared
by other traps and at least one STOP trap will happen eventually.
STOP trap is synchronization point for event notification and the
tracer can determine the current group stop state by looking at the
signal number portion of exit code (si_status from waitid(2) or
si_code from PTRACE_GETSIGINFO).

Notifications are generated both on start and end of group stops but,
because group stop participation always happens before STOP trap, this
doesn't cause an extra trap while tracee is participating in group
stop.  The symmetry will be useful later.

Note that this notification works iff tracee is not trapped.
Currently there is no way to be notified of group stop state changes
while tracee is trapped.  This will be addressed by a later patch.

An example program follows.

  #define PTRACE_SEIZE		0x4206
  #define PTRACE_INTERRUPT	0x4207

  #define PTRACE_SEIZE_DEVEL	0x80000000

  static const struct timespec ts1s = { .tv_sec = 1 };

  int main(int argc, char **argv)
  {
	  pid_t tracee, tracer;
	  int i;

	  tracee = fork();
	  if (!tracee)
		  while (1)
			  pause();

	  tracer = fork();
	  if (!tracer) {
		  siginfo_t si;

		  ptrace(PTRACE_SEIZE, tracee, NULL,
			 (void *)(unsigned long)PTRACE_SEIZE_DEVEL);
		  ptrace(PTRACE_INTERRUPT, tracee, NULL, NULL);
	  repeat:
		  waitid(P_PID, tracee, NULL, WSTOPPED);

		  ptrace(PTRACE_GETSIGINFO, tracee, NULL, &si);
		  if (!si.si_code) {
			  printf("tracer: SIG %d\n", si.si_signo);
			  ptrace(PTRACE_CONT, tracee, NULL,
				 (void *)(unsigned long)si.si_signo);
			  goto repeat;
		  }
		  printf("tracer: stopped=%d signo=%d\n",
			 si.si_signo != SIGTRAP, si.si_signo);
		  ptrace(PTRACE_CONT, tracee, NULL, NULL);
		  goto repeat;
	  }

	  for (i = 0; i < 3; i++) {
		  nanosleep(&ts1s, NULL);
		  printf("mother: SIGSTOP\n");
		  kill(tracee, SIGSTOP);
		  nanosleep(&ts1s, NULL);
		  printf("mother: SIGCONT\n");
		  kill(tracee, SIGCONT);
	  }
	  nanosleep(&ts1s, NULL);

	  kill(tracer, SIGKILL);
	  kill(tracee, SIGKILL);
	  return 0;
  }

In the above program, tracer keeps tracee running and gets
notification of each group stop state changes.

  # ./test-notify
  tracer: stopped=0 signo=5
  mother: SIGSTOP
  tracer: SIG 19
  tracer: stopped=1 signo=19
  mother: SIGCONT
  tracer: stopped=0 signo=5
  tracer: SIG 18
  mother: SIGSTOP
  tracer: SIG 19
  tracer: stopped=1 signo=19
  mother: SIGCONT
  tracer: stopped=0 signo=5
  tracer: SIG 18
  mother: SIGSTOP
  tracer: SIG 19
  tracer: stopped=1 signo=19
  mother: SIGCONT
  tracer: stopped=0 signo=5
  tracer: SIG 18
Signed-off-by: NTejun Heo <tj@kernel.org>
Cc: Oleg Nesterov <oleg@redhat.com>

PTRACE_ATTACH implicitly issues SIGSTOP on attach which has side
effects on tracee signal and job control states.  This patch
implements a new ptrace request PTRACE_SEIZE which attaches a tracee
without trapping it or affecting its signal and job control states.

The usage is the same with PTRACE_ATTACH but it takes PTRACE_SEIZE_*
flags in @data.  Currently, the only defined flag is
PTRACE_SEIZE_DEVEL which is a temporary flag to enable PTRACE_SEIZE.
PTRACE_SEIZE will change ptrace behaviors outside of attach itself.
The changes will be implemented gradually and the DEVEL flag is to
prevent programs which expect full SEIZE behavior from using it before
all the behavior modifications are complete while allowing unit
testing.  The flag will be removed once SEIZE behaviors are completely
implemented.

* PTRACE_SEIZE, unlike ATTACH, doesn't force tracee to trap.  After
  attaching tracee continues to run unless a trap condition occurs.

* PTRACE_SEIZE doesn't affect signal or group stop state.

* If PTRACE_SEIZE'd, group stop uses PTRACE_EVENT_STOP trap which uses
  exit_code of (signr | PTRACE_EVENT_STOP << 8) where signr is one of
  the stopping signals if group stop is in effect or SIGTRAP
  otherwise, and returns usual trap siginfo on PTRACE_GETSIGINFO
  instead of NULL.

Seizing sets PT_SEIZED in ->ptrace of the tracee.  This flag will be
used to determine whether new SEIZE behaviors should be enabled.

Test program follows.

  #define PTRACE_SEIZE		0x4206
  #define PTRACE_SEIZE_DEVEL	0x80000000

  static const struct timespec ts100ms = { .tv_nsec = 100000000 };
  static const struct timespec ts1s = { .tv_sec = 1 };
  static const struct timespec ts3s = { .tv_sec = 3 };

  int main(int argc, char **argv)
  {
	  pid_t tracee;

	  tracee = fork();
	  if (tracee == 0) {
		  nanosleep(&ts100ms, NULL);
		  while (1) {
			  printf("tracee: alive\n");
			  nanosleep(&ts1s, NULL);
		  }
	  }

	  if (argc > 1)
		  kill(tracee, SIGSTOP);

	  nanosleep(&ts100ms, NULL);

	  ptrace(PTRACE_SEIZE, tracee, NULL,
		 (void *)(unsigned long)PTRACE_SEIZE_DEVEL);
	  if (argc > 1) {
		  waitid(P_PID, tracee, NULL, WSTOPPED);
		  ptrace(PTRACE_CONT, tracee, NULL, NULL);
	  }
	  nanosleep(&ts3s, NULL);
	  printf("tracer: exiting\n");
	  return 0;
  }

When the above program is called w/o argument, tracee is seized while
running and remains running.  When tracer exits, tracee continues to
run and print out messages.

  # ./test-seize-simple
  tracee: alive
  tracee: alive
  tracee: alive
  tracer: exiting
  tracee: alive
  tracee: alive

When called with an argument, tracee is seized from stopped state and
continued, and returns to stopped state when tracer exits.

  # ./test-seize
  tracee: alive
  tracee: alive
  tracee: alive
  tracer: exiting
  # ps -el|grep test-seize
  1 T     0  4720     1  0  80   0 -   941 signal ttyS0    00:00:00 test-seize

-v2: SEIZE doesn't schedule TRAP_STOP and leaves tracee running as Jan
     suggested.

-v3: PTRACE_EVENT_STOP traps now report group stop state by signr.  If
     group stop is in effect the stop signal number is returned as
     part of exit_code; otherwise, SIGTRAP.  This was suggested by
     Denys and Oleg.
Signed-off-by: NTejun Heo <tj@kernel.org>
Cc: Jan Kratochvil <jan.kratochvil@redhat.com>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Oleg Nesterov <oleg@redhat.com>

do_signal_stop() implemented both normal group stop and trap for group
stop while ptraced.  This approach has been enough but scheduled
changes require trap mechanism which can be used in more generic
manner and using group stop trap for generic trap site simplifies both
userland visible interface and implementation.

This patch adds a new jobctl flag - JOBCTL_TRAP_STOP.  When set, it
triggers a trap site, which behaves like group stop trap, in
get_signal_to_deliver() after checking for pending signals.  While
ptraced, do_signal_stop() doesn't stop itself.  It initiates group
stop if requested and schedules JOBCTL_TRAP_STOP and returns.  The
caller - get_signal_to_deliver() - is responsible for checking whether
TRAP_STOP is pending afterwards and handling it.

ptrace_attach() is updated to use JOBCTL_TRAP_STOP instead of
JOBCTL_STOP_PENDING and __ptrace_unlink() to clear all pending trap
bits and TRAPPING so that TRAP_STOP and future trap bits don't linger
after detach.

While at it, add proper function comment to do_signal_stop() and make
it return bool.

-v2: __ptrace_unlink() updated to clear JOBCTL_TRAP_MASK and TRAPPING
     instead of JOBCTL_PENDING_MASK.  This avoids accidentally
     clearing JOBCTL_STOP_CONSUME.  Spotted by Oleg.

-v3: do_signal_stop() updated to return %false without dropping
     siglock while ptraced and TRAP_STOP check moved inside for(;;)
     loop after group stop participation.  This avoids unnecessary
     relocking and also will help avoiding unnecessary traps by
     consuming group stop before handling pending traps.

-v4: Jobctl trap handling moved into a separate function -
     do_jobctl_trap().
Signed-off-by: NTejun Heo <tj@kernel.org>
Cc: Oleg Nesterov <oleg@redhat.com>

Fix kernel-doc warnings in signal.c:

  Warning(kernel/signal.c:2374): No description found for parameter 'nset'
  Warning(kernel/signal.c:2374): Excess function parameter 'set' description in 'sys_rt_sigprocmask'
Signed-off-by: NRandy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Remove the following three noop tracehooks in signals.c.

* tracehook_force_sigpending()
* tracehook_get_signal()
* tracehook_finish_jctl()

The code area is about to be updated and these hooks don't do anything
other than obfuscating the logic.
Signed-off-by: NTejun Heo <tj@kernel.org>
Signed-off-by: NOleg Nesterov <oleg@redhat.com>

ptracer->signal->wait_chldexit was used to wait for TRAPPING; however,
->wait_chldexit was already complicated with waker-side filtering
without adding TRAPPING wait on top of it.  Also, it unnecessarily
made TRAPPING clearing depend on the current ptrace relationship - if
the ptracee is detached, wakeup is lost.

There is no reason to use signal->wait_chldexit here.  We're just
waiting for JOBCTL_TRAPPING bit to clear and given the relatively
infrequent use of ptrace, bit_waitqueue can serve it perfectly.

This patch makes JOBCTL_TRAPPING wait use bit_waitqueue instead of
signal->wait_chldexit.

-v2: Use JOBCTL_*_BIT macros instead of ilog2() as suggested by Linus.
Signed-off-by: NTejun Heo <tj@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: NOleg Nesterov <oleg@redhat.com>

task->jobctl currently hosts JOBCTL_STOP_PENDING and will host TRAP
pending bits too.  Setting pending conditions on a dying task may make
the task unkillable.  Currently, each setting site is responsible for
checking for the condition but with to-be-added job control traps this
becomes too fragile.

This patch adds task_set_jobctl_pending() which should be used when
setting task->jobctl bits to schedule a stop or trap.  The function
performs the followings to ease setting pending bits.

* Sanity checks.

* If fatal signal is pending or PF_EXITING is set, no bit is set.

* STOP_SIGMASK is automatically cleared if new value is being set.

do_signal_stop() and ptrace_attach() are updated to use
task_set_jobctl_pending() instead of setting STOP_PENDING explicitly.
The surrounding structures around setting are changed to fit
task_set_jobctl_pending() better but there should be no userland
visible behavior difference.
Signed-off-by: NTejun Heo <tj@kernel.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: NOleg Nesterov <oleg@redhat.com>

JOBCTL_TRAPPING indicates that ptracer is waiting for tracee to
(re)transit into TRACED.  task_clear_jobctl_pending() must be called
when either tracee enters TRACED or the transition is cancelled for
some reason.  The former is achieved by explicitly calling
task_clear_jobctl_pending() in ptrace_stop() and the latter by calling
it at the end of do_signal_stop().

Calling task_clear_jobctl_trapping() at the end of do_signal_stop()
limits the scope TRAPPING can be used and is fragile in that seemingly
unrelated changes to tracee's control flow can lead to stuck TRAPPING.

We already have task_clear_jobctl_pending() calls on those cancelling
events to clear JOBCTL_STOP_PENDING.  Cancellations can be handled by
making those call sites use JOBCTL_PENDING_MASK instead and updating
task_clear_jobctl_pending() such that task_clear_jobctl_trapping() is
called automatically if no stop/trap is pending.

This patch makes the above changes and removes the fallback
task_clear_jobctl_trapping() call from do_signal_stop().
Signed-off-by: NTejun Heo <tj@kernel.org>
Signed-off-by: NOleg Nesterov <oleg@redhat.com>

This patch introduces JOBCTL_PENDING_MASK and replaces
task_clear_jobctl_stop_pending() with task_clear_jobctl_pending()
which takes an extra @mask argument.

JOBCTL_PENDING_MASK is currently equal to JOBCTL_STOP_PENDING but
future patches will add more bits.  recalc_sigpending_tsk() is updated
to use JOBCTL_PENDING_MASK instead.

task_clear_jobctl_pending() takes @mask which in subset of
JOBCTL_PENDING_MASK and clears the relevant jobctl bits.  If
JOBCTL_STOP_PENDING is set, other STOP bits are cleared together.  All
task_clear_jobctl_stop_pending() users are updated to call
task_clear_jobctl_pending() with JOBCTL_STOP_PENDING which is
functionally identical to task_clear_jobctl_stop_pending().

This patch doesn't cause any functional change.
Signed-off-by: NTejun Heo <tj@kernel.org>
Signed-off-by: NOleg Nesterov <oleg@redhat.com>

In ptrace_stop(), after arch hook is done, the task state and jobctl
bits are updated while holding siglock.  The ordering requirement
there is that TASK_TRACED is set before JOBCTL_TRAPPING is cleared to
prevent ptracer waiting on TRAPPING doesn't end up waking up TRACED is
actually set and sees TASK_RUNNING in wait(2).

Move set_current_state(TASK_TRACED) to the top of the block and
reorganize comments.  This makes the ordering more obvious
(TASK_TRACED before other updates) and helps future updates to group
stop participation.

This patch doesn't cause any functional change.
Signed-off-by: NTejun Heo <tj@kernel.org>
Signed-off-by: NOleg Nesterov <oleg@redhat.com>

signal->group_stop currently hosts mostly group stop related flags;
however, it's gonna be used for wider purposes and the GROUP_STOP_
flag prefix becomes confusing.  Rename signal->group_stop to
signal->jobctl and rename all GROUP_STOP_* flags to JOBCTL_*.

Bit position macros JOBCTL_*_BIT are defined and JOBCTL_* flags are
defined in terms of them to allow using bitops later.

While at it, reassign JOBCTL_TRAPPING to bit 22 to better accomodate
future additions.

This doesn't cause any functional change.

-v2: JOBCTL_*_BIT macros added as suggested by Linus.
Signed-off-by: NTejun Heo <tj@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: NOleg Nesterov <oleg@redhat.com>

ERESTART* is always wrong without TIF_SIGPENDING. Teach sys_pause()
to handle the spurious wakeup correctly.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>

GROUP_STOP_TRAPPING waiting mechanism piggybacks on
signal->wait_chldexit which is primarily used to implement waiting for
wait(2) and friends.  When do_wait() waits on signal->wait_chldexit,
it uses a custom wake up callback, child_wait_callback(), which
expects the child task which is waking up the parent to be passed in
as @key to filter out spurious wakeups.

task_clear_group_stop_trapping() used __wake_up_sync() which uses NULL
@key causing the following oops if the parent was doing do_wait().

  BUG: unable to handle kernel NULL pointer dereference at 00000000000002d8
  IP: [<ffffffff810499f9>] child_wait_callback+0x29/0x80
  PGD 1d899067 PUD 1e418067 PMD 0
  Oops: 0000 [#1] PREEMPT SMP
  last sysfs file: /sys/devices/pci0000:00/0000:00:03.0/local_cpus
  CPU 2
  Modules linked in:

  Pid: 4498, comm: test-continued Not tainted 2.6.39-rc6-work+ #32 Bochs Bochs
  RIP: 0010:[<ffffffff810499f9>]  [<ffffffff810499f9>] child_wait_callback+0x29/0x80
  RSP: 0000:ffff88001b889bf8  EFLAGS: 00010046
  RAX: 0000000000000000 RBX: ffff88001fab3af8 RCX: 0000000000000000
  RDX: 0000000000000001 RSI: 0000000000000002 RDI: ffff88001d91df20
  RBP: ffff88001b889c08 R08: 0000000000000000 R09: 0000000000000000
  R10: 0000000000000000 R11: 0000000000000001 R12: 0000000000000000
  R13: ffff88001fb70550 R14: 0000000000000000 R15: 0000000000000001
  FS:  00007f26ccae4700(0000) GS:ffff88001fd00000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
  CR2: 00000000000002d8 CR3: 000000001b8ac000 CR4: 00000000000006e0
  DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
  DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
  Process test-continued (pid: 4498, threadinfo ffff88001b888000, task ffff88001fb88000)
  Stack:
   ffff88001b889c18 ffff88001fb70538 ffff88001b889c58 ffffffff810312f9
   0000000000000001 0000000200000001 ffff88001b889c58 ffff88001fb70518
   0000000000000002 0000000000000082 0000000000000001 0000000000000000
  Call Trace:
   [<ffffffff810312f9>] __wake_up_common+0x59/0x90
   [<ffffffff81035263>] __wake_up_sync_key+0x53/0x80
   [<ffffffff810352a0>] __wake_up_sync+0x10/0x20
   [<ffffffff8105a984>] task_clear_jobctl_trapping+0x44/0x50
   [<ffffffff8105bcbc>] ptrace_stop+0x7c/0x290
   [<ffffffff8105c20a>] do_signal_stop+0x28a/0x2d0
   [<ffffffff8105d27f>] get_signal_to_deliver+0x14f/0x5a0
   [<ffffffff81002175>] do_signal+0x75/0x7b0
   [<ffffffff8100292d>] do_notify_resume+0x5d/0x70
   [<ffffffff8182e36a>] retint_signal+0x46/0x8c
  Code: 00 00 55 48 89 e5 53 48 83 ec 08 0f 1f 44 00 00 8b 47 d8 83 f8 03 74 3a 85 c0 49 89 c8 75 23 89 c0 48 8b 5f e0 4c 8d 0c 40 31 c0 <4b> 39 9c c8 d8 02 00 00 74 1d 48 83 c4 08 5b c9 c3 66 0f 1f 44

Fix it by using __wake_up_sync_key() and passing in the child as @key.

I still think it's a mistake to piggyback on wait_chldexit for this.
Given the relative low frequency of ptrace use, we would be much
better off leaving already complex wait_chldexit alone and using bit
waitqueue.
Signed-off-by: NTejun Heo <tj@kernel.org>
Reviewed-by: NOleg Nesterov <oleg@redhat.com>

sys_sigprocmask() changes current->blocked by hand. Convert this code
to use set_current_blocked().
Signed-off-by: NOleg Nesterov <oleg@redhat.com>

Cleanup. Remove the unneeded goto's, we can simply read blocked.sig[0]
unconditionally and then copy-to-user it if oset != NULL.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Acked-by: NTejun Heo <tj@kernel.org>
Reviewed-by: NMatt Fleming <matt.fleming@linux.intel.com>

As Tejun and Linus pointed out, "nand" is the wrong name for "x & ~y",
it should be "andn". Rename signandsets() as suggested.
Suggested-by: NTejun Heo <tj@kernel.org>
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Acked-by: NTejun Heo <tj@kernel.org>

do_sigtimedwait() changes current->blocked and thus it needs
set_current_blocked()->retarget_shared_pending().

We could use set_current_blocked() directly. It is fine to change
->real_blocked from all-zeroes to ->blocked and vice versa lockless,
but this is not immediately clear, looks racy, and needs a huge
comment to explain why this is correct.

To keep the things simple this patch adds the new static helper,
__set_task_blocked() which should be called with ->siglock held. This
way we can change both ->real_blocked and ->blocked atomically under
->siglock as the current code does. This is more understandable.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Acked-by: NTejun Heo <tj@kernel.org>
Reviewed-by: NMatt Fleming <matt.fleming@linux.intel.com>

Factor out the common code in sys_rt_sigtimedwait/compat_sys_rt_sigtimedwait
to the new helper, do_sigtimedwait().

Add the comment to document the extra tick we add to timespec_to_jiffies(ts),
thanks to Linus who explained this to me.

Perhaps it would be better to move compat_sys_rt_sigtimedwait() into
signal.c under CONFIG_COMPAT, then we can make do_sigtimedwait() static.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Acked-by: NTejun Heo <tj@kernel.org>
Reviewed-by: NMatt Fleming <matt.fleming@linux.intel.com>

No functional changes, cleanup compat_sys_rt_sigtimedwait() and
sys_rt_sigtimedwait().

Calculate the timeout before we take ->siglock, this simplifies and
lessens the code. Use timespec_valid() to check the timespec.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Acked-by: NTejun Heo <tj@kernel.org>
Reviewed-by: NMatt Fleming <matt.fleming@linux.intel.com>

sys_rt_sigprocmask() looks unnecessarily complicated, simplify it.
We can just read current->blocked lockless unconditionally before
anything else and then copy-to-user it if needed.  At worst we
copy 4 words on mips.

We could copy-to-user the old mask first and simplify the code even
more, but the patch tries to keep the current behaviour: we change
current->block even if copy_to_user(oset) fails.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Reviewed-by: NMatt Fleming <matt.fleming@linux.intel.com>
Acked-by: NTejun Heo <tj@kernel.org>

In short, almost every changing of current->blocked is wrong, or at least
can lead to the unexpected results.

For example. Two threads T1 and T2, T1 sleeps in sigtimedwait/pause/etc.
kill(tgid, SIG) can pick T2 for TIF_SIGPENDING. If T2 calls sigprocmask()
and blocks SIG before it notices the pending signal, nobody else can handle
this pending shared signal.

I am not sure this is bug, but at least this looks strange imho. T1 should
not sleep forever, there is a signal which should wake it up.

This patch moves the code which actually changes ->blocked into the new
helper, set_current_blocked() and changes this code to call
retarget_shared_pending() as exit_signals() does. We should only care about
the signals we just blocked, we use "newset & ~current->blocked" as a mask.

We do not check !sigisemptyset(newblocked), retarget_shared_pending() is
cheap unless mask & shared_pending.

Note: for this particular case we could simply change sigprocmask() to
return -EINTR if signal_pending(), but then we should change other callers
and, more importantly, if we need this fix then set_current_blocked() will
have more callers and some of them can't restart. See the next patch as a
random example.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Reviewed-by: NMatt Fleming <matt.fleming@linux.intel.com>
Acked-by: NTejun Heo <tj@kernel.org>

No functional changes, preparation to simplify the review of the next change.

1. We can read current->block lockless, nobody else can ever change this mask.

2. Calculate the resulting sigset_t outside of ->siglock into the temporary
   variable, then take ->siglock and change ->blocked.

Also, kill the stale comment about BKL.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Reviewed-by: NMatt Fleming <matt.fleming@linux.intel.com>
Acked-by: NTejun Heo <tj@kernel.org>

retarget_shared_pending() blindly does recalc_sigpending_and_wake() for
every sub-thread, this is suboptimal. We can check t->blocked and stop
looping once every bit in shared_pending has the new target.

Note: we do not take task_is_stopped_or_traced(t) into account, we are
not trying to speed up the signal delivery or to avoid the unnecessary
(but harmless) signal_wake_up(0) in this unlikely case.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Reviewed-by: NMatt Fleming <matt.fleming@linux.intel.com>
Acked-by: NTejun Heo <tj@kernel.org>

exit_signals() checks signal_pending() before retarget_shared_pending() but
this is suboptimal. We can avoid the while_each_thread() loop in case when
there are no shared signals visible to us.

Add the "shared_pending.signal & ~blocked" check. We don't use tsk->blocked
directly but pass ~blocked as an argument, this is needed for the next patch.

Note: we can optimize this more. while_each_thread(t) can check t->blocked
into account and stop after every pending signal has the new target, see the
next patch.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Reviewed-by: NMatt Fleming <matt.fleming@linux.intel.com>
Acked-by: NTejun Heo <tj@kernel.org>

No functional changes. Move the notify-other-threads code from exit_signals()
to the new helper, retarget_shared_pending().
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Reviewed-by: NMatt Fleming <matt.fleming@linux.intel.com>
Acked-by: NTejun Heo <tj@kernel.org>

Fix erroneous syscall kernel-doc comments in kernel/signal.c.
Reported-by: NMatt Fleming <matt@console-pimps.org>
Signed-off-by: NRandy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Add kernel-doc to syscalls in signal.c.
Signed-off-by: NRandy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

General coding style and comment fixes; no code changes:

 - Use multi-line-comment coding style.
 - Put some function signatures completely on one line.
 - Hyphenate some words.
 - Spell Posix as POSIX.
 - Correct typos & spellos in some comments.
 - Drop trailing whitespace.
 - End sentences with periods.
Signed-off-by: NRandy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This patch moves SIGNAL_STOP_DEQUEUED from signal_struct->flags to
task_struct->group_stop, and thus makes it per-thread.

Like SIGNAL_STOP_DEQUEUED, GROUP_STOP_DEQUEUED can be false-positive
after return from get_signal_to_deliver(), this is fine. The only
purpose of this bit is: we can drop ->siglock after __dequeue_signal()
returns the sig_kernel_stop() signal and before we call
do_signal_stop(), in this case we must not miss SIGCONT if it comes in
between.

But, unlike SIGNAL_STOP_DEQUEUED, GROUP_STOP_DEQUEUED can not be
false-positive in do_signal_stop() if multiple threads dequeue the
sig_kernel_stop() signal at the same time.

Consider two threads T1 and T2, SIGTTIN has a hanlder.

	- T1 dequeues SIGTSTP and sets SIGNAL_STOP_DEQUEUED, then
	  it drops ->siglock

	- SIGCONT comes and clears SIGNAL_STOP_DEQUEUED, SIGTSTP
	  should be cancelled.

	- T2 dequeues SIGTTIN and sets SIGNAL_STOP_DEQUEUED again.
	  Since we have a handler we should not stop, T2 returns
	  to usermode to run the handler.

	- T1 continues, calls do_signal_stop() and wrongly starts
	  the group stop because SIGNAL_STOP_DEQUEUED was restored
	  in between.

With or without this change:

	- we need to do something with ptrace_signal() which can
	  return SIGSTOP, but this needs another discussion

	- SIGSTOP can be lost if it races with the mt exec, will
	  be fixed later.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Signed-off-by: NTejun Heo <tj@kernel.org>

PF_EXITING or TASK_STOPPED has already called task_participate_group_stop()
and cleared its ->group_stop. No need to do task_clear_group_stop_pending()
when we start the new group stop.

Add a small comment to explain the !task_is_stopped() check. Note that this
check is not exactly right and it can lead to unnecessary stop later if the
thread is TASK_PTRACED. What we need is task_participated_in_group_stop(),
this will be solved later.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Signed-off-by: NTejun Heo <tj@kernel.org>

prepare_signal(SIGCONT) should never set TIF_SIGPENDING or wake up
the TASK_INTERRUPTIBLE threads. We are going to call complete_signal()
which should pick the right thread correctly. All we need is to wake
up the TASK_STOPPED threads.

If the task was stopped, it can't return to usermode without taking
->siglock. Otherwise we don't care, and the spurious TIF_SIGPENDING
can't be useful.

The comment says:

	* If there is a handler for SIGCONT, we must make
	* sure that no thread returns to user mode before
	* we post the signal

It is not clear what this means. Probably, "when there's only a single
thread" and this continues to be true. Otherwise, even if this SIGCONT
is not private, with or without this change only one thread can dequeue
SIGCONT, other threads can happily return to user mode before before
that thread handles this signal.

Note also that wake_up_state(t, __TASK_STOPPED) can't race with the task
which changes its state, TASK_STOPPED state is protected by ->siglock as
well.

In short: when it comes to signal delivery, SIGCONT is the normal signal
and does not need any special support.
Signed-off-by: NOleg Nesterov <oleg@redhat.com>
Signed-off-by: NTejun Heo <tj@kernel.org>

Fixes generated by 'codespell' and manually reviewed.
Signed-off-by: NLucas De Marchi <lucas.demarchi@profusion.mobi>