diff mbox series

[6/7] Block all signals on timer_create thread (BZ#10815)

Message ID 20191210183221.26912-6-adhemerval.zanella@linaro.org
State New
Headers show
Series [v3,1/7] Fix __libc_signal_block_all on sparc64 | expand

Commit Message

Adhemerval Zanella Dec. 10, 2019, 6:32 p.m. UTC
The behavior of the signal mask on threads created by timer_create
for SIGEV_THREAD timers are implementation defined and glibc explicit
unblocks all signals before calling the user-defined function.

This behavior, although not incorrect standard-wise, opens a race if a
program using a blocked rt-signal plus sigwaitinfo (and without an
installed signal handler for the rt-signal) receives the signal while
executing the timer threads created by SIGEV_THREAD.

A better alternative discussed in bug report is to rather block all
signals (besides the internal ones not available to application
usage).

This patch fixes by only unblock the SIGSETXID (used on set*uid).
The SIGTIMER is the same as SIGCANCEL and it will be handled by
sigwaitinfo on the helper thread (thus it can be masked as well).

Checked on x86_64-linux-gnu and i686-linux-gnu.
---
 rt/Makefile                                   |  7 +-
 rt/tst-timer-sigmask.c                        | 96 +++++++++++++++++++
 sysdeps/unix/sysv/linux/internal-signals.h    | 14 +++
 sysdeps/unix/sysv/linux/kernel-posix-timers.h |  1 -
 sysdeps/unix/sysv/linux/timer_routines.c      | 96 +++++++------------
 5 files changed, 148 insertions(+), 66 deletions(-)
 create mode 100644 rt/tst-timer-sigmask.c

-- 
2.17.1

Comments

Florian Weimer Dec. 12, 2019, 1:10 p.m. UTC | #1
* Adhemerval Zanella:

> The behavior of the signal mask on threads created by timer_create

> for SIGEV_THREAD timers are implementation defined and glibc explicit

                              implementation-defined           explictlly

> unblocks all signals before calling the user-defined function.

>

> This behavior, although not incorrect standard-wise, opens a race if a

> program using a blocked rt-signal plus sigwaitinfo (and without an

> installed signal handler for the rt-signal) receives the signal while

> executing the timer threads created by SIGEV_THREAD.


while executing the user-defined function for SIGEV_THREAD?

> A better alternative discussed in bug report is to rather block all

> signals (besides the internal ones not available to application

> usage).

>

> This patch fixes by only unblock the SIGSETXID (used on set*uid).

             fixes this issiue by only unblicking SIGSETXID

> The SIGTIMER is the same as SIGCANCEL and it will be handled by


No “The”?

> sigwaitinfo on the helper thread (thus it can be masked as well).

>

> Checked on x86_64-linux-gnu and i686-linux-gnu.


Please also try to build this on Hurd.

> diff --git a/rt/tst-timer-sigmask.c b/rt/tst-timer-sigmask.c

> new file mode 100644

> index 0000000000..1030e4c79f

> --- /dev/null

> +++ b/rt/tst-timer-sigmask.c

> @@ -0,0 +1,96 @@


> +static void

> +thread_handler (union sigval sv)

> +{

> +  bool failure = false;

> +

> +  sigset_t ss;

> +  sigprocmask (SIG_BLOCK, NULL, &ss);

> +  if (test_verbose > 0)

> +    printf ("%s: blocked signal mask = { ", __func__);

> +  for (int sig = 1; sig < NSIG; sig++)

> +    {

> +#ifdef __linux__

> +      /* POSIX timers threads created to handle SIGEV_THREAD blocks

> +	 all signals except SIGKILL, SIGSTOP, and SIGSETXID.  */

> +      if (!sigismember (&ss, sig)

> +	  && (sig != SIGSETXID && sig != SIGKILL && sig != SIGSTOP))

> +	{

> +	  failure = true;

> +	}

> +#endif

> +      if (test_verbose && sigismember (&ss, sig))

> +	printf ("%d, ", sig);

> +    }

> +  if (test_verbose > 0)

> +    printf ("}\n");

> +

> +  pthread_mutex_lock (&lock);

> +  thread_handler_check = failure

> +			 ? THREAD_SIGNAL_CHECK_FAIL : THREAD_SIGNAL_CHECK_OK;

> +  pthread_cond_signal (&cond);

> +  pthread_mutex_unlock (&lock);


Should you use the x variants?  xpthread_sigmask, xpthread_mutex_lock
etc.?

The synchronization might more easily expressed using a POSIX barrier.

> diff --git a/sysdeps/unix/sysv/linux/kernel-posix-timers.h b/sysdeps/unix/sysv/linux/kernel-posix-timers.h

> index 1ded4df51a..d60cb95f80 100644

> --- a/sysdeps/unix/sysv/linux/kernel-posix-timers.h

> +++ b/sysdeps/unix/sysv/linux/kernel-posix-timers.h

> @@ -43,7 +43,6 @@ extern pthread_mutex_t __active_timer_sigev_thread_lock attribute_hidden;

>  /* Type of timers in the kernel.  */

>  typedef int kernel_timer_t;

>  

> -

>  /* Internal representation of timer.  */

>  struct timer

>  {


Unrelated change?

> +		  /* This is the signal we are waiting for.  */

> +		  td->thrfunc = tk->thrfunc;

> +		  td->sival = tk->sival;

>  

> +		  pthread_t th;

> +		  pthread_create (&th, &tk->attr, timer_sigev_thread, td);

> +		}


I think this does not unblock SIGCANCEL on the new thread with the
current pthread_create implementation.  This will change with the
block-signals-around-clone patch I submitted, though.

Thanks,
Florian
Adhemerval Zanella Dec. 12, 2019, 2:53 p.m. UTC | #2
On 12/12/2019 10:10, Florian Weimer wrote:
> * Adhemerval Zanella:

> 

>> The behavior of the signal mask on threads created by timer_create

>> for SIGEV_THREAD timers are implementation defined and glibc explicit

>                               implementation-defined           explictlly


Ack.

> 

>> unblocks all signals before calling the user-defined function.

>>

>> This behavior, although not incorrect standard-wise, opens a race if a

>> program using a blocked rt-signal plus sigwaitinfo (and without an

>> installed signal handler for the rt-signal) receives the signal while

>> executing the timer threads created by SIGEV_THREAD.

> 

> while executing the user-defined function for SIGEV_THREAD?


Ack.

> 

>> A better alternative discussed in bug report is to rather block all

>> signals (besides the internal ones not available to application

>> usage).

>>

>> This patch fixes by only unblock the SIGSETXID (used on set*uid).

>              fixes this issiue by only unblicking SIGSETXID

> 

>> The SIGTIMER is the same as SIGCANCEL and it will be handled by

> 

> No “The”?


Ack.

> 

>> sigwaitinfo on the helper thread (thus it can be masked as well).

>>

>> Checked on x86_64-linux-gnu and i686-linux-gnu.

> 

> Please also try to build this on Hurd.


I usually do, both build and rt/tests builds fine on i686-gnu.

> 

>> diff --git a/rt/tst-timer-sigmask.c b/rt/tst-timer-sigmask.c

>> new file mode 100644

>> index 0000000000..1030e4c79f

>> --- /dev/null

>> +++ b/rt/tst-timer-sigmask.c

>> @@ -0,0 +1,96 @@

> 

>> +static void

>> +thread_handler (union sigval sv)

>> +{

>> +  bool failure = false;

>> +

>> +  sigset_t ss;

>> +  sigprocmask (SIG_BLOCK, NULL, &ss);

>> +  if (test_verbose > 0)

>> +    printf ("%s: blocked signal mask = { ", __func__);

>> +  for (int sig = 1; sig < NSIG; sig++)

>> +    {

>> +#ifdef __linux__

>> +      /* POSIX timers threads created to handle SIGEV_THREAD blocks

>> +	 all signals except SIGKILL, SIGSTOP, and SIGSETXID.  */

>> +      if (!sigismember (&ss, sig)

>> +	  && (sig != SIGSETXID && sig != SIGKILL && sig != SIGSTOP))

>> +	{

>> +	  failure = true;

>> +	}

>> +#endif

>> +      if (test_verbose && sigismember (&ss, sig))

>> +	printf ("%d, ", sig);

>> +    }

>> +  if (test_verbose > 0)

>> +    printf ("}\n");

>> +

>> +  pthread_mutex_lock (&lock);

>> +  thread_handler_check = failure

>> +			 ? THREAD_SIGNAL_CHECK_FAIL : THREAD_SIGNAL_CHECK_OK;

>> +  pthread_cond_signal (&cond);

>> +  pthread_mutex_unlock (&lock);

> 

> Should you use the x variants?  xpthread_sigmask, xpthread_mutex_lock

> etc.?

> 

> The synchronization might more easily expressed using a POSIX barrier.


Ack, I changed to use xpthread_barrier_*.

> 

>> diff --git a/sysdeps/unix/sysv/linux/kernel-posix-timers.h b/sysdeps/unix/sysv/linux/kernel-posix-timers.h

>> index 1ded4df51a..d60cb95f80 100644

>> --- a/sysdeps/unix/sysv/linux/kernel-posix-timers.h

>> +++ b/sysdeps/unix/sysv/linux/kernel-posix-timers.h

>> @@ -43,7 +43,6 @@ extern pthread_mutex_t __active_timer_sigev_thread_lock attribute_hidden;

>>  /* Type of timers in the kernel.  */

>>  typedef int kernel_timer_t;

>>  

>> -

>>  /* Internal representation of timer.  */

>>  struct timer

>>  {

> 

> Unrelated change?


Ack.

> 

>> +		  /* This is the signal we are waiting for.  */

>> +		  td->thrfunc = tk->thrfunc;

>> +		  td->sival = tk->sival;

>>  

>> +		  pthread_t th;

>> +		  pthread_create (&th, &tk->attr, timer_sigev_thread, td);

>> +		}

> 

> I think this does not unblock SIGCANCEL on the new thread with the

> current pthread_create implementation.  This will change with the

> block-signals-around-clone patch I submitted, though.


It does not indeed, although not sure how valid is the scenario to
pthread cancel a thread not created explicitly by a pthread_create
issued by the user (since POSIX also state it is not valid to call
pthread_join() timer thread).

The sigcancel is explicit blocked with __libc_signal_block_sigtimer
on __start_helper_thread, so sigwaitinfo would be the only point
where the timer is triggered.  I don't think it is correct to unblock
it before calling pthread_create, so the created thread inherit the
unblock SIGCANCEL, for the same reason the SIGCANCEL is blocked on the
helper thread that calls sigwaitinfo (possible lose wakeups).

Another possibility is decouple SIGTIMER from SIGCANCEL, so the
helper sigwaitinfo thread is created with SIGTIMER blocked, but
not SIGCANCEL.
Florian Weimer Dec. 12, 2019, 3:14 p.m. UTC | #3
* Adhemerval Zanella:

>> I think this does not unblock SIGCANCEL on the new thread with the

>> current pthread_create implementation.  This will change with the

>> block-signals-around-clone patch I submitted, though.

>

> It does not indeed, although not sure how valid is the scenario to

> pthread cancel a thread not created explicitly by a pthread_create

> issued by the user (since POSIX also state it is not valid to call

> pthread_join() timer thread).


I don't see any restrictions on calling pthread_cancel or pthread_exit.

> The sigcancel is explicit blocked with __libc_signal_block_sigtimer

> on __start_helper_thread, so sigwaitinfo would be the only point

> where the timer is triggered.  I don't think it is correct to unblock

> it before calling pthread_create, so the created thread inherit the

> unblock SIGCANCEL, for the same reason the SIGCANCEL is blocked on the

> helper thread that calls sigwaitinfo (possible lose wakeups).


We can wrap the thread start function with another function that
unblocks SIGCANCEL.  Or note the semantic difference on the other change
(which I assume we still want for other reasons).

Thanks,
Florian
Adhemerval Zanella Dec. 12, 2019, 4:01 p.m. UTC | #4
On 12/12/2019 12:14, Florian Weimer wrote:
> * Adhemerval Zanella:

> 

>>> I think this does not unblock SIGCANCEL on the new thread with the

>>> current pthread_create implementation.  This will change with the

>>> block-signals-around-clone patch I submitted, though.

>>

>> It does not indeed, although not sure how valid is the scenario to

>> pthread cancel a thread not created explicitly by a pthread_create

>> issued by the user (since POSIX also state it is not valid to call

>> pthread_join() timer thread).

> 

> I don't see any restrictions on calling pthread_cancel or pthread_exit.

> 

>> The sigcancel is explicit blocked with __libc_signal_block_sigtimer

>> on __start_helper_thread, so sigwaitinfo would be the only point

>> where the timer is triggered.  I don't think it is correct to unblock

>> it before calling pthread_create, so the created thread inherit the

>> unblock SIGCANCEL, for the same reason the SIGCANCEL is blocked on the

>> helper thread that calls sigwaitinfo (possible lose wakeups).

> 

> We can wrap the thread start function with another function that

> unblocks SIGCANCEL.  Or note the semantic difference on the other change

> (which I assume we still want for other reasons).


We can add such call ontimer_sigev_thread, but it does not really
help the 'rt' side of the function with another syscall. Do you
consider this a block? Should we reinstate the sigprocmask to
unblock SIGCANCEL on timer_sigev_thread?
diff mbox series

Patch

diff --git a/rt/Makefile b/rt/Makefile
index 6c8365e0c0..a7a82879c7 100644
--- a/rt/Makefile
+++ b/rt/Makefile
@@ -47,6 +47,7 @@  tests := tst-shm tst-timer tst-timer2 \
 	 tst-timer3 tst-timer4 tst-timer5 \
 	 tst-cpuclock2 tst-cputimer1 tst-cputimer2 tst-cputimer3 \
 	 tst-shm-cancel
+tests-internal := tst-timer-sigmask
 
 extra-libs := librt
 extra-libs-others := $(extra-libs)
@@ -63,9 +64,11 @@  LDFLAGS-rt.so = -Wl,--enable-new-dtags,-z,nodelete
 $(objpfx)librt.so: $(shared-thread-library)
 
 ifeq (yes,$(build-shared))
-$(addprefix $(objpfx),$(tests)): $(objpfx)librt.so $(shared-thread-library)
+$(addprefix $(objpfx),$(tests) $(tests-internal)): \
+	$(objpfx)librt.so $(shared-thread-library)
 else
-$(addprefix $(objpfx),$(tests)): $(objpfx)librt.a $(static-thread-library)
+$(addprefix $(objpfx),$(tests)) $(tests-internal): \
+	$(objpfx)librt.a $(static-thread-library)
 endif
 
 tst-mqueue7-ARGS = -- $(host-test-program-cmd)
diff --git a/rt/tst-timer-sigmask.c b/rt/tst-timer-sigmask.c
new file mode 100644
index 0000000000..1030e4c79f
--- /dev/null
+++ b/rt/tst-timer-sigmask.c
@@ -0,0 +1,96 @@ 
+/* Check resulting signal mask from POSIX time using SIGEV_THREAD.
+   Copyright (C) 2019 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <https://www.gnu.org/licenses/>.  */
+
+#include <stdio.h>
+#include <time.h>
+#include <signal.h>
+#include <pthread.h>
+#include <stdbool.h>
+
+#include <support/check.h>
+#include <support/test-driver.h>
+
+#include <internal-signals.h>
+
+static pthread_mutex_t lock = PTHREAD_MUTEX_INITIALIZER;
+static pthread_cond_t cond = PTHREAD_COND_INITIALIZER;
+static enum
+{
+  THREAD_SIGNAL_CHECK_INIT = 0,
+  THREAD_SIGNAL_CHECK_OK,
+  THREAD_SIGNAL_CHECK_FAIL
+} thread_handler_check = THREAD_SIGNAL_CHECK_INIT;
+
+static void
+thread_handler (union sigval sv)
+{
+  bool failure = false;
+
+  sigset_t ss;
+  sigprocmask (SIG_BLOCK, NULL, &ss);
+  if (test_verbose > 0)
+    printf ("%s: blocked signal mask = { ", __func__);
+  for (int sig = 1; sig < NSIG; sig++)
+    {
+#ifdef __linux__
+      /* POSIX timers threads created to handle SIGEV_THREAD blocks
+	 all signals except SIGKILL, SIGSTOP, and SIGSETXID.  */
+      if (!sigismember (&ss, sig)
+	  && (sig != SIGSETXID && sig != SIGKILL && sig != SIGSTOP))
+	{
+	  failure = true;
+	}
+#endif
+      if (test_verbose && sigismember (&ss, sig))
+	printf ("%d, ", sig);
+    }
+  if (test_verbose > 0)
+    printf ("}\n");
+
+  pthread_mutex_lock (&lock);
+  thread_handler_check = failure
+			 ? THREAD_SIGNAL_CHECK_FAIL : THREAD_SIGNAL_CHECK_OK;
+  pthread_cond_signal (&cond);
+  pthread_mutex_unlock (&lock);
+}
+
+static int
+do_test (void)
+{
+  struct sigevent sev = { 0 };
+  sev.sigev_notify = SIGEV_THREAD;
+  sev.sigev_notify_function = &thread_handler;
+
+  timer_t timerid;
+  TEST_COMPARE (timer_create (CLOCK_REALTIME, &sev, &timerid), 0);
+
+  struct itimerspec trigger = { 0 };
+  trigger.it_value.tv_nsec = 1000000;
+  TEST_COMPARE (timer_settime (timerid, 0, &trigger, NULL), 0);
+
+  pthread_mutex_lock (&lock);
+  while (thread_handler_check == THREAD_SIGNAL_CHECK_INIT)
+    pthread_cond_wait (&cond, &lock);
+  pthread_mutex_unlock (&lock);
+
+  TEST_COMPARE (thread_handler_check, THREAD_SIGNAL_CHECK_OK);
+
+  return 0;
+}
+
+#include <support/test-driver.c>
diff --git a/sysdeps/unix/sysv/linux/internal-signals.h b/sysdeps/unix/sysv/linux/internal-signals.h
index 04e1ec4f0a..f155a8cd32 100644
--- a/sysdeps/unix/sysv/linux/internal-signals.h
+++ b/sysdeps/unix/sysv/linux/internal-signals.h
@@ -70,6 +70,11 @@  static const sigset_t sigapp_set = {
 #endif
 };
 
+static const sigset_t sigtimer_set = {
+  .__val = { [0]                      = __sigmask (SIGTIMER),
+	     [1 ... _SIGSET_NWORDS-1] = 0 }
+};
+
 /* Block all signals, including internal glibc ones.  */
 static inline void
 __libc_signal_block_all (sigset_t *set)
@@ -88,6 +93,15 @@  __libc_signal_block_app (sigset_t *set)
 			 _NSIG / 8);
 }
 
+/* Block only the SIGTIMER set.  */
+static inline void
+__libc_signal_block_sigtimer (sigset_t *set)
+{
+  INTERNAL_SYSCALL_DECL (err);
+  INTERNAL_SYSCALL_CALL (rt_sigprocmask, err, SIG_BLOCK, &sigtimer_set, set,
+			 _NSIG / 8);
+}
+
 /* Restore current process signal mask.  */
 static inline void
 __libc_signal_restore_set (const sigset_t *set)
diff --git a/sysdeps/unix/sysv/linux/kernel-posix-timers.h b/sysdeps/unix/sysv/linux/kernel-posix-timers.h
index 1ded4df51a..d60cb95f80 100644
--- a/sysdeps/unix/sysv/linux/kernel-posix-timers.h
+++ b/sysdeps/unix/sysv/linux/kernel-posix-timers.h
@@ -43,7 +43,6 @@  extern pthread_mutex_t __active_timer_sigev_thread_lock attribute_hidden;
 /* Type of timers in the kernel.  */
 typedef int kernel_timer_t;
 
-
 /* Internal representation of timer.  */
 struct timer
 {
diff --git a/sysdeps/unix/sysv/linux/timer_routines.c b/sysdeps/unix/sysv/linux/timer_routines.c
index d96d963177..98bbe077fa 100644
--- a/sysdeps/unix/sysv/linux/timer_routines.c
+++ b/sysdeps/unix/sysv/linux/timer_routines.c
@@ -42,14 +42,6 @@  struct thread_start_data
 static void *
 timer_sigev_thread (void *arg)
 {
-  /* The parent thread has all signals blocked.  This is a bit
-     surprising for user code, although valid.  We unblock all
-     signals.  */
-  sigset_t ss;
-  sigemptyset (&ss);
-  INTERNAL_SYSCALL_DECL (err);
-  INTERNAL_SYSCALL (rt_sigprocmask, err, 4, SIG_SETMASK, &ss, NULL, _NSIG / 8);
-
   struct thread_start_data *td = (struct thread_start_data *) arg;
 
   void (*thrfunc) (sigval_t) = td->thrfunc;
@@ -69,65 +61,49 @@  timer_sigev_thread (void *arg)
 static void *
 timer_helper_thread (void *arg)
 {
-  /* Wait for the SIGTIMER signal, allowing the setXid signal, and
-     none else.  */
-  sigset_t ss;
-  sigemptyset (&ss);
-  __sigaddset (&ss, SIGTIMER);
-
   /* Endless loop of waiting for signals.  The loop is only ended when
      the thread is canceled.  */
   while (1)
     {
       siginfo_t si;
 
-      /* sigwaitinfo cannot be used here, since it deletes
-	 SIGCANCEL == SIGTIMER from the set.  */
-
-      /* XXX The size argument hopefully will have to be changed to the
-	 real size of the user-level sigset_t.  */
-      int result = SYSCALL_CANCEL (rt_sigtimedwait, &ss, &si, NULL, _NSIG / 8);
-
-      if (result > 0)
+      while (sigwaitinfo (&sigtimer_set, &si) < 0);
+      if (si.si_code == SI_TIMER)
 	{
-	  if (si.si_code == SI_TIMER)
-	    {
-	      struct timer *tk = (struct timer *) si.si_ptr;
+	  struct timer *tk = (struct timer *) si.si_ptr;
+
+	  /* Check the timer is still used and will not go away
+	     while we are reading the values here.  */
+	  pthread_mutex_lock (&__active_timer_sigev_thread_lock);
 
-	      /* Check the timer is still used and will not go away
-		 while we are reading the values here.  */
-	      pthread_mutex_lock (&__active_timer_sigev_thread_lock);
+	  struct timer *runp = __active_timer_sigev_thread;
+	  while (runp != NULL)
+	    if (runp == tk)
+	      break;
+	  else
+	    runp = runp->next;
 
-	      struct timer *runp = __active_timer_sigev_thread;
-	      while (runp != NULL)
-		if (runp == tk)
-		  break;
-		else
-		  runp = runp->next;
+	  if (runp != NULL)
+	    {
+	      struct thread_start_data *td = malloc (sizeof (*td));
 
-	      if (runp != NULL)
+	      /* There is not much we can do if the allocation fails.  */
+	      if (td != NULL)
 		{
-		  struct thread_start_data *td = malloc (sizeof (*td));
-
-		  /* There is not much we can do if the allocation fails.  */
-		  if (td != NULL)
-		    {
-		      /* This is the signal we are waiting for.  */
-		      td->thrfunc = tk->thrfunc;
-		      td->sival = tk->sival;
-
-		      pthread_t th;
-		      (void) pthread_create (&th, &tk->attr,
-					     timer_sigev_thread, td);
-		    }
-		}
+		  /* This is the signal we are waiting for.  */
+		  td->thrfunc = tk->thrfunc;
+		  td->sival = tk->sival;
 
-	      pthread_mutex_unlock (&__active_timer_sigev_thread_lock);
+		  pthread_t th;
+		  pthread_create (&th, &tk->attr, timer_sigev_thread, td);
+		}
 	    }
-	  else if (si.si_code == SI_TKILL)
-	    /* The thread is canceled.  */
-	    pthread_exit (NULL);
+
+	  pthread_mutex_unlock (&__active_timer_sigev_thread_lock);
 	}
+      else if (si.si_code == SI_TKILL)
+	/* The thread is canceled.  */
+	pthread_exit (NULL);
     }
 }
 
@@ -161,15 +137,10 @@  __start_helper_thread (void)
 
   /* Block all signals in the helper thread but SIGSETXID.  To do this
      thoroughly we temporarily have to block all signals here.  The
-     helper can lose wakeups if SIGCANCEL is not blocked throughout,
-     but sigfillset omits it SIGSETXID.  So, we add SIGCANCEL back
-     explicitly here.  */
+     helper can lose wakeups if SIGCANCEL is not blocked throughout.  */
   sigset_t ss;
-  sigset_t oss;
-  sigfillset (&ss);
-  __sigaddset (&ss, SIGCANCEL);
-  INTERNAL_SYSCALL_DECL (err);
-  INTERNAL_SYSCALL (rt_sigprocmask, err, 4, SIG_SETMASK, &ss, &oss, _NSIG / 8);
+  __libc_signal_block_app (&ss);
+  __libc_signal_block_sigtimer (NULL);
 
   /* Create the helper thread for this timer.  */
   pthread_t th;
@@ -179,8 +150,7 @@  __start_helper_thread (void)
     __helper_tid = ((struct pthread *) th)->tid;
 
   /* Restore the signal mask.  */
-  INTERNAL_SYSCALL (rt_sigprocmask, err, 4, SIG_SETMASK, &oss, NULL,
-		    _NSIG / 8);
+  __libc_signal_restore_set (&ss);
 
   /* No need for the attribute anymore.  */
   (void) pthread_attr_destroy (&attr);