diff mbox series

nptl: Fix __libc_cleanup_pop_restore asynchronous restore (BZ#29214)

Message ID 20220531202810.3864051-1-adhemerval.zanella@linaro.org
State New
Headers show
Series nptl: Fix __libc_cleanup_pop_restore asynchronous restore (BZ#29214) | expand

Commit Message

Adhemerval Zanella Netto May 31, 2022, 8:28 p.m. UTC
This was due a wrong revert done on 404656009b459658.

Checked on x86_64-linux-gnu.
---
 nptl/libc-cleanup.c            |  3 +-
 sysdeps/pthread/Makefile       |  1 +
 sysdeps/pthread/tst-cancel30.c | 75 ++++++++++++++++++++++++++++++++++
 3 files changed, 78 insertions(+), 1 deletion(-)
 create mode 100644 sysdeps/pthread/tst-cancel30.c

Comments

Andreas Schwab June 7, 2022, 10:27 a.m. UTC | #1
On Mai 31 2022, Adhemerval Zanella via Libc-alpha wrote:

> diff --git a/sysdeps/pthread/tst-cancel30.c b/sysdeps/pthread/tst-cancel30.c
> new file mode 100644
> index 0000000000..ef3f187e71
> --- /dev/null
> +++ b/sysdeps/pthread/tst-cancel30.c
> @@ -0,0 +1,75 @@
> +/* Check if printf like functions does not disable asynchronous cancellation
> +   mode (BZ#29214).
> +
> +   Copyright (C) 2022 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 <support/check.h>
> +#include <support/xthread.h>
> +#include <support/xstdio.h>
> +
> +static pthread_barrier_t b;
> +
> +static void *
> +tf (void *arg)
> +{
> +  int old;
> +
> +  TEST_COMPARE (pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, NULL), 0);
> +  
> +  TEST_COMPARE (pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, &old), 0);
> +  TEST_COMPARE (old, PTHREAD_CANCEL_ASYNCHRONOUS);
> +
> +  /* Check if internal lock cleanup routines restore the cancellation type
> +     correctly.  */
> +  printf ("...\n");
> +  TEST_COMPARE (pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, &old), 0);
> +  TEST_COMPARE (old, PTHREAD_CANCEL_ASYNCHRONOUS);
> +
> +  xpthread_barrier_wait (&b);
> +
> +  /* Wait indefinitely for cancellation, which only works if asynchronous
> +     cancellation is enabled.  */
> +#ifdef SYS_futex
> +  syscall (SYS_futex, &(int){0}, FUTEX_WAIT_PRIVATE, 0, NULL, NULL, 0);
> +#else
> +  for (;;);
> +#endif

Should that handle SYS_futex_time64?
Florian Weimer June 7, 2022, 11:20 a.m. UTC | #2
* Andreas Schwab via Libc-alpha:

>> +  /* Wait indefinitely for cancellation, which only works if asynchronous
>> +     cancellation is enabled.  */
>> +#ifdef SYS_futex
>> +  syscall (SYS_futex, &(int){0}, FUTEX_WAIT_PRIVATE, 0, NULL, NULL, 0);
>> +#else
>> +  for (;;);
>> +#endif
>
> Should that handle SYS_futex_time64?

syscall (SYS_pause) should be architecture-agnostic, and like any
syscall-invoked system call, it's not a cancellation point.

Thanks,
Florian
Adhemerval Zanella Netto June 7, 2022, 12:47 p.m. UTC | #3
On 07/06/2022 08:20, Florian Weimer via Libc-alpha wrote:
> * Andreas Schwab via Libc-alpha:
> 
>>> +  /* Wait indefinitely for cancellation, which only works if asynchronous
>>> +     cancellation is enabled.  */
>>> +#ifdef SYS_futex
>>> +  syscall (SYS_futex, &(int){0}, FUTEX_WAIT_PRIVATE, 0, NULL, NULL, 0);
>>> +#else
>>> +  for (;;);
>>> +#endif
>>
>> Should that handle SYS_futex_time64?

Indeed, riscv32 for instance is using the loop instead.

> 
> syscall (SYS_pause) should be architecture-agnostic, and like any
> syscall-invoked system call, it's not a cancellation point.

It is not in unfortunately:

$ python3 ./sysdeps/unix/sysv/linux/glibcsyscalls.py query-syscall pause
pause:
  defined: arm hppa i386 m68k microblaze mips/mips32 mips/mips64/n32 mips/mips64/n64 powerpc/powerpc32 powerpc/powerpc64 s390/s390-32 s390/s390-64 sh sparc/s
parc32 sparc/sparc64 x86_64/64 x86_64/x32
  undefined: aarch64 alpha arc csky ia64 nios2 or1k riscv/rv32 riscv/rv64

So we will need to use ppoll and also ppoll_time64.  I think it would
be better to add a libspport to abstract and use glibc internal definition
that handle the ppoll/poll_time64 (since there is no need to handle the
timeout here).
diff mbox series

Patch

diff --git a/nptl/libc-cleanup.c b/nptl/libc-cleanup.c
index c4a83591bf..2ce59388d4 100644
--- a/nptl/libc-cleanup.c
+++ b/nptl/libc-cleanup.c
@@ -57,7 +57,8 @@  __libc_cleanup_pop_restore (struct _pthread_cleanup_buffer *buffer)
   THREAD_SETMEM (self, cleanup, buffer->__prev);
 
   int cancelhandling = atomic_load_relaxed (&self->cancelhandling);
-  if (cancelhandling & CANCELTYPE_BITMASK)
+  if (buffer->__canceltype != PTHREAD_CANCEL_DEFERRED
+      && (cancelhandling & CANCELTYPE_BITMASK) == 0)
     {
       int newval;
       do
diff --git a/sysdeps/pthread/Makefile b/sysdeps/pthread/Makefile
index 8cebe7a784..7d1670da87 100644
--- a/sysdeps/pthread/Makefile
+++ b/sysdeps/pthread/Makefile
@@ -126,6 +126,7 @@  tests += tst-cnd-basic tst-mtx-trylock tst-cnd-broadcast \
 	 tst-pthread-raise-blocked-self \
 	 tst-pthread_kill-exited \
 	 tst-pthread_kill-exiting \
+	 tst-cancel30 \
 	 # tests
 
 tests-time64 += \
diff --git a/sysdeps/pthread/tst-cancel30.c b/sysdeps/pthread/tst-cancel30.c
new file mode 100644
index 0000000000..ef3f187e71
--- /dev/null
+++ b/sysdeps/pthread/tst-cancel30.c
@@ -0,0 +1,75 @@ 
+/* Check if printf like functions does not disable asynchronous cancellation
+   mode (BZ#29214).
+
+   Copyright (C) 2022 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 <support/check.h>
+#include <support/xthread.h>
+#include <support/xstdio.h>
+
+static pthread_barrier_t b;
+
+static void *
+tf (void *arg)
+{
+  int old;
+
+  TEST_COMPARE (pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, NULL), 0);
+  
+  TEST_COMPARE (pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, &old), 0);
+  TEST_COMPARE (old, PTHREAD_CANCEL_ASYNCHRONOUS);
+
+  /* Check if internal lock cleanup routines restore the cancellation type
+     correctly.  */
+  printf ("...\n");
+  TEST_COMPARE (pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, &old), 0);
+  TEST_COMPARE (old, PTHREAD_CANCEL_ASYNCHRONOUS);
+
+  xpthread_barrier_wait (&b);
+
+  /* Wait indefinitely for cancellation, which only works if asynchronous
+     cancellation is enabled.  */
+#ifdef SYS_futex
+  syscall (SYS_futex, &(int){0}, FUTEX_WAIT_PRIVATE, 0, NULL, NULL, 0);
+#else
+  for (;;);
+#endif
+
+  return 0;
+}
+
+static int
+do_test (void)
+{
+  xpthread_barrier_init (&b, NULL, 2);
+
+  pthread_t th = xpthread_create (NULL, tf, NULL);
+
+  xpthread_barrier_wait (&b);
+
+  xpthread_cancel (th);
+
+  void *status = xpthread_join (th);
+  TEST_VERIFY (status == PTHREAD_CANCELED);
+
+  return 0;
+}
+
+/* There is no need to wait full TIMEOUT if asynchronous is not working.  */
+#define TIMEOUT 3
+#include <support/test-driver.c>