Re: [PATCH 05/18] linux: Only use 64-bit syscall if required for pselect

[Lukasz Majewski <lukma@denx.de>]

[-- Attachment #1: Type: text/plain, Size: 10320 bytes --]

On Thu, 17 Jun 2021 08:50:51 -0300
Adhemerval Zanella <adhemerval.zanella@linaro.org> wrote:

> For !__ASSUME_TIME64_SYSCALLS there is no need to issue a 64-bit
> syscall if the provided timeout fits in a 32-bit one.  The 64-bit
> usage should be rare since the timeout is a relative one.  This also
> avoids the need to use supports_time64() (which breaks the usage case
> of live migration like CRIU or similar).
> 
> Checked on i686-linux-gnu on a 4.15 kernel and on a 5.11 kernel
> (with and without --enable-kernel=5.1) and on x86_64-linux-gnu.
> ---
>  misc/Makefile                                 |   9 ++
>  misc/tst-pselect.c                            | 120
> ++++++++---------- .../unix/sysv/linux/microblaze/pselect32.c    |
> 3 +- sysdeps/unix/sysv/linux/pselect.c             |  47 ++++---
>  sysdeps/unix/sysv/linux/pselect32.c           |   6 -
>  5 files changed, 95 insertions(+), 90 deletions(-)
> 
> diff --git a/misc/Makefile b/misc/Makefile
> index f34ab09fe3..fa40bf0e11 100644
> --- a/misc/Makefile
> +++ b/misc/Makefile
> @@ -148,6 +148,12 @@ CFLAGS-brk.op = $(no-stack-protector)
>  
>  include ../Rules
>  
> +ifeq (yes,$(build-shared))
> +librt = $(common-objpfx)rt/librt.so
> +else
> +librt = $(common-objpfx)rt/librt.a
> +endif
> +
>  $(objpfx)libg.a: $(dep-dummy-lib); $(make-dummy-lib)
>  
>  $(objpfx)tst-tsearch: $(libm)
> @@ -162,3 +168,6 @@ tst-allocate_once-ENV =
> MALLOC_TRACE=$(objpfx)tst-allocate_once.mtrace
> $(objpfx)tst-allocate_once-mem.out: $(objpfx)tst-allocate_once.out
> $(common-objpfx)malloc/mtrace $(objpfx)tst-allocate_once.mtrace > $@;
> \ $(evaluate-test) +
> +$(objpfx)tst-pselect: $(librt)
> +$(objpfx)tst-pselect-time64: $(librt)
> diff --git a/misc/tst-pselect.c b/misc/tst-pselect.c
> index c89176b058..f8c404c275 100644
> --- a/misc/tst-pselect.c
> +++ b/misc/tst-pselect.c
> @@ -16,14 +16,14 @@
>     <https://www.gnu.org/licenses/>.  */
>  
>  #include <errno.h>
> -#include <signal.h>
> -#include <stdio.h>
> -#include <unistd.h>
> -#include <sys/select.h>
> -#include <sys/wait.h>
> +#include <intprops.h>
> +#include <support/check.h>
> +#include <support/support.h>
> +#include <support/xsignal.h>
> +#include <support/xunistd.h>
> +#include <support/xtime.h>
>  #include <stdlib.h>
>  
> -
>  static volatile int handler_called;
>  
>  static void
> @@ -33,59 +33,43 @@ handler (int sig)
>  }
>  
>  
> -static int
> -do_test (void)
> +static void
> +test_pselect_basic (void)
>  {
>    struct sigaction sa;
>    sa.sa_handler = handler;
>    sa.sa_flags = 0;
>    sigemptyset (&sa.sa_mask);
>  
> -  if (sigaction (SIGUSR1, &sa, NULL) != 0)
> -    {
> -      puts ("sigaction failed");
> -      return 1;
> -    }
> +  xsigaction (SIGUSR1, &sa, NULL);
>  
>    sa.sa_handler = SIG_IGN;
> -  if (sigaction (SIGCHLD, &sa, NULL) != 0)
> -    {
> -      puts ("2nd sigaction failed");
> -      return 1;
> -    }
> +  xsigaction (SIGCHLD, &sa, NULL);
>  
>    sigset_t ss_usr1;
>    sigemptyset (&ss_usr1);
>    sigaddset (&ss_usr1, SIGUSR1);
> -  if (sigprocmask (SIG_BLOCK, &ss_usr1, NULL) != 0)
> -    {
> -      puts ("sigprocmask failed");
> -      return 1;
> -    }
> +  TEST_COMPARE (sigprocmask (SIG_BLOCK, &ss_usr1, NULL), 0);
>  
>    int fds[2][2];
> -
> -  if (pipe (fds[0]) != 0 || pipe (fds[1]) != 0)
> -    {
> -      puts ("pipe failed");
> -      return 1;
> -    }
> +  xpipe (fds[0]);
> +  xpipe (fds[1]);
>  
>    fd_set rfds;
>    FD_ZERO (&rfds);
>  
>    sigset_t ss;
> -  sigprocmask (SIG_SETMASK, NULL, &ss);
> +  TEST_COMPARE (sigprocmask (SIG_SETMASK, NULL, &ss), 0);
>    sigdelset (&ss, SIGUSR1);
>  
>    struct timespec to = { .tv_sec = 0, .tv_nsec = 500000000 };
>  
>    pid_t parent = getpid ();
> -  pid_t p = fork ();
> +  pid_t p = xfork ();
>    if (p == 0)
>      {
> -      close (fds[0][1]);
> -      close (fds[1][0]);
> +      xclose (fds[0][1]);
> +      xclose (fds[1][0]);
>  
>        FD_SET (fds[0][0], &rfds);
>  
> @@ -93,55 +77,63 @@ do_test (void)
>        do
>  	{
>  	  if (getppid () != parent)
> -	    exit (2);
> +	    FAIL_EXIT1 ("getppid()=%d != parent=%d", getppid(),
> parent); 
>  	  errno = 0;
>  	  e = pselect (fds[0][0] + 1, &rfds, NULL, NULL, &to, &ss);
>  	}
>        while (e == 0);
>  
> -      if (e != -1)
> -	{
> -	  puts ("child: pselect did not fail");
> -	  return 0;
> -	}
> -      if (errno != EINTR)
> -	{
> -	  puts ("child: pselect did not set errno to EINTR");
> -	  return 0;
> -	}
> +      TEST_COMPARE (e, -1);
> +      TEST_COMPARE (errno, EINTR);
>  
>        TEMP_FAILURE_RETRY (write (fds[1][1], "foo", 3));
>  
>        exit (0);
>      }
>  
> -  close (fds[0][0]);
> -  close (fds[1][1]);
> +  xclose (fds[0][0]);
> +  xclose (fds[1][1]);
>  
>    FD_SET (fds[1][0], &rfds);
>  
> -  kill (p, SIGUSR1);
> +  TEST_COMPARE (kill (p, SIGUSR1), 0);
>  
>    int e = pselect (fds[1][0] + 1, &rfds, NULL, NULL, NULL, &ss);
> -  if (e == -1)
> -    {
> -      puts ("parent: pselect failed");
> -      return 1;
> -    }
> -  if (e != 1)
> -    {
> -      puts ("parent: pselect did not report readable fd");
> -      return 1;
> -    }
> -  if (!FD_ISSET (fds[1][0], &rfds))
> -    {
> -      puts ("parent: pselect reports wrong fd");
> -      return 1;
> -    }
> +  TEST_COMPARE (e, 1);
> +  TEST_VERIFY (FD_ISSET (fds[1][0], &rfds));
> +}
> +
> +static void
> +test_pselect_large_timeout (void)
> +{
> +  support_create_timer (0, 100000000, false, NULL);
> +
> +  int fds[2];
> +  xpipe (fds);
> +
> +  fd_set rfds;
> +  FD_ZERO (&rfds);
> +  FD_SET (fds[0], &rfds);
> +
> +  sigset_t ss;
> +  TEST_COMPARE (sigprocmask (SIG_SETMASK, NULL, &ss), 0);
> +  sigdelset (&ss, SIGALRM);
> +
> +  struct timespec ts = { TYPE_MAXIMUM (time_t), 0 };
> +
> +  TEST_COMPARE (pselect (fds[0] + 1, &rfds, NULL, NULL, &ts, &ss),
> -1);
> +  TEST_VERIFY (errno == EINTR || errno == EOVERFLOW);
> +}
> +
> +static int
> +do_test (void)
> +{
> +  test_pselect_basic ();
> +
> +  test_pselect_large_timeout ();
>  
>    return 0;
>  }
>  
> -#define TEST_FUNCTION do_test ()
> -#include "../test-skeleton.c"
> +#include <support/test-driver.c>
> diff --git a/sysdeps/unix/sysv/linux/microblaze/pselect32.c
> b/sysdeps/unix/sysv/linux/microblaze/pselect32.c index
> a4f436462b..70b7b52a48 100644 ---
> a/sysdeps/unix/sysv/linux/microblaze/pselect32.c +++
> b/sysdeps/unix/sysv/linux/microblaze/pselect32.c @@ -35,8 +35,7 @@
> __pselect32 (int nfds, fd_set *readfds, fd_set *writefds, struct
> timeval tv32, *ptv32 = NULL; if (timeout != NULL)
>      {
> -      if (! in_time_t_range (timeout->tv_sec)
> -	  || ! valid_nanoseconds (timeout->tv_nsec))
> +      if (! valid_nanoseconds (timeout->tv_nsec))
>  	{
>  	  __set_errno (EINVAL);
>  	  return -1;
> diff --git a/sysdeps/unix/sysv/linux/pselect.c
> b/sysdeps/unix/sysv/linux/pselect.c index 5e8a0cc2dc..79e8584ea7
> 100644 --- a/sysdeps/unix/sysv/linux/pselect.c
> +++ b/sysdeps/unix/sysv/linux/pselect.c
> @@ -18,7 +18,23 @@
>  
>  #include <sys/select.h>
>  #include <sysdep-cancel.h>
> -#include <time64-support.h>
> +
> +static int
> +pselect64_syscall (int nfds, fd_set *readfds, fd_set *writefds,
> +		   fd_set *exceptfds, const struct __timespec64
> *timeout,
> +		   const sigset_t *sigmask)
> +{
> +#ifndef __NR_pselect6_time64
> +# define __NR_pselect6_time64 __NR_pselect6
> +#endif
> +  /* NB: This is required by ARGIFY used in x32 internal_syscallN.
> */
> +  __syscall_ulong_t data[2] =
> +    {
> +      (uintptr_t) sigmask, __NSIG_BYTES
> +    };
> +  return SYSCALL_CANCEL (pselect6_time64, nfds, readfds, writefds,
> exceptfds,
> +			 timeout, data);
> +}
>  
>  int
>  __pselect64 (int nfds, fd_set *readfds, fd_set *writefds, fd_set
> *exceptfds, @@ -37,30 +53,25 @@ __pselect64 (int nfds, fd_set
> *readfds, fd_set *writefds, fd_set *exceptfds, we can only pass in 6
> directly.  If there is an architecture with support for more
> parameters a new version of this file needs to be created.  */
> -
> -#ifndef __NR_pselect6_time64
> -# define __NR_pselect6_time64 __NR_pselect6
> -#endif
> +#ifdef __ASSUME_TIME64_SYSCALLS
> +  return pselect64_syscall (nfds, readfds, writefds, exceptfds,
> timeout,
> +			    sigmask);
> +#else
> +  bool is32bit = timeout != NULL
> +		 ? in_time_t_range (timeout->tv_sec) : true;
>    int r;
> -  if (supports_time64 ())
> +  if (!is32bit)
>      {
> -      /* NB: This is required by ARGIFY used in x32
> internal_syscallN.  */
> -      __syscall_ulong_t data[2] =
> -	{
> -	  (uintptr_t) sigmask, __NSIG_BYTES
> -	};
> -      r = SYSCALL_CANCEL (pselect6_time64, nfds, readfds, writefds,
> exceptfds,
> -			  timeout, data);
> +      r = pselect64_syscall (nfds, readfds, writefds, exceptfds,
> timeout,
> +			     sigmask);
>        if (r == 0 || errno != ENOSYS)
>  	return r;
> -
> -      mark_time64_unsupported ();
> +      __set_errno (EOVERFLOW);
> +      return -1;
>      }
>  
> -#ifndef __ASSUME_TIME64_SYSCALLS
> -  r = __pselect32 (nfds, readfds, writefds, exceptfds, timeout,
> sigmask);
> +  return __pselect32 (nfds, readfds, writefds, exceptfds, timeout,
> sigmask); #endif
> -  return r;
>  }
>  
>  #if __TIMESIZE != 64
> diff --git a/sysdeps/unix/sysv/linux/pselect32.c
> b/sysdeps/unix/sysv/linux/pselect32.c index eb59b51cdf..48724d8727
> 100644 --- a/sysdeps/unix/sysv/linux/pselect32.c
> +++ b/sysdeps/unix/sysv/linux/pselect32.c
> @@ -29,12 +29,6 @@ __pselect32 (int nfds, fd_set *readfds, fd_set
> *writefds, struct timespec ts32, *pts32 = NULL;
>    if (timeout != NULL)
>      {
> -      if (! in_time_t_range (timeout->tv_sec))
> -	{
> -	  __set_errno (EINVAL);
> -	  return -1;
> -	}
> -
>        ts32 = valid_timespec64_to_timespec (*timeout);
>        pts32 = &ts32;
>      }

Reviewed-by: Lukasz Majewski <lukma@denx.de>


Best regards,

Lukasz Majewski

--

DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-59 Fax: (+49)-8142-66989-80 Email: lukma@denx.de

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]