unofficial mirror of libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Alistair Francis <alistair23@gmail.com>
To: Lukasz Majewski <lukma@denx.de>
Cc: Alistair Francis <alistair.francis@wdc.com>,
	GNU C Library <libc-alpha@sourceware.org>,
	 Paul Eggert <eggert@cs.ucla.edu>,
	Joseph Myers <joseph@codesourcery.com>
Subject: Re: [PATCH v5] sysdeps/clock_nanosleep: Use clock_nanosleep_time64 if avaliable
Date: Fri, 8 Nov 2019 08:56:00 -0800	[thread overview]
Message-ID: <CAKmqyKOMKO3V2ZiTpNZp0_F3om5UE4b0X0Hvh=uuwwAzMnyMVA@mail.gmail.com> (raw)
In-Reply-To: <20191108135432.06b5ff95@jawa>

On Fri, Nov 8, 2019 at 4:54 AM Lukasz Majewski <lukma@denx.de> wrote:
>
> Hi Alistair,
>
> Please find my comments regarding checking of NULL pointers.
>
> > The clock_nanosleep syscall is not supported on newer 32-bit
> > platforms (such as RV32). To fix this issue let's use
> > clock_nanosleep_time64 if it is avaliable.
> > ---
> > This was patch was runtime tested with RV32 and RV64
> > It was build tested using the ./scripts/build-many-glibcs.py script.
> >
> > I also ran:
> > $ make ; make install ; make check
> > tests on native ARM (32-bit) with the following three confiugrations:
> >  - 4.19 Kernel and 4.19 Headers
> >  - 5.2 Kernel and 4.19 Headers
> >  - 5.2 Kernel and 5.2 Headers
> > and didn't see any regressions
> >
> > v5:
> >  - Fix clock_nanosleep syscall
> >  - Rebase on master
> >
> > v4:
> >  - Rebase on master
> >  - Use __clock_nanosleep to avoid duplicate implementations
> >  - Fix the error handling when a syscall fails
> > v2:
> >  - Explicitly include `#include <kernel-features.h>`
> >
> >  include/time.h                            | 20 +++++++
> >  sysdeps/unix/sysv/linux/clock_nanosleep.c | 66
> > +++++++++++++++++++++-- 2 files changed, 81 insertions(+), 5
> > deletions(-)
> >
> > diff --git a/include/time.h b/include/time.h
> > index b3e635395db..03389bda290 100644
> > --- a/include/time.h
> > +++ b/include/time.h
> > @@ -209,6 +209,26 @@ libc_hidden_proto (__difftime64)
> >
> >  extern double __difftime (time_t time1, time_t time0);
> >
> > +#if __TIMESIZE == 64
> > +# define __thrd_sleep_time64 thrd_sleep
> > +# define __clock_nanosleep_time64 __clock_nanosleep
> > +# define __nanosleep_time64 __nanosleep
> > +# define __nanosleep_nocancel_time64 __nanosleep_nocancel
> > +#else
> > +extern int __thrd_sleep_time64 (const struct __timespec64*
> > time_point,
> > +                                struct __timespec64* remaining);
> > +libc_hidden_proto (__thrd_sleep_time64)
> > +extern int __clock_nanosleep_time64 (clockid_t clock_id,
> > +                                     int flags, const struct
> > __timespec64 *req,
> > +                                     struct __timespec64 *rem);
> > +libc_hidden_proto (__clock_nanosleep_time64)
> > +extern int __nanosleep_time64 (const struct __timespec64
> > *requested_time,
> > +                                struct __timespec64 *remaining);
> > +libc_hidden_proto (__nanosleep_time64)
> > +extern int __nanosleep_nocancel_time64 (const struct __timespec64
> > *requested_time,
> > +                                        struct __timespec64
> > *remaining); +libc_hidden_proto (__nanosleep_nocancel_time64)
> > +#endif
> >
> >  /* Use in the clock_* functions.  Size of the field representing the
> >     actual clock ID.  */
> > diff --git a/sysdeps/unix/sysv/linux/clock_nanosleep.c
> > b/sysdeps/unix/sysv/linux/clock_nanosleep.c index
> > f3c6fd2d5f7..7212dcf9c6d 100644 ---
> > a/sysdeps/unix/sysv/linux/clock_nanosleep.c +++
> > b/sysdeps/unix/sysv/linux/clock_nanosleep.c @@ -16,6 +16,7 @@
> >     <https://www.gnu.org/licenses/>.  */
> >
> >  #include <time.h>
> > +#include <kernel-features.h>
> >  #include <errno.h>
> >
> >  #include <sysdep-cancel.h>
> > @@ -26,9 +27,11 @@
> >  /* We can simply use the syscall.  The CPU clocks are not supported
> >     with this function.  */
> >  int
> > -__clock_nanosleep (clockid_t clock_id, int flags, const struct
> > timespec *req,
> > -                struct timespec *rem)
> > +__clock_nanosleep_time64 (clockid_t clock_id, int flags, const
> > struct __timespec64 *req,
> > +                          struct __timespec64 *rem)
> >  {
> > +  int r;
> > +
> >    if (clock_id == CLOCK_THREAD_CPUTIME_ID)
> >      return EINVAL;
> >    if (clock_id == CLOCK_PROCESS_CPUTIME_ID)
> > @@ -37,11 +40,64 @@ __clock_nanosleep (clockid_t clock_id, int flags,
> > const struct timespec *req, /* If the call is interrupted by a signal
> > handler or encounters an error, it returns a positive value similar
> > to errno.  */ INTERNAL_SYSCALL_DECL (err);
> > -  int r = INTERNAL_SYSCALL_CANCEL (clock_nanosleep, err, clock_id,
> > flags,
> > -                                req, rem);
> > +
> > +#ifdef __ASSUME_TIME64_SYSCALLS
> > +# ifndef __NR_clock_nanosleep_time64
> > +#  define __NR_clock_nanosleep_time64 __NR_clock_nanosleep
> > +# endif
> > +  r = INTERNAL_SYSCALL_CANCEL (clock_nanosleep_time64, err, clock_id,
> > +                               flags, req, rem);
> > +#else
> > +# ifdef __NR_clock_nanosleep_time64
> > +  r = INTERNAL_SYSCALL_CANCEL (clock_nanosleep_time64, err, clock_id,
> > +                               flags, req, rem);
> > +
> > +  if (r == 0 || errno != ENOSYS)
> > +    {
> > +      return (INTERNAL_SYSCALL_ERROR_P (r, err)
> > +              ? INTERNAL_SYSCALL_ERRNO (r, err) : 0);
> > +    }
> > +# endif /* __NR_clock_nanosleep_time64 */
> > +  struct timespec ts32, tr32;
> > +
> > +  if (! in_time_t_range (req->tv_sec))
> > +    {
> > +      __set_errno (EOVERFLOW);
> > +      return -1;
> > +    }
> > +
> > +  ts32 = valid_timespec64_to_timespec (*req);
>
> Here we don't need to check req == NULL, as we will go into this piece
> of code only for 32 bit systems (and kernel < 5.1) with glibc's local
> copy of struct __timespec64.
>
> > +  r =  INTERNAL_SYSCALL_CANCEL (clock_nanosleep, err, clock_id,
> > flags,
> > +                                &ts32, &tr32);
> > +
> > +  if ((r == 0 || errno != ENOSYS) && rem)
> > +    *rem = valid_timespec_to_timespec64 (tr32);
> > +#endif /* __ASSUME_TIME64_SYSCALLS */
> > +
> >    return (INTERNAL_SYSCALL_ERROR_P (r, err)
> > -       ? INTERNAL_SYSCALL_ERRNO (r, err) : 0);
> > +          ? INTERNAL_SYSCALL_ERRNO (r, err) : 0);
> >  }
> > +
> > +#if __TIMESIZE != 64
> > +int
> > +__clock_nanosleep (clockid_t clock_id, int flags, const struct
> > timespec *req,
> > +                   struct timespec *rem)
> > +{
> > +  int r;
> > +  struct __timespec64 treq64, trem64;
> > +
>
> ^^^^^^^^^^^^ - [*]
>
> > +  treq64 = valid_timespec_to_timespec64 (*req);
>
> The above line would cause regression. Please find below explanation:
>
> The check on *rqtp (in Linux kernel's clock_nanosleep syscall handling):
> https://elixir.bootlin.com/linux/v5.4-rc5/source/kernel/time/posix-timers.c#L1220
>
> When compiling test program with clock_nanosleep() passing *request [1]
> as NULL [3] - the clock_nanosleep returns 14, which corresponds to
> #define EFAULT 14 [2] (which matches [1]).
>
> Hence we should have following code in [*]:
>
> if (req == NULL)
>   {
>     __set_errno(EFAULT)
>     returm 1;
>   }
>
>
> Otherwise we will break in glibc before we enter the clock_nanosleep
> syscall (which would then return with proper error code).

I think you are right. I have updated the patch.

Alistair

>
>
> Links:
> [1] - http://man7.org/linux/man-pages/man2/clock_nanosleep.2.html
> [2] - cpp -dM /usr/include/errno.h | grep 'define E' | sort -n -k 3
> [3] - test program:
>
> int main(int argc, char **argv)
> {
>   int result;
>
>   result = clock_nanosleep(CLOCK_REALTIME, 0 /* relative */,
>     NULL, NULL);
>
>   if (result)
>     printf("clock_nanosleep: %d\n", result);
>
>   return 0;
> }
>
> > +  r = __clock_nanosleep_time64 (clock_id, flags, &treq64, &trem64);
> > +
> > +  if (r == 0 || errno != ENOSYS)
> > +    {
> > +      if (rem)
> > +        *rem = valid_timespec64_to_timespec (trem64);
> > +    }
> > +
> > +  return r;
> > +}
> > +#endif
> >  libc_hidden_def (__clock_nanosleep)
> >  versioned_symbol (libc, __clock_nanosleep, clock_nanosleep,
> > GLIBC_2_17); /* clock_nanosleep moved to libc in version 2.17;
>
>
>
>
> 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

  reply	other threads:[~2019-11-08 16:56 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-11-07 18:11 [PATCH v5] sysdeps/clock_nanosleep: Use clock_nanosleep_time64 if avaliable Alistair Francis
2019-11-07 19:15 ` Adhemerval Zanella
2019-11-07 19:23   ` Alistair Francis
2019-11-07 19:37     ` Alistair Francis
2019-11-08 13:49     ` Adhemerval Zanella
2019-11-08 17:08       ` Alistair Francis
2019-11-08 12:54 ` Lukasz Majewski
2019-11-08 16:56   ` Alistair Francis [this message]
2019-11-08 17:07     ` Alistair Francis

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://www.gnu.org/software/libc/involved.html

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='CAKmqyKOMKO3V2ZiTpNZp0_F3om5UE4b0X0Hvh=uuwwAzMnyMVA@mail.gmail.com' \
    --to=alistair23@gmail.com \
    --cc=alistair.francis@wdc.com \
    --cc=eggert@cs.ucla.edu \
    --cc=joseph@codesourcery.com \
    --cc=libc-alpha@sourceware.org \
    --cc=lukma@denx.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).