This is the mail archive of the mailing list for the glibc project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: glibc utimes glitch with coreutils 'touch'

Paul Eggert <eggert@CS.UCLA.EDU> wrote:
> I'll submit an alternate patch to coreutils so that it works around
> this glibc glitch.  The simplest workaround is to not invoke utimes
> when building with glibc; perhaps I can improve on that, but I don't
> know.

We don't have to condemn the coreutils always
to use the inferior utime interface on glibc systems.
Instead, I'm adding a configure-time test of utimes, so that
if it works, coreutils will use it.

Here's the C program it'll compile and run.
So far it compiles and exits successfully on all systems listed
except for the one with Debian libc6 version 2.3.2-1.

/* tested with gcc on the following:
   AIX 4.3  (both gcc and cc)
   openbsd 3.2
   solaris 9
   netbsd 1.6
   OSF1 V4.0 (both gcc and cc)
   linux 2.4.20, 2.4.21
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/time.h>
#include <unistd.h>
#include <stdlib.h>
#include <stdio.h>
#include <utime.h>

main ()
  static struct timeval timeval[2] = {{9, 10}, {11, 12}};
  struct stat sbuf;
  char const *file = "x";
  FILE *f;

  exit ( ! ((f = fopen (file, "w"))
	    && fclose (f) == 0
	    && utimes (file, timeval) == 0
	    && lstat (file, &sbuf) == 0
	    && sbuf.st_atime == timeval[0].tv_sec
	    && sbuf.st_mtime == timeval[1].tv_sec) );

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]