Re: [PATCH v2 7/8] Use PRINTF_FORTIFY instead of _IO_FLAGS2_FORTIFY.

["Gabriel F. T. Gomes" <gabriel at inconstante dot eti dot br>]

On Mon, 29 Oct 2018, Gabriel F. T. Gomes wrote:

>From: Zack Weinberg <zackw@panix.com>
>
> int
>-__vdprintf_chk (int d, int flags, const char *format, va_list arg)
>+__vdprintf_chk (int d, int flag, const char *format, va_list ap)
> {
>
> [...]
>
>-  done = __vfprintf_internal (&tmpfil.file, format, arg, 0);
>
> [...]
>
>-  return done;
>+  return __vdprintf_internal (d, format, ap, mode);

While reviewing the first version of this patch, I noticed that it could
have an effect on bug 20231, because, after this patch, __vdprintf_chk
will call __vdprintf_internal, which has the extra check for EOF (as
reported in the bug).  However, the bug is marked as unconfirmed, and I
was unable to reproduce it with the following test case:

    #include <stdio.h>
    #include <stdarg.h>
    #include <sys/types.h>
    #include <sys/stat.h>
    #include <fcntl.h>
    #include <unistd.h>
    int
    main (void)
    {
      int fd, libret, sysret;
      va_list ap;
      fd = open ("/tmp/blablabla", O_RDWR | O_TRUNC | O_CREAT, S_IRWXU);
      if (fd == -1)
        perror (NULL);
      sysret = close (fd);
      if (sysret)
        perror (NULL);
      libret = vdprintf (fd, "blablabla", ap);
      if (libret != EOF)
        printf ("Bug 20231 reproduced\n");
      return 0;
    }

Maybe the test case is wrong, in which case I could fix it, then mark this
patch as solving bug 20231.  On the other hand, if the test case is
correct, we could just close bug 20231.