[PATCH v3 1/2] sysdeps/stat: Handle 64-bit ino_t types on 32-bit hosts

[Joseph Myers]

On Thu, 3 Oct 2019, Alistair Francis wrote:

> diff --git a/sysdeps/unix/sysv/linux/generic/bits/stat.h b/sysdeps/unix/sysv/linux/generic/bits/stat.h
> index 1565f3f8248..0c1850322be 100644
> --- a/sysdeps/unix/sysv/linux/generic/bits/stat.h
> +++ b/sysdeps/unix/sysv/linux/generic/bits/stat.h
> @@ -40,7 +40,10 @@
>  /* Versions of the `xmknod' interface.  */
>  #define _MKNOD_VER_LINUX	0
>  
> -#if defined __USE_FILE_OFFSET64
> +#if defined __USE_FILE_OFFSET64 || defined __INO_T_MATCHES_INO64_T
> +# if defined __INO_T_MATCHES_INO64_T && !defined __OFF_T_MATCHES_OFF64_T
> +#  error "ino_t and off_t must both be the same type"
> +# endif
>  # define __field64(type, type64, name) type64 name
>  #elif __WORDSIZE == 64

Logically it feels like this actually should be added to the __WORDSIZE == 
64 case rather than the __USE_FILE_OFFSET64, so that the type names such 
as __ino_t get used to define the fields rather than those such as 
__ino64_t.  (If that makes a difference to anything in practice, it should 
just be to what typedef name debug info says was used to declare the 
fields.)

There are actually three types involved, not two.  If there's no other 
place that needs to check whether blkcnt_t and blkcnt64_t match and so 
needs a macro for that, I suppose some convenient .c file in linux/generic 
could do

#if __INO_T_MATCHES_INO64_T
_Static_assert (sizeof (__blkcnt_t) == sizeof (__blkcnt64_t),
		"__blkcnt_t and __blkcnt64_t must match");
#endif

to make sure the build fails if anyone tries to create a broken 
configuration where some types match but not others.

-- 
Joseph S. Myers
joseph@codesourcery.com