Re: [PATCH v2] libelf: Use int64_t for offsets in libelf.h

[Mark Wielaard <mjw at redhat dot com>]

On Fri, 2015-10-09 at 17:40 -0700, Josh Stone wrote:
> Some systems don't have loff_t, like FreeBSD where off_t always supports
> large files.  We need a standardized 64-bit signed type for the public
> header, without depending on configuration... OK, just use int64_t.

I note that this is a deliberate difference between the elfutils libelf
interface and other libelf implementations, which use off_t in the
public interface and don't have LFS support on 32bit systems. We were
lucky enough to use off64_t, then loff_t, from the start, so we always
support LFS as far as possible.

I do think that the type should kept signed to not accidentally change
any assumptions of programs using the current loff_t elements. So
int64_t is a good choice.

I don't really like that changing it to int64_t makes it less clear we
are talking about offsets here. But I don't have a good suggestion what
to do different. Maybe introduce a typedef int64_t elf_off_t? Or maybe
detect we don't have loff_t and define it ourselves somehow? But that is
not easy to do with the preprocessor.

Keeping the sys/types.h include seems correct since we need some other
types for the Elf_Arhdr struct.

Cheers,

Mark