Re: readelf reporting of e_shstrndx is slightly wrong

[Egeyar Bagcioglu]

Hi,


On 08/21/2018 05:31 PM, Nick Clifton wrote:
>       if (filedata->section_headers != NULL
>           && header->e_shstrndx == (SHN_XINDEX & 0xffff))
>         printf (" (%u)", filedata->section_headers[0].sh_link);
>        else if (header->e_shstrndx != SHN_UNDEF
>                && header->e_shstrndx >= header->e_shnum)
>         printf (_(" <corrupt: out of range>"));
>     
> [...]
>
> Note - the use if "& 0xffff" in the above code is confusing, and looks
> surplus to me, but I do not think that it makes any difference to the
> behaviour.

I haven't checked the location of this quoted code; however, the use of 
"& 0xffff" is probably due to a clever hack made internally:

Elf[32|64]_External_Ehdr in include/elf/external.h declares e_shstrndx 
as 2 bytes,
while Elf_Internal_Ehdr in include/elf/internal.h  declares it as 4 
bytes because we would like to be able to keep the overflowing values 
within the same variable without extra checks.

We would like to be able to store values that are reserved or higher 
than 0xffff as valid indices internally in the 4-byte e_shstrndx. 
Therefore, some constants including SHN_XINDEX are re-defined in 
elf/internal.h as 4 byte values. These redefinitions hold the new bounds 
on such indices. Because of this internal simplification, though, we 
need to use "& 0xffff" to get the 2-byte value that's used when reading 
values in or outputting them.

Regards,
Egeyar