"make check" failures on Linux/x86

[Jie Zhang]

Alan Modra wrote:

> 	* readelf.c (dynamic_nent): New variable.

The new dynamic_nent clarify the confusion caused by the two different 
meanings of dynamic_size.

> 	(get_32bit_dynamic_section): Set it.
> 	(get_64bit_dynamic_section): Here too.
> 	(process_dynamic_section): Use it instead of dynamic_size.
> 	(process_syminfo): Likewise.

However, there is still a problem. For some MIPS targets, e.g. IRIX5, 
the PT_DYNAMIC segment includes .dynstr, .dynsym, and .hash sections 
besides the .dynamic sections. The dynamic_size we get in 
process_program_headers () is the size of the whole PT_DYNAMIC segment 
of an executable file that has no section header table. In this case we 
cannot take the dynamic_size as the size of dynamic section. We need use 
the DT_NULL tag to determine the actual number of entries in the dynamic 
section.


-- 
Jie