BFD and global variables

[Brian Blietz]

Sorry for my last post.
I have figured it out now.

DUH!  The VMA of the .data section is not going to correspond to
anything in the binary, becuase the .data section doesn't really exist!

I have used the section->filepos to correct my offset calculation.

Brian

Brian Blietz wrote:

> Is there anyway to change the .data section of a file
> using BFD?
>
> I have the following code that does not work:
> file_buffer is an array filled with fread() of the file that the BFD 
> corresponds to.
>
> ----------------------------------------------------------------------------------------------- 
>
> #define ELF_BASE_VMA 0x8048000
> unsigned int *data_section_ptr;
> int offset;
>
> offset = symbol_table[i]->section->vma + symbol_table[i]->value
>                         - ELF_BASE_VMA;
> (unsigned int)data_section_ptr = (unsigned int)file_buffer + offset;
> *data_section_ptr = some_value;
> ----------------------------------------------------------------------------------------------- 
>
>
> BUT, this code does work for changing the .text section:
> it changes the value that a pushl instruction will use.
> ----------------------------------------------------------------------------------------------- 
>
> unsigned int *pushl_id_ptr;
> int offset;
>
> offset = symbol_table[i]->section->vma + symbol_table[i]->value
>                         - ELF_BASE_VMA;
> (unsigned int)pushl_id_ptr = (unsigned int)file_buffer + offset + 4;
> *pushl_id_ptr = rand();           
> ----------------------------------------------------------------------------------------------- 
>
>
> After I have updated file_buffer, I write it off to another file with 
> fwrite().
>
> For some reason, my pointer into the file array is pointing at
> a bunch of 0's.  If I do objdump -s and look at the .data section,
> there is plenty of (non-zero) data there.
>
>
>
> Regards,
> Brian


-- 
Brian Blietz
bblietz@iastate.edu

"There are 10 types of people in the world...
Those who understand binary, and those who don't."