Question about the differences between objdump and objcopy

XU SHENG Sheng.Xu@alcatel-lucent.com
Tue May 27 05:16:00 GMT 2008


Hi Alan,
	In fact, my doubt is how to translate an ELF file to raw binary file without the usage of bfd library. My work failed because of the unexpected zeroed bytes padded in binary file output using bfd library (just like the output of objcopy).I have looked up some SPECs relative to ELF, but no detail information about this topic found. 

Best Regards,
Edison

-----Original Message-----
From: Alan Modra [mailto:amodra@bigpond.net.au] 
Sent: 2008年5月26日 14:30
To: XU SHENG
Cc: binutils@sources.redhat.com
Subject: Re: Question about the differences between objdump and objcopy

On Mon, May 26, 2008 at 11:01:27AM +0800, XU SHENG wrote:
> Program Headers:
>   Type           Offset   VirtAddr   PhysAddr   FileSiz MemSiz  Flg Align
>   LOAD           0x000000 0x08000000 0x08000000 0x00000 0x0815c RW  0x10000
>   LOAD           0x010000 0xfe000000 0xfe000000 0x24000 0x24050 RWE 0x10000
>   LOAD           0x040000 0xfe040000 0xfe040000 0x3ac000 0x3ac000 R   0x10000
>   GNU_STACK      0x000000 0x00000000 0x00000000 0x00000 0x00000 RWE 0x4
> 
> 	The Vir/Phy addr of all loadable segment are the same, but there is still many many pad bytes in binary output. 

Well, you would expect some padding, wouldn't you?  You must have gaps between loaded sections to get these program headers.  Please see bfd/binary.c:binary_set_section_contents if you need more details.

--
Alan Modra
Australia Development Lab, IBM



More information about the Binutils mailing list