OSDev.org

Did you test it out on Linux/CYGWIN? I will be trying this out on Linux but need to reboot to do that

No one else who tried this?

Neo wrote:Did you test it out on Linux/CYGWIN?

I tested it on Cygwin. I have a VM with Mandrake 9.2 on it, but I don't remember if I tested with that or not. It has exactly the same verions of binutils and gcc as my Cygwin installation does.

I remember this happening to me a few months back (the non-constant load base error), and although it's starting to fade a little I'm pretty sure the problem had something to do with missing sections.

In my case, I think my final object file had a .rodata section but I wasn't telling the linker where to put it in the script. Check to see if you have any sections in your object file(s) that are being linked that might not have a home in your linker script.

Anyone know where I can get version 2.17 of binutils from? Need to check this out further.

Neo wrote: Anyone know where I can get version 2.17 of binutils from? Need to check this out further.

You probably want to get a cvs snapshot since there hasn't been an official stable release. Of course, some people have created releases from the snapshots for various systems, byt cvs is probably your best option

Actually, try adding *(COMMON) to .bss. Also, for one reason or other, I'm using *(.rodata*) with the extra star in my script.

mystran wrote: Also, for one reason or other, I'm using *(.rodata*) with the extra star in my script.

I had to add one to HigherHalfBareBones (and my own kernel) when I started compiling with optimizations enabled. The compiler added new sections called .rodata.something (I forget what the something was). I think it contained string constants, but I don't remember exactly...

mystran wrote: Actually, try adding *(COMMON) to .bss. Also, for one reason or other, I'm using *(.rodata*) with the extra star in my script.

The *(.rodata*) and *(COMMON) are both there in my script.

Ahm, ok, sorry, must have looked at the other script posted to this thread.