Re: [SSI-devel] Re: [Etherboot-developers] BUG: Large ramdisks break ELF image

[<ke...@et...>]

>> Hmm, this points to the ELF header from the first block of the boot file
>> which is loaded at where mkelf-linux instructs Etherboot to, usually low
>> memory. But this theory is easy to test. Could someone change first32.c
>> to make a copy of seg[S_SETUP]->p_paddr before the ramdisk move and jump
>> to that? If so I'll make the patch.
>
>I don't happen to have a machine handy I can run this on but the code
>builds ok.  
>
>It turns out we already had a copy of seg[S_SETUP]->p_paddr

I suspect you are right about this. I was thinking of tagged images,
where the header also specifies where the header block is to go in
memory.  mknbi specifies that it goes into low memory. In ELF images,
there is no such constraint so the header block is located in
Etherboot's memory. So it looks like I'll have to put out 1.4.3 soon. Of
course the real mystery is why it didn't break sooner.