[CFT] ELF Relocatable x86 and x86_64 bzImages

[ebiederm at xmission.com (Eric W. Biederman)]

"H. Peter Anvin" <hpa at zytor.com> writes:

> Eric W. Biederman wrote:
>>
>>> Ok. I am decompressing the kernel to 16MB and after reducing 1MB of
>>> CONFIG_PHYSICAL_START I am left with 15MB which is not 4M aligned
>>> hence I seems to be running into it.
>>>
>>> I changed it to
>>>
>>> if ((u32)output) & 0x3fffff)
>>>
>>> and kdump kernel booted fine. But this will run into issues if I load
>>> kernel at 1MB.
>>>
>>> I got a dump question. Why do I have to load the kernel at 4MB alignment?
>>> Existing kernel boots loads at 1MB, which is non 4MB aligned and it works
>>> fine?
>> 4MB is a little harsh, but I haven't worked through what the exact rules
>> are, I know 4MB is the worst case alignment for arch/i386.
>>
>
> 4 MB would be worst case for i386; 2 MB for x86-64.  Actually the x86-64 worst
> case would be gigabyte, but that's more than a little bit extreme.

Yep and that is what a test for, except for the gigabyte case which we don't
currently implement.  Although I can imagine that gigabyte pages might
be interesting for the identity mapped part of the page table.

Eric