Re: [coreboot] [patch] mkelfimage intelligent placement of ramdisk

[Myles Watson]

> I hope others test this patch so we can be certain it doesn't break
> anyone where it previously worked.  It shouldn't.

I was hoping it would fix qemu's problem with a LAB kernel, but it doesn't
change the behavior.
> > > if (e820_map[i].addr + info->real_mode->initrd_size >=
>
> 0x38000000)
> > > 	continue;
> >
> > ..what is this 3.5MB limit about? Also might this condition be
> > sensitive to an overflow error?
>
> Peter isn't sure why 3.5MB specifically.  He said it is known to work
> for many years in other boot loaders.  There may have been a historic
> reason for this particular value but he doesn't know.

I read that value as 896MB, for what it's worth.

Thanks,
Myles