When Depthcharge boots a Multiboot kernel, it checks the ramdisk image for being in the Fuchsia/Zircon ZBI (previously known as BOOTDATA) format.  In this case, it adds some information to the image, increasing its size.  It updates the ZBI headers inside the ramdisk to reflect the new size correctly.  But it fails to update the Multiboot module headers to match, so the Multiboot protocol says the ramdisk is smaller than it actually is.

This means the Zircon Multiboot entry code (now actually a trampoline loader, no longer in the Zircon kernel proper) can't do sanity checks of the Multiboot protocol header-reported size against the size encoded in the ZBI header inside the ramdisk.

We've worked around this by just disabling that sanity check.  But it would be nice to have this fixed and rolled out to Eve devices at some point so we can put that check in.

I have an untested but straightforward fix for this that I'll send to code review shortly.