Since there seems to be a renewed interest in LiME and it's format, I think it's time that we extend the format to include the storage of optional metadata. Anything that can be collected at acquisition time saves the need for scanning and other possibly more complex processes during analysis. Formats like AFF4 are great, but are too complex to implement in the kernel.

I'd like to crowdsource opinions on how the LiME format should be extended. When contributing thoughts please keep the following in mind.

Changes to the format should not break existing parsers
To minimize the number of future changes, the enhancements should be as flexible as possible. For example, I've heard rumors that the LiME format is being used in acquisition tools that target more than just Linux, so I'd prefer a generic key/value store for metadata rather than any hardcoded solution.
Whatever we collect during acquisition time (at least in LiME) must be easily accessible from a kernel module

What are your thoughts? What metadata should be collected? Obvious answers that come to mind are DTB, kernel virtual base address, and KASLR slide, but I'm sure there are more.