From: Charles D. <ch...@dy...> - 2017-10-21 20:51:11
|
Howdy -- I'm trying to ensure that, given the same set of input files, I get a byte-for-byte identical squashfs image created. This makes it easier to prove that an image is correct (making it possible to prove that it represents a given set of inputs and hasn't been subtly modified by comparing the hash against that of a new image generated from those same inputs), and is also part of an effort to minimize the size of binary deltas between subsequent builds when distributed with casync or similar tooling. One source of presently-unavoidable delta is the various places where mksquashfs calls time(). A 3rd-party patch for this exists in various trees, and appears to have been submitted to this mailing list in late 2015, but I'm not finding any contemporaneous followup or indication of what would have needed to happen for it to be merged. This patch is a little more aggressive than what I immediately need (ignoring mtimes taken from input files as well as those generated from whole cloth), but as it's been helpful to others, it seems a reasonable starting point. Is there work that should happen for this to be acceptable upstream? |