[Clfs-dev] Embedded: Is e2fsprogs really needed in core embedded book?

[Andrew Bradford]

> I don't remember the original reason for it being there in the first
> place. Is there anything from e2fsprogs or util-linux being used in the
> bootscripts?

The rc.d/startup bootscript does run fsck if fsck exists and is
executable.  If fsck doesn't exist, it is simply skipped and there are
no errors or issues (provided your file system is clean).

Through some testing on my BeagleBoard, I've found that BusyBox isn't
quite as full featured in the fsck department as I had hoped.  It
provides fsck, but only as the front end to the real fsck executables.
 This is the same way fsck works in the real e2fsprogs, it's simply a
front end that can be called and calls the proper fsck utility (such
as fsck.ext3).

If we want fsck ability for ext2/3/4, it looks like we'll have to keep
e2fsprogs.  Granted, some embedded systems will be using non-ext2/3/4
file systems such as jffs2 (which I'm lead to believe has no explicit
fsck), so having e2fsprogs be an integral part of the book may not be
required.  It might fit better in the "beyond" section of the embedded
book but be highly recommended for those using an ext2/3/4 file
system.

If e2fsprogs gets moved out of the core part of the embedded book, the
core's starting to look a little sparse.  It'll just have BusyBox,
Iana-Etc, and Zlib.  I'm of the impression that Zlib isn't really a
true requirement until the "beyond" section.  This all begs the
question, is simply booting with BusyBox and Iana-Etc enough to be
considered the basic entirety of the core part of the book?  This type
of discussion should probably be another thread rather than continuing
this one...

-Andrew