> OK, I bit the bullet and read up about what group descriptors are. > This reference turned out to be quite good, in case you're collecting > links and you don't already have it... > > http://www.science.unitn.it/~fiorella/guidelinux/tlk/node95.html Already known here -- a great big picture document. > So what tool would you use to examine a group descriptor, in the > same way dumpe2fs is showing me the superblock? Forensic and handmade tools :( It's not that hard, though. Just find your way to the inodes and dump their contents. > And if I point mount > to a different superblock, does that make it also use a different > group descriptor? No. The group descriptors are not very redundant, anyway. > > 2) Use forensic tools. A search on freshmeat.net for 'forensic' > >should produce lots of those. I think this is exactly what you need > >now. > > Like "The Coroner's Toolkit"? This is going to be a steep learning > curve. Yes. Fortunately you seem to have evaded this route. > Apart from that... is now the time to ask around the e2fsck, dumpe2fs, > and debuge2fs people what's up with their tools? Can I wander into > their mailing lists asserting the existence of bugs yet? :-) By all means, do so! Explain the whole mess to them and ask for an analysis and, if then necessary, explanations why debugfs didn't work.