Comment by aap_
2 months ago
Hi, this is me. I'm still hacking on it but ran into some hard to understand kernel bugs. once i mount more than the root filesystem (say /usr/man) there are issues with inode allocation/freeing. mixing and matching v4 and v5 stuff in various ways can also lead to other interesting bugs but often an allocated inode ends up on the freelist, and things break.
Otoh it's so much fun to hack and fiddle with the unix kernel :) very zen
> but ran into some hard to understand kernel bugs
Are the bugs in the original, or somehow artifacts of how we got it? (Or, phrased differently: Were these bugs present at the University of Utah in 1974, or are they "new"?)
That's the puzzling thing. i find it hard to believe they sent out an operating system that can't deal with multiple file systems. yet i can't get them to work correctly. The pre-v4 nsys kernel is another piece in the puzzle. it doesn't have pipes implemented yet but aside from that (i put them in) it also shows these "busy i" bugs, but even when running on a single disk. Maybe there's more i'm doing wrong there since it's running on the fs from the v4 tape. But that i'm getting such similar bugs in different situations suggests there is something wrong that i'm not seeing yet. gotta debug more.
If it turns out to be a timing-related bug it may be that the bug was much less obvious on real hardware.
a) Do these inode issues also happen with the supplied (v4) kernel? b) Do these inode issues also happen with a rebuilded kernel which uses the original lib1 and lib2?
I once had strange effects on V6 if lib1 and/or lib2 were rebuild by me.
Should be not hard to test.
1 reply →