Subject: Re: LFS -> 1.6 kernel failure.
To: Chris Tribo <email@example.com>
From: Konrad Schroder <firstname.lastname@example.org>
Date: 06/13/2002 17:28:06
On Thu, 13 Jun 2002, Chris Tribo wrote:
> panic: getblk: block size invariant failed
> Stopped in pid 105 (lfs_cleanerd) at cpu_Debugger+0x4: leave
> lfs_markv: 8192 > 1024 (fixed)
I think there may be a problem with the cleaner's interaction with growing
files, which is most troublesome with growing fragments. The "(fixed)"
message in itself shouldn't be anything to worry about, but the "block
size invariant" message means that the filesystem is getting confused
about what size this fragment really is.
I'm working on possible solutions for this, but in the meantime I'll
change the coalescing part of lfs_cleanerd not to coalesce files with
fewer than NDADDR blocks, which should help a bit.
It might be a good idea to build LFSes with fsize == bsize until I can
figure out a good solution for this problem.