itojun@iijlab.net wrote:

> 	Maybe very dumb question (I'm not a filesystem guy)...
> 
> 	Now update(8) is included in the kernel, and turned on by default.
> 	I would like to know how does it interact with fsck on root partition.
> 	When we do fsck on root partition to recover from fatal filesystem
> 	crash, we do:
> 	- make sure in single user mode, no update(8) running
> 	- run fsck on root partition
> 	- NEVER type sync, immediately type halt
> 	to avoid buffer cache writeback (if the above process itself is
> 	incorrect, kick me).  What happens when we always have update(8)
> 	in kernel?  What is the suggested process for fsck'ing root partition?

First guess is that because root is mounted r/o on boot is that the
in-kernel trickle-sync doesn't do anything...

Simon.