Hi Ryusuke,


On Monday 13 June 2011 16:40:48 you wrote:
> So, you hit the bug on a raw hadrware.
> In my hardware, /sys/kernel/mm/cleancache/* didn't change.
> 
> Is it reproducible without zcache?
> 
> If not, we should share the steps with zcache.


I think I've got it reproductable. I believe it shouldn't matter whether it's 
bare metal or virtualized OS.

First off, attached is kernel log of the crash. Kernel version is 3.0-rc2 with 
a slight change for zcache -- more on that in next mail.


Steps to reproduce:
1) enable zcache (important details in the next mail)
2) hack NILFS2 to add cleancache (in next mail, again)
3) compile, boot etc.
4) mount NILFS2
5) (important!) read content of some file(s) from NILFS2 (it seems to me 
cleancache currently handles only file data, not metadata; but I may be wrong 
here
6) echo 3 > /proc/sys/vm/drop_caches

As far as I remember, the `echo 3 > /proc/sys/vm/drop_caches' it worked a-OK 
with EXT4 alone. Want me to verify that?


Thanks for your patience ;-)
-- 
dexen deVries

``One can't proceed from the informal to the formal by formal means.''