Russell Cattelan writes:
=> Daniel Moore wrote:
=> > Anyway, there it is. Anyone have any thoughts?
=>
=> Yup, I've been looking into this.
=> What are you doing to blow the 128k cache?
To test other changes I've been making, I've been doing nasty
things like making 20000 files and then unlinking every second
one, then unmounting and remounting.
=> A "clean" file-system shouldn't be in reovery code, you must
=> have crashed the system at some point.
To check that a FS is clean, the head and tail of the log need
to be located. xlog_recover gets called, and from it xlog_find_tail
and xlog_find_head get called.
Once the head and tail are located, the cleanliness of the FS
is known, but the log needs to be checked to find the ends...
=> kmalloc failures are a know problem...
yup.
-----------------------------------------------------
Daniel Moore dxm@xxxxxxx
R&D Software Engineer Phone: +61-3-98348209
SGI Performance Tools Group Fax: +61-3-98132378
-----------------------------------------------------
|