> >If you want to include the other attack I mentioned (I just tried it,
> >got up to > 160000 vnodes), then you have to exclude vnodes that are
> >only live because of v_cache_src entries from the count.
> 
> It should probably only count vnodes in "actual" use.

There's no counter for that currently.  Counting non-zero v_usecount
vs. non-zero v_holdcnt separately would be reasonably easy.

Actual use obviously shouldn't include cached data.  Can you say off
the top of your head whether v_holdcnt applies to anything other than
v_cache_src and non-VM buffer-cache (struct buf) stuff?

If not, then v_usecount == 0 could be considered non-use without
worrying about v_holdcnt, since most vnodes with cached data are going
to have an associated vm_object holding a real reference.

> >BTW: You still haven't committed the v_id patch I sent you in May.  Is
> >there any specific reason for this?
> 
> I seem to remember we stalled on some detail which wouldn't or
> couldn't work was it NFS ?

No, there was a completely unrelated NFS bug I ran into while looking
into it (which has been fixed), the last comment from you seemed to
imply that you were going to commit the patch.

There were lots of details that needed further attention in my
experiments for actually deallocating vnodes, but those patches
certainly weren't supposed to be committed.


To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message

Reply via email to