DragonFly kernel List (threaded) for 2011-03
I have a high level question on maxvnodes limit.
One of the gsoc project description says;
"further caching of the related VM object even after the vnode is
thrown away by associating it with a mount point and inode number"
I assume the reason for this is that;
1. vnodes maybe released when there is no real memory pressure.
2. keeping VM objects may still be beneficial in case the file is
If that is the case, why not just increase the maxvnodes limit? What
is the downside of having a too large a limit? How to determine the
On the implementation side, I am thinking of passing the VM objects
to the root vnode when vnodes are reclaimed. And the root vnode
becomes responsible for freeing the VM objects in a particular order
and passing VM objects back to new vnodes (based on the inode
number?). Is this the right approach?