dirty pages count of a vnode

[Reinoud Zandijk <reinoud%NetBSD.org@localhost>]

Dear folks,

as you might know, the out of discspace detection in UDF is not that good. I
tried various scheme's to count the number of still to be written sectors.
They all failed since a sector can be written twice before writeout thus
leading to a miscount.

A better scheme i thought might be is to use UVM/UBC's dirty page counting
information to have an indication before and after the vop_write() operation
of the number of dirty pages and thus of the sectors still to be written out.

Hoever, i couldn't find the number of dirty pages in the UVM object associated
with the vnode nor couldn't find it in genfs.

Any ideas as to tackle this problem in general or how to count the number of
dirty pages an object has? Shouldn't that be handy in general?

With regards,
Reinoud