On Thu, Jun 20, 2002 at 03:58:20PM +0200, Wojciech Puchar wrote:
> > > >
> > > > seconds sound cool. but does it mean per-process queues???
> > >
> > > I don't think so.  Write ordering should be preserved, but if a process
> > > tries to schedule a write when no buffers are available to it (but
> > > would be available to other processes) then it would block.  A process
> > > would need keep track of how many buffers it's using for I/O so that
> 
> it means counting not-yet-written buffers or all buffers per process?
> 
> both is quite simple to implement

Bear in mind that I'm not really familiar with UBC, or buffer allocation
policy in general...  That said, my thought was "not-yet-written buffers".
If a page isn't dirty, or isn't destined to disk, it isn't going to cause
I/O starvation: there's no I/O associated with it.

--aidan