Re: Getting Started with Hurd-L4

[Sam Mason]

Marcus Brinkmann wrote:
>mmap() only will create an entry in the database of the pager that
>indicates which file should be mapped at which place.  Only when a
>page fault arises is an actual mapping established.  At that point
>(and not earlier), the container stuff will happen.  The memory in the
>container is then mapped from the physmem task with a generic
>map-something-from-a-container interface.

OK, I misunderstood something someone else had said about the
container getting "locked".  I thought that only one task could have a
container mapped into its address space at one time.  I didn't quite
realise how low-level the container stuff is going to be so assumed it
would work diferently.

One thing I did just think of is something that Linux calls
"over-commit".  Not sure if this system will be affected by it, but
when I read the above paragraph that's what immediatly sprang into
mind.  If memory pages are given out on a demand only basis, are tasks
going to be affected by physmem making promises that it can't keep?

Cheers,
  Sam