Re: Unmapping fpages

[Espen Skoglund]

[R Koot]
> L4_Unmap is the only syscall that hasn't changed since v2, I believe
> it is showing some defects and it should be considered to fix this
> in the kernel instead of user-mode.

Or maybe Unmap has not been changed considerably because it serves its
purpose very well...

However, we do have several proposed changes for Unmap in the
pipeline; one of them being the tagged unmaps I mentioned in an
earlier mail.  We have not sorted out all the consequences of these
changes yet, though, and in some cases (as with the tagged unmap) we
haven't even tried to specify the semantics properly (i.e., by
updating the reference manual).

That said, your comment about considering "to fix this in the kernel
instead of user-mode" does not sit well with me.  The L4 philosophy
has always been to sort out problems in user-mode rather than
in-kernel if possible.  The reason to include something in the kernel
has been that either the functionality is not possible to implement at
user-level, or can only be implemented with a prohibitively expensive
performance overhead for some common usage scenario.

        eSk