Bug#129559: libpager robustness

[Thomas Bushnell, BSG]

Marcus Brinkmann <Marcus.Brinkmann@ruhr-uni-bochum.de> writes:

> I remember reading about this in the TODO list etc.  Are there any concrete
> other unrobustnesses you remember?

There are a jillion.  The filesystem basically assumes that it's
talking to the kernel.  If it's not, then the user could be bollixing
things up horribly in various ways; since the pager is essentially
shared by all users, it can affect others too.

One example would be to fail to properly respond to sync and flush
requests. 

> BTW, I did not consider the return value too much, but just assumed that
> returning the same as before would be fine.  I now verified that the return
> value is ignored in mach_msg_server (in glibc), and that current practice
> seems to be that the demux function returns 0 if it could not process the
> message and 1 if it could and did.  As we are not processing the message,
> but just synchronise our state, I don't see a problem (even semantically).

Yes, that's right.