On 9/11/05, Matthias Scheler <tron@zhadum.de> wrote:
> In article <6b2d1e190509101248368131f9@mail.gmail.com>,
>         "Julio M. Merino Vidal" <jmmv84@gmail.com> writes:
> > I'm pleased to announce that tmpfs, an (efficient) memory file-system,
> > has been added to NetBSD's main source tree.
>=20
> That's great news. Thanks a lot.
>=20
> I've just tried it and wonder if I have found a bug:

Yes, sure it's a bug, although it looks certainly strange.  I have to
review the parts that control file system and node usage, because
they are not efficient nor correct (as seems from this bug).
Furthermore, I suspect there is a (related) bug in the error handling
path, because these kind of errors (ENOSPC) are often followed
by a panic :-/

I can reproduce this by unpacking a pkgsrc tarball on my 1GB
RAM + 2GB swap machine with an automatically sized tmpfs; it
breaks half way (triggering an assertion inside UVM code).

I'll look at this after the NFS changes I have in the tree are flushed.

Thanks for the report and for testing this!

--=20
Julio M. Merino Vidal <jmmv84@gmail.com>
http://www.livejournal.com/users/jmmv/
The NetBSD Project - http://www.NetBSD.org/