Re: Out-of-Line memory deallocation: MIG vs. glibc

[olafBuddenhagen]

And a last "Hi again"

On Tue, Apr 11, 2006 at 06:45:04AM -0400, Thomas Schwinge wrote:

> Now we have to figure out if the new fakeroot actually only obfuscates
> the issues you uncovered, which might very well be, because of the
> radically decreased (factor 300) `fdslen'.

I'm not sure what you mean by "obfuscates" in this context. The fact
that it doesn't transfer this excessive number of fds anymore, is very
clearly the reason why it doesn't trigger the Hurd/MIG bug (which I
tracked down and fixed) anymore.

> Is the Hurd supposed to cope with a `fdslen' of around a thousand; are
> the valid uses for that?

I don't think it happens very often that a program opens >512 files and
then exec()s. Nor am I sure whether this is likely to happen very often
with any other kind of port rights. But it could. The Hurd is supposed
not to have any such arbitrary limits, right?

And it any case, it definitely shouldn't crash.

-antrik-