Cong Wang [off-list ref] wrote:

quoted

In order to prevent breaking userspace, perhaps make it so that the
kernel caps cfg.max at twice that value?  Would allow storing up to
16777216 addresses with an average chain depth of 16 (which is quite
large).  We could increase the max limit in case someone presents a use
case.

Not sure if I understand this, I don't see how cap'ing cfg->max could
help prevent breaking user-space? Are you suggesting to cap it with
HASHLIMIT_MAX_SIZE too? Something like below?

+       if (cfg->max > 2 * HASHLIMIT_MAX_SIZE)
+               cfg->max = 2 * HASHLIMIT_MAX_SIZE;

Yes, thats what I meant, cap the user-provided value to something thats
going to be less of a problem.

But now that I read it, the "2 *" part looks really silly, so I suggst
to go with " > FOO_MAX", else its not a maximum value after all.