Chuck Silvers <[email protected]> writes:
> I guess your kernel was built with DIAGNOSTIC disabled? there's an assertion
> in pool_init() that you would have hit that would have made the problem more
> obvious:
>
> KASSERTMSG((prsize <= palloc->pa_pagesz),
> "%s: [%s] pool item size (%zu) larger than page size (%u)",
> __func__, wchan, prsize, palloc->pa_pagesz);
Wow, thanks for the explanation.
After it settles out if the different-IPL case can work, it would be
good to
- update the man page
- have pool_cache_init fail if what's passed will not work correctly,
even if DIAGNOSTIC is not enabled