Mouse <[email protected]> writes: >> 2) It's unclear to me why we have limits. Right now there is some >> (fairly small) amount of wired memory proportional to these limits. >> If it were dynamically allocated, so that asking for another >> semaphore just worked, that would be even better. > > Except doesn't that open you up to a DoS attack, running the kernel out > of memory by creating a zillion semaphores?
I meant to say, and maybe I didn't manage to, that you'd still want a max limit, but that it would be much higher than the amount you preallocate. More like the ulimit datasize.
