On Tue, Oct 29, 2013 at 7:08 PM, Niko Matsakis <n...@alum.mit.edu> wrote:
> If I understand correctly, what you are proposing is to offer fixed > size stacks and to use a guard page to check for stack overflow (vs a > stack preamble)? > > My two thoughts are: > > 1. I do think segmented stacks offer several tangible benefits: > > - Recursion limited only by available memory / address space > - Avoid chewing up address space on 32 bit builds > > However, I can accept that on balance they are not worth the price, > given that point #2 is probably not that important for 64-bit systems. > > It is sad to lose limitless recursion but typically one can rewrite > routines that recurse arbitrarily deep to use a stack on the heap, > though sometimes the code is very unnatural, and using a stack on the > heap will not play as well with lifetimes, since the compiler doesn't > know that you are obeying a stack discipline. Fixed stacks can be expanded once you have precise collection ( help will come from LLVM eventually...). What i think hasnt been discussed ( though it has previously) is how much of a penalty segmented stacks represent and what it could be .. In terms of memory allocations a faster expansion or greater reuse from a segment pool ( Why does the scheduler segment pool start empty. I would have thought you would have 2Meg worth pre allocated - one runs in the context of a microbench the other allocated before hand ( like C stacks) ) . a 4^N expansion of each stack would result in more contigous stacks. And if stack is unused for a long time hand some segments back. In terms of the stack size check , as has been mentioned you can do more with some analysis , each method gives an indication of worst cases stack growth , these can then be added together to reduces checks by 80% and hence significantly reduce the performance impact . Though obviously for production release soon you will get there faster with fixed stacks earlier. Ben
_______________________________________________ Rust-dev mailing list Rust-dev@mozilla.org https://mail.mozilla.org/listinfo/rust-dev
