On Tue, Oct 29, 2013 at 7:08 AM, 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. > > 2. I think that our official semantics for stack overflow should be > task failure, not abort. There are some technical challenges to be > overcome with regard to the best way to signal/detect stack overflow, > and I'm fine with this being a "todo" item (possibly for a long time). > But abort is wrong. > > One non-technical difficulty to failing on overflow is how to handle > user-defined destructors when there is no stack to run them on -- but > I think this is adequately addressed by keeping a red zone (so that > simple dtors work) and implementing Graydon's plan for handling > recursive failure otherwise. We also have to modify drop glue to not > be recursive (see point #1 about the convenience of limitless > recursion -- though of course drop glue must be ready for OOM as > well). > If we want to unwind on task failure, we'll need to disable the `prune-eh` pass that bubbles up `nounwind` since every function will be able to unwind. I think it will cause a very significant increase in code size.
_______________________________________________ Rust-dev mailing list Rust-dev@mozilla.org https://mail.mozilla.org/listinfo/rust-dev
