#13944: Introduce synchronized FFI -------------------------------------+------------------------------------- Reporter: winter | Owner: (none) Type: feature request | Status: new Priority: normal | Milestone: 8.4.1 Component: Compiler | Version: 8.0.1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: None/Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by bgamari):
Yes, i mean all the thread pause stuff. I want to understand what is the minimum preparation before we can make FFI calls which doesn't get in GC's way. I believe this helps to reduce GC pause.
I don't believe there is anything that we do currently that can simply be omitted without careful consideration at very least. Have you confirmed that `safe` is really as slow as you think it is? I think that we should better characterise the problem before we begin discussing possible solutions. I think it would be helpful if you had a benchmark and could use it answer a few questions, a. What overhead is incurred by a `unsafe` call? b. What overhead is incurred by a `safe` call? c. Where specifically is the additional overhead of the `safe` call coming from? -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/13944#comment:10 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler