[GHC] #8703: Use guard pages rather than heap checks

[GHC]

#8703: Use guard pages rather than heap checks
-------------------------------------+------------------------------------
        Reporter:  schyler           |            Owner:  simonmar
            Type:  feature request   |           Status:  closed
        Priority:  normal            |        Milestone:
       Component:  Runtime System    |          Version:
      Resolution:  wontfix           |         Keywords:
Operating System:  Unknown/Multiple  |     Architecture:  Unknown/Multiple
 Type of failure:  None/Unknown      |       Difficulty:  Unknown
       Test Case:                    |       Blocked By:
        Blocking:                    |  Related Tickets:
-------------------------------------+------------------------------------
Changes (by simonmar):

 * status:  new => closed
 * resolution:   => wontfix


Comment:

 This is a lot harder to do than you might imagine - the problem is not the
 memory protection itself, but how to handle the fault when the invalid
 page is accessed.  This requires all kinds of platform-specific hackery.

 GHC's heap is constructed of linked lists of pages, so at the end of each
 page we have to swing the heap pointer to the next page in the list, which
 is often not contiguous with the previous page.  Using the page fault
 trick is even harder when the memory must grow through non-contiguous
 pages, because the fault handler has to also update the current heap
 pointer.

 I'm not saying it can't be done, but it is (a) very tricky, (b) non-
 portable, and (c) after you've surmounted all the obstacles, it probably
 doesn't save that much time, if any.  And because it's non-portable, you
 still have to keep the old way of doing things.

 I'm closing the ticket.  But if you want to implement this and demonstrate
 that it is indeed a win, be my guest!

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8703#comment:1>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler