GitLab

Forcing large allocations here can creates serious fragmentation in
some cases, and since the large allocations are only a small
optimisation we should allow the nursery to hoover up small blocks
before allocating large chunks.

This makes compiling DynFlags a lot quicker

This squashes the "out of spill slots" panic that occasionally happens
on x86, by adding instructions to bump and retreat the C stack pointer
as necessary.  The panic has become more common since the new codegen,
because we lump code into larger blocks, and the register allocator
isn't very good at reusing stack slots for spilling (see Note [extra
spill slots]).

I assume that this is what is intended, as it is used with foldl'

and creating one is quite slow

It was always False before

Also added a couple of comments.

StgWord is a newtyped Word64, as it needed to be something that
has a UArray instance.

It's now a newtyped Integer. Perhaps a newtyped Word32 would make more
sense, though.

I'm not sure if there's a reason why the HeapRep constructor takes
2 WordOffs rather than 2 StgHalfWords.

Overlap the main thread's clearNursery() with the other threads.

I made it strict, as an incidental consequence of this patch:

  commit 5bae803a
  Author: Simon Peyton Jones <simonpj@microsoft.com>
  Date:   Sat Sep 15 23:09:25 2012 +0100
      Fix UNPACK with -fomit-interface-pragmas.

But it's very important that chooseBoxingStrategy is lazy, else
(in bigger programs with lots of recursion in types) GHC can
loop. This showed up in Data.Sequence; and I think it was making
haddock loop as well.

Anyway this patch makes it lazy again.