It now just warns about files it doesn't recognise, rather than giving
an error. This means that random text files etc in the same directory
as the bindists don't make it fall over.

We weren't seting the _DO_HADDOCK variables early enough.

I don't think it's necessary to build ghc-pkg that early.

Dependency problem was discovered by int-e.

I've also added some comments about what's going on.

Also a couple of other fixes and sanity checks along the way.

We now leave making installers to the Haskell Platform.

The file no longer exists

The external core chapter had an abstract, which seems to reset the
page count to 0.

Documentation is done too

The desugarer was generating a redundant box/unbox pair on the
LHS of a RULE, which in turn made matching fail.

See Note [Simple coercions] in DsBinds.

It was checking to see whether the hscTarget is HscC or HscLlvm, and
changing it to HscC otherwise. But this meant that as well as rejecting
HscAsm, it also rejected HscInterpreted and HscNothing.

Now we just check whether it's HscAsm.

Darwin x86 has inconsistent PIC base register, so splitting (which happened before)
ensures that each cmm procedure only has one entry point (namely the first block).

This reverts commit 024df664.

Of course I gaff on my last day...

This includes selector, ap, and constructor thunks. They are still
guarded by the -ticky-dyn-thk flag.

per cmm procedure on Darwin/PPC, because of splitting.

x86 should be treated the same way, I'll come back to that later.

size_expr now ignores RealWorld lambdas, arguments, and applications.

Worker-wrapper previously removed all lambdas from a function, if they
were all unused. Removing *all* value lambdas is no longer
allowed. Instead (\_ -> E) will become (\_void -> E), where it used to
become E. The previous behavior can be recovered via the new
-ffun-to-thunk flag.

Nofib notables:

----------------------------------------------------------------
        Program               O2          O2 newly ignoring RealWorld
                                          and not turning function
                                          closures into thunks
----------------------------------------------------------------

 Allocations

  comp_lab_zift            333090392%           -5.0%
reverse-complem            155188304%           -3.2%

        rewrite             15380888%           +4.0%
         boyer2              3901064%           +7.5%

rewrite previously benefited from fortunate LoopBreaker choice that is
now disrupted.

A function in boyer2 goes from $wonewayunify1 size 700 to size 650,
thus gets inlined into rewritelemmas, thus exposing a parameter
scrutinisation, thus allowing SpecConstr, which unfortunately involves
reboxing.

Run Time

 fannkuch-redux                 7.89%          -15.9%

            hpg                 0.25%           +5.6%
           wang                 0.21%           +5.8%

/shrug

Patch from wgmitchener.

From the ticket:
The two addresses (adjustorStub and code) contain the same memory
(double-mapped), but one is writable while the other is executable.
This is how libffi works around the SELinux restrictions. On
non-SELinux systems the code and data addresses are probably the same.

I tried making mkJumpToAddr return [Word32] on all platforms,
but it went wrong on x86 (possibly due to alignment?). Rather than
chasing the bug, I've just used an Either type for now.