GitLab

patch from #878

Not much has changed really: just the removal of the overlap
restriction, and the re-instatement of the requirement that
-package-name must be used when compiling a package now.

The following packages are now "core" packages:

  base, Cabal, haskell98, readline, regex-base, regex-compat
  regex-posix, stm, template-haskell, unix, Win32

Core packages are those packages required to bootstrap GHC, or are
closely tied to GHC (stm, template-haskell).  These are the packages
that will be provided in a source distribution from now on.

All other packages are classified as "extra" packages.  As far as
binary distributions and nightly builds go, nothing will change -
we'll still build and include all these packages in the distributions.

NOTE: 'sh darcs-all get' will now get the core packages only.  To get
the extra packages too, use 'sh darcs-all --extra get'.

This bug causes crashse on Sparc when calling foreign functions with
more than 13 arguments.

These modules are always home modules, by definition, so the package
name is redundant.

For the benefit of old GHCs that don't understand {-# UNPACK #-}

The ordering of items in the parsed export list is now correct, so we
have to compensate to get the right output again.

From David's email:
The problem is that the inline assembler code was placing the result
of an operation in a register that is used as input later in the code.
At the bottom of this message I've extracted a short short code
fragment that you can run through gcc (on a powerpc machine) to see
the generated assembly output.

The changes to fix the problem are fairly simple.  The first adds an
ampersand to the output list of the assembly fragment ("=r" (result)
--> "=&r" (result)) The ampersand just tells gcc that result can not
be placed in a register used for any of the input parameters (o, n, or
p).  Otherwise, it feels free to place output parameters in the same
registers used by the inputs -- but because of the flow of control
here we need everything in a distinct register.  This change fixes the
TVar program above.

The second change adds a clobber list (the :"cc", "memory").  This
tells gcc that the condition code (due to the compare) and memory (due
to the store) might be changed during the asm execution.  The lack of
a clobber list did not seem to be causing any trouble, but without it
gcc is free to assume that no state is changed during the execution.

For some reason this was only enabled when $(bootstrapped)=YES.  This
would be one reason why the stage1 compiler is slow.  I guess we'll
find out if anything goes wrong.