Major Llvm refactoring

This combined patch reworks the LLVM backend in a number of ways:

1. Most prominently, we introduce a LlvmM monad carrying the contents of
   the old LlvmEnv around. This patch completely removes LlvmEnv and
   refactors towards standard library monad combinators wherever possible.

2. Support for streaming - we can now generate chunks of Llvm for Cmm as
   it comes in. This might improve our speed.

3. To allow streaming, we need a more flexible way to handle forward
   references. The solution (getGlobalPtr) unifies LlvmCodeGen.Data
   and getHsFunc as well.

4. Skip alloca-allocation for registers that are actually never written.
   LLVM will automatically eliminate these, but output is smaller and
   friendlier to human eyes this way.

5. We use LlvmM to collect references for llvm.used. This allows places
   other than cmmProcLlvmGens to generate entries.

1 files changed, 1 insertions, 5 deletions

diff --git a/compiler/main/CodeOutput.lhs b/compiler/main/CodeOutput.lhs
index f94030306d..b8b187241b 100644
--- a/compiler/main/CodeOutput.lhs
+++ b/compiler/main/CodeOutput.lhs
@@ -168,13 +168,9 @@ outputLlvm :: DynFlags -> FilePath -> Stream IO RawCmmGroup () -> IO ()
 outputLlvm dflags filenm cmm_stream
   = do ncg_uniqs <- mkSplitUniqSupply 'n'
 
-       -- ToDo: make the LLVM backend consume the C-- incrementally,
-       -- by pushing the cmm_stream inside (c.f. nativeCodeGen)
-       rawcmms <- Stream.collect cmm_stream
-
        {-# SCC "llvm_output" #-} doOutput filenm $
            \f -> {-# SCC "llvm_CodeGen" #-}
-                 llvmCodeGen dflags f ncg_uniqs rawcmms
+                 llvmCodeGen dflags f ncg_uniqs cmm_stream
 \end{code}