Allow top-level string literals in Core (#8472)

This commits relaxes the invariants of the Core syntax so that a
top-level variable can be bound to a primitive string literal of type
Addr#.

This commit:

* Relaxes the invatiants of the Core, and allows top-level bindings whose
  type is Addr# as long as their RHS is either a primitive string literal or
  another variable.

* Allows the simplifier and the full-laziness transformer to float out
  primitive string literals to the top leve.

* Introduces the new StgGenTopBinding type to accomodate top-level Addr#
  bindings.

* Introduces a new type of labels in the object code, with the suffix "_bytes",
  for exported top-level Addr# bindings.

* Makes some built-in rules more robust. This was necessary to keep them
  functional after the above changes.

This is a continuation of D2554.

Rebasing notes:
This had two slightly suspicious performance regressions:

* T12425: bytes allocated regressed by roughly 5%
* T4029: bytes allocated regressed by a bit over 1%
* T13035: bytes allocated regressed by a bit over 5%

These deserve additional investigation.

Rebased by: bgamari.

Test Plan: ./validate --slow

Reviewers: goldfire, trofi, simonmar, simonpj, austin, hvr, bgamari

Reviewed By: trofi, simonpj, bgamari

Subscribers: trofi, simonpj, gridaphobe, thomie

Differential Revision: https://phabricator.haskell.org/D2605

GHC Trac Issues: #8472

1 files changed, 13 insertions, 0 deletions

diff --git a/compiler/cmm/CLabel.hs b/compiler/cmm/CLabel.hs
index 0f3410a66e..ee87ef1b37 100644
--- a/compiler/cmm/CLabel.hs
+++ b/compiler/cmm/CLabel.hs
@@ -26,6 +26,7 @@ module CLabel (
         mkApEntryLabel,
         mkApInfoTableLabel,
         mkClosureTableLabel,
+        mkBytesLabel,
 
         mkLocalClosureLabel,
         mkLocalInfoTableLabel,
@@ -389,6 +390,9 @@ data IdLabelInfo
 
   | ClosureTable        -- ^ Table of closures for Enum tycons
 
+  | Bytes               -- ^ Content of a string literal. See
+                        -- Note [Bytes label].
+
   deriving (Eq, Ord)
 
 
@@ -474,6 +478,7 @@ mkClosureTableLabel         :: Name -> CafInfo -> CLabel
 mkLocalConInfoTableLabel    :: CafInfo -> Name -> CLabel
 mkLocalConEntryLabel        :: CafInfo -> Name -> CLabel
 mkConInfoTableLabel         :: Name -> CafInfo -> CLabel
+mkBytesLabel                :: Name -> CLabel
 mkClosureLabel name         c     = IdLabel name c Closure
 mkInfoTableLabel name       c     = IdLabel name c InfoTable
 mkEntryLabel name           c     = IdLabel name c Entry
@@ -481,6 +486,7 @@ mkClosureTableLabel name    c     = IdLabel name c ClosureTable
 mkLocalConInfoTableLabel    c con = IdLabel con c ConInfoTable
 mkLocalConEntryLabel        c con = IdLabel con c ConEntry
 mkConInfoTableLabel name    c     = IdLabel name c ConInfoTable
+mkBytesLabel name                 = IdLabel name NoCafRefs Bytes
 
 mkConEntryLabel       :: Name -> CafInfo -> CLabel
 mkConEntryLabel name        c     = IdLabel name c ConEntry
@@ -935,6 +941,7 @@ idInfoLabelType info =
     ConInfoTable  -> DataLabel
     ClosureTable  -> DataLabel
     RednCounts    -> DataLabel
+    Bytes         -> DataLabel
     _             -> CodeLabel
 
 
@@ -1056,6 +1063,11 @@ export this because in other modules we either have
        * A saturated call 'Just x'; allocate using Just_con_info
 Not exporting these Just_info labels reduces the number of symbols
 somewhat.
+
+Note [Bytes label]
+~~~~~~~~~~~~~~~~~~
+For a top-level string literal 'foo', we have just one symbol 'foo_bytes', which
+points to a static data block containing the content of the literal.
 -}
 
 instance Outputable CLabel where
@@ -1234,6 +1246,7 @@ ppIdFlavor x = pp_cSEP <>
                        ConEntry         -> text "con_entry"
                        ConInfoTable     -> text "con_info"
                        ClosureTable     -> text "closure_tbl"
+                       Bytes            -> text "bytes"
                       )