Generalise reallyUnsafePtrEquality# and use it

fixes #9192 and #17126
updates containers submodule

1. Changes the type of the primop `reallyUnsafePtrEquality#` to the most
general version possible (heterogeneous as well as levity-polymorphic):

> reallyUnsafePtrEquality#
>   :: forall {l :: Levity} {k :: Levity}
>        (a :: TYPE (BoxedRep l)) (b :: TYPE (BoxedRep k))
>   . a -> b -> Int#

2. Adds a new internal module, `GHC.Ext.PtrEq`, which contains pointer
equality operations that are now subsumed by `reallyUnsafePtrEquality#`.
These functions are then re-exported by `GHC.Exts` (so that no function
goes missing from the export list of `GHC.Exts`, which is user-facing).
More specifically, `GHC.Ext.PtrEq` defines:

  - A new function:
    * reallyUnsafePtrEquality :: forall (a :: Type). a -> a -> Int#

  - Library definitions of ex-primops:
     * `sameMutableArray#`
     * `sameSmallMutableArray`
     * `sameMutableByteArray#`
     * `sameMutableArrayArray#`
     * `sameMutVar#`
     * `sameTVar#`
     * `sameMVar#`
     * `sameIOPort#`
     * `eqStableName#`

  - New functions for comparing non-mutable arrays:
     * `sameArray#`
     * `sameSmallArray#`
     * `sameByteArray#`
     * `sameArrayArray#`

  These were requested in #9192.

Generally speaking, existing libraries that
use `reallyUnsafePtrEquality#` will continue to work with the new,
levity-polymorphic version. But not all!
Some (`containers`, `unordered-containers`, `dependent-map`) contain
the following:

> unsafeCoerce# reallyUnsafePtrEquality# a b

If we make `reallyUnsafePtrEquality#` levity-polymorphic, this code
fails the current GHC representation-polymorphism checks.
We agreed that the right solution here is to modify the library;
in this case by deleting the call to `unsafeCoerce#`,
since `reallyUnsafePtrEquality#` is now type-heterogeneous too.

1 files changed, 2 insertions, 25 deletions

diff --git a/compiler/GHC/StgToCmm/Prim.hs b/compiler/GHC/StgToCmm/Prim.hs
index 542372105e..d61880a0e2 100644
--- a/compiler/GHC/StgToCmm/Prim.hs
+++ b/compiler/GHC/StgToCmm/Prim.hs
@@ -342,6 +342,8 @@ emitPrimOp dflags primop = case primop of
   StableNameToIntOp -> \[arg] -> opIntoRegs $ \[res] ->
     emitAssign (CmmLocal res) (cmmLoadIndexW platform arg (fixedHdrSizeW profile) (bWord platform))
 
+  EqStablePtrOp -> \args -> opTranslate args (mo_wordEq platform)
+
   ReallyUnsafePtrEqualityOp -> \[arg1, arg2] -> opIntoRegs $ \[res] ->
     emitAssign (CmmLocal res) (CmmMachOp (mo_wordEq platform) [arg1,arg2])
 
@@ -1462,20 +1464,6 @@ emitPrimOp dflags primop = case primop of
   FloatToDoubleOp -> \args -> opTranslate args (MO_FF_Conv W32 W64)
   DoubleToFloatOp -> \args -> opTranslate args (MO_FF_Conv W64 W32)
 
--- Word comparisons masquerading as more exotic things.
-
-  SameMutVarOp            -> \args -> opTranslate args (mo_wordEq platform)
-  SameMVarOp              -> \args -> opTranslate args (mo_wordEq platform)
-  SameIOPortOp            -> \args -> opTranslate args (mo_wordEq platform)
-  SameMutableArrayOp      -> \args -> opTranslate args (mo_wordEq platform)
-  SameMutableByteArrayOp  -> \args -> opTranslate args (mo_wordEq platform)
-  SameMutableArrayArrayOp -> \args -> opTranslate args (mo_wordEq platform)
-  SameSmallMutableArrayOp -> \args -> opTranslate args (mo_wordEq platform)
-  SameTVarOp              -> \args -> opTranslate args (mo_wordEq platform)
-  EqStablePtrOp           -> \args -> opTranslate args (mo_wordEq platform)
--- See Note [Comparing stable names]
-  EqStableNameOp          -> \args -> opTranslate args (mo_wordEq platform)
-
   IntQuotRemOp -> \args -> opCallishHandledLater args $
     if ncg && (x86ish || ppc) && not (quotRemCanBeOptimized args)
     then Left (MO_S_QuotRem  (wordWidth platform))
@@ -2092,17 +2080,6 @@ genericFabsOp w [res_r] [aa]
 
 genericFabsOp _ _ _ = panic "genericFabsOp"
 
--- Note [Comparing stable names]
--- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
---
--- A StableName# is actually a pointer to a stable name object (SNO)
--- containing an index into the stable name table (SNT). We
--- used to compare StableName#s by following the pointers to the
--- SNOs and checking whether they held the same SNT indices. However,
--- this is not necessary: there is a one-to-one correspondence
--- between SNOs and entries in the SNT, so simple pointer equality
--- does the trick.
-
 ------------------------------------------------------------------------------
 -- Helpers for translating various minor variants of array indexing.