nativeGen: Allow -fregs-graph to be used

Previously the flag was silently ignored due the #7679 and #8657. This,
however, seems unnecessarily brutal and makes experimentation unduly
difficult for users.

Test Plan: Validate

Reviewers: austin, simonmar

Subscribers: thomie

Differential Revision: https://phabricator.haskell.org/D2335

GHC Trac Issues: #7679, #8657

(cherry picked from commit 6a5d13c4ade5bbb84873970065a1acd1546f6c31)

diff --git a/compiler/nativeGen/AsmCodeGen.hs b/compiler/nativeGen/AsmCodeGen.hs
index 2285d94..9e90c43 100644 (file)
--- a/compiler/nativeGen/AsmCodeGen.hs
+++ b/compiler/nativeGen/AsmCodeGen.hs
@@ -538,10 +538,8 @@ cmmNativeGen dflags this_mod modLoc ncgImpl us fileIds dbgMap cmm count
 
         -- allocate registers
         (alloced, usAlloc, ppr_raStatsColor, ppr_raStatsLinear) <-
-         if False
-           -- Disabled, see #7679, #8657
-           --  ( gopt Opt_RegsGraph dflags
-           --  || gopt Opt_RegsIterative dflags)
+         if ( gopt Opt_RegsGraph dflags
+           || gopt Opt_RegsIterative dflags )
           then do
                 -- the regs usable for allocation
                 let (alloc_regs :: UniqFM (UniqSet RealReg))

diff --git a/docs/users_guide/using-optimisation.rst b/docs/users_guide/using-optimisation.rst
index 5e4995d..7047f4c 100644 (file)
--- a/docs/users_guide/using-optimisation.rst
+++ b/docs/users_guide/using-optimisation.rst
@@ -412,20 +412,26 @@ list.
 
 .. ghc-flag:: -fregs-graph
 
-    *Off by default due to a performance regression bug. Only applies in
-    combination with the native code generator.* Use the graph colouring
-    register allocator for register allocation in the native code
-    generator. By default, GHC uses a simpler, faster linear register
-    allocator. The downside being that the linear register allocator
-    usually generates worse code.
+    :default: off due to a performance regression bug (:ghc-ticket:`7679`)
+
+    *Only applies in combination with the native code generator.* Use the graph
+    colouring register allocator for register allocation in the native code
+    generator. By default, GHC uses a simpler, faster linear register allocator.
+    The downside being that the linear register allocator usually generates
+    worse code.
+
+    Note that the graph colouring allocator is a bit experimental and may fail
+    when faced with code with high register pressure :ghc-ticket:`8657`.
 
 .. ghc-flag:: -fregs-iterative
 
-    *Off by default, only applies in combination with the native code
-    generator.* Use the iterative coalescing graph colouring register
-    allocator for register allocation in the native code generator. This
-    is the same register allocator as the ``-fregs-graph`` one but also
-    enables iterative coalescing during register allocation.
+    :default: off
+
+    *Only applies in combination with the native code generator.* Use the
+    iterative coalescing graph colouring register allocator for register
+    allocation in the native code generator. This is the same register allocator
+    as the :ghc-flag:`-fregs-graph` one but also enables iterative coalescing
+    during register allocation.
 
 .. ghc-flag:: -fsimplifier-phases=<n>