Pointer Tagging

  
This patch implements pointer tagging as per our ICFP'07 paper "Faster
laziness using dynamic pointer tagging".  It improves performance by
10-15% for most workloads, including GHC itself.

The original patches were by Alexey Rodriguez Yakushev
<mrchebas@gmail.com>, with additions and improvements by me.  I've
re-recorded the development as a single patch.

The basic idea is this: we use the low 2 bits of a pointer to a heap
object (3 bits on a 64-bit architecture) to encode some information
about the object pointed to.  For a constructor, we encode the "tag"
of the constructor (e.g. True vs. False), for a function closure its
arity.  This enables some decisions to be made without dereferencing
the pointer, which speeds up some common operations.  In particular it
enables us to avoid costly indirect jumps in many cases.

More information in the commentary:

http://hackage.haskell.org/trac/ghc/wiki/Commentary/Rts/HaskellExecution/PointerTagging

1 files changed, 24 insertions, 5 deletions

diff --git a/rts/StgStdThunks.cmm b/rts/StgStdThunks.cmm
index db9c254233..20ceb6aaba 100644
--- a/rts/StgStdThunks.cmm
+++ b/rts/StgStdThunks.cmm
@@ -39,10 +39,23 @@
 #define RET_PARAMS
 #endif
 
+/*
+ * TODO: On return, we can use a more efficient
+ *       untagging (we know the constructor tag).
+ * 
+ * When entering stg_sel_#_upd, we know R1 points to its closure,
+ * so it's untagged.
+ * The payload might be a thunk or a constructor,
+ * so we enter it.
+ *
+ * When returning, we know for sure it is a constructor,
+ * so we untag it before accessing the field.
+ *
+ */
 #define SELECTOR_CODE_UPD(offset) \
   INFO_TABLE_RET(stg_sel_ret_##offset##_upd, RET_SMALL, RET_PARAMS)	\
   {									\
-      R1 = StgClosure_payload(R1,offset);				\
+      R1 = StgClosure_payload(UNTAG(R1),offset);			\
       GET_SAVED_CCCS;							\
       Sp = Sp + SIZEOF_StgHeader;					\
       ENTER();								\
@@ -58,8 +71,11 @@
       ENTER_CCS_THUNK(R1);						\
       SAVE_CCCS(WITHUPD_FRAME_SIZE);					\
       W_[Sp-WITHUPD_FRAME_SIZE] = stg_sel_ret_##offset##_upd_info;	\
-      R1 = StgThunk_payload(R1,0);					\
       Sp = Sp - WITHUPD_FRAME_SIZE;					\
+      R1 = StgThunk_payload(R1,0);				        \
+      if (GETTAG(R1) != 0) {                                            \
+          jump RET_LBL(stg_sel_ret_##offset##_upd);		        \
+      }                                                                 \
       jump %GET_ENTRY(R1);						\
   }
   /* NOTE: no need to ENTER() here, we know the closure cannot evaluate to a function,
@@ -85,10 +101,10 @@ SELECTOR_CODE_UPD(15)
 #define SELECTOR_CODE_NOUPD(offset) \
   INFO_TABLE_RET(stg_sel_ret_##offset##_noupd, RET_SMALL, RET_PARAMS)	\
   {									\
-      R1 = StgClosure_payload(R1,offset);				\
+      R1 = StgClosure_payload(UNTAG(R1),offset);			\
       GET_SAVED_CCCS;							\
       Sp = Sp + SIZEOF_StgHeader;					\
-      jump %GET_ENTRY(R1);						\
+      ENTER();								\
   }									\
 									\
   INFO_TABLE_SELECTOR(stg_sel_##offset##_noupd, offset, THUNK_SELECTOR, "stg_sel_noupd", "stg_sel_noupd")\
@@ -101,8 +117,11 @@ SELECTOR_CODE_UPD(15)
       ENTER_CCS_THUNK(R1);						\
       SAVE_CCCS(NOUPD_FRAME_SIZE);					\
       W_[Sp-NOUPD_FRAME_SIZE] = stg_sel_ret_##offset##_noupd_info;	\
-      R1 = StgThunk_payload(R1,0);					\
       Sp = Sp - NOUPD_FRAME_SIZE;					\
+      R1 = StgThunk_payload(R1,0);					\
+      if (GETTAG(R1) != 0) {                                            \
+          jump RET_LBL(stg_sel_ret_##offset##_noupd);		        \
+      }                                                                 \
       jump %GET_ENTRY(R1);						\
   }