Improve accuracy of get/setAllocationCounter

Summary:
get/setAllocationCounter didn't take into account allocations in the
current block. This was known at the time, but it turns out to be
important to have more accuracy when using these in a fine-grained
way.

Test Plan:
New unit test to test incrementally larger allocaitons.  Before I got
results like this:

```
+0
+0
+0
+0
+0
+4096
+0
+0
+0
+0
+0
+4064
+0
+0
+4088
+4056
+0
+0
+0
+4088
+4096
+4056
+4096
```

Notice how the results aren't always monotonically increasing.  After
this patch:

```
+344
+416
+488
+560
+632
+704
+776
+848
+920
+992
+1064
+1136
+1208
+1280
+1352
+1424
+1496
+1568
+1640
+1712
+1784
+1856
+1928
+2000
+2072
+2144
```

Reviewers: hvr, erikd, simonmar, jrtc27, trommler

Reviewed By: simonmar

Subscribers: trommler, jrtc27, rwbarton, thomie, carter

Differential Revision: https://phabricator.haskell.org/D4363

3 files changed, 5 insertions, 2 deletions

diff --git a/includes/Cmm.h b/includes/Cmm.h
index 57d78ccaa5..18b2aaf324 100644
--- a/includes/Cmm.h
+++ b/includes/Cmm.h
@@ -161,9 +161,11 @@
 
 /* TO_W_(n) converts n to W_ type from a smaller type */
 #if SIZEOF_W == 4
+#define TO_I64(x) %sx64(x)
 #define TO_W_(x) %sx32(x)
 #define HALF_W_(x) %lobits16(x)
 #elif SIZEOF_W == 8
+#define TO_I64(x) (x)
 #define TO_W_(x) %sx64(x)
 #define HALF_W_(x) %lobits32(x)
 #endif
diff --git a/includes/rts/Threads.h b/includes/rts/Threads.h
index fceacdc75d..f72f5ed121 100644
--- a/includes/rts/Threads.h
+++ b/includes/rts/Threads.h
@@ -43,8 +43,6 @@ StgRegTable * resumeThread  (void *);
 //
 int     cmp_thread                       (StgPtr tso1, StgPtr tso2);
 int     rts_getThreadId                  (StgPtr tso);
-HsInt64 rts_getThreadAllocationCounter   (StgPtr tso);
-void    rts_setThreadAllocationCounter   (StgPtr tso, HsInt64 i);
 void    rts_enableThreadAllocationLimit  (StgPtr tso);
 void    rts_disableThreadAllocationLimit (StgPtr tso);
 
diff --git a/includes/stg/MiscClosures.h b/includes/stg/MiscClosures.h
index 76cfbd6c8c..1fbfab9fbe 100644
--- a/includes/stg/MiscClosures.h
+++ b/includes/stg/MiscClosures.h
@@ -468,6 +468,9 @@ RTS_FUN_DECL(stg_traceCcszh);
 RTS_FUN_DECL(stg_clearCCSzh);
 RTS_FUN_DECL(stg_traceEventzh);
 RTS_FUN_DECL(stg_traceMarkerzh);
+RTS_FUN_DECL(stg_getThreadAllocationCounterzh);
+RTS_FUN_DECL(stg_setThreadAllocationCounterzh);
+
 
 /* Other misc stuff */
 // See wiki:Commentary/Compiler/Backends/PprC#Prototypes