[PATCH] D20578: [esan|wset] Iterate all memory to compute the total working set

[Derek Bruening via llvm-commits]

bruening added inline comments.

================
Comment at: lib/esan/working_set.cpp:35
@@ +34,3 @@
+// Our shadow memory assumes that the line size is 64.
+static const u32 CacheLineSize = 64;
+
----------------
filcab wrote:
> Should this be in the header?
We can move it to the header once another file needs it.

================
Comment at: lib/esan/working_set.cpp:82
@@ -77,1 +81,3 @@
 
+// This routine will word-align ShadowStart and ShadowEnd prior to scanning.
+static u32 countAndClearShadowValues(u32 BitIdx, uptr ShadowStart,
----------------
filcab wrote:
> We should probably mention it might count and clear a few bytes more than `[ShadowStart,ShadowEnd)`
That is what the word-align comment implies.


http://reviews.llvm.org/D20578