Land change to notify valgrind when we modify code on x86. This does

not seem to impact performance and makes it a lot easier for people
embedding V8 to run their application under valgrind.

If we observe a negative performance impact, we will need to revise
this again.

Review URL: http://codereview.chromium.org/13612.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2662 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

LICENSE

@@ -20,6 +20,12 @@ are:
     copyrighted by Douglas Crockford and Baruch Even and released under
     an MIT license.
 
+  - Valgrind client API header, located at third_party/valgrind/valgrind.h
+    This is release under the BSD license.
+
+  - Valgrind client API header, located at third_party/valgrind/valgrind.h
+    This is release under the BSD license.
+
 These libraries have their own licenses; we recommend you read them,
 as their terms may differ from the terms below.
 

src/ia32/cpu-ia32.cc

@@ -27,6 +27,10 @@
 
 // CPU specific code for ia32 independent of OS goes here.
 
+#ifdef __GNUC__
+#include "third_party/valgrind/valgrind.h"
+#endif
+
 #include "v8.h"
 
 #include "cpu.h"
@@ -49,6 +53,15 @@ void CPU::FlushICache(void* start, size_t size) {
 
   // If flushing of the instruction cache becomes necessary Windows has the
   // API function FlushInstructionCache.
+
+  // By default, valgrind only checks the stack for writes that might need to
+  // invalidate already cached translated code.  This leads to random
+  // instability when code patches or moves are sometimes unnoticed.  One
+  // solution is to run valgrind with --smc-check=all, but this comes at a big
+  // performance cost.  We can notify valgrind to invalidate its cache.
+#ifdef VALGRIND_DISCARD_TRANSLATIONS
+  VALGRIND_DISCARD_TRANSLATIONS(start, size);
+#endif
 }