On x86: enable by default, expensive-when-needed instrumentation of Add32.

This reduces Memcheck's false-positive level a lot on clang -O2 generated
code.

diff --git a/memcheck/mc_translate.c b/memcheck/mc_translate.c
index 666719160092d44d9e054e0fffb9ef0949e29aa4..5ed39ae21e8ea6b428f318921f53f753038bf6ee 100644 (file)
--- a/memcheck/mc_translate.c
+++ b/memcheck/mc_translate.c
@@ -8104,6 +8104,7 @@ IRSB* MC_(instrument) ( VgCallbackClosure* closure,
          are used only to construct memory addresses, which is an
          approximation to the above, and is self-contained.*/
 #     if defined(VGA_x86)
+      mce.dlbo.dl_Add32           = DLauto;
       mce.dlbo.dl_CmpEQ32_CmpNE32 = DLexpensive;
 #     elif defined(VGA_amd64)
       mce.dlbo.dl_Add64           = DLauto;