Greedy allocator: perf work

src/disassembler.cc

Index: src/disassembler.cc
diff --git a/src/disassembler.cc b/src/disassembler.cc
index fbdda546465f190fcd3924ddcb4002739cdacc67..1c2e138b8cb91f7c5491b878d5a2fc5e4f223c8a 100644
--- a/src/disassembler.cc
+++ b/src/disassembler.cc
@@ -146,7 +146,7 @@ static int DecodeIt(Isolate* isolate, std::ostream* os,
     }
 
     // Instruction address and instruction offset.
-    out.AddFormatted("%p  %4d  ", prev_pc, prev_pc - begin);
+    out.AddFormatted("%p  %4X  ", prev_pc, prev_pc - begin);

[Jarin, 2015/06/03 15:06:16]

Why did you change this? 

I am worried that tools relying on formatting here might be unhappy.

[Mircea Trofin, 2015/06/04 03:38:39]

"perf" (the tool) outputs offsets as hex, so this makes perf analysis much
easier. I share your worry though. I'll change to having a flag perhaps?

[Jarin, 2015/06/08 13:32:49]

Maybe use this only locally, or change the formatting in a separate change list.
It looks like your change does not format jump targets consistently, so we
should not really commit it as is.

 
     // Instruction.
     out.AddFormatted("%s", decode_buffer.start());