|
From: <sv...@va...> - 2005-11-10 12:41:11
|
Author: njn
Date: 2005-11-10 03:40:36 +0000 (Thu, 10 Nov 2005)
New Revision: 5064
Log:
need --trace-children when self-hosting
Modified:
trunk/README_DEVELOPERS
Modified: trunk/README_DEVELOPERS
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
--- trunk/README_DEVELOPERS 2005-11-10 03:38:56 UTC (rev 5063)
+++ trunk/README_DEVELOPERS 2005-11-10 03:40:36 UTC (rev 5064)
@@ -66,18 +66,20 @@
=20
(4) Choose a very simple program (date) and try
=20
- outer/.../bin/valgrind --weird-hacks=3Denable-outer \
+ outer/.../bin/valgrind --weird-hacks=3Denable-outer --trace-children=
=3Dyes \
--tool=3Dcachegrind -v inner/.../bin/valgrind --tool=3Dnone -v pr=
og
=20
-It's fragile, confusing and slow, but it does work well enough for
-you to get some useful performance data. The inner Valgrind has most of
-its output (ie. those lines beginning with "=3D=3D<pid>=3D=3D") prefixed=
with a
-'>', which helps a lot.
+If you omit the --trace-children=3Dyes, you'll only valgrind inner's lau=
ncher
+program, not its stage2. The whole thing is fragile, confusing and slow=
,
+but it does work well enough for you to get some useful performance data=
.
+The inner Valgrind has most of its output (ie. those lines beginning wit=
h
+"=3D=3D<pid>=3D=3D") prefixed with a '>', which helps a lot.
=20
At the time of writing the allocator is not annotated with client reques=
ts
so Memcheck is not as useful as it could be. It also has not been teste=
d
much, so don't be surprised if you hit problems.
=20
+
Printing out problematic blocks
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
If you want to print out a disassembly of a particular block that
|