Add a README for Valgrind binaries.

Add a README for Valgrind binaries.

BUG=http://code.google.com/p/nativeclient/issues/detail?id=1605
TEST=None
Committed: http://code.google.com/p/nativeclient/source/detail?r=4983

[eugenis, 2011-04-12 07:28:05]

Please take another look.

[Mark Seaborn, 2011-04-13 06:26:48]

You could commit this but it could be clearer...

http://codereview.chromium.org/6816010/diff/3001/README
File README (right):

http://codereview.chromium.org/6816010/diff/3001/README#newcode10
README:10: To find out the exact revision of the binaries, see "svn log".
OK, if I do that, I find:

> r3497 | eugenis@google.com | 2010-10-19 04:30:44 -0700 (Tue, 19 Oct 2010) | 5
lines
>
> Update memcheck and tsan binaries to tsan@2519 + valgrind-variant@62+68+71.

I don't understand what "@62+68+71" means.  Does it mean r62 with r68 and r71
cherry-picked?  (That seems unlikely if it is build automatically by a
buildbot.)

It's not immediately obvious that tsan@2519 refers to the directory
http://data-race-test.googlecode.com/svn/trunk/tsan/.  It wouldn't hurt to spell
this out.

[Evgeniy Stepanov, 2011-04-13 12:48:20]

http://codereview.chromium.org/6816010/diff/3001/README
File README (right):

http://codereview.chromium.org/6816010/diff/3001/README#newcode10
README:10: To find out the exact revision of the binaries, see "svn log".
On 2011/04/13 06:26:48, Mark Seaborn wrote:
> OK, if I do that, I find:
> 
> > r3497 | mailto:eugenis@google.com | 2010-10-19 04:30:44 -0700 (Tue, 19 Oct
2010) | 5
> lines
> >
> > Update memcheck and tsan binaries to tsan@2519 + valgrind-variant@62+68+71.
> 
> I don't understand what "@62+68+71" means.  Does it mean r62 with r68 and r71
> cherry-picked?  (That seems unlikely if it is build automatically by a
> buildbot.)

Yes, the current binary was built manually with a couple of cherry-picked
changes. It should not happen in the future. Newest binaries from the buildbot
should be fine.

> It's not immediately obvious that tsan@2519 refers to the directory
> http://data-race-test.googlecode.com/svn/trunk/tsan/.  It wouldn't hurt to
spell
> this out.

Yes, ideally commit messages should look like "Update memcheck and tsan binaries
to http://data-race-test.googlecode.com/svn/trunk@rXXXX." Valgrind revision is
pinned in third_party/update_valgrind.sh script.

Do you want me to fix the comment by changing SVN history (not sure if that's
allowed in NaCl)?

I plan to work on NaCl/DSO support in Valgrind and TSan in Q2. I'll update the
binaries with a proper comment then.

[eugenis, 2011-04-21 08:27:40]

On 2011/04/13 12:48:20, Evgeniy Stepanov wrote:
> http://codereview.chromium.org/6816010/diff/3001/README
> File README (right):
> 
> http://codereview.chromium.org/6816010/diff/3001/README#newcode10
> README:10: To find out the exact revision of the binaries, see "svn log".
> On 2011/04/13 06:26:48, Mark Seaborn wrote:
> > OK, if I do that, I find:
> > 
> > > r3497 | mailto:eugenis@google.com | 2010-10-19 04:30:44 -0700 (Tue, 19 Oct
> 2010) | 5
> > lines
> > >
> > > Update memcheck and tsan binaries to tsan@2519 +
valgrind-variant@62+68+71.
> > 
> > I don't understand what "@62+68+71" means.  Does it mean r62 with r68 and
r71
> > cherry-picked?  (That seems unlikely if it is build automatically by a
> > buildbot.)
> 
> Yes, the current binary was built manually with a couple of cherry-picked
> changes. It should not happen in the future. Newest binaries from the buildbot
> should be fine.
> 
> > It's not immediately obvious that tsan@2519 refers to the directory
> > http://data-race-test.googlecode.com/svn/trunk/tsan/.  It wouldn't hurt to
> spell
> > this out.
> 
> Yes, ideally commit messages should look like "Update memcheck and tsan
binaries
> to http://data-race-test.googlecode.com/svn/trunk%40rXXXX.%22 Valgrind
revision is
> pinned in third_party/update_valgrind.sh script.
> 
> Do you want me to fix the comment by changing SVN history (not sure if that's
> allowed in NaCl)?
> 
> I plan to work on NaCl/DSO support in Valgrind and TSan in Q2. I'll update the
> binaries with a proper comment then.

Ping. Does it LGTY?

[pasko-google - do not use, 2011-04-22 12:55:54]

LGTM

Less detective work is required to find out sources, by svn-blaming the file
this discussion would be found.

Optional: put a TODO(eugenis): distribute the memcheck binary without having to
perform tricky cherry-picks.