memory leak about leveldb and replicant?

[李俊良]

I tested the value over 100M, found memory leak.

        comparisons_(new uint64_t[n]),
        heap_(new unsigned[n]),

So need free them when destruct

virtual ~MergingIterator() {
    delete[] children_;
    delete[] comparisons_;
    delete[] heap_;
}

In addition, ./client/pending_cond_follow.cc:handle_response

140                 if (*m_data)
141                 {
142                     *m_data = static_cast<char*>(realloc(*m_data, data.size()));
143                 }
144                 else
145                 {
146                     *m_data = static_cast<char*>(malloc(data.size()));
147                 }
148

m_data not free, This problem can be detected from cppcheck or valgrind

[Robert Escriva]

Thank you for catching these!

I've updated the code in Git with the appropriate fixes.

-Robert

> --
> You received this message because you are subscribed to the Google Groups
> "hyperdex-discuss" group.
> To unsubscribe from this group and stop receiving emails from it, send an email
> to hyperdex-discu...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[李俊良]

m_data is char ** ,  it should free *m_data

在 2016年2月17日星期三 UTC+8上午12:53:17，Robert Escriva写道：

[Robert Escriva]

Good catch! Thank you.

-Robert

[李俊良]

the pending of pending_cond_follow in client.cc could be inserted into map, maybe the best way to explicitly release

在 2016年2月17日星期三 UTC+8上午12:53:17，Robert Escriva写道：

Thank you for catching these!

[Robert Escriva]

I'm currently working on running valgrind behind every test in the
Replicant suite, and some tests that will exercise this and other paths.
I don't want to keep playing whack-a-mole.

-Robert

[李俊良]

how to run gremlin test? :)

在 2016年2月17日星期三 UTC+8上午11:30:02，Robert Escriva写道：

[Robert Escriva]

I've pushed that go GitHub here: https://github.com/rescrv/gremlin

Add the "gremlin" file in that repository to your path and make sure you
have Python 3 installed. Then everything should work with "make check".

To run additional tests, you should run configure like this:

./configure --enable-example-state-machines

I'm still cleaning up the valgrind errors in Replicant, so feel free to
look around, but be aware I'll be changing things for the better. :-)

-Robert

[李俊良]

thanks :)

在 2016年2月17日星期三 UTC+8下午9:50:12，Robert Escriva写道：