[Openais] cpg valgrind error.

[Andrew Beekhof]

On Tue, Sep 2, 2008 at 20:59, Steven Dake <sdake at redhat.com> wrote:
> Alan,
>
> Personally I would ignore those errors relating to sendmsg having an
> uninitialized variable in the iovector.  I think it is just a bug with
> valgrind or something innocuous.  I am not sure what causes it, but I am
> aware they are there.  If you have a suppression file or have a patch to
> remove the valgrind warning I'll happily merge either.

I finally tracked this one down (I tinkered with a 20 line C file
until it went away).
Basically the error is spot on and there are a number of fields in the
msghdr struct that have not been initialized.

The good news is that the final version of the new ipc code in
whitetank initializes the complete struct (I just checked) and so we
shouldn't see these again.

> On Tue, 2008-09-02 at 15:21 -0400, Alan Conway wrote:
>> Is this a known suppressable valgrind error from cpg?
>>
>> ==27614== Syscall param socketcall.sendmsg(msg.msg_iov[i]) points to
>> uninitialised byte(s)
>> ==27614==    at 0x3B6DED289B: (within /lib64/libc-2.5.so)
>> ==27614==    by 0x511B3B2: (within /usr/lib64/openais/libcpg.so.2.0.0)
>> ==27614==    by 0x511B6C9: (within /usr/lib64/openais/libcpg.so.2.0.0)
>> ==27614==    by 0x511C178: cpg_initialize
>> (in /usr/lib64/openais/libcpg.so.2.0.0)
>> ==27614==    by 0x4ED7152:
>> qpid::cluster::Cpg::Cpg(qpid::cluster::Cpg::Handler&) (Cpg.cpp:73)
>>
>>
>> Its happening in cpg_initialize, so its hard to see what could have been
>> left un-initialized in my code at that point.