Re: [Valgrind-users] non-terminating processes

[Nicholas N. <nj...@cs...>]

On Tue, 23 Jan 2007, Tom Hughes wrote:

>> I am using Valgrind 3.0.1 (Red Hat 9.0).
>>
>> Anyway, I am doing a --leak-check=full on my valgrind command which,
>> when the process exits, does a leak check.  Problem is, the process I am
>> running valgrind on is like a system process that pretty much is in an endless
>> loop, waiting for messages and processing them.  It never terminates.
>>
>> I know in purify if you killed the process with a certain signal
>> (like -15 but not -9) purify would terminate the process and do all
>> the leak checking.  However, this does not seem to work with
>> valgrind.
>
> I think you will find that if you use a less prehistoric version
> of valgrind then it will do a leak check when you kill it.

3.2.2 was released yesterday, you might like to try that.

Also, you can embed client requests in your code to do leak checking at 
certain points without terminating the program. I think the relevant request 
is called VALGRIND_DO_LEAK_CHECK -- see the Memcheck section of the manual.

Nick