On Fri, Jul 31, 2015 at 7:13 PM, Heikki Linnakangas <hlinnaka@iki.fi> wrote:
> On 07/31/2015 11:05 AM, Haribabu Kommi wrote:
>>>
>>> Thanks for the defect details.
>>> I am able to reproduce the issue on Head.
>>>
>>> This is because of many unnecessary "Expr Context" that are created in
>>> function "domain_check_input"
>>> using CreateStandaloneExprContext function under TopMemoryContext.
>>>
>>> This memory context is reset for future use by storing the context
>>> pointer
>>> in "fcinfo->flinfo->fn_extra".
>>> But this pointer always set as NULL for every call. Because of this
>>> reason
>>> the memory was leaked.
>>
>>
>> Adding to this point, In "PLyObject_ToComposite" function the context
>> pointer
>> which was allocated earlier is freed without removing the context.
>> Because of this
>> reason for the next record, it again allocates the context and thus it
>> leading to a
>> memory leak.
>
>
> Yep.
>
>> Instead of reset the context in "domain_check_input" function, we can free
>> that
>> context, which can solve this problem. But I am not sure, is it right fix?
>>
>> Instead of the above fix, either we need to teach plpython functions to
>> taking
>> care of reusing the pointer or something?
>
>
> The problem is that perm_fmgr_info() is a crock, as explained in its
> comments. That crock leads to this kind of memory leaks.
>
> I'm not sure why we use that crock. It doesn't seem hard to just use a more
> short-lived memory context. I hacked together the attached patch, which
> fixes this particular test case, but I just used TopMemoryContext in most
> other places so this doesn't plug all the leaks. But I think we want
> something like this, but using an appropriate memory context in each
> PLy_typeinfo_init call. For some call sites, I think we do need to create a
> new memory context that can be easily freed in PLy_typeinfo_dealloc().
>
> Would you like to finish up that patch?
Thanks for providing details.
I find similar perm_fmgr_info() function in both plperl and pltcl modules also.
I will check the function impact on those two modules also and finish up
the patch.
Regards,
Hari Babu
Fujitsu Australia