topic Leaks in mainThread() in Operating System - HP-UX

Leaks in mainThread()

MJ_Techi — Wed, 01 Dec 2010 06:10:29 GMT

Hi,

I am new to GDB(version 6.1). I would like to know about some errors like when I am finding memory leaks through gdb I am only getting the results like:-

20 bytes leaked in 1 blocks
No. Total bytes Blocks Address Function
0 20 1 0x4106c310 mainThread()

And :-

20 bytes leaked at 0x4106c310 (100.00% of all bytes leaked)
#0 0x200000007e8280d2 in mainThread() at ../cm_main.c:4258
#1 0x04000bc2 in main() at cm.cpp:22
#2 0x60000000c00427b2 in main_opd_entry() from /usr/lib/hpux32/dld.so

which I think cannot be possible or may be it is not finding the leaks at all and stuck at some point.

If anybody come across such issues please help me out.

thanks
MJ

Re: Leaks in mainThread()

Dennis Handly — Wed, 01 Dec 2010 19:17:03 GMT

What is is doing on line 4258 of cm_main.c?

Do you ever free that space?

Re: Leaks in mainThread()

MJ_Techi — Thu, 02 Dec 2010 10:41:04 GMT

cm is an executable. we do not have the source code for cm_main.c.

But we have shared object(.so) files linked to cm.

These .so files contain custom API's

These API are invoked using custom workbenches(you can say dlopen)

We are able to successfully set the breakpoints within this API.

Sometimes the leaks are captured within this custom API's

But sometimes it goes to the cm_main.c and captures the above leaks(100% leak in mainThread()) and does not scan the API at all

Why is it so inconsistent?

Re: Leaks in mainThread()

Dennis Handly — Fri, 03 Dec 2010 07:19:10 GMT

>But sometimes it goes to the cm_main.c and captures the above leaks (100% leak in mainThread()) and does not scan the API at all
Why is it so inconsistent?

I'm not sure how that can happen. The allocation site for memory should be very accurate. There should be a call to malloc at cm_main.c:4258. Perhaps sometimes this is freed by a call inside your API?