Stack trace in leak summary

Anonymous — Wed, 05 Apr 2006 12:49:12 -0000

Dear Developpers,
I used your product before but I couldn't used it.
Yesterday I faced a problem that Valgrind could not
report some memory leaks in application. I tried
several leak detectors in Internet but none of them
could work with our application too. I tried pre-load
feature of google-perftools but It didn't work neather.

I changed the way I worked with perftools, I wrote a
single thread application and added HeapLeakChecker to
it. Thank you very much, I resolved several memory
leaks with help of reports. However the main memory
leak exists .

There is a memory allocator function in one of my 3rd
party libraries that caused leakage. there are really
very much functions in 3rd party library that called
this memory allocator function and pprof reports only
this function to me. For example, if it reported me
stack traces of memory allocations I could fix it very
fast but I could not fix it now (because the code is
really complicated and in none way I reach good results).

I saw stack trace libraries in perftools, is there any
way for me to get stack trace of memory allocations?

Thanks in advance
* Roozbeh (roozbeh@ipronto.net)

CPUPROFILE on larger runs, unreadable output file

DonWebber — Thu, 31 Mar 2005 21:25:46 -0000

I'm impressed with these profilers. They will help a
lot with
my analysis and visualization of the execution.

I'm having a problem when I tried to run a larger example.
The profile output file was not readable by pprof. It's
got non-ascii characters in it, so clearly something very
bad has happened to it.

This example takes about 5 minutes, and 700 MBytes.
Is this a problem size that you have run before ?
I've had no problem running smaller examples.

CPUPROFILE worked fine on smaller examples.

HEAPPROFILE worked fine on all of these examples.

I'm running Red Hat 9, gcc 3.4.3 .

Recent changes to support-requests

Stack trace in leak summary

CPUPROFILE on larger runs, unreadable output file