Re: [UniMRCP] Crash observed in uniMRCP server

[Sean DiSanti]

Just throwing this out there, are you sure it's actually a leak and not a pool of memory that's being re-used? 

On Fri, Jul 2, 2021 at 1:23 AM Vaibhav Sawke <vaibha...@gmail.com> wrote:

Hello All,

I am writing a custom Azure SR plugin for UniMRCP.

We ran Valgrind, and it showed the below leaks in the report.

SIPp load run was (r = 15 l =80 m=300 (80 concurrent calls and hit rate is 15 calls per second.)

==10157== 32,768 (14,848 direct, 17,920 indirect) bytes in 32 blocks are definitely lost in loss record 310 of 311
==10157== at 0x4C33B25: calloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==10157== by 0x59E0519: su_home_new (su_alloc.c:559)
==10157== by 0x59CB3BA: soa_clone (soa.c:286)
==10157== by 0x59A44E6: nua_stack_set_params (nua_params.c:552)
==10157== by 0x599B2AB: nua_stack_init_handle (nua_client.c:1634)
==10157== by 0x599B3C0: nua_client_init_request0 (nua_client.c:477)|
==10157== by 0x599B3C0: nua_client_init_request (nua_client.c:442)
==10157== by 0x5998392: nua_stack_signal (nua_stack.c:592)
==10157== by 0x59ED213: su_base_port_execute_msgs (su_base_port.c:280)
==10157== by 0x59ED848: su_base_port_run (su_base_port.c:343)
==10157== by 0x59EE034: su_pthread_port_clone_main (su_pthread_port.c:324)
==10157== by 0x622A6DA: start_thread (pthread_create.c:463)
==10157== by 0x53F371E: clone (clone.S:95)

We have checked the code multiple times and confirmed that streams and recognizers in the plugin are closed in every case.

Though there was no side-effect of this leak, i.e., we could run 4Lacs+ of call without any increase in RAM usage.

So the query is, should we concerned about this reported leak?¨

Note:- To reproduce the above, we had removed the cognitive service APIs and put a sleep of 2 seconds in the stream_write function of demo_recog, and Valgrind observed leaks.

-Vaibhav Sawke

--
You received this message because you are subscribed to the Google Groups "UniMRCP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to unimrcp+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/unimrcp/06f4e7e0-55f6-4aac-83d8-8b9244a28dd4n%40googlegroups.com.