Project Openings

[#419] potential memory leak in discovery

Date:
2015-08-17 14:38

Priority:
3

State:
Open

Submitted by:
Antoine Hincelin (ahincelin)

Assigned to:
Stéphane ROUGES (srouges)

Target Fix Version:
none

Product:
DPWS Core

Operating System:
All

Component:
none

Version:
v2.5.1

Severity:
normal

Resolution:
Fixed

Hardware:
none

URL:

Summary:
potential memory leak in discovery

Detailed description
Under some circumstances, handles attached to discovery memory are not all released, which creates a memory leak when calling deleteHandlePool() cf DPWS.log below handles of type 32 (BYE msg) are created 2 times, but there are deleted only once. It seems that reactor didn't have time to execute last item: send_bye_callback. (delayed item?), which should release the handle. In deleteHandlePool(), maybe DC_FREE should be applied to handle pool as well, not only to recycling pool?
Under some circumstances, handles attached to discovery memory are not all released, which creates a memory leak when calling deleteHandlePool() cf DPWS.log below handles of type 32 (BYE msg) are created 2 times, but there are deleted only once. It seems that reactor didn't have time to execute last item: send_bye_callback. (delayed item?), which should release the handle. In deleteHandlePool(), maybe DC_FREE should be applied to handle pool as well, not only to recycling pool?

Comments
Attachments
Changes

Message
Date: 2015-08-31 15:58 Sender: Antoine Hincelin Thanks for the fix Stéphane. Best regards, Antoine
Date: 2015-08-31 12:36 Sender: Stéphane ROUGES Hi Antoine, I posted a fix. I did detect a memory leak in the handle pool when stopping a sample. The fix seems to fix it. Hope you will be able to check it in your use-case. Best regards, Stéphane
Date: 2015-08-31 12:27 Sender: Antoine Hincelin Hi Stéphane, Here is the shutdown sequence: dpws_stop_server(200); dpws_server_shutdown(NULL); dpws_shutdown(); Regards, Antoine
Date: 2015-08-31 10:23 Sender: Stéphane ROUGES Hi Antoine, What you say makes sense. To be sure, I guess you are stopping the stack with a sequence dpws_stop_server_XXX() (with a timeout or maybe none) + dpws_shutdown() ? Regards, Stéphane

Attachments:

Size	Name	Date	By	Download
31 KiB	DPWS.log	2015-08-17 14:38	ahincelin	DPWS.log

Field	Old Value	Date	By
Resolution	None	2015-08-31 12:36	srouges
assigned_to	none	2015-08-31 10:23	srouges
File Added	69: DPWS.log	2015-08-17 14:38	ahincelin