SACSegmentation pointer handling issue

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

SACSegmentation pointer handling issue

koma
Hi all,

So I'm compiling the plane segmentation code verbatim (
http://www.pointclouds.org/documentation/tutorials/planar_segmentation.php#planar-segmentation
) and running into some REALLY weird runtime issues. It seems like either
PCL or the way PCL uses Eigen is causing memory to be free'd multiple times.
Parsing through the code I found it difficult to figure out exactly why this
is happening, so hoping some more experienced developers could help.

I've also noticed that this issue only happens on some of my computers, even
though I'm running the exact same build.

Boost 1.54
Eigen 3.2
VTK 5.10.1
PCL 1.8

Another peculiar thing is that when I'm using the vanilla plane segmentation
example code, the g++ compiler flags make a difference. When I compile
without optimization ( -O0 ), the issue occurs. But when I use ( -O3 ), the
example code works.

It's also the case that when using ( -O3 ), if I malloc another pointcloud,
the code will have the same issues (regardless of whether or not I use the
pointcloud for anything). Simply adding something like



will replicate the issue for me.

I've uploaded the Valgring memory leak trace which shows the double free
happening somewhere.
debug.txt <http://www.pcl-users.org/file/t499162/debug.txt>  

Here's the bt (NOTE, I enabled L_DEBUG for PCL):



Any ideas?




--
Sent from: http://www.pcl-users.org/
_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: SACSegmentation pointer handling issue

koma
Sorry for the second post, but it seems like my backtrace didn't send. I've
attached it here as well
backtrace.txt <http://www.pcl-users.org/file/t499162/backtrace.txt>  




--
Sent from: http://www.pcl-users.org/
_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: SACSegmentation pointer handling issue

koma
I believe I found the issue. While using the latest Eigen (3.3) I would have
this issue, but after reverting to Eigen 3.2 it's been solved. Does anybody
have this working with Eigen 3.3?



--
Sent from: http://www.pcl-users.org/
_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users