Greetings,

I have a small C program that creates a window (Using GTK+) and uses that window to play files using gstreamer. It plays some videos, and puts images between videos for a few seconds. Kind of media player, but very simple. Since I wanted it to be respawned if it failed (If some video is not found, for the moment, C program simply fails, respawns, and play the next file -Plays a list of movies-) I put the C program inside a while in a bash script. The bash script calls the C program, and when this exits, waits 5 seconds and respawns the program. Usually, this works like a charm. But I have found a problem, since with some videos I get some errors with MemMgr_Free. This seems to be a sign of memory leakage (Sorry if I am being terribly wrong, still a newbie with just windows experience). Well, point is, I tried to kill the process, since with this fault, program does not end, just freeze. But when I kill the program after this has happened, it stays "zombie" (Seems to be the usual word for defunct processes). Googling it points to the father as the one to blame, but... Since it works when C program behaves normally... I am not sure.

The machine running this program is far away, so I can only access via SSH. Any hint, pointers, anything else? As I have already said, I am just a windows newbie, and electronics engineer, not I.T. guy, so this scares me a lot. Specially when involving zombies

Oh, setup of the system:

Hardware: Pandaboard (I know that I should ask there, but does not seem to me as hardware-related)
Software: Ubuntu 11.10 (Kernel 3.1.0)

Lastly, cat /proc/meminfo shows "MemFree: 289664 kB", so the memory leakage (If it's that) has not left the system without any memory.

As with my other posts, please forgive the mistakes in the written text. Feel free to point anything that is wrongly said, as I am not english speaker, but always want to improve my skills.

Thanks in advance!

usually it means your c code has died during a kernel call. So you need to kill the kernel to clean it up (that means a reboot). I can suggest you only to update/change/replace the player.

Well, the player is being written by me. So... If I understand you correctly, it is my code what causes the "defunct" status.

Since the code just creates the window, and then builds a gstreamer pipeline to play the videos, and this works ok with some videos and fails with others... May it be a codec issue? I don't know if codecs do kernel calls, but since this player uses hardware acceleration of the processor, probably this means that yes, that there are kernel calls.

no, the real player is a codec somewhere inside gstreamer. All of the programs need the kernel, because all the resources available only by kernel calls. A user space app cannot access directly the hardware. So yes, it can be a codec issue.

1 members found this post helpful.

Ok, thank you very much.

Will ask at gstreamer mail-list.

Also remember what the "zombie" status is. When a process dies, it goes into "zombie" state specifically so that its parent can rendezvous with it and collect its final status without creating a race-condition. Only then does the process-table entry disappear. (If the parent dies, the children are parented to init (pid=1), "the one process that never dies," which reaps them.) Your parent script must reap its children.