Comment 10 for bug 1466273

Here is what smemstat reports for one of the spinning processes:

25375 0.0 B 11.0 M 12.5 M 19.1 M phablet /usr/lib/arm-linux-gnueabihf/thumbnailer/vs-thumb

That's while the process is eating around 110% CPU. So, swap doesn't seem to be the issue.

After running kill -9 25375, top continues to show the process as running, still using 110% CPU, and the wait in the script doesn't return. However, running smemstat again after sending the SIGKILL no longer produces any output. It looks like smemstat thinks the process isn't there anymore (even though top continues to show it).

Then, 30 seconds after having sent the kill, the process finally terminates.