Each time myth frontend crashes (or you simply exit without reboot), duplicate evilwm processes are created, each using a signifigant amount of CPU. It appears that the .xinitrc file daemonizes evilwm, which is not killed when X restarts. There are a couple of possible solutions I am looking into. Reproducible: Always Steps to Reproduce: 1. Let mythfrontend auto-launch. 2. look at the output of ps -A | grep evilwm. 3. Exit mythfrontend without a reboot, mythfrontend restarts. 4. look at the output of ps -A | grep evilwm. Actual Results: There are twice as many evilwm processes as before. Expected Results: There should not be duplicate evilwm processes.
I just happened to come across this bug while researching a similar problem. When X11 exits, all existing window managers should shut down automatically. However, I was seeing evilwm spinning, chewing up 100% of the CPU time. Upgrading to evilwm 1.0.1 resolved this issue for me. The changelog for evilwm has the following for 1.0.1: 2009-03-30: Released 1.0.1. Fixes a problem while quitting with recent versions of X.org. http://www.6809.org.uk/evilwm/ Please note that evilwm is currently marked as ~x86 / ~amd86 . Given the fact that this is causing people problems, I would suggest that this bug be refocused on marking evilwm 1.0.1 as stable. Of course, it's possible for the mythtv gentoo team to work around this issue, but fundamentally, I don't see anything that mythtv is doing wrong...
Consider resolving as duplicate of bug 285141 ?
evilwm 1.0.1 has been marked as stable on x86 / amd64. (bug 285141) Perhaps the ebuild for mythtv should have a dependency on evilwm>=1.0.1
I spoke to lack about this a while back to get 1.0.1 stable which he's taken care of. When the last of the arch teams catch up, he'll remove 1.0.0 from the tree. Its tough to really say that mythtv should depend on 1.0.1 since its really an interaction between a new X.org stack and evilwm using some old code. I think the reality of it is that 1.0.0 should just die from the tree.