On Mon, Apr 02 2007, Marcelo Tosatti wrote: > On Mon, Apr 02, 2007 at 06:20:44PM +0200, Jens Axboe wrote: > > On Mon, Apr 02 2007, Jim Gettys wrote: > > > You won't get anyone to take that bet.... Care to work on it? The > > > people who've been working the most on providing a decent user interface > > > experience in the face of OOM are the Nokia folks; I can send you some > > > contacts. > > > > Didn't figure I would :-) And yeah, I don't mind taking a look at low > > memory and OOM situations. > > I've been working on a SIGDANGER equivalent for Linux - so that we > can signal applications once a certain lower limit of memory has been > reached. I'm currently busy with the resume work, but plan to get back > to SIGDANGER once we're finished with PM.
Cool. BTW, where do you store your resume stuff? > We need certainly need to tune the OOM killer - I've done some testing > with recent v2.6 kernels on our machines and it does behave very badly, > killing daemons which are not the memory hog in question, for example > > /proc/pid/oom_adj and oom_score need to be adjusted. Yeah, those are the most trivial bits. In my experience the OOM killer always gets it wrong, but that's largely because it's never customized. -- Jens Axboe _______________________________________________ Devel mailing list [email protected] http://mailman.laptop.org/mailman/listinfo/devel
