hmm, on Sun, Nov 29, 2009 at 05:43:28PM +0000, Edd Barrett said that > On Sun, Nov 29, 2009 at 12:41:41PM -0500, Brad wrote: > > On Sunday 29 November 2009 12:29:20 Edd Barrett wrote: > > > On Sun, Nov 29, 2009 at 12:41:41AM +0100, frantisek holop wrote: > > > > $ yt http://www.youtube.com/watch?v=0QRO3gKj3qw > > > > $ mplayer ___youtube_____-_larry_predicts_chrome_os___.mp4 > > > > > > > > either crashes, or reboots openbsd. > > > > can anybody reproduce this? > > > > > Track down and resolve the issue with OpenBSD first. > > Perhaps ktrace might give some info?
apart from the rtorrent/mmap issue, i can't remember openbsd being ever rebooted by a user space process... the whole mplayer issue is fishy. it has worked for years without any problems. i think that the kern.shminfo.shmall=32768 is just delaying/hiding the issues here. during my multiple tries at playing that movie, there is one new scenario: the whole xsession stops, and xdm comes up again, exactly as if pressing ctrl+alt+backspace -- killing the X process. i am still guessing this must be a hw issue... but none of the other eeepc owners report anything like this. (although it took over a year for the others to confirm that the usb emulated sd card reader cannot write to the card actually, so it's not good metrics).. i saw some commits to the intel driver so i'll try a more recent snapshot again soon.. i will also try ktrace and an mplayer svn build asap, but since i have the sysctl setting, this was the first movie that misbehaved so it's not easy to reproduce. -f -- i thank my lucky stars i'm not superstitious.