* Oliver <oli...@8.c.9.b.0.7.4.0.1.0.0.2.ip6.arpa> [07.01.2013 21:46]:
> > this is not the idea: > > face the fact, that a kernel can panic seldom during boot, > > even before preinit, even if the kernel is good. > > So if the kernel is good, evidently something else isn't... If you're > shipping > images where things are so messed up that the kernel will panic before it > even > gets to executing init, I'd look to fix the underlying symptoms rather than > simply trying to work around it with this. please read again. even a good kernel can stop working/oops + hang. simply reboot your device several tousand times and wait. it will occour. ofcourse there is no reboot-loop, because on next boot there will be no oops again. if you have e.g. a wireless driver which is oopsing, it's not worth shipping it 8-) and developers can switch off via menuconfig this behavior. > As for panic on OOM... again, bad idea - let the userspace tasks get killed > and ensure you've got a watcher daemon that will spin it back up. have seen a lot of oom-killer-action: ofcourse murphy will be with you and kills the daemon AND the watcher for the daemon, so again you have a box which is...unreachable. if you ran out of memory that something is completely wrong. (nice read: http://www.landley.net/notes.html#01-01-2013) bye, bastian. PS: here some nice oom-kills: http://intercity-vpn.de/crashlog/?id=1357116422 http://intercity-vpn.de/crashlog/?id=1357528406 http://intercity-vpn.de/crashlog/?id=1357355934 http://intercity-vpn.de/crashlog/?id=1357355632 http://intercity-vpn.de/crashlog/?id=1357355452 _______________________________________________ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/mailman/listinfo/openwrt-devel
