Grzegorz Nosek wrote: > 2008/5/17 Daniel Hokka Zakrisson <daniel@hozac.com>: >> Grzegorz Nosek wrote: >>> Hi all, >>> >>> (sorry if you receive this message twice) >>> >>> I've been experiencing some weird hangs (boom and it's dead, no panic, >>> not even a softlockup or lockdep warning). It happens about once a >>> month >>> (of course, in production only), usually under some load (I suspected >>> I/O but it might be simple CPU usage too). It happened on several very >>> different machines (2-way pentium3, 4-way opteron 270). >>> >>> After enabling the nmi watchdog I could trace it back to schedule(), or >>> at least the nmi watchdog felt the need to kill the machine right in >>> the >>> middle of a schedule() call. >> >> So I guess it can happen on vanilla kernels as well... This was noticed >> and fixed on PlanetLab, and the patches are already fed upstream (see >> http://vserver.13thfloor.at/Experimental/delta-sched-fix0{4,5,6}.diff), >> to >> be included in the next release. > > The fix06.diff patch looks like a proper fix, although (guessing) it > could still lock the scheduler when all contexts are paused (is this > even possible? probably not). Paused processes are placed on the hold queue. > I'm not really sure how it can happen on vanilla kernels, as it's a > loop added by the vserver patch (both the label and the goto). Care to > enlighten me? :) Vanilla referred to vanilla Linux-VServer, i.e. without the PlanetLab patches. -- Daniel Hokka Zakrisson