On Wed, 2015-10-14 at 17:54 +0200, Dario Faggioli wrote: So, about this:
> Note that it is safe to get rid of the locking in > schedule_cpu_switch() as the pCPU being switched is, at > the time of the switch, not a valid member of any cpupool, > so no scheduling event should be expected on it, locking > or not. > I changed my mind about removing this locking. In fact... > @@ -1509,8 +1507,6 @@ int schedule_cpu_switch(unsigned int cpu, > struct cpupool *c) > return -ENOMEM; > } > > - lock = pcpu_schedule_lock_irqsave(cpu, &flags); > - > SCHED_OP(old_ops, tick_suspend, cpu); > vpriv_old = idle->sched_priv; > idle->sched_priv = vpriv; > @@ -1520,8 +1516,6 @@ int schedule_cpu_switch(unsigned int cpu, > struct cpupool *c) > SCHED_OP(new_ops, tick_resume, cpu); > SCHED_OP(new_ops, insert_vcpu, idle); > > - pcpu_schedule_unlock_irqrestore(lock, flags, cpu); > - > SCHED_OP(old_ops, free_vdata, vpriv_old); > SCHED_OP(old_ops, free_pdata, ppriv_old, cpu); > ... while reworking and testing my other series (the one about Credit2 runqueues) I'm seeing some issues (even without the patches from that series applied) that may be related to this lock not being taken any longer. Also (less strong an argument, but still), when splitting allocation and initialization of per-pCPU data (which is what that series does) I'd probably have to reinstate some similar locking again. :-/ I'll investigate more and respin this series. Thanks and Regards, Dario -- <<This happens because I choose it to happen!>> (Raistlin Majere) ----------------------------------------------------------------- Dario Faggioli, Ph.D, http://about.me/dario.faggioli Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)
signature.asc
Description: This is a digitally signed message part
_______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel
