Hi, On 07/16/2011 02:34 AM, Peter Zijlstra wrote: > On Sat, 2011-07-16 at 09:38 +0200, Andre Heider wrote: >> Hi, >> >> when I boot a recent kernel I'm getting hangs early in the boot process. >> >> The kernel boots most of the time, but when /sbin/init kicks in it >> waits forever for something. I only get a few lines of output, mostly >> udev related. >> When the kernel does not boot, it seems to hang somewhere when mapping >> the irqs (at least that's what the last lines of ps3fb output >> suggest). >> I can run into both situation with the same kernel binary. It's also >> consistent with two userlands, I tried debian stable and testing. >> >> When this happens, I can't interact with the system, so I don't have >> much more info. >> >> I bisected this to: >> >> commit 317f394160e9beb97d19a84c39b7e5eb3d7815a8 >> Author: Peter Zijlstra <a.p.zijls...@chello.nl> >> Date: Tue Apr 5 17:23:58 2011 +0200 >> >> sched: Move the second half of ttwu() to the remote cpu >> >> All kernels including this patch only work for me when booted with 'nosmp'.
I verified that indeed 317f394160e9beb97d19a84c39b7e5eb3d7815a8 'sched: Move the second half of ttwu() to the remote cpu' introduces the hang. >> Any ideas? > > Verify 184748cc50b2dceb8287f9fb657eda48ff8fcfe7 does indeed cover your > PPC flavour. It has some ppc changes, but I could have missed PS3 if its > 'special'. I don't think PS3 is special. The IPI code is here: http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=blob;f=arch/powerpc/platforms/ps3/smp.c;hb=HEAD > Another thing to check is if your sched IPI handler calls > irq_enter()/irq_exit(), if not try that. I tried adding some these, but no change. I tried to figure out what is happening, but I can't seem to. It seems when ttwu_queue_remote() is used the pending scheduling is not performed. I can't say for sure what is happening. With the test patch below the system boots OK. Also, if I add a udbg_printf(".") statement in the body of ps3's do_message_pass() the system boots OK. Any help would be greatly appreciated. -Geoff diff --git a/kernel/sched.c b/kernel/sched.c index 9e3ede1..c16a35a 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2561,21 +2561,21 @@ static void ttwu_queue_remote(struct task_struct *p, int cpu) if (!next) smp_send_reschedule(cpu); } #endif static void ttwu_queue(struct task_struct *p, int cpu) { struct rq *rq = cpu_rq(cpu); -#if defined(CONFIG_SMP) && defined(CONFIG_SCHED_TTWU_QUEUE) +#if 0 //defined(CONFIG_SMP) && defined(CONFIG_SCHED_TTWU_QUEUE) if (sched_feat(TTWU_QUEUE) && cpu != smp_processor_id()) { ttwu_queue_remote(p, cpu); return; } #endif raw_spin_lock(&rq->lock); ttwu_do_activate(rq, p, 0); raw_spin_unlock(&rq->lock); } _______________________________________________ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
