Re: [PATCH] fix rcu vs hotplug race

[Ingo Molnar <mingo@elte.hu>]

* Dhaval Giani <dhaval@linux.vnet.ibm.com> wrote:

> On running kernel compiles in parallel with cpu hotplug,
> 
> ------------[ cut here ]------------
> WARNING: at arch/x86/kernel/smp.c:118
> native_smp_send_reschedule+0x21/0x36()
> Modules linked in:
> Pid: 27483, comm: cc1 Not tainted 2.6.26-rc7 #1
>  [<c01217d9>] warn_on_slowpath+0x41/0x5d
>  [<c01515b7>] ? generic_file_aio_read+0x10f/0x137
>  [<c0151340>] ? file_read_actor+0x0/0xf7
>  [<c013ae4c>] ? validate_chain+0xaa/0x29c
>  [<c013c854>] ? __lock_acquire+0x612/0x666
>  [<c013c854>] ? __lock_acquire+0x612/0x666
>  [<c013ae4c>] ? validate_chain+0xaa/0x29c
>  [<c01715d3>] ? file_kill+0x2d/0x30
>  [<c013cbd7>] ? __lock_release+0x4b/0x51
>  [<c01715d3>] ? file_kill+0x2d/0x30
>  [<c0110355>] native_smp_send_reschedule+0x21/0x36
>  [<c014fe8f>] force_quiescent_state+0x47/0x57
>  [<c014fef0>] call_rcu+0x51/0x6d
>  [<c01713b3>] __fput+0x130/0x158
>  [<c0171231>] fput+0x17/0x19
>  [<c016fd99>] filp_close+0x4d/0x57
>  [<c016fdff>] sys_close+0x5c/0x97
>  [<c0103861>] sysenter_past_esp+0x6a/0xb1
>  =======================
> ---[ end trace aa35f3913ddf2d06 ]---
> 
> This is because a reschedule is sent to a CPU which is offline.
> Just ensure that the CPU we send the smp_send_reschedule is actually
> online.
> 
> Signed-off-by: Dhaval Giani <dhaval@linux.vnet.ibm.com>
> ---
>  kernel/rcuclassic.c |    3 ++-
>  1 files changed, 2 insertions(+), 1 deletion(-)
> 
> Index: linux-2.6.26-rc7/kernel/rcuclassic.c
> ===================================================================
> --- linux-2.6.26-rc7.orig/kernel/rcuclassic.c
> +++ linux-2.6.26-rc7/kernel/rcuclassic.c
> @@ -93,7 +93,8 @@ static void force_quiescent_state(struct
>  		cpumask = rcp->cpumask;
>  		cpu_clear(rdp->cpu, cpumask);
>  		for_each_cpu_mask(cpu, cpumask)
> -			smp_send_reschedule(cpu);
> +			if (cpu_online(cpu))
> +				smp_send_reschedule(cpu);
>  	}

hm, not sure - we might just be fighting the symptom and we might now 
create a silent resource leak instead. Isnt a full RCU quiescent state 
forced (on all CPUs) before a CPU is cleared out of cpu_online_map? That 
way the to-be-offlined CPU should never actually show up in 
rcp->cpumask.

	Ingo