[RFC,1/2] PCI: correctly flush workqueue when destroy pcie hotplug controller

When destroying a PCIe hotplug controller, all work items associated with
that controller should be flushed.  Function pcie_cleanup_slot() calls
cancel_delayed_work() and flush_workqueue() to achieve that.
Function flush_workqueue() will flush all work items already submitted,
but new work items submitted by those already submitted work items may
still be in live state when returning from flush_workqueue().

For the extreme case, pciehp driver may expierence following calling path:
1) pcie_isr() -> pciehp_handle_xxx() -> queue_interrupt_event()->queue_work()
2) interrupt_event_handler() -> handle_button_press_event() ->
   queue_delayed_work()
3) pciehp_queue_pushbutton_work() -> queue_work()

So enhance pcie_cleanup_slot() to correctly flush workqueue when destroying
PCIe hotplug controller.

Signed-off-by: Jiang Liu <jiang.liu@huawei.com>
---
 drivers/pci/hotplug/pciehp_hpc.c |   13 ++++++++++++-
 1 files changed, 12 insertions(+), 1 deletions(-)

I think you're rignt.

Reviewed-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>

(2012/04/08 3:18), Jiang Liu wrote:
> When destroying a PCIe hotplug controller, all work items associated with
> that controller should be flushed.  Function pcie_cleanup_slot() calls
> cancel_delayed_work() and flush_workqueue() to achieve that.
> Function flush_workqueue() will flush all work items already submitted,
> but new work items submitted by those already submitted work items may
> still be in live state when returning from flush_workqueue().
> 
> For the extreme case, pciehp driver may expierence following calling path:
> 1) pcie_isr() ->  pciehp_handle_xxx() ->  queue_interrupt_event()->queue_work()
> 2) interrupt_event_handler() ->  handle_button_press_event() ->
>     queue_delayed_work()
> 3) pciehp_queue_pushbutton_work() ->  queue_work()
> 
> So enhance pcie_cleanup_slot() to correctly flush workqueue when destroying
> PCIe hotplug controller.
> 
> Signed-off-by: Jiang Liu<jiang.liu@huawei.com>
> ---
>   drivers/pci/hotplug/pciehp_hpc.c |   13 ++++++++++++-
>   1 files changed, 12 insertions(+), 1 deletions(-)
> 
> diff --git a/drivers/pci/hotplug/pciehp_hpc.c b/drivers/pci/hotplug/pciehp_hpc.c
> index d4fa705..9a48a51 100644
> --- a/drivers/pci/hotplug/pciehp_hpc.c
> +++ b/drivers/pci/hotplug/pciehp_hpc.c
> @@ -890,8 +890,19 @@ static int pcie_init_slot(struct controller *ctrl)
>   static void pcie_cleanup_slot(struct controller *ctrl)
>   {
>   	struct slot *slot = ctrl->slot;
> -	cancel_delayed_work(&slot->work);
> +
> +	/*
> +	 * Following workqueue flushing logic is to deal with the special call path:
> +	 * 1) pcie_isr() ->  pciehp_handle_xxx() ->
> +	 *    queue_interrupt_event(pciehp_wq_event)->queue_work(pciehp_wq)
> +	 * 2) interrupt_event_handler() ->  handle_button_press_event() ->
> +	 *    queue_delayed_work(pciehp_wq)
> +	 * 3) pciehp_queue_pushbutton_work() ->  queue_work(pciehp_wq)
> +	 */
>   	flush_workqueue(pciehp_wq);
> +	cancel_delayed_work_sync(&slot->work);
> +	flush_workqueue(pciehp_wq);
> +
>   	kfree(slot);
>   }
> 

--
To unsubscribe from this list: send the line "unsubscribe linux-pci" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

On Tue, Apr 10, 2012 at 4:43 AM, Kenji Kaneshige
<kaneshige.kenji@jp.fujitsu.com> wrote:
> I think you're rignt.
>
> Reviewed-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
>
> (2012/04/08 3:18), Jiang Liu wrote:
>> When destroying a PCIe hotplug controller, all work items associated with
>> that controller should be flushed.  Function pcie_cleanup_slot() calls
>> cancel_delayed_work() and flush_workqueue() to achieve that.
>> Function flush_workqueue() will flush all work items already submitted,
>> but new work items submitted by those already submitted work items may
>> still be in live state when returning from flush_workqueue().
>>
>> For the extreme case, pciehp driver may expierence following calling path:
>> 1) pcie_isr() ->  pciehp_handle_xxx() ->  queue_interrupt_event()->queue_work()
>> 2) interrupt_event_handler() ->  handle_button_press_event() ->
>>     queue_delayed_work()
>> 3) pciehp_queue_pushbutton_work() ->  queue_work()
>>
>> So enhance pcie_cleanup_slot() to correctly flush workqueue when destroying
>> PCIe hotplug controller.
>>
>> Signed-off-by: Jiang Liu<jiang.liu@huawei.com>
>> ---
>>   drivers/pci/hotplug/pciehp_hpc.c |   13 ++++++++++++-
>>   1 files changed, 12 insertions(+), 1 deletions(-)
>>
>> diff --git a/drivers/pci/hotplug/pciehp_hpc.c b/drivers/pci/hotplug/pciehp_hpc.c
>> index d4fa705..9a48a51 100644
>> --- a/drivers/pci/hotplug/pciehp_hpc.c
>> +++ b/drivers/pci/hotplug/pciehp_hpc.c
>> @@ -890,8 +890,19 @@ static int pcie_init_slot(struct controller *ctrl)
>>   static void pcie_cleanup_slot(struct controller *ctrl)
>>   {
>>       struct slot *slot = ctrl->slot;
>> -     cancel_delayed_work(&slot->work);
>> +
>> +     /*
>> +      * Following workqueue flushing logic is to deal with the special call path:
>> +      * 1) pcie_isr() ->  pciehp_handle_xxx() ->
>> +      *    queue_interrupt_event(pciehp_wq_event)->queue_work(pciehp_wq)
>> +      * 2) interrupt_event_handler() ->  handle_button_press_event() ->
>> +      *    queue_delayed_work(pciehp_wq)
>> +      * 3) pciehp_queue_pushbutton_work() ->  queue_work(pciehp_wq)
>> +      */
>>       flush_workqueue(pciehp_wq);
>> +     cancel_delayed_work_sync(&slot->work);
>> +     flush_workqueue(pciehp_wq);
>> +
>>       kfree(slot);
>>   }

Jiang, I'm ignoring this (and the next [2/2] patch) because I assume
they're superceded by the "[PATCH RFC 00/17] Introduce a global lock
to serialize all PCI hotplug" series of April 16.  Let me know if
otherwise.

Bjorn
--
To unsubscribe from this list: send the line "unsubscribe linux-pci" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html