Re: [PATCH RFC v2 1/3] drivers/platform/x86: add cpu physically hotplug driver

[Gu Zheng <guz.fnst@cn.fujitsu.com>]

On 06/06/2013 09:40 AM, liguang wrote:

> this driver will support cpu phyical add/removal automatically
> after online/offline. if cpu hotpluged, cpu will not
> online automatically, and for cpu offline, we try to
> do actually eject if allowed for cpu like
> "echo 1 > /sys/bus/acpi/devices/LNXCPU\:0X/eject"
> this "echo ..." is only present for recent kernel
> (sorry, can't figure out since when), for a little
> older kernel, there's not such approach AFAICS.
> 
> Signed-off-by: liguang <lig.fnst@cn.fujitsu.com>
> ---
>  drivers/platform/x86/Kconfig              |    8 ++++
>  drivers/platform/x86/Makefile             |    1 +
>  drivers/platform/x86/cpu_physic_hotplug.c |   60 +++++++++++++++++++++++++++++
>  3 files changed, 69 insertions(+), 0 deletions(-)
>  create mode 100644 drivers/platform/x86/cpu_physic_hotplug.c
> 
> diff --git a/drivers/platform/x86/Kconfig b/drivers/platform/x86/Kconfig
> index 8577261..39b2392 100644
> --- a/drivers/platform/x86/Kconfig
> +++ b/drivers/platform/x86/Kconfig
> @@ -789,4 +789,12 @@ config PVPANIC
>  	  a paravirtualized device provided by QEMU; it lets a virtual machine
>  	  (guest) communicate panic events to the host.
>  
> +config QEMU_CPU_PHYSIC_HOTPLUG
> +	tristate "physically add/remove cpu after cpu onlined/offlined"
> +	depends on ACPI_HOTPLUG_CPU
> +	---help---
> +	  This driver will support physically remove a cpu after
> +	  it offlined for QEMU automatically. someone may require this feature
> +	  to do a physically removal for a cpu.
> +
>  endif # X86_PLATFORM_DEVICES
> diff --git a/drivers/platform/x86/Makefile b/drivers/platform/x86/Makefile
> index ef0ec74..2e669b0 100644
> --- a/drivers/platform/x86/Makefile
> +++ b/drivers/platform/x86/Makefile
> @@ -53,3 +53,4 @@ obj-$(CONFIG_APPLE_GMUX)	+= apple-gmux.o
>  obj-$(CONFIG_CHROMEOS_LAPTOP)	+= chromeos_laptop.o
>  
>  obj-$(CONFIG_PVPANIC)           += pvpanic.o
> +obj-$(CONFIG_QEMU_CPU_PHYSIC_HOTPLUG)	+= cpu_physic_hotplug.o
> diff --git a/drivers/platform/x86/cpu_physic_hotplug.c b/drivers/platform/x86/cpu_physic_hotplug.c
> new file mode 100644
> index 0000000..a52c042
> --- /dev/null
> +++ b/drivers/platform/x86/cpu_physic_hotplug.c
> @@ -0,0 +1,60 @@
> +#include <linux/kernel.h>
> +#include <linux/module.h>
> +#include <linux/acpi.h>
> +#include <linux/cpu.h>
> +#include <linux/notifier.h>
> +#include <acpi/processor.h>
> +
> +MODULE_AUTHOR("Li Guang");
> +MODULE_DESCRIPTION("CPU physically hot-plug/unplug Driver");
> +MODULE_LICENSE("GPL");
> +
> +static int cpu_logic_hotplug_notify(struct notifier_block *nfb,
> +				unsigned long action, void *hcpu)
> +{
> +	unsigned int cpu = (unsigned long)hcpu;
> +	struct acpi_processor *pr = per_cpu(processors, cpu);
> +
> +	if (pr) {
> +		switch (action) {
> +		case CPU_ONLINE:
> +			break;
> +		case CPU_DEAD:
> +			break;
> +		default:
> +			break;
> +		}
> +	}
> +	return NOTIFY_OK;
> +}
> +
> +static struct notifier_block cpu_logic_hotplug_notifier =
> +{
> +	.notifier_call = cpu_logic_hotplug_notify,
> +};
> +
> +static int cpu_physic_hotplug_notify(struct notifier_block *nfb,
> +				 unsigned char *s)
> +{
> +}

Hi guang,
	Maybe you need to define the callback function in the right format at the beginning,
if so, no need to correct it later.:)

Thanks,
Gu


> +
> +static struct notifier_block cpu_physic_hotplug_notifier =
> +{
> +	.notifier_call = cpu_physic_hotplug_notify,
> +};
> +
> +static int __init cpu_qemu_hotplug_init(void)
> +{
> +	register_hotcpu_notifier(&cpu_logic_hotplug_notifier);
> +	register_ec_gpe_notifier(&cpu_physic_hotplug_notifier);


As the [PATCH 2/3] has no dependence on this one, so you can set [PATCH 2/3] to [PATCH 1/3] and this one
to [PATCH 2/3]. Then you can use the xxx_ec_space_notifier directly here.

> +	return 0;
> +}
> +
> +static void __exit cpu_qemu_hotplug_exit(void)
> +{
> +	unregister_hotcpu_notifier(&cpu_logic_hotplug_notifier);
> +	unregister_ec_gpe_notifier(&cpu_physic_hotplug_notifier);
> +}
> +
> +module_init(cpu_qemu_hotplug_init);
> +module_exit(cpu_qemu_hotplug_exit);