Re: [PATCH v2] mm/hotplug, x86: Disable ARCH_MEMORY_PROBE by default

[KOSAKI Motohiro]

(7/22/13 1:12 PM), Toshi Kani wrote:
> On Mon, 2013-07-22 at 10:37 +0200, Ingo Molnar wrote:
> > * Toshi Kani <toshi.kani@xxxxxx> wrote:
> >
> > > CONFIG_ARCH_MEMORY_PROBE enables /sys/devices/system/memory/probe
> > > interface, which allows a given memory address to be hot-added as
> > > follows. (See Documentation/memory-hotplug.txt for more detail.)
> > >
> > > # echo start_address_of_new_memory > /sys/devices/system/memory/probe
> > >
> > > This probe interface is required on powerpc. On x86, however, ACPI
> > > notifies a memory hotplug event to the kernel, which performs its
> > > hotplug operation as the result. Therefore, regular users do not need
> > > this interface on x86. This probe interface is also error-prone and
> > > misleading that the kernel blindly adds a given memory address without
> > > checking if the memory is present on the system; no probing is done
> > > despite of its name. The kernel crashes when a user requests to online
> > > a memory block that is not present on the system. This interface is
> > > currently used for testing as it can fake a hotplug event.
> > >
> > > This patch disables CONFIG_ARCH_MEMORY_PROBE by default on x86, adds
> > > its Kconfig menu entry on x86, and clarifies its use in Documentation/
> > > memory-hotplug.txt.
> >
> > Could we please also fix it to never crash the kernel, even if stupid
> > ranges are provided?
>
> Yes, this probe interface can be enhanced to verify the firmware
> information before adding a given memory address.  However, such change
> would interfere its test use of "fake" hotplug, which is only the known
> use-case of this interface on x86.
>
> In order to verify if a given memory address is enabled at run-time (as
> opposed to boot-time), we need to check with ACPI memory device objects
> on x86.  However, system vendors tend to not implement memory device
> objects unless their systems support memory hotplug.  Dave Hansen is
> using this interface for his testing as a way to fake a hotplug event on
> a system that does not support memory hotplug.

One of possible option is to return EINVAL when system has real hotplug device.
I mean this interface is only useful when system don't have proper hardware
feature and doesn't work correctly hardware property and this interface command
are not consistent.

Dave, What do you think?

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/