On Wed, 21 Oct 2015 17:18:36 +0800
Tang Chen <tangchen@xxxxxxxxxxxxxx> wrote:
>
> On 10/21/2015 05:00 PM, Vlastimil Babka wrote:
> > On 10/21/2015 04:20 AM, Changsheng Liu wrote:
> >>
> >>
> >> 在 2015/10/15 0:18, Vlastimil Babka 写道:
> >>> On 10/12/2015 08:58 AM, Changsheng Liu wrote:
> >>>> From: Changsheng Liu <liuchangcheng@xxxxxxxxxx>
> >>>>
> >>>> After the user config CONFIG_MOVABLE_NODE,
> >>>> When the memory is hot added, should_add_memory_movable() return 0
> >>>> because all zones including ZONE_MOVABLE are empty,
> >>>> so the memory that was hot added will be assigned to ZONE_NORMAL
> >>>> and ZONE_NORMAL will be created firstly.
> >>>> But we want the whole node to be added to ZONE_MOVABLE by default.
> >>>>
> >>>> So we change should_add_memory_movable(): if the user config
> >>>> CONFIG_MOVABLE_NODE and sysctl parameter hotadd_memory_as_movable is 1
> >>>> and the ZONE_NORMAL is empty or the pfn of the hot-added memory
> >>>> is after the end of the ZONE_NORMAL it will always return 1
> >>>> and then the whole node will be added to ZONE_MOVABLE by default.
> >>>> If we want the node to be assigned to ZONE_NORMAL,
> >>>> we can do it as follows:
> >>>> "echo online_kernel > /sys/devices/system/memory/memoryXXX/state"
> >>>>
> >>>> By the patch, the behavious of kernel is changed by sysctl,
> >>>> user can automatically create movable memory
> >>>> by only the following udev rule:
> >>>> SUBSYSTEM=="memory", ACTION=="add",
> >>>> ATTR{state}=="offline", ATTR{state}="online"
> >> I'm sorry for replying you so late due to the busy business trip.
> >>> So just to be clear, we are adding a new sysctl, because the existing
> >>> movable_node kernel option, which is checked by
> >>> movable_node_is_enabled(), and
> >>> does the same thing for non-hot-added-memory (?) cannot be reused
> >>> for hot-added
> >>> memory, as that would be a potentially surprising behavior change?
> >>> Correct? Then
> >>> this should be mentioned in the changelog too, and wherever
> >>> "movable_node" is
> >>> documented should also mention the new sysctl. Personally, I would
> >>> expect
> >>> movable_node to affect hot-added memory as well, and would be
> >>> surprised that it
> >>> doesn't...
> >> I think it can let the user decides when to use this feature.
> >> The user can enable the feature when making the hot_added memory
> >> of a node movable and
> >> make the feature disable to assign the hot_added memory of the
> >> next
> >> node to ZONE_NORMAL .
> >
> > So you mean sysctl is more flexible than boot option. OK, but wasn't
> > such flexibility already provided by "echo online_kernel" vs "echo
> > online_movable"? It doesn't sound like a strong reason for a new
> > sysctl? Not doing surprising behavior change maybe does...
> > .
> >
> Hi Vlastimil,
>
> The current kernel will add hot-added memory to ZONE_NORMAL by default.
> If users use a udev rule as below:
>
> SUBSYSTEM=="memory", ACTION=="add", ATTR{state}=="offline", ATTR{state}="online"
>
> it will online the memory as normal memory, which will not be hotpluggable.
>
> Please refer to: https://lkml.org/lkml/2015/10/9/58
>
> I think this is the root motivation of the patch.
>
> But BTW, I'm quite familiar with udev rules, but can something like this
> work ?
>
> SUBSYSTEM=="memory", ACTION=="add", ATTR{state}=="offline", ATTR{state}="online_movable"
>
> I'm not sure. I added Ishimatu in.
I think the udev rules fails to online memory as movable.
When hot adding memory, the memory is managed as ZONE_NORMAL.
And add events of memory section are notified to udev in ascending
order, like 0->1->2->3. Thus udev starts to online memory from section 0.
But to change zone from ZONE_NORMAL to ZONE_MOVALBE, udev onlines memory
in descending order, like 3->2->1->0. So the udev rules cannot online
memory as movable.
Thanks,
Yasuaki Ishimatsu
>
> For now, I think, if the above rule works, we don't need this patch. If
> not, maybe we should just change the kernel behavior to make the
> hot-added memory be added to ZONE_MOVABLE by default.
>
> I don't have objection. But a sysctl doesn't sound necessary.
>
> Thanks.
>
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>