LinuxLists.cc - [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On 23.07.20 10:45, Roger Pau Monne wrote:
> Add an extra option to add_memory_resource that overrides the memory
> hotplug online behavior in order to force onlining of memory from
> add_memory_resource unconditionally.
>
> This is required for the Xen balloon driver, that must run the
> online page callback in order to correctly process the newly added
> memory region, note this is an unpopulated region that is used by Linux
> to either hotplug RAM or to map foreign pages from other domains, and
> hence memory hotplug when running on Xen can be used even without the
> user explicitly requesting it, as part of the normal operations of the
> OS when attempting to map memory from a different domain.
>
> Setting a different default value of memhp_default_online_type when
> attaching the balloon driver is not a robust solution, as the user (or
> distro init scripts) could still change it and thus break the Xen
> balloon driver.

I think we discussed this a couple of times before (even triggered by my
request), and this is responsibility of user space to configure. Usually
distros have udev rules to online memory automatically. Especially, user
space should eb able to configure *how* to online memory.

It's the admin/distro responsibility to configure this properly. In case
this doesn't happen (or as you say, users change it), bad luck.

E.g., virtio-mem takes care to not add more memory in case it is not
getting onlined. I remember hyper-v has similar code to at least wait a
bit for memory to get onlined.

Nacked-by: David Hildenbrand <[email protected]>

--
Thanks,

David / dhildenb

2020-07-23 11:53:19

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On 23.07.20 13:37, David Hildenbrand wrote:
> On 23.07.20 10:45, Roger Pau Monne wrote:
>> Add an extra option to add_memory_resource that overrides the memory
>> hotplug online behavior in order to force onlining of memory from
>> add_memory_resource unconditionally.
>>
>> This is required for the Xen balloon driver, that must run the
>> online page callback in order to correctly process the newly added
>> memory region, note this is an unpopulated region that is used by Linux
>> to either hotplug RAM or to map foreign pages from other domains, and
>> hence memory hotplug when running on Xen can be used even without the
>> user explicitly requesting it, as part of the normal operations of the
>> OS when attempting to map memory from a different domain.
>>
>> Setting a different default value of memhp_default_online_type when
>> attaching the balloon driver is not a robust solution, as the user (or
>> distro init scripts) could still change it and thus break the Xen
>> balloon driver.
>
> I think we discussed this a couple of times before (even triggered by my
> request), and this is responsibility of user space to configure. Usually
> distros have udev rules to online memory automatically. Especially, user
> space should eb able to configure *how* to online memory.
>
> It's the admin/distro responsibility to configure this properly. In case
> this doesn't happen (or as you say, users change it), bad luck.
>
> E.g., virtio-mem takes care to not add more memory in case it is not
> getting onlined. I remember hyper-v has similar code to at least wait a
> bit for memory to get onlined.
>
> Nacked-by: David Hildenbrand <[email protected]>
>

Oh, BTW, I removed that "online" parameter in

commit f29d8e9c0191a2a02500945db505e5c89159c3f4
Author: David Hildenbrand <[email protected]>
Date: Fri Dec 28 00:35:36 2018 -0800

mm/memory_hotplug: drop "online" parameter from add_memory_resource()

Userspace should always be in charge of how to online memory and if memory
should be onlined automatically in the kernel. Let's drop the parameter
to overwrite this - XEN passes memhp_auto_online, just like add_memory(),
so we can directly use that instead internally.

Xen was passing "memhp_auto_online" since

commit 703fc13a3f6615e29ce3eb862275d7b58a5d03ba
Author: Vitaly Kuznetsov <[email protected]>
Date: Tue Mar 15 14:56:52 2016 -0700

xen_balloon: support memory auto onlining policy

Add support for the newly added kernel memory auto onlining policy to
Xen ballon driver.

And before that I assume XEN was completely relying on udev rules to handle it. Parameter was introduced in

commit 31bc3858ea3ebcc3157b3f5f0e624c5962f5a7a6
Author: Vitaly Kuznetsov <[email protected]>
Date: Tue Mar 15 14:56:48 2016 -0700

memory-hotplug: add automatic onlining policy for the newly added memory

Currently, all newly added memory blocks remain in 'offline' state
unless someone onlines them, some linux distributions carry special udev
rules like:

SUBSYSTEM=="memory", ACTION=="add", ATTR{state}=="offline", ATTR{state}="online"

--
Thanks,

David / dhildenb

2020-07-23 12:24:04

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On Thu, Jul 23, 2020 at 01:37:03PM +0200, David Hildenbrand wrote:
> On 23.07.20 10:45, Roger Pau Monne wrote:
> > Add an extra option to add_memory_resource that overrides the memory
> > hotplug online behavior in order to force onlining of memory from
> > add_memory_resource unconditionally.
> >
> > This is required for the Xen balloon driver, that must run the
> > online page callback in order to correctly process the newly added
> > memory region, note this is an unpopulated region that is used by Linux
> > to either hotplug RAM or to map foreign pages from other domains, and
> > hence memory hotplug when running on Xen can be used even without the
> > user explicitly requesting it, as part of the normal operations of the
> > OS when attempting to map memory from a different domain.
> >
> > Setting a different default value of memhp_default_online_type when
> > attaching the balloon driver is not a robust solution, as the user (or
> > distro init scripts) could still change it and thus break the Xen
> > balloon driver.
>
> I think we discussed this a couple of times before (even triggered by my
> request), and this is responsibility of user space to configure. Usually
> distros have udev rules to online memory automatically. Especially, user
> space should eb able to configure *how* to online memory.

Note (as per the commit message) that in the specific case I'm
referring to the memory hotplugged by the Xen balloon driver will be
an unpopulated range to be used internally by certain Xen subsystems,
like the xen-blkback or the privcmd drivers. The addition of such
blocks of (unpopulated) memory can happen without the user explicitly
requesting it, and hence not even aware such hotplug process is taking
place. To be clear: no actual RAM will be added to the system.

Failure to online such blocks using the Xen specific online handler
(which does not handle back the memory to the allocator in any way)
will result in the system getting stuck and malfunctioning.

> It's the admin/distro responsibility to configure this properly. In case
> this doesn't happen (or as you say, users change it), bad luck.
>
> E.g., virtio-mem takes care to not add more memory in case it is not
> getting onlined. I remember hyper-v has similar code to at least wait a
> bit for memory to get onlined.

I don't think VirtIO or Hyper-V use the hotplug system in the same way
as Xen, as said this is done to add unpopulated memory regions that
will be used to map foreign memory (from other domains) by Xen drivers
on the system.

Maybe this should somehow use a different mechanism to hotplug such
empty memory blocks? I don't mind doing this differently, but I would
need some pointers. Allowing user-space to change a (seemingly
unrelated) parameter and as a result produce failures on Xen drivers
is not an acceptable solution IMO.

Thanks, Roger.

2020-07-23 12:31:28

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On 23.07.20 14:23, Roger Pau Monné wrote:
> On Thu, Jul 23, 2020 at 01:37:03PM +0200, David Hildenbrand wrote:
>> On 23.07.20 10:45, Roger Pau Monne wrote:
>>> Add an extra option to add_memory_resource that overrides the memory
>>> hotplug online behavior in order to force onlining of memory from
>>> add_memory_resource unconditionally.
>>>
>>> This is required for the Xen balloon driver, that must run the
>>> online page callback in order to correctly process the newly added
>>> memory region, note this is an unpopulated region that is used by Linux
>>> to either hotplug RAM or to map foreign pages from other domains, and
>>> hence memory hotplug when running on Xen can be used even without the
>>> user explicitly requesting it, as part of the normal operations of the
>>> OS when attempting to map memory from a different domain.
>>>
>>> Setting a different default value of memhp_default_online_type when
>>> attaching the balloon driver is not a robust solution, as the user (or
>>> distro init scripts) could still change it and thus break the Xen
>>> balloon driver.
>>
>> I think we discussed this a couple of times before (even triggered by my
>> request), and this is responsibility of user space to configure. Usually
>> distros have udev rules to online memory automatically. Especially, user
>> space should eb able to configure *how* to online memory.
>
> Note (as per the commit message) that in the specific case I'm
> referring to the memory hotplugged by the Xen balloon driver will be
> an unpopulated range to be used internally by certain Xen subsystems,
> like the xen-blkback or the privcmd drivers. The addition of such
> blocks of (unpopulated) memory can happen without the user explicitly
> requesting it, and hence not even aware such hotplug process is taking
> place. To be clear: no actual RAM will be added to the system.
>
> Failure to online such blocks using the Xen specific online handler
> (which does not handle back the memory to the allocator in any way)
> will result in the system getting stuck and malfunctioning.
>
>> It's the admin/distro responsibility to configure this properly. In case
>> this doesn't happen (or as you say, users change it), bad luck.
>>
>> E.g., virtio-mem takes care to not add more memory in case it is not
>> getting onlined. I remember hyper-v has similar code to at least wait a
>> bit for memory to get onlined.
>
> I don't think VirtIO or Hyper-V use the hotplug system in the same way
> as Xen, as said this is done to add unpopulated memory regions that
> will be used to map foreign memory (from other domains) by Xen drivers
> on the system.
>
> Maybe this should somehow use a different mechanism to hotplug such
> empty memory blocks? I don't mind doing this differently, but I would
> need some pointers. Allowing user-space to change a (seemingly
> unrelated) parameter and as a result produce failures on Xen drivers
> is not an acceptable solution IMO.

Maybe we can use the same approach as Xen PV-domains: pre-allocate a
region in the memory map to be used for mapping foreign pages. For the
kernel it will look like pre-ballooned memory, so it will create struct
page for the region (which is what we are after), but it won't give the
memory to the allocator.

Juergen

2020-07-23 12:34:34

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On 23.07.20 14:28, Jürgen Groß wrote:
> On 23.07.20 14:23, Roger Pau Monné wrote:
>> On Thu, Jul 23, 2020 at 01:37:03PM +0200, David Hildenbrand wrote:
>>> On 23.07.20 10:45, Roger Pau Monne wrote:
>>>> Add an extra option to add_memory_resource that overrides the memory
>>>> hotplug online behavior in order to force onlining of memory from
>>>> add_memory_resource unconditionally.
>>>>
>>>> This is required for the Xen balloon driver, that must run the
>>>> online page callback in order to correctly process the newly added
>>>> memory region, note this is an unpopulated region that is used by Linux
>>>> to either hotplug RAM or to map foreign pages from other domains, and
>>>> hence memory hotplug when running on Xen can be used even without the
>>>> user explicitly requesting it, as part of the normal operations of the
>>>> OS when attempting to map memory from a different domain.
>>>>
>>>> Setting a different default value of memhp_default_online_type when
>>>> attaching the balloon driver is not a robust solution, as the user (or
>>>> distro init scripts) could still change it and thus break the Xen
>>>> balloon driver.
>>>
>>> I think we discussed this a couple of times before (even triggered by my
>>> request), and this is responsibility of user space to configure. Usually
>>> distros have udev rules to online memory automatically. Especially, user
>>> space should eb able to configure *how* to online memory.
>>
>> Note (as per the commit message) that in the specific case I'm
>> referring to the memory hotplugged by the Xen balloon driver will be
>> an unpopulated range to be used internally by certain Xen subsystems,
>> like the xen-blkback or the privcmd drivers. The addition of such
>> blocks of (unpopulated) memory can happen without the user explicitly
>> requesting it, and hence not even aware such hotplug process is taking
>> place. To be clear: no actual RAM will be added to the system.
>>
>> Failure to online such blocks using the Xen specific online handler
>> (which does not handle back the memory to the allocator in any way)
>> will result in the system getting stuck and malfunctioning.
>>
>>> It's the admin/distro responsibility to configure this properly. In case
>>> this doesn't happen (or as you say, users change it), bad luck.
>>>
>>> E.g., virtio-mem takes care to not add more memory in case it is not
>>> getting onlined. I remember hyper-v has similar code to at least wait a
>>> bit for memory to get onlined.
>>
>> I don't think VirtIO or Hyper-V use the hotplug system in the same way
>> as Xen, as said this is done to add unpopulated memory regions that
>> will be used to map foreign memory (from other domains) by Xen drivers
>> on the system.
>>
>> Maybe this should somehow use a different mechanism to hotplug such
>> empty memory blocks? I don't mind doing this differently, but I would
>> need some pointers. Allowing user-space to change a (seemingly
>> unrelated) parameter and as a result produce failures on Xen drivers
>> is not an acceptable solution IMO.
>
> Maybe we can use the same approach as Xen PV-domains: pre-allocate a
> region in the memory map to be used for mapping foreign pages. For the
> kernel it will look like pre-ballooned memory, so it will create struct
> page for the region (which is what we are after), but it won't give the
> memory to the allocator.

Something like that sounds a lot cleaner to me than abusing the memory
hotplug mechanism (which the xen balloon also uses to just expose more
memory). Because there are other issues in case you "really want memory
to be onlined". What if onlining fails (nacked by a notifier, e.g., kasan)?

--
Thanks,

David / dhildenb

2020-07-23 13:10:10

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On Thu, Jul 23, 2020 at 02:28:13PM +0200, Jürgen Groß wrote:
> On 23.07.20 14:23, Roger Pau Monné wrote:
> > On Thu, Jul 23, 2020 at 01:37:03PM +0200, David Hildenbrand wrote:
> > > On 23.07.20 10:45, Roger Pau Monne wrote:
> > > > Add an extra option to add_memory_resource that overrides the memory
> > > > hotplug online behavior in order to force onlining of memory from
> > > > add_memory_resource unconditionally.
> > > >
> > > > This is required for the Xen balloon driver, that must run the
> > > > online page callback in order to correctly process the newly added
> > > > memory region, note this is an unpopulated region that is used by Linux
> > > > to either hotplug RAM or to map foreign pages from other domains, and
> > > > hence memory hotplug when running on Xen can be used even without the
> > > > user explicitly requesting it, as part of the normal operations of the
> > > > OS when attempting to map memory from a different domain.
> > > >
> > > > Setting a different default value of memhp_default_online_type when
> > > > attaching the balloon driver is not a robust solution, as the user (or
> > > > distro init scripts) could still change it and thus break the Xen
> > > > balloon driver.
> > >
> > > I think we discussed this a couple of times before (even triggered by my
> > > request), and this is responsibility of user space to configure. Usually
> > > distros have udev rules to online memory automatically. Especially, user
> > > space should eb able to configure *how* to online memory.
> >
> > Note (as per the commit message) that in the specific case I'm
> > referring to the memory hotplugged by the Xen balloon driver will be
> > an unpopulated range to be used internally by certain Xen subsystems,
> > like the xen-blkback or the privcmd drivers. The addition of such
> > blocks of (unpopulated) memory can happen without the user explicitly
> > requesting it, and hence not even aware such hotplug process is taking
> > place. To be clear: no actual RAM will be added to the system.
> >
> > Failure to online such blocks using the Xen specific online handler
> > (which does not handle back the memory to the allocator in any way)
> > will result in the system getting stuck and malfunctioning.
> >
> > > It's the admin/distro responsibility to configure this properly. In case
> > > this doesn't happen (or as you say, users change it), bad luck.
> > >
> > > E.g., virtio-mem takes care to not add more memory in case it is not
> > > getting onlined. I remember hyper-v has similar code to at least wait a
> > > bit for memory to get onlined.
> >
> > I don't think VirtIO or Hyper-V use the hotplug system in the same way
> > as Xen, as said this is done to add unpopulated memory regions that
> > will be used to map foreign memory (from other domains) by Xen drivers
> > on the system.
> >
> > Maybe this should somehow use a different mechanism to hotplug such
> > empty memory blocks? I don't mind doing this differently, but I would
> > need some pointers. Allowing user-space to change a (seemingly
> > unrelated) parameter and as a result produce failures on Xen drivers
> > is not an acceptable solution IMO.
>
> Maybe we can use the same approach as Xen PV-domains: pre-allocate a
> region in the memory map to be used for mapping foreign pages. For the
> kernel it will look like pre-ballooned memory, so it will create struct
> page for the region (which is what we are after), but it won't give the
> memory to the allocator.

IMO using something similar to memory hotplug would give us more
flexibility, and TBH the logic is already there in the balloon driver.
It seems quite wasteful to allocate such region(s) beforehand for all
domains, even when most of them won't end up using foreign mappings at
all.

Anyway, I'm going to take a look at how to do that, I guess it's going
to involve playing with the memory map and reserving some space.

I suggest we should remove the Xen balloon hotplug logic, as it's not
working properly and we don't have a plan to fix it.

Thanks, Roger.

2020-07-23 13:17:44

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On 23.07.20 15:08, Roger Pau Monné wrote:
> On Thu, Jul 23, 2020 at 02:28:13PM +0200, Jürgen Groß wrote:
>> On 23.07.20 14:23, Roger Pau Monné wrote:
>>> On Thu, Jul 23, 2020 at 01:37:03PM +0200, David Hildenbrand wrote:
>>>> On 23.07.20 10:45, Roger Pau Monne wrote:
>>>>> Add an extra option to add_memory_resource that overrides the memory
>>>>> hotplug online behavior in order to force onlining of memory from
>>>>> add_memory_resource unconditionally.
>>>>>
>>>>> This is required for the Xen balloon driver, that must run the
>>>>> online page callback in order to correctly process the newly added
>>>>> memory region, note this is an unpopulated region that is used by Linux
>>>>> to either hotplug RAM or to map foreign pages from other domains, and
>>>>> hence memory hotplug when running on Xen can be used even without the
>>>>> user explicitly requesting it, as part of the normal operations of the
>>>>> OS when attempting to map memory from a different domain.
>>>>>
>>>>> Setting a different default value of memhp_default_online_type when
>>>>> attaching the balloon driver is not a robust solution, as the user (or
>>>>> distro init scripts) could still change it and thus break the Xen
>>>>> balloon driver.
>>>>
>>>> I think we discussed this a couple of times before (even triggered by my
>>>> request), and this is responsibility of user space to configure. Usually
>>>> distros have udev rules to online memory automatically. Especially, user
>>>> space should eb able to configure *how* to online memory.
>>>
>>> Note (as per the commit message) that in the specific case I'm
>>> referring to the memory hotplugged by the Xen balloon driver will be
>>> an unpopulated range to be used internally by certain Xen subsystems,
>>> like the xen-blkback or the privcmd drivers. The addition of such
>>> blocks of (unpopulated) memory can happen without the user explicitly
>>> requesting it, and hence not even aware such hotplug process is taking
>>> place. To be clear: no actual RAM will be added to the system.
>>>
>>> Failure to online such blocks using the Xen specific online handler
>>> (which does not handle back the memory to the allocator in any way)
>>> will result in the system getting stuck and malfunctioning.
>>>
>>>> It's the admin/distro responsibility to configure this properly. In case
>>>> this doesn't happen (or as you say, users change it), bad luck.
>>>>
>>>> E.g., virtio-mem takes care to not add more memory in case it is not
>>>> getting onlined. I remember hyper-v has similar code to at least wait a
>>>> bit for memory to get onlined.
>>>
>>> I don't think VirtIO or Hyper-V use the hotplug system in the same way
>>> as Xen, as said this is done to add unpopulated memory regions that
>>> will be used to map foreign memory (from other domains) by Xen drivers
>>> on the system.
>>>
>>> Maybe this should somehow use a different mechanism to hotplug such
>>> empty memory blocks? I don't mind doing this differently, but I would
>>> need some pointers. Allowing user-space to change a (seemingly
>>> unrelated) parameter and as a result produce failures on Xen drivers
>>> is not an acceptable solution IMO.
>>
>> Maybe we can use the same approach as Xen PV-domains: pre-allocate a
>> region in the memory map to be used for mapping foreign pages. For the
>> kernel it will look like pre-ballooned memory, so it will create struct
>> page for the region (which is what we are after), but it won't give the
>> memory to the allocator.
>
> IMO using something similar to memory hotplug would give us more
> flexibility, and TBH the logic is already there in the balloon driver.
> It seems quite wasteful to allocate such region(s) beforehand for all
> domains, even when most of them won't end up using foreign mappings at
> all.

I do wonder why these issues you describe start to pop up now, literally
years after this stuff has been implemented - or am I missing something
important?

>
> Anyway, I'm going to take a look at how to do that, I guess it's going
> to involve playing with the memory map and reserving some space.
>
> I suggest we should remove the Xen balloon hotplug logic, as it's not
> working properly and we don't have a plan to fix it.

Which exact hotplug logic are you referring to?

--
Thanks,

David / dhildenb

2020-07-23 13:23:43

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On 23.07.20 15:08, Roger Pau Monné wrote:
> On Thu, Jul 23, 2020 at 02:28:13PM +0200, Jürgen Groß wrote:
>> On 23.07.20 14:23, Roger Pau Monné wrote:
>>> On Thu, Jul 23, 2020 at 01:37:03PM +0200, David Hildenbrand wrote:
>>>> On 23.07.20 10:45, Roger Pau Monne wrote:
>>>>> Add an extra option to add_memory_resource that overrides the memory
>>>>> hotplug online behavior in order to force onlining of memory from
>>>>> add_memory_resource unconditionally.
>>>>>
>>>>> This is required for the Xen balloon driver, that must run the
>>>>> online page callback in order to correctly process the newly added
>>>>> memory region, note this is an unpopulated region that is used by Linux
>>>>> to either hotplug RAM or to map foreign pages from other domains, and
>>>>> hence memory hotplug when running on Xen can be used even without the
>>>>> user explicitly requesting it, as part of the normal operations of the
>>>>> OS when attempting to map memory from a different domain.
>>>>>
>>>>> Setting a different default value of memhp_default_online_type when
>>>>> attaching the balloon driver is not a robust solution, as the user (or
>>>>> distro init scripts) could still change it and thus break the Xen
>>>>> balloon driver.
>>>>
>>>> I think we discussed this a couple of times before (even triggered by my
>>>> request), and this is responsibility of user space to configure. Usually
>>>> distros have udev rules to online memory automatically. Especially, user
>>>> space should eb able to configure *how* to online memory.
>>>
>>> Note (as per the commit message) that in the specific case I'm
>>> referring to the memory hotplugged by the Xen balloon driver will be
>>> an unpopulated range to be used internally by certain Xen subsystems,
>>> like the xen-blkback or the privcmd drivers. The addition of such
>>> blocks of (unpopulated) memory can happen without the user explicitly
>>> requesting it, and hence not even aware such hotplug process is taking
>>> place. To be clear: no actual RAM will be added to the system.
>>>
>>> Failure to online such blocks using the Xen specific online handler
>>> (which does not handle back the memory to the allocator in any way)
>>> will result in the system getting stuck and malfunctioning.
>>>
>>>> It's the admin/distro responsibility to configure this properly. In case
>>>> this doesn't happen (or as you say, users change it), bad luck.
>>>>
>>>> E.g., virtio-mem takes care to not add more memory in case it is not
>>>> getting onlined. I remember hyper-v has similar code to at least wait a
>>>> bit for memory to get onlined.
>>>
>>> I don't think VirtIO or Hyper-V use the hotplug system in the same way
>>> as Xen, as said this is done to add unpopulated memory regions that
>>> will be used to map foreign memory (from other domains) by Xen drivers
>>> on the system.
>>>
>>> Maybe this should somehow use a different mechanism to hotplug such
>>> empty memory blocks? I don't mind doing this differently, but I would
>>> need some pointers. Allowing user-space to change a (seemingly
>>> unrelated) parameter and as a result produce failures on Xen drivers
>>> is not an acceptable solution IMO.
>>
>> Maybe we can use the same approach as Xen PV-domains: pre-allocate a
>> region in the memory map to be used for mapping foreign pages. For the
>> kernel it will look like pre-ballooned memory, so it will create struct
>> page for the region (which is what we are after), but it won't give the
>> memory to the allocator.
>
> IMO using something similar to memory hotplug would give us more
> flexibility, and TBH the logic is already there in the balloon driver.
> It seems quite wasteful to allocate such region(s) beforehand for all
> domains, even when most of them won't end up using foreign mappings at
> all.

We can do it for dom0 only per default, and add a boot parameter e.g.
for driver domains.

And the logic is already there (just pv-only right now).

>
> Anyway, I'm going to take a look at how to do that, I guess it's going
> to involve playing with the memory map and reserving some space.

Look at arch/x86/xen/setup.c (xen_add_extra_mem() and its usage).

>
> I suggest we should remove the Xen balloon hotplug logic, as it's not
> working properly and we don't have a plan to fix it.

I have used memory hotplug successfully not very long ago.

Juergen

2020-07-23 13:25:42

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On 23.07.20 14:23, Roger Pau Monné wrote:
> On Thu, Jul 23, 2020 at 01:37:03PM +0200, David Hildenbrand wrote:
>> On 23.07.20 10:45, Roger Pau Monne wrote:
>>> Add an extra option to add_memory_resource that overrides the memory
>>> hotplug online behavior in order to force onlining of memory from
>>> add_memory_resource unconditionally.
>>>
>>> This is required for the Xen balloon driver, that must run the
>>> online page callback in order to correctly process the newly added
>>> memory region, note this is an unpopulated region that is used by Linux
>>> to either hotplug RAM or to map foreign pages from other domains, and
>>> hence memory hotplug when running on Xen can be used even without the
>>> user explicitly requesting it, as part of the normal operations of the
>>> OS when attempting to map memory from a different domain.
>>>
>>> Setting a different default value of memhp_default_online_type when
>>> attaching the balloon driver is not a robust solution, as the user (or
>>> distro init scripts) could still change it and thus break the Xen
>>> balloon driver.
>>
>> I think we discussed this a couple of times before (even triggered by my
>> request), and this is responsibility of user space to configure. Usually
>> distros have udev rules to online memory automatically. Especially, user
>> space should eb able to configure *how* to online memory.
>
> Note (as per the commit message) that in the specific case I'm
> referring to the memory hotplugged by the Xen balloon driver will be
> an unpopulated range to be used internally by certain Xen subsystems,
> like the xen-blkback or the privcmd drivers. The addition of such
> blocks of (unpopulated) memory can happen without the user explicitly
> requesting it, and hence not even aware such hotplug process is taking
> place. To be clear: no actual RAM will be added to the system.

Okay, but there is also the case where XEN will actually hotplug memory
using this same handler IIRC (at least I've read papers about it). Both
are using the same handler, correct?

>
>> It's the admin/distro responsibility to configure this properly. In case
>> this doesn't happen (or as you say, users change it), bad luck.
>>
>> E.g., virtio-mem takes care to not add more memory in case it is not
>> getting onlined. I remember hyper-v has similar code to at least wait a
>> bit for memory to get onlined.
>
> I don't think VirtIO or Hyper-V use the hotplug system in the same way
> as Xen, as said this is done to add unpopulated memory regions that
> will be used to map foreign memory (from other domains) by Xen drivers
> on the system.

Indeed, if the memory is never exposed to the buddy (and all you need is
struct pages + a kernel virtual mapping), I wonder if
memremap/ZONE_DEVICE is what you want? Then you won't have user-visible
memory blocks created with unclear online semantics, partially involving
the buddy.

--
Thanks,

David / dhildenb

2020-07-23 13:27:57

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On Thu, Jul 23, 2020 at 03:14:31PM +0200, David Hildenbrand wrote:
> On 23.07.20 15:08, Roger Pau Monné wrote:
> > On Thu, Jul 23, 2020 at 02:28:13PM +0200, Jürgen Groß wrote:
> >> On 23.07.20 14:23, Roger Pau Monné wrote:
> >>> On Thu, Jul 23, 2020 at 01:37:03PM +0200, David Hildenbrand wrote:
> >>>> On 23.07.20 10:45, Roger Pau Monne wrote:
> >>>>> Add an extra option to add_memory_resource that overrides the memory
> >>>>> hotplug online behavior in order to force onlining of memory from
> >>>>> add_memory_resource unconditionally.
> >>>>>
> >>>>> This is required for the Xen balloon driver, that must run the
> >>>>> online page callback in order to correctly process the newly added
> >>>>> memory region, note this is an unpopulated region that is used by Linux
> >>>>> to either hotplug RAM or to map foreign pages from other domains, and
> >>>>> hence memory hotplug when running on Xen can be used even without the
> >>>>> user explicitly requesting it, as part of the normal operations of the
> >>>>> OS when attempting to map memory from a different domain.
> >>>>>
> >>>>> Setting a different default value of memhp_default_online_type when
> >>>>> attaching the balloon driver is not a robust solution, as the user (or
> >>>>> distro init scripts) could still change it and thus break the Xen
> >>>>> balloon driver.
> >>>>
> >>>> I think we discussed this a couple of times before (even triggered by my
> >>>> request), and this is responsibility of user space to configure. Usually
> >>>> distros have udev rules to online memory automatically. Especially, user
> >>>> space should eb able to configure *how* to online memory.
> >>>
> >>> Note (as per the commit message) that in the specific case I'm
> >>> referring to the memory hotplugged by the Xen balloon driver will be
> >>> an unpopulated range to be used internally by certain Xen subsystems,
> >>> like the xen-blkback or the privcmd drivers. The addition of such
> >>> blocks of (unpopulated) memory can happen without the user explicitly
> >>> requesting it, and hence not even aware such hotplug process is taking
> >>> place. To be clear: no actual RAM will be added to the system.
> >>>
> >>> Failure to online such blocks using the Xen specific online handler
> >>> (which does not handle back the memory to the allocator in any way)
> >>> will result in the system getting stuck and malfunctioning.
> >>>
> >>>> It's the admin/distro responsibility to configure this properly. In case
> >>>> this doesn't happen (or as you say, users change it), bad luck.
> >>>>
> >>>> E.g., virtio-mem takes care to not add more memory in case it is not
> >>>> getting onlined. I remember hyper-v has similar code to at least wait a
> >>>> bit for memory to get onlined.
> >>>
> >>> I don't think VirtIO or Hyper-V use the hotplug system in the same way
> >>> as Xen, as said this is done to add unpopulated memory regions that
> >>> will be used to map foreign memory (from other domains) by Xen drivers
> >>> on the system.
> >>>
> >>> Maybe this should somehow use a different mechanism to hotplug such
> >>> empty memory blocks? I don't mind doing this differently, but I would
> >>> need some pointers. Allowing user-space to change a (seemingly
> >>> unrelated) parameter and as a result produce failures on Xen drivers
> >>> is not an acceptable solution IMO.
> >>
> >> Maybe we can use the same approach as Xen PV-domains: pre-allocate a
> >> region in the memory map to be used for mapping foreign pages. For the
> >> kernel it will look like pre-ballooned memory, so it will create struct
> >> page for the region (which is what we are after), but it won't give the
> >> memory to the allocator.
> >
> > IMO using something similar to memory hotplug would give us more
> > flexibility, and TBH the logic is already there in the balloon driver.
> > It seems quite wasteful to allocate such region(s) beforehand for all
> > domains, even when most of them won't end up using foreign mappings at
> > all.
>
> I do wonder why these issues you describe start to pop up now, literally
> years after this stuff has been implemented - or am I missing something
> important?

We are (very slowly) implementing support to switch to a PVH dom0
(something similar to a fully emulated guest as dom0), and that kind
of guest no longer has a pre-allocated memory region for mappings,
likely because we mostly use the native path when dealing with
memory, and such reservation was done by the PV specific setup path
that deals with the memory map.

> >
> > Anyway, I'm going to take a look at how to do that, I guess it's going
> > to involve playing with the memory map and reserving some space.
> >
> > I suggest we should remove the Xen balloon hotplug logic, as it's not
> > working properly and we don't have a plan to fix it.
>
> Which exact hotplug logic are you referring to?

There are some sections in the Xen balloon driver protected with
CONFIG_XEN_BALLOON_MEMORY_HOTPLUG.

When xen_hotplug_unpopulated is enabled but the default policy for
memory hotplug is to not online the added blocks certain operations
like alloc_xenballooned_pages would block forever waiting for the
hotplugged memory to be onlined in order to be usable.

Thanks, Roger.

2020-07-23 13:40:46

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On Thu, Jul 23, 2020 at 03:20:55PM +0200, Jürgen Groß wrote:
> On 23.07.20 15:08, Roger Pau Monné wrote:
> > On Thu, Jul 23, 2020 at 02:28:13PM +0200, Jürgen Groß wrote:
> > > On 23.07.20 14:23, Roger Pau Monné wrote:
> > > > On Thu, Jul 23, 2020 at 01:37:03PM +0200, David Hildenbrand wrote:
> > > > > On 23.07.20 10:45, Roger Pau Monne wrote:
> > > > > > Add an extra option to add_memory_resource that overrides the memory
> > > > > > hotplug online behavior in order to force onlining of memory from
> > > > > > add_memory_resource unconditionally.
> > > > > >
> > > > > > This is required for the Xen balloon driver, that must run the
> > > > > > online page callback in order to correctly process the newly added
> > > > > > memory region, note this is an unpopulated region that is used by Linux
> > > > > > to either hotplug RAM or to map foreign pages from other domains, and
> > > > > > hence memory hotplug when running on Xen can be used even without the
> > > > > > user explicitly requesting it, as part of the normal operations of the
> > > > > > OS when attempting to map memory from a different domain.
> > > > > >
> > > > > > Setting a different default value of memhp_default_online_type when
> > > > > > attaching the balloon driver is not a robust solution, as the user (or
> > > > > > distro init scripts) could still change it and thus break the Xen
> > > > > > balloon driver.
> > > > >
> > > > > I think we discussed this a couple of times before (even triggered by my
> > > > > request), and this is responsibility of user space to configure. Usually
> > > > > distros have udev rules to online memory automatically. Especially, user
> > > > > space should eb able to configure *how* to online memory.
> > > >
> > > > Note (as per the commit message) that in the specific case I'm
> > > > referring to the memory hotplugged by the Xen balloon driver will be
> > > > an unpopulated range to be used internally by certain Xen subsystems,
> > > > like the xen-blkback or the privcmd drivers. The addition of such
> > > > blocks of (unpopulated) memory can happen without the user explicitly
> > > > requesting it, and hence not even aware such hotplug process is taking
> > > > place. To be clear: no actual RAM will be added to the system.
> > > >
> > > > Failure to online such blocks using the Xen specific online handler
> > > > (which does not handle back the memory to the allocator in any way)
> > > > will result in the system getting stuck and malfunctioning.
> > > >
> > > > > It's the admin/distro responsibility to configure this properly. In case
> > > > > this doesn't happen (or as you say, users change it), bad luck.
> > > > >
> > > > > E.g., virtio-mem takes care to not add more memory in case it is not
> > > > > getting onlined. I remember hyper-v has similar code to at least wait a
> > > > > bit for memory to get onlined.
> > > >
> > > > I don't think VirtIO or Hyper-V use the hotplug system in the same way
> > > > as Xen, as said this is done to add unpopulated memory regions that
> > > > will be used to map foreign memory (from other domains) by Xen drivers
> > > > on the system.
> > > >
> > > > Maybe this should somehow use a different mechanism to hotplug such
> > > > empty memory blocks? I don't mind doing this differently, but I would
> > > > need some pointers. Allowing user-space to change a (seemingly
> > > > unrelated) parameter and as a result produce failures on Xen drivers
> > > > is not an acceptable solution IMO.
> > >
> > > Maybe we can use the same approach as Xen PV-domains: pre-allocate a
> > > region in the memory map to be used for mapping foreign pages. For the
> > > kernel it will look like pre-ballooned memory, so it will create struct
> > > page for the region (which is what we are after), but it won't give the
> > > memory to the allocator.
> >
> > IMO using something similar to memory hotplug would give us more
> > flexibility, and TBH the logic is already there in the balloon driver.
> > It seems quite wasteful to allocate such region(s) beforehand for all
> > domains, even when most of them won't end up using foreign mappings at
> > all.
>
> We can do it for dom0 only per default, and add a boot parameter e.g.
> for driver domains.
>
> And the logic is already there (just pv-only right now).
>
> >
> > Anyway, I'm going to take a look at how to do that, I guess it's going
> > to involve playing with the memory map and reserving some space.
>
> Look at arch/x86/xen/setup.c (xen_add_extra_mem() and its usage).

Yes, I've taken a look. It's my rough understanding that I would need
to add a hook for HVM/PVH that modifies the memory map in order to add
an extra region (or regions) that would be marked as reserved using
memblock_reserve by xen_add_extra_mem.

Adding such hook for PVH guests booted using the PVH entry point and
fetching the memory map using the hypercall interface
(mem_map_via_hcall) seems feasible, however I'm not sure dealing with
other guests types is that easy.

> >
> > I suggest we should remove the Xen balloon hotplug logic, as it's not
> > working properly and we don't have a plan to fix it.
>
> I have used memory hotplug successfully not very long ago.

Right, but it requires a certain set of enabled options, which IMO is
not obvious. For example enabling xen_hotplug_unpopulated without also
setting the default memory hotplug policy to online the added blocks
will result in processes getting stuck. This is IMO too fragile.

Roger.

2020-07-23 13:48:25

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On 23.07.20 15:22, David Hildenbrand wrote:
> On 23.07.20 14:23, Roger Pau Monné wrote:
>> On Thu, Jul 23, 2020 at 01:37:03PM +0200, David Hildenbrand wrote:
>>> On 23.07.20 10:45, Roger Pau Monne wrote:
>>>> Add an extra option to add_memory_resource that overrides the memory
>>>> hotplug online behavior in order to force onlining of memory from
>>>> add_memory_resource unconditionally.
>>>>
>>>> This is required for the Xen balloon driver, that must run the
>>>> online page callback in order to correctly process the newly added
>>>> memory region, note this is an unpopulated region that is used by Linux
>>>> to either hotplug RAM or to map foreign pages from other domains, and
>>>> hence memory hotplug when running on Xen can be used even without the
>>>> user explicitly requesting it, as part of the normal operations of the
>>>> OS when attempting to map memory from a different domain.
>>>>
>>>> Setting a different default value of memhp_default_online_type when
>>>> attaching the balloon driver is not a robust solution, as the user (or
>>>> distro init scripts) could still change it and thus break the Xen
>>>> balloon driver.
>>>
>>> I think we discussed this a couple of times before (even triggered by my
>>> request), and this is responsibility of user space to configure. Usually
>>> distros have udev rules to online memory automatically. Especially, user
>>> space should eb able to configure *how* to online memory.
>>
>> Note (as per the commit message) that in the specific case I'm
>> referring to the memory hotplugged by the Xen balloon driver will be
>> an unpopulated range to be used internally by certain Xen subsystems,
>> like the xen-blkback or the privcmd drivers. The addition of such
>> blocks of (unpopulated) memory can happen without the user explicitly
>> requesting it, and hence not even aware such hotplug process is taking
>> place. To be clear: no actual RAM will be added to the system.
>
> Okay, but there is also the case where XEN will actually hotplug memory
> using this same handler IIRC (at least I've read papers about it). Both
> are using the same handler, correct?
>
>>
>>> It's the admin/distro responsibility to configure this properly. In case
>>> this doesn't happen (or as you say, users change it), bad luck.
>>>
>>> E.g., virtio-mem takes care to not add more memory in case it is not
>>> getting onlined. I remember hyper-v has similar code to at least wait a
>>> bit for memory to get onlined.
>>
>> I don't think VirtIO or Hyper-V use the hotplug system in the same way
>> as Xen, as said this is done to add unpopulated memory regions that
>> will be used to map foreign memory (from other domains) by Xen drivers
>> on the system.
>
> Indeed, if the memory is never exposed to the buddy (and all you need is
> struct pages + a kernel virtual mapping), I wonder if
> memremap/ZONE_DEVICE is what you want? Then you won't have user-visible
> memory blocks created with unclear online semantics, partially involving
> the buddy.

And just a note that there is also DCSS on s390x / z/VM which allows to
map segments into the VM physical address space (e.g., you can share
segments between VMs). They don't need any memmap (struct page) for that
memory, though. All they do is create the identity mapping in the kernel
virtual address space manually. Not sure what the exact requirements on
the XEN side are. I assume you need a memmap for this memory.

--
Thanks,

David / dhildenb

2020-07-23 13:50:03

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On 23.07.20 15:39, Roger Pau Monné wrote:
> On Thu, Jul 23, 2020 at 03:20:55PM +0200, Jürgen Groß wrote:
>> On 23.07.20 15:08, Roger Pau Monné wrote:
>>> On Thu, Jul 23, 2020 at 02:28:13PM +0200, Jürgen Groß wrote:
>>>> On 23.07.20 14:23, Roger Pau Monné wrote:
>>>>> On Thu, Jul 23, 2020 at 01:37:03PM +0200, David Hildenbrand wrote:
>>>>>> On 23.07.20 10:45, Roger Pau Monne wrote:
>>>>>>> Add an extra option to add_memory_resource that overrides the memory
>>>>>>> hotplug online behavior in order to force onlining of memory from
>>>>>>> add_memory_resource unconditionally.
>>>>>>>
>>>>>>> This is required for the Xen balloon driver, that must run the
>>>>>>> online page callback in order to correctly process the newly added
>>>>>>> memory region, note this is an unpopulated region that is used by Linux
>>>>>>> to either hotplug RAM or to map foreign pages from other domains, and
>>>>>>> hence memory hotplug when running on Xen can be used even without the
>>>>>>> user explicitly requesting it, as part of the normal operations of the
>>>>>>> OS when attempting to map memory from a different domain.
>>>>>>>
>>>>>>> Setting a different default value of memhp_default_online_type when
>>>>>>> attaching the balloon driver is not a robust solution, as the user (or
>>>>>>> distro init scripts) could still change it and thus break the Xen
>>>>>>> balloon driver.
>>>>>>
>>>>>> I think we discussed this a couple of times before (even triggered by my
>>>>>> request), and this is responsibility of user space to configure. Usually
>>>>>> distros have udev rules to online memory automatically. Especially, user
>>>>>> space should eb able to configure *how* to online memory.
>>>>>
>>>>> Note (as per the commit message) that in the specific case I'm
>>>>> referring to the memory hotplugged by the Xen balloon driver will be
>>>>> an unpopulated range to be used internally by certain Xen subsystems,
>>>>> like the xen-blkback or the privcmd drivers. The addition of such
>>>>> blocks of (unpopulated) memory can happen without the user explicitly
>>>>> requesting it, and hence not even aware such hotplug process is taking
>>>>> place. To be clear: no actual RAM will be added to the system.
>>>>>
>>>>> Failure to online such blocks using the Xen specific online handler
>>>>> (which does not handle back the memory to the allocator in any way)
>>>>> will result in the system getting stuck and malfunctioning.
>>>>>
>>>>>> It's the admin/distro responsibility to configure this properly. In case
>>>>>> this doesn't happen (or as you say, users change it), bad luck.
>>>>>>
>>>>>> E.g., virtio-mem takes care to not add more memory in case it is not
>>>>>> getting onlined. I remember hyper-v has similar code to at least wait a
>>>>>> bit for memory to get onlined.
>>>>>
>>>>> I don't think VirtIO or Hyper-V use the hotplug system in the same way
>>>>> as Xen, as said this is done to add unpopulated memory regions that
>>>>> will be used to map foreign memory (from other domains) by Xen drivers
>>>>> on the system.
>>>>>
>>>>> Maybe this should somehow use a different mechanism to hotplug such
>>>>> empty memory blocks? I don't mind doing this differently, but I would
>>>>> need some pointers. Allowing user-space to change a (seemingly
>>>>> unrelated) parameter and as a result produce failures on Xen drivers
>>>>> is not an acceptable solution IMO.
>>>>
>>>> Maybe we can use the same approach as Xen PV-domains: pre-allocate a
>>>> region in the memory map to be used for mapping foreign pages. For the
>>>> kernel it will look like pre-ballooned memory, so it will create struct
>>>> page for the region (which is what we are after), but it won't give the
>>>> memory to the allocator.
>>>
>>> IMO using something similar to memory hotplug would give us more
>>> flexibility, and TBH the logic is already there in the balloon driver.
>>> It seems quite wasteful to allocate such region(s) beforehand for all
>>> domains, even when most of them won't end up using foreign mappings at
>>> all.
>>
>> We can do it for dom0 only per default, and add a boot parameter e.g.
>> for driver domains.
>>
>> And the logic is already there (just pv-only right now).
>>
>>>
>>> Anyway, I'm going to take a look at how to do that, I guess it's going
>>> to involve playing with the memory map and reserving some space.
>>
>> Look at arch/x86/xen/setup.c (xen_add_extra_mem() and its usage).
>
> Yes, I've taken a look. It's my rough understanding that I would need
> to add a hook for HVM/PVH that modifies the memory map in order to add
> an extra region (or regions) that would be marked as reserved using
> memblock_reserve by xen_add_extra_mem.
>
> Adding such hook for PVH guests booted using the PVH entry point and
> fetching the memory map using the hypercall interface
> (mem_map_via_hcall) seems feasible, however I'm not sure dealing with
> other guests types is that easy.

I think for dom0 we can just use the existing logic using the host
memory map for selecting which region to use (possibly the size could
be specified as boot parameter in order to overwrite the default).

For domUs we'd need a boot parameter specifying either just the size
(resulting in a possible clash in case of pci passthrough) or
specifying the guest physical region for that additional area.

>
>>>
>>> I suggest we should remove the Xen balloon hotplug logic, as it's not
>>> working properly and we don't have a plan to fix it.
>>
>> I have used memory hotplug successfully not very long ago.
>
> Right, but it requires a certain set of enabled options, which IMO is
> not obvious. For example enabling xen_hotplug_unpopulated without also
> setting the default memory hotplug policy to online the added blocks
> will result in processes getting stuck. This is IMO too fragile.

Yes, memory hotplug is an item on my todo-list for some years now.

Juergen

2020-07-23 13:56:35

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On 23.07.20 15:47, David Hildenbrand wrote:
> On 23.07.20 15:22, David Hildenbrand wrote:
>> On 23.07.20 14:23, Roger Pau Monné wrote:
>>> On Thu, Jul 23, 2020 at 01:37:03PM +0200, David Hildenbrand wrote:
>>>> On 23.07.20 10:45, Roger Pau Monne wrote:
>>>>> Add an extra option to add_memory_resource that overrides the memory
>>>>> hotplug online behavior in order to force onlining of memory from
>>>>> add_memory_resource unconditionally.
>>>>>
>>>>> This is required for the Xen balloon driver, that must run the
>>>>> online page callback in order to correctly process the newly added
>>>>> memory region, note this is an unpopulated region that is used by Linux
>>>>> to either hotplug RAM or to map foreign pages from other domains, and
>>>>> hence memory hotplug when running on Xen can be used even without the
>>>>> user explicitly requesting it, as part of the normal operations of the
>>>>> OS when attempting to map memory from a different domain.
>>>>>
>>>>> Setting a different default value of memhp_default_online_type when
>>>>> attaching the balloon driver is not a robust solution, as the user (or
>>>>> distro init scripts) could still change it and thus break the Xen
>>>>> balloon driver.
>>>>
>>>> I think we discussed this a couple of times before (even triggered by my
>>>> request), and this is responsibility of user space to configure. Usually
>>>> distros have udev rules to online memory automatically. Especially, user
>>>> space should eb able to configure *how* to online memory.
>>>
>>> Note (as per the commit message) that in the specific case I'm
>>> referring to the memory hotplugged by the Xen balloon driver will be
>>> an unpopulated range to be used internally by certain Xen subsystems,
>>> like the xen-blkback or the privcmd drivers. The addition of such
>>> blocks of (unpopulated) memory can happen without the user explicitly
>>> requesting it, and hence not even aware such hotplug process is taking
>>> place. To be clear: no actual RAM will be added to the system.
>>
>> Okay, but there is also the case where XEN will actually hotplug memory
>> using this same handler IIRC (at least I've read papers about it). Both
>> are using the same handler, correct?
>>
>>>
>>>> It's the admin/distro responsibility to configure this properly. In case
>>>> this doesn't happen (or as you say, users change it), bad luck.
>>>>
>>>> E.g., virtio-mem takes care to not add more memory in case it is not
>>>> getting onlined. I remember hyper-v has similar code to at least wait a
>>>> bit for memory to get onlined.
>>>
>>> I don't think VirtIO or Hyper-V use the hotplug system in the same way
>>> as Xen, as said this is done to add unpopulated memory regions that
>>> will be used to map foreign memory (from other domains) by Xen drivers
>>> on the system.
>>
>> Indeed, if the memory is never exposed to the buddy (and all you need is
>> struct pages + a kernel virtual mapping), I wonder if
>> memremap/ZONE_DEVICE is what you want? Then you won't have user-visible
>> memory blocks created with unclear online semantics, partially involving
>> the buddy.
>
> And just a note that there is also DCSS on s390x / z/VM which allows to
> map segments into the VM physical address space (e.g., you can share
> segments between VMs). They don't need any memmap (struct page) for that
> memory, though. All they do is create the identity mapping in the kernel
> virtual address space manually. Not sure what the exact requirements on
> the XEN side are. I assume you need a memmap for this memory.

We need to be able to do I/O with that memory via normal drivers and we
need to be able to map it, both from user land and from the kernel.

Juergen

2020-07-23 14:03:36

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On Thu, Jul 23, 2020 at 03:22:49PM +0200, David Hildenbrand wrote:
> On 23.07.20 14:23, Roger Pau Monné wrote:
> > On Thu, Jul 23, 2020 at 01:37:03PM +0200, David Hildenbrand wrote:
> >> On 23.07.20 10:45, Roger Pau Monne wrote:
> >>> Add an extra option to add_memory_resource that overrides the memory
> >>> hotplug online behavior in order to force onlining of memory from
> >>> add_memory_resource unconditionally.
> >>>
> >>> This is required for the Xen balloon driver, that must run the
> >>> online page callback in order to correctly process the newly added
> >>> memory region, note this is an unpopulated region that is used by Linux
> >>> to either hotplug RAM or to map foreign pages from other domains, and
> >>> hence memory hotplug when running on Xen can be used even without the
> >>> user explicitly requesting it, as part of the normal operations of the
> >>> OS when attempting to map memory from a different domain.
> >>>
> >>> Setting a different default value of memhp_default_online_type when
> >>> attaching the balloon driver is not a robust solution, as the user (or
> >>> distro init scripts) could still change it and thus break the Xen
> >>> balloon driver.
> >>
> >> I think we discussed this a couple of times before (even triggered by my
> >> request), and this is responsibility of user space to configure. Usually
> >> distros have udev rules to online memory automatically. Especially, user
> >> space should eb able to configure *how* to online memory.
> >
> > Note (as per the commit message) that in the specific case I'm
> > referring to the memory hotplugged by the Xen balloon driver will be
> > an unpopulated range to be used internally by certain Xen subsystems,
> > like the xen-blkback or the privcmd drivers. The addition of such
> > blocks of (unpopulated) memory can happen without the user explicitly
> > requesting it, and hence not even aware such hotplug process is taking
> > place. To be clear: no actual RAM will be added to the system.
>
> Okay, but there is also the case where XEN will actually hotplug memory
> using this same handler IIRC (at least I've read papers about it). Both
> are using the same handler, correct?

Yes, it's used by this dual purpose, which I have to admit I don't
like that much either.

One set of pages should be clearly used for RAM memory hotplug, and
the other to map foreign pages that are not related to memory hotplug,
it's just that we happen to need a physical region with backing struct
pages.

> >
> >> It's the admin/distro responsibility to configure this properly. In case
> >> this doesn't happen (or as you say, users change it), bad luck.
> >>
> >> E.g., virtio-mem takes care to not add more memory in case it is not
> >> getting onlined. I remember hyper-v has similar code to at least wait a
> >> bit for memory to get onlined.
> >
> > I don't think VirtIO or Hyper-V use the hotplug system in the same way
> > as Xen, as said this is done to add unpopulated memory regions that
> > will be used to map foreign memory (from other domains) by Xen drivers
> > on the system.
>
> Indeed, if the memory is never exposed to the buddy (and all you need is
> struct pages + a kernel virtual mapping), I wonder if
> memremap/ZONE_DEVICE is what you want?

I'm certainly not familiar with the Linux memory subsystem, but if
that gets us a backing struct page and a kernel mapping then I would
say yes.

> Then you won't have user-visible
> memory blocks created with unclear online semantics, partially involving
> the buddy.

Seems like a fine solution.

Juergen: would you be OK to use a separate page-list for
alloc_xenballooned_pages on HVM/PVH using the logic described by
David?

I guess I would leave PV as-is, since it already has this reserved
region to map foreign pages.

Thanks, Roger.

2020-07-23 15:11:38

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On 23.07.20 15:59, Roger Pau Monné wrote:
> On Thu, Jul 23, 2020 at 03:22:49PM +0200, David Hildenbrand wrote:
>> On 23.07.20 14:23, Roger Pau Monné wrote:
>>> On Thu, Jul 23, 2020 at 01:37:03PM +0200, David Hildenbrand wrote:
>>>> On 23.07.20 10:45, Roger Pau Monne wrote:
>>>>> Add an extra option to add_memory_resource that overrides the memory
>>>>> hotplug online behavior in order to force onlining of memory from
>>>>> add_memory_resource unconditionally.
>>>>>
>>>>> This is required for the Xen balloon driver, that must run the
>>>>> online page callback in order to correctly process the newly added
>>>>> memory region, note this is an unpopulated region that is used by Linux
>>>>> to either hotplug RAM or to map foreign pages from other domains, and
>>>>> hence memory hotplug when running on Xen can be used even without the
>>>>> user explicitly requesting it, as part of the normal operations of the
>>>>> OS when attempting to map memory from a different domain.
>>>>>
>>>>> Setting a different default value of memhp_default_online_type when
>>>>> attaching the balloon driver is not a robust solution, as the user (or
>>>>> distro init scripts) could still change it and thus break the Xen
>>>>> balloon driver.
>>>>
>>>> I think we discussed this a couple of times before (even triggered by my
>>>> request), and this is responsibility of user space to configure. Usually
>>>> distros have udev rules to online memory automatically. Especially, user
>>>> space should eb able to configure *how* to online memory.
>>>
>>> Note (as per the commit message) that in the specific case I'm
>>> referring to the memory hotplugged by the Xen balloon driver will be
>>> an unpopulated range to be used internally by certain Xen subsystems,
>>> like the xen-blkback or the privcmd drivers. The addition of such
>>> blocks of (unpopulated) memory can happen without the user explicitly
>>> requesting it, and hence not even aware such hotplug process is taking
>>> place. To be clear: no actual RAM will be added to the system.
>>
>> Okay, but there is also the case where XEN will actually hotplug memory
>> using this same handler IIRC (at least I've read papers about it). Both
>> are using the same handler, correct?
>
> Yes, it's used by this dual purpose, which I have to admit I don't
> like that much either.
>
> One set of pages should be clearly used for RAM memory hotplug, and
> the other to map foreign pages that are not related to memory hotplug,
> it's just that we happen to need a physical region with backing struct
> pages.
>
>>>
>>>> It's the admin/distro responsibility to configure this properly. In case
>>>> this doesn't happen (or as you say, users change it), bad luck.
>>>>
>>>> E.g., virtio-mem takes care to not add more memory in case it is not
>>>> getting onlined. I remember hyper-v has similar code to at least wait a
>>>> bit for memory to get onlined.
>>>
>>> I don't think VirtIO or Hyper-V use the hotplug system in the same way
>>> as Xen, as said this is done to add unpopulated memory regions that
>>> will be used to map foreign memory (from other domains) by Xen drivers
>>> on the system.
>>
>> Indeed, if the memory is never exposed to the buddy (and all you need is
>> struct pages + a kernel virtual mapping), I wonder if
>> memremap/ZONE_DEVICE is what you want?
>
> I'm certainly not familiar with the Linux memory subsystem, but if
> that gets us a backing struct page and a kernel mapping then I would
> say yes.
>
>> Then you won't have user-visible
>> memory blocks created with unclear online semantics, partially involving
>> the buddy.
>
> Seems like a fine solution.
>
> Juergen: would you be OK to use a separate page-list for
> alloc_xenballooned_pages on HVM/PVH using the logic described by
> David?
>
> I guess I would leave PV as-is, since it already has this reserved
> region to map foreign pages.

I would really like a common solution, especially as it would enable
pv driver domains to use that feature, too.

And finding a region for this memory zone in PVH dom0 should be common
with PV dom0 after all. We don't want to collide with either PCI space
or hotplug memory.

Juergen

2020-07-23 16:06:14

by Andrew Cooper

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On 23/07/2020 16:10, Jürgen Groß wrote:
> On 23.07.20 15:59, Roger Pau Monné wrote:
>> On Thu, Jul 23, 2020 at 03:22:49PM +0200, David Hildenbrand wrote:
>>> On 23.07.20 14:23, Roger Pau Monné wrote:
>>>> On Thu, Jul 23, 2020 at 01:37:03PM +0200, David Hildenbrand wrote:
>>>>> On 23.07.20 10:45, Roger Pau Monne wrote:
>>>>>> Add an extra option to add_memory_resource that overrides the memory
>>>>>> hotplug online behavior in order to force onlining of memory from
>>>>>> add_memory_resource unconditionally.
>>>>>>
>>>>>> This is required for the Xen balloon driver, that must run the
>>>>>> online page callback in order to correctly process the newly added
>>>>>> memory region, note this is an unpopulated region that is used by
>>>>>> Linux
>>>>>> to either hotplug RAM or to map foreign pages from other domains,
>>>>>> and
>>>>>> hence memory hotplug when running on Xen can be used even without
>>>>>> the
>>>>>> user explicitly requesting it, as part of the normal operations
>>>>>> of the
>>>>>> OS when attempting to map memory from a different domain.
>>>>>>
>>>>>> Setting a different default value of memhp_default_online_type when
>>>>>> attaching the balloon driver is not a robust solution, as the
>>>>>> user (or
>>>>>> distro init scripts) could still change it and thus break the Xen
>>>>>> balloon driver.
>>>>>
>>>>> I think we discussed this a couple of times before (even triggered
>>>>> by my
>>>>> request), and this is responsibility of user space to configure.
>>>>> Usually
>>>>> distros have udev rules to online memory automatically.
>>>>> Especially, user
>>>>> space should eb able to configure *how* to online memory.
>>>>
>>>> Note (as per the commit message) that in the specific case I'm
>>>> referring to the memory hotplugged by the Xen balloon driver will be
>>>> an unpopulated range to be used internally by certain Xen subsystems,
>>>> like the xen-blkback or the privcmd drivers. The addition of such
>>>> blocks of (unpopulated) memory can happen without the user explicitly
>>>> requesting it, and hence not even aware such hotplug process is taking
>>>> place. To be clear: no actual RAM will be added to the system.
>>>
>>> Okay, but there is also the case where XEN will actually hotplug memory
>>> using this same handler IIRC (at least I've read papers about it). Both
>>> are using the same handler, correct?
>>
>> Yes, it's used by this dual purpose, which I have to admit I don't
>> like that much either.
>>
>> One set of pages should be clearly used for RAM memory hotplug, and
>> the other to map foreign pages that are not related to memory hotplug,
>> it's just that we happen to need a physical region with backing struct
>> pages.
>>
>>>>
>>>>> It's the admin/distro responsibility to configure this properly.
>>>>> In case
>>>>> this doesn't happen (or as you say, users change it), bad luck.
>>>>>
>>>>> E.g., virtio-mem takes care to not add more memory in case it is not
>>>>> getting onlined. I remember hyper-v has similar code to at least
>>>>> wait a
>>>>> bit for memory to get onlined.
>>>>
>>>> I don't think VirtIO or Hyper-V use the hotplug system in the same way
>>>> as Xen, as said this is done to add unpopulated memory regions that
>>>> will be used to map foreign memory (from other domains) by Xen drivers
>>>> on the system.
>>>
>>> Indeed, if the memory is never exposed to the buddy (and all you
>>> need is
>>> struct pages + a kernel virtual mapping), I wonder if
>>> memremap/ZONE_DEVICE is what you want?
>>
>> I'm certainly not familiar with the Linux memory subsystem, but if
>> that gets us a backing struct page and a kernel mapping then I would
>> say yes.
>>
>>> Then you won't have user-visible
>>> memory blocks created with unclear online semantics, partially
>>> involving
>>> the buddy.
>>
>> Seems like a fine solution.
>>
>> Juergen: would you be OK to use a separate page-list for
>> alloc_xenballooned_pages on HVM/PVH using the logic described by
>> David?
>>
>> I guess I would leave PV as-is, since it already has this reserved
>> region to map foreign pages.
>
> I would really like a common solution, especially as it would enable
> pv driver domains to use that feature, too.
>
> And finding a region for this memory zone in PVH dom0 should be common
> with PV dom0 after all. We don't want to collide with either PCI space
> or hotplug memory.

While I agree with goal here, these are two very different things, due
to the completely different nature of PV and HVM/PVH guests.

HVM/PVH guests have a concrete guest physical address space. Linux
needs to pick some gfn's to use which aren't used by anything else (and
Xen's behaviour of not providing any help here is deeply unhelpful, and
needs fixing), and get struct page_info's for them.

PV is totally different. Linux still needs page_info's for them, but
there is no concept of a guest physical address space. You can
literally gain access to foreign mappings or grant maps by asking Xen to
modify a PTE. For convenience with the core code, Linux tries to map
this concept back into a 1:1 pfn space, but it is quite fictitious.

~Andrew

2020-07-23 16:25:48

[permalink] [raw]

Subject: Re: [PATCH 3/3] memory: introduce an option to force onlining of hotplug memory

On Thu, Jul 23, 2020 at 05:10:03PM +0200, Jürgen Groß wrote:
> On 23.07.20 15:59, Roger Pau Monné wrote:
> > On Thu, Jul 23, 2020 at 03:22:49PM +0200, David Hildenbrand wrote:
> > > On 23.07.20 14:23, Roger Pau Monné wrote:
> > > > On Thu, Jul 23, 2020 at 01:37:03PM +0200, David Hildenbrand wrote:
> > > > > On 23.07.20 10:45, Roger Pau Monne wrote:
> > > > > > Add an extra option to add_memory_resource that overrides the memory
> > > > > > hotplug online behavior in order to force onlining of memory from
> > > > > > add_memory_resource unconditionally.
> > > > > >
> > > > > > This is required for the Xen balloon driver, that must run the
> > > > > > online page callback in order to correctly process the newly added
> > > > > > memory region, note this is an unpopulated region that is used by Linux
> > > > > > to either hotplug RAM or to map foreign pages from other domains, and
> > > > > > hence memory hotplug when running on Xen can be used even without the
> > > > > > user explicitly requesting it, as part of the normal operations of the
> > > > > > OS when attempting to map memory from a different domain.
> > > > > >
> > > > > > Setting a different default value of memhp_default_online_type when
> > > > > > attaching the balloon driver is not a robust solution, as the user (or
> > > > > > distro init scripts) could still change it and thus break the Xen
> > > > > > balloon driver.
> > > > >
> > > > > I think we discussed this a couple of times before (even triggered by my
> > > > > request), and this is responsibility of user space to configure. Usually
> > > > > distros have udev rules to online memory automatically. Especially, user
> > > > > space should eb able to configure *how* to online memory.
> > > >
> > > > Note (as per the commit message) that in the specific case I'm
> > > > referring to the memory hotplugged by the Xen balloon driver will be
> > > > an unpopulated range to be used internally by certain Xen subsystems,
> > > > like the xen-blkback or the privcmd drivers. The addition of such
> > > > blocks of (unpopulated) memory can happen without the user explicitly
> > > > requesting it, and hence not even aware such hotplug process is taking
> > > > place. To be clear: no actual RAM will be added to the system.
> > >
> > > Okay, but there is also the case where XEN will actually hotplug memory
> > > using this same handler IIRC (at least I've read papers about it). Both
> > > are using the same handler, correct?
> >
> > Yes, it's used by this dual purpose, which I have to admit I don't
> > like that much either.
> >
> > One set of pages should be clearly used for RAM memory hotplug, and
> > the other to map foreign pages that are not related to memory hotplug,
> > it's just that we happen to need a physical region with backing struct
> > pages.
> >
> > > >
> > > > > It's the admin/distro responsibility to configure this properly. In case
> > > > > this doesn't happen (or as you say, users change it), bad luck.
> > > > >
> > > > > E.g., virtio-mem takes care to not add more memory in case it is not
> > > > > getting onlined. I remember hyper-v has similar code to at least wait a
> > > > > bit for memory to get onlined.
> > > >
> > > > I don't think VirtIO or Hyper-V use the hotplug system in the same way
> > > > as Xen, as said this is done to add unpopulated memory regions that
> > > > will be used to map foreign memory (from other domains) by Xen drivers
> > > > on the system.
> > >
> > > Indeed, if the memory is never exposed to the buddy (and all you need is
> > > struct pages + a kernel virtual mapping), I wonder if
> > > memremap/ZONE_DEVICE is what you want?
> >
> > I'm certainly not familiar with the Linux memory subsystem, but if
> > that gets us a backing struct page and a kernel mapping then I would
> > say yes.
> >
> > > Then you won't have user-visible
> > > memory blocks created with unclear online semantics, partially involving
> > > the buddy.
> >
> > Seems like a fine solution.
> >
> > Juergen: would you be OK to use a separate page-list for
> > alloc_xenballooned_pages on HVM/PVH using the logic described by
> > David?
> >
> > I guess I would leave PV as-is, since it already has this reserved
> > region to map foreign pages.
>
> I would really like a common solution, especially as it would enable
> pv driver domains to use that feature, too.

I think PV is much more easy on that regard, as it doesn't have MMIO
holes except when using PCI passthrough, and in that case it's
trivial to identify those.

However on HVM/PVH this is not so trivial. I'm certainly not opposing
to a solution that covers both, but ATM I would really like to get
something working for PVH dom0, or else it's not usable on Linux IMO.

> And finding a region for this memory zone in PVH dom0 should be common
> with PV dom0 after all. We don't want to collide with either PCI space
> or hotplug memory.

Right, we could use the native memory map for that on dom0, and maybe
create a custom resource for the Xen balloon driver instead of
allocating from iomem_resource?

DomUs are more tricky as a guest has no idea where mappings can be
safely placed, maybe we will have to resort to iomem_resource in that
case, as I don't see much other option due to the lack of information
from Xen.

I also think that ZONE_DEVICE will need some adjustments, for once the
types in memory_type don't seem to be suitable for Xen, as they are
either specific to DAX or PCI. I gave a try at using allocate_resource
plus memremap_pages but that didn't seem to fly, I will need to
investigate further.

Maybe we can resort to something even simpler than memremap_pages? I
certainly have very little idea of how this is supposed to be used,
but dev_pagemap seems overly complex for what we are trying to
achieve.

Thanks, Roger.

2020-07-23 17:40:38