Let's try to unplug completely offline big blocks first. Then, (if
enabled via unplug_offline) try to offline and remove whole big blocks.
No locking necessary - we can deal with concurrent onlining/offlining
just fine.
Note1: This is sub-optimal and might be dangerous in some environments: we
could end up in an infinite loop when offlining (e.g., long-term pinnings),
similar as with DIMMs. We'll introduce safe memory hotunplug via
fake-offlining next, and use this basic mode only when explicitly enabled.
Note2: Without ZONE_MOVABLE, memory unplug will be extremely unreliable
with bigger block sizes.
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David Hildenbrand <david@redhat.com>
Link: https://lore.kernel.org/r/20201112133815.13332-29-david@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Currently, we do not support device block sizes that exceed the Linux
memory block size. For example, having a device block size of 1 GiB (e.g.,
gigantic pages in the hypervisor) won't work with 128 MiB Linux memory
blocks.
Let's implement Big Block Mode (BBM), whereby we add/remove at least
one Linux memory block at a time. With a 1 GiB device block size, a Big
Block (BB) will cover 8 Linux memory blocks.
We'll keep registering the online_page_callback machinery, it will be used
for safe memory hotunplug in BBM next.
Note: BBM is properly prepared for variable-sized Linux memory
blocks that we might see in the future. So we won't care how many Linux
memory blocks a big block actually spans, and how the memory notifier is
called.
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David Hildenbrand <david@redhat.com>
Link: https://lore.kernel.org/r/20201112133815.13332-26-david@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Let's use wrappers for the low-level functions that dev_dbg/dev_warn
and work on addr + size, such that we can reuse them for adding/removing
in other granularity.
We only warn when adding memory failed, because that's something to pay
attention to. We won't warn when removing failed, we'll reuse that in
racy context soon (and we do have proper BUG_ON() statements in the
current cases where it must never happen).
Reviewed-by: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Link: https://lore.kernel.org/r/20201112133815.13332-25-david@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
We don't want to add too much memory when it's not getting onlined
immediately, to avoid running OOM. Generalize the handling, to avoid
making use of memory block states. Use a threshold of 1 GiB for now.
Properly adjust the offline size when adding/removing memory. As we are
not always protected by a lock when touching the offline size, use an
atomic64_t. We don't care about races (e.g., someone offlining memory
while we are adding more), only about consistent values.
(1 GiB needs a memmap of ~16MiB - which sounds reasonable even for
setups with little boot memory and (possibly) one virtio-mem device per
node)
We don't want to retrigger when onlining is caused immediately by our
action (e.g., adding memory which immediately gets onlined), so use a
flag to indicate if the workqueue is active and use that as an
indicator whether to trigger a retry. This will also be especially relevant
for Big Block Mode (BBM), whereby we might re-online memory in case
offlining of another memory block failed.
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Link: https://lore.kernel.org/r/20201112133815.13332-17-david@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Let's trigger from offlining code only when we're not allowed to unplug
online memory. Handle the other case (memmap possibly freeing up another
memory block) when actually removing memory. We now also properly handle
the case when removing already offline memory blocks via
virtio_mem_mb_remove(). When removing via virtio_mem_remove(), when
unloading the driver, virtio_mem_retry() is a NOP and safe to use.
While at it, move retry handling when offlining out of
virtio_mem_notify_offline(), to share it with Big Block Mode (BBM)
soon.
This is a preparation for Big Block Mode (BBM), whereby we can see some
temporary offlining of memory blocks without actually making progress.
Imagine you have a Big Block that spans to Linux memory blocks. Assume
the first Linux memory blocks has no unmovable data on it. When we would
call offline_and_remove_memory() on the big block, we would
1. Try to offline the first block. Works, notifiers triggered.
virtio_mem_retry() called.
2. Try to offline the second block. Does not work.
3. Re-online first block.
4. Exit to main loop, exit workqueue.
5. Retry immediately (due to virtio_mem_retry()), go to 1.
The result are endless retries.
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Link: https://lore.kernel.org/r/20201112133815.13332-16-david@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Pull virtio updates from Michael Tsirkin:
"vhost, vdpa, and virtio cleanups and fixes
A very quiet cycle, no new features"
* tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost:
MAINTAINERS: add URL for virtio-mem
vhost_vdpa: remove unnecessary spin_lock in vhost_vring_call
vringh: fix __vringh_iov() when riov and wiov are different
vdpa/mlx5: Setup driver only if VIRTIO_CONFIG_S_DRIVER_OK
s390: virtio: PV needs VIRTIO I/O device protection
virtio: let arch advertise guest's memory access restrictions
vhost_vdpa: Fix duplicate included kernel.h
vhost: reduce stack usage in log_used
virtio-mem: Constify mem_id_table
virtio_input: Constify id_table
virtio-balloon: Constify id_table
vdpa/mlx5: Fix failure to bring link up
vdpa/mlx5: Make use of a specific 16 bit endianness API
virtio-mem adds memory in memory block granularity, to be able to remove
it in the same granularity again later, and to grow slowly on demand.
This, however, results in quite a lot of resources when adding a lot of
memory. Resources are effectively stored in a list-based tree. Having a
lot of resources not only wastes memory, it also makes traversing that
tree more expensive, and makes /proc/iomem explode in size (e.g.,
requiring kexec-tools to manually merge resources later when e.g., trying
to create a kdump header).
Before this patch, we get (/proc/iomem) when hotplugging 2G via virtio-mem
on x86-64:
[...]
100000000-13fffffff : System RAM
140000000-33fffffff : virtio0
140000000-147ffffff : System RAM (virtio_mem)
148000000-14fffffff : System RAM (virtio_mem)
150000000-157ffffff : System RAM (virtio_mem)
158000000-15fffffff : System RAM (virtio_mem)
160000000-167ffffff : System RAM (virtio_mem)
168000000-16fffffff : System RAM (virtio_mem)
170000000-177ffffff : System RAM (virtio_mem)
178000000-17fffffff : System RAM (virtio_mem)
180000000-187ffffff : System RAM (virtio_mem)
188000000-18fffffff : System RAM (virtio_mem)
190000000-197ffffff : System RAM (virtio_mem)
198000000-19fffffff : System RAM (virtio_mem)
1a0000000-1a7ffffff : System RAM (virtio_mem)
1a8000000-1afffffff : System RAM (virtio_mem)
1b0000000-1b7ffffff : System RAM (virtio_mem)
1b8000000-1bfffffff : System RAM (virtio_mem)
3280000000-32ffffffff : PCI Bus 0000:00
With this patch, we get (/proc/iomem):
[...]
fffc0000-ffffffff : Reserved
100000000-13fffffff : System RAM
140000000-33fffffff : virtio0
140000000-1bfffffff : System RAM (virtio_mem)
3280000000-32ffffffff : PCI Bus 0000:00
Of course, with more hotplugged memory, it gets worse. When unplugging
memory blocks again, try_remove_memory() (via offline_and_remove_memory())
will properly split the resource up again.
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Wei Yang <richardw.yang@linux.intel.com>
Cc: Anton Blanchard <anton@ozlabs.org>
Cc: Ard Biesheuvel <ardb@kernel.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Haiyang Zhang <haiyangz@microsoft.com>
Cc: Heiko Carstens <hca@linux.ibm.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: Juergen Gross <jgross@suse.com>
Cc: Julien Grall <julien@xen.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: "K. Y. Srinivasan" <kys@microsoft.com>
Cc: Len Brown <lenb@kernel.org>
Cc: Leonardo Bras <leobras.c@gmail.com>
Cc: Libor Pechacek <lpechacek@suse.cz>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Nathan Lynch <nathanl@linux.ibm.com>
Cc: "Oliver O'Halloran" <oohall@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Pingfan Liu <kernelfans@gmail.com>
Cc: "Rafael J. Wysocki" <rjw@rjwysocki.net>
Cc: Roger Pau Monné <roger.pau@citrix.com>
Cc: Stefano Stabellini <sstabellini@kernel.org>
Cc: Stephen Hemminger <sthemmin@microsoft.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vasily Gorbik <gor@linux.ibm.com>
Cc: Vishal Verma <vishal.l.verma@intel.com>
Cc: Wei Liu <wei.liu@kernel.org>
Link: https://lkml.kernel.org/r/20200911103459.10306-7-david@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When introducing virtio-mem, the semantics of ZONE_MOVABLE were rather
unclear, which is why we special-cased ZONE_MOVABLE such that partially
plugged blocks would never end up in ZONE_MOVABLE.
Now that the semantics are much clearer (and will be documented in a
follow-up patch including the new virtio-mem behavior), let's allow to
online partially plugged memory blocks to ZONE_MOVABLE and also consider
memory blocks that were onlined to ZONE_MOVABLE when unplugging memory.
While unplugged memory pages are, in general, unmovable, they can be
skipped when offlining memory.
virtio-mem only unplugs fairly big chunks (in the megabyte range) and
rather tries to shrink the memory region than randomly choosing memory.
In theory, if all other pages in the movable zone would be movable,
virtio-mem would only shrink that zone and not create any kind of
fragmentation.
In the future, we might want to remember the zone again and use the
information when (un)plugging memory. For now, let's keep it simple.
Note: Support for defragmentation is planned, to deal with fragmentation
after unplug due to memory chunks within memory blocks that could not get
unplugged before (e.g., somebody pinning pages within ZONE_MOVABLE for a
longer time).
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Qian Cai <cai@lca.pw>
Link: http://lkml.kernel.org/r/20200816125333.7434-6-david@redhat.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Virtio-mem managed memory is always detected and added by the virtio-mem
driver, never using something like the firmware-provided memory map.
This is the case after an ordinary system reboot, and has to be guaranteed
after kexec. Especially, virtio-mem added memory resources can contain
inaccessible parts ("unblocked memory blocks"), blindly forwarding them
to a kexec kernel is dangerous, as unplugged memory will get accessed
(esp. written).
Let's use the new way of adding special driver-managed memory introduced
in commit 7b7b27214b ("mm/memory_hotplug: introduce
add_memory_driver_managed()").
This will result in no entries in /sys/firmware/memmap ("raw firmware-
provided memory map"), the memory resource will be flagged
IORESOURCE_MEM_DRIVER_MANAGED (esp., kexec_file_load() will not place
kexec images on this memory), and it is exposed as "System RAM
(virtio_mem)" in /proc/iomem, so esp. kexec-tools can properly handle it.
Example /proc/iomem before this change:
[...]
140000000-333ffffff : virtio0
140000000-147ffffff : System RAM
334000000-533ffffff : virtio1
338000000-33fffffff : System RAM
340000000-347ffffff : System RAM
348000000-34fffffff : System RAM
[...]
Example /proc/iomem after this change:
[...]
140000000-333ffffff : virtio0
140000000-147ffffff : System RAM (virtio_mem)
334000000-533ffffff : virtio1
338000000-33fffffff : System RAM (virtio_mem)
340000000-347ffffff : System RAM (virtio_mem)
348000000-34fffffff : System RAM (virtio_mem)
[...]
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: teawater <teawaterz@linux.alibaba.com>
Fixes: 5f1f79bbc9 ("virtio-mem: Paravirtualized memory hotplug")
Signed-off-by: David Hildenbrand <david@redhat.com>
Link: https://lore.kernel.org/r/20200611093518.5737-1-david@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Smatch complains that "rc" can be uninitialized if we hit the "break;"
statement on the first iteration through the loop. I suspect that this
can't happen in real life, but returning a zero literal is cleaner and
silence the static checker warning.
Fixes: 5f1f79bbc9 ("virtio-mem: Paravirtualized memory hotplug")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Link: https://lore.kernel.org/r/20200610085911.GC5439@mwanda
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
If subblock size is large (e.g. 1G) 32 bit math involving it
can overflow. Rather than try to catch all instances of that,
let's tweak block size to 64 bit.
It ripples through UAPI which is an ABI change, but it's not too late to
make it, and it will allow supporting >4Gbyte blocks while might
become necessary down the road.
Fixes: 5f1f79bbc9 ("virtio-mem: Paravirtualized memory hotplug")
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: David Hildenbrand <david@redhat.com>
rc is initialized to -ENIVAL but that's never used. Drop it.
Fixes: 5f1f79bbc9 ("virtio-mem: Paravirtualized memory hotplug")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: David Hildenbrand <david@redhat.com>
Right now, we always try to unplug single subblocks when processing an
online memory block. Let's try to unplug the complete online memory block
first, in case it is fully plugged and the unplug request is large
enough. Fallback to single subblocks in case the memory block cannot get
unplugged as a whole.
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Link: https://lore.kernel.org/r/20200507140139.17083-16-david@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Let's add a parent resource, named after the virtio device (inspired by
drivers/dax/kmem.c). This allows user space to identify which memory
belongs to which virtio-mem device.
With this change and two virtio-mem devices:
:/# cat /proc/iomem
00000000-00000fff : Reserved
00001000-0009fbff : System RAM
[...]
140000000-333ffffff : virtio0
140000000-147ffffff : System RAM
148000000-14fffffff : System RAM
150000000-157ffffff : System RAM
[...]
334000000-3033ffffff : virtio1
338000000-33fffffff : System RAM
340000000-347ffffff : System RAM
348000000-34fffffff : System RAM
[...]
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Link: https://lore.kernel.org/r/20200507140139.17083-12-david@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
We also want to unplug online memory (contained in online memory blocks
and, therefore, managed by the buddy), and eventually replug it later.
When requested to unplug memory, we use alloc_contig_range() to allocate
subblocks in online memory blocks (so we are the owner) and send them to
our hypervisor. When requested to plug memory, we can replug such memory
using free_contig_range() after asking our hypervisor.
We also want to mark all allocated pages PG_offline, so nobody will
touch them. To differentiate pages that were never onlined when
onlining the memory block from pages allocated via alloc_contig_range(), we
use PageDirty(). Based on this flag, virtio_mem_fake_online() can either
online the pages for the first time or use free_contig_range().
It is worth noting that there are no guarantees on how much memory can
actually get unplugged again. All device memory might completely be
fragmented with unmovable data, such that no subblock can get unplugged.
We are not touching the ZONE_MOVABLE. If memory is onlined to the
ZONE_MOVABLE, it can only get unplugged after that memory was offlined
manually by user space. In normal operation, virtio-mem memory is
suggested to be onlined to ZONE_NORMAL. In the future, we will try to
make unplug more likely to succeed.
Add a module parameter to control if online memory shall be touched.
As we want to access alloc_contig_range()/free_contig_range() from
kernel module context, export the symbols.
Note: Whenever virtio-mem uses alloc_contig_range(), all affected pages
are on the same node, in the same zone, and contain no holes.
Acked-by: Michal Hocko <mhocko@suse.com> # to export contig range allocator API
Tested-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Dave Young <dyoung@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com>
Cc: Alexander Potapenko <glider@google.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Link: https://lore.kernel.org/r/20200507140139.17083-6-david@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
