mm: Consider non-anon swap cache folios in folio_expected_ref_count()
Subsystems (2)
PATCH 1/1 mm: Consider non-anon swap cache folios in folio_expected_ref_count()
Currently, folio_expected_ref_count() only adds references for the swap cache if the folio is anonymous. However, according to the comment above the definition of PG_swapcache in enum pageflags, shmem folios can also have PG_swapcache set. This patch makes sure references for the swap cache are added if folio_test_swapcache(folio) is true.
This issue was found when trying to hot-unplug memory in a QEMU/KVM virtual machine. When initiating hot-unplug when most of the guest memory is allocated, hot-unplug hangs partway through removal due to migration failures. The following message would be printed several times, and would be printed again about every five seconds:
[ 49.641309] migrating pfn b12f25 failed ret:7 [ 49.641310] page: refcount:2 mapcount:0 mapping:0000000033bd8fe2 index:0x7f404d925 pfn:0xb12f25 [ 49.641311] aops:swap_aops [ 49.641313] flags: 0x300000000030508(uptodate|active|owner_priv_1|reclaim|swapbacked|node=0|zone=3) [ 49.641314] raw: 0300000000030508 ffffed312c4bc908 ffffed312c4bc9c8 0000000000000000 [ 49.641315] raw: 00000007f404d925 00000000000c823b 00000002ffffffff 0000000000000000 [ 49.641315] page dumped because: migration failure
When debugging this, I found that these migration failures were due to __migrate_folio() returning -EAGAIN for a small set of folios because the expected reference count it calculates via folio_expected_ref_count() is one less than the actual reference count of the folios. Furthermore, all of the affected folios were not anonymous, but had the PG_swapcache flag set, inspiring this patch. After applying this patch, the memory hot-unplug behaves as expected.
I tested this on a machine running Ubuntu 24.04 with kernel version 6.8.0-90-generic and 64GB of memory. The guest VM is managed by libvirt and runs Ubuntu 24.04 with kernel version 6.18 (though the head of the mm-unstable branch as a Dec 16, 2025 was also tested and behaves the same) and 48GB of memory. The libvirt XML definition for the VM can be found at [1]. CONFIG_MHP_DEFAULT_ONLINE_TYPE_ONLINE_MOVABLE is set in the guest kernel so the hot-pluggable memory is automatically onlined.
Below are the steps to reproduce this behavior:
1) Define and start and virtual machine
host$ virsh -c qemu:///system define ./test_vm.xml # test_vm.xml from [1] host$ virsh -c qemu:///system start test_vm
2) Setup swap in the guest
guest$ sudo fallocate -l 32G /swapfile
guest$ sudo chmod 0600 /swapfile
guest$ sudo mkswap /swapfile
guest$ sudo swapon /swapfile
3) Use alloc_data [2] to allocate most of the remaining guest memory guest$ ./alloc_data 45
4) In a separate guest terminal, monitor the amount of used memory guest$ watch -n1 free -h
5) When alloc_data has finished allocating, initiate the memory hot-unplug using the provided xml file [3]
host$ virsh -c qemu:///system detach-device test_vm ./remove.xml --live
After initiating the memory hot-unplug, you should see the amount of available memory in the guest decrease, and the amount of used swap data increase. If everything works as expected, when all of the memory is unplugged, there should be around 8.5-9GB of data in swap. If the unplugging is unsuccessful, the amount of used swap data will settle below that. If that happens, you should be able to see log messages in dmesg similar to the one posted above.
[1] https://github.com/BijanT/linux_patch_files/blob/main/test_vm.xml
[2] https://github.com/BijanT/linux_patch_files/blob/main/alloc_data.c
[3] https://github.com/BijanT/linux_patch_files/blob/main/remove.xml
---
I am not very familiar with the memory hot-(un)plug or swapping code, so I am not 100% certain if this patch actually solves the root of the problem. I believe the issue is from shmem folios, in which case I believe this patch is correct. However, I couldn't think of an easy way to confirm that the affected folios were from shmem. I guess it could be possible that the root cause could be from some bug where some anonymous pages do not return true to folio_test_anon(). I don't think that's the case, but figured the MM maintainers would have a better idea of what's going on.
--- include/linux/mm.h | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
include/linux/mm.h +4 -4
@@ -2459,10 +2459,10 @@ static inline int folio_expected_ref_count(const struct folio *folio)if (WARN_ON_ONCE (page_has_type (&folio->page) && !folio_test_hugetlb (folio)))return 0 ; -if (folio_test_anon (folio)) { -/* One reference per page from the swapcache. */ - ref_count +=folio_test_swapcache (folio) << order; - }else { +/* One reference per page from the swapcache. */ + ref_count +=folio_test_swapcache (folio) << order; + +if (!folio_test_anon (folio)) {/* One reference per page from the pagecache. */ ref_count += !!folio->mapping << order;/* One reference from PG_private. */
Signature · 1 line
2.43.0
Quoted 96 lines · Bijan Tabatabai
Currently,
folio_expected_ref_count()only adds references for the swap cache if the folio is anonymous. However, according to the comment above the definition ofPG_swapcachein enum pageflags, shmem folios can also havePG_swapcacheset. This patch makes sure references for the swap cache are added iffolio_test_swapcache(folio)is true.This issue was found when trying to hot-unplug memory in a QEMU/KVM virtual machine. When initiating hot-unplug when most of the guest memory is allocated, hot-unplug hangs partway through removal due to migration failures. The following message would be printed several times, and would be printed again about every five seconds:
[ 49.641309] migrating pfn b12f25 failed ret:7 [ 49.641310] page: refcount:2 mapcount:0 mapping:0000000033bd8fe2 index:0x7f404d925 pfn:0xb12f25 [ 49.641311] aops:swap_aops [ 49.641313] flags: 0x300000000030508(uptodate|active|owner_priv_1|reclaim|swapbacked|node=0|zone=3) [ 49.641314] raw: 0300000000030508 ffffed312c4bc908 ffffed312c4bc9c8 0000000000000000 [ 49.641315] raw: 00000007f404d925 00000000000c823b 00000002ffffffff 0000000000000000 [ 49.641315] page dumped because: migration failureWhen debugging this, I found that these migration failures were due to
__migrate_folio()returning-EAGAINfor a small set of folios because the expected reference count it calculates viafolio_expected_ref_count()is one less than the actual reference count of the folios. Furthermore, all of the affected folios were not anonymous, but had thePG_swapcacheflag set, inspiring this patch. After applying this patch, the memory hot-unplug behaves as expected.I tested this on a machine running Ubuntu 24.04 with kernel version 6.8.0-90-generic and 64GB of memory. The guest VM is managed by libvirt and runs Ubuntu 24.04 with kernel version 6.18 (though the head of the mm-unstable branch as a Dec 16, 2025 was also tested and behaves the same) and 48GB of memory. The libvirt XML definition for the VM can be found at [1].
CONFIG_MHP_DEFAULT_ONLINE_TYPE_ONLINE_MOVABLEis set in the guest kernel so the hot-pluggable memory is automatically onlined.Below are the steps to reproduce this behavior:
1) Define and start and virtual machine
host$ virsh -c qemu:///system define ./test_vm.xml # test_vm.xml from [1] host$ virsh -c qemu:///system starttest_vm2) Setup swap in the guest
guest$ sudo fallocate -l 32G /swapfile
guest$ sudo chmod 0600 /swapfile
guest$ sudo mkswap /swapfile
guest$ sudo swapon /swapfile3) Use
alloc_data[2] to allocate most of the remaining guest memory guest$ ./alloc_data 454) In a separate guest terminal, monitor the amount of used memory guest$ watch -n1 free -h
5) When
alloc_datahas finished allocating, initiate the memory hot-unplug using the provided xml file [3]
host$ virsh -c qemu:///system detach-devicetest_vm./remove.xml --liveAfter initiating the memory hot-unplug, you should see the amount of available memory in the guest decrease, and the amount of used swap data increase. If everything works as expected, when all of the memory is unplugged, there should be around 8.5-9GB of data in swap. If the unplugging is unsuccessful, the amount of used swap data will settle below that. If that happens, you should be able to see log messages in dmesg similar to the one posted above.
[1] https://github.com/BijanT/linux_patch_files/blob/main/test_vm.xml
[2] https://github.com/BijanT/linux_patch_files/blob/main/alloc_data.c
[3] https://github.com/BijanT/linux_patch_files/blob/main/remove.xmlFixes 86ebd50224c0 ("mm: add folio_expected_ref_count() for reference count calculation")Signed-off-by Bijan Tabatabai <[email protected]>---
I am not very familiar with the memory hot-(un)plug or swapping code, so I am not 100% certain if this patch actually solves the root of the problem. I believe the issue is from shmem folios, in which case I believe this patch is correct. However, I couldn't think of an easy way to confirm that the affected folios were from shmem. I guess it could be possible that the root cause could be from some bug where some anonymous pages do not return true to
folio_test_anon(). I don't think that's the case, but figured the MM maintainers would have a better idea of what's going on.--- include/linux/mm.h | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)include/linux/mm.h +0
-1@@ -2459,10 +2459,10 @@ static inline int folio_expected_ref_count(const struct folio *folio)if (WARN_ON_ONCE (page_has_type (&folio->page) && !folio_test_hugetlb (folio)))return 0 ; -if (folio_test_anon (folio)) {
include/linux/mm.h static inline int folio_expected_ref_count(const struct folio *folio) · line 2468
Quoted 10 lines
include/linux/mm.h +4
-3-/* One reference per page from the swapcache. */ - ref_count +=folio_test_swapcache (folio) << order; - }else { +/* One reference per page from the swapcache. */ + ref_count +=folio_test_swapcache (folio) << order; + +if (!folio_test_anon (folio)) {/* One reference per page from the pagecache. */ ref_count += !!folio->mapping << order;/* One reference from PG_private. */
We discussed that recently [1] and I think Zi wanted to send a patch. We were a bit confused about the semantics of folio_test_swapcache(), but concluded that it should be fine when called against pagecache folios.
So far I thought 86ebd50224c0 did not result in the issue because it replaced
-static int folio_expected_refs(struct address_space *mapping,
- struct folio *folio)
-{
- int refs = 1;
- if (!mapping)
- return refs;
-
- refs += folio_nr_pages(folio);
- if (folio_test_private(folio))
- refs++;
-
- return refs;
-}
in migration code where !mapping would have only have returned 1 (reference held by the caller) that folio_expected_ref_count() now expects to be added in the caller.
But looking again, in the caller, we obtain
mapping = folio_mapping(src)
Which returns the swap_address_space() for folios in the swapcache.
So it indeed looks like 86ebd50224c0 introduced the issue.
Thanks!
We should cc: stable
[1] https://lore.kernel.org/all/[email protected]/
[2] https://lore.kernel.org/all/[email protected]/
Signature · 3 lines
Cheers David
Quoted 75 lines · David Hildenbrand (Red Hat)
Quoted 74 lines · Bijan Tabatabai
Currently,
folio_expected_ref_count()only adds references for the swap cache if the folio is anonymous. However, according to the comment above the definition ofPG_swapcachein enum pageflags, shmem folios can also havePG_swapcacheset. This patch makes sure references for the swap cache are added iffolio_test_swapcache(folio)is true.This issue was found when trying to hot-unplug memory in a QEMU/KVM virtual machine. When initiating hot-unplug when most of the guest memory is allocated, hot-unplug hangs partway through removal due to migration failures. The following message would be printed several times, and would be printed again about every five seconds:
[ 49.641309] migrating pfn b12f25 failed ret:7 [ 49.641310] page: refcount:2 mapcount:0 mapping:0000000033bd8fe2 index:0x7f404d925 pfn:0xb12f25 [ 49.641311] aops:swap_aops [ 49.641313] flags: 0x300000000030508(uptodate|active|owner_priv_1|reclaim|swapbacked|node=0|zone=3) [ 49.641314] raw: 0300000000030508 ffffed312c4bc908 ffffed312c4bc9c8 0000000000000000 [ 49.641315] raw: 00000007f404d925 00000000000c823b 00000002ffffffff 0000000000000000 [ 49.641315] page dumped because: migration failureWhen debugging this, I found that these migration failures were due to
__migrate_folio()returning-EAGAINfor a small set of folios because the expected reference count it calculates viafolio_expected_ref_count()is one less than the actual reference count of the folios. Furthermore, all of the affected folios were not anonymous, but had thePG_swapcacheflag set, inspiring this patch. After applying this patch, the memory hot-unplug behaves as expected.I tested this on a machine running Ubuntu 24.04 with kernel version 6.8.0-90-generic and 64GB of memory. The guest VM is managed by libvirt and runs Ubuntu 24.04 with kernel version 6.18 (though the head of the mm-unstable branch as a Dec 16, 2025 was also tested and behaves the same) and 48GB of memory. The libvirt XML definition for the VM can be found at [1].
CONFIG_MHP_DEFAULT_ONLINE_TYPE_ONLINE_MOVABLEis set in the guest kernel so the hot-pluggable memory is automatically onlined.Below are the steps to reproduce this behavior:
1) Define and start and virtual machine
host$ virsh -c qemu:///system define ./test_vm.xml # test_vm.xml from [1] host$ virsh -c qemu:///system starttest_vm2) Setup swap in the guest
guest$ sudo fallocate -l 32G /swapfile
guest$ sudo chmod 0600 /swapfile
guest$ sudo mkswap /swapfile
guest$ sudo swapon /swapfile3) Use
alloc_data[2] to allocate most of the remaining guest memory guest$ ./alloc_data 454) In a separate guest terminal, monitor the amount of used memory guest$ watch -n1 free -h
5) When
alloc_datahas finished allocating, initiate the memory hot-unplug using the provided xml file [3]
host$ virsh -c qemu:///system detach-devicetest_vm./remove.xml --liveAfter initiating the memory hot-unplug, you should see the amount of available memory in the guest decrease, and the amount of used swap data increase. If everything works as expected, when all of the memory is unplugged, there should be around 8.5-9GB of data in swap. If the unplugging is unsuccessful, the amount of used swap data will settle below that. If that happens, you should be able to see log messages in dmesg similar to the one posted above.
[1] https://github.com/BijanT/linux_patch_files/blob/main/test_vm.xml
[2] https://github.com/BijanT/linux_patch_files/blob/main/alloc_data.c
[3] https://github.com/BijanT/linux_patch_files/blob/main/remove.xmlFixes 86ebd50224c0 ("mm: add folio_expected_ref_count() for reference count calculation")Signed-off-by Bijan Tabatabai <[email protected]>---
Quoted 8 lines
Quoted 8 lines
I am not very familiar with the memory hot-(un)plug or swapping code, so I am not 100% certain if this patch actually solves the root of the problem. I believe the issue is from shmem folios, in which case I believe this patch is correct. However, I couldn't think of an easy way to confirm that the affected folios were from shmem. I guess it could be possible that the root cause could be from some bug where some anonymous pages do not return true to
folio_test_anon(). I don't think that's the case, but figured the MM maintainers would have a better idea of what's going on.
I am not sure about if shmem in swapcache causes the issue, since the above setup does not involve shmem. +Baolin and Hugh for some insight.
But David also mentioned that in __read_swap_cache_async() there is a chance that anon folio in swapcache can have anon flag not set yet. +Chris and Kairui for more analysis.
Quoted 15 lines
Quoted 15 lines
--- include/linux/mm.h | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)include/linux/mm.h +0
-2@@ -2459,10 +2459,10 @@ static inline int folio_expected_ref_count(const struct folio *folio)if (WARN_ON_ONCE (page_has_type (&folio->page) && !folio_test_hugetlb (folio)))return 0 ; - if (folio_test_anon(folio)) { -/* One reference per page from the swapcache. */ - ref_count +=folio_test_swapcache (folio) << order;
Quoted 8 lines
include/linux/mm.h static inline int folio_expected_ref_count(const struct folio *folio) · line 2468
Quoted 8 lines
include/linux/mm.h +4
-1- }else { +/* One reference per page from the swapcache. */ + ref_count +=folio_test_swapcache (folio) << order; + +if (!folio_test_anon (folio)) {/* One reference per page from the pagecache. */ ref_count += !!folio->mapping << order;/* One reference from PG_private. */
This change is almost the same as what I proposed in [1] during my discussion with David.
Quoted 33 lines
We discussed that recently [1] and I think Zi wanted to send a patch. We were a bit confused about the semantics of
folio_test_swapcache(), but concluded that it should be fine when called against pagecache folios.So far I thought
86ebd50224c0did not result in the issue because it replaced(quoted diff) +0
-13-static int folio_expected_refs(struct address_space *mapping, - struct folio *folio) -{ - int refs = 1; - if (!mapping) - return refs; - - refs += folio_nr_pages(folio); - if (folio_test_private(folio)) - refs++; - - return refs; -}in migration code where !mapping would have only have returned 1 (reference held by the caller) that
folio_expected_ref_count()now expects to be added in the caller.But looking again, in the caller, we obtain
mapping =
folio_mapping(src)Which returns the
swap_address_space()for folios in the swapcache.So it indeed looks like
86ebd50224c0introduced the issue.Thanks!
Quoted 9 lines
We should cc: stable
[1] https://lore.kernel.org/all/[email protected]/
[2] https://lore.kernel.org/all/[email protected]/
I agree with David. Acked-by: Zi Yan <[email protected]>
Best Regards,
Yan, Zi
David Hildenbrand (Red Hat)
maintainerQuoted 10 lines
Quoted 9 lines
Quoted 9 lines
I am not very familiar with the memory hot-(un)plug or swapping code, so I am not 100% certain if this patch actually solves the root of the problem. I believe the issue is from shmem folios, in which case I believe this patch is correct. However, I couldn't think of an easy way to confirm that the affected folios were from shmem. I guess it could be possible that the root cause could be from some bug where some anonymous pages do not return true to
folio_test_anon(). I don't think that's the case, but figured the MM maintainers would have a better idea of what's going on.
Quoted 2 lines
I am not sure about if shmem in swapcache causes the issue, since the above setup does not involve shmem. +Baolin and Hugh for some insight.
We might just push out another unrelated shmem page to swap as we create memory pressure in the system I think.
Quoted 4 lines
But David also mentioned that in
__read_swap_cache_async()there is a chance that anon folio in swapcache can have anon flag not set yet. +Chris and Kairui for more analysis.
Right, when we swapin an anon folio and did not map it into the page table yet. Likely we can trigger something similar when we proactively read a shmem page from swap into the swapcache.
So it's unclear "where" a swapcache page belongs to until we move it to its owner (anon / shmem), which is also why I cannot judge easily from
[ 49.641309] migrating pfn b12f25 failed ret:7 [ 49.641310] page: refcount:2 mapcount:0 mapping:0000000033bd8fe2 index:0x7f404d925 pfn:0xb12f25 [ 49.641311] aops:swap_aops [ 49.641313] flags: 0x300000000030508(uptodate|active|owner_priv_1|reclaim|swapbacked|node=0|zone=3) [ 49.641314] raw: 0300000000030508 ffffed312c4bc908 ffffed312c4bc9c8 0000000000000000 [ 49.641315] raw: 00000007f404d925 00000000000c823b 00000002ffffffff 0000000000000000 [ 49.641315] page dumped because: migration failure
What exactly that was.
It was certainly an order-0 folio.
[...]
Quoted 2 lines
I agree with David. Acked-by: Zi Yan <[email protected]>
Thanks for the fast review :)
Signature · 3 lines
Cheers David
Baolin Wang
Reviewed-byQuoted 32 lines · David Hildenbrand (Red Hat)
Quoted 10 lines
Quoted 9 lines
Quoted 9 lines
I am not very familiar with the memory hot-(un)plug or swapping code, so I am not 100% certain if this patch actually solves the root of the problem. I believe the issue is from shmem folios, in which case I believe this patch is correct. However, I couldn't think of an easy way to confirm that the affected folios were from shmem. I guess it could be possible that the root cause could be from some bug where some anonymous pages do not return true to
folio_test_anon(). I don't think that's the case, but figured the MM maintainers would have a better idea of what's going on.Quoted 2 lines
I am not sure about if shmem in swapcache causes the issue, since the above setup does not involve shmem. +Baolin and Hugh for some insight.
We might just push out another unrelated shmem page to swap as we create memory pressure in the system I think.
Quoted 4 lines
But David also mentioned that in
__read_swap_cache_async()there is a chance that anon folio in swapcache can have anon flag not set yet. +Chris and Kairui for more analysis.Right, when we swapin an anon folio and did not map it into the page table yet. Likely we can trigger something similar when we proactively read a shmem page from swap into the swapcache.
So it's unclear "where" a swapcache page belongs to until we move it to its owner (anon / shmem), which is also why I cannot judge easily from
[ 49.641309] migrating pfn b12f25 failed ret:7
[ 49.641310] page: refcount:2 mapcount:0 mapping:0000000033bd8fe2 index:0x7f404d925 pfn:0xb12f25 [ 49.641311] aops:swap_aops
[ 49.641313] flags: 0x300000000030508(uptodate|active|owner_priv_1| reclaim|swapbacked|node=0|zone=3) [ 49.641314] raw: 0300000000030508ffffed312c4bc908ffffed312c4bc9c80000000000000000 [ 49.641315] raw:00000007f404d92500000000000c823b00000002ffffffff0000000000000000 [ 49.641315] page dumped because: migration failure
Quoted 3 lines
What exactly that was.
It was certainly an order-0 folio.
Thanks David for the explanation. It completely makes sense to me. So feel free to add:
Quoted 10 lines · David Hildenbrand (Red Hat)
Quoted 10 lines
Quoted 9 lines
Quoted 9 lines
I am not very familiar with the memory hot-(un)plug or swapping code, so I am not 100% certain if this patch actually solves the root of the problem. I believe the issue is from shmem folios, in which case I believe this patch is correct. However, I couldn't think of an easy way to confirm that the affected folios were from shmem. I guess it could be possible that the root cause could be from some bug where some anonymous pages do not return true to
folio_test_anon(). I don't think that's the case, but figured the MM maintainers would have a better idea of what's going on.
Quoted 6 lines
Quoted 2 lines
I am not sure about if shmem in swapcache causes the issue, since the above setup does not involve shmem. +Baolin and Hugh for some insight.
We might just push out another unrelated shmem page to swap as we create memory pressure in the system I think.
One trivial question: currently we only put anon/shmem folio in swapcache, right?
Quoted 38 lines
Quoted 4 lines
But David also mentioned that in
__read_swap_cache_async()there is a chance that anon folio in swapcache can have anon flag not set yet. +Chris and Kairui for more analysis.Right, when we swapin an anon folio and did not map it into the page table yet. Likely we can trigger something similar when we proactively read a shmem page from swap into the swapcache.
So it's unclear "where" a swapcache page belongs to until we move it to its owner (anon / shmem), which is also why I cannot judge easily from
[ 49.641309] migrating pfn b12f25 failed ret:7 [ 49.641310] page: refcount:2 mapcount:0 mapping:0000000033bd8fe2 index:0x7f404d925 pfn:0xb12f25 [ 49.641311] aops:swap_aops [ 49.641313] flags: 0x300000000030508(uptodate|active|owner_priv_1|reclaim|swapbacked|node=0|zone=3) [ 49.641314] raw: 0300000000030508 ffffed312c4bc908 ffffed312c4bc9c8 0000000000000000 [ 49.641315] raw: 00000007f404d925 00000000000c823b 00000002ffffffff 0000000000000000 [ 49.641315] page dumped because: migration failureWhat exactly that was.
It was certainly an order-0 folio.
[...]
Quoted 2 lines
I agree with David. Acked-by: Zi Yan <[email protected]>
Thanks for the fast review :)
Signature · 3 lines
Cheers David
Signature · 2 lines
Wei Yang Help you, Help me
Quoted 14 lines · Wei Yang
Quoted 13 lines · David Hildenbrand (Red Hat)
Quoted 12 lines
Quoted 9 lines
Quoted 9 lines
I am not very familiar with the memory hot-(un)plug or swapping code, so I am not 100% certain if this patch actually solves the root of the problem. I believe the issue is from shmem folios, in which case I believe this patch is correct. However, I couldn't think of an easy way to confirm that the affected folios were from shmem. I guess it could be possible that the root cause could be from some bug where some anonymous pages do not return true to
folio_test_anon(). I don't think that's the case, but figured the MM maintainers would have a better idea of what's going on.I am not sure about if shmem in swapcache causes the issue, since the above setup does not involve shmem. +Baolin and Hugh for some insight.
Quoted 6 lines
Quoted 3 lines
We might just push out another unrelated shmem page to swap as we create memory pressure in the system I think.
One trivial question: currently we only put anon/shmem folio in swapcache, right?
AFAICT, Yes (note a special case for anonymous folios: lazyfree anonymous folios will be directly freed instead of being swapped out).
Quoted 15 lines · Wei Yang
Quoted 13 lines · David Hildenbrand (Red Hat)
Quoted 12 lines
Quoted 9 lines
Quoted 9 lines
I am not very familiar with the memory hot-(un)plug or swapping code, so I am not 100% certain if this patch actually solves the root of the problem. I believe the issue is from shmem folios, in which case I believe this patch is correct. However, I couldn't think of an easy way to confirm that the affected folios were from shmem. I guess it could be possible that the root cause could be from some bug where some anonymous pages do not return true to
folio_test_anon(). I don't think that's the case, but figured the MM maintainers would have a better idea of what's going on.I am not sure about if shmem in swapcache causes the issue, since the above setup does not involve shmem. +Baolin and Hugh for some insight.
Quoted 6 lines
Quoted 3 lines
We might just push out another unrelated shmem page to swap as we create memory pressure in the system I think.
One trivial question: currently we only put anon/shmem folio in swapcache, right?
For swapout, yes, the entry point to move a folio to swap space is folio_alloc_swap, only anon and shmem can do that (vmscan.c -> folio_test_anon && folio_test_swapbacked, and shmem.c).
Swapin is a bit different because of readahead, readahead folios are not marked as anon / shmem (folio->mapping) until used, they do belong to anon / shmem though, but we don't add them to the mapping until that mapping does a swap cache lookup and use the cached folio.
Also maybe worth mentioning, swap cache lookup convention requires the caller to lock the folio and double check folio still matches the swap entry before use (folio_matches_swap_entry), folios there are unstable and could no longer be a valid swap cache folio unless locked.
Quoted 20 lines · Kairui Song
Quoted 19 lines · Wei Yang
Quoted 16 lines · David Hildenbrand (Red Hat)
Quoted 10 lines
Quoted 9 lines
Quoted 9 lines
I am not very familiar with the memory hot-(un)plug or swapping code, so I am not 100% certain if this patch actually solves the root of the problem. I believe the issue is from shmem folios, in which case I believe this patch is correct. However, I couldn't think of an easy way to confirm that the affected folios were from shmem. I guess it could be possible that the root cause could be from some bug where some anonymous pages do not return true to
folio_test_anon(). I don't think that's the case, but figured the MM maintainers would have a better idea of what's going on.Quoted 2 lines
I am not sure about if shmem in swapcache causes the issue, since the above setup does not involve shmem. +Baolin and Hugh for some insight.
We might just push out another unrelated shmem page to swap as we create memory pressure in the system I think.
Quoted 7 lines
Quoted 2 lines
One trivial question: currently we only put anon/shmem folio in swapcache, right?
For swapout, yes, the entry point to move a folio to swap space is
folio_alloc_swap, only anon and shmem can do that (vmscan.c->folio_test_anon&&folio_test_swapbacked, andshmem.c).
Thanks for this information.
Quoted 5 lines
Swapin is a bit different because of readahead, readahead folios are not marked as anon / shmem (
folio->mapping) until used, they do belong to anon / shmem though, but we don't add them to the mapping until that mapping does a swap cache lookup and use the cached folio.
I saw this. So there is some folio which is in swapcache but no sure is anon/shmem yet.
Quoted 4 lines
Also maybe worth mentioning, swap cache lookup convention requires the caller to lock the folio and double check folio still matches the swap entry before use (
folio_matches_swap_entry), folios there are unstable and could no longer be a valid swap cache folio unless locked.
Thanks for this notice, will pay attention to this.
Signature · 2 lines
Wei Yang Help you, Help me
Quoted 87 lines · Zi Yan
Quoted 75 lines · David Hildenbrand (Red Hat)
Quoted 74 lines · Bijan Tabatabai
Currently,
folio_expected_ref_count()only adds references for the swap cache if the folio is anonymous. However, according to the comment above the definition ofPG_swapcachein enum pageflags, shmem folios can also havePG_swapcacheset. This patch makes sure references for the swap cache are added iffolio_test_swapcache(folio)is true.This issue was found when trying to hot-unplug memory in a QEMU/KVM virtual machine. When initiating hot-unplug when most of the guest memory is allocated, hot-unplug hangs partway through removal due to migration failures. The following message would be printed several times, and would be printed again about every five seconds:
[ 49.641309] migrating pfn b12f25 failed ret:7 [ 49.641310] page: refcount:2 mapcount:0 mapping:0000000033bd8fe2 index:0x7f404d925 pfn:0xb12f25 [ 49.641311] aops:swap_aops [ 49.641313] flags: 0x300000000030508(uptodate|active|owner_priv_1|reclaim|swapbacked|node=0|zone=3) [ 49.641314] raw: 0300000000030508 ffffed312c4bc908 ffffed312c4bc9c8 0000000000000000 [ 49.641315] raw: 00000007f404d925 00000000000c823b 00000002ffffffff 0000000000000000 [ 49.641315] page dumped because: migration failureWhen debugging this, I found that these migration failures were due to
__migrate_folio()returning-EAGAINfor a small set of folios because the expected reference count it calculates viafolio_expected_ref_count()is one less than the actual reference count of the folios. Furthermore, all of the affected folios were not anonymous, but had thePG_swapcacheflag set, inspiring this patch. After applying this patch, the memory hot-unplug behaves as expected.I tested this on a machine running Ubuntu 24.04 with kernel version 6.8.0-90-generic and 64GB of memory. The guest VM is managed by libvirt and runs Ubuntu 24.04 with kernel version 6.18 (though the head of the mm-unstable branch as a Dec 16, 2025 was also tested and behaves the same) and 48GB of memory. The libvirt XML definition for the VM can be found at [1].
CONFIG_MHP_DEFAULT_ONLINE_TYPE_ONLINE_MOVABLEis set in the guest kernel so the hot-pluggable memory is automatically onlined.Below are the steps to reproduce this behavior:
1) Define and start and virtual machine
host$ virsh -c qemu:///system define ./test_vm.xml # test_vm.xml from [1] host$ virsh -c qemu:///system starttest_vm2) Setup swap in the guest
guest$ sudo fallocate -l 32G /swapfile
guest$ sudo chmod 0600 /swapfile
guest$ sudo mkswap /swapfile
guest$ sudo swapon /swapfile3) Use
alloc_data[2] to allocate most of the remaining guest memory guest$ ./alloc_data 454) In a separate guest terminal, monitor the amount of used memory guest$ watch -n1 free -h
5) When
alloc_datahas finished allocating, initiate the memory hot-unplug using the provided xml file [3]
host$ virsh -c qemu:///system detach-devicetest_vm./remove.xml --liveAfter initiating the memory hot-unplug, you should see the amount of available memory in the guest decrease, and the amount of used swap data increase. If everything works as expected, when all of the memory is unplugged, there should be around 8.5-9GB of data in swap. If the unplugging is unsuccessful, the amount of used swap data will settle below that. If that happens, you should be able to see log messages in dmesg similar to the one posted above.
[1] https://github.com/BijanT/linux_patch_files/blob/main/test_vm.xml
[2] https://github.com/BijanT/linux_patch_files/blob/main/alloc_data.c
[3] https://github.com/BijanT/linux_patch_files/blob/main/remove.xmlFixes 86ebd50224c0 ("mm: add folio_expected_ref_count() for reference count calculation")Signed-off-by Bijan Tabatabai <[email protected]>---
Quoted 8 lines
Quoted 8 lines
I am not very familiar with the memory hot-(un)plug or swapping code, so I am not 100% certain if this patch actually solves the root of the problem. I believe the issue is from shmem folios, in which case I believe this patch is correct. However, I couldn't think of an easy way to confirm that the affected folios were from shmem. I guess it could be possible that the root cause could be from some bug where some anonymous pages do not return true to
folio_test_anon(). I don't think that's the case, but figured the MM maintainers would have a better idea of what's going on.
Quoted 6 lines
I am not sure about if shmem in swapcache causes the issue, since the above setup does not involve shmem. +Baolin and Hugh for some insight.
But David also mentioned that in
__read_swap_cache_async()there is a chance that anon folio in swapcache can have anon flag not set yet. +Chris and Kairui for more analysis.
Yeah, that's possible, a typical case is swap readahead will alloc and add folios into swap cache, but won't add it to anon/shmem mapping. Anon/shmem will use the folio in swapcache upon page fault, and make it anon/shmem folio by then.
This change looks good to me too, thanks for Ccing me.