Several HMM tests hardcode TWOMEG as the THP size. This is wrong on
architectures where the PMD size is not 2MB such as arm64 with 64K base
pages where THP is 512MB. Fix this by using read_pmd_pagesize() from
vm_util instead.
While here also replace the custom file_read_ulong() helper used to
parse the default hugetlbfs page size from /proc/meminfo with the
existing default_huge_page_size() from vm_util.
Link: https://lore.kernel.org/20260331063445.3551404-3-apopple@nvidia.com
Link: https://lore.kernel.org/linux-mm/8bd0396a-8997-4d2e-a13f-5aac033083d7@linux.dev/
Fixes: fee9f6d1b8 ("mm/hmm/test: add selftests for HMM")
Fixes: 519071529d ("selftests/mm/hmm-tests: new tests for zone device THP migration")
Signed-off-by: Alistair Popple <apopple@nvidia.com>
Reported-by: Zenghui Yu <zenghui.yu@linux.dev>
Closes: https://lore.kernel.org/linux-mm/8bd0396a-8997-4d2e-a13f-5aac033083d7@linux.dev/
Reviewed-by: Balbir Singh <balbirs@nvidia.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: Leon Romanovsky <leon@kernel.org>
Cc: Liam Howlett <liam.howlett@oracle.com>
Cc: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: <stable@vger,kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "Minor hmm_test fixes and cleanups".
Two bugfixes a cleanup for the HMM kernel selftests. These were mostly
reported by Zenghui Yu with special thanks to Lorenzo for analysing and
pointing out the problems.
This patch (of 3):
When dmirror_fops_release() is called it frees the dmirror struct but
doesn't migrate device private pages back to system memory first. This
leaves those pages with a dangling zone_device_data pointer to the freed
dmirror.
If a subsequent fault occurs on those pages (eg. during coredump) the
dmirror_devmem_fault() callback dereferences the stale pointer causing a
kernel panic. This was reported [1] when running mm/ksft_hmm.sh on arm64,
where a test failure triggered SIGABRT and the resulting coredump walked
the VMAs faulting in the stale device private pages.
Fix this by calling dmirror_device_evict_chunk() for each devmem chunk in
dmirror_fops_release() to migrate all device private pages back to system
memory before freeing the dmirror struct. The function is moved earlier
in the file to avoid a forward declaration.
Link: https://lore.kernel.org/20260331063445.3551404-1-apopple@nvidia.com
Link: https://lore.kernel.org/20260331063445.3551404-2-apopple@nvidia.com
Fixes: b2ef9f5a5c ("mm/hmm/test: add selftest driver for HMM")
Signed-off-by: Alistair Popple <apopple@nvidia.com>
Reported-by: Zenghui Yu <zenghui.yu@linux.dev>
Closes: https://lore.kernel.org/linux-mm/8bd0396a-8997-4d2e-a13f-5aac033083d7@linux.dev/
Reviewed-by: Balbir Singh <balbirs@nvidia.com>
Tested-by: Zenghui Yu <zenghui.yu@linux.dev>
Cc: David Hildenbrand <david@kernel.org>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: Leon Romanovsky <leon@kernel.org>
Cc: Liam Howlett <liam.howlett@oracle.com>
Cc: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Zenghui Yu <zenghui.yu@linux.dev>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
hugetlb_dio test uses sub-page offsets (pagesize / 2) to verify that
hugepages used as DIO user buffers are correctly unpinned at completion.
However, on filesystems with a logical block size larger than half the
page size (e.g., 4K-sector block devices), these unaligned DIO writes are
rejected with -EINVAL, causing the test to fail unexpectedly.
Add get_dio_alignment() to query the filesystem's required DIO alignment
via statx(STATX_DIOALIGN) and skip individual test cases whose file offset
or write size is not a multiple of that alignment. Aligned cases continue
to run so the core coverage is preserved.
While here, open the temporary file once in main() and share the fd across
all test cases instead of reopening it in each invocation.
=== Reproduce Steps ===
# dd if=/dev/zero of=/tmp/test.img bs=1M count=512
# losetup --sector-size 4096 /dev/loop0 /tmp/test.img
# mkfs.xfs /dev/loop0
# mkdir -p /mnt/dio_test
# mount /dev/loop0 /mnt/dio_test
// Modify test to open /mnt/dio_test and rebuild it:
- fd = open("/tmp", O_TMPFILE | O_RDWR | O_DIRECT, 0664);
+ fd = open("/mnt/dio_test", O_TMPFILE | O_RDWR | O_DIRECT, 0664);
# getconf PAGESIZE
4096
# echo 100 >/proc/sys/vm/nr_hugepages
# ./hugetlb_dio
TAP version 13
1..4
# No. Free pages before allocation : 100
# No. Free pages after munmap : 100
ok 1 free huge pages from 0-12288
Bail out! Error writing to file
: Invalid argument (22)
# Planned tests != run tests (4 != 1)
# Totals: pass:1 fail:0 xfail:0 xpass:0 skip:0 error:0
Link: https://lore.kernel.org/20260401090520.24018-1-liwang@redhat.com
Signed-off-by: Li Wang <liwang@redhat.com>
Suggested-by: Mike Rapoport <rppt@kernel.org>
Suggested-by: David Hildenbrand <david@kernel.org>
Acked-by: David Hildenbrand (Arm) <david@kernel.org>
Cc: Liam Howlett <liam.howlett@oracle.com>
Cc: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Commit 2697dd8ae7 ("mm/mseal: update VMA end correctly on merge") fixed
an issue in the loop which iterates through VMAs applying mseal, which was
triggered by mseal()'ing a range of VMAs where the second was mseal()'d
and the first mergeable with it, once mseal()'d.
Add a regression test to assert that this behaviour is correct. We place
it in the merge selftests as this is strictly an issue with merging (via a
vma_modify() invocation).
It also asserts that mseal()'d ranges are correctly merged as you'd
expect.
The test is implemented such that it is skipped if mseal() is not
available on the system.
[rppt@kernel.org: fix inclusions, to fix handle_uprobe_upon_merged_vma()]
Link: https://lore.kernel.org/ac_mCIUQWRAbuH8F@kernel.org
[ljs@kernel.org: simplifications per Pedro]
Link: https://lore.kernel.org/1c9c922d-5cb5-4cff-9273-b737cdb57ca1@lucifer.local
Link: https://lore.kernel.org/20260331073627.50010-1-ljs@kernel.org
Signed-off-by: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
Signed-off-by: Mike Rapoport <rppt@kernel.org>
Cc: David Hildenbrand <david@kernel.org>
Cc: Jann Horn <jannh@google.com>
Cc: Liam Howlett <liam.howlett@oracle.com>
Cc: Lorenzo Stoakes <ljs@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Pedro Falcato <pfalcato@suse.de>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
weighted_interleave_auto_store() fetches old_wi_state inside the if
(!input) block only. This causes two memory leaks:
1. When a user writes "false" and the current mode is already manual,
the function returns early without freeing the freshly allocated
new_wi_state.
2. When a user writes "true", old_wi_state stays NULL because the
fetch is skipped entirely. The old state is then overwritten by
rcu_assign_pointer() but never freed, since the cleanup path is
gated on old_wi_state being non-NULL. A user can trigger this
repeatedly by writing "1" in a loop.
Fix both leaks by moving the old_wi_state fetch before the input check,
making it unconditional. This also allows a unified early return for both
"true" and "false" when the requested mode matches the current mode.
Link: https://lore.kernel.org/20260401005702.7096-1-liu.yun@linux.dev
Link: https://sashiko.dev/#/patchset/20260331100740.84906-1-liu.yun@linux.dev
Fixes: e341f9c3c8 ("mm/mempolicy: Weighted Interleave Auto-tuning")
Signed-off-by: Jackie Liu <liuyun01@kylinos.cn>
Reviewed-by: Joshua Hahn <joshua.hahnjy@gmail.com>
Reviewed by: Donet Tom <donettom@linux.ibm.com>
Cc: Gregory Price <gourry@gourry.net>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Byungchul Park <byungchul@sk.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: <stable@vger.kernel.org> # v6.16+
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
DAMON_LRU_SORT handles commit_inputs request inside kdamond thread,
reading the module parameters. If the user updates the module
parameters while the kdamond thread is reading those, races can happen.
To avoid this, the commit_inputs parameter shows whether it is still in
the progress, assuming users wouldn't update parameters in the middle of
the work. Some users might ignore that. Add a warning about the
behavior.
The issue was discovered in [1] by sashiko.
Link: https://lore.kernel.org/20260329153052.46657-3-sj@kernel.org
Link: https://lore.kernel.org/20260319161620.189392-2-objecting@objecting.org [1]
Fixes: 6acfcd0d75 ("Docs/admin-guide/damon: add a document for DAMON_LRU_SORT")
Signed-off-by: SeongJae Park <sj@kernel.org>
Cc: <stable@vger.kernel.org> # 6.0.x
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "Docs/admin-guide/mm/damon: warn commit_inputs vs other
params race".
Writing 'Y' to the commit_inputs parameter of DAMON_RECLAIM and
DAMON_LRU_SORT, and writing other parameters before the commit_inputs
request is completely processed can cause race conditions. While the
consequence can be bad, the documentation is not clearly describing that.
Add clear warnings.
The issue was discovered [1,2] by sashiko.
This patch (of 2):
DAMON_RECLAIM handles commit_inputs request inside kdamond thread,
reading the module parameters. If the user updates the module
parameters while the kdamond thread is reading those, races can happen.
To avoid this, the commit_inputs parameter shows whether it is still in
the progress, assuming users wouldn't update parameters in the middle of
the work. Some users might ignore that. Add a warning about the
behavior.
The issue was discovered in [1] by sashiko.
Link: https://lore.kernel.org/20260329153052.46657-2-sj@kernel.org
Link: https://lore.kernel.org/20260319161620.189392-3-objecting@objecting.org [1]
Link: https://lore.kernel.org/20260319161620.189392-2-objecting@objecting.org [3]
Fixes: 81a84182c3 ("Docs/admin-guide/mm/damon/reclaim: document 'commit_inputs' parameter")
Signed-off-by: SeongJae Park <sj@kernel.org>
Cc: <stable@vger.kernel.org> # 5.19.x
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
damos_adjust_quota() uses time_after_eq() to show if it is time to start a
new quota charge window, comparing the current jiffies and the scheduled
next charge window start time. If it is, the next charge window start
time is updated and the new charge window starts.
The time check and next window start time update is skipped while the
scheme is deactivated by the watermarks. Let's suppose the deactivation
is kept more than LONG_MAX jiffies (assuming CONFIG_HZ of 250, more than
99 days in 32 bit systems and more than one billion years in 64 bit
systems), resulting in having the jiffies larger than the next charge
window start time + LONG_MAX. Then, the time_after_eq() call can return
false until another LONG_MAX jiffies are passed.
This means the scheme can continue working after being reactivated by the
watermarks. But, soon, the quota will be exceeded and the scheme will
again effectively stop working until the next charge window starts.
Because the current charge window is extended to up to LONG_MAX jiffies,
however, it will look like it stopped unexpectedly and indefinitely, from
the user's perspective.
Fix this by using !time_in_range_open() instead.
The issue was discovered [1] by sashiko.
Link: https://lore.kernel.org/20260329152306.45796-1-sj@kernel.org
Link: https://lore.kernel.org/20260324040722.57944-1-sj@kernel.org [1]
Fixes: ee801b7dd7 ("mm/damon/schemes: activate schemes based on a watermarks mechanism")
Signed-off-by: SeongJae Park <sj@kernel.org>
Cc: <stable@vger.kernel.org> # 5.16.x
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Users can set damos_quota_goal->nid with arbitrary value for
node_memcg_{used,free}_bp. But DAMON core is using those for NODE-DATA()
without a validation of the value. This can result in out of bounds
memory access. The issue can actually triggered using DAMON user-space
tool (damo), like below.
$ sudo mkdir /sys/fs/cgroup/foo
$ sudo ./damo start --damos_action stat --damos_quota_interval 1s \
--damos_quota_goal node_memcg_used_bp 50% -1 /foo
$ sudo dmseg
[...]
[ 524.181426] Unable to handle kernel paging request at virtual address 0000000000002c00
Fix this issue by adding the validation of the given node id. If an
invalid node id is given, it returns 0% for used memory ratio, and 100%
for free memory ratio.
Link: https://lore.kernel.org/20260329043902.46163-3-sj@kernel.org
Fixes: b74a120bcf ("mm/damon/core: implement DAMOS_QUOTA_NODE_MEMCG_USED_BP")
Signed-off-by: SeongJae Park <sj@kernel.org>
Cc: <stable@vger.kernel.org> # 6.19.x
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "mm/damon/core: validate damos_quota_goal->nid".
node_mem[cg]_{used,free}_bp DAMOS quota goals receive the node id. The
node id is used for si_meminfo_node() and NODE_DATA() without proper
validation. As a result, privileged users can trigger an out of bounds
memory access using DAMON_SYSFS. Fix the issues.
The issue was originally reported [1] with a fix by another author. The
original author announced [2] that they will stop working including the
fix that was still in the review stage. Hence I'm restarting this.
This patch (of 2):
Users can set damos_quota_goal->nid with arbitrary value for
node_mem_{used,free}_bp. But DAMON core is using those for
si_meminfo_node() without the validation of the value. This can result in
out of bounds memory access. The issue can actually triggered using DAMON
user-space tool (damo), like below.
$ sudo ./damo start --damos_action stat \
--damos_quota_goal node_mem_used_bp 50% -1 \
--damos_quota_interval 1s
$ sudo dmesg
[...]
[ 65.565986] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000098
Fix this issue by adding the validation of the given node. If an invalid
node id is given, it returns 0% for used memory ratio, and 100% for free
memory ratio.
Link: https://lore.kernel.org/20260329043902.46163-2-sj@kernel.org
Link: https://lore.kernel.org/20260325073034.140353-1-objecting@objecting.org [1]
Link: https://lore.kernel.org/20260327040924.68553-1-sj@kernel.org [2]
Fixes: 0e1c773b50 ("mm/damon/core: introduce damos quota goal metrics for memory node utilization")
Signed-off-by: SeongJae Park <sj@kernel.org>
Cc: <stable@vger.kernel.org> # 6.16.x
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Destroy the DAMON context and reset the global pointer when damon_start()
fails. Otherwise, the context allocated by damon_stat_build_ctx() is
leaked, and the stale damon_stat_context pointer will be overwritten on
the next enable attempt, making the old allocation permanently
unreachable.
Link: https://lore.kernel.org/20260331101553.88422-1-liu.yun@linux.dev
Fixes: 369c415e60 ("mm/damon: introduce DAMON_STAT module")
Signed-off-by: Jackie Liu <liuyun01@kylinos.cn>
Reviewed-by: SeongJae Park <sj@kernel.org>
Cc: <stable@vger.kernel.org> # 6.17.x
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
When kdamond_fn() main loop is finished, the function cancels remaining
damos_walk() request and unset the damon_ctx->kdamond so that API callers
and API functions themselves can show the context is terminated.
damos_walk() adds the caller's request to the queue first. After that, it
shows if the kdamond of the damon_ctx is still running (damon_ctx->kdamond
is set). Only if the kdamond is running, damos_walk() starts waiting for
the kdamond's handling of the newly added request.
The damos_walk() requests registration and damon_ctx->kdamond unset are
protected by different mutexes, though. Hence, damos_walk() could race
with damon_ctx->kdamond unset, and result in deadlocks.
For example, let's suppose kdamond successfully finished the damow_walk()
request cancelling. Right after that, damos_walk() is called for the
context. It registers the new request, and shows the context is still
running, because damon_ctx->kdamond unset is not yet done. Hence the
damos_walk() caller starts waiting for the handling of the request.
However, the kdamond is already on the termination steps, so it never
handles the new request. As a result, the damos_walk() caller thread
infinitely waits.
Fix this by introducing another damon_ctx field, namely
walk_control_obsolete. It is protected by the
damon_ctx->walk_control_lock, which protects damos_walk() request
registration. Initialize (unset) it in kdamond_fn() before letting
damon_start() returns and set it just before the cancelling of the
remaining damos_walk() request is executed. damos_walk() reads the
obsolete field under the lock and avoids adding a new request.
After this change, only requests that are guaranteed to be handled or
cancelled are registered. Hence the after-registration DAMON context
termination check is no longer needed. Remove it together.
The issue is found by sashiko [1].
Link: https://lore.kernel.org/20260327233319.3528-3-sj@kernel.org
Link: https://lore.kernel.org/20260325141956.87144-1-sj@kernel.org [1]
Fixes: bf0eaba0ff ("mm/damon/core: implement damos_walk()")
Signed-off-by: SeongJae Park <sj@kernel.org>
Cc: <stable@vger.kernel.org> # 6.14.x
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "mm/damon/core: fix damon_call()/damos_walk() vs kdmond exit
race".
damon_call() and damos_walk() can leak memory and/or deadlock when they
race with kdamond terminations. Fix those.
This patch (of 2);
When kdamond_fn() main loop is finished, the function cancels all
remaining damon_call() requests and unset the damon_ctx->kdamond so that
API callers and API functions themselves can know the context is
terminated. damon_call() adds the caller's request to the queue first.
After that, it shows if the kdamond of the damon_ctx is still running
(damon_ctx->kdamond is set). Only if the kdamond is running, damon_call()
starts waiting for the kdamond's handling of the newly added request.
The damon_call() requests registration and damon_ctx->kdamond unset are
protected by different mutexes, though. Hence, damon_call() could race
with damon_ctx->kdamond unset, and result in deadlocks.
For example, let's suppose kdamond successfully finished the damon_call()
requests cancelling. Right after that, damon_call() is called for the
context. It registers the new request, and shows the context is still
running, because damon_ctx->kdamond unset is not yet done. Hence the
damon_call() caller starts waiting for the handling of the request.
However, the kdamond is already on the termination steps, so it never
handles the new request. As a result, the damon_call() caller threads
infinitely waits.
Fix this by introducing another damon_ctx field, namely
call_controls_obsolete. It is protected by the
damon_ctx->call_controls_lock, which protects damon_call() requests
registration. Initialize (unset) it in kdamond_fn() before letting
damon_start() returns and set it just before the cancelling of remaining
damon_call() requests is executed. damon_call() reads the obsolete field
under the lock and avoids adding a new request.
After this change, only requests that are guaranteed to be handled or
cancelled are registered. Hence the after-registration DAMON context
termination check is no longer needed. Remove it together.
Note that the deadlock will not happen when damon_call() is called for
repeat mode request. In tis case, damon_call() returns instead of waiting
for the handling when the request registration succeeds and it shows the
kdamond is running. However, if the request also has dealloc_on_cancel,
the request memory would be leaked.
The issue is found by sashiko [1].
Link: https://lore.kernel.org/20260327233319.3528-1-sj@kernel.org
Link: https://lore.kernel.org/20260327233319.3528-2-sj@kernel.org
Link: https://lore.kernel.org/20260325141956.87144-1-sj@kernel.org [1]
Fixes: 42b7491af1 ("mm/damon/core: introduce damon_call()")
Signed-off-by: SeongJae Park <sj@kernel.org>
Cc: <stable@vger.kernel.org> # 6.14.x
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Currently, per-CPU acomp_ctx are allocated on pool creation and/or CPU
hotplug, and destroyed on pool destruction or CPU hotunplug. This
complicates the lifetime management to save memory while a CPU is
offlined, which is not very common.
Simplify lifetime management by allocating per-CPU acomp_ctx once on pool
creation (or CPU hotplug for CPUs onlined later), and keeping them
allocated until the pool is destroyed.
Refactor cleanup code from zswap_cpu_comp_dead() into acomp_ctx_free() to
be used elsewhere.
The main benefit of using the CPU hotplug multi state instance startup
callback to allocate the acomp_ctx resources is that it prevents the cores
from being offlined until the multi state instance addition call returns.
From Documentation/core-api/cpu_hotplug.rst:
"The node list add/remove operations and the callback invocations are
serialized against CPU hotplug operations."
Furthermore, zswap_[de]compress() cannot contend with
zswap_cpu_comp_prepare() because:
- During pool creation/deletion, the pool is not in the zswap_pools
list.
- During CPU hot[un]plug, the CPU is not yet online, as Yosry pointed
out. zswap_cpu_comp_prepare() will be run on a control CPU,
since CPUHP_MM_ZSWP_POOL_PREPARE is in the PREPARE section of "enum
cpuhp_state".
In both these cases, any recursions into zswap reclaim from
zswap_cpu_comp_prepare() will be handled by the old pool.
The above two observations enable the following simplifications:
1) zswap_cpu_comp_prepare():
a) acomp_ctx mutex locking:
If the process gets migrated while zswap_cpu_comp_prepare() is
running, it will complete on the new CPU. In case of failures, we
pass the acomp_ctx pointer obtained at the start of
zswap_cpu_comp_prepare() to acomp_ctx_free(), which again, can
only undergo migration. There appear to be no contention
scenarios that might cause inconsistent values of acomp_ctx's
members. Hence, it seems there is no need for
mutex_lock(&acomp_ctx->mutex) in zswap_cpu_comp_prepare().
b) acomp_ctx mutex initialization:
Since the pool is not yet on zswap_pools list, we don't need to
initialize the per-CPU acomp_ctx mutex in
zswap_pool_create(). This has been restored to occur in
zswap_cpu_comp_prepare().
c) Subsequent CPU offline-online transitions:
zswap_cpu_comp_prepare() checks upfront if acomp_ctx->acomp is
valid. If so, it returns success. This should handle any CPU
hotplug online-offline transitions after pool creation is done.
2) CPU offline vis-a-vis zswap ops:
Let's suppose the process is migrated to another CPU before the
current CPU is dysfunctional. If zswap_[de]compress() holds the
acomp_ctx->mutex lock of the offlined CPU, that mutex will be
released once it completes on the new CPU. Since there is no
teardown callback, there is no possibility of UAF.
3) Pool creation/deletion and process migration to another CPU:
During pool creation/deletion, the pool is not in the zswap_pools
list. Hence it cannot contend with zswap ops on that CPU. However,
the process can get migrated.
a) Pool creation --> zswap_cpu_comp_prepare()
--> process migrated:
* Old CPU offline: no-op.
* zswap_cpu_comp_prepare() continues
to run on the new CPU to finish
allocating acomp_ctx resources for
the offlined CPU.
b) Pool deletion --> acomp_ctx_free()
--> process migrated:
* Old CPU offline: no-op.
* acomp_ctx_free() continues
to run on the new CPU to finish
de-allocating acomp_ctx resources
for the offlined CPU.
4) Pool deletion vis-a-vis CPU onlining:
The call to cpuhp_state_remove_instance() cannot race with
zswap_cpu_comp_prepare() because of hotplug synchronization.
The current acomp_ctx_get_cpu_lock()/acomp_ctx_put_unlock() are deleted.
Instead, zswap_[de]compress() directly call
mutex_[un]lock(&acomp_ctx->mutex).
The per-CPU memory cost of not deleting the acomp_ctx resources upon CPU
offlining, and only deleting them when the pool is destroyed, is 8.28 KB
on x86_64. This cost is only paid when a CPU is offlined, until it is
onlined again.
Link: https://lore.kernel.org/20260331183351.29844-3-kanchanapsridhar2026@gmail.com
Co-developed-by: Kanchana P. Sridhar <kanchanapsridhar2026@gmail.com>
Signed-off-by: Kanchana P. Sridhar <kanchanapsridhar2026@gmail.com>
Signed-off-by: Kanchana P Sridhar <kanchana.p.sridhar@intel.com>
Acked-by: Yosry Ahmed <yosry@kernel.org>
Cc: Chengming Zhou <chengming.zhou@linux.dev>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Sergey Senozhatsky <senozhatsky@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "zswap pool per-CPU acomp_ctx simplifications", v3.
This patchset first removes redundant checks on the acomp_ctx and its
"req" member in zswap_cpu_comp_dead().
Next, it persists the zswap pool's per-CPU acomp_ctx resources to last
until the pool is destroyed. It then simplifies the per-CPU acomp_ctx
mutex locking in zswap_compress()/zswap_decompress().
Code comments added after allocation and before checking to deallocate the
per-CPU acomp_ctx's members, based on expected crypto API return values
and zswap changes this patchset makes.
Patch 2 is an independent submission of patch 23 from [1], to
facilitate merging.
This patch (of 2):
There are presently redundant checks on the per-CPU acomp_ctx and it's
"req" member in zswap_cpu_comp_dead(): redundant because they are
inconsistent with zswap_pool_create() handling of failure in allocating
the acomp_ctx, and with the expected NULL return value from the
acomp_request_alloc() API when it fails to allocate an acomp_req.
Fix these by converting to them to be NULL checks.
Add comments in zswap_cpu_comp_prepare() clarifying the expected return
values of the crypto_alloc_acomp_node() and acomp_request_alloc() API.
Link: https://lore.kernel.org/20260331183351.29844-2-kanchanapsridhar2026@gmail.com
Link: https://patchwork.kernel.org/project/linux-mm/list/?series=1046677
Signed-off-by: Kanchana P. Sridhar <kanchanapsridhar2026@gmail.com>
Suggested-by: Yosry Ahmed <yosry@kernel.org>
Acked-by: Yosry Ahmed <yosry@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
When shrinking lruvec, MGLRU pins address space before walking it. This
is excessive since all it needs for walking the page range is a stable
mm_struct to be able to take and release mmap_read_lock and a stable
mm->mm_mt tree to walk. This address space pinning results in delays when
releasing the memory of a dying process. This also prevents mm reapers
(both in-kernel oom-reaper and userspace process_mrelease()) from doing
their job during MGLRU scan because they check task_will_free_mem() which
will yield negative result due to the elevated mm->mm_users.
This affects the system in the sense that if the MM of the killed
process is being reclaimed by kswapd then reapers won't be able to reap
it. Even the process itself (which might have higher-priority than
kswapd) will not free its memory until kswapd drops the last reference.
IOW, we delay freeing the memory because kswapd is reclaiming it. In
Android the visible result for us is that process_mrelease() (userspace
reaper) skips MM in such cases and we see process memory not released
for an unusually long time (secs).
Replace unnecessary address space pinning with mm_struct pinning by
replacing mmget/mmput with mmgrab/mmdrop calls. mm_mt is contained within
mm_struct itself, therefore it won't be freed as long as mm_struct is
stable and it won't change during the walk because mmap_read_lock is being
held.
Link: https://lore.kernel.org/20260322070843.941997-1-surenb@google.com
Fixes: bd74fdaea1 ("mm: multi-gen LRU: support page table walks")
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Reviewed-by: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Liam Howlett <liam.howlett@oracle.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Qi Zheng <zhengqi.arch@bytedance.com>
Cc: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Wei Xu <weixugc@google.com>
Cc: Yuanchu Xie <yuanchu@google.com>
Cc: Yu Zhao <yuzhao@google.com>
Cc: Kalesh Singh <kaleshsingh@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Stop pinning modules indefinitely upon file handler registration.
Instead, dynamically increment the module reference count only when a live
update session actively uses the file handler (e.g., during preservation
or deserialization), and release it when the session ends.
This allows modules providing live update handlers to be gracefully
unloaded when no live update is in progress.
Link: https://lore.kernel.org/20260327033335.696621-11-pasha.tatashin@soleen.com
Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Reviewed-by: Pratyush Yadav (Google) <pratyush@kernel.org>
Cc: David Matlack <dmatlack@google.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Samiullah Khawaja <skhawaja@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Change liveupdate_unregister_file_handler and liveupdate_unregister_flb to
return void instead of an error code. This follows the design principle
that unregistration during module unload should not fail, as the unload
cannot be stopped at that point.
Link: https://lore.kernel.org/20260327033335.696621-10-pasha.tatashin@soleen.com
Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Reviewed-by: Pratyush Yadav (Google) <pratyush@kernel.org>
Cc: David Matlack <dmatlack@google.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Samiullah Khawaja <skhawaja@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Now that file handler unregistration automatically unregisters all
associated file handlers (FLBs), the liveupdate_test_unregister() function
is no longer needed. Remove it along with its usages and declarations.
Link: https://lore.kernel.org/20260327033335.696621-9-pasha.tatashin@soleen.com
Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Reviewed-by: Pratyush Yadav (Google) <pratyush@kernel.org>
Cc: David Matlack <dmatlack@google.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Samiullah Khawaja <skhawaja@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
To ensure that unregistration is always successful and doesn't leave
dangling resources, introduce auto-unregistration of FLBs: when a file
handler is unregistered, all FLBs associated with it are automatically
unregistered.
Introduce a new helper luo_flb_unregister_all() which unregisters all FLBs
linked to the given file handler.
Link: https://lore.kernel.org/20260327033335.696621-8-pasha.tatashin@soleen.com
Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Reviewed-by: Pratyush Yadav (Google) <pratyush@kernel.org>
Cc: David Matlack <dmatlack@google.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Samiullah Khawaja <skhawaja@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Now that FLB module references are handled dynamically during active
sessions, we can safely remove the luo_session_quiesce() and
luo_session_resume() mechanism.
Link: https://lore.kernel.org/20260327033335.696621-7-pasha.tatashin@soleen.com
Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Reviewed-by: Pratyush Yadav (Google) <pratyush@kernel.org>
Cc: David Matlack <dmatlack@google.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Samiullah Khawaja <skhawaja@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Stop pinning modules indefinitely upon FLB registration. Instead,
dynamically take a module reference when the FLB is actively used in a
session (e.g., during preserve and retrieve) and release it when the
session concludes.
This allows modules providing FLB operations to be cleanly unloaded when
not in active use by the live update orchestrator.
Link: https://lore.kernel.org/20260327033335.696621-6-pasha.tatashin@soleen.com
Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Reviewed-by: Samiullah Khawaja <skhawaja@google.com>
Reviewed-by: Pratyush Yadav (Google) <pratyush@kernel.org>
Cc: David Matlack <dmatlack@google.com>
Cc: Mike Rapoport <rppt@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Because liveupdate FLB objects will soon drop their persistent module
references when registered, list traversals must be protected against
concurrent module unloading.
To provide this protection, utilize the global luo_register_rwlock. It
protects the global registry of FLBs and the handler's specific list of
FLB dependencies.
Read locks are used during concurrent list traversals (e.g., during
preservation and serialization). Write locks are taken during
registration and unregistration.
Link: https://lore.kernel.org/20260327033335.696621-5-pasha.tatashin@soleen.com
Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Reviewed-by: Pratyush Yadav (Google) <pratyush@kernel.org>
Cc: David Matlack <dmatlack@google.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Samiullah Khawaja <skhawaja@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Because liveupdate file handlers will no longer hold a module reference
when registered, we must ensure that the access to the handler list is
protected against concurrent module unloading.
Utilize the global luo_register_rwlock to protect the global registry of
file handlers. Read locks are taken during list traversals in
luo_preserve_file() and luo_file_deserialize(). Write locks are taken
during registration and unregistration.
Link: https://lore.kernel.org/20260327033335.696621-4-pasha.tatashin@soleen.com
Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Reviewed-by: Pratyush Yadav (Google) <pratyush@kernel.org>
Cc: David Matlack <dmatlack@google.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Samiullah Khawaja <skhawaja@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The luo_flb_get_private() function, which is responsible for lazily
initializing the private state of FLB objects, can be called concurrently
from multiple threads. This creates a data race on the 'initialized' flag
and can lead to multiple executions of mutex_init() and INIT_LIST_HEAD()
on the same memory.
Introduce a static spinlock (luo_flb_init_lock) local to the function to
synchronize the initialization path. Use smp_load_acquire() and
smp_store_release() for memory ordering between the fast path and the slow
path.
Link: https://lore.kernel.org/20260327033335.696621-3-pasha.tatashin@soleen.com
Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Reviewed-by: Pratyush Yadav <pratyush@kernel.org>
Cc: David Matlack <dmatlack@google.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Samiullah Khawaja <skhawaja@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "liveupdate: Fix module unloading and unregister API", v3.
This patch series addresses an issue with how LUO handles module reference
counting and unregistration during a module unload (e.g., via rmmod).
Currently, modules that register live update file handlers are pinned for
the entire duration they are registered. This prevents the modules from
being unloaded gracefully, even when no live update session is in
progress.
Furthermore, if a module is forcefully unloaded, the unregistration
functions return an error (e.g. -EBUSY) if a session is active, which is
ignored by the kernel's module unload path, leaving dangling pointers in
the LUO global lists.
To resolve these issues, this series introduces the following changes:
1. Adds a global read-write semaphore (luo_register_rwlock) to protect
the registration lists for both file handlers and FLBs.
2. Reduces the scope of module reference counting for file handlers and
FLBs. Instead of pinning modules indefinitely upon registration,
references are now taken only when they are actively used in a live
update session (e.g., during preservation, retrieval, or
deserialization).
3. Removes the global luo_session_quiesce() mechanism since module
unload behavior now handles active sessions implicitly.
4. Introduces auto-unregistration of FLBs during file handler
unregistration to prevent leaving dangling resources.
5. Changes the unregistration functions to return void instead of
an error code.
6. Fixes a data race in luo_flb_get_private() by introducing a spinlock
for thread-safe lazy initialization.
7. Strengthens security by using %.*s when printing untrusted deserialized
compatible strings and session names to prevent out-of-bounds reads.
This patch (of 10):
Deserialized strings from KHO data (such as file handler compatible
strings and session names) are provided by the previous kernel and might
not be null-terminated if the data is corrupted or maliciously crafted.
When printing these strings in error messages, use the %.*s format
specifier with the maximum buffer size to prevent out-of-bounds reads into
adjacent kernel memory.
Link: https://lore.kernel.org/20260327033335.696621-1-pasha.tatashin@soleen.com
Link: https://lore.kernel.org/20260327033335.696621-2-pasha.tatashin@soleen.com
Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Reviewed-by: Pratyush Yadav (Google) <pratyush@kernel.org>
Cc: David Matlack <dmatlack@google.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Samiullah Khawaja <skhawaja@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The balance_dirty_pages() won't do the dirty folios throttling on
cgroupv1. See commit 9badce000e ("cgroup, writeback: don't enable
cgroup writeback on traditional hierarchies").
Moreover, after commit 6b0dfabb35 ("fs: Remove aops->writepage"), we no
longer attempt to write back filesystem folios through reclaim.
On large memory systems, the flusher may not be able to write back quickly
enough. Consequently, MGLRU will encounter many folios that are already
under writeback. Since we cannot reclaim these dirty folios, the system
may run out of memory and trigger the OOM killer.
Hence, for cgroup v1, let's throttle reclaim after waking up the flusher,
which is similar to commit 81a70c21d9 ("mm/cgroup/reclaim: fix dirty
pages throttling on cgroup v1"), to avoid unnecessary OOM.
The following test program can easily reproduce the OOM issue. With this
patch applied, the test passes successfully.
$mkdir /sys/fs/cgroup/memory/test
$echo 256M > /sys/fs/cgroup/memory/test/memory.limit_in_bytes
$echo $$ > /sys/fs/cgroup/memory/test/cgroup.procs
$dd if=/dev/zero of=/mnt/data.bin bs=1M count=800
Link: https://lore.kernel.org/3445af0f09e8ca945492e052e82594f8c4f2e2f6.1774606060.git.baolin.wang@linux.alibaba.com
Fixes: ac35a49023 ("mm: multi-gen LRU: minimal implementation")
Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>
Reviewed-by: Barry Song <baohua@kernel.org>
Reviewed-by: Kairui Song <kasong@tencent.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Qi Zheng <zhengqi.arch@bytedance.com>
Cc: Wei Xu <weixugc@google.com>
Cc: Yuanchu Xie <yuanchu@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Verify that a file can only be preserved once across all active sessions.
Attempting to preserve it a second time, whether in the same or a
different session, should fail with EBUSY.
Link: https://lore.kernel.org/20260326163943.574070-4-pasha.tatashin@soleen.com
Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Reviewed-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
Reviewed-by: Samiullah Khawaja <skhawaja@google.com>
Cc: David Matlack <dmatlack@google.com>
Cc: Pratyush Yadav <pratyush@kernel.org>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Christian Brauner <brauner@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Memfds are identified by their underlying inode. Implement get_id for
memfd_luo to return the inode pointer. This prevents the same memfd from
being managed twice by LUO if the same inode is pointed by multiple file
objects.
Link: https://lore.kernel.org/20260326163943.574070-3-pasha.tatashin@soleen.com
Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Reviewed-by: Pratyush Yadav (Google) <pratyush@kernel.org>
Cc: David Matlack <dmatlack@google.com>
Cc: Mike Rapoport (Microsoft) <rppt@kernel.org>
Cc: Samiullah Khawaja <skhawaja@google.com>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Christian Brauner <brauner@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "liveupdate: prevent double preservation", v4.
Currently, LUO does not prevent the same file from being managed twice
across different active sessions.
Because LUO preserves files of absolutely different types: memfd, and
upcoming vfiofd [1], iommufd [2], guestmefd (and possible kvmfd/cpufd).
There is no common private data or guarantee on how to prevent that the
same file is not preserved twice beside using inode or some slower and
expensive method like hashtables.
This patch (of 4)
Currently, LUO does not prevent the same file from being managed twice
across different active sessions.
Use a global xarray luo_preserved_files to keep track of file identifiers
being preserved by LUO. Update luo_preserve_file() to check and insert
the file identifier into this xarray when it is preserved, and erase it in
luo_file_unpreserve_files() when it is released.
To allow handlers to define what constitutes a "unique" file (e.g.,
different struct file objects pointing to the same hardware resource), add
a get_id() callback to struct liveupdate_file_ops. If not provided, the
default identifier is the struct file pointer itself.
This ensures that the same file (or resource) cannot be managed by
multiple sessions. If another session attempts to preserve an already
managed file, it will now fail with -EBUSY.
Link: https://lore.kernel.org/20260326163943.574070-1-pasha.tatashin@soleen.com
Link: https://lore.kernel.org/20260326163943.574070-2-pasha.tatashin@soleen.com
Link: https://lore.kernel.org/all/20260129212510.967611-1-dmatlack@google.com [1]
Link: https://lore.kernel.org/all/20260203220948.2176157-1-skhawaja@google.com [2]
Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Reviewed-by: Samiullah Khawaja <skhawaja@google.com>
Reviewed-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
Cc: David Matlack <dmatlack@google.com>
Cc: Pratyush Yadav <pratyush@kernel.org>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Christian Brauner <brauner@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Add documentation for the kexec-metadata feature that tracks the previous
kernel version and kexec boot count across kexec reboots. This helps
diagnose bugs that only reproduce when kexecing from specific kernel
versions.
Link: https://lore.kernel.org/20260316-kho-v9-6-ed6dcd951988@debian.org
Signed-off-by: Breno Leitao <leitao@debian.org>
Suggested-by: Mike Rapoport <rppt@kernel.org>
Reviewed-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
Reviewed-by: Pratyush Yadav <pratyush@kernel.org>
Cc: Alexander Graf <graf@amazon.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: "Liam R. Howlett" <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes <ljs@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Pasha Tatashin <pasha.tatashin@soleen.com>
Cc: SeongJae Park <sj@kernel.org>
Cc: Shuah Khan <skhan@linuxfoundation.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Use Kexec Handover (KHO) to pass the previous kernel's version string and
the number of kexec reboots since the last cold boot to the next kernel,
and print it at boot time.
Example output:
[ 0.000000] KHO: exec from: 6.19.0-rc4-next-20260107 (count 1)
Motivation
==========
Bugs that only reproduce when kexecing from specific kernel versions are
difficult to diagnose. These issues occur when a buggy kernel kexecs into
a new kernel, with the bug manifesting only in the second kernel.
Recent examples include the following commits:
* commit eb22663125 ("x86/boot: Fix page table access in
5-level to 4-level paging transition")
* commit 77d48d39e9 ("efistub/tpm: Use ACPI reclaim memory
for event log to avoid corruption")
* commit 64b45dd46e ("x86/efi: skip memattr table on kexec
boot")
As kexec-based reboots become more common, these version-dependent bugs
are appearing more frequently. At scale, correlating crashes to the
previous kernel version is challenging, especially when issues only occur
in specific transition scenarios.
Implementation
==============
The kexec metadata is stored as a plain C struct (struct
kho_kexec_metadata) rather than FDT format, for simplicity and direct
field access. It is registered via kho_add_subtree() as a separate
subtree, keeping it independent from the core KHO ABI. This design
choice:
- Keeps the core KHO ABI minimal and stable
- Allows the metadata format to evolve independently
- Avoids requiring version bumps for all KHO consumers (LUO, etc.)
when the metadata format changes
The struct kho_kexec_metadata contains two fields:
- previous_release: The kernel version that initiated the kexec
- kexec_count: Number of kexec boots since last cold boot
On cold boot, kexec_count starts at 0 and increments with each kexec. The
count helps identify issues that only manifest after multiple consecutive
kexec reboots.
[leitao@debian.org: call kho_kexec_metadata_init() for both boot paths]
Link: https://lore.kernel.org/all/20260309-kho-v8-5-c3abcf4ac750@debian.org/ [1]
Link: https://lore.kernel.org/20260409-kho_fix_merge_issue-v1-1-710c84ceaa85@debian.org
Link: https://lore.kernel.org/20260316-kho-v9-5-ed6dcd951988@debian.org
Signed-off-by: Breno Leitao <leitao@debian.org>
Acked-by: SeongJae Park <sj@kernel.org>
Reviewed-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
Reviewed-by: Pratyush Yadav <pratyush@kernel.org>
Cc: Alexander Graf <graf@amazon.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: "Liam R. Howlett" <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes <ljs@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Pasha Tatashin <pasha.tatashin@soleen.com>
Cc: Shuah Khan <skhan@linuxfoundation.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
kho_in_debugfs_init() calls fdt_totalsize() to determine blob sizes, which
assumes all blobs are FDTs. This breaks for non-FDT blobs like struct
kho_kexec_metadata.
Fix this by reading the "blob-size" property from the FDT (persisted by
kho_add_subtree()) instead of calling fdt_totalsize(). Also rename local
variables from fdt_phys/sub_fdt to blob_phys/blob for consistency with the
non-FDT-specific naming.
Link: https://lore.kernel.org/20260316-kho-v9-4-ed6dcd951988@debian.org
Signed-off-by: Breno Leitao <leitao@debian.org>
Reviewed-by: Pratyush Yadav <pratyush@kernel.org>
Cc: Alexander Graf <graf@amazon.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: "Liam R. Howlett" <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes <ljs@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Mike Rapoport (Microsoft) <rppt@kernel.org>
Cc: Pasha Tatashin <pasha.tatashin@soleen.com>
Cc: SeongJae Park <sj@kernel.org>
Cc: Shuah Khan <skhan@linuxfoundation.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
kho_add_subtree() accepts a size parameter but only forwards it to
debugfs. The size is not persisted in the KHO FDT, so it is lost across
kexec. This makes it impossible for the incoming kernel to determine the
blob size without understanding the blob format.
Store the blob size as a "blob-size" property in the KHO FDT alongside the
"preserved-data" physical address. This allows the receiving kernel to
recover the size for any blob regardless of format.
Also extend kho_retrieve_subtree() with an optional size output parameter
so callers can learn the blob size without needing to understand the blob
format. Update all callers to pass NULL for the new parameter.
Link: https://lore.kernel.org/20260316-kho-v9-3-ed6dcd951988@debian.org
Signed-off-by: Breno Leitao <leitao@debian.org>
Reviewed-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
Reviewed-by: Pratyush Yadav <pratyush@kernel.org>
Cc: Alexander Graf <graf@amazon.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: "Liam R. Howlett" <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes <ljs@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Pasha Tatashin <pasha.tatashin@soleen.com>
Cc: SeongJae Park <sj@kernel.org>
Cc: Shuah Khan <skhan@linuxfoundation.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Since kho_add_subtree() now accepts arbitrary data blobs (not just FDTs),
rename the parameter from 'fdt' to 'blob' to better reflect its purpose.
Apply the same rename to kho_remove_subtree() for consistency.
Also rename kho_debugfs_fdt_add() and kho_debugfs_fdt_remove() to
kho_debugfs_blob_add() and kho_debugfs_blob_remove() respectively, with
the same parameter rename from 'fdt' to 'blob'.
Link: https://lore.kernel.org/20260316-kho-v9-2-ed6dcd951988@debian.org
Signed-off-by: Breno Leitao <leitao@debian.org>
Reviewed-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
Reviewed-by: Pratyush Yadav <pratyush@kernel.org>
Cc: Alexander Graf <graf@amazon.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: "Liam R. Howlett" <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes <ljs@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Pasha Tatashin <pasha.tatashin@soleen.com>
Cc: SeongJae Park <sj@kernel.org>
Cc: Shuah Khan <skhan@linuxfoundation.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "kho: history: track previous kernel version and kexec boot
count", v9.
Use Kexec Handover (KHO) to pass the previous kernel's version string and
the number of kexec reboots since the last cold boot to the next kernel,
and print it at boot time.
Example
=======
[ 0.000000] Linux version 6.19.0-rc3-upstream-00047-ge5d992347849
...
[ 0.000000] KHO: exec from: 6.19.0-rc4-next-20260107upstream-00004-g3071b0dc4498 (count 1)
Motivation
==========
Bugs that only reproduce when kexecing from specific kernel versions are
difficult to diagnose. These issues occur when a buggy kernel kexecs into
a new kernel, with the bug manifesting only in the second kernel.
Recent examples include:
* eb22663125 ("x86/boot: Fix page table access in 5-level to 4-level paging transition")
* 77d48d39e9 ("efistub/tpm: Use ACPI reclaim memory for event log to avoid corruption")
* 64b45dd46e ("x86/efi: skip memattr table on kexec boot")
As kexec-based reboots become more common, these version-dependent bugs
are appearing more frequently. At scale, correlating crashes to the
previous kernel version is challenging, especially when issues only occur
in specific transition scenarios.
Some bugs manifest only after multiple consecutive kexec reboots.
Tracking the kexec count helps identify these cases (this metric is
already used by live update sub-system).
KHO provides a reliable mechanism to pass information between kernels. By
carrying the previous kernel's release string and kexec count forward, we
can print this context at boot time to aid debugging.
The goal of this feature is to have this information being printed in
early boot, so, users can trace back kernel releases in kexec. Systemd is
not helpful because we cannot assume that the previous kernel has systemd
or even write access to the disk (common when using Linux as bootloaders)
This patch (of 6):
kho_add_subtree() assumes the fdt argument is always an FDT and calls
fdt_totalsize() on it in the debugfs code path. This assumption will
break if a caller passes arbitrary data instead of an FDT.
When CONFIG_KEXEC_HANDOVER_DEBUGFS is enabled, kho_debugfs_fdt_add() calls
__kho_debugfs_fdt_add(), which executes:
f->wrapper.size = fdt_totalsize(fdt);
Fix this by adding an explicit size parameter to kho_add_subtree() so
callers specify the blob size. This allows subtrees to contain arbitrary
data formats, not just FDTs. Update all callers:
- memblock.c: use fdt_totalsize(fdt)
- luo_core.c: use fdt_totalsize(fdt_out)
- test_kho.c: use fdt_totalsize()
- kexec_handover.c (root fdt): use fdt_totalsize(kho_out.fdt)
Also update __kho_debugfs_fdt_add() to receive the size explicitly instead
of computing it internally via fdt_totalsize(). In kho_in_debugfs_init(),
pass fdt_totalsize() for the root FDT and sub-blobs since all current
users are FDTs. A subsequent patch will persist the size in the KHO FDT
so the incoming side can handle non-FDT blobs correctly.
Link: https://lore.kernel.org/20260323110747.193569-1-duanchenghao@kylinos.cn
Link: https://lore.kernel.org/20260316-kho-v9-1-ed6dcd951988@debian.org
Signed-off-by: Breno Leitao <leitao@debian.org>
Suggested-by: Pratyush Yadav <pratyush@kernel.org>
Reviewed-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
Reviewed-by: Pratyush Yadav <pratyush@kernel.org>
Cc: Alexander Graf <graf@amazon.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: "Liam R. Howlett" <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes <ljs@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Pasha Tatashin <pasha.tatashin@soleen.com>
Cc: SeongJae Park <sj@kernel.org>
Cc: Shuah Khan <skhan@linuxfoundation.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Add a Kconfig option to default kmemleak verbose mode on at build time.
This option depends on DEBUG_KMEMLEAK_AUTO_SCAN since verbose reporting is
only meaningful when the automatic scanning thread is running.
When enabled, kmemleak prints full details (backtrace, hex dump, address)
of unreferenced objects to dmesg as they are detected during scanning,
removing the need to manually read /sys/kernel/debug/kmemleak.
Making this a compile-time option rather than a boot parameter allows
debug kernel flavors to enable verbose kmemleak reporting by default
without requiring changes to boot arguments. A machine can simply swap to
a debug kernel and benefit from kmemleak reporting automatically.
By surfacing leak reports directly in dmesg, they are automatically
forwarded through any kernel logging infrastructure and can be easily
captured by log aggregation tooling, making it practical to monitor memory
leaks across large fleets.
The verbose setting can still be toggled at runtime via
/sys/module/kmemleak/parameters/verbose.
Link: https://lore.kernel.org/20260323-kmemleak_report-v1-1-ba2cdd9c11b9@debian.org
Signed-off-by: Breno Leitao <leitao@debian.org>
Acked-by: SeongJae Park <sj@kernel.org>
Acked-by: Vlastimil Babka (SUSE) <vbabka@kernel.org>
Reviewed-by: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Liam Howlett <liam.howlett@oracle.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
We are moving to a far more proactive model of maintainership within mm
and thus put a great deal of emphasis on sub-maintainers being active
within the community both in terms of code contributions and review.
The MGLRU has not had much activity since being added to the kernel and
the current maintainers who kindly stepped up have unfortunately not been
able to contribute a great deal to it for over a year, nor engage all that
heavily in review.
As a result, and within no negative connotations implied whatsoever, it
seems appropriate to downgrade the current maintainers to reviewers.
At this time nobody is quite exercising the maintainer role in this area
of the kernel, but there is encouraging activity from a number of people
who are trusted elsewhere in the kernel, and who have contributed relevant
work or review.
Therefore add further reviewers, and at this stage - to reflect the
reality on the ground - we will not have any sub-maintainers listed at
all.
Each of the files listed are shared with other sections in MAINTAINERS, so
this doesn't reduce sub-maintainer coverage.
Link: https://lore.kernel.org/20260326185629.355476-1-ljs@kernel.org
Signed-off-by: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
Acked-by: Axel Rasmussen <axelrasmussen@google.com>
Acked-by: Vlastimil Babka (SUSE) <vbabka@kernel.org>
Acked-by: David Hildenbrand (Arm) <david@kernel.org>
Acked-by: Barry Song <baohua@kernel.org>
Acked-by: SeongJae Park <sj@kernel.org>
Acked-by: Kairui Song <kasong@tencent.com>
Acked-by: Qi Zheng <qi.zheng@linux.dev>
Acked-by: Yuanchu Xie <yuanchu@google.com>
Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Wei Xu <weixugc@google.com>
Cc: Kalesh Singh <kaleshsingh@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The nr_pages parameter of mem_cgroup_update_lru_size() represents a page
count. During the reparenting of LRU folios, the value passed to it can
potentially exceed the maximum value of a 32-bit integer. It should be
declared as long instead of int to match the types used in lruvec size
accounting and to prevent possible overflow.
Update the parameter type to long to ensure correctness.
Link: https://lore.kernel.org/fd4140de44fa0a3978e4e2426731187fe8625f0b.1774604356.git.zhengqi.arch@bytedance.com
Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com>
Reviewed-by: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
Reviewed-by: Harry Yoo (Oracle) <harry@kernel.org>
Cc: Allen Pais <apais@linux.microsoft.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Hamza Mahfooz <hamzamahfooz@linux.microsoft.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Imran Khan <imran.f.khan@oracle.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kamalesh Babulal <kamalesh.babulal@oracle.com>
Cc: Lance Yang <lance.yang@linux.dev>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Michal Koutný <mkoutny@suse.com>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Usama Arif <usamaarif642@gmail.com>
Cc: Wei Xu <weixugc@google.com>
Cc: Yuanchu Xie <yuanchu@google.com>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The __mod_memcg_state() and __mod_memcg_lruvec_state() functions are also
used to reparent non-hierarchical stats. In this scenario, the values
passed to them are accumulated statistics that might be extremely large
and exceed the upper limit of a 32-bit integer.
Change the val parameter type from int to long in these functions and
their corresponding tracepoints (memcg_rstat_stats) to prevent potential
overflow issues.
After that, in memcg_state_val_in_pages(), if the passed val is negative,
the expression val * unit / PAGE_SIZE could be implicitly converted to a
massive positive number when compared with 1UL in the max() macro. This
leads to returning an incorrect massive positive value.
Fix this by using abs(val) to calculate the magnitude first, and then
restoring the sign of the value before returning the result.
Additionally, use mult_frac() to prevent potential overflow during the
multiplication of val and unit.
Link: https://lore.kernel.org/70a9440e49c464b4dca88bcabc6b491bd335c9f0.1774604356.git.zhengqi.arch@bytedance.com
Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com>
Reported-by: Harry Yoo (Oracle) <harry@kernel.org>
Reviewed-by: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
Reviewed-by: Harry Yoo (Oracle) <harry@kernel.org>
Cc: Allen Pais <apais@linux.microsoft.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Hamza Mahfooz <hamzamahfooz@linux.microsoft.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Imran Khan <imran.f.khan@oracle.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kamalesh Babulal <kamalesh.babulal@oracle.com>
Cc: Lance Yang <lance.yang@linux.dev>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Michal Koutný <mkoutny@suse.com>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Usama Arif <usamaarif642@gmail.com>
Cc: Wei Xu <weixugc@google.com>
Cc: Yuanchu Xie <yuanchu@google.com>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "fix unexpected type conversions and potential overflows",
v3.
As Harry Yoo pointed out [1], in scenarios where massive state updates
occur (e.g., during the reparenting of LRU folios), the values passed to
memcg stat update functions can accumulate and exceed the upper limit of a
32-bit integer.
If the parameter types are not large enough (like 'int') or are handled
incorrectly, it can lead to severe truncation, potential overflow issues,
and unexpected type conversion bugs.
This series aims to address these issues by correcting the parameter types
in the relevant functions, and by fixing an implicit conversion bug in
memcg_state_val_in_pages().
This patch (of 3):
The memcg_rstat_updated() tracks updates for vmstats_percpu->state and
lruvec_stats_percpu->state. Since these state values are of type long,
change the val parameter passed to memcg_rstat_updated() to long as well.
Correspondingly, change the type of stats_updates in struct
memcg_vmstats_percpu and struct memcg_vmstats from unsigned int and
atomic_t to unsigned long and atomic_long_t respectively to prevent
potential overflow when handling large state updates during the
reparenting of LRU folios.
Link: https://lore.kernel.org/cover.1774604356.git.zhengqi.arch@bytedance.com
Link: https://lore.kernel.org/a5b0b468e7b4fe5f26c50e36d5d016f16d92f98f.1774604356.git.zhengqi.arch@bytedance.com
Link: https://lore.kernel.org/all/acDxaEgnqPI-Z4be@hyeyoo/ [1]
Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com>
Reviewed-by: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
Reviewed-by: Harry Yoo (Oracle) <harry@kernel.org>
Cc: Allen Pais <apais@linux.microsoft.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Hamza Mahfooz <hamzamahfooz@linux.microsoft.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Imran Khan <imran.f.khan@oracle.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kamalesh Babulal <kamalesh.babulal@oracle.com>
Cc: Lance Yang <lance.yang@linux.dev>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Michal Koutný <mkoutny@suse.com>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Usama Arif <usamaarif642@gmail.com>
Cc: Wei Xu <weixugc@google.com>
Cc: Yuanchu Xie <yuanchu@google.com>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
We must ensure the folio is deleted from or added to the correct lruvec
list. So, add VM_WARN_ON_ONCE_FOLIO() to catch invalid users. The
VM_BUG_ON_PAGE() in move_pages_to_lru() can be removed as
add_page_to_lru_list() will perform the necessary check.
Link: https://lore.kernel.org/2c90fc006d9d730331a3caeef96f7e5dabe2036d.1772711148.git.zhengqi.arch@bytedance.com
Signed-off-by: Muchun Song <songmuchun@bytedance.com>
Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com>
Acked-by: Roman Gushchin <roman.gushchin@linux.dev>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Allen Pais <apais@linux.microsoft.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Chengming Zhou <chengming.zhou@linux.dev>
Cc: Chen Ridong <chenridong@huawei.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Hamza Mahfooz <hamzamahfooz@linux.microsoft.com>
Cc: Harry Yoo <harry.yoo@oracle.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Imran Khan <imran.f.khan@oracle.com>
Cc: Kamalesh Babulal <kamalesh.babulal@oracle.com>
Cc: Lance Yang <lance.yang@linux.dev>
Cc: Liam Howlett <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michal Koutný <mkoutny@suse.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Usama Arif <usamaarif642@gmail.com>
Cc: Vlastimil Babka <vbabka@kernel.org>
Cc: Wei Xu <weixugc@google.com>
Cc: Yosry Ahmed <yosry@kernel.org>
Cc: Yuanchu Xie <yuanchu@google.com>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Now that everything is set up, switch folio->memcg_data pointers to
objcgs, update the accessors, and execute reparenting on cgroup death.
Finally, folio->memcg_data of LRU folios and kmem folios will always point
to an object cgroup pointer. The folio->memcg_data of slab folios will
point to an vector of object cgroups.
Link: https://lore.kernel.org/80cb7af198dc6f2173fe616d1207a4c315ece141.1772711148.git.zhengqi.arch@bytedance.com
Signed-off-by: Muchun Song <songmuchun@bytedance.com>
Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com>
Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Allen Pais <apais@linux.microsoft.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Chengming Zhou <chengming.zhou@linux.dev>
Cc: Chen Ridong <chenridong@huawei.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Hamza Mahfooz <hamzamahfooz@linux.microsoft.com>
Cc: Harry Yoo <harry.yoo@oracle.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Imran Khan <imran.f.khan@oracle.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kamalesh Babulal <kamalesh.babulal@oracle.com>
Cc: Lance Yang <lance.yang@linux.dev>
Cc: Liam Howlett <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michal Koutný <mkoutny@suse.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Usama Arif <usamaarif642@gmail.com>
Cc: Vlastimil Babka <vbabka@kernel.org>
Cc: Wei Xu <weixugc@google.com>
Cc: Yosry Ahmed <yosry@kernel.org>
Cc: Yuanchu Xie <yuanchu@google.com>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Convert objcg to be per-memcg per-node type, so that when reparent LRU
folios later, we can hold the lru lock at the node level, thus avoiding
holding too many lru locks at once.
[zhengqi.arch@bytedance.com: reset pn->orig_objcg to NULL]
Link: https://lore.kernel.org/20260309112939.31937-1-qi.zheng@linux.dev
[akpm@linux-foundation.org: fix comment typo, per Usama. Reflow comment to 80 cols]
[devnexen@gmail.com: fix obj_cgroup leak in mem_cgroup_css_online() error path]
Link: https://lore.kernel.org/20260322193631.45457-1-devnexen@gmail.com
[devnexen@gmail.com: add newline, per Qi Zheng]
Link: https://lore.kernel.org/20260323063007.7783-1-devnexen@gmail.com
Link: https://lore.kernel.org/56c04b1c5d54f75ccdc12896df6c1ca35403ecc3.1772711148.git.zhengqi.arch@bytedance.com
Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com>
Signed-off-by: David Carlier <devnexen@gmail.com>
Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Allen Pais <apais@linux.microsoft.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Chengming Zhou <chengming.zhou@linux.dev>
Cc: Chen Ridong <chenridong@huawei.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Hamza Mahfooz <hamzamahfooz@linux.microsoft.com>
Cc: Harry Yoo <harry.yoo@oracle.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Imran Khan <imran.f.khan@oracle.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kamalesh Babulal <kamalesh.babulal@oracle.com>
Cc: Lance Yang <lance.yang@linux.dev>
Cc: Liam Howlett <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michal Koutný <mkoutny@suse.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Muchun Song <songmuchun@bytedance.com>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Usama Arif <usamaarif642@gmail.com>
Cc: Vlastimil Babka <vbabka@kernel.org>
Cc: Wei Xu <weixugc@google.com>
Cc: Yosry Ahmed <yosry@kernel.org>
Cc: Yuanchu Xie <yuanchu@google.com>
Cc: Zi Yan <ziy@nvidia.com>
Cc: Usama Arif <usama.arif@linux.dev>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
To resolve the dying memcg issue, we need to reparent LRU folios of child
memcg to its parent memcg. This could cause problems for non-hierarchical
stats.
As Yosry Ahmed pointed out:
In short, if memory is charged to a dying cgroup at the time of
reparenting, when the memory gets uncharged the stats updates will occur
at the parent. This will update both hierarchical and non-hierarchical
stats of the parent, which would corrupt the parent's non-hierarchical
stats (because those counters were never incremented when the memory was
charged).
Now we have the following two types of non-hierarchical stats, and they
are only used in CONFIG_MEMCG_V1:
a. memcg->vmstats->state_local[i]
b. pn->lruvec_stats->state_local[i]
To ensure that these non-hierarchical stats work properly, we need to
reparent these non-hierarchical stats after reparenting LRU folios. To
this end, this commit makes the following preparations:
1. implement reparent_state_local() to reparent non-hierarchical stats
2. make css_killed_work_fn() to be called in rcu work, and implement
get_non_dying_memcg_start() and get_non_dying_memcg_end() to avoid race
between mod_memcg_state()/mod_memcg_lruvec_state()
and reparent_state_local()
Link: https://lore.kernel.org/e862995c45a7101a541284b6ebee5e5c32c89066.1772711148.git.zhengqi.arch@bytedance.com
Co-developed-by: Yosry Ahmed <yosry@kernel.org>
Signed-off-by: Yosry Ahmed <yosry@kernel.org>
Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com>
Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Allen Pais <apais@linux.microsoft.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Chengming Zhou <chengming.zhou@linux.dev>
Cc: Chen Ridong <chenridong@huawei.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Hamza Mahfooz <hamzamahfooz@linux.microsoft.com>
Cc: Harry Yoo <harry.yoo@oracle.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Imran Khan <imran.f.khan@oracle.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kamalesh Babulal <kamalesh.babulal@oracle.com>
Cc: Lance Yang <lance.yang@linux.dev>
Cc: Liam Howlett <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michal Koutný <mkoutny@suse.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Muchun Song <songmuchun@bytedance.com>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Usama Arif <usamaarif642@gmail.com>
Cc: Vlastimil Babka <vbabka@kernel.org>
Cc: Wei Xu <weixugc@google.com>
Cc: Yuanchu Xie <yuanchu@google.com>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
For cgroup v2, count_shadow_nodes() is the only place to read
non-hierarchical stats (lruvec_stats->state_local). To avoid the need to
consider cgroup v2 during subsequent non-hierarchical stats reparenting,
use lruvec_lru_size() instead of lruvec_page_state_local() to get the
number of lru pages.
For NR_SLAB_RECLAIMABLE_B and NR_SLAB_UNRECLAIMABLE_B cases, it appears
that the statistics here have already been problematic for a while since
slab pages have been reparented. So just ignore it for now.
Link: https://lore.kernel.org/b1d448c667a8fb377c3390d9aba43bdb7e4d5739.1772711148.git.zhengqi.arch@bytedance.com
Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com>
Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
Acked-by: Muchun Song <muchun.song@linux.dev>
Cc: Allen Pais <apais@linux.microsoft.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Chengming Zhou <chengming.zhou@linux.dev>
Cc: Chen Ridong <chenridong@huawei.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: Hamza Mahfooz <hamzamahfooz@linux.microsoft.com>
Cc: Harry Yoo <harry.yoo@oracle.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Imran Khan <imran.f.khan@oracle.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kamalesh Babulal <kamalesh.babulal@oracle.com>
Cc: Lance Yang <lance.yang@linux.dev>
Cc: Liam Howlett <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes (Oracle) <ljs@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michal Koutný <mkoutny@suse.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Muchun Song <songmuchun@bytedance.com>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Usama Arif <usamaarif642@gmail.com>
Cc: Vlastimil Babka <vbabka@kernel.org>
Cc: Wei Xu <weixugc@google.com>
Cc: Yosry Ahmed <yosry@kernel.org>
Cc: Yuanchu Xie <yuanchu@google.com>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Alistair Popple