PT-2026-43892 · Linux · Linux
Sashiko
·
Published
2026-05-27
·
Updated
2026-05-27
·
CVE-2026-46025
None
No severity ratings or metrics are available. When they are, we'll update the corresponding info on the page.
In the Linux kernel, the following vulnerability has been resolved:
mm/damon/core: fix damon call() vs kdamond fn() exit race
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].
Found an issue in the description? Have something to add? Feel free to write us 👾
Related Identifiers
Affected Products
Linux