| CVE |
Vendors |
Products |
Updated |
CVSS v3.1 |
| In the Linux kernel, the following vulnerability has been resolved:
mm/mglru: don't use stack pointers when walking PUD and PMD tables
MGLRU passes stack copies of page-table entries in walk_pud_range() and
walk_pmd_range() when walking to the next level. These copies can become
stale when khugepaged or MADV_DONTNEED (CONFIG_PT_RECLAIM) removes a PTE
table concurrently.
This pattern dates back to commit bd74fdaea146 ("mm: multi-gen LRU:
support page table walks"). Likely, earlier locking masked the latent
bug. Later changes exposed it: commit 1d65b771bc08 ("mm/khugepaged:
retract_page_tables() without mmap or vma lock") allowed concurrent
retraction, and commit cf4a65539c13 ("memcg-v1: no need for memcg locking
for MGLRU") dropped the walker's outer RCU protection.
The fix is simple, so just fix the original commit so we don't have to
speculate. To fix it, pass the actual PMD/PUD pointers. This restores the
RCU-protected lookup and validation against the live PMD. Note that the PUD
handling is required as well to handle folded page tables correctly in all
cases.
Fix generated by an LLM and sent by Josef Bacik in private.
[ david: handle walk_pud_range() as well, clarify+simplify description,
adjust Fixes: ] |
| In the Linux kernel, the following vulnerability has been resolved:
mm/huge_memory: fix truncate racing with PMD zapping
zap_huge_pmd_folio() removes the rmap before calling folio_mark_dirty().
If this was the last mapping, truncation can skip unmapping the folio
and proceed without waiting for the PMD lock, racing with dirtying.
ifs_set_range_dirty() can oops if MADV_DONTNEED races with
ftruncate() on XFS, when we get the folio set dirty after it was found
to not be mapped anymore.
Move folio_mark_dirty() and folio_mark_accessed() before rmap removal,
matching the PTE zap path. This keeps the folio mapped while updating
its state, so truncation must wait for the page table lock.
Fix generated by an LLM and sent by Chris Mason in private.
[ david: simplify comment and description ] |
| In the Linux kernel, the following vulnerability has been resolved:
mm/khugepaged: always recheck the pmd in try_collapse_pte_mapped_thp()
try_collapse_pte_mapped_thp() drops the PTE lock after it has cleared the
page table, and takes the PMD lock and the PTE lock to remove it.
In between, MADV_DONTNEED with CONFIG_PT_RECLAIM can reclaim the
page table. The pmd_same() check is supposed to handle this, but it is
only done with CONFIG_SPLIT_PTE_PTLOCKS.
Fix it by performing the check unconditionally.
Fix generated by an LLM and sent by Josef Bacik in private.
[ david: adjust description ] |
| Partially overwritten session files can lead to remote code execution.
The files session save handler, which is the default, writes updated session data into the existing session file without first clearing it. If a write is interrupted, the file can contain the beginning of the new session followed by the end of the old one, and bytes of an ordinary string can later be unserialized as objects. An attacker who can influence the session data, for example through session upload progress, and interrupt a write at a suitable point may inject objects, which can lead to code execution when the application contains a suitable gadget chain. Session data can also be corrupted.
This issue affects PHP versions before 8.2.36, 8.3.* before 8.3.37, 8.4.* before 8.4.28 and 8.5.* before 8.5.13. |
| @fastify/rate-limit is the rate-limiting plugin for the Fastify web framework. Starting in version 11.1.0, the plugin documented a manual flow for limiting failed login attempts that peeks at the limiter, runs the password check, and consumes a slot only on failure. Because the peek and the consume are two separate store operations with the password check in between, concurrent requests can all pass the peek before any slot is consumed, so in versions 11.1.0 through 11.2.2 a burst of requests runs far more failed attempts than the configured limit, and the sequential path also allowed one attempt more than configured. The default per-request limiter is not affected. The issue is fixed in @fastify/rate-limit 11.3.0, which makes the manual flow atomic by reserving a slot in a single operation and refunding it only on success, and users should upgrade to 11.3.0 or later. |
| memtomem-stm is a short-term memory proxy gateway with proactive memory surfacing for AI agents. From 0.1.2 until 0.1.34, concurrent surfacing RPC failures can start more than one reconnect for the same dirty session because concurrent LTM recovery is not uniquely owned by a session generation. Cancellation during a post-reconnect retry can also leave the replacement transport reusable after an interrupted protocol exchange, causing session-state integrity loss or local service degradation. The duplicate-reconnect subcase was introduced in 0.1.24, while the cancellation-recovery subcase affects the broader listed range. This issue is fixed in version 0.1.34. |
| Better Auth (better-auth, @better-auth/kysely-adapter, and @better-auth/drizzle-adapter) before 1.7.16 and 1.8.0-beta.0 before 1.8.0-beta.5 contains a race condition with MySQL or MariaDB serial IDs because the inserted row's ID is read via a separate LAST_INSERT_ID() query that may run on a different pooled connection. Under concurrent load, a user signing in can receive another user's session and act as that user. |
| better-auth from 0.8.1-beta.5 before 1.7.16 and 1.8.0-beta.0 before 1.8.0-beta.5, and @better-auth/passkey from 1.6.25 before 1.7.16 and 1.8.0-beta.0 before 1.8.0-beta.5, let revoked sessions stay valid because secondary storage session lists are updated non-atomically and some routes trust cookie-cached sessions. An attacker holding a valid session can evade bulk revocation or bans, or use a revoked cached session to register a passkey or create a lasting session. |
| Flyto2 Core is an execution kernel for automation and AI-agent workflows. Prior to 2.26.8, validate_url_ssrf in src/core/utils.py resolves and checks a hostname but returns the hostname URL rather than the validated address, and session.get in http.get performs an independent DNS lookup when connecting. A low-privileged caller with control of the hostname's DNS can make the first lookup return a public address and the connection-time lookup return an internal address, bypassing the SSRF guard for all modules that call it and exposing readable internal or metadata-service responses. Successful exploitation requires attacker-controlled DNS and a successful timing-dependent answer change between the two lookups. This issue is fixed in version 2.26.8. |
| In JetBrains TeamCity before 2026.2 privilege escalation of project-scoped access tokens via concurrent project deletion was possible |
| NVIDIA TensorRT-LLM for any platform contains a vulnerability where a user may cause a race condition by running concurrent requests. A successful exploit of this vulnerability may lead to data tampering. |
| FlowIntel's alerting module contained an authorization bypass in the alert access control logic. When an alert was linked to a private case, the system relied solely on connector ownership (user or organization) to determine access, without verifying that the requesting user still held valid membership in the associated private case.
This meant that a user who was a connector owner of an alert could continue to read alert details, review history, raw payloads, and action comments, and could perform review-status updates or case-creation actions, even after their access to the linked private case had been revoked or they were never a member of that case.
Additionally, a time-of-check to time-of-use (TOCTOU) gap existed: after a review-status update or a competing case-creation claim was committed, the response could disclose private case identifiers or review history to a user whose access had been concurrently removed by another session linking the alert to a private case.
Impact:
- Unauthorized disclosure of private case data (review comments, raw payloads, action history) through alert endpoints.
- Unauthorized modification of alert review status and creation of cases from alerts the user should no longer access.
- Leakage of a competing case identifier to an unauthorized user during concurrent case-creation races.
Affected versions: FlowIntel prior to 4.1.0. |
| Multiple protocol dissector crashes in 4.6.0 to 4.6.9 and 4.4.0 to 4.4.19 allows denial of service |
| The Okta .NET SDK Abstractions library systematically writes per-request authentication credentials to a shared HTTP client instance, rather than being isolated to individual outbound request objects. When using scoped child clients, a subsequent scoped request can overwrite the shared client configuration. Because the original parent client credential has not been restored, subsequent outbound requests may be transmitted using the newly written credential, which can result in subsequent actions being executed under the context of an incorrect security principal in the Okta Management API and improper attribution in audit logs. |
| The Okta .NET SDK Abstractions library improperly synchronizes authentication credentials set on a shared HTTP client during request execution. When concurrent requests are dispatched for different principals, one principal's credential may be transmitted on an outbound request intended for another principal, which can result in credential exposure and lead to unauthorized actions performed as the affected principal at the Okta Management API. |
| In Autonomy Logic OpenPLC Runtime versions 4.2.4 and prior, a race condition allows multiple concurrent requests to pass the existence check before any user is committed, resulting in several unauthenticated admin accounts being created. |
| In the Linux kernel, the following vulnerability has been resolved:
mm/gup: recheck the PUD under PTL in follow_pud_mask()
follow_pud_mask() reads the PUD without a lock. If it is a leaf, it takes
the lock and calls follow_huge_pud(), which reads the entry again, but
only tests whether it is present.
A leaf can be zapped concurrently and a page fault can fill the empty
entry with a page table.
This currently only applies when device-dax makes use of PUD entries,
and the fault after e.g., MADV_DONTNEED would find THPs to be disabled.
That requires a sequence like
1) Fault in the PUD
2) Disable THP
3) Have GUP race against concurrent MADV_DONTNEED + refault
Which is rather unlikely to happen in practice, which is why we never
received a report about such a problem.
Before commit 1b1676180246 ("mm/gup: handle huge pud for
follow_pud_mask()"), we performed that check in follow_devmap_pud() by
testing for pud_devmap().
Let's fix it by just structuring follow_pud_mask() similar to
follow_pmd_mask(), minus protnone stuff that does not apply and minus
FOLL_SPLIT_PMD handling. That makes it much easier to spot any such
inconsistencies.
Perform the pud_bad() check early in follow_pmd_mask() and just forward
the PUD value. We can drop the pud_present() check in follow_huge_pud(),
just like we have in follow_huge_pmd().
Even though we are performing more changes than strictly necessary, the
fix should be easy to backport. |
| In the Linux kernel, the following vulnerability has been resolved:
io_uring: defer foreign io-wq enqueues to task_work
An IOPOLL ring without DEFER_TASKRUN can be reaped by a task other than
the one that submitted the I/O. If a polled request is marked
REQ_F_REISSUE, io_free_batch_list() calls io_queue_iowq() with the
original tctx. Ordinary task exit can tear down tctx->io_wq after
draining only the tracked requests. The live request pins the task, but
not its io-wq, so the foreign task can race io_wq_put_and_exit() and
enqueue through a freed or NULL queue.
Forward foreign userspace enqueues to the owning task before accessing
tctx->io_wq. Use a dedicated task_work callback so reissues still go
straight to io-wq, and fail them if task_work is running in cancellation
mode. Preserve the existing PF_KTHREAD cancellation path. This also
covers the other callers of io_queue_iowq() without changing the
worker-owned enqueue path. |
| Dell Alienware Command Center (AWCC), versions prior to 6.14.105.0, contain a Time-of-check Time-of-use (TOCTOU) Race Condition vulnerability. A low privileged attacker with local access could potentially exploit this vulnerability, leading to Code Execution and Elevation of Privilege. |
| Use after free in Windows Win32K allows an authorized attacker to elevate privileges locally. |