CVE-2026-89987
Low risk· EPSS 11%Exploitation Probability (EPSS)
Low risk11th percentile - higher than 11% of all known CVEs
Summary
A vulnerability in the Linux kernel's handling of huge pages for tmpfs was found. During unmapping (zap), the dirty bit from the PMD entry is not transferred to the folio, leading to loss of write information. As a result, data in MAP_SHARED tmpfs segments can be silently lost during unmap if the page is PMD-mapped and swap is enabled.
Risk Assessment
The organization may experience silent data loss in shared memory (e.g., /dev/shm) used by processes, potentially causing application crashes or data corruption. The issue only occurs when swap is enabled and THP for tmpfs is enabled.
Recommendation
Apply the kernel patch that transfers the dirty bit from PMD to folio during zap operations immediately. Until updated, consider disabling THP for tmpfs (e.g., mount with huge=never) or disabling swap to reduce risk.
Other vulnerabilities in Linux kernel
See all- CVE-2026-99259Unknown
In the Linux kernel, the soc/tegra: pmc driver has an uninitialised early_pmc->rate variable for legacy 32-bit Tegra devices. The variable is never configured and remains 0, triggering a warning in tegra_pmc_enter_suspend_mode(). The issue appeared after the PMC driver was changed to create the PMC context dynamically, updating rate only in the new context.
- CVE-2026-99258Unknown
In the Linux kernel, hv_apicid_to_vp_index() uses per-CPU hypercall input and output pages. In an SEV-SNP CoCo VM without a paravisor the output page is not allocated, so the hypervisor stores the resulting VP index into memory not allocated by the guest. The fix uses the input page for both input and output.
- CVE-2026-99257Unknown
In the Linux kernel, the shared GPIO mechanism based on platform lookup tables requires an associated struct device. An fwnode-based lookup with no associated device currently causes a NULL-pointer dereference. The fix adds a check that lets the kernel gracefully handle this situation with an error.
- CVE-2026-99256Unknown
In the Linux kernel, the dm cache module used a rw_semaphore for background_work_lock, although write locks on rw_semaphores have strict owner semantics. The locking process was not always the same as the unlocking one, triggering a warning with CONFIG_DEBUG_RWSEMS during suspend/resume of a cache device. The fix replaces the rw_semaphore with a spinlock and a wait queue.
- CVE-2026-99251Unknown
In the Linux kernel, init_vp_index() contains a retry loop that strips CPUs from a node if they are already in the allocated cpumask or outside the HK_TYPE_MANAGED_IRQ housekeeping cpumask. If the housekeeping cpumask contains no CPU in that node, the loop becomes infinite. The fix skips NUMA nodes with no housekeeping CPU and aborts early when no online housekeeping CPUs exist.
- CVE-2026-99250Unknown
In the Linux kernel, the NBCON console printing kthreads are started very early during boot via early_initcall(). Asynchronously enabling/disabling interrupts during early boot may interfere with early SMP setup on some hardware platforms, causing hangs. The fix postpones kthread startup until after all arch initcalls have completed.
- CVE-2026-99249Unknown
In the Linux kernel, zap_pid_ns_processes() waiting for PID namespace cleanup was not freezable. A tracer outside the namespace that does not reap its traced zombies keeps PIDs allocated, making the wait unbounded and causing every suspend attempt to fail. The fix makes the wait loops use TASK_FREEZABLE states so the freezer can freeze waiting tasks.
- CVE-2026-99248Unknown
In the Linux kernel, the i2c: xiic driver was changed to restore non-managed runtime PM instead of the devres-based approach to fix a clk WARN flood. The managed release tore down runtime PM in the wrong order, causing the clock to be disabled twice and triggering "clkN already disabled" warnings on every teardown.
- CVE-2026-99247Unknown
In the Linux kernel, if trace buffer allocation fails during early boot, tracing_disabled remains set and the global trace array events list stays uninitialized. tracer_init_tracefs_work_func() runs asynchronously without checking this state and calls event_trace_init() and init_tracer_tracefs(), leading to a NULL pointer dereference Oops during boot. The fix checks tracing_disabled at the start of the work function.
- CVE-2026-99215Unknown
In the Linux kernel power supply subsystem, a use-after-free vulnerability exists. If a power supply is unregistered while its external_power_changed() callback is still running, the driver may free data that the callback is still accessing, leading to a kernel crash (NULL pointer dereference).
Original NVD description (English source)
In the Linux kernel, the following vulnerability has been resolved: mm/huge_memory: transfer the pmd dirty bit to the folio on zap zap_huge_pmd_folio() propagates the pmd young bit to the folio for the file case, but not the dirty bit. The pte path does propagate it, in zap_present_folio_ptes() and so does the pmd split path, in __split_huge_pmd_locked(). For most file mappings the omission is harmless, because writing to a shared file mapping goes through page_mkwrite(), which dirties the folio. tmpfs is different: it has no page_mkwrite(), and vma_wants_writenotify() is false for it, so a *read* fault on a MAP_SHARED tmpfs mapping installs a writable pmd via do_read_fault(). do_read_fault() does not call fault_dirty_shared_page(), so subsequent stores through that mapping set only the hardware dirty bit in the pmd and never call folio_mark_dirty(). A shmem folio allocated by a fault is marked uptodate but not dirty (see the clear: block in shmem_get_folio_gfp()), so PG_dirty is never set at all. Unmapping such a folio - munmap(), or exit_mmap() when the process dies - then loses the only record that it was written, because zap_huge_pmd() drops the pmd without transferring the dirty bit. Reclaim afterwards sees a clean shmem folio: the whole swap-out block in shrink_folio_list() is inside "if (folio_test_dirty(folio))", so pageout() is skipped and the folio falls into __remove_mapping(). There, folio_is_file_lru() is false for a swapbacked folio, so no shadow entry is created and __filemap_remove_folio(folio, NULL) simply empties the i_pages slot. The data is freed without ever being written to swap, and the next fault on that index returns a freshly zeroed folio. This is silent data loss for any process that keeps state in a MAP_SHARED tmpfs segment across an unmap - for example a cache handed from one process generation to the next through /dev/shm. It requires the folio to be PMD-mapped, so it only shows up once shmem THP is enabled (which is what we did in Meta fleet and started noticing crashes); with THP off the pte path transfers the dirty bit correctly. It also only becomes visible when swap is enabled, because with no swap device shmem folios (which are on the anon LRU) are not scanned by reclaim at all, so the clean folio is never dropped. Reproduced on x86_64 with a tmpfs mounted huge=within_size: read-fault a 2MB-backed region, write a known pattern through the resulting mapping, munmap, force reclaim of the cgroup, then re-map and read back. Without this patch the region reads back as zeros and vmstat shows zswpout 0 - the data was discarded rather than swapped. With this patch the region reads back correctly and the pages are swapped out as expected. With huge=never, or when the first touch is a write, the test passes either way.
Vulnerability data from NVD (NIST) · CISA KEV · EPSS