Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB
CVE-2026-93253 Overview
CVE-2026-93253 is a Linux kernel issue in the sched/isolation subsystem. The flaw triggers a warning when the kernel frees cpumask memblock memory before the memory map finishes deferred initialization. The condition is reached during boot when the nohz_full kernel parameter is set, causing housekeeping_init() to call memblock_phys_free() too early.
The upstream fix defers the free operation to an initcall by queuing freed memory on a lockless list and processing it later via a new housekeeping_late_init() helper. The issue is boot-time only and does not provide a known path to remote compromise.
Critical Impact
The vulnerability produces a kernel warning during early boot on systems using CPU isolation, with potential downstream memory-handling side effects if the fix is absent.
Affected Products
- Linux kernel builds containing the deferred memory map initialization code and the sched/isolation housekeeping logic
- Systems booted with the nohz_full kernel parameter
- Kernel trees that include commit 59bd1d914bb5 ("memblock: warn when freeing reserved memory before memory map is initialized")
Discovery Timeline
- 2026-09-24 - CVE-2026-93253 published to NVD
- 2026-09-24 - Last updated in NVD database
Technical Details for CVE-2026-93253
Vulnerability Analysis
The Linux kernel scheduler isolation code allocates cpumask structures from memblock during early boot. When CPU isolation is requested through nohz_full, housekeeping_init() later releases that memory via memblock_phys_free(). The release happens before deferred_init_memmap() finishes populating the memory map, which triggers a warning at mm/memblock.c:904 inside __free_reserved_area().
The call chain reproduces as start_kernel() → housekeeping_init() → memblock_phys_free() → warning. Freeing reserved memory before the memory map is initialized can leave page metadata in an inconsistent state, and prior reports on the same code path included a KASAN use-after-free signature.
Root Cause
The root cause is an ordering issue between memblock-based allocation and the deferred memory map initialization. The housekeeping subsystem releases cpumask memory synchronously inside its init routine, which runs before the memory map is ready to accept frees.
Attack Vector
This is a local, boot-time condition triggered by kernel configuration. An attacker cannot reach the code path remotely. Triggering the warning requires control of boot parameters, specifically setting nohz_full on a kernel that includes the memblock warning commit without the deferred-free fix.
The vulnerability manifests only during kernel initialization. See the Linux Kernel Commit Details and Linux Kernel Commit Update for the full patch content.
Detection Methods for CVE-2026-93253
Indicators of Compromise
- Kernel log entry: Cannot free reserved memory because of deferred initialization of the memory map
- Warning at mm/memblock.c:904 in __free_reserved_area during boot
- Call trace including memblock_phys_free and housekeeping_init on CPU0 (swapper/0/0)
Detection Strategies
- Parse dmesg and persistent boot logs for the memblock reserved-memory warning string
- Correlate the warning with the presence of nohz_full in /proc/cmdline
- Audit kernel versions deployed in isolation-sensitive workloads to confirm the fix commit is included
Monitoring Recommendations
- Forward boot-time kernel logs from Linux fleets into a centralized log store for pattern matching
- Alert on any WARNING: CPU: entries referencing __free_reserved_area or memblock_phys_free
- Track kernel build hashes against the fixed commits 2b58c749b8c5 and 811fdac2d1bd
How to Mitigate CVE-2026-93253
Immediate Actions Required
- Update to a Linux kernel release that includes the deferred housekeeping_late_init() free logic
- Confirm the dependent memblock fix commit 7c2eee9c1367 is also present to avoid the related KASAN report
- Rebuild and reboot any affected hosts that boot with nohz_full
Patch Information
The fix is distributed as two upstream commits. See the Linux Kernel Commit Details and the Linux Kernel Commit Update. The change saves housekeeping cpumask memblock memory to a llist free list in housekeeping_init() and releases it from a new housekeeping_late_init() initcall.
Workarounds
- Remove the nohz_full kernel boot parameter if CPU isolation is not required
- Pin affected hosts to a prior kernel that does not include the warning commit 59bd1d914bb5 until the fix is deployed
- Rebuild the kernel with the two referenced commits backported if a vendor update is not yet available
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB

