{"id":"CVE-2026-90247","published":"2026-09-17T17:17:20.803","lastModified":"2026-09-17T17:17:20.803","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nbpf: Fix mmap_lock leak in irq_work path\n\nstack_map_get_build_id_offset() introduced a per-CPU irq_work to defer\nmmap_read_unlock() from NMI context, and bpf_find_vma() later reused the\nsame mmap_unlock_work. Both callers only check whether the work is busy\nbefore taking mmap_lock, so a nested caller can reuse the slot before the\nfirst caller queues it. Two read locks may then be acquired while only one\ndeferred unlock runs, leaking a read lock and blocking exit_mmap().\n\nReserve the per-CPU slot before mmap_read_trylock(). Use the same wrapper\nin stackmap and bpf_find_vma() so both callers release the reservation on\ntrylock failure. Keep rejecting the slot while the irq_work remains busy.\nRelease it after the irq_work callback unlocks the mm.","cvssScore":null,"cvssSeverity":null,"cvssVector":null,"cwes":[],"vendors":[],"products":[],"references":[{"url":"https://git.kernel.org/stable/c/051f2da26ce377f683b938a3382e0f16d02f3139","tags":[]},{"url":"https://git.kernel.org/stable/c/a052ad5edccf5319f50ed955de4368f8318a9f20","tags":[]},{"url":"https://git.kernel.org/stable/c/fa9dcacdcdf487f0ffef64bf67622f1caed509f1","tags":[]}],"exploitRefs":[],"hasPoc":false,"ai":null}