{"id":"CVE-2026-89978","published":"2026-09-16T11:17:08.667","lastModified":"2026-09-16T11:17:08.667","description":"In the Linux kernel, the following vulnerability has been resolved:\n\naccel/amdxdna: return early from a zero-length flush\n\nSYNC_BO does not constrain its size, so a request for zero bytes reaches\ndrm_clflush_virt_range(), which ends with an unconditional\nclflushopt(end - 1). For an empty range that is the byte before the\nmapping, and abo->mem.kva comes from vmap(), so the access lands in the\nguard page below the vmalloc area and faults:\n\n  BUG: unable to handle page fault for address: ffffd16fbbc70fff\n  #PF: supervisor read access in kernel mode\n  Oops: Oops: 0000 [#1] SMP NOPTI\n  CPU: 7 UID: 1000 Comm: sync_bo_probe\n  RIP: 0010:drm_clflush_virt_range+0x3c/0x70\n  Call Trace:\n   amdxdna_drm_sync_bo_ioctl+0x124/0x430 [amdxdna]\n   drm_ioctl+0x301/0x4c0\n   __x64_sys_ioctl+0x115/0x2f0\n   do_syscall_64+0xa6/0x3d0\n\nAny process that can open the render node can do this. Reproduced 3 of 3\ntimes on a Strix Point NPU (1022:17f0), by calling SYNC_BO with size 0 on\nan AMDXDNA_BO_SHARE object. The import arm takes the same request but\nflushes the whole scatterlist, so it survives it.\n\nNothing needs flushing for an empty range, so answer before choosing a\npath.","cvssScore":null,"cvssSeverity":null,"cvssVector":null,"cwes":[],"vendors":[],"products":[],"references":[{"url":"https://git.kernel.org/stable/c/45962da5821d0a691f638ccb13842889156d8969","tags":[]},{"url":"https://git.kernel.org/stable/c/dc14753664240cedf669623b27ae9922b0618b25","tags":[]},{"url":"https://git.kernel.org/stable/c/f00def884a831dc49eb659dac2dd09dbfd21e67f","tags":[]}],"exploitRefs":[],"hasPoc":false,"ai":null}