{"id":"CVE-2026-98859","kind":"vulnerability","aliases":["CVE-2026-98859"],"related":[],"title":"Linux kernel: HID multitouch release timer can use freed memory on unbind","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nHID: multitouch: stop the release timer from being rearmed on remove\n\nmt_remove() quiesces the sticky-finger timer before stopping the\nhardware:\n\n    timer_delete_sync(&td->release_timer);\n\n    sysfs_remove_group(&hdev->dev.kobj, &mt_attribute_group);\n    hid_hw_stop(hdev);\n\ntimer_delete_sync() waits for a running callback and dequeues the timer,\nbut it does not stop the timer from being armed again. The transport is\nstill delivering reports at that point, and the report path rearms it:\n\n    if (app->quirks & MT_QUIRK_STICKY_FINGERS) {\n        if (td->mt_io_flags & MT_IO_SLOTS_MASK)\n            mod_timer(&td->release_timer,\n                      jiffies + msecs_to_jiffies(100));\n\nA report that arrives after timer_delete_sync() has returned therefore\nleaves the timer queued. td is allocated with devm_kzalloc() against\nhdev->dev, so it is freed when the driver is unbound, after mt_remove()\nreturns. When the timer fires afterwards, mt_expired_timeout()\ndereferences the freed td:\n\n    struct mt_device *td = timer_container_of(td, t, release_timer);\n    struct hid_device *hdev = td->hdev;\n\n    if (test_and_set_bit_lock(MT_IO_FLAGS_RUNNING, &td->mt_io_flags))\n\nSimply moving the teardown after hid_hw_stop() does not fix this on its\nown, because mt_expired_timeout() calls mt_release_contacts(), which\nwalks hdev->inputs; the timer still has to be quiesced before\nhid_hw_stop() tears the input devices down.\n\nUse timer_shutdown_sync() instead, which additionally makes any later\nmod_timer() a no-op, so neither ordering constraint has to be traded off\nagainst the other. This is the final-teardown pattern the function was\nintroduced for, and hid-wiimote already uses it for the same reason.","summary":"The Linux kernel HID multitouch driver can re-arm its sticky-finger release timer after removal, so the timer later reads memory already freed when the driver unbinds. The sources do not state what an attacker gains beyond a kernel memory use-after-free.","severity":"unknown","cvss":null,"epss":null,"exploited":false,"exploited_at":null,"products":[],"ecosystems":[],"tags":["hid","kernel","laptops","linux","multitouch","runtime","touchscreens"],"not_affected_if":["no-multitouch-hid-device"],"how_to_check":"- Check the kernel version with `uname -r` and whether the `hid_multitouch` module is loaded with `lsmod | grep hid_multitouch`\n- unknown for whether the sources list affected kernel version ranges","fix":"- Update to a kernel release that includes the upstream commit titled 'HID: multitouch: stop the release timer from being rearmed on remove' (version unknown from sources)\n- Until patched, unknown whether unloading or unbinding the device driver can be avoided","status":"open","published_at":"2026-10-18T16:17:40+00:00","first_seen_at":"2026-10-23T21:57:17+00:00","triaged_at":"2026-10-25T05:12:56+00:00","updated_at":"2026-10-25T05:12:56+00:00","category":"runtime","confidence":0.5,"triage":{"at":"2026-10-25T05:12:56Z","model":"claude-haiku-5-5","output":{"fix":"- Update to a kernel release that includes the upstream commit titled 'HID: multitouch: stop the release timer from being rearmed on remove' (version unknown from sources)\n- Until patched, unknown whether unloading or unbinding the device driver can be avoided","reach":"local","title":"Linux kernel: HID multitouch release timer can use freed memory on unbind","impact":"high","summary":"The Linux kernel HID multitouch driver can re-arm its sticky-finger release timer after removal, so the timer later reads memory already freed when the driver unbinds. The sources do not state what an attacker gains beyond a kernel memory use-after-free.","category":"runtime","nicknames":[],"who_cares":["linux","kernel","hid","multitouch","laptops","touchscreens"],"confidence":0.5,"needs_auth":"unknown","worst_case":"kernel memory use-after-free may crash the system or run code","analyst_note":"Find the affected kernel version ranges and fixed commit in the NVD entry or the upstream kernel git history for the HID multitouch driver.","how_to_check":"- Check the kernel version with `uname -r` and whether the `hid_multitouch` module is loaded with `lsmod | grep hid_multitouch`\n- unknown for whether the sources list affected kernel version ranges","needs_analyst":true,"not_affected_if":["no-multitouch-hid-device"],"who_is_affected":"Systems running a Linux kernel with the HID multitouch driver bound to a touch device, such as touchscreens and touchpads, where the device keeps sending reports during driver removal. Machines without multitouch HID devices, or without the module loaded, are unlikely to be exposed. Whether an attacker can trigger the timing from outside the machine is unknown from the sources.","user_interaction":"unknown","needs_special_setup":"unknown","severity_if_unknown":"unknown"},"cost_usd":0.000716,"input_tokens":3912,"output_tokens":649,"severity_before":"unknown"},"analyst_status":null,"analyst_note":null,"search":"'-2026':14A,17A '-98859':15A,18A '100':187C 'addit':317C 'afterward':229C 'alloc':205C 'alreadi':50B,355C 'app':163C 'arm':35B,145C 'arriv':191C 'attack':63B 'attribut':118C 'beyond':65B 'bit':258C 'call':287C 'callback':131C 'constraint':330C 'contact':290C 'contain':242C 'cve':13A,16A 'delet':107C,125C,194C 'deliv':151C 'dequeu':133C 'derefer':233C 'dev':211C 'dev.kobj':116C 'devic':239C,250C,309C 'devm':207C 'driver':31B,54B,218C 'expir':231C,285C 'final':343C 'final-teardown':342C 'finger':39B,100C,168C 'fire':228C 'fix':278C 'flag':173C,262C,267C 'follow':78C 'free':72B 'freed':9A,51B,215C,235C 'function':347C 'gain':64B 'group':114C,119C 'hardwar':105C 'hdev':115C,123C,210C,251C,253C,293C 'hid':3A,19B,29B,83C,120C,249C,273C,303C,353C 'hid-wiimot':352C 'hw':121C,274C,304C 'input':294C,308C 'instead':315C 'introduc':349C 'io':172C,175C,261C,266C 'jiffi':183C,186C 'kernel':2A,20B,28B,67B,76C 'kzalloc':208C 'laptop':21B 'later':47B,320C 'leav':199C 'linux':1A,22B,27B,75C 'lock':259C 'make':318C 'mask':177C 'memori':10A,49B,68B 'mod':178C,321C 'move':269C 'msec':184C 'mt':94C,117C,165C,171C,174C,222C,230C,238C,260C,265C,284C,288C 'multitouch':4A,23B,30B,84C 'neither':328C 'no-op':324C 'op':326C 'order':329C 'path':159C 'pattern':345C 'point':155C 'queu':202C 'quiesc':96C,301C 'quirk':164C,166C 're':34B 're-arm':33B 'read':48B 'rearm':91C,160C 'reason':361C 'releas':5A,40B,87C,110C,181C,246C,289C 'remov':43B,93C,95C,113C,223C 'report':152C,158C,189C 'resolv':82C 'return':197C,224C 'run':130C,263C 'runtim':24B 'set':257C 'shutdown':313C 'simpli':268C 'slot':176C 'sourc':57B 'state':60B 'sticki':38B,99C,167C 'sticky-fing':37B,98C 'still':150C,297C 'stop':85C,103C,122C,140C,275C,305C 'struct':237C,248C 'sync':108C,126C,195C,314C 'sysf':112C 'td':109C,170C,180C,203C,236C,240C,244C,252C,264C 'tear':306C 'teardown':271C,344C 'test':255C 'therefor':198C 'timeout':232C,286C 'timer':6A,41B,46B,88C,101C,106C,111C,124C,135C,142C,179C,182C,193C,201C,227C,241C,247C,296C,312C,322C 'touchscreen':25B 'trade':334C 'transport':148C 'unbind':12A,55B 'unbound':220C 'use':8A,70B,311C,356C 'use-after-fre':69B 'vulner':79C 'wait':127C 'walk':292C 'wiimot':354C","nicknames":[],"mitigations":[],"who_affected":"Systems running a Linux kernel with the HID multitouch driver bound to a touch device, such as touchscreens and touchpads, where the device keeps sending reports during driver removal. Machines without multitouch HID devices, or without the module loaded, are unlikely to be exposed. Whether an attacker can trigger the timing from outside the machine is unknown from the sources.","tier":"C","verified":{},"analyst_revisit_at":null,"first_public_at":"2026-10-18T16:17:40+00:00","kev":null,"cvss_vector":null,"cvss_by":null,"poc_count":0,"sources":[{"source":"nvd","external_id":"CVE-2026-98859","url":"https://nvd.nist.gov/vuln/detail/CVE-2026-98859","severity":null,"published_at":"2026-10-18T16:17:40+00:00","modified_at":"2026-10-21T13:21:50+00:00","first_seen_at":"2026-10-23T21:57:17+00:00","fetched_at":"2026-10-23T22:28:56+00:00","note":null}],"documents":[],"notice":"Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai"}