{"id":"CVE-2026-99229","kind":"vulnerability","aliases":["CVE-2026-99229"],"related":[],"title":"Linux kernel: io_uring wait loop can spin forever and ignore kill signals","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nio_uring/wait: don't spin on a stuck overflow list\n\nAs long as IO_CHECK_CQ_OVERFLOW_BIT is set, io_cqring_wait_schedule()\nreturns right away, before it gets to look at signals or arms the timeout,\nfor io_cqring_wait() to flush the overflow list. If that doesn't empty the\nlist, and the CQ ring doesn't hold the number of events the caller wants,\nthe loop starts over, with the same result. The task spins in the kernel,\nand can't be killed.\n\nThe assumption is that this can't happen: if CQEs can't be flushed, the\nring is full, and min_events is capped to the size of the ring. But:\n\n- With IORING_SETUP_CQE_MIXED a 32b CQE needs two contiguous entries. It\n  is stuck on the list if a single one is free, and then the ring holds\n  less than cq_entries events. A single threaded application that asks\n  for as many events as the ring has entries never gets out of\n  io_uring_enter().\n\n- Whether the ring is full is up to the CQ head, which the application\n  can write to. While one thread waits for events on an empty ring,\n  another one can move the head back so that the ring looks full and\n  have a CQE overflow. The waiter can't be stopped any more, not even by\n  SIGKILL.\n\nOnly the application can make room in the CQ ring. Return to it if the\nlist could not be flushed.","summary":"A bug in the Linux kernel's io_uring wait path can leave a task spinning in kernel mode when completion-queue overflow entries cannot be flushed. The stuck task cannot be killed, even by SIGKILL, which lets a local or application-level actor tie up a process indefinitely.","severity":"unknown","cvss":null,"epss":null,"exploited":false,"exploited_at":null,"products":[],"ecosystems":[],"tags":["io_uring","linux-kernel","runtime","self-hosted"],"not_affected_if":["io_uring-disabled","no-cqe-mixed"],"how_to_check":"- Check the kernel version with `uname -r` and compare it against the fixed release in the kernel stable tree referenced by NVD.\n- Find processes that use io_uring with `IORING_SETUP_CQE_MIXED` or that spin in kernel mode (state `R` with high CPU in `top`).","fix":null,"status":"open","published_at":"2026-11-02T16:17:47+00:00","first_seen_at":"2026-11-02T16:24:06+00:00","triaged_at":"2026-11-02T16:26:54+00:00","updated_at":"2026-11-02T16:26:54+00:00","category":"runtime","confidence":0.5,"triage":{"at":"2026-11-02T16:26:54Z","model":"claude-haiku-5-5","output":{"fix":"unknown","reach":"local","title":"Linux kernel: io_uring wait loop can spin forever and ignore kill signals","impact":"low","summary":"A bug in the Linux kernel's io_uring wait path can leave a task spinning in kernel mode when completion-queue overflow entries cannot be flushed. The stuck task cannot be killed, even by SIGKILL, which lets a local or application-level actor tie up a process indefinitely.","audience":"businesses","category":"runtime","nicknames":[],"who_cares":["linux-kernel","io_uring","self-hosted"],"confidence":0.5,"fix_status":"unknown","needs_auth":"unknown","worst_case":"A process becomes unkillable and spins a CPU core in the kernel","analyst_note":"Confirm affected kernel version ranges and the fixed commit from the NVD entry or the kernel stable tree.","how_to_check":"- Check the kernel version with `uname -r` and compare it against the fixed release in the kernel stable tree referenced by NVD.\n- Find processes that use io_uring with `IORING_SETUP_CQE_MIXED` or that spin in kernel mode (state `R` with high CPU in `top`).\n- unknown for the exact affected version ranges, since the sources given do not state them.","plain_action":"","needs_analyst":true,"not_affected_if":["io_uring-disabled","no-cqe-mixed"],"who_is_affected":"Systems running a Linux kernel with io_uring enabled, especially workloads that use IORING_SETUP_CQE_MIXED or let one thread write the completion-queue head while another waits. The sources do not list the affected kernel version ranges. Hosts that do not use io_uring are not exposed.","user_interaction":"no","needs_special_setup":"yes","severity_if_unknown":"medium"},"cost_usd":0.000766,"input_tokens":4505,"output_tokens":630,"severity_before":"unknown"},"analyst_status":null,"analyst_note":null,"search":"'-2026':15A,18A '-99229':16A,19A '32b':213C 'actor':74B 'anoth':290C 'applic':72B,244C,276C,322C 'application-level':71B 'arm':125C 'ask':246C 'assumpt':178C 'away':116C 'back':296C 'bit':107C 'bug':30B 'caller':156C 'cannot':54B,60B 'cap':199C 'check':104C 'complet':50B 'completion-queu':49B 'contigu':217C 'could':336C 'cq':105C,146C,238C,272C,328C 'cqe':210C,214C,306C 'cqes':186C 'cqring':111C,130C 'cve':14A,17A 'doesn':139C,148C 'empti':141C,288C 'enter':262C 'entri':53B,218C,239C,255C 'even':63B,317C 'event':154C,197C,240C,250C,285C 'flush':56B,133C,190C,339C 'follow':85C 'forev':9A 'free':230C 'full':194C,267C,302C 'get':119C,257C 'happen':184C 'head':273C,295C 'hold':150C,235C 'host':28B 'ignor':11A 'indefinit':79B 'io':3A,20B,36B,90C,103C,110C,129C,260C 'ior':208C 'kernel':2A,24B,34B,46B,83C,171C 'kill':12A,62B,176C 'leav':41B 'less':236C 'let':67B 'level':73B 'linux':1A,23B,33B,82C 'linux-kernel':22B 'list':99C,136C,143C,224C,335C 'local':69B 'long':101C 'look':121C,301C 'loop':6A,159C 'make':324C 'mani':249C 'min':196C 'mix':211C 'mode':47B 'move':293C 'need':215C 'never':256C 'number':152C 'one':228C,281C,291C 'overflow':52B,98C,106C,135C,307C 'path':39B 'process':78B 'queue':51B 'resolv':89C 'result':165C 'return':114C,330C 'right':115C 'ring':147C,192C,205C,234C,253C,265C,289C,300C,329C 'room':325C 'runtim':25B 'schedul':113C 'self':27B 'self-host':26B 'set':109C 'setup':209C 'sigkil':65B,319C 'signal':13A,123C 'singl':227C,242C 'size':202C 'spin':8A,44B,94C,168C 'start':160C 'stop':313C 'stuck':58B,97C,221C 'task':43B,59B,167C 'thread':243C,282C 'tie':75B 'timeout':127C 'two':216C 'ure':4A,21B,37B,261C 'uring/wait':91C 'vulner':86C 'wait':5A,38B,112C,131C,283C 'waiter':309C 'want':157C 'whether':263C 'write':278C","nicknames":[],"mitigations":[],"who_affected":"Systems running a Linux kernel with io_uring enabled, especially workloads that use IORING_SETUP_CQE_MIXED or let one thread write the completion-queue head while another waits. The sources do not list the affected kernel version ranges. Hosts that do not use io_uring are not exposed.","tier":"C","verified":{},"analyst_revisit_at":null,"fix_status":"unknown","audience":"businesses","plain_action":null,"pkeys":[],"ptokens":[],"labels":["iouring","linuxkernel","runtime","selfhosted","unknown"],"first_public_at":"2026-11-02T16:17:47+00:00","kev":null,"cvss_vector":null,"cvss_by":null,"poc_count":0,"sources":[{"source":"nvd","external_id":"CVE-2026-99229","url":"https://nvd.nist.gov/vuln/detail/CVE-2026-99229","severity":null,"published_at":"2026-11-02T16:17:47+00:00","modified_at":"2026-11-02T16:17:47+00:00","first_seen_at":"2026-11-02T16:24:06+00:00","fetched_at":"2026-11-02T16:24:06+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"}