{"id":"AZL-115103","summary":"CVE-2026-99561 affecting package kernel 6.6.161.1-1","details":"In the Linux kernel, the following vulnerability has been resolved:\n\nfork: don't create io threads once PF_POSTCOREDUMP is set\n\nzap_process() skips every thread that already has PF_POSTCOREDUMP set.\nSuch a thread is past synchronize_group_exit(), so a coredump can't\ncatch it anymore. It isn't counted in core_state-\u003ethreads_remaining and\nit isn't sent SIGKILL.\n\ndo_exit() sets PF_POSTCOREDUMP in synchronize_group_exit() and calls\nio_uring_files_cancel() right after that. The cancellation runs task\nwork and a create_worker_cb() that io-wq queued before the exit creates\na new io-wq worker from there. If another thread started a coredump in\nthe meantime that worker joins a thread-group which is already being\ndumped. zap_process() never saw it so it was never counted. But when it\nexits it sees signal-\u003ecore_state in synchronize_group_exit() and\ndecrements threads_remaining like any other thread.\n\nSo the count reaches zero one thread early and the dumper leaves\ncoredump_wait_inactive() while a thread it counted is still running.\n\nThe task has no fatal signal pending because zap_process() deliberately\ndidn't send it one, and it never went through get_signal() so it doesn't\nhave PF_SIGNALED either.\n\nRefuse to create an io thread when the creator has PF_POSTCOREDUMP.\nThat costs nothing. io_uring_files_cancel() raises IO_WQ_BIT_EXIT\nbefore it runs any task work, so a worker created from there only ever\ngets to exit again, and io_should_retry_thread() doesn't retry -EINTR.\n\nClearing PF_POSTCOREDUMP for the new thread in copy_process() was the\nother option. It makes the worker a thread like any other, but it only\nhelps while the dump hasn't started. A worker born after zap_process()\nhas run is invisible to it whatever its flags say and still decrements\nthreads_remaining on the way out. Refusing to create it covers both, and\nthen no task is ever born with the flag, so there is nothing to clear.\n\nMoving io_uring_files_cancel() ahead of synchronize_group_exit() closes\nthe same window. It runs the cancellation, and whatever that can block\non, before the thread announces itself to the dumper. A thread that\nblocks there never reaches coredump_task_exit() at all, so the dumper\nends up waiting for a thread that never parks. Refusing the creation\nleaves the cancellation where it is. But it's very ugly to run io_uring\nwork even before we did all the generic exit work.","modified":"2026-11-20T14:17:30.885787299Z","published":"2026-11-19T15:18:12Z","upstream":["CVE-2026-99561"],"references":[{"type":"WEB","url":"https://nvd.nist.gov/vuln/detail/CVE-2026-99561"}],"affected":[{"package":{"name":"kernel","ecosystem":"Azure Linux:3","purl":"pkg:rpm/azure-linux/kernel"},"ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"0"},{"last_affected":"6.6.161.1-1"}]}],"database_specific":{"source":"https://github.com/microsoft/AzureLinuxVulnerabilityData/blob/main/osv/AZL-115103.json"}}],"schema_version":"1.9.2"}