{"id":"CVE-2026-93208","title":"In the Linux kernel, the following vulnerability has been resolved:\n\nkasan: fix cache shrink race with CPU hotplug\n\nkasan_quarantine_remove_cache() first invokes per_cpu_remove_cache() on\nall online CPUs","summary":"In the Linux kernel, the following vulnerability has been resolved:\n\nkasan: fix cache shrink race with CPU hotplug\n\nkasan_quarantine_remove_cache() first invokes per_cpu_remove_cache() on\nall online CPUs.  Each callback moves objects bel…","severity":"none","vendor":"Linux","product":"Linux","affected":["Linux >= 07d067e4f2ceb72b9f681995cc53828caaba9e6e < 7cd164f0e1bab0f8dd125c92ed4f19bbd6b92a4a","Linux >= 07d067e4f2ceb72b9f681995cc53828caaba9e6e < 709c3646545e0a1f99a5816384c633f6552c5a98","Linux >= 07d067e4f2ceb72b9f681995cc53828caaba9e6e < 30e8cb8598aa41b1b9f8803081d2ae5e5369c0f3","Linux >= 07d067e4f2ceb72b9f681995cc53828caaba9e6e < 3119d58e4ef8719d669911d38a53fc00086ac48b","Linux >= 07d067e4f2ceb72b9f681995cc53828caaba9e6e < 8790303cbaac52a11dfed4aab261f8ea60682525","Linux 5.19"],"published":"2026-09-24","updated":"2026-09-24","sourceUpdated":"2026-09-24T16:17:15.497","source":"NVD","sourceUrl":"https://nvd.nist.gov/vuln/detail/CVE-2026-93208","references":[{"url":"https://git.kernel.org/stable/c/30e8cb8598aa41b1b9f8803081d2ae5e5369c0f3","label":"416baaa9-dc9f-4396-8d5f-8c081fb06d67"},{"url":"https://git.kernel.org/stable/c/3119d58e4ef8719d669911d38a53fc00086ac48b","label":"416baaa9-dc9f-4396-8d5f-8c081fb06d67"},{"url":"https://git.kernel.org/stable/c/709c3646545e0a1f99a5816384c633f6552c5a98","label":"416baaa9-dc9f-4396-8d5f-8c081fb06d67"},{"url":"https://git.kernel.org/stable/c/7cd164f0e1bab0f8dd125c92ed4f19bbd6b92a4a","label":"416baaa9-dc9f-4396-8d5f-8c081fb06d67"},{"url":"https://git.kernel.org/stable/c/8790303cbaac52a11dfed4aab261f8ea60682525","label":"416baaa9-dc9f-4396-8d5f-8c081fb06d67"}],"tags":["nvd","cve.org"],"ingestedAt":"2026-09-24T15:45:56.672Z","slug":"CVE-2026-93208","body":"## Overview\n\nIn the Linux kernel, the following vulnerability has been resolved:\n\nkasan: fix cache shrink race with CPU hotplug\n\nkasan_quarantine_remove_cache() first invokes per_cpu_remove_cache() on\nall online CPUs.  Each callback moves objects belonging to the cache from\ncpu_quarantine to the CPU's shrink_qlist, where they can later be freed\nfrom task context.\n\nkmem_cache_destroy() invokes the quarantine removal path while holding\ncpus_read_lock(), but kmem_cache_shrink() does not.  The latter can\ntherefore race with CPU offlining as follows:\n\n  kmem_cache_shrink()             CPU hotplug\n  -------------------             -----------\n  on_each_cpu()\n    CPU1 moves objects to\n    CPU1's shrink_qlist\n  on_each_cpu() returns\n                                  CPU1 goes offline\n                                  kasan_cpu_offline()\n                                    drains cpu_quarantine\n                                    leaves shrink_qlist untouched\n  for_each_online_cpu()\n    skips CPU1\n\nThe objects left on CPU1's shrink_qlist are not returned to the slab\nallocator.  This may prevent kmem_cache_shrink() from releasing slabs that\nwould otherwise become empty.  If CPU1 remains offline, a later\nkmem_cache_destroy() also skips the list and can report that the cache\nstill contains objects.\n\nAn intermittent occurrence was observed with a virtio-9p filesystem.  The\nmount and umount commands both returned 0, but the kernel logged the\nfollowing during the userspace-triggered teardown:\n\n  [  2994.380134][  T111] BUG 9p-fcall-cache-1 (Tainted: G    B              ): Objects remaining on __kmem_cache_shutdown()\n  [  2994.381140][  T111] Object 0xff11000004361118 @offset=4376\n  [  2994.381607][  T111] Allocated in p9_fcall_init+0x201/0x400 age=19564 cpu=1 pid=104\n  [  2994.382591][  T111]  p9_fcall_init+0x201/0x400\n  [  2994.382810][  T111]  p9_tag_alloc+0x12f/0x700\n  [  2994.382982][  T111]  p9_client_prepare_req+0x102/0x3e0\n  [  2994.383165][  T111]  p9_client_rpc+0x1ab/0xa50\n  [  2994.383334][  T111]  p9_client_getattr_dotl+0xb0/0x1a0\n  [  2994.383515][  T111]  v9fs_vfs_getattr_dotl+0x115/0x360\n  [  2994.383719][  T111]  vfs_getattr_nosec+0x22c/0x3a0\n  [  2994.383910][  T111]  vfs_statx+0xd7/0x170\n  [  2994.384062][  T111]  vfs_fstatat+0x45/0x80\n  [  2994.384215][  T111]  __do_sys_newfstatat+0x84/0xe0\n  [  2994.384386][  T111]  do_syscall_64+0x115/0x6a0\n  [  2994.384566][  T111]  entry_SYSCALL_64_after_hwframe+0x77/0x7f\n  [  2994.399720][  T111] WARNING: mm/slub.c:1244 at __kmem_cache_shutdown+0x363/0x500, CPU#0: busybox/111\n  [  2994.405655][  T111] Call Trace:\n  [  2994.406325][  T111]  kmem_cache_destroy+0x73/0x1b0\n  [  2994.406630][  T111]  p9_client_destroy+0x271/0x3c0\n  [  2994.407210][  T111]  v9fs_session_close+0x3c/0x260\n  [  2994.407409][  T111]  v9fs_kill_super+0x48/0x90\n  [  2994.407584][  T111]  deactivate_locked_super+0xa3/0x160\n  [  2994.407778][  T111]  cleanup_mnt+0x1dd/0x3e0\n\nThus, a successful umount left objects in the 9p fcall cache and prevented\nthe cache from being destroyed cleanly.\n\nPer-CPU shrink_qlist storage exists for every possible CPU, and each list\nis protected by its own raw spinlock.  Iterate over possible CPUs so that\na list populated before its CPU went offline is drained as well.\n\nfor_each_possible_cpu() can do more work than for_each_online_cpu(), but\nthis change only affects CONFIG_KASAN_GENERIC kernels.  The extra work is\nlimited to cache shrink and cache destruction paths and does not affect\nthe normal allocation/free fast path.  It adds one raw-spinlock-protected\nscan of each possible CPU's shrink list.  These lists are normally empty;\na non-empty list is traversed to remove objects belonging to the cache\nbeing shrunk or destroyed.\n\n## Remediation\n\nRefer to the linked advisories for vendor-supplied fixes and affected version ranges.","depth":"sunlit","depthScore":3,"depthScoreParts":{"impact":2.8,"likelihood":0,"exploitation":0,"ransomware":0},"changes":[]}