{"id":"CVE-2026-99084","title":"readdir: take no inode lock on an immutable directory","summary":"In the Linux kernel, the following vulnerability has been resolved:\n\nreaddir: take no inode lock on an immutable directory\n\niterate_dir() takes the directory's i_rwsem shared and holds it across\n->iterate_shared(). For a directory that n…","severity":"none","vendor":"Linux","product":"Linux","affected":["Linux >= 9d4e752a24f740b31ca827bfab07010e4e7f34b0 < 6d30b399ae28e5166d62484f81ccaefde6fe7996","Linux >= 9d4e752a24f740b31ca827bfab07010e4e7f34b0 < e08f57a7dcb83818c058e415988c23beb6c5a47f","Linux 7.1"],"published":"2026-10-23","updated":"2026-10-23","sourceUpdated":"2026-10-23T12:45:02.631Z","source":"CVEORG","sourceUrl":"https://www.cve.org/CVERecord?id=CVE-2026-99084","references":[{"url":"https://git.kernel.org/stable/c/6d30b399ae28e5166d62484f81ccaefde6fe7996"},{"url":"https://git.kernel.org/stable/c/e08f57a7dcb83818c058e415988c23beb6c5a47f"}],"tags":["cve.org"],"ingestedAt":"2026-10-23T13:24:55.509Z","slug":"CVE-2026-99084","body":"## Overview\n\nIn the Linux kernel, the following vulnerability has been resolved:\n\nreaddir: take no inode lock on an immutable directory\n\niterate_dir() takes the directory's i_rwsem shared and holds it across\n->iterate_shared(). For a directory that never has an entry and is\nnever removed the lock keeps nothing still, it only orders every reader\nand every writer of that inode behind each other.\n\nFor the directory of a nullfs instance that matters. The instance of\nthe initial mount namespace is the root of every empty mount namespace\nand the private instance is the root of every kernel thread, so one\ninode is shared across users who have nothing else in common. And a\nreader can hold the lock for as long as it likes: back the getdents()\nbuffer with a mapping of a file on a FUSE mount of your own, let the\ncopy of \".\" and \"..\" fault and let the server wait. Queue an exclusive\ntaker behind it, a mkdir() in that directory goes through start_dirop()\nbefore the read-only mount is reported, and from then on every lookup\nthat misses the dcache in that directory, every create and every mount\non it waits until the server answers. One user of an empty mount\nnamespace stalls all the others.\n\nAdd FOP_IMMUTABLE for the file operations of a directory that never\nchanges and is never removed and let iterate_dir() skip the lock for\nit. The flag never changes for a file, ->f_pos is protected by\nf_pos_lock since directories are FMODE_ATOMIC_POS, IS_DEADDIR can't be\nset on such a directory and neither touch_atime() nor fsnotify take\ni_rwsem. Set it on the nullfs directory. The placeholder directories of\nlibfs never have an entry either but their owners remove them, so they\nkeep the lock.\n\n## Affected\n\n- `Linux >= 9d4e752a24f740b31ca827bfab07010e4e7f34b0 < 6d30b399ae28e5166d62484f81ccaefde6fe7996`\n- `Linux >= 9d4e752a24f740b31ca827bfab07010e4e7f34b0 < e08f57a7dcb83818c058e415988c23beb6c5a47f`\n- `Linux 7.1`\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":[]}