commit e09610b541f2725a6382137c168e4efe208917c3 Author: Greg Kroah-Hartman Date: Fri Oct 30 13:08:34 2026 +0100 Linux 7.2.14 Link: https://lore.kernel.org/r/20261028165942.107695352@linuxfoundation.org Tested-by: Brett A C Sheffield Tested-by: Peter Schneider Tested-by: Salvatore Bonaccorso Tested-by: Pavel Machek (CIP) Tested-by: Miguel Ojeda Tested-by: Ron Economos Tested-by: SJ Park Tested-by: Shuah Khan Tested-by: Barry K. Nathan Tested-by: Florian Fainelli Tested-by: Sean Lyon Signed-off-by: Greg Kroah-Hartman commit 3d15ac36ce174477efa5365e424dbd47950fd2b9 Author: Christian Brauner Date: Mon Aug 24 14:08:20 2026 +0200 pid_namespace: prevent TIF_NOTIFY_SIGNAL from interrupting the reaper commit 5c0985dd3c32fc2bbd0e8beaf13cf52cd97376a2 upstream. Since commit 7fea700e04bd ("zap_pid_ns_processes: clear TIF_NOTIFY_SIGNAL along with TIF_SIGPENDING") zap_pid_ns_processes() clears TIF_NOTIFY_SIGNAL before every kernel_wait4() call. Otherwise kernel_wait4() doesn't sleep and returns -ERESTARTSYS if signal_pending() is true and the reaper busy-waits. But clearing TIF_NOTIFY_SIGNAL doesn't fix it. It can get raised again. For example, klp_send_signals() sets it raw against any task that hasn't transitioned yet. And it keeps sending it until the livepatch finishes. The sleep loop waiting for pid_allocated to drop has the same problem and it doesn't even clear anything. So a schedule() in TASK_INTERRUPTIBLE returns immediately. Use a no_notify_signal guard. signal_pending() will ignore TIF_NOTIFY_SIGNAL and the pid namespace reaper sleeps until a child exits or free_pid() wakes it. Link: https://patch.msgid.link/20260824-work-tif_notify_signal-v2-5-6609e42b3157@kernel.org Fixes: 7fea700e04bd ("zap_pid_ns_processes: clear TIF_NOTIFY_SIGNAL along with TIF_SIGPENDING") Reviewed-by: Oleg Nesterov Cc: stable@vger.kernel.org Suggested-by: Oleg Nesterov Signed-off-by: Christian Brauner (Amutable) Signed-off-by: Greg Kroah-Hartman commit de4ce7a7e3ba266bb753252bb6360619b8727b48 Author: Christian Brauner Date: Mon Aug 24 14:08:17 2026 +0200 coredump: prevent TIF_NOTIFY_SIGNAL from interrupting coredumps commit a58d647acfdc62561f3570d114025b3bf38ae41b upstream. TIF_NOTIFY_SIGNAL is used to kick a task in uninterruptible sleep to return to userspace and run task work and then go back to sleep. This mechanism works well but breaks coredumps. dump_interrupted() only allows fatal signals to interrupt a coredump and the whole regular write path going to actual filesystems is impervious to TIF_NOTIFY_SIGNAL as well. However, both the usermodehelper pipe and the coredump socket will bail early on TIF_NOTIFY_SIGNAL. This affects the following codepaths: - coredump_sock_recv() -> unix_stream_read_generic() The request/ack handshake is abandoned before any coredump data is sent. - anon_pipe_write() returning -ERESTARTSYS Once the pipe is full this truncates the coredump. - unix_stream_sendmsg() returning -ERESTARTSYS Once the send buffer is full this truncates the coredump. - coredump_sock_wait() -> __kernel_read() This reports a failure that didn't happen. - wait_for_dump_helpers() -> wait_event_interruptible() This stops waiting for the coredump helpers. Such truncation is entirely invisible to userspace and all uapi bits still indicate that a successful coredump happened. A crashing process with a bunch of file backed mappings and io_uring thrown in loses most of the coredump data. If the NT_FILE note goes past PAGE_SIZE mappings it's gonzo. TIF_NOTIFY_SIGNAL is sent by io_uring for the common case. And it uses poll without sleeping so a completion callback runs task_work_add() from interrupt context against the task that submitted the request. This is the task that is running the coredump. Since that task hasn't set work_exited (it hasn't exited yet after all) TIF_NOTIFY_SIGNAL keeps reappearing. A coredumping task doesn't return to userspace. The task work is run at exit. So interrupting it doesn't buy anything and just loses the coredump which is quite valuable. Note that this isn't specific to io_uring. There's also klp_send_signals(), bpf_task_work_schedule_signal(), landlock's tsync and then technically, kthread_stop() and the printk kunit test set the bit raw. Link: https://patch.msgid.link/20260824-work-tif_notify_signal-v2-2-6609e42b3157@kernel.org Fixes: 12db8b690010 ("entry: Add support for TIF_NOTIFY_SIGNAL") Reviewed-by: Oleg Nesterov Cc: stable@vger.kernel.org Signed-off-by: Christian Brauner (Amutable) Signed-off-by: Greg Kroah-Hartman commit 649c6e593ad62a90b0e1cf457ad1531a588e7b62 Author: Christian Brauner Date: Mon Aug 24 14:08:16 2026 +0200 signal: allow taks to temporarily block TIF_NOTIFY_SIGNAL commit 22b7f491be8f93ff72de0a3fe02b8ee9e64d3f21 upstream. TIF_NOTIFY_SIGNAL is used to kick a task in uninterruptible sleep to return to userspace and run task work and then go back to sleep. This mechanism works well but breaks e.g., coredumps. dump_interrupted() only allows fatal signals to interrupt a coredump and the whole regular write path going to actual filesystems is impervious to TIF_NOTIFY_SIGNAL as well. Add PF_NO_NOTIFY_SIGNAL and helpers to raise and restore it. This is the same approach as memalloc_nofs_save(). signal_pending() will not report a fake pending signal via TIF_NOTIFY_SIGNAL if inside a PF_NO_NOTIFY_SIGNAL section. No functional changes. Link: https://patch.msgid.link/20260824-work-tif_notify_signal-v2-1-6609e42b3157@kernel.org Reviewed-by: Oleg Nesterov Cc: stable@vger.kernel.org Signed-off-by: Christian Brauner (Amutable) Signed-off-by: Greg Kroah-Hartman commit 0182428fb2ae8bd56baf6f0a118fa45bf97514e3 Author: Christian Brauner Date: Mon Sep 21 15:44:57 2026 +0200 exit: hang up the tty before closing the files commit ad016f0131ebfec250a3360ea75d35128b3f4aa8 upstream. do_exit() closes the task's files in exit_files() and hangs up the controlling tty of a session leader in disassociate_ctty(1) after that. Since commit d99d38540bf0 ("fs: make close_files() synchronous") the final __fput() of every file runs inside exit_files(). So when the session leader holds the last open of its tty the tty is released before disassociate_ctty() runs. tty_release() clears signal->tty for the whole session in session_clear_tty() once the count drops to zero and sends no signal doing so. disassociate_ctty(1) then finds neither a tty nor a tty_old_pgrp and does nothing. The foreground process group loses its SIGHUP: do_exit() exit_files() close_files() tty_release() tty->count == 0 session_clear_tty() signal->tty = NULL, no signal disassociate_ctty(1) get_current_tty() NULL signal->tty_old_pgrp NULL, nothing sent That only affects real ttys. For a pty the master's open keeps the slave's count above zero. And it only affects a foreground job that holds no descriptor to the tty anymore while its session leader exits. Everything else is unchanged. The DTR drop on the last close happens in tty_port_shutdown() regardless, stopped jobs get their SIGHUP from kill_orphaned_pgrp() and signal->tty is cleared either way. Before that commit the final __fput() ran from exit_task_work() which comes after disassociate_ctty(). That order isn't old. Until v3.14 exit_task_work() came right after exit_files() and before v3.6 fput() was synchronous, so the tty was always released first. Commit c39df5fa37b0 ("exit: call disassociate_ctty() before exit_task_namespaces()") moved disassociate_ctty() up to fix a pppd crash and in front of exit_task_work() as a side effect. The hangup in this case has worked since then and that's eleven years of userspace being able to rely on it. Hang the tty up before closing the files. This is the ordinary hangup with the file still open: __tty_hangup() swaps in hung_up_tty_fops and tty_release() runs from the close afterwards as it does when a modem drops the line. disassociate_ctty() stays in front of exit_task_namespaces() which the pppd fix needs. Reported-by: Chris Mason Link: https://patch.msgid.link/20260921-work-coredump-fixes-v3-8-8e4adb1619e6@kernel.org Acked-by: Oleg Nesterov Signed-off-by: Christian Brauner (Amutable) Signed-off-by: Greg Kroah-Hartman commit 14c7de6ca239326b79fd3f7aee55dfada1bb72ab Author: Christian Brauner Date: Mon Sep 21 15:44:56 2026 +0200 fork: release the files of a failed fork after sched_cancel_fork() commit 296a98c1fa50778c3b31eebed1a3f903f9684349 upstream. Reorder copy_process() cleanup so sched_fork() taking scx_fork_rwsem scx_pre_fork() is safe and isn't held around exiting files. Right now, a fork that fails while another thread closed the bpf link fd will deadlock against its own read side. It also blocks every fork on the system: copy_process() holds scx_fork_rwsem for read exit_files() close_files() filp_close_sync() bpf_scx_unreg() kthread_flush_work() waits for the disable work scx_root_disable() percpu_down_write(&scx_fork_rwsem) Simply release the child's files after sched_cancel_fork() dropped the lock. The task starts out as a copy of its parent so p->files points to the parent's table until copy_files() replaces it. exit_files() must not run for a fork that failed before that. Clear p->files up front so exit_files() is a no-op for those and make copy_files() set it explicitly for CLONE_FILES. Reported-by: Chris Mason Link: https://patch.msgid.link/20260921-work-coredump-fixes-v3-7-8e4adb1619e6@kernel.org Signed-off-by: Christian Brauner (Amutable) Signed-off-by: Greg Kroah-Hartman commit faa706f087f9a985f7f0878b00e8595d51c5926c Author: Christian Brauner Date: Mon Sep 21 15:44:58 2026 +0200 ptrace: refuse to change the signal mask of a user worker commit cabc249e6560bee7bdfee9fbcb7ce4e951dcfd41 upstream. A user worker is created with every signal other than SIGKILL and SIGSTOP blocked. It never runs user code and that mask is what keeps get_signal() from dequeuing anything else for it. complete_signal() never picks a thread that blocks the signal, tkill() queues on the worker but nothing dequeues it. ptrace_signal() requeues an injected signal that the tracee blocks. PTRACE_SETSIGMASK is the only way to change that mask from the outside. A tracer that clears it makes the worker eligible for every signal. Whatever the worker then dequeues it can only act on by leaving. Refuse PTRACE_SETSIGMASK for a user worker with -EPERM, the way ptrace_attach() refuses a kernel thread. A worker is guaranteed to only ever dequeue SIGKILL or SIGSTOP and an injected signal stays pending on it, which is what already happens when the mask isn't touched. Taking the request and quietly leaving the mask alone was the other option, the way set_current_blocked() keeps SIGKILL and SIGSTOP unblocked whatever userspace asks for. But then a tracer gets success back with nothing changed and no way to tell that apart from a mask that took effect. Fixes: e8b33b8cfafc ("Revert "kernel: treat PF_IO_WORKER like PF_KTHREAD for ptrace/signals"") Cc: stable@vger.kernel.org Link: https://patch.msgid.link/20260921-work-coredump-fixes-v3-9-8e4adb1619e6@kernel.org Acked-by: Oleg Nesterov Signed-off-by: Christian Brauner (Amutable) Signed-off-by: Greg Kroah-Hartman commit 95990342844411d672859a1a7173e4d9235043b6 Author: Christian Brauner Date: Tue Sep 22 15:26:16 2026 +0200 signal: only SIGKILL and the freezers interrupt a coredumping task commit 563d606b93d7a1d23512832fc252d32c68860d87 upstream. For quite a while now coredump has allowed either SIGKILL or freezing to interrupt an ongoing coredump. Since we seem to like it complicated "freezing" can mean a lot of things: (1) Power Management induced freezing (2) cgroup v1 freezer controller induced freezing (3) cgroup v2 freezer controller induced freezing So (1) and (2) are handled by checking freezing() but (3) isn't. (3) uses cgroup_freeze_task() which sets JOBCTL_TRAP_FREEZE and calls signal_wake_up() on every task in the cgroup. The trap isn't handled by the coredump code. Which means the current logic is inconsistent. Power and cgroup v1 freeze and abort the coredump. With cgroup v2 it depends on where the core goes. A dump to a file completes and the cgroup v2 freeze has to wait for it to finish. When dumping to a pipe or socket the dump is interrupted at the next write because of trap's TIF_SIGPENDING. Which is it? Let's be consistent and align (1)-(3): freezing interrupts an ongoing coredump and doesn't make it wait until the dump is done. All three freezers go through signal_wake_up() and so set TIF_SIGPENDING on the dumper. And once zap_threads() has set core_state that flag can't mean anything else anymore. prepare_signal() drops every signal but SIGKILL for the group, the dying threads don't retarget shared signals to the dumper, and zap_process() cleared the pending job control traps. So the flag is the thing to check. Let dump_interrupted() test task_sigpending() and nothing else. That is also what the pipe and socket writes already fail on. TIF_NOTIFY_SIGNAL is left out on purpose, task work must not cut a dump short. A seized tracee that gets PTRACE_INTERRUPT during the dump is cut short the same way. It already was for pipes and sockets. The one freeze that check can miss is one requested before the dump started. zap_threads() clears TIF_SIGPENDING and the cgroup v2 freezer only wakes a task once while its JOBCTL_TRAP_FREEZE isn't part of JOBCTL_PENDING_MASK and survives zap_process(). So refuse to start the dump when freezing() is true or the trap is pending. The task then exits without dumping, which is what happens today when the freeze interrupts the dump at the first write. Fixes: 403bad72b67d ("coredump: only SIGKILL should interrupt the coredumping task") Fixes: 76f969e8948d ("cgroup: cgroup v2 freezer") Cc: stable@vger.kernel.org Suggested-by: Oleg Nesterov Link: https://lore.kernel.org/r/arJ6zZuI59BJnXpl@redhat.com Link: https://patch.msgid.link/20260921-work-coredump-fixes-v3-2-8e4adb1619e6@kernel.org Signed-off-by: Christian Brauner (Amutable) Signed-off-by: Greg Kroah-Hartman commit 7a6afc7fa9750a1e6c54ae311e6fe2b97985ef43 Author: Christian Brauner Date: Mon Sep 21 15:44:54 2026 +0200 signal: don't retarget shared signals in a dying thread group commit c0e659511d4f49087da095a95799edb5b775a705 upstream. If the calling thread has blocked a signal then retarget_shared_pending() slingshots a shared pending signal to a sibling thread. Today it only skips PF_EXITING threads but not a coredumping thread. But if SIGNAL_GROUP_EXIT is set no thread in the thread-group can dequeue such signals anymore as get_signal() sends SIGKILL to every thread and prepare_signal() drops any other signal. exit_signals() also returns early on SIGNAL_GROUP_EXIT. But neither sigprocmask() nor the mask restore in sigtimedwait() do. Say a thread sleeps in sigtimedwait() and a sibling thread crashes. The sleeping thread gets woken by the zap call. It restores its mask on the way out and retargets a signal that was queued for the coredumping process. That means TIF_SIGPENDING is set on the coredumping task which zap_threads() had cleared. So a blocking write sees TIF_SIGPENDING and truncates the dump. Stop retargeting signals when SIGNAL_GROUP_EXIT is set and avoid needlessly truncating coredumps. Fixes: e6fa16ab9c1e ("signal: sigprocmask() should do retarget_shared_pending()") Cc: stable@vger.kernel.org Acked-by: Oleg Nesterov Link: https://patch.msgid.link/20260921-work-coredump-fixes-v3-5-8e4adb1619e6@kernel.org Signed-off-by: Christian Brauner (Amutable) Signed-off-by: Greg Kroah-Hartman commit a0f09ff4c6b93312e565cfe75d76c3a551a3ec4c Author: Christian Brauner Date: Mon Sep 21 15:44:50 2026 +0200 coredump: hold RCU while releasing parked threads commit 61d58591aa0188bf2463e263962256316ff8b576 upstream. coredump_finish() releases the parked threads by clearing ->task in their core_thread entry and calling wake_up_process() on each of them. It does that without holding a reference on the task and without being inside rcu. But calling wake_up_process(task) without rcu here isn't safe. A parked thread doesn't need that wakeup to leave. A spurious wakeup or a preemption after the store is enough for the task to go away. So if the coredump client is preempted between the store and wake_up_process() the thread can exit and be freed in the meantime and try_to_wake_up() takes pi_lock in freed memory. Hold rcu across the loop. Fixes: a94e2d408eae ("coredump: kill mm->core_done") Cc: stable@vger.kernel.org Acked-by: Oleg Nesterov Link: https://patch.msgid.link/20260921-work-coredump-fixes-v3-1-8e4adb1619e6@kernel.org Signed-off-by: Christian Brauner (Amutable) Signed-off-by: Greg Kroah-Hartman commit 7271c064dd60d4456c5b90da757acd9420d06697 Author: Aaron Ahmed Date: Fri Oct 16 23:04:03 2026 +0000 ftrace: Initialize modname and offset in print_rec() commit e1f339ff3b83857b93baf8b88186275a413d445e upstream. print_rec() leaves 'modname' and 'offset' uninitialized and kallsyms_lookup() only sets them on success. When the lookup fails like when a module is being loaded or unloaded, 'modname' holds a garbage value, passes the 'if (modname)' check, and is dereferenced by seq_printf(). Oops: general protection fault RIP: 0010:string+0x48/0xe0 Call Trace: vsnprintf+0x1f5/0x650 seq_printf+0x8e/0xb0 print_rec+0x6c/0xc0 t_show+0x4c/0x440 seq_read_iter+0x119/0x480 This can be triggered by reading available_filter_functions while a module is being loaded or unloaded. Initialize both variables so a failed lookup skips the module name. Cc: stable@vger.kernel.org Fixes: b39181f7c690 ("ftrace: Add FTRACE_MCOUNT_MAX_OFFSET to avoid adding weak function") Link: https://patch.msgid.link/20261016230403.4916-1-aarnahmd@amazon.com Tested-by: Gavin Inglis Signed-off-by: Aaron Ahmed Signed-off-by: Steven Rostedt Signed-off-by: Greg Kroah-Hartman commit 33a078f04e264eb71aa9f209cb334e4010123818 Author: Weiming Shi Date: Sun Sep 6 19:01:53 2026 +0800 asymmetric_keys: reject trust keys without IDs commit 9695df3b1db4ac4534274d38ee16e3b9b0bf910e upstream. The PKCS#8 parser deliberately leaves the asym_key_ids payload slot empty. key_or_keyring_common() unconditionally dereferences that slot when a PKCS#8 key is supplied to a key_or_keyring restriction. Both the plain and :chain forms reach this branch, allowing an unprivileged caller to trigger a NULL pointer dereference through KEYCTL_RESTRICT_KEYRING followed by add_key(). Oops: general protection fault, probably for non-canonical address 0xdffffc0000000000 KASAN: null-ptr-deref in range [0x0000000000000000-0x0000000000000007] RIP: 0010:key_or_keyring_common (crypto/asymmetric_keys/restrict.c:205 crypto/asymmetric_keys/restrict.c:279) Call Trace: __key_create_or_update (security/keys/key.c:884) key_create_or_update (security/keys/key.c:1021) __do_sys_add_key (security/keys/keyctl.c:134) do_syscall_64 (arch/x86/entry/common.c:52 arch/x86/entry/common.c:83) entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:130) Kernel panic - not syncing: Fatal exception Reject an asymmetric restriction key without key IDs. A PKCS#8 private key cannot identify the signer of a candidate certificate and is not a valid trust anchor. Fixes: 3c58b2362ba8 ("KEYS: Implement PKCS#8 RSA Private Key parser [ver #2]") Cc: stable@vger.kernel.org Reported-by: Xiang Mei Link: https://lore.kernel.org/r/20260429181629.110802-2-bestswngs@gmail.com Assisted-by: Claude:claude-opus-4-8 Signed-off-by: Weiming Shi Signed-off-by: Herbert Xu Signed-off-by: Greg Kroah-Hartman commit 63b19ec2863b2cb5279b746ecd3d6210b1a93945 Author: Srish Srinivasan Date: Sat Sep 12 11:59:49 2026 +0530 keys/trusted_keys: return immediately after TPM unseal failure commit f92d0ccb05bf9089072cb067dda0dffd5394c4ef upstream. trusted_tpm_unseal() proceeds to pcrlock() when the TPM unseal operation fails. If pcrlock() succeeds, its return value overwrites the unseal error, causing key instantiation to succeed. Return immediately when unseal fails to preserve the original error. Cc: stable@vger.kernel.org # v5.15+ Fixes: 5d0682be3189 ("KEYS: trusted: Add generic trusted keys framework") Signed-off-by: Srish Srinivasan Reviewed-by: Jarkko Sakkinen Link: https://lore.kernel.org/r/20260912062950.279104-2-ssrish@linux.ibm.com Signed-off-by: Jarkko Sakkinen Signed-off-by: Greg Kroah-Hartman commit be761640fbf427997f26c17ab3670f52753572a0 Author: Stefano Garzarella Date: Wed Sep 23 19:35:06 2026 +0200 KEYS: trusted: Fix blob allocation size in tpm2_key_decode() commit 11077cbcc2695f71c944a84a5f08c6672ab759f1 upstream. tpm2_key_decode() allocates 4 bytes more than needed. The ASN.1 callbacks tpm2_key_priv() and tpm2_key_pub() provide the lengths of TPM2B_PRIVATE and TPM2B_PUBLIC, so ctx.priv_len and ctx.pub_len already account for the 2-byte `size` field each of those structures starts with. I noticed this while reviewing commit 114f00d738f1 ("KEYS: trusted: Fix tpm2_load_cmd() boundary check"), which correctly reports ctx.priv_len + ctx.pub_len as the decoded blob size [1]. Let's allocate exactly that amount, matching the data copied into the blob. [1] https://lore.kernel.org/linux-integrity/apfoKo-BdwaLXtkT@sgarzare-redhat/ Cc: stable@vger.kernel.org # v5.13+ Fixes: f2219745250f ("security: keys: trusted: use ASN.1 TPM2 key format for the blobs") Signed-off-by: Stefano Garzarella Link: https://lore.kernel.org/r/20260923173506.41519-1-sgarzare@redhat.com Reviewed-by: Jarkko Sakkinen Signed-off-by: Jarkko Sakkinen Signed-off-by: Greg Kroah-Hartman commit 6fc0c818b1cac66003e937c8a0d3bfb23fd3f130 Author: Adrian Martinez Date: Sat Oct 10 23:40:00 2026 +0300 KEYS: Fix key_reject_and_link() race with keyring restriction commit 0b157166e612b70f98a332b6c7be7eac6f8e9593 upstream. The restrict_link check in key_reject_and_link() is done before taking the keyring semaphore, racing with a concurrent keyring_restrict() that installs the restriction while holding the semaphore in write mode. Move the check to after __key_link_lock(), so it is done under the semaphore, matching the pattern used in __key_create_or_update(). Cc: stable@vger.kernel.org Fixes: 5ac7eace2d00 ("KEYS: Add a facility to restrict new links into a keyring") Reviewed-by: Jarkko Sakkinen Signed-off-by: Adrian Martinez Link: https://lore.kernel.org/r/CACO1+RT+m==BY+ZPkE+fo536D0qu3wog4xE1dfuXQ56g+=Qv=A@mail.gmail.com Signed-off-by: Jarkko Sakkinen Signed-off-by: Greg Kroah-Hartman commit cf1af7df80e38bf58bd74c2cccce32fcbd1a596a Author: Cen Zhang Date: Mon Oct 19 22:57:21 2026 +0800 keys: Protect the read method check against type removal commit 925c84e3cc21caf09de92d9bcafab0147790debf upstream. keyctl_read_key() checks key->type->read without holding key->sem. A reference keeps the key alive, but does not pin the module containing its type. A surviving RxRPC key can therefore be read while AF_RXRPC is being unloaded: KEYCTL_READ task Module unload and key GC ---------------- ------------------------ lookup_user_key() permission / possession checks load key->type af_rxrpc_exit() unregister_key_type() GC: down_write(key->sem) GC: retype to key_type_dead GC: up_write(key->sem) unregister_key_type() returns free_module() read saved type's read member The final access can read freed module storage. The read invocation in __keyctl_read_key() is already protected by key->sem, but the earlier method check is outside that critical section. Move the method check into __keyctl_read_key(), after key_validate() and under its existing read hold on key->sem. The check and invocation then use the same protected type for length queries and every read retry. The GC cannot retype the key until the semaphore is released, and type unregistration waits for that GC pass before the module can be freed. Return -EOPNOTSUPP when validation succeeds and the type has no read method. A temporary buffer may now be allocated before this result; allocation and copying to userspace remain outside the semaphore. Assisted-by: LLM Cc: stable@vger.kernel.org # v5.10+ Fixes: d3ec10aa9581 ("KEYS: Don't write out to userspace while holding key semaphore") Suggested-by: Jarkko Sakkinen Signed-off-by: Cen Zhang Reviewed-by: Jarkko Sakkinen Reviewed-by: David Howells Link: https://lore.kernel.org/r/pm-key-management-objects-candidate-0010-v7-fc85721ff5cfb9e33ac6@gmail.com Signed-off-by: Jarkko Sakkinen Signed-off-by: Greg Kroah-Hartman commit 84f7c699f8d139f743c7dc11ef41d085690955bb Author: Kyle Zeng Date: Tue Oct 13 16:12:51 2026 -0700 keys: make keyring indexes independent of kernel addresses commit de891c73c1cf41ac8de12c07fa2c84d6a912ad0c upstream. KEYCTL_READ returns the serial numbers in a keyring in associative-array traversal order. The first word of the index is a public, deterministic hash of the type pointer, domain pointer and description. An unprivileged user can therefore insert keys with known descriptions, predict their order for each possible KASLR slide, and recover the kernel image address. On an x86-64 build with 16 MiB alignment, 45 user keys suffice to distinguish the candidate slides. Remove the addresses from the whole index, not just its first hash word. Assign each key type and domain a nonzero 64-bit ID and hash those IDs and the description with a once-initialized SipHash key. Use the full IDs in the later index chunks as well. Truncating or hashing the IDs there would lose the guarantee that distinct type/domain/description tuples have a differing index bit. Allocate IDs lazily since private key types need not be registered. Cache them in the index key so tree walks and copies of an existing index do not need to dereference a type descriptor that may have been unregistered. Keep the 32-bit chunk layout and diff_objects() bit offsets consistent, and retain the separate root slot for keyrings used by recursive search. The key comparison, permissions and userspace data format are unchanged; only the address-dependent in-memory ordering is replaced. Cc: stable@vger.kernel.org Fixes: b2a4df200d57 ("KEYS: Expand the capacity of a keyring") Assisted-by: Codex:gpt-6-astra Signed-off-by: Kyle Zeng Reviewed-by: Jarkko Sakkinen Link: https://lore.kernel.org/r/20261013231251.56170-1-kylebot@openai.com Signed-off-by: Jarkko Sakkinen Signed-off-by: Greg Kroah-Hartman commit dea367df18e3a4faac393e3a4e818ee03af89b19 Author: Cen Zhang Date: Fri Oct 9 13:11:03 2026 +0800 keys: Protect the type name while describing a key commit fced7238b7e80cf8aa688d5af976e4aa6e33c329 upstream. keyctl_describe_key() must keep the key type's name alive until kasprintf() has consumed it. The retained key reference keeps the key and its description alive, but does not pin the module providing its type. The name is passed to kasprintf() without holding key->sem. kvasprintf() uses the saved name pointer in two formatting passes, with a GFP_KERNEL allocation between them. With AF_RXRPC=m, a task that passes View permission before type retirement can overlap a privileged module unload in this order: 1. KEYCTL_DESCRIBE looks up the key, saves key->type->name in the kasprintf() arguments and completes the first formatting pass. 2. While the formatter is delayed before its second pass, af_rxrpc_exit() calls unregister_key_type(), which removes the type from the registry and waits for key_gc_keytype(). 3. The collector takes key->sem for writing, retypes the retained key to key_type_dead and completes retirement. Unregistration and module exit return, allowing free_module() to release the type and name. 4. The formatter resumes its second pass and reads the saved name from released module storage, which can fault in string(). Hold key->sem for reading across kasprintf(). The collector then cannot retype the key or complete retirement until both formatting passes have finished. If retirement wins the lock first, formatting uses the core key_type_dead name instead. Release the semaphore immediately after kasprintf(), including on allocation failure, before copying to userspace. Oops report as below: BUG: unable to handle page fault for address: fffffbfff8054794 #PF: supervisor read access in kernel mode #PF: error_code(0x0000) - not-present page RIP: 0010:string+0x23f/0x470 Call Trace: vsnprintf+0x330/0x1110 Cc: stable@vger.kernel.org Fixes: aa9d4437893f ("KEYS: Fix the size of the key description passed to/from userspace") Assisted-by: LLM Signed-off-by: Cen Zhang Reviewed-by: Jarkko Sakkinen Link: https://lore.kernel.org/r/pm-key-management-objects-candidate-0009-v4-6b537c5732fd957c33fa@gmail.com Signed-off-by: Jarkko Sakkinen Signed-off-by: Greg Kroah-Hartman commit b4dfda438fee3c6faa15d9b421c373de6a0e476f Author: Cen Zhang Date: Fri Oct 9 13:52:16 2026 +0800 keys: Check the authorization payload in process keyring searches commit 0a0ec45bf577a9b4d4960567f20d3877ef480de9 upstream. An authorization payload and its saved credentials must remain valid while search_process_keyrings_rcu() searches the requestor's keyrings. The function validates the authorization key, then separately loads payload.data[0] and unconditionally reads rka->cred. Validation does not prevent request_key_auth_revoke() from clearing the payload. A request-key helper can assume authority and fork a descendant that inherits the authorization key. If the descendant misses in its own keyrings, it falls back to the saved requestor credentials. Successful target instantiation invalidates the authorization key without clearing the payload. After the helper exits, UMH_WAIT_PROC returns to the requester, which calls complete_request_key() and revokes that key. With the descendant paused after validation, the paths can interleave as follows: 1. The descendant enters search_process_keyrings_rcu() under RCU and key_validate() succeeds. 2. The helper instantiates the target, invalidates the authorization key, and exits. 3. The requester resumes from UMH_WAIT_PROC, calls complete_request_key(), and revokes the authorization key. request_key_auth_revoke() sets its payload pointer to NULL. 4. The descendant loads NULL and dereferences rka->cred. This NULL dereference can oops the kernel. RCU delays payload disposal but does not prevent the pointer from being cleared. Load the payload with dereference_key_rcu() and require a non-NULL result before validating the authorization key and searching its saved credentials. A non-NULL payload and its credentials remain alive for the caller's RCU read-side section; a cleared payload skips the fallback search and follows the existing error selection. KASAN report as below: Oops: general protection fault, probably for non-canonical address 0xdffffc0000000005: 0000 [#1] SMP KASAN NOPTI KASAN: null-ptr-deref in range [0x0000000000000028-0x000000000000002f] RIP: 0010:search_process_keyrings_rcu+0x32b/0x790 Cc: stable@vger.kernel.org Fixes: e59428f721ee ("keys: Move the RCU locks outwards from the keyring search functions") Assisted-by: LLM Signed-off-by: Cen Zhang Reviewed-by: Jarkko Sakkinen Link: https://lore.kernel.org/r/pm-key-management-objects-candidate-0004-v3-598361863b76bc25d58d@gmail.com Signed-off-by: Jarkko Sakkinen Signed-off-by: Greg Kroah-Hartman commit 5c6a9512887a6cc694d594d9c0b57c0841d215b5 Author: Cen Zhang Date: Fri Oct 9 13:11:59 2026 +0800 keys: Avoid the owner account dereference in named keyring lookup commit 51409aaa01f8602552c336009fd22b226b0cdf1d upstream. Named keyring lookup must keep the storage containing the owner UID alive through the namespace mapping check. find_keyring_by_name() reads keyring->user->uid under keyring_name_lock, but that lock protects the keyring's name entry and allocation, not its separate key_user. A named session-keyring join can overlap a privileged KEYCTL_CHOWN on another CPU. If the keyring holds the last reference to its old dynamic key_user and the ownership transfer succeeds, the following ordering is possible: Named join Chown find_keyring_by_name() keyctl_chown_key() read_lock(keyring_name_lock) down_write(key->sem) load old keyring->user replace key->user and key->uid up_write(key->sem) key_put(key) key_user_put(old user): free read old user->uid read_unlock(keyring_name_lock) Chown neither takes keyring_name_lock nor key_session_mutex, so it can free the old account between the pointer load and the UID read. The lookup then reads freed memory even though the keyring itself is alive. Use keyring->uid for the mapping check. key_alloc() initializes this inline owner UID and keyctl_chown_key() updates it on chown. The existing name lock keeps its containing keyring allocated throughout the check, so lookup no longer depends on the account's lifetime. This also matches the owner UID used by the permission check. KASAN report as below: BUG: KASAN: slab-use-after-free in find_keyring_by_name+0x577/0x5d0 Read of size 4 at addr ffff8881128ae2ec by task keycase/500 Cc: stable@vger.kernel.org Fixes: 2ea190d0a006 ("keys: skip keys from another user namespace") Assisted-by: LLM Signed-off-by: Cen Zhang Reviewed-by: Jarkko Sakkinen Link: https://lore.kernel.org/r/pm-key-management-objects-candidate-0002-v3-2baa8cad2bb554364e10@gmail.com Signed-off-by: Jarkko Sakkinen Signed-off-by: Greg Kroah-Hartman commit 72debb893a4bb99ec2c26cf32f5fad60ac1ea26e Author: Jens Axboe Date: Tue Oct 27 12:02:04 2026 -0600 io_uring: keep the drain active when a deferred drain request is started Commit 2295613cd3713beffc8665e956135a5d48dfc208 upstream. ecb6817b769f ends the drain when the dispatched drain request has posted its CQE and nothing else is deferred. But the deferred list is processed in between, and if the request it starts is itself a drain request, the drain is over before that one has even run. A request submitted after that point then runs ahead of it: 1. P0 in flight, D1 and D2 (both IOSQE_IO_DRAIN) deferred behind it 2. P0 completes, D1 is started 3. D1 completes, drain_done is set, D2 is started, the list is empty 4. IO_RING_F_DRAIN_ACTIVE is cleared with D2 still in flight Anything submitted now bypasses the drain. The same happens with D2 as the head of a link, as those aren't tracked in ->drain_req at all. Have io_queue_deferred() report whether it started a drain request, and don't end the drain if it did. Starting non-drain requests and ending it is still fine, they were only waiting on the drain request that just completed. Fixes: ecb6817b769f ("io_uring: end the drain when the drain request completes") Reported-by: Karl Mehltretter Link: https://lore.kernel.org/io-uring/20261027043323.98916-1-kmehltretter@gmail.com/ Cc: stable@vger.kernel.org Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 7be9e9d997e0d105b9e65aac35447ad3a854436a Author: Jens Axboe Date: Mon Oct 19 13:30:51 2026 -0600 io_uring: end the drain when the drain request completes Commit ecb6817b769fd7d71af2c9f446188649020b2e53 upstream. Since requests are drained based on allocated requests, the drain stays active after the drain request has posted its CQE, until it is freed and a new request finds the defer list empty. A drain request run by io-wq is freed only after the worker drops its reference, so a request submitted after the drain completed can still be deferred behind it, and get issued much later than the application expects. A drain request dispatched from the defer list runs alone. Track it, and end the drain when its CQE is flushed and nothing else is deferred. Have io-wq complete drain requests through task_work, so that this happens under ->uring_lock before the CQE is visible. Linked drain requests keep waiting until the link is freed. Reported-by: Wei Gao Link: https://lore.kernel.org/io-uring/20261018070154.16944-1-wegao@suse.com/ Fixes: 8fb7aee05591 ("io_uring: drain based on allocates reqs") Cc: stable@vger.kernel.org Tested-by: Wei Gao Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 857aa2012b91cad6fdef54a4b6c632d9d0b6e64f Author: Junyeong Lee Date: Tue Oct 20 00:21:00 2026 +0900 io_uring/waitid: prevent cancel reference wrap Commit 98787d3fa4e99564d2b2bc21255b21f670c450d0 upstream. io_waitid packs the cancellation state into bit 31 of ->refs and uses bits 30:0 as a 31-bit reference counter: #define IO_WAITID_CANCEL_FLAG BIT(31) #define IO_WAITID_REF_MASK GENMASK(30, 0) __io_waitid_cancel() marks a request canceled, then increments the low bits of ->refs to claim completion ownership. When another path owns the request, the increment remains even though cancellation returns false. The request remains discoverable by cancellation, so repeated failed attempts can wrap the 31-bit reference field to zero. A later cancellation can claim ownership and complete the request a second time. Use atomic_fetch_or() when setting the cancel flag and inspect the old flag value. The first cancel attempt continues to the existing ownership increment, preserving its wake and issuer handoff semantics. Later attempts see the cancel flag and return without incrementing the reference field. This makes repeated failed cancellation attempts idempotent while leaving the first-cancel state transitions unchanged. Fixes: f31ecf671ddc ("io_uring: add IORING_OP_WAITID support") Cc: stable@vger.kernel.org Assisted-by: LLM Signed-off-by: Junyeong Lee Link: https://patch.msgid.link/20261019152100.63565-1-jylee5498@gmail.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 4bfec43cf844ca25fc5d5717effea22bb4b6eae6 Author: Austin Schuh Date: Sat Oct 17 10:17:26 2026 -0700 io_uring/wait: wake real-time waiters from hard interrupt context Commit 5b01d5955ac206e38097f777eff6ab2a8b948c41 upstream. Since commit cebf123c634a ("io_uring: implement our own schedule timeout handling"), a CQ wait with a timeout sleeps on io_uring's own hrtimer instead of an hrtimer sleeper. On PREEMPT_RT, an hrtimer that isn't marked for hard interrupt expiry expires in the per-CPU ktimers thread, which runs at SCHED_FIFO 1. A real-time task waiting in io_uring_enter() is therefore not woken at its deadline while any other real-time task is runnable on its CPU: the wakeup waits until every one of them has blocked. hrtimer sleepers, which schedule_hrtimeout_range_clock() used to give io_uring, don't have this problem: __hrtimer_setup_sleeper() marks the timer hard when the sleeping task is in a real-time or deadline class. Both of io_uring's callbacks are fine in hard interrupt context: io_cqring_timer_wakeup() sets a flag and calls wake_up_process(), and io_cqring_min_timer_wakeup() only reads ring state, under RCU, before doing the same or re-arming the timer. Apply the sleeper rule to them. Measured on a Jetson Orin Nano running 6.18.39-rt with this change backported: a SCHED_FIFO 40 task waits in io_uring_enter() for 40,000 deadlines 500us apart, while a SCHED_FIFO 20 task runs bursts of up to 1ms on the same CPU. Before, 6,647 wakeups were more than 100us late, the latest by 1,040us, and 2,758 deadlines passed without a wakeup at all. After, none were more than 100us late and the latest was 18us. This mirrors epoll: epoll_pwait2() waits on an hrtimer sleeper, so its timeout is already hard for real-time tasks, and on the same setup it was never more than 17us late. Fixes: cebf123c634a ("io_uring: implement our own schedule timeout handling") Cc: stable@vger.kernel.org Assisted-by: LLM Signed-off-by: Austin Schuh Link: https://patch.msgid.link/20261017171726.601443-1-austin.linux@gmail.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 6cc019eadf12836dbeeeacc1adbfc35be49c7a39 Author: Tamaki Yanagawa Date: Sun Oct 18 04:22:32 2026 +0000 io_uring/net: keep recv buffer select length within MAX_RW_COUNT Commit 96262cfab554cf11412de7a6247f0e82c94e130f upstream. Since commit dcbd1c054848 ("io_uring/net: let io_recv_buf_select return the length of the buffer region"), io_recv_buf_select() returns the selected length as an int. For a non-bundle recv with sqe->len == 0 the length comes from the ring provided buffer, which can be up to 4G, so a buffer of 2G or more makes io_recv() fail the request with a bogus negative result (e.g. -2147483648) instead of receiving into it. Clamp the length to MAX_RW_COUNT, which import_ubuf() limits the transfer to anyway. Fixes: dcbd1c054848 ("io_uring/net: let io_recv_buf_select return the length of the buffer region") Cc: stable@vger.kernel.org Signed-off-by: Tamaki Yanagawa Assisted-by: Claude:claude-opus-5-5 Link: https://patch.msgid.link/20261018042227.1105-1-ty@000ty.net Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit d379a8db88e5c6847b4db5ba8b4880463d051113 Author: Jens Axboe Date: Fri Oct 16 09:15:13 2026 -0600 io_uring/poll: don't loop on HUP with multiple references Commit 67bb1ad9a517bedf1e1599fe5448620b9f38bf2a upstream. io_poll_check_events() drops down to one reference when a multishot request sees HUP with more than one, so that the loop goes around once more. It did so by subtracting one less than it had. With the cmpxchg that only takes the local count down: the exchange fails, the next pass reads the unchanged count, and goes through the same motions again. The loop then only ends once io_poll_issue() stops the request. Drop the extra references from the counter instead, and loop once more without trying the exchange. The next pass sees one reference, like it did before, and ends unless a wake up came in. Fixes: 937d38dcacd6 ("io_uring/poll: avoid retry wakeup loss when dropping ownership") Signed-off-by: Jens Axboe Co-Authored-By: Claude Fable 5.1 Signed-off-by: Greg Kroah-Hartman commit df705a274da10da7bb546e08b49fa80a6ef19c78 Author: Jens Axboe Date: Thu Oct 15 16:38:45 2026 -0600 io_uring/timeout: don't lose a cancel as io-wq arms a timeout Commit 89f193dc3ef462059387e9691bf6af4b1808379a upstream. An IORING_OP_TIMEOUT run by io-wq, because of IOSQE_ASYNC, has the same problem as poll, futex and waitid. io_wq_submit_work() looks at IO_WQ_WORK_CANCEL once, before it issues the request. A cancel in between flags the work, gets -EALREADY, and searches ->timeout_list in vain, while the worker then adds the request and starts the timer: io-wq worker io_try_cancel() io_wq_submit_work() IO_WQ_WORK_CANCEL isn't set io_async_cancel_one() sets IO_WQ_WORK_CANCEL = -EALREADY io_timeout_cancel() = -ENOENT io_timeout() adds the request to ->timeout_list hrtimer_start() = IOU_ISSUE_SKIP_COMPLETE Nobody looks at the flag again, and the timeout runs its course. At ring exit, io_uring_try_cancel_requests() cancels io-wq work and then kills the timeouts the same way, so an exiting task can get stuck for the full duration of such a timeout. Look at the flag under ->timeout_lock, which the search takes as well, and fail the request if it is set. If the search came first, the flag is visible. If it comes later, it finds the request on the list. Fixes: ce35a47a3a02 ("io_uring: add IOSQE_ASYNC") Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 9d5ba7e121dc8c4cfada7a9ac8864345798eb229 Author: Jens Axboe Date: Thu Oct 15 14:25:00 2026 -0600 io_uring/timeout: don't flush with the sequence behind the last flush Commit 85f85d573ffa7772e1ad4504d0cec5113a5c5cf1 upstream. io_flush_timeouts() measures the events since the last flush as the distance of the current sequence from ->cq_last_tm_flush. A fired timeout bumps ->cq_timeouts before its CQE is posted, so until it is the sequence can be behind the last flush, and the distance wraps: last flush: seq 100, cq_last_tm_flush = 100 Y: IORING_OP_TIMEOUT, off 50, target_seq 150 pure timeout A fires, cq_timeouts = 1 task_work posts A's CQE, cached_cq_tail = 101 pure timeout B fires before the flush, cq_timeouts = 2 io_flush_timeouts(): seq = 99, events_got = 0xffffffff Y, and every sequence timeout behind it, is ended with -ETIME although no event happened. Two timeouts expiring close together is all it takes, a completion with IOSQE_CQE_SKIP_SUCCESS after a single one does it too. Nothing but those in flight CQEs separates the two, so count from the last flush when the sequence is behind it, as io_timeout() does. Fixes: f010505b78a4 ("io_uring: flush timeouts that should already have expired") Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit db2cdb40d9f8f17f62346e4e601a2160c37d3138 Author: Jens Axboe Date: Thu Oct 15 14:18:22 2026 -0600 io_uring/timeout: don't put a sequence target behind the last flush Commit fdfce905a58edea053ee7b68b5bd4348dd1b36e0 upstream. io_timeout_fn() and io_kill_timeout() bump ->cq_timeouts when a timeout fires, but its CQE is only posted later from task_work. Until then the sequence io_timeout() computes is one below the last flush per fired timeout. Now that ->cq_last_tm_flush is kept while sequence timeouts are pending, a timeout armed in that window with a small offset gets a target behind the base. Its distance from the base wraps, and it's never completed by events, only by its timer. With DEFER_TASKRUN that window is as wide as the application makes it, as submission runs before the local task_work. Nothing but those in flight CQEs separates the two, so use the last flush as the base for the target when the sequence is behind it. Fixes: c6c0b3eba88c ("io_uring/timeout: don't skip over reached sequences") Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit eb2a50d8235e4580fea9aa20f3fbe1a002c38cb1 Author: Juanlu Herrero Date: Tue Oct 13 15:37:20 2026 -0700 io_uring/poll: avoid retry wakeup loss when dropping ownership Commit 937d38dcacd675dc0572d6c9b6743960df3128ed upstream. If a wakeup sets IO_POLL_RETRY_FLAG between the owning task consuming the existing instance of IO_POLL_RETRY_FLAG and the loop check of io_poll_check_events(), then we stall on processing the request until an unrelated wakeup comes along. Using atomic_cmpxchg instead of atomic_sub_return to guarantee both the count v and poll_refs are 0 and no wake-ups are missed. Cc: stable@vger.kernel.org Fixes: a26a35e9019f ("io_uring: make poll refs more robust") Signed-off-by: Juanlu Herrero Assisted-by: LLM Suggested-by: Anand Natarajan Link: https://patch.msgid.link/20261013-pu-cmpxchg-v1-1-4ddc10fd8a03@fastmail.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit a01f35f52314a8d5d8c7273dac934aab4143ddb1 Author: Josef Bacik Date: Thu Oct 15 19:31:07 2026 +0000 io_uring: look up the file before queueing to io-wq Commit 98bc044d5a5374f8d3605fe55a8e2df6d6fc3f82 upstream. io_prep_async_work() decides by the file of a request what kind of work it is for io-wq. I/O on regular files and block devices is bounded work, limited to min(SQ entries, 4 * CPUs) workers, for anything else that may block indefinitely the limit is RLIMIT_NPROC. Writes to a regular file are hashed by the inode, so that they are run one after the other instead of having a bunch of workers fight for the inode lock. Since the file is only assigned when a request is issued, a request that is queued without an attempt to issue it, REQ_F_FORCE_ASYNC, doesn't have one at that point. io_prep_async_work() takes that for "not a regular file or block device": - Reads and writes end up as unbound work. 32 reads with IOSQE_ASYNC from a slow block device on a ring with 4 SQ entries get 32 workers, not 4. - Writes aren't hashed. Four buffered writes to one file get four workers, three of which do nothing but wait for the lock. The late assignment is for the benefit of links, where a request may use a fixed file that one of its predecessors installs. By the time a request gets to io_queue_iowq() these are done, it is as good as issued. Assign the file there. io_assign_file() moves up and is forced inline. With a third caller the compiler may stop inlining it, and every request that is issued would pay for a call. IORING_OP_SPLICE is left as it is, unbound and not hashed. The file of the request is where it writes to, which says nothing about how long it takes: a splice from an empty pipe or a socket into a regular file waits for its input. As bounded work, four of these on a ring with 4 SQ entries keep all other bounded work of the task from running, and as hashed work one of them blocks all writes to that file. REQ_F_FORCE_ASYNC is not only set for IOSQE_ASYNC. What changes, and brings back what was the case before the file assignment was deferred: - Requests with IOSQE_ASYNC or IOSQE_IO_DRAIN, and all requests that are submitted while a drain is active: reads and writes on regular files and block devices count against the limit for bounded workers, including the one set with IORING_REGISTER_IOWQ_MAX_WORKERS, so fewer of them run at the same time. - Of these, writes to one regular file are run one at a time, unless they are O_DIRECT and the filesystem has FOP_DIO_PARALLEL_WRITE. - IORING_OP_FALLOCATE and IORING_OP_FTRUNCATE, which always go to io-wq, are run one at a time per file, and in turn with these writes. - The descriptor or the fixed file slot is looked up when the request is queued, not when a worker gets around to it, so closing or replacing it in between no longer has an effect on the request. Fixes: 6bf9c47a3989 ("io_uring: defer file assignment") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-43-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit d402e2da23b890042b34c7b5ef5259a19e9336d1 Author: Josef Bacik Date: Thu Oct 15 19:31:06 2026 +0000 io_uring/timeout: don't end a linked timeout with 1 Commit c74193430febdd120e35936dda3e5aeec3a54405 upstream. A linked timeout that fires cancels the request it is linked to, and completes with -ETIME if that worked out: ret = io_try_cancel(req->tctx, &cd, 0); ... io_req_set_res(req, ret ?: -ETIME, 0); io_try_cancel() returns 0 or an error, except for requests that are parked on ->futex_list or ->waitid_list. io_futex_cancel() and io_waitid_cancel() return the number of requests that they have cancelled, and io_try_cancel() hands it on. So the linked timeout of an IORING_OP_FUTEX_WAIT, IORING_OP_FUTEX_WAITV or IORING_OP_WAITID that it has cancelled completes with 1, each time. The waitid test of liburing checks for that 1, and has to be changed to expect -ETIME, which is what the linked timeout of any other request completes with. Fixes: f31ecf671ddc ("io_uring: add IORING_OP_WAITID support") Fixes: 194bb58c6090 ("io_uring: add support for futex wake and wait") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-42-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 17a03003bdd7614e8111634e3c4a7c9de96005b6 Author: Josef Bacik Date: Thu Oct 15 19:31:05 2026 +0000 io_uring/timeout: no ABS update of multishot timeouts Commit b088ae34d4da730dc647261b893a6eb622fe5e1f upstream. IORING_TIMEOUT_MULTISHOT can't be combined with IORING_TIMEOUT_ABS, prep rejects that, as the request is armed again with the same value each time it fires. IORING_TIMEOUT_UPDATE doesn't check and takes an absolute time for a multishot timeout. The first expiry is as asked for, but io_timeout_complete() then arms the timer with ->mode of the original request, i.e. it uses the absolute time as the interval. With CLOCK_MONOTONIC the next expiry is an uptime away, with CLOCK_REALTIME more than fifty years. Fail such an update with -EINVAL, and do so before the timeout is touched. Fixes: ea97f6c8558e ("io_uring: add support for multishot timeouts") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-41-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 97b8dc36ba3cb176b74a7929140aeaeb40d5cdc0 Author: Josef Bacik Date: Thu Oct 15 19:31:04 2026 +0000 io_uring/net: don't retry recv on uncommitted buffers Commit 8444de681a179e41a4eb92d5bbe770a0677074a6 upstream. Commit 41b70df5b38b ("io_uring/net: commit partial buffers on retry") commits ring provided buffers when a receive is going to be retried after a partial transfer, so that another request using the same buffer group can't be handed the memory the first one keeps receiving into. It commits the number of bytes received so far. That does the job for a single buffer which is consumed as a whole, but the request hangs on to all it has mapped: - For a buffer ring with IOU_PBUF_RING_INC only these bytes are consumed. The rest of the buffer stays at the head of the ring. - For a bundle only the buffers which hold these bytes are consumed, the others that were peeked stay in the ring. In both cases the next receive with that buffer group gets memory that the retried request is still going to write to, and whoever comes last overwrites the data of the other. What the retry fills isn't committed at all, as the buffer list isn't known any more at that point. So the kernel hands it out again later on, while the CQE tells the application that it was used, and for an incremental buffer that it is done with. The remainder can't be committed in advance: if the request ends up receiving less, there is no way to tell the application that more was consumed than the CQE accounts for. Complete the request with what it has got instead of retrying it. MSG_WAITALL is no longer honoured for receives into incremental buffers or multi buffer bundles, they can return a short count, with the request marked as failed like for other short MSG_WAITALL transfers. Fixes: 41b70df5b38b ("io_uring/net: commit partial buffers on retry") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-40-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit ee0bf68a63fe3d353f9c1850900c5e3718899ba2 Author: Josef Bacik Date: Thu Oct 15 19:31:03 2026 +0000 io_uring: don't pass restart error codes to userspace Commit e1a64bc64897f4c2fe460922c345f2ed8f889b5f upstream. -ERESTARTSYS and friends are instructions for the system call exit code, which either restarts the call or turns them into -EINTR. They are not supposed to be seen by applications. A request isn't a system call, nobody looks at what it returns before that is put into a CQE. The read/write and the networking opcodes convert these codes to -EINTR by hand. The others don't. An io-wq worker that is blocked in the open of a FIFO, a tty or some other device, or in a splice from an empty pipe, and that gets interrupted because the request is cancelled, completes it with -512: IORING_OP_OPENAT, IOSQE_ASYNC // FIFO without a writer IORING_OP_ASYNC_CANCEL // -EALREADY, worker is signalled CQE of the open: res == -512 There are more candidates, like whatever a file system or driver returns from ->flush(), ->fsync(), ->fallocate() or ->uring_cmd(), or breaking a lease or delegation. Take care of it in io_issue_sqe(), which all of them go through. The helper is moved from rw.c as is. The cost is a compare and a branch that is not taken per request. Fixes: 15b71abe7b52 ("io_uring: add support for IORING_OP_OPENAT") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-39-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 7299a32764ab875443b58581d929027954dcf24c Author: Josef Bacik Date: Thu Oct 15 19:31:02 2026 +0000 io_uring/openclose: keep the lock in io_pipe_fixed() Commit a3d7ee11289f657aa9af826d8f0844221292a27a upstream. io_pipe_fixed() installs both ends of the pipe in the fixed file table, drops ->uring_lock, and copies the slot numbers to the application. If that faults, it takes the lock again and removes whatever is in these slots. If the request is run by io-wq the lock really is dropped in between, and the slots are visible to everybody else. By the time the error path gets to them, they may have been closed and reused, by IORING_OP_CLOSE and a direct open or accept, or by IORING_REGISTER_FILES_UPDATE. Then a file is removed from the table that has nothing to do with the failed request. Do the copy with the lock held. Faulting in user memory under ->uring_lock is fine, the registration paths do it all the time. Fixes: 53db8a71ecb4 ("io_uring: add support for IORING_OP_PIPE") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-38-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit e3e19f7019847089302b50cf682c970d013fd223 Author: Josef Bacik Date: Thu Oct 15 19:31:01 2026 +0000 io_uring/sqpoll: let the thread sleep on disabled rings Commit d492dcbb3b81c1c2d0aabef41fa1307093f3a556 upstream. __io_sq_thread() doesn't submit anything as long as the ring has IORING_SETUP_R_DISABLED set. The check that keeps the SQPOLL thread from going to sleep while the SQ ring holds entries doesn't know about that. If an application fills in SQEs first and enables the ring later, the thread finds that there is work each time it is about to sleep, goes around, doesn't submit it, and so on. It uses up a CPU until the ring gets enabled. Entries in the SQ ring of a disabled ring are no reason to stay awake. io_register_enable_rings() clears the flag and then wakes the thread, and the flag is looked at after prepare_to_wait(), so the thread can't miss that. Fixes: 0298ef969a11 ("io_uring: clean R_DISABLED startup mess") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-37-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 21feda9465760a8e4487385deabdaeba9eb7ee0b Author: Josef Bacik Date: Thu Oct 15 19:31:00 2026 +0000 io_uring/kbuf: handle index wrap in last buffer check Commit 6e773b5161506a4d0e739eb3aa4667d28f20219c upstream. io_ring_buffer_select() flags the request with REQ_F_BL_EMPTY if it takes the last buffer of the ring. Head and tail are 16-bit counters, but "head + 1" is computed as an int. If head is 65535 and tail has wrapped to 0, that compares 65536 to 0 and the flag isn't set. A bundle receive relies on the flag to tell that there is no point in trying to get more data. Without it, it goes for another round, finds no buffer, and the request, which has received data, is marked as failed on account of the -ENOBUFS. That severs a link, and ends a multishot. Fixes: c3f9109dbc9e ("io_uring/kbuf: flag request if buffer pool is empty after buffer pick") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-36-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 51f4ee29bf0564e6ae7d3540693a9bc1dd2c8fd7 Author: Josef Bacik Date: Thu Oct 15 19:30:59 2026 +0000 io_uring/zcrx: fix buffer offsets in areas above 4GB Commit ab4536bf2e5617b04f2d9396f06393c7bbfb071d upstream. io_zcrx_queue_cqe() computes the offset of the buffer in its area from net_iov_idx(), which is an unsigned int, so the shift is done in 32 bits and only the result is widened to the u64 that goes into the CQE. An area can be up to 1TB in size. For every niov beyond the first 4GB the CQE points the application at a buffer 4GB, or a multiple of that, below the one which has the data. It reads the wrong data, and then returns a buffer it doesn't own via the refill queue while the real one is never given back. Fixes: 11ed914bbf94 ("io_uring/zcrx: add io_recvzc request") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-35-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit b779d47fdae809b68b93159c54e7e4b95959e5c6 Author: Josef Bacik Date: Thu Oct 15 19:30:58 2026 +0000 io_uring/napi: don't busy poll past an expired timeout Commit 21ed976c12f6b77d7ea028a4fe9a08e69a56179f upstream. __io_napi_busy_loop() caps the busy poll time to what is left until the timeout of the wait. The difference is negative if the timeout has passed already, which is what an application gets if it uses IORING_ENTER_ABS_TIMER with a time that isn't in the future any more. The min_t(u64, ...) turns that into a huge value, so the cap has no effect and the wait, which should return right away, spins for the full busy poll time first. Fixes: d5cce407e4f5 ("io_uring/napi: postpone napi timeout adjustment") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-34-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 981621d6eadcd232d4ea748aaeb046337f10d3c1 Author: Josef Bacik Date: Thu Oct 15 19:30:57 2026 +0000 io_uring/napi: make stale entries time out after 60s Commit 9c151648780d89e5446e5f89f3fa2e416c3bb911 upstream. With dynamic tracking a NAPI id that hasn't been seen for a minute is meant to be dropped from the list the ring busy polls. The timeout is added to jiffies, but it is defined as 60 * SEC_CONVERSION, and SEC_CONVERSION isn't the number of jiffies in a second. It is the internal factor of __timespec64_to_jiffies(), HZ scaled up by SEC_JIFFIE_SC bits. For HZ=1000 that is 1000 << 22, which puts the timeout at about eight years. On 32-bit the multiplication overflows. So entries are never removed, and every id that a socket of the ring ever had keeps being polled on each wait. Fixes: 8d0c12a80cde ("io-uring: add napi busy poll support") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-33-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit c1c41f740888cf0a4d4f2a60ebd80bf95650bfd2 Author: Josef Bacik Date: Thu Oct 15 19:30:56 2026 +0000 io_uring/wait: no early -ETIME if min_wait finds work Commit 83cce67298ab1f4d2e9205a49f56975475cdb154 upstream. When the min wait time is over and there is work to run, which for DEFER_TASKRUN includes task_work that was queued lazily and didn't wake the waiter, io_cqring_min_timer_wakeup() ends the wait via io_cqring_timer_wakeup(). That flags it as timed out. io_cqring_wait() runs the work, leaves the loop, and returns 0 if the CQ ring holds something. Running task_work doesn't have to result in a CQE, though. A poll wake up can turn out to be spurious, for instance if somebody else has read the data by the time the request is retried. The application then gets -ETIME once the min wait time has passed, no matter how far away the timeout it asked for is. Wake the task without declaring a timeout. From now on a single CQE satisfies the wait, like in the case where the timer is switched over to the normal timeout. If running the work doesn't post one, the loop in io_cqring_wait() goes back to sleep with the normal timeout. Fixes: 1100c4a2656d ("io_uring: add support for batch wait timeout") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-32-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit b0e0256dd45a536a1e7ba0aaefe413073e9535b8 Author: Josef Bacik Date: Thu Oct 15 19:30:55 2026 +0000 io_uring/timeout: use the right clock for ABS updates Commit dad82eb1805d76c555224e4030319e3844642b64 upstream. io_parse_user_time() converts an absolute time from the time namespace of the caller to the host's view, for the clock given by the flags. The flags of IORING_TIMEOUT_UPDATE and IORING_LINK_TIMEOUT_UPDATE can't name a clock, the updated timer keeps running on the clock the timeout was set up with. So the time is always converted as if it was CLOCK_MONOTONIC. In a time namespace with a monotonic offset, that moves the new expiry of a CLOCK_REALTIME timeout by that offset, although the clock isn't affected by time namespaces at all, and the one of a CLOCK_BOOTTIME timeout by the difference between the two offsets. Do the conversion when the timeout has been looked up and its clock is known. The update is run by the submitter, its SQPOLL thread or one of its io-wq workers, all of which share the time namespace. Fixes: 9cc6bac1bebf ("io_uring/timeout: honour caller's time namespace for IORING_TIMEOUT_ABS") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-31-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit ba1d5cf3c948c6e265e04df479e4bb102e56d51c Author: Josef Bacik Date: Thu Oct 15 19:30:54 2026 +0000 io_uring/timeout: keep repeat count on multishot update Commit 3dc2a32cd356ee382aa4a8c7d844c53b2c0ad83b upstream. IORING_TIMEOUT_UPDATE clears ->off of the timeout it updates, to turn a timeout that waits for a number of completions into a pure one. For a multishot timeout ->off isn't a sequence though, it tells whether the number of repeats is limited: io_timeout_finish() only looks at ->repeats if ->off is set. A multishot timeout that was set up to fire N times hence fires forever once its interval has been updated. Leave ->off alone for those, they are never treated as sequence timeouts anyway. Fixes: ea97f6c8558e ("io_uring: add support for multishot timeouts") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-30-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 4f229e93dffe6a2f4fbe1bc2c0807d3150a913fb Author: Josef Bacik Date: Thu Oct 15 19:30:53 2026 +0000 io_uring/timeout: don't skip over reached sequences Commit c6c0b3eba88c3b13ca11ea112404982e86d58f7a upstream. io_flush_timeouts() copes with the wrap of the CQ sequence by looking at the distance from ->cq_last_tm_flush. That relies on the target of each pending timeout being at or ahead of the last flush. io_timeout() breaks that. It sets ->cq_last_tm_flush to the current sequence, on the assumption that nothing has been posted since the last flush. That isn't true. A multishot request posts its CQEs right when it is issued or retried, but they are committed, and the timeouts flushed, at the end of the submission or task_work run: X: IORING_OP_TIMEOUT, off == 1 target_seq == N + 1 io_submit_sqes() multishot recv, posts two CQEs sequence == N + 2 Y: IORING_OP_TIMEOUT, off == 100 ->cq_last_tm_flush = N + 2 io_flush_timeouts() X: events_needed = N + 1 - (N + 2) // 0xffffffff events_got = 0 // not yet X isn't completed although the completions it was waiting for have happened, only its timer can end it now. And as the list is sorted and the walk ends at the first timeout that isn't due, it gets in the way of the ones behind it. The insertion sort has the same problem, it measures the distance of the pending targets from the current sequence. ->cq_last_tm_flush only needs to be brought up to date if no timeout is waiting for a sequence, as io_flush_timeouts() isn't called before the first one is set up. Otherwise it is current enough. Timeouts with a sequence come first in the list, so looking at the first entry will do. Use the last flush as the base for the insertion sort as well. Fixes: f010505b78a4 ("io_uring: flush timeouts that should already have expired") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-29-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 40ce56dcf1a78ea5d721b99670c3ab05327be136 Author: Josef Bacik Date: Thu Oct 15 19:30:52 2026 +0000 io_uring/fdinfo: fix CQE dump of CQE32 rings Commit e4507a62c247081fffe299622c04c6f7c3d98e92 upstream. Since the CQ ring dump was changed to deal with IORING_SETUP_CQE_MIXED, a 32b CQE is taken to occupy two entries of the ring. That is how it works for mixed rings. With IORING_SETUP_CQE32 each entry has twice the size: head and tail count 32b CQEs, and entry n is found at cqes[n << 1]. For such rings the dump reads cqes[n], and skips every other entry. It shows half of the pending CQEs at best, and what it shows is taken from the wrong place unless the head happens to be at the start of the ring. The extra data of the last entry isn't shown at all, as the check that keeps the dump of a mixed ring from reading beyond the array is applied to these rings as well. [ 7.2: without the last-slot check, which 7.2 does not have ] Fixes: 82ceb7fcc5ff ("io_uring/fdinfo: handle mixed sized CQEs") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-28-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Co-Authored-By: Claude Fable 5.1 Signed-off-by: Greg Kroah-Hartman commit a3bc9b2e7b3c09b9a19c87d6e798a924cf630a9b Author: Josef Bacik Date: Thu Oct 15 19:30:51 2026 +0000 io_uring/register: keep 128b SQEs intact on resize Commit 0feb5bcd2d4261d656729892aef4d643461ea5ac upstream. On a ring with IORING_SETUP_SQE_MIXED a 128b SQE takes two entries that have to be contiguous in memory, io_init_req() fails it with -EINVAL if it starts on the last entry of the SQ ring. The entry a pending SQE ends up in after IORING_REGISTER_RESIZE_RINGS depends on the size of the new ring, and nothing takes care of that. With four entries instead of eight, a 128b SQE which was fine in entries 3 and 4 is split up between entry 3 and entry 0. The resize succeeds, and the request fails later on. If the ring has an SQ array, the second half is also taken from the wrong place unless the array happens to map the following index to the following entry. The submission side doesn't look at the array for the second half. Do what is done for 32b CQEs on IORING_SETUP_CQE_MIXED rings: fail the resize with -EOVERFLOW if a 128b SQE would wrap. And copy the second half from where it is. Fixes: 1cba30bf9fdd ("io_uring: add support for IORING_SETUP_SQE_MIXED") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-27-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 4b7aaf3ce9d53c27281da3216c200547a621e315 Author: Josef Bacik Date: Thu Oct 15 19:30:50 2026 +0000 io_uring/register: handle index wrap when resizing Commit e8dee8d5498c710a22a84076d8ecb661627032b4 upstream. The ring heads and tails are free running 32-bit counters. io_register_resize_rings() takes that into account when it checks whether the pending entries fit into the new rings, but then copies them with for (i = old_head; i < tail; i++) If the tail has wrapped and the head hasn't, that copies nothing. The new rings get the old head and tail, so the application finds whatever is in the freshly allocated memory, zeroes, in place of the CQEs it hasn't reaped yet, or the kernel in place of the SQEs that have not been submitted yet. A ring that completes a million requests per second gets there after little more than an hour. Loop until the tail is reached, the number of iterations has been bounded by the size of the new ring at that point. Fixes: 79cfe9e59c2a ("io_uring/register: add IORING_REGISTER_RESIZE_RINGS") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-26-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 37d8caedace34a8ba198ac7e6e51910e1da05209 Author: Josef Bacik Date: Thu Oct 15 19:30:49 2026 +0000 io_uring/register: don't wrap 32b CQEs when resizing Commit 0edf1930b8ed3c398d3cbdbafd8f72dbd1a3d9e4 upstream. On an IORING_SETUP_CQE_MIXED ring, a 32b CQE uses two contiguous CQ entries and is never allowed to wrap the end of the ring. When posting, io_cqe_cache_refill() inserts a dummy IORING_CQE_F_SKIP entry if only one entry is left before the ring wraps, and the application relies on that to read a 32b CQE as one contiguous 32 byte slot. io_register_resize_rings() copies pending CQEs to the new ring one 16b entry at a time, placing entry i at i & (new_entries - 1). When the CQ ring is shrunk, a pending 32b CQE that was contiguous in the old ring can land with its first half on the last entry of the new ring and its second half on entry 0. For example, with 8 old entries, 4 new entries and a 32b CQE at index 3, the halves end up at entries 3 and 0. The application then reads the extra data of that CQE from past the end of the CQ array, and the CQE content it sees is wrong. This was found by code inspection and has not been reproduced. Fix by failing the resize with -EOVERFLOW if a pending 32b CQE would wrap the new ring, the same way the resize fails when the pending CQEs don't fit. Fixes: e26dca67fde1 ("io_uring: add support for IORING_SETUP_CQE_MIXED") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-25-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit a08a28e125b34e130a62830fe141ff2d6c9de6d6 Author: Josef Bacik Date: Thu Oct 15 19:30:48 2026 +0000 io_uring: don't split 128b SQEs at the SQPOLL cap Commit 79956bb39b4d372eb146955ea159ecb6fba37f5b upstream. If an SQPOLL thread serves more than one ring, __io_sq_thread() submits no more than IORING_SQPOLL_CAP_ENTRIES_VALUE entries per ring and round, for fairness. On a ring with IORING_SETUP_SQE_MIXED the cap may fall between the two entries of a 128b SQE. io_init_req() finds that just one entry is left of what it was asked to submit and fails the request with -EINVAL, although the second entry is sitting in the SQ ring. In the next round the second entry is taken for an SQE of its own, so on top of the failed request the application gets whatever its payload amounts to when it is read as an SQE. The application has done nothing wrong, and it can't know where the cap is going to be applied. For SQPOLL, let the request have its second entry if the SQ ring holds it, in excess of the number of entries that were to be submitted. Fixes: 1cba30bf9fdd ("io_uring: add support for IORING_SETUP_SQE_MIXED") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-24-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 89f3d0dfef794bace554794e3c264b1e1cd727e8 Author: Josef Bacik Date: Thu Oct 15 19:30:47 2026 +0000 io_uring: drop the CQE cache if a refill fails Commit 78f9c9cffef02a290b6e97672aaa9c75df0d21da upstream. ->cqe_cached and ->cqe_sentinel delimit the CQ entries which can be handed out without looking at the ring. io_get_cqe_overflow() only calls io_cqe_cache_refill() if that range is too small. Without IORING_SETUP_CQE_MIXED too small means empty. With it, a 32b CQE needs two entries, and one may be left. If the refill then fails, the entry stays in the cache, and the next 16b CQE gets it without any checks. That CQE overtakes the 32b one, which went to the overflow list. The test for IO_CHECK_CQ_OVERFLOW_BIT, which is there to prevent that, is part of the refill: CQ ring with 4 entries, head == 0 16b CQE, 32b CQE tail == 3, one entry cached 32b CQE no room, the CQE overflows 16b CQE written to entry 3, ahead of the 32b CQE Forget about the rest of the cache when it is refilled. Fixes: e26dca67fde1 ("io_uring: add support for IORING_SETUP_CQE_MIXED") Reported-by: Leonid Ovchinnikov Closes: https://lore.kernel.org/io-uring/CAECZYD5OMD9vMTzjteFMMP40Dn+SWKj9SFj=43U6uO3XMnxJbw@mail.gmail.com/ Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-23-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 27cee9c0f7789ff74bf262989928909d80d1ac70 Author: Josef Bacik Date: Thu Oct 15 19:30:46 2026 +0000 io_uring/futex: allocate private hash before waiting Commit 3eff68bce25584cded3fb6ec033ff9a1c9913d04 upstream. The futex code relies on there being no private futex waiters by the time the first private hash of a mm is installed: __futex_pivot_hash() moves the waiters from the previous private hash to the new one, but it does not look for them in the global hash which is used as long as there is no private hash. That is why every clone(CLONE_VM) allocates it up front: a task which is alone in its mm cannot block in futex_wait() and create a thread or invoke prctl(PR_FUTEX_HASH) at the same time. IORING_OP_FUTEX_WAIT and IORING_OP_FUTEX_WAITV do not block the submitter though: io_uring_enter() io_futex_wait() futex_wait_setup() // no private hash: queued in the global hash pthread_create() futex_hash_allocate_default() __futex_pivot_hash() // first private hash, nothing is moved futex(FUTEX_WAKE_PRIVATE) futex_hash() // private hash: no waiter The wakeup is lost and the request only completes when it is cancelled. prctl(PR_FUTEX_HASH, PR_FUTEX_HASH_SET_SLOTS, n) instead of pthread_create() has the same effect. Allocate the default private hash when a wait request on a private futex is prepared, which happens in the context of the submitter. This is a RCU protected pointer check once the hash exists. A failure to allocate it fails the request with -ENOMEM. Fixes: 80367ad01d93 ("futex: Add basic infrastructure for local task local hash") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-22-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 34eef9b254f8c970f06b94c7738db0ecb67fb30a Author: Josef Bacik Date: Thu Oct 15 19:30:45 2026 +0000 io_uring/poll: drop stale double poll entry on rearm Commit f2dbe87968a3aef74a11c025776d293b80964d3f upstream. A file that uses two wait queues for poll, e.g. a tty, gets a second io_poll entry, which IORING_OP_POLL_ADD keeps in ->async_data until the request is freed. io_poll_add() can run more than once for a request though: IORING_OP_POLL_REMOVE with one of the update flags disarms and rearms it, and io_poll_task_func() reissues it if a wake up turns out to be spurious. The second poll_wait() of the new round then finds *poll_ptr set, with ->head cleared by the disarm, and __io_queue_proc() fails what it takes for a third wait queue with -EINVAL. For an update that means that the poll request is completed with -ECANCELED, while the update itself reports success. For a spurious wake up, e.g. tcsetattr() on the polled tty, the poll request fails with -EINVAL. Free the old entry and clear the poll flags before arming, which is what io_arm_apoll() does for internal polling. All entries have been removed from their wait queues at that point. Fixes: b69de288e913 ("io_uring: allow events and user_data update of running poll requests") Fixes: 6e5aedb9324a ("io_uring/poll: attempt request issue after racy poll wakeup") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-21-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 4ce1acd22448310a124d40fcf71dea49e31625c6 Author: Josef Bacik Date: Thu Oct 15 19:30:44 2026 +0000 io_uring/wait: look at all CQ waiters on wake up Commit 356f92f2412686f822383b784fa102924407457a upstream. io_wake_function() returns -1 if the waiter it is called for doesn't have enough events yet, which makes __wake_up_common() stop walking the wait queue. That assumes that the waiters behind it can't be satisfied either. Each waiter has its own target though, computed from the min_complete it passed in: thread A: io_uring_enter(GETEVENTS, min_complete = 10) // queued first thread B: io_uring_enter(GETEVENTS, min_complete = 1) one CQE is posted: A isn't woken, and B isn't even considered B keeps sleeping with the event it asked for sitting in the CQ ring, until A's demand is met or a timeout or signal ends the wait. Return 0 to carry on with the next waiter. Only waiters whose condition is met are woken, like before. Fixes: bda521624e75 ("io_uring: make CQ ring wakeups be more efficient") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-20-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 92bfadcbe0a9a001beb5e58a2abaf040b873deb6 Author: Josef Bacik Date: Thu Oct 15 19:30:43 2026 +0000 io_uring: wake ring pollers for local task_work Commit b88902d8745a701d24275d012f0830e3628e2a22 upstream. With IORING_SETUP_DEFER_TASKRUN completions are queued as local work, and only become CQEs when the submitter enters the kernel to wait for events. For an application that finds out via poll() or epoll on the ring fd when it is time to do that, io_uring_poll() reports EPOLLIN if local work is pending. That only helps if the work is there at the time of the call. When DEFER_TASKRUN was added, io_req_local_work_add() did io_cqring_wake(), which took care of tasks sleeping in poll() as well. Then polling got a wait queue of its own, ->poll_wq, and the wake up in io_req_local_work_add() was replaced by waking the submitter directly. Since then ->poll_wq is only woken when CQEs are committed, which for such a ring is done by the very task that is waiting to be told to: submitter completion poll(ring_fd) io_uring_poll() // nothing yet schedule() io_req_local_work_add() // ->cq_wait_nr: nobody waits // sleeps on Wake ->poll_wq when the first entry is put on the work list. Like for the eventfd, that must not be done right away if the work is queued from a wake up handler, as the chain of wake ups could get back to the wait queue whose lock is held. Use an RCU callback in that case, with a reference to the ring. Fixes: 130bd686d9be ("io_uring: waitqueue-less cq waiting") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-19-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 11ca374f18cd4243bce34893a58412a57fad4b8c Author: Josef Bacik Date: Thu Oct 15 19:30:42 2026 +0000 io_uring/cancel: let CANCEL_ALL find parked requests Commit fe6ca13c98b1223f6005f32913ebd32e4df64b77 upstream. A cancellation with IORING_ASYNC_CANCEL_ALL or _ANY goes on until it finds nothing more. A request may stay where it was found for a while after it has been cancelled, so each one that matches is stamped with the sequence number of the cancellation, and is skipped from then on. io_cancel_cb() does so for io-wq work as well. There is no need to: pending work is taken off the list, and the workers are looked at once per call. And it does harm. A request that is run by io-wq may be parked while it still is the current work of the worker: poll gets armed for it, or it is put on ->futex_list or ->waitid_list. io-wq worker io_try_cancel() io_wq_submit_work() io_arm_poll_handler() io_poll_add_hash() io_async_cancel_one() io_cancel_cb() stamps the request sets IO_WQ_WORK_CANCEL = -EALREADY io_poll_cancel() skips the request = -ENOENT io_assign_current_work(NULL) The flag comes too late for the worker to see it, and the search that would have found the request takes it for done. The cancellation returns, and the request stays. With a multishot poll, IOSQE_ASYNC, and IORING_OP_ASYNC_CANCEL with IORING_ASYNC_CANCEL_ALL up to 63us later, that happens to 4 of 1,000 requests (4 CPUs, KVM). Take the stamp off again if the work matches. If the request is found on one of the lists it gets a new one, and io_cancel_cb() doesn't match it a second time. __io_async_cancel() tries the io-wq of the caller twice. So far the stamp has kept it from counting work that is running there the second time. Skip it instead, so that the number that is returned stays the same. Fixes: 8e29da69fead ("io_uring: add support for IORING_ASYNC_CANCEL_ALL") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-18-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 8ed3d8d7d9ec65b602c1e4f5cfd460ca004f8b9b Author: Josef Bacik Date: Thu Oct 15 19:30:41 2026 +0000 io_uring/cancel: recognize "no timeout" on 32-bit Commit 3774e4d4c8533da2caae95556c4d635c7c4f45b8 upstream. IORING_REGISTER_SYNC_CANCEL waits without a limit if both fields of io_uring_sync_cancel_reg::timeout are -1. io_sync_cancel() compares them with -1UL. They are of type long long. With a long of 32 bits -1UL is 4294967295, which is what it stays when it is converted for the comparison, and the fields are never equal to it. So { -1, -1 } is taken for a timeout. ktime_set() takes the nanoseconds as an unsigned long, so what becomes of it is -1s + 4294967295ns = 3.29s. If the request is being run by io-wq and can't be interrupted, the call returns -ETIME after that time, rather than waiting for the request to be gone. On i386, with a buffered read from a null_blk device that takes 4s to complete a request, IOSQE_ASYNC, that is what each call does. Fixes: 78a861b94959 ("io_uring: add sync cancelation API through io_uring_register()") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-17-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit a93aa29cbcbed7d38657d00f1ad240cd0d9fac55 Author: Josef Bacik Date: Thu Oct 15 19:30:40 2026 +0000 io_uring/cancel: don't forget that the work was running Commit 8aa0e5db4845fb5ac11abe40d7b1cac0656c3ee3 upstream. If the request to be cancelled is being run by io-wq, io_async_cancel_one() flags the work, signals the worker and returns -EALREADY. Since commit ccbf726171b7 ("io_uring: perform poll removal even if async work removal is successful") io_try_cancel() goes on regardless, as poll may be armed for the request. If it isn't, and nothing else turns up, the result is the -ENOENT of the last search. __io_async_cancel() makes up for that. On -ENOENT it tries the io-wq of every task that uses the ring, and that of the caller is one of them. So the work is found a second time, and the result is -EALREADY after all. Unless the worker is done with it by then. If the work is flagged before io_wq_submit_work() looks at IO_WQ_WORK_CANCEL, all the worker does is to queue the request to be failed by task work. Then it isn't anywhere to be found until the task work has run, and - IORING_REGISTER_SYNC_CANCEL returns -ENOENT, - IORING_OP_ASYNC_CANCEL completes with -ENOENT, for a request that they have just cancelled, and that completes with -ECANCELED. With a read of an empty pipe, IOSQE_ASYNC, and a sync cancel right behind it, 1 to 3 of 1,000,000 calls return -ENOENT, and 50 to 110 with IORING_SETUP_DEFER_TASKRUN (4 CPUs, KVM). Nothing is written to the pipe, and all of these reads end with -ECANCELED. Remember that the work was found. IORING_OP_ASYNC_CANCEL completes with -EALREADY, as it does if the worker is a bit slower. For IORING_REGISTER_SYNC_CANCEL it means to try again, which ends with 0. The result of a linked timeout, which calls io_try_cancel() by itself, and the number that is returned for IORING_ASYNC_CANCEL_ALL and _ANY don't change. Fixes: ccbf726171b7 ("io_uring: perform poll removal even if async work removal is successful") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-16-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 6773939d93fa6c983acf63ca45de5ca2318d8781 Author: Josef Bacik Date: Thu Oct 15 19:30:39 2026 +0000 io_uring: don't lose a cancel as io-wq parks a request Commit 5cb049d2203d7a81a1c1a737c8df4575e71078e3 upstream. IORING_OP_FUTEX_WAIT, IORING_OP_FUTEX_WAITV and IORING_OP_WAITID don't block. They put the request on a list of the ring, where a wake up or a cancellation finds it, and return IOU_ISSUE_SKIP_COMPLETE. They do so as well if they are run by io-wq, e.g. because of IOSQE_ASYNC. io_wq_submit_work() looks at IO_WQ_WORK_CANCEL once, before it issues the request. Somebody who wants to cancel the request after that, and before it is on the list, finds it running: io-wq worker io_try_cancel() io_wq_submit_work() IO_WQ_WORK_CANCEL isn't set io_async_cancel_one() sets IO_WQ_WORK_CANCEL signals the worker = -EALREADY ... io_futex_cancel() = -ENOENT ... io_futex_wait() adds the request to ->futex_list = IOU_ISSUE_SKIP_COMPLETE Nobody looks at the flag again. IORING_OP_ASYNC_CANCEL returns -EALREADY and the request stays. It is the problem that "io_uring/poll: don't lose a cancel as io-wq arms poll" deals with for poll. Look at the flag once the request is on the list, and cancel it if it is set. The lists are searched and the requests are added with ->uring_lock held, and the flag is set before the search. So either the search finds the request, or it came first and the flag is seen. What the canceller is told doesn't change, see the other patch, also for IORING_ASYNC_CANCEL_ALL and _ANY. Fixes: 194bb58c6090 ("io_uring: add support for futex wake and wait") Fixes: 8f350194d5cf ("io_uring: add support for vectored futex waits") Fixes: f31ecf671ddc ("io_uring: add IORING_OP_WAITID support") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-15-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit f1a2919e0d834fcbb9372ff29738360fe0ca0acb Author: Josef Bacik Date: Thu Oct 15 19:30:38 2026 +0000 io_uring/poll: don't lose a cancel as io-wq arms poll Commit 6d2d4c2530e95ff6ddeb98195d49d99dbca14369 upstream. A request on a pollable file that was forced to io-wq is issued without blocking there. If that ends in -EAGAIN, io_wq_submit_work() arms poll for it and returns. It looks at IO_WQ_WORK_CANCEL once, before it issues the request. Somebody who wants to cancel the request in between finds it running: io-wq worker io_try_cancel() io_wq_submit_work() IO_WQ_WORK_CANCEL isn't set io_issue_sqe() = -EAGAIN io_async_cancel_one() sets IO_WQ_WORK_CANCEL signals the worker = -EALREADY io_poll_cancel() = -ENOENT ... = -ENOENT io_arm_poll_handler() = IO_APOLL_OK return Nobody looks at the flag again, and the signal doesn't matter to a worker that doesn't sleep. The request is armed as if nothing had happened: - a linked timeout completes with -ENOENT, and the request it was meant to limit waits for as long as it takes. - IORING_OP_ASYNC_CANCEL returns -EALREADY, and the request stays. - IORING_REGISTER_SYNC_CANCEL waits for the request to complete if it gets -EALREADY. Since commit 5623eb1ed035 ("io_uring/tctx: add separate lock for list of tctx's in ctx") it doesn't sleep but spins, so its next attempt finds the armed poll. The same goes for a multishot poll, recv or accept that is armed from io-wq. With a read of an empty pipe, IOSQE_ASYNC, and a linked timeout of 5ms or a cancel in the same submission, in a guest that is slow enough (TCG, KASAN), the request is lost in 21 of 24 rounds of 40 or 100 attempts. While it arms poll from io-wq, __io_arm_poll_handler() owns the request. Look at the flag once the request is in the hash table. The table is searched and the request is added with ->uring_lock held, and the flag is set before the search. So either the search finds the request, or it came first and the flag is seen. With the request marked as cancelled, the task work that is queued, right away or because ownership can't be released, fails it with -ECANCELED. What the canceller is told doesn't change. In this window a linked timeout still completes with -ENOENT rather than -ETIME, a cancel request with -EALREADY, and one with IORING_ASYNC_CANCEL_ALL or _ANY with a count that doesn't include the request. But the request is cancelled, so what is new is a request that completes with -ECANCELED after such a result. This doesn't help in two cases: - IORING_ASYNC_CANCEL_ALL and _ANY, if the work is flagged after the test added here, while io-wq still has it as the current work of the worker. io_cancel_cb() stamps it with the sequence of the cancel, so the search of the hash table on behalf of the same cancel takes it for done and skips it. - requests that are parked elsewhere by io-wq, like IORING_OP_FUTEX_WAIT and IORING_OP_WAITID. They have the same problem. Fixes: 90fa02883f06 ("io_uring: implement async hybrid mode for pollable requests") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-14-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 300665d10747810ea605a36f9d0802a161c1f146 Author: Josef Bacik Date: Thu Oct 15 19:30:37 2026 +0000 io_uring/io-wq: let cancel find picked up hashed work Commit e9efeb4fc97b0879512a3c3b995ed448f2487e12 upstream. io_get_next_work() takes all the work items that share a hash off the work list in one go, and the worker runs them one after the other. Only the one it is working on is visible as ->cur_work. The others are neither on the list nor current, so io_wq_cancel_cb() can't find them. IORING_OP_ASYNC_CANCEL and IORING_REGISTER_SYNC_CANCEL return -ENOENT for a request that has not even been started, e.g. a buffered write queued behind another write to the same file that is blocked, and the request is carried out. The cancellation done for a task or ring that goes away gets there in the end, as it keeps trying, but it has to wait for each item to become current. Keep a pointer to the remainder of the chain next to ->cur_work, and have io_wq_worker_cancel() look at it. The items on it belong to the worker, which needs ->lock to move on to the next one. Matching work is flagged with IO_WQ_WORK_CANCEL and fails with -ECANCELED when its turn comes. The worker isn't signalled for it, that would interrupt the unrelated work it is busy with. As the request isn't gone yet, it is reported like running work, with -EALREADY. Fixes: 86f3cd1b589a ("io-wq: handle hashed writes in chains") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-13-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 50eb342b1c452ea6d21edfb0aa2cffbf06c9f1c9 Author: Josef Bacik Date: Thu Oct 15 19:30:36 2026 +0000 io_uring/io-wq: don't exit with a creation pending Commit e231d2e8f80d3649ecd3851b7f2abdfbe606c04b upstream. A worker that is the last one to go to sleep while work is pending asks for another worker. io_wq_dec_running() bumps ->nr_running on behalf of the new worker and queues create_worker_cb() as task_work for the owner of the io-wq. Until that has run, everybody else takes it that a worker is running which is going to look after the work list: io_wq_enqueue() doesn't create one, and other workers that go to sleep don't find themselves to be the last. If the worker that made the request exits, io_worker_exit() cancels the task_work, and io_worker_cancel_cb() drops ->nr_running without a look at the work list: worker A worker B io_wq_enqueue() blocks, work is pending nr_running++ queues create_worker_cb() wakes, handles the work idle timeout, decides to exit picks up work 1 queues work 2 B is on the free list still: wakes B work 1 blocks nr_running 2 -> 1 io_worker_exit() io_worker_cancel_cb() nr_running 1 -> 0 Work 2 has to wait for work 1, although there is room for more workers. It takes an owner that doesn't get around to run its task_work for the time of the idle timeout, less if the worker exits because its affinity was changed or the io-wq is marked exit-on-idle. Let the worker stay around until its request has been dealt with. Only the worker itself sets ->create_state, so it can't be set behind its back once it has decided to leave. Fixes: 1d5f5ea7cb7d ("io-wq: remove worker to owner tw dependency") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-12-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 18bf64cb927aa3743bc1e02bf57e74bbbbe0d6f2 Author: Josef Bacik Date: Thu Oct 15 19:30:35 2026 +0000 io_uring/io-wq: don't strand work when a worker exits Commit ad41f325be1271b22d99353819c90f0d5e45acc8 upstream. A worker exits if its idle timeout fired and it isn't the last one, if its affinity was changed, or if the wq is marked exit-on-idle. It decides so after it found the work list empty, and until io_worker_exit() is through it stays on the free list and in ->nr_running: worker io_wq_enqueue() io_acct_run_queue() == false queues work io_acct_activate_free_worker() wake_up_process(worker) // found one, nothing to create acct->nr_workers-- break io_worker_exit() Nobody is going to look at the work. If other workers exist, it has to wait until one of them finishes what it is blocked on or hits its idle timeout. If not, which is possible for the affinity and the exit-on-idle case, it sits there until something else is queued. For exit-on-idle it takes a task that closes its last ring and sets up a new one while the workers are on their way out. Make the worker take itself off the free list when it gives up its ->nr_workers slot, then leave ->nr_running, and only then take the final look at the work list. That look and queueing work are both done under acct->lock, so either the worker sees the work and stays, or the enqueue side sees that the worker is gone and wakes or creates another one. If it wants to stay and finds that its slot has been taken, a worker has been created in its place, which is going to look at the work list, and it can leave after all. Fixes: 38aa434ab933 ("io_uring/io-wq: add exit-on-idle state") Fixes: 05c5f4ee4da7 ("io-wq: get rid of FIXED worker flag") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-11-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit d0494e51aa02b405655538ac8b9f3b1ed7fa7d29 Author: Josef Bacik Date: Thu Oct 15 19:30:34 2026 +0000 io_uring/io-wq: don't go to sleep on pending work Commit f19c38feb96802ee53a99649d740d2bf630bcce7 upstream. io_wq_enqueue() wakes a worker off the free list. If there is none it creates one, unless ->nr_running says that a worker is running. That relies on whoever takes ->nr_running to zero to check for pending work, which io_wq_dec_running() no longer does for a worker without ->cur_work, on the grounds that an idle worker is either woken right away or someone else has created a worker. Neither has to be the case: worker A (idle) worker B (idle) io_wq_enqueue() woken for work 1 woken (e.g. timeout) takes work 1 work list is empty queues work 2 B is first on the free list: wake_up_process(B) leaves the free list work 1 blocks nr_running 2 -> 1 schedule_timeout() nr_running 1 -> 0 ->cur_work == NULL Work 2 isn't looked at until work 1 is done or A's idle timeout fires, five seconds later. ->cur_work is also NULL while a worker blocks on ->uring_lock in io_wq_put_work(), with the same result if it was the only one running. Let the idle worker leave ->nr_running by itself before it takes its last look at the work list. Both that and queueing work are done under acct->lock, so either the worker sees the work, or the enqueue side sees that it can't count on this worker. This doesn't bring back the request for a new worker which the check was meant to avoid, and as idle workers don't get to io_wq_dec_running() any more, the check can go. Fixes: 8343cae362e1 ("io_uring/io-wq: ignore non-busy worker going to sleep") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-10-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 8c94677e2b6145393d4c129fe0f5364b0e5000a5 Author: Josef Bacik Date: Thu Oct 15 19:30:33 2026 +0000 io_uring/mpscq: disable preemption while pushing Commit 68a95c3968895cd198f42baadb23e635fecc5a66 upstream. A push makes the node the new tail, and then links it to its predecessor. In between the queue isn't empty, but mpscq_pop() can't get at the node. The consumers deal with that by trying again, with a cond_resched() in the loop to give a preempted producer the chance to finish. cond_resched() doesn't do that if the consumer has a higher priority than the producer and both have to use the same CPU: producer (SCHED_OTHER) consumer (SCHED_FIFO), same CPU io_req_local_work_add() mpscq_push() prev = xchg(&q->tail, node) io_cqring_wait() __io_run_local_work() again: cond_resched() // nothing to do mpscq_pop() // NULL // work pending, no events yet goto again The producer can be any task that is able to queue task_work for the ring: an io-wq worker, or somebody who sends a message with IORING_OP_MSG_RING. The consumer spins, holding ->uring_lock, until the scheduler takes pity on the fair class, which takes about a second with the default settings of the fair server, and forever without it. The loops in tctx_task_work_run() and io_cancel_local_task_work() have the same problem. Don't allow preemption between the two steps. That bounds the time the window is open on the CPU of the producer by what interrupts take, and a consumer that waits for that is on another CPU. The cost is a preempt_disable()/preempt_enable() pair per push. Fixes: 50cb44bd0d5f ("io_uring/mpscq: add lockless multi-producer, single-consumer FIFO queue") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-9-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit aaa06bb9fc03384f7cf2dbc3e43f2939cc7791b2 Author: Josef Bacik Date: Thu Oct 15 19:30:32 2026 +0000 io_uring/cancel: rearm local work wake up before sleep Commit 76a82c666abf155ad52d5f41043eac77dd3e41d8 upstream. io_uring_try_cancel_requests() sets ->cq_wait_nr to 1 so that the next io_req_local_work_add() wakes the cancelling task. The add that takes the count to zero is the only one that wakes, later ones see a count <= 0 and return. io_uring_cancel_generic() however only becomes TASK_INTERRUPTIBLE much later, so the one wake up can be spent on a running task. That is fine as long as the entry of that add is still queued, as the io_local_work_pending() check then skips the sleep. It is not if the add was delayed after queueing its entry: CPU0 (submitter in exec or exit) CPU1 (io_req_local_work_add) mpscq_push(req A) io_uring_try_cancel_requests() io_run_local_work() // runs req A io_uring_try_cancel_requests() atomic_set(&ctx->cq_wait_nr, 1) atomic_read(&ctx->cq_wait_nr) == 1 count taken, wake_up_state() // task is running, no-op prepare_to_wait(TASK_INTERRUPTIBLE) io_local_work_pending() == false schedule() mpscq_push(req B) cq_wait_nr <= 0, return Only the sleeping task can run req B, so it is stuck in execve() until a signal happens to wake it. An exiting task doesn't even get that. Arm the count again once the task state is set and before the work list is checked, for the rings whose local work this task is the one to run. [ 7.2: the loop still has the WARN_ON_ONCE() that commit fddcf4c5321b replaced in 7.3; the new check makes it unreachable, so it is dropped ] Fixes: d46ab2c98aba ("io_uring: switch local task_work to a mpscq") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-8-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Co-Authored-By: Claude Fable 5.1 Signed-off-by: Greg Kroah-Hartman commit d29be0d1a67241ce5a87148ad26396a89825d89e Author: Josef Bacik Date: Thu Oct 15 19:30:31 2026 +0000 io_uring: don't lose the DEFER_TASKRUN wake up Commit 8257f6ec45ee07c6ed933c59d8982b03cc7bc1ca upstream. Since local task_work moved to the mpscq, ->cq_wait_nr is a credit: the add that counts it down to zero, or exchanges it for IO_CQ_WAKE_INIT, issues the only wake up of the wait cycle, and every later add returns early as the count is <= 0. An add looks at the count after it has queued its entry, so an add that gets delayed in between can take the credit of a later cycle. That was assumed to cause an early wake up at worst, but the waiter publishes the count before it changes its task state: CPU0 (submitter, io_cqring_wait) CPU1 (io_req_local_work_add) mpscq_push(req A) io_run_local_work() // runs req A atomic_set(&ctx->cq_wait_nr, 1) atomic_read(&ctx->cq_wait_nr) == 1 atomic_xchg(.., IO_CQ_WAKE_INIT) wake_up_state(.., TASK_INTERRUPTIBLE) // still TASK_RUNNING, no-op set_current_state(TASK_INTERRUPTIBLE) io_local_work_pending() == false schedule() mpscq_push(req B) cq_wait_nr <= 0, return The task now sleeps with req B queued, until a signal or a timeout ends the wait. Set the task state first and only then make the count visible, so that whoever takes the credit finds a task it can wake. The full barrier that orders the count against the check of the work list is kept. Do the same for the wait of the BPF driven loop, which copied the sequence. Fixes: d46ab2c98aba ("io_uring: switch local task_work to a mpscq") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-7-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit abdc57c7ebabaaf0b622178a6a85834e1e54ff80 Author: Josef Bacik Date: Thu Oct 15 19:30:30 2026 +0000 io_uring: don't flood a cancelling task with task_work Commit 22e643e61b9b6299b42ff7d88bab48bccada8a5d upstream. io_ring_exit_work() asks each task which has used the ring to remove its tctx node, by task_work, and repeats that until the list is empty. io_tctx_exit_cb() doesn't touch the node while the task is in io_uring_cancel_generic(), which removes all the nodes at its end. So the work item queues the callback again as soon as it has completed, for nothing. That is more than a waste of time. task_work_run() only returns when it finds the list empty. If the rings are released from inside the cancellation, because requests of other rings held the last references to their files, up to 64 work items keep sending callbacks to the one task which is running them. It never finds the list empty, never gets back to the cancellation, the nodes stay, and so it goes on: a task with 60 pairs of rings which poll each other ran 280000 callbacks in 5 seconds, 20 seconds after exit_group(). It ends when all the senders happen to be late at the same time. Until then the task keeps a CPU and the work items busy. An exiting task can't be killed meanwhile. execve() cancels the requests in the same way, so a live task can get stuck there as well, for minutes. If it takes more than five, the work items run into the WARN_ON_ONCE(time_after(jiffies, timeout)) of io_ring_exit_work(). No privilege is needed for any of this. If the callback has met a task in cancellation, wait a little before it is sent again. Nobody has to wake the work item up for that: either the node is gone by then, or it is the turn of the next round. Fixes: d56d938b4bef ("io_uring: do ctx initiated file note removal") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-6-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 92e20e9c72bdf6cfa6f84eded7b33399bc7061da Author: Josef Bacik Date: Thu Oct 15 19:30:29 2026 +0000 io_uring/io-wq: don't hide a link from cancellations Commit 436d48a2ed13112d2d27085325838e5bce6b4507 upstream. A task that exits cancels its tracked requests and waits until they are gone. io_uring_cancel_generic() looks everywhere a request can be, and goes to sleep if it found nothing to cancel. It is woken up each time one of the requests of the task is completed by task work or freed. When io-wq is done with a request that has a link, there is a time in which the next request of the link can't be found: io_worker_handle_work() io_assign_current_work(worker, NULL) io_wq_free_work(work) io_req_find_next() // takes the link off the request io_free_req() // task work for the task io_assign_current_work(worker, linked) io_run_task_work() cond_resched() worker->cur_work = linked It is not pending, it is not the current work of a worker, and it is not yet on the list that it is about to be parked on. If the pass that the exiting task makes because of io_free_req() falls into that time, it finds nothing and the task goes to sleep. The worker then runs the link, which IO_WQ_BIT_EXIT doesn't prevent for bounded work. If that parks the request nothing is completed or freed, and nobody wakes up the task. If it is the process that exits it can't be killed any more either. NOP, IOSQE_ASYNC | IOSQE_IO_LINK FUTEX_WAIT, FUTEX2_PRIVATE exit of the thread, or of the process, up to 63us later With four of these per round 5 to 8 of 10,000 rounds hang (4 CPUs, KVM): io_uring_cancel_generic do_exit Commit f8b632e89a10 ("io_uring: tighten task exit cancellations") is about the same gap, and makes the task look again for whatever happens before io_free_req(). Free the request once the link is the current work of the worker, or pending. Whoever is woken up by that can find the link, until it is parked where cancellations look as well. To do so split io_wq_free_work() in two. [ 7.2: io_wq_put_work() keeps the link check of 7.2, without the REQ_F_CQE_SKIP condition that commit 5530a5102e3c adds in 7.3 ] Fixes: 768134d4f481 ("io_uring: don't do flush cancel under inflight_lock") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-5-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Co-Authored-By: Claude Fable 5.1 Signed-off-by: Greg Kroah-Hartman commit fdb771015c1e13677b57a4015f0c938748f13781 Author: Josef Bacik Date: Thu Oct 15 19:30:28 2026 +0000 io_uring/io-wq: don't lose the signal that ends work Commit 63c103279a95e82280092f2f4232e91e8117a252 upstream. When a thread exits, only its tracked requests are cancelled. Then its io-wq is torn down. io_wq_exit_workers() signals every worker, to get it out of an interruptible sleep, and waits for all of them to exit. Before they do, they go through the work that is left. Unbounded work is cancelled. Bounded work is run, as it is expected to end by itself. Not all of it does. An open has no file yet, so it counts as bounded, and the open of a FIFO waits for the other end. That is what the signal is for. But the workers are signalled once, and a worker that hasn't started the work by then loses the signal: io_wq_exit_workers() io_worker_handle_work() io_wq_worker_wake() set TIF_NOTIFY_SIGNAL io_assign_current_work() io_run_task_work() clear TIF_NOTIFY_SIGNAL io_wq_submit_work() fifo_open() wait_for_partner() wait_for_completion_timeout() The thread never gets out of do_exit(): IORING_OP_OPENAT of a FIFO without a writer, IOSQE_ASYNC exit of the thread that has submitted it hangs in more than 9 of 10 attempts (4 CPUs, KVM), until the process is killed. Set TIF_NOTIFY_SIGNAL for bounded work that is started with IO_WQ_BIT_EXIT set. The bit is set before the workers are signalled, and it is tested after the flag has been cleared, with clear_notify_signal() providing the barrier. So either the flag stays, or the bit is seen. Work that is started while the io-wq exits is treated like work that is being run at that time: it isn't interrupted unless it waits interruptibly. Fixes: afcc4015d1bf ("io-wq: provide an io_wq_put_and_exit() helper") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-4-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit c347691e5dc5481b2d6943c123619e04847fb0dc Author: Josef Bacik Date: Thu Oct 15 19:30:27 2026 +0000 io_uring/wait: don't spin on a stuck overflow list Commit 34f5a664c34d3cb62f6bec7f4156754594a7d25b upstream. As long as IO_CHECK_CQ_OVERFLOW_BIT is set, io_cqring_wait_schedule() returns right away, before it gets to look at signals or arms the timeout, for io_cqring_wait() to flush the overflow list. If that doesn't empty the list, and the CQ ring doesn't hold the number of events the caller wants, the loop starts over, with the same result. The task spins in the kernel, and can't be killed. The assumption is that this can't happen: if CQEs can't be flushed, the ring is full, and min_events is capped to the size of the ring. But: - With IORING_SETUP_CQE_MIXED a 32b CQE needs two contiguous entries. It is stuck on the list if a single one is free, and then the ring holds less than cq_entries events. A single threaded application that asks for as many events as the ring has entries never gets out of io_uring_enter(). - Whether the ring is full is up to the CQ head, which the application can write to. While one thread waits for events on an empty ring, another one can move the head back so that the ring looks full and have a CQE overflow. The waiter can't be stopped any more, not even by SIGKILL. Only the application can make room in the CQ ring. Return to it if the list could not be flushed. Fixes: 3fcf19d592d5 ("io_uring: parse check_cq out of wq waiting") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-3-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 685b49483200384d7f49d338af476e26e6824637 Author: Josef Bacik Date: Thu Oct 15 19:30:26 2026 +0000 io_uring/cmd: don't leave list poison in ->hash_node Commit 4c83e39f4d1944d7146fd06094f2c8053238d55c upstream. io_uring_cmd_del_cancelable() uses hlist_del() to take a command off ->cancelable_uring_cmd, which leaves LIST_POISON1/2 behind in ->hash_node. Requests are recycled through the request cache without being reinitialised, and not every user of ->hash_node sets it up before it is used: If a vectored futex wait is woken while futex_wait_multiple_setup() is still queueing the futexes, io_futexv_wait() doesn't add the request to ->futex_list, and leaves the completion to the task_work queued by the wake up. io_futexv_complete() ends in __io_futex_complete(), whose hlist_del_init() is fine for a request that is unhashed, which is how all other users of the node leave it. If the request was a cancelable uring_cmd in its previous life, e.g. a ublk FETCH_REQ, it writes through LIST_POISON2: Oops: general protection fault, probably for non-canonical address KASAN: maybe wild-memory-access in range [0xdead000000000120-...] RIP: 0010:io_futexv_complete+0x219/0x420 Call Trace: tctx_task_work_run+0x804/0x950 tctx_task_work+0x71/0xa0 task_work_run+0x118/0x200 Leave the node unhashed. Fixes: 93b8cc60c37b ("io_uring: cancelable uring_cmd") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-2-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit c10ad2813e15186fb51b37a4d17fafddc99e89a5 Author: Josef Bacik Date: Thu Oct 15 19:30:25 2026 +0000 io_uring/waitid: own the request before clearing ->head Commit c254eec4364b9f4b4cbdce43be3fb4df5a651a70 upstream. io_waitid_remove_wq() only takes the wait queue lock if ->head is set. io_waitid_wait() clears ->head first thing, before it tries to become the owner of the request by bumping ->refs. If somebody else owns the request at that point, nothing keeps it from completing the request while the wake up handler is still going to touch it: CPU0 (io_waitid_wait) CPU1 (io_waitid, holds a ref) list_del_init(&wait->entry) smp_store_release(&iw->head, NULL) __do_wait() finds another child io_waitid_remove_wq() // ->head == NULL, no locking io_waitid_finish(), IOU_COMPLETE // request freed and reused atomic_fetch_inc(&iw->refs) The same goes for __io_waitid_cancel() if it claims the request in that window. ->refs is part of the per opcode data of the request, so this increments whatever the next user of the request keeps there. If that happens to be zero, the handler concludes that it owns the request, and queues io_waitid_cb() for it. With the window held open by a delay, and the request reused for a receive that is waiting for data: BUG: KASAN: slab-use-after-free in __wake_up_common+0x1ca/0x1f0 __wake_up_sync_key+0x36/0x50 sock_def_readable+0xf1/0x420 unix_stream_sendmsg+0x60b/0x1110 Allocated by task 87: io_arm_apoll+0x1ed/0x940 Freed by task 87: __io_submit_flush_completions+0x118d/0x1a10 tctx_task_work_run+0x327/0x950 Only clear ->head once the handler owns the request. If it doesn't, the owner finds ->head set and has to wait for the lock, which is held by the caller of the handler. Deleting the already deleted entry again is harmless. Fixes: ab673c1bcaf2 ("io_uring/waitid: use io_waitid_remove_wq() consistently") Assisted-by: LLM Signed-off-by: Josef Bacik Link: https://patch.msgid.link/20261015-b4-io-uring-fixes-v1-1-c1556f5d7bf5@toxicpanda.com Signed-off-by: Jens Axboe Signed-off-by: Greg Kroah-Hartman commit 75395db7a9b40a2e826c838e648fef8ed1a2d174 Author: Christian Brauner Date: Mon Sep 21 15:44:53 2026 +0200 io-wq: order the exit bit against worker creation task work Commit 0e5663856b69219ef2f02e04bac9f5c5be19718a upstream. io_wq_exit_workers() cancels task work for queued worker creation with task_work_cancel_match(). That has a plain load of task->task_works. io_queue_worker_create() adds the new entry and tests IO_WQ_BIT_EXIT to cancel it if the workqueue is already on its way out. That must be ordered. task_work_add() has a full barrier via cmpxchg(). But io_wq_exit_start() sets the bit with set_bit() which doesn't have any memory ordering. So afaict, on weakly ordered architectures the exiting task may load task_works before the store of the bit is visible. The other side tests the bit before the store is visible as well. So the entry remains queued with worker_refs and the exiting task waits on worker_done indefinitely. If the task still has rings then io_uring_del_tctx_node() provides the barrier via test_and_set_bit() in io_wq_set_exit_on_idle(). When it has closed all rings though that barrier is gone. Add the barrier after the set_bit(). Fixes: 71a85387546e ("io-wq: check for wq exit after adding new worker task_work") Cc: stable@vger.kernel.org Reviewed-by: Jens Axboe Acked-by: Oleg Nesterov Link: https://patch.msgid.link/20260921-work-coredump-fixes-v3-4-8e4adb1619e6@kernel.org Signed-off-by: Christian Brauner (Amutable) Signed-off-by: Greg Kroah-Hartman commit e33e2718c62a602001ab549642a1aed0be797105 Author: Haoran Wang Date: Fri Oct 16 11:33:13 2026 +0800 tracing/syscalls: Use copy_from_user for the faultable syscall buffer commit 097fd0050089df17b032f2b852b9b65e548e342a upstream. syscall_copy_user_array() copies the buffer argument of faultable syscalls (write, pwrite64, ...) with __copy_from_user(), which skips access_ok(). Use copy_from_user() instead so a non-user address is rejected and the entry is recorded as faulted. It could be a security concern for machines that trust perf a bit too much. Cc: stable@vger.kernel.org Fixes: baa031b7bd2ce ("tracing: Allow syscall trace events to read more than one user parameter") Link: https://patch.msgid.link/6ad19afd.a938ce27.16851e.3440@mx.google.com Signed-off-by: Haoran Wang Signed-off-by: Steven Rostedt Signed-off-by: Greg Kroah-Hartman commit 5597bb8db6855fcb97a67196d227fea359bcd81c Author: Lee Jones Date: Mon Oct 19 12:07:29 2026 +0000 tracing/histograms: Verify event_call identity when updating synth_event refcount commit 3403d9bb27ebc3572bd048620746b4414e073bf9 upstream. In event_hist_trigger_parse() and hist_unreg_all(), the histogram registration and unregistration paths look up find_synth_event() using trace_event_name(file->event_call) and adjust se->ref if a match is found. Because find_synth_event() only compares the event basename, registering or removing a histogram on a non-synthetic trace event whose basename matches a synthetic event erroneously modifies the synthetic event's reference count. If a histogram is registered on the non-synthetic event before the synthetic event is created and removed afterwards, se->ref is decremented without a matching increment, allowing the synthetic event to be freed while an active histogram action still holds a pointer to it. Introduce find_synth_event_call() to verify that &se->call matches file->event_call before adjusting se->ref. Fixes: 4b147936fa50 ("tracing: Add support for 'synthetic' events") Cc: stable@vger.kernel.org Link: https://patch.msgid.link/20261019120730.2342739-1-lee@kernel.org Signed-off-by: Lee Jones Signed-off-by: Steven Rostedt Signed-off-by: Greg Kroah-Hartman commit 3ec17e0b23a84cc64e08fb831a6fce458e20a01a Author: Donggeun Yoo Date: Thu Sep 17 10:55:32 2026 +0900 tracing: Fix 64-bit and signed histogram fields on 32-bit kernels commit 443d23176812803d0fa182b8d05ef066a7b8271f upstream. DEFINE_HIST_FIELD_FN() reads a trace event field and widens it to the u64 the histogram uses as a key or a value: type *addr = (type *)(event + hist_field->field->offset); return (u64)(unsigned long)*addr; unsigned long is 32 bits on a 32-bit kernel, so the intermediate cast discards the upper half of a 64-bit field, and widens a signed field by zero extension from 32 bits rather than sign extension from its own width. Five of the eight instantiations are wrong there: s64, u64, s32, s16 and s8. The cast came in with commit 79e577cbce4c ("tracing: Support string type key properly"), which also added hist_field_dynstring() and hist_field_pstring(). Those convert a char * to u64 and do warn without an unsigned long in between on 32-bit. DEFINE_HIST_FIELD_FN() converts an integer and never a pointer, so it gains nothing from the cast and only loses information. Drop the intermediate cast and let the usual arithmetic conversions widen *addr according to its own type. unsigned long is 64 bits on LP64, so the two forms are identical there and 64-bit kernels are unaffected. Cc: stable@vger.kernel.org Link: https://patch.msgid.link/20260917015532.103081-1-donggeunyoo.kernel@gmail.com Fixes: 79e577cbce4c ("tracing: Support string type key properly") Reported-by: sashiko-bot@kernel.org Closes: https://lore.kernel.org/all/20260914054614.82A9B1F000FF@smtp.kernel.org/ Signed-off-by: Donggeun Yoo Signed-off-by: Steven Rostedt Signed-off-by: Greg Kroah-Hartman commit 5afb9317e2eeb1c7021a82995ce244fda5a4bd4f Author: Donggeun Yoo Date: Mon Sep 14 19:40:36 2026 +0900 tracing: Unregister field variable histograms when the command fails commit 91c14d6c13096091c56304120aaa26e479130b8b upstream. An action that names a field on another event needs that field as a variable, so create_field_var_hist() registers a second hist trigger on that event to provide it, and records it in target_hist_data->field_var_hists[]. If a later part of the same command fails, create_actions() returns an error and event_hist_trigger_parse() jumps to out_free. The write fails with -EINVAL and sched_switch gets no trigger, but: # echo 'hist:keys=pid:ts0=common_timestamp.usecs' >> \ events/sched/sched_waking/trigger # echo 'my_synth u64 lat; int a; int b' >> synthetic_events # echo 'hist:keys=next_pid:wakeup_lat=common_timestamp.usecs-$ts0:\ onmatch(sched.sched_waking).my_synth($wakeup_lat,prio,nosuchfld)'\ >> events/sched/sched_switch/trigger # cat events/sched/sched_waking/trigger hist:keys=pid:vals=hitcount:ts0=common_timestamp.usecs:sort=hitcount:size=2048:clock=global [active] hist:keys=pid:vals=hitcount:synthetic_prio=prio:sort=hitcount:size=2048 [active] The second histogram is the one created for 'prio'. It is still active and still counting. out_free reaches destroy_field_var_hists(), which frees the field_var_hists[] entries, but never unregister_field_var_hists(), which is what removes the triggers and runs only from event_hist_trigger_free(). The entry is freed and the trigger is left with nothing tracking it. Unregister the field variable histograms in out_free as well, and record each one as soon as its trigger is registered rather than after the lookup that follows it, so no exit can leave a registered trigger untracked. The entry then belongs to field_var_hists[] and is released by destroy_field_var_hists(), so freeing it on that exit would be a double free. Cc: stable@vger.kernel.org Link: https://patch.msgid.link/20260914104036.1768529-1-donggeunyoo.kernel@gmail.com Fixes: 02205a6752f2 ("tracing: Add support for 'field variables'") Reported-by: sashiko-bot@kernel.org Closes: https://lore.kernel.org/all/20260913204302.DB2551F000FF@smtp.kernel.org/ Signed-off-by: Donggeun Yoo Assisted-by: Claude:claude-fable-5 Signed-off-by: Steven Rostedt Signed-off-by: Greg Kroah-Hartman commit 9631dbccd3c649b567a5aafffea6b00634183997 Author: Donggeun Yoo Date: Tue Sep 15 08:47:54 2026 +0900 tracing: Save the event file instead of the compatible histogram commit d5c42664fc094899af9aae5300b27fe8fdb658ef upstream. create_field_var_hist() creates a histogram on the matched event to supply a field variable, and records it on the target so it can be torn down later: hist_data = find_compatible_hist(target_hist_data, file); ... /* Save the compatible histogram information */ var_hist->hist_data = hist_data; hist_data is not the histogram it creates. It is one that was already there, whose keys the new one copies. The saved pointer has a single user, unregister_field_var_hists(), which runs when the target is removed: file = hist_data->field_var_hists[i]->hist_data->event_file; find_compatible_hist() matches on keys alone, so it can pick one with no variables of its own. check_var_refs() then returns false and the user can remove it while the target still points at it: BUG: KASAN: slab-use-after-free in event_hist_trigger_free+0x2b2/0x320 Read of size 8 at addr ffff8880093b90e0 by task init/1 Freed by task 1: kfree+0x154/0x420 event_hist_trigger_free+0x1d5/0x320 event_hist_trigger_parse+0x35f3/0x69e0 trigger_process_regex+0x1a6/0x250 Save the event file instead. It is all the dereference ever wanted, and it does not go away when the user removes a trigger. Cc: stable@vger.kernel.org Reported-by: Sashiko Link: https://lore.kernel.org/all/20260914110753.221E61F000FF@smtp.kernel.org/ Link: https://patch.msgid.link/20260914234754.3676905-1-donggeunyoo.kernel@gmail.com Fixes: 02205a6752f2 ("tracing: Add support for 'field variables'") Signed-off-by: Donggeun Yoo Assisted-by: Claude:claude-fable-5 Acked-by: Tom Zanussi Signed-off-by: Steven Rostedt Signed-off-by: Greg Kroah-Hartman commit 64993d970dde2a2da353e9e99595a4fa88ec27d9 Author: Kyle Zeng Date: Fri Oct 23 10:46:20 2026 -0700 tracing: Reject incomplete event filter expressions commit 1fdc4164e488916402badddbb1f51e9380dabb91 upstream. A non-empty expression normally finishes in predicate_parse()'s inner loop, which resolves the pending logical operators before jumping to finalization. If the input ends after an operator, possibly followed by unary '!' tokens, the outer loop instead falls through with unresolved operator state. For example, "common_pid == 1 || common_pid == 2 &&" leaves prog[0].target equal to -1. The second pass then reads prog[-1] before the forward-jump check can reject it. Depending on adjacent slab contents, it can also copy an out-of-bounds word into a branch target that passes the forward-only check. A self-targeted perf tracepoint counting event can reach this path through PERF_EVENT_IOC_SET_FILTER. Reject the outer-loop fallthrough when a predicate has already been parsed. This ensures that non-empty expressions only reach finalization after the inner loop has resolved their operators. Keep the existing no-filter error for input that has no predicates. Add a regression test for trailing logical operators and unary negations. Fixes: 80765597bc587 ("tracing: Rewrite filter logic to be simpler and faster") Cc: stable@vger.kernel.org Link: https://patch.msgid.link/20261023174620.91969-1-kylebot@openai.com Assisted-by: Codex:gpt-6-astra Acked-by: Masami Hiramatsu (Google) Tested-by: JongHyuk Kim(MalHyuk) Signed-off-by: Kyle Zeng Signed-off-by: Steven Rostedt Signed-off-by: Greg Kroah-Hartman commit d6d991afe775ab7fcf89523a2502b7ca3be56ffd Author: David Carlier Date: Fri Oct 16 07:16:35 2026 +0100 tracing: Hold trace_types_lock when resetting last boot data from readers commit 5a6bd62f09147b49b31fd19d4eba122cb5f941a1 upstream. update_last_data_if_empty() is called from tracing_open() with O_TRUNC, tracing_read_pipe() and tracing_buffers_read() without trace_types_lock. Two readers can both pass the LAST_BOOT test and free module_delta twice, and a module notifier can update module_delta while it is being freed. Take trace_types_lock in those paths and keep the callers that already hold it on an unlocked variant. Cc: stable@vger.kernel.org Fixes: 32dc0042528d ("tracing: Reset last-boot buffers when reading out all cpu buffers") Link: https://patch.msgid.link/20261016061638.214847-2-devnexen@gmail.com Signed-off-by: David Carlier Signed-off-by: Steven Rostedt Signed-off-by: Greg Kroah-Hartman commit dab2bd4d67ad5eaa95019a9511f8d379401b5021 Author: David Carlier Date: Fri Oct 16 07:16:34 2026 +0100 tracing: Hold trace_types_lock when updating modules in persistent instances commit 66877a312ae3c2c3d849e82e882f642ffd4ec3e2 upstream. trace_module_record() walks ftrace_trace_arrays from the module notifier without trace_types_lock. A concurrent rmdir of an instance removes it from the list and frees it under that lock, so a module load or unload can walk into a freed trace_array. Cc: stable@vger.kernel.org Fixes: 5f3719f697c3 ("tracing: Update modules to persistent instances when loaded") Link: https://patch.msgid.link/20261016061638.214847-1-devnexen@gmail.com Signed-off-by: David Carlier Signed-off-by: Steven Rostedt Signed-off-by: Greg Kroah-Hartman commit 00d90a21662fef723f904ef7423f7a0cfa5f8171 Author: Sasha Levin Date: Sun Oct 11 18:33:31 2026 -0400 tools/dma: fix building with O= commit 9441e7eca18c8004d1afe57b2f9b27717549562a upstream. With O= set, "all" depends on $(OUTPUT)dma_map_benchmark, but the only rule builds dma_map_benchmark in the source directory, and built-in rules are disabled: make: *** No rule to make target '/dma_map_benchmark', needed by 'all'. Stop. Build $(OUTPUT)dma_map_benchmark. Fixes: f74ee32963f1b ("tools/dma: move dma_map_benchmark from selftests to tools/dma") Cc: stable@vger.kernel.org Assisted-by: LLM Signed-off-by: Sasha Levin Acked-by: Qinxin Xia Link: https://lore.kernel.org/r/20261011223358.2327625-3-sashal@kernel.org Signed-off-by: Marek Szyprowski Signed-off-by: Greg Kroah-Hartman commit 73820561d8cbfd3364a29b8fa77f1707f26029da Author: Pauli Virtanen Date: Wed Sep 2 00:04:36 2026 +0300 Bluetooth: L2CAP: refuse __l2cap_chan_add if chan already has conn [ Upstream commit 86b55991d3cd690e73c5ef8d20b75ae4905b357a ] l2cap_chan may be linked to l2cap_conn at most once. This is assumed in several places, eg l2cap_chan_del cleanup. There is a TOCTOU race where the invariant is violated: [Task 1] [Task 2] l2cap_chan_connect l2cap_sock_bind l2cap_chan_lock lock_sock l2cap_state_change if (sk->sk_state != BT_OPEN) chan->state = BT_CONNECT l2cap_sock_state_change_cb chan->state = BT_BOUND sk->sk_state = BT_BOUND lock_sock <------------------ release_sock sk->sk_state = BT_CONNECT l2cap_sock_connect() does not check sk->sk_state, so since chan->state is now BT_BOUND, subsequent connect() ends up with second __l2cap_chan_add. Explicitly document and check the invariant in __l2cap_chan_add with WARN_ON_ONCE. The only callsite where it could be hit is l2cap_chan_connect, so add pre-check there to avoid relying on chan->state. chan->state read/write is not properly guarded currently so there can be other TOCTOUC problems. Add l2cap_lock_chan in l2cap_sock_bind() to guard chan->state write. Fixes: b66774b48dd9 ("Bluetooth: L2CAP: Fix UAF in channel timeout by holding conn ref") Assisted-by: deepseek-4-flash # finding the race condition Signed-off-by: Pauli Virtanen Signed-off-by: Luiz Augusto von Dentz Signed-off-by: Sasha Levin commit 4e7f5dd22b5d9c0712e580c86b3fd37f4fb29cca Author: Hui Su Date: Mon Oct 26 11:41:09 2026 +0100 sched/fair: Avoid overflow in place_entity() [ Upstream commit 2f93bcb3f9aaf6ceee1aec0a97d4fe386d48e781 ] place_entity() inflates an entity's virtual lag with lag * (load + weight) / load Commit 4823725d9d1d ("sched/fair: Increase weight bits for avg_vruntime") replaced the scaled-down weights used in this calculation with avg_vruntime_weight(), removing the arithmetic headroom provided by scale_load_down(). The multiplication can overflow s64 before the division even when the final quotient is representable. A task's hierarchical effective weight can be as small as 2 after avg_vruntime_weight() scaling under group scheduling, while entity_lag() bounds |vlag| by calc_delta_fair(cfs_rq_max_slice(cfs_rq) + TICK_NSEC, se) For example, with HZ=1000, NICE_0_LOAD=1048576, and the minimum scaled hierarchical weight scale_load(2)=2048, this bound is 891289600. With 114 maximum-weight entities, the runqueue load is 10361604096. The intermediate product is: 891289600 * (10361604096 + 2) = 9235189971864780800 > S64_MAX The signed overflow corrupts the entity's placement. Rewrite lag * (load + weight) / load as lag + lag * weight / load The two expressions are equivalent, but the latter avoids multiplying the virtual lag by the total runqueue weight. entity_lag() bounds vlag against the entity's effective weight, so the remaining lag * weight product stays bounded. rescale_entity() preserves this scale across h_load changes and already relies on the same product when rescaling vlag. Keep the original zero-load warning semantics by multiplying lag by weight after replacing the divisor with one. Fixes: 4823725d9d1d ("sched/fair: Increase weight bits for avg_vruntime") Signed-off-by: Hui Su Signed-off-by: Peter Zijlstra (Intel) Link: https://patch.msgid.link/20260928124754.1692077-1-sh_def@163.com [ Backport to v7.2 which doesn't have flat hierarchy ] Signed-off-by: Vincent Guittot Cc: stable@vger.kernel.org Signed-off-by: Sasha Levin commit cc190c4073f570fd78a9d00dd3d13a71eced5403 Author: SJ Park Date: Mon Oct 26 06:36:43 2026 -0700 mm/damon/core: error damos_commit_quota_goal() for zero target_value [ Upstream commit 610c4438fb66e06d9e693c0ad3b2c1be54c0d61f ] Patch series "mm/damon: move zero damos quota target_value handling to the core layer". Having zero DAMOS quota target value can cause division by zero. DAMON API callers are checking the target value parameters to avoid that. It is easy to make mistakes in some of the multiple API callers. Move that to the core layer. Patch 1 adds the corner case handling into the core layer DAMON parameters validation logic. Patches 2 and 3 remove no more needed DAMON API callers side handling of the corner case in DAMON_LRU_SORT and DMON_SAMPLIE_MTIER, respectively. This patch (of 3): If a DAMOS scheme has a damos_quota_goal of zero target_value, damos_quota_goal() could trigger division-by-zero error. Hence each DAMON API callers should do the zero target_value validation. It is easy to make mistakes. Actually such bugs in DAMON_LRU_SORT and DAMON_SAMPLE_MTIER were found and fixed [1]. It is better to handle the corner case only once in the core layer, instead of multiple places in all DAMON API callers. One straightforward option is using an alternative denominator for the corner case in the damos_quota_goal(). However, the zero target_value is meaningless. In this case, the quota goal is always evaluated as achieved or over-achieved. The quota will only keep being reduced. Simply avoid using zero target_value by adding a check in the core layer DAMOS quota goal parameters validation/commit path, damos_commit_quota_goal(). Update it to return an error in the case. Also update its caller to propagate the error. Notes for stable kernel porting. The upstream commit doesn't have Fixes: and therefore not ported to stable kernels. It turned out it fixes commit 8b549a4fd3c5 ("mm/damon/sysfs-schemes: commit damos quota goals user input to DAMOS") which was merged into v6.8. The user impact is a race condition that can result in divide-by-zero. Hence backport. Reproducing the issue requires sysfs write permission, certain DAMON ABI usages that are supposed to be uncommon, and a rare race timing, though. The issue was discovered [2] by Sashiko as a part [3] of its bug report. Also, the upstream commit on this stable kernel can fix the issue for parameters commit paths but DAMON start path. That's because DAMON start path does not invoke damon_commit_quota_goals(). To fix the path with minimum change, add a new small validation of quota goal target value before damon_start() call. [2] https://sashiko.dev/#/bug/linux-7f814c9a-986f-4103-a55a-ef9537dcf62b [3] https://lore.kernel.org/20261021005234.94107-1-sj@kernel.org Link: https://lore.kernel.org/20260903010722.94244-1-sj@kernel.org Link: https://lore.kernel.org/20260903010722.94244-2-sj@kernel.org Link: https://lore.kernel.org/20260803134034.15217-1-sj@kernel.org [1] Signed-off-by: SJ Park Signed-off-by: Andrew Morton Signed-off-by: SJ Park Signed-off-by: Sasha Levin commit 57c0d5281666d150c73e56e6795a4eef3ede8693 Author: Hongling Zeng Date: Mon Oct 26 10:55:33 2026 +0900 ntfs: take invalidate_lock in ntfs_filemap_page_mkwrite() [ Upstream commit 0fecc393f2060e6bc25138df32cb923ec7071c6b ] ntfs_filemap_page_mkwrite() calls iomap_page_mkwrite() without holding mapping->invalidate_lock, so a concurrent truncate or fallocate can be in the middle of invalidating pagecache and rewriting the runlist while the write fault maps blocks and dirties the folio. This races with ntfs_attr_fallocate(), which merges clusters into the in-memory runlist, drops the runlist lock, and only afterwards zeroes the newly allocated clusters on disk; and with the punch-hole/insert/collapse paths that free clusters after truncating the cache. Per Documentation/filesystems/locking.rst, ->page_mkwrite() must ensure there are no truncate/invalidate races, "usually mapping->invalidate_lock is suitable for proper serialization". xfs takes its mmaplock (= the invalidate_lock rwsem) shared in exactly this path. Take invalidate_lock shared around iomap_page_mkwrite(). The read-only fault path is already covered because filemap_fault() itself grabs invalidate_lock shared on instantiation/read paths; only page_mkwrite was bypassing it in this driver. Fixes: 9c87959601e8 ("ntfs: update file operations") Cc: stable@vger.kernel.org Reviewed-by: Hyunchul Lee Reviewed-by: Baolin Liu Signed-off-by: Hongling Zeng Co-developed-by: Namjae Jeon Signed-off-by: Namjae Jeon Signed-off-by: Hyunchul Lee Signed-off-by: Sasha Levin commit 7dd1c2fec6ebdd1a88a02cb58ff515a258eeb327 Author: Hongling Zeng Date: Mon Oct 26 10:55:32 2026 +0900 ntfs: fix race between fallocate and mmap reads [ Upstream commit 67aded1da114dc44808315f249bd9e7e440f799d ] The fallocate implementation only takes invalidate_lock for punch hole, collapse range, and insert range operations. For standard allocation modes (mode == 0, FALLOC_FL_KEEP_SIZE), the lock is not held. During ntfs_attr_fallocate(), new clusters are mapped to the runlist via ntfs_attr_map_cluster() before being zeroed by ntfs_dio_zero_range(). This creates a window where concurrent mmap page faults can read uninitialized disk data. Since mmap uses filemap_fault() which takes invalidate_lock in shared mode, it can fault in pages during this window and expose old disk contents to userspace. This is an information leak and data integrity issue. Fix by taking invalidate_lock for all fallocate operations, not just for punch/collapse/insert modes. This prevents concurrent page faults from accessing unzeroed clusters during the allocation window. Fixes: 495e90fa3348 ("ntfs: update attrib operations") Cc: stable@vger.kernel.org Reviewed-by: Baolin Liu Reviewed-by: Hyunchul Lee Signed-off-by: Hongling Zeng Signed-off-by: Namjae Jeon Stable-dep-of: 0fecc393f206 ("ntfs: take invalidate_lock in ntfs_filemap_page_mkwrite") Signed-off-by: Hyunchul Lee Signed-off-by: Sasha Levin commit cc2839946328c1618b3870d521b32bb9483d49d0 Author: Kuba Piecuch Date: Sat Oct 24 17:25:13 2026 +0800 sched_ext: Wait for SCX_OPSS_DISPATCHING before moving tasks off DSQs [ Upstream commit 2935062be77113c1c4565ca9d0c72180449aafee ] A task can be found on a non-local DSQ while the CPU which dispatched it is still running ops.dequeue() and hasn't cleared SCX_OPSS_DISPATCHING. The DSQ consume and move paths don't wait for the dispatch to finish and can operate on the task concurrently with the dispatcher. Call scx_wait_dispatching() in scx_move_local_task_to_local_dsq(), unlink_dsq_and_switch_rq_lock() and move_task_between_dsqs() before they operate on the task. In scx_move_local_task_to_local_dsq(), wait before scx_resolve_local_dsq() as it may update p->scx.flags. Fixes: ebf1ccff79c4 ("sched_ext: Fix ops.dequeue() semantics") Fixes: cb86607ada73 ("sched_ext: Don't run ops.dequeue() with a DSQ lock held") Cc: stable@vger.kernel.org # v7.1+ Reviewed-by: Andrea Righi Assisted-by: Claude:claude-opus-5.5 Signed-off-by: Kuba Piecuch Signed-off-by: Tejun Heo [ Inlined scx_wait_dispatching(), which 7.2.y doesn't have, into move_local_task_to_local_dsq(), unlink_dsq_and_lock_src_rq() and move_task_between_dsqs(), as was done for the reenqueue paths in 7.2.y. 7.2.y has no scx_resolve_local_dsq(), so move_local_task_to_local_dsq() waits before linking the task. ] Signed-off-by: Cheng-Yang Chou Signed-off-by: Sasha Levin