In the Linux kernel, the following vulnerability has been resolved: optee: ffa: fix sleep in atomic context The OP-TEE driver registers the function notifcallback() for FF-A notifications. However, this function is called in an atomic context leading to errors like this when processing asynchronous notifications: | BUG: sleeping function called from invalid context at kernel/locking/mutex.c:258 | inatomic(): 1, irqsdisabled(): 1, nonblock: 0, pid: 9, name: kworker/0:0 | preemptcount: 1, expected: 0 | RCU nest depth: 0, expected: 0 | CPU: 0 UID: 0 PID: 9 Comm: kworker/0:0 Not tainted 6.14.0-00019-g657536ebe0aa #13 | Hardware name: linux,dummy-virt (DT) | Workqueue: ffapcpuirqnotification notifpcpuirqworkfn | Call trace: | showstack+0x18/0x24 (C) | dumpstacklvl+0x78/0x90 | dumpstack+0x18/0x24 | _mightresched+0x114/0x170 | _mightsleep+0x48/0x98 | mutexlock+0x24/0x80 | opteegetmsgarg+0x7c/0x21c | simplecallwitharg+0x50/0xc0 | opteedobottomhalf+0x14/0x20 | notifcallback+0x3c/0x48 | handlenotifcallbacks+0x9c/0xe0 | notifgetandhandle+0x40/0x88 | genericexecsingle+0x80/0xc0 | smpcallfunctionsingle+0xfc/0x1a0 | notifpcpuirqworkfn+0x2c/0x38 | processonework+0x14c/0x2b4 | workerthread+0x2e4/0x3e0 | kthread+0x13c/0x210 | retfromfork+0x10/0x20 Fix this by adding work queue to process the notification in a non-atomic context.