In the Linux kernel, the following vulnerability has been resolved:
bpf: Use rawspinlockt in ringbuf
The function _bpfringbufreserve is invoked from a tracepoint, which disables preemption. Using spinlockt in this context can lead to a "sleep in atomic" warning in the RT variant. This issue is illustrated in the example below:
BUG: sleeping function called from invalid context at kernel/locking/spinlockrt.c:48 inatomic(): 1, irqsdisabled(): 0, nonblock: 0, pid: 556208, name: testprogs preemptcount: 1, expected: 0 RCU nest depth: 1, expected: 1 INFO: lockdep is turned off. Preemption disabled at: [<ffffd33a5c88ea44>] migrateenable+0xc0/0x39c CPU: 7 PID: 556208 Comm: testprogs Tainted: G Hardware name: Qualcomm SA8775P Ride (DT) Call trace: dumpbacktrace+0xac/0x130 showstack+0x1c/0x30 dumpstacklvl+0xac/0xe8 dumpstack+0x18/0x30 _mightresched+0x3bc/0x4fc rtspinlock+0x8c/0x1a4 _bpfringbufreserve+0xc4/0x254 bpfringbufreservedynptr+0x5c/0xdc bpfprogac3d15160d62622atestreadwrite+0x104/0x238 tracecallbpf+0x238/0x774 perfcallbpfenter.isra.0+0x104/0x194 perfsyscallenter+0x2f8/0x510 tracesysenter+0x39c/0x564 syscalltraceenter+0x220/0x3c0 doel0svc+0x138/0x1dc el0svc+0x54/0x130 el0t64synchandler+0x134/0x150 el0t64_sync+0x17c/0x180
Switch the spinlock to rawspinlockt to avoid this error.