CVE-2022-49265

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> PM: domains: Fix sleep-in-atomic bug caused by genpd_debug_remove()<br /> <br /> When a genpd with GENPD_FLAG_IRQ_SAFE gets removed, the following<br /> sleep-in-atomic bug will be seen, as genpd_debug_remove() will be called<br /> with a spinlock being held.<br /> <br /> [ 0.029183] BUG: sleeping function called from invalid context at kernel/locking/rwsem.c:1460<br /> [ 0.029204] in_atomic(): 1, irqs_disabled(): 128, non_block: 0, pid: 1, name: swapper/0<br /> [ 0.029219] preempt_count: 1, expected: 0<br /> [ 0.029230] CPU: 1 PID: 1 Comm: swapper/0 Not tainted 5.17.0-rc4+ #489<br /> [ 0.029245] Hardware name: Thundercomm TurboX CM2290 (DT)<br /> [ 0.029256] Call trace:<br /> [ 0.029265] dump_backtrace.part.0+0xbc/0xd0<br /> [ 0.029285] show_stack+0x3c/0xa0<br /> [ 0.029298] dump_stack_lvl+0x7c/0xa0<br /> [ 0.029311] dump_stack+0x18/0x34<br /> [ 0.029323] __might_resched+0x10c/0x13c<br /> [ 0.029338] __might_sleep+0x4c/0x80<br /> [ 0.029351] down_read+0x24/0xd0<br /> [ 0.029363] lookup_one_len_unlocked+0x9c/0xcc<br /> [ 0.029379] lookup_positive_unlocked+0x10/0x50<br /> [ 0.029392] debugfs_lookup+0x68/0xac<br /> [ 0.029406] genpd_remove.part.0+0x12c/0x1b4<br /> [ 0.029419] of_genpd_remove_last+0xa8/0xd4<br /> [ 0.029434] psci_cpuidle_domain_probe+0x174/0x53c<br /> [ 0.029449] platform_probe+0x68/0xe0<br /> [ 0.029462] really_probe+0x190/0x430<br /> [ 0.029473] __driver_probe_device+0x90/0x18c<br /> [ 0.029485] driver_probe_device+0x40/0xe0<br /> [ 0.029497] __driver_attach+0xf4/0x1d0<br /> [ 0.029508] bus_for_each_dev+0x70/0xd0<br /> [ 0.029523] driver_attach+0x24/0x30<br /> [ 0.029534] bus_add_driver+0x164/0x22c<br /> [ 0.029545] driver_register+0x78/0x130<br /> [ 0.029556] __platform_driver_register+0x28/0x34<br /> [ 0.029569] psci_idle_init_domains+0x1c/0x28<br /> [ 0.029583] do_one_initcall+0x50/0x1b0<br /> [ 0.029595] kernel_init_freeable+0x214/0x280<br /> [ 0.029609] kernel_init+0x2c/0x13c<br /> [ 0.029622] ret_from_fork+0x10/0x20<br /> <br /> It doesn&amp;#39;t seem necessary to call genpd_debug_remove() with the lock, so<br /> move it out from locking to fix the problem.