CVE-2024-42290

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> irqchip/imx-irqsteer: Handle runtime power management correctly<br /> <br /> The power domain is automatically activated from clk_prepare(). However, on<br /> certain platforms like i.MX8QM and i.MX8QXP, the power-on handling invokes<br /> sleeping functions, which triggers the &amp;#39;scheduling while atomic&amp;#39; bug in the<br /> context switch path during device probing:<br /> <br /> BUG: scheduling while atomic: kworker/u13:1/48/0x00000002<br /> Call trace:<br /> __schedule_bug+0x54/0x6c<br /> __schedule+0x7f0/0xa94<br /> schedule+0x5c/0xc4<br /> schedule_preempt_disabled+0x24/0x40<br /> __mutex_lock.constprop.0+0x2c0/0x540<br /> __mutex_lock_slowpath+0x14/0x20<br /> mutex_lock+0x48/0x54<br /> clk_prepare_lock+0x44/0xa0<br /> clk_prepare+0x20/0x44<br /> imx_irqsteer_resume+0x28/0xe0<br /> pm_generic_runtime_resume+0x2c/0x44<br /> __genpd_runtime_resume+0x30/0x80<br /> genpd_runtime_resume+0xc8/0x2c0<br /> __rpm_callback+0x48/0x1d8<br /> rpm_callback+0x6c/0x78<br /> rpm_resume+0x490/0x6b4<br /> __pm_runtime_resume+0x50/0x94<br /> irq_chip_pm_get+0x2c/0xa0<br /> __irq_do_set_handler+0x178/0x24c<br /> irq_set_chained_handler_and_data+0x60/0xa4<br /> mxc_gpio_probe+0x160/0x4b0<br /> <br /> Cure this by implementing the irq_bus_lock/sync_unlock() interrupt chip<br /> callbacks and handle power management in them as they are invoked from<br /> non-atomic context.<br /> <br /> [ tglx: Rewrote change log, added Fixes tag ]