CVE-2026-43311: soc/tegra: pmc: Fix unsafe generic_handle_irq() call

In the Linux kernel, the following vulnerability has been resolved:

soc/tegra: pmc: Fix unsafe generichandleirq() call

Currently, when resuming from system suspend on Tegra platforms, the following warning is observed:

WARNING: CPU: 0 PID: 14459 at kernel/irq/irqdesc.c:666 Call trace: handleirqdesc+0x20/0x58 (P) tegra186pmcwakesyscoreresume+0xe4/0x15c syscoreresume+0x3c/0xb8 suspenddevicesandenter+0x510/0x540 pmsuspend+0x16c/0x1d8

The warning occurs because generichandleirq() is being called from a non-interrupt context which is considered as unsafe.

Fix this warning by deferring generichandleirq() call to an IRQ work which gets executed in hard IRQ context where generichandleirq() can be called safely.

When PREEMPTRT kernels are used, regular IRQ work (initialized with initirqwork) is deferred to run in per-CPU kthreads in preemptible context rather than hard IRQ context. Hence, use the IRQWORKINITHARD variant so that with PREEMPTRT kernels, the IRQ work is processed in hardirq context instead of being deferred to a thread which is required for calling generichandleirq().

On non-PREEMPTRT kernels, both initirqwork() and IRQWORKINITHARD() execute in IRQ context, so this change has no functional impact for standard kernel configurations.

[treding@nvidia.com: miscellaneous cleanups]