Hi Sasha,
On 10.07.26 23:03, Sasha Levin wrote:
On Thu, Jul 09, 2026 at 11:45:14AM +0200, Florian Fainelli wrote:
On ARCH_BRCMSTB using 32-bit and 64-bit ARM kernels, same build failure as 6.1 for MIPS:
arch/mips/kernel/smp.c: In function 'stop_this_cpu': arch/mips/kernel/smp.c:396:2: error: implicit declaration of function 'rcutree_report_cpu_dead'; did you mean 'rcutree_prepare_cpu'? [-Werror=implicit-function-declaration] rcutree_report_cpu_dead(); ^~~~~~~~~~~~~~~~~~~~~~~ rcutree_prepare_cpu
The offending backport was dropped before 6.6.144 was released, so the build failure itself is gone.
That still leaves 6.6.y and 6.1.y without 9f3f3bdc6d9d ("MIPS: smp: report dying CPU to RCU in stop_this_cpu()") while both trees already ship 91840be8f710 ("irq_work: Fix use-after-free in irq_work_single() on PREEMPT_RT"), which is what exposes the MIPS reboot hang that 9f3f3bdc6d9d fixes.
Jonas, could you send backports of 9f3f3bdc6d9d for 6.6.y and 6.1.y with that adaptation? I'd rather take the author's version than adapt it here.
Sure, I adopted it accordingly and sent patches for 6.6.y and 6.1.y.
Best regards, Jonas