Re: [PATCH] ARM: enable interrupts when arm_notify_die() is handling user mode errors

[Sebastian Andrzej Siewior <bigeasy@linutronix.de>]

On 2026-06-25 20:26:12 [+0800], Xie Yuanbin wrote:
> ```c
> 	if (!inf->fn(addr, ifsr | FSR_LNX_PF, regs))
> 		return;
> 
> 	if (likely(user_mode(regs)))
> 		local_irq_enable();
> 
> 	pr_alert("8<--- cut here ---\n");
> ```
> or
> ```c
> 	if (!inf->fn(addr, ifsr | FSR_LNX_PF, regs))
> 		return;
> 
> 	if (likely(interrupts_enabled(regs)))
> 		local_irq_enable();
> 
> 	pr_alert("8<--- cut here ---\n");
> ```
> 
> Which one do you prefer? I prefer the first one, because for kernel
> fault, kernel may have encountered a serious issue,
> and enabling interrupts may be not appropriate.

Why would the latter be not appropriate?. 
Anyway, in the kernel case you do die() which disables interrupts as of
oops_begin(). It does later restore the state in oops_end() and invokes
make_task_dead(). This one will complain if either preemption or
interrupts are disabled and reset both.

Should you get that far and not panic() earlier (due to in_interrupt()
for instance) then interrupts will be later enabled before that kernel
thread is killed. So it could be done earlier or not, at this point the
system is pretty much done.

Sebastian