Re: [RFT] final cur of tr based current for -ac8

[Keith Owens]

On Sat, 10 Nov 2001 17:33:31 -0500, 
Benjamin LaHaise <[e-mail]> wrote:
>diff -ur kernels/2.4/v2.4.13-ac8/arch/i386/kernel/nmi.c v2.4.13-ac8+tr.4/arch/i386/kernel/nmi.c
>--- kernels/2.4/v2.4.13-ac8/arch/i386/kernel/nmi.c	Tue Nov  6 20:43:22 2001
>+++ v2.4.13-ac8+tr.4/arch/i386/kernel/nmi.c	Sat Nov 10 14:00:33 2001
>@@ -264,7 +264,7 @@
> 	/*
> 	 * NMI can interrupt page faults, use hard_get_current.
> 	 */
>-	int sum, cpu = hard_get_current()->processor;
>+	int sum, cpu = hard_smp_processor_id();
> 
> 	sum = apic_timer_irqs[cpu];

I am still unhappy with that NMI code.  The NMI handler can use generic
code including printk, the generic code will use the standard
get_current.  Why does nmi.c not do set_current(hard_get_current());