Re: [PATCH] AMD Thermal Interrupt Support

[Andi Kleen <andi@firstfloor.org>]

> As to the question of Athlon RevG, it turns out that it has the same
> thermal erratum as RevF.  Therefore the fact that this code only works
> with Barcelona and up is not a defficiency.

Thanks.

> 
> There are 2 pending issues, to which I have received insufficient feedback:
> 
> 1. Andi would like to eliminate the trampoline in mce_thermal.c, but
> no one has responded to my proposed disgusting hack on entry_64.S in
> order to do so.

There are two ways to do it: 

either duplicate the entry point in entry_64.S and set different
vectors or let the asm glue jump through a call vector (use 
apicinterrupt ...,*thermal_vector(%rip) ) Later is probably better.

> 
> 2. Ingo pointed out that a given config file did not build.  But when

The patch that was in git-x86 didn't build on 32bit. Perhaps it was a 
32bit config file? Run it under "linux32" or equivalent on your distro.

> +	 */
> +	if (therm_throt_process(1))
> +		mce_log_therm_throt_event(cpu, 1);
> +	/*
> +	 * We'll still get subsequent interrupts even if we don't clear the
> +	 * status bit in THERM_CTL_F3X64.  Take advantage of this fact to avoid
> +	 * touching PCI space.  (If this assumption fails at some point, we'll
> +	 * need to schedule_work() in order to enter a process context, so that
> +	 * PCI locks can be asserted for proper access.  This requirement, in

PCI locks are spinlocks so they don't need process context. In PREEMPT-RT
kernels they can sleep, but there the handler will be likely already
a thread. Touching config space from interrupt context is legal, although
not very popular because it tends to be slow (but a few drivers do it) 

> +	/*
> +	 * If any of the northbridges has PCI ID 0x1103, then its thermal
> +	 * hardware suffers from an erratum which prevents this code from
> +	 * working, so abort.

Please add 

 * This implies it only works on Family 10h aka AMD Quad Core.

Otherwise I can just see the support questions of people asking why this
doesn't work for them.

Anyways I'm unsure about the blacklist here -- white list would
probably have been better. k8_northbridges[] will certainly include
Griffin northtbridges and who knows if TT will work there or not.
[sorry for mentioning that not earlier]


-Andi