Re: [PATCH] x86/thermal: Fix LVT thermal setup for SMI delivery mode

From: Thomas Gleixner <tglx@linutronix.de>
To: Borislav Petkov <bp@suse.de>, James Feeney <james@nurealm.net>
Cc: linux-smp@vger.kernel.org, Jens Axboe <axboe@kernel.dk>,
	lkml <linux-kernel@vger.kernel.org>,
	Zhang Rui <rui.zhang@intel.com>,
	Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>,
	x86-ml <x86@kernel.org>
Subject: Re: [PATCH] x86/thermal: Fix LVT thermal setup for SMI delivery mode
Date: Thu, 27 May 2021 13:49:42 +0200	[thread overview]
Message-ID: <87h7io8kh5.ffs@nanos.tec.linutronix.de> (raw)
In-Reply-To: <YK905sC/2cVOYo6I@zn.tnic>

On Thu, May 27 2021 at 12:31, Borislav Petkov wrote:
> @@ -1226,6 +1227,14 @@ void __init setup_arch(char **cmdline_p)
>  
>  	x86_init.timers.wallclock_init();
>  
> +	/*
> +	 * This needs to run before setup_local_APIC() which soft-disables the
> +	 * local APIC temporarily and that masks the thermal LVT interrupt,
> +	 * leading to softlockups on machines which have configured SMI
> +	 * interrupt delivery.
> +	 */
> +	therm_lvt_init();

That works, but TBH, it's a hack....

What I really fail to understand is how disabling that LVT entry makes
the machine lock up. 

Also if disabling this entry is causing the BIOS/SMM gunk to go south, then
disabling CONFIG_X86_THERMAL_VECTOR should have the same effect.

Which made me look at other places like lapic_suspend/resume which does
the save/restore Kconfig conditional as well.

Thanks,

        tglx