Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Tue, 11 Jul 2023 14:55:57 +0200
From:   Peter Zijlstra <peterz@infradead.org>
To:     Bruno Goncalves <bgoncalv@redhat.com>
Cc:     arjan@linux.intel.com, rafael.j.wysocki@intel.com, x86@kernel.org,
        LKML <linux-kernel@vger.kernel.org>
Subject: Re: [6.5.0-rc1] unchecked MSR access error: RDMSR from 0xe2 at rIP:
 0xffffffff87090227 (native_read_msr+0x7/0x40) (intel_idle_init_cstates_icpu)
Message-ID: <20230711125557.GM3062772@hirez.programming.kicks-ass.net>
References: <CA+QYu4qSBdhEgFURu+ouAf2d_JNPbZgCSUaxCLoGzMqDQOLWsQ@mail.gmail.com>
 <20230711112421.GI3062772@hirez.programming.kicks-ass.net>
 <CA+QYu4qzJgiiU1qsjtWb9OU3=C=hb_c-Ag5Y4c=Xp_ObfGH=hg@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CA+QYu4qzJgiiU1qsjtWb9OU3=C=hb_c-Ag5Y4c=Xp_ObfGH=hg@mail.gmail.com>
Precedence: bulk

On Tue, Jul 11, 2023 at 02:25:51PM +0200, Bruno Goncalves wrote:

> R640 2x  Intel(R) Xeon(R) Silver 4116 CPU @ 2.10GHz

Gawd, I hate our naming :/ Google tells me that is a skylake.

> and start the VM with:
> 
> -accel kvm -cpu
> Skylake-Server-IBRS,ss=on,vmx=on,pdcm=on,hypervisor=on,tsc-adjust=on,clflushopt=on,umip=on,pku=on,md-clear=on,stibp=on,arch-capabilities=on,ssbd=on,xsaves=on,ibpb=on,ibrs=on,amd-stibp=on,amd-ssbd=on,rsba=on,skip-l1dfl-vmentry=on,pschange-mc-no=on

You tell it to be a skylake

> The decoded call trace:

> /builds/4626306068/workdir/drivers/idle/intel_idle.c:1820

And that's skx_idle_state_table_update() reading
MSR_PKG_CST_CONFIG_CONTROL and that code has been around since 2021.

So things are somewhat consistent. But I find it weird that intel_idle
gets selected for a guest, I'm not exactly sure what's up with that.

Oohh, this vm-guest mode is new :/ But it doesn't make sense, that
commit babbles something about waking CPUs from idle to do TLB
invalidate, but that shouldn't be the case, that's what we have
kvm_flush_tlb_multi() for, it should avoid the IPI and flush on
vcpu-enter.

Arjan, what is the actual problem you're trying to solve any why hide
this in intel_idle ?