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:   Sat, 15 Jul 2023 00:02:20 +0200
From:   Peter Zijlstra <peterz@infradead.org>
To:     Arjan van de Ven <arjan@linux.intel.com>
Cc:     "Wysocki, Rafael J" <rafael.j.wysocki@intel.com>,
        Bruno Goncalves <bgoncalv@redhat.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: <20230714220220.GC3273303@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>
 <20230711125557.GM3062772@hirez.programming.kicks-ass.net>
 <20230711132553.GN3062772@hirez.programming.kicks-ass.net>
 <0837a34c-f66e-aa04-d4a7-b032d3dbab7a@intel.com>
 <b672c500-5f2f-d60c-84da-80269b8b420f@linux.intel.com>
 <20230714211109.GC3275140@hirez.programming.kicks-ass.net>
 <6ce7c4d8-3320-7712-a562-a43224c2602b@linux.intel.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <6ce7c4d8-3320-7712-a562-a43224c2602b@linux.intel.com>
Precedence: bulk

On Fri, Jul 14, 2023 at 02:47:24PM -0700, Arjan van de Ven wrote:

> > I still don't know why this needs to be in intel_idle.
> 
> we can do a seperate idle driver; it'll still be x86 specific (since
> idle really is arch specific)...  and then the umwait parts of this
> will be Intel specific.. as well any future idle methods ..  and I'm
> not sure the AMD folks would even want it used .... at which point it
> ends up Intel specific fully and we now have 2 Intel idle drivers.  I
> don't see how that makes sense.

intel-idle is huge, the last thing it needs is yet another little driver
hidding inside it.

Creating a new simple and small x86-guest cpuidle driver seems, well,
simpler.

And the whole umwait / mwaitx thing really isn't that hard, this is all
indirect functions anyway. All that thing needs to do is amortize the
VMEXIT cost.