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;
Message-ID: <c38179ec-ff16-7b38-8124-87d009aafac8@quicinc.com>
Date:   Wed, 12 Jul 2023 16:03:36 +0800
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:102.0) Gecko/20100101
 Thunderbird/102.11.2
Subject: Re: [PATCH] arm64: Add the arm64.nolse_atomics command line option
To:     Marc Zyngier <maz@kernel.org>
CC:     <will@kernel.org>, <corbet@lwn.net>, <catalin.marinas@arm.com>,
        <quic_pkondeti@quicinc.com>, <quic_kaushalk@quicinc.com>,
        <quic_satyap@quicinc.com>, <quic_shashim@quicinc.com>,
        <quic_songxue@quicinc.com>, <linux-doc@vger.kernel.org>,
        <linux-kernel@vger.kernel.org>,
        <linux-arm-kernel@lists.infradead.org>
References: <20230710055955.36551-1-quic_aiquny@quicinc.com>
 <875y6s8bwb.wl-maz@kernel.org>
 <32f442e3-3d5c-4cec-9791-0da039f88287@quicinc.com>
 <874jmc8654.wl-maz@kernel.org>
 <6e07ad52-2629-346e-6217-ec07777ebc5b@quicinc.com>
 <86lefnvsto.wl-maz@kernel.org>
 <8a950aa5-fdd8-f983-0411-4b39ade596f4@quicinc.com>
 <86cz0ywx5p.wl-maz@kernel.org>
 <9e69158f-1e67-344b-fea9-85bb5e21183f@quicinc.com>
 <87h6q9a8pu.wl-maz@kernel.org>
From:   "Aiqun(Maria) Yu" <quic_aiquny@quicinc.com>
In-Reply-To: <87h6q9a8pu.wl-maz@kernel.org>
Content-Type: text/plain; charset="UTF-8"; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 7/12/2023 3:29 PM, Marc Zyngier wrote:
> On Wed, 12 Jul 2023 03:47:55 +0100,
> "Aiqun(Maria) Yu" <quic_aiquny@quicinc.com> wrote:
>>
>> On 7/11/2023 6:38 PM, Marc Zyngier wrote:
>>> On Tue, 11 Jul 2023 11:12:48 +0100,
>>> "Aiqun(Maria) Yu" <quic_aiquny@quicinc.com> wrote:
>>>>
>>>> For the KVM part, per my understanding, as long as the current feature
>>>> id being overriden, the KVM system also get the current vcpu without
>>>> the lse atomic feature enabled.
>>>> KVM vcpu will read the sys reg from host arm64_ftr_regs which is
>>>> already been controled by the idreg_overrides.
>>>
>>> You're completely missing the point.
>>>
>>> The guest is free to map memory as non-cacheable *and* to use LSE
>>> atomics even if the idregs pretend this is not available. At which
>> The guest also can have the current linux kernel mechanism of LSE
>> ATOMIC way.
> 
> [snip useless diagrams]
> 
> Yes, the guest can do the right thing. The guest, a totally
> unprivileged piece of SW, can also ignore the idregs and take the
> whole machine down because your HW is broken.
> 
if the guest ignore the idregs, it is not supported by the current Linux 
KVM id reg emulation as well. The similar rule is applied to other cpu 
feature as well.

So it can be an expected machine down because of this.

We want to support/utilize the current HW with current inline runtime 
patching for lse atomic ops.
>> Just like other KVM vcpu cpu features, lse atomic can be a feature
>> inherit from the pysical cpu features for the KVM vcpus.
> 
> See above. Your reasoning applies to a well behaved guest, which is
> the *wrong* way to reason about these things.

The feature supported is not always that *freely* even for current cpu 
features as well.
Our current target is that the software can utilize the HW as best as 
software can.

The current HW can be possible with Generic common Image with other cpu 
which support lse atomic. So the Image can have inline runtime patching 
for lse atomic operations. And from software side it can have option to 
support this.

For example, for current newer memory controller the far lse atomic 
operations is supported, and the atomic operation is not limited to 
non-cached memory mapping as well.

Also the lse atomic instead of FWB performance in specific scenarios can 
be different with current hardware design as well.

we are trying to do possible improvement with HW design change instead 
of ruin it.
Feel free to comment if it is not same understanding.
> 
> 	M.
> 

-- 
Thx and BRs,
Aiqun(Maria) Yu