public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Chao Gao <chao.gao@intel.com>
To: Dave Hansen <dave.hansen@intel.com>
Cc: <tglx@linutronix.de>, <x86@kernel.org>, <seanjc@google.com>,
	<pbonzini@redhat.com>, <linux-kernel@vger.kernel.org>,
	<kvm@vger.kernel.org>, <peterz@infradead.org>,
	<rick.p.edgecombe@intel.com>, <weijiang.yang@intel.com>,
	<john.allen@amd.com>, <bp@alien8.de>
Subject: Re: [PATCH v3 00/10] Introduce CET supervisor state support
Date: Tue, 18 Mar 2025 23:24:39 +0800	[thread overview]
Message-ID: <Z9mQN3gPpcWPGTwA@intel.com> (raw)
In-Reply-To: <b9389b35-c1ef-4a53-9eb2-051df0aaf33d@intel.com>

On Fri, Mar 07, 2025 at 11:09:42AM -0800, Dave Hansen wrote:
>On 3/7/25 08:41, Chao Gao wrote:
>> case |IA32_XSS[12] | Space | RFBM[12] | Drop%	
>> -----+-------------+-------+----------+------
>>   1  |	   0	   | None  |	0     |  0.0%
>>   2  |	   1	   | None  |	0     |  0.2%
>>   3  |	   1	   | 24B?  |	1     |  0.2%
>
>So, 0.2% is still, what, dozens of cycles? Are you sure that it really
>takes the CPU dozens of cycles to skip over the feature during XSAVE?
>
>If it really turns out to be this measurable, we should probably follow
>up with the folks that implement XSAVE and see what's going on under the
>covers.

I reran the performance tests and observed a run-to-run variation of 0.4%
to 0.7%. So, I don't think there is any measurable performance difference.
I will update the performance statements in the cover letter.

>
>On a separate note, I was bugging Thomas a bit on IRC. His memory was
>that the AMX-era FPU rework only expected KVM to support user features.
>You might want to dig through the history a bit and see if _that_ was
>ever properly addressed because that would change the problem you're
>trying to solve.

I went through the email discussions and found only one relevant thread:

https://lore.kernel.org/kvm/87wnmf66m5.ffs@tglx/#t

where Thomas mentioned that guest_perm would be set as follows:

	guest_fpu::__state_perm = supported_xcr0 & xstate_get_group_perm();

If implemented this way, KVM would only support user features. However, the
committed change is:

980fe2fddcff ("x86/fpu: Extend fpu_xstate_prctl() with guest permissions")

In this change, fpu->guest_perm is copied from fpu->perm:

+	/* Same defaults for guests */
+	fpu->guest_perm = fpu->perm;

There are indeed some issues with enabling supervisor features for guest
FPUs, but they have been addressed by recent changes in the tip-tree ([1],
[2]) and patch 1 of this series.

[1]: https://lore.kernel.org/all/20250218141045.85201-1-stanspas@amazon.de/
[2]: https://lore.kernel.org/all/20250317140613.1761633-1-chao.gao@intel.com/

      reply	other threads:[~2025-03-18 15:24 UTC|newest]

Thread overview: 40+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-03-07 16:41 [PATCH v3 00/10] Introduce CET supervisor state support Chao Gao
2025-03-07 16:41 ` [PATCH v3 01/10] x86/fpu/xstate: Always preserve non-user xfeatures/flags in __state_perm Chao Gao
2025-03-07 16:41 ` [PATCH v3 02/10] x86/fpu/xstate: Drop @perm from guest pseudo FPU container Chao Gao
2025-03-07 16:41 ` [PATCH v3 03/10] x86/fpu/xstate: Correct xfeatures cache in guest pseudo fpu container Chao Gao
2025-03-07 17:48   ` Dave Hansen
2025-03-08  2:44     ` Chao Gao
2025-03-07 16:41 ` [PATCH v3 04/10] x86/fpu/xstate: Correct guest fpstate size calculation Chao Gao
2025-03-07 17:53   ` Dave Hansen
2025-03-08  2:56     ` Chao Gao
2025-03-07 21:37   ` Chang S. Bae
2025-03-08  2:49     ` Chao Gao
2025-03-09 22:06       ` Chang S. Bae
2025-03-10  1:33         ` Chao Gao
2025-03-10  5:21           ` Chang S. Bae
2025-03-10  7:06             ` Chao Gao
2025-03-10 17:33               ` Chang S. Bae
2025-03-11 12:09                 ` Chao Gao
2025-03-12  1:03                   ` Chang S. Bae
2025-03-07 16:41 ` [PATCH v3 05/10] x86/fpu/xstate: Introduce guest FPU configuration Chao Gao
2025-03-07 18:06   ` Dave Hansen
2025-03-08  3:00     ` Chao Gao
2025-03-07 16:41 ` [PATCH v3 06/10] x86/fpu/xstate: Initialize guest perm with fpu_guest_cfg Chao Gao
2025-03-07 18:14   ` Dave Hansen
2025-03-08  3:14     ` Chao Gao
2025-03-07 16:41 ` [PATCH v3 07/10] x86/fpu/xstate: Initialize guest fpstate with fpu_guest_config Chao Gao
2025-03-07 18:41   ` Dave Hansen
2025-03-08  3:38     ` Chao Gao
2025-03-07 16:41 ` [PATCH v3 08/10] x86/fpu/xstate: Add CET supervisor xfeature support Chao Gao
2025-03-07 18:39   ` Dave Hansen
2025-03-08  3:24     ` Chao Gao
2025-03-07 16:41 ` [PATCH v3 09/10] x86/fpu/xstate: Introduce XFEATURE_MASK_KERNEL_DYNAMIC xfeature set Chao Gao
2025-03-09 22:06   ` Chang S. Bae
2025-03-10  3:49     ` Chao Gao
2025-03-10  5:20       ` Chang S. Bae
2025-03-10  5:53         ` Chao Gao
2025-03-11 12:27           ` Chao Gao
2025-03-12  1:03             ` Chang S. Bae
2025-03-07 16:41 ` [PATCH v3 10/10] x86/fpu/xstate: Warn if CET supervisor state is detected in normal fpstate Chao Gao
2025-03-07 19:09 ` [PATCH v3 00/10] Introduce CET supervisor state support Dave Hansen
2025-03-18 15:24   ` Chao Gao [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=Z9mQN3gPpcWPGTwA@intel.com \
    --to=chao.gao@intel.com \
    --cc=bp@alien8.de \
    --cc=dave.hansen@intel.com \
    --cc=john.allen@amd.com \
    --cc=kvm@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=pbonzini@redhat.com \
    --cc=peterz@infradead.org \
    --cc=rick.p.edgecombe@intel.com \
    --cc=seanjc@google.com \
    --cc=tglx@linutronix.de \
    --cc=weijiang.yang@intel.com \
    --cc=x86@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox