Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Thu, 09 Sep 2021 21:51:55 -0700
From:   "H. Peter Anvin" <hpa@zytor.com>
To:     Lai Jiangshan <laijs@linux.alibaba.com>,
        Lai Jiangshan <jiangshanlai@gmail.com>,
        linux-kernel@vger.kernel.org
CC:     Andy Lutomirski <luto@kernel.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
        x86@kernel.org, Joerg Roedel <jroedel@suse.de>,
        Youquan Song <youquan.song@intel.com>,
        Tony Luck <tony.luck@intel.com>,
        Juergen Gross <jgross@suse.com>,
        "Peter Zijlstra (Intel)" <peterz@infradead.org>
Subject: Re: [PATCH 17/24] x86/entry: Introduce struct ist_regs
User-Agent: K-9 Mail for Android
In-Reply-To: <286DCB70-B36E-4229-966E-BE45F2AEA703@zytor.com>
References: <20210831175025.27570-1-jiangshanlai@gmail.com> <20210831175025.27570-18-jiangshanlai@gmail.com> <eb294b5d-82f2-be80-b3e3-db556c155d95@linux.alibaba.com> <286DCB70-B36E-4229-966E-BE45F2AEA703@zytor.com>
Message-ID: <4C8A910F-679F-4F49-AAA5-AA3B9BE5F2BD@zytor.com>
MIME-Version: 1.0
Content-Type: text/plain;
 charset=utf-8
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Note: with FRED paranoid entry is no longer needed since the hardware enfor=
ces gs_base mode consistency=2E

On September 9, 2021 9:50:40 PM PDT, "H=2E Peter Anvin" <hpa@zytor=2Ecom> =
wrote:
>It may affect your design, I don't know, but FRED exception delivery (as =
currently architected per draft 2=2E0; it is subject to change still) pushe=
s several fields above the conventional stack frame (and thus above pt_regs=
=2E)
>
>On September 9, 2021 5:18:47 PM PDT, Lai Jiangshan <laijs@linux=2Ealibaba=
=2Ecom> wrote:
>>
>>
>>On 2021/9/1 01:50, Lai Jiangshan wrote:
>>> From: Lai Jiangshan <laijs@linux=2Ealibaba=2Ecom>
>>>=20
>>> struct ist_regs is the upmost stack frame for IST interrupt, it
>>> consists of struct pt_regs and other fields which will be added in
>>> later patch=2E
>>>=20
>>> Make vc_switch_off_ist() take ist_regs as its argument and it will swi=
tch
>>> the whole ist_regs if needed=2E
>>>=20
>>> Make the frame before calling paranoid_entry() and paranoid_exit() be
>>> struct ist_regs=2E
>>>=20
>>> This patch is prepared for converting paranoid_entry() and paranoid_ex=
it()
>>> into C code which will need the additinal fields to store the results =
in
>>> paranoid_entry() and to use them in paranoid_exit()=2E
>>
>>This patch was over designed=2E
>>
>>In ASM code, we can easily save results in the callee-saved registers=2E
>>For example, rc3 is saved in %r14, gsbase info is saved in %rbx=2E
>>
>>And in C code, we can't save results in registers=2E  And I thought ther=
e was
>>no place to save the results because the CR3 and gsbase are not kernel's=
=2E
>>So I extended the pt_regs to ist_regs to save the results=2E
>>
>>But it was incorrect=2E  The results can be saved in percpu data at the =
end of
>>paranoid_entry() after the CR3/gsbase are settled down=2E  And the resul=
ts
>>can be read at the beginning of paranoid_exit() before the CR3/gsbase ar=
e
>>switched to the interrupted context's=2E
>>
>>sigh=2E
>>
>>>=20
>>> The C interrupt handlers don't use struct ist_regs due to they don't n=
eed
>>> the additional fields in the struct ist_regs, and they still use pt_re=
gs=2E
>>>=20
>>
>

--=20
Sent from my Android device with K-9 Mail=2E Please excuse my brevity=2E