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;
From:   Daniel Axtens <dja@axtens.net>
To:     Christophe Leroy <christophe.leroy@csgroup.eu>,
        Benjamin Herrenschmidt <benh@kernel.crashing.org>,
        Paul Mackerras <paulus@samba.org>,
        Michael Ellerman <mpe@ellerman.id.au>
Cc:     linuxppc-dev@lists.ozlabs.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] powerpc/64: Avoid link stack corruption in kexec_wait()
In-Reply-To: <9f43118b-ef81-9a80-0e0b-5e74433e0b8c@csgroup.eu>
References: <3ffe7775f3fcda8e5fca6a7bc7db0b8251153c67.1629705147.git.christophe.leroy@csgroup.eu>
 <87ilzm6str.fsf@dja-thinkpad.axtens.net>
 <9f43118b-ef81-9a80-0e0b-5e74433e0b8c@csgroup.eu>
Date:   Tue, 31 Aug 2021 22:43:04 +1000
Message-ID: <87fsup7pk7.fsf@dja-thinkpad.axtens.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Christophe Leroy <christophe.leroy@csgroup.eu> writes:

> Le 31/08/2021 =C3=A0 08:17, Daniel Axtens a =C3=A9crit=C2=A0:
>> Hi Christophe,
>>=20
>>> Use bcl 20,31,+4 instead of bl in order to preserve link stack.
>>>
>>> See commit c974809a26a1 ("powerpc/vdso: Avoid link stack corruption
>>> in __get_datapage()") for details.
>>=20
>>  From my understanding of that commit message, the change helps to keep
>> the link stack correctly balanced which is helpful for performance,
>> rather than for correctness. If I understand correctly, kexec_wait is
>> not in a hot path - rather it is where CPUs spin while waiting for
>> kexec. Is there any benefit in using the more complicated opcode in this
>> situation?
>
> AFAICS the main benefit is to keep things consistent over the kernel and =
not have to wonder "is it a=20
> hot path or not ? If it is I use bcl 20,31, if it is not I use bl". The b=
est way to keep things in=20
> order is to always use the right instruction.

Yeah, Nick Piggin convinced me of this offline as well.

>
>>=20
>>> Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
>>> ---
>>>   arch/powerpc/kernel/misc_64.S | 2 +-
>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/arch/powerpc/kernel/misc_64.S b/arch/powerpc/kernel/misc_6=
4.S
>>> index 4b761a18a74d..613509907166 100644
>>> --- a/arch/powerpc/kernel/misc_64.S
>>> +++ b/arch/powerpc/kernel/misc_64.S
>>> @@ -255,7 +255,7 @@ _GLOBAL(scom970_write)
>>>    * Physical (hardware) cpu id should be in r3.
>>>    */
>>>   _GLOBAL(kexec_wait)
>>> -	bl	1f
>>> +	bcl	20,31,1f
>>>   1:	mflr	r5
>>=20
>> Would it be better to create a macro of some sort to wrap this unusual
>> special form so that the meaning is more clear?
>
> Not sure, I think people working with assembly will easily recognise that=
 form whereas an obscure=20
> macro is always puzzling.
>
> I like macros when they allow you to not repeat again and again the same =
sequence of several=20
> instructions, but here it is a single quite simple instruction which is n=
ot worth a macro in my mind.
>


Sure - I was mostly thinking specifically of the bcl; mflr situation but
I agree that for the single instruction it's not needed.

In short, I am convinced, and so:
Reviewed-by: Daniel Axtens <dja@axtens.net>

Kind regards,
Daniel

> Christophe