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:   Wed, 10 Feb 2021 12:00:20 +1000
From:   Nicholas Piggin <npiggin@gmail.com>
Subject: Re: [PATCH v5 20/22] powerpc/syscall: Avoid storing 'current' in
 another pointer
To:     Christophe Leroy <christophe.leroy@csgroup.eu>,
        David Laight <David.Laight@ACULAB.COM>,
        'Segher Boessenkool' <segher@kernel.crashing.org>
Cc:     "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "linuxppc-dev@lists.ozlabs.org" <linuxppc-dev@lists.ozlabs.org>,
        "msuchanek@suse.de" <msuchanek@suse.de>,
        Paul Mackerras <paulus@samba.org>
References: <cover.1612796617.git.christophe.leroy@csgroup.eu>
        <24804747098369ebcdac38970b8f7a1260bdd248.1612796617.git.christophe.leroy@csgroup.eu>
        <1612838134.rvncv9kzls.astroid@bobo.none>
        <20210209135053.GD27854@gate.crashing.org>
        <d35cc46eca474b2c9f94a4de269321e7@AcuMS.aculab.com>
        <f6ae2e77-3a89-c294-9a6e-58d84fbb46b1@csgroup.eu>
In-Reply-To: <f6ae2e77-3a89-c294-9a6e-58d84fbb46b1@csgroup.eu>
MIME-Version: 1.0
Message-Id: <1612922312.mnpowzgd0r.astroid@bobo.none>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Excerpts from Christophe Leroy's message of February 10, 2021 3:03 am:
>=20
>=20
> Le 09/02/2021 =C3=A0 15:31, David Laight a =C3=A9crit=C2=A0:
>> From: Segher Boessenkool
>>> Sent: 09 February 2021 13:51
>>>
>>> On Tue, Feb 09, 2021 at 12:36:20PM +1000, Nicholas Piggin wrote:
>>>> What if you did this?
>>>
>>>> +static inline struct task_struct *get_current(void)
>>>> +{
>>>> +	register struct task_struct *task asm ("r2");
>>>> +
>>>> +	return task;
>>>> +}
>>>
>>> Local register asm variables are *only* guaranteed to live in that
>>> register as operands to an asm.  See
>>>    https://gcc.gnu.org/onlinedocs/gcc/Local-Register-Variables.html#Loc=
al-Register-Variables
>>> ("The only supported use" etc.)
>>>
>>> You can do something like
>>>
>>> static inline struct task_struct *get_current(void)
>>> {
>>> 	register struct task_struct *task asm ("r2");
>>>
>>> 	asm("" : "+r"(task));
>>>
>>> 	return task;
>>> }
>>>
>>> which makes sure that "task" actually is in r2 at the point of that asm=
.
>>=20
>> If "r2" always contains current (and is never assigned by the compiler)
>> why not use a global register variable for it?
>>=20
>=20
>=20
> The change proposed by Nick doesn't solve the issue.

It seemed to change code generation in a simple test case, oh well.

>=20
> The problem is that at the begining of the function we have:
>=20
> 	unsigned long *ti_flagsp =3D &current_thread_info()->flags;
>=20
> When the function uses ti_flagsp for the first time, it does use 112(r2)
>=20
> Then the function calls some other functions.
>=20
> Most likely because the function could update 'current', GCC copies r2 in=
to r30, so that if r2 get=20
> changed by the called function, ti_flagsp is still based on the previous =
value of current.
>=20
> Allthough we know r2 wont change, GCC doesn't know it. And in order to sa=
ve r2 into r30, it needs to=20
> save r30 in the stack.
>=20
>=20
> By using &current_thread_info()->flags directly instead of this intermedi=
aite ti_flagsp pointer, GCC=20
> uses r2 instead instead of doing a copy.
>=20
>=20
> Nick, I don't understand the reason why you need that 'ti_flagsp' local v=
ar.

Just to save typing, I don't mind your patch I was just wondering if=20
current could be improved in general.

Thanks,
Nick