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;
Date:   Tue, 24 May 2022 16:01:48 +0530
From:   "Naveen N. Rao" <naveen.n.rao@linux.vnet.ibm.com>
Subject: Re: [RFC PATCH 1/4] objtool: Add --mnop as an option to --mcount
To:     Christophe Leroy <christophe.leroy@csgroup.eu>,
        "linuxppc-dev@lists.ozlabs.org" <linuxppc-dev@lists.ozlabs.org>,
        Sathvika Vasireddy <sv@linux.ibm.com>
Cc:     "aik@ozlabs.ru" <aik@ozlabs.ru>,
        "jpoimboe@redhat.com" <jpoimboe@redhat.com>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "mbenes@suse.cz" <mbenes@suse.cz>,
        "mpe@ellerman.id.au" <mpe@ellerman.id.au>,
        "peterz@infradead.org" <peterz@infradead.org>,
        "rostedt@goodmis.org" <rostedt@goodmis.org>
References: <20220523175548.922671-1-sv@linux.ibm.com>
        <20220523175548.922671-2-sv@linux.ibm.com>
        <26c7bfc8-3089-034a-70c0-8857d7cd3a99@csgroup.eu>
        <1653386854.o7nss9hzc9.naveen@linux.ibm.com>
        <d45030be-3f6b-ebeb-3d63-bf7a96d3ff3b@csgroup.eu>
In-Reply-To: <d45030be-3f6b-ebeb-3d63-bf7a96d3ff3b@csgroup.eu>
MIME-Version: 1.0
User-Agent: astroid/4d6b06ad (https://github.com/astroidmail/astroid)
Message-Id: <1653388084.w21cyb07gc.naveen@linux.ibm.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Christophe Leroy wrote:
>=20
>=20
> Le 24/05/2022 =C3=A0 12:15, Naveen N. Rao a =C3=A9crit=C2=A0:
>> Christophe Leroy wrote:
>>>
>>>
>>> Le 23/05/2022 =C3=A0 19:55, Sathvika Vasireddy a =C3=A9crit=C2=A0:
>>>> Architectures can select HAVE_NOP_MCOUNT if they choose
>>>> to nop out mcount call sites. If that config option is
>>>> selected, then --mnop is passed as an option to objtool,
>>>> along with --mcount.
>>>>
>>>
>>> Is there a reason not to nop out mcount call sites on powerpc as well ?
>>=20
>> Yes, if there are functions that are out of range of _mcount(), then the=
=20
>> linker would have inserted long branch trampolines. We detect such cases=
=20
>> during boot. But, if we nop out the _mcount call sites during build=20
>> time, we will need some other way to identify these.
>>=20
>=20
> But does it really matter whether _mcount is reachable or not ?
>=20
> _mcount is never used, and the function we want to call in lieu of=20
> _mcount might be reachable while _mcount is not or might be unreachable=20
> while _mcount is.

For the most part, we will end up having to go to ftrace_caller or=20
ftrace_regs_caller, both of which will usually be close to _mcount.

We need to know for sure either way. Nop'ing out the _mcount locations=20
at boot allows us to discover existing long branch trampolines. If we=20
want to avoid it, we need to note down those locations during build=20
time.

Do you have a different approach in mind?


- Naveen