From:   Michael Ellerman <mpe@ellerman.id.au>
To:     "Naveen N. Rao" <naveen.n.rao@linux.vnet.ibm.com>,
        kernel test robot <lkp@intel.com>,
        Sathvika Vasireddy <sv@linux.ibm.com>
Cc:     Christophe Leroy <christophe.leroy@csgroup.eu>,
        linux-kernel@vger.kernel.org, oe-kbuild-all@lists.linux.dev,
        linuxppc-dev@lists.ozlabs.org, npiggin@gmail.com
Subject: Re: arch/powerpc/kernel/head_85xx.o: warning: objtool:
 .head.text+0x1a6c: unannotated intra-function call
In-Reply-To: <1674631223.9e09lbzzb6.naveen@linux.ibm.com>
References: <202301161955.38kK6ksW-lkp@intel.com>
 <b2273730-f885-7658-7ec4-12fb5bfc515b@linux.ibm.com>
 <1674631223.9e09lbzzb6.naveen@linux.ibm.com>
Date:   Wed, 25 Jan 2023 19:31:46 +1100
Message-ID: <87v8kvow0d.fsf@mpe.ellerman.id.au>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

"Naveen N. Rao" <naveen.n.rao@linux.vnet.ibm.com> writes:
> Sathvika Vasireddy wrote:
>>=20
>>>>> arch/powerpc/kvm/booke.o: warning: objtool: kvmppc_fill_pt_regs+0x30:=
 unannotated intra-function call
>>=20
>> As an attempt to fix it, I tried expanding ANNOTATE_INTRA_FUNCTION_CALL=
=20
>> macro to indicate that the branch target is valid. It then threw another=
=20
>> warning (arch/powerpc/kvm/booke.o: warning: objtool:=20
>> kvmppc_fill_pt_regs+0x38: intra_function_call not a direct call). The=20
>> below diff just removes the warnings for me, but I'm not very sure if=20
>> this is the best way to fix the objtool warnings seen with this=20
>> particular file. Please let me know if there are any better ways to fix =
it.
>>=20
>> diff --git a/arch/powerpc/kvm/booke.c b/arch/powerpc/kvm/booke.c
>> index 0dce93ccaadf..b6a413824b98 100644
>> --- a/arch/powerpc/kvm/booke.c
>> +++ b/arch/powerpc/kvm/booke.c
>> @@ -917,7 +917,9 @@ static void kvmppc_fill_pt_regs(struct pt_regs *regs)
>>  =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 asm("mr %0, 1" : "=3Dr"(r1));
>>  =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 asm("mflr %0" : "=3Dr"(lr));
>>  =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 asm("mfmsr %0" : "=3Dr"(msr)=
);
>> +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 asm(".pushsection .discard.intra_f=
unction_calls; .long 999f;=20
>> .popsection; 999:");
>>  =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 asm("bl 1f; 1: mflr %0" : "=
=3Dr"(ip));
>
> I don't think you can assume that there won't be anything in between two=
=20
> asm statements.

Yeah, compiler could interleave something theoretically.

> Even if that works, I don't think it is good to expand the macro here.=20=
=20
> That asm statement looks to be trying to grab the current nip. I don't=20
> know enough about that code, and someone who knows more about KVM may be=
=20
> able to help, but it looks like we should be able to simply set 'ip' to=20
> the address of kvmppc_fill_pt_regs()?

There is _THIS_IP_ which should be sufficient.

cheers