Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Subject: Re: [PATCH 10/12] powerpc/entry32: Blacklist exception entry points
 for kprobe.
From:   Christophe Leroy <christophe.leroy@c-s.fr>
To:     "Naveen N. Rao" <naveen.n.rao@linux.vnet.ibm.com>,
        Benjamin Herrenschmidt <benh@kernel.crashing.org>,
        Michael Ellerman <mpe@ellerman.id.au>,
        Paul Mackerras <paulus@samba.org>
Cc:     linuxppc-dev@lists.ozlabs.org, linux-kernel@vger.kernel.org
References: <dff05b59a161434a546010507000816750073f28.1585474724.git.christophe.leroy@c-s.fr>
 <aea027844b12fcbc29ea78d26c5848a6794d1688.1585474724.git.christophe.leroy@c-s.fr>
 <1585588031.jvow7mwq4x.naveen@linux.ibm.com>
 <7f367f35-1bb8-bbb6-f399-8e911f76e043@c-s.fr>
Message-ID: <83053ddf-9ba6-d551-6711-890c3f3810b5@c-s.fr>
Date:   Tue, 31 Mar 2020 07:51:35 +0200
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:68.0) Gecko/20100101
 Thunderbird/68.6.0
MIME-Version: 1.0
In-Reply-To: <7f367f35-1bb8-bbb6-f399-8e911f76e043@c-s.fr>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Language: fr
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk


Le 30/03/2020 à 20:33, Christophe Leroy a écrit :
> 
> 
> Le 30/03/2020 à 19:08, Naveen N. Rao a écrit :
>> Christophe Leroy wrote:
>>> kprobe does not handle events happening in real mode.
>>>
>>> As exception entry points are running with MMU disabled,
>>> blacklist them.
>>>
>>> Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
>>> ---
>>>  arch/powerpc/kernel/entry_32.S | 7 +++++++
>>>  1 file changed, 7 insertions(+)
>>>
>>> diff --git a/arch/powerpc/kernel/entry_32.S 
>>> b/arch/powerpc/kernel/entry_32.S
>>> index 94f78c03cb79..9a1a45d6038a 100644
>>> --- a/arch/powerpc/kernel/entry_32.S
>>> +++ b/arch/powerpc/kernel/entry_32.S
>>> @@ -51,6 +51,7 @@ mcheck_transfer_to_handler:
>>>      mfspr    r0,SPRN_DSRR1
>>>      stw    r0,_DSRR1(r11)
>>>      /* fall through */
>>> +_ASM_NOKPROBE_SYMBOL(mcheck_transfer_to_handler)
>>>
>>>      .globl    debug_transfer_to_handler
>>>  debug_transfer_to_handler:
>>> @@ -59,6 +60,7 @@ debug_transfer_to_handler:
>>>      mfspr    r0,SPRN_CSRR1
>>>      stw    r0,_CSRR1(r11)
>>>      /* fall through */
>>> +_ASM_NOKPROBE_SYMBOL(debug_transfer_to_handler)
>>>
>>>      .globl    crit_transfer_to_handler
>>>  crit_transfer_to_handler:
>>> @@ -94,6 +96,7 @@ crit_transfer_to_handler:
>>>      rlwinm    r0,r1,0,0,(31 - THREAD_SHIFT)
>>>      stw    r0,KSP_LIMIT(r8)
>>>      /* fall through */
>>> +_ASM_NOKPROBE_SYMBOL(crit_transfer_to_handler)
>>>  #endif
>>>
>>>  #ifdef CONFIG_40x
>>> @@ -115,6 +118,7 @@ crit_transfer_to_handler:
>>>      rlwinm    r0,r1,0,0,(31 - THREAD_SHIFT)
>>>      stw    r0,KSP_LIMIT(r8)
>>>      /* fall through */
>>> +_ASM_NOKPROBE_SYMBOL(crit_transfer_to_handler)
>>>  #endif
>>>
>>>  /*
>>> @@ -127,6 +131,7 @@ crit_transfer_to_handler:
>>>      .globl    transfer_to_handler_full
>>>  transfer_to_handler_full:
>>>      SAVE_NVGPRS(r11)
>>> +_ASM_NOKPROBE_SYMBOL(transfer_to_handler_full)
>>>      /* fall through */
>>>
>>>      .globl    transfer_to_handler
>>> @@ -286,6 +291,8 @@ reenable_mmu:
>>>      lwz    r2, GPR2(r11)
>>>      b    fast_exception_return
>>>  #endif
>>> +_ASM_NOKPROBE_SYMBOL(transfer_to_handler)
>>> +_ASM_NOKPROBE_SYMBOL(transfer_to_handler_cont)
>>
>> These are added after 'reenable_mmu', which is itself not blacklisted. 
>> Is that intentional?
> 
> Yes I put it as the complete end of the entry part, ie just before 
> stack_ovf which is a function by itself.
> 
> Note that reenable_mmu is inside an #ifdef CONFIG_TRACE_IRQFLAGS.
> 
> I'm not completely sure where to put the _ASM_NOKPROBE_SYMBOL()s, that's 
> the reason why I put it close to the symbol itself in my first series.
> 
> Could you have a look at the code and tell me what looks the most 
> appropriate as a location to you ?
> 
> https://elixir.bootlin.com/linux/v5.6/source/arch/powerpc/kernel/entry_32.S#L230 

Ok, thinking about it once more, I guess we have a problem as everything 
after that reenable_mmu will be visible.

So I'll respin

Christophe