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] powerpc: Perform a bounds check in arch_add_memory
To:     Alastair D'Silva <alastair@au1.ibm.com>,
        Michal Hocko <mhocko@kernel.org>
Cc:     Benjamin Herrenschmidt <benh@kernel.crashing.org>,
        Paul Mackerras <paulus@samba.org>,
        Michael Ellerman <mpe@ellerman.id.au>,
        Christophe Leroy <christophe.leroy@c-s.fr>,
        Andrew Morton <akpm@linux-foundation.org>,
        Mike Rapoport <rppt@linux.vnet.ibm.com>,
        "Aneesh Kumar K.V" <aneesh.kumar@linux.ibm.com>,
        linuxppc-dev@lists.ozlabs.org, linux-kernel@vger.kernel.org
References: <20190827052047.31547-1-alastair@au1.ibm.com>
 <20190827062844.GQ7538@dhcp22.suse.cz>
 <ea9e43fff6b6531af0620f9df62e015af66d4535.camel@au1.ibm.com>
 <1f2d967a-57a1-d3a3-4eb7-306b43709fee@redhat.com>
 <7e8e34ece6386bd3b0703f218a3b4688c83886d7.camel@au1.ibm.com>
 <5ce9f700-8e70-c669-32fa-9598a01eb8ba@redhat.com>
 <6823c187257b5033bca8905dda17b6c79a4944a6.camel@au1.ibm.com>
From:   David Hildenbrand <david@redhat.com>
Openpgp: preference=signencrypt
Autocrypt: addr=david@redhat.com; prefer-encrypt=mutual; keydata=
 xsFNBFXLn5EBEAC+zYvAFJxCBY9Tr1xZgcESmxVNI/0ffzE/ZQOiHJl6mGkmA1R7/uUpiCjJ
 dBrn+lhhOYjjNefFQou6478faXE6o2AhmebqT4KiQoUQFV4R7y1KMEKoSyy8hQaK1umALTdL
 QZLQMzNE74ap+GDK0wnacPQFpcG1AE9RMq3aeErY5tujekBS32jfC/7AnH7I0v1v1TbbK3Gp
 XNeiN4QroO+5qaSr0ID2sz5jtBLRb15RMre27E1ImpaIv2Jw8NJgW0k/D1RyKCwaTsgRdwuK
 Kx/Y91XuSBdz0uOyU/S8kM1+ag0wvsGlpBVxRR/xw/E8M7TEwuCZQArqqTCmkG6HGcXFT0V9
 PXFNNgV5jXMQRwU0O/ztJIQqsE5LsUomE//bLwzj9IVsaQpKDqW6TAPjcdBDPLHvriq7kGjt
 WhVhdl0qEYB8lkBEU7V2Yb+SYhmhpDrti9Fq1EsmhiHSkxJcGREoMK/63r9WLZYI3+4W2rAc
 UucZa4OT27U5ZISjNg3Ev0rxU5UH2/pT4wJCfxwocmqaRr6UYmrtZmND89X0KigoFD/XSeVv
 jwBRNjPAubK9/k5NoRrYqztM9W6sJqrH8+UWZ1Idd/DdmogJh0gNC0+N42Za9yBRURfIdKSb
 B3JfpUqcWwE7vUaYrHG1nw54pLUoPG6sAA7Mehl3nd4pZUALHwARAQABzSREYXZpZCBIaWxk
 ZW5icmFuZCA8ZGF2aWRAcmVkaGF0LmNvbT7CwX4EEwECACgFAljj9eoCGwMFCQlmAYAGCwkI
 BwMCBhUIAgkKCwQWAgMBAh4BAheAAAoJEE3eEPcA/4Na5IIP/3T/FIQMxIfNzZshIq687qgG
 8UbspuE/YSUDdv7r5szYTK6KPTlqN8NAcSfheywbuYD9A4ZeSBWD3/NAVUdrCaRP2IvFyELj
 xoMvfJccbq45BxzgEspg/bVahNbyuBpLBVjVWwRtFCUEXkyazksSv8pdTMAs9IucChvFmmq3
 jJ2vlaz9lYt/lxN246fIVceckPMiUveimngvXZw21VOAhfQ+/sofXF8JCFv2mFcBDoa7eYob
 s0FLpmqFaeNRHAlzMWgSsP80qx5nWWEvRLdKWi533N2vC/EyunN3HcBwVrXH4hxRBMco3jvM
 m8VKLKao9wKj82qSivUnkPIwsAGNPdFoPbgghCQiBjBe6A75Z2xHFrzo7t1jg7nQfIyNC7ez
 MZBJ59sqA9EDMEJPlLNIeJmqslXPjmMFnE7Mby/+335WJYDulsRybN+W5rLT5aMvhC6x6POK
 z55fMNKrMASCzBJum2Fwjf/VnuGRYkhKCqqZ8gJ3OvmR50tInDV2jZ1DQgc3i550T5JDpToh
 dPBxZocIhzg+MBSRDXcJmHOx/7nQm3iQ6iLuwmXsRC6f5FbFefk9EjuTKcLMvBsEx+2DEx0E
 UnmJ4hVg7u1PQ+2Oy+Lh/opK/BDiqlQ8Pz2jiXv5xkECvr/3Sv59hlOCZMOaiLTTjtOIU7Tq
 7ut6OL64oAq+zsFNBFXLn5EBEADn1959INH2cwYJv0tsxf5MUCghCj/CA/lc/LMthqQ773ga
 uB9mN+F1rE9cyyXb6jyOGn+GUjMbnq1o121Vm0+neKHUCBtHyseBfDXHA6m4B3mUTWo13nid
 0e4AM71r0DS8+KYh6zvweLX/LL5kQS9GQeT+QNroXcC1NzWbitts6TZ+IrPOwT1hfB4WNC+X
 2n4AzDqp3+ILiVST2DT4VBc11Gz6jijpC/KI5Al8ZDhRwG47LUiuQmt3yqrmN63V9wzaPhC+
 xbwIsNZlLUvuRnmBPkTJwwrFRZvwu5GPHNndBjVpAfaSTOfppyKBTccu2AXJXWAE1Xjh6GOC
 8mlFjZwLxWFqdPHR1n2aPVgoiTLk34LR/bXO+e0GpzFXT7enwyvFFFyAS0Nk1q/7EChPcbRb
 hJqEBpRNZemxmg55zC3GLvgLKd5A09MOM2BrMea+l0FUR+PuTenh2YmnmLRTro6eZ/qYwWkC
 u8FFIw4pT0OUDMyLgi+GI1aMpVogTZJ70FgV0pUAlpmrzk/bLbRkF3TwgucpyPtcpmQtTkWS
 gDS50QG9DR/1As3LLLcNkwJBZzBG6PWbvcOyrwMQUF1nl4SSPV0LLH63+BrrHasfJzxKXzqg
 rW28CTAE2x8qi7e/6M/+XXhrsMYG+uaViM7n2je3qKe7ofum3s4vq7oFCPsOgwARAQABwsFl
 BBgBAgAPBQJVy5+RAhsMBQkJZgGAAAoJEE3eEPcA/4NagOsP/jPoIBb/iXVbM+fmSHOjEshl
 KMwEl/m5iLj3iHnHPVLBUWrXPdS7iQijJA/VLxjnFknhaS60hkUNWexDMxVVP/6lbOrs4bDZ
 NEWDMktAeqJaFtxackPszlcpRVkAs6Msn9tu8hlvB517pyUgvuD7ZS9gGOMmYwFQDyytpepo
 YApVV00P0u3AaE0Cj/o71STqGJKZxcVhPaZ+LR+UCBZOyKfEyq+ZN311VpOJZ1IvTExf+S/5
 lqnciDtbO3I4Wq0ArLX1gs1q1XlXLaVaA3yVqeC8E7kOchDNinD3hJS4OX0e1gdsx/e6COvy
 qNg5aL5n0Kl4fcVqM0LdIhsubVs4eiNCa5XMSYpXmVi3HAuFyg9dN+x8thSwI836FoMASwOl
 C7tHsTjnSGufB+D7F7ZBT61BffNBBIm1KdMxcxqLUVXpBQHHlGkbwI+3Ye+nE6HmZH7IwLwV
 W+Ajl7oYF+jeKaH4DZFtgLYGLtZ1LDwKPjX7VAsa4Yx7S5+EBAaZGxK510MjIx6SGrZWBrrV
 TEvdV00F2MnQoeXKzD7O4WFbL55hhyGgfWTHwZ457iN9SgYi1JLPqWkZB0JRXIEtjd4JEQcx
 +8Umfre0Xt4713VxMygW0PnQt5aSQdMD58jHFxTk092mU+yIHj5LeYgvwSgZN4airXk5yRXl
 SE+xAvmumFBY
Organization: Red Hat GmbH
Message-ID: <6ffaaaaf-eaed-0409-61e8-7333f8a43f5a@redhat.com>
Date:   Thu, 5 Sep 2019 09:52:44 +0200
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101
 Thunderbird/60.8.0
MIME-Version: 1.0
In-Reply-To: <6823c187257b5033bca8905dda17b6c79a4944a6.camel@au1.ibm.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 04.09.19 07:25, Alastair D'Silva wrote:
> On Mon, 2019-09-02 at 09:28 +0200, David Hildenbrand wrote:
>> On 02.09.19 01:54, Alastair D'Silva wrote:
>>> On Tue, 2019-08-27 at 09:13 +0200, David Hildenbrand wrote:
>>>> On 27.08.19 08:39, Alastair D'Silva wrote:
>>>>> On Tue, 2019-08-27 at 08:28 +0200, Michal Hocko wrote:
>>>>>> On Tue 27-08-19 15:20:46, Alastair D'Silva wrote:
>>>>>>> From: Alastair D'Silva <alastair@d-silva.org>
>>>>>>>
>>>>>>> It is possible for firmware to allocate memory ranges
>>>>>>> outside
>>>>>>> the range of physical memory that we support
>>>>>>> (MAX_PHYSMEM_BITS).
>>>>>>
>>>>>> Doesn't that count as a FW bug? Do you have any evidence of
>>>>>> that
>>>>>> in
>>>>>> the
>>>>>> field? Just wondering...
>>>>>>
>>>>>
>>>>> Not outside our lab, but OpenCAPI attached LPC memory is
>>>>> assigned
>>>>> addresses based on the slot/NPU it is connected to. These
>>>>> addresses
>>>>> prior to:
>>>>> 4ffe713b7587 ("powerpc/mm: Increase the max addressable memory
>>>>> to
>>>>> 2PB")
>>>>> were inaccessible and resulted in bogus sections - see our
>>>>> discussion
>>>>> on 'mm: Trigger bug on if a section is not found in
>>>>> __section_nr'.
>>>>> Doing this check here was your suggestion :)
>>>>>
>>>>> It's entirely possible that a similar problem will occur in the
>>>>> future,
>>>>> and it's cheap to guard against, which is why I've added this.
>>>>>
>>>>
>>>> If you keep it here, I guess this should be wrapped by a
>>>> WARN_ON_ONCE().
>>>>
>>>> If we move it to common code (e.g., __add_pages() or
>>>> add_memory()),
>>>> then
>>>> probably not. I can see that s390x allows to configure
>>>> MAX_PHYSMEM_BITS,
>>>> so the check could actually make sense.
>>>>
>>>
>>> I couldn't see a nice platform indepedent way to determine the
>>> allowable address range, but if there is, then I'll move this to
>>> the
>>> generic code instead.
>>>
>>
>> At least on the !ZONE_DEVICE path we have
>>
>> __add_memory() -> register_memory_resource() ...
>>
>> return ERR_PTR(-E2BIG);
>>
>>
>> I was thinking about something like
>>
>> int add_pages()
>> {
>> 	if ((start + size - 1) >> MAX_PHYSMEM_BITS)
>> 		return -E2BIG;	
>>
>> 	return arch_add_memory(...)
>> }
>>
>> And switching users of arch_add_memory() to add_pages(). However, x86
>> already has an add_pages() function, so that would need some more
>> thought.
>>
>> Maybe simply renaming the existing add_pages() to arch_add_pages().
>>
>> add_pages(): Create virtual mapping
>> __add_pages(): Don't create virtual mapping
>>
>> arch_add_memory(): Arch backend for add_pages()
>> arch_add_pages(): Arch backend for __add_pages()
>>
>> It would be even more consistent if we would have arch_add_pages()
>> vs.
>> __arch_add_pages().
> 
> Looking a bit further, I think a good course of action would be to add
> the check to memory_hotplug.c:check_hotplug_memory_range().
> 
> This would be the least invasive, and could check both
> MAX_POSSIBLE_PHYSMEM_BITS and MAX_PHYSMEM_BITS.

You won't be able to catch the memremap path that way, just saying. But
at least it would be an easy change.

> 
> With that in mind, we can drop this patch.
> 


-- 

Thanks,

David / dhildenb