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;
Subject: Re: [PATCH v2 4/8] powerpc/mm: protect linear mapping modifications
 by a mutex
To:     Oscar Salvador <osalvador@suse.de>
Cc:     linux-kernel@vger.kernel.org, linux-mm@kvack.org,
        linuxppc-dev@lists.ozlabs.org,
        Michael Ellerman <mpe@ellerman.id.au>,
        Benjamin Herrenschmidt <benh@kernel.crashing.org>,
        Paul Mackerras <paulus@samba.org>,
        Rashmica Gupta <rashmica.g@gmail.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Mike Rapoport <rppt@kernel.org>,
        Michal Hocko <mhocko@suse.com>,
        Wei Yang <richard.weiyang@linux.alibaba.com>
References: <20201111145322.15793-1-david@redhat.com>
 <20201111145322.15793-5-david@redhat.com> <20201117153750.GD15987@linux>
From:   David Hildenbrand <david@redhat.com>
Organization: Red Hat GmbH
Message-ID: <274264b2-9ec8-bde0-5725-184c8fd5f05e@redhat.com>
Date:   Tue, 17 Nov 2020 16:46:40 +0100
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.6.0
MIME-Version: 1.0
In-Reply-To: <20201117153750.GD15987@linux>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 17.11.20 16:37, Oscar Salvador wrote:
> On Wed, Nov 11, 2020 at 03:53:18PM +0100, David Hildenbrand wrote:
>> @@ -144,7 +147,9 @@ void __ref arch_remove_linear_mapping(u64 start, u64 size)
>>   	start = (unsigned long)__va(start);
>>   	flush_dcache_range_chunked(start, start + size, FLUSH_CHUNK_SIZE);
>>   
>> +	mutex_lock(&linear_mapping_mutex);
>>   	ret = remove_section_mapping(start, start + size);
>> +	mutex_unlock(&linear_mapping_mutex);
>>   	WARN_ON_ONCE(ret);
> 
> My expertise in this area is low, so bear with me.
> 
> Why we do not need to protect flush_dcache_range_chunked and
> vm_unmap_aliases?
> 

vm_unmap_aliases does own locking and can handle concurrent calls.


flush_dcache_range_chunked()->flush_dcache_range() ends up as a sequence 
of memory barriers paired with dcbf instructions.

dcbf: Copies modified cache blocks to main storage and invalidates the 
copy in the data cache.

It's called from various places and no global variables seem to be 
involved, so it looks like it doesn't need any kind of locking.

-- 
Thanks,

David / dhildenb