Message-ID: <06e4428b-860e-1e66-defd-77666fcfa0c5@gmail.com>
Date:   Tue, 16 Oct 2018 17:06:48 -0400
From:   Pavel Tatashin <pasha.tatashin@...il.com>
To:     Alexander Duyck <alexander.h.duyck@...ux.intel.com>,
        linux-mm@...ck.org, akpm@...ux-foundation.org
Cc:     pavel.tatashin@...rosoft.com, mhocko@...e.com,
        dave.jiang@...el.com, linux-kernel@...r.kernel.org,
        willy@...radead.org, davem@...emloft.net,
        yi.z.zhang@...ux.intel.com, khalid.aziz@...cle.com,
        rppt@...ux.vnet.ibm.com, vbabka@...e.cz,
        sparclinux@...r.kernel.org, dan.j.williams@...el.com,
        ldufour@...ux.vnet.ibm.com, mgorman@...hsingularity.net,
        mingo@...nel.org, kirill.shutemov@...ux.intel.com
Subject: Re: [mm PATCH v3 2/6] mm: Drop meminit_pfn_in_nid as it is redundant



On 10/16/18 4:49 PM, Alexander Duyck wrote:
> On 10/16/2018 1:33 PM, Pavel Tatashin wrote:
>>
>>
>> On 10/15/18 4:27 PM, Alexander Duyck wrote:
>>> As best as I can tell the meminit_pfn_in_nid call is completely
>>> redundant.
>>> The deferred memory initialization is already making use of
>>> for_each_free_mem_range which in turn will call into __next_mem_range
>>> which
>>> will only return a memory range if it matches the node ID provided
>>> assuming
>>> it is not NUMA_NO_NODE.
>>>
>>> I am operating on the assumption that there are no zones or pgdata_t
>>> structures that have a NUMA node of NUMA_NO_NODE associated with
>>> them. If
>>> that is the case then __next_mem_range will never return a memory range
>>> that doesn't match the zone's node ID and as such the check is
>>> redundant.
>>>
>>> So one piece I would like to verfy on this is if this works for ia64.
>>> Technically it was using a different approach to get the node ID, but it
>>> seems to have the node ID also encoded into the memblock. So I am
>>> assuming this is okay, but would like to get confirmation on that.
>>>
>>> Signed-off-by: Alexander Duyck <alexander.h.duyck@...ux.intel.com>
>>
>> If I am not mistaken, this code is for systems with memory interleaving.
>> Quick looks shows that x86, powerpc, s390, and sparc have it set.
>>
>> I am not sure about other arches, but at least on SPARC, there are some
>> processors with memory interleaving feature:
>>
>> http://www.fujitsu.com/global/products/computing/servers/unix/sparc-enterprise/technology/performance/memory.html
>>
>>
>> Pavel
> 
> I get what it is for. However as best I can tell the check is actually
> redundant. In the case of the deferred page initialization we are
> already pulling the memory regions via "for_each_free_mem_range". That
> function is already passed a NUMA node ID. Because of that we are
> already checking the memory range to determine if it is in the node or
> not. As such it doesn't really make sense to go through for each PFN and
> then go back to the memory range and see if the node matches or not.
> 

Agree, it looks redundant, nice clean-up, I like it.

Reviewed-by: Pavel Tatashin <pavel.tatashin@...rosoft.com>

Thank you,
Pavel


> You can take a look at __next_mem_range which is called by
> for_each_free_mem_range and passed &memblock.memory and
> &memblock.reserved to avoid:
> https://elixir.bootlin.com/linux/latest/source/mm/memblock.c#L899
> 
> Then you can work your way through:
> meminit_pfn_in_nid(pfn, node, state)
>  __early_pfn_to_nid(pfn, state)
>   memblock_search_pfn_nid(pfn, &start_pfn, &end_pfn)
>    memblock_search(&memblock.memory, pfn)
> 
> From what I can tell the deferred init is going back through the
> memblock.memory list we pulled this range from and just validating it
> against itself. This makes sense for the standard init as that is just
> going from start_pfn->end_pfn, but for the deferred init we are pulling
> the memory ranges ahead of time so we shouldn't need to re-validate the
> memory that is contained within that range.

Powered by blists - more mailing lists