Date:   Fri, 25 Aug 2023 16:00:28 +1000
From:   Alistair Popple <apopple@...dia.com>
To:     "Huang, Ying" <ying.huang@...el.com>
Cc:     Andrew Morton <akpm@...ux-foundation.org>, linux-mm@...ck.org,
        linux-kernel@...r.kernel.org, linux-cxl@...r.kernel.org,
        nvdimm@...ts.linux.dev, linux-acpi@...r.kernel.org,
        "Aneesh Kumar K . V" <aneesh.kumar@...ux.ibm.com>,
        Wei Xu <weixugc@...gle.com>,
        Dan Williams <dan.j.williams@...el.com>,
        Dave Hansen <dave.hansen@...el.com>,
        Davidlohr Bueso <dave@...olabs.net>,
        Johannes Weiner <hannes@...xchg.org>,
        Jonathan Cameron <Jonathan.Cameron@...wei.com>,
        Michal Hocko <mhocko@...nel.org>,
        Yang Shi <shy828301@...il.com>,
        Rafael J Wysocki <rafael.j.wysocki@...el.com>
Subject: Re: [PATCH RESEND 4/4] dax, kmem: calculate abstract distance with
 general interface


"Huang, Ying" <ying.huang@...el.com> writes:

> Alistair Popple <apopple@...dia.com> writes:
>
>> "Huang, Ying" <ying.huang@...el.com> writes:
>>
>>> Alistair Popple <apopple@...dia.com> writes:
>>>
>>>> "Huang, Ying" <ying.huang@...el.com> writes:
>>>>
>>>>> Alistair Popple <apopple@...dia.com> writes:
>>>>>
>>>>>> Huang Ying <ying.huang@...el.com> writes:
>>>>>>
>>>>>>> Previously, a fixed abstract distance MEMTIER_DEFAULT_DAX_ADISTANCE is
>>>>>>> used for slow memory type in kmem driver.  This limits the usage of
>>>>>>> kmem driver, for example, it cannot be used for HBM (high bandwidth
>>>>>>> memory).
>>>>>>>
>>>>>>> So, we use the general abstract distance calculation mechanism in kmem
>>>>>>> drivers to get more accurate abstract distance on systems with proper
>>>>>>> support.  The original MEMTIER_DEFAULT_DAX_ADISTANCE is used as
>>>>>>> fallback only.
>>>>>>>
>>>>>>> Now, multiple memory types may be managed by kmem.  These memory types
>>>>>>> are put into the "kmem_memory_types" list and protected by
>>>>>>> kmem_memory_type_lock.
>>>>>>
>>>>>> See below but I wonder if kmem_memory_types could be a common helper
>>>>>> rather than kdax specific?
>>>>>>
>>>>>>> Signed-off-by: "Huang, Ying" <ying.huang@...el.com>
>>>>>>> Cc: Aneesh Kumar K.V <aneesh.kumar@...ux.ibm.com>
>>>>>>> Cc: Wei Xu <weixugc@...gle.com>
>>>>>>> Cc: Alistair Popple <apopple@...dia.com>
>>>>>>> Cc: Dan Williams <dan.j.williams@...el.com>
>>>>>>> Cc: Dave Hansen <dave.hansen@...el.com>
>>>>>>> Cc: Davidlohr Bueso <dave@...olabs.net>
>>>>>>> Cc: Johannes Weiner <hannes@...xchg.org>
>>>>>>> Cc: Jonathan Cameron <Jonathan.Cameron@...wei.com>
>>>>>>> Cc: Michal Hocko <mhocko@...nel.org>
>>>>>>> Cc: Yang Shi <shy828301@...il.com>
>>>>>>> Cc: Rafael J Wysocki <rafael.j.wysocki@...el.com>
>>>>>>> ---
>>>>>>>  drivers/dax/kmem.c           | 54 +++++++++++++++++++++++++++---------
>>>>>>>  include/linux/memory-tiers.h |  2 ++
>>>>>>>  mm/memory-tiers.c            |  2 +-
>>>>>>>  3 files changed, 44 insertions(+), 14 deletions(-)
>>>>>>>
>>>>>>> diff --git a/drivers/dax/kmem.c b/drivers/dax/kmem.c
>>>>>>> index 898ca9505754..837165037231 100644
>>>>>>> --- a/drivers/dax/kmem.c
>>>>>>> +++ b/drivers/dax/kmem.c
>>>>>>> @@ -49,14 +49,40 @@ struct dax_kmem_data {
>>>>>>>  	struct resource *res[];
>>>>>>>  };
>>>>>>>  
>>>>>>> -static struct memory_dev_type *dax_slowmem_type;
>>>>>>> +static DEFINE_MUTEX(kmem_memory_type_lock);
>>>>>>> +static LIST_HEAD(kmem_memory_types);
>>>>>>> +
>>>>>>> +static struct memory_dev_type *kmem_find_alloc_memorty_type(int adist)
>>>>>>> +{
>>>>>>> +	bool found = false;
>>>>>>> +	struct memory_dev_type *mtype;
>>>>>>> +
>>>>>>> +	mutex_lock(&kmem_memory_type_lock);
>>>>>>> +	list_for_each_entry(mtype, &kmem_memory_types, list) {
>>>>>>> +		if (mtype->adistance == adist) {
>>>>>>> +			found = true;
>>>>>>> +			break;
>>>>>>> +		}
>>>>>>> +	}
>>>>>>> +	if (!found) {
>>>>>>> +		mtype = alloc_memory_type(adist);
>>>>>>> +		if (!IS_ERR(mtype))
>>>>>>> +			list_add(&mtype->list, &kmem_memory_types);
>>>>>>> +	}
>>>>>>> +	mutex_unlock(&kmem_memory_type_lock);
>>>>>>> +
>>>>>>> +	return mtype;
>>>>>>> +}
>>>>>>> +
>>>>>>>  static int dev_dax_kmem_probe(struct dev_dax *dev_dax)
>>>>>>>  {
>>>>>>>  	struct device *dev = &dev_dax->dev;
>>>>>>>  	unsigned long total_len = 0;
>>>>>>>  	struct dax_kmem_data *data;
>>>>>>> +	struct memory_dev_type *mtype;
>>>>>>>  	int i, rc, mapped = 0;
>>>>>>>  	int numa_node;
>>>>>>> +	int adist = MEMTIER_DEFAULT_DAX_ADISTANCE;
>>>>>>>  
>>>>>>>  	/*
>>>>>>>  	 * Ensure good NUMA information for the persistent memory.
>>>>>>> @@ -71,6 +97,11 @@ static int dev_dax_kmem_probe(struct dev_dax *dev_dax)
>>>>>>>  		return -EINVAL;
>>>>>>>  	}
>>>>>>>  
>>>>>>> +	mt_calc_adistance(numa_node, &adist);
>>>>>>> +	mtype = kmem_find_alloc_memorty_type(adist);
>>>>>>> +	if (IS_ERR(mtype))
>>>>>>> +		return PTR_ERR(mtype);
>>>>>>> +
>>>>>>
>>>>>> I wrote my own quick and dirty module to test this and wrote basically
>>>>>> the same code sequence.
>>>>>>
>>>>>> I notice your using a list of memory types here though. I think it would
>>>>>> be nice to have a common helper that other users could call to do the
>>>>>> mt_calc_adistance() / kmem_find_alloc_memory_type() /
>>>>>> init_node_memory_type() sequence and cleanup as my naive approach would
>>>>>> result in a new memory_dev_type per device even though adist might be
>>>>>> the same. A common helper would make it easy to de-dup those.
>>>>>
>>>>> If it's useful, we can move kmem_find_alloc_memory_type() to
>>>>> memory-tier.c after some revision.  But I tend to move it after we have
>>>>> the second user.  What do you think about that?
>>>>
>>>> Usually I would agree, but this series already introduces a general
>>>> interface for calculating adist even though there's only one user and
>>>> implementation. So if we're going to add a general interface I think it
>>>> would be better to make it more usable now rather than after variations
>>>> of it have been cut and pasted into other drivers.
>>>
>>> In general, I would like to introduce complexity when necessary.  So, we
>>> can discuss the necessity of the general interface firstly.  We can do
>>> that in [1/4] of the series.
>>
>> Do we need one memory_dev_type per adistance or per adistance+device?
>>
>> If IUC correctly I think it's the former. Logically that means
>> memory_dev_types should be managed by the memory-tiering subsystem
>> because they are system wide rather than driver specific resources. That
>> we need to add the list field to struct memory_dev_type specifically for
>> use by dax/kmem supports that idea.
>
> In the original design (page 9/10/11 of [1]), memory_dev_type (Memory
> Type) is driver specific.

Oh fair enough. I was making these comments based on the incorrect
understanding that these were a global rather than driver specific
resource. Thanks for correcting that!

>> Also I'm not sure why you consider moving the
>> kmem_memory_types/kmem_find_alloc_memory_type()/etc. functions into
>> mm/memory-tiers.c to add complexity. Isn't it just moving code around or
>> am I missing some other subtlety that makes this hard? I really think
>> logically memory-tiering.c is where management of the various
>> memory_dev_types belongs.
>
> IMHO, it depends on whether these functions are shared by at least 2
> drivers.  If so, we can put them in mm/memory-tiers.c.  Otherwise, we
> should keep them in the driver.

Ok. Not sure I entirely agree because I suspect it would still make the
code clearer even for a single user. But generally you're correct and as
these memory_dev_type's are *supposed* to be driver specific (rather
than one per adistance) I don't think it's such a big issue.

Powered by blists - more mailing lists