diff mbox series

[-V3,4/4] dax, kmem: calculate abstract distance with general interface

Message ID 20230912082101.342002-5-ying.huang@intel.com (mailing list archive)
State New
Headers show
Series memory tiering: calculate abstract distance based on ACPI HMAT | expand

Commit Message

Huang, Ying Sept. 12, 2023, 8:21 a.m. UTC 
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.

Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Tested-by: Bharata B Rao <bharata@amd.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Wei Xu <weixugc@google.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Yang Shi <shy828301@gmail.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: Rafael J Wysocki <rafael.j.wysocki@intel.com>
---
 drivers/dax/kmem.c           | 62 ++++++++++++++++++++++++++++--------
 include/linux/memory-tiers.h |  2 ++
 mm/memory-tiers.c            |  2 +-
 3 files changed, 52 insertions(+), 14 deletions(-)

Comments

Dave Jiang Sept. 14, 2023, 5:31 p.m. UTC | #1 
On 9/12/23 01:21, Huang Ying wrote:
> 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.
> 
> Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
> Tested-by: Bharata B Rao <bharata@amd.com>
> Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
> Cc: Wei Xu <weixugc@google.com>
> Cc: Alistair Popple <apopple@nvidia.com>
> Cc: Dan Williams <dan.j.williams@intel.com>
> Cc: Dave Hansen <dave.hansen@intel.com>
> Cc: Davidlohr Bueso <dave@stgolabs.net>
> Cc: Johannes Weiner <hannes@cmpxchg.org>
> Cc: Jonathan Cameron <Jonathan.Cameron@huawei.com>
> Cc: Michal Hocko <mhocko@kernel.org>
> Cc: Yang Shi <shy828301@gmail.com>
> Cc: Dave Jiang <dave.jiang@intel.com>
> Cc: Rafael J Wysocki <rafael.j.wysocki@intel.com>

Reviewed-by: Dave Jiang <dave.jiang@intel.com>

> ---
>  drivers/dax/kmem.c           | 62 ++++++++++++++++++++++++++++--------
>  include/linux/memory-tiers.h |  2 ++
>  mm/memory-tiers.c            |  2 +-
>  3 files changed, 52 insertions(+), 14 deletions(-)
> 
> diff --git a/drivers/dax/kmem.c b/drivers/dax/kmem.c
> index c57acb73e3db..369c698b7706 100644
> --- a/drivers/dax/kmem.c
> +++ b/drivers/dax/kmem.c
> @@ -49,14 +49,52 @@ 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_memory_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 void kmem_put_memory_types(void)
> +{
> +	struct memory_dev_type *mtype, *mtn;
> +
> +	mutex_lock(&kmem_memory_type_lock);
> +	list_for_each_entry_safe(mtype, mtn, &kmem_memory_types, list) {
> +		list_del(&mtype->list);
> +		put_memory_type(mtype);
> +	}
> +	mutex_unlock(&kmem_memory_type_lock);
> +}
> +
>  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 +109,11 @@ static int dev_dax_kmem_probe(struct dev_dax *dev_dax)
>  		return -EINVAL;
>  	}
>  
> +	mt_calc_adistance(numa_node, &adist);
> +	mtype = kmem_find_alloc_memory_type(adist);
> +	if (IS_ERR(mtype))
> +		return PTR_ERR(mtype);
> +
>  	for (i = 0; i < dev_dax->nr_range; i++) {
>  		struct range range;
>  
> @@ -88,7 +131,7 @@ static int dev_dax_kmem_probe(struct dev_dax *dev_dax)
>  		return -EINVAL;
>  	}
>  
> -	init_node_memory_type(numa_node, dax_slowmem_type);
> +	init_node_memory_type(numa_node, mtype);
>  
>  	rc = -ENOMEM;
>  	data = kzalloc(struct_size(data, res, dev_dax->nr_range), GFP_KERNEL);
> @@ -167,7 +210,7 @@ static int dev_dax_kmem_probe(struct dev_dax *dev_dax)
>  err_res_name:
>  	kfree(data);
>  err_dax_kmem_data:
> -	clear_node_memory_type(numa_node, dax_slowmem_type);
> +	clear_node_memory_type(numa_node, mtype);
>  	return rc;
>  }
>  
> @@ -219,7 +262,7 @@ static void dev_dax_kmem_remove(struct dev_dax *dev_dax)
>  		 * for that. This implies this reference will be around
>  		 * till next reboot.
>  		 */
> -		clear_node_memory_type(node, dax_slowmem_type);
> +		clear_node_memory_type(node, NULL);
>  	}
>  }
>  #else
> @@ -251,12 +294,6 @@ static int __init dax_kmem_init(void)
>  	if (!kmem_name)
>  		return -ENOMEM;
>  
> -	dax_slowmem_type = alloc_memory_type(MEMTIER_DEFAULT_DAX_ADISTANCE);
> -	if (IS_ERR(dax_slowmem_type)) {
> -		rc = PTR_ERR(dax_slowmem_type);
> -		goto err_dax_slowmem_type;
> -	}
> -
>  	rc = dax_driver_register(&device_dax_kmem_driver);
>  	if (rc)
>  		goto error_dax_driver;
> @@ -264,8 +301,7 @@ static int __init dax_kmem_init(void)
>  	return rc;
>  
>  error_dax_driver:
> -	put_memory_type(dax_slowmem_type);
> -err_dax_slowmem_type:
> +	kmem_put_memory_types();
>  	kfree_const(kmem_name);
>  	return rc;
>  }
> @@ -275,7 +311,7 @@ static void __exit dax_kmem_exit(void)
>  	dax_driver_unregister(&device_dax_kmem_driver);
>  	if (!any_hotremove_failed)
>  		kfree_const(kmem_name);
> -	put_memory_type(dax_slowmem_type);
> +	kmem_put_memory_types();
>  }
>  
>  MODULE_AUTHOR("Intel Corporation");
> diff --git a/include/linux/memory-tiers.h b/include/linux/memory-tiers.h
> index 9d27ca3b143e..ab6651402d7e 100644
> --- a/include/linux/memory-tiers.h
> +++ b/include/linux/memory-tiers.h
> @@ -24,6 +24,8 @@ struct memory_tier;
>  struct memory_dev_type {
>  	/* list of memory types that are part of same tier as this type */
>  	struct list_head tier_sibiling;
> +	/* list of memory types that are managed by one driver */
> +	struct list_head list;
>  	/* abstract distance for this specific memory type */
>  	int adistance;
>  	/* Nodes of same abstract distance */
> diff --git a/mm/memory-tiers.c b/mm/memory-tiers.c
> index fa1a8b418f9a..ca68ef17554b 100644
> --- a/mm/memory-tiers.c
> +++ b/mm/memory-tiers.c
> @@ -586,7 +586,7 @@ EXPORT_SYMBOL_GPL(init_node_memory_type);
>  void clear_node_memory_type(int node, struct memory_dev_type *memtype)
>  {
>  	mutex_lock(&memory_tier_lock);
> -	if (node_memory_types[node].memtype == memtype)
> +	if (node_memory_types[node].memtype == memtype || !memtype)
>  		node_memory_types[node].map_count--;
>  	/*
>  	 * If we umapped all the attached devices to this node,
Alistair Popple Sept. 19, 2023, 5:31 a.m. UTC | #2 
Huang Ying <ying.huang@intel.com> writes:

> diff --git a/mm/memory-tiers.c b/mm/memory-tiers.c
> index fa1a8b418f9a..ca68ef17554b 100644
> --- a/mm/memory-tiers.c
> +++ b/mm/memory-tiers.c
> @@ -586,7 +586,7 @@ EXPORT_SYMBOL_GPL(init_node_memory_type);
>  void clear_node_memory_type(int node, struct memory_dev_type *memtype)
>  {
>  	mutex_lock(&memory_tier_lock);
> -	if (node_memory_types[node].memtype == memtype)
> +	if (node_memory_types[node].memtype == memtype || !memtype)
>  		node_memory_types[node].map_count--;
>
>  	/*
>  	 * If we umapped all the attached devices to this node,

This implies it's possible memtype == NULL. Yet we have this:

	 * clear the node memory type.
	 */
	if (!node_memory_types[node].map_count) {
		node_memory_types[node].memtype = NULL;
		put_memory_type(memtype);
	}

It's not safe to call put_memory_type(NULL), so what condition guarantees
map_count > 1 when called with memtype == NULL? Thanks.

 - Alistair
Huang, Ying Sept. 19, 2023, 5:56 a.m. UTC | #3 
Alistair Popple <apopple@nvidia.com> writes:

> Huang Ying <ying.huang@intel.com> writes:
>
>> diff --git a/mm/memory-tiers.c b/mm/memory-tiers.c
>> index fa1a8b418f9a..ca68ef17554b 100644
>> --- a/mm/memory-tiers.c
>> +++ b/mm/memory-tiers.c
>> @@ -586,7 +586,7 @@ EXPORT_SYMBOL_GPL(init_node_memory_type);
>>  void clear_node_memory_type(int node, struct memory_dev_type *memtype)
>>  {
>>  	mutex_lock(&memory_tier_lock);
>> -	if (node_memory_types[node].memtype == memtype)
>> +	if (node_memory_types[node].memtype == memtype || !memtype)
>>  		node_memory_types[node].map_count--;
>>
>>  	/*
>>  	 * If we umapped all the attached devices to this node,
>
> This implies it's possible memtype == NULL. Yet we have this:
>
> 	 * clear the node memory type.
> 	 */
> 	if (!node_memory_types[node].map_count) {
> 		node_memory_types[node].memtype = NULL;
> 		put_memory_type(memtype);
> 	}
>
> It's not safe to call put_memory_type(NULL), so what condition guarantees
> map_count > 1 when called with memtype == NULL? Thanks.

Nothing guarantees that.  Thanks for pointing this out.  I will revise
the code to put_memory_type(node_memory_types[node].memtype) firstly.

--
Best Regards,
Huang, Ying
diff mbox series

Patch

diff --git a/drivers/dax/kmem.c b/drivers/dax/kmem.c
index c57acb73e3db..369c698b7706 100644
--- a/drivers/dax/kmem.c
+++ b/drivers/dax/kmem.c
@@ -49,14 +49,52 @@  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_memory_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 void kmem_put_memory_types(void)
+{
+	struct memory_dev_type *mtype, *mtn;
+
+	mutex_lock(&kmem_memory_type_lock);
+	list_for_each_entry_safe(mtype, mtn, &kmem_memory_types, list) {
+		list_del(&mtype->list);
+		put_memory_type(mtype);
+	}
+	mutex_unlock(&kmem_memory_type_lock);
+}
+
 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 +109,11 @@  static int dev_dax_kmem_probe(struct dev_dax *dev_dax)
 		return -EINVAL;
 	}
 
+	mt_calc_adistance(numa_node, &adist);
+	mtype = kmem_find_alloc_memory_type(adist);
+	if (IS_ERR(mtype))
+		return PTR_ERR(mtype);
+
 	for (i = 0; i < dev_dax->nr_range; i++) {
 		struct range range;
 
@@ -88,7 +131,7 @@  static int dev_dax_kmem_probe(struct dev_dax *dev_dax)
 		return -EINVAL;
 	}
 
-	init_node_memory_type(numa_node, dax_slowmem_type);
+	init_node_memory_type(numa_node, mtype);
 
 	rc = -ENOMEM;
 	data = kzalloc(struct_size(data, res, dev_dax->nr_range), GFP_KERNEL);
@@ -167,7 +210,7 @@  static int dev_dax_kmem_probe(struct dev_dax *dev_dax)
 err_res_name:
 	kfree(data);
 err_dax_kmem_data:
-	clear_node_memory_type(numa_node, dax_slowmem_type);
+	clear_node_memory_type(numa_node, mtype);
 	return rc;
 }
 
@@ -219,7 +262,7 @@  static void dev_dax_kmem_remove(struct dev_dax *dev_dax)
 		 * for that. This implies this reference will be around
 		 * till next reboot.
 		 */
-		clear_node_memory_type(node, dax_slowmem_type);
+		clear_node_memory_type(node, NULL);
 	}
 }
 #else
@@ -251,12 +294,6 @@  static int __init dax_kmem_init(void)
 	if (!kmem_name)
 		return -ENOMEM;
 
-	dax_slowmem_type = alloc_memory_type(MEMTIER_DEFAULT_DAX_ADISTANCE);
-	if (IS_ERR(dax_slowmem_type)) {
-		rc = PTR_ERR(dax_slowmem_type);
-		goto err_dax_slowmem_type;
-	}
-
 	rc = dax_driver_register(&device_dax_kmem_driver);
 	if (rc)
 		goto error_dax_driver;
@@ -264,8 +301,7 @@  static int __init dax_kmem_init(void)
 	return rc;
 
 error_dax_driver:
-	put_memory_type(dax_slowmem_type);
-err_dax_slowmem_type:
+	kmem_put_memory_types();
 	kfree_const(kmem_name);
 	return rc;
 }
@@ -275,7 +311,7 @@  static void __exit dax_kmem_exit(void)
 	dax_driver_unregister(&device_dax_kmem_driver);
 	if (!any_hotremove_failed)
 		kfree_const(kmem_name);
-	put_memory_type(dax_slowmem_type);
+	kmem_put_memory_types();
 }
 
 MODULE_AUTHOR("Intel Corporation");
diff --git a/include/linux/memory-tiers.h b/include/linux/memory-tiers.h
index 9d27ca3b143e..ab6651402d7e 100644
--- a/include/linux/memory-tiers.h
+++ b/include/linux/memory-tiers.h
@@ -24,6 +24,8 @@  struct memory_tier;
 struct memory_dev_type {
 	/* list of memory types that are part of same tier as this type */
 	struct list_head tier_sibiling;
+	/* list of memory types that are managed by one driver */
+	struct list_head list;
 	/* abstract distance for this specific memory type */
 	int adistance;
 	/* Nodes of same abstract distance */
diff --git a/mm/memory-tiers.c b/mm/memory-tiers.c
index fa1a8b418f9a..ca68ef17554b 100644
--- a/mm/memory-tiers.c
+++ b/mm/memory-tiers.c
@@ -586,7 +586,7 @@  EXPORT_SYMBOL_GPL(init_node_memory_type);
 void clear_node_memory_type(int node, struct memory_dev_type *memtype)
 {
 	mutex_lock(&memory_tier_lock);
-	if (node_memory_types[node].memtype == memtype)
+	if (node_memory_types[node].memtype == memtype || !memtype)
 		node_memory_types[node].map_count--;
 	/*
 	 * If we umapped all the attached devices to this node,