[v2,2/2] dax: add a sysfs knob to control memmap_on_memory behavior

Message ID 20231206-vv-dax_abi-v2-2-f4f4f2336d08@intel.com
State New
Headers
Series Add DAX ABI for memmap_on_memory |

Commit Message

Verma, Vishal L Dec. 7, 2023, 4:36 a.m. UTC
  Add a sysfs knob for dax devices to control the memmap_on_memory setting
if the dax device were to be hotplugged as system memory.

The default memmap_on_memory setting for dax devices originating via
pmem or hmem is set to 'false' - i.e. no memmap_on_memory semantics, to
preserve legacy behavior. For dax devices via CXL, the default is on.
The sysfs control allows the administrator to override the above
defaults if needed.

Cc: David Hildenbrand <david@redhat.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Huang Ying <ying.huang@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
---
 drivers/dax/bus.c                       | 40 +++++++++++++++++++++++++++++++++
 Documentation/ABI/testing/sysfs-bus-dax | 13 +++++++++++
 2 files changed, 53 insertions(+)
  

Comments

Zhijian Li (Fujitsu) Dec. 7, 2023, 8:29 a.m. UTC | #1
Hi Vishal,


On 07/12/2023 12:36, Vishal Verma wrote:
> +
> +What:		/sys/bus/dax/devices/daxX.Y/memmap_on_memory
> +Date:		October, 2023
> +KernelVersion:	v6.8
> +Contact:	nvdimm@lists.linux.dev
> +Description:
> +		(RW) Control the memmap_on_memory setting if the dax device
> +		were to be hotplugged as system memory. This determines whether
> +		the 'altmap' for the hotplugged memory will be placed on the
> +		device being hotplugged (memmap_on+memory=1) or if it will be

s/memmap_on+memory=1/memmap_on_memory=1


I have a question here

What relationship about memmap_on_memory and 'ndctl-create-namespace -M' option which
specifies where is the vmemmap backed memory.
I'm confused that memmap_on_memory=1 and '-M dev' are the same for nvdimm devdax mode ?

ndctl-create-namespace
...
        -M, --map=
            A pmem namespace in "fsdax" or "devdax" mode requires allocation of per-page metadata. The allocation
            can be drawn from either:

            •   "mem": typical system memory

            •   "dev": persistent memory reserved from the namespace :: Given relative capacities of "Persistent
                Memory" to "System RAM" the allocation defaults to reserving space out of the namespace directly
                ("--map=dev"). The overhead is 64-bytes per 4K (16GB per 1TB) on x86.


Thanks
Zhijian

> +		placed on regular memory (memmap_on_memory=0). This attribute
> +		must be set before the device is handed over to the 'kmem'
  
Verma, Vishal L Dec. 7, 2023, 7:25 p.m. UTC | #2
On Thu, 2023-12-07 at 08:29 +0000, Zhijian Li (Fujitsu) wrote:
> Hi Vishal,
> 
> 
> On 07/12/2023 12:36, Vishal Verma wrote:
> > +
> > +What:          /sys/bus/dax/devices/daxX.Y/memmap_on_memory
> > +Date:          October, 2023
> > +KernelVersion: v6.8
> > +Contact:       nvdimm@lists.linux.dev
> > +Description:
> > +               (RW) Control the memmap_on_memory setting if the dax device
> > +               were to be hotplugged as system memory. This determines whether
> > +               the 'altmap' for the hotplugged memory will be placed on the
> > +               device being hotplugged (memmap_on+memory=1) or if it will be
> 
> s/memmap_on+memory=1/memmap_on_memory=1

Thanks, will fix.
> 
> 
> I have a question here
> 
> What relationship about memmap_on_memory and 'ndctl-create-namespace
> -M' option which
> specifies where is the vmemmap backed memory.
> I'm confused that memmap_on_memory=1 and '-M dev' are the same for
> nvdimm devdax mode ?
> 
The main difference is that memory that comes from non-nvdimm sources,
such as hmem, or cxl, doesn't have a chance to set up the altmaps as
pmem can with '-M dev'. For these, memmap_on_memory does this as part
of the memory hotplug.
  
Huang, Ying Dec. 8, 2023, 3:13 a.m. UTC | #3
Vishal Verma <vishal.l.verma@intel.com> writes:

> Add a sysfs knob for dax devices to control the memmap_on_memory setting
> if the dax device were to be hotplugged as system memory.
>
> The default memmap_on_memory setting for dax devices originating via
> pmem or hmem is set to 'false' - i.e. no memmap_on_memory semantics, to
> preserve legacy behavior. For dax devices via CXL, the default is on.
> The sysfs control allows the administrator to override the above
> defaults if needed.
>
> Cc: David Hildenbrand <david@redhat.com>
> Cc: Dan Williams <dan.j.williams@intel.com>
> Cc: Dave Jiang <dave.jiang@intel.com>
> Cc: Dave Hansen <dave.hansen@linux.intel.com>
> Cc: Huang Ying <ying.huang@intel.com>
> Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
> Reviewed-by: David Hildenbrand <david@redhat.com>
> Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
> ---
>  drivers/dax/bus.c                       | 40 +++++++++++++++++++++++++++++++++
>  Documentation/ABI/testing/sysfs-bus-dax | 13 +++++++++++
>  2 files changed, 53 insertions(+)
>
> diff --git a/drivers/dax/bus.c b/drivers/dax/bus.c
> index 1ff1ab5fa105..11abb57cc031 100644
> --- a/drivers/dax/bus.c
> +++ b/drivers/dax/bus.c
> @@ -1270,6 +1270,45 @@ static ssize_t numa_node_show(struct device *dev,
>  }
>  static DEVICE_ATTR_RO(numa_node);
>  
> +static ssize_t memmap_on_memory_show(struct device *dev,
> +				     struct device_attribute *attr, char *buf)
> +{
> +	struct dev_dax *dev_dax = to_dev_dax(dev);
> +
> +	return sprintf(buf, "%d\n", dev_dax->memmap_on_memory);
> +}
> +
> +static ssize_t memmap_on_memory_store(struct device *dev,
> +				      struct device_attribute *attr,
> +				      const char *buf, size_t len)
> +{
> +	struct dev_dax *dev_dax = to_dev_dax(dev);
> +	struct dax_region *dax_region = dev_dax->region;
> +	ssize_t rc;
> +	bool val;
> +
> +	rc = kstrtobool(buf, &val);
> +	if (rc)
> +		return rc;
> +
> +	if (dev_dax->memmap_on_memory == val)
> +		return len;
> +
> +	device_lock(dax_region->dev);
> +	if (!dax_region->dev->driver) {

This still doesn't look right.  Can we check whether the current driver
is kmem?  And only allow change if it's not kmem?

--
Best Regards,
Huang, Ying

> +		device_unlock(dax_region->dev);
> +		return -ENXIO;
> +	}
> +
> +	device_lock(dev);
> +	dev_dax->memmap_on_memory = val;
> +	device_unlock(dev);
> +
> +	device_unlock(dax_region->dev);
> +	return rc == 0 ? len : rc;
> +}
> +static DEVICE_ATTR_RW(memmap_on_memory);
> +
>  static umode_t dev_dax_visible(struct kobject *kobj, struct attribute *a, int n)
>  {
>  	struct device *dev = container_of(kobj, struct device, kobj);
> @@ -1296,6 +1335,7 @@ static struct attribute *dev_dax_attributes[] = {
>  	&dev_attr_align.attr,
>  	&dev_attr_resource.attr,
>  	&dev_attr_numa_node.attr,
> +	&dev_attr_memmap_on_memory.attr,
>  	NULL,
>  };
>  
> diff --git a/Documentation/ABI/testing/sysfs-bus-dax b/Documentation/ABI/testing/sysfs-bus-dax
> index a61a7b186017..bb063a004e41 100644
> --- a/Documentation/ABI/testing/sysfs-bus-dax
> +++ b/Documentation/ABI/testing/sysfs-bus-dax
> @@ -149,3 +149,16 @@ KernelVersion:	v5.1
>  Contact:	nvdimm@lists.linux.dev
>  Description:
>  		(RO) The id attribute indicates the region id of a dax region.
> +
> +What:		/sys/bus/dax/devices/daxX.Y/memmap_on_memory
> +Date:		October, 2023
> +KernelVersion:	v6.8
> +Contact:	nvdimm@lists.linux.dev
> +Description:
> +		(RW) Control the memmap_on_memory setting if the dax device
> +		were to be hotplugged as system memory. This determines whether
> +		the 'altmap' for the hotplugged memory will be placed on the
> +		device being hotplugged (memmap_on+memory=1) or if it will be
> +		placed on regular memory (memmap_on_memory=0). This attribute
> +		must be set before the device is handed over to the 'kmem'
> +		driver (i.e.  hotplugged into system-ram).
  
Zhijian Li (Fujitsu) Dec. 8, 2023, 9:20 a.m. UTC | #4
On 08/12/2023 03:25, Verma, Vishal L wrote:
> On Thu, 2023-12-07 at 08:29 +0000, Zhijian Li (Fujitsu) wrote:
>> Hi Vishal,
>>
>>
>> On 07/12/2023 12:36, Vishal Verma wrote:
>>> +
>>> +What:          /sys/bus/dax/devices/daxX.Y/memmap_on_memory
>>> +Date:          October, 2023
>>> +KernelVersion: v6.8
>>> +Contact:       nvdimm@lists.linux.dev
>>> +Description:
>>> +               (RW) Control the memmap_on_memory setting if the dax device
>>> +               were to be hotplugged as system memory. This determines whether
>>> +               the 'altmap' for the hotplugged memory will be placed on the
>>> +               device being hotplugged (memmap_on+memory=1) or if it will be
>>
>> s/memmap_on+memory=1/memmap_on_memory=1
> 
> Thanks, will fix.
>>
>>
>> I have a question here
>>
>> What relationship about memmap_on_memory and 'ndctl-create-namespace
>> -M' option which
>> specifies where is the vmemmap backed memory.
>> I'm confused that memmap_on_memory=1 and '-M dev' are the same for
>> nvdimm devdax mode ?
>>
> The main difference is that memory that comes from non-nvdimm sources,
> such as hmem, or cxl, doesn't have a chance to set up the altmaps as
> pmem can with '-M dev'. For these, memmap_on_memory does this as part
> of the memory hotplug.

Thanks for your explanation.
(I wrongly thought nvdimm.kmem was also controlled by '-M dev' before)

feel free add:
Tested-by: Li Zhijian <lizhijian@fujitsu.com>
  
David Hildenbrand Dec. 8, 2023, 11:36 a.m. UTC | #5
On 07.12.23 05:36, Vishal Verma wrote:
> Add a sysfs knob for dax devices to control the memmap_on_memory setting
> if the dax device were to be hotplugged as system memory.
> 
> The default memmap_on_memory setting for dax devices originating via
> pmem or hmem is set to 'false' - i.e. no memmap_on_memory semantics, to
> preserve legacy behavior. For dax devices via CXL, the default is on.
> The sysfs control allows the administrator to override the above
> defaults if needed.
> 
> Cc: David Hildenbrand <david@redhat.com>
> Cc: Dan Williams <dan.j.williams@intel.com>
> Cc: Dave Jiang <dave.jiang@intel.com>
> Cc: Dave Hansen <dave.hansen@linux.intel.com>
> Cc: Huang Ying <ying.huang@intel.com>
> Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
> Reviewed-by: David Hildenbrand <david@redhat.com>
> Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
> ---
>   drivers/dax/bus.c                       | 40 +++++++++++++++++++++++++++++++++
>   Documentation/ABI/testing/sysfs-bus-dax | 13 +++++++++++
>   2 files changed, 53 insertions(+)
> 
> diff --git a/drivers/dax/bus.c b/drivers/dax/bus.c
> index 1ff1ab5fa105..11abb57cc031 100644
> --- a/drivers/dax/bus.c
> +++ b/drivers/dax/bus.c
> @@ -1270,6 +1270,45 @@ static ssize_t numa_node_show(struct device *dev,
>   }
>   static DEVICE_ATTR_RO(numa_node);
>   
> +static ssize_t memmap_on_memory_show(struct device *dev,
> +				     struct device_attribute *attr, char *buf)
> +{
> +	struct dev_dax *dev_dax = to_dev_dax(dev);
> +
> +	return sprintf(buf, "%d\n", dev_dax->memmap_on_memory);
> +}
> +
> +static ssize_t memmap_on_memory_store(struct device *dev,
> +				      struct device_attribute *attr,
> +				      const char *buf, size_t len)
> +{
> +	struct dev_dax *dev_dax = to_dev_dax(dev);
> +	struct dax_region *dax_region = dev_dax->region;
> +	ssize_t rc;
> +	bool val;
> +
> +	rc = kstrtobool(buf, &val);
> +	if (rc)
> +		return rc;
> +
> +	if (dev_dax->memmap_on_memory == val)
> +		return len;
> +
> +	device_lock(dax_region->dev);
> +	if (!dax_region->dev->driver) {
> +		device_unlock(dax_region->dev);
> +		return -ENXIO;
> +	}
> +
> +	device_lock(dev);
> +	dev_dax->memmap_on_memory = val;
> +	device_unlock(dev);
> +
> +	device_unlock(dax_region->dev);
> +	return rc == 0 ? len : rc;
> +}
> +static DEVICE_ATTR_RW(memmap_on_memory);
> +
>   static umode_t dev_dax_visible(struct kobject *kobj, struct attribute *a, int n)
>   {
>   	struct device *dev = container_of(kobj, struct device, kobj);
> @@ -1296,6 +1335,7 @@ static struct attribute *dev_dax_attributes[] = {
>   	&dev_attr_align.attr,
>   	&dev_attr_resource.attr,
>   	&dev_attr_numa_node.attr,
> +	&dev_attr_memmap_on_memory.attr,
>   	NULL,
>   };
>   
> diff --git a/Documentation/ABI/testing/sysfs-bus-dax b/Documentation/ABI/testing/sysfs-bus-dax
> index a61a7b186017..bb063a004e41 100644
> --- a/Documentation/ABI/testing/sysfs-bus-dax
> +++ b/Documentation/ABI/testing/sysfs-bus-dax
> @@ -149,3 +149,16 @@ KernelVersion:	v5.1
>   Contact:	nvdimm@lists.linux.dev
>   Description:
>   		(RO) The id attribute indicates the region id of a dax region.
> +
> +What:		/sys/bus/dax/devices/daxX.Y/memmap_on_memory
> +Date:		October, 2023
> +KernelVersion:	v6.8
> +Contact:	nvdimm@lists.linux.dev
> +Description:
> +		(RW) Control the memmap_on_memory setting if the dax device
> +		were to be hotplugged as system memory. This determines whether
> +		the 'altmap' for the hotplugged memory will be placed on the
> +		device being hotplugged (memmap_on+memory=1) or if it will be
> +		placed on regular memory (memmap_on_memory=0). This attribute
> +		must be set before the device is handed over to the 'kmem'
> +		driver (i.e.  hotplugged into system-ram).
> 

Should we note the dependency on other factors as given in 
mhp_supports_memmap_on_memory(), especially, the system-wide setting and 
some weird kernel configurations?
  
Verma, Vishal L Dec. 8, 2023, 9:24 p.m. UTC | #6
On Fri, 2023-12-08 at 09:20 +0000, Zhijian Li (Fujitsu) wrote:
> 
> 
> On 08/12/2023 03:25, Verma, Vishal L wrote:
> > On Thu, 2023-12-07 at 08:29 +0000, Zhijian Li (Fujitsu) wrote:
> > > Hi Vishal,
> > > 
> > > 
> > > On 07/12/2023 12:36, Vishal Verma wrote:
> > > > +
> > > > +What:          /sys/bus/dax/devices/daxX.Y/memmap_on_memory
> > > > +Date:          October, 2023
> > > > +KernelVersion: v6.8
> > > > +Contact:       nvdimm@lists.linux.dev
> > > > +Description:
> > > > +               (RW) Control the memmap_on_memory setting if the dax device
> > > > +               were to be hotplugged as system memory. This determines whether
> > > > +               the 'altmap' for the hotplugged memory will be placed on the
> > > > +               device being hotplugged (memmap_on+memory=1) or if it will be
> > > 
> > > s/memmap_on+memory=1/memmap_on_memory=1
> > 
> > Thanks, will fix.
> > > 
> > > 
> > > I have a question here
> > > 
> > > What relationship about memmap_on_memory and 'ndctl-create-namespace
> > > -M' option which
> > > specifies where is the vmemmap backed memory.
> > > I'm confused that memmap_on_memory=1 and '-M dev' are the same for
> > > nvdimm devdax mode ?
> > > 
> > The main difference is that memory that comes from non-nvdimm sources,
> > such as hmem, or cxl, doesn't have a chance to set up the altmaps as
> > pmem can with '-M dev'. For these, memmap_on_memory does this as part
> > of the memory hotplug.
> 
> Thanks for your explanation.
> (I wrongly thought nvdimm.kmem was also controlled by '-M dev' before)
> 
> feel free add:
> Tested-by: Li Zhijian <lizhijian@fujitsu.com>

Thank you Zhijian!
  
Verma, Vishal L Dec. 8, 2023, 9:25 p.m. UTC | #7
On Fri, 2023-12-08 at 11:13 +0800, Huang, Ying wrote:
> Vishal Verma <vishal.l.verma@intel.com> writes:
> 
[..]
> > 
> > +
> > +static ssize_t memmap_on_memory_store(struct device *dev,
> > +                                     struct device_attribute *attr,
> > +                                     const char *buf, size_t len)
> > +{
> > +       struct dev_dax *dev_dax = to_dev_dax(dev);
> > +       struct dax_region *dax_region = dev_dax->region;
> > +       ssize_t rc;
> > +       bool val;
> > +
> > +       rc = kstrtobool(buf, &val);
> > +       if (rc)
> > +               return rc;
> > +
> > +       if (dev_dax->memmap_on_memory == val)
> > +               return len;
> > +
> > +       device_lock(dax_region->dev);
> > +       if (!dax_region->dev->driver) {
> 
> This still doesn't look right.  Can we check whether the current driver
> is kmem?  And only allow change if it's not kmem?

Ah yes I lost track of this todo between revisions when I split this
out. Let me fix that for the next revision.
  
Verma, Vishal L Dec. 8, 2023, 9:26 p.m. UTC | #8
On Fri, 2023-12-08 at 12:36 +0100, David Hildenbrand wrote:
> On 07.12.23 05:36, Vishal Verma wrote:
[..]
> > 
> > +
> > +What:          /sys/bus/dax/devices/daxX.Y/memmap_on_memory
> > +Date:          October, 2023
> > +KernelVersion: v6.8
> > +Contact:       nvdimm@lists.linux.dev
> > +Description:
> > +               (RW) Control the memmap_on_memory setting if the dax device
> > +               were to be hotplugged as system memory. This determines whether
> > +               the 'altmap' for the hotplugged memory will be placed on the
> > +               device being hotplugged (memmap_on+memory=1) or if it will be
> > +               placed on regular memory (memmap_on_memory=0). This attribute
> > +               must be set before the device is handed over to the 'kmem'
> > +               driver (i.e.  hotplugged into system-ram).
> > 
> 
> Should we note the dependency on other factors as given in 
> mhp_supports_memmap_on_memory(), especially, the system-wide setting and 
> some weird kernel configurations?
> 
Yep good idea, I'll make a note of those for v3.
  

Patch

diff --git a/drivers/dax/bus.c b/drivers/dax/bus.c
index 1ff1ab5fa105..11abb57cc031 100644
--- a/drivers/dax/bus.c
+++ b/drivers/dax/bus.c
@@ -1270,6 +1270,45 @@  static ssize_t numa_node_show(struct device *dev,
 }
 static DEVICE_ATTR_RO(numa_node);
 
+static ssize_t memmap_on_memory_show(struct device *dev,
+				     struct device_attribute *attr, char *buf)
+{
+	struct dev_dax *dev_dax = to_dev_dax(dev);
+
+	return sprintf(buf, "%d\n", dev_dax->memmap_on_memory);
+}
+
+static ssize_t memmap_on_memory_store(struct device *dev,
+				      struct device_attribute *attr,
+				      const char *buf, size_t len)
+{
+	struct dev_dax *dev_dax = to_dev_dax(dev);
+	struct dax_region *dax_region = dev_dax->region;
+	ssize_t rc;
+	bool val;
+
+	rc = kstrtobool(buf, &val);
+	if (rc)
+		return rc;
+
+	if (dev_dax->memmap_on_memory == val)
+		return len;
+
+	device_lock(dax_region->dev);
+	if (!dax_region->dev->driver) {
+		device_unlock(dax_region->dev);
+		return -ENXIO;
+	}
+
+	device_lock(dev);
+	dev_dax->memmap_on_memory = val;
+	device_unlock(dev);
+
+	device_unlock(dax_region->dev);
+	return rc == 0 ? len : rc;
+}
+static DEVICE_ATTR_RW(memmap_on_memory);
+
 static umode_t dev_dax_visible(struct kobject *kobj, struct attribute *a, int n)
 {
 	struct device *dev = container_of(kobj, struct device, kobj);
@@ -1296,6 +1335,7 @@  static struct attribute *dev_dax_attributes[] = {
 	&dev_attr_align.attr,
 	&dev_attr_resource.attr,
 	&dev_attr_numa_node.attr,
+	&dev_attr_memmap_on_memory.attr,
 	NULL,
 };
 
diff --git a/Documentation/ABI/testing/sysfs-bus-dax b/Documentation/ABI/testing/sysfs-bus-dax
index a61a7b186017..bb063a004e41 100644
--- a/Documentation/ABI/testing/sysfs-bus-dax
+++ b/Documentation/ABI/testing/sysfs-bus-dax
@@ -149,3 +149,16 @@  KernelVersion:	v5.1
 Contact:	nvdimm@lists.linux.dev
 Description:
 		(RO) The id attribute indicates the region id of a dax region.
+
+What:		/sys/bus/dax/devices/daxX.Y/memmap_on_memory
+Date:		October, 2023
+KernelVersion:	v6.8
+Contact:	nvdimm@lists.linux.dev
+Description:
+		(RW) Control the memmap_on_memory setting if the dax device
+		were to be hotplugged as system memory. This determines whether
+		the 'altmap' for the hotplugged memory will be placed on the
+		device being hotplugged (memmap_on+memory=1) or if it will be
+		placed on regular memory (memmap_on_memory=0). This attribute
+		must be set before the device is handed over to the 'kmem'
+		driver (i.e.  hotplugged into system-ram).