drivers/base/memory: clarify some memory block properties

All warnings (new ones prefixed by >>):

drivers/base/memory.c: In function 'init_memory_block':

Quoted 1 line

Quoted 1 line

drivers/base/memory.c:573:16: warning: variable 'start_pfn' set but not used [-Wunused-but-set-variable]

      573 |  unsigned long start_pfn;
          |                ^~~~~~~~~

No need to store the value for each and every memory block, as we can easily query the value at runtime. Reshuffle the members to optimize the memory layout. Also, let's clarify what the interface once was used for and why it's legacy nowadays.

"phys_device" was used on s390x in older versions of lsmem[2]/chmem[3], back when they were still part of s390x-tools. They were later replaced by the variants in linux-utils. For example, RHEL6 and RHEL7 contain lsmem/chmem from s390-utils. RHEL8 switched to versions from util-linux on s390x [4].

"phys_device" was added with sysfs support for memory hotplug in commit 3947be1969a9 ("[PATCH] memory hotplug: sysfs and add/remove functions") in 2005. It always returned 0.

s390x started returning something != 0 on some setups (if sclp.rzm is set by HW) in 2010 via commit 57b552ba0b2f ("memory hotplug/s390: set phys_device").

For s390x, it allowed for identifying which memory block devices belong to the same storage increment (RZM). Only if all memory block devices comprising a single storage increment were offline, the memory could actually be removed in the hypervisor.

Since commit e5d709bb5fb7 ("s390/memory hotplug: provide memory_block_size_bytes() function") in 2013 a memory block devices spans at least one storage increment - which is why the interface isn't really helpful/used anymore (except by old lsmem/chmem tools).

There were once RFC patches to make use of "phys_device" in ACPI context; however, the underlying problem could be solved using different interfaces [1].

[1] https://patchwork.kernel.org/patch/2163871/
[2] https://github.com/ibm-s390-tools/s390-tools/blob/v2.1.0/zconf/lsmem
[3] https://github.com/ibm-s390-tools/s390-tools/blob/v2.1.0/zconf/chmem
[4] https://bugzilla.redhat.com/show_bug.cgi?id=1504134

Cc Andrew Morton <[email protected]>

Cc Dave Hansen <[email protected]>

Cc Michal Hocko <[email protected]>

Cc Oscar Salvador <[email protected]>

Cc Greg Kroah-Hartman <[email protected]>

Cc Gerald Schaefer <[email protected]>

Cc Jonathan Corbet <[email protected]>

Cc "Rafael J. Wysocki" <[email protected]>

Cc Mauro Carvalho Chehab <[email protected]>

Cc Ilya Dryomov <[email protected]>

Cc Vaibhav Jain <[email protected]>

Cc Tom Rix <[email protected]>

Cc Geert Uytterhoeven <[email protected]>

Cc [email protected]

Signed-off-by David Hildenbrand <[email protected]>

---
 .../ABI/testing/sysfs-devices-memory          |  5 ++--
 .../admin-guide/mm/memory-hotplug.rst         |  4 ++--
 drivers/base/memory.c                         | 23 ++++++++-----------
 include/linux/memory.h                        |  3 +--
 4 files changed, 15 insertions(+), 20 deletions(-)

Documentation/ABI/testing/sysfs-devices-memory +3 -2

index 246a45b96d22..58dbc592bc57 100644
@@ -26,8 +26,9 @@ Date:		September 2008
 Contact:	Badari Pulavarty <[email protected]>
 Description:
 		The file /sys/devices/system/memory/memoryX/phys_device
-		is read-only and is designed to show the name of physical
-		memory device.  Implementation is currently incomplete.
+		is read-only;  it is a legacy interface only ever used on s390x
+		to expose the covered storage increment.
+Users:		Legacy s390-tools lsmem/chmem
 
 What:		/sys/devices/system/memory/memoryX/phys_index
 Date:		September 2008

Documentation/admin-guide/mm/memory-hotplug.rst +2 -2

index 5c4432c96c4b..245739f55ac7 100644
@@ -160,8 +160,8 @@ Under each memory block, you can see 5 files:
 
                     "online_movable", "online", "offline" command
                     which will be performed on all sections in the block.
-``phys_device``     read-only: designed to show the name of physical memory
-                    device.  This is not well implemented now.
+``phys_device``	    read-only: legacy interface only ever used on s390x to
+		    expose the covered storage increment.
 ``removable``       read-only: contains an integer value indicating
                     whether the memory block is removable or not
                     removable.  A value of 1 indicates that the memory

drivers/base/memory.c +9 -14

index 901e379676be..16959d339172 100644
@@ -290,20 +290,20 @@ static ssize_t state_store(struct device *dev, struct device_attribute *attr,
 }
 
 /*
- * phys_device is a bad name for this.  What I really want
- * is a way to differentiate between memory ranges that
- * are part of physical devices that constitute
- * a complete removable unit or fru.
- * i.e. do these ranges belong to the same physical device,
- * s.t. if I offline all of these sections I can then
- * remove the physical device?
+ * Legacy interface that we cannot remove: s390x exposes the storage increment
+ * covered by a memory block, allowing for identifying which memory blocks
+ * comprise a storage increment. Since a memory block spans complete
+ * storage increments nowadays, this interface is basically unused. Other
+ * archs never exposed != 0.
  */
 static ssize_t phys_device_show(struct device *dev,
 				struct device_attribute *attr, char *buf)
 {
 	struct memory_block *mem = to_memory_block(dev);
+	unsigned long start_pfn = section_nr_to_pfn(mem->start_section_nr);
 
-	return sysfs_emit(buf, "%d\n", mem->phys_device);
+	return sysfs_emit(buf, "%d\n",
+			  arch_get_memory_phys_device(start_pfn));
 }
 
 #ifdef CONFIG_MEMORY_HOTREMOVE

@@ -488,11 +488,7 @@ static DEVICE_ATTR_WO(soft_offline_page);
 static DEVICE_ATTR_WO(hard_offline_page);
 #endif
 
-/*
- * Note that phys_device is optional.  It is here to allow for
- * differentiation between which *physical* devices each
- * section belongs to...
- */
+/* See phys_device_show(). */
 int __weak arch_get_memory_phys_device(unsigned long start_pfn)
 {
 	return 0;

@@ -589,7 +585,6 @@ static int init_memory_block(unsigned long block_id, unsigned long state)
 	mem->start_section_nr = block_id * sections_per_block;
 	mem->state = state;
 	start_pfn = section_nr_to_pfn(mem->start_section_nr);
-	mem->phys_device = arch_get_memory_phys_device(start_pfn);
 	mem->nid = NUMA_NO_NODE;
 
 	ret = register_memory(mem);

include/linux/memory.h +1 -2

index 439a89e758d8..4da95e684e20 100644
@@ -27,9 +27,8 @@ struct memory_block {
 	unsigned long start_section_nr;
 	unsigned long state;		/* serialized by the dev->lock */
 	int online_type;		/* for passing data to online routine */
-	int phys_device;		/* to which fru does this belong? */
-	struct device dev;
 	int nid;			/* NID for this memory block */
+	struct device dev;
 };
 
 int arch_get_memory_phys_device(unsigned long start_pfn);

Signature · 1 line

2.29.2

In commit 53cdc1cb29e8 ("drivers/base/memory.c: indicate all memory blocks as removable") we changed the output of the "removable" property of memory devices to return "1" if and only if the kernel supports memory offlining.

Let's update documentation, stating that the interface is legacy. Also update documentation of the "state" property and "valid_zones" properties.

Cc Andrew Morton <[email protected]>

Cc Dave Hansen <[email protected]>

Cc Michal Hocko <[email protected]>

Cc Oscar Salvador <[email protected]>

Cc Jonathan Corbet <[email protected]>

Cc David Hildenbrand <[email protected]>

Cc Greg Kroah-Hartman <[email protected]>

Cc Jonathan Cameron <[email protected]>

Cc Ilya Dryomov <[email protected]>

Cc Mauro Carvalho Chehab <[email protected]>

Cc Geert Uytterhoeven <[email protected]>

Cc [email protected]

Signed-off-by David Hildenbrand <[email protected]>

---
 .../ABI/testing/sysfs-devices-memory          | 53 ++++++++++++-------
 .../admin-guide/mm/memory-hotplug.rst         | 16 +++---
 2 files changed, 41 insertions(+), 28 deletions(-)

Documentation/ABI/testing/sysfs-devices-memory +33 -20

index 58dbc592bc57..d8b0f80b9e33 100644
@@ -13,13 +13,13 @@ What:		/sys/devices/system/memory/memoryX/removable
 Date:		June 2008
 Contact:	Badari Pulavarty <[email protected]>
 Description:
-		The file /sys/devices/system/memory/memoryX/removable
-		indicates whether this memory block is removable or not.
-		This is useful for a user-level agent to determine
-		identify removable sections of the memory before attempting
-		potentially expensive hot-remove memory operation
+		The file /sys/devices/system/memory/memoryX/removable is a
+		legacy interface used to indicated whether a memory block is
+		likely to be offlineable or not.  Newer kernel versions return
+		"1" if and only if the kernel supports memory offlining.
 Users:		hotplug memory remove tools
 		http://www.ibm.com/developerworks/wikis/display/LinuxP/powerpc-utils
+		lsmem/chmem part of util-linux
 
 What:		/sys/devices/system/memory/memoryX/phys_device
 Date:		September 2008

@@ -44,23 +44,25 @@ Date:		September 2008
 Contact:	Badari Pulavarty <[email protected]>
 Description:
 		The file /sys/devices/system/memory/memoryX/state
-		is read-write.  When read, its contents show the
-		online/offline state of the memory section.  When written,
-		root can toggle the the online/offline state of a removable
-		memory section (see removable file description above)
-		using the following commands::
+		is read-write.  When read, it returns the online/offline
+		state of the memory block.  When written, root can toggle
+		the online/offline state of a memory block using the following
+		commands::
 
 		  # echo online > /sys/devices/system/memory/memoryX/state
 		  # echo offline > /sys/devices/system/memory/memoryX/state
 
-		For example, if /sys/devices/system/memory/memory22/removable
-		contains a value of 1 and
-		/sys/devices/system/memory/memory22/state contains the
-		string "online" the following command can be executed by
-		by root to offline that section::
-
-		  # echo offline > /sys/devices/system/memory/memory22/state
-
+		On newer kernel versions, advanced states can be specified
+		when onlining to select a target zone: "online_movable"
+		selects the movable zone.  "online_kernel" selects the
+		applicable kernel zone (DMA, DMA32, or Normal).  However,
+		after successfully setting one of the advanced states,
+		reading the file will return "online"; the zone information
+		can be obtained via "valid_zones" instead.
+
+		While onlining is unlikely to fail, there are no guarantees
+		that offlining will succeed.  Offlining is more likely to
+		succeed if "valid_zones" indicates "Movable".
 Users:		hotplug memory remove tools
 		http://www.ibm.com/developerworks/wikis/display/LinuxP/powerpc-utils
 

@@ -70,8 +72,19 @@ Date:           July 2014
 Contact:	Zhang Zhen <[email protected]>
 Description:
 		The file /sys/devices/system/memory/memoryX/valid_zones	is
-		read-only and is designed to show which zone this memory
-		block can be onlined to.
+		read-only.
+
+		For online memory blocks, it returns in which zone memory
+		provided by a memory block is managed.  If multiple zones
+		apply (not applicable for hotplugged memory), "None" is returned
+		and the memory block cannot be offlined.
+
+		For offline memory blocks, it returns by which zone memory
+		provided by a memory block can be managed when onlining.
+		The first returned zone ("default") will be used when setting
+		the state of an offline memory block to "online".  Only one of
+		the kernel zones (DMA, DMA32, Normal) is applicable for a single
+		memory block.
 
 What:		/sys/devices/system/memoryX/nodeY
 Date:		October 2009

Documentation/admin-guide/mm/memory-hotplug.rst +8 -8

index 245739f55ac7..5307f90738aa 100644
@@ -162,14 +162,14 @@ Under each memory block, you can see 5 files:
                     which will be performed on all sections in the block.
 ``phys_device``	    read-only: legacy interface only ever used on s390x to
 		    expose the covered storage increment.
-``removable``       read-only: contains an integer value indicating
-                    whether the memory block is removable or not
-                    removable.  A value of 1 indicates that the memory
-                    block is removable and a value of 0 indicates that
-                    it is not removable. A memory block is removable only if
-                    every section in the block is removable.
-``valid_zones``     read-only: designed to show which zones this memory block
-		    can be onlined to.
+``removable``	    read-only: legacy interface that indicated whether a memory
+		    block was likely to be offlineable or not.  Newer kernel
+		    versions return "1" if and only if the kernel supports
+		    memory offlining.
+``valid_zones``     read-only: designed to show by which zone memory provided by
+		    a memory block is managed, and to show by which zone memory
+		    provided by an offline memory block could be managed when
+		    onlining.
 
 		    The first column shows it`s default zone.
 

Signature · 1 line

2.29.2