342
ACPI-Defined Health Functions (_NCH, _NBS)
The ACPI specification includes two vendor-agnostic methods for operating systems and
management software to call for determining the health of a persistent memory device.
Get NVDIMM Current Health Information (_NCH) can be called by the operating
systems at boot time to get the current health of the persistent memory device and
take appropriate action. The values reported by _NCH can change during runtime and
should be monitored for changes. _NCH contains health information that shows if
• The persistent memory requires maintenance
• The persistent memory device performance is degraded
• The operating system can assume write persistency loss on
subsequent power events
• The operating system can assume all data will be lost on subsequent
power events
Get NVDIMM Boot Status (_NBS) allows operating systems a vendor-agnostic method
to discover persistent memory health status that does not change during runtime. The
most significant attribute reported by _NBS is Data Loss Count (DLC). Data Loss Count is
expected to be used by applications and operating systems to help identify the rare case
where a persistent memory dirty shutdown has occurred. See “Unsafe/Dirty Shutdown”
later in this chapter for more information on how to properly use this attribute.
Vendor-Specific Device Health (_DSMs)
Many vendors may want to add further health attributes beyond what exists in _NBS
and _NCH. Vendors are free to design their own ACPI persistent memory device-specific
methods (_DSM) to be called by the operating system and privileged applications.
Although vendors implement persistent memory health discovery differently, a few
common health attributes are likely to exist to determine if a persistent memory device
requires service. These health attributes may include information such as an overall health
summary of the persistent memory, current persistent memory temperature, persistent
media error counts, and total device lifetime utilization. Many operating systems, such as
Linux, include support to retrieve and report the vendor-unique health statistics through
tools such as ndctl. The Intel persistent memory _DSM interface document can be found
under the “Related Specification” section of https://docs.pmem.io/.
Chapter 17 reliability, availability, and ServiCeability (raS)
ACPI-Defined Health Functions (_NCH, _NBS)
The ACPI specification includes two vendor-agnostic methods for operating systems and
management software to call for determining the health of a persistent memory device.
Get NVDIMM Current Health Information (_NCH) can be called by the operating
systems at boot time to get the current health of the persistent memory device and
take appropriate action. The values reported by _NCH can change during runtime and
should be monitored for changes. _NCH contains health information that shows if
• The persistent memory requires maintenance
• The persistent memory device performance is degraded
• The operating system can assume write persistency loss on
subsequent power events
• The operating system can assume all data will be lost on subsequent
power events
Get NVDIMM Boot Status (_NBS) allows operating systems a vendor-agnostic method
to discover persistent memory health status that does not change during runtime. The
most significant attribute reported by _NBS is Data Loss Count (DLC). Data Loss Count is
expected to be used by applications and operating systems to help identify the rare case
where a persistent memory dirty shutdown has occurred. See “Unsafe/Dirty Shutdown”
later in this chapter for more information on how to properly use this attribute.
Vendor-Specific Device Health (_DSMs)
Many vendors may want to add further health attributes beyond what exists in _NBS
and _NCH. Vendors are free to design their own ACPI persistent memory device-specific
methods (_DSM) to be called by the operating system and privileged applications.
Although vendors implement persistent memory health discovery differently, a few
common health attributes are likely to exist to determine if a persistent memory device
requires service. These health attributes may include information such as an overall health
summary of the persistent memory, current persistent memory temperature, persistent
media error counts, and total device lifetime utilization. Many operating systems, such as
Linux, include support to retrieve and report the vendor-unique health statistics through
tools such as ndctl. The Intel persistent memory _DSM interface document can be found
under the “Related Specification” section of https://docs.pmem.io/.
Chapter 17 reliability, availability, and ServiCeability (raS)