12ceb3fb0SAlex ChiangWhat: /sys/devices/system/cpu/ 22ceb3fb0SAlex ChiangDate: pre-git history 32ceb3fb0SAlex ChiangContact: Linux kernel mailing list <linux-kernel@vger.kernel.org> 42ceb3fb0SAlex ChiangDescription: 52ceb3fb0SAlex Chiang A collection of both global and individual CPU attributes 62ceb3fb0SAlex Chiang 72ceb3fb0SAlex Chiang Individual CPU attributes are contained in subdirectories 82ceb3fb0SAlex Chiang named by the kernel's logical CPU number, e.g.: 92ceb3fb0SAlex Chiang 102ceb3fb0SAlex Chiang /sys/devices/system/cpu/cpu#/ 112ceb3fb0SAlex Chiang 12d93fc863SAlex ChiangWhat: /sys/devices/system/cpu/kernel_max 13d93fc863SAlex Chiang /sys/devices/system/cpu/offline 14d93fc863SAlex Chiang /sys/devices/system/cpu/online 15d93fc863SAlex Chiang /sys/devices/system/cpu/possible 16d93fc863SAlex Chiang /sys/devices/system/cpu/present 17d93fc863SAlex ChiangDate: December 2008 18d93fc863SAlex ChiangContact: Linux kernel mailing list <linux-kernel@vger.kernel.org> 19d93fc863SAlex ChiangDescription: CPU topology files that describe kernel limits related to 20d93fc863SAlex Chiang hotplug. Briefly: 21d93fc863SAlex Chiang 22d93fc863SAlex Chiang kernel_max: the maximum cpu index allowed by the kernel 23d93fc863SAlex Chiang configuration. 24d93fc863SAlex Chiang 25d93fc863SAlex Chiang offline: cpus that are not online because they have been 26d93fc863SAlex Chiang HOTPLUGGED off or exceed the limit of cpus allowed by the 27d93fc863SAlex Chiang kernel configuration (kernel_max above). 28d93fc863SAlex Chiang 29d93fc863SAlex Chiang online: cpus that are online and being scheduled. 30d93fc863SAlex Chiang 31d93fc863SAlex Chiang possible: cpus that have been allocated resources and can be 32d93fc863SAlex Chiang brought online if they are present. 33d93fc863SAlex Chiang 34d93fc863SAlex Chiang present: cpus that have been identified as being present in 35d93fc863SAlex Chiang the system. 36d93fc863SAlex Chiang 374f4cfa6cSMauro Carvalho Chehab See Documentation/admin-guide/cputopology.rst for more information. 38d93fc863SAlex Chiang 39d93fc863SAlex Chiang 4012633e80SNathan FontenotWhat: /sys/devices/system/cpu/probe 4112633e80SNathan Fontenot /sys/devices/system/cpu/release 4212633e80SNathan FontenotDate: November 2009 4312633e80SNathan FontenotContact: Linux kernel mailing list <linux-kernel@vger.kernel.org> 4412633e80SNathan FontenotDescription: Dynamic addition and removal of CPU's. This is not hotplug 4512633e80SNathan Fontenot removal, this is meant complete removal/addition of the CPU 4612633e80SNathan Fontenot from the system. 4712633e80SNathan Fontenot 4812633e80SNathan Fontenot probe: writes to this file will dynamically add a CPU to the 4912633e80SNathan Fontenot system. Information written to the file to add CPU's is 5012633e80SNathan Fontenot architecture specific. 5112633e80SNathan Fontenot 5212633e80SNathan Fontenot release: writes to this file dynamically remove a CPU from 5312633e80SNathan Fontenot the system. Information writtento the file to remove CPU's 5412633e80SNathan Fontenot is architecture specific. 55657348a0SAlex Chiang 56657348a0SAlex ChiangWhat: /sys/devices/system/cpu/cpu#/node 57657348a0SAlex ChiangDate: October 2009 58657348a0SAlex ChiangContact: Linux memory management mailing list <linux-mm@kvack.org> 59657348a0SAlex ChiangDescription: Discover NUMA node a CPU belongs to 60657348a0SAlex Chiang 61657348a0SAlex Chiang When CONFIG_NUMA is enabled, a symbolic link that points 62657348a0SAlex Chiang to the corresponding NUMA node directory. 63657348a0SAlex Chiang 64657348a0SAlex Chiang For example, the following symlink is created for cpu42 65657348a0SAlex Chiang in NUMA node 2: 66657348a0SAlex Chiang 67657348a0SAlex Chiang /sys/devices/system/cpu/cpu42/node2 -> ../../node/node2 68657348a0SAlex Chiang 69657348a0SAlex Chiang 70663fb2fcSAlex ChiangWhat: /sys/devices/system/cpu/cpu#/topology/core_id 71663fb2fcSAlex Chiang /sys/devices/system/cpu/cpu#/topology/core_siblings 72663fb2fcSAlex Chiang /sys/devices/system/cpu/cpu#/topology/core_siblings_list 73663fb2fcSAlex Chiang /sys/devices/system/cpu/cpu#/topology/physical_package_id 74663fb2fcSAlex Chiang /sys/devices/system/cpu/cpu#/topology/thread_siblings 75663fb2fcSAlex Chiang /sys/devices/system/cpu/cpu#/topology/thread_siblings_list 76663fb2fcSAlex ChiangDate: December 2008 77663fb2fcSAlex ChiangContact: Linux kernel mailing list <linux-kernel@vger.kernel.org> 78663fb2fcSAlex ChiangDescription: CPU topology files that describe a logical CPU's relationship 79663fb2fcSAlex Chiang to other cores and threads in the same physical package. 80663fb2fcSAlex Chiang 81663fb2fcSAlex Chiang One cpu# directory is created per logical CPU in the system, 82663fb2fcSAlex Chiang e.g. /sys/devices/system/cpu/cpu42/. 83663fb2fcSAlex Chiang 84663fb2fcSAlex Chiang Briefly, the files above are: 85663fb2fcSAlex Chiang 86663fb2fcSAlex Chiang core_id: the CPU core ID of cpu#. Typically it is the 87663fb2fcSAlex Chiang hardware platform's identifier (rather than the kernel's). 88663fb2fcSAlex Chiang The actual value is architecture and platform dependent. 89663fb2fcSAlex Chiang 90663fb2fcSAlex Chiang core_siblings: internal kernel map of cpu#'s hardware threads 91663fb2fcSAlex Chiang within the same physical_package_id. 92663fb2fcSAlex Chiang 93663fb2fcSAlex Chiang core_siblings_list: human-readable list of the logical CPU 94663fb2fcSAlex Chiang numbers within the same physical_package_id as cpu#. 95663fb2fcSAlex Chiang 96663fb2fcSAlex Chiang physical_package_id: physical package id of cpu#. Typically 97663fb2fcSAlex Chiang corresponds to a physical socket number, but the actual value 98663fb2fcSAlex Chiang is architecture and platform dependent. 99663fb2fcSAlex Chiang 100663fb2fcSAlex Chiang thread_siblings: internel kernel map of cpu#'s hardware 101663fb2fcSAlex Chiang threads within the same core as cpu# 102663fb2fcSAlex Chiang 103663fb2fcSAlex Chiang thread_siblings_list: human-readable list of cpu#'s hardware 104663fb2fcSAlex Chiang threads within the same core as cpu# 105663fb2fcSAlex Chiang 1064f4cfa6cSMauro Carvalho Chehab See Documentation/admin-guide/cputopology.rst for more information. 107663fb2fcSAlex Chiang 108663fb2fcSAlex Chiang 109c1fb5c47SAlex ChiangWhat: /sys/devices/system/cpu/cpuidle/current_driver 110c1fb5c47SAlex Chiang /sys/devices/system/cpu/cpuidle/current_governer_ro 111b6d8ef86SAishwarya Pant /sys/devices/system/cpu/cpuidle/available_governors 112b6d8ef86SAishwarya Pant /sys/devices/system/cpu/cpuidle/current_governor 113c1fb5c47SAlex ChiangDate: September 2007 114c1fb5c47SAlex ChiangContact: Linux kernel mailing list <linux-kernel@vger.kernel.org> 115c1fb5c47SAlex ChiangDescription: Discover cpuidle policy and mechanism 116c1fb5c47SAlex Chiang 117c1fb5c47SAlex Chiang Various CPUs today support multiple idle levels that are 118c1fb5c47SAlex Chiang differentiated by varying exit latencies and power 119c1fb5c47SAlex Chiang consumption during idle. 120c1fb5c47SAlex Chiang 121c1fb5c47SAlex Chiang Idle policy (governor) is differentiated from idle mechanism 122c1fb5c47SAlex Chiang (driver) 123c1fb5c47SAlex Chiang 124b6d8ef86SAishwarya Pant current_driver: (RO) displays current idle mechanism 125c1fb5c47SAlex Chiang 126b6d8ef86SAishwarya Pant current_governor_ro: (RO) displays current idle policy 127b6d8ef86SAishwarya Pant 128b6d8ef86SAishwarya Pant With the cpuidle_sysfs_switch boot option enabled (meant for 129b6d8ef86SAishwarya Pant developer testing), the following three attributes are visible 130b6d8ef86SAishwarya Pant instead: 131b6d8ef86SAishwarya Pant 132b6d8ef86SAishwarya Pant current_driver: same as described above 133b6d8ef86SAishwarya Pant 134b6d8ef86SAishwarya Pant available_governors: (RO) displays a space separated list of 135b6d8ef86SAishwarya Pant available governors 136b6d8ef86SAishwarya Pant 137b6d8ef86SAishwarya Pant current_governor: (RW) displays current idle policy. Users can 138b6d8ef86SAishwarya Pant switch the governor at runtime by writing to this file. 139c1fb5c47SAlex Chiang 140671c3095SMauro Carvalho Chehab See Documentation/admin-guide/pm/cpuidle.rst and 141671c3095SMauro Carvalho Chehab Documentation/driver-api/pm/cpuidle.rst for more information. 142c1fb5c47SAlex Chiang 143c1fb5c47SAlex Chiang 144b6d8ef86SAishwarya PantWhat: /sys/devices/system/cpu/cpuX/cpuidle/stateN/name 145b6d8ef86SAishwarya Pant /sys/devices/system/cpu/cpuX/cpuidle/stateN/latency 146b6d8ef86SAishwarya Pant /sys/devices/system/cpu/cpuX/cpuidle/stateN/power 147b6d8ef86SAishwarya Pant /sys/devices/system/cpu/cpuX/cpuidle/stateN/time 148b6d8ef86SAishwarya Pant /sys/devices/system/cpu/cpuX/cpuidle/stateN/usage 14904dab58aSRafael J. Wysocki /sys/devices/system/cpu/cpuX/cpuidle/stateN/above 15004dab58aSRafael J. Wysocki /sys/devices/system/cpu/cpuX/cpuidle/stateN/below 151b6d8ef86SAishwarya PantDate: September 2007 152b6d8ef86SAishwarya PantKernelVersion: v2.6.24 153b6d8ef86SAishwarya PantContact: Linux power management list <linux-pm@vger.kernel.org> 154b6d8ef86SAishwarya PantDescription: 155b6d8ef86SAishwarya Pant The directory /sys/devices/system/cpu/cpuX/cpuidle contains per 156b6d8ef86SAishwarya Pant logical CPU specific cpuidle information for each online cpu X. 157b6d8ef86SAishwarya Pant The processor idle states which are available for use have the 158b6d8ef86SAishwarya Pant following attributes: 159b6d8ef86SAishwarya Pant 160b6d8ef86SAishwarya Pant name: (RO) Name of the idle state (string). 161b6d8ef86SAishwarya Pant 162b6d8ef86SAishwarya Pant latency: (RO) The latency to exit out of this idle state (in 163b6d8ef86SAishwarya Pant microseconds). 164b6d8ef86SAishwarya Pant 165b6d8ef86SAishwarya Pant power: (RO) The power consumed while in this idle state (in 166b6d8ef86SAishwarya Pant milliwatts). 167b6d8ef86SAishwarya Pant 168b6d8ef86SAishwarya Pant time: (RO) The total time spent in this idle state (in microseconds). 169b6d8ef86SAishwarya Pant 170b6d8ef86SAishwarya Pant usage: (RO) Number of times this state was entered (a count). 171b6d8ef86SAishwarya Pant 17204dab58aSRafael J. Wysocki above: (RO) Number of times this state was entered, but the 17304dab58aSRafael J. Wysocki observed CPU idle duration was too short for it (a count). 17404dab58aSRafael J. Wysocki 17504dab58aSRafael J. Wysocki below: (RO) Number of times this state was entered, but the 17604dab58aSRafael J. Wysocki observed CPU idle duration was too long for it (a count). 177b6d8ef86SAishwarya Pant 178b6d8ef86SAishwarya PantWhat: /sys/devices/system/cpu/cpuX/cpuidle/stateN/desc 179b6d8ef86SAishwarya PantDate: February 2008 180b6d8ef86SAishwarya PantKernelVersion: v2.6.25 181b6d8ef86SAishwarya PantContact: Linux power management list <linux-pm@vger.kernel.org> 182b6d8ef86SAishwarya PantDescription: 183b6d8ef86SAishwarya Pant (RO) A small description about the idle state (string). 184b6d8ef86SAishwarya Pant 185b6d8ef86SAishwarya Pant 186b6d8ef86SAishwarya PantWhat: /sys/devices/system/cpu/cpuX/cpuidle/stateN/disable 187b6d8ef86SAishwarya PantDate: March 2012 188b6d8ef86SAishwarya PantKernelVersion: v3.10 189b6d8ef86SAishwarya PantContact: Linux power management list <linux-pm@vger.kernel.org> 190b6d8ef86SAishwarya PantDescription: 191b6d8ef86SAishwarya Pant (RW) Option to disable this idle state (bool). The behavior and 192b6d8ef86SAishwarya Pant the effect of the disable variable depends on the implementation 193b6d8ef86SAishwarya Pant of a particular governor. In the ladder governor, for example, 194b6d8ef86SAishwarya Pant it is not coherent, i.e. if one is disabling a light state, then 195b6d8ef86SAishwarya Pant all deeper states are disabled as well, but the disable variable 196b6d8ef86SAishwarya Pant does not reflect it. Likewise, if one enables a deep state but a 197b6d8ef86SAishwarya Pant lighter state still is disabled, then this has no effect. 198b6d8ef86SAishwarya Pant 199b6d8ef86SAishwarya Pant 200b6d8ef86SAishwarya PantWhat: /sys/devices/system/cpu/cpuX/cpuidle/stateN/residency 201b6d8ef86SAishwarya PantDate: March 2014 202b6d8ef86SAishwarya PantKernelVersion: v3.15 203b6d8ef86SAishwarya PantContact: Linux power management list <linux-pm@vger.kernel.org> 204b6d8ef86SAishwarya PantDescription: 205b6d8ef86SAishwarya Pant (RO) Display the target residency i.e. the minimum amount of 206b6d8ef86SAishwarya Pant time (in microseconds) this cpu should spend in this idle state 207b6d8ef86SAishwarya Pant to make the transition worth the effort. 208b6d8ef86SAishwarya Pant 20964bdff69SRafael J. WysockiWhat: /sys/devices/system/cpu/cpuX/cpuidle/stateN/s2idle/ 21064bdff69SRafael J. WysockiDate: March 2018 21164bdff69SRafael J. WysockiKernelVersion: v4.17 21264bdff69SRafael J. WysockiContact: Linux power management list <linux-pm@vger.kernel.org> 21364bdff69SRafael J. WysockiDescription: 21464bdff69SRafael J. Wysocki Idle state usage statistics related to suspend-to-idle. 21564bdff69SRafael J. Wysocki 21664bdff69SRafael J. Wysocki This attribute group is only present for states that can be 21764bdff69SRafael J. Wysocki used in suspend-to-idle with suspended timekeeping. 21864bdff69SRafael J. Wysocki 21964bdff69SRafael J. WysockiWhat: /sys/devices/system/cpu/cpuX/cpuidle/stateN/s2idle/time 22064bdff69SRafael J. WysockiDate: March 2018 22164bdff69SRafael J. WysockiKernelVersion: v4.17 22264bdff69SRafael J. WysockiContact: Linux power management list <linux-pm@vger.kernel.org> 22364bdff69SRafael J. WysockiDescription: 22464bdff69SRafael J. Wysocki Total time spent by the CPU in suspend-to-idle (with scheduler 22564bdff69SRafael J. Wysocki tick suspended) after requesting this state. 22664bdff69SRafael J. Wysocki 22764bdff69SRafael J. WysockiWhat: /sys/devices/system/cpu/cpuX/cpuidle/stateN/s2idle/usage 22864bdff69SRafael J. WysockiDate: March 2018 22964bdff69SRafael J. WysockiKernelVersion: v4.17 23064bdff69SRafael J. WysockiContact: Linux power management list <linux-pm@vger.kernel.org> 23164bdff69SRafael J. WysockiDescription: 23264bdff69SRafael J. Wysocki Total number of times this state has been requested by the CPU 23364bdff69SRafael J. Wysocki while entering suspend-to-idle. 234b6d8ef86SAishwarya Pant 2350cda8b91SAlex ChiangWhat: /sys/devices/system/cpu/cpu#/cpufreq/* 2360cda8b91SAlex ChiangDate: pre-git history 237dec102aaSViresh KumarContact: linux-pm@vger.kernel.org 2380cda8b91SAlex ChiangDescription: Discover and change clock speed of CPUs 2390cda8b91SAlex Chiang 2400cda8b91SAlex Chiang Clock scaling allows you to change the clock speed of the 2410cda8b91SAlex Chiang CPUs on the fly. This is a nice method to save battery 2420cda8b91SAlex Chiang power, because the lower the clock speed, the less power 2430cda8b91SAlex Chiang the CPU consumes. 2440cda8b91SAlex Chiang 2450cda8b91SAlex Chiang There are many knobs to tweak in this directory. 2460cda8b91SAlex Chiang 2470cda8b91SAlex Chiang See files in Documentation/cpu-freq/ for more information. 2480cda8b91SAlex Chiang 2490cda8b91SAlex Chiang 250f4fd3797SLan TianyuWhat: /sys/devices/system/cpu/cpu#/cpufreq/freqdomain_cpus 251f4fd3797SLan TianyuDate: June 2013 252dec102aaSViresh KumarContact: linux-pm@vger.kernel.org 253f4fd3797SLan TianyuDescription: Discover CPUs in the same CPU frequency coordination domain 254f4fd3797SLan Tianyu 255f4fd3797SLan Tianyu freqdomain_cpus is the list of CPUs (online+offline) that share 256f4fd3797SLan Tianyu the same clock/freq domain (possibly at the hardware level). 257f4fd3797SLan Tianyu That information may be hidden from the cpufreq core and the 258f4fd3797SLan Tianyu value of related_cpus may be different from freqdomain_cpus. This 259f4fd3797SLan Tianyu attribute is useful for user space DVFS controllers to get better 260f4fd3797SLan Tianyu power/performance results for platforms using acpi-cpufreq. 261f4fd3797SLan Tianyu 262f4fd3797SLan Tianyu This file is only present if the acpi-cpufreq driver is in use. 263f4fd3797SLan Tianyu 264f4fd3797SLan Tianyu 265eecaaba5SBorislav PetkovWhat: /sys/devices/system/cpu/cpu*/cache/index3/cache_disable_{0,1} 266468727abSAlex ChiangDate: August 2008 267468727abSAlex ChiangKernelVersion: 2.6.27 268ea8e080bSAravind GopalakrishnanContact: Linux kernel mailing list <linux-kernel@vger.kernel.org> 269eecaaba5SBorislav PetkovDescription: Disable L3 cache indices 270468727abSAlex Chiang 271eecaaba5SBorislav Petkov These files exist in every CPU's cache/index3 directory. Each 272eecaaba5SBorislav Petkov cache_disable_{0,1} file corresponds to one disable slot which 273eecaaba5SBorislav Petkov can be used to disable a cache index. Reading from these files 274eecaaba5SBorislav Petkov on a processor with this functionality will return the currently 275eecaaba5SBorislav Petkov disabled index for that node. There is one L3 structure per 276eecaaba5SBorislav Petkov node, or per internal node on MCM machines. Writing a valid 277eecaaba5SBorislav Petkov index to one of these files will cause the specificed cache 278eecaaba5SBorislav Petkov index to be disabled. 279eecaaba5SBorislav Petkov 280eecaaba5SBorislav Petkov All AMD processors with L3 caches provide this functionality. 281eecaaba5SBorislav Petkov For details, see BKDGs at 282eecaaba5SBorislav Petkov http://developer.amd.com/documentation/guides/Pages/default.aspx 283615b7300SAndre Przywara 284615b7300SAndre Przywara 285615b7300SAndre PrzywaraWhat: /sys/devices/system/cpu/cpufreq/boost 286615b7300SAndre PrzywaraDate: August 2012 287615b7300SAndre PrzywaraContact: Linux kernel mailing list <linux-kernel@vger.kernel.org> 288615b7300SAndre PrzywaraDescription: Processor frequency boosting control 289615b7300SAndre Przywara 290615b7300SAndre Przywara This switch controls the boost setting for the whole system. 291615b7300SAndre Przywara Boosting allows the CPU and the firmware to run at a frequency 292615b7300SAndre Przywara beyound it's nominal limit. 2933ba9b1b8STom Saeger More details can be found in 2943ba9b1b8STom Saeger Documentation/admin-guide/pm/cpufreq.rst 295c4fd675fSZhang Yanfei 296c4fd675fSZhang Yanfei 297c4fd675fSZhang YanfeiWhat: /sys/devices/system/cpu/cpu#/crash_notes 298c4fd675fSZhang Yanfei /sys/devices/system/cpu/cpu#/crash_notes_size 299c4fd675fSZhang YanfeiDate: April 2013 300c4fd675fSZhang YanfeiContact: kexec@lists.infradead.org 301c4fd675fSZhang YanfeiDescription: address and size of the percpu note. 302c4fd675fSZhang Yanfei 303c4fd675fSZhang Yanfei crash_notes: the physical address of the memory that holds the 304c4fd675fSZhang Yanfei note of cpu#. 305c4fd675fSZhang Yanfei 306c4fd675fSZhang Yanfei crash_notes_size: size of the note of cpu#. 307fbe299e0SRamkumar Ramachandra 308fbe299e0SRamkumar Ramachandra 309fbe299e0SRamkumar RamachandraWhat: /sys/devices/system/cpu/intel_pstate/max_perf_pct 310fbe299e0SRamkumar Ramachandra /sys/devices/system/cpu/intel_pstate/min_perf_pct 311fbe299e0SRamkumar Ramachandra /sys/devices/system/cpu/intel_pstate/no_turbo 312fbe299e0SRamkumar RamachandraDate: February 2013 313fbe299e0SRamkumar RamachandraContact: linux-pm@vger.kernel.org 314fbe299e0SRamkumar RamachandraDescription: Parameters for the Intel P-state driver 315fbe299e0SRamkumar Ramachandra 316fbe299e0SRamkumar Ramachandra Logic for selecting the current P-state in Intel 317fbe299e0SRamkumar Ramachandra Sandybridge+ processors. The three knobs control 318fbe299e0SRamkumar Ramachandra limits for the P-state that will be requested by the 319fbe299e0SRamkumar Ramachandra driver. 320fbe299e0SRamkumar Ramachandra 321fbe299e0SRamkumar Ramachandra max_perf_pct: limits the maximum P state that will be requested by 322fbe299e0SRamkumar Ramachandra the driver stated as a percentage of the available performance. 323fbe299e0SRamkumar Ramachandra 324fbe299e0SRamkumar Ramachandra min_perf_pct: limits the minimum P state that will be requested by 325fbe299e0SRamkumar Ramachandra the driver stated as a percentage of the available performance. 326fbe299e0SRamkumar Ramachandra 327fbe299e0SRamkumar Ramachandra no_turbo: limits the driver to selecting P states below the turbo 328fbe299e0SRamkumar Ramachandra frequency range. 329fbe299e0SRamkumar Ramachandra 3303ba9b1b8STom Saeger More details can be found in 3313ba9b1b8STom Saeger Documentation/admin-guide/pm/intel_pstate.rst 332246246cbSSudeep Holla 333246246cbSSudeep HollaWhat: /sys/devices/system/cpu/cpu*/cache/index*/<set_of_attributes_mentioned_below> 334246246cbSSudeep HollaDate: July 2014(documented, existed before August 2008) 335246246cbSSudeep HollaContact: Sudeep Holla <sudeep.holla@arm.com> 336246246cbSSudeep Holla Linux kernel mailing list <linux-kernel@vger.kernel.org> 337246246cbSSudeep HollaDescription: Parameters for the CPU cache attributes 338246246cbSSudeep Holla 339246246cbSSudeep Holla allocation_policy: 340246246cbSSudeep Holla - WriteAllocate: allocate a memory location to a cache line 341246246cbSSudeep Holla on a cache miss because of a write 342246246cbSSudeep Holla - ReadAllocate: allocate a memory location to a cache line 343246246cbSSudeep Holla on a cache miss because of a read 344246246cbSSudeep Holla - ReadWriteAllocate: both writeallocate and readallocate 345246246cbSSudeep Holla 346246246cbSSudeep Holla attributes: LEGACY used only on IA64 and is same as write_policy 347246246cbSSudeep Holla 348246246cbSSudeep Holla coherency_line_size: the minimum amount of data in bytes that gets 349246246cbSSudeep Holla transferred from memory to cache 350246246cbSSudeep Holla 3512539b258SWill Deacon level: the cache hierarchy in the multi-level cache configuration 352246246cbSSudeep Holla 353246246cbSSudeep Holla number_of_sets: total number of sets in the cache, a set is a 354246246cbSSudeep Holla collection of cache lines with the same cache index 355246246cbSSudeep Holla 356246246cbSSudeep Holla physical_line_partition: number of physical cache line per cache tag 357246246cbSSudeep Holla 358246246cbSSudeep Holla shared_cpu_list: the list of logical cpus sharing the cache 359246246cbSSudeep Holla 360246246cbSSudeep Holla shared_cpu_map: logical cpu mask containing the list of cpus sharing 361246246cbSSudeep Holla the cache 362246246cbSSudeep Holla 363246246cbSSudeep Holla size: the total cache size in kB 364246246cbSSudeep Holla 365246246cbSSudeep Holla type: 366246246cbSSudeep Holla - Instruction: cache that only holds instructions 367246246cbSSudeep Holla - Data: cache that only caches data 368246246cbSSudeep Holla - Unified: cache that holds both data and instructions 369246246cbSSudeep Holla 370246246cbSSudeep Holla ways_of_associativity: degree of freedom in placing a particular block 371246246cbSSudeep Holla of memory in the cache 372246246cbSSudeep Holla 373246246cbSSudeep Holla write_policy: 374246246cbSSudeep Holla - WriteThrough: data is written to both the cache line 375246246cbSSudeep Holla and to the block in the lower-level memory 376246246cbSSudeep Holla - WriteBack: data is written only to the cache line and 377246246cbSSudeep Holla the modified cache line is written to main 378246246cbSSudeep Holla memory only when it is replaced 3791b028984SShilpasri G Bhat 3801d78dc59STony Luck 3811d78dc59STony LuckWhat: /sys/devices/system/cpu/cpu*/cache/index*/id 3821d78dc59STony LuckDate: September 2016 3831d78dc59STony LuckContact: Linux kernel mailing list <linux-kernel@vger.kernel.org> 3841d78dc59STony LuckDescription: Cache id 3851d78dc59STony Luck 3861d78dc59STony Luck The id provides a unique number for a specific instance of 3871d78dc59STony Luck a cache of a particular type. E.g. there may be a level 3881d78dc59STony Luck 3 unified cache on each socket in a server and we may 3891d78dc59STony Luck assign them ids 0, 1, 2, ... 3901d78dc59STony Luck 3911d78dc59STony Luck Note that id value can be non-contiguous. E.g. level 1 3921d78dc59STony Luck caches typically exist per core, but there may not be a 3931d78dc59STony Luck power of two cores on a socket, so these caches may be 3941d78dc59STony Luck numbered 0, 1, 2, 3, 4, 5, 8, 9, 10, ... 3951d78dc59STony Luck 3961b028984SShilpasri G BhatWhat: /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats 3971b028984SShilpasri G Bhat /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/turbo_stat 3981b028984SShilpasri G Bhat /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/sub_turbo_stat 3991b028984SShilpasri G Bhat /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/unthrottle 4001b028984SShilpasri G Bhat /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/powercap 4011b028984SShilpasri G Bhat /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/overtemp 4021b028984SShilpasri G Bhat /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/supply_fault 4031b028984SShilpasri G Bhat /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/overcurrent 4041b028984SShilpasri G Bhat /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/occ_reset 4051b028984SShilpasri G BhatDate: March 2016 4061b028984SShilpasri G BhatContact: Linux kernel mailing list <linux-kernel@vger.kernel.org> 4071b028984SShilpasri G Bhat Linux for PowerPC mailing list <linuxppc-dev@ozlabs.org> 4081b028984SShilpasri G BhatDescription: POWERNV CPUFreq driver's frequency throttle stats directory and 4091b028984SShilpasri G Bhat attributes 4101b028984SShilpasri G Bhat 4111b028984SShilpasri G Bhat 'cpuX/cpufreq/throttle_stats' directory contains the CPU frequency 4121b028984SShilpasri G Bhat throttle stat attributes for the chip. The throttle stats of a cpu 4131b028984SShilpasri G Bhat is common across all the cpus belonging to a chip. Below are the 4141b028984SShilpasri G Bhat throttle attributes exported in the 'throttle_stats' directory: 4151b028984SShilpasri G Bhat 4161b028984SShilpasri G Bhat - turbo_stat : This file gives the total number of times the max 4171b028984SShilpasri G Bhat frequency is throttled to lower frequency in turbo (at and above 4181b028984SShilpasri G Bhat nominal frequency) range of frequencies. 4191b028984SShilpasri G Bhat 4201b028984SShilpasri G Bhat - sub_turbo_stat : This file gives the total number of times the 4211b028984SShilpasri G Bhat max frequency is throttled to lower frequency in sub-turbo(below 4221b028984SShilpasri G Bhat nominal frequency) range of frequencies. 4231b028984SShilpasri G Bhat 4241b028984SShilpasri G Bhat - unthrottle : This file gives the total number of times the max 4251b028984SShilpasri G Bhat frequency is unthrottled after being throttled. 4261b028984SShilpasri G Bhat 4271b028984SShilpasri G Bhat - powercap : This file gives the total number of times the max 4281b028984SShilpasri G Bhat frequency is throttled due to 'Power Capping'. 4291b028984SShilpasri G Bhat 4301b028984SShilpasri G Bhat - overtemp : This file gives the total number of times the max 4311b028984SShilpasri G Bhat frequency is throttled due to 'CPU Over Temperature'. 4321b028984SShilpasri G Bhat 4331b028984SShilpasri G Bhat - supply_fault : This file gives the total number of times the 4341b028984SShilpasri G Bhat max frequency is throttled due to 'Power Supply Failure'. 4351b028984SShilpasri G Bhat 4361b028984SShilpasri G Bhat - overcurrent : This file gives the total number of times the 4371b028984SShilpasri G Bhat max frequency is throttled due to 'Overcurrent'. 4381b028984SShilpasri G Bhat 4391b028984SShilpasri G Bhat - occ_reset : This file gives the total number of times the max 4401b028984SShilpasri G Bhat frequency is throttled due to 'OCC Reset'. 4411b028984SShilpasri G Bhat 4421b028984SShilpasri G Bhat The sysfs attributes representing different throttle reasons like 4431b028984SShilpasri G Bhat powercap, overtemp, supply_fault, overcurrent and occ_reset map to 4441b028984SShilpasri G Bhat the reasons provided by OCC firmware for throttling the frequency. 4451b028984SShilpasri G Bhat 4461b028984SShilpasri G BhatWhat: /sys/devices/system/cpu/cpufreq/policyX/throttle_stats 4471b028984SShilpasri G Bhat /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/turbo_stat 4481b028984SShilpasri G Bhat /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/sub_turbo_stat 4491b028984SShilpasri G Bhat /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/unthrottle 4501b028984SShilpasri G Bhat /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/powercap 4511b028984SShilpasri G Bhat /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/overtemp 4521b028984SShilpasri G Bhat /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/supply_fault 4531b028984SShilpasri G Bhat /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/overcurrent 4541b028984SShilpasri G Bhat /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/occ_reset 4551b028984SShilpasri G BhatDate: March 2016 4561b028984SShilpasri G BhatContact: Linux kernel mailing list <linux-kernel@vger.kernel.org> 4571b028984SShilpasri G Bhat Linux for PowerPC mailing list <linuxppc-dev@ozlabs.org> 4581b028984SShilpasri G BhatDescription: POWERNV CPUFreq driver's frequency throttle stats directory and 4591b028984SShilpasri G Bhat attributes 4601b028984SShilpasri G Bhat 4611b028984SShilpasri G Bhat 'policyX/throttle_stats' directory and all the attributes are same as 4621b028984SShilpasri G Bhat the /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats directory and 4631b028984SShilpasri G Bhat attributes which give the frequency throttle information of the chip. 464f8d9f924SSteve Capper 465f8d9f924SSteve CapperWhat: /sys/devices/system/cpu/cpuX/regs/ 466f8d9f924SSteve Capper /sys/devices/system/cpu/cpuX/regs/identification/ 467f8d9f924SSteve Capper /sys/devices/system/cpu/cpuX/regs/identification/midr_el1 468f8d9f924SSteve Capper /sys/devices/system/cpu/cpuX/regs/identification/revidr_el1 469f8d9f924SSteve CapperDate: June 2016 470f8d9f924SSteve CapperContact: Linux ARM Kernel Mailing list <linux-arm-kernel@lists.infradead.org> 471f8d9f924SSteve CapperDescription: AArch64 CPU registers 472f8d9f924SSteve Capper 'identification' directory exposes the CPU ID registers for 473f8d9f924SSteve Capper identifying model and revision of the CPU. 474a2b60670SJuri Lelli 475a2b60670SJuri LelliWhat: /sys/devices/system/cpu/cpu#/cpu_capacity 476a2b60670SJuri LelliDate: December 2016 477a2b60670SJuri LelliContact: Linux kernel mailing list <linux-kernel@vger.kernel.org> 478a2b60670SJuri LelliDescription: information about CPUs heterogeneity. 479a2b60670SJuri Lelli 480a2b60670SJuri Lelli cpu_capacity: capacity of cpu#. 48187590ce6SThomas Gleixner 48287590ce6SThomas GleixnerWhat: /sys/devices/system/cpu/vulnerabilities 48387590ce6SThomas Gleixner /sys/devices/system/cpu/vulnerabilities/meltdown 48487590ce6SThomas Gleixner /sys/devices/system/cpu/vulnerabilities/spectre_v1 48587590ce6SThomas Gleixner /sys/devices/system/cpu/vulnerabilities/spectre_v2 486c456442cSKonrad Rzeszutek Wilk /sys/devices/system/cpu/vulnerabilities/spec_store_bypass 487d90a7a0eSJiri Kosina /sys/devices/system/cpu/vulnerabilities/l1tf 4888a4b06d3SThomas Gleixner /sys/devices/system/cpu/vulnerabilities/mds 489a7a248c5SPawan Gupta /sys/devices/system/cpu/vulnerabilities/tsx_async_abort 490db4d30fbSVineela Tummalapalli /sys/devices/system/cpu/vulnerabilities/itlb_multihit 4919ecccfaaSDavid WoodhouseDate: January 2018 49287590ce6SThomas GleixnerContact: Linux kernel mailing list <linux-kernel@vger.kernel.org> 49387590ce6SThomas GleixnerDescription: Information about CPU vulnerabilities 49487590ce6SThomas Gleixner 49587590ce6SThomas Gleixner The files are named after the code names of CPU 49687590ce6SThomas Gleixner vulnerabilities. The output of those files reflects the 49787590ce6SThomas Gleixner state of the CPUs in the system. Possible output values: 49887590ce6SThomas Gleixner 49987590ce6SThomas Gleixner "Not affected" CPU is not affected by the vulnerability 50087590ce6SThomas Gleixner "Vulnerable" CPU is affected and no mitigation in effect 5019ecccfaaSDavid Woodhouse "Mitigation: $M" CPU is affected and mitigation $M is in effect 50205736e4aSThomas Gleixner 5035999bbe7SThomas Gleixner See also: Documentation/admin-guide/hw-vuln/index.rst 504d90a7a0eSJiri Kosina 50505736e4aSThomas GleixnerWhat: /sys/devices/system/cpu/smt 50605736e4aSThomas Gleixner /sys/devices/system/cpu/smt/active 50705736e4aSThomas Gleixner /sys/devices/system/cpu/smt/control 50805736e4aSThomas GleixnerDate: June 2018 50905736e4aSThomas GleixnerContact: Linux kernel mailing list <linux-kernel@vger.kernel.org> 51005736e4aSThomas GleixnerDescription: Control Symetric Multi Threading (SMT) 51105736e4aSThomas Gleixner 51205736e4aSThomas Gleixner active: Tells whether SMT is active (enabled and siblings online) 51305736e4aSThomas Gleixner 51405736e4aSThomas Gleixner control: Read/write interface to control SMT. Possible 51505736e4aSThomas Gleixner values: 51605736e4aSThomas Gleixner 51705736e4aSThomas Gleixner "on" SMT is enabled 51805736e4aSThomas Gleixner "off" SMT is disabled 51905736e4aSThomas Gleixner "forceoff" SMT is force disabled. Cannot be changed. 52005736e4aSThomas Gleixner "notsupported" SMT is not supported by the CPU 521de7b77e5SJosh Poimboeuf "notimplemented" SMT runtime toggling is not 522de7b77e5SJosh Poimboeuf implemented for the architecture 52305736e4aSThomas Gleixner 52405736e4aSThomas Gleixner If control status is "forceoff" or "notsupported" writes 52505736e4aSThomas Gleixner are rejected. 526b9c273baSRafael J. Wysocki 527b9c273baSRafael J. WysockiWhat: /sys/devices/system/cpu/cpu#/power/energy_perf_bias 528b9c273baSRafael J. WysockiDate: March 2019 529b9c273baSRafael J. WysockiContact: linux-pm@vger.kernel.org 530b9c273baSRafael J. WysockiDescription: Intel Energy and Performance Bias Hint (EPB) 531b9c273baSRafael J. Wysocki 532b9c273baSRafael J. Wysocki EPB for the given CPU in a sliding scale 0 - 15, where a value 533b9c273baSRafael J. Wysocki of 0 corresponds to a hint preference for highest performance 534b9c273baSRafael J. Wysocki and a value of 15 corresponds to the maximum energy savings. 535b9c273baSRafael J. Wysocki 536b9c273baSRafael J. Wysocki In order to change the EPB value for the CPU, write either 537b9c273baSRafael J. Wysocki a number in the 0 - 15 sliding scale above, or one of the 538b9c273baSRafael J. Wysocki strings: "performance", "balance-performance", "normal", 539b9c273baSRafael J. Wysocki "balance-power", "power" (that represent values reflected by 540b9c273baSRafael J. Wysocki their meaning), to this attribute. 541b9c273baSRafael J. Wysocki 542b9c273baSRafael J. Wysocki This attribute is present for all online CPUs supporting the 543b9c273baSRafael J. Wysocki Intel EPB feature. 544203dffacSFenghua Yu 545203dffacSFenghua YuWhat: /sys/devices/system/cpu/umwait_control 546203dffacSFenghua Yu /sys/devices/system/cpu/umwait_control/enable_c02 547203dffacSFenghua Yu /sys/devices/system/cpu/umwait_control/max_time 548203dffacSFenghua YuDate: May 2019 549203dffacSFenghua YuContact: Linux kernel mailing list <linux-kernel@vger.kernel.org> 550203dffacSFenghua YuDescription: Umwait control 551203dffacSFenghua Yu 552203dffacSFenghua Yu enable_c02: Read/write interface to control umwait C0.2 state 553203dffacSFenghua Yu Read returns C0.2 state status: 554203dffacSFenghua Yu 0: C0.2 is disabled 555203dffacSFenghua Yu 1: C0.2 is enabled 556203dffacSFenghua Yu 557203dffacSFenghua Yu Write 'y' or '1' or 'on' to enable C0.2 state. 558203dffacSFenghua Yu Write 'n' or '0' or 'off' to disable C0.2 state. 559203dffacSFenghua Yu 560203dffacSFenghua Yu The interface is case insensitive. 561203dffacSFenghua Yu 562203dffacSFenghua Yu max_time: Read/write interface to control umwait maximum time 563203dffacSFenghua Yu in TSC-quanta that the CPU can reside in either C0.1 564203dffacSFenghua Yu or C0.2 state. The time is an unsigned 32-bit number. 565203dffacSFenghua Yu Note that a value of zero means there is no limit. 566203dffacSFenghua Yu Low order two bits must be zero. 567734560acSRyan Grimm 568734560acSRyan GrimmWhat: /sys/devices/system/cpu/svm 569734560acSRyan GrimmDate: August 2019 570734560acSRyan GrimmContact: Linux kernel mailing list <linux-kernel@vger.kernel.org> 571734560acSRyan Grimm Linux for PowerPC mailing list <linuxppc-dev@ozlabs.org> 572734560acSRyan GrimmDescription: Secure Virtual Machine 573734560acSRyan Grimm 574734560acSRyan Grimm If 1, it means the system is using the Protected Execution 575734560acSRyan Grimm Facility in POWER9 and newer processors. i.e., it is a Secure 576734560acSRyan Grimm Virtual Machine. 577