ABI/testing/sysfs-devices-system-cpu

   1 What:           /sys/devices/system/cpu/
   2 Date:           pre-git history
   3 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
   4 Description:
   5                 A collection of both global and individual CPU attributes
   6
   7                 Individual CPU attributes are contained in subdirectories
   8                 named by the kernel's logical CPU number, e.g.:
   9
  10                 /sys/devices/system/cpu/cpuX/
  11
  12 What:           /sys/devices/system/cpu/kernel_max
  13                 /sys/devices/system/cpu/offline
  14                 /sys/devices/system/cpu/online
  15                 /sys/devices/system/cpu/possible
  16                 /sys/devices/system/cpu/present
  17 Date:           December 2008
  18 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
  19 Description:    CPU topology files that describe kernel limits related to
  20                 hotplug. Briefly:
  21
  22                 kernel_max: the maximum cpu index allowed by the kernel
  23                 configuration.
  24
  25                 offline: cpus that are not online because they have been
  26                 HOTPLUGGED off or exceed the limit of cpus allowed by the
  27                 kernel configuration (kernel_max above).
  28
  29                 online: cpus that are online and being scheduled.
  30
  31                 possible: cpus that have been allocated resources and can be
  32                 brought online if they are present.
  33
  34                 present: cpus that have been identified as being present in
  35                 the system.
  36
  37                 See Documentation/admin-guide/cputopology.rst for more information.
  38
  39
  40 What:           /sys/devices/system/cpu/probe
  41                 /sys/devices/system/cpu/release
  42 Date:           November 2009
  43 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
  44 Description:    Dynamic addition and removal of CPU's.  This is not hotplug
  45                 removal, this is meant complete removal/addition of the CPU
  46                 from the system.
  47
  48                 probe: writes to this file will dynamically add a CPU to the
  49                 system.  Information written to the file to add CPU's is
  50                 architecture specific.
  51
  52                 release: writes to this file dynamically remove a CPU from
  53                 the system.  Information written to the file to remove CPU's
  54                 is architecture specific.
  55
  56 What:           /sys/devices/system/cpu/cpuX/node
  57 Date:           October 2009
  58 Contact:        Linux memory management mailing list <linux-mm@kvack.org>
  59 Description:    Discover NUMA node a CPU belongs to
  60
  61                 When CONFIG_NUMA is enabled, a symbolic link that points
  62                 to the corresponding NUMA node directory.
  63
  64                 For example, the following symlink is created for cpu42
  65                 in NUMA node 2:
  66
  67                 /sys/devices/system/cpu/cpu42/node2 -> ../../node/node2
  68
  69
  70 What:           /sys/devices/system/cpu/cpuX/topology/core_id
  71                 /sys/devices/system/cpu/cpuX/topology/core_siblings
  72                 /sys/devices/system/cpu/cpuX/topology/core_siblings_list
  73                 /sys/devices/system/cpu/cpuX/topology/physical_package_id
  74                 /sys/devices/system/cpu/cpuX/topology/thread_siblings
  75                 /sys/devices/system/cpu/cpuX/topology/thread_siblings_list
  76                 /sys/devices/system/cpu/cpuX/topology/ppin
  77 Date:           December 2008
  78 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
  79 Description:    CPU topology files that describe a logical CPU's relationship
  80                 to other cores and threads in the same physical package.
  81
  82                 One cpuX directory is created per logical CPU in the system,
  83                 e.g. /sys/devices/system/cpu/cpu42/.
  84
  85                 Briefly, the files above are:
  86
  87                 core_id: the CPU core ID of cpuX. Typically it is the
  88                 hardware platform's identifier (rather than the kernel's).
  89                 The actual value is architecture and platform dependent.
  90
  91                 core_siblings: internal kernel map of cpuX's hardware threads
  92                 within the same physical_package_id.
  93
  94                 core_siblings_list: human-readable list of the logical CPU
  95                 numbers within the same physical_package_id as cpuX.
  96
  97                 physical_package_id: physical package id of cpuX. Typically
  98                 corresponds to a physical socket number, but the actual value
  99                 is architecture and platform dependent.
 100
 101                 thread_siblings: internal kernel map of cpuX's hardware
 102                 threads within the same core as cpuX
 103
 104                 thread_siblings_list: human-readable list of cpuX's hardware
 105                 threads within the same core as cpuX
 106
 107                 ppin: human-readable Protected Processor Identification
 108                 Number of the socket the cpu# belongs to. There should be
 109                 one per physical_package_id. File is readable only to
 110                 admin.
 111
 112                 See Documentation/admin-guide/cputopology.rst for more information.
 113
 114
 115 What:           /sys/devices/system/cpu/cpuidle/available_governors
 116                 /sys/devices/system/cpu/cpuidle/current_driver
 117                 /sys/devices/system/cpu/cpuidle/current_governor
 118                 /sys/devices/system/cpu/cpuidle/current_governer_ro
 119 Date:           September 2007
 120 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 121 Description:    Discover cpuidle policy and mechanism
 122
 123                 Various CPUs today support multiple idle levels that are
 124                 differentiated by varying exit latencies and power
 125                 consumption during idle.
 126
 127                 Idle policy (governor) is differentiated from idle mechanism
 128                 (driver).
 129
 130                 available_governors: (RO) displays a space separated list of
 131                 available governors.
 132
 133                 current_driver: (RO) displays current idle mechanism.
 134
 135                 current_governor: (RW) displays current idle policy. Users can
 136                 switch the governor at runtime by writing to this file.
 137
 138                 current_governor_ro: (RO) displays current idle policy.
 139
 140                 See Documentation/admin-guide/pm/cpuidle.rst and
 141                 Documentation/driver-api/pm/cpuidle.rst for more information.
 142
 143
 144 What:           /sys/devices/system/cpu/cpuX/cpuidle/state<N>/name
 145                 /sys/devices/system/cpu/cpuX/cpuidle/stateN/latency
 146                 /sys/devices/system/cpu/cpuX/cpuidle/stateN/power
 147                 /sys/devices/system/cpu/cpuX/cpuidle/stateN/time
 148                 /sys/devices/system/cpu/cpuX/cpuidle/stateN/usage
 149                 /sys/devices/system/cpu/cpuX/cpuidle/stateN/above
 150                 /sys/devices/system/cpu/cpuX/cpuidle/stateN/below
 151 Date:           September 2007
 152 KernelVersion:  v2.6.24
 153 Contact:        Linux power management list <linux-pm@vger.kernel.org>
 154 Description:
 155                 The directory /sys/devices/system/cpu/cpuX/cpuidle contains per
 156                 logical CPU specific cpuidle information for each online cpu X.
 157                 The processor idle states which are available for use have the
 158                 following attributes:
 159
 160                 ======== ==== =================================================
 161                 name:    (RO) Name of the idle state (string).
 162
 163                 latency: (RO) The latency to exit out of this idle state (in
 164                               microseconds).
 165
 166                 power:   (RO) The power consumed while in this idle state (in
 167                               milliwatts).
 168
 169                 time:    (RO) The total time spent in this idle state
 170                               (in microseconds).
 171
 172                 usage:   (RO) Number of times this state was entered (a count).
 173
 174                 above:   (RO) Number of times this state was entered, but the
 175                               observed CPU idle duration was too short for it
 176                               (a count).
 177
 178                 below:   (RO) Number of times this state was entered, but the
 179                               observed CPU idle duration was too long for it
 180                               (a count).
 181                 ======== ==== =================================================
 182
 183 What:           /sys/devices/system/cpu/cpuX/cpuidle/state<N>/desc
 184 Date:           February 2008
 185 KernelVersion:  v2.6.25
 186 Contact:        Linux power management list <linux-pm@vger.kernel.org>
 187 Description:
 188                 (RO) A small description about the idle state (string).
 189
 190
 191 What:           /sys/devices/system/cpu/cpuX/cpuidle/state<N>/disable
 192 Date:           March 2012
 193 KernelVersion:  v3.10
 194 Contact:        Linux power management list <linux-pm@vger.kernel.org>
 195 Description:
 196                 (RW) Option to disable this idle state (bool). The behavior and
 197                 the effect of the disable variable depends on the implementation
 198                 of a particular governor. In the ladder governor, for example,
 199                 it is not coherent, i.e. if one is disabling a light state, then
 200                 all deeper states are disabled as well, but the disable variable
 201                 does not reflect it. Likewise, if one enables a deep state but a
 202                 lighter state still is disabled, then this has no effect.
 203
 204 What:           /sys/devices/system/cpu/cpuX/cpuidle/state<N>/default_status
 205 Date:           December 2019
 206 KernelVersion:  v5.6
 207 Contact:        Linux power management list <linux-pm@vger.kernel.org>
 208 Description:
 209                 (RO) The default status of this state, "enabled" or "disabled".
 210
 211 What:           /sys/devices/system/cpu/cpuX/cpuidle/state<N>/residency
 212 Date:           March 2014
 213 KernelVersion:  v3.15
 214 Contact:        Linux power management list <linux-pm@vger.kernel.org>
 215 Description:
 216                 (RO) Display the target residency i.e. the minimum amount of
 217                 time (in microseconds) this cpu should spend in this idle state
 218                 to make the transition worth the effort.
 219
 220 What:           /sys/devices/system/cpu/cpuX/cpuidle/state<N>/s2idle/
 221 Date:           March 2018
 222 KernelVersion:  v4.17
 223 Contact:        Linux power management list <linux-pm@vger.kernel.org>
 224 Description:
 225                 Idle state usage statistics related to suspend-to-idle.
 226
 227                 This attribute group is only present for states that can be
 228                 used in suspend-to-idle with suspended timekeeping.
 229
 230 What:           /sys/devices/system/cpu/cpuX/cpuidle/state<N>/s2idle/time
 231 Date:           March 2018
 232 KernelVersion:  v4.17
 233 Contact:        Linux power management list <linux-pm@vger.kernel.org>
 234 Description:
 235                 Total time spent by the CPU in suspend-to-idle (with scheduler
 236                 tick suspended) after requesting this state.
 237
 238 What:           /sys/devices/system/cpu/cpuX/cpuidle/state<N>/s2idle/usage
 239 Date:           March 2018
 240 KernelVersion:  v4.17
 241 Contact:        Linux power management list <linux-pm@vger.kernel.org>
 242 Description:
 243                 Total number of times this state has been requested by the CPU
 244                 while entering suspend-to-idle.
 245
 246 What:           /sys/devices/system/cpu/cpuX/cpufreq/*
 247 Date:           pre-git history
 248 Contact:        linux-pm@vger.kernel.org
 249 Description:    Discover and change clock speed of CPUs
 250
 251                 Clock scaling allows you to change the clock speed of the
 252                 CPUs on the fly. This is a nice method to save battery
 253                 power, because the lower the clock speed, the less power
 254                 the CPU consumes.
 255
 256                 There are many knobs to tweak in this directory.
 257
 258                 See files in Documentation/cpu-freq/ for more information.
 259
 260
 261 What:           /sys/devices/system/cpu/cpuX/cpufreq/freqdomain_cpus
 262 Date:           June 2013
 263 Contact:        linux-pm@vger.kernel.org
 264 Description:    Discover CPUs in the same CPU frequency coordination domain
 265
 266                 freqdomain_cpus is the list of CPUs (online+offline) that share
 267                 the same clock/freq domain (possibly at the hardware level).
 268                 That information may be hidden from the cpufreq core and the
 269                 value of related_cpus may be different from freqdomain_cpus. This
 270                 attribute is useful for user space DVFS controllers to get better
 271                 power/performance results for platforms using acpi-cpufreq.
 272
 273                 This file is only present if the acpi-cpufreq or the cppc-cpufreq
 274                 drivers are in use.
 275
 276
 277 What:           /sys/devices/system/cpu/cpu*/cache/index3/cache_disable_{0,1}
 278 Date:           August 2008
 279 KernelVersion:  2.6.27
 280 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 281 Description:    Disable L3 cache indices
 282
 283                 These files exist in every CPU's cache/index3 directory. Each
 284                 cache_disable_{0,1} file corresponds to one disable slot which
 285                 can be used to disable a cache index. Reading from these files
 286                 on a processor with this functionality will return the currently
 287                 disabled index for that node. There is one L3 structure per
 288                 node, or per internal node on MCM machines. Writing a valid
 289                 index to one of these files will cause the specified cache
 290                 index to be disabled.
 291
 292                 All AMD processors with L3 caches provide this functionality.
 293                 For details, see BKDGs at
 294                 https://www.amd.com/en/support/tech-docs?keyword=bios+kernel
 295
 296
 297 What:           /sys/devices/system/cpu/cpufreq/boost
 298 Date:           August 2012
 299 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 300 Description:    Processor frequency boosting control
 301
 302                 This switch controls the boost setting for the whole system.
 303                 Boosting allows the CPU and the firmware to run at a frequency
 304                 beyond it's nominal limit.
 305
 306                 More details can be found in
 307                 Documentation/admin-guide/pm/cpufreq.rst
 308
 309
 310 What:           /sys/devices/system/cpu/cpuX/crash_notes
 311                 /sys/devices/system/cpu/cpuX/crash_notes_size
 312 Date:           April 2013
 313 Contact:        kexec@lists.infradead.org
 314 Description:    address and size of the percpu note.
 315
 316                 crash_notes: the physical address of the memory that holds the
 317                 note of cpuX.
 318
 319                 crash_notes_size: size of the note of cpuX.
 320
 321
 322 What:           /sys/devices/system/cpu/intel_pstate/max_perf_pct
 323                 /sys/devices/system/cpu/intel_pstate/min_perf_pct
 324                 /sys/devices/system/cpu/intel_pstate/no_turbo
 325 Date:           February 2013
 326 Contact:        linux-pm@vger.kernel.org
 327 Description:    Parameters for the Intel P-state driver
 328
 329                 Logic for selecting the current P-state in Intel
 330                 Sandybridge+ processors. The three knobs control
 331                 limits for the P-state that will be requested by the
 332                 driver.
 333
 334                 max_perf_pct: limits the maximum P state that will be requested by
 335                 the driver stated as a percentage of the available performance.
 336
 337                 min_perf_pct: limits the minimum P state that will be requested by
 338                 the driver stated as a percentage of the available performance.
 339
 340                 no_turbo: limits the driver to selecting P states below the turbo
 341                 frequency range.
 342
 343                 More details can be found in
 344                 Documentation/admin-guide/pm/intel_pstate.rst
 345
 346 What:           /sys/devices/system/cpu/cpu*/cache/index*/<set_of_attributes_mentioned_below>
 347 Date:           July 2014(documented, existed before August 2008)
 348 Contact:        Sudeep Holla <sudeep.holla@arm.com>
 349                 Linux kernel mailing list <linux-kernel@vger.kernel.org>
 350 Description:    Parameters for the CPU cache attributes
 351
 352                 allocation_policy:
 353                         - WriteAllocate:
 354                                         allocate a memory location to a cache line
 355                                         on a cache miss because of a write
 356                         - ReadAllocate:
 357                                         allocate a memory location to a cache line
 358                                         on a cache miss because of a read
 359                         - ReadWriteAllocate:
 360                                         both writeallocate and readallocate
 361
 362                 attributes:
 363                             LEGACY used only on IA64 and is same as write_policy
 364
 365                 coherency_line_size:
 366                                      the minimum amount of data in bytes that gets
 367                                      transferred from memory to cache
 368
 369                 level:
 370                         the cache hierarchy in the multi-level cache configuration
 371
 372                 number_of_sets:
 373                                 total number of sets in the cache, a set is a
 374                                 collection of cache lines with the same cache index
 375
 376                 physical_line_partition:
 377                                 number of physical cache line per cache tag
 378
 379                 shared_cpu_list:
 380                                 the list of logical cpus sharing the cache
 381
 382                 shared_cpu_map:
 383                                 logical cpu mask containing the list of cpus sharing
 384                                 the cache
 385
 386                 size:
 387                         the total cache size in kB
 388
 389                 type:
 390                         - Instruction: cache that only holds instructions
 391                         - Data: cache that only caches data
 392                         - Unified: cache that holds both data and instructions
 393
 394                 ways_of_associativity:
 395                         degree of freedom in placing a particular block
 396                         of memory in the cache
 397
 398                 write_policy:
 399                         - WriteThrough:
 400                                         data is written to both the cache line
 401                                         and to the block in the lower-level memory
 402                         - WriteBack:
 403                                      data is written only to the cache line and
 404                                      the modified cache line is written to main
 405                                      memory only when it is replaced
 406
 407
 408 What:           /sys/devices/system/cpu/cpu*/cache/index*/id
 409 Date:           September 2016
 410 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 411 Description:    Cache id
 412
 413                 The id provides a unique number for a specific instance of
 414                 a cache of a particular type. E.g. there may be a level
 415                 3 unified cache on each socket in a server and we may
 416                 assign them ids 0, 1, 2, ...
 417
 418                 Note that id value can be non-contiguous. E.g. level 1
 419                 caches typically exist per core, but there may not be a
 420                 power of two cores on a socket, so these caches may be
 421                 numbered 0, 1, 2, 3, 4, 5, 8, 9, 10, ...
 422
 423 What:           /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats
 424                 /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/turbo_stat
 425                 /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/sub_turbo_stat
 426                 /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/unthrottle
 427                 /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/powercap
 428                 /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/overtemp
 429                 /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/supply_fault
 430                 /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/overcurrent
 431                 /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/occ_reset
 432 Date:           March 2016
 433 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 434                 Linux for PowerPC mailing list <linuxppc-dev@ozlabs.org>
 435 Description:    POWERNV CPUFreq driver's frequency throttle stats directory and
 436                 attributes
 437
 438                 'cpuX/cpufreq/throttle_stats' directory contains the CPU frequency
 439                 throttle stat attributes for the chip. The throttle stats of a cpu
 440                 is common across all the cpus belonging to a chip. Below are the
 441                 throttle attributes exported in the 'throttle_stats' directory:
 442
 443                 - turbo_stat : This file gives the total number of times the max
 444                   frequency is throttled to lower frequency in turbo (at and above
 445                   nominal frequency) range of frequencies.
 446
 447                 - sub_turbo_stat : This file gives the total number of times the
 448                   max frequency is throttled to lower frequency in sub-turbo(below
 449                   nominal frequency) range of frequencies.
 450
 451                 - unthrottle : This file gives the total number of times the max
 452                   frequency is unthrottled after being throttled.
 453
 454                 - powercap : This file gives the total number of times the max
 455                   frequency is throttled due to 'Power Capping'.
 456
 457                 - overtemp : This file gives the total number of times the max
 458                   frequency is throttled due to 'CPU Over Temperature'.
 459
 460                 - supply_fault : This file gives the total number of times the
 461                   max frequency is throttled due to 'Power Supply Failure'.
 462
 463                 - overcurrent : This file gives the total number of times the
 464                   max frequency is throttled due to 'Overcurrent'.
 465
 466                 - occ_reset : This file gives the total number of times the max
 467                   frequency is throttled due to 'OCC Reset'.
 468
 469                 The sysfs attributes representing different throttle reasons like
 470                 powercap, overtemp, supply_fault, overcurrent and occ_reset map to
 471                 the reasons provided by OCC firmware for throttling the frequency.
 472
 473 What:           /sys/devices/system/cpu/cpufreq/policyX/throttle_stats
 474                 /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/turbo_stat
 475                 /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/sub_turbo_stat
 476                 /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/unthrottle
 477                 /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/powercap
 478                 /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/overtemp
 479                 /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/supply_fault
 480                 /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/overcurrent
 481                 /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/occ_reset
 482 Date:           March 2016
 483 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 484                 Linux for PowerPC mailing list <linuxppc-dev@ozlabs.org>
 485 Description:    POWERNV CPUFreq driver's frequency throttle stats directory and
 486                 attributes
 487
 488                 'policyX/throttle_stats' directory and all the attributes are same as
 489                 the /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats directory and
 490                 attributes which give the frequency throttle information of the chip.
 491
 492 What:           /sys/devices/system/cpu/cpuX/regs/
 493                 /sys/devices/system/cpu/cpuX/regs/identification/
 494                 /sys/devices/system/cpu/cpuX/regs/identification/midr_el1
 495                 /sys/devices/system/cpu/cpuX/regs/identification/revidr_el1
 496 Date:           June 2016
 497 Contact:        Linux ARM Kernel Mailing list <linux-arm-kernel@lists.infradead.org>
 498 Description:    AArch64 CPU registers
 499
 500                 'identification' directory exposes the CPU ID registers for
 501                 identifying model and revision of the CPU.
 502
 503 What:           /sys/devices/system/cpu/aarch32_el0
 504 Date:           May 2021
 505 Contact:        Linux ARM Kernel Mailing list <linux-arm-kernel@lists.infradead.org>
 506 Description:    Identifies the subset of CPUs in the system that can execute
 507                 AArch32 (32-bit ARM) applications. If present, the same format as
 508                 /sys/devices/system/cpu/{offline,online,possible,present} is used.
 509                 If absent, then all or none of the CPUs can execute AArch32
 510                 applications and execve() will behave accordingly.
 511
 512 What:           /sys/devices/system/cpu/cpuX/cpu_capacity
 513 Date:           December 2016
 514 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 515 Description:    information about CPUs heterogeneity.
 516
 517                 cpu_capacity: capacity of cpuX.
 518
 519 What:           /sys/devices/system/cpu/vulnerabilities
 520                 /sys/devices/system/cpu/vulnerabilities/meltdown
 521                 /sys/devices/system/cpu/vulnerabilities/spectre_v1
 522                 /sys/devices/system/cpu/vulnerabilities/spectre_v2
 523                 /sys/devices/system/cpu/vulnerabilities/spec_store_bypass
 524                 /sys/devices/system/cpu/vulnerabilities/l1tf
 525                 /sys/devices/system/cpu/vulnerabilities/mds
 526                 /sys/devices/system/cpu/vulnerabilities/srbds
 527                 /sys/devices/system/cpu/vulnerabilities/tsx_async_abort
 528                 /sys/devices/system/cpu/vulnerabilities/itlb_multihit
 529                 /sys/devices/system/cpu/vulnerabilities/mmio_stale_data
 530 Date:           January 2018
 531 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 532 Description:    Information about CPU vulnerabilities
 533
 534                 The files are named after the code names of CPU
 535                 vulnerabilities. The output of those files reflects the
 536                 state of the CPUs in the system. Possible output values:
 537
 538                 ================  ==============================================
 539                 "Not affected"    CPU is not affected by the vulnerability
 540                 "Vulnerable"      CPU is affected and no mitigation in effect
 541                 "Mitigation: $M"  CPU is affected and mitigation $M is in effect
 542                 ================  ==============================================
 543
 544                 See also: Documentation/admin-guide/hw-vuln/index.rst
 545
 546 What:           /sys/devices/system/cpu/smt
 547                 /sys/devices/system/cpu/smt/active
 548                 /sys/devices/system/cpu/smt/control
 549 Date:           June 2018
 550 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 551 Description:    Control Symmetric Multi Threading (SMT)
 552
 553                 active:  Tells whether SMT is active (enabled and siblings online)
 554
 555                 control: Read/write interface to control SMT. Possible
 556                          values:
 557
 558                          ================ =========================================
 559                          "on"             SMT is enabled
 560                          "off"            SMT is disabled
 561                          "forceoff"       SMT is force disabled. Cannot be changed.
 562                          "notsupported"   SMT is not supported by the CPU
 563                          "notimplemented" SMT runtime toggling is not
 564                                           implemented for the architecture
 565                          ================ =========================================
 566
 567                          If control status is "forceoff" or "notsupported" writes
 568                          are rejected.
 569
 570 What:           /sys/devices/system/cpu/cpuX/power/energy_perf_bias
 571 Date:           March 2019
 572 Contact:        linux-pm@vger.kernel.org
 573 Description:    Intel Energy and Performance Bias Hint (EPB)
 574
 575                 EPB for the given CPU in a sliding scale 0 - 15, where a value
 576                 of 0 corresponds to a hint preference for highest performance
 577                 and a value of 15 corresponds to the maximum energy savings.
 578
 579                 In order to change the EPB value for the CPU, write either
 580                 a number in the 0 - 15 sliding scale above, or one of the
 581                 strings: "performance", "balance-performance", "normal",
 582                 "balance-power", "power" (that represent values reflected by
 583                 their meaning), to this attribute.
 584
 585                 This attribute is present for all online CPUs supporting the
 586                 Intel EPB feature.
 587
 588 What:           /sys/devices/system/cpu/umwait_control
 589                 /sys/devices/system/cpu/umwait_control/enable_c02
 590                 /sys/devices/system/cpu/umwait_control/max_time
 591 Date:           May 2019
 592 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 593 Description:    Umwait control
 594
 595                 enable_c02: Read/write interface to control umwait C0.2 state
 596                         Read returns C0.2 state status:
 597                                 0: C0.2 is disabled
 598                                 1: C0.2 is enabled
 599
 600                         Write 'y' or '1'  or 'on' to enable C0.2 state.
 601                         Write 'n' or '0'  or 'off' to disable C0.2 state.
 602
 603                         The interface is case insensitive.
 604
 605                 max_time: Read/write interface to control umwait maximum time
 606                           in TSC-quanta that the CPU can reside in either C0.1
 607                           or C0.2 state. The time is an unsigned 32-bit number.
 608                           Note that a value of zero means there is no limit.
 609                           Low order two bits must be zero.
 610
 611 What:           /sys/devices/system/cpu/svm
 612 Date:           August 2019
 613 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 614                 Linux for PowerPC mailing list <linuxppc-dev@ozlabs.org>
 615 Description:    Secure Virtual Machine
 616
 617                 If 1, it means the system is using the Protected Execution
 618                 Facility in POWER9 and newer processors. i.e., it is a Secure
 619                 Virtual Machine.
 620
 621 What:           /sys/devices/system/cpu/cpuX/purr
 622 Date:           Apr 2005
 623 Contact:        Linux for PowerPC mailing list <linuxppc-dev@ozlabs.org>
 624 Description:    PURR ticks for this CPU since the system boot.
 625
 626                 The Processor Utilization Resources Register (PURR) is
 627                 a 64-bit counter which provides an estimate of the
 628                 resources used by the CPU thread. The contents of this
 629                 register increases monotonically. This sysfs interface
 630                 exposes the number of PURR ticks for cpuX.
 631
 632 What:           /sys/devices/system/cpu/cpuX/spurr
 633 Date:           Dec 2006
 634 Contact:        Linux for PowerPC mailing list <linuxppc-dev@ozlabs.org>
 635 Description:    SPURR ticks for this CPU since the system boot.
 636
 637                 The Scaled Processor Utilization Resources Register
 638                 (SPURR) is a 64-bit counter that provides a frequency
 639                 invariant estimate of the resources used by the CPU
 640                 thread. The contents of this register increases
 641                 monotonically. This sysfs interface exposes the number
 642                 of SPURR ticks for cpuX.
 643
 644 What:           /sys/devices/system/cpu/cpuX/idle_purr
 645 Date:           Apr 2020
 646 Contact:        Linux for PowerPC mailing list <linuxppc-dev@ozlabs.org>
 647 Description:    PURR ticks for cpuX when it was idle.
 648
 649                 This sysfs interface exposes the number of PURR ticks
 650                 for cpuX when it was idle.
 651
 652 What:           /sys/devices/system/cpu/cpuX/idle_spurr
 653 Date:           Apr 2020
 654 Contact:        Linux for PowerPC mailing list <linuxppc-dev@ozlabs.org>
 655 Description:    SPURR ticks for cpuX when it was idle.
 656
 657                 This sysfs interface exposes the number of SPURR ticks
 658                 for cpuX when it was idle.
 659
 660 What:           /sys/devices/system/cpu/cpuX/mte_tcf_preferred
 661 Date:           July 2021
 662 Contact:        Linux ARM Kernel Mailing list <linux-arm-kernel@lists.infradead.org>
 663 Description:    Preferred MTE tag checking mode
 664
 665                 When a user program specifies more than one MTE tag checking
 666                 mode, this sysfs node is used to specify which mode should
 667                 be preferred when scheduling a task on that CPU. Possible
 668                 values:
 669
 670                 ================  ==============================================
 671                 "sync"            Prefer synchronous mode
 672                 "asymm"           Prefer asymmetric mode
 673                 "async"           Prefer asynchronous mode
 674                 ================  ==============================================
 675
 676                 See also: Documentation/arm64/memory-tagging-extension.rst
 677
 678 What:           /sys/devices/system/cpu/nohz_full
 679 Date:           Apr 2015
 680 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 681 Description:
 682                 (RO) the list of CPUs that are in nohz_full mode.
 683                 These CPUs are set by boot parameter "nohz_full=".
 684
 685 What:           /sys/devices/system/cpu/isolated
 686 Date:           Apr 2015
 687 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 688 Description:
 689                 (RO) the list of CPUs that are isolated and don't
 690                 participate in load balancing. These CPUs are set by
 691                 boot parameter "isolcpus=".