Documentation/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_siblings
  71                 /sys/devices/system/cpu/cpuX/topology/core_siblings_list
  72                 /sys/devices/system/cpu/cpuX/topology/physical_package_id
  73                 /sys/devices/system/cpu/cpuX/topology/thread_siblings
  74                 /sys/devices/system/cpu/cpuX/topology/thread_siblings_list
  75                 /sys/devices/system/cpu/cpuX/topology/ppin
  76 Date:           December 2008
  77 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
  78 Description:    CPU topology files that describe a logical CPU's relationship
  79                 to other cores and threads in the same physical package.
  80
  81                 One cpuX directory is created per logical CPU in the system,
  82                 e.g. /sys/devices/system/cpu/cpu42/.
  83
  84                 Briefly, the files above are:
  85
  86                 core_siblings: internal kernel map of cpuX's hardware threads
  87                 within the same physical_package_id.
  88
  89                 core_siblings_list: human-readable list of the logical CPU
  90                 numbers within the same physical_package_id as cpuX.
  91
  92                 physical_package_id: physical package id of cpuX. Typically
  93                 corresponds to a physical socket number, but the actual value
  94                 is architecture and platform dependent.
  95
  96                 thread_siblings: internal kernel map of cpuX's hardware
  97                 threads within the same core as cpuX
  98
  99                 thread_siblings_list: human-readable list of cpuX's hardware
 100                 threads within the same core as cpuX
 101
 102                 ppin: human-readable Protected Processor Identification
 103                 Number of the socket the cpu# belongs to. There should be
 104                 one per physical_package_id. File is readable only to
 105                 admin.
 106
 107                 See Documentation/admin-guide/cputopology.rst for more information.
 108
 109
 110 What:           /sys/devices/system/cpu/cpuidle/available_governors
 111                 /sys/devices/system/cpu/cpuidle/current_driver
 112                 /sys/devices/system/cpu/cpuidle/current_governor
 113                 /sys/devices/system/cpu/cpuidle/current_governer_ro
 114 Date:           September 2007
 115 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 116 Description:    Discover cpuidle policy and mechanism
 117
 118                 Various CPUs today support multiple idle levels that are
 119                 differentiated by varying exit latencies and power
 120                 consumption during idle.
 121
 122                 Idle policy (governor) is differentiated from idle mechanism
 123                 (driver).
 124
 125                 available_governors: (RO) displays a space separated list of
 126                 available governors.
 127
 128                 current_driver: (RO) displays current idle mechanism.
 129
 130                 current_governor: (RW) displays current idle policy. Users can
 131                 switch the governor at runtime by writing to this file.
 132
 133                 current_governor_ro: (RO) displays current idle policy.
 134
 135                 See Documentation/admin-guide/pm/cpuidle.rst and
 136                 Documentation/driver-api/pm/cpuidle.rst for more information.
 137
 138
 139 What:           /sys/devices/system/cpu/cpuX/cpuidle/state<N>/name
 140                 /sys/devices/system/cpu/cpuX/cpuidle/stateN/latency
 141                 /sys/devices/system/cpu/cpuX/cpuidle/stateN/power
 142                 /sys/devices/system/cpu/cpuX/cpuidle/stateN/time
 143                 /sys/devices/system/cpu/cpuX/cpuidle/stateN/usage
 144                 /sys/devices/system/cpu/cpuX/cpuidle/stateN/above
 145                 /sys/devices/system/cpu/cpuX/cpuidle/stateN/below
 146 Date:           September 2007
 147 KernelVersion:  v2.6.24
 148 Contact:        Linux power management list <linux-pm@vger.kernel.org>
 149 Description:
 150                 The directory /sys/devices/system/cpu/cpuX/cpuidle contains per
 151                 logical CPU specific cpuidle information for each online cpu X.
 152                 The processor idle states which are available for use have the
 153                 following attributes:
 154
 155                 ======== ==== =================================================
 156                 name:    (RO) Name of the idle state (string).
 157
 158                 latency: (RO) The latency to exit out of this idle state (in
 159                               microseconds).
 160
 161                 power:   (RO) The power consumed while in this idle state (in
 162                               milliwatts).
 163
 164                 time:    (RO) The total time spent in this idle state
 165                               (in microseconds).
 166
 167                 usage:   (RO) Number of times this state was entered (a count).
 168
 169                 above:   (RO) Number of times this state was entered, but the
 170                               observed CPU idle duration was too short for it
 171                               (a count).
 172
 173                 below:   (RO) Number of times this state was entered, but the
 174                               observed CPU idle duration was too long for it
 175                               (a count).
 176                 ======== ==== =================================================
 177
 178 What:           /sys/devices/system/cpu/cpuX/cpuidle/state<N>/desc
 179 Date:           February 2008
 180 KernelVersion:  v2.6.25
 181 Contact:        Linux power management list <linux-pm@vger.kernel.org>
 182 Description:
 183                 (RO) A small description about the idle state (string).
 184
 185
 186 What:           /sys/devices/system/cpu/cpuX/cpuidle/state<N>/disable
 187 Date:           March 2012
 188 KernelVersion:  v3.10
 189 Contact:        Linux power management list <linux-pm@vger.kernel.org>
 190 Description:
 191                 (RW) Option to disable this idle state (bool). The behavior and
 192                 the effect of the disable variable depends on the implementation
 193                 of a particular governor. In the ladder governor, for example,
 194                 it is not coherent, i.e. if one is disabling a light state, then
 195                 all deeper states are disabled as well, but the disable variable
 196                 does not reflect it. Likewise, if one enables a deep state but a
 197                 lighter state still is disabled, then this has no effect.
 198
 199 What:           /sys/devices/system/cpu/cpuX/cpuidle/state<N>/default_status
 200 Date:           December 2019
 201 KernelVersion:  v5.6
 202 Contact:        Linux power management list <linux-pm@vger.kernel.org>
 203 Description:
 204                 (RO) The default status of this state, "enabled" or "disabled".
 205
 206 What:           /sys/devices/system/cpu/cpuX/cpuidle/state<N>/residency
 207 Date:           March 2014
 208 KernelVersion:  v3.15
 209 Contact:        Linux power management list <linux-pm@vger.kernel.org>
 210 Description:
 211                 (RO) Display the target residency i.e. the minimum amount of
 212                 time (in microseconds) this cpu should spend in this idle state
 213                 to make the transition worth the effort.
 214
 215 What:           /sys/devices/system/cpu/cpuX/cpuidle/state<N>/s2idle/
 216 Date:           March 2018
 217 KernelVersion:  v4.17
 218 Contact:        Linux power management list <linux-pm@vger.kernel.org>
 219 Description:
 220                 Idle state usage statistics related to suspend-to-idle.
 221
 222                 This attribute group is only present for states that can be
 223                 used in suspend-to-idle with suspended timekeeping.
 224
 225 What:           /sys/devices/system/cpu/cpuX/cpuidle/state<N>/s2idle/time
 226 Date:           March 2018
 227 KernelVersion:  v4.17
 228 Contact:        Linux power management list <linux-pm@vger.kernel.org>
 229 Description:
 230                 Total time spent by the CPU in suspend-to-idle (with scheduler
 231                 tick suspended) after requesting this state.
 232
 233 What:           /sys/devices/system/cpu/cpuX/cpuidle/state<N>/s2idle/usage
 234 Date:           March 2018
 235 KernelVersion:  v4.17
 236 Contact:        Linux power management list <linux-pm@vger.kernel.org>
 237 Description:
 238                 Total number of times this state has been requested by the CPU
 239                 while entering suspend-to-idle.
 240
 241 What:           /sys/devices/system/cpu/cpuX/cpufreq/*
 242 Date:           pre-git history
 243 Contact:        linux-pm@vger.kernel.org
 244 Description:    Discover and change clock speed of CPUs
 245
 246                 Clock scaling allows you to change the clock speed of the
 247                 CPUs on the fly. This is a nice method to save battery
 248                 power, because the lower the clock speed, the less power
 249                 the CPU consumes.
 250
 251                 There are many knobs to tweak in this directory.
 252
 253                 See files in Documentation/cpu-freq/ for more information.
 254
 255
 256 What:           /sys/devices/system/cpu/cpuX/cpufreq/freqdomain_cpus
 257 Date:           June 2013
 258 Contact:        linux-pm@vger.kernel.org
 259 Description:    Discover CPUs in the same CPU frequency coordination domain
 260
 261                 freqdomain_cpus is the list of CPUs (online+offline) that share
 262                 the same clock/freq domain (possibly at the hardware level).
 263                 That information may be hidden from the cpufreq core and the
 264                 value of related_cpus may be different from freqdomain_cpus. This
 265                 attribute is useful for user space DVFS controllers to get better
 266                 power/performance results for platforms using acpi-cpufreq.
 267
 268                 This file is only present if the acpi-cpufreq or the cppc-cpufreq
 269                 drivers are in use.
 270
 271
 272 What:           /sys/devices/system/cpu/cpu*/cache/index3/cache_disable_{0,1}
 273 Date:           August 2008
 274 KernelVersion:  2.6.27
 275 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 276 Description:    Disable L3 cache indices
 277
 278                 These files exist in every CPU's cache/index3 directory. Each
 279                 cache_disable_{0,1} file corresponds to one disable slot which
 280                 can be used to disable a cache index. Reading from these files
 281                 on a processor with this functionality will return the currently
 282                 disabled index for that node. There is one L3 structure per
 283                 node, or per internal node on MCM machines. Writing a valid
 284                 index to one of these files will cause the specified cache
 285                 index to be disabled.
 286
 287                 All AMD processors with L3 caches provide this functionality.
 288                 For details, see BKDGs at
 289                 https://www.amd.com/en/support/tech-docs?keyword=bios+kernel
 290
 291
 292 What:           /sys/devices/system/cpu/cpufreq/boost
 293 Date:           August 2012
 294 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 295 Description:    Processor frequency boosting control
 296
 297                 This switch controls the boost setting for the whole system.
 298                 Boosting allows the CPU and the firmware to run at a frequency
 299                 beyond its nominal limit.
 300
 301                 More details can be found in
 302                 Documentation/admin-guide/pm/cpufreq.rst
 303
 304
 305 What:           /sys/devices/system/cpu/cpuX/crash_notes
 306                 /sys/devices/system/cpu/cpuX/crash_notes_size
 307 Date:           April 2013
 308 Contact:        kexec@lists.infradead.org
 309 Description:    address and size of the percpu note.
 310
 311                 crash_notes: the physical address of the memory that holds the
 312                 note of cpuX.
 313
 314                 crash_notes_size: size of the note of cpuX.
 315
 316
 317 What:           /sys/devices/system/cpu/intel_pstate/max_perf_pct
 318                 /sys/devices/system/cpu/intel_pstate/min_perf_pct
 319                 /sys/devices/system/cpu/intel_pstate/no_turbo
 320 Date:           February 2013
 321 Contact:        linux-pm@vger.kernel.org
 322 Description:    Parameters for the Intel P-state driver
 323
 324                 Logic for selecting the current P-state in Intel
 325                 Sandybridge+ processors. The three knobs control
 326                 limits for the P-state that will be requested by the
 327                 driver.
 328
 329                 max_perf_pct: limits the maximum P state that will be requested by
 330                 the driver stated as a percentage of the available performance.
 331
 332                 min_perf_pct: limits the minimum P state that will be requested by
 333                 the driver stated as a percentage of the available performance.
 334
 335                 no_turbo: limits the driver to selecting P states below the turbo
 336                 frequency range.
 337
 338                 More details can be found in
 339                 Documentation/admin-guide/pm/intel_pstate.rst
 340
 341 What:           /sys/devices/system/cpu/cpu*/cache/index*/<set_of_attributes_mentioned_below>
 342 Date:           July 2014(documented, existed before August 2008)
 343 Contact:        Sudeep Holla <sudeep.holla@arm.com>
 344                 Linux kernel mailing list <linux-kernel@vger.kernel.org>
 345 Description:    Parameters for the CPU cache attributes
 346
 347                 allocation_policy:
 348                         - WriteAllocate:
 349                                         allocate a memory location to a cache line
 350                                         on a cache miss because of a write
 351                         - ReadAllocate:
 352                                         allocate a memory location to a cache line
 353                                         on a cache miss because of a read
 354                         - ReadWriteAllocate:
 355                                         both writeallocate and readallocate
 356
 357                 coherency_line_size:
 358                                      the minimum amount of data in bytes that gets
 359                                      transferred from memory to cache
 360
 361                 level:
 362                         the cache hierarchy in the multi-level cache configuration
 363
 364                 number_of_sets:
 365                                 total number of sets in the cache, a set is a
 366                                 collection of cache lines with the same cache index
 367
 368                 physical_line_partition:
 369                                 number of physical cache line per cache tag
 370
 371                 shared_cpu_list:
 372                                 the list of logical cpus sharing the cache
 373
 374                 shared_cpu_map:
 375                                 logical cpu mask containing the list of cpus sharing
 376                                 the cache
 377
 378                 size:
 379                         the total cache size in kB
 380
 381                 type:
 382                         - Instruction: cache that only holds instructions
 383                         - Data: cache that only caches data
 384                         - Unified: cache that holds both data and instructions
 385
 386                 ways_of_associativity:
 387                         degree of freedom in placing a particular block
 388                         of memory in the cache
 389
 390                 write_policy:
 391                         - WriteThrough:
 392                                         data is written to both the cache line
 393                                         and to the block in the lower-level memory
 394                         - WriteBack:
 395                                      data is written only to the cache line and
 396                                      the modified cache line is written to main
 397                                      memory only when it is replaced
 398
 399
 400 What:           /sys/devices/system/cpu/cpu*/cache/index*/id
 401 Date:           September 2016
 402 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 403 Description:    Cache id
 404
 405                 The id provides a unique number for a specific instance of
 406                 a cache of a particular type. E.g. there may be a level
 407                 3 unified cache on each socket in a server and we may
 408                 assign them ids 0, 1, 2, ...
 409
 410                 Note that id value can be non-contiguous. E.g. level 1
 411                 caches typically exist per core, but there may not be a
 412                 power of two cores on a socket, so these caches may be
 413                 numbered 0, 1, 2, 3, 4, 5, 8, 9, 10, ...
 414
 415 What:           /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats
 416                 /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/turbo_stat
 417                 /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/sub_turbo_stat
 418                 /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/unthrottle
 419                 /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/powercap
 420                 /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/overtemp
 421                 /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/supply_fault
 422                 /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/overcurrent
 423                 /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats/occ_reset
 424 Date:           March 2016
 425 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 426                 Linux for PowerPC mailing list <linuxppc-dev@lists.ozlabs.org>
 427 Description:    POWERNV CPUFreq driver's frequency throttle stats directory and
 428                 attributes
 429
 430                 'cpuX/cpufreq/throttle_stats' directory contains the CPU frequency
 431                 throttle stat attributes for the chip. The throttle stats of a cpu
 432                 is common across all the cpus belonging to a chip. Below are the
 433                 throttle attributes exported in the 'throttle_stats' directory:
 434
 435                 - turbo_stat : This file gives the total number of times the max
 436                   frequency is throttled to lower frequency in turbo (at and above
 437                   nominal frequency) range of frequencies.
 438
 439                 - sub_turbo_stat : This file gives the total number of times the
 440                   max frequency is throttled to lower frequency in sub-turbo(below
 441                   nominal frequency) range of frequencies.
 442
 443                 - unthrottle : This file gives the total number of times the max
 444                   frequency is unthrottled after being throttled.
 445
 446                 - powercap : This file gives the total number of times the max
 447                   frequency is throttled due to 'Power Capping'.
 448
 449                 - overtemp : This file gives the total number of times the max
 450                   frequency is throttled due to 'CPU Over Temperature'.
 451
 452                 - supply_fault : This file gives the total number of times the
 453                   max frequency is throttled due to 'Power Supply Failure'.
 454
 455                 - overcurrent : This file gives the total number of times the
 456                   max frequency is throttled due to 'Overcurrent'.
 457
 458                 - occ_reset : This file gives the total number of times the max
 459                   frequency is throttled due to 'OCC Reset'.
 460
 461                 The sysfs attributes representing different throttle reasons like
 462                 powercap, overtemp, supply_fault, overcurrent and occ_reset map to
 463                 the reasons provided by OCC firmware for throttling the frequency.
 464
 465 What:           /sys/devices/system/cpu/cpufreq/policyX/throttle_stats
 466                 /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/turbo_stat
 467                 /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/sub_turbo_stat
 468                 /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/unthrottle
 469                 /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/powercap
 470                 /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/overtemp
 471                 /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/supply_fault
 472                 /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/overcurrent
 473                 /sys/devices/system/cpu/cpufreq/policyX/throttle_stats/occ_reset
 474 Date:           March 2016
 475 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 476                 Linux for PowerPC mailing list <linuxppc-dev@lists.ozlabs.org>
 477 Description:    POWERNV CPUFreq driver's frequency throttle stats directory and
 478                 attributes
 479
 480                 'policyX/throttle_stats' directory and all the attributes are same as
 481                 the /sys/devices/system/cpu/cpuX/cpufreq/throttle_stats directory and
 482                 attributes which give the frequency throttle information of the chip.
 483
 484 What:           /sys/devices/system/cpu/cpuX/regs/
 485                 /sys/devices/system/cpu/cpuX/regs/identification/
 486                 /sys/devices/system/cpu/cpuX/regs/identification/midr_el1
 487                 /sys/devices/system/cpu/cpuX/regs/identification/revidr_el1
 488                 /sys/devices/system/cpu/cpuX/regs/identification/smidr_el1
 489 Date:           June 2016
 490 Contact:        Linux ARM Kernel Mailing list <linux-arm-kernel@lists.infradead.org>
 491 Description:    AArch64 CPU registers
 492
 493                 'identification' directory exposes the CPU ID registers for
 494                 identifying model and revision of the CPU and SMCU.
 495
 496 What:           /sys/devices/system/cpu/aarch32_el0
 497 Date:           May 2021
 498 Contact:        Linux ARM Kernel Mailing list <linux-arm-kernel@lists.infradead.org>
 499 Description:    Identifies the subset of CPUs in the system that can execute
 500                 AArch32 (32-bit ARM) applications. If present, the same format as
 501                 /sys/devices/system/cpu/{offline,online,possible,present} is used.
 502                 If absent, then all or none of the CPUs can execute AArch32
 503                 applications and execve() will behave accordingly.
 504
 505 What:           /sys/devices/system/cpu/cpuX/cpu_capacity
 506 Date:           December 2016
 507 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 508 Description:    information about CPUs heterogeneity.
 509
 510                 cpu_capacity: capacity of cpuX.
 511
 512 What:           /sys/devices/system/cpu/vulnerabilities
 513                 /sys/devices/system/cpu/vulnerabilities/gather_data_sampling
 514                 /sys/devices/system/cpu/vulnerabilities/itlb_multihit
 515                 /sys/devices/system/cpu/vulnerabilities/l1tf
 516                 /sys/devices/system/cpu/vulnerabilities/mds
 517                 /sys/devices/system/cpu/vulnerabilities/meltdown
 518                 /sys/devices/system/cpu/vulnerabilities/mmio_stale_data
 519                 /sys/devices/system/cpu/vulnerabilities/reg_file_data_sampling
 520                 /sys/devices/system/cpu/vulnerabilities/retbleed
 521                 /sys/devices/system/cpu/vulnerabilities/spec_store_bypass
 522                 /sys/devices/system/cpu/vulnerabilities/spectre_v1
 523                 /sys/devices/system/cpu/vulnerabilities/spectre_v2
 524                 /sys/devices/system/cpu/vulnerabilities/srbds
 525                 /sys/devices/system/cpu/vulnerabilities/tsx_async_abort
 526 Date:           January 2018
 527 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 528 Description:    Information about CPU vulnerabilities
 529
 530                 The files are named after the code names of CPU
 531                 vulnerabilities. The output of those files reflects the
 532                 state of the CPUs in the system. Possible output values:
 533
 534                 ================  ==============================================
 535                 "Not affected"    CPU is not affected by the vulnerability
 536                 "Vulnerable"      CPU is affected and no mitigation in effect
 537                 "Mitigation: $M"  CPU is affected and mitigation $M is in effect
 538                 ================  ==============================================
 539
 540                 See also: Documentation/admin-guide/hw-vuln/index.rst
 541
 542 What:           /sys/devices/system/cpu/smt
 543                 /sys/devices/system/cpu/smt/active
 544                 /sys/devices/system/cpu/smt/control
 545 Date:           June 2018
 546 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 547 Description:    Control Symmetric Multi Threading (SMT)
 548
 549                 active:  Tells whether SMT is active (enabled and siblings online)
 550
 551                 control: Read/write interface to control SMT. Possible
 552                          values:
 553
 554                          ================ =========================================
 555                          "on"             SMT is enabled
 556                          "off"            SMT is disabled
 557                          "<N>"            SMT is enabled with N threads per core.
 558                          "forceoff"       SMT is force disabled. Cannot be changed.
 559                          "notsupported"   SMT is not supported by the CPU
 560                          "notimplemented" SMT runtime toggling is not
 561                                           implemented for the architecture
 562                          ================ =========================================
 563
 564                          If control status is "forceoff" or "notsupported" writes
 565                          are rejected.
 566
 567 What:           /sys/devices/system/cpu/cpuX/power/energy_perf_bias
 568 Date:           March 2019
 569 Contact:        linux-pm@vger.kernel.org
 570 Description:    Intel Energy and Performance Bias Hint (EPB)
 571
 572                 EPB for the given CPU in a sliding scale 0 - 15, where a value
 573                 of 0 corresponds to a hint preference for highest performance
 574                 and a value of 15 corresponds to the maximum energy savings.
 575
 576                 In order to change the EPB value for the CPU, write either
 577                 a number in the 0 - 15 sliding scale above, or one of the
 578                 strings: "performance", "balance-performance", "normal",
 579                 "balance-power", "power" (that represent values reflected by
 580                 their meaning), to this attribute.
 581
 582                 This attribute is present for all online CPUs supporting the
 583                 Intel EPB feature.
 584
 585 What:           /sys/devices/system/cpu/umwait_control
 586                 /sys/devices/system/cpu/umwait_control/enable_c02
 587                 /sys/devices/system/cpu/umwait_control/max_time
 588 Date:           May 2019
 589 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 590 Description:    Umwait control
 591
 592                 enable_c02: Read/write interface to control umwait C0.2 state
 593                         Read returns C0.2 state status:
 594                                 0: C0.2 is disabled
 595                                 1: C0.2 is enabled
 596
 597                         Write 'y' or '1'  or 'on' to enable C0.2 state.
 598                         Write 'n' or '0'  or 'off' to disable C0.2 state.
 599
 600                         The interface is case insensitive.
 601
 602                 max_time: Read/write interface to control umwait maximum time
 603                           in TSC-quanta that the CPU can reside in either C0.1
 604                           or C0.2 state. The time is an unsigned 32-bit number.
 605                           Note that a value of zero means there is no limit.
 606                           Low order two bits must be zero.
 607
 608 What:           /sys/devices/system/cpu/svm
 609 Date:           August 2019
 610 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 611                 Linux for PowerPC mailing list <linuxppc-dev@lists.ozlabs.org>
 612 Description:    Secure Virtual Machine
 613
 614                 If 1, it means the system is using the Protected Execution
 615                 Facility in POWER9 and newer processors. i.e., it is a Secure
 616                 Virtual Machine.
 617
 618 What:           /sys/devices/system/cpu/cpuX/purr
 619 Date:           Apr 2005
 620 Contact:        Linux for PowerPC mailing list <linuxppc-dev@lists.ozlabs.org>
 621 Description:    PURR ticks for this CPU since the system boot.
 622
 623                 The Processor Utilization Resources Register (PURR) is
 624                 a 64-bit counter which provides an estimate of the
 625                 resources used by the CPU thread. The contents of this
 626                 register increases monotonically. This sysfs interface
 627                 exposes the number of PURR ticks for cpuX.
 628
 629 What:           /sys/devices/system/cpu/cpuX/spurr
 630 Date:           Dec 2006
 631 Contact:        Linux for PowerPC mailing list <linuxppc-dev@lists.ozlabs.org>
 632 Description:    SPURR ticks for this CPU since the system boot.
 633
 634                 The Scaled Processor Utilization Resources Register
 635                 (SPURR) is a 64-bit counter that provides a frequency
 636                 invariant estimate of the resources used by the CPU
 637                 thread. The contents of this register increases
 638                 monotonically. This sysfs interface exposes the number
 639                 of SPURR ticks for cpuX.
 640
 641 What:           /sys/devices/system/cpu/cpuX/idle_purr
 642 Date:           Apr 2020
 643 Contact:        Linux for PowerPC mailing list <linuxppc-dev@lists.ozlabs.org>
 644 Description:    PURR ticks for cpuX when it was idle.
 645
 646                 This sysfs interface exposes the number of PURR ticks
 647                 for cpuX when it was idle.
 648
 649 What:           /sys/devices/system/cpu/cpuX/idle_spurr
 650 Date:           Apr 2020
 651 Contact:        Linux for PowerPC mailing list <linuxppc-dev@lists.ozlabs.org>
 652 Description:    SPURR ticks for cpuX when it was idle.
 653
 654                 This sysfs interface exposes the number of SPURR ticks
 655                 for cpuX when it was idle.
 656
 657 What:           /sys/devices/system/cpu/cpuX/mte_tcf_preferred
 658 Date:           July 2021
 659 Contact:        Linux ARM Kernel Mailing list <linux-arm-kernel@lists.infradead.org>
 660 Description:    Preferred MTE tag checking mode
 661
 662                 When a user program specifies more than one MTE tag checking
 663                 mode, this sysfs node is used to specify which mode should
 664                 be preferred when scheduling a task on that CPU. Possible
 665                 values:
 666
 667                 ================  ==============================================
 668                 "sync"            Prefer synchronous mode
 669                 "asymm"           Prefer asymmetric mode
 670                 "async"           Prefer asynchronous mode
 671                 ================  ==============================================
 672
 673                 See also: Documentation/arch/arm64/memory-tagging-extension.rst
 674
 675 What:           /sys/devices/system/cpu/nohz_full
 676 Date:           Apr 2015
 677 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 678 Description:
 679                 (RO) the list of CPUs that are in nohz_full mode.
 680                 These CPUs are set by boot parameter "nohz_full=".
 681
 682 What:           /sys/devices/system/cpu/isolated
 683 Date:           Apr 2015
 684 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 685 Description:
 686                 (RO) the list of CPUs that are isolated and don't
 687                 participate in load balancing. These CPUs are set by
 688                 boot parameter "isolcpus=".
 689
 690 What:           /sys/devices/system/cpu/crash_hotplug
 691 Date:           Aug 2023
 692 Contact:        Linux kernel mailing list <linux-kernel@vger.kernel.org>
 693 Description:
 694                 (RO) indicates whether or not the kernel directly supports
 695                 modifying the crash elfcorehdr for CPU hot un/plug and/or
 696                 on/offline changes.