1 accept_memory= [MM] 2 Format: { eager | lazy } 3 default: lazy 4 By default, unaccepted memory is accepted lazily to 5 avoid prolonged boot times. The lazy option will add 6 some runtime overhead until all memory is eventually 7 accepted. In most cases the overhead is negligible. 8 For some workloads or for debugging purposes 9 accept_memory=eager can be used to accept all memory 10 at once during boot. 11 12 acpi= [HW,ACPI,X86,ARM64,RISCV64,EARLY] 13 Advanced Configuration and Power Interface 14 Format: { force | on | off | strict | noirq | rsdt | 15 copy_dsdt } 16 force -- enable ACPI if default was off 17 on -- enable ACPI but allow fallback to DT [arm64,riscv64] 18 off -- disable ACPI if default was on 19 noirq -- do not use ACPI for IRQ routing 20 strict -- Be less tolerant of platforms that are not 21 strictly ACPI specification compliant. 22 rsdt -- prefer RSDT over (default) XSDT 23 copy_dsdt -- copy DSDT to memory 24 For ARM64 and RISCV64, ONLY "acpi=off", "acpi=on" or 25 "acpi=force" are available 26 27 See also Documentation/power/runtime_pm.rst, pci=noacpi 28 29 acpi_apic_instance= [ACPI,IOAPIC,EARLY] 30 Format: <int> 31 2: use 2nd APIC table, if available 32 1,0: use 1st APIC table 33 default: 0 34 35 acpi_backlight= [HW,ACPI] 36 { vendor | video | native | none } 37 If set to vendor, prefer vendor-specific driver 38 (e.g. thinkpad_acpi, sony_acpi, etc.) instead 39 of the ACPI video.ko driver. 40 If set to video, use the ACPI video.ko driver. 41 If set to native, use the device's native backlight mode. 42 If set to none, disable the ACPI backlight interface. 43 44 acpi_force_32bit_fadt_addr [ACPI,EARLY] 45 force FADT to use 32 bit addresses rather than the 46 64 bit X_* addresses. Some firmware have broken 64 47 bit addresses for force ACPI ignore these and use 48 the older legacy 32 bit addresses. 49 50 acpica_no_return_repair [HW, ACPI] 51 Disable AML predefined validation mechanism 52 This mechanism can repair the evaluation result to make 53 the return objects more ACPI specification compliant. 54 This option is useful for developers to identify the 55 root cause of an AML interpreter issue when the issue 56 has something to do with the repair mechanism. 57 58 acpi.debug_layer= [HW,ACPI,ACPI_DEBUG] 59 acpi.debug_level= [HW,ACPI,ACPI_DEBUG] 60 Format: <int> 61 CONFIG_ACPI_DEBUG must be enabled to produce any ACPI 62 debug output. Bits in debug_layer correspond to a 63 _COMPONENT in an ACPI source file, e.g., 64 #define _COMPONENT ACPI_EVENTS 65 Bits in debug_level correspond to a level in 66 ACPI_DEBUG_PRINT statements, e.g., 67 ACPI_DEBUG_PRINT((ACPI_DB_INFO, ... 68 The debug_level mask defaults to "info". See 69 Documentation/firmware-guide/acpi/debug.rst for more information about 70 debug layers and levels. 71 72 Enable processor driver info messages: 73 acpi.debug_layer=0x20000000 74 Enable AML "Debug" output, i.e., stores to the Debug 75 object while interpreting AML: 76 acpi.debug_layer=0xffffffff acpi.debug_level=0x2 77 Enable all messages related to ACPI hardware: 78 acpi.debug_layer=0x2 acpi.debug_level=0xffffffff 79 80 Some values produce so much output that the system is 81 unusable. The "log_buf_len" parameter may be useful 82 if you need to capture more output. 83 84 acpi_enforce_resources= [ACPI] 85 { strict | lax | no } 86 Check for resource conflicts between native drivers 87 and ACPI OperationRegions (SystemIO and SystemMemory 88 only). IO ports and memory declared in ACPI might be 89 used by the ACPI subsystem in arbitrary AML code and 90 can interfere with legacy drivers. 91 strict (default): access to resources claimed by ACPI 92 is denied; legacy drivers trying to access reserved 93 resources will fail to bind to device using them. 94 lax: access to resources claimed by ACPI is allowed; 95 legacy drivers trying to access reserved resources 96 will bind successfully but a warning message is logged. 97 no: ACPI OperationRegions are not marked as reserved, 98 no further checks are performed. 99 100 acpi_force_table_verification [HW,ACPI,EARLY] 101 Enable table checksum verification during early stage. 102 By default, this is disabled due to x86 early mapping 103 size limitation. 104 105 acpi_irq_balance [HW,ACPI] 106 ACPI will balance active IRQs 107 default in APIC mode 108 109 acpi_irq_nobalance [HW,ACPI] 110 ACPI will not move active IRQs (default) 111 default in PIC mode 112 113 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA 114 Format: <irq>,<irq>... 115 116 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for 117 use by PCI 118 Format: <irq>,<irq>... 119 120 acpi_mask_gpe= [HW,ACPI] 121 Due to the existence of _Lxx/_Exx, some GPEs triggered 122 by unsupported hardware/firmware features can result in 123 GPE floodings that cannot be automatically disabled by 124 the GPE dispatcher. 125 This facility can be used to prevent such uncontrolled 126 GPE floodings. 127 Format: <byte> or <bitmap-list> 128 129 acpi_no_auto_serialize [HW,ACPI] 130 Disable auto-serialization of AML methods 131 AML control methods that contain the opcodes to create 132 named objects will be marked as "Serialized" by the 133 auto-serialization feature. 134 This feature is enabled by default. 135 This option allows to turn off the feature. 136 137 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump 138 kernels. 139 140 acpi_no_static_ssdt [HW,ACPI,EARLY] 141 Disable installation of static SSDTs at early boot time 142 By default, SSDTs contained in the RSDT/XSDT will be 143 installed automatically and they will appear under 144 /sys/firmware/acpi/tables. 145 This option turns off this feature. 146 Note that specifying this option does not affect 147 dynamic table installation which will install SSDT 148 tables to /sys/firmware/acpi/tables/dynamic. 149 150 acpi_no_watchdog [HW,ACPI,WDT] 151 Ignore the ACPI-based watchdog interface (WDAT) and let 152 a native driver control the watchdog device instead. 153 154 acpi_rsdp= [ACPI,EFI,KEXEC,EARLY] 155 Pass the RSDP address to the kernel, mostly used 156 on machines running EFI runtime service to boot the 157 second kernel for kdump. 158 159 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS 160 Format: To spoof as Windows 98: ="Microsoft Windows" 161 162 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead 163 of 2 which is mandated by ACPI 6) as the supported ACPI 164 specification revision (when using this switch, it may 165 be necessary to carry out a cold reboot _twice_ in a 166 row to make it take effect on the platform firmware). 167 168 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings 169 acpi_osi="string1" # add string1 170 acpi_osi="!string2" # remove string2 171 acpi_osi=!* # remove all strings 172 acpi_osi=! # disable all built-in OS vendor 173 strings 174 acpi_osi=!! # enable all built-in OS vendor 175 strings 176 acpi_osi= # disable all strings 177 178 'acpi_osi=!' can be used in combination with single or 179 multiple 'acpi_osi="string1"' to support specific OS 180 vendor string(s). Note that such command can only 181 affect the default state of the OS vendor strings, thus 182 it cannot affect the default state of the feature group 183 strings and the current state of the OS vendor strings, 184 specifying it multiple times through kernel command line 185 is meaningless. This command is useful when one do not 186 care about the state of the feature group strings which 187 should be controlled by the OSPM. 188 Examples: 189 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent 190 to 'acpi_osi="Windows 2000" acpi_osi=!', they all 191 can make '_OSI("Windows 2000")' TRUE. 192 193 'acpi_osi=' cannot be used in combination with other 194 'acpi_osi=' command lines, the _OSI method will not 195 exist in the ACPI namespace. NOTE that such command can 196 only affect the _OSI support state, thus specifying it 197 multiple times through kernel command line is also 198 meaningless. 199 Examples: 200 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)' 201 FALSE. 202 203 'acpi_osi=!*' can be used in combination with single or 204 multiple 'acpi_osi="string1"' to support specific 205 string(s). Note that such command can affect the 206 current state of both the OS vendor strings and the 207 feature group strings, thus specifying it multiple times 208 through kernel command line is meaningful. But it may 209 still not able to affect the final state of a string if 210 there are quirks related to this string. This command 211 is useful when one want to control the state of the 212 feature group strings to debug BIOS issues related to 213 the OSPM features. 214 Examples: 215 1. 'acpi_osi="Module Device" acpi_osi=!*' can make 216 '_OSI("Module Device")' FALSE. 217 2. 'acpi_osi=!* acpi_osi="Module Device"' can make 218 '_OSI("Module Device")' TRUE. 219 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is 220 equivalent to 221 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"' 222 and 223 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!', 224 they all will make '_OSI("Windows 2000")' TRUE. 225 226 acpi_pm_good [X86] 227 Override the pmtimer bug detection: force the kernel 228 to assume that this machine's pmtimer latches its value 229 and always returns good values. 230 231 acpi_sci= [HW,ACPI,EARLY] ACPI System Control Interrupt trigger mode 232 Format: { level | edge | high | low } 233 234 acpi_skip_timer_override [HW,ACPI,EARLY] 235 Recognize and ignore IRQ0/pin2 Interrupt Override. 236 For broken nForce2 BIOS resulting in XT-PIC timer. 237 238 acpi_sleep= [HW,ACPI] Sleep options 239 Format: { s3_bios, s3_mode, s3_beep, s4_hwsig, 240 s4_nohwsig, old_ordering, nonvs, 241 sci_force_enable, nobl } 242 See Documentation/power/video.rst for information on 243 s3_bios and s3_mode. 244 s3_beep is for debugging; it makes the PC's speaker beep 245 as soon as the kernel's real-mode entry point is called. 246 s4_hwsig causes the kernel to check the ACPI hardware 247 signature during resume from hibernation, and gracefully 248 refuse to resume if it has changed. This complies with 249 the ACPI specification but not with reality, since 250 Windows does not do this and many laptops do change it 251 on docking. So the default behaviour is to allow resume 252 and simply warn when the signature changes, unless the 253 s4_hwsig option is enabled. 254 s4_nohwsig prevents ACPI hardware signature from being 255 used (or even warned about) during resume. 256 old_ordering causes the ACPI 1.0 ordering of the _PTS 257 control method, with respect to putting devices into 258 low power states, to be enforced (the ACPI 2.0 ordering 259 of _PTS is used by default). 260 nonvs prevents the kernel from saving/restoring the 261 ACPI NVS memory during suspend/hibernation and resume. 262 sci_force_enable causes the kernel to set SCI_EN directly 263 on resume from S1/S3 (which is against the ACPI spec, 264 but some broken systems don't work without it). 265 nobl causes the internal blacklist of systems known to 266 behave incorrectly in some ways with respect to system 267 suspend and resume to be ignored (use wisely). 268 269 acpi_use_timer_override [HW,ACPI,EARLY] 270 Use timer override. For some broken Nvidia NF5 boards 271 that require a timer override, but don't have HPET 272 273 add_efi_memmap [EFI,X86,EARLY] Include EFI memory map in 274 kernel's map of available physical RAM. 275 276 agp= [AGP] 277 { off | try_unsupported } 278 off: disable AGP support 279 try_unsupported: try to drive unsupported chipsets 280 (may crash computer or cause data corruption) 281 282 ALSA [HW,ALSA] 283 See Documentation/sound/alsa-configuration.rst 284 285 alignment= [KNL,ARM] 286 Allow the default userspace alignment fault handler 287 behaviour to be specified. Bit 0 enables warnings, 288 bit 1 enables fixups, and bit 2 sends a segfault. 289 290 align_va_addr= [X86-64] 291 Align virtual addresses by clearing slice [14:12] when 292 allocating a VMA at process creation time. This option 293 gives you up to 3% performance improvement on AMD F15h 294 machines (where it is enabled by default) for a 295 CPU-intensive style benchmark, and it can vary highly in 296 a microbenchmark depending on workload and compiler. 297 298 32: only for 32-bit processes 299 64: only for 64-bit processes 300 on: enable for both 32- and 64-bit processes 301 off: disable for both 32- and 64-bit processes 302 303 alloc_snapshot [FTRACE] 304 Allocate the ftrace snapshot buffer on boot up when the 305 main buffer is allocated. This is handy if debugging 306 and you need to use tracing_snapshot() on boot up, and 307 do not want to use tracing_snapshot_alloc() as it needs 308 to be done where GFP_KERNEL allocations are allowed. 309 310 allow_mismatched_32bit_el0 [ARM64,EARLY] 311 Allow execve() of 32-bit applications and setting of the 312 PER_LINUX32 personality on systems where only a strict 313 subset of the CPUs support 32-bit EL0. When this 314 parameter is present, the set of CPUs supporting 32-bit 315 EL0 is indicated by /sys/devices/system/cpu/aarch32_el0 316 and hot-unplug operations may be restricted. 317 318 See Documentation/arch/arm64/asymmetric-32bit.rst for more 319 information. 320 321 amd_iommu= [HW,X86-64] 322 Pass parameters to the AMD IOMMU driver in the system. 323 Possible values are: 324 fullflush - Deprecated, equivalent to iommu.strict=1 325 off - do not initialize any AMD IOMMU found in 326 the system 327 force_isolation - Force device isolation for all 328 devices. The IOMMU driver is not 329 allowed anymore to lift isolation 330 requirements as needed. This option 331 does not override iommu=pt 332 force_enable - Force enable the IOMMU on platforms known 333 to be buggy with IOMMU enabled. Use this 334 option with care. 335 pgtbl_v1 - Use v1 page table for DMA-API (Default). 336 pgtbl_v2 - Use v2 page table for DMA-API. 337 irtcachedis - Disable Interrupt Remapping Table (IRT) caching. 338 339 amd_iommu_dump= [HW,X86-64] 340 Enable AMD IOMMU driver option to dump the ACPI table 341 for AMD IOMMU. With this option enabled, AMD IOMMU 342 driver will print ACPI tables for AMD IOMMU during 343 IOMMU initialization. 344 345 amd_iommu_intr= [HW,X86-64] 346 Specifies one of the following AMD IOMMU interrupt 347 remapping modes: 348 legacy - Use legacy interrupt remapping mode. 349 vapic - Use virtual APIC mode, which allows IOMMU 350 to inject interrupts directly into guest. 351 This mode requires kvm-amd.avic=1. 352 (Default when IOMMU HW support is present.) 353 354 amd_pstate= [X86,EARLY] 355 disable 356 Do not enable amd_pstate as the default 357 scaling driver for the supported processors 358 passive 359 Use amd_pstate with passive mode as a scaling driver. 360 In this mode autonomous selection is disabled. 361 Driver requests a desired performance level and platform 362 tries to match the same performance level if it is 363 satisfied by guaranteed performance level. 364 active 365 Use amd_pstate_epp driver instance as the scaling driver, 366 driver provides a hint to the hardware if software wants 367 to bias toward performance (0x0) or energy efficiency (0xff) 368 to the CPPC firmware. then CPPC power algorithm will 369 calculate the runtime workload and adjust the realtime cores 370 frequency. 371 guided 372 Activate guided autonomous mode. Driver requests minimum and 373 maximum performance level and the platform autonomously 374 selects a performance level in this range and appropriate 375 to the current workload. 376 377 amd_prefcore= 378 [X86] 379 disable 380 Disable amd-pstate preferred core. 381 382 amijoy.map= [HW,JOY] Amiga joystick support 383 Map of devices attached to JOY0DAT and JOY1DAT 384 Format: <a>,<b> 385 See also Documentation/input/joydev/joystick.rst 386 387 analog.map= [HW,JOY] Analog joystick and gamepad support 388 Specifies type or capabilities of an analog joystick 389 connected to one of 16 gameports 390 Format: <type1>,<type2>,..<type16> 391 392 apc= [HW,SPARC] 393 Power management functions (SPARCstation-4/5 + deriv.) 394 Format: noidle 395 Disable APC CPU standby support. SPARCstation-Fox does 396 not play well with APC CPU idle - disable it if you have 397 APC and your system crashes randomly. 398 399 apic= [APIC,X86,EARLY] Advanced Programmable Interrupt Controller 400 Change the output verbosity while booting 401 Format: { quiet (default) | verbose | debug } 402 Change the amount of debugging information output 403 when initialising the APIC and IO-APIC components. 404 For X86-32, this can also be used to specify an APIC 405 driver name. 406 Format: apic=driver_name 407 Examples: apic=bigsmp 408 409 apic_extnmi= [APIC,X86,EARLY] External NMI delivery setting 410 Format: { bsp (default) | all | none } 411 bsp: External NMI is delivered only to CPU 0 412 all: External NMIs are broadcast to all CPUs as a 413 backup of CPU 0 414 none: External NMI is masked for all CPUs. This is 415 useful so that a dump capture kernel won't be 416 shot down by NMI 417 418 autoconf= [IPV6] 419 See Documentation/networking/ipv6.rst. 420 421 apm= [APM] Advanced Power Management 422 See header of arch/x86/kernel/apm_32.c. 423 424 apparmor= [APPARMOR] Disable or enable AppArmor at boot time 425 Format: { "0" | "1" } 426 See security/apparmor/Kconfig help text 427 0 -- disable. 428 1 -- enable. 429 Default value is set via kernel config option. 430 431 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards 432 Format: <io>,<irq>,<nodeID> 433 434 arm64.no32bit_el0 [ARM64] Unconditionally disable the execution of 435 32 bit applications. 436 437 arm64.nobti [ARM64] Unconditionally disable Branch Target 438 Identification support 439 440 arm64.nomops [ARM64] Unconditionally disable Memory Copy and Memory 441 Set instructions support 442 443 arm64.nomte [ARM64] Unconditionally disable Memory Tagging Extension 444 support 445 446 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication 447 support 448 449 arm64.nosme [ARM64] Unconditionally disable Scalable Matrix 450 Extension support 451 452 arm64.nosve [ARM64] Unconditionally disable Scalable Vector 453 Extension support 454 455 ataflop= [HW,M68k] 456 457 atarimouse= [HW,MOUSE] Atari Mouse 458 459 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess, 460 EzKey and similar keyboards 461 462 atkbd.reset= [HW] Reset keyboard during initialization 463 464 atkbd.set= [HW] Select keyboard code set 465 Format: <int> (2 = AT (default), 3 = PS/2) 466 467 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar 468 keyboards 469 470 atkbd.softraw= [HW] Choose between synthetic and real raw mode 471 Format: <bool> (0 = real, 1 = synthetic (default)) 472 473 atkbd.softrepeat= [HW] 474 Use software keyboard repeat 475 476 audit= [KNL] Enable the audit sub-system 477 Format: { "0" | "1" | "off" | "on" } 478 0 | off - kernel audit is disabled and can not be 479 enabled until the next reboot 480 unset - kernel audit is initialized but disabled and 481 will be fully enabled by the userspace auditd. 482 1 | on - kernel audit is initialized and partially 483 enabled, storing at most audit_backlog_limit 484 messages in RAM until it is fully enabled by the 485 userspace auditd. 486 Default: unset 487 488 audit_backlog_limit= [KNL] Set the audit queue size limit. 489 Format: <int> (must be >=0) 490 Default: 64 491 492 bau= [X86_UV] Enable the BAU on SGI UV. The default 493 behavior is to disable the BAU (i.e. bau=0). 494 Format: { "0" | "1" } 495 0 - Disable the BAU. 496 1 - Enable the BAU. 497 unset - Disable the BAU. 498 499 baycom_epp= [HW,AX25] 500 Format: <io>,<mode> 501 502 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem 503 Format: <io>,<mode> 504 See header of drivers/net/hamradio/baycom_par.c. 505 506 baycom_ser_fdx= [HW,AX25] 507 BayCom Serial Port AX.25 Modem (Full Duplex Mode) 508 Format: <io>,<irq>,<mode>[,<baud>] 509 See header of drivers/net/hamradio/baycom_ser_fdx.c. 510 511 baycom_ser_hdx= [HW,AX25] 512 BayCom Serial Port AX.25 Modem (Half Duplex Mode) 513 Format: <io>,<irq>,<mode> 514 See header of drivers/net/hamradio/baycom_ser_hdx.c. 515 516 bert_disable [ACPI] 517 Disable BERT OS support on buggy BIOSes. 518 519 bgrt_disable [ACPI,X86,EARLY] 520 Disable BGRT to avoid flickering OEM logo. 521 522 blkdevparts= Manual partition parsing of block device(s) for 523 embedded devices based on command line input. 524 See Documentation/block/cmdline-partition.rst 525 526 boot_delay= [KNL,EARLY] 527 Milliseconds to delay each printk during boot. 528 Only works if CONFIG_BOOT_PRINTK_DELAY is enabled, 529 and you may also have to specify "lpj=". Boot_delay 530 values larger than 10 seconds (10000) are assumed 531 erroneous and ignored. 532 Format: integer 533 534 bootconfig [KNL,EARLY] 535 Extended command line options can be added to an initrd 536 and this will cause the kernel to look for it. 537 538 See Documentation/admin-guide/bootconfig.rst 539 540 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards) 541 bttv.radio= Most important insmod options are available as 542 kernel args too. 543 bttv.pll= See Documentation/admin-guide/media/bttv.rst 544 bttv.tuner= 545 546 bulk_remove=off [PPC] This parameter disables the use of the pSeries 547 firmware feature for flushing multiple hpte entries 548 at a time. 549 550 c101= [NET] Moxa C101 synchronous serial card 551 552 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection. 553 Sometimes CPU hardware bugs make them report the cache 554 size incorrectly. The kernel will attempt work arounds 555 to fix known problems, but for some CPUs it is not 556 possible to determine what the correct size should be. 557 This option provides an override for these situations. 558 559 carrier_timeout= 560 [NET] Specifies amount of time (in seconds) that 561 the kernel should wait for a network carrier. By default 562 it waits 120 seconds. 563 564 ca_keys= [KEYS] This parameter identifies a specific key(s) on 565 the system trusted keyring to be used for certificate 566 trust validation. 567 format: { id:<keyid> | builtin } 568 569 cca= [MIPS,EARLY] Override the kernel pages' cache coherency 570 algorithm. Accepted values range from 0 to 7 571 inclusive. See arch/mips/include/asm/pgtable-bits.h 572 for platform specific values (SB1, Loongson3 and 573 others). 574 575 ccw_timeout_log [S390] 576 See Documentation/arch/s390/common_io.rst for details. 577 578 cgroup_disable= [KNL] Disable a particular controller or optional feature 579 Format: {name of the controller(s) or feature(s) to disable} 580 The effects of cgroup_disable=foo are: 581 - foo isn't auto-mounted if you mount all cgroups in 582 a single hierarchy 583 - foo isn't visible as an individually mountable 584 subsystem 585 - if foo is an optional feature then the feature is 586 disabled and corresponding cgroup files are not 587 created 588 {Currently only "memory" controller deal with this and 589 cut the overhead, others just disable the usage. So 590 only cgroup_disable=memory is actually worthy} 591 Specifying "pressure" disables per-cgroup pressure 592 stall information accounting feature 593 594 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1 595 Format: { { controller | "all" | "named" } 596 [,{ controller | "all" | "named" }...] } 597 Like cgroup_disable, but only applies to cgroup v1; 598 the blacklisted controllers remain available in cgroup2. 599 "all" blacklists all controllers and "named" disables 600 named mounts. Specifying both "all" and "named" disables 601 all v1 hierarchies. 602 603 cgroup_favordynmods= [KNL] Enable or Disable favordynmods. 604 Format: { "true" | "false" } 605 Defaults to the value of CONFIG_CGROUP_FAVOR_DYNMODS. 606 607 cgroup.memory= [KNL] Pass options to the cgroup memory controller. 608 Format: <string> 609 nosocket -- Disable socket memory accounting. 610 nokmem -- Disable kernel memory accounting. 611 nobpf -- Disable BPF memory accounting. 612 613 checkreqprot= [SELINUX] Set initial checkreqprot flag value. 614 Format: { "0" | "1" } 615 See security/selinux/Kconfig help text. 616 0 -- check protection applied by kernel (includes 617 any implied execute protection). 618 1 -- check protection requested by application. 619 Default value is set via a kernel config option. 620 Value can be changed at runtime via 621 /sys/fs/selinux/checkreqprot. 622 Setting checkreqprot to 1 is deprecated. 623 624 cio_ignore= [S390] 625 See Documentation/arch/s390/common_io.rst for details. 626 627 clearcpuid=X[,X...] [X86] 628 Disable CPUID feature X for the kernel. See 629 arch/x86/include/asm/cpufeatures.h for the valid bit 630 numbers X. Note the Linux-specific bits are not necessarily 631 stable over kernel options, but the vendor-specific 632 ones should be. 633 X can also be a string as appearing in the flags: line 634 in /proc/cpuinfo which does not have the above 635 instability issue. However, not all features have names 636 in /proc/cpuinfo. 637 Note that using this option will taint your kernel. 638 Also note that user programs calling CPUID directly 639 or using the feature without checking anything 640 will still see it. This just prevents it from 641 being used by the kernel or shown in /proc/cpuinfo. 642 Also note the kernel might malfunction if you disable 643 some critical bits. 644 645 clk_ignore_unused 646 [CLK] 647 Prevents the clock framework from automatically gating 648 clocks that have not been explicitly enabled by a Linux 649 device driver but are enabled in hardware at reset or 650 by the bootloader/firmware. Note that this does not 651 force such clocks to be always-on nor does it reserve 652 those clocks in any way. This parameter is useful for 653 debug and development, but should not be needed on a 654 platform with proper driver support. For more 655 information, see Documentation/driver-api/clk.rst. 656 657 clock= [BUGS=X86-32, HW] gettimeofday clocksource override. 658 [Deprecated] 659 Forces specified clocksource (if available) to be used 660 when calculating gettimeofday(). If specified 661 clocksource is not available, it defaults to PIT. 662 Format: { pit | tsc | cyclone | pmtmr } 663 664 clocksource= Override the default clocksource 665 Format: <string> 666 Override the default clocksource and use the clocksource 667 with the name specified. 668 Some clocksource names to choose from, depending on 669 the platform: 670 [all] jiffies (this is the base, fallback clocksource) 671 [ACPI] acpi_pm 672 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2, 673 pxa_timer,timer3,32k_counter,timer0_1 674 [X86-32] pit,hpet,tsc; 675 scx200_hrt on Geode; cyclone on IBM x440 676 [MIPS] MIPS 677 [PARISC] cr16 678 [S390] tod 679 [SH] SuperH 680 [SPARC64] tick 681 [X86-64] hpet,tsc 682 683 clocksource.arm_arch_timer.evtstrm= 684 [ARM,ARM64,EARLY] 685 Format: <bool> 686 Enable/disable the eventstream feature of the ARM 687 architected timer so that code using WFE-based polling 688 loops can be debugged more effectively on production 689 systems. 690 691 clocksource.verify_n_cpus= [KNL] 692 Limit the number of CPUs checked for clocksources 693 marked with CLOCK_SOURCE_VERIFY_PERCPU that 694 are marked unstable due to excessive skew. 695 A negative value says to check all CPUs, while 696 zero says not to check any. Values larger than 697 nr_cpu_ids are silently truncated to nr_cpu_ids. 698 The actual CPUs are chosen randomly, with 699 no replacement if the same CPU is chosen twice. 700 701 clocksource-wdtest.holdoff= [KNL] 702 Set the time in seconds that the clocksource 703 watchdog test waits before commencing its tests. 704 Defaults to zero when built as a module and to 705 10 seconds when built into the kernel. 706 707 cma=nn[MG]@[start[MG][-end[MG]]] 708 [KNL,CMA,EARLY] 709 Sets the size of kernel global memory area for 710 contiguous memory allocations and optionally the 711 placement constraint by the physical address range of 712 memory allocations. A value of 0 disables CMA 713 altogether. For more information, see 714 kernel/dma/contiguous.c 715 716 cma_pernuma=nn[MG] 717 [KNL,CMA,EARLY] 718 Sets the size of kernel per-numa memory area for 719 contiguous memory allocations. A value of 0 disables 720 per-numa CMA altogether. And If this option is not 721 specified, the default value is 0. 722 With per-numa CMA enabled, DMA users on node nid will 723 first try to allocate buffer from the pernuma area 724 which is located in node nid, if the allocation fails, 725 they will fallback to the global default memory area. 726 727 numa_cma=<node>:nn[MG][,<node>:nn[MG]] 728 [KNL,CMA,EARLY] 729 Sets the size of kernel numa memory area for 730 contiguous memory allocations. It will reserve CMA 731 area for the specified node. 732 733 With numa CMA enabled, DMA users on node nid will 734 first try to allocate buffer from the numa area 735 which is located in node nid, if the allocation fails, 736 they will fallback to the global default memory area. 737 738 cmo_free_hint= [PPC] Format: { yes | no } 739 Specify whether pages are marked as being inactive 740 when they are freed. This is used in CMO environments 741 to determine OS memory pressure for page stealing by 742 a hypervisor. 743 Default: yes 744 745 coherent_pool=nn[KMG] [ARM,KNL,EARLY] 746 Sets the size of memory pool for coherent, atomic dma 747 allocations, by default set to 256K. 748 749 com20020= [HW,NET] ARCnet - COM20020 chipset 750 Format: 751 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]] 752 753 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers) 754 Format: <io>[,<irq>] 755 756 com90xx= [HW,NET] 757 ARCnet - COM90xx chipset (memory-mapped buffers) 758 Format: <io>[,<irq>[,<memstart>]] 759 760 condev= [HW,S390] console device 761 conmode= 762 763 con3215_drop= [S390,EARLY] 3215 console drop mode. 764 Format: y|n|Y|N|1|0 765 When set to true, drop data on the 3215 console when 766 the console buffer is full. In this case the 767 operator using a 3270 terminal emulator (for example 768 x3270) does not have to enter the clear key for the 769 console output to advance and the kernel to continue. 770 This leads to a much faster boot time when a 3270 771 terminal emulator is active. If no 3270 terminal 772 emulator is used, this parameter has no effect. 773 774 console= [KNL] Output console device and options. 775 776 tty<n> Use the virtual console device <n>. 777 778 ttyS<n>[,options] 779 ttyUSB0[,options] 780 Use the specified serial port. The options are of 781 the form "bbbbpnf", where "bbbb" is the baud rate, 782 "p" is parity ("n", "o", or "e"), "n" is number of 783 bits, and "f" is flow control ("r" for RTS or 784 omit it). Default is "9600n8". 785 786 See Documentation/admin-guide/serial-console.rst for more 787 information. See 788 Documentation/networking/netconsole.rst for an 789 alternative. 790 791 <DEVNAME>:<n>.<n>[,options] 792 Use the specified serial port on the serial core bus. 793 The addressing uses DEVNAME of the physical serial port 794 device, followed by the serial core controller instance, 795 and the serial port instance. The options are the same 796 as documented for the ttyS addressing above. 797 798 The mapping of the serial ports to the tty instances 799 can be viewed with: 800 801 $ ls -d /sys/bus/serial-base/devices/*:*.*/tty/* 802 /sys/bus/serial-base/devices/00:04:0.0/tty/ttyS0 803 804 In the above example, the console can be addressed with 805 console=00:04:0.0. Note that a console addressed this 806 way will only get added when the related device driver 807 is ready. The use of an earlycon parameter in addition to 808 the console may be desired for console output early on. 809 810 uart[8250],io,<addr>[,options] 811 uart[8250],mmio,<addr>[,options] 812 uart[8250],mmio16,<addr>[,options] 813 uart[8250],mmio32,<addr>[,options] 814 uart[8250],0x<addr>[,options] 815 Start an early, polled-mode console on the 8250/16550 816 UART at the specified I/O port or MMIO address, 817 switching to the matching ttyS device later. 818 MMIO inter-register address stride is either 8-bit 819 (mmio), 16-bit (mmio16), or 32-bit (mmio32). 820 If none of [io|mmio|mmio16|mmio32], <addr> is assumed 821 to be equivalent to 'mmio'. 'options' are specified in 822 the same format described for ttyS above; if unspecified, 823 the h/w is not re-initialized. 824 825 hvc<n> Use the hypervisor console device <n>. This is for 826 both Xen and PowerPC hypervisors. 827 828 { null | "" } 829 Use to disable console output, i.e., to have kernel 830 console messages discarded. 831 This must be the only console= parameter used on the 832 kernel command line. 833 834 If the device connected to the port is not a TTY but a braille 835 device, prepend "brl," before the device type, for instance 836 console=brl,ttyS0 837 For now, only VisioBraille is supported. 838 839 console_msg_format= 840 [KNL] Change console messages format 841 default 842 By default we print messages on consoles in 843 "[time stamp] text\n" format (time stamp may not be 844 printed, depending on CONFIG_PRINTK_TIME or 845 `printk_time' param). 846 syslog 847 Switch to syslog format: "<%u>[time stamp] text\n" 848 IOW, each message will have a facility and loglevel 849 prefix. The format is similar to one used by syslog() 850 syscall, or to executing "dmesg -S --raw" or to reading 851 from /proc/kmsg. 852 853 consoleblank= [KNL] The console blank (screen saver) timeout in 854 seconds. A value of 0 disables the blank timer. 855 Defaults to 0. 856 857 coredump_filter= 858 [KNL] Change the default value for 859 /proc/<pid>/coredump_filter. 860 See also Documentation/filesystems/proc.rst. 861 862 coresight_cpu_debug.enable 863 [ARM,ARM64] 864 Format: <bool> 865 Enable/disable the CPU sampling based debugging. 866 0: default value, disable debugging 867 1: enable debugging at boot time 868 869 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver 870 Format: 871 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>] 872 873 cpuidle.off=1 [CPU_IDLE] 874 disable the cpuidle sub-system 875 876 cpuidle.governor= 877 [CPU_IDLE] Name of the cpuidle governor to use. 878 879 cpufreq.off=1 [CPU_FREQ] 880 disable the cpufreq sub-system 881 882 cpufreq.default_governor= 883 [CPU_FREQ] Name of the default cpufreq governor or 884 policy to use. This governor must be registered in the 885 kernel before the cpufreq driver probes. 886 887 cpu_init_udelay=N 888 [X86,EARLY] Delay for N microsec between assert and de-assert 889 of APIC INIT to start processors. This delay occurs 890 on every CPU online, such as boot, and resume from suspend. 891 Default: 10000 892 893 cpuhp.parallel= 894 [SMP] Enable/disable parallel bringup of secondary CPUs 895 Format: <bool> 896 Default is enabled if CONFIG_HOTPLUG_PARALLEL=y. Otherwise 897 the parameter has no effect. 898 899 crash_kexec_post_notifiers 900 Run kdump after running panic-notifiers and dumping 901 kmsg. This only for the users who doubt kdump always 902 succeeds in any situation. 903 Note that this also increases risks of kdump failure, 904 because some panic notifiers can make the crashed 905 kernel more unstable. 906 907 crashkernel=size[KMG][@offset[KMG]] 908 [KNL,EARLY] Using kexec, Linux can switch to a 'crash kernel' 909 upon panic. This parameter reserves the physical 910 memory region [offset, offset + size] for that kernel 911 image. If '@offset' is omitted, then a suitable offset 912 is selected automatically. 913 [KNL, X86-64, ARM64, RISCV, LoongArch] Select a region 914 under 4G first, and fall back to reserve region above 915 4G when '@offset' hasn't been specified. 916 See Documentation/admin-guide/kdump/kdump.rst for further details. 917 918 crashkernel=range1:size1[,range2:size2,...][@offset] 919 [KNL] Same as above, but depends on the memory 920 in the running system. The syntax of range is 921 start-[end] where start and end are both 922 a memory unit (amount[KMG]). See also 923 Documentation/admin-guide/kdump/kdump.rst for an example. 924 925 crashkernel=size[KMG],high 926 [KNL, X86-64, ARM64, RISCV, LoongArch] range could be 927 above 4G. 928 Allow kernel to allocate physical memory region from top, 929 so could be above 4G if system have more than 4G ram 930 installed. Otherwise memory region will be allocated 931 below 4G, if available. 932 It will be ignored if crashkernel=X is specified. 933 crashkernel=size[KMG],low 934 [KNL, X86-64, ARM64, RISCV, LoongArch] range under 4G. 935 When crashkernel=X,high is passed, kernel could allocate 936 physical memory region above 4G, that cause second kernel 937 crash on system that require some amount of low memory, 938 e.g. swiotlb requires at least 64M+32K low memory, also 939 enough extra low memory is needed to make sure DMA buffers 940 for 32-bit devices won't run out. Kernel would try to allocate 941 default size of memory below 4G automatically. The default 942 size is platform dependent. 943 --> x86: max(swiotlb_size_or_default() + 8MiB, 256MiB) 944 --> arm64: 128MiB 945 --> riscv: 128MiB 946 --> loongarch: 128MiB 947 This one lets the user specify own low range under 4G 948 for second kernel instead. 949 0: to disable low allocation. 950 It will be ignored when crashkernel=X,high is not used 951 or memory reserved is below 4G. 952 953 cryptomgr.notests 954 [KNL] Disable crypto self-tests 955 956 cs89x0_dma= [HW,NET] 957 Format: <dma> 958 959 cs89x0_media= [HW,NET] 960 Format: { rj45 | aui | bnc } 961 962 csdlock_debug= [KNL] Enable or disable debug add-ons of cross-CPU 963 function call handling. When switched on, 964 additional debug data is printed to the console 965 in case a hanging CPU is detected, and that 966 CPU is pinged again in order to try to resolve 967 the hang situation. The default value of this 968 option depends on the CSD_LOCK_WAIT_DEBUG_DEFAULT 969 Kconfig option. 970 971 dasd= [HW,NET] 972 See header of drivers/s390/block/dasd_devmap.c. 973 974 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port 975 (one device per port) 976 Format: <port#>,<type> 977 See also Documentation/input/devices/joystick-parport.rst 978 979 debug [KNL,EARLY] Enable kernel debugging (events log level). 980 981 debug_boot_weak_hash 982 [KNL,EARLY] Enable printing [hashed] pointers early in the 983 boot sequence. If enabled, we use a weak hash instead 984 of siphash to hash pointers. Use this option if you are 985 seeing instances of '(___ptrval___)') and need to see a 986 value (hashed pointer) instead. Cryptographically 987 insecure, please do not use on production kernels. 988 989 debug_locks_verbose= 990 [KNL] verbose locking self-tests 991 Format: <int> 992 Print debugging info while doing the locking API 993 self-tests. 994 Bitmask for the various LOCKTYPE_ tests. Defaults to 0 995 (no extra messages), setting it to -1 (all bits set) 996 will print _a_lot_ more information - normally only 997 useful to lockdep developers. 998 999 debug_objects [KNL,EARLY] Enable object debugging 1000 1001 debug_guardpage_minorder= 1002 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this 1003 parameter allows control of the order of pages that will 1004 be intentionally kept free (and hence protected) by the 1005 buddy allocator. Bigger value increase the probability 1006 of catching random memory corruption, but reduce the 1007 amount of memory for normal system use. The maximum 1008 possible value is MAX_PAGE_ORDER/2. Setting this 1009 parameter to 1 or 2 should be enough to identify most 1010 random memory corruption problems caused by bugs in 1011 kernel or driver code when a CPU writes to (or reads 1012 from) a random memory location. Note that there exists 1013 a class of memory corruptions problems caused by buggy 1014 H/W or F/W or by drivers badly programming DMA 1015 (basically when memory is written at bus level and the 1016 CPU MMU is bypassed) which are not detectable by 1017 CONFIG_DEBUG_PAGEALLOC, hence this option will not 1018 help tracking down these problems. 1019 1020 debug_pagealloc= 1021 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this parameter 1022 enables the feature at boot time. By default, it is 1023 disabled and the system will work mostly the same as a 1024 kernel built without CONFIG_DEBUG_PAGEALLOC. 1025 Note: to get most of debug_pagealloc error reports, it's 1026 useful to also enable the page_owner functionality. 1027 on: enable the feature 1028 1029 debugfs= [KNL,EARLY] This parameter enables what is exposed to 1030 userspace and debugfs internal clients. 1031 Format: { on, no-mount, off } 1032 on: All functions are enabled. 1033 no-mount: 1034 Filesystem is not registered but kernel clients can 1035 access APIs and a crashkernel can be used to read 1036 its content. There is nothing to mount. 1037 off: Filesystem is not registered and clients 1038 get a -EPERM as result when trying to register files 1039 or directories within debugfs. 1040 This is equivalent of the runtime functionality if 1041 debugfs was not enabled in the kernel at all. 1042 Default value is set in build-time with a kernel configuration. 1043 1044 debugpat [X86] Enable PAT debugging 1045 1046 default_hugepagesz= 1047 [HW] The size of the default HugeTLB page. This is 1048 the size represented by the legacy /proc/ hugepages 1049 APIs. In addition, this is the default hugetlb size 1050 used for shmget(), mmap() and mounting hugetlbfs 1051 filesystems. If not specified, defaults to the 1052 architecture's default huge page size. Huge page 1053 sizes are architecture dependent. See also 1054 Documentation/admin-guide/mm/hugetlbpage.rst. 1055 Format: size[KMG] 1056 1057 deferred_probe_timeout= 1058 [KNL] Debugging option to set a timeout in seconds for 1059 deferred probe to give up waiting on dependencies to 1060 probe. Only specific dependencies (subsystems or 1061 drivers) that have opted in will be ignored. A timeout 1062 of 0 will timeout at the end of initcalls. If the time 1063 out hasn't expired, it'll be restarted by each 1064 successful driver registration. This option will also 1065 dump out devices still on the deferred probe list after 1066 retrying. 1067 1068 delayacct [KNL] Enable per-task delay accounting 1069 1070 dell_smm_hwmon.ignore_dmi= 1071 [HW] Continue probing hardware even if DMI data 1072 indicates that the driver is running on unsupported 1073 hardware. 1074 1075 dell_smm_hwmon.force= 1076 [HW] Activate driver even if SMM BIOS signature does 1077 not match list of supported models and enable otherwise 1078 blacklisted features. 1079 1080 dell_smm_hwmon.power_status= 1081 [HW] Report power status in /proc/i8k 1082 (disabled by default). 1083 1084 dell_smm_hwmon.restricted= 1085 [HW] Allow controlling fans only if SYS_ADMIN 1086 capability is set. 1087 1088 dell_smm_hwmon.fan_mult= 1089 [HW] Factor to multiply fan speed with. 1090 1091 dell_smm_hwmon.fan_max= 1092 [HW] Maximum configurable fan speed. 1093 1094 dfltcc= [HW,S390] 1095 Format: { on | off | def_only | inf_only | always } 1096 on: s390 zlib hardware support for compression on 1097 level 1 and decompression (default) 1098 off: No s390 zlib hardware support 1099 def_only: s390 zlib hardware support for deflate 1100 only (compression on level 1) 1101 inf_only: s390 zlib hardware support for inflate 1102 only (decompression) 1103 always: Same as 'on' but ignores the selected compression 1104 level always using hardware support (used for debugging) 1105 1106 dhash_entries= [KNL] 1107 Set number of hash buckets for dentry cache. 1108 1109 disable_1tb_segments [PPC,EARLY] 1110 Disables the use of 1TB hash page table segments. This 1111 causes the kernel to fall back to 256MB segments which 1112 can be useful when debugging issues that require an SLB 1113 miss to occur. 1114 1115 disable= [IPV6] 1116 See Documentation/networking/ipv6.rst. 1117 1118 disable_radix [PPC,EARLY] 1119 Disable RADIX MMU mode on POWER9 1120 1121 disable_tlbie [PPC] 1122 Disable TLBIE instruction. Currently does not work 1123 with KVM, with HASH MMU, or with coherent accelerators. 1124 1125 disable_ddw [PPC/PSERIES,EARLY] 1126 Disable Dynamic DMA Window support. Use this 1127 to workaround buggy firmware. 1128 1129 disable_ipv6= [IPV6] 1130 See Documentation/networking/ipv6.rst. 1131 1132 disable_mtrr_cleanup [X86,EARLY] 1133 The kernel tries to adjust MTRR layout from continuous 1134 to discrete, to make X server driver able to add WB 1135 entry later. This parameter disables that. 1136 1137 disable_mtrr_trim [X86, Intel and AMD only,EARLY] 1138 By default the kernel will trim any uncacheable 1139 memory out of your available memory pool based on 1140 MTRR settings. This parameter disables that behavior, 1141 possibly causing your machine to run very slowly. 1142 1143 disable_timer_pin_1 [X86,EARLY] 1144 Disable PIN 1 of APIC timer 1145 Can be useful to work around chipset bugs. 1146 1147 dis_ucode_ldr [X86] Disable the microcode loader. 1148 1149 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support, 1150 this option disables the debugging code at boot. 1151 1152 dma_debug_entries=<number> 1153 This option allows to tune the number of preallocated 1154 entries for DMA-API debugging code. One entry is 1155 required per DMA-API allocation. Use this if the 1156 DMA-API debugging code disables itself because the 1157 architectural default is too low. 1158 1159 dma_debug_driver=<driver_name> 1160 With this option the DMA-API debugging driver 1161 filter feature can be enabled at boot time. Just 1162 pass the driver to filter for as the parameter. 1163 The filter can be disabled or changed to another 1164 driver later using sysfs. 1165 1166 reg_file_data_sampling= 1167 [X86] Controls mitigation for Register File Data 1168 Sampling (RFDS) vulnerability. RFDS is a CPU 1169 vulnerability which may allow userspace to infer 1170 kernel data values previously stored in floating point 1171 registers, vector registers, or integer registers. 1172 RFDS only affects Intel Atom processors. 1173 1174 on: Turns ON the mitigation. 1175 off: Turns OFF the mitigation. 1176 1177 This parameter overrides the compile time default set 1178 by CONFIG_MITIGATION_RFDS. Mitigation cannot be 1179 disabled when other VERW based mitigations (like MDS) 1180 are enabled. In order to disable RFDS mitigation all 1181 VERW based mitigations need to be disabled. 1182 1183 For details see: 1184 Documentation/admin-guide/hw-vuln/reg-file-data-sampling.rst 1185 1186 driver_async_probe= [KNL] 1187 List of driver names to be probed asynchronously. * 1188 matches with all driver names. If * is specified, the 1189 rest of the listed driver names are those that will NOT 1190 match the *. 1191 Format: <driver_name1>,<driver_name2>... 1192 1193 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>] 1194 Broken monitors, graphic adapters, KVMs and EDIDless 1195 panels may send no or incorrect EDID data sets. 1196 This parameter allows to specify an EDID data sets 1197 in the /lib/firmware directory that are used instead. 1198 An EDID data set will only be used for a particular 1199 connector, if its name and a colon are prepended to 1200 the EDID name. Each connector may use a unique EDID 1201 data set by separating the files with a comma. An EDID 1202 data set with no connector name will be used for 1203 any connectors not explicitly specified. 1204 1205 dscc4.setup= [NET] 1206 1207 dt_cpu_ftrs= [PPC,EARLY] 1208 Format: {"off" | "known"} 1209 Control how the dt_cpu_ftrs device-tree binding is 1210 used for CPU feature discovery and setup (if it 1211 exists). 1212 off: Do not use it, fall back to legacy cpu table. 1213 known: Do not pass through unknown features to guests 1214 or userspace, only those that the kernel is aware of. 1215 1216 dump_apple_properties [X86] 1217 Dump name and content of EFI device properties on 1218 x86 Macs. Useful for driver authors to determine 1219 what data is available or for reverse-engineering. 1220 1221 dyndbg[="val"] [KNL,DYNAMIC_DEBUG] 1222 <module>.dyndbg[="val"] 1223 Enable debug messages at boot time. See 1224 Documentation/admin-guide/dynamic-debug-howto.rst 1225 for details. 1226 1227 early_ioremap_debug [KNL,EARLY] 1228 Enable debug messages in early_ioremap support. This 1229 is useful for tracking down temporary early mappings 1230 which are not unmapped. 1231 1232 earlycon= [KNL,EARLY] Output early console device and options. 1233 1234 When used with no options, the early console is 1235 determined by stdout-path property in device tree's 1236 chosen node or the ACPI SPCR table if supported by 1237 the platform. 1238 1239 cdns,<addr>[,options] 1240 Start an early, polled-mode console on a Cadence 1241 (xuartps) serial port at the specified address. Only 1242 supported option is baud rate. If baud rate is not 1243 specified, the serial port must already be setup and 1244 configured. 1245 1246 uart[8250],io,<addr>[,options[,uartclk]] 1247 uart[8250],mmio,<addr>[,options[,uartclk]] 1248 uart[8250],mmio32,<addr>[,options[,uartclk]] 1249 uart[8250],mmio32be,<addr>[,options[,uartclk]] 1250 uart[8250],0x<addr>[,options] 1251 Start an early, polled-mode console on the 8250/16550 1252 UART at the specified I/O port or MMIO address. 1253 MMIO inter-register address stride is either 8-bit 1254 (mmio) or 32-bit (mmio32 or mmio32be). 1255 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed 1256 to be equivalent to 'mmio'. 'options' are specified 1257 in the same format described for "console=ttyS<n>"; if 1258 unspecified, the h/w is not initialized. 'uartclk' is 1259 the uart clock frequency; if unspecified, it is set 1260 to 'BASE_BAUD' * 16. 1261 1262 pl011,<addr> 1263 pl011,mmio32,<addr> 1264 Start an early, polled-mode console on a pl011 serial 1265 port at the specified address. The pl011 serial port 1266 must already be setup and configured. Options are not 1267 yet supported. If 'mmio32' is specified, then only 1268 the driver will use only 32-bit accessors to read/write 1269 the device registers. 1270 1271 liteuart,<addr> 1272 Start an early console on a litex serial port at the 1273 specified address. The serial port must already be 1274 setup and configured. Options are not yet supported. 1275 1276 meson,<addr> 1277 Start an early, polled-mode console on a meson serial 1278 port at the specified address. The serial port must 1279 already be setup and configured. Options are not yet 1280 supported. 1281 1282 msm_serial,<addr> 1283 Start an early, polled-mode console on an msm serial 1284 port at the specified address. The serial port 1285 must already be setup and configured. Options are not 1286 yet supported. 1287 1288 msm_serial_dm,<addr> 1289 Start an early, polled-mode console on an msm serial 1290 dm port at the specified address. The serial port 1291 must already be setup and configured. Options are not 1292 yet supported. 1293 1294 owl,<addr> 1295 Start an early, polled-mode console on a serial port 1296 of an Actions Semi SoC, such as S500 or S900, at the 1297 specified address. The serial port must already be 1298 setup and configured. Options are not yet supported. 1299 1300 rda,<addr> 1301 Start an early, polled-mode console on a serial port 1302 of an RDA Micro SoC, such as RDA8810PL, at the 1303 specified address. The serial port must already be 1304 setup and configured. Options are not yet supported. 1305 1306 sbi 1307 Use RISC-V SBI (Supervisor Binary Interface) for early 1308 console. 1309 1310 smh Use ARM semihosting calls for early console. 1311 1312 s3c2410,<addr> 1313 s3c2412,<addr> 1314 s3c2440,<addr> 1315 s3c6400,<addr> 1316 s5pv210,<addr> 1317 exynos4210,<addr> 1318 Use early console provided by serial driver available 1319 on Samsung SoCs, requires selecting proper type and 1320 a correct base address of the selected UART port. The 1321 serial port must already be setup and configured. 1322 Options are not yet supported. 1323 1324 lantiq,<addr> 1325 Start an early, polled-mode console on a lantiq serial 1326 (lqasc) port at the specified address. The serial port 1327 must already be setup and configured. Options are not 1328 yet supported. 1329 1330 lpuart,<addr> 1331 lpuart32,<addr> 1332 Use early console provided by Freescale LP UART driver 1333 found on Freescale Vybrid and QorIQ LS1021A processors. 1334 A valid base address must be provided, and the serial 1335 port must already be setup and configured. 1336 1337 ec_imx21,<addr> 1338 ec_imx6q,<addr> 1339 Start an early, polled-mode, output-only console on the 1340 Freescale i.MX UART at the specified address. The UART 1341 must already be setup and configured. 1342 1343 ar3700_uart,<addr> 1344 Start an early, polled-mode console on the 1345 Armada 3700 serial port at the specified 1346 address. The serial port must already be setup 1347 and configured. Options are not yet supported. 1348 1349 qcom_geni,<addr> 1350 Start an early, polled-mode console on a Qualcomm 1351 Generic Interface (GENI) based serial port at the 1352 specified address. The serial port must already be 1353 setup and configured. Options are not yet supported. 1354 1355 efifb,[options] 1356 Start an early, unaccelerated console on the EFI 1357 memory mapped framebuffer (if available). On cache 1358 coherent non-x86 systems that use system memory for 1359 the framebuffer, pass the 'ram' option so that it is 1360 mapped with the correct attributes. 1361 1362 linflex,<addr> 1363 Use early console provided by Freescale LINFlexD UART 1364 serial driver for NXP S32V234 SoCs. A valid base 1365 address must be provided, and the serial port must 1366 already be setup and configured. 1367 1368 earlyprintk= [X86,SH,ARM,M68k,S390,UM,EARLY] 1369 earlyprintk=vga 1370 earlyprintk=sclp 1371 earlyprintk=xen 1372 earlyprintk=serial[,ttySn[,baudrate]] 1373 earlyprintk=serial[,0x...[,baudrate]] 1374 earlyprintk=ttySn[,baudrate] 1375 earlyprintk=dbgp[debugController#] 1376 earlyprintk=pciserial[,force],bus:device.function[,baudrate] 1377 earlyprintk=xdbc[xhciController#] 1378 earlyprintk=bios 1379 1380 earlyprintk is useful when the kernel crashes before 1381 the normal console is initialized. It is not enabled by 1382 default because it has some cosmetic problems. 1383 1384 Append ",keep" to not disable it when the real console 1385 takes over. 1386 1387 Only one of vga, serial, or usb debug port can 1388 be used at a time. 1389 1390 Currently only ttyS0 and ttyS1 may be specified by 1391 name. Other I/O ports may be explicitly specified 1392 on some architectures (x86 and arm at least) by 1393 replacing ttySn with an I/O port address, like this: 1394 earlyprintk=serial,0x1008,115200 1395 You can find the port for a given device in 1396 /proc/tty/driver/serial: 1397 2: uart:ST16650V2 port:00001008 irq:18 ... 1398 1399 Interaction with the standard serial driver is not 1400 very good. 1401 1402 The VGA output is eventually overwritten by 1403 the real console. 1404 1405 The xen option can only be used in Xen domains. 1406 1407 The sclp output can only be used on s390. 1408 1409 The bios output can only be used on SuperH. 1410 1411 The optional "force" to "pciserial" enables use of a 1412 PCI device even when its classcode is not of the 1413 UART class. 1414 1415 edac_report= [HW,EDAC] Control how to report EDAC event 1416 Format: {"on" | "off" | "force"} 1417 on: enable EDAC to report H/W event. May be overridden 1418 by other higher priority error reporting module. 1419 off: disable H/W event reporting through EDAC. 1420 force: enforce the use of EDAC to report H/W event. 1421 default: on. 1422 1423 edd= [EDD] 1424 Format: {"off" | "on" | "skip[mbr]"} 1425 1426 efi= [EFI,EARLY] 1427 Format: { "debug", "disable_early_pci_dma", 1428 "nochunk", "noruntime", "nosoftreserve", 1429 "novamap", "no_disable_early_pci_dma" } 1430 debug: enable misc debug output. 1431 disable_early_pci_dma: disable the busmaster bit on all 1432 PCI bridges while in the EFI boot stub. 1433 nochunk: disable reading files in "chunks" in the EFI 1434 boot stub, as chunking can cause problems with some 1435 firmware implementations. 1436 noruntime : disable EFI runtime services support 1437 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose) 1438 attribute may cause the kernel to reserve the 1439 memory range for a memory mapping driver to 1440 claim. Specify efi=nosoftreserve to disable this 1441 reservation and treat the memory by its base type 1442 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM"). 1443 novamap: do not call SetVirtualAddressMap(). 1444 no_disable_early_pci_dma: Leave the busmaster bit set 1445 on all PCI bridges while in the EFI boot stub 1446 1447 efi_no_storage_paranoia [EFI,X86,EARLY] 1448 Using this parameter you can use more than 50% of 1449 your efi variable storage. Use this parameter only if 1450 you are really sure that your UEFI does sane gc and 1451 fulfills the spec otherwise your board may brick. 1452 1453 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI,X86,EARLY] 1454 Add arbitrary attribute to specific memory range by 1455 updating original EFI memory map. 1456 Region of memory which aa attribute is added to is 1457 from ss to ss+nn. 1458 1459 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000 1460 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000) 1461 attribute is added to range 0x100000000-0x180000000 and 1462 0x10a0000000-0x1120000000. 1463 1464 If efi_fake_mem=8G@9G:0x40000 is specified, the 1465 EFI_MEMORY_SP(0x40000) attribute is added to 1466 range 0x240000000-0x43fffffff. 1467 1468 Using this parameter you can do debugging of EFI memmap 1469 related features. For example, you can do debugging of 1470 Address Range Mirroring feature even if your box 1471 doesn't support it, or mark specific memory as 1472 "soft reserved". 1473 1474 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT 1475 that is to be dynamically loaded by Linux. If there are 1476 multiple variables with the same name but with different 1477 vendor GUIDs, all of them will be loaded. See 1478 Documentation/admin-guide/acpi/ssdt-overlays.rst for details. 1479 1480 1481 eisa_irq_edge= [PARISC,HW] 1482 See header of drivers/parisc/eisa.c. 1483 1484 ekgdboc= [X86,KGDB,EARLY] Allow early kernel console debugging 1485 Format: ekgdboc=kbd 1486 1487 This is designed to be used in conjunction with 1488 the boot argument: earlyprintk=vga 1489 1490 This parameter works in place of the kgdboc parameter 1491 but can only be used if the backing tty is available 1492 very early in the boot process. For early debugging 1493 via a serial port see kgdboc_earlycon instead. 1494 1495 elanfreq= [X86-32] 1496 See comment before function elanfreq_setup() in 1497 arch/x86/kernel/cpu/cpufreq/elanfreq.c. 1498 1499 elfcorehdr=[size[KMG]@]offset[KMG] [PPC,SH,X86,S390,EARLY] 1500 Specifies physical address of start of kernel core 1501 image elf header and optionally the size. Generally 1502 kexec loader will pass this option to capture kernel. 1503 See Documentation/admin-guide/kdump/kdump.rst for details. 1504 1505 enable_mtrr_cleanup [X86,EARLY] 1506 The kernel tries to adjust MTRR layout from continuous 1507 to discrete, to make X server driver able to add WB 1508 entry later. This parameter enables that. 1509 1510 enable_timer_pin_1 [X86] 1511 Enable PIN 1 of APIC timer 1512 Can be useful to work around chipset bugs 1513 (in particular on some ATI chipsets). 1514 The kernel tries to set a reasonable default. 1515 1516 enforcing= [SELINUX] Set initial enforcing status. 1517 Format: {"0" | "1"} 1518 See security/selinux/Kconfig help text. 1519 0 -- permissive (log only, no denials). 1520 1 -- enforcing (deny and log). 1521 Default value is 0. 1522 Value can be changed at runtime via 1523 /sys/fs/selinux/enforce. 1524 1525 erst_disable [ACPI] 1526 Disable Error Record Serialization Table (ERST) 1527 support. 1528 1529 ether= [HW,NET] Ethernet cards parameters 1530 This option is obsoleted by the "netdev=" option, which 1531 has equivalent usage. See its documentation for details. 1532 1533 evm= [EVM] 1534 Format: { "fix" } 1535 Permit 'security.evm' to be updated regardless of 1536 current integrity status. 1537 1538 early_page_ext [KNL,EARLY] Enforces page_ext initialization to earlier 1539 stages so cover more early boot allocations. 1540 Please note that as side effect some optimizations 1541 might be disabled to achieve that (e.g. parallelized 1542 memory initialization is disabled) so the boot process 1543 might take longer, especially on systems with a lot of 1544 memory. Available with CONFIG_PAGE_EXTENSION=y. 1545 1546 failslab= 1547 fail_usercopy= 1548 fail_page_alloc= 1549 fail_make_request=[KNL] 1550 General fault injection mechanism. 1551 Format: <interval>,<probability>,<space>,<times> 1552 See also Documentation/fault-injection/. 1553 1554 fb_tunnels= [NET] 1555 Format: { initns | none } 1556 See Documentation/admin-guide/sysctl/net.rst for 1557 fb_tunnels_only_for_init_ns 1558 1559 floppy= [HW] 1560 See Documentation/admin-guide/blockdev/floppy.rst. 1561 1562 forcepae [X86-32] 1563 Forcefully enable Physical Address Extension (PAE). 1564 Many Pentium M systems disable PAE but may have a 1565 functionally usable PAE implementation. 1566 Warning: use of this parameter will taint the kernel 1567 and may cause unknown problems. 1568 1569 fred= [X86-64] 1570 Enable/disable Flexible Return and Event Delivery. 1571 Format: { on | off } 1572 on: enable FRED when it's present. 1573 off: disable FRED, the default setting. 1574 1575 ftrace=[tracer] 1576 [FTRACE] will set and start the specified tracer 1577 as early as possible in order to facilitate early 1578 boot debugging. 1579 1580 ftrace_boot_snapshot 1581 [FTRACE] On boot up, a snapshot will be taken of the 1582 ftrace ring buffer that can be read at: 1583 /sys/kernel/tracing/snapshot. 1584 This is useful if you need tracing information from kernel 1585 boot up that is likely to be overridden by user space 1586 start up functionality. 1587 1588 Optionally, the snapshot can also be defined for a tracing 1589 instance that was created by the trace_instance= command 1590 line parameter. 1591 1592 trace_instance=foo,sched_switch ftrace_boot_snapshot=foo 1593 1594 The above will cause the "foo" tracing instance to trigger 1595 a snapshot at the end of boot up. 1596 1597 ftrace_dump_on_oops[=2(orig_cpu) | =<instance>][,<instance> | 1598 ,<instance>=2(orig_cpu)] 1599 [FTRACE] will dump the trace buffers on oops. 1600 If no parameter is passed, ftrace will dump global 1601 buffers of all CPUs, if you pass 2 or orig_cpu, it 1602 will dump only the buffer of the CPU that triggered 1603 the oops, or the specific instance will be dumped if 1604 its name is passed. Multiple instance dump is also 1605 supported, and instances are separated by commas. Each 1606 instance supports only dump on CPU that triggered the 1607 oops by passing 2 or orig_cpu to it. 1608 1609 ftrace_dump_on_oops=foo=orig_cpu 1610 1611 The above will dump only the buffer of "foo" instance 1612 on CPU that triggered the oops. 1613 1614 ftrace_dump_on_oops,foo,bar=orig_cpu 1615 1616 The above will dump global buffer on all CPUs, the 1617 buffer of "foo" instance on all CPUs and the buffer 1618 of "bar" instance on CPU that triggered the oops. 1619 1620 ftrace_filter=[function-list] 1621 [FTRACE] Limit the functions traced by the function 1622 tracer at boot up. function-list is a comma-separated 1623 list of functions. This list can be changed at run 1624 time by the set_ftrace_filter file in the debugfs 1625 tracing directory. 1626 1627 ftrace_notrace=[function-list] 1628 [FTRACE] Do not trace the functions specified in 1629 function-list. This list can be changed at run time 1630 by the set_ftrace_notrace file in the debugfs 1631 tracing directory. 1632 1633 ftrace_graph_filter=[function-list] 1634 [FTRACE] Limit the top level callers functions traced 1635 by the function graph tracer at boot up. 1636 function-list is a comma-separated list of functions 1637 that can be changed at run time by the 1638 set_graph_function file in the debugfs tracing directory. 1639 1640 ftrace_graph_notrace=[function-list] 1641 [FTRACE] Do not trace from the functions specified in 1642 function-list. This list is a comma-separated list of 1643 functions that can be changed at run time by the 1644 set_graph_notrace file in the debugfs tracing directory. 1645 1646 ftrace_graph_max_depth=<uint> 1647 [FTRACE] Used with the function graph tracer. This is 1648 the max depth it will trace into a function. This value 1649 can be changed at run time by the max_graph_depth file 1650 in the tracefs tracing directory. default: 0 (no limit) 1651 1652 fw_devlink= [KNL,EARLY] Create device links between consumer and supplier 1653 devices by scanning the firmware to infer the 1654 consumer/supplier relationships. This feature is 1655 especially useful when drivers are loaded as modules as 1656 it ensures proper ordering of tasks like device probing 1657 (suppliers first, then consumers), supplier boot state 1658 clean up (only after all consumers have probed), 1659 suspend/resume & runtime PM (consumers first, then 1660 suppliers). 1661 Format: { off | permissive | on | rpm } 1662 off -- Don't create device links from firmware info. 1663 permissive -- Create device links from firmware info 1664 but use it only for ordering boot state clean 1665 up (sync_state() calls). 1666 on -- Create device links from firmware info and use it 1667 to enforce probe and suspend/resume ordering. 1668 rpm -- Like "on", but also use to order runtime PM. 1669 1670 fw_devlink.strict=<bool> 1671 [KNL,EARLY] Treat all inferred dependencies as mandatory 1672 dependencies. This only applies for fw_devlink=on|rpm. 1673 Format: <bool> 1674 1675 fw_devlink.sync_state = 1676 [KNL,EARLY] When all devices that could probe have finished 1677 probing, this parameter controls what to do with 1678 devices that haven't yet received their sync_state() 1679 calls. 1680 Format: { strict | timeout } 1681 strict -- Default. Continue waiting on consumers to 1682 probe successfully. 1683 timeout -- Give up waiting on consumers and call 1684 sync_state() on any devices that haven't yet 1685 received their sync_state() calls after 1686 deferred_probe_timeout has expired or by 1687 late_initcall() if !CONFIG_MODULES. 1688 1689 gamecon.map[2|3]= 1690 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad 1691 support via parallel port (up to 5 devices per port) 1692 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5> 1693 See also Documentation/input/devices/joystick-parport.rst 1694 1695 gamma= [HW,DRM] 1696 1697 gart_fix_e820= [X86-64,EARLY] disable the fix e820 for K8 GART 1698 Format: off | on 1699 default: on 1700 1701 gather_data_sampling= 1702 [X86,INTEL,EARLY] Control the Gather Data Sampling (GDS) 1703 mitigation. 1704 1705 Gather Data Sampling is a hardware vulnerability which 1706 allows unprivileged speculative access to data which was 1707 previously stored in vector registers. 1708 1709 This issue is mitigated by default in updated microcode. 1710 The mitigation may have a performance impact but can be 1711 disabled. On systems without the microcode mitigation 1712 disabling AVX serves as a mitigation. 1713 1714 force: Disable AVX to mitigate systems without 1715 microcode mitigation. No effect if the microcode 1716 mitigation is present. Known to cause crashes in 1717 userspace with buggy AVX enumeration. 1718 1719 off: Disable GDS mitigation. 1720 1721 gcov_persist= [GCOV] When non-zero (default), profiling data for 1722 kernel modules is saved and remains accessible via 1723 debugfs, even when the module is unloaded/reloaded. 1724 When zero, profiling data is discarded and associated 1725 debugfs files are removed at module unload time. 1726 1727 goldfish [X86] Enable the goldfish android emulator platform. 1728 Don't use this when you are not running on the 1729 android emulator 1730 1731 gpio-mockup.gpio_mockup_ranges 1732 [HW] Sets the ranges of gpiochip of for this device. 1733 Format: <start1>,<end1>,<start2>,<end2>... 1734 gpio-mockup.gpio_mockup_named_lines 1735 [HW] Let the driver know GPIO lines should be named. 1736 1737 gpt [EFI] Forces disk with valid GPT signature but 1738 invalid Protective MBR to be treated as GPT. If the 1739 primary GPT is corrupted, it enables the backup/alternate 1740 GPT to be used instead. 1741 1742 grcan.enable0= [HW] Configuration of physical interface 0. Determines 1743 the "Enable 0" bit of the configuration register. 1744 Format: 0 | 1 1745 Default: 0 1746 grcan.enable1= [HW] Configuration of physical interface 1. Determines 1747 the "Enable 0" bit of the configuration register. 1748 Format: 0 | 1 1749 Default: 0 1750 grcan.select= [HW] Select which physical interface to use. 1751 Format: 0 | 1 1752 Default: 0 1753 grcan.txsize= [HW] Sets the size of the tx buffer. 1754 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0. 1755 Default: 1024 1756 grcan.rxsize= [HW] Sets the size of the rx buffer. 1757 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0. 1758 Default: 1024 1759 1760 hardened_usercopy= 1761 [KNL] Under CONFIG_HARDENED_USERCOPY, whether 1762 hardening is enabled for this boot. Hardened 1763 usercopy checking is used to protect the kernel 1764 from reading or writing beyond known memory 1765 allocation boundaries as a proactive defense 1766 against bounds-checking flaws in the kernel's 1767 copy_to_user()/copy_from_user() interface. 1768 on Perform hardened usercopy checks (default). 1769 off Disable hardened usercopy checks. 1770 1771 hardlockup_all_cpu_backtrace= 1772 [KNL] Should the hard-lockup detector generate 1773 backtraces on all cpus. 1774 Format: 0 | 1 1775 1776 hashdist= [KNL,NUMA] Large hashes allocated during boot 1777 are distributed across NUMA nodes. Defaults on 1778 for 64-bit NUMA, off otherwise. 1779 Format: 0 | 1 (for off | on) 1780 1781 hcl= [IA-64] SGI's Hardware Graph compatibility layer 1782 1783 hd= [EIDE] (E)IDE hard drive subsystem geometry 1784 Format: <cyl>,<head>,<sect> 1785 1786 hest_disable [ACPI] 1787 Disable Hardware Error Source Table (HEST) support; 1788 corresponding firmware-first mode error processing 1789 logic will be disabled. 1790 1791 hibernate= [HIBERNATION] 1792 noresume Don't check if there's a hibernation image 1793 present during boot. 1794 nocompress Don't compress/decompress hibernation images. 1795 no Disable hibernation and resume. 1796 protect_image Turn on image protection during restoration 1797 (that will set all pages holding image data 1798 during restoration read-only). 1799 1800 hibernate.compressor= [HIBERNATION] Compression algorithm to be 1801 used with hibernation. 1802 Format: { lzo | lz4 } 1803 Default: lzo 1804 1805 lzo: Select LZO compression algorithm to 1806 compress/decompress hibernation image. 1807 1808 lz4: Select LZ4 compression algorithm to 1809 compress/decompress hibernation image. 1810 1811 highmem=nn[KMG] [KNL,BOOT,EARLY] forces the highmem zone to have an exact 1812 size of <nn>. This works even on boxes that have no 1813 highmem otherwise. This also works to reduce highmem 1814 size on bigger boxes. 1815 1816 highres= [KNL] Enable/disable high resolution timer mode. 1817 Valid parameters: "on", "off" 1818 Default: "on" 1819 1820 hlt [BUGS=ARM,SH] 1821 1822 hostname= [KNL,EARLY] Set the hostname (aka UTS nodename). 1823 Format: <string> 1824 This allows setting the system's hostname during early 1825 startup. This sets the name returned by gethostname. 1826 Using this parameter to set the hostname makes it 1827 possible to ensure the hostname is correctly set before 1828 any userspace processes run, avoiding the possibility 1829 that a process may call gethostname before the hostname 1830 has been explicitly set, resulting in the calling 1831 process getting an incorrect result. The string must 1832 not exceed the maximum allowed hostname length (usually 1833 64 characters) and will be truncated otherwise. 1834 1835 hpet= [X86-32,HPET] option to control HPET usage 1836 Format: { enable (default) | disable | force | 1837 verbose } 1838 disable: disable HPET and use PIT instead 1839 force: allow force enabled of undocumented chips (ICH4, 1840 VIA, nVidia) 1841 verbose: show contents of HPET registers during setup 1842 1843 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET 1844 registers. Default set by CONFIG_HPET_MMAP_DEFAULT. 1845 1846 hugepages= [HW] Number of HugeTLB pages to allocate at boot. 1847 If this follows hugepagesz (below), it specifies 1848 the number of pages of hugepagesz to be allocated. 1849 If this is the first HugeTLB parameter on the command 1850 line, it specifies the number of pages to allocate for 1851 the default huge page size. If using node format, the 1852 number of pages to allocate per-node can be specified. 1853 See also Documentation/admin-guide/mm/hugetlbpage.rst. 1854 Format: <integer> or (node format) 1855 <node>:<integer>[,<node>:<integer>] 1856 1857 hugepagesz= 1858 [HW] The size of the HugeTLB pages. This is used in 1859 conjunction with hugepages (above) to allocate huge 1860 pages of a specific size at boot. The pair 1861 hugepagesz=X hugepages=Y can be specified once for 1862 each supported huge page size. Huge page sizes are 1863 architecture dependent. See also 1864 Documentation/admin-guide/mm/hugetlbpage.rst. 1865 Format: size[KMG] 1866 1867 hugetlb_cma= [HW,CMA,EARLY] The size of a CMA area used for allocation 1868 of gigantic hugepages. Or using node format, the size 1869 of a CMA area per node can be specified. 1870 Format: nn[KMGTPE] or (node format) 1871 <node>:nn[KMGTPE][,<node>:nn[KMGTPE]] 1872 1873 Reserve a CMA area of given size and allocate gigantic 1874 hugepages using the CMA allocator. If enabled, the 1875 boot-time allocation of gigantic hugepages is skipped. 1876 1877 hugetlb_free_vmemmap= 1878 [KNL] Requires CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP 1879 enabled. 1880 Control if HugeTLB Vmemmap Optimization (HVO) is enabled. 1881 Allows heavy hugetlb users to free up some more 1882 memory (7 * PAGE_SIZE for each 2MB hugetlb page). 1883 Format: { on | off (default) } 1884 1885 on: enable HVO 1886 off: disable HVO 1887 1888 Built with CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP_DEFAULT_ON=y, 1889 the default is on. 1890 1891 Note that the vmemmap pages may be allocated from the added 1892 memory block itself when memory_hotplug.memmap_on_memory is 1893 enabled, those vmemmap pages cannot be optimized even if this 1894 feature is enabled. Other vmemmap pages not allocated from 1895 the added memory block itself do not be affected. 1896 1897 hung_task_panic= 1898 [KNL] Should the hung task detector generate panics. 1899 Format: 0 | 1 1900 1901 A value of 1 instructs the kernel to panic when a 1902 hung task is detected. The default value is controlled 1903 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time 1904 option. The value selected by this boot parameter can 1905 be changed later by the kernel.hung_task_panic sysctl. 1906 1907 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC) 1908 terminal devices. Valid values: 0..8 1909 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs. 1910 If specified, z/VM IUCV HVC accepts connections 1911 from listed z/VM user IDs only. 1912 1913 hv_nopvspin [X86,HYPER_V,EARLY] 1914 Disables the paravirt spinlock optimizations 1915 which allow the hypervisor to 'idle' the guest 1916 on lock contention. 1917 1918 i2c_bus= [HW] Override the default board specific I2C bus speed 1919 or register an additional I2C bus that is not 1920 registered from board initialization code. 1921 Format: 1922 <bus_id>,<clkrate> 1923 1924 i2c_touchscreen_props= [HW,ACPI,X86] 1925 Set device-properties for ACPI-enumerated I2C-attached 1926 touchscreen, to e.g. fix coordinates of upside-down 1927 mounted touchscreens. If you need this option please 1928 submit a drivers/platform/x86/touchscreen_dmi.c patch 1929 adding a DMI quirk for this. 1930 1931 Format: 1932 <ACPI_HW_ID>:<prop_name>=<val>[:prop_name=val][:...] 1933 Where <val> is one of: 1934 Omit "=<val>" entirely Set a boolean device-property 1935 Unsigned number Set a u32 device-property 1936 Anything else Set a string device-property 1937 1938 Examples (split over multiple lines): 1939 i2c_touchscreen_props=GDIX1001:touchscreen-inverted-x: 1940 touchscreen-inverted-y 1941 1942 i2c_touchscreen_props=MSSL1680:touchscreen-size-x=1920: 1943 touchscreen-size-y=1080:touchscreen-inverted-y: 1944 firmware-name=gsl1680-vendor-model.fw:silead,home-button 1945 1946 i8042.debug [HW] Toggle i8042 debug mode 1947 i8042.unmask_kbd_data 1948 [HW] Enable printing of interrupt data from the KBD port 1949 (disabled by default, and as a pre-condition 1950 requires that i8042.debug=1 be enabled) 1951 i8042.direct [HW] Put keyboard port into non-translated mode 1952 i8042.dumbkbd [HW] Pretend that controller can only read data from 1953 keyboard and cannot control its state 1954 (Don't attempt to blink the leds) 1955 i8042.noaux [HW] Don't check for auxiliary (== mouse) port 1956 i8042.nokbd [HW] Don't check/create keyboard port 1957 i8042.noloop [HW] Disable the AUX Loopback command while probing 1958 for the AUX port 1959 i8042.nomux [HW] Don't check presence of an active multiplexing 1960 controller 1961 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX 1962 controllers 1963 i8042.notimeout [HW] Ignore timeout condition signalled by controller 1964 i8042.reset [HW] Reset the controller during init, cleanup and 1965 suspend-to-ram transitions, only during s2r 1966 transitions, or never reset 1967 Format: { 1 | Y | y | 0 | N | n } 1968 1, Y, y: always reset controller 1969 0, N, n: don't ever reset controller 1970 Default: only on s2r transitions on x86; most other 1971 architectures force reset to be always executed 1972 i8042.unlock [HW] Unlock (ignore) the keylock 1973 i8042.kbdreset [HW] Reset device connected to KBD port 1974 i8042.probe_defer 1975 [HW] Allow deferred probing upon i8042 probe errors 1976 1977 i810= [HW,DRM] 1978 1979 i915.invert_brightness= 1980 [DRM] Invert the sense of the variable that is used to 1981 set the brightness of the panel backlight. Normally a 1982 brightness value of 0 indicates backlight switched off, 1983 and the maximum of the brightness value sets the backlight 1984 to maximum brightness. If this parameter is set to 0 1985 (default) and the machine requires it, or this parameter 1986 is set to 1, a brightness value of 0 sets the backlight 1987 to maximum brightness, and the maximum of the brightness 1988 value switches the backlight off. 1989 -1 -- never invert brightness 1990 0 -- machine default 1991 1 -- force brightness inversion 1992 1993 ia32_emulation= [X86-64] 1994 Format: <bool> 1995 When true, allows loading 32-bit programs and executing 32-bit 1996 syscalls, essentially overriding IA32_EMULATION_DEFAULT_DISABLED at 1997 boot time. When false, unconditionally disables IA32 emulation. 1998 1999 icn= [HW,ISDN] 2000 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]] 2001 2002 2003 idle= [X86,EARLY] 2004 Format: idle=poll, idle=halt, idle=nomwait 2005 Poll forces a polling idle loop that can slightly 2006 improve the performance of waking up a idle CPU, but 2007 will use a lot of power and make the system run hot. 2008 Not recommended. 2009 idle=halt: Halt is forced to be used for CPU idle. 2010 In such case C2/C3 won't be used again. 2011 idle=nomwait: Disable mwait for CPU C-states 2012 2013 idxd.sva= [HW] 2014 Format: <bool> 2015 Allow force disabling of Shared Virtual Memory (SVA) 2016 support for the idxd driver. By default it is set to 2017 true (1). 2018 2019 idxd.tc_override= [HW] 2020 Format: <bool> 2021 Allow override of default traffic class configuration 2022 for the device. By default it is set to false (0). 2023 2024 ieee754= [MIPS] Select IEEE Std 754 conformance mode 2025 Format: { strict | legacy | 2008 | relaxed } 2026 Default: strict 2027 2028 Choose which programs will be accepted for execution 2029 based on the IEEE 754 NaN encoding(s) supported by 2030 the FPU and the NaN encoding requested with the value 2031 of an ELF file header flag individually set by each 2032 binary. Hardware implementations are permitted to 2033 support either or both of the legacy and the 2008 NaN 2034 encoding mode. 2035 2036 Available settings are as follows: 2037 strict accept binaries that request a NaN encoding 2038 supported by the FPU 2039 legacy only accept legacy-NaN binaries, if supported 2040 by the FPU 2041 2008 only accept 2008-NaN binaries, if supported 2042 by the FPU 2043 relaxed accept any binaries regardless of whether 2044 supported by the FPU 2045 2046 The FPU emulator is always able to support both NaN 2047 encodings, so if no FPU hardware is present or it has 2048 been disabled with 'nofpu', then the settings of 2049 'legacy' and '2008' strap the emulator accordingly, 2050 'relaxed' straps the emulator for both legacy-NaN and 2051 2008-NaN, whereas 'strict' enables legacy-NaN only on 2052 legacy processors and both NaN encodings on MIPS32 or 2053 MIPS64 CPUs. 2054 2055 The setting for ABS.fmt/NEG.fmt instruction execution 2056 mode generally follows that for the NaN encoding, 2057 except where unsupported by hardware. 2058 2059 ignore_loglevel [KNL,EARLY] 2060 Ignore loglevel setting - this will print /all/ 2061 kernel messages to the console. Useful for debugging. 2062 We also add it as printk module parameter, so users 2063 could change it dynamically, usually by 2064 /sys/module/printk/parameters/ignore_loglevel. 2065 2066 ignore_rlimit_data 2067 Ignore RLIMIT_DATA setting for data mappings, 2068 print warning at first misuse. Can be changed via 2069 /sys/module/kernel/parameters/ignore_rlimit_data. 2070 2071 ihash_entries= [KNL] 2072 Set number of hash buckets for inode cache. 2073 2074 ima_appraise= [IMA] appraise integrity measurements 2075 Format: { "off" | "enforce" | "fix" | "log" } 2076 default: "enforce" 2077 2078 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead. 2079 The builtin appraise policy appraises all files 2080 owned by uid=0. 2081 2082 ima_canonical_fmt [IMA] 2083 Use the canonical format for the binary runtime 2084 measurements, instead of host native format. 2085 2086 ima_hash= [IMA] 2087 Format: { md5 | sha1 | rmd160 | sha256 | sha384 2088 | sha512 | ... } 2089 default: "sha1" 2090 2091 The list of supported hash algorithms is defined 2092 in crypto/hash_info.h. 2093 2094 ima_policy= [IMA] 2095 The builtin policies to load during IMA setup. 2096 Format: "tcb | appraise_tcb | secure_boot | 2097 fail_securely | critical_data" 2098 2099 The "tcb" policy measures all programs exec'd, files 2100 mmap'd for exec, and all files opened with the read 2101 mode bit set by either the effective uid (euid=0) or 2102 uid=0. 2103 2104 The "appraise_tcb" policy appraises the integrity of 2105 all files owned by root. 2106 2107 The "secure_boot" policy appraises the integrity 2108 of files (eg. kexec kernel image, kernel modules, 2109 firmware, policy, etc) based on file signatures. 2110 2111 The "fail_securely" policy forces file signature 2112 verification failure also on privileged mounted 2113 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE 2114 flag. 2115 2116 The "critical_data" policy measures kernel integrity 2117 critical data. 2118 2119 ima_tcb [IMA] Deprecated. Use ima_policy= instead. 2120 Load a policy which meets the needs of the Trusted 2121 Computing Base. This means IMA will measure all 2122 programs exec'd, files mmap'd for exec, and all files 2123 opened for read by uid=0. 2124 2125 ima_template= [IMA] 2126 Select one of defined IMA measurements template formats. 2127 Formats: { "ima" | "ima-ng" | "ima-ngv2" | "ima-sig" | 2128 "ima-sigv2" } 2129 Default: "ima-ng" 2130 2131 ima_template_fmt= 2132 [IMA] Define a custom template format. 2133 Format: { "field1|...|fieldN" } 2134 2135 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage 2136 Format: <min_file_size> 2137 Set the minimal file size for using asynchronous hash. 2138 If left unspecified, ahash usage is disabled. 2139 2140 ahash performance varies for different data sizes on 2141 different crypto accelerators. This option can be used 2142 to achieve the best performance for a particular HW. 2143 2144 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size 2145 Format: <bufsize> 2146 Set hashing buffer size. Default: 4k. 2147 2148 ahash performance varies for different chunk sizes on 2149 different crypto accelerators. This option can be used 2150 to achieve best performance for particular HW. 2151 2152 init= [KNL] 2153 Format: <full_path> 2154 Run specified binary instead of /sbin/init as init 2155 process. 2156 2157 initcall_debug [KNL] Trace initcalls as they are executed. Useful 2158 for working out where the kernel is dying during 2159 startup. 2160 2161 initcall_blacklist= [KNL] Do not execute a comma-separated list of 2162 initcall functions. Useful for debugging built-in 2163 modules and initcalls. 2164 2165 initramfs_async= [KNL] 2166 Format: <bool> 2167 Default: 1 2168 This parameter controls whether the initramfs 2169 image is unpacked asynchronously, concurrently 2170 with devices being probed and 2171 initialized. This should normally just work, 2172 but as a debugging aid, one can get the 2173 historical behaviour of the initramfs 2174 unpacking being completed before device_ and 2175 late_ initcalls. 2176 2177 initrd= [BOOT,EARLY] Specify the location of the initial ramdisk 2178 2179 initrdmem= [KNL,EARLY] Specify a physical address and size from which to 2180 load the initrd. If an initrd is compiled in or 2181 specified in the bootparams, it takes priority over this 2182 setting. 2183 Format: ss[KMG],nn[KMG] 2184 Default is 0, 0 2185 2186 init_on_alloc= [MM,EARLY] Fill newly allocated pages and heap objects with 2187 zeroes. 2188 Format: 0 | 1 2189 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON. 2190 2191 init_on_free= [MM,EARLY] Fill freed pages and heap objects with zeroes. 2192 Format: 0 | 1 2193 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON. 2194 2195 init_mlocked_on_free= [MM] Fill freed userspace memory with zeroes if 2196 it was mlock'ed and not explicitly munlock'ed 2197 afterwards. 2198 Format: 0 | 1 2199 Default set by CONFIG_INIT_MLOCKED_ON_FREE_DEFAULT_ON 2200 2201 init_pkru= [X86] Specify the default memory protection keys rights 2202 register contents for all processes. 0x55555554 by 2203 default (disallow access to all but pkey 0). Can 2204 override in debugfs after boot. 2205 2206 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver 2207 Format: <irq> 2208 2209 int_pln_enable [X86] Enable power limit notification interrupt 2210 2211 integrity_audit=[IMA] 2212 Format: { "0" | "1" } 2213 0 -- basic integrity auditing messages. (Default) 2214 1 -- additional integrity auditing messages. 2215 2216 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option 2217 on 2218 Enable intel iommu driver. 2219 off 2220 Disable intel iommu driver. 2221 igfx_off [Default Off] 2222 By default, gfx is mapped as normal device. If a gfx 2223 device has a dedicated DMAR unit, the DMAR unit is 2224 bypassed by not enabling DMAR with this option. In 2225 this case, gfx device will use physical address for 2226 DMA. 2227 strict [Default Off] 2228 Deprecated, equivalent to iommu.strict=1. 2229 sp_off [Default Off] 2230 By default, super page will be supported if Intel IOMMU 2231 has the capability. With this option, super page will 2232 not be supported. 2233 sm_on 2234 Enable the Intel IOMMU scalable mode if the hardware 2235 advertises that it has support for the scalable mode 2236 translation. 2237 sm_off 2238 Disallow use of the Intel IOMMU scalable mode. 2239 tboot_noforce [Default Off] 2240 Do not force the Intel IOMMU enabled under tboot. 2241 By default, tboot will force Intel IOMMU on, which 2242 could harm performance of some high-throughput 2243 devices like 40GBit network cards, even if identity 2244 mapping is enabled. 2245 Note that using this option lowers the security 2246 provided by tboot because it makes the system 2247 vulnerable to DMA attacks. 2248 2249 intel_idle.max_cstate= [KNL,HW,ACPI,X86] 2250 0 disables intel_idle and fall back on acpi_idle. 2251 1 to 9 specify maximum depth of C-state. 2252 2253 intel_pstate= [X86,EARLY] 2254 disable 2255 Do not enable intel_pstate as the default 2256 scaling driver for the supported processors 2257 active 2258 Use intel_pstate driver to bypass the scaling 2259 governors layer of cpufreq and provides it own 2260 algorithms for p-state selection. There are two 2261 P-state selection algorithms provided by 2262 intel_pstate in the active mode: powersave and 2263 performance. The way they both operate depends 2264 on whether or not the hardware managed P-states 2265 (HWP) feature has been enabled in the processor 2266 and possibly on the processor model. 2267 passive 2268 Use intel_pstate as a scaling driver, but configure it 2269 to work with generic cpufreq governors (instead of 2270 enabling its internal governor). This mode cannot be 2271 used along with the hardware-managed P-states (HWP) 2272 feature. 2273 force 2274 Enable intel_pstate on systems that prohibit it by default 2275 in favor of acpi-cpufreq. Forcing the intel_pstate driver 2276 instead of acpi-cpufreq may disable platform features, such 2277 as thermal controls and power capping, that rely on ACPI 2278 P-States information being indicated to OSPM and therefore 2279 should be used with caution. This option does not work with 2280 processors that aren't supported by the intel_pstate driver 2281 or on platforms that use pcc-cpufreq instead of acpi-cpufreq. 2282 no_hwp 2283 Do not enable hardware P state control (HWP) 2284 if available. 2285 hwp_only 2286 Only load intel_pstate on systems which support 2287 hardware P state control (HWP) if available. 2288 support_acpi_ppc 2289 Enforce ACPI _PPC performance limits. If the Fixed ACPI 2290 Description Table, specifies preferred power management 2291 profile as "Enterprise Server" or "Performance Server", 2292 then this feature is turned on by default. 2293 per_cpu_perf_limits 2294 Allow per-logical-CPU P-State performance control limits using 2295 cpufreq sysfs interface 2296 2297 intremap= [X86-64,Intel-IOMMU,EARLY] 2298 on enable Interrupt Remapping (default) 2299 off disable Interrupt Remapping 2300 nosid disable Source ID checking 2301 no_x2apic_optout 2302 BIOS x2APIC opt-out request will be ignored 2303 nopost disable Interrupt Posting 2304 posted_msi 2305 enable MSIs delivered as posted interrupts 2306 2307 iomem= Disable strict checking of access to MMIO memory 2308 strict regions from userspace. 2309 relaxed 2310 2311 iommu= [X86,EARLY] 2312 off 2313 force 2314 noforce 2315 biomerge 2316 panic 2317 nopanic 2318 merge 2319 nomerge 2320 soft 2321 pt [X86] 2322 nopt [X86] 2323 nobypass [PPC/POWERNV] 2324 Disable IOMMU bypass, using IOMMU for PCI devices. 2325 2326 iommu.forcedac= [ARM64,X86,EARLY] Control IOVA allocation for PCI devices. 2327 Format: { "0" | "1" } 2328 0 - Try to allocate a 32-bit DMA address first, before 2329 falling back to the full range if needed. 2330 1 - Allocate directly from the full usable range, 2331 forcing Dual Address Cycle for PCI cards supporting 2332 greater than 32-bit addressing. 2333 2334 iommu.strict= [ARM64,X86,S390,EARLY] Configure TLB invalidation behaviour 2335 Format: { "0" | "1" } 2336 0 - Lazy mode. 2337 Request that DMA unmap operations use deferred 2338 invalidation of hardware TLBs, for increased 2339 throughput at the cost of reduced device isolation. 2340 Will fall back to strict mode if not supported by 2341 the relevant IOMMU driver. 2342 1 - Strict mode. 2343 DMA unmap operations invalidate IOMMU hardware TLBs 2344 synchronously. 2345 unset - Use value of CONFIG_IOMMU_DEFAULT_DMA_{LAZY,STRICT}. 2346 Note: on x86, strict mode specified via one of the 2347 legacy driver-specific options takes precedence. 2348 2349 iommu.passthrough= 2350 [ARM64,X86,EARLY] Configure DMA to bypass the IOMMU by default. 2351 Format: { "0" | "1" } 2352 0 - Use IOMMU translation for DMA. 2353 1 - Bypass the IOMMU for DMA. 2354 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH. 2355 2356 io7= [HW] IO7 for Marvel-based Alpha systems 2357 See comment before marvel_specify_io7 in 2358 arch/alpha/kernel/core_marvel.c. 2359 2360 io_delay= [X86,EARLY] I/O delay method 2361 0x80 2362 Standard port 0x80 based delay 2363 0xed 2364 Alternate port 0xed based delay (needed on some systems) 2365 udelay 2366 Simple two microseconds delay 2367 none 2368 No delay 2369 2370 ip= [IP_PNP] 2371 See Documentation/admin-guide/nfs/nfsroot.rst. 2372 2373 ipcmni_extend [KNL,EARLY] Extend the maximum number of unique System V 2374 IPC identifiers from 32,768 to 16,777,216. 2375 2376 irqaffinity= [SMP] Set the default irq affinity mask 2377 The argument is a cpu list, as described above. 2378 2379 irqchip.gicv2_force_probe= 2380 [ARM,ARM64,EARLY] 2381 Format: <bool> 2382 Force the kernel to look for the second 4kB page 2383 of a GICv2 controller even if the memory range 2384 exposed by the device tree is too small. 2385 2386 irqchip.gicv3_nolpi= 2387 [ARM,ARM64,EARLY] 2388 Force the kernel to ignore the availability of 2389 LPIs (and by consequence ITSs). Intended for system 2390 that use the kernel as a bootloader, and thus want 2391 to let secondary kernels in charge of setting up 2392 LPIs. 2393 2394 irqchip.gicv3_pseudo_nmi= [ARM64,EARLY] 2395 Enables support for pseudo-NMIs in the kernel. This 2396 requires the kernel to be built with 2397 CONFIG_ARM64_PSEUDO_NMI. 2398 2399 irqfixup [HW] 2400 When an interrupt is not handled search all handlers 2401 for it. Intended to get systems with badly broken 2402 firmware running. 2403 2404 irqpoll [HW] 2405 When an interrupt is not handled search all handlers 2406 for it. Also check all handlers each timer 2407 interrupt. Intended to get systems with badly broken 2408 firmware running. 2409 2410 isapnp= [ISAPNP] 2411 Format: <RDP>,<reset>,<pci_scan>,<verbosity> 2412 2413 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance. 2414 [Deprecated - use cpusets instead] 2415 Format: [flag-list,]<cpu-list> 2416 2417 Specify one or more CPUs to isolate from disturbances 2418 specified in the flag list (default: domain): 2419 2420 nohz 2421 Disable the tick when a single task runs. 2422 2423 A residual 1Hz tick is offloaded to workqueues, which you 2424 need to affine to housekeeping through the global 2425 workqueue's affinity configured via the 2426 /sys/devices/virtual/workqueue/cpumask sysfs file, or 2427 by using the 'domain' flag described below. 2428 2429 NOTE: by default the global workqueue runs on all CPUs, 2430 so to protect individual CPUs the 'cpumask' file has to 2431 be configured manually after bootup. 2432 2433 domain 2434 Isolate from the general SMP balancing and scheduling 2435 algorithms. Note that performing domain isolation this way 2436 is irreversible: it's not possible to bring back a CPU to 2437 the domains once isolated through isolcpus. It's strongly 2438 advised to use cpusets instead to disable scheduler load 2439 balancing through the "cpuset.sched_load_balance" file. 2440 It offers a much more flexible interface where CPUs can 2441 move in and out of an isolated set anytime. 2442 2443 You can move a process onto or off an "isolated" CPU via 2444 the CPU affinity syscalls or cpuset. 2445 <cpu number> begins at 0 and the maximum value is 2446 "number of CPUs in system - 1". 2447 2448 managed_irq 2449 2450 Isolate from being targeted by managed interrupts 2451 which have an interrupt mask containing isolated 2452 CPUs. The affinity of managed interrupts is 2453 handled by the kernel and cannot be changed via 2454 the /proc/irq/* interfaces. 2455 2456 This isolation is best effort and only effective 2457 if the automatically assigned interrupt mask of a 2458 device queue contains isolated and housekeeping 2459 CPUs. If housekeeping CPUs are online then such 2460 interrupts are directed to the housekeeping CPU 2461 so that IO submitted on the housekeeping CPU 2462 cannot disturb the isolated CPU. 2463 2464 If a queue's affinity mask contains only isolated 2465 CPUs then this parameter has no effect on the 2466 interrupt routing decision, though interrupts are 2467 only delivered when tasks running on those 2468 isolated CPUs submit IO. IO submitted on 2469 housekeeping CPUs has no influence on those 2470 queues. 2471 2472 The format of <cpu-list> is described above. 2473 2474 iucv= [HW,NET] 2475 2476 ivrs_ioapic [HW,X86-64] 2477 Provide an override to the IOAPIC-ID<->DEVICE-ID 2478 mapping provided in the IVRS ACPI table. 2479 By default, PCI segment is 0, and can be omitted. 2480 2481 For example, to map IOAPIC-ID decimal 10 to 2482 PCI segment 0x1 and PCI device 00:14.0, 2483 write the parameter as: 2484 ivrs_ioapic=10@0001:00:14.0 2485 2486 Deprecated formats: 2487 * To map IOAPIC-ID decimal 10 to PCI device 00:14.0 2488 write the parameter as: 2489 ivrs_ioapic[10]=00:14.0 2490 * To map IOAPIC-ID decimal 10 to PCI segment 0x1 and 2491 PCI device 00:14.0 write the parameter as: 2492 ivrs_ioapic[10]=0001:00:14.0 2493 2494 ivrs_hpet [HW,X86-64] 2495 Provide an override to the HPET-ID<->DEVICE-ID 2496 mapping provided in the IVRS ACPI table. 2497 By default, PCI segment is 0, and can be omitted. 2498 2499 For example, to map HPET-ID decimal 10 to 2500 PCI segment 0x1 and PCI device 00:14.0, 2501 write the parameter as: 2502 ivrs_hpet=10@0001:00:14.0 2503 2504 Deprecated formats: 2505 * To map HPET-ID decimal 0 to PCI device 00:14.0 2506 write the parameter as: 2507 ivrs_hpet[0]=00:14.0 2508 * To map HPET-ID decimal 10 to PCI segment 0x1 and 2509 PCI device 00:14.0 write the parameter as: 2510 ivrs_ioapic[10]=0001:00:14.0 2511 2512 ivrs_acpihid [HW,X86-64] 2513 Provide an override to the ACPI-HID:UID<->DEVICE-ID 2514 mapping provided in the IVRS ACPI table. 2515 By default, PCI segment is 0, and can be omitted. 2516 2517 For example, to map UART-HID:UID AMD0020:0 to 2518 PCI segment 0x1 and PCI device ID 00:14.5, 2519 write the parameter as: 2520 ivrs_acpihid=AMD0020:0@0001:00:14.5 2521 2522 Deprecated formats: 2523 * To map UART-HID:UID AMD0020:0 to PCI segment is 0, 2524 PCI device ID 00:14.5, write the parameter as: 2525 ivrs_acpihid[00:14.5]=AMD0020:0 2526 * To map UART-HID:UID AMD0020:0 to PCI segment 0x1 and 2527 PCI device ID 00:14.5, write the parameter as: 2528 ivrs_acpihid[0001:00:14.5]=AMD0020:0 2529 2530 js= [HW,JOY] Analog joystick 2531 See Documentation/input/joydev/joystick.rst. 2532 2533 kasan_multi_shot 2534 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print 2535 report on every invalid memory access. Without this 2536 parameter KASAN will print report only for the first 2537 invalid access. 2538 2539 keep_bootcon [KNL,EARLY] 2540 Do not unregister boot console at start. This is only 2541 useful for debugging when something happens in the window 2542 between unregistering the boot console and initializing 2543 the real console. 2544 2545 keepinitrd [HW,ARM] See retain_initrd. 2546 2547 kernelcore= [KNL,X86,IA-64,PPC,EARLY] 2548 Format: nn[KMGTPE] | nn% | "mirror" 2549 This parameter specifies the amount of memory usable by 2550 the kernel for non-movable allocations. The requested 2551 amount is spread evenly throughout all nodes in the 2552 system as ZONE_NORMAL. The remaining memory is used for 2553 movable memory in its own zone, ZONE_MOVABLE. In the 2554 event, a node is too small to have both ZONE_NORMAL and 2555 ZONE_MOVABLE, kernelcore memory will take priority and 2556 other nodes will have a larger ZONE_MOVABLE. 2557 2558 ZONE_MOVABLE is used for the allocation of pages that 2559 may be reclaimed or moved by the page migration 2560 subsystem. Note that allocations like PTEs-from-HighMem 2561 still use the HighMem zone if it exists, and the Normal 2562 zone if it does not. 2563 2564 It is possible to specify the exact amount of memory in 2565 the form of "nn[KMGTPE]", a percentage of total system 2566 memory in the form of "nn%", or "mirror". If "mirror" 2567 option is specified, mirrored (reliable) memory is used 2568 for non-movable allocations and remaining memory is used 2569 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror" 2570 are exclusive, so you cannot specify multiple forms. 2571 2572 kgdbdbgp= [KGDB,HW,EARLY] kgdb over EHCI usb debug port. 2573 Format: <Controller#>[,poll interval] 2574 The controller # is the number of the ehci usb debug 2575 port as it is probed via PCI. The poll interval is 2576 optional and is the number seconds in between 2577 each poll cycle to the debug port in case you need 2578 the functionality for interrupting the kernel with 2579 gdb or control-c on the dbgp connection. When 2580 not using this parameter you use sysrq-g to break into 2581 the kernel debugger. 2582 2583 kgdboc= [KGDB,HW] kgdb over consoles. 2584 Requires a tty driver that supports console polling, 2585 or a supported polling keyboard driver (non-usb). 2586 Serial only format: <serial_device>[,baud] 2587 keyboard only format: kbd 2588 keyboard and serial format: kbd,<serial_device>[,baud] 2589 Optional Kernel mode setting: 2590 kms, kbd format: kms,kbd 2591 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud] 2592 2593 kgdboc_earlycon= [KGDB,HW,EARLY] 2594 If the boot console provides the ability to read 2595 characters and can work in polling mode, you can use 2596 this parameter to tell kgdb to use it as a backend 2597 until the normal console is registered. Intended to 2598 be used together with the kgdboc parameter which 2599 specifies the normal console to transition to. 2600 2601 The name of the early console should be specified 2602 as the value of this parameter. Note that the name of 2603 the early console might be different than the tty 2604 name passed to kgdboc. It's OK to leave the value 2605 blank and the first boot console that implements 2606 read() will be picked. 2607 2608 kgdbwait [KGDB,EARLY] Stop kernel execution and enter the 2609 kernel debugger at the earliest opportunity. 2610 2611 kmac= [MIPS] Korina ethernet MAC address. 2612 Configure the RouterBoard 532 series on-chip 2613 Ethernet adapter MAC address. 2614 2615 kmemleak= [KNL,EARLY] Boot-time kmemleak enable/disable 2616 Valid arguments: on, off 2617 Default: on 2618 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y, 2619 the default is off. 2620 2621 kprobe_event=[probe-list] 2622 [FTRACE] Add kprobe events and enable at boot time. 2623 The probe-list is a semicolon delimited list of probe 2624 definitions. Each definition is same as kprobe_events 2625 interface, but the parameters are comma delimited. 2626 For example, to add a kprobe event on vfs_read with 2627 arg1 and arg2, add to the command line; 2628 2629 kprobe_event=p,vfs_read,$arg1,$arg2 2630 2631 See also Documentation/trace/kprobetrace.rst "Kernel 2632 Boot Parameter" section. 2633 2634 kpti= [ARM64,EARLY] Control page table isolation of 2635 user and kernel address spaces. 2636 Default: enabled on cores which need mitigation. 2637 0: force disabled 2638 1: force enabled 2639 2640 kunit.enable= [KUNIT] Enable executing KUnit tests. Requires 2641 CONFIG_KUNIT to be set to be fully enabled. The 2642 default value can be overridden via 2643 KUNIT_DEFAULT_ENABLED. 2644 Default is 1 (enabled) 2645 2646 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs. 2647 Default is 0 (don't ignore, but inject #GP) 2648 2649 kvm.eager_page_split= 2650 [KVM,X86] Controls whether or not KVM will try to 2651 proactively split all huge pages during dirty logging. 2652 Eager page splitting reduces interruptions to vCPU 2653 execution by eliminating the write-protection faults 2654 and MMU lock contention that would otherwise be 2655 required to split huge pages lazily. 2656 2657 VM workloads that rarely perform writes or that write 2658 only to a small region of VM memory may benefit from 2659 disabling eager page splitting to allow huge pages to 2660 still be used for reads. 2661 2662 The behavior of eager page splitting depends on whether 2663 KVM_DIRTY_LOG_INITIALLY_SET is enabled or disabled. If 2664 disabled, all huge pages in a memslot will be eagerly 2665 split when dirty logging is enabled on that memslot. If 2666 enabled, eager page splitting will be performed during 2667 the KVM_CLEAR_DIRTY ioctl, and only for the pages being 2668 cleared. 2669 2670 Eager page splitting is only supported when kvm.tdp_mmu=Y. 2671 2672 Default is Y (on). 2673 2674 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface. 2675 Default is false (don't support). 2676 2677 kvm.nx_huge_pages= 2678 [KVM] Controls the software workaround for the 2679 X86_BUG_ITLB_MULTIHIT bug. 2680 force : Always deploy workaround. 2681 off : Never deploy workaround. 2682 auto : Deploy workaround based on the presence of 2683 X86_BUG_ITLB_MULTIHIT. 2684 2685 Default is 'auto'. 2686 2687 If the software workaround is enabled for the host, 2688 guests do need not to enable it for nested guests. 2689 2690 kvm.nx_huge_pages_recovery_ratio= 2691 [KVM] Controls how many 4KiB pages are periodically zapped 2692 back to huge pages. 0 disables the recovery, otherwise if 2693 the value is N KVM will zap 1/Nth of the 4KiB pages every 2694 period (see below). The default is 60. 2695 2696 kvm.nx_huge_pages_recovery_period_ms= 2697 [KVM] Controls the time period at which KVM zaps 4KiB pages 2698 back to huge pages. If the value is a non-zero N, KVM will 2699 zap a portion (see ratio above) of the pages every N msecs. 2700 If the value is 0 (the default), KVM will pick a period based 2701 on the ratio, such that a page is zapped after 1 hour on average. 2702 2703 kvm-amd.nested= [KVM,AMD] Control nested virtualization feature in 2704 KVM/SVM. Default is 1 (enabled). 2705 2706 kvm-amd.npt= [KVM,AMD] Control KVM's use of Nested Page Tables, 2707 a.k.a. Two-Dimensional Page Tables. Default is 1 2708 (enabled). Disable by KVM if hardware lacks support 2709 for NPT. 2710 2711 kvm-arm.mode= 2712 [KVM,ARM,EARLY] Select one of KVM/arm64's modes of 2713 operation. 2714 2715 none: Forcefully disable KVM. 2716 2717 nvhe: Standard nVHE-based mode, without support for 2718 protected guests. 2719 2720 protected: nVHE-based mode with support for guests whose 2721 state is kept private from the host. 2722 2723 nested: VHE-based mode with support for nested 2724 virtualization. Requires at least ARMv8.3 2725 hardware. 2726 2727 Defaults to VHE/nVHE based on hardware support. Setting 2728 mode to "protected" will disable kexec and hibernation 2729 for the host. "nested" is experimental and should be 2730 used with extreme caution. 2731 2732 kvm-arm.vgic_v3_group0_trap= 2733 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-0 2734 system registers 2735 2736 kvm-arm.vgic_v3_group1_trap= 2737 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-1 2738 system registers 2739 2740 kvm-arm.vgic_v3_common_trap= 2741 [KVM,ARM,EARLY] Trap guest accesses to GICv3 common 2742 system registers 2743 2744 kvm-arm.vgic_v4_enable= 2745 [KVM,ARM,EARLY] Allow use of GICv4 for direct 2746 injection of LPIs. 2747 2748 kvm_cma_resv_ratio=n [PPC,EARLY] 2749 Reserves given percentage from system memory area for 2750 contiguous memory allocation for KVM hash pagetable 2751 allocation. 2752 By default it reserves 5% of total system memory. 2753 Format: <integer> 2754 Default: 5 2755 2756 kvm-intel.ept= [KVM,Intel] Control KVM's use of Extended Page Tables, 2757 a.k.a. Two-Dimensional Page Tables. Default is 1 2758 (enabled). Disable by KVM if hardware lacks support 2759 for EPT. 2760 2761 kvm-intel.emulate_invalid_guest_state= 2762 [KVM,Intel] Control whether to emulate invalid guest 2763 state. Ignored if kvm-intel.enable_unrestricted_guest=1, 2764 as guest state is never invalid for unrestricted 2765 guests. This param doesn't apply to nested guests (L2), 2766 as KVM never emulates invalid L2 guest state. 2767 Default is 1 (enabled). 2768 2769 kvm-intel.flexpriority= 2770 [KVM,Intel] Control KVM's use of FlexPriority feature 2771 (TPR shadow). Default is 1 (enabled). Disable by KVM if 2772 hardware lacks support for it. 2773 2774 kvm-intel.nested= 2775 [KVM,Intel] Control nested virtualization feature in 2776 KVM/VMX. Default is 1 (enabled). 2777 2778 kvm-intel.unrestricted_guest= 2779 [KVM,Intel] Control KVM's use of unrestricted guest 2780 feature (virtualized real and unpaged mode). Default 2781 is 1 (enabled). Disable by KVM if EPT is disabled or 2782 hardware lacks support for it. 2783 2784 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault 2785 CVE-2018-3620. 2786 2787 Valid arguments: never, cond, always 2788 2789 always: L1D cache flush on every VMENTER. 2790 cond: Flush L1D on VMENTER only when the code between 2791 VMEXIT and VMENTER can leak host memory. 2792 never: Disables the mitigation 2793 2794 Default is cond (do L1 cache flush in specific instances) 2795 2796 kvm-intel.vpid= [KVM,Intel] Control KVM's use of Virtual Processor 2797 Identification feature (tagged TLBs). Default is 1 2798 (enabled). Disable by KVM if hardware lacks support 2799 for it. 2800 2801 l1d_flush= [X86,INTEL,EARLY] 2802 Control mitigation for L1D based snooping vulnerability. 2803 2804 Certain CPUs are vulnerable to an exploit against CPU 2805 internal buffers which can forward information to a 2806 disclosure gadget under certain conditions. 2807 2808 In vulnerable processors, the speculatively 2809 forwarded data can be used in a cache side channel 2810 attack, to access data to which the attacker does 2811 not have direct access. 2812 2813 This parameter controls the mitigation. The 2814 options are: 2815 2816 on - enable the interface for the mitigation 2817 2818 l1tf= [X86,EARLY] Control mitigation of the L1TF vulnerability on 2819 affected CPUs 2820 2821 The kernel PTE inversion protection is unconditionally 2822 enabled and cannot be disabled. 2823 2824 full 2825 Provides all available mitigations for the 2826 L1TF vulnerability. Disables SMT and 2827 enables all mitigations in the 2828 hypervisors, i.e. unconditional L1D flush. 2829 2830 SMT control and L1D flush control via the 2831 sysfs interface is still possible after 2832 boot. Hypervisors will issue a warning 2833 when the first VM is started in a 2834 potentially insecure configuration, 2835 i.e. SMT enabled or L1D flush disabled. 2836 2837 full,force 2838 Same as 'full', but disables SMT and L1D 2839 flush runtime control. Implies the 2840 'nosmt=force' command line option. 2841 (i.e. sysfs control of SMT is disabled.) 2842 2843 flush 2844 Leaves SMT enabled and enables the default 2845 hypervisor mitigation, i.e. conditional 2846 L1D flush. 2847 2848 SMT control and L1D flush control via the 2849 sysfs interface is still possible after 2850 boot. Hypervisors will issue a warning 2851 when the first VM is started in a 2852 potentially insecure configuration, 2853 i.e. SMT enabled or L1D flush disabled. 2854 2855 flush,nosmt 2856 2857 Disables SMT and enables the default 2858 hypervisor mitigation. 2859 2860 SMT control and L1D flush control via the 2861 sysfs interface is still possible after 2862 boot. Hypervisors will issue a warning 2863 when the first VM is started in a 2864 potentially insecure configuration, 2865 i.e. SMT enabled or L1D flush disabled. 2866 2867 flush,nowarn 2868 Same as 'flush', but hypervisors will not 2869 warn when a VM is started in a potentially 2870 insecure configuration. 2871 2872 off 2873 Disables hypervisor mitigations and doesn't 2874 emit any warnings. 2875 It also drops the swap size and available 2876 RAM limit restriction on both hypervisor and 2877 bare metal. 2878 2879 Default is 'flush'. 2880 2881 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst 2882 2883 l2cr= [PPC] 2884 2885 l3cr= [PPC] 2886 2887 lapic [X86-32,APIC,EARLY] Enable the local APIC even if BIOS 2888 disabled it. 2889 2890 lapic= [X86,APIC] Do not use TSC deadline 2891 value for LAPIC timer one-shot implementation. Default 2892 back to the programmable timer unit in the LAPIC. 2893 Format: notscdeadline 2894 2895 lapic_timer_c2_ok [X86,APIC,EARLY] trust the local apic timer 2896 in C2 power state. 2897 2898 libata.dma= [LIBATA] DMA control 2899 libata.dma=0 Disable all PATA and SATA DMA 2900 libata.dma=1 PATA and SATA Disk DMA only 2901 libata.dma=2 ATAPI (CDROM) DMA only 2902 libata.dma=4 Compact Flash DMA only 2903 Combinations also work, so libata.dma=3 enables DMA 2904 for disks and CDROMs, but not CFs. 2905 2906 libata.ignore_hpa= [LIBATA] Ignore HPA limit 2907 libata.ignore_hpa=0 keep BIOS limits (default) 2908 libata.ignore_hpa=1 ignore limits, using full disk 2909 2910 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume 2911 when set. 2912 Format: <int> 2913 2914 libata.force= [LIBATA] Force configurations. The format is a comma- 2915 separated list of "[ID:]VAL" where ID is PORT[.DEVICE]. 2916 PORT and DEVICE are decimal numbers matching port, link 2917 or device. Basically, it matches the ATA ID string 2918 printed on console by libata. If the whole ID part is 2919 omitted, the last PORT and DEVICE values are used. If 2920 ID hasn't been specified yet, the configuration applies 2921 to all ports, links and devices. 2922 2923 If only DEVICE is omitted, the parameter applies to 2924 the port and all links and devices behind it. DEVICE 2925 number of 0 either selects the first device or the 2926 first fan-out link behind PMP device. It does not 2927 select the host link. DEVICE number of 15 selects the 2928 host link and device attached to it. 2929 2930 The VAL specifies the configuration to force. As long 2931 as there is no ambiguity, shortcut notation is allowed. 2932 For example, both 1.5 and 1.5G would work for 1.5Gbps. 2933 The following configurations can be forced. 2934 2935 * Cable type: 40c, 80c, short40c, unk, ign or sata. 2936 Any ID with matching PORT is used. 2937 2938 * SATA link speed limit: 1.5Gbps or 3.0Gbps. 2939 2940 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7]. 2941 udma[/][16,25,33,44,66,100,133] notation is also 2942 allowed. 2943 2944 * nohrst, nosrst, norst: suppress hard, soft and both 2945 resets. 2946 2947 * rstonce: only attempt one reset during hot-unplug 2948 link recovery. 2949 2950 * [no]dbdelay: Enable or disable the extra 200ms delay 2951 before debouncing a link PHY and device presence 2952 detection. 2953 2954 * [no]ncq: Turn on or off NCQ. 2955 2956 * [no]ncqtrim: Enable or disable queued DSM TRIM. 2957 2958 * [no]ncqati: Enable or disable NCQ trim on ATI chipset. 2959 2960 * [no]trim: Enable or disable (unqueued) TRIM. 2961 2962 * trim_zero: Indicate that TRIM command zeroes data. 2963 2964 * max_trim_128m: Set 128M maximum trim size limit. 2965 2966 * [no]dma: Turn on or off DMA transfers. 2967 2968 * atapi_dmadir: Enable ATAPI DMADIR bridge support. 2969 2970 * atapi_mod16_dma: Enable the use of ATAPI DMA for 2971 commands that are not a multiple of 16 bytes. 2972 2973 * [no]dmalog: Enable or disable the use of the 2974 READ LOG DMA EXT command to access logs. 2975 2976 * [no]iddevlog: Enable or disable access to the 2977 identify device data log. 2978 2979 * [no]logdir: Enable or disable access to the general 2980 purpose log directory. 2981 2982 * max_sec_128: Set transfer size limit to 128 sectors. 2983 2984 * max_sec_1024: Set or clear transfer size limit to 2985 1024 sectors. 2986 2987 * max_sec_lba48: Set or clear transfer size limit to 2988 65535 sectors. 2989 2990 * [no]lpm: Enable or disable link power management. 2991 2992 * [no]setxfer: Indicate if transfer speed mode setting 2993 should be skipped. 2994 2995 * [no]fua: Disable or enable FUA (Force Unit Access) 2996 support for devices supporting this feature. 2997 2998 * dump_id: Dump IDENTIFY data. 2999 3000 * disable: Disable this device. 3001 3002 If there are multiple matching configurations changing 3003 the same attribute, the last one is used. 3004 3005 load_ramdisk= [RAM] [Deprecated] 3006 3007 lockd.nlm_grace_period=P [NFS] Assign grace period. 3008 Format: <integer> 3009 3010 lockd.nlm_tcpport=N [NFS] Assign TCP port. 3011 Format: <integer> 3012 3013 lockd.nlm_timeout=T [NFS] Assign timeout value. 3014 Format: <integer> 3015 3016 lockd.nlm_udpport=M [NFS] Assign UDP port. 3017 Format: <integer> 3018 3019 lockdown= [SECURITY,EARLY] 3020 { integrity | confidentiality } 3021 Enable the kernel lockdown feature. If set to 3022 integrity, kernel features that allow userland to 3023 modify the running kernel are disabled. If set to 3024 confidentiality, kernel features that allow userland 3025 to extract confidential information from the kernel 3026 are also disabled. 3027 3028 locktorture.acq_writer_lim= [KNL] 3029 Set the time limit in jiffies for a lock 3030 acquisition. Acquisitions exceeding this limit 3031 will result in a splat once they do complete. 3032 3033 locktorture.bind_readers= [KNL] 3034 Specify the list of CPUs to which the readers are 3035 to be bound. 3036 3037 locktorture.bind_writers= [KNL] 3038 Specify the list of CPUs to which the writers are 3039 to be bound. 3040 3041 locktorture.call_rcu_chains= [KNL] 3042 Specify the number of self-propagating call_rcu() 3043 chains to set up. These are used to ensure that 3044 there is a high probability of an RCU grace period 3045 in progress at any given time. Defaults to 0, 3046 which disables these call_rcu() chains. 3047 3048 locktorture.long_hold= [KNL] 3049 Specify the duration in milliseconds for the 3050 occasional long-duration lock hold time. Defaults 3051 to 100 milliseconds. Select 0 to disable. 3052 3053 locktorture.nested_locks= [KNL] 3054 Specify the maximum lock nesting depth that 3055 locktorture is to exercise, up to a limit of 8 3056 (MAX_NESTED_LOCKS). Specify zero to disable. 3057 Note that this parameter is ineffective on types 3058 of locks that do not support nested acquisition. 3059 3060 locktorture.nreaders_stress= [KNL] 3061 Set the number of locking read-acquisition kthreads. 3062 Defaults to being automatically set based on the 3063 number of online CPUs. 3064 3065 locktorture.nwriters_stress= [KNL] 3066 Set the number of locking write-acquisition kthreads. 3067 3068 locktorture.onoff_holdoff= [KNL] 3069 Set time (s) after boot for CPU-hotplug testing. 3070 3071 locktorture.onoff_interval= [KNL] 3072 Set time (s) between CPU-hotplug operations, or 3073 zero to disable CPU-hotplug testing. 3074 3075 locktorture.rt_boost= [KNL] 3076 Do periodic testing of real-time lock priority 3077 boosting. Select 0 to disable, 1 to boost 3078 only rt_mutex, and 2 to boost unconditionally. 3079 Defaults to 2, which might seem to be an 3080 odd choice, but which should be harmless for 3081 non-real-time spinlocks, due to their disabling 3082 of preemption. Note that non-realtime mutexes 3083 disable boosting. 3084 3085 locktorture.rt_boost_factor= [KNL] 3086 Number that determines how often and for how 3087 long priority boosting is exercised. This is 3088 scaled down by the number of writers, so that the 3089 number of boosts per unit time remains roughly 3090 constant as the number of writers increases. 3091 On the other hand, the duration of each boost 3092 increases with the number of writers. 3093 3094 locktorture.shuffle_interval= [KNL] 3095 Set task-shuffle interval (jiffies). Shuffling 3096 tasks allows some CPUs to go into dyntick-idle 3097 mode during the locktorture test. 3098 3099 locktorture.shutdown_secs= [KNL] 3100 Set time (s) after boot system shutdown. This 3101 is useful for hands-off automated testing. 3102 3103 locktorture.stat_interval= [KNL] 3104 Time (s) between statistics printk()s. 3105 3106 locktorture.stutter= [KNL] 3107 Time (s) to stutter testing, for example, 3108 specifying five seconds causes the test to run for 3109 five seconds, wait for five seconds, and so on. 3110 This tests the locking primitive's ability to 3111 transition abruptly to and from idle. 3112 3113 locktorture.torture_type= [KNL] 3114 Specify the locking implementation to test. 3115 3116 locktorture.verbose= [KNL] 3117 Enable additional printk() statements. 3118 3119 locktorture.writer_fifo= [KNL] 3120 Run the write-side locktorture kthreads at 3121 sched_set_fifo() real-time priority. 3122 3123 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver 3124 Format: <irq> 3125 3126 loglevel= [KNL,EARLY] 3127 All Kernel Messages with a loglevel smaller than the 3128 console loglevel will be printed to the console. It can 3129 also be changed with klogd or other programs. The 3130 loglevels are defined as follows: 3131 3132 0 (KERN_EMERG) system is unusable 3133 1 (KERN_ALERT) action must be taken immediately 3134 2 (KERN_CRIT) critical conditions 3135 3 (KERN_ERR) error conditions 3136 4 (KERN_WARNING) warning conditions 3137 5 (KERN_NOTICE) normal but significant condition 3138 6 (KERN_INFO) informational 3139 7 (KERN_DEBUG) debug-level messages 3140 3141 log_buf_len=n[KMG] [KNL,EARLY] 3142 Sets the size of the printk ring buffer, in bytes. 3143 n must be a power of two and greater than the 3144 minimal size. The minimal size is defined by 3145 LOG_BUF_SHIFT kernel config parameter. There 3146 is also CONFIG_LOG_CPU_MAX_BUF_SHIFT config 3147 parameter that allows to increase the default size 3148 depending on the number of CPUs. See init/Kconfig 3149 for more details. 3150 3151 logo.nologo [FB] Disables display of the built-in Linux logo. 3152 This may be used to provide more screen space for 3153 kernel log messages and is useful when debugging 3154 kernel boot problems. 3155 3156 lp=0 [LP] Specify parallel ports to use, e.g, 3157 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses 3158 lp=reset first parallel port). 'lp=0' disables the 3159 lp=auto printer driver. 'lp=reset' (which can be 3160 specified in addition to the ports) causes 3161 attached printers to be reset. Using 3162 lp=port1,port2,... specifies the parallel ports 3163 to associate lp devices with, starting with 3164 lp0. A port specification may be 'none' to skip 3165 that lp device, or a parport name such as 3166 'parport0'. Specifying 'lp=auto' instead of a 3167 port specification list means that device IDs 3168 from each port should be examined, to see if 3169 an IEEE 1284-compliant printer is attached; if 3170 so, the driver will manage that printer. 3171 See also header of drivers/char/lp.c. 3172 3173 lpj=n [KNL] 3174 Sets loops_per_jiffy to given constant, thus avoiding 3175 time-consuming boot-time autodetection (up to 250 ms per 3176 CPU). 0 enables autodetection (default). To determine 3177 the correct value for your kernel, boot with normal 3178 autodetection and see what value is printed. Note that 3179 on SMP systems the preset will be applied to all CPUs, 3180 which is likely to cause problems if your CPUs need 3181 significantly divergent settings. An incorrect value 3182 will cause delays in the kernel to be wrong, leading to 3183 unpredictable I/O errors and other breakage. Although 3184 unlikely, in the extreme case this might damage your 3185 hardware. 3186 3187 ltpc= [NET] 3188 Format: <io>,<irq>,<dma> 3189 3190 lsm.debug [SECURITY] Enable LSM initialization debugging output. 3191 3192 lsm=lsm1,...,lsmN 3193 [SECURITY] Choose order of LSM initialization. This 3194 overrides CONFIG_LSM, and the "security=" parameter. 3195 3196 machvec= [IA-64] Force the use of a particular machine-vector 3197 (machvec) in a generic kernel. 3198 Example: machvec=hpzx1 3199 3200 machtype= [Loongson] Share the same kernel image file between 3201 different yeeloong laptops. 3202 Example: machtype=lemote-yeeloong-2f-7inch 3203 3204 max_addr=nn[KMG] [KNL,BOOT,IA-64] All physical memory greater 3205 than or equal to this physical address is ignored. 3206 3207 maxcpus= [SMP,EARLY] Maximum number of processors that an SMP kernel 3208 will bring up during bootup. maxcpus=n : n >= 0 limits 3209 the kernel to bring up 'n' processors. Surely after 3210 bootup you can bring up the other plugged cpu by executing 3211 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus 3212 only takes effect during system bootup. 3213 While n=0 is a special case, it is equivalent to "nosmp", 3214 which also disables the IO APIC. 3215 3216 max_loop= [LOOP] The number of loop block devices that get 3217 (loop.max_loop) unconditionally pre-created at init time. The default 3218 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead 3219 of statically allocating a predefined number, loop 3220 devices can be requested on-demand with the 3221 /dev/loop-control interface. 3222 3223 mce [X86-32] Machine Check Exception 3224 3225 mce=option [X86-64] See Documentation/arch/x86/x86_64/boot-options.rst 3226 3227 md= [HW] RAID subsystems devices and level 3228 See Documentation/admin-guide/md.rst. 3229 3230 mdacon= [MDA] 3231 Format: <first>,<last> 3232 Specifies range of consoles to be captured by the MDA. 3233 3234 mds= [X86,INTEL,EARLY] 3235 Control mitigation for the Micro-architectural Data 3236 Sampling (MDS) vulnerability. 3237 3238 Certain CPUs are vulnerable to an exploit against CPU 3239 internal buffers which can forward information to a 3240 disclosure gadget under certain conditions. 3241 3242 In vulnerable processors, the speculatively 3243 forwarded data can be used in a cache side channel 3244 attack, to access data to which the attacker does 3245 not have direct access. 3246 3247 This parameter controls the MDS mitigation. The 3248 options are: 3249 3250 full - Enable MDS mitigation on vulnerable CPUs 3251 full,nosmt - Enable MDS mitigation and disable 3252 SMT on vulnerable CPUs 3253 off - Unconditionally disable MDS mitigation 3254 3255 On TAA-affected machines, mds=off can be prevented by 3256 an active TAA mitigation as both vulnerabilities are 3257 mitigated with the same mechanism so in order to disable 3258 this mitigation, you need to specify tsx_async_abort=off 3259 too. 3260 3261 Not specifying this option is equivalent to 3262 mds=full. 3263 3264 For details see: Documentation/admin-guide/hw-vuln/mds.rst 3265 3266 mem=nn[KMG] [HEXAGON,EARLY] Set the memory size. 3267 Must be specified, otherwise memory size will be 0. 3268 3269 mem=nn[KMG] [KNL,BOOT,EARLY] Force usage of a specific amount 3270 of memory Amount of memory to be used in cases 3271 as follows: 3272 3273 1 for test; 3274 2 when the kernel is not able to see the whole system memory; 3275 3 memory that lies after 'mem=' boundary is excluded from 3276 the hypervisor, then assigned to KVM guests. 3277 4 to limit the memory available for kdump kernel. 3278 3279 [ARC,MICROBLAZE] - the limit applies only to low memory, 3280 high memory is not affected. 3281 3282 [ARM64] - only limits memory covered by the linear 3283 mapping. The NOMAP regions are not affected. 3284 3285 [X86] Work as limiting max address. Use together 3286 with memmap= to avoid physical address space collisions. 3287 Without memmap= PCI devices could be placed at addresses 3288 belonging to unused RAM. 3289 3290 Note that this only takes effects during boot time since 3291 in above case 3, memory may need be hot added after boot 3292 if system memory of hypervisor is not sufficient. 3293 3294 mem=nn[KMG]@ss[KMG] 3295 [ARM,MIPS,EARLY] - override the memory layout 3296 reported by firmware. 3297 Define a memory region of size nn[KMG] starting at 3298 ss[KMG]. 3299 Multiple different regions can be specified with 3300 multiple mem= parameters on the command line. 3301 3302 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel 3303 memory. 3304 3305 memblock=debug [KNL,EARLY] Enable memblock debug messages. 3306 3307 memchunk=nn[KMG] 3308 [KNL,SH] Allow user to override the default size for 3309 per-device physically contiguous DMA buffers. 3310 3311 memhp_default_state=online/offline/online_kernel/online_movable 3312 [KNL] Set the initial state for the memory hotplug 3313 onlining policy. If not specified, the default value is 3314 set according to the 3315 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config 3316 option. 3317 See Documentation/admin-guide/mm/memory-hotplug.rst. 3318 3319 memmap=exactmap [KNL,X86,EARLY] Enable setting of an exact 3320 E820 memory map, as specified by the user. 3321 Such memmap=exactmap lines can be constructed based on 3322 BIOS output or other requirements. See the memmap=nn@ss 3323 option description. 3324 3325 memmap=nn[KMG]@ss[KMG] 3326 [KNL, X86,MIPS,XTENSA,EARLY] Force usage of a specific region of memory. 3327 Region of memory to be used is from ss to ss+nn. 3328 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG], 3329 which limits max address to nn[KMG]. 3330 Multiple different regions can be specified, 3331 comma delimited. 3332 Example: 3333 memmap=100M@2G,100M#3G,1G!1024G 3334 3335 memmap=nn[KMG]#ss[KMG] 3336 [KNL,ACPI,EARLY] Mark specific memory as ACPI data. 3337 Region of memory to be marked is from ss to ss+nn. 3338 3339 memmap=nn[KMG]$ss[KMG] 3340 [KNL,ACPI,EARLY] Mark specific memory as reserved. 3341 Region of memory to be reserved is from ss to ss+nn. 3342 Example: Exclude memory from 0x18690000-0x1869ffff 3343 memmap=64K$0x18690000 3344 or 3345 memmap=0x10000$0x18690000 3346 Some bootloaders may need an escape character before '$', 3347 like Grub2, otherwise '$' and the following number 3348 will be eaten. 3349 3350 memmap=nn[KMG]!ss[KMG,EARLY] 3351 [KNL,X86] Mark specific memory as protected. 3352 Region of memory to be used, from ss to ss+nn. 3353 The memory region may be marked as e820 type 12 (0xc) 3354 and is NVDIMM or ADR memory. 3355 3356 memmap=<size>%<offset>-<oldtype>+<newtype> 3357 [KNL,ACPI,EARLY] Convert memory within the specified region 3358 from <oldtype> to <newtype>. If "-<oldtype>" is left 3359 out, the whole region will be marked as <newtype>, 3360 even if previously unavailable. If "+<newtype>" is left 3361 out, matching memory will be removed. Types are 3362 specified as e820 types, e.g., 1 = RAM, 2 = reserved, 3363 3 = ACPI, 12 = PRAM. 3364 3365 memory_corruption_check=0/1 [X86,EARLY] 3366 Some BIOSes seem to corrupt the first 64k of 3367 memory when doing things like suspend/resume. 3368 Setting this option will scan the memory 3369 looking for corruption. Enabling this will 3370 both detect corruption and prevent the kernel 3371 from using the memory being corrupted. 3372 However, its intended as a diagnostic tool; if 3373 repeatable BIOS-originated corruption always 3374 affects the same memory, you can use memmap= 3375 to prevent the kernel from using that memory. 3376 3377 memory_corruption_check_size=size [X86,EARLY] 3378 By default it checks for corruption in the low 3379 64k, making this memory unavailable for normal 3380 use. Use this parameter to scan for 3381 corruption in more or less memory. 3382 3383 memory_corruption_check_period=seconds [X86,EARLY] 3384 By default it checks for corruption every 60 3385 seconds. Use this parameter to check at some 3386 other rate. 0 disables periodic checking. 3387 3388 memory_hotplug.memmap_on_memory 3389 [KNL,X86,ARM] Boolean flag to enable this feature. 3390 Format: {on | off (default)} 3391 When enabled, runtime hotplugged memory will 3392 allocate its internal metadata (struct pages, 3393 those vmemmap pages cannot be optimized even 3394 if hugetlb_free_vmemmap is enabled) from the 3395 hotadded memory which will allow to hotadd a 3396 lot of memory without requiring additional 3397 memory to do so. 3398 This feature is disabled by default because it 3399 has some implication on large (e.g. GB) 3400 allocations in some configurations (e.g. small 3401 memory blocks). 3402 The state of the flag can be read in 3403 /sys/module/memory_hotplug/parameters/memmap_on_memory. 3404 Note that even when enabled, there are a few cases where 3405 the feature is not effective. 3406 3407 memtest= [KNL,X86,ARM,M68K,PPC,RISCV,EARLY] Enable memtest 3408 Format: <integer> 3409 default : 0 <disable> 3410 Specifies the number of memtest passes to be 3411 performed. Each pass selects another test 3412 pattern from a given set of patterns. Memtest 3413 fills the memory with this pattern, validates 3414 memory contents and reserves bad memory 3415 regions that are detected. 3416 3417 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control 3418 Valid arguments: on, off 3419 Default: off 3420 mem_encrypt=on: Activate SME 3421 mem_encrypt=off: Do not activate SME 3422 3423 Refer to Documentation/virt/kvm/x86/amd-memory-encryption.rst 3424 for details on when memory encryption can be activated. 3425 3426 mem_sleep_default= [SUSPEND] Default system suspend mode: 3427 s2idle - Suspend-To-Idle 3428 shallow - Power-On Suspend or equivalent (if supported) 3429 deep - Suspend-To-RAM or equivalent (if supported) 3430 See Documentation/admin-guide/pm/sleep-states.rst. 3431 3432 mfgpt_irq= [IA-32] Specify the IRQ to use for the 3433 Multi-Function General Purpose Timers on AMD Geode 3434 platforms. 3435 3436 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when 3437 the BIOS has incorrectly applied a workaround. TinyBIOS 3438 version 0.98 is known to be affected, 0.99 fixes the 3439 problem by letting the user disable the workaround. 3440 3441 mga= [HW,DRM] 3442 3443 microcode.force_minrev= [X86] 3444 Format: <bool> 3445 Enable or disable the microcode minimal revision 3446 enforcement for the runtime microcode loader. 3447 3448 min_addr=nn[KMG] [KNL,BOOT,IA-64] All physical memory below this 3449 physical address is ignored. 3450 3451 mini2440= [ARM,HW,KNL] 3452 Format:[0..2][b][c][t] 3453 Default: "0tb" 3454 MINI2440 configuration specification: 3455 0 - The attached screen is the 3.5" TFT 3456 1 - The attached screen is the 7" TFT 3457 2 - The VGA Shield is attached (1024x768) 3458 Leaving out the screen size parameter will not load 3459 the TFT driver, and the framebuffer will be left 3460 unconfigured. 3461 b - Enable backlight. The TFT backlight pin will be 3462 linked to the kernel VESA blanking code and a GPIO 3463 LED. This parameter is not necessary when using the 3464 VGA shield. 3465 c - Enable the s3c camera interface. 3466 t - Reserved for enabling touchscreen support. The 3467 touchscreen support is not enabled in the mainstream 3468 kernel as of 2.6.30, a preliminary port can be found 3469 in the "bleeding edge" mini2440 support kernel at 3470 https://repo.or.cz/w/linux-2.6/mini2440.git 3471 3472 mitigations= 3473 [X86,PPC,S390,ARM64,EARLY] Control optional mitigations for 3474 CPU vulnerabilities. This is a set of curated, 3475 arch-independent options, each of which is an 3476 aggregation of existing arch-specific options. 3477 3478 Note, "mitigations" is supported if and only if the 3479 kernel was built with CPU_MITIGATIONS=y. 3480 3481 off 3482 Disable all optional CPU mitigations. This 3483 improves system performance, but it may also 3484 expose users to several CPU vulnerabilities. 3485 Equivalent to: if nokaslr then kpti=0 [ARM64] 3486 gather_data_sampling=off [X86] 3487 kvm.nx_huge_pages=off [X86] 3488 l1tf=off [X86] 3489 mds=off [X86] 3490 mmio_stale_data=off [X86] 3491 no_entry_flush [PPC] 3492 no_uaccess_flush [PPC] 3493 nobp=0 [S390] 3494 nopti [X86,PPC] 3495 nospectre_bhb [ARM64] 3496 nospectre_v1 [X86,PPC] 3497 nospectre_v2 [X86,PPC,S390,ARM64] 3498 reg_file_data_sampling=off [X86] 3499 retbleed=off [X86] 3500 spec_rstack_overflow=off [X86] 3501 spec_store_bypass_disable=off [X86,PPC] 3502 spectre_bhi=off [X86] 3503 spectre_v2_user=off [X86] 3504 srbds=off [X86,INTEL] 3505 ssbd=force-off [ARM64] 3506 tsx_async_abort=off [X86] 3507 3508 Exceptions: 3509 This does not have any effect on 3510 kvm.nx_huge_pages when 3511 kvm.nx_huge_pages=force. 3512 3513 auto (default) 3514 Mitigate all CPU vulnerabilities, but leave SMT 3515 enabled, even if it's vulnerable. This is for 3516 users who don't want to be surprised by SMT 3517 getting disabled across kernel upgrades, or who 3518 have other ways of avoiding SMT-based attacks. 3519 Equivalent to: (default behavior) 3520 3521 auto,nosmt 3522 Mitigate all CPU vulnerabilities, disabling SMT 3523 if needed. This is for users who always want to 3524 be fully mitigated, even if it means losing SMT. 3525 Equivalent to: l1tf=flush,nosmt [X86] 3526 mds=full,nosmt [X86] 3527 tsx_async_abort=full,nosmt [X86] 3528 mmio_stale_data=full,nosmt [X86] 3529 retbleed=auto,nosmt [X86] 3530 3531 mminit_loglevel= 3532 [KNL,EARLY] When CONFIG_DEBUG_MEMORY_INIT is set, this 3533 parameter allows control of the logging verbosity for 3534 the additional memory initialisation checks. A value 3535 of 0 disables mminit logging and a level of 4 will 3536 log everything. Information is printed at KERN_DEBUG 3537 so loglevel=8 may also need to be specified. 3538 3539 mmio_stale_data= 3540 [X86,INTEL,EARLY] Control mitigation for the Processor 3541 MMIO Stale Data vulnerabilities. 3542 3543 Processor MMIO Stale Data is a class of 3544 vulnerabilities that may expose data after an MMIO 3545 operation. Exposed data could originate or end in 3546 the same CPU buffers as affected by MDS and TAA. 3547 Therefore, similar to MDS and TAA, the mitigation 3548 is to clear the affected CPU buffers. 3549 3550 This parameter controls the mitigation. The 3551 options are: 3552 3553 full - Enable mitigation on vulnerable CPUs 3554 3555 full,nosmt - Enable mitigation and disable SMT on 3556 vulnerable CPUs. 3557 3558 off - Unconditionally disable mitigation 3559 3560 On MDS or TAA affected machines, 3561 mmio_stale_data=off can be prevented by an active 3562 MDS or TAA mitigation as these vulnerabilities are 3563 mitigated with the same mechanism so in order to 3564 disable this mitigation, you need to specify 3565 mds=off and tsx_async_abort=off too. 3566 3567 Not specifying this option is equivalent to 3568 mmio_stale_data=full. 3569 3570 For details see: 3571 Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst 3572 3573 <module>.async_probe[=<bool>] [KNL] 3574 If no <bool> value is specified or if the value 3575 specified is not a valid <bool>, enable asynchronous 3576 probe on this module. Otherwise, enable/disable 3577 asynchronous probe on this module as indicated by the 3578 <bool> value. See also: module.async_probe 3579 3580 module.async_probe=<bool> 3581 [KNL] When set to true, modules will use async probing 3582 by default. To enable/disable async probing for a 3583 specific module, use the module specific control that 3584 is documented under <module>.async_probe. When both 3585 module.async_probe and <module>.async_probe are 3586 specified, <module>.async_probe takes precedence for 3587 the specific module. 3588 3589 module.enable_dups_trace 3590 [KNL] When CONFIG_MODULE_DEBUG_AUTOLOAD_DUPS is set, 3591 this means that duplicate request_module() calls will 3592 trigger a WARN_ON() instead of a pr_warn(). Note that 3593 if MODULE_DEBUG_AUTOLOAD_DUPS_TRACE is set, WARN_ON()s 3594 will always be issued and this option does nothing. 3595 module.sig_enforce 3596 [KNL] When CONFIG_MODULE_SIG is set, this means that 3597 modules without (valid) signatures will fail to load. 3598 Note that if CONFIG_MODULE_SIG_FORCE is set, that 3599 is always true, so this option does nothing. 3600 3601 module_blacklist= [KNL] Do not load a comma-separated list of 3602 modules. Useful for debugging problem modules. 3603 3604 mousedev.tap_time= 3605 [MOUSE] Maximum time between finger touching and 3606 leaving touchpad surface for touch to be considered 3607 a tap and be reported as a left button click (for 3608 touchpads working in absolute mode only). 3609 Format: <msecs> 3610 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices 3611 reporting absolute coordinates, such as tablets 3612 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices 3613 reporting absolute coordinates, such as tablets 3614 3615 movablecore= [KNL,X86,IA-64,PPC,EARLY] 3616 Format: nn[KMGTPE] | nn% 3617 This parameter is the complement to kernelcore=, it 3618 specifies the amount of memory used for migratable 3619 allocations. If both kernelcore and movablecore is 3620 specified, then kernelcore will be at *least* the 3621 specified value but may be more. If movablecore on its 3622 own is specified, the administrator must be careful 3623 that the amount of memory usable for all allocations 3624 is not too small. 3625 3626 movable_node [KNL,EARLY] Boot-time switch to make hotplugable memory 3627 NUMA nodes to be movable. This means that the memory 3628 of such nodes will be usable only for movable 3629 allocations which rules out almost all kernel 3630 allocations. Use with caution! 3631 3632 MTD_Partition= [MTD] 3633 Format: <name>,<region-number>,<size>,<offset> 3634 3635 MTD_Region= [MTD] Format: 3636 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>] 3637 3638 mtdparts= [MTD] 3639 See drivers/mtd/parsers/cmdlinepart.c 3640 3641 mtdset= [ARM] 3642 ARM/S3C2412 JIVE boot control 3643 3644 See arch/arm/mach-s3c/mach-jive.c 3645 3646 mtouchusb.raw_coordinates= 3647 [HW] Make the MicroTouch USB driver use raw coordinates 3648 ('y', default) or cooked coordinates ('n') 3649 3650 mtrr=debug [X86,EARLY] 3651 Enable printing debug information related to MTRR 3652 registers at boot time. 3653 3654 mtrr_chunk_size=nn[KMG,X86,EARLY] 3655 used for mtrr cleanup. It is largest continuous chunk 3656 that could hold holes aka. UC entries. 3657 3658 mtrr_gran_size=nn[KMG,X86,EARLY] 3659 Used for mtrr cleanup. It is granularity of mtrr block. 3660 Default is 1. 3661 Large value could prevent small alignment from 3662 using up MTRRs. 3663 3664 mtrr_spare_reg_nr=n [X86,EARLY] 3665 Format: <integer> 3666 Range: 0,7 : spare reg number 3667 Default : 1 3668 Used for mtrr cleanup. It is spare mtrr entries number. 3669 Set to 2 or more if your graphical card needs more. 3670 3671 multitce=off [PPC] This parameter disables the use of the pSeries 3672 firmware feature for updating multiple TCE entries 3673 at a time. 3674 3675 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card 3676 3677 netdev= [NET] Network devices parameters 3678 Format: <irq>,<io>,<mem_start>,<mem_end>,<name> 3679 Note that mem_start is often overloaded to mean 3680 something different and driver-specific. 3681 This usage is only documented in each driver source 3682 file if at all. 3683 3684 netpoll.carrier_timeout= 3685 [NET] Specifies amount of time (in seconds) that 3686 netpoll should wait for a carrier. By default netpoll 3687 waits 4 seconds. 3688 3689 nf_conntrack.acct= 3690 [NETFILTER] Enable connection tracking flow accounting 3691 0 to disable accounting 3692 1 to enable accounting 3693 Default value is 0. 3694 3695 nfs.cache_getent= 3696 [NFS] sets the pathname to the program which is used 3697 to update the NFS client cache entries. 3698 3699 nfs.cache_getent_timeout= 3700 [NFS] sets the timeout after which an attempt to 3701 update a cache entry is deemed to have failed. 3702 3703 nfs.callback_nr_threads= 3704 [NFSv4] set the total number of threads that the 3705 NFS client will assign to service NFSv4 callback 3706 requests. 3707 3708 nfs.callback_tcpport= 3709 [NFS] set the TCP port on which the NFSv4 callback 3710 channel should listen. 3711 3712 nfs.delay_retrans= 3713 [NFS] specifies the number of times the NFSv4 client 3714 retries the request before returning an EAGAIN error, 3715 after a reply of NFS4ERR_DELAY from the server. 3716 Only applies if the softerr mount option is enabled, 3717 and the specified value is >= 0. 3718 3719 nfs.enable_ino64= 3720 [NFS] enable 64-bit inode numbers. 3721 If zero, the NFS client will fake up a 32-bit inode 3722 number for the readdir() and stat() syscalls instead 3723 of returning the full 64-bit number. 3724 The default is to return 64-bit inode numbers. 3725 3726 nfs.idmap_cache_timeout= 3727 [NFS] set the maximum lifetime for idmapper cache 3728 entries. 3729 3730 nfs.max_session_cb_slots= 3731 [NFSv4.1] Sets the maximum number of session 3732 slots the client will assign to the callback 3733 channel. This determines the maximum number of 3734 callbacks the client will process in parallel for 3735 a particular server. 3736 3737 nfs.max_session_slots= 3738 [NFSv4.1] Sets the maximum number of session slots 3739 the client will attempt to negotiate with the server. 3740 This limits the number of simultaneous RPC requests 3741 that the client can send to the NFSv4.1 server. 3742 Note that there is little point in setting this 3743 value higher than the max_tcp_slot_table_limit. 3744 3745 nfs.nfs4_disable_idmapping= 3746 [NFSv4] When set to the default of '1', this option 3747 ensures that both the RPC level authentication 3748 scheme and the NFS level operations agree to use 3749 numeric uids/gids if the mount is using the 3750 'sec=sys' security flavour. In effect it is 3751 disabling idmapping, which can make migration from 3752 legacy NFSv2/v3 systems to NFSv4 easier. 3753 Servers that do not support this mode of operation 3754 will be autodetected by the client, and it will fall 3755 back to using the idmapper. 3756 To turn off this behaviour, set the value to '0'. 3757 3758 nfs.nfs4_unique_id= 3759 [NFS4] Specify an additional fixed unique ident- 3760 ification string that NFSv4 clients can insert into 3761 their nfs_client_id4 string. This is typically a 3762 UUID that is generated at system install time. 3763 3764 nfs.recover_lost_locks= 3765 [NFSv4] Attempt to recover locks that were lost due 3766 to a lease timeout on the server. Please note that 3767 doing this risks data corruption, since there are 3768 no guarantees that the file will remain unchanged 3769 after the locks are lost. 3770 If you want to enable the kernel legacy behaviour of 3771 attempting to recover these locks, then set this 3772 parameter to '1'. 3773 The default parameter value of '0' causes the kernel 3774 not to attempt recovery of lost locks. 3775 3776 nfs.send_implementation_id= 3777 [NFSv4.1] Send client implementation identification 3778 information in exchange_id requests. 3779 If zero, no implementation identification information 3780 will be sent. 3781 The default is to send the implementation identification 3782 information. 3783 3784 nfs4.layoutstats_timer= 3785 [NFSv4.2] Change the rate at which the kernel sends 3786 layoutstats to the pNFS metadata server. 3787 3788 Setting this to value to 0 causes the kernel to use 3789 whatever value is the default set by the layout 3790 driver. A non-zero value sets the minimum interval 3791 in seconds between layoutstats transmissions. 3792 3793 nfsd.inter_copy_offload_enable= 3794 [NFSv4.2] When set to 1, the server will support 3795 server-to-server copies for which this server is 3796 the destination of the copy. 3797 3798 nfsd.nfs4_disable_idmapping= 3799 [NFSv4] When set to the default of '1', the NFSv4 3800 server will return only numeric uids and gids to 3801 clients using auth_sys, and will accept numeric uids 3802 and gids from such clients. This is intended to ease 3803 migration from NFSv2/v3. 3804 3805 nfsd.nfsd4_ssc_umount_timeout= 3806 [NFSv4.2] When used as the destination of a 3807 server-to-server copy, knfsd temporarily mounts 3808 the source server. It caches the mount in case 3809 it will be needed again, and discards it if not 3810 used for the number of milliseconds specified by 3811 this parameter. 3812 3813 nfsaddrs= [NFS] Deprecated. Use ip= instead. 3814 See Documentation/admin-guide/nfs/nfsroot.rst. 3815 3816 nfsroot= [NFS] nfs root filesystem for disk-less boxes. 3817 See Documentation/admin-guide/nfs/nfsroot.rst. 3818 3819 nfsrootdebug [NFS] enable nfsroot debugging messages. 3820 See Documentation/admin-guide/nfs/nfsroot.rst. 3821 3822 nmi_backtrace.backtrace_idle [KNL] 3823 Dump stacks even of idle CPUs in response to an 3824 NMI stack-backtrace request. 3825 3826 nmi_debug= [KNL,SH] Specify one or more actions to take 3827 when a NMI is triggered. 3828 Format: [state][,regs][,debounce][,die] 3829 3830 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels 3831 Format: [panic,][nopanic,][rNNN,][num] 3832 Valid num: 0 or 1 3833 0 - turn hardlockup detector in nmi_watchdog off 3834 1 - turn hardlockup detector in nmi_watchdog on 3835 rNNN - configure the watchdog with raw perf event 0xNNN 3836 3837 When panic is specified, panic when an NMI watchdog 3838 timeout occurs (or 'nopanic' to not panic on an NMI 3839 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set) 3840 To disable both hard and soft lockup detectors, 3841 please see 'nowatchdog'. 3842 This is useful when you use a panic=... timeout and 3843 need the box quickly up again. 3844 3845 These settings can be accessed at runtime via 3846 the nmi_watchdog and hardlockup_panic sysctls. 3847 3848 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths 3849 emulation library even if a 387 maths coprocessor 3850 is present. 3851 3852 no4lvl [RISCV,EARLY] Disable 4-level and 5-level paging modes. 3853 Forces kernel to use 3-level paging instead. 3854 3855 no5lvl [X86-64,RISCV,EARLY] Disable 5-level paging mode. Forces 3856 kernel to use 4-level paging instead. 3857 3858 noalign [KNL,ARM] 3859 3860 noaltinstr [S390,EARLY] Disables alternative instructions 3861 patching (CPU alternatives feature). 3862 3863 noapic [SMP,APIC,EARLY] Tells the kernel to not make use of any 3864 IOAPICs that may be present in the system. 3865 3866 noautogroup Disable scheduler automatic task group creation. 3867 3868 nocache [ARM,EARLY] 3869 3870 no_console_suspend 3871 [HW] Never suspend the console 3872 Disable suspending of consoles during suspend and 3873 hibernate operations. Once disabled, debugging 3874 messages can reach various consoles while the rest 3875 of the system is being put to sleep (ie, while 3876 debugging driver suspend/resume hooks). This may 3877 not work reliably with all consoles, but is known 3878 to work with serial and VGA consoles. 3879 To facilitate more flexible debugging, we also add 3880 console_suspend, a printk module parameter to control 3881 it. Users could use console_suspend (usually 3882 /sys/module/printk/parameters/console_suspend) to 3883 turn on/off it dynamically. 3884 3885 no_debug_objects 3886 [KNL,EARLY] Disable object debugging 3887 3888 nodsp [SH] Disable hardware DSP at boot time. 3889 3890 noefi [EFI,EARLY] Disable EFI runtime services support. 3891 3892 no_entry_flush [PPC,EARLY] Don't flush the L1-D cache when entering the kernel. 3893 3894 noexec [IA-64] 3895 3896 noexec32 [X86-64] 3897 This affects only 32-bit executables. 3898 noexec32=on: enable non-executable mappings (default) 3899 read doesn't imply executable mappings 3900 noexec32=off: disable non-executable mappings 3901 read implies executable mappings 3902 3903 no_file_caps Tells the kernel not to honor file capabilities. The 3904 only way then for a file to be executed with privilege 3905 is to be setuid root or executed by root. 3906 3907 nofpu [MIPS,SH] Disable hardware FPU at boot time. 3908 3909 nofsgsbase [X86] Disables FSGSBASE instructions. 3910 3911 nofxsr [BUGS=X86-32] Disables x86 floating point extended 3912 register save and restore. The kernel will only save 3913 legacy floating-point registers on task switch. 3914 3915 nohalt [IA-64] Tells the kernel not to use the power saving 3916 function PAL_HALT_LIGHT when idle. This increases 3917 power-consumption. On the positive side, it reduces 3918 interrupt wake-up latency, which may improve performance 3919 in certain environments such as networked servers or 3920 real-time systems. 3921 3922 no_hash_pointers 3923 [KNL,EARLY] 3924 Force pointers printed to the console or buffers to be 3925 unhashed. By default, when a pointer is printed via %p 3926 format string, that pointer is "hashed", i.e. obscured 3927 by hashing the pointer value. This is a security feature 3928 that hides actual kernel addresses from unprivileged 3929 users, but it also makes debugging the kernel more 3930 difficult since unequal pointers can no longer be 3931 compared. However, if this command-line option is 3932 specified, then all normal pointers will have their true 3933 value printed. This option should only be specified when 3934 debugging the kernel. Please do not use on production 3935 kernels. 3936 3937 nohibernate [HIBERNATION] Disable hibernation and resume. 3938 3939 nohlt [ARM,ARM64,MICROBLAZE,MIPS,PPC,SH] Forces the kernel to 3940 busy wait in do_idle() and not use the arch_cpu_idle() 3941 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP 3942 to be effective. This is useful on platforms where the 3943 sleep(SH) or wfi(ARM,ARM64) instructions do not work 3944 correctly or when doing power measurements to evaluate 3945 the impact of the sleep instructions. This is also 3946 useful when using JTAG debugger. 3947 3948 nohugeiomap [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge I/O mappings. 3949 3950 nohugevmalloc [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge vmalloc mappings. 3951 3952 nohz= [KNL] Boottime enable/disable dynamic ticks 3953 Valid arguments: on, off 3954 Default: on 3955 3956 nohz_full= [KNL,BOOT,SMP,ISOL] 3957 The argument is a cpu list, as described above. 3958 In kernels built with CONFIG_NO_HZ_FULL=y, set 3959 the specified list of CPUs whose tick will be stopped 3960 whenever possible. The boot CPU will be forced outside 3961 the range to maintain the timekeeping. Any CPUs 3962 in this list will have their RCU callbacks offloaded, 3963 just as if they had also been called out in the 3964 rcu_nocbs= boot parameter. 3965 3966 Note that this argument takes precedence over 3967 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option. 3968 3969 noinitrd [RAM] Tells the kernel not to load any configured 3970 initial RAM disk. 3971 3972 nointremap [X86-64,Intel-IOMMU,EARLY] Do not enable interrupt 3973 remapping. 3974 [Deprecated - use intremap=off] 3975 3976 nointroute [IA-64] 3977 3978 noinvpcid [X86,EARLY] Disable the INVPCID cpu feature. 3979 3980 noiotrap [SH] Disables trapped I/O port accesses. 3981 3982 noirqdebug [X86-32] Disables the code which attempts to detect and 3983 disable unhandled interrupt sources. 3984 3985 noisapnp [ISAPNP] Disables ISA PnP code. 3986 3987 nojitter [IA-64] Disables jitter checking for ITC timers. 3988 3989 nokaslr [KNL,EARLY] 3990 When CONFIG_RANDOMIZE_BASE is set, this disables 3991 kernel and module base offset ASLR (Address Space 3992 Layout Randomization). 3993 3994 no-kvmapf [X86,KVM,EARLY] Disable paravirtualized asynchronous page 3995 fault handling. 3996 3997 no-kvmclock [X86,KVM,EARLY] Disable paravirtualized KVM clock driver 3998 3999 nolapic [X86-32,APIC,EARLY] Do not enable or use the local APIC. 4000 4001 nolapic_timer [X86-32,APIC,EARLY] Do not use the local APIC timer. 4002 4003 nomca [IA-64] Disable machine check abort handling 4004 4005 nomce [X86-32] Disable Machine Check Exception 4006 4007 nomfgpt [X86-32] Disable Multi-Function General Purpose 4008 Timer usage (for AMD Geode machines). 4009 4010 nomodeset Disable kernel modesetting. Most systems' firmware 4011 sets up a display mode and provides framebuffer memory 4012 for output. With nomodeset, DRM and fbdev drivers will 4013 not load if they could possibly displace the pre- 4014 initialized output. Only the system framebuffer will 4015 be available for use. The respective drivers will not 4016 perform display-mode changes or accelerated rendering. 4017 4018 Useful as error fallback, or for testing and debugging. 4019 4020 nomodule Disable module load 4021 4022 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to 4023 shutdown the other cpus. Instead use the REBOOT_VECTOR 4024 irq. 4025 4026 nopat [X86,EARLY] Disable PAT (page attribute table extension of 4027 pagetables) support. 4028 4029 nopcid [X86-64,EARLY] Disable the PCID cpu feature. 4030 4031 nopku [X86] Disable Memory Protection Keys CPU feature found 4032 in some Intel CPUs. 4033 4034 nopti [X86-64,EARLY] 4035 Equivalent to pti=off 4036 4037 nopv= [X86,XEN,KVM,HYPER_V,VMWARE,EARLY] 4038 Disables the PV optimizations forcing the guest to run 4039 as generic guest with no PV drivers. Currently support 4040 XEN HVM, KVM, HYPER_V and VMWARE guest. 4041 4042 nopvspin [X86,XEN,KVM,EARLY] 4043 Disables the qspinlock slow path using PV optimizations 4044 which allow the hypervisor to 'idle' the guest on lock 4045 contention. 4046 4047 norandmaps Don't use address space randomization. Equivalent to 4048 echo 0 > /proc/sys/kernel/randomize_va_space 4049 4050 noreplace-smp [X86-32,SMP] Don't replace SMP instructions 4051 with UP alternatives 4052 4053 noresume [SWSUSP] Disables resume and restores original swap 4054 space. 4055 4056 nosbagart [IA-64] 4057 4058 no-scroll [VGA] Disables scrollback. 4059 This is required for the Braillex ib80-piezo Braille 4060 reader made by F.H. Papenmeier (Germany). 4061 4062 nosgx [X86-64,SGX,EARLY] Disables Intel SGX kernel support. 4063 4064 nosmap [PPC,EARLY] 4065 Disable SMAP (Supervisor Mode Access Prevention) 4066 even if it is supported by processor. 4067 4068 nosmep [PPC64s,EARLY] 4069 Disable SMEP (Supervisor Mode Execution Prevention) 4070 even if it is supported by processor. 4071 4072 nosmp [SMP,EARLY] Tells an SMP kernel to act as a UP kernel, 4073 and disable the IO APIC. legacy for "maxcpus=0". 4074 4075 nosmt [KNL,MIPS,PPC,S390,EARLY] Disable symmetric multithreading (SMT). 4076 Equivalent to smt=1. 4077 4078 [KNL,X86,PPC] Disable symmetric multithreading (SMT). 4079 nosmt=force: Force disable SMT, cannot be undone 4080 via the sysfs control file. 4081 4082 nosoftlockup [KNL] Disable the soft-lockup detector. 4083 4084 nospec_store_bypass_disable 4085 [HW,EARLY] Disable all mitigations for the Speculative 4086 Store Bypass vulnerability 4087 4088 nospectre_bhb [ARM64,EARLY] Disable all mitigations for Spectre-BHB (branch 4089 history injection) vulnerability. System may allow data leaks 4090 with this option. 4091 4092 nospectre_v1 [X86,PPC,EARLY] Disable mitigations for Spectre Variant 1 4093 (bounds check bypass). With this option data leaks are 4094 possible in the system. 4095 4096 nospectre_v2 [X86,PPC_E500,ARM64,EARLY] Disable all mitigations 4097 for the Spectre variant 2 (indirect branch 4098 prediction) vulnerability. System may allow data 4099 leaks with this option. 4100 4101 no-steal-acc [X86,PV_OPS,ARM64,PPC/PSERIES,RISCV,EARLY] Disable 4102 paravirtualized steal time accounting. steal time is 4103 computed, but won't influence scheduler behaviour 4104 4105 nosync [HW,M68K] Disables sync negotiation for all devices. 4106 4107 no_timer_check [X86,APIC] Disables the code which tests for 4108 broken timer IRQ sources. 4109 4110 no_uaccess_flush 4111 [PPC,EARLY] Don't flush the L1-D cache after accessing user data. 4112 4113 novmcoredd [KNL,KDUMP] 4114 Disable device dump. Device dump allows drivers to 4115 append dump data to vmcore so you can collect driver 4116 specified debug info. Drivers can append the data 4117 without any limit and this data is stored in memory, 4118 so this may cause significant memory stress. Disabling 4119 device dump can help save memory but the driver debug 4120 data will be no longer available. This parameter 4121 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP 4122 is set. 4123 4124 no-vmw-sched-clock 4125 [X86,PV_OPS,EARLY] Disable paravirtualized VMware 4126 scheduler clock and use the default one. 4127 4128 nowatchdog [KNL] Disable both lockup detectors, i.e. 4129 soft-lockup and NMI watchdog (hard-lockup). 4130 4131 nowb [ARM,EARLY] 4132 4133 nox2apic [X86-64,APIC,EARLY] Do not enable x2APIC mode. 4134 4135 NOTE: this parameter will be ignored on systems with the 4136 LEGACY_XAPIC_DISABLED bit set in the 4137 IA32_XAPIC_DISABLE_STATUS MSR. 4138 4139 noxsave [BUGS=X86] Disables x86 extended register state save 4140 and restore using xsave. The kernel will fallback to 4141 enabling legacy floating-point and sse state. 4142 4143 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended 4144 register states. The kernel will fall back to use 4145 xsave to save the states. By using this parameter, 4146 performance of saving the states is degraded because 4147 xsave doesn't support modified optimization while 4148 xsaveopt supports it on xsaveopt enabled systems. 4149 4150 noxsaves [X86] Disables xsaves and xrstors used in saving and 4151 restoring x86 extended register state in compacted 4152 form of xsave area. The kernel will fall back to use 4153 xsaveopt and xrstor to save and restore the states 4154 in standard form of xsave area. By using this 4155 parameter, xsave area per process might occupy more 4156 memory on xsaves enabled systems. 4157 4158 nps_mtm_hs_ctr= [KNL,ARC] 4159 This parameter sets the maximum duration, in 4160 cycles, each HW thread of the CTOP can run 4161 without interruptions, before HW switches it. 4162 The actual maximum duration is 16 times this 4163 parameter's value. 4164 Format: integer between 1 and 255 4165 Default: 255 4166 4167 nptcg= [IA-64] Override max number of concurrent global TLB 4168 purges which is reported from either PAL_VM_SUMMARY or 4169 SAL PALO. 4170 4171 nr_cpus= [SMP,EARLY] Maximum number of processors that an SMP kernel 4172 could support. nr_cpus=n : n >= 1 limits the kernel to 4173 support 'n' processors. It could be larger than the 4174 number of already plugged CPU during bootup, later in 4175 runtime you can physically add extra cpu until it reaches 4176 n. So during boot up some boot time memory for per-cpu 4177 variables need be pre-allocated for later physical cpu 4178 hot plugging. 4179 4180 nr_uarts= [SERIAL] maximum number of UARTs to be registered. 4181 4182 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86, EARLY] 4183 Disable NUMA, Only set up a single NUMA node 4184 spanning all memory. 4185 4186 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic 4187 NUMA balancing. 4188 Allowed values are enable and disable 4189 4190 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA. 4191 'node', 'default' can be specified 4192 This can be set from sysctl after boot. 4193 See Documentation/admin-guide/sysctl/vm.rst for details. 4194 4195 ohci1394_dma=early [HW,EARLY] enable debugging via the ohci1394 driver. 4196 See Documentation/core-api/debugging-via-ohci1394.rst for more 4197 info. 4198 4199 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands 4200 Rather than timing out after 20 ms if an EC 4201 command is not properly ACKed, override the length 4202 of the timeout. We have interrupts disabled while 4203 waiting for the ACK, so if this is set too high 4204 interrupts *may* be lost! 4205 4206 omap_mux= [OMAP] Override bootloader pin multiplexing. 4207 Format: <mux_mode0.mode_name=value>... 4208 For example, to override I2C bus2: 4209 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100 4210 4211 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration 4212 4213 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock] 4214 4215 boundary - index of last SLC block on Flex-OneNAND. 4216 The remaining blocks are configured as MLC blocks. 4217 lock - Configure if Flex-OneNAND boundary should be locked. 4218 Once locked, the boundary cannot be changed. 4219 1 indicates lock status, 0 indicates unlock status. 4220 4221 oops=panic [KNL,EARLY] 4222 Always panic on oopses. Default is to just kill the 4223 process, but there is a small probability of 4224 deadlocking the machine. 4225 This will also cause panics on machine check exceptions. 4226 Useful together with panic=30 to trigger a reboot. 4227 4228 page_alloc.shuffle= 4229 [KNL] Boolean flag to control whether the page allocator 4230 should randomize its free lists. This parameter can be 4231 used to enable/disable page randomization. The state of 4232 the flag can be read from sysfs at: 4233 /sys/module/page_alloc/parameters/shuffle. 4234 This parameter is only available if CONFIG_SHUFFLE_PAGE_ALLOCATOR=y. 4235 4236 page_owner= [KNL,EARLY] Boot-time page_owner enabling option. 4237 Storage of the information about who allocated 4238 each page is disabled in default. With this switch, 4239 we can turn it on. 4240 on: enable the feature 4241 4242 page_poison= [KNL,EARLY] Boot-time parameter changing the state of 4243 poisoning on the buddy allocator, available with 4244 CONFIG_PAGE_POISONING=y. 4245 off: turn off poisoning (default) 4246 on: turn on poisoning 4247 4248 page_reporting.page_reporting_order= 4249 [KNL] Minimal page reporting order 4250 Format: <integer> 4251 Adjust the minimal page reporting order. The page 4252 reporting is disabled when it exceeds MAX_PAGE_ORDER. 4253 4254 panic= [KNL] Kernel behaviour on panic: delay <timeout> 4255 timeout > 0: seconds before rebooting 4256 timeout = 0: wait forever 4257 timeout < 0: reboot immediately 4258 Format: <timeout> 4259 4260 panic_on_taint= [KNL,EARLY] 4261 Bitmask for conditionally calling panic() in add_taint() 4262 Format: <hex>[,nousertaint] 4263 Hexadecimal bitmask representing the set of TAINT flags 4264 that will cause the kernel to panic when add_taint() is 4265 called with any of the flags in this set. 4266 The optional switch "nousertaint" can be utilized to 4267 prevent userspace forced crashes by writing to sysctl 4268 /proc/sys/kernel/tainted any flagset matching with the 4269 bitmask set on panic_on_taint. 4270 See Documentation/admin-guide/tainted-kernels.rst for 4271 extra details on the taint flags that users can pick 4272 to compose the bitmask to assign to panic_on_taint. 4273 4274 panic_on_warn=1 panic() instead of WARN(). Useful to cause kdump 4275 on a WARN(). 4276 4277 panic_print= Bitmask for printing system info when panic happens. 4278 User can chose combination of the following bits: 4279 bit 0: print all tasks info 4280 bit 1: print system memory info 4281 bit 2: print timer info 4282 bit 3: print locks info if CONFIG_LOCKDEP is on 4283 bit 4: print ftrace buffer 4284 bit 5: print all printk messages in buffer 4285 bit 6: print all CPUs backtrace (if available in the arch) 4286 bit 7: print only tasks in uninterruptible (blocked) state 4287 *Be aware* that this option may print a _lot_ of lines, 4288 so there are risks of losing older messages in the log. 4289 Use this option carefully, maybe worth to setup a 4290 bigger log buffer with "log_buf_len" along with this. 4291 4292 parkbd.port= [HW] Parallel port number the keyboard adapter is 4293 connected to, default is 0. 4294 Format: <parport#> 4295 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation, 4296 0 for XT, 1 for AT (default is AT). 4297 Format: <mode> 4298 4299 parport= [HW,PPT] Specify parallel ports. 0 disables. 4300 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] } 4301 Use 'auto' to force the driver to use any 4302 IRQ/DMA settings detected (the default is to 4303 ignore detected IRQ/DMA settings because of 4304 possible conflicts). You can specify the base 4305 address, IRQ, and DMA settings; IRQ and DMA 4306 should be numbers, or 'auto' (for using detected 4307 settings on that particular port), or 'nofifo' 4308 (to avoid using a FIFO even if it is detected). 4309 Parallel ports are assigned in the order they 4310 are specified on the command line, starting 4311 with parport0. 4312 4313 parport_init_mode= [HW,PPT] 4314 Configure VIA parallel port to operate in 4315 a specific mode. This is necessary on Pegasos 4316 computer where firmware has no options for setting 4317 up parallel port mode and sets it to spp. 4318 Currently this function knows 686a and 8231 chips. 4319 Format: [spp|ps2|epp|ecp|ecpepp] 4320 4321 pata_legacy.all= [HW,LIBATA] 4322 Format: <int> 4323 Set to non-zero to probe primary and secondary ISA 4324 port ranges on PCI systems where no PCI PATA device 4325 has been found at either range. Disabled by default. 4326 4327 pata_legacy.autospeed= [HW,LIBATA] 4328 Format: <int> 4329 Set to non-zero if a chip is present that snoops speed 4330 changes. Disabled by default. 4331 4332 pata_legacy.ht6560a= [HW,LIBATA] 4333 Format: <int> 4334 Set to 1, 2, or 3 for HT 6560A on the primary channel, 4335 the secondary channel, or both channels respectively. 4336 Disabled by default. 4337 4338 pata_legacy.ht6560b= [HW,LIBATA] 4339 Format: <int> 4340 Set to 1, 2, or 3 for HT 6560B on the primary channel, 4341 the secondary channel, or both channels respectively. 4342 Disabled by default. 4343 4344 pata_legacy.iordy_mask= [HW,LIBATA] 4345 Format: <int> 4346 IORDY enable mask. Set individual bits to allow IORDY 4347 for the respective channel. Bit 0 is for the first 4348 legacy channel handled by this driver, bit 1 is for 4349 the second channel, and so on. The sequence will often 4350 correspond to the primary legacy channel, the secondary 4351 legacy channel, and so on, but the handling of a PCI 4352 bus and the use of other driver options may interfere 4353 with the sequence. By default IORDY is allowed across 4354 all channels. 4355 4356 pata_legacy.opti82c46x= [HW,LIBATA] 4357 Format: <int> 4358 Set to 1, 2, or 3 for Opti 82c611A on the primary 4359 channel, the secondary channel, or both channels 4360 respectively. Disabled by default. 4361 4362 pata_legacy.opti82c611a= [HW,LIBATA] 4363 Format: <int> 4364 Set to 1, 2, or 3 for Opti 82c465MV on the primary 4365 channel, the secondary channel, or both channels 4366 respectively. Disabled by default. 4367 4368 pata_legacy.pio_mask= [HW,LIBATA] 4369 Format: <int> 4370 PIO mode mask for autospeed devices. Set individual 4371 bits to allow the use of the respective PIO modes. 4372 Bit 0 is for mode 0, bit 1 is for mode 1, and so on. 4373 All modes allowed by default. 4374 4375 pata_legacy.probe_all= [HW,LIBATA] 4376 Format: <int> 4377 Set to non-zero to probe tertiary and further ISA 4378 port ranges on PCI systems. Disabled by default. 4379 4380 pata_legacy.probe_mask= [HW,LIBATA] 4381 Format: <int> 4382 Probe mask for legacy ISA PATA ports. Depending on 4383 platform configuration and the use of other driver 4384 options up to 6 legacy ports are supported: 0x1f0, 4385 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing 4386 of individual ports can be disabled by setting the 4387 corresponding bits in the mask to 1. Bit 0 is for 4388 the first port in the list above (0x1f0), and so on. 4389 By default all supported ports are probed. 4390 4391 pata_legacy.qdi= [HW,LIBATA] 4392 Format: <int> 4393 Set to non-zero to probe QDI controllers. By default 4394 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise. 4395 4396 pata_legacy.winbond= [HW,LIBATA] 4397 Format: <int> 4398 Set to non-zero to probe Winbond controllers. Use 4399 the standard I/O port (0x130) if 1, otherwise the 4400 value given is the I/O port to use (typically 0x1b0). 4401 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE, 4402 0 otherwise. 4403 4404 pata_platform.pio_mask= [HW,LIBATA] 4405 Format: <int> 4406 Supported PIO mode mask. Set individual bits to allow 4407 the use of the respective PIO modes. Bit 0 is for 4408 mode 0, bit 1 is for mode 1, and so on. Mode 0 only 4409 allowed by default. 4410 4411 pause_on_oops=<int> 4412 Halt all CPUs after the first oops has been printed for 4413 the specified number of seconds. This is to be used if 4414 your oopses keep scrolling off the screen. 4415 4416 pcbit= [HW,ISDN] 4417 4418 pci=option[,option...] [PCI,EARLY] various PCI subsystem options. 4419 4420 Some options herein operate on a specific device 4421 or a set of devices (<pci_dev>). These are 4422 specified in one of the following formats: 4423 4424 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]* 4425 pci:<vendor>:<device>[:<subvendor>:<subdevice>] 4426 4427 Note: the first format specifies a PCI 4428 bus/device/function address which may change 4429 if new hardware is inserted, if motherboard 4430 firmware changes, or due to changes caused 4431 by other kernel parameters. If the 4432 domain is left unspecified, it is 4433 taken to be zero. Optionally, a path 4434 to a device through multiple device/function 4435 addresses can be specified after the base 4436 address (this is more robust against 4437 renumbering issues). The second format 4438 selects devices using IDs from the 4439 configuration space which may match multiple 4440 devices in the system. 4441 4442 earlydump dump PCI config space before the kernel 4443 changes anything 4444 off [X86] don't probe for the PCI bus 4445 bios [X86-32] force use of PCI BIOS, don't access 4446 the hardware directly. Use this if your machine 4447 has a non-standard PCI host bridge. 4448 nobios [X86-32] disallow use of PCI BIOS, only direct 4449 hardware access methods are allowed. Use this 4450 if you experience crashes upon bootup and you 4451 suspect they are caused by the BIOS. 4452 conf1 [X86] Force use of PCI Configuration Access 4453 Mechanism 1 (config address in IO port 0xCF8, 4454 data in IO port 0xCFC, both 32-bit). 4455 conf2 [X86] Force use of PCI Configuration Access 4456 Mechanism 2 (IO port 0xCF8 is an 8-bit port for 4457 the function, IO port 0xCFA, also 8-bit, sets 4458 bus number. The config space is then accessed 4459 through ports 0xC000-0xCFFF). 4460 See http://wiki.osdev.org/PCI for more info 4461 on the configuration access mechanisms. 4462 noaer [PCIE] If the PCIEAER kernel config parameter is 4463 enabled, this kernel boot option can be used to 4464 disable the use of PCIE advanced error reporting. 4465 nodomains [PCI] Disable support for multiple PCI 4466 root domains (aka PCI segments, in ACPI-speak). 4467 nommconf [X86] Disable use of MMCONFIG for PCI 4468 Configuration 4469 check_enable_amd_mmconf [X86] check for and enable 4470 properly configured MMIO access to PCI 4471 config space on AMD family 10h CPU 4472 nomsi [MSI] If the PCI_MSI kernel config parameter is 4473 enabled, this kernel boot option can be used to 4474 disable the use of MSI interrupts system-wide. 4475 noioapicquirk [APIC] Disable all boot interrupt quirks. 4476 Safety option to keep boot IRQs enabled. This 4477 should never be necessary. 4478 ioapicreroute [APIC] Enable rerouting of boot IRQs to the 4479 primary IO-APIC for bridges that cannot disable 4480 boot IRQs. This fixes a source of spurious IRQs 4481 when the system masks IRQs. 4482 noioapicreroute [APIC] Disable workaround that uses the 4483 boot IRQ equivalent of an IRQ that connects to 4484 a chipset where boot IRQs cannot be disabled. 4485 The opposite of ioapicreroute. 4486 biosirq [X86-32] Use PCI BIOS calls to get the interrupt 4487 routing table. These calls are known to be buggy 4488 on several machines and they hang the machine 4489 when used, but on other computers it's the only 4490 way to get the interrupt routing table. Try 4491 this option if the kernel is unable to allocate 4492 IRQs or discover secondary PCI buses on your 4493 motherboard. 4494 rom [X86] Assign address space to expansion ROMs. 4495 Use with caution as certain devices share 4496 address decoders between ROMs and other 4497 resources. 4498 norom [X86] Do not assign address space to 4499 expansion ROMs that do not already have 4500 BIOS assigned address ranges. 4501 nobar [X86] Do not assign address space to the 4502 BARs that weren't assigned by the BIOS. 4503 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be 4504 assigned automatically to PCI devices. You can 4505 make the kernel exclude IRQs of your ISA cards 4506 this way. 4507 pirqaddr=0xAAAAA [X86] Specify the physical address 4508 of the PIRQ table (normally generated 4509 by the BIOS) if it is outside the 4510 F0000h-100000h range. 4511 lastbus=N [X86] Scan all buses thru bus #N. Can be 4512 useful if the kernel is unable to find your 4513 secondary buses and you want to tell it 4514 explicitly which ones they are. 4515 assign-busses [X86] Always assign all PCI bus 4516 numbers ourselves, overriding 4517 whatever the firmware may have done. 4518 usepirqmask [X86] Honor the possible IRQ mask stored 4519 in the BIOS $PIR table. This is needed on 4520 some systems with broken BIOSes, notably 4521 some HP Pavilion N5400 and Omnibook XE3 4522 notebooks. This will have no effect if ACPI 4523 IRQ routing is enabled. 4524 noacpi [X86] Do not use ACPI for IRQ routing 4525 or for PCI scanning. 4526 use_crs [X86] Use PCI host bridge window information 4527 from ACPI. On BIOSes from 2008 or later, this 4528 is enabled by default. If you need to use this, 4529 please report a bug. 4530 nocrs [X86] Ignore PCI host bridge windows from ACPI. 4531 If you need to use this, please report a bug. 4532 use_e820 [X86] Use E820 reservations to exclude parts of 4533 PCI host bridge windows. This is a workaround 4534 for BIOS defects in host bridge _CRS methods. 4535 If you need to use this, please report a bug to 4536 <linux-pci@vger.kernel.org>. 4537 no_e820 [X86] Ignore E820 reservations for PCI host 4538 bridge windows. This is the default on modern 4539 hardware. If you need to use this, please report 4540 a bug to <linux-pci@vger.kernel.org>. 4541 routeirq Do IRQ routing for all PCI devices. 4542 This is normally done in pci_enable_device(), 4543 so this option is a temporary workaround 4544 for broken drivers that don't call it. 4545 skip_isa_align [X86] do not align io start addr, so can 4546 handle more pci cards 4547 noearly [X86] Don't do any early type 1 scanning. 4548 This might help on some broken boards which 4549 machine check when some devices' config space 4550 is read. But various workarounds are disabled 4551 and some IOMMU drivers will not work. 4552 bfsort Sort PCI devices into breadth-first order. 4553 This sorting is done to get a device 4554 order compatible with older (<= 2.4) kernels. 4555 nobfsort Don't sort PCI devices into breadth-first order. 4556 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size) 4557 tuning and use the BIOS-configured MPS defaults. 4558 pcie_bus_safe Set every device's MPS to the largest value 4559 supported by all devices below the root complex. 4560 pcie_bus_perf Set device MPS to the largest allowable MPS 4561 based on its parent bus. Also set MRRS (Max 4562 Read Request Size) to the largest supported 4563 value (no larger than the MPS that the device 4564 or bus can support) for best performance. 4565 pcie_bus_peer2peer Set every device's MPS to 128B, which 4566 every device is guaranteed to support. This 4567 configuration allows peer-to-peer DMA between 4568 any pair of devices, possibly at the cost of 4569 reduced performance. This also guarantees 4570 that hot-added devices will work. 4571 cbiosize=nn[KMG] The fixed amount of bus space which is 4572 reserved for the CardBus bridge's IO window. 4573 The default value is 256 bytes. 4574 cbmemsize=nn[KMG] The fixed amount of bus space which is 4575 reserved for the CardBus bridge's memory 4576 window. The default value is 64 megabytes. 4577 resource_alignment= 4578 Format: 4579 [<order of align>@]<pci_dev>[; ...] 4580 Specifies alignment and device to reassign 4581 aligned memory resources. How to 4582 specify the device is described above. 4583 If <order of align> is not specified, 4584 PAGE_SIZE is used as alignment. 4585 A PCI-PCI bridge can be specified if resource 4586 windows need to be expanded. 4587 To specify the alignment for several 4588 instances of a device, the PCI vendor, 4589 device, subvendor, and subdevice may be 4590 specified, e.g., 12@pci:8086:9c22:103c:198f 4591 for 4096-byte alignment. 4592 ecrc= Enable/disable PCIe ECRC (transaction layer 4593 end-to-end CRC checking). Only effective if 4594 OS has native AER control (either granted by 4595 ACPI _OSC or forced via "pcie_ports=native") 4596 bios: Use BIOS/firmware settings. This is the 4597 the default. 4598 off: Turn ECRC off 4599 on: Turn ECRC on. 4600 hpiosize=nn[KMG] The fixed amount of bus space which is 4601 reserved for hotplug bridge's IO window. 4602 Default size is 256 bytes. 4603 hpmmiosize=nn[KMG] The fixed amount of bus space which is 4604 reserved for hotplug bridge's MMIO window. 4605 Default size is 2 megabytes. 4606 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is 4607 reserved for hotplug bridge's MMIO_PREF window. 4608 Default size is 2 megabytes. 4609 hpmemsize=nn[KMG] The fixed amount of bus space which is 4610 reserved for hotplug bridge's MMIO and 4611 MMIO_PREF window. 4612 Default size is 2 megabytes. 4613 hpbussize=nn The minimum amount of additional bus numbers 4614 reserved for buses below a hotplug bridge. 4615 Default is 1. 4616 realloc= Enable/disable reallocating PCI bridge resources 4617 if allocations done by BIOS are too small to 4618 accommodate resources required by all child 4619 devices. 4620 off: Turn realloc off 4621 on: Turn realloc on 4622 realloc same as realloc=on 4623 noari do not use PCIe ARI. 4624 noats [PCIE, Intel-IOMMU, AMD-IOMMU] 4625 do not use PCIe ATS (and IOMMU device IOTLB). 4626 pcie_scan_all Scan all possible PCIe devices. Otherwise we 4627 only look for one device below a PCIe downstream 4628 port. 4629 big_root_window Try to add a big 64bit memory window to the PCIe 4630 root complex on AMD CPUs. Some GFX hardware 4631 can resize a BAR to allow access to all VRAM. 4632 Adding the window is slightly risky (it may 4633 conflict with unreported devices), so this 4634 taints the kernel. 4635 disable_acs_redir=<pci_dev>[; ...] 4636 Specify one or more PCI devices (in the format 4637 specified above) separated by semicolons. 4638 Each device specified will have the PCI ACS 4639 redirect capabilities forced off which will 4640 allow P2P traffic between devices through 4641 bridges without forcing it upstream. Note: 4642 this removes isolation between devices and 4643 may put more devices in an IOMMU group. 4644 force_floating [S390] Force usage of floating interrupts. 4645 nomio [S390] Do not use MIO instructions. 4646 norid [S390] ignore the RID field and force use of 4647 one PCI domain per PCI function 4648 4649 pcie_aspm= [PCIE] Forcibly enable or ignore PCIe Active State Power 4650 Management. 4651 off Don't touch ASPM configuration at all. Leave any 4652 configuration done by firmware unchanged. 4653 force Enable ASPM even on devices that claim not to support it. 4654 WARNING: Forcing ASPM on may cause system lockups. 4655 4656 pcie_ports= [PCIE] PCIe port services handling: 4657 native Use native PCIe services (PME, AER, DPC, PCIe hotplug) 4658 even if the platform doesn't give the OS permission to 4659 use them. This may cause conflicts if the platform 4660 also tries to use these services. 4661 dpc-native Use native PCIe service for DPC only. May 4662 cause conflicts if firmware uses AER or DPC. 4663 compat Disable native PCIe services (PME, AER, DPC, PCIe 4664 hotplug). 4665 4666 pcie_port_pm= [PCIE] PCIe port power management handling: 4667 off Disable power management of all PCIe ports 4668 force Forcibly enable power management of all PCIe ports 4669 4670 pcie_pme= [PCIE,PM] Native PCIe PME signaling options: 4671 nomsi Do not use MSI for native PCIe PME signaling (this makes 4672 all PCIe root ports use INTx for all services). 4673 4674 pcmv= [HW,PCMCIA] BadgePAD 4 4675 4676 pd_ignore_unused 4677 [PM] 4678 Keep all power-domains already enabled by bootloader on, 4679 even if no driver has claimed them. This is useful 4680 for debug and development, but should not be 4681 needed on a platform with proper driver support. 4682 4683 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at 4684 boot time. 4685 Format: { 0 | 1 } 4686 See arch/parisc/kernel/pdc_chassis.c 4687 4688 percpu_alloc= [MM,EARLY] 4689 Select which percpu first chunk allocator to use. 4690 Currently supported values are "embed" and "page". 4691 Archs may support subset or none of the selections. 4692 See comments in mm/percpu.c for details on each 4693 allocator. This parameter is primarily for debugging 4694 and performance comparison. 4695 4696 pirq= [SMP,APIC] Manual mp-table setup 4697 See Documentation/arch/x86/i386/IO-APIC.rst. 4698 4699 plip= [PPT,NET] Parallel port network link 4700 Format: { parport<nr> | timid | 0 } 4701 See also Documentation/admin-guide/parport.rst. 4702 4703 pmtmr= [X86] Manual setup of pmtmr I/O Port. 4704 Override pmtimer IOPort with a hex value. 4705 e.g. pmtmr=0x508 4706 4707 pmu_override= [PPC] Override the PMU. 4708 This option takes over the PMU facility, so it is no 4709 longer usable by perf. Setting this option starts the 4710 PMU counters by setting MMCR0 to 0 (the FC bit is 4711 cleared). If a number is given, then MMCR1 is set to 4712 that number, otherwise (e.g., 'pmu_override=on'), MMCR1 4713 remains 0. 4714 4715 pm_debug_messages [SUSPEND,KNL] 4716 Enable suspend/resume debug messages during boot up. 4717 4718 pnp.debug=1 [PNP] 4719 Enable PNP debug messages (depends on the 4720 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time 4721 via /sys/module/pnp/parameters/debug. We always show 4722 current resource usage; turning this on also shows 4723 possible settings and some assignment information. 4724 4725 pnpacpi= [ACPI] 4726 { off } 4727 4728 pnpbios= [ISAPNP] 4729 { on | off | curr | res | no-curr | no-res } 4730 4731 pnp_reserve_irq= 4732 [ISAPNP] Exclude IRQs for the autoconfiguration 4733 4734 pnp_reserve_dma= 4735 [ISAPNP] Exclude DMAs for the autoconfiguration 4736 4737 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration 4738 Ranges are in pairs (I/O port base and size). 4739 4740 pnp_reserve_mem= 4741 [ISAPNP] Exclude memory regions for the 4742 autoconfiguration. 4743 Ranges are in pairs (memory base and size). 4744 4745 ports= [IP_VS_FTP] IPVS ftp helper module 4746 Default is 21. 4747 Up to 8 (IP_VS_APP_MAX_PORTS) ports 4748 may be specified. 4749 Format: <port>,<port>.... 4750 4751 possible_cpus= [SMP,S390,X86] 4752 Format: <unsigned int> 4753 Set the number of possible CPUs, overriding the 4754 regular discovery mechanisms (such as ACPI/FW, etc). 4755 4756 powersave=off [PPC] This option disables power saving features. 4757 It specifically disables cpuidle and sets the 4758 platform machine description specific power_save 4759 function to NULL. On Idle the CPU just reduces 4760 execution priority. 4761 4762 ppc_strict_facility_enable 4763 [PPC,ENABLE] This option catches any kernel floating point, 4764 Altivec, VSX and SPE outside of regions specifically 4765 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()). 4766 There is some performance impact when enabling this. 4767 4768 ppc_tm= [PPC,EARLY] 4769 Format: {"off"} 4770 Disable Hardware Transactional Memory 4771 4772 preempt= [KNL] 4773 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC 4774 none - Limited to cond_resched() calls 4775 voluntary - Limited to cond_resched() and might_sleep() calls 4776 full - Any section that isn't explicitly preempt disabled 4777 can be preempted anytime. 4778 4779 print-fatal-signals= 4780 [KNL] debug: print fatal signals 4781 4782 If enabled, warn about various signal handling 4783 related application anomalies: too many signals, 4784 too many POSIX.1 timers, fatal signals causing a 4785 coredump - etc. 4786 4787 If you hit the warning due to signal overflow, 4788 you might want to try "ulimit -i unlimited". 4789 4790 default: off. 4791 4792 printk.always_kmsg_dump= 4793 Trigger kmsg_dump for cases other than kernel oops or 4794 panics 4795 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 4796 default: disabled 4797 4798 printk.console_no_auto_verbose= 4799 Disable console loglevel raise on oops, panic 4800 or lockdep-detected issues (only if lock debug is on). 4801 With an exception to setups with low baudrate on 4802 serial console, keeping this 0 is a good choice 4803 in order to provide more debug information. 4804 Format: <bool> 4805 default: 0 (auto_verbose is enabled) 4806 4807 printk.devkmsg={on,off,ratelimit} 4808 Control writing to /dev/kmsg. 4809 on - unlimited logging to /dev/kmsg from userspace 4810 off - logging to /dev/kmsg disabled 4811 ratelimit - ratelimit the logging 4812 Default: ratelimit 4813 4814 printk.time= Show timing data prefixed to each printk message line 4815 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 4816 4817 processor.max_cstate= [HW,ACPI] 4818 Limit processor to maximum C-state 4819 max_cstate=9 overrides any DMI blacklist limit. 4820 4821 processor.nocst [HW,ACPI] 4822 Ignore the _CST method to determine C-states, 4823 instead using the legacy FADT method 4824 4825 profile= [KNL] Enable kernel profiling via /proc/profile 4826 Format: [<profiletype>,]<number> 4827 Param: <profiletype>: "schedule", "sleep", or "kvm" 4828 [defaults to kernel profiling] 4829 Param: "schedule" - profile schedule points. 4830 Param: "sleep" - profile D-state sleeping (millisecs). 4831 Requires CONFIG_SCHEDSTATS 4832 Param: "kvm" - profile VM exits. 4833 Param: <number> - step/bucket size as a power of 2 for 4834 statistical time based profiling. 4835 4836 prompt_ramdisk= [RAM] [Deprecated] 4837 4838 prot_virt= [S390] enable hosting protected virtual machines 4839 isolated from the hypervisor (if hardware supports 4840 that). If enabled, the default kernel base address 4841 might be overridden even when Kernel Address Space 4842 Layout Randomization is disabled. 4843 Format: <bool> 4844 4845 psi= [KNL] Enable or disable pressure stall information 4846 tracking. 4847 Format: <bool> 4848 4849 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to 4850 probe for; one of (bare|imps|exps|lifebook|any). 4851 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports 4852 per second. 4853 psmouse.resetafter= [HW,MOUSE] 4854 Try to reset the device after so many bad packets 4855 (0 = never). 4856 psmouse.resolution= 4857 [HW,MOUSE] Set desired mouse resolution, in dpi. 4858 psmouse.smartscroll= 4859 [HW,MOUSE] Controls Logitech smartscroll autorepeat. 4860 0 = disabled, 1 = enabled (default). 4861 4862 pstore.backend= Specify the name of the pstore backend to use 4863 4864 pti= [X86-64] Control Page Table Isolation of user and 4865 kernel address spaces. Disabling this feature 4866 removes hardening, but improves performance of 4867 system calls and interrupts. 4868 4869 on - unconditionally enable 4870 off - unconditionally disable 4871 auto - kernel detects whether your CPU model is 4872 vulnerable to issues that PTI mitigates 4873 4874 Not specifying this option is equivalent to pti=auto. 4875 4876 pty.legacy_count= 4877 [KNL] Number of legacy pty's. Overwrites compiled-in 4878 default number. 4879 4880 quiet [KNL,EARLY] Disable most log messages 4881 4882 r128= [HW,DRM] 4883 4884 radix_hcall_invalidate=on [PPC/PSERIES] 4885 Disable RADIX GTSE feature and use hcall for TLB 4886 invalidate. 4887 4888 raid= [HW,RAID] 4889 See Documentation/admin-guide/md.rst. 4890 4891 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes 4892 See Documentation/admin-guide/blockdev/ramdisk.rst. 4893 4894 ramdisk_start= [RAM] RAM disk image start address 4895 4896 random.trust_cpu=off 4897 [KNL,EARLY] Disable trusting the use of the CPU's 4898 random number generator (if available) to 4899 initialize the kernel's RNG. 4900 4901 random.trust_bootloader=off 4902 [KNL,EARLY] Disable trusting the use of the a seed 4903 passed by the bootloader (if available) to 4904 initialize the kernel's RNG. 4905 4906 randomize_kstack_offset= 4907 [KNL,EARLY] Enable or disable kernel stack offset 4908 randomization, which provides roughly 5 bits of 4909 entropy, frustrating memory corruption attacks 4910 that depend on stack address determinism or 4911 cross-syscall address exposures. This is only 4912 available on architectures that have defined 4913 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET. 4914 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 4915 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT. 4916 4917 ras=option[,option,...] [KNL] RAS-specific options 4918 4919 cec_disable [X86] 4920 Disable the Correctable Errors Collector, 4921 see CONFIG_RAS_CEC help text. 4922 4923 rcu_nocbs[=cpu-list] 4924 [KNL] The optional argument is a cpu list, 4925 as described above. 4926 4927 In kernels built with CONFIG_RCU_NOCB_CPU=y, 4928 enable the no-callback CPU mode, which prevents 4929 such CPUs' callbacks from being invoked in 4930 softirq context. Invocation of such CPUs' RCU 4931 callbacks will instead be offloaded to "rcuox/N" 4932 kthreads created for that purpose, where "x" is 4933 "p" for RCU-preempt, "s" for RCU-sched, and "g" 4934 for the kthreads that mediate grace periods; and 4935 "N" is the CPU number. This reduces OS jitter on 4936 the offloaded CPUs, which can be useful for HPC 4937 and real-time workloads. It can also improve 4938 energy efficiency for asymmetric multiprocessors. 4939 4940 If a cpulist is passed as an argument, the specified 4941 list of CPUs is set to no-callback mode from boot. 4942 4943 Otherwise, if the '=' sign and the cpulist 4944 arguments are omitted, no CPU will be set to 4945 no-callback mode from boot but the mode may be 4946 toggled at runtime via cpusets. 4947 4948 Note that this argument takes precedence over 4949 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option. 4950 4951 rcu_nocb_poll [KNL] 4952 Rather than requiring that offloaded CPUs 4953 (specified by rcu_nocbs= above) explicitly 4954 awaken the corresponding "rcuoN" kthreads, 4955 make these kthreads poll for callbacks. 4956 This improves the real-time response for the 4957 offloaded CPUs by relieving them of the need to 4958 wake up the corresponding kthread, but degrades 4959 energy efficiency by requiring that the kthreads 4960 periodically wake up to do the polling. 4961 4962 rcutree.blimit= [KNL] 4963 Set maximum number of finished RCU callbacks to 4964 process in one batch. 4965 4966 rcutree.do_rcu_barrier= [KNL] 4967 Request a call to rcu_barrier(). This is 4968 throttled so that userspace tests can safely 4969 hammer on the sysfs variable if they so choose. 4970 If triggered before the RCU grace-period machinery 4971 is fully active, this will error out with EAGAIN. 4972 4973 rcutree.dump_tree= [KNL] 4974 Dump the structure of the rcu_node combining tree 4975 out at early boot. This is used for diagnostic 4976 purposes, to verify correct tree setup. 4977 4978 rcutree.gp_cleanup_delay= [KNL] 4979 Set the number of jiffies to delay each step of 4980 RCU grace-period cleanup. 4981 4982 rcutree.gp_init_delay= [KNL] 4983 Set the number of jiffies to delay each step of 4984 RCU grace-period initialization. 4985 4986 rcutree.gp_preinit_delay= [KNL] 4987 Set the number of jiffies to delay each step of 4988 RCU grace-period pre-initialization, that is, 4989 the propagation of recent CPU-hotplug changes up 4990 the rcu_node combining tree. 4991 4992 rcutree.jiffies_till_first_fqs= [KNL] 4993 Set delay from grace-period initialization to 4994 first attempt to force quiescent states. 4995 Units are jiffies, minimum value is zero, 4996 and maximum value is HZ. 4997 4998 rcutree.jiffies_till_next_fqs= [KNL] 4999 Set delay between subsequent attempts to force 5000 quiescent states. Units are jiffies, minimum 5001 value is one, and maximum value is HZ. 5002 5003 rcutree.jiffies_till_sched_qs= [KNL] 5004 Set required age in jiffies for a 5005 given grace period before RCU starts 5006 soliciting quiescent-state help from 5007 rcu_note_context_switch() and cond_resched(). 5008 If not specified, the kernel will calculate 5009 a value based on the most recent settings 5010 of rcutree.jiffies_till_first_fqs 5011 and rcutree.jiffies_till_next_fqs. 5012 This calculated value may be viewed in 5013 rcutree.jiffies_to_sched_qs. Any attempt to set 5014 rcutree.jiffies_to_sched_qs will be cheerfully 5015 overwritten. 5016 5017 rcutree.kthread_prio= [KNL,BOOT] 5018 Set the SCHED_FIFO priority of the RCU per-CPU 5019 kthreads (rcuc/N). This value is also used for 5020 the priority of the RCU boost threads (rcub/N) 5021 and for the RCU grace-period kthreads (rcu_bh, 5022 rcu_preempt, and rcu_sched). If RCU_BOOST is 5023 set, valid values are 1-99 and the default is 1 5024 (the least-favored priority). Otherwise, when 5025 RCU_BOOST is not set, valid values are 0-99 and 5026 the default is zero (non-realtime operation). 5027 When RCU_NOCB_CPU is set, also adjust the 5028 priority of NOCB callback kthreads. 5029 5030 rcutree.nocb_nobypass_lim_per_jiffy= [KNL] 5031 On callback-offloaded (rcu_nocbs) CPUs, 5032 RCU reduces the lock contention that would 5033 otherwise be caused by callback floods through 5034 use of the ->nocb_bypass list. However, in the 5035 common non-flooded case, RCU queues directly to 5036 the main ->cblist in order to avoid the extra 5037 overhead of the ->nocb_bypass list and its lock. 5038 But if there are too many callbacks queued during 5039 a single jiffy, RCU pre-queues the callbacks into 5040 the ->nocb_bypass queue. The definition of "too 5041 many" is supplied by this kernel boot parameter. 5042 5043 rcutree.qhimark= [KNL] 5044 Set threshold of queued RCU callbacks beyond which 5045 batch limiting is disabled. 5046 5047 rcutree.qlowmark= [KNL] 5048 Set threshold of queued RCU callbacks below which 5049 batch limiting is re-enabled. 5050 5051 rcutree.qovld= [KNL] 5052 Set threshold of queued RCU callbacks beyond which 5053 RCU's force-quiescent-state scan will aggressively 5054 enlist help from cond_resched() and sched IPIs to 5055 help CPUs more quickly reach quiescent states. 5056 Set to less than zero to make this be set based 5057 on rcutree.qhimark at boot time and to zero to 5058 disable more aggressive help enlistment. 5059 5060 rcutree.rcu_delay_page_cache_fill_msec= [KNL] 5061 Set the page-cache refill delay (in milliseconds) 5062 in response to low-memory conditions. The range 5063 of permitted values is in the range 0:100000. 5064 5065 rcutree.rcu_divisor= [KNL] 5066 Set the shift-right count to use to compute 5067 the callback-invocation batch limit bl from 5068 the number of callbacks queued on this CPU. 5069 The result will be bounded below by the value of 5070 the rcutree.blimit kernel parameter. Every bl 5071 callbacks, the softirq handler will exit in 5072 order to allow the CPU to do other work. 5073 5074 Please note that this callback-invocation batch 5075 limit applies only to non-offloaded callback 5076 invocation. Offloaded callbacks are instead 5077 invoked in the context of an rcuoc kthread, which 5078 scheduler will preempt as it does any other task. 5079 5080 rcutree.rcu_fanout_exact= [KNL] 5081 Disable autobalancing of the rcu_node combining 5082 tree. This is used by rcutorture, and might 5083 possibly be useful for architectures having high 5084 cache-to-cache transfer latencies. 5085 5086 rcutree.rcu_fanout_leaf= [KNL] 5087 Change the number of CPUs assigned to each 5088 leaf rcu_node structure. Useful for very 5089 large systems, which will choose the value 64, 5090 and for NUMA systems with large remote-access 5091 latencies, which will choose a value aligned 5092 with the appropriate hardware boundaries. 5093 5094 rcutree.rcu_min_cached_objs= [KNL] 5095 Minimum number of objects which are cached and 5096 maintained per one CPU. Object size is equal 5097 to PAGE_SIZE. The cache allows to reduce the 5098 pressure to page allocator, also it makes the 5099 whole algorithm to behave better in low memory 5100 condition. 5101 5102 rcutree.rcu_nocb_gp_stride= [KNL] 5103 Set the number of NOCB callback kthreads in 5104 each group, which defaults to the square root 5105 of the number of CPUs. Larger numbers reduce 5106 the wakeup overhead on the global grace-period 5107 kthread, but increases that same overhead on 5108 each group's NOCB grace-period kthread. 5109 5110 rcutree.rcu_kick_kthreads= [KNL] 5111 Cause the grace-period kthread to get an extra 5112 wake_up() if it sleeps three times longer than 5113 it should at force-quiescent-state time. 5114 This wake_up() will be accompanied by a 5115 WARN_ONCE() splat and an ftrace_dump(). 5116 5117 rcutree.rcu_resched_ns= [KNL] 5118 Limit the time spend invoking a batch of RCU 5119 callbacks to the specified number of nanoseconds. 5120 By default, this limit is checked only once 5121 every 32 callbacks in order to limit the pain 5122 inflicted by local_clock() overhead. 5123 5124 rcutree.rcu_unlock_delay= [KNL] 5125 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels, 5126 this specifies an rcu_read_unlock()-time delay 5127 in microseconds. This defaults to zero. 5128 Larger delays increase the probability of 5129 catching RCU pointer leaks, that is, buggy use 5130 of RCU-protected pointers after the relevant 5131 rcu_read_unlock() has completed. 5132 5133 rcutree.sysrq_rcu= [KNL] 5134 Commandeer a sysrq key to dump out Tree RCU's 5135 rcu_node tree with an eye towards determining 5136 why a new grace period has not yet started. 5137 5138 rcutree.use_softirq= [KNL] 5139 If set to zero, move all RCU_SOFTIRQ processing to 5140 per-CPU rcuc kthreads. Defaults to a non-zero 5141 value, meaning that RCU_SOFTIRQ is used by default. 5142 Specify rcutree.use_softirq=0 to use rcuc kthreads. 5143 5144 But note that CONFIG_PREEMPT_RT=y kernels disable 5145 this kernel boot parameter, forcibly setting it 5146 to zero. 5147 5148 rcutree.enable_rcu_lazy= [KNL] 5149 To save power, batch RCU callbacks and flush after 5150 delay, memory pressure or callback list growing too 5151 big. 5152 5153 rcutree.rcu_normal_wake_from_gp= [KNL] 5154 Reduces a latency of synchronize_rcu() call. This approach 5155 maintains its own track of synchronize_rcu() callers, so it 5156 does not interact with regular callbacks because it does not 5157 use a call_rcu[_hurry]() path. Please note, this is for a 5158 normal grace period. 5159 5160 How to enable it: 5161 5162 echo 1 > /sys/module/rcutree/parameters/rcu_normal_wake_from_gp 5163 or pass a boot parameter "rcutree.rcu_normal_wake_from_gp=1" 5164 5165 Default is 0. 5166 5167 rcuscale.gp_async= [KNL] 5168 Measure performance of asynchronous 5169 grace-period primitives such as call_rcu(). 5170 5171 rcuscale.gp_async_max= [KNL] 5172 Specify the maximum number of outstanding 5173 callbacks per writer thread. When a writer 5174 thread exceeds this limit, it invokes the 5175 corresponding flavor of rcu_barrier() to allow 5176 previously posted callbacks to drain. 5177 5178 rcuscale.gp_exp= [KNL] 5179 Measure performance of expedited synchronous 5180 grace-period primitives. 5181 5182 rcuscale.holdoff= [KNL] 5183 Set test-start holdoff period. The purpose of 5184 this parameter is to delay the start of the 5185 test until boot completes in order to avoid 5186 interference. 5187 5188 rcuscale.kfree_by_call_rcu= [KNL] 5189 In kernels built with CONFIG_RCU_LAZY=y, test 5190 call_rcu() instead of kfree_rcu(). 5191 5192 rcuscale.kfree_mult= [KNL] 5193 Instead of allocating an object of size kfree_obj, 5194 allocate one of kfree_mult * sizeof(kfree_obj). 5195 Defaults to 1. 5196 5197 rcuscale.kfree_rcu_test= [KNL] 5198 Set to measure performance of kfree_rcu() flooding. 5199 5200 rcuscale.kfree_rcu_test_double= [KNL] 5201 Test the double-argument variant of kfree_rcu(). 5202 If this parameter has the same value as 5203 rcuscale.kfree_rcu_test_single, both the single- 5204 and double-argument variants are tested. 5205 5206 rcuscale.kfree_rcu_test_single= [KNL] 5207 Test the single-argument variant of kfree_rcu(). 5208 If this parameter has the same value as 5209 rcuscale.kfree_rcu_test_double, both the single- 5210 and double-argument variants are tested. 5211 5212 rcuscale.kfree_nthreads= [KNL] 5213 The number of threads running loops of kfree_rcu(). 5214 5215 rcuscale.kfree_alloc_num= [KNL] 5216 Number of allocations and frees done in an iteration. 5217 5218 rcuscale.kfree_loops= [KNL] 5219 Number of loops doing rcuscale.kfree_alloc_num number 5220 of allocations and frees. 5221 5222 rcuscale.minruntime= [KNL] 5223 Set the minimum test run time in seconds. This 5224 does not affect the data-collection interval, 5225 but instead allows better measurement of things 5226 like CPU consumption. 5227 5228 rcuscale.nreaders= [KNL] 5229 Set number of RCU readers. The value -1 selects 5230 N, where N is the number of CPUs. A value 5231 "n" less than -1 selects N-n+1, where N is again 5232 the number of CPUs. For example, -2 selects N 5233 (the number of CPUs), -3 selects N+1, and so on. 5234 A value of "n" less than or equal to -N selects 5235 a single reader. 5236 5237 rcuscale.nwriters= [KNL] 5238 Set number of RCU writers. The values operate 5239 the same as for rcuscale.nreaders. 5240 N, where N is the number of CPUs 5241 5242 rcuscale.scale_type= [KNL] 5243 Specify the RCU implementation to test. 5244 5245 rcuscale.shutdown= [KNL] 5246 Shut the system down after performance tests 5247 complete. This is useful for hands-off automated 5248 testing. 5249 5250 rcuscale.verbose= [KNL] 5251 Enable additional printk() statements. 5252 5253 rcuscale.writer_holdoff= [KNL] 5254 Write-side holdoff between grace periods, 5255 in microseconds. The default of zero says 5256 no holdoff. 5257 5258 rcuscale.writer_holdoff_jiffies= [KNL] 5259 Additional write-side holdoff between grace 5260 periods, but in jiffies. The default of zero 5261 says no holdoff. 5262 5263 rcutorture.fqs_duration= [KNL] 5264 Set duration of force_quiescent_state bursts 5265 in microseconds. 5266 5267 rcutorture.fqs_holdoff= [KNL] 5268 Set holdoff time within force_quiescent_state bursts 5269 in microseconds. 5270 5271 rcutorture.fqs_stutter= [KNL] 5272 Set wait time between force_quiescent_state bursts 5273 in seconds. 5274 5275 rcutorture.fwd_progress= [KNL] 5276 Specifies the number of kthreads to be used 5277 for RCU grace-period forward-progress testing 5278 for the types of RCU supporting this notion. 5279 Defaults to 1 kthread, values less than zero or 5280 greater than the number of CPUs cause the number 5281 of CPUs to be used. 5282 5283 rcutorture.fwd_progress_div= [KNL] 5284 Specify the fraction of a CPU-stall-warning 5285 period to do tight-loop forward-progress testing. 5286 5287 rcutorture.fwd_progress_holdoff= [KNL] 5288 Number of seconds to wait between successive 5289 forward-progress tests. 5290 5291 rcutorture.fwd_progress_need_resched= [KNL] 5292 Enclose cond_resched() calls within checks for 5293 need_resched() during tight-loop forward-progress 5294 testing. 5295 5296 rcutorture.gp_cond= [KNL] 5297 Use conditional/asynchronous update-side 5298 primitives, if available. 5299 5300 rcutorture.gp_exp= [KNL] 5301 Use expedited update-side primitives, if available. 5302 5303 rcutorture.gp_normal= [KNL] 5304 Use normal (non-expedited) asynchronous 5305 update-side primitives, if available. 5306 5307 rcutorture.gp_sync= [KNL] 5308 Use normal (non-expedited) synchronous 5309 update-side primitives, if available. If all 5310 of rcutorture.gp_cond=, rcutorture.gp_exp=, 5311 rcutorture.gp_normal=, and rcutorture.gp_sync= 5312 are zero, rcutorture acts as if is interpreted 5313 they are all non-zero. 5314 5315 rcutorture.irqreader= [KNL] 5316 Run RCU readers from irq handlers, or, more 5317 accurately, from a timer handler. Not all RCU 5318 flavors take kindly to this sort of thing. 5319 5320 rcutorture.leakpointer= [KNL] 5321 Leak an RCU-protected pointer out of the reader. 5322 This can of course result in splats, and is 5323 intended to test the ability of things like 5324 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect 5325 such leaks. 5326 5327 rcutorture.n_barrier_cbs= [KNL] 5328 Set callbacks/threads for rcu_barrier() testing. 5329 5330 rcutorture.nfakewriters= [KNL] 5331 Set number of concurrent RCU writers. These just 5332 stress RCU, they don't participate in the actual 5333 test, hence the "fake". 5334 5335 rcutorture.nocbs_nthreads= [KNL] 5336 Set number of RCU callback-offload togglers. 5337 Zero (the default) disables toggling. 5338 5339 rcutorture.nocbs_toggle= [KNL] 5340 Set the delay in milliseconds between successive 5341 callback-offload toggling attempts. 5342 5343 rcutorture.nreaders= [KNL] 5344 Set number of RCU readers. The value -1 selects 5345 N-1, where N is the number of CPUs. A value 5346 "n" less than -1 selects N-n-2, where N is again 5347 the number of CPUs. For example, -2 selects N 5348 (the number of CPUs), -3 selects N+1, and so on. 5349 5350 rcutorture.object_debug= [KNL] 5351 Enable debug-object double-call_rcu() testing. 5352 5353 rcutorture.onoff_holdoff= [KNL] 5354 Set time (s) after boot for CPU-hotplug testing. 5355 5356 rcutorture.onoff_interval= [KNL] 5357 Set time (jiffies) between CPU-hotplug operations, 5358 or zero to disable CPU-hotplug testing. 5359 5360 rcutorture.read_exit= [KNL] 5361 Set the number of read-then-exit kthreads used 5362 to test the interaction of RCU updaters and 5363 task-exit processing. 5364 5365 rcutorture.read_exit_burst= [KNL] 5366 The number of times in a given read-then-exit 5367 episode that a set of read-then-exit kthreads 5368 is spawned. 5369 5370 rcutorture.read_exit_delay= [KNL] 5371 The delay, in seconds, between successive 5372 read-then-exit testing episodes. 5373 5374 rcutorture.shuffle_interval= [KNL] 5375 Set task-shuffle interval (s). Shuffling tasks 5376 allows some CPUs to go into dyntick-idle mode 5377 during the rcutorture test. 5378 5379 rcutorture.shutdown_secs= [KNL] 5380 Set time (s) after boot system shutdown. This 5381 is useful for hands-off automated testing. 5382 5383 rcutorture.stall_cpu= [KNL] 5384 Duration of CPU stall (s) to test RCU CPU stall 5385 warnings, zero to disable. 5386 5387 rcutorture.stall_cpu_block= [KNL] 5388 Sleep while stalling if set. This will result 5389 in warnings from preemptible RCU in addition to 5390 any other stall-related activity. Note that 5391 in kernels built with CONFIG_PREEMPTION=n and 5392 CONFIG_PREEMPT_COUNT=y, this parameter will 5393 cause the CPU to pass through a quiescent state. 5394 Given CONFIG_PREEMPTION=n, this will suppress 5395 RCU CPU stall warnings, but will instead result 5396 in scheduling-while-atomic splats. 5397 5398 Use of this module parameter results in splats. 5399 5400 5401 rcutorture.stall_cpu_holdoff= [KNL] 5402 Time to wait (s) after boot before inducing stall. 5403 5404 rcutorture.stall_cpu_irqsoff= [KNL] 5405 Disable interrupts while stalling if set. 5406 5407 rcutorture.stall_gp_kthread= [KNL] 5408 Duration (s) of forced sleep within RCU 5409 grace-period kthread to test RCU CPU stall 5410 warnings, zero to disable. If both stall_cpu 5411 and stall_gp_kthread are specified, the 5412 kthread is starved first, then the CPU. 5413 5414 rcutorture.stat_interval= [KNL] 5415 Time (s) between statistics printk()s. 5416 5417 rcutorture.stutter= [KNL] 5418 Time (s) to stutter testing, for example, specifying 5419 five seconds causes the test to run for five seconds, 5420 wait for five seconds, and so on. This tests RCU's 5421 ability to transition abruptly to and from idle. 5422 5423 rcutorture.test_boost= [KNL] 5424 Test RCU priority boosting? 0=no, 1=maybe, 2=yes. 5425 "Maybe" means test if the RCU implementation 5426 under test support RCU priority boosting. 5427 5428 rcutorture.test_boost_duration= [KNL] 5429 Duration (s) of each individual boost test. 5430 5431 rcutorture.test_boost_interval= [KNL] 5432 Interval (s) between each boost test. 5433 5434 rcutorture.test_no_idle_hz= [KNL] 5435 Test RCU's dyntick-idle handling. See also the 5436 rcutorture.shuffle_interval parameter. 5437 5438 rcutorture.torture_type= [KNL] 5439 Specify the RCU implementation to test. 5440 5441 rcutorture.verbose= [KNL] 5442 Enable additional printk() statements. 5443 5444 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL] 5445 Dump ftrace buffer after reporting RCU CPU 5446 stall warning. 5447 5448 rcupdate.rcu_cpu_stall_notifiers= [KNL] 5449 Provide RCU CPU stall notifiers, but see the 5450 warnings in the RCU_CPU_STALL_NOTIFIER Kconfig 5451 option's help text. TL;DR: You almost certainly 5452 do not want rcupdate.rcu_cpu_stall_notifiers. 5453 5454 rcupdate.rcu_cpu_stall_suppress= [KNL] 5455 Suppress RCU CPU stall warning messages. 5456 5457 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL] 5458 Suppress RCU CPU stall warning messages and 5459 rcutorture writer stall warnings that occur 5460 during early boot, that is, during the time 5461 before the init task is spawned. 5462 5463 rcupdate.rcu_cpu_stall_timeout= [KNL] 5464 Set timeout for RCU CPU stall warning messages. 5465 The value is in seconds and the maximum allowed 5466 value is 300 seconds. 5467 5468 rcupdate.rcu_exp_cpu_stall_timeout= [KNL] 5469 Set timeout for expedited RCU CPU stall warning 5470 messages. The value is in milliseconds 5471 and the maximum allowed value is 21000 5472 milliseconds. Please note that this value is 5473 adjusted to an arch timer tick resolution. 5474 Setting this to zero causes the value from 5475 rcupdate.rcu_cpu_stall_timeout to be used (after 5476 conversion from seconds to milliseconds). 5477 5478 rcupdate.rcu_cpu_stall_cputime= [KNL] 5479 Provide statistics on the cputime and count of 5480 interrupts and tasks during the sampling period. For 5481 multiple continuous RCU stalls, all sampling periods 5482 begin at half of the first RCU stall timeout. 5483 5484 rcupdate.rcu_exp_stall_task_details= [KNL] 5485 Print stack dumps of any tasks blocking the 5486 current expedited RCU grace period during an 5487 expedited RCU CPU stall warning. 5488 5489 rcupdate.rcu_expedited= [KNL] 5490 Use expedited grace-period primitives, for 5491 example, synchronize_rcu_expedited() instead 5492 of synchronize_rcu(). This reduces latency, 5493 but can increase CPU utilization, degrade 5494 real-time latency, and degrade energy efficiency. 5495 No effect on CONFIG_TINY_RCU kernels. 5496 5497 rcupdate.rcu_normal= [KNL] 5498 Use only normal grace-period primitives, 5499 for example, synchronize_rcu() instead of 5500 synchronize_rcu_expedited(). This improves 5501 real-time latency, CPU utilization, and 5502 energy efficiency, but can expose users to 5503 increased grace-period latency. This parameter 5504 overrides rcupdate.rcu_expedited. No effect on 5505 CONFIG_TINY_RCU kernels. 5506 5507 rcupdate.rcu_normal_after_boot= [KNL] 5508 Once boot has completed (that is, after 5509 rcu_end_inkernel_boot() has been invoked), use 5510 only normal grace-period primitives. No effect 5511 on CONFIG_TINY_RCU kernels. 5512 5513 But note that CONFIG_PREEMPT_RT=y kernels enables 5514 this kernel boot parameter, forcibly setting 5515 it to the value one, that is, converting any 5516 post-boot attempt at an expedited RCU grace 5517 period to instead use normal non-expedited 5518 grace-period processing. 5519 5520 rcupdate.rcu_task_collapse_lim= [KNL] 5521 Set the maximum number of callbacks present 5522 at the beginning of a grace period that allows 5523 the RCU Tasks flavors to collapse back to using 5524 a single callback queue. This switching only 5525 occurs when rcupdate.rcu_task_enqueue_lim is 5526 set to the default value of -1. 5527 5528 rcupdate.rcu_task_contend_lim= [KNL] 5529 Set the minimum number of callback-queuing-time 5530 lock-contention events per jiffy required to 5531 cause the RCU Tasks flavors to switch to per-CPU 5532 callback queuing. This switching only occurs 5533 when rcupdate.rcu_task_enqueue_lim is set to 5534 the default value of -1. 5535 5536 rcupdate.rcu_task_enqueue_lim= [KNL] 5537 Set the number of callback queues to use for the 5538 RCU Tasks family of RCU flavors. The default 5539 of -1 allows this to be automatically (and 5540 dynamically) adjusted. This parameter is intended 5541 for use in testing. 5542 5543 rcupdate.rcu_task_ipi_delay= [KNL] 5544 Set time in jiffies during which RCU tasks will 5545 avoid sending IPIs, starting with the beginning 5546 of a given grace period. Setting a large 5547 number avoids disturbing real-time workloads, 5548 but lengthens grace periods. 5549 5550 rcupdate.rcu_task_lazy_lim= [KNL] 5551 Number of callbacks on a given CPU that will 5552 cancel laziness on that CPU. Use -1 to disable 5553 cancellation of laziness, but be advised that 5554 doing so increases the danger of OOM due to 5555 callback flooding. 5556 5557 rcupdate.rcu_task_stall_info= [KNL] 5558 Set initial timeout in jiffies for RCU task stall 5559 informational messages, which give some indication 5560 of the problem for those not patient enough to 5561 wait for ten minutes. Informational messages are 5562 only printed prior to the stall-warning message 5563 for a given grace period. Disable with a value 5564 less than or equal to zero. Defaults to ten 5565 seconds. A change in value does not take effect 5566 until the beginning of the next grace period. 5567 5568 rcupdate.rcu_task_stall_info_mult= [KNL] 5569 Multiplier for time interval between successive 5570 RCU task stall informational messages for a given 5571 RCU tasks grace period. This value is clamped 5572 to one through ten, inclusive. It defaults to 5573 the value three, so that the first informational 5574 message is printed 10 seconds into the grace 5575 period, the second at 40 seconds, the third at 5576 160 seconds, and then the stall warning at 600 5577 seconds would prevent a fourth at 640 seconds. 5578 5579 rcupdate.rcu_task_stall_timeout= [KNL] 5580 Set timeout in jiffies for RCU task stall 5581 warning messages. Disable with a value less 5582 than or equal to zero. Defaults to ten minutes. 5583 A change in value does not take effect until 5584 the beginning of the next grace period. 5585 5586 rcupdate.rcu_tasks_lazy_ms= [KNL] 5587 Set timeout in milliseconds RCU Tasks asynchronous 5588 callback batching for call_rcu_tasks(). 5589 A negative value will take the default. A value 5590 of zero will disable batching. Batching is 5591 always disabled for synchronize_rcu_tasks(). 5592 5593 rcupdate.rcu_tasks_rude_lazy_ms= [KNL] 5594 Set timeout in milliseconds RCU Tasks 5595 Rude asynchronous callback batching for 5596 call_rcu_tasks_rude(). A negative value 5597 will take the default. A value of zero will 5598 disable batching. Batching is always disabled 5599 for synchronize_rcu_tasks_rude(). 5600 5601 rcupdate.rcu_tasks_trace_lazy_ms= [KNL] 5602 Set timeout in milliseconds RCU Tasks 5603 Trace asynchronous callback batching for 5604 call_rcu_tasks_trace(). A negative value 5605 will take the default. A value of zero will 5606 disable batching. Batching is always disabled 5607 for synchronize_rcu_tasks_trace(). 5608 5609 rcupdate.rcu_self_test= [KNL] 5610 Run the RCU early boot self tests 5611 5612 rdinit= [KNL] 5613 Format: <full_path> 5614 Run specified binary instead of /init from the ramdisk, 5615 used for early userspace startup. See initrd. 5616 5617 rdrand= [X86,EARLY] 5618 force - Override the decision by the kernel to hide the 5619 advertisement of RDRAND support (this affects 5620 certain AMD processors because of buggy BIOS 5621 support, specifically around the suspend/resume 5622 path). 5623 5624 rdt= [HW,X86,RDT] 5625 Turn on/off individual RDT features. List is: 5626 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp, 5627 mba, smba, bmec. 5628 E.g. to turn on cmt and turn off mba use: 5629 rdt=cmt,!mba 5630 5631 reboot= [KNL] 5632 Format (x86 or x86_64): 5633 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] | d[efault] \ 5634 [[,]s[mp]#### \ 5635 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \ 5636 [[,]f[orce] 5637 Where reboot_mode is one of warm (soft) or cold (hard) or gpio 5638 (prefix with 'panic_' to set mode for panic 5639 reboot only), 5640 reboot_type is one of bios, acpi, kbd, triple, efi, or pci, 5641 reboot_force is either force or not specified, 5642 reboot_cpu is s[mp]#### with #### being the processor 5643 to be used for rebooting. 5644 5645 refscale.holdoff= [KNL] 5646 Set test-start holdoff period. The purpose of 5647 this parameter is to delay the start of the 5648 test until boot completes in order to avoid 5649 interference. 5650 5651 refscale.lookup_instances= [KNL] 5652 Number of data elements to use for the forms of 5653 SLAB_TYPESAFE_BY_RCU testing. A negative number 5654 is negated and multiplied by nr_cpu_ids, while 5655 zero specifies nr_cpu_ids. 5656 5657 refscale.loops= [KNL] 5658 Set the number of loops over the synchronization 5659 primitive under test. Increasing this number 5660 reduces noise due to loop start/end overhead, 5661 but the default has already reduced the per-pass 5662 noise to a handful of picoseconds on ca. 2020 5663 x86 laptops. 5664 5665 refscale.nreaders= [KNL] 5666 Set number of readers. The default value of -1 5667 selects N, where N is roughly 75% of the number 5668 of CPUs. A value of zero is an interesting choice. 5669 5670 refscale.nruns= [KNL] 5671 Set number of runs, each of which is dumped onto 5672 the console log. 5673 5674 refscale.readdelay= [KNL] 5675 Set the read-side critical-section duration, 5676 measured in microseconds. 5677 5678 refscale.scale_type= [KNL] 5679 Specify the read-protection implementation to test. 5680 5681 refscale.shutdown= [KNL] 5682 Shut down the system at the end of the performance 5683 test. This defaults to 1 (shut it down) when 5684 refscale is built into the kernel and to 0 (leave 5685 it running) when refscale is built as a module. 5686 5687 refscale.verbose= [KNL] 5688 Enable additional printk() statements. 5689 5690 refscale.verbose_batched= [KNL] 5691 Batch the additional printk() statements. If zero 5692 (the default) or negative, print everything. Otherwise, 5693 print every Nth verbose statement, where N is the value 5694 specified. 5695 5696 regulator_ignore_unused 5697 [REGULATOR] 5698 Prevents regulator framework from disabling regulators 5699 that are unused, due no driver claiming them. This may 5700 be useful for debug and development, but should not be 5701 needed on a platform with proper driver support. 5702 5703 relax_domain_level= 5704 [KNL, SMP] Set scheduler's default relax_domain_level. 5705 See Documentation/admin-guide/cgroup-v1/cpusets.rst. 5706 5707 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory 5708 Format: <base1>,<size1>[,<base2>,<size2>,...] 5709 Reserve I/O ports or memory so the kernel won't use 5710 them. If <base> is less than 0x10000, the region 5711 is assumed to be I/O ports; otherwise it is memory. 5712 5713 reservetop= [X86-32,EARLY] 5714 Format: nn[KMG] 5715 Reserves a hole at the top of the kernel virtual 5716 address space. 5717 5718 reset_devices [KNL] Force drivers to reset the underlying device 5719 during initialization. 5720 5721 resume= [SWSUSP] 5722 Specify the partition device for software suspend 5723 Format: 5724 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>} 5725 5726 resume_offset= [SWSUSP] 5727 Specify the offset from the beginning of the partition 5728 given by "resume=" at which the swap header is located, 5729 in <PAGE_SIZE> units (needed only for swap files). 5730 See Documentation/power/swsusp-and-swap-files.rst 5731 5732 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to 5733 read the resume files 5734 5735 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up. 5736 Useful for devices that are detected asynchronously 5737 (e.g. USB and MMC devices). 5738 5739 retain_initrd [RAM] Keep initrd memory after extraction. After boot, it will 5740 be accessible via /sys/firmware/initrd. 5741 5742 retbleed= [X86] Control mitigation of RETBleed (Arbitrary 5743 Speculative Code Execution with Return Instructions) 5744 vulnerability. 5745 5746 AMD-based UNRET and IBPB mitigations alone do not stop 5747 sibling threads from influencing the predictions of other 5748 sibling threads. For that reason, STIBP is used on pro- 5749 cessors that support it, and mitigate SMT on processors 5750 that don't. 5751 5752 off - no mitigation 5753 auto - automatically select a migitation 5754 auto,nosmt - automatically select a mitigation, 5755 disabling SMT if necessary for 5756 the full mitigation (only on Zen1 5757 and older without STIBP). 5758 ibpb - On AMD, mitigate short speculation 5759 windows on basic block boundaries too. 5760 Safe, highest perf impact. It also 5761 enables STIBP if present. Not suitable 5762 on Intel. 5763 ibpb,nosmt - Like "ibpb" above but will disable SMT 5764 when STIBP is not available. This is 5765 the alternative for systems which do not 5766 have STIBP. 5767 unret - Force enable untrained return thunks, 5768 only effective on AMD f15h-f17h based 5769 systems. 5770 unret,nosmt - Like unret, but will disable SMT when STIBP 5771 is not available. This is the alternative for 5772 systems which do not have STIBP. 5773 5774 Selecting 'auto' will choose a mitigation method at run 5775 time according to the CPU. 5776 5777 Not specifying this option is equivalent to retbleed=auto. 5778 5779 rfkill.default_state= 5780 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm, 5781 etc. communication is blocked by default. 5782 1 Unblocked. 5783 5784 rfkill.master_switch_mode= 5785 0 The "airplane mode" button does nothing. 5786 1 The "airplane mode" button toggles between everything 5787 blocked and the previous configuration. 5788 2 The "airplane mode" button toggles between everything 5789 blocked and everything unblocked. 5790 5791 rhash_entries= [KNL,NET] 5792 Set number of hash buckets for route cache 5793 5794 ring3mwait=disable 5795 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported 5796 CPUs. 5797 5798 riscv_isa_fallback [RISCV,EARLY] 5799 When CONFIG_RISCV_ISA_FALLBACK is not enabled, permit 5800 falling back to detecting extension support by parsing 5801 "riscv,isa" property on devicetree systems when the 5802 replacement properties are not found. See the Kconfig 5803 entry for RISCV_ISA_FALLBACK. 5804 5805 ro [KNL] Mount root device read-only on boot 5806 5807 rodata= [KNL,EARLY] 5808 on Mark read-only kernel memory as read-only (default). 5809 off Leave read-only kernel memory writable for debugging. 5810 full Mark read-only kernel memory and aliases as read-only 5811 [arm64] 5812 5813 rockchip.usb_uart 5814 [EARLY] 5815 Enable the uart passthrough on the designated usb port 5816 on Rockchip SoCs. When active, the signals of the 5817 debug-uart get routed to the D+ and D- pins of the usb 5818 port and the regular usb controller gets disabled. 5819 5820 root= [KNL] Root filesystem 5821 Usually this a a block device specifier of some kind, 5822 see the early_lookup_bdev comment in 5823 block/early-lookup.c for details. 5824 Alternatively this can be "ram" for the legacy initial 5825 ramdisk, "nfs" and "cifs" for root on a network file 5826 system, or "mtd" and "ubi" for mounting from raw flash. 5827 5828 rootdelay= [KNL] Delay (in seconds) to pause before attempting to 5829 mount the root filesystem 5830 5831 rootflags= [KNL] Set root filesystem mount option string 5832 5833 rootfstype= [KNL] Set root filesystem type 5834 5835 rootwait [KNL] Wait (indefinitely) for root device to show up. 5836 Useful for devices that are detected asynchronously 5837 (e.g. USB and MMC devices). 5838 5839 rootwait= [KNL] Maximum time (in seconds) to wait for root device 5840 to show up before attempting to mount the root 5841 filesystem. 5842 5843 rproc_mem=nn[KMG][@address] 5844 [KNL,ARM,CMA] Remoteproc physical memory block. 5845 Memory area to be used by remote processor image, 5846 managed by CMA. 5847 5848 rw [KNL] Mount root device read-write on boot 5849 5850 S [KNL] Run init in single mode 5851 5852 s390_iommu= [HW,S390] 5853 Set s390 IOTLB flushing mode 5854 strict 5855 With strict flushing every unmap operation will result 5856 in an IOTLB flush. Default is lazy flushing before 5857 reuse, which is faster. Deprecated, equivalent to 5858 iommu.strict=1. 5859 5860 s390_iommu_aperture= [KNL,S390] 5861 Specifies the size of the per device DMA address space 5862 accessible through the DMA and IOMMU APIs as a decimal 5863 factor of the size of main memory. 5864 The default is 1 meaning that one can concurrently use 5865 as many DMA addresses as physical memory is installed, 5866 if supported by hardware, and thus map all of memory 5867 once. With a value of 2 one can map all of memory twice 5868 and so on. As a special case a factor of 0 imposes no 5869 restrictions other than those given by hardware at the 5870 cost of significant additional memory use for tables. 5871 5872 sa1100ir [NET] 5873 See drivers/net/irda/sa1100_ir.c. 5874 5875 sched_verbose [KNL,EARLY] Enables verbose scheduler debug messages. 5876 5877 schedstats= [KNL,X86] Enable or disable scheduled statistics. 5878 Allowed values are enable and disable. This feature 5879 incurs a small amount of overhead in the scheduler 5880 but is useful for debugging and performance tuning. 5881 5882 sched_thermal_decay_shift= 5883 [Deprecated] 5884 [KNL, SMP] Set a decay shift for scheduler thermal 5885 pressure signal. Thermal pressure signal follows the 5886 default decay period of other scheduler pelt 5887 signals(usually 32 ms but configurable). Setting 5888 sched_thermal_decay_shift will left shift the decay 5889 period for the thermal pressure signal by the shift 5890 value. 5891 i.e. with the default pelt decay period of 32 ms 5892 sched_thermal_decay_shift thermal pressure decay pr 5893 1 64 ms 5894 2 128 ms 5895 and so on. 5896 Format: integer between 0 and 10 5897 Default is 0. 5898 5899 scftorture.holdoff= [KNL] 5900 Number of seconds to hold off before starting 5901 test. Defaults to zero for module insertion and 5902 to 10 seconds for built-in smp_call_function() 5903 tests. 5904 5905 scftorture.longwait= [KNL] 5906 Request ridiculously long waits randomly selected 5907 up to the chosen limit in seconds. Zero (the 5908 default) disables this feature. Please note 5909 that requesting even small non-zero numbers of 5910 seconds can result in RCU CPU stall warnings, 5911 softlockup complaints, and so on. 5912 5913 scftorture.nthreads= [KNL] 5914 Number of kthreads to spawn to invoke the 5915 smp_call_function() family of functions. 5916 The default of -1 specifies a number of kthreads 5917 equal to the number of CPUs. 5918 5919 scftorture.onoff_holdoff= [KNL] 5920 Number seconds to wait after the start of the 5921 test before initiating CPU-hotplug operations. 5922 5923 scftorture.onoff_interval= [KNL] 5924 Number seconds to wait between successive 5925 CPU-hotplug operations. Specifying zero (which 5926 is the default) disables CPU-hotplug operations. 5927 5928 scftorture.shutdown_secs= [KNL] 5929 The number of seconds following the start of the 5930 test after which to shut down the system. The 5931 default of zero avoids shutting down the system. 5932 Non-zero values are useful for automated tests. 5933 5934 scftorture.stat_interval= [KNL] 5935 The number of seconds between outputting the 5936 current test statistics to the console. A value 5937 of zero disables statistics output. 5938 5939 scftorture.stutter_cpus= [KNL] 5940 The number of jiffies to wait between each change 5941 to the set of CPUs under test. 5942 5943 scftorture.use_cpus_read_lock= [KNL] 5944 Use use_cpus_read_lock() instead of the default 5945 preempt_disable() to disable CPU hotplug 5946 while invoking one of the smp_call_function*() 5947 functions. 5948 5949 scftorture.verbose= [KNL] 5950 Enable additional printk() statements. 5951 5952 scftorture.weight_single= [KNL] 5953 The probability weighting to use for the 5954 smp_call_function_single() function with a zero 5955 "wait" parameter. A value of -1 selects the 5956 default if all other weights are -1. However, 5957 if at least one weight has some other value, a 5958 value of -1 will instead select a weight of zero. 5959 5960 scftorture.weight_single_wait= [KNL] 5961 The probability weighting to use for the 5962 smp_call_function_single() function with a 5963 non-zero "wait" parameter. See weight_single. 5964 5965 scftorture.weight_many= [KNL] 5966 The probability weighting to use for the 5967 smp_call_function_many() function with a zero 5968 "wait" parameter. See weight_single. 5969 Note well that setting a high probability for 5970 this weighting can place serious IPI load 5971 on the system. 5972 5973 scftorture.weight_many_wait= [KNL] 5974 The probability weighting to use for the 5975 smp_call_function_many() function with a 5976 non-zero "wait" parameter. See weight_single 5977 and weight_many. 5978 5979 scftorture.weight_all= [KNL] 5980 The probability weighting to use for the 5981 smp_call_function_all() function with a zero 5982 "wait" parameter. See weight_single and 5983 weight_many. 5984 5985 scftorture.weight_all_wait= [KNL] 5986 The probability weighting to use for the 5987 smp_call_function_all() function with a 5988 non-zero "wait" parameter. See weight_single 5989 and weight_many. 5990 5991 skew_tick= [KNL,EARLY] Offset the periodic timer tick per cpu to mitigate 5992 xtime_lock contention on larger systems, and/or RCU lock 5993 contention on all systems with CONFIG_MAXSMP set. 5994 Format: { "0" | "1" } 5995 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1" 5996 1 -- enable. 5997 Note: increases power consumption, thus should only be 5998 enabled if running jitter sensitive (HPC/RT) workloads. 5999 6000 security= [SECURITY] Choose a legacy "major" security module to 6001 enable at boot. This has been deprecated by the 6002 "lsm=" parameter. 6003 6004 selinux= [SELINUX] Disable or enable SELinux at boot time. 6005 Format: { "0" | "1" } 6006 See security/selinux/Kconfig help text. 6007 0 -- disable. 6008 1 -- enable. 6009 Default value is 1. 6010 6011 serialnumber [BUGS=X86-32] 6012 6013 sev=option[,option...] [X86-64] See Documentation/arch/x86/x86_64/boot-options.rst 6014 6015 shapers= [NET] 6016 Maximal number of shapers. 6017 6018 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller 6019 Limit apic dumping. The parameter defines the maximal 6020 number of local apics being dumped. Also it is possible 6021 to set it to "all" by meaning -- no limit here. 6022 Format: { 1 (default) | 2 | ... | all }. 6023 The parameter valid if only apic=debug or 6024 apic=verbose is specified. 6025 Example: apic=debug show_lapic=all 6026 6027 simeth= [IA-64] 6028 simscsi= 6029 6030 slab_debug[=options[,slabs][;[options[,slabs]]...] [MM] 6031 Enabling slab_debug allows one to determine the 6032 culprit if slab objects become corrupted. Enabling 6033 slab_debug can create guard zones around objects and 6034 may poison objects when not in use. Also tracks the 6035 last alloc / free. For more information see 6036 Documentation/mm/slub.rst. 6037 (slub_debug legacy name also accepted for now) 6038 6039 slab_max_order= [MM] 6040 Determines the maximum allowed order for slabs. 6041 A high setting may cause OOMs due to memory 6042 fragmentation. For more information see 6043 Documentation/mm/slub.rst. 6044 (slub_max_order legacy name also accepted for now) 6045 6046 slab_merge [MM] 6047 Enable merging of slabs with similar size when the 6048 kernel is built without CONFIG_SLAB_MERGE_DEFAULT. 6049 (slub_merge legacy name also accepted for now) 6050 6051 slab_min_objects= [MM] 6052 The minimum number of objects per slab. SLUB will 6053 increase the slab order up to slab_max_order to 6054 generate a sufficiently large slab able to contain 6055 the number of objects indicated. The higher the number 6056 of objects the smaller the overhead of tracking slabs 6057 and the less frequently locks need to be acquired. 6058 For more information see Documentation/mm/slub.rst. 6059 (slub_min_objects legacy name also accepted for now) 6060 6061 slab_min_order= [MM] 6062 Determines the minimum page order for slabs. Must be 6063 lower or equal to slab_max_order. For more information see 6064 Documentation/mm/slub.rst. 6065 (slub_min_order legacy name also accepted for now) 6066 6067 slab_nomerge [MM] 6068 Disable merging of slabs with similar size. May be 6069 necessary if there is some reason to distinguish 6070 allocs to different slabs, especially in hardened 6071 environments where the risk of heap overflows and 6072 layout control by attackers can usually be 6073 frustrated by disabling merging. This will reduce 6074 most of the exposure of a heap attack to a single 6075 cache (risks via metadata attacks are mostly 6076 unchanged). Debug options disable merging on their 6077 own. 6078 For more information see Documentation/mm/slub.rst. 6079 (slub_nomerge legacy name also accepted for now) 6080 6081 slram= [HW,MTD] 6082 6083 smart2= [HW] 6084 Format: <io1>[,<io2>[,...,<io8>]] 6085 6086 smp.csd_lock_timeout= [KNL] 6087 Specify the period of time in milliseconds 6088 that smp_call_function() and friends will wait 6089 for a CPU to release the CSD lock. This is 6090 useful when diagnosing bugs involving CPUs 6091 disabling interrupts for extended periods 6092 of time. Defaults to 5,000 milliseconds, and 6093 setting a value of zero disables this feature. 6094 This feature may be more efficiently disabled 6095 using the csdlock_debug- kernel parameter. 6096 6097 smp.panic_on_ipistall= [KNL] 6098 If a csd_lock_timeout extends for more than 6099 the specified number of milliseconds, panic the 6100 system. By default, let CSD-lock acquisition 6101 take as long as they take. Specifying 300,000 6102 for this value provides a 5-minute timeout. 6103 6104 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices 6105 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port 6106 smsc-ircc2.ircc_sir= [HW] SIR base I/O port 6107 smsc-ircc2.ircc_fir= [HW] FIR base I/O port 6108 smsc-ircc2.ircc_irq= [HW] IRQ line 6109 smsc-ircc2.ircc_dma= [HW] DMA channel 6110 smsc-ircc2.ircc_transceiver= [HW] Transceiver type: 6111 0: Toshiba Satellite 1800 (GP data pin select) 6112 1: Fast pin select (default) 6113 2: ATC IRMode 6114 6115 smt= [KNL,MIPS,S390,EARLY] Set the maximum number of threads 6116 (logical CPUs) to use per physical CPU on systems 6117 capable of symmetric multithreading (SMT). Will 6118 be capped to the actual hardware limit. 6119 Format: <integer> 6120 Default: -1 (no limit) 6121 6122 softlockup_panic= 6123 [KNL] Should the soft-lockup detector generate panics. 6124 Format: 0 | 1 6125 6126 A value of 1 instructs the soft-lockup detector 6127 to panic the machine when a soft-lockup occurs. It is 6128 also controlled by the kernel.softlockup_panic sysctl 6129 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the 6130 respective build-time switch to that functionality. 6131 6132 softlockup_all_cpu_backtrace= 6133 [KNL] Should the soft-lockup detector generate 6134 backtraces on all cpus. 6135 Format: 0 | 1 6136 6137 sonypi.*= [HW] Sony Programmable I/O Control Device driver 6138 See Documentation/admin-guide/laptops/sonypi.rst 6139 6140 spectre_bhi= [X86] Control mitigation of Branch History Injection 6141 (BHI) vulnerability. This setting affects the 6142 deployment of the HW BHI control and the SW BHB 6143 clearing sequence. 6144 6145 on - (default) Enable the HW or SW mitigation 6146 as needed. 6147 off - Disable the mitigation. 6148 6149 spectre_v2= [X86,EARLY] Control mitigation of Spectre variant 2 6150 (indirect branch speculation) vulnerability. 6151 The default operation protects the kernel from 6152 user space attacks. 6153 6154 on - unconditionally enable, implies 6155 spectre_v2_user=on 6156 off - unconditionally disable, implies 6157 spectre_v2_user=off 6158 auto - kernel detects whether your CPU model is 6159 vulnerable 6160 6161 Selecting 'on' will, and 'auto' may, choose a 6162 mitigation method at run time according to the 6163 CPU, the available microcode, the setting of the 6164 CONFIG_MITIGATION_RETPOLINE configuration option, 6165 and the compiler with which the kernel was built. 6166 6167 Selecting 'on' will also enable the mitigation 6168 against user space to user space task attacks. 6169 6170 Selecting 'off' will disable both the kernel and 6171 the user space protections. 6172 6173 Specific mitigations can also be selected manually: 6174 6175 retpoline - replace indirect branches 6176 retpoline,generic - Retpolines 6177 retpoline,lfence - LFENCE; indirect branch 6178 retpoline,amd - alias for retpoline,lfence 6179 eibrs - Enhanced/Auto IBRS 6180 eibrs,retpoline - Enhanced/Auto IBRS + Retpolines 6181 eibrs,lfence - Enhanced/Auto IBRS + LFENCE 6182 ibrs - use IBRS to protect kernel 6183 6184 Not specifying this option is equivalent to 6185 spectre_v2=auto. 6186 6187 spectre_v2_user= 6188 [X86] Control mitigation of Spectre variant 2 6189 (indirect branch speculation) vulnerability between 6190 user space tasks 6191 6192 on - Unconditionally enable mitigations. Is 6193 enforced by spectre_v2=on 6194 6195 off - Unconditionally disable mitigations. Is 6196 enforced by spectre_v2=off 6197 6198 prctl - Indirect branch speculation is enabled, 6199 but mitigation can be enabled via prctl 6200 per thread. The mitigation control state 6201 is inherited on fork. 6202 6203 prctl,ibpb 6204 - Like "prctl" above, but only STIBP is 6205 controlled per thread. IBPB is issued 6206 always when switching between different user 6207 space processes. 6208 6209 seccomp 6210 - Same as "prctl" above, but all seccomp 6211 threads will enable the mitigation unless 6212 they explicitly opt out. 6213 6214 seccomp,ibpb 6215 - Like "seccomp" above, but only STIBP is 6216 controlled per thread. IBPB is issued 6217 always when switching between different 6218 user space processes. 6219 6220 auto - Kernel selects the mitigation depending on 6221 the available CPU features and vulnerability. 6222 6223 Default mitigation: "prctl" 6224 6225 Not specifying this option is equivalent to 6226 spectre_v2_user=auto. 6227 6228 spec_rstack_overflow= 6229 [X86,EARLY] Control RAS overflow mitigation on AMD Zen CPUs 6230 6231 off - Disable mitigation 6232 microcode - Enable microcode mitigation only 6233 safe-ret - Enable sw-only safe RET mitigation (default) 6234 ibpb - Enable mitigation by issuing IBPB on 6235 kernel entry 6236 ibpb-vmexit - Issue IBPB only on VMEXIT 6237 (cloud-specific mitigation) 6238 6239 spec_store_bypass_disable= 6240 [HW,EARLY] Control Speculative Store Bypass (SSB) Disable mitigation 6241 (Speculative Store Bypass vulnerability) 6242 6243 Certain CPUs are vulnerable to an exploit against a 6244 a common industry wide performance optimization known 6245 as "Speculative Store Bypass" in which recent stores 6246 to the same memory location may not be observed by 6247 later loads during speculative execution. The idea 6248 is that such stores are unlikely and that they can 6249 be detected prior to instruction retirement at the 6250 end of a particular speculation execution window. 6251 6252 In vulnerable processors, the speculatively forwarded 6253 store can be used in a cache side channel attack, for 6254 example to read memory to which the attacker does not 6255 directly have access (e.g. inside sandboxed code). 6256 6257 This parameter controls whether the Speculative Store 6258 Bypass optimization is used. 6259 6260 On x86 the options are: 6261 6262 on - Unconditionally disable Speculative Store Bypass 6263 off - Unconditionally enable Speculative Store Bypass 6264 auto - Kernel detects whether the CPU model contains an 6265 implementation of Speculative Store Bypass and 6266 picks the most appropriate mitigation. If the 6267 CPU is not vulnerable, "off" is selected. If the 6268 CPU is vulnerable the default mitigation is 6269 architecture and Kconfig dependent. See below. 6270 prctl - Control Speculative Store Bypass per thread 6271 via prctl. Speculative Store Bypass is enabled 6272 for a process by default. The state of the control 6273 is inherited on fork. 6274 seccomp - Same as "prctl" above, but all seccomp threads 6275 will disable SSB unless they explicitly opt out. 6276 6277 Default mitigations: 6278 X86: "prctl" 6279 6280 On powerpc the options are: 6281 6282 on,auto - On Power8 and Power9 insert a store-forwarding 6283 barrier on kernel entry and exit. On Power7 6284 perform a software flush on kernel entry and 6285 exit. 6286 off - No action. 6287 6288 Not specifying this option is equivalent to 6289 spec_store_bypass_disable=auto. 6290 6291 spia_io_base= [HW,MTD] 6292 spia_fio_base= 6293 spia_pedr= 6294 spia_peddr= 6295 6296 split_lock_detect= 6297 [X86] Enable split lock detection or bus lock detection 6298 6299 When enabled (and if hardware support is present), atomic 6300 instructions that access data across cache line 6301 boundaries will result in an alignment check exception 6302 for split lock detection or a debug exception for 6303 bus lock detection. 6304 6305 off - not enabled 6306 6307 warn - the kernel will emit rate-limited warnings 6308 about applications triggering the #AC 6309 exception or the #DB exception. This mode is 6310 the default on CPUs that support split lock 6311 detection or bus lock detection. Default 6312 behavior is by #AC if both features are 6313 enabled in hardware. 6314 6315 fatal - the kernel will send SIGBUS to applications 6316 that trigger the #AC exception or the #DB 6317 exception. Default behavior is by #AC if 6318 both features are enabled in hardware. 6319 6320 ratelimit:N - 6321 Set system wide rate limit to N bus locks 6322 per second for bus lock detection. 6323 0 < N <= 1000. 6324 6325 N/A for split lock detection. 6326 6327 6328 If an #AC exception is hit in the kernel or in 6329 firmware (i.e. not while executing in user mode) 6330 the kernel will oops in either "warn" or "fatal" 6331 mode. 6332 6333 #DB exception for bus lock is triggered only when 6334 CPL > 0. 6335 6336 srbds= [X86,INTEL,EARLY] 6337 Control the Special Register Buffer Data Sampling 6338 (SRBDS) mitigation. 6339 6340 Certain CPUs are vulnerable to an MDS-like 6341 exploit which can leak bits from the random 6342 number generator. 6343 6344 By default, this issue is mitigated by 6345 microcode. However, the microcode fix can cause 6346 the RDRAND and RDSEED instructions to become 6347 much slower. Among other effects, this will 6348 result in reduced throughput from /dev/urandom. 6349 6350 The microcode mitigation can be disabled with 6351 the following option: 6352 6353 off: Disable mitigation and remove 6354 performance impact to RDRAND and RDSEED 6355 6356 srcutree.big_cpu_lim [KNL] 6357 Specifies the number of CPUs constituting a 6358 large system, such that srcu_struct structures 6359 should immediately allocate an srcu_node array. 6360 This kernel-boot parameter defaults to 128, 6361 but takes effect only when the low-order four 6362 bits of srcutree.convert_to_big is equal to 3 6363 (decide at boot). 6364 6365 srcutree.convert_to_big [KNL] 6366 Specifies under what conditions an SRCU tree 6367 srcu_struct structure will be converted to big 6368 form, that is, with an rcu_node tree: 6369 6370 0: Never. 6371 1: At init_srcu_struct() time. 6372 2: When rcutorture decides to. 6373 3: Decide at boot time (default). 6374 0x1X: Above plus if high contention. 6375 6376 Either way, the srcu_node tree will be sized based 6377 on the actual runtime number of CPUs (nr_cpu_ids) 6378 instead of the compile-time CONFIG_NR_CPUS. 6379 6380 srcutree.counter_wrap_check [KNL] 6381 Specifies how frequently to check for 6382 grace-period sequence counter wrap for the 6383 srcu_data structure's ->srcu_gp_seq_needed field. 6384 The greater the number of bits set in this kernel 6385 parameter, the less frequently counter wrap will 6386 be checked for. Note that the bottom two bits 6387 are ignored. 6388 6389 srcutree.exp_holdoff [KNL] 6390 Specifies how many nanoseconds must elapse 6391 since the end of the last SRCU grace period for 6392 a given srcu_struct until the next normal SRCU 6393 grace period will be considered for automatic 6394 expediting. Set to zero to disable automatic 6395 expediting. 6396 6397 srcutree.srcu_max_nodelay [KNL] 6398 Specifies the number of no-delay instances 6399 per jiffy for which the SRCU grace period 6400 worker thread will be rescheduled with zero 6401 delay. Beyond this limit, worker thread will 6402 be rescheduled with a sleep delay of one jiffy. 6403 6404 srcutree.srcu_max_nodelay_phase [KNL] 6405 Specifies the per-grace-period phase, number of 6406 non-sleeping polls of readers. Beyond this limit, 6407 grace period worker thread will be rescheduled 6408 with a sleep delay of one jiffy, between each 6409 rescan of the readers, for a grace period phase. 6410 6411 srcutree.srcu_retry_check_delay [KNL] 6412 Specifies number of microseconds of non-sleeping 6413 delay between each non-sleeping poll of readers. 6414 6415 srcutree.small_contention_lim [KNL] 6416 Specifies the number of update-side contention 6417 events per jiffy will be tolerated before 6418 initiating a conversion of an srcu_struct 6419 structure to big form. Note that the value of 6420 srcutree.convert_to_big must have the 0x10 bit 6421 set for contention-based conversions to occur. 6422 6423 ssbd= [ARM64,HW,EARLY] 6424 Speculative Store Bypass Disable control 6425 6426 On CPUs that are vulnerable to the Speculative 6427 Store Bypass vulnerability and offer a 6428 firmware based mitigation, this parameter 6429 indicates how the mitigation should be used: 6430 6431 force-on: Unconditionally enable mitigation for 6432 for both kernel and userspace 6433 force-off: Unconditionally disable mitigation for 6434 for both kernel and userspace 6435 kernel: Always enable mitigation in the 6436 kernel, and offer a prctl interface 6437 to allow userspace to register its 6438 interest in being mitigated too. 6439 6440 stack_guard_gap= [MM] 6441 override the default stack gap protection. The value 6442 is in page units and it defines how many pages prior 6443 to (for stacks growing down) resp. after (for stacks 6444 growing up) the main stack are reserved for no other 6445 mapping. Default value is 256 pages. 6446 6447 stack_depot_disable= [KNL,EARLY] 6448 Setting this to true through kernel command line will 6449 disable the stack depot thereby saving the static memory 6450 consumed by the stack hash table. By default this is set 6451 to false. 6452 6453 stacktrace [FTRACE] 6454 Enabled the stack tracer on boot up. 6455 6456 stacktrace_filter=[function-list] 6457 [FTRACE] Limit the functions that the stack tracer 6458 will trace at boot up. function-list is a comma-separated 6459 list of functions. This list can be changed at run 6460 time by the stack_trace_filter file in the debugfs 6461 tracing directory. Note, this enables stack tracing 6462 and the stacktrace above is not needed. 6463 6464 sti= [PARISC,HW] 6465 Format: <num> 6466 Set the STI (builtin display/keyboard on the HP-PARISC 6467 machines) console (graphic card) which should be used 6468 as the initial boot-console. 6469 See also comment in drivers/video/console/sticore.c. 6470 6471 sti_font= [HW] 6472 See comment in drivers/video/console/sticore.c. 6473 6474 stifb= [HW] 6475 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]] 6476 6477 strict_sas_size= 6478 [X86] 6479 Format: <bool> 6480 Enable or disable strict sigaltstack size checks 6481 against the required signal frame size which 6482 depends on the supported FPU features. This can 6483 be used to filter out binaries which have 6484 not yet been made aware of AT_MINSIGSTKSZ. 6485 6486 stress_hpt [PPC,EARLY] 6487 Limits the number of kernel HPT entries in the hash 6488 page table to increase the rate of hash page table 6489 faults on kernel addresses. 6490 6491 stress_slb [PPC,EARLY] 6492 Limits the number of kernel SLB entries, and flushes 6493 them frequently to increase the rate of SLB faults 6494 on kernel addresses. 6495 6496 sunrpc.min_resvport= 6497 sunrpc.max_resvport= 6498 [NFS,SUNRPC] 6499 SunRPC servers often require that client requests 6500 originate from a privileged port (i.e. a port in the 6501 range 0 < portnr < 1024). 6502 An administrator who wishes to reserve some of these 6503 ports for other uses may adjust the range that the 6504 kernel's sunrpc client considers to be privileged 6505 using these two parameters to set the minimum and 6506 maximum port values. 6507 6508 sunrpc.svc_rpc_per_connection_limit= 6509 [NFS,SUNRPC] 6510 Limit the number of requests that the server will 6511 process in parallel from a single connection. 6512 The default value is 0 (no limit). 6513 6514 sunrpc.pool_mode= 6515 [NFS] 6516 Control how the NFS server code allocates CPUs to 6517 service thread pools. Depending on how many NICs 6518 you have and where their interrupts are bound, this 6519 option will affect which CPUs will do NFS serving. 6520 Note: this parameter cannot be changed while the 6521 NFS server is running. 6522 6523 auto the server chooses an appropriate mode 6524 automatically using heuristics 6525 global a single global pool contains all CPUs 6526 percpu one pool for each CPU 6527 pernode one pool for each NUMA node (equivalent 6528 to global on non-NUMA machines) 6529 6530 sunrpc.tcp_slot_table_entries= 6531 sunrpc.udp_slot_table_entries= 6532 [NFS,SUNRPC] 6533 Sets the upper limit on the number of simultaneous 6534 RPC calls that can be sent from the client to a 6535 server. Increasing these values may allow you to 6536 improve throughput, but will also increase the 6537 amount of memory reserved for use by the client. 6538 6539 suspend.pm_test_delay= 6540 [SUSPEND] 6541 Sets the number of seconds to remain in a suspend test 6542 mode before resuming the system (see 6543 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG 6544 is set. Default value is 5. 6545 6546 svm= [PPC] 6547 Format: { on | off | y | n | 1 | 0 } 6548 This parameter controls use of the Protected 6549 Execution Facility on pSeries. 6550 6551 swiotlb= [ARM,IA-64,PPC,MIPS,X86,EARLY] 6552 Format: { <int> [,<int>] | force | noforce } 6553 <int> -- Number of I/O TLB slabs 6554 <int> -- Second integer after comma. Number of swiotlb 6555 areas with their own lock. Will be rounded up 6556 to a power of 2. 6557 force -- force using of bounce buffers even if they 6558 wouldn't be automatically used by the kernel 6559 noforce -- Never use bounce buffers (for debugging) 6560 6561 switches= [HW,M68k,EARLY] 6562 6563 sysctl.*= [KNL] 6564 Set a sysctl parameter, right before loading the init 6565 process, as if the value was written to the respective 6566 /proc/sys/... file. Both '.' and '/' are recognized as 6567 separators. Unrecognized parameters and invalid values 6568 are reported in the kernel log. Sysctls registered 6569 later by a loaded module cannot be set this way. 6570 Example: sysctl.vm.swappiness=40 6571 6572 sysrq_always_enabled 6573 [KNL] 6574 Ignore sysrq setting - this boot parameter will 6575 neutralize any effect of /proc/sys/kernel/sysrq. 6576 Useful for debugging. 6577 6578 tcpmhash_entries= [KNL,NET] 6579 Set the number of tcp_metrics_hash slots. 6580 Default value is 8192 or 16384 depending on total 6581 ram pages. This is used to specify the TCP metrics 6582 cache size. See Documentation/networking/ip-sysctl.rst 6583 "tcp_no_metrics_save" section for more details. 6584 6585 tdfx= [HW,DRM] 6586 6587 test_suspend= [SUSPEND] 6588 Format: { "mem" | "standby" | "freeze" }[,N] 6589 Specify "mem" (for Suspend-to-RAM) or "standby" (for 6590 standby suspend) or "freeze" (for suspend type freeze) 6591 as the system sleep state during system startup with 6592 the optional capability to repeat N number of times. 6593 The system is woken from this state using a 6594 wakeup-capable RTC alarm. 6595 6596 thash_entries= [KNL,NET] 6597 Set number of hash buckets for TCP connection 6598 6599 thermal.act= [HW,ACPI] 6600 -1: disable all active trip points in all thermal zones 6601 <degrees C>: override all lowest active trip points 6602 6603 thermal.crt= [HW,ACPI] 6604 -1: disable all critical trip points in all thermal zones 6605 <degrees C>: override all critical trip points 6606 6607 thermal.off= [HW,ACPI] 6608 1: disable ACPI thermal control 6609 6610 thermal.psv= [HW,ACPI] 6611 -1: disable all passive trip points 6612 <degrees C>: override all passive trip points to this 6613 value 6614 6615 thermal.tzp= [HW,ACPI] 6616 Specify global default ACPI thermal zone polling rate 6617 <deci-seconds>: poll all this frequency 6618 0: no polling (default) 6619 6620 threadirqs [KNL,EARLY] 6621 Force threading of all interrupt handlers except those 6622 marked explicitly IRQF_NO_THREAD. 6623 6624 topology= [S390,EARLY] 6625 Format: {off | on} 6626 Specify if the kernel should make use of the cpu 6627 topology information if the hardware supports this. 6628 The scheduler will make use of this information and 6629 e.g. base its process migration decisions on it. 6630 Default is on. 6631 6632 topology_updates= [KNL, PPC, NUMA] 6633 Format: {off} 6634 Specify if the kernel should ignore (off) 6635 topology updates sent by the hypervisor to this 6636 LPAR. 6637 6638 torture.disable_onoff_at_boot= [KNL] 6639 Prevent the CPU-hotplug component of torturing 6640 until after init has spawned. 6641 6642 torture.ftrace_dump_at_shutdown= [KNL] 6643 Dump the ftrace buffer at torture-test shutdown, 6644 even if there were no errors. This can be a 6645 very costly operation when many torture tests 6646 are running concurrently, especially on systems 6647 with rotating-rust storage. 6648 6649 torture.verbose_sleep_frequency= [KNL] 6650 Specifies how many verbose printk()s should be 6651 emitted between each sleep. The default of zero 6652 disables verbose-printk() sleeping. 6653 6654 torture.verbose_sleep_duration= [KNL] 6655 Duration of each verbose-printk() sleep in jiffies. 6656 6657 tp720= [HW,PS2] 6658 6659 tpm_suspend_pcr=[HW,TPM] 6660 Format: integer pcr id 6661 Specify that at suspend time, the tpm driver 6662 should extend the specified pcr with zeros, 6663 as a workaround for some chips which fail to 6664 flush the last written pcr on TPM_SaveState. 6665 This will guarantee that all the other pcrs 6666 are saved. 6667 6668 tpm_tis.interrupts= [HW,TPM] 6669 Enable interrupts for the MMIO based physical layer 6670 for the FIFO interface. By default it is set to false 6671 (0). For more information about TPM hardware interfaces 6672 defined by Trusted Computing Group (TCG) see 6673 https://trustedcomputinggroup.org/resource/pc-client-platform-tpm-profile-ptp-specification/ 6674 6675 tp_printk [FTRACE] 6676 Have the tracepoints sent to printk as well as the 6677 tracing ring buffer. This is useful for early boot up 6678 where the system hangs or reboots and does not give the 6679 option for reading the tracing buffer or performing a 6680 ftrace_dump_on_oops. 6681 6682 To turn off having tracepoints sent to printk, 6683 echo 0 > /proc/sys/kernel/tracepoint_printk 6684 Note, echoing 1 into this file without the 6685 tp_printk kernel cmdline option has no effect. 6686 6687 The tp_printk_stop_on_boot (see below) can also be used 6688 to stop the printing of events to console at 6689 late_initcall_sync. 6690 6691 ** CAUTION ** 6692 6693 Having tracepoints sent to printk() and activating high 6694 frequency tracepoints such as irq or sched, can cause 6695 the system to live lock. 6696 6697 tp_printk_stop_on_boot [FTRACE] 6698 When tp_printk (above) is set, it can cause a lot of noise 6699 on the console. It may be useful to only include the 6700 printing of events during boot up, as user space may 6701 make the system inoperable. 6702 6703 This command line option will stop the printing of events 6704 to console at the late_initcall_sync() time frame. 6705 6706 trace_buf_size=nn[KMG] 6707 [FTRACE] will set tracing buffer size on each cpu. 6708 6709 trace_clock= [FTRACE] Set the clock used for tracing events 6710 at boot up. 6711 local - Use the per CPU time stamp counter 6712 (converted into nanoseconds). Fast, but 6713 depending on the architecture, may not be 6714 in sync between CPUs. 6715 global - Event time stamps are synchronize across 6716 CPUs. May be slower than the local clock, 6717 but better for some race conditions. 6718 counter - Simple counting of events (1, 2, ..) 6719 note, some counts may be skipped due to the 6720 infrastructure grabbing the clock more than 6721 once per event. 6722 uptime - Use jiffies as the time stamp. 6723 perf - Use the same clock that perf uses. 6724 mono - Use ktime_get_mono_fast_ns() for time stamps. 6725 mono_raw - Use ktime_get_raw_fast_ns() for time 6726 stamps. 6727 boot - Use ktime_get_boot_fast_ns() for time stamps. 6728 Architectures may add more clocks. See 6729 Documentation/trace/ftrace.rst for more details. 6730 6731 trace_event=[event-list] 6732 [FTRACE] Set and start specified trace events in order 6733 to facilitate early boot debugging. The event-list is a 6734 comma-separated list of trace events to enable. See 6735 also Documentation/trace/events.rst 6736 6737 trace_instance=[instance-info] 6738 [FTRACE] Create a ring buffer instance early in boot up. 6739 This will be listed in: 6740 6741 /sys/kernel/tracing/instances 6742 6743 Events can be enabled at the time the instance is created 6744 via: 6745 6746 trace_instance=<name>,<system1>:<event1>,<system2>:<event2> 6747 6748 Note, the "<system*>:" portion is optional if the event is 6749 unique. 6750 6751 trace_instance=foo,sched:sched_switch,irq_handler_entry,initcall 6752 6753 will enable the "sched_switch" event (note, the "sched:" is optional, and 6754 the same thing would happen if it was left off). The irq_handler_entry 6755 event, and all events under the "initcall" system. 6756 6757 trace_options=[option-list] 6758 [FTRACE] Enable or disable tracer options at boot. 6759 The option-list is a comma delimited list of options 6760 that can be enabled or disabled just as if you were 6761 to echo the option name into 6762 6763 /sys/kernel/tracing/trace_options 6764 6765 For example, to enable stacktrace option (to dump the 6766 stack trace of each event), add to the command line: 6767 6768 trace_options=stacktrace 6769 6770 See also Documentation/trace/ftrace.rst "trace options" 6771 section. 6772 6773 trace_trigger=[trigger-list] 6774 [FTRACE] Add a event trigger on specific events. 6775 Set a trigger on top of a specific event, with an optional 6776 filter. 6777 6778 The format is is "trace_trigger=<event>.<trigger>[ if <filter>],..." 6779 Where more than one trigger may be specified that are comma deliminated. 6780 6781 For example: 6782 6783 trace_trigger="sched_switch.stacktrace if prev_state == 2" 6784 6785 The above will enable the "stacktrace" trigger on the "sched_switch" 6786 event but only trigger it if the "prev_state" of the "sched_switch" 6787 event is "2" (TASK_UNINTERUPTIBLE). 6788 6789 See also "Event triggers" in Documentation/trace/events.rst 6790 6791 6792 traceoff_on_warning 6793 [FTRACE] enable this option to disable tracing when a 6794 warning is hit. This turns off "tracing_on". Tracing can 6795 be enabled again by echoing '1' into the "tracing_on" 6796 file located in /sys/kernel/tracing/ 6797 6798 This option is useful, as it disables the trace before 6799 the WARNING dump is called, which prevents the trace to 6800 be filled with content caused by the warning output. 6801 6802 This option can also be set at run time via the sysctl 6803 option: kernel/traceoff_on_warning 6804 6805 transparent_hugepage= 6806 [KNL] 6807 Format: [always|madvise|never] 6808 Can be used to control the default behavior of the system 6809 with respect to transparent hugepages. 6810 See Documentation/admin-guide/mm/transhuge.rst 6811 for more details. 6812 6813 trusted.source= [KEYS] 6814 Format: <string> 6815 This parameter identifies the trust source as a backend 6816 for trusted keys implementation. Supported trust 6817 sources: 6818 - "tpm" 6819 - "tee" 6820 - "caam" 6821 - "dcp" 6822 If not specified then it defaults to iterating through 6823 the trust source list starting with TPM and assigns the 6824 first trust source as a backend which is initialized 6825 successfully during iteration. 6826 6827 trusted.rng= [KEYS] 6828 Format: <string> 6829 The RNG used to generate key material for trusted keys. 6830 Can be one of: 6831 - "kernel" 6832 - the same value as trusted.source: "tpm" or "tee" 6833 - "default" 6834 If not specified, "default" is used. In this case, 6835 the RNG's choice is left to each individual trust source. 6836 6837 trusted.dcp_use_otp_key 6838 This is intended to be used in combination with 6839 trusted.source=dcp and will select the DCP OTP key 6840 instead of the DCP UNIQUE key blob encryption. 6841 6842 trusted.dcp_skip_zk_test 6843 This is intended to be used in combination with 6844 trusted.source=dcp and will disable the check if the 6845 blob key is all zeros. This is helpful for situations where 6846 having this key zero'ed is acceptable. E.g. in testing 6847 scenarios. 6848 6849 tsc= Disable clocksource stability checks for TSC. 6850 Format: <string> 6851 [x86] reliable: mark tsc clocksource as reliable, this 6852 disables clocksource verification at runtime, as well 6853 as the stability checks done at bootup. Used to enable 6854 high-resolution timer mode on older hardware, and in 6855 virtualized environment. 6856 [x86] noirqtime: Do not use TSC to do irq accounting. 6857 Used to run time disable IRQ_TIME_ACCOUNTING on any 6858 platforms where RDTSC is slow and this accounting 6859 can add overhead. 6860 [x86] unstable: mark the TSC clocksource as unstable, this 6861 marks the TSC unconditionally unstable at bootup and 6862 avoids any further wobbles once the TSC watchdog notices. 6863 [x86] nowatchdog: disable clocksource watchdog. Used 6864 in situations with strict latency requirements (where 6865 interruptions from clocksource watchdog are not 6866 acceptable). 6867 [x86] recalibrate: force recalibration against a HW timer 6868 (HPET or PM timer) on systems whose TSC frequency was 6869 obtained from HW or FW using either an MSR or CPUID(0x15). 6870 Warn if the difference is more than 500 ppm. 6871 [x86] watchdog: Use TSC as the watchdog clocksource with 6872 which to check other HW timers (HPET or PM timer), but 6873 only on systems where TSC has been deemed trustworthy. 6874 This will be suppressed by an earlier tsc=nowatchdog and 6875 can be overridden by a later tsc=nowatchdog. A console 6876 message will flag any such suppression or overriding. 6877 6878 tsc_early_khz= [X86,EARLY] Skip early TSC calibration and use the given 6879 value instead. Useful when the early TSC frequency discovery 6880 procedure is not reliable, such as on overclocked systems 6881 with CPUID.16h support and partial CPUID.15h support. 6882 Format: <unsigned int> 6883 6884 tsx= [X86] Control Transactional Synchronization 6885 Extensions (TSX) feature in Intel processors that 6886 support TSX control. 6887 6888 This parameter controls the TSX feature. The options are: 6889 6890 on - Enable TSX on the system. Although there are 6891 mitigations for all known security vulnerabilities, 6892 TSX has been known to be an accelerator for 6893 several previous speculation-related CVEs, and 6894 so there may be unknown security risks associated 6895 with leaving it enabled. 6896 6897 off - Disable TSX on the system. (Note that this 6898 option takes effect only on newer CPUs which are 6899 not vulnerable to MDS, i.e., have 6900 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get 6901 the new IA32_TSX_CTRL MSR through a microcode 6902 update. This new MSR allows for the reliable 6903 deactivation of the TSX functionality.) 6904 6905 auto - Disable TSX if X86_BUG_TAA is present, 6906 otherwise enable TSX on the system. 6907 6908 Not specifying this option is equivalent to tsx=off. 6909 6910 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst 6911 for more details. 6912 6913 tsx_async_abort= [X86,INTEL,EARLY] Control mitigation for the TSX Async 6914 Abort (TAA) vulnerability. 6915 6916 Similar to Micro-architectural Data Sampling (MDS) 6917 certain CPUs that support Transactional 6918 Synchronization Extensions (TSX) are vulnerable to an 6919 exploit against CPU internal buffers which can forward 6920 information to a disclosure gadget under certain 6921 conditions. 6922 6923 In vulnerable processors, the speculatively forwarded 6924 data can be used in a cache side channel attack, to 6925 access data to which the attacker does not have direct 6926 access. 6927 6928 This parameter controls the TAA mitigation. The 6929 options are: 6930 6931 full - Enable TAA mitigation on vulnerable CPUs 6932 if TSX is enabled. 6933 6934 full,nosmt - Enable TAA mitigation and disable SMT on 6935 vulnerable CPUs. If TSX is disabled, SMT 6936 is not disabled because CPU is not 6937 vulnerable to cross-thread TAA attacks. 6938 off - Unconditionally disable TAA mitigation 6939 6940 On MDS-affected machines, tsx_async_abort=off can be 6941 prevented by an active MDS mitigation as both vulnerabilities 6942 are mitigated with the same mechanism so in order to disable 6943 this mitigation, you need to specify mds=off too. 6944 6945 Not specifying this option is equivalent to 6946 tsx_async_abort=full. On CPUs which are MDS affected 6947 and deploy MDS mitigation, TAA mitigation is not 6948 required and doesn't provide any additional 6949 mitigation. 6950 6951 For details see: 6952 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst 6953 6954 turbografx.map[2|3]= [HW,JOY] 6955 TurboGraFX parallel port interface 6956 Format: 6957 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7> 6958 See also Documentation/input/devices/joystick-parport.rst 6959 6960 udbg-immortal [PPC] When debugging early kernel crashes that 6961 happen after console_init() and before a proper 6962 console driver takes over, this boot options might 6963 help "seeing" what's going on. 6964 6965 uhash_entries= [KNL,NET] 6966 Set number of hash buckets for UDP/UDP-Lite connections 6967 6968 uhci-hcd.ignore_oc= 6969 [USB] Ignore overcurrent events (default N). 6970 Some badly-designed motherboards generate lots of 6971 bogus events, for ports that aren't wired to 6972 anything. Set this parameter to avoid log spamming. 6973 Note that genuine overcurrent events won't be 6974 reported either. 6975 6976 unknown_nmi_panic 6977 [X86] Cause panic on unknown NMI. 6978 6979 unwind_debug [X86-64,EARLY] 6980 Enable unwinder debug output. This can be 6981 useful for debugging certain unwinder error 6982 conditions, including corrupt stacks and 6983 bad/missing unwinder metadata. 6984 6985 usbcore.authorized_default= 6986 [USB] Default USB device authorization: 6987 (default -1 = authorized (same as 1), 6988 0 = not authorized, 1 = authorized, 2 = authorized 6989 if device connected to internal port) 6990 6991 usbcore.autosuspend= 6992 [USB] The autosuspend time delay (in seconds) used 6993 for newly-detected USB devices (default 2). This 6994 is the time required before an idle device will be 6995 autosuspended. Devices for which the delay is set 6996 to a negative value won't be autosuspended at all. 6997 6998 usbcore.usbfs_snoop= 6999 [USB] Set to log all usbfs traffic (default 0 = off). 7000 7001 usbcore.usbfs_snoop_max= 7002 [USB] Maximum number of bytes to snoop in each URB 7003 (default = 65536). 7004 7005 usbcore.blinkenlights= 7006 [USB] Set to cycle leds on hubs (default 0 = off). 7007 7008 usbcore.old_scheme_first= 7009 [USB] Start with the old device initialization 7010 scheme (default 0 = off). 7011 7012 usbcore.usbfs_memory_mb= 7013 [USB] Memory limit (in MB) for buffers allocated by 7014 usbfs (default = 16, 0 = max = 2047). 7015 7016 usbcore.use_both_schemes= 7017 [USB] Try the other device initialization scheme 7018 if the first one fails (default 1 = enabled). 7019 7020 usbcore.initial_descriptor_timeout= 7021 [USB] Specifies timeout for the initial 64-byte 7022 USB_REQ_GET_DESCRIPTOR request in milliseconds 7023 (default 5000 = 5.0 seconds). 7024 7025 usbcore.nousb [USB] Disable the USB subsystem 7026 7027 usbcore.quirks= 7028 [USB] A list of quirk entries to augment the built-in 7029 usb core quirk list. List entries are separated by 7030 commas. Each entry has the form 7031 VendorID:ProductID:Flags. The IDs are 4-digit hex 7032 numbers and Flags is a set of letters. Each letter 7033 will change the built-in quirk; setting it if it is 7034 clear and clearing it if it is set. The letters have 7035 the following meanings: 7036 a = USB_QUIRK_STRING_FETCH_255 (string 7037 descriptors must not be fetched using 7038 a 255-byte read); 7039 b = USB_QUIRK_RESET_RESUME (device can't resume 7040 correctly so reset it instead); 7041 c = USB_QUIRK_NO_SET_INTF (device can't handle 7042 Set-Interface requests); 7043 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't 7044 handle its Configuration or Interface 7045 strings); 7046 e = USB_QUIRK_RESET (device can't be reset 7047 (e.g morph devices), don't use reset); 7048 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has 7049 more interface descriptions than the 7050 bNumInterfaces count, and can't handle 7051 talking to these interfaces); 7052 g = USB_QUIRK_DELAY_INIT (device needs a pause 7053 during initialization, after we read 7054 the device descriptor); 7055 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For 7056 high speed and super speed interrupt 7057 endpoints, the USB 2.0 and USB 3.0 spec 7058 require the interval in microframes (1 7059 microframe = 125 microseconds) to be 7060 calculated as interval = 2 ^ 7061 (bInterval-1). 7062 Devices with this quirk report their 7063 bInterval as the result of this 7064 calculation instead of the exponent 7065 variable used in the calculation); 7066 i = USB_QUIRK_DEVICE_QUALIFIER (device can't 7067 handle device_qualifier descriptor 7068 requests); 7069 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device 7070 generates spurious wakeup, ignore 7071 remote wakeup capability); 7072 k = USB_QUIRK_NO_LPM (device can't handle Link 7073 Power Management); 7074 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL 7075 (Device reports its bInterval as linear 7076 frames instead of the USB 2.0 7077 calculation); 7078 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs 7079 to be disconnected before suspend to 7080 prevent spurious wakeup); 7081 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a 7082 pause after every control message); 7083 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra 7084 delay after resetting its port); 7085 p = USB_QUIRK_SHORT_SET_ADDRESS_REQ_TIMEOUT 7086 (Reduce timeout of the SET_ADDRESS 7087 request from 5000 ms to 500 ms); 7088 Example: quirks=0781:5580:bk,0a5c:5834:gij 7089 7090 usbhid.mousepoll= 7091 [USBHID] The interval which mice are to be polled at. 7092 7093 usbhid.jspoll= 7094 [USBHID] The interval which joysticks are to be polled at. 7095 7096 usbhid.kbpoll= 7097 [USBHID] The interval which keyboards are to be polled at. 7098 7099 usb-storage.delay_use= 7100 [UMS] The delay in seconds before a new device is 7101 scanned for Logical Units (default 1). 7102 7103 usb-storage.quirks= 7104 [UMS] A list of quirks entries to supplement or 7105 override the built-in unusual_devs list. List 7106 entries are separated by commas. Each entry has 7107 the form VID:PID:Flags where VID and PID are Vendor 7108 and Product ID values (4-digit hex numbers) and 7109 Flags is a set of characters, each corresponding 7110 to a common usb-storage quirk flag as follows: 7111 a = SANE_SENSE (collect more than 18 bytes 7112 of sense data, not on uas); 7113 b = BAD_SENSE (don't collect more than 18 7114 bytes of sense data, not on uas); 7115 c = FIX_CAPACITY (decrease the reported 7116 device capacity by one sector); 7117 d = NO_READ_DISC_INFO (don't use 7118 READ_DISC_INFO command, not on uas); 7119 e = NO_READ_CAPACITY_16 (don't use 7120 READ_CAPACITY_16 command); 7121 f = NO_REPORT_OPCODES (don't use report opcodes 7122 command, uas only); 7123 g = MAX_SECTORS_240 (don't transfer more than 7124 240 sectors at a time, uas only); 7125 h = CAPACITY_HEURISTICS (decrease the 7126 reported device capacity by one 7127 sector if the number is odd); 7128 i = IGNORE_DEVICE (don't bind to this 7129 device); 7130 j = NO_REPORT_LUNS (don't use report luns 7131 command, uas only); 7132 k = NO_SAME (do not use WRITE_SAME, uas only) 7133 l = NOT_LOCKABLE (don't try to lock and 7134 unlock ejectable media, not on uas); 7135 m = MAX_SECTORS_64 (don't transfer more 7136 than 64 sectors = 32 KB at a time, 7137 not on uas); 7138 n = INITIAL_READ10 (force a retry of the 7139 initial READ(10) command, not on uas); 7140 o = CAPACITY_OK (accept the capacity 7141 reported by the device, not on uas); 7142 p = WRITE_CACHE (the device cache is ON 7143 by default, not on uas); 7144 r = IGNORE_RESIDUE (the device reports 7145 bogus residue values, not on uas); 7146 s = SINGLE_LUN (the device has only one 7147 Logical Unit); 7148 t = NO_ATA_1X (don't allow ATA(12) and ATA(16) 7149 commands, uas only); 7150 u = IGNORE_UAS (don't bind to the uas driver); 7151 w = NO_WP_DETECT (don't test whether the 7152 medium is write-protected). 7153 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE 7154 even if the device claims no cache, 7155 not on uas) 7156 Example: quirks=0419:aaf5:rl,0421:0433:rc 7157 7158 user_debug= [KNL,ARM] 7159 Format: <int> 7160 See arch/arm/Kconfig.debug help text. 7161 1 - undefined instruction events 7162 2 - system calls 7163 4 - invalid data aborts 7164 8 - SIGSEGV faults 7165 16 - SIGBUS faults 7166 Example: user_debug=31 7167 7168 userpte= 7169 [X86,EARLY] Flags controlling user PTE allocations. 7170 7171 nohigh = do not allocate PTE pages in 7172 HIGHMEM regardless of setting 7173 of CONFIG_HIGHPTE. 7174 7175 vdso= [X86,SH,SPARC] 7176 On X86_32, this is an alias for vdso32=. Otherwise: 7177 7178 vdso=1: enable VDSO (the default) 7179 vdso=0: disable VDSO mapping 7180 7181 vdso32= [X86] Control the 32-bit vDSO 7182 vdso32=1: enable 32-bit VDSO 7183 vdso32=0 or vdso32=2: disable 32-bit VDSO 7184 7185 See the help text for CONFIG_COMPAT_VDSO for more 7186 details. If CONFIG_COMPAT_VDSO is set, the default is 7187 vdso32=0; otherwise, the default is vdso32=1. 7188 7189 For compatibility with older kernels, vdso32=2 is an 7190 alias for vdso32=0. 7191 7192 Try vdso32=0 if you encounter an error that says: 7193 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed! 7194 7195 vector= [IA-64,SMP] 7196 vector=percpu: enable percpu vector domain 7197 7198 video= [FB,EARLY] Frame buffer configuration 7199 See Documentation/fb/modedb.rst. 7200 7201 video.brightness_switch_enabled= [ACPI] 7202 Format: [0|1] 7203 If set to 1, on receiving an ACPI notify event 7204 generated by hotkey, video driver will adjust brightness 7205 level and then send out the event to user space through 7206 the allocated input device. If set to 0, video driver 7207 will only send out the event without touching backlight 7208 brightness level. 7209 default: 1 7210 7211 virtio_mmio.device= 7212 [VMMIO] Memory mapped virtio (platform) device. 7213 7214 <size>@<baseaddr>:<irq>[:<id>] 7215 where: 7216 <size> := size (can use standard suffixes 7217 like K, M and G) 7218 <baseaddr> := physical base address 7219 <irq> := interrupt number (as passed to 7220 request_irq()) 7221 <id> := (optional) platform device id 7222 example: 7223 virtio_mmio.device=1K@0x100b0000:48:7 7224 7225 Can be used multiple times for multiple devices. 7226 7227 vga= [BOOT,X86-32] Select a particular video mode 7228 See Documentation/arch/x86/boot.rst and 7229 Documentation/admin-guide/svga.rst. 7230 Use vga=ask for menu. 7231 This is actually a boot loader parameter; the value is 7232 passed to the kernel using a special protocol. 7233 7234 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y. 7235 May slow down system boot speed, especially when 7236 enabled on systems with a large amount of memory. 7237 All options are enabled by default, and this 7238 interface is meant to allow for selectively 7239 enabling or disabling specific virtual memory 7240 debugging features. 7241 7242 Available options are: 7243 P Enable page structure init time poisoning 7244 - Disable all of the above options 7245 7246 vmalloc=nn[KMG] [KNL,BOOT,EARLY] Forces the vmalloc area to have an 7247 exact size of <nn>. This can be used to increase 7248 the minimum size (128MB on x86). It can also be 7249 used to decrease the size and leave more room 7250 for directly mapped kernel RAM. 7251 7252 vmcp_cma=nn[MG] [KNL,S390,EARLY] 7253 Sets the memory size reserved for contiguous memory 7254 allocations for the vmcp device driver. 7255 7256 vmhalt= [KNL,S390] Perform z/VM CP command after system halt. 7257 Format: <command> 7258 7259 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic. 7260 Format: <command> 7261 7262 vmpoff= [KNL,S390] Perform z/VM CP command after power off. 7263 Format: <command> 7264 7265 vsyscall= [X86-64,EARLY] 7266 Controls the behavior of vsyscalls (i.e. calls to 7267 fixed addresses of 0xffffffffff600x00 from legacy 7268 code). Most statically-linked binaries and older 7269 versions of glibc use these calls. Because these 7270 functions are at fixed addresses, they make nice 7271 targets for exploits that can control RIP. 7272 7273 emulate Vsyscalls turn into traps and are emulated 7274 reasonably safely. The vsyscall page is 7275 readable. 7276 7277 xonly [default] Vsyscalls turn into traps and are 7278 emulated reasonably safely. The vsyscall 7279 page is not readable. 7280 7281 none Vsyscalls don't work at all. This makes 7282 them quite hard to use for exploits but 7283 might break your system. 7284 7285 vt.color= [VT] Default text color. 7286 Format: 0xYX, X = foreground, Y = background. 7287 Default: 0x07 = light gray on black. 7288 7289 vt.cur_default= [VT] Default cursor shape. 7290 Format: 0xCCBBAA, where AA, BB, and CC are the same as 7291 the parameters of the <Esc>[?A;B;Cc escape sequence; 7292 see VGA-softcursor.txt. Default: 2 = underline. 7293 7294 vt.default_blu= [VT] 7295 Format: <blue0>,<blue1>,<blue2>,...,<blue15> 7296 Change the default blue palette of the console. 7297 This is a 16-member array composed of values 7298 ranging from 0-255. 7299 7300 vt.default_grn= [VT] 7301 Format: <green0>,<green1>,<green2>,...,<green15> 7302 Change the default green palette of the console. 7303 This is a 16-member array composed of values 7304 ranging from 0-255. 7305 7306 vt.default_red= [VT] 7307 Format: <red0>,<red1>,<red2>,...,<red15> 7308 Change the default red palette of the console. 7309 This is a 16-member array composed of values 7310 ranging from 0-255. 7311 7312 vt.default_utf8= 7313 [VT] 7314 Format=<0|1> 7315 Set system-wide default UTF-8 mode for all tty's. 7316 Default is 1, i.e. UTF-8 mode is enabled for all 7317 newly opened terminals. 7318 7319 vt.global_cursor_default= 7320 [VT] 7321 Format=<-1|0|1> 7322 Set system-wide default for whether a cursor 7323 is shown on new VTs. Default is -1, 7324 i.e. cursors will be created by default unless 7325 overridden by individual drivers. 0 will hide 7326 cursors, 1 will display them. 7327 7328 vt.italic= [VT] Default color for italic text; 0-15. 7329 Default: 2 = green. 7330 7331 vt.underline= [VT] Default color for underlined text; 0-15. 7332 Default: 3 = cyan. 7333 7334 watchdog timers [HW,WDT] For information on watchdog timers, 7335 see Documentation/watchdog/watchdog-parameters.rst 7336 or other driver-specific files in the 7337 Documentation/watchdog/ directory. 7338 7339 watchdog_thresh= 7340 [KNL] 7341 Set the hard lockup detector stall duration 7342 threshold in seconds. The soft lockup detector 7343 threshold is set to twice the value. A value of 0 7344 disables both lockup detectors. Default is 10 7345 seconds. 7346 7347 workqueue.unbound_cpus= 7348 [KNL,SMP] Specify to constrain one or some CPUs 7349 to use in unbound workqueues. 7350 Format: <cpu-list> 7351 By default, all online CPUs are available for 7352 unbound workqueues. 7353 7354 workqueue.watchdog_thresh= 7355 If CONFIG_WQ_WATCHDOG is configured, workqueue can 7356 warn stall conditions and dump internal state to 7357 help debugging. 0 disables workqueue stall 7358 detection; otherwise, it's the stall threshold 7359 duration in seconds. The default value is 30 and 7360 it can be updated at runtime by writing to the 7361 corresponding sysfs file. 7362 7363 workqueue.cpu_intensive_thresh_us= 7364 Per-cpu work items which run for longer than this 7365 threshold are automatically considered CPU intensive 7366 and excluded from concurrency management to prevent 7367 them from noticeably delaying other per-cpu work 7368 items. Default is 10000 (10ms). 7369 7370 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel 7371 will report the work functions which violate this 7372 threshold repeatedly. They are likely good 7373 candidates for using WQ_UNBOUND workqueues instead. 7374 7375 workqueue.cpu_intensive_warning_thresh=<uint> 7376 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel 7377 will report the work functions which violate the 7378 intensive_threshold_us repeatedly. In order to prevent 7379 spurious warnings, start printing only after a work 7380 function has violated this threshold number of times. 7381 7382 The default is 4 times. 0 disables the warning. 7383 7384 workqueue.power_efficient 7385 Per-cpu workqueues are generally preferred because 7386 they show better performance thanks to cache 7387 locality; unfortunately, per-cpu workqueues tend to 7388 be more power hungry than unbound workqueues. 7389 7390 Enabling this makes the per-cpu workqueues which 7391 were observed to contribute significantly to power 7392 consumption unbound, leading to measurably lower 7393 power usage at the cost of small performance 7394 overhead. 7395 7396 The default value of this parameter is determined by 7397 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT. 7398 7399 workqueue.default_affinity_scope= 7400 Select the default affinity scope to use for unbound 7401 workqueues. Can be one of "cpu", "smt", "cache", 7402 "numa" and "system". Default is "cache". For more 7403 information, see the Affinity Scopes section in 7404 Documentation/core-api/workqueue.rst. 7405 7406 This can be changed after boot by writing to the 7407 matching /sys/module/workqueue/parameters file. All 7408 workqueues with the "default" affinity scope will be 7409 updated accordingly. 7410 7411 workqueue.debug_force_rr_cpu 7412 Workqueue used to implicitly guarantee that work 7413 items queued without explicit CPU specified are put 7414 on the local CPU. This guarantee is no longer true 7415 and while local CPU is still preferred work items 7416 may be put on foreign CPUs. This debug option 7417 forces round-robin CPU selection to flush out 7418 usages which depend on the now broken guarantee. 7419 When enabled, memory and cache locality will be 7420 impacted. 7421 7422 writecombine= [LOONGARCH,EARLY] Control the MAT (Memory Access 7423 Type) of ioremap_wc(). 7424 7425 on - Enable writecombine, use WUC for ioremap_wc() 7426 off - Disable writecombine, use SUC for ioremap_wc() 7427 7428 x2apic_phys [X86-64,APIC,EARLY] Use x2apic physical mode instead of 7429 default x2apic cluster mode on platforms 7430 supporting x2apic. 7431 7432 xen_512gb_limit [KNL,X86-64,XEN] 7433 Restricts the kernel running paravirtualized under Xen 7434 to use only up to 512 GB of RAM. The reason to do so is 7435 crash analysis tools and Xen tools for doing domain 7436 save/restore/migration must be enabled to handle larger 7437 domains. 7438 7439 xen_emul_unplug= [HW,X86,XEN,EARLY] 7440 Unplug Xen emulated devices 7441 Format: [unplug0,][unplug1] 7442 ide-disks -- unplug primary master IDE devices 7443 aux-ide-disks -- unplug non-primary-master IDE devices 7444 nics -- unplug network devices 7445 all -- unplug all emulated devices (NICs and IDE disks) 7446 unnecessary -- unplugging emulated devices is 7447 unnecessary even if the host did not respond to 7448 the unplug protocol 7449 never -- do not unplug even if version check succeeds 7450 7451 xen_legacy_crash [X86,XEN,EARLY] 7452 Crash from Xen panic notifier, without executing late 7453 panic() code such as dumping handler. 7454 7455 xen_msr_safe= [X86,XEN,EARLY] 7456 Format: <bool> 7457 Select whether to always use non-faulting (safe) MSR 7458 access functions when running as Xen PV guest. The 7459 default value is controlled by CONFIG_XEN_PV_MSR_SAFE. 7460 7461 xen_nopvspin [X86,XEN,EARLY] 7462 Disables the qspinlock slowpath using Xen PV optimizations. 7463 This parameter is obsoleted by "nopvspin" parameter, which 7464 has equivalent effect for XEN platform. 7465 7466 xen_nopv [X86] 7467 Disables the PV optimizations forcing the HVM guest to 7468 run as generic HVM guest with no PV drivers. 7469 This option is obsoleted by the "nopv" option, which 7470 has equivalent effect for XEN platform. 7471 7472 xen_no_vector_callback 7473 [KNL,X86,XEN,EARLY] Disable the vector callback for Xen 7474 event channel interrupts. 7475 7476 xen_scrub_pages= [XEN] 7477 Boolean option to control scrubbing pages before giving them back 7478 to Xen, for use by other domains. Can be also changed at runtime 7479 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages. 7480 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT. 7481 7482 xen_timer_slop= [X86-64,XEN,EARLY] 7483 Set the timer slop (in nanoseconds) for the virtual Xen 7484 timers (default is 100000). This adjusts the minimum 7485 delta of virtualized Xen timers, where lower values 7486 improve timer resolution at the expense of processing 7487 more timer interrupts. 7488 7489 xen.balloon_boot_timeout= [XEN] 7490 The time (in seconds) to wait before giving up to boot 7491 in case initial ballooning fails to free enough memory. 7492 Applies only when running as HVM or PVH guest and 7493 started with less memory configured than allowed at 7494 max. Default is 180. 7495 7496 xen.event_eoi_delay= [XEN] 7497 How long to delay EOI handling in case of event 7498 storms (jiffies). Default is 10. 7499 7500 xen.event_loop_timeout= [XEN] 7501 After which time (jiffies) the event handling loop 7502 should start to delay EOI handling. Default is 2. 7503 7504 xen.fifo_events= [XEN] 7505 Boolean parameter to disable using fifo event handling 7506 even if available. Normally fifo event handling is 7507 preferred over the 2-level event handling, as it is 7508 fairer and the number of possible event channels is 7509 much higher. Default is on (use fifo events). 7510 7511 xirc2ps_cs= [NET,PCMCIA] 7512 Format: 7513 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]] 7514 7515 xive= [PPC] 7516 By default on POWER9 and above, the kernel will 7517 natively use the XIVE interrupt controller. This option 7518 allows the fallback firmware mode to be used: 7519 7520 off Fallback to firmware control of XIVE interrupt 7521 controller on both pseries and powernv 7522 platforms. Only useful on POWER9 and above. 7523 7524 xive.store-eoi=off [PPC] 7525 By default on POWER10 and above, the kernel will use 7526 stores for EOI handling when the XIVE interrupt mode 7527 is active. This option allows the XIVE driver to use 7528 loads instead, as on POWER9. 7529 7530 xhci-hcd.quirks [USB,KNL] 7531 A hex value specifying bitmask with supplemental xhci 7532 host controller quirks. Meaning of each bit can be 7533 consulted in header drivers/usb/host/xhci.h. 7534 7535 xmon [PPC,EARLY] 7536 Format: { early | on | rw | ro | off } 7537 Controls if xmon debugger is enabled. Default is off. 7538 Passing only "xmon" is equivalent to "xmon=early". 7539 early Call xmon as early as possible on boot; xmon 7540 debugger is called from setup_arch(). 7541 on xmon debugger hooks will be installed so xmon 7542 is only called on a kernel crash. Default mode, 7543 i.e. either "ro" or "rw" mode, is controlled 7544 with CONFIG_XMON_DEFAULT_RO_MODE. 7545 rw xmon debugger hooks will be installed so xmon 7546 is called only on a kernel crash, mode is write, 7547 meaning SPR registers, memory and, other data 7548 can be written using xmon commands. 7549 ro same as "rw" option above but SPR registers, 7550 memory, and other data can't be written using 7551 xmon commands. 7552 off xmon is disabled. 7553