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 | nospcr } 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 nocmcff -- Disable firmware first mode for corrected 25 errors. This disables parsing the HEST CMC error 26 source to check if firmware has set the FF flag. This 27 may result in duplicate corrected error reports. 28 nospcr -- disable console in ACPI SPCR table as 29 default _serial_ console on ARM64 30 For ARM64, ONLY "acpi=off", "acpi=on", "acpi=force" or 31 "acpi=nospcr" are available 32 For RISCV64, ONLY "acpi=off", "acpi=on" or "acpi=force" 33 are available 34 35 See also Documentation/power/runtime_pm.rst, pci=noacpi 36 37 acpi_apic_instance= [ACPI,IOAPIC,EARLY] 38 Format: <int> 39 2: use 2nd APIC table, if available 40 1,0: use 1st APIC table 41 default: 0 42 43 acpi_backlight= [HW,ACPI] 44 { vendor | video | native | none } 45 If set to vendor, prefer vendor-specific driver 46 (e.g. thinkpad_acpi, sony_acpi, etc.) instead 47 of the ACPI video.ko driver. 48 If set to video, use the ACPI video.ko driver. 49 If set to native, use the device's native backlight mode. 50 If set to none, disable the ACPI backlight interface. 51 52 acpi_force_32bit_fadt_addr [ACPI,EARLY] 53 force FADT to use 32 bit addresses rather than the 54 64 bit X_* addresses. Some firmware have broken 64 55 bit addresses for force ACPI ignore these and use 56 the older legacy 32 bit addresses. 57 58 acpica_no_return_repair [HW, ACPI] 59 Disable AML predefined validation mechanism 60 This mechanism can repair the evaluation result to make 61 the return objects more ACPI specification compliant. 62 This option is useful for developers to identify the 63 root cause of an AML interpreter issue when the issue 64 has something to do with the repair mechanism. 65 66 acpi.debug_layer= [HW,ACPI,ACPI_DEBUG] 67 acpi.debug_level= [HW,ACPI,ACPI_DEBUG] 68 Format: <int> 69 CONFIG_ACPI_DEBUG must be enabled to produce any ACPI 70 debug output. Bits in debug_layer correspond to a 71 _COMPONENT in an ACPI source file, e.g., 72 #define _COMPONENT ACPI_EVENTS 73 Bits in debug_level correspond to a level in 74 ACPI_DEBUG_PRINT statements, e.g., 75 ACPI_DEBUG_PRINT((ACPI_DB_INFO, ... 76 The debug_level mask defaults to "info". See 77 Documentation/firmware-guide/acpi/debug.rst for more information about 78 debug layers and levels. 79 80 Enable processor driver info messages: 81 acpi.debug_layer=0x20000000 82 Enable AML "Debug" output, i.e., stores to the Debug 83 object while interpreting AML: 84 acpi.debug_layer=0xffffffff acpi.debug_level=0x2 85 Enable all messages related to ACPI hardware: 86 acpi.debug_layer=0x2 acpi.debug_level=0xffffffff 87 88 Some values produce so much output that the system is 89 unusable. The "log_buf_len" parameter may be useful 90 if you need to capture more output. 91 92 acpi_enforce_resources= [ACPI] 93 { strict | lax | no } 94 Check for resource conflicts between native drivers 95 and ACPI OperationRegions (SystemIO and SystemMemory 96 only). IO ports and memory declared in ACPI might be 97 used by the ACPI subsystem in arbitrary AML code and 98 can interfere with legacy drivers. 99 strict (default): access to resources claimed by ACPI 100 is denied; legacy drivers trying to access reserved 101 resources will fail to bind to device using them. 102 lax: access to resources claimed by ACPI is allowed; 103 legacy drivers trying to access reserved resources 104 will bind successfully but a warning message is logged. 105 no: ACPI OperationRegions are not marked as reserved, 106 no further checks are performed. 107 108 acpi_force_table_verification [HW,ACPI,EARLY] 109 Enable table checksum verification during early stage. 110 By default, this is disabled due to x86 early mapping 111 size limitation. 112 113 acpi_irq_balance [HW,ACPI] 114 ACPI will balance active IRQs 115 default in APIC mode 116 117 acpi_irq_nobalance [HW,ACPI] 118 ACPI will not move active IRQs (default) 119 default in PIC mode 120 121 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA 122 Format: <irq>,<irq>... 123 124 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for 125 use by PCI 126 Format: <irq>,<irq>... 127 128 acpi_mask_gpe= [HW,ACPI] 129 Due to the existence of _Lxx/_Exx, some GPEs triggered 130 by unsupported hardware/firmware features can result in 131 GPE floodings that cannot be automatically disabled by 132 the GPE dispatcher. 133 This facility can be used to prevent such uncontrolled 134 GPE floodings. 135 Format: <byte> or <bitmap-list> 136 137 acpi_no_auto_serialize [HW,ACPI] 138 Disable auto-serialization of AML methods 139 AML control methods that contain the opcodes to create 140 named objects will be marked as "Serialized" by the 141 auto-serialization feature. 142 This feature is enabled by default. 143 This option allows to turn off the feature. 144 145 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump 146 kernels. 147 148 acpi_no_static_ssdt [HW,ACPI,EARLY] 149 Disable installation of static SSDTs at early boot time 150 By default, SSDTs contained in the RSDT/XSDT will be 151 installed automatically and they will appear under 152 /sys/firmware/acpi/tables. 153 This option turns off this feature. 154 Note that specifying this option does not affect 155 dynamic table installation which will install SSDT 156 tables to /sys/firmware/acpi/tables/dynamic. 157 158 acpi_no_watchdog [HW,ACPI,WDT] 159 Ignore the ACPI-based watchdog interface (WDAT) and let 160 a native driver control the watchdog device instead. 161 162 acpi_rsdp= [ACPI,EFI,KEXEC,EARLY] 163 Pass the RSDP address to the kernel, mostly used 164 on machines running EFI runtime service to boot the 165 second kernel for kdump. 166 167 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS 168 Format: To spoof as Windows 98: ="Microsoft Windows" 169 170 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead 171 of 2 which is mandated by ACPI 6) as the supported ACPI 172 specification revision (when using this switch, it may 173 be necessary to carry out a cold reboot _twice_ in a 174 row to make it take effect on the platform firmware). 175 176 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings 177 acpi_osi="string1" # add string1 178 acpi_osi="!string2" # remove string2 179 acpi_osi=!* # remove all strings 180 acpi_osi=! # disable all built-in OS vendor 181 strings 182 acpi_osi=!! # enable all built-in OS vendor 183 strings 184 acpi_osi= # disable all strings 185 186 'acpi_osi=!' can be used in combination with single or 187 multiple 'acpi_osi="string1"' to support specific OS 188 vendor string(s). Note that such command can only 189 affect the default state of the OS vendor strings, thus 190 it cannot affect the default state of the feature group 191 strings and the current state of the OS vendor strings, 192 specifying it multiple times through kernel command line 193 is meaningless. This command is useful when one do not 194 care about the state of the feature group strings which 195 should be controlled by the OSPM. 196 Examples: 197 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent 198 to 'acpi_osi="Windows 2000" acpi_osi=!', they all 199 can make '_OSI("Windows 2000")' TRUE. 200 201 'acpi_osi=' cannot be used in combination with other 202 'acpi_osi=' command lines, the _OSI method will not 203 exist in the ACPI namespace. NOTE that such command can 204 only affect the _OSI support state, thus specifying it 205 multiple times through kernel command line is also 206 meaningless. 207 Examples: 208 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)' 209 FALSE. 210 211 'acpi_osi=!*' can be used in combination with single or 212 multiple 'acpi_osi="string1"' to support specific 213 string(s). Note that such command can affect the 214 current state of both the OS vendor strings and the 215 feature group strings, thus specifying it multiple times 216 through kernel command line is meaningful. But it may 217 still not able to affect the final state of a string if 218 there are quirks related to this string. This command 219 is useful when one want to control the state of the 220 feature group strings to debug BIOS issues related to 221 the OSPM features. 222 Examples: 223 1. 'acpi_osi="Module Device" acpi_osi=!*' can make 224 '_OSI("Module Device")' FALSE. 225 2. 'acpi_osi=!* acpi_osi="Module Device"' can make 226 '_OSI("Module Device")' TRUE. 227 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is 228 equivalent to 229 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"' 230 and 231 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!', 232 they all will make '_OSI("Windows 2000")' TRUE. 233 234 acpi_pm_good [X86] 235 Override the pmtimer bug detection: force the kernel 236 to assume that this machine's pmtimer latches its value 237 and always returns good values. 238 239 acpi_sci= [HW,ACPI,EARLY] ACPI System Control Interrupt trigger mode 240 Format: { level | edge | high | low } 241 242 acpi_skip_timer_override [HW,ACPI,EARLY] 243 Recognize and ignore IRQ0/pin2 Interrupt Override. 244 For broken nForce2 BIOS resulting in XT-PIC timer. 245 246 acpi_sleep= [HW,ACPI] Sleep options 247 Format: { s3_bios, s3_mode, s3_beep, s4_hwsig, 248 s4_nohwsig, old_ordering, nonvs, 249 sci_force_enable, nobl } 250 See Documentation/power/video.rst for information on 251 s3_bios and s3_mode. 252 s3_beep is for debugging; it makes the PC's speaker beep 253 as soon as the kernel's real-mode entry point is called. 254 s4_hwsig causes the kernel to check the ACPI hardware 255 signature during resume from hibernation, and gracefully 256 refuse to resume if it has changed. This complies with 257 the ACPI specification but not with reality, since 258 Windows does not do this and many laptops do change it 259 on docking. So the default behaviour is to allow resume 260 and simply warn when the signature changes, unless the 261 s4_hwsig option is enabled. 262 s4_nohwsig prevents ACPI hardware signature from being 263 used (or even warned about) during resume. 264 old_ordering causes the ACPI 1.0 ordering of the _PTS 265 control method, with respect to putting devices into 266 low power states, to be enforced (the ACPI 2.0 ordering 267 of _PTS is used by default). 268 nonvs prevents the kernel from saving/restoring the 269 ACPI NVS memory during suspend/hibernation and resume. 270 sci_force_enable causes the kernel to set SCI_EN directly 271 on resume from S1/S3 (which is against the ACPI spec, 272 but some broken systems don't work without it). 273 nobl causes the internal blacklist of systems known to 274 behave incorrectly in some ways with respect to system 275 suspend and resume to be ignored (use wisely). 276 277 acpi_use_timer_override [HW,ACPI,EARLY] 278 Use timer override. For some broken Nvidia NF5 boards 279 that require a timer override, but don't have HPET 280 281 add_efi_memmap [EFI,X86,EARLY] Include EFI memory map in 282 kernel's map of available physical RAM. 283 284 agp= [AGP] 285 { off | try_unsupported } 286 off: disable AGP support 287 try_unsupported: try to drive unsupported chipsets 288 (may crash computer or cause data corruption) 289 290 ALSA [HW,ALSA] 291 See Documentation/sound/alsa-configuration.rst 292 293 alignment= [KNL,ARM] 294 Allow the default userspace alignment fault handler 295 behaviour to be specified. Bit 0 enables warnings, 296 bit 1 enables fixups, and bit 2 sends a segfault. 297 298 align_va_addr= [X86-64] 299 Align virtual addresses by clearing slice [14:12] when 300 allocating a VMA at process creation time. This option 301 gives you up to 3% performance improvement on AMD F15h 302 machines (where it is enabled by default) for a 303 CPU-intensive style benchmark, and it can vary highly in 304 a microbenchmark depending on workload and compiler. 305 306 32: only for 32-bit processes 307 64: only for 64-bit processes 308 on: enable for both 32- and 64-bit processes 309 off: disable for both 32- and 64-bit processes 310 311 alloc_snapshot [FTRACE] 312 Allocate the ftrace snapshot buffer on boot up when the 313 main buffer is allocated. This is handy if debugging 314 and you need to use tracing_snapshot() on boot up, and 315 do not want to use tracing_snapshot_alloc() as it needs 316 to be done where GFP_KERNEL allocations are allowed. 317 318 allow_mismatched_32bit_el0 [ARM64,EARLY] 319 Allow execve() of 32-bit applications and setting of the 320 PER_LINUX32 personality on systems where only a strict 321 subset of the CPUs support 32-bit EL0. When this 322 parameter is present, the set of CPUs supporting 32-bit 323 EL0 is indicated by /sys/devices/system/cpu/aarch32_el0 324 and hot-unplug operations may be restricted. 325 326 See Documentation/arch/arm64/asymmetric-32bit.rst for more 327 information. 328 329 amd_iommu= [HW,X86-64] 330 Pass parameters to the AMD IOMMU driver in the system. 331 Possible values are: 332 fullflush - Deprecated, equivalent to iommu.strict=1 333 off - do not initialize any AMD IOMMU found in 334 the system 335 force_isolation - Force device isolation for all 336 devices. The IOMMU driver is not 337 allowed anymore to lift isolation 338 requirements as needed. This option 339 does not override iommu=pt 340 force_enable - Force enable the IOMMU on platforms known 341 to be buggy with IOMMU enabled. Use this 342 option with care. 343 pgtbl_v1 - Use v1 page table for DMA-API (Default). 344 pgtbl_v2 - Use v2 page table for DMA-API. 345 irtcachedis - Disable Interrupt Remapping Table (IRT) caching. 346 nohugepages - Limit page-sizes used for v1 page-tables 347 to 4 KiB. 348 v2_pgsizes_only - Limit page-sizes used for v1 page-tables 349 to 4KiB/2Mib/1GiB. 350 351 352 amd_iommu_dump= [HW,X86-64] 353 Enable AMD IOMMU driver option to dump the ACPI table 354 for AMD IOMMU. With this option enabled, AMD IOMMU 355 driver will print ACPI tables for AMD IOMMU during 356 IOMMU initialization. 357 358 amd_iommu_intr= [HW,X86-64] 359 Specifies one of the following AMD IOMMU interrupt 360 remapping modes: 361 legacy - Use legacy interrupt remapping mode. 362 vapic - Use virtual APIC mode, which allows IOMMU 363 to inject interrupts directly into guest. 364 This mode requires kvm-amd.avic=1. 365 (Default when IOMMU HW support is present.) 366 367 amd_pstate= [X86,EARLY] 368 disable 369 Do not enable amd_pstate as the default 370 scaling driver for the supported processors 371 passive 372 Use amd_pstate with passive mode as a scaling driver. 373 In this mode autonomous selection is disabled. 374 Driver requests a desired performance level and platform 375 tries to match the same performance level if it is 376 satisfied by guaranteed performance level. 377 active 378 Use amd_pstate_epp driver instance as the scaling driver, 379 driver provides a hint to the hardware if software wants 380 to bias toward performance (0x0) or energy efficiency (0xff) 381 to the CPPC firmware. then CPPC power algorithm will 382 calculate the runtime workload and adjust the realtime cores 383 frequency. 384 guided 385 Activate guided autonomous mode. Driver requests minimum and 386 maximum performance level and the platform autonomously 387 selects a performance level in this range and appropriate 388 to the current workload. 389 390 amd_prefcore= 391 [X86] 392 disable 393 Disable amd-pstate preferred core. 394 395 amijoy.map= [HW,JOY] Amiga joystick support 396 Map of devices attached to JOY0DAT and JOY1DAT 397 Format: <a>,<b> 398 See also Documentation/input/joydev/joystick.rst 399 400 analog.map= [HW,JOY] Analog joystick and gamepad support 401 Specifies type or capabilities of an analog joystick 402 connected to one of 16 gameports 403 Format: <type1>,<type2>,..<type16> 404 405 apc= [HW,SPARC] 406 Power management functions (SPARCstation-4/5 + deriv.) 407 Format: noidle 408 Disable APC CPU standby support. SPARCstation-Fox does 409 not play well with APC CPU idle - disable it if you have 410 APC and your system crashes randomly. 411 412 apic [APIC,X86-64] Use IO-APIC. Default. 413 414 apic= [APIC,X86,EARLY] Advanced Programmable Interrupt Controller 415 Change the output verbosity while booting 416 Format: { quiet (default) | verbose | debug } 417 Change the amount of debugging information output 418 when initialising the APIC and IO-APIC components. 419 420 apic_extnmi= [APIC,X86,EARLY] External NMI delivery setting 421 Format: { bsp (default) | all | none } 422 bsp: External NMI is delivered only to CPU 0 423 all: External NMIs are broadcast to all CPUs as a 424 backup of CPU 0 425 none: External NMI is masked for all CPUs. This is 426 useful so that a dump capture kernel won't be 427 shot down by NMI 428 429 apicpmtimer Do APIC timer calibration using the pmtimer. Implies 430 apicmaintimer. Useful when your PIT timer is totally 431 broken. 432 433 autoconf= [IPV6] 434 See Documentation/networking/ipv6.rst. 435 436 apm= [APM] Advanced Power Management 437 See header of arch/x86/kernel/apm_32.c. 438 439 apparmor= [APPARMOR] Disable or enable AppArmor at boot time 440 Format: { "0" | "1" } 441 See security/apparmor/Kconfig help text 442 0 -- disable. 443 1 -- enable. 444 Default value is set via kernel config option. 445 446 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards 447 Format: <io>,<irq>,<nodeID> 448 449 arm64.no32bit_el0 [ARM64] Unconditionally disable the execution of 450 32 bit applications. 451 452 arm64.nobti [ARM64] Unconditionally disable Branch Target 453 Identification support 454 455 arm64.nogcs [ARM64] Unconditionally disable Guarded Control Stack 456 support 457 458 arm64.nomops [ARM64] Unconditionally disable Memory Copy and Memory 459 Set instructions support 460 461 arm64.nompam [ARM64] Unconditionally disable Memory Partitioning And 462 Monitoring support 463 464 arm64.nomte [ARM64] Unconditionally disable Memory Tagging Extension 465 support 466 467 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication 468 support 469 470 arm64.nosme [ARM64] Unconditionally disable Scalable Matrix 471 Extension support 472 473 arm64.nosve [ARM64] Unconditionally disable Scalable Vector 474 Extension support 475 476 ataflop= [HW,M68k] 477 478 atarimouse= [HW,MOUSE] Atari Mouse 479 480 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess, 481 EzKey and similar keyboards 482 483 atkbd.reset= [HW] Reset keyboard during initialization 484 485 atkbd.set= [HW] Select keyboard code set 486 Format: <int> (2 = AT (default), 3 = PS/2) 487 488 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar 489 keyboards 490 491 atkbd.softraw= [HW] Choose between synthetic and real raw mode 492 Format: <bool> (0 = real, 1 = synthetic (default)) 493 494 atkbd.softrepeat= [HW] 495 Use software keyboard repeat 496 497 audit= [KNL] Enable the audit sub-system 498 Format: { "0" | "1" | "off" | "on" } 499 0 | off - kernel audit is disabled and can not be 500 enabled until the next reboot 501 unset - kernel audit is initialized but disabled and 502 will be fully enabled by the userspace auditd. 503 1 | on - kernel audit is initialized and partially 504 enabled, storing at most audit_backlog_limit 505 messages in RAM until it is fully enabled by the 506 userspace auditd. 507 Default: unset 508 509 audit_backlog_limit= [KNL] Set the audit queue size limit. 510 Format: <int> (must be >=0) 511 Default: 64 512 513 bau= [X86_UV] Enable the BAU on SGI UV. The default 514 behavior is to disable the BAU (i.e. bau=0). 515 Format: { "0" | "1" } 516 0 - Disable the BAU. 517 1 - Enable the BAU. 518 unset - Disable the BAU. 519 520 baycom_epp= [HW,AX25] 521 Format: <io>,<mode> 522 523 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem 524 Format: <io>,<mode> 525 See header of drivers/net/hamradio/baycom_par.c. 526 527 baycom_ser_fdx= [HW,AX25] 528 BayCom Serial Port AX.25 Modem (Full Duplex Mode) 529 Format: <io>,<irq>,<mode>[,<baud>] 530 See header of drivers/net/hamradio/baycom_ser_fdx.c. 531 532 baycom_ser_hdx= [HW,AX25] 533 BayCom Serial Port AX.25 Modem (Half Duplex Mode) 534 Format: <io>,<irq>,<mode> 535 See header of drivers/net/hamradio/baycom_ser_hdx.c. 536 537 bdev_allow_write_mounted= 538 Format: <bool> 539 Control the ability to open a mounted block device 540 for writing, i.e., allow / disallow writes that bypass 541 the FS. This was implemented as a means to prevent 542 fuzzers from crashing the kernel by overwriting the 543 metadata underneath a mounted FS without its awareness. 544 This also prevents destructive formatting of mounted 545 filesystems by naive storage tooling that don't use 546 O_EXCL. Default is Y and can be changed through the 547 Kconfig option CONFIG_BLK_DEV_WRITE_MOUNTED. 548 549 bert_disable [ACPI] 550 Disable BERT OS support on buggy BIOSes. 551 552 bgrt_disable [ACPI,X86,EARLY] 553 Disable BGRT to avoid flickering OEM logo. 554 555 blkdevparts= Manual partition parsing of block device(s) for 556 embedded devices based on command line input. 557 See Documentation/block/cmdline-partition.rst 558 559 boot_delay= [KNL,EARLY] 560 Milliseconds to delay each printk during boot. 561 Only works if CONFIG_BOOT_PRINTK_DELAY is enabled, 562 and you may also have to specify "lpj=". Boot_delay 563 values larger than 10 seconds (10000) are assumed 564 erroneous and ignored. 565 Format: integer 566 567 bootconfig [KNL,EARLY] 568 Extended command line options can be added to an initrd 569 and this will cause the kernel to look for it. 570 571 See Documentation/admin-guide/bootconfig.rst 572 573 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards) 574 bttv.radio= Most important insmod options are available as 575 kernel args too. 576 bttv.pll= See Documentation/admin-guide/media/bttv.rst 577 bttv.tuner= 578 579 bulk_remove=off [PPC] This parameter disables the use of the pSeries 580 firmware feature for flushing multiple hpte entries 581 at a time. 582 583 c101= [NET] Moxa C101 synchronous serial card 584 585 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection. 586 Sometimes CPU hardware bugs make them report the cache 587 size incorrectly. The kernel will attempt work arounds 588 to fix known problems, but for some CPUs it is not 589 possible to determine what the correct size should be. 590 This option provides an override for these situations. 591 592 carrier_timeout= 593 [NET] Specifies amount of time (in seconds) that 594 the kernel should wait for a network carrier. By default 595 it waits 120 seconds. 596 597 ca_keys= [KEYS] This parameter identifies a specific key(s) on 598 the system trusted keyring to be used for certificate 599 trust validation. 600 format: { id:<keyid> | builtin } 601 602 cca= [MIPS,EARLY] Override the kernel pages' cache coherency 603 algorithm. Accepted values range from 0 to 7 604 inclusive. See arch/mips/include/asm/pgtable-bits.h 605 for platform specific values (SB1, Loongson3 and 606 others). 607 608 ccw_timeout_log [S390] 609 See Documentation/arch/s390/common_io.rst for details. 610 611 cgroup_disable= [KNL] Disable a particular controller or optional feature 612 Format: {name of the controller(s) or feature(s) to disable} 613 The effects of cgroup_disable=foo are: 614 - foo isn't auto-mounted if you mount all cgroups in 615 a single hierarchy 616 - foo isn't visible as an individually mountable 617 subsystem 618 - if foo is an optional feature then the feature is 619 disabled and corresponding cgroup files are not 620 created 621 {Currently only "memory" controller deal with this and 622 cut the overhead, others just disable the usage. So 623 only cgroup_disable=memory is actually worthy} 624 Specifying "pressure" disables per-cgroup pressure 625 stall information accounting feature 626 627 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1 628 Format: { { controller | "all" | "named" } 629 [,{ controller | "all" | "named" }...] } 630 Like cgroup_disable, but only applies to cgroup v1; 631 the blacklisted controllers remain available in cgroup2. 632 "all" blacklists all controllers and "named" disables 633 named mounts. Specifying both "all" and "named" disables 634 all v1 hierarchies. 635 636 cgroup_v1_proc= [KNL] Show also missing controllers in /proc/cgroups 637 Format: { "true" | "false" } 638 /proc/cgroups lists only v1 controllers by default. 639 This compatibility option enables listing also v2 640 controllers (whose v1 code is not compiled!), so that 641 semi-legacy software can check this file to decide 642 about usage of v2 (sic) controllers. 643 644 cgroup_favordynmods= [KNL] Enable or Disable favordynmods. 645 Format: { "true" | "false" } 646 Defaults to the value of CONFIG_CGROUP_FAVOR_DYNMODS. 647 648 cgroup.memory= [KNL] Pass options to the cgroup memory controller. 649 Format: <string> 650 nosocket -- Disable socket memory accounting. 651 nokmem -- Disable kernel memory accounting. 652 nobpf -- Disable BPF memory accounting. 653 654 checkreqprot= [SELINUX] Set initial checkreqprot flag value. 655 Format: { "0" | "1" } 656 See security/selinux/Kconfig help text. 657 0 -- check protection applied by kernel (includes 658 any implied execute protection). 659 1 -- check protection requested by application. 660 Default value is set via a kernel config option. 661 Value can be changed at runtime via 662 /sys/fs/selinux/checkreqprot. 663 Setting checkreqprot to 1 is deprecated. 664 665 cio_ignore= [S390] 666 See Documentation/arch/s390/common_io.rst for details. 667 668 clearcpuid=X[,X...] [X86] 669 Disable CPUID feature X for the kernel. See 670 arch/x86/include/asm/cpufeatures.h for the valid bit 671 numbers X. Note the Linux-specific bits are not necessarily 672 stable over kernel options, but the vendor-specific 673 ones should be. 674 X can also be a string as appearing in the flags: line 675 in /proc/cpuinfo which does not have the above 676 instability issue. However, not all features have names 677 in /proc/cpuinfo. 678 Note that using this option will taint your kernel. 679 Also note that user programs calling CPUID directly 680 or using the feature without checking anything 681 will still see it. This just prevents it from 682 being used by the kernel or shown in /proc/cpuinfo. 683 Also note the kernel might malfunction if you disable 684 some critical bits. 685 686 clk_ignore_unused 687 [CLK] 688 Prevents the clock framework from automatically gating 689 clocks that have not been explicitly enabled by a Linux 690 device driver but are enabled in hardware at reset or 691 by the bootloader/firmware. Note that this does not 692 force such clocks to be always-on nor does it reserve 693 those clocks in any way. This parameter is useful for 694 debug and development, but should not be needed on a 695 platform with proper driver support. For more 696 information, see Documentation/driver-api/clk.rst. 697 698 clock= [BUGS=X86-32, HW] gettimeofday clocksource override. 699 [Deprecated] 700 Forces specified clocksource (if available) to be used 701 when calculating gettimeofday(). If specified 702 clocksource is not available, it defaults to PIT. 703 Format: { pit | tsc | cyclone | pmtmr } 704 705 clocksource= Override the default clocksource 706 Format: <string> 707 Override the default clocksource and use the clocksource 708 with the name specified. 709 Some clocksource names to choose from, depending on 710 the platform: 711 [all] jiffies (this is the base, fallback clocksource) 712 [ACPI] acpi_pm 713 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2, 714 pxa_timer,timer3,32k_counter,timer0_1 715 [X86-32] pit,hpet,tsc; 716 scx200_hrt on Geode; cyclone on IBM x440 717 [MIPS] MIPS 718 [PARISC] cr16 719 [S390] tod 720 [SH] SuperH 721 [SPARC64] tick 722 [X86-64] hpet,tsc 723 724 clocksource.arm_arch_timer.evtstrm= 725 [ARM,ARM64,EARLY] 726 Format: <bool> 727 Enable/disable the eventstream feature of the ARM 728 architected timer so that code using WFE-based polling 729 loops can be debugged more effectively on production 730 systems. 731 732 clocksource.verify_n_cpus= [KNL] 733 Limit the number of CPUs checked for clocksources 734 marked with CLOCK_SOURCE_VERIFY_PERCPU that 735 are marked unstable due to excessive skew. 736 A negative value says to check all CPUs, while 737 zero says not to check any. Values larger than 738 nr_cpu_ids are silently truncated to nr_cpu_ids. 739 The actual CPUs are chosen randomly, with 740 no replacement if the same CPU is chosen twice. 741 742 clocksource-wdtest.holdoff= [KNL] 743 Set the time in seconds that the clocksource 744 watchdog test waits before commencing its tests. 745 Defaults to zero when built as a module and to 746 10 seconds when built into the kernel. 747 748 cma=nn[MG]@[start[MG][-end[MG]]] 749 [KNL,CMA,EARLY] 750 Sets the size of kernel global memory area for 751 contiguous memory allocations and optionally the 752 placement constraint by the physical address range of 753 memory allocations. A value of 0 disables CMA 754 altogether. For more information, see 755 kernel/dma/contiguous.c 756 757 cma_pernuma=nn[MG] 758 [KNL,CMA,EARLY] 759 Sets the size of kernel per-numa memory area for 760 contiguous memory allocations. A value of 0 disables 761 per-numa CMA altogether. And If this option is not 762 specified, the default value is 0. 763 With per-numa CMA enabled, DMA users on node nid will 764 first try to allocate buffer from the pernuma area 765 which is located in node nid, if the allocation fails, 766 they will fallback to the global default memory area. 767 768 numa_cma=<node>:nn[MG][,<node>:nn[MG]] 769 [KNL,CMA,EARLY] 770 Sets the size of kernel numa memory area for 771 contiguous memory allocations. It will reserve CMA 772 area for the specified node. 773 774 With numa CMA enabled, DMA users on node nid will 775 first try to allocate buffer from the numa area 776 which is located in node nid, if the allocation fails, 777 they will fallback to the global default memory area. 778 779 cmo_free_hint= [PPC] Format: { yes | no } 780 Specify whether pages are marked as being inactive 781 when they are freed. This is used in CMO environments 782 to determine OS memory pressure for page stealing by 783 a hypervisor. 784 Default: yes 785 786 coherent_pool=nn[KMG] [ARM,KNL,EARLY] 787 Sets the size of memory pool for coherent, atomic dma 788 allocations, by default set to 256K. 789 790 com20020= [HW,NET] ARCnet - COM20020 chipset 791 Format: 792 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]] 793 794 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers) 795 Format: <io>[,<irq>] 796 797 com90xx= [HW,NET] 798 ARCnet - COM90xx chipset (memory-mapped buffers) 799 Format: <io>[,<irq>[,<memstart>]] 800 801 condev= [HW,S390] console device 802 conmode= 803 804 con3215_drop= [S390,EARLY] 3215 console drop mode. 805 Format: y|n|Y|N|1|0 806 When set to true, drop data on the 3215 console when 807 the console buffer is full. In this case the 808 operator using a 3270 terminal emulator (for example 809 x3270) does not have to enter the clear key for the 810 console output to advance and the kernel to continue. 811 This leads to a much faster boot time when a 3270 812 terminal emulator is active. If no 3270 terminal 813 emulator is used, this parameter has no effect. 814 815 console= [KNL] Output console device and options. 816 817 tty<n> Use the virtual console device <n>. 818 819 ttyS<n>[,options] 820 ttyUSB0[,options] 821 Use the specified serial port. The options are of 822 the form "bbbbpnf", where "bbbb" is the baud rate, 823 "p" is parity ("n", "o", or "e"), "n" is number of 824 bits, and "f" is flow control ("r" for RTS or 825 omit it). Default is "9600n8". 826 827 See Documentation/admin-guide/serial-console.rst for more 828 information. See 829 Documentation/networking/netconsole.rst for an 830 alternative. 831 832 <DEVNAME>:<n>.<n>[,options] 833 Use the specified serial port on the serial core bus. 834 The addressing uses DEVNAME of the physical serial port 835 device, followed by the serial core controller instance, 836 and the serial port instance. The options are the same 837 as documented for the ttyS addressing above. 838 839 The mapping of the serial ports to the tty instances 840 can be viewed with: 841 842 $ ls -d /sys/bus/serial-base/devices/*:*.*/tty/* 843 /sys/bus/serial-base/devices/00:04:0.0/tty/ttyS0 844 845 In the above example, the console can be addressed with 846 console=00:04:0.0. Note that a console addressed this 847 way will only get added when the related device driver 848 is ready. The use of an earlycon parameter in addition to 849 the console may be desired for console output early on. 850 851 uart[8250],io,<addr>[,options] 852 uart[8250],mmio,<addr>[,options] 853 uart[8250],mmio16,<addr>[,options] 854 uart[8250],mmio32,<addr>[,options] 855 uart[8250],0x<addr>[,options] 856 Start an early, polled-mode console on the 8250/16550 857 UART at the specified I/O port or MMIO address, 858 switching to the matching ttyS device later. 859 MMIO inter-register address stride is either 8-bit 860 (mmio), 16-bit (mmio16), or 32-bit (mmio32). 861 If none of [io|mmio|mmio16|mmio32], <addr> is assumed 862 to be equivalent to 'mmio'. 'options' are specified in 863 the same format described for ttyS above; if unspecified, 864 the h/w is not re-initialized. 865 866 hvc<n> Use the hypervisor console device <n>. This is for 867 both Xen and PowerPC hypervisors. 868 869 { null | "" } 870 Use to disable console output, i.e., to have kernel 871 console messages discarded. 872 This must be the only console= parameter used on the 873 kernel command line. 874 875 If the device connected to the port is not a TTY but a braille 876 device, prepend "brl," before the device type, for instance 877 console=brl,ttyS0 878 For now, only VisioBraille is supported. 879 880 console_msg_format= 881 [KNL] Change console messages format 882 default 883 By default we print messages on consoles in 884 "[time stamp] text\n" format (time stamp may not be 885 printed, depending on CONFIG_PRINTK_TIME or 886 `printk_time' param). 887 syslog 888 Switch to syslog format: "<%u>[time stamp] text\n" 889 IOW, each message will have a facility and loglevel 890 prefix. The format is similar to one used by syslog() 891 syscall, or to executing "dmesg -S --raw" or to reading 892 from /proc/kmsg. 893 894 consoleblank= [KNL] The console blank (screen saver) timeout in 895 seconds. A value of 0 disables the blank timer. 896 Defaults to 0. 897 898 coredump_filter= 899 [KNL] Change the default value for 900 /proc/<pid>/coredump_filter. 901 See also Documentation/filesystems/proc.rst. 902 903 coresight_cpu_debug.enable 904 [ARM,ARM64] 905 Format: <bool> 906 Enable/disable the CPU sampling based debugging. 907 0: default value, disable debugging 908 1: enable debugging at boot time 909 910 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver 911 Format: 912 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>] 913 914 cpuidle.off=1 [CPU_IDLE] 915 disable the cpuidle sub-system 916 917 cpuidle.governor= 918 [CPU_IDLE] Name of the cpuidle governor to use. 919 920 cpufreq.off=1 [CPU_FREQ] 921 disable the cpufreq sub-system 922 923 cpufreq.default_governor= 924 [CPU_FREQ] Name of the default cpufreq governor or 925 policy to use. This governor must be registered in the 926 kernel before the cpufreq driver probes. 927 928 cpu_init_udelay=N 929 [X86,EARLY] Delay for N microsec between assert and de-assert 930 of APIC INIT to start processors. This delay occurs 931 on every CPU online, such as boot, and resume from suspend. 932 Default: 10000 933 934 cpuhp.parallel= 935 [SMP] Enable/disable parallel bringup of secondary CPUs 936 Format: <bool> 937 Default is enabled if CONFIG_HOTPLUG_PARALLEL=y. Otherwise 938 the parameter has no effect. 939 940 crash_kexec_post_notifiers 941 Only jump to kdump kernel after running the panic 942 notifiers and dumping kmsg. This option increases 943 the risks of a kdump failure, since some panic 944 notifiers can make the crashed kernel more unstable. 945 In configurations where kdump may not be reliable, 946 running the panic notifiers could allow collecting 947 more data on dmesg, like stack traces from other CPUS 948 or extra data dumped by panic_print. Note that some 949 configurations enable this option unconditionally, 950 like Hyper-V, PowerPC (fadump) and AMD SEV-SNP. 951 952 crashkernel=size[KMG][@offset[KMG]] 953 [KNL,EARLY] Using kexec, Linux can switch to a 'crash kernel' 954 upon panic. This parameter reserves the physical 955 memory region [offset, offset + size] for that kernel 956 image. If '@offset' is omitted, then a suitable offset 957 is selected automatically. 958 [KNL, X86-64, ARM64, RISCV, LoongArch] Select a region 959 under 4G first, and fall back to reserve region above 960 4G when '@offset' hasn't been specified. 961 See Documentation/admin-guide/kdump/kdump.rst for further details. 962 963 crashkernel=range1:size1[,range2:size2,...][@offset] 964 [KNL] Same as above, but depends on the memory 965 in the running system. The syntax of range is 966 start-[end] where start and end are both 967 a memory unit (amount[KMG]). See also 968 Documentation/admin-guide/kdump/kdump.rst for an example. 969 970 crashkernel=size[KMG],high 971 [KNL, X86-64, ARM64, RISCV, LoongArch] range could be 972 above 4G. 973 Allow kernel to allocate physical memory region from top, 974 so could be above 4G if system have more than 4G ram 975 installed. Otherwise memory region will be allocated 976 below 4G, if available. 977 It will be ignored if crashkernel=X is specified. 978 crashkernel=size[KMG],low 979 [KNL, X86-64, ARM64, RISCV, LoongArch] range under 4G. 980 When crashkernel=X,high is passed, kernel could allocate 981 physical memory region above 4G, that cause second kernel 982 crash on system that require some amount of low memory, 983 e.g. swiotlb requires at least 64M+32K low memory, also 984 enough extra low memory is needed to make sure DMA buffers 985 for 32-bit devices won't run out. Kernel would try to allocate 986 default size of memory below 4G automatically. The default 987 size is platform dependent. 988 --> x86: max(swiotlb_size_or_default() + 8MiB, 256MiB) 989 --> arm64: 128MiB 990 --> riscv: 128MiB 991 --> loongarch: 128MiB 992 This one lets the user specify own low range under 4G 993 for second kernel instead. 994 0: to disable low allocation. 995 It will be ignored when crashkernel=X,high is not used 996 or memory reserved is below 4G. 997 crashkernel=size[KMG],cma 998 [KNL, X86] Reserve additional crash kernel memory from 999 CMA. This reservation is usable by the first system's 1000 userspace memory and kernel movable allocations (memory 1001 balloon, zswap). Pages allocated from this memory range 1002 will not be included in the vmcore so this should not 1003 be used if dumping of userspace memory is intended and 1004 it has to be expected that some movable kernel pages 1005 may be missing from the dump. 1006 1007 A standard crashkernel reservation, as described above, 1008 is still needed to hold the crash kernel and initrd. 1009 1010 This option increases the risk of a kdump failure: DMA 1011 transfers configured by the first kernel may end up 1012 corrupting the second kernel's memory. 1013 1014 This reservation method is intended for systems that 1015 can't afford to sacrifice enough memory for standard 1016 crashkernel reservation and where less reliable and 1017 possibly incomplete kdump is preferable to no kdump at 1018 all. 1019 1020 cryptomgr.notests 1021 [KNL] Disable crypto self-tests 1022 1023 cs89x0_dma= [HW,NET] 1024 Format: <dma> 1025 1026 cs89x0_media= [HW,NET] 1027 Format: { rj45 | aui | bnc } 1028 1029 csdlock_debug= [KNL] Enable or disable debug add-ons of cross-CPU 1030 function call handling. When switched on, 1031 additional debug data is printed to the console 1032 in case a hanging CPU is detected, and that 1033 CPU is pinged again in order to try to resolve 1034 the hang situation. The default value of this 1035 option depends on the CSD_LOCK_WAIT_DEBUG_DEFAULT 1036 Kconfig option. 1037 1038 dasd= [HW,NET] 1039 See header of drivers/s390/block/dasd_devmap.c. 1040 1041 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port 1042 (one device per port) 1043 Format: <port#>,<type> 1044 See also Documentation/input/devices/joystick-parport.rst 1045 1046 debug [KNL,EARLY] Enable kernel debugging (events log level). 1047 1048 debug_boot_weak_hash 1049 [KNL,EARLY] Enable printing [hashed] pointers early in the 1050 boot sequence. If enabled, we use a weak hash instead 1051 of siphash to hash pointers. Use this option if you are 1052 seeing instances of '(___ptrval___)') and need to see a 1053 value (hashed pointer) instead. Cryptographically 1054 insecure, please do not use on production kernels. 1055 1056 debug_locks_verbose= 1057 [KNL] verbose locking self-tests 1058 Format: <int> 1059 Print debugging info while doing the locking API 1060 self-tests. 1061 Bitmask for the various LOCKTYPE_ tests. Defaults to 0 1062 (no extra messages), setting it to -1 (all bits set) 1063 will print _a_lot_ more information - normally only 1064 useful to lockdep developers. 1065 1066 debug_objects [KNL,EARLY] Enable object debugging 1067 1068 debug_guardpage_minorder= 1069 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this 1070 parameter allows control of the order of pages that will 1071 be intentionally kept free (and hence protected) by the 1072 buddy allocator. Bigger value increase the probability 1073 of catching random memory corruption, but reduce the 1074 amount of memory for normal system use. The maximum 1075 possible value is MAX_PAGE_ORDER/2. Setting this 1076 parameter to 1 or 2 should be enough to identify most 1077 random memory corruption problems caused by bugs in 1078 kernel or driver code when a CPU writes to (or reads 1079 from) a random memory location. Note that there exists 1080 a class of memory corruptions problems caused by buggy 1081 H/W or F/W or by drivers badly programming DMA 1082 (basically when memory is written at bus level and the 1083 CPU MMU is bypassed) which are not detectable by 1084 CONFIG_DEBUG_PAGEALLOC, hence this option will not 1085 help tracking down these problems. 1086 1087 debug_pagealloc= 1088 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this parameter 1089 enables the feature at boot time. By default, it is 1090 disabled and the system will work mostly the same as a 1091 kernel built without CONFIG_DEBUG_PAGEALLOC. 1092 Note: to get most of debug_pagealloc error reports, it's 1093 useful to also enable the page_owner functionality. 1094 on: enable the feature 1095 1096 debugfs= [KNL,EARLY] This parameter enables what is exposed to 1097 userspace and debugfs internal clients. 1098 Format: { on, no-mount, off } 1099 on: All functions are enabled. 1100 no-mount: 1101 Filesystem is not registered but kernel clients can 1102 access APIs and a crashkernel can be used to read 1103 its content. There is nothing to mount. 1104 off: Filesystem is not registered and clients 1105 get a -EPERM as result when trying to register files 1106 or directories within debugfs. 1107 This is equivalent of the runtime functionality if 1108 debugfs was not enabled in the kernel at all. 1109 Default value is set in build-time with a kernel configuration. 1110 1111 debugpat [X86] Enable PAT debugging 1112 1113 default_hugepagesz= 1114 [HW] The size of the default HugeTLB page. This is 1115 the size represented by the legacy /proc/ hugepages 1116 APIs. In addition, this is the default hugetlb size 1117 used for shmget(), mmap() and mounting hugetlbfs 1118 filesystems. If not specified, defaults to the 1119 architecture's default huge page size. Huge page 1120 sizes are architecture dependent. See also 1121 Documentation/admin-guide/mm/hugetlbpage.rst. 1122 Format: size[KMG] 1123 1124 deferred_probe_timeout= 1125 [KNL] Debugging option to set a timeout in seconds for 1126 deferred probe to give up waiting on dependencies to 1127 probe. Only specific dependencies (subsystems or 1128 drivers) that have opted in will be ignored. A timeout 1129 of 0 will timeout at the end of initcalls. If the time 1130 out hasn't expired, it'll be restarted by each 1131 successful driver registration. This option will also 1132 dump out devices still on the deferred probe list after 1133 retrying. 1134 1135 delayacct [KNL] Enable per-task delay accounting 1136 1137 dell_smm_hwmon.ignore_dmi= 1138 [HW] Continue probing hardware even if DMI data 1139 indicates that the driver is running on unsupported 1140 hardware. 1141 1142 dell_smm_hwmon.force= 1143 [HW] Activate driver even if SMM BIOS signature does 1144 not match list of supported models and enable otherwise 1145 blacklisted features. 1146 1147 dell_smm_hwmon.power_status= 1148 [HW] Report power status in /proc/i8k 1149 (disabled by default). 1150 1151 dell_smm_hwmon.restricted= 1152 [HW] Allow controlling fans only if SYS_ADMIN 1153 capability is set. 1154 1155 dell_smm_hwmon.fan_mult= 1156 [HW] Factor to multiply fan speed with. 1157 1158 dell_smm_hwmon.fan_max= 1159 [HW] Maximum configurable fan speed. 1160 1161 dfltcc= [HW,S390] 1162 Format: { on | off | def_only | inf_only | always } 1163 on: s390 zlib hardware support for compression on 1164 level 1 and decompression (default) 1165 off: No s390 zlib hardware support 1166 def_only: s390 zlib hardware support for deflate 1167 only (compression on level 1) 1168 inf_only: s390 zlib hardware support for inflate 1169 only (decompression) 1170 always: Same as 'on' but ignores the selected compression 1171 level always using hardware support (used for debugging) 1172 1173 dhash_entries= [KNL] 1174 Set number of hash buckets for dentry cache. 1175 1176 disable_1tb_segments [PPC,EARLY] 1177 Disables the use of 1TB hash page table segments. This 1178 causes the kernel to fall back to 256MB segments which 1179 can be useful when debugging issues that require an SLB 1180 miss to occur. 1181 1182 disable= [IPV6] 1183 See Documentation/networking/ipv6.rst. 1184 1185 disable_radix [PPC,EARLY] 1186 Disable RADIX MMU mode on POWER9 1187 1188 disable_tlbie [PPC] 1189 Disable TLBIE instruction. Currently does not work 1190 with KVM, with HASH MMU, or with coherent accelerators. 1191 1192 disable_ddw [PPC/PSERIES,EARLY] 1193 Disable Dynamic DMA Window support. Use this 1194 to workaround buggy firmware. 1195 1196 disable_ipv6= [IPV6] 1197 See Documentation/networking/ipv6.rst. 1198 1199 disable_mtrr_cleanup [X86,EARLY] 1200 The kernel tries to adjust MTRR layout from continuous 1201 to discrete, to make X server driver able to add WB 1202 entry later. This parameter disables that. 1203 1204 disable_mtrr_trim [X86, Intel and AMD only,EARLY] 1205 By default the kernel will trim any uncacheable 1206 memory out of your available memory pool based on 1207 MTRR settings. This parameter disables that behavior, 1208 possibly causing your machine to run very slowly. 1209 1210 disable_timer_pin_1 [X86,EARLY] 1211 Disable PIN 1 of APIC timer 1212 Can be useful to work around chipset bugs. 1213 1214 dis_ucode_ldr [X86] Disable the microcode loader. 1215 1216 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support, 1217 this option disables the debugging code at boot. 1218 1219 dma_debug_entries=<number> 1220 This option allows to tune the number of preallocated 1221 entries for DMA-API debugging code. One entry is 1222 required per DMA-API allocation. Use this if the 1223 DMA-API debugging code disables itself because the 1224 architectural default is too low. 1225 1226 dma_debug_driver=<driver_name> 1227 With this option the DMA-API debugging driver 1228 filter feature can be enabled at boot time. Just 1229 pass the driver to filter for as the parameter. 1230 The filter can be disabled or changed to another 1231 driver later using sysfs. 1232 1233 reg_file_data_sampling= 1234 [X86] Controls mitigation for Register File Data 1235 Sampling (RFDS) vulnerability. RFDS is a CPU 1236 vulnerability which may allow userspace to infer 1237 kernel data values previously stored in floating point 1238 registers, vector registers, or integer registers. 1239 RFDS only affects Intel Atom processors. 1240 1241 on: Turns ON the mitigation. 1242 off: Turns OFF the mitigation. 1243 1244 This parameter overrides the compile time default set 1245 by CONFIG_MITIGATION_RFDS. Mitigation cannot be 1246 disabled when other VERW based mitigations (like MDS) 1247 are enabled. In order to disable RFDS mitigation all 1248 VERW based mitigations need to be disabled. 1249 1250 For details see: 1251 Documentation/admin-guide/hw-vuln/reg-file-data-sampling.rst 1252 1253 driver_async_probe= [KNL] 1254 List of driver names to be probed asynchronously. * 1255 matches with all driver names. If * is specified, the 1256 rest of the listed driver names are those that will NOT 1257 match the *. 1258 Format: <driver_name1>,<driver_name2>... 1259 1260 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>] 1261 Broken monitors, graphic adapters, KVMs and EDIDless 1262 panels may send no or incorrect EDID data sets. 1263 This parameter allows to specify an EDID data sets 1264 in the /lib/firmware directory that are used instead. 1265 An EDID data set will only be used for a particular 1266 connector, if its name and a colon are prepended to 1267 the EDID name. Each connector may use a unique EDID 1268 data set by separating the files with a comma. An EDID 1269 data set with no connector name will be used for 1270 any connectors not explicitly specified. 1271 1272 dscc4.setup= [NET] 1273 1274 dt_cpu_ftrs= [PPC,EARLY] 1275 Format: {"off" | "known"} 1276 Control how the dt_cpu_ftrs device-tree binding is 1277 used for CPU feature discovery and setup (if it 1278 exists). 1279 off: Do not use it, fall back to legacy cpu table. 1280 known: Do not pass through unknown features to guests 1281 or userspace, only those that the kernel is aware of. 1282 1283 dump_apple_properties [X86] 1284 Dump name and content of EFI device properties on 1285 x86 Macs. Useful for driver authors to determine 1286 what data is available or for reverse-engineering. 1287 1288 dyndbg[="val"] [KNL,DYNAMIC_DEBUG] 1289 <module>.dyndbg[="val"] 1290 Enable debug messages at boot time. See 1291 Documentation/admin-guide/dynamic-debug-howto.rst 1292 for details. 1293 1294 early_ioremap_debug [KNL,EARLY] 1295 Enable debug messages in early_ioremap support. This 1296 is useful for tracking down temporary early mappings 1297 which are not unmapped. 1298 1299 earlycon= [KNL,EARLY] Output early console device and options. 1300 1301 When used with no options, the early console is 1302 determined by stdout-path property in device tree's 1303 chosen node or the ACPI SPCR table if supported by 1304 the platform. 1305 1306 cdns,<addr>[,options] 1307 Start an early, polled-mode console on a Cadence 1308 (xuartps) serial port at the specified address. Only 1309 supported option is baud rate. If baud rate is not 1310 specified, the serial port must already be setup and 1311 configured. 1312 1313 uart[8250],io,<addr>[,options[,uartclk]] 1314 uart[8250],mmio,<addr>[,options[,uartclk]] 1315 uart[8250],mmio32,<addr>[,options[,uartclk]] 1316 uart[8250],mmio32be,<addr>[,options[,uartclk]] 1317 uart[8250],0x<addr>[,options] 1318 Start an early, polled-mode console on the 8250/16550 1319 UART at the specified I/O port or MMIO address. 1320 MMIO inter-register address stride is either 8-bit 1321 (mmio) or 32-bit (mmio32 or mmio32be). 1322 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed 1323 to be equivalent to 'mmio'. 'options' are specified 1324 in the same format described for "console=ttyS<n>"; if 1325 unspecified, the h/w is not initialized. 'uartclk' is 1326 the uart clock frequency; if unspecified, it is set 1327 to 'BASE_BAUD' * 16. 1328 1329 pl011,<addr> 1330 pl011,mmio32,<addr> 1331 Start an early, polled-mode console on a pl011 serial 1332 port at the specified address. The pl011 serial port 1333 must already be setup and configured. Options are not 1334 yet supported. If 'mmio32' is specified, then only 1335 the driver will use only 32-bit accessors to read/write 1336 the device registers. 1337 1338 liteuart,<addr> 1339 Start an early console on a litex serial port at the 1340 specified address. The serial port must already be 1341 setup and configured. Options are not yet supported. 1342 1343 meson,<addr> 1344 Start an early, polled-mode console on a meson serial 1345 port at the specified address. The serial port must 1346 already be setup and configured. Options are not yet 1347 supported. 1348 1349 msm_serial,<addr> 1350 Start an early, polled-mode console on an msm serial 1351 port at the specified address. The serial port 1352 must already be setup and configured. Options are not 1353 yet supported. 1354 1355 msm_serial_dm,<addr> 1356 Start an early, polled-mode console on an msm serial 1357 dm port at the specified address. The serial port 1358 must already be setup and configured. Options are not 1359 yet supported. 1360 1361 owl,<addr> 1362 Start an early, polled-mode console on a serial port 1363 of an Actions Semi SoC, such as S500 or S900, at the 1364 specified address. The serial port must already be 1365 setup and configured. Options are not yet supported. 1366 1367 rda,<addr> 1368 Start an early, polled-mode console on a serial port 1369 of an RDA Micro SoC, such as RDA8810PL, at the 1370 specified address. The serial port must already be 1371 setup and configured. Options are not yet supported. 1372 1373 sbi 1374 Use RISC-V SBI (Supervisor Binary Interface) for early 1375 console. 1376 1377 smh Use ARM semihosting calls for early console. 1378 1379 s3c2410,<addr> 1380 s3c2412,<addr> 1381 s3c2440,<addr> 1382 s3c6400,<addr> 1383 s5pv210,<addr> 1384 exynos4210,<addr> 1385 Use early console provided by serial driver available 1386 on Samsung SoCs, requires selecting proper type and 1387 a correct base address of the selected UART port. The 1388 serial port must already be setup and configured. 1389 Options are not yet supported. 1390 1391 lantiq,<addr> 1392 Start an early, polled-mode console on a lantiq serial 1393 (lqasc) port at the specified address. The serial port 1394 must already be setup and configured. Options are not 1395 yet supported. 1396 1397 lpuart,<addr> 1398 lpuart32,<addr> 1399 Use early console provided by Freescale LP UART driver 1400 found on Freescale Vybrid and QorIQ LS1021A processors. 1401 A valid base address must be provided, and the serial 1402 port must already be setup and configured. 1403 1404 ec_imx21,<addr> 1405 ec_imx6q,<addr> 1406 Start an early, polled-mode, output-only console on the 1407 Freescale i.MX UART at the specified address. The UART 1408 must already be setup and configured. 1409 1410 ar3700_uart,<addr> 1411 Start an early, polled-mode console on the 1412 Armada 3700 serial port at the specified 1413 address. The serial port must already be setup 1414 and configured. Options are not yet supported. 1415 1416 qcom_geni,<addr> 1417 Start an early, polled-mode console on a Qualcomm 1418 Generic Interface (GENI) based serial port at the 1419 specified address. The serial port must already be 1420 setup and configured. Options are not yet supported. 1421 1422 efifb,[options] 1423 Start an early, unaccelerated console on the EFI 1424 memory mapped framebuffer (if available). On cache 1425 coherent non-x86 systems that use system memory for 1426 the framebuffer, pass the 'ram' option so that it is 1427 mapped with the correct attributes. 1428 1429 linflex,<addr> 1430 Use early console provided by Freescale LINFlexD UART 1431 serial driver for NXP S32V234 SoCs. A valid base 1432 address must be provided, and the serial port must 1433 already be setup and configured. 1434 1435 earlyprintk= [X86,SH,ARM,M68k,S390,UM,EARLY] 1436 earlyprintk=vga 1437 earlyprintk=sclp 1438 earlyprintk=xen 1439 earlyprintk=serial[,ttySn[,baudrate]] 1440 earlyprintk=serial[,0x...[,baudrate]] 1441 earlyprintk=ttySn[,baudrate] 1442 earlyprintk=dbgp[debugController#] 1443 earlyprintk=mmio32,membase[,{nocfg|baudrate}] 1444 earlyprintk=pciserial[,force],bus:device.function[,{nocfg|baudrate}] 1445 earlyprintk=xdbc[xhciController#] 1446 earlyprintk=bios 1447 1448 earlyprintk is useful when the kernel crashes before 1449 the normal console is initialized. It is not enabled by 1450 default because it has some cosmetic problems. 1451 1452 Use "nocfg" to skip UART configuration, assume 1453 BIOS/firmware has configured UART correctly. 1454 1455 Append ",keep" to not disable it when the real console 1456 takes over. 1457 1458 Only one of vga, serial, or usb debug port can 1459 be used at a time. 1460 1461 Currently only ttyS0 and ttyS1 may be specified by 1462 name. Other I/O ports may be explicitly specified 1463 on some architectures (x86 and arm at least) by 1464 replacing ttySn with an I/O port address, like this: 1465 earlyprintk=serial,0x1008,115200 1466 You can find the port for a given device in 1467 /proc/tty/driver/serial: 1468 2: uart:ST16650V2 port:00001008 irq:18 ... 1469 1470 Interaction with the standard serial driver is not 1471 very good. 1472 1473 The VGA output is eventually overwritten by 1474 the real console. 1475 1476 The xen option can only be used in Xen domains. 1477 1478 The sclp output can only be used on s390. 1479 1480 The bios output can only be used on SuperH. 1481 1482 The optional "force" to "pciserial" enables use of a 1483 PCI device even when its classcode is not of the 1484 UART class. 1485 1486 edac_report= [HW,EDAC] Control how to report EDAC event 1487 Format: {"on" | "off" | "force"} 1488 on: enable EDAC to report H/W event. May be overridden 1489 by other higher priority error reporting module. 1490 off: disable H/W event reporting through EDAC. 1491 force: enforce the use of EDAC to report H/W event. 1492 default: on. 1493 1494 edd= [EDD] 1495 Format: {"off" | "on" | "skip[mbr]"} 1496 1497 efi= [EFI,EARLY] 1498 Format: { "debug", "disable_early_pci_dma", 1499 "nochunk", "noruntime", "nosoftreserve", 1500 "novamap", "no_disable_early_pci_dma" } 1501 debug: enable misc debug output. 1502 disable_early_pci_dma: disable the busmaster bit on all 1503 PCI bridges while in the EFI boot stub. 1504 nochunk: disable reading files in "chunks" in the EFI 1505 boot stub, as chunking can cause problems with some 1506 firmware implementations. 1507 noruntime : disable EFI runtime services support 1508 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose) 1509 attribute may cause the kernel to reserve the 1510 memory range for a memory mapping driver to 1511 claim. Specify efi=nosoftreserve to disable this 1512 reservation and treat the memory by its base type 1513 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM"). 1514 novamap: do not call SetVirtualAddressMap(). 1515 no_disable_early_pci_dma: Leave the busmaster bit set 1516 on all PCI bridges while in the EFI boot stub 1517 1518 efi_no_storage_paranoia [EFI,X86,EARLY] 1519 Using this parameter you can use more than 50% of 1520 your efi variable storage. Use this parameter only if 1521 you are really sure that your UEFI does sane gc and 1522 fulfills the spec otherwise your board may brick. 1523 1524 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT 1525 that is to be dynamically loaded by Linux. If there are 1526 multiple variables with the same name but with different 1527 vendor GUIDs, all of them will be loaded. See 1528 Documentation/admin-guide/acpi/ssdt-overlays.rst for details. 1529 1530 1531 eisa_irq_edge= [PARISC,HW] 1532 See header of drivers/parisc/eisa.c. 1533 1534 ekgdboc= [X86,KGDB,EARLY] Allow early kernel console debugging 1535 Format: ekgdboc=kbd 1536 1537 This is designed to be used in conjunction with 1538 the boot argument: earlyprintk=vga 1539 1540 This parameter works in place of the kgdboc parameter 1541 but can only be used if the backing tty is available 1542 very early in the boot process. For early debugging 1543 via a serial port see kgdboc_earlycon instead. 1544 1545 elanfreq= [X86-32] 1546 See comment before function elanfreq_setup() in 1547 arch/x86/kernel/cpu/cpufreq/elanfreq.c. 1548 1549 elfcorehdr=[size[KMG]@]offset[KMG] [PPC,SH,X86,S390,EARLY] 1550 Specifies physical address of start of kernel core 1551 image elf header and optionally the size. Generally 1552 kexec loader will pass this option to capture kernel. 1553 See Documentation/admin-guide/kdump/kdump.rst for details. 1554 1555 enable_mtrr_cleanup [X86,EARLY] 1556 The kernel tries to adjust MTRR layout from continuous 1557 to discrete, to make X server driver able to add WB 1558 entry later. This parameter enables that. 1559 1560 enable_timer_pin_1 [X86] 1561 Enable PIN 1 of APIC timer 1562 Can be useful to work around chipset bugs 1563 (in particular on some ATI chipsets). 1564 The kernel tries to set a reasonable default. 1565 1566 enforcing= [SELINUX] Set initial enforcing status. 1567 Format: {"0" | "1"} 1568 See security/selinux/Kconfig help text. 1569 0 -- permissive (log only, no denials). 1570 1 -- enforcing (deny and log). 1571 Default value is 0. 1572 Value can be changed at runtime via 1573 /sys/fs/selinux/enforce. 1574 1575 erst_disable [ACPI] 1576 Disable Error Record Serialization Table (ERST) 1577 support. 1578 1579 ether= [HW,NET] Ethernet cards parameters 1580 This option is obsoleted by the "netdev=" option, which 1581 has equivalent usage. See its documentation for details. 1582 1583 evm= [EVM] 1584 Format: { "fix" } 1585 Permit 'security.evm' to be updated regardless of 1586 current integrity status. 1587 1588 early_page_ext [KNL,EARLY] Enforces page_ext initialization to earlier 1589 stages so cover more early boot allocations. 1590 Please note that as side effect some optimizations 1591 might be disabled to achieve that (e.g. parallelized 1592 memory initialization is disabled) so the boot process 1593 might take longer, especially on systems with a lot of 1594 memory. Available with CONFIG_PAGE_EXTENSION=y. 1595 1596 failslab= 1597 fail_usercopy= 1598 fail_page_alloc= 1599 fail_skb_realloc= 1600 fail_make_request=[KNL] 1601 General fault injection mechanism. 1602 Format: <interval>,<probability>,<space>,<times> 1603 See also Documentation/fault-injection/. 1604 1605 fb_tunnels= [NET] 1606 Format: { initns | none } 1607 See Documentation/admin-guide/sysctl/net.rst for 1608 fb_tunnels_only_for_init_ns 1609 1610 floppy= [HW] 1611 See Documentation/admin-guide/blockdev/floppy.rst. 1612 1613 forcepae [X86-32] 1614 Forcefully enable Physical Address Extension (PAE). 1615 Many Pentium M systems disable PAE but may have a 1616 functionally usable PAE implementation. 1617 Warning: use of this parameter will taint the kernel 1618 and may cause unknown problems. 1619 1620 fred= [X86-64] 1621 Enable/disable Flexible Return and Event Delivery. 1622 Format: { on | off } 1623 on: enable FRED when it's present. 1624 off: disable FRED, the default setting. 1625 1626 ftrace=[tracer] 1627 [FTRACE] will set and start the specified tracer 1628 as early as possible in order to facilitate early 1629 boot debugging. 1630 1631 ftrace_boot_snapshot 1632 [FTRACE] On boot up, a snapshot will be taken of the 1633 ftrace ring buffer that can be read at: 1634 /sys/kernel/tracing/snapshot. 1635 This is useful if you need tracing information from kernel 1636 boot up that is likely to be overridden by user space 1637 start up functionality. 1638 1639 Optionally, the snapshot can also be defined for a tracing 1640 instance that was created by the trace_instance= command 1641 line parameter. 1642 1643 trace_instance=foo,sched_switch ftrace_boot_snapshot=foo 1644 1645 The above will cause the "foo" tracing instance to trigger 1646 a snapshot at the end of boot up. 1647 1648 ftrace_dump_on_oops[=2(orig_cpu) | =<instance>][,<instance> | 1649 ,<instance>=2(orig_cpu)] 1650 [FTRACE] will dump the trace buffers on oops. 1651 If no parameter is passed, ftrace will dump global 1652 buffers of all CPUs, if you pass 2 or orig_cpu, it 1653 will dump only the buffer of the CPU that triggered 1654 the oops, or the specific instance will be dumped if 1655 its name is passed. Multiple instance dump is also 1656 supported, and instances are separated by commas. Each 1657 instance supports only dump on CPU that triggered the 1658 oops by passing 2 or orig_cpu to it. 1659 1660 ftrace_dump_on_oops=foo=orig_cpu 1661 1662 The above will dump only the buffer of "foo" instance 1663 on CPU that triggered the oops. 1664 1665 ftrace_dump_on_oops,foo,bar=orig_cpu 1666 1667 The above will dump global buffer on all CPUs, the 1668 buffer of "foo" instance on all CPUs and the buffer 1669 of "bar" instance on CPU that triggered the oops. 1670 1671 ftrace_filter=[function-list] 1672 [FTRACE] Limit the functions traced by the function 1673 tracer at boot up. function-list is a comma-separated 1674 list of functions. This list can be changed at run 1675 time by the set_ftrace_filter file in the debugfs 1676 tracing directory. 1677 1678 ftrace_notrace=[function-list] 1679 [FTRACE] Do not trace the functions specified in 1680 function-list. This list can be changed at run time 1681 by the set_ftrace_notrace file in the debugfs 1682 tracing directory. 1683 1684 ftrace_graph_filter=[function-list] 1685 [FTRACE] Limit the top level callers functions traced 1686 by the function graph tracer at boot up. 1687 function-list is a comma-separated list of functions 1688 that can be changed at run time by the 1689 set_graph_function file in the debugfs tracing directory. 1690 1691 ftrace_graph_notrace=[function-list] 1692 [FTRACE] Do not trace from the functions specified in 1693 function-list. This list is a comma-separated list of 1694 functions that can be changed at run time by the 1695 set_graph_notrace file in the debugfs tracing directory. 1696 1697 ftrace_graph_max_depth=<uint> 1698 [FTRACE] Used with the function graph tracer. This is 1699 the max depth it will trace into a function. This value 1700 can be changed at run time by the max_graph_depth file 1701 in the tracefs tracing directory. default: 0 (no limit) 1702 1703 fw_devlink= [KNL,EARLY] Create device links between consumer and supplier 1704 devices by scanning the firmware to infer the 1705 consumer/supplier relationships. This feature is 1706 especially useful when drivers are loaded as modules as 1707 it ensures proper ordering of tasks like device probing 1708 (suppliers first, then consumers), supplier boot state 1709 clean up (only after all consumers have probed), 1710 suspend/resume & runtime PM (consumers first, then 1711 suppliers). 1712 Format: { off | permissive | on | rpm } 1713 off -- Don't create device links from firmware info. 1714 permissive -- Create device links from firmware info 1715 but use it only for ordering boot state clean 1716 up (sync_state() calls). 1717 on -- Create device links from firmware info and use it 1718 to enforce probe and suspend/resume ordering. 1719 rpm -- Like "on", but also use to order runtime PM. 1720 1721 fw_devlink.strict=<bool> 1722 [KNL,EARLY] Treat all inferred dependencies as mandatory 1723 dependencies. This only applies for fw_devlink=on|rpm. 1724 Format: <bool> 1725 1726 fw_devlink.sync_state = 1727 [KNL,EARLY] When all devices that could probe have finished 1728 probing, this parameter controls what to do with 1729 devices that haven't yet received their sync_state() 1730 calls. 1731 Format: { strict | timeout } 1732 strict -- Default. Continue waiting on consumers to 1733 probe successfully. 1734 timeout -- Give up waiting on consumers and call 1735 sync_state() on any devices that haven't yet 1736 received their sync_state() calls after 1737 deferred_probe_timeout has expired or by 1738 late_initcall() if !CONFIG_MODULES. 1739 1740 gamecon.map[2|3]= 1741 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad 1742 support via parallel port (up to 5 devices per port) 1743 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5> 1744 See also Documentation/input/devices/joystick-parport.rst 1745 1746 gamma= [HW,DRM] 1747 1748 gart_fix_e820= [X86-64,EARLY] disable the fix e820 for K8 GART 1749 Format: off | on 1750 default: on 1751 1752 gather_data_sampling= 1753 [X86,INTEL,EARLY] Control the Gather Data Sampling (GDS) 1754 mitigation. 1755 1756 Gather Data Sampling is a hardware vulnerability which 1757 allows unprivileged speculative access to data which was 1758 previously stored in vector registers. 1759 1760 This issue is mitigated by default in updated microcode. 1761 The mitigation may have a performance impact but can be 1762 disabled. On systems without the microcode mitigation 1763 disabling AVX serves as a mitigation. 1764 1765 force: Disable AVX to mitigate systems without 1766 microcode mitigation. No effect if the microcode 1767 mitigation is present. Known to cause crashes in 1768 userspace with buggy AVX enumeration. 1769 1770 off: Disable GDS mitigation. 1771 1772 gbpages [X86] Use GB pages for kernel direct mappings. 1773 1774 gcov_persist= [GCOV] When non-zero (default), profiling data for 1775 kernel modules is saved and remains accessible via 1776 debugfs, even when the module is unloaded/reloaded. 1777 When zero, profiling data is discarded and associated 1778 debugfs files are removed at module unload time. 1779 1780 goldfish [X86] Enable the goldfish android emulator platform. 1781 Don't use this when you are not running on the 1782 android emulator 1783 1784 gpio-mockup.gpio_mockup_ranges 1785 [HW] Sets the ranges of gpiochip of for this device. 1786 Format: <start1>,<end1>,<start2>,<end2>... 1787 gpio-mockup.gpio_mockup_named_lines 1788 [HW] Let the driver know GPIO lines should be named. 1789 1790 gpt [EFI] Forces disk with valid GPT signature but 1791 invalid Protective MBR to be treated as GPT. If the 1792 primary GPT is corrupted, it enables the backup/alternate 1793 GPT to be used instead. 1794 1795 grcan.enable0= [HW] Configuration of physical interface 0. Determines 1796 the "Enable 0" bit of the configuration register. 1797 Format: 0 | 1 1798 Default: 0 1799 grcan.enable1= [HW] Configuration of physical interface 1. Determines 1800 the "Enable 0" bit of the configuration register. 1801 Format: 0 | 1 1802 Default: 0 1803 grcan.select= [HW] Select which physical interface to use. 1804 Format: 0 | 1 1805 Default: 0 1806 grcan.txsize= [HW] Sets the size of the tx buffer. 1807 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0. 1808 Default: 1024 1809 grcan.rxsize= [HW] Sets the size of the rx buffer. 1810 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0. 1811 Default: 1024 1812 1813 hardened_usercopy= 1814 [KNL] Under CONFIG_HARDENED_USERCOPY, whether 1815 hardening is enabled for this boot. Hardened 1816 usercopy checking is used to protect the kernel 1817 from reading or writing beyond known memory 1818 allocation boundaries as a proactive defense 1819 against bounds-checking flaws in the kernel's 1820 copy_to_user()/copy_from_user() interface. 1821 The default is determined by 1822 CONFIG_HARDENED_USERCOPY_DEFAULT_ON. 1823 on Perform hardened usercopy checks. 1824 off Disable hardened usercopy checks. 1825 1826 hardlockup_all_cpu_backtrace= 1827 [KNL] Should the hard-lockup detector generate 1828 backtraces on all cpus. 1829 Format: 0 | 1 1830 1831 hash_pointers= 1832 [KNL,EARLY] 1833 By default, when pointers are printed to the console 1834 or buffers via the %p format string, that pointer is 1835 "hashed", i.e. obscured by hashing the pointer value. 1836 This is a security feature that hides actual kernel 1837 addresses from unprivileged users, but it also makes 1838 debugging the kernel more difficult since unequal 1839 pointers can no longer be compared. The choices are: 1840 Format: { auto | always | never } 1841 Default: auto 1842 1843 auto - Hash pointers unless slab_debug is enabled. 1844 always - Always hash pointers (even if slab_debug is 1845 enabled). 1846 never - Never hash pointers. This option should only 1847 be specified when debugging the kernel. Do 1848 not use on production kernels. The boot 1849 param "no_hash_pointers" is an alias for 1850 this mode. 1851 1852 hashdist= [KNL,NUMA] Large hashes allocated during boot 1853 are distributed across NUMA nodes. Defaults on 1854 for 64-bit NUMA, off otherwise. 1855 Format: 0 | 1 (for off | on) 1856 1857 hd= [EIDE] (E)IDE hard drive subsystem geometry 1858 Format: <cyl>,<head>,<sect> 1859 1860 hest_disable [ACPI] 1861 Disable Hardware Error Source Table (HEST) support; 1862 corresponding firmware-first mode error processing 1863 logic will be disabled. 1864 1865 hibernate= [HIBERNATION] 1866 noresume Don't check if there's a hibernation image 1867 present during boot. 1868 nocompress Don't compress/decompress hibernation images. 1869 no Disable hibernation and resume. 1870 protect_image Turn on image protection during restoration 1871 (that will set all pages holding image data 1872 during restoration read-only). 1873 1874 hibernate.compressor= [HIBERNATION] Compression algorithm to be 1875 used with hibernation. 1876 Format: { lzo | lz4 } 1877 Default: lzo 1878 1879 lzo: Select LZO compression algorithm to 1880 compress/decompress hibernation image. 1881 1882 lz4: Select LZ4 compression algorithm to 1883 compress/decompress hibernation image. 1884 1885 hibernate.pm_test_delay= 1886 [HIBERNATION] 1887 Sets the number of seconds to remain in a hibernation test 1888 mode before resuming the system (see 1889 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG 1890 is set. Default value is 5. 1891 1892 highmem=nn[KMG] [KNL,BOOT,EARLY] forces the highmem zone to have an exact 1893 size of <nn>. This works even on boxes that have no 1894 highmem otherwise. This also works to reduce highmem 1895 size on bigger boxes. 1896 1897 highres= [KNL] Enable/disable high resolution timer mode. 1898 Valid parameters: "on", "off" 1899 Default: "on" 1900 1901 hlt [BUGS=ARM,SH] 1902 1903 hostname= [KNL,EARLY] Set the hostname (aka UTS nodename). 1904 Format: <string> 1905 This allows setting the system's hostname during early 1906 startup. This sets the name returned by gethostname. 1907 Using this parameter to set the hostname makes it 1908 possible to ensure the hostname is correctly set before 1909 any userspace processes run, avoiding the possibility 1910 that a process may call gethostname before the hostname 1911 has been explicitly set, resulting in the calling 1912 process getting an incorrect result. The string must 1913 not exceed the maximum allowed hostname length (usually 1914 64 characters) and will be truncated otherwise. 1915 1916 hpet= [X86-32,HPET] option to control HPET usage 1917 Format: { enable (default) | disable | force | 1918 verbose } 1919 disable: disable HPET and use PIT instead 1920 force: allow force enabled of undocumented chips (ICH4, 1921 VIA, nVidia) 1922 verbose: show contents of HPET registers during setup 1923 1924 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET 1925 registers. Default set by CONFIG_HPET_MMAP_DEFAULT. 1926 1927 hugepages= [HW,EARLY] Number of HugeTLB pages to allocate at boot. 1928 If this follows hugepagesz (below), it specifies 1929 the number of pages of hugepagesz to be allocated. 1930 If this is the first HugeTLB parameter on the command 1931 line, it specifies the number of pages to allocate for 1932 the default huge page size. If using node format, the 1933 number of pages to allocate per-node can be specified. 1934 See also Documentation/admin-guide/mm/hugetlbpage.rst. 1935 Format: <integer> or (node format) 1936 <node>:<integer>[,<node>:<integer>] 1937 1938 hugepagesz= 1939 [HW,EARLY] The size of the HugeTLB pages. This is 1940 used in conjunction with hugepages (above) to 1941 allocate huge pages of a specific size at boot. The 1942 pair hugepagesz=X hugepages=Y can be specified once 1943 for each supported huge page size. Huge page sizes 1944 are architecture dependent. See also 1945 Documentation/admin-guide/mm/hugetlbpage.rst. 1946 Format: size[KMG] 1947 1948 hugepage_alloc_threads= 1949 [HW] The number of threads that should be used to 1950 allocate hugepages during boot. This option can be 1951 used to improve system bootup time when allocating 1952 a large amount of huge pages. 1953 The default value is 25% of the available hardware threads. 1954 1955 Note that this parameter only applies to non-gigantic huge pages. 1956 1957 hugetlb_cma= [HW,CMA,EARLY] The size of a CMA area used for allocation 1958 of gigantic hugepages. Or using node format, the size 1959 of a CMA area per node can be specified. 1960 Format: nn[KMGTPE] or (node format) 1961 <node>:nn[KMGTPE][,<node>:nn[KMGTPE]] 1962 1963 Reserve a CMA area of given size and allocate gigantic 1964 hugepages using the CMA allocator. If enabled, the 1965 boot-time allocation of gigantic hugepages is skipped. 1966 1967 hugetlb_cma_only= 1968 [HW,CMA,EARLY] When allocating new HugeTLB pages, only 1969 try to allocate from the CMA areas. 1970 1971 This option does nothing if hugetlb_cma= is not also 1972 specified. 1973 1974 hugetlb_free_vmemmap= 1975 [KNL] Requires CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP 1976 enabled. 1977 Control if HugeTLB Vmemmap Optimization (HVO) is enabled. 1978 Allows heavy hugetlb users to free up some more 1979 memory (7 * PAGE_SIZE for each 2MB hugetlb page). 1980 Format: { on | off (default) } 1981 1982 on: enable HVO 1983 off: disable HVO 1984 1985 Built with CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP_DEFAULT_ON=y, 1986 the default is on. 1987 1988 Note that the vmemmap pages may be allocated from the added 1989 memory block itself when memory_hotplug.memmap_on_memory is 1990 enabled, those vmemmap pages cannot be optimized even if this 1991 feature is enabled. Other vmemmap pages not allocated from 1992 the added memory block itself do not be affected. 1993 1994 hung_task_panic= 1995 [KNL] Should the hung task detector generate panics. 1996 Format: 0 | 1 1997 1998 A value of 1 instructs the kernel to panic when a 1999 hung task is detected. The default value is controlled 2000 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time 2001 option. The value selected by this boot parameter can 2002 be changed later by the kernel.hung_task_panic sysctl. 2003 2004 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC) 2005 terminal devices. Valid values: 0..8 2006 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs. 2007 If specified, z/VM IUCV HVC accepts connections 2008 from listed z/VM user IDs only. 2009 2010 hv_nopvspin [X86,HYPER_V,EARLY] 2011 Disables the paravirt spinlock optimizations 2012 which allow the hypervisor to 'idle' the guest 2013 on lock contention. 2014 2015 hw_protection= [HW] 2016 Format: reboot | shutdown 2017 2018 Hardware protection action taken on critical events like 2019 overtemperature or imminent voltage loss. 2020 2021 i2c_bus= [HW] Override the default board specific I2C bus speed 2022 or register an additional I2C bus that is not 2023 registered from board initialization code. 2024 Format: 2025 <bus_id>,<clkrate> 2026 2027 i2c_touchscreen_props= [HW,ACPI,X86] 2028 Set device-properties for ACPI-enumerated I2C-attached 2029 touchscreen, to e.g. fix coordinates of upside-down 2030 mounted touchscreens. If you need this option please 2031 submit a drivers/platform/x86/touchscreen_dmi.c patch 2032 adding a DMI quirk for this. 2033 2034 Format: 2035 <ACPI_HW_ID>:<prop_name>=<val>[:prop_name=val][:...] 2036 Where <val> is one of: 2037 Omit "=<val>" entirely Set a boolean device-property 2038 Unsigned number Set a u32 device-property 2039 Anything else Set a string device-property 2040 2041 Examples (split over multiple lines): 2042 i2c_touchscreen_props=GDIX1001:touchscreen-inverted-x: 2043 touchscreen-inverted-y 2044 2045 i2c_touchscreen_props=MSSL1680:touchscreen-size-x=1920: 2046 touchscreen-size-y=1080:touchscreen-inverted-y: 2047 firmware-name=gsl1680-vendor-model.fw:silead,home-button 2048 2049 i8042.debug [HW] Toggle i8042 debug mode 2050 i8042.unmask_kbd_data 2051 [HW] Enable printing of interrupt data from the KBD port 2052 (disabled by default, and as a pre-condition 2053 requires that i8042.debug=1 be enabled) 2054 i8042.direct [HW] Put keyboard port into non-translated mode 2055 i8042.dumbkbd [HW] Pretend that controller can only read data from 2056 keyboard and cannot control its state 2057 (Don't attempt to blink the leds) 2058 i8042.noaux [HW] Don't check for auxiliary (== mouse) port 2059 i8042.nokbd [HW] Don't check/create keyboard port 2060 i8042.noloop [HW] Disable the AUX Loopback command while probing 2061 for the AUX port 2062 i8042.nomux [HW] Don't check presence of an active multiplexing 2063 controller 2064 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX 2065 controllers 2066 i8042.notimeout [HW] Ignore timeout condition signalled by controller 2067 i8042.reset [HW] Reset the controller during init, cleanup and 2068 suspend-to-ram transitions, only during s2r 2069 transitions, or never reset 2070 Format: { 1 | Y | y | 0 | N | n } 2071 1, Y, y: always reset controller 2072 0, N, n: don't ever reset controller 2073 Default: only on s2r transitions on x86; most other 2074 architectures force reset to be always executed 2075 i8042.unlock [HW] Unlock (ignore) the keylock 2076 i8042.kbdreset [HW] Reset device connected to KBD port 2077 i8042.probe_defer 2078 [HW] Allow deferred probing upon i8042 probe errors 2079 2080 i810= [HW,DRM] 2081 2082 i915.invert_brightness= 2083 [DRM] Invert the sense of the variable that is used to 2084 set the brightness of the panel backlight. Normally a 2085 brightness value of 0 indicates backlight switched off, 2086 and the maximum of the brightness value sets the backlight 2087 to maximum brightness. If this parameter is set to 0 2088 (default) and the machine requires it, or this parameter 2089 is set to 1, a brightness value of 0 sets the backlight 2090 to maximum brightness, and the maximum of the brightness 2091 value switches the backlight off. 2092 -1 -- never invert brightness 2093 0 -- machine default 2094 1 -- force brightness inversion 2095 2096 ia32_emulation= [X86-64] 2097 Format: <bool> 2098 When true, allows loading 32-bit programs and executing 32-bit 2099 syscalls, essentially overriding IA32_EMULATION_DEFAULT_DISABLED at 2100 boot time. When false, unconditionally disables IA32 emulation. 2101 2102 icn= [HW,ISDN] 2103 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]] 2104 2105 2106 idle= [X86,EARLY] 2107 Format: idle=poll, idle=halt, idle=nomwait 2108 2109 idle=poll: Don't do power saving in the idle loop 2110 using HLT, but poll for rescheduling event. This will 2111 make the CPUs eat a lot more power, but may be useful 2112 to get slightly better performance in multiprocessor 2113 benchmarks. It also makes some profiling using 2114 performance counters more accurate. Please note that 2115 on systems with MONITOR/MWAIT support (like Intel 2116 EM64T CPUs) this option has no performance advantage 2117 over the normal idle loop. It may also interact badly 2118 with hyperthreading. 2119 2120 idle=halt: Halt is forced to be used for CPU idle. 2121 In such case C2/C3 won't be used again. 2122 2123 idle=nomwait: Disable mwait for CPU C-states 2124 2125 idxd.sva= [HW] 2126 Format: <bool> 2127 Allow force disabling of Shared Virtual Memory (SVA) 2128 support for the idxd driver. By default it is set to 2129 true (1). 2130 2131 idxd.tc_override= [HW] 2132 Format: <bool> 2133 Allow override of default traffic class configuration 2134 for the device. By default it is set to false (0). 2135 2136 ieee754= [MIPS] Select IEEE Std 754 conformance mode 2137 Format: { strict | legacy | 2008 | relaxed | emulated } 2138 Default: strict 2139 2140 Choose which programs will be accepted for execution 2141 based on the IEEE 754 NaN encoding(s) supported by 2142 the FPU and the NaN encoding requested with the value 2143 of an ELF file header flag individually set by each 2144 binary. Hardware implementations are permitted to 2145 support either or both of the legacy and the 2008 NaN 2146 encoding mode. 2147 2148 Available settings are as follows: 2149 strict accept binaries that request a NaN encoding 2150 supported by the FPU 2151 legacy only accept legacy-NaN binaries, if supported 2152 by the FPU 2153 2008 only accept 2008-NaN binaries, if supported 2154 by the FPU 2155 relaxed accept any binaries regardless of whether 2156 supported by the FPU 2157 emulated accept any binaries but enable FPU emulator 2158 if binary mode is unsupported by the FPU. 2159 2160 The FPU emulator is always able to support both NaN 2161 encodings, so if no FPU hardware is present or it has 2162 been disabled with 'nofpu', then the settings of 2163 'legacy' and '2008' strap the emulator accordingly, 2164 'relaxed' straps the emulator for both legacy-NaN and 2165 2008-NaN, whereas 'strict' enables legacy-NaN only on 2166 legacy processors and both NaN encodings on MIPS32 or 2167 MIPS64 CPUs. 2168 2169 The setting for ABS.fmt/NEG.fmt instruction execution 2170 mode generally follows that for the NaN encoding, 2171 except where unsupported by hardware. 2172 2173 ignore_loglevel [KNL,EARLY] 2174 Ignore loglevel setting - this will print /all/ 2175 kernel messages to the console. Useful for debugging. 2176 We also add it as printk module parameter, so users 2177 could change it dynamically, usually by 2178 /sys/module/printk/parameters/ignore_loglevel. 2179 2180 ignore_rlimit_data 2181 Ignore RLIMIT_DATA setting for data mappings, 2182 print warning at first misuse. Can be changed via 2183 /sys/module/kernel/parameters/ignore_rlimit_data. 2184 2185 ihash_entries= [KNL] 2186 Set number of hash buckets for inode cache. 2187 2188 ima_appraise= [IMA] appraise integrity measurements 2189 Format: { "off" | "enforce" | "fix" | "log" } 2190 default: "enforce" 2191 2192 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead. 2193 The builtin appraise policy appraises all files 2194 owned by uid=0. 2195 2196 ima_canonical_fmt [IMA] 2197 Use the canonical format for the binary runtime 2198 measurements, instead of host native format. 2199 2200 ima_hash= [IMA] 2201 Format: { md5 | sha1 | rmd160 | sha256 | sha384 2202 | sha512 | ... } 2203 default: "sha1" 2204 2205 The list of supported hash algorithms is defined 2206 in crypto/hash_info.h. 2207 2208 ima_policy= [IMA] 2209 The builtin policies to load during IMA setup. 2210 Format: "tcb | appraise_tcb | secure_boot | 2211 fail_securely | critical_data" 2212 2213 The "tcb" policy measures all programs exec'd, files 2214 mmap'd for exec, and all files opened with the read 2215 mode bit set by either the effective uid (euid=0) or 2216 uid=0. 2217 2218 The "appraise_tcb" policy appraises the integrity of 2219 all files owned by root. 2220 2221 The "secure_boot" policy appraises the integrity 2222 of files (eg. kexec kernel image, kernel modules, 2223 firmware, policy, etc) based on file signatures. 2224 2225 The "fail_securely" policy forces file signature 2226 verification failure also on privileged mounted 2227 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE 2228 flag. 2229 2230 The "critical_data" policy measures kernel integrity 2231 critical data. 2232 2233 ima_tcb [IMA] Deprecated. Use ima_policy= instead. 2234 Load a policy which meets the needs of the Trusted 2235 Computing Base. This means IMA will measure all 2236 programs exec'd, files mmap'd for exec, and all files 2237 opened for read by uid=0. 2238 2239 ima_template= [IMA] 2240 Select one of defined IMA measurements template formats. 2241 Formats: { "ima" | "ima-ng" | "ima-ngv2" | "ima-sig" | 2242 "ima-sigv2" } 2243 Default: "ima-ng" 2244 2245 ima_template_fmt= 2246 [IMA] Define a custom template format. 2247 Format: { "field1|...|fieldN" } 2248 2249 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage 2250 Format: <min_file_size> 2251 Set the minimal file size for using asynchronous hash. 2252 If left unspecified, ahash usage is disabled. 2253 2254 ahash performance varies for different data sizes on 2255 different crypto accelerators. This option can be used 2256 to achieve the best performance for a particular HW. 2257 2258 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size 2259 Format: <bufsize> 2260 Set hashing buffer size. Default: 4k. 2261 2262 ahash performance varies for different chunk sizes on 2263 different crypto accelerators. This option can be used 2264 to achieve best performance for particular HW. 2265 2266 ima= [IMA] Enable or disable IMA 2267 Format: { "off" | "on" } 2268 Default: "on" 2269 Note that disabling IMA is limited to kdump kernel. 2270 2271 indirect_target_selection= [X86,Intel] Mitigation control for Indirect 2272 Target Selection(ITS) bug in Intel CPUs. Updated 2273 microcode is also required for a fix in IBPB. 2274 2275 on: Enable mitigation (default). 2276 off: Disable mitigation. 2277 force: Force the ITS bug and deploy default 2278 mitigation. 2279 vmexit: Only deploy mitigation if CPU is affected by 2280 guest/host isolation part of ITS. 2281 stuff: Deploy RSB-fill mitigation when retpoline is 2282 also deployed. Otherwise, deploy the default 2283 mitigation. 2284 2285 For details see: 2286 Documentation/admin-guide/hw-vuln/indirect-target-selection.rst 2287 2288 init= [KNL] 2289 Format: <full_path> 2290 Run specified binary instead of /sbin/init as init 2291 process. 2292 2293 initcall_debug [KNL] Trace initcalls as they are executed. Useful 2294 for working out where the kernel is dying during 2295 startup. 2296 2297 initcall_blacklist= [KNL] Do not execute a comma-separated list of 2298 initcall functions. Useful for debugging built-in 2299 modules and initcalls. 2300 2301 initramfs_async= [KNL] 2302 Format: <bool> 2303 Default: 1 2304 This parameter controls whether the initramfs 2305 image is unpacked asynchronously, concurrently 2306 with devices being probed and 2307 initialized. This should normally just work, 2308 but as a debugging aid, one can get the 2309 historical behaviour of the initramfs 2310 unpacking being completed before device_ and 2311 late_ initcalls. 2312 2313 initrd= [BOOT,EARLY] Specify the location of the initial ramdisk 2314 2315 initrdmem= [KNL,EARLY] Specify a physical address and size from which to 2316 load the initrd. If an initrd is compiled in or 2317 specified in the bootparams, it takes priority over this 2318 setting. 2319 Format: ss[KMG],nn[KMG] 2320 Default is 0, 0 2321 2322 init_on_alloc= [MM,EARLY] Fill newly allocated pages and heap objects with 2323 zeroes. 2324 Format: 0 | 1 2325 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON. 2326 2327 init_on_free= [MM,EARLY] Fill freed pages and heap objects with zeroes. 2328 Format: 0 | 1 2329 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON. 2330 2331 init_pkru= [X86] Specify the default memory protection keys rights 2332 register contents for all processes. 0x55555554 by 2333 default (disallow access to all but pkey 0). Can 2334 override in debugfs after boot. 2335 2336 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver 2337 Format: <irq> 2338 2339 int_pln_enable [X86] Enable power limit notification interrupt 2340 2341 integrity_audit=[IMA] 2342 Format: { "0" | "1" } 2343 0 -- basic integrity auditing messages. (Default) 2344 1 -- additional integrity auditing messages. 2345 2346 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option 2347 on 2348 Enable intel iommu driver. 2349 off 2350 Disable intel iommu driver. 2351 igfx_off [Default Off] 2352 By default, gfx is mapped as normal device. If a gfx 2353 device has a dedicated DMAR unit, the DMAR unit is 2354 bypassed by not enabling DMAR with this option. In 2355 this case, gfx device will use physical address for 2356 DMA. 2357 strict [Default Off] 2358 Deprecated, equivalent to iommu.strict=1. 2359 sp_off [Default Off] 2360 By default, super page will be supported if Intel IOMMU 2361 has the capability. With this option, super page will 2362 not be supported. 2363 sm_on 2364 Enable the Intel IOMMU scalable mode if the hardware 2365 advertises that it has support for the scalable mode 2366 translation. 2367 sm_off 2368 Disallow use of the Intel IOMMU scalable mode. 2369 tboot_noforce [Default Off] 2370 Do not force the Intel IOMMU enabled under tboot. 2371 By default, tboot will force Intel IOMMU on, which 2372 could harm performance of some high-throughput 2373 devices like 40GBit network cards, even if identity 2374 mapping is enabled. 2375 Note that using this option lowers the security 2376 provided by tboot because it makes the system 2377 vulnerable to DMA attacks. 2378 2379 intel_idle.max_cstate= [KNL,HW,ACPI,X86] 2380 0 disables intel_idle and fall back on acpi_idle. 2381 1 to 9 specify maximum depth of C-state. 2382 2383 intel_pstate= [X86,EARLY] 2384 disable 2385 Do not enable intel_pstate as the default 2386 scaling driver for the supported processors 2387 active 2388 Use intel_pstate driver to bypass the scaling 2389 governors layer of cpufreq and provides it own 2390 algorithms for p-state selection. There are two 2391 P-state selection algorithms provided by 2392 intel_pstate in the active mode: powersave and 2393 performance. The way they both operate depends 2394 on whether or not the hardware managed P-states 2395 (HWP) feature has been enabled in the processor 2396 and possibly on the processor model. 2397 passive 2398 Use intel_pstate as a scaling driver, but configure it 2399 to work with generic cpufreq governors (instead of 2400 enabling its internal governor). This mode cannot be 2401 used along with the hardware-managed P-states (HWP) 2402 feature. 2403 force 2404 Enable intel_pstate on systems that prohibit it by default 2405 in favor of acpi-cpufreq. Forcing the intel_pstate driver 2406 instead of acpi-cpufreq may disable platform features, such 2407 as thermal controls and power capping, that rely on ACPI 2408 P-States information being indicated to OSPM and therefore 2409 should be used with caution. This option does not work with 2410 processors that aren't supported by the intel_pstate driver 2411 or on platforms that use pcc-cpufreq instead of acpi-cpufreq. 2412 no_hwp 2413 Do not enable hardware P state control (HWP) 2414 if available. 2415 hwp_only 2416 Only load intel_pstate on systems which support 2417 hardware P state control (HWP) if available. 2418 support_acpi_ppc 2419 Enforce ACPI _PPC performance limits. If the Fixed ACPI 2420 Description Table, specifies preferred power management 2421 profile as "Enterprise Server" or "Performance Server", 2422 then this feature is turned on by default. 2423 per_cpu_perf_limits 2424 Allow per-logical-CPU P-State performance control limits using 2425 cpufreq sysfs interface 2426 no_cas 2427 Do not enable capacity-aware scheduling (CAS) on 2428 hybrid systems 2429 2430 intremap= [X86-64,Intel-IOMMU,EARLY] 2431 on enable Interrupt Remapping (default) 2432 off disable Interrupt Remapping 2433 nosid disable Source ID checking 2434 no_x2apic_optout 2435 BIOS x2APIC opt-out request will be ignored 2436 nopost disable Interrupt Posting 2437 posted_msi 2438 enable MSIs delivered as posted interrupts 2439 2440 iomem= Disable strict checking of access to MMIO memory 2441 strict regions from userspace. 2442 relaxed 2443 2444 iommu= [X86,EARLY] 2445 2446 off 2447 Don't initialize and use any kind of IOMMU. 2448 2449 force 2450 Force the use of the hardware IOMMU even when 2451 it is not actually needed (e.g. because < 3 GB 2452 memory). 2453 2454 noforce 2455 Don't force hardware IOMMU usage when it is not 2456 needed. (default). 2457 2458 biomerge 2459 panic 2460 nopanic 2461 merge 2462 nomerge 2463 2464 soft 2465 Use software bounce buffering (SWIOTLB) (default for 2466 Intel machines). This can be used to prevent the usage 2467 of an available hardware IOMMU. 2468 2469 [X86] 2470 pt 2471 [X86] 2472 nopt 2473 [PPC/POWERNV] 2474 nobypass 2475 Disable IOMMU bypass, using IOMMU for PCI devices. 2476 2477 [X86] 2478 AMD Gart HW IOMMU-specific options: 2479 2480 <size> 2481 Set the size of the remapping area in bytes. 2482 2483 allowed 2484 Overwrite iommu off workarounds for specific chipsets 2485 2486 fullflush 2487 Flush IOMMU on each allocation (default). 2488 2489 nofullflush 2490 Don't use IOMMU fullflush. 2491 2492 memaper[=<order>] 2493 Allocate an own aperture over RAM with size 2494 32MB<<order. (default: order=1, i.e. 64MB) 2495 2496 merge 2497 Do scatter-gather (SG) merging. Implies "force" 2498 (experimental). 2499 2500 nomerge 2501 Don't do scatter-gather (SG) merging. 2502 2503 noaperture 2504 Ask the IOMMU not to touch the aperture for AGP. 2505 2506 noagp 2507 Don't initialize the AGP driver and use full aperture. 2508 2509 panic 2510 Always panic when IOMMU overflows. 2511 2512 iommu.forcedac= [ARM64,X86,EARLY] Control IOVA allocation for PCI devices. 2513 Format: { "0" | "1" } 2514 0 - Try to allocate a 32-bit DMA address first, before 2515 falling back to the full range if needed. 2516 1 - Allocate directly from the full usable range, 2517 forcing Dual Address Cycle for PCI cards supporting 2518 greater than 32-bit addressing. 2519 2520 iommu.strict= [ARM64,X86,S390,EARLY] Configure TLB invalidation behaviour 2521 Format: { "0" | "1" } 2522 0 - Lazy mode. 2523 Request that DMA unmap operations use deferred 2524 invalidation of hardware TLBs, for increased 2525 throughput at the cost of reduced device isolation. 2526 Will fall back to strict mode if not supported by 2527 the relevant IOMMU driver. 2528 1 - Strict mode. 2529 DMA unmap operations invalidate IOMMU hardware TLBs 2530 synchronously. 2531 unset - Use value of CONFIG_IOMMU_DEFAULT_DMA_{LAZY,STRICT}. 2532 Note: on x86, strict mode specified via one of the 2533 legacy driver-specific options takes precedence. 2534 2535 iommu.passthrough= 2536 [ARM64,X86,EARLY] Configure DMA to bypass the IOMMU by default. 2537 Format: { "0" | "1" } 2538 0 - Use IOMMU translation for DMA. 2539 1 - Bypass the IOMMU for DMA. 2540 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH. 2541 2542 io7= [HW] IO7 for Marvel-based Alpha systems 2543 See comment before marvel_specify_io7 in 2544 arch/alpha/kernel/core_marvel.c. 2545 2546 io_delay= [X86,EARLY] I/O delay method 2547 0x80 2548 Standard port 0x80 based delay 2549 0xed 2550 Alternate port 0xed based delay (needed on some systems) 2551 udelay 2552 Simple two microseconds delay 2553 none 2554 No delay 2555 2556 ip= [IP_PNP] 2557 See Documentation/admin-guide/nfs/nfsroot.rst. 2558 2559 ipcmni_extend [KNL,EARLY] Extend the maximum number of unique System V 2560 IPC identifiers from 32,768 to 16,777,216. 2561 2562 ipe.enforce= [IPE] 2563 Format: <bool> 2564 Determine whether IPE starts in permissive (0) or 2565 enforce (1) mode. The default is enforce. 2566 2567 ipe.success_audit= 2568 [IPE] 2569 Format: <bool> 2570 Start IPE with success auditing enabled, emitting 2571 an audit event when a binary is allowed. The default 2572 is 0. 2573 2574 irqaffinity= [SMP] Set the default irq affinity mask 2575 The argument is a cpu list, as described above. 2576 2577 irqchip.gicv2_force_probe= 2578 [ARM,ARM64,EARLY] 2579 Format: <bool> 2580 Force the kernel to look for the second 4kB page 2581 of a GICv2 controller even if the memory range 2582 exposed by the device tree is too small. 2583 2584 irqchip.gicv3_nolpi= 2585 [ARM,ARM64,EARLY] 2586 Force the kernel to ignore the availability of 2587 LPIs (and by consequence ITSs). Intended for system 2588 that use the kernel as a bootloader, and thus want 2589 to let secondary kernels in charge of setting up 2590 LPIs. 2591 2592 irqchip.gicv3_pseudo_nmi= [ARM64,EARLY] 2593 Enables support for pseudo-NMIs in the kernel. This 2594 requires the kernel to be built with 2595 CONFIG_ARM64_PSEUDO_NMI. 2596 2597 irqchip.riscv_imsic_noipi 2598 [RISC-V,EARLY] 2599 Force the kernel to not use IMSIC software injected MSIs 2600 as IPIs. Intended for system where IMSIC is trap-n-emulated, 2601 and thus want to reduce MMIO traps when triggering IPIs 2602 to multiple harts. 2603 2604 irqfixup [HW] 2605 When an interrupt is not handled search all handlers 2606 for it. Intended to get systems with badly broken 2607 firmware running. 2608 2609 irqpoll [HW] 2610 When an interrupt is not handled search all handlers 2611 for it. Also check all handlers each timer 2612 interrupt. Intended to get systems with badly broken 2613 firmware running. 2614 2615 isapnp= [ISAPNP] 2616 Format: <RDP>,<reset>,<pci_scan>,<verbosity> 2617 2618 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance. 2619 [Deprecated - use cpusets instead] 2620 Format: [flag-list,]<cpu-list> 2621 2622 Specify one or more CPUs to isolate from disturbances 2623 specified in the flag list (default: domain): 2624 2625 nohz 2626 Disable the tick when a single task runs as well as 2627 disabling other kernel noises like having RCU callbacks 2628 offloaded. This is equivalent to the nohz_full parameter. 2629 2630 A residual 1Hz tick is offloaded to workqueues, which you 2631 need to affine to housekeeping through the global 2632 workqueue's affinity configured via the 2633 /sys/devices/virtual/workqueue/cpumask sysfs file, or 2634 by using the 'domain' flag described below. 2635 2636 NOTE: by default the global workqueue runs on all CPUs, 2637 so to protect individual CPUs the 'cpumask' file has to 2638 be configured manually after bootup. 2639 2640 domain 2641 Isolate from the general SMP balancing and scheduling 2642 algorithms. Note that performing domain isolation this way 2643 is irreversible: it's not possible to bring back a CPU to 2644 the domains once isolated through isolcpus. It's strongly 2645 advised to use cpusets instead to disable scheduler load 2646 balancing through the "cpuset.sched_load_balance" file. 2647 It offers a much more flexible interface where CPUs can 2648 move in and out of an isolated set anytime. 2649 2650 You can move a process onto or off an "isolated" CPU via 2651 the CPU affinity syscalls or cpuset. 2652 <cpu number> begins at 0 and the maximum value is 2653 "number of CPUs in system - 1". 2654 2655 managed_irq 2656 2657 Isolate from being targeted by managed interrupts 2658 which have an interrupt mask containing isolated 2659 CPUs. The affinity of managed interrupts is 2660 handled by the kernel and cannot be changed via 2661 the /proc/irq/* interfaces. 2662 2663 This isolation is best effort and only effective 2664 if the automatically assigned interrupt mask of a 2665 device queue contains isolated and housekeeping 2666 CPUs. If housekeeping CPUs are online then such 2667 interrupts are directed to the housekeeping CPU 2668 so that IO submitted on the housekeeping CPU 2669 cannot disturb the isolated CPU. 2670 2671 If a queue's affinity mask contains only isolated 2672 CPUs then this parameter has no effect on the 2673 interrupt routing decision, though interrupts are 2674 only delivered when tasks running on those 2675 isolated CPUs submit IO. IO submitted on 2676 housekeeping CPUs has no influence on those 2677 queues. 2678 2679 The format of <cpu-list> is described above. 2680 2681 iucv= [HW,NET] 2682 2683 ivrs_ioapic [HW,X86-64] 2684 Provide an override to the IOAPIC-ID<->DEVICE-ID 2685 mapping provided in the IVRS ACPI table. 2686 By default, PCI segment is 0, and can be omitted. 2687 2688 For example, to map IOAPIC-ID decimal 10 to 2689 PCI segment 0x1 and PCI device 00:14.0, 2690 write the parameter as: 2691 ivrs_ioapic=10@0001:00:14.0 2692 2693 Deprecated formats: 2694 * To map IOAPIC-ID decimal 10 to PCI device 00:14.0 2695 write the parameter as: 2696 ivrs_ioapic[10]=00:14.0 2697 * To map IOAPIC-ID decimal 10 to PCI segment 0x1 and 2698 PCI device 00:14.0 write the parameter as: 2699 ivrs_ioapic[10]=0001:00:14.0 2700 2701 ivrs_hpet [HW,X86-64] 2702 Provide an override to the HPET-ID<->DEVICE-ID 2703 mapping provided in the IVRS ACPI table. 2704 By default, PCI segment is 0, and can be omitted. 2705 2706 For example, to map HPET-ID decimal 10 to 2707 PCI segment 0x1 and PCI device 00:14.0, 2708 write the parameter as: 2709 ivrs_hpet=10@0001:00:14.0 2710 2711 Deprecated formats: 2712 * To map HPET-ID decimal 0 to PCI device 00:14.0 2713 write the parameter as: 2714 ivrs_hpet[0]=00:14.0 2715 * To map HPET-ID decimal 10 to PCI segment 0x1 and 2716 PCI device 00:14.0 write the parameter as: 2717 ivrs_ioapic[10]=0001:00:14.0 2718 2719 ivrs_acpihid [HW,X86-64] 2720 Provide an override to the ACPI-HID:UID<->DEVICE-ID 2721 mapping provided in the IVRS ACPI table. 2722 By default, PCI segment is 0, and can be omitted. 2723 2724 For example, to map UART-HID:UID AMD0020:0 to 2725 PCI segment 0x1 and PCI device ID 00:14.5, 2726 write the parameter as: 2727 ivrs_acpihid=AMD0020:0@0001:00:14.5 2728 2729 Deprecated formats: 2730 * To map UART-HID:UID AMD0020:0 to PCI segment is 0, 2731 PCI device ID 00:14.5, write the parameter as: 2732 ivrs_acpihid[00:14.5]=AMD0020:0 2733 * To map UART-HID:UID AMD0020:0 to PCI segment 0x1 and 2734 PCI device ID 00:14.5, write the parameter as: 2735 ivrs_acpihid[0001:00:14.5]=AMD0020:0 2736 2737 js= [HW,JOY] Analog joystick 2738 See Documentation/input/joydev/joystick.rst. 2739 2740 kasan_multi_shot 2741 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print 2742 report on every invalid memory access. Without this 2743 parameter KASAN will print report only for the first 2744 invalid access. 2745 2746 keep_bootcon [KNL,EARLY] 2747 Do not unregister boot console at start. This is only 2748 useful for debugging when something happens in the window 2749 between unregistering the boot console and initializing 2750 the real console. 2751 2752 keepinitrd [HW,ARM] See retain_initrd. 2753 2754 kernelcore= [KNL,X86,PPC,EARLY] 2755 Format: nn[KMGTPE] | nn% | "mirror" 2756 This parameter specifies the amount of memory usable by 2757 the kernel for non-movable allocations. The requested 2758 amount is spread evenly throughout all nodes in the 2759 system as ZONE_NORMAL. The remaining memory is used for 2760 movable memory in its own zone, ZONE_MOVABLE. In the 2761 event, a node is too small to have both ZONE_NORMAL and 2762 ZONE_MOVABLE, kernelcore memory will take priority and 2763 other nodes will have a larger ZONE_MOVABLE. 2764 2765 ZONE_MOVABLE is used for the allocation of pages that 2766 may be reclaimed or moved by the page migration 2767 subsystem. Note that allocations like PTEs-from-HighMem 2768 still use the HighMem zone if it exists, and the Normal 2769 zone if it does not. 2770 2771 It is possible to specify the exact amount of memory in 2772 the form of "nn[KMGTPE]", a percentage of total system 2773 memory in the form of "nn%", or "mirror". If "mirror" 2774 option is specified, mirrored (reliable) memory is used 2775 for non-movable allocations and remaining memory is used 2776 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror" 2777 are exclusive, so you cannot specify multiple forms. 2778 2779 kgdbdbgp= [KGDB,HW,EARLY] kgdb over EHCI usb debug port. 2780 Format: <Controller#>[,poll interval] 2781 The controller # is the number of the ehci usb debug 2782 port as it is probed via PCI. The poll interval is 2783 optional and is the number seconds in between 2784 each poll cycle to the debug port in case you need 2785 the functionality for interrupting the kernel with 2786 gdb or control-c on the dbgp connection. When 2787 not using this parameter you use sysrq-g to break into 2788 the kernel debugger. 2789 2790 kgdboc= [KGDB,HW] kgdb over consoles. 2791 Requires a tty driver that supports console polling, 2792 or a supported polling keyboard driver (non-usb). 2793 Serial only format: <serial_device>[,baud] 2794 keyboard only format: kbd 2795 keyboard and serial format: kbd,<serial_device>[,baud] 2796 Optional Kernel mode setting: 2797 kms, kbd format: kms,kbd 2798 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud] 2799 2800 kgdboc_earlycon= [KGDB,HW,EARLY] 2801 If the boot console provides the ability to read 2802 characters and can work in polling mode, you can use 2803 this parameter to tell kgdb to use it as a backend 2804 until the normal console is registered. Intended to 2805 be used together with the kgdboc parameter which 2806 specifies the normal console to transition to. 2807 2808 The name of the early console should be specified 2809 as the value of this parameter. Note that the name of 2810 the early console might be different than the tty 2811 name passed to kgdboc. It's OK to leave the value 2812 blank and the first boot console that implements 2813 read() will be picked. 2814 2815 kgdbwait [KGDB,EARLY] Stop kernel execution and enter the 2816 kernel debugger at the earliest opportunity. 2817 2818 kho= [KEXEC,EARLY] 2819 Format: { "0" | "1" | "off" | "on" | "y" | "n" } 2820 Enables or disables Kexec HandOver. 2821 "0" | "off" | "n" - kexec handover is disabled 2822 "1" | "on" | "y" - kexec handover is enabled 2823 2824 kho_scratch= [KEXEC,EARLY] 2825 Format: ll[KMG],mm[KMG],nn[KMG] | nn% 2826 Defines the size of the KHO scratch region. The KHO 2827 scratch regions are physically contiguous memory 2828 ranges that can only be used for non-kernel 2829 allocations. That way, even when memory is heavily 2830 fragmented with handed over memory, the kexeced 2831 kernel will always have enough contiguous ranges to 2832 bootstrap itself. 2833 2834 It is possible to specify the exact amount of 2835 memory in the form of "ll[KMG],mm[KMG],nn[KMG]" 2836 where the first parameter defines the size of a low 2837 memory scratch area, the second parameter defines 2838 the size of a global scratch area and the third 2839 parameter defines the size of additional per-node 2840 scratch areas. The form "nn%" defines scale factor 2841 (in percents) of memory that was used during boot. 2842 2843 kmac= [MIPS] Korina ethernet MAC address. 2844 Configure the RouterBoard 532 series on-chip 2845 Ethernet adapter MAC address. 2846 2847 kmemleak= [KNL,EARLY] Boot-time kmemleak enable/disable 2848 Valid arguments: on, off 2849 Default: on 2850 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y, 2851 the default is off. 2852 2853 kprobe_event=[probe-list] 2854 [FTRACE] Add kprobe events and enable at boot time. 2855 The probe-list is a semicolon delimited list of probe 2856 definitions. Each definition is same as kprobe_events 2857 interface, but the parameters are comma delimited. 2858 For example, to add a kprobe event on vfs_read with 2859 arg1 and arg2, add to the command line; 2860 2861 kprobe_event=p,vfs_read,$arg1,$arg2 2862 2863 See also Documentation/trace/kprobetrace.rst "Kernel 2864 Boot Parameter" section. 2865 2866 kpti= [ARM64,EARLY] Control page table isolation of 2867 user and kernel address spaces. 2868 Default: enabled on cores which need mitigation. 2869 0: force disabled 2870 1: force enabled 2871 2872 kunit.enable= [KUNIT] Enable executing KUnit tests. Requires 2873 CONFIG_KUNIT to be set to be fully enabled. The 2874 default value can be overridden via 2875 KUNIT_DEFAULT_ENABLED. 2876 Default is 1 (enabled) 2877 2878 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs. 2879 Default is 0 (don't ignore, but inject #GP) 2880 2881 kvm.eager_page_split= 2882 [KVM,X86] Controls whether or not KVM will try to 2883 proactively split all huge pages during dirty logging. 2884 Eager page splitting reduces interruptions to vCPU 2885 execution by eliminating the write-protection faults 2886 and MMU lock contention that would otherwise be 2887 required to split huge pages lazily. 2888 2889 VM workloads that rarely perform writes or that write 2890 only to a small region of VM memory may benefit from 2891 disabling eager page splitting to allow huge pages to 2892 still be used for reads. 2893 2894 The behavior of eager page splitting depends on whether 2895 KVM_DIRTY_LOG_INITIALLY_SET is enabled or disabled. If 2896 disabled, all huge pages in a memslot will be eagerly 2897 split when dirty logging is enabled on that memslot. If 2898 enabled, eager page splitting will be performed during 2899 the KVM_CLEAR_DIRTY ioctl, and only for the pages being 2900 cleared. 2901 2902 Eager page splitting is only supported when kvm.tdp_mmu=Y. 2903 2904 Default is Y (on). 2905 2906 kvm.enable_virt_at_load=[KVM,ARM64,LOONGARCH,MIPS,RISCV,X86] 2907 If enabled, KVM will enable virtualization in hardware 2908 when KVM is loaded, and disable virtualization when KVM 2909 is unloaded (if KVM is built as a module). 2910 2911 If disabled, KVM will dynamically enable and disable 2912 virtualization on-demand when creating and destroying 2913 VMs, i.e. on the 0=>1 and 1=>0 transitions of the 2914 number of VMs. 2915 2916 Enabling virtualization at module load avoids potential 2917 latency for creation of the 0=>1 VM, as KVM serializes 2918 virtualization enabling across all online CPUs. The 2919 "cost" of enabling virtualization when KVM is loaded, 2920 is that doing so may interfere with using out-of-tree 2921 hypervisors that want to "own" virtualization hardware. 2922 2923 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface. 2924 Default is false (don't support). 2925 2926 kvm.nx_huge_pages= 2927 [KVM] Controls the software workaround for the 2928 X86_BUG_ITLB_MULTIHIT bug. 2929 force : Always deploy workaround. 2930 off : Never deploy workaround. 2931 auto : Deploy workaround based on the presence of 2932 X86_BUG_ITLB_MULTIHIT. 2933 2934 Default is 'auto'. 2935 2936 If the software workaround is enabled for the host, 2937 guests do need not to enable it for nested guests. 2938 2939 kvm.nx_huge_pages_recovery_ratio= 2940 [KVM] Controls how many 4KiB pages are periodically zapped 2941 back to huge pages. 0 disables the recovery, otherwise if 2942 the value is N KVM will zap 1/Nth of the 4KiB pages every 2943 period (see below). The default is 60. 2944 2945 kvm.nx_huge_pages_recovery_period_ms= 2946 [KVM] Controls the time period at which KVM zaps 4KiB pages 2947 back to huge pages. If the value is a non-zero N, KVM will 2948 zap a portion (see ratio above) of the pages every N msecs. 2949 If the value is 0 (the default), KVM will pick a period based 2950 on the ratio, such that a page is zapped after 1 hour on average. 2951 2952 kvm-amd.nested= [KVM,AMD] Control nested virtualization feature in 2953 KVM/SVM. Default is 1 (enabled). 2954 2955 kvm-amd.npt= [KVM,AMD] Control KVM's use of Nested Page Tables, 2956 a.k.a. Two-Dimensional Page Tables. Default is 1 2957 (enabled). Disable by KVM if hardware lacks support 2958 for NPT. 2959 2960 kvm-arm.mode= 2961 [KVM,ARM,EARLY] Select one of KVM/arm64's modes of 2962 operation. 2963 2964 none: Forcefully disable KVM. 2965 2966 nvhe: Standard nVHE-based mode, without support for 2967 protected guests. 2968 2969 protected: Mode with support for guests whose state is 2970 kept private from the host, using VHE or 2971 nVHE depending on HW support. 2972 2973 nested: VHE-based mode with support for nested 2974 virtualization. Requires at least ARMv8.4 2975 hardware (with FEAT_NV2). 2976 2977 Defaults to VHE/nVHE based on hardware support. Setting 2978 mode to "protected" will disable kexec and hibernation 2979 for the host. To force nVHE on VHE hardware, add 2980 "arm64_sw.hvhe=0 id_aa64mmfr1.vh=0" to the 2981 command-line. 2982 "nested" is experimental and should be used with 2983 extreme caution. 2984 2985 kvm-arm.vgic_v3_group0_trap= 2986 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-0 2987 system registers 2988 2989 kvm-arm.vgic_v3_group1_trap= 2990 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-1 2991 system registers 2992 2993 kvm-arm.vgic_v3_common_trap= 2994 [KVM,ARM,EARLY] Trap guest accesses to GICv3 common 2995 system registers 2996 2997 kvm-arm.vgic_v4_enable= 2998 [KVM,ARM,EARLY] Allow use of GICv4 for direct 2999 injection of LPIs. 3000 3001 kvm-arm.wfe_trap_policy= 3002 [KVM,ARM] Control when to set WFE instruction trap for 3003 KVM VMs. Traps are allowed but not guaranteed by the 3004 CPU architecture. 3005 3006 trap: set WFE instruction trap 3007 3008 notrap: clear WFE instruction trap 3009 3010 kvm-arm.wfi_trap_policy= 3011 [KVM,ARM] Control when to set WFI instruction trap for 3012 KVM VMs. Traps are allowed but not guaranteed by the 3013 CPU architecture. 3014 3015 trap: set WFI instruction trap 3016 3017 notrap: clear WFI instruction trap 3018 3019 kvm_cma_resv_ratio=n [PPC,EARLY] 3020 Reserves given percentage from system memory area for 3021 contiguous memory allocation for KVM hash pagetable 3022 allocation. 3023 By default it reserves 5% of total system memory. 3024 Format: <integer> 3025 Default: 5 3026 3027 kvm-intel.ept= [KVM,Intel] Control KVM's use of Extended Page Tables, 3028 a.k.a. Two-Dimensional Page Tables. Default is 1 3029 (enabled). Disable by KVM if hardware lacks support 3030 for EPT. 3031 3032 kvm-intel.emulate_invalid_guest_state= 3033 [KVM,Intel] Control whether to emulate invalid guest 3034 state. Ignored if kvm-intel.enable_unrestricted_guest=1, 3035 as guest state is never invalid for unrestricted 3036 guests. This param doesn't apply to nested guests (L2), 3037 as KVM never emulates invalid L2 guest state. 3038 Default is 1 (enabled). 3039 3040 kvm-intel.flexpriority= 3041 [KVM,Intel] Control KVM's use of FlexPriority feature 3042 (TPR shadow). Default is 1 (enabled). Disable by KVM if 3043 hardware lacks support for it. 3044 3045 kvm-intel.nested= 3046 [KVM,Intel] Control nested virtualization feature in 3047 KVM/VMX. Default is 1 (enabled). 3048 3049 kvm-intel.unrestricted_guest= 3050 [KVM,Intel] Control KVM's use of unrestricted guest 3051 feature (virtualized real and unpaged mode). Default 3052 is 1 (enabled). Disable by KVM if EPT is disabled or 3053 hardware lacks support for it. 3054 3055 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault 3056 CVE-2018-3620. 3057 3058 Valid arguments: never, cond, always 3059 3060 always: L1D cache flush on every VMENTER. 3061 cond: Flush L1D on VMENTER only when the code between 3062 VMEXIT and VMENTER can leak host memory. 3063 never: Disables the mitigation 3064 3065 Default is cond (do L1 cache flush in specific instances) 3066 3067 kvm-intel.vpid= [KVM,Intel] Control KVM's use of Virtual Processor 3068 Identification feature (tagged TLBs). Default is 1 3069 (enabled). Disable by KVM if hardware lacks support 3070 for it. 3071 3072 l1d_flush= [X86,INTEL,EARLY] 3073 Control mitigation for L1D based snooping vulnerability. 3074 3075 Certain CPUs are vulnerable to an exploit against CPU 3076 internal buffers which can forward information to a 3077 disclosure gadget under certain conditions. 3078 3079 In vulnerable processors, the speculatively 3080 forwarded data can be used in a cache side channel 3081 attack, to access data to which the attacker does 3082 not have direct access. 3083 3084 This parameter controls the mitigation. The 3085 options are: 3086 3087 on - enable the interface for the mitigation 3088 3089 l1tf= [X86,EARLY] Control mitigation of the L1TF vulnerability on 3090 affected CPUs 3091 3092 The kernel PTE inversion protection is unconditionally 3093 enabled and cannot be disabled. 3094 3095 full 3096 Provides all available mitigations for the 3097 L1TF vulnerability. Disables SMT and 3098 enables all mitigations in the 3099 hypervisors, i.e. unconditional L1D flush. 3100 3101 SMT control and L1D flush control via the 3102 sysfs interface is still possible after 3103 boot. Hypervisors will issue a warning 3104 when the first VM is started in a 3105 potentially insecure configuration, 3106 i.e. SMT enabled or L1D flush disabled. 3107 3108 full,force 3109 Same as 'full', but disables SMT and L1D 3110 flush runtime control. Implies the 3111 'nosmt=force' command line option. 3112 (i.e. sysfs control of SMT is disabled.) 3113 3114 flush 3115 Leaves SMT enabled and enables the default 3116 hypervisor mitigation, i.e. conditional 3117 L1D flush. 3118 3119 SMT control and L1D flush control via the 3120 sysfs interface is still possible after 3121 boot. Hypervisors will issue a warning 3122 when the first VM is started in a 3123 potentially insecure configuration, 3124 i.e. SMT enabled or L1D flush disabled. 3125 3126 flush,nosmt 3127 3128 Disables SMT and enables the default 3129 hypervisor mitigation. 3130 3131 SMT control and L1D flush control via the 3132 sysfs interface is still possible after 3133 boot. Hypervisors will issue a warning 3134 when the first VM is started in a 3135 potentially insecure configuration, 3136 i.e. SMT enabled or L1D flush disabled. 3137 3138 flush,nowarn 3139 Same as 'flush', but hypervisors will not 3140 warn when a VM is started in a potentially 3141 insecure configuration. 3142 3143 off 3144 Disables hypervisor mitigations and doesn't 3145 emit any warnings. 3146 It also drops the swap size and available 3147 RAM limit restriction on both hypervisor and 3148 bare metal. 3149 3150 Default is 'flush'. 3151 3152 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst 3153 3154 l2cr= [PPC] 3155 3156 l3cr= [PPC] 3157 3158 lapic [X86-32,APIC,EARLY] Enable the local APIC even if BIOS 3159 disabled it. 3160 3161 lapic= [X86,APIC] Do not use TSC deadline 3162 value for LAPIC timer one-shot implementation. Default 3163 back to the programmable timer unit in the LAPIC. 3164 Format: notscdeadline 3165 3166 lapic_timer_c2_ok [X86,APIC,EARLY] trust the local apic timer 3167 in C2 power state. 3168 3169 libata.dma= [LIBATA] DMA control 3170 libata.dma=0 Disable all PATA and SATA DMA 3171 libata.dma=1 PATA and SATA Disk DMA only 3172 libata.dma=2 ATAPI (CDROM) DMA only 3173 libata.dma=4 Compact Flash DMA only 3174 Combinations also work, so libata.dma=3 enables DMA 3175 for disks and CDROMs, but not CFs. 3176 3177 libata.ignore_hpa= [LIBATA] Ignore HPA limit 3178 libata.ignore_hpa=0 keep BIOS limits (default) 3179 libata.ignore_hpa=1 ignore limits, using full disk 3180 3181 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume 3182 when set. 3183 Format: <int> 3184 3185 libata.force= [LIBATA] Force configurations. The format is a comma- 3186 separated list of "[ID:]VAL" where ID is PORT[.DEVICE]. 3187 PORT and DEVICE are decimal numbers matching port, link 3188 or device. Basically, it matches the ATA ID string 3189 printed on console by libata. If the whole ID part is 3190 omitted, the last PORT and DEVICE values are used. If 3191 ID hasn't been specified yet, the configuration applies 3192 to all ports, links and devices. 3193 3194 If only DEVICE is omitted, the parameter applies to 3195 the port and all links and devices behind it. DEVICE 3196 number of 0 either selects the first device or the 3197 first fan-out link behind PMP device. It does not 3198 select the host link. DEVICE number of 15 selects the 3199 host link and device attached to it. 3200 3201 The VAL specifies the configuration to force. As long 3202 as there is no ambiguity, shortcut notation is allowed. 3203 For example, both 1.5 and 1.5G would work for 1.5Gbps. 3204 The following configurations can be forced. 3205 3206 * Cable type: 40c, 80c, short40c, unk, ign or sata. 3207 Any ID with matching PORT is used. 3208 3209 * SATA link speed limit: 1.5Gbps or 3.0Gbps. 3210 3211 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7]. 3212 udma[/][16,25,33,44,66,100,133] notation is also 3213 allowed. 3214 3215 * nohrst, nosrst, norst: suppress hard, soft and both 3216 resets. 3217 3218 * rstonce: only attempt one reset during hot-unplug 3219 link recovery. 3220 3221 * [no]dbdelay: Enable or disable the extra 200ms delay 3222 before debouncing a link PHY and device presence 3223 detection. 3224 3225 * [no]ncq: Turn on or off NCQ. 3226 3227 * [no]ncqtrim: Enable or disable queued DSM TRIM. 3228 3229 * [no]ncqati: Enable or disable NCQ trim on ATI chipset. 3230 3231 * [no]trim: Enable or disable (unqueued) TRIM. 3232 3233 * trim_zero: Indicate that TRIM command zeroes data. 3234 3235 * max_trim_128m: Set 128M maximum trim size limit. 3236 3237 * [no]dma: Turn on or off DMA transfers. 3238 3239 * atapi_dmadir: Enable ATAPI DMADIR bridge support. 3240 3241 * atapi_mod16_dma: Enable the use of ATAPI DMA for 3242 commands that are not a multiple of 16 bytes. 3243 3244 * [no]dmalog: Enable or disable the use of the 3245 READ LOG DMA EXT command to access logs. 3246 3247 * [no]iddevlog: Enable or disable access to the 3248 identify device data log. 3249 3250 * [no]logdir: Enable or disable access to the general 3251 purpose log directory. 3252 3253 * max_sec_128: Set transfer size limit to 128 sectors. 3254 3255 * max_sec_1024: Set or clear transfer size limit to 3256 1024 sectors. 3257 3258 * max_sec_lba48: Set or clear transfer size limit to 3259 65535 sectors. 3260 3261 * external: Mark port as external (hotplug-capable). 3262 3263 * [no]lpm: Enable or disable link power management. 3264 3265 * [no]setxfer: Indicate if transfer speed mode setting 3266 should be skipped. 3267 3268 * [no]fua: Disable or enable FUA (Force Unit Access) 3269 support for devices supporting this feature. 3270 3271 * dump_id: Dump IDENTIFY data. 3272 3273 * disable: Disable this device. 3274 3275 If there are multiple matching configurations changing 3276 the same attribute, the last one is used. 3277 3278 load_ramdisk= [RAM] [Deprecated] 3279 3280 lockd.nlm_grace_period=P [NFS] Assign grace period. 3281 Format: <integer> 3282 3283 lockd.nlm_tcpport=N [NFS] Assign TCP port. 3284 Format: <integer> 3285 3286 lockd.nlm_timeout=T [NFS] Assign timeout value. 3287 Format: <integer> 3288 3289 lockd.nlm_udpport=M [NFS] Assign UDP port. 3290 Format: <integer> 3291 3292 lockdown= [SECURITY,EARLY] 3293 { integrity | confidentiality } 3294 Enable the kernel lockdown feature. If set to 3295 integrity, kernel features that allow userland to 3296 modify the running kernel are disabled. If set to 3297 confidentiality, kernel features that allow userland 3298 to extract confidential information from the kernel 3299 are also disabled. 3300 3301 locktorture.acq_writer_lim= [KNL] 3302 Set the time limit in jiffies for a lock 3303 acquisition. Acquisitions exceeding this limit 3304 will result in a splat once they do complete. 3305 3306 locktorture.bind_readers= [KNL] 3307 Specify the list of CPUs to which the readers are 3308 to be bound. 3309 3310 locktorture.bind_writers= [KNL] 3311 Specify the list of CPUs to which the writers are 3312 to be bound. 3313 3314 locktorture.call_rcu_chains= [KNL] 3315 Specify the number of self-propagating call_rcu() 3316 chains to set up. These are used to ensure that 3317 there is a high probability of an RCU grace period 3318 in progress at any given time. Defaults to 0, 3319 which disables these call_rcu() chains. 3320 3321 locktorture.long_hold= [KNL] 3322 Specify the duration in milliseconds for the 3323 occasional long-duration lock hold time. Defaults 3324 to 100 milliseconds. Select 0 to disable. 3325 3326 locktorture.nested_locks= [KNL] 3327 Specify the maximum lock nesting depth that 3328 locktorture is to exercise, up to a limit of 8 3329 (MAX_NESTED_LOCKS). Specify zero to disable. 3330 Note that this parameter is ineffective on types 3331 of locks that do not support nested acquisition. 3332 3333 locktorture.nreaders_stress= [KNL] 3334 Set the number of locking read-acquisition kthreads. 3335 Defaults to being automatically set based on the 3336 number of online CPUs. 3337 3338 locktorture.nwriters_stress= [KNL] 3339 Set the number of locking write-acquisition kthreads. 3340 3341 locktorture.onoff_holdoff= [KNL] 3342 Set time (s) after boot for CPU-hotplug testing. 3343 3344 locktorture.onoff_interval= [KNL] 3345 Set time (s) between CPU-hotplug operations, or 3346 zero to disable CPU-hotplug testing. 3347 3348 locktorture.rt_boost= [KNL] 3349 Do periodic testing of real-time lock priority 3350 boosting. Select 0 to disable, 1 to boost 3351 only rt_mutex, and 2 to boost unconditionally. 3352 Defaults to 2, which might seem to be an 3353 odd choice, but which should be harmless for 3354 non-real-time spinlocks, due to their disabling 3355 of preemption. Note that non-realtime mutexes 3356 disable boosting. 3357 3358 locktorture.rt_boost_factor= [KNL] 3359 Number that determines how often and for how 3360 long priority boosting is exercised. This is 3361 scaled down by the number of writers, so that the 3362 number of boosts per unit time remains roughly 3363 constant as the number of writers increases. 3364 On the other hand, the duration of each boost 3365 increases with the number of writers. 3366 3367 locktorture.shuffle_interval= [KNL] 3368 Set task-shuffle interval (jiffies). Shuffling 3369 tasks allows some CPUs to go into dyntick-idle 3370 mode during the locktorture test. 3371 3372 locktorture.shutdown_secs= [KNL] 3373 Set time (s) after boot system shutdown. This 3374 is useful for hands-off automated testing. 3375 3376 locktorture.stat_interval= [KNL] 3377 Time (s) between statistics printk()s. 3378 3379 locktorture.stutter= [KNL] 3380 Time (s) to stutter testing, for example, 3381 specifying five seconds causes the test to run for 3382 five seconds, wait for five seconds, and so on. 3383 This tests the locking primitive's ability to 3384 transition abruptly to and from idle. 3385 3386 locktorture.torture_type= [KNL] 3387 Specify the locking implementation to test. 3388 3389 locktorture.verbose= [KNL] 3390 Enable additional printk() statements. 3391 3392 locktorture.writer_fifo= [KNL] 3393 Run the write-side locktorture kthreads at 3394 sched_set_fifo() real-time priority. 3395 3396 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver 3397 Format: <irq> 3398 3399 loglevel= [KNL,EARLY] 3400 All Kernel Messages with a loglevel smaller than the 3401 console loglevel will be printed to the console. It can 3402 also be changed with klogd or other programs. The 3403 loglevels are defined as follows: 3404 3405 0 (KERN_EMERG) system is unusable 3406 1 (KERN_ALERT) action must be taken immediately 3407 2 (KERN_CRIT) critical conditions 3408 3 (KERN_ERR) error conditions 3409 4 (KERN_WARNING) warning conditions 3410 5 (KERN_NOTICE) normal but significant condition 3411 6 (KERN_INFO) informational 3412 7 (KERN_DEBUG) debug-level messages 3413 3414 log_buf_len=n[KMG] [KNL,EARLY] 3415 Sets the size of the printk ring buffer, in bytes. 3416 n must be a power of two and greater than the 3417 minimal size. The minimal size is defined by 3418 LOG_BUF_SHIFT kernel config parameter. There 3419 is also CONFIG_LOG_CPU_MAX_BUF_SHIFT config 3420 parameter that allows to increase the default size 3421 depending on the number of CPUs. See init/Kconfig 3422 for more details. 3423 3424 logo.nologo [FB] Disables display of the built-in Linux logo. 3425 This may be used to provide more screen space for 3426 kernel log messages and is useful when debugging 3427 kernel boot problems. 3428 3429 lp=0 [LP] Specify parallel ports to use, e.g, 3430 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses 3431 lp=reset first parallel port). 'lp=0' disables the 3432 lp=auto printer driver. 'lp=reset' (which can be 3433 specified in addition to the ports) causes 3434 attached printers to be reset. Using 3435 lp=port1,port2,... specifies the parallel ports 3436 to associate lp devices with, starting with 3437 lp0. A port specification may be 'none' to skip 3438 that lp device, or a parport name such as 3439 'parport0'. Specifying 'lp=auto' instead of a 3440 port specification list means that device IDs 3441 from each port should be examined, to see if 3442 an IEEE 1284-compliant printer is attached; if 3443 so, the driver will manage that printer. 3444 See also header of drivers/char/lp.c. 3445 3446 lpj=n [KNL] 3447 Sets loops_per_jiffy to given constant, thus avoiding 3448 time-consuming boot-time autodetection (up to 250 ms per 3449 CPU). 0 enables autodetection (default). To determine 3450 the correct value for your kernel, boot with normal 3451 autodetection and see what value is printed. Note that 3452 on SMP systems the preset will be applied to all CPUs, 3453 which is likely to cause problems if your CPUs need 3454 significantly divergent settings. An incorrect value 3455 will cause delays in the kernel to be wrong, leading to 3456 unpredictable I/O errors and other breakage. Although 3457 unlikely, in the extreme case this might damage your 3458 hardware. 3459 3460 lsm.debug [SECURITY] Enable LSM initialization debugging output. 3461 3462 lsm=lsm1,...,lsmN 3463 [SECURITY] Choose order of LSM initialization. This 3464 overrides CONFIG_LSM, and the "security=" parameter. 3465 3466 machtype= [Loongson] Share the same kernel image file between 3467 different yeeloong laptops. 3468 Example: machtype=lemote-yeeloong-2f-7inch 3469 3470 maxcpus= [SMP,EARLY] Maximum number of processors that an SMP kernel 3471 will bring up during bootup. maxcpus=n : n >= 0 limits 3472 the kernel to bring up 'n' processors. Surely after 3473 bootup you can bring up the other plugged cpu by executing 3474 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus 3475 only takes effect during system bootup. 3476 While n=0 is a special case, it is equivalent to "nosmp", 3477 which also disables the IO APIC. 3478 3479 max_loop= [LOOP] The number of loop block devices that get 3480 (loop.max_loop) unconditionally pre-created at init time. The default 3481 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead 3482 of statically allocating a predefined number, loop 3483 devices can be requested on-demand with the 3484 /dev/loop-control interface. 3485 3486 mce= [X86-{32,64}] 3487 3488 Please see Documentation/arch/x86/x86_64/machinecheck.rst for sysfs runtime tunables. 3489 3490 off 3491 disable machine check 3492 3493 no_cmci 3494 disable CMCI(Corrected Machine Check Interrupt) that 3495 Intel processor supports. Usually this disablement is 3496 not recommended, but it might be handy if your 3497 hardware is misbehaving. 3498 3499 Note that you'll get more problems without CMCI than 3500 with due to the shared banks, i.e. you might get 3501 duplicated error logs. 3502 3503 dont_log_ce 3504 don't make logs for corrected errors. All events 3505 reported as corrected are silently cleared by OS. This 3506 option will be useful if you have no interest in any 3507 of corrected errors. 3508 3509 ignore_ce 3510 disable features for corrected errors, e.g. 3511 polling timer and CMCI. All events reported as 3512 corrected are not cleared by OS and remained in its 3513 error banks. 3514 3515 Usually this disablement is not recommended, however 3516 if there is an agent checking/clearing corrected 3517 errors (e.g. BIOS or hardware monitoring 3518 applications), conflicting with OS's error handling, 3519 and you cannot deactivate the agent, then this option 3520 will be a help. 3521 3522 no_lmce 3523 do not opt-in to Local MCE delivery. Use legacy method 3524 to broadcast MCEs. 3525 3526 bootlog 3527 enable logging of machine checks left over from 3528 booting. Disabled by default on AMD Fam10h and older 3529 because some BIOS leave bogus ones. 3530 3531 If your BIOS doesn't do that it's a good idea to 3532 enable though to make sure you log even machine check 3533 events that result in a reboot. On Intel systems it is 3534 enabled by default. 3535 3536 nobootlog 3537 disable boot machine check logging. 3538 3539 monarchtimeout (number) 3540 sets the time in us to wait for other CPUs on machine 3541 checks. 0 to disable. 3542 3543 bios_cmci_threshold 3544 don't overwrite the bios-set CMCI threshold. This boot 3545 option prevents Linux from overwriting the CMCI 3546 threshold set by the bios. Without this option, Linux 3547 always sets the CMCI threshold to 1. Enabling this may 3548 make memory predictive failure analysis less effective 3549 if the bios sets thresholds for memory errors since we 3550 will not see details for all errors. 3551 3552 recovery 3553 force-enable recoverable machine check code paths 3554 3555 Everything else is in sysfs now. 3556 3557 3558 md= [HW] RAID subsystems devices and level 3559 See Documentation/admin-guide/md.rst. 3560 3561 mdacon= [MDA] 3562 Format: <first>,<last> 3563 Specifies range of consoles to be captured by the MDA. 3564 3565 mds= [X86,INTEL,EARLY] 3566 Control mitigation for the Micro-architectural Data 3567 Sampling (MDS) vulnerability. 3568 3569 Certain CPUs are vulnerable to an exploit against CPU 3570 internal buffers which can forward information to a 3571 disclosure gadget under certain conditions. 3572 3573 In vulnerable processors, the speculatively 3574 forwarded data can be used in a cache side channel 3575 attack, to access data to which the attacker does 3576 not have direct access. 3577 3578 This parameter controls the MDS mitigation. The 3579 options are: 3580 3581 full - Enable MDS mitigation on vulnerable CPUs 3582 full,nosmt - Enable MDS mitigation and disable 3583 SMT on vulnerable CPUs 3584 off - Unconditionally disable MDS mitigation 3585 3586 On TAA-affected machines, mds=off can be prevented by 3587 an active TAA mitigation as both vulnerabilities are 3588 mitigated with the same mechanism so in order to disable 3589 this mitigation, you need to specify tsx_async_abort=off 3590 too. 3591 3592 Not specifying this option is equivalent to 3593 mds=full. 3594 3595 For details see: Documentation/admin-guide/hw-vuln/mds.rst 3596 3597 mem=nn[KMG] [HEXAGON,EARLY] Set the memory size. 3598 Must be specified, otherwise memory size will be 0. 3599 3600 mem=nn[KMG] [KNL,BOOT,EARLY] Force usage of a specific amount 3601 of memory Amount of memory to be used in cases 3602 as follows: 3603 3604 1 for test; 3605 2 when the kernel is not able to see the whole system memory; 3606 3 memory that lies after 'mem=' boundary is excluded from 3607 the hypervisor, then assigned to KVM guests. 3608 4 to limit the memory available for kdump kernel. 3609 3610 [ARC,MICROBLAZE] - the limit applies only to low memory, 3611 high memory is not affected. 3612 3613 [ARM64] - only limits memory covered by the linear 3614 mapping. The NOMAP regions are not affected. 3615 3616 [X86] Work as limiting max address. Use together 3617 with memmap= to avoid physical address space collisions. 3618 Without memmap= PCI devices could be placed at addresses 3619 belonging to unused RAM. 3620 3621 Note that this only takes effects during boot time since 3622 in above case 3, memory may need be hot added after boot 3623 if system memory of hypervisor is not sufficient. 3624 3625 mem=nn[KMG]@ss[KMG] 3626 [ARM,MIPS,EARLY] - override the memory layout 3627 reported by firmware. 3628 Define a memory region of size nn[KMG] starting at 3629 ss[KMG]. 3630 Multiple different regions can be specified with 3631 multiple mem= parameters on the command line. 3632 3633 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel 3634 memory. 3635 3636 memblock=debug [KNL,EARLY] Enable memblock debug messages. 3637 3638 memchunk=nn[KMG] 3639 [KNL,SH] Allow user to override the default size for 3640 per-device physically contiguous DMA buffers. 3641 3642 memhp_default_state=online/offline/online_kernel/online_movable 3643 [KNL] Set the initial state for the memory hotplug 3644 onlining policy. If not specified, the default value is 3645 set according to the 3646 CONFIG_MHP_DEFAULT_ONLINE_TYPE kernel config 3647 options. 3648 See Documentation/admin-guide/mm/memory-hotplug.rst. 3649 3650 memmap=exactmap [KNL,X86,EARLY] Enable setting of an exact 3651 E820 memory map, as specified by the user. 3652 Such memmap=exactmap lines can be constructed based on 3653 BIOS output or other requirements. See the memmap=nn@ss 3654 option description. 3655 3656 memmap=nn[KMG]@ss[KMG] 3657 [KNL, X86,MIPS,XTENSA,EARLY] Force usage of a specific region of memory. 3658 Region of memory to be used is from ss to ss+nn. 3659 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG], 3660 which limits max address to nn[KMG]. 3661 Multiple different regions can be specified, 3662 comma delimited. 3663 Example: 3664 memmap=100M@2G,100M#3G,1G!1024G 3665 3666 memmap=nn[KMG]#ss[KMG] 3667 [KNL,ACPI,EARLY] Mark specific memory as ACPI data. 3668 Region of memory to be marked is from ss to ss+nn. 3669 3670 memmap=nn[KMG]$ss[KMG] 3671 [KNL,ACPI,EARLY] Mark specific memory as reserved. 3672 Region of memory to be reserved is from ss to ss+nn. 3673 Example: Exclude memory from 0x18690000-0x1869ffff 3674 memmap=64K$0x18690000 3675 or 3676 memmap=0x10000$0x18690000 3677 Some bootloaders may need an escape character before '$', 3678 like Grub2, otherwise '$' and the following number 3679 will be eaten. 3680 3681 memmap=nn[KMG]!ss[KMG,EARLY] 3682 [KNL,X86] Mark specific memory as protected. 3683 Region of memory to be used, from ss to ss+nn. 3684 The memory region may be marked as e820 type 12 (0xc) 3685 and is NVDIMM or ADR memory. 3686 3687 memmap=<size>%<offset>-<oldtype>+<newtype> 3688 [KNL,ACPI,EARLY] Convert memory within the specified region 3689 from <oldtype> to <newtype>. If "-<oldtype>" is left 3690 out, the whole region will be marked as <newtype>, 3691 even if previously unavailable. If "+<newtype>" is left 3692 out, matching memory will be removed. Types are 3693 specified as e820 types, e.g., 1 = RAM, 2 = reserved, 3694 3 = ACPI, 12 = PRAM. 3695 3696 memory_corruption_check=0/1 [X86,EARLY] 3697 Some BIOSes seem to corrupt the first 64k of 3698 memory when doing things like suspend/resume. 3699 Setting this option will scan the memory 3700 looking for corruption. Enabling this will 3701 both detect corruption and prevent the kernel 3702 from using the memory being corrupted. 3703 However, its intended as a diagnostic tool; if 3704 repeatable BIOS-originated corruption always 3705 affects the same memory, you can use memmap= 3706 to prevent the kernel from using that memory. 3707 3708 memory_corruption_check_size=size [X86,EARLY] 3709 By default it checks for corruption in the low 3710 64k, making this memory unavailable for normal 3711 use. Use this parameter to scan for 3712 corruption in more or less memory. 3713 3714 memory_corruption_check_period=seconds [X86,EARLY] 3715 By default it checks for corruption every 60 3716 seconds. Use this parameter to check at some 3717 other rate. 0 disables periodic checking. 3718 3719 memory_hotplug.memmap_on_memory 3720 [KNL,X86,ARM] Boolean flag to enable this feature. 3721 Format: {on | off (default)} 3722 When enabled, runtime hotplugged memory will 3723 allocate its internal metadata (struct pages, 3724 those vmemmap pages cannot be optimized even 3725 if hugetlb_free_vmemmap is enabled) from the 3726 hotadded memory which will allow to hotadd a 3727 lot of memory without requiring additional 3728 memory to do so. 3729 This feature is disabled by default because it 3730 has some implication on large (e.g. GB) 3731 allocations in some configurations (e.g. small 3732 memory blocks). 3733 The state of the flag can be read in 3734 /sys/module/memory_hotplug/parameters/memmap_on_memory. 3735 Note that even when enabled, there are a few cases where 3736 the feature is not effective. 3737 3738 memtest= [KNL,X86,ARM,M68K,PPC,RISCV,EARLY] Enable memtest 3739 Format: <integer> 3740 default : 0 <disable> 3741 Specifies the number of memtest passes to be 3742 performed. Each pass selects another test 3743 pattern from a given set of patterns. Memtest 3744 fills the memory with this pattern, validates 3745 memory contents and reserves bad memory 3746 regions that are detected. 3747 3748 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control 3749 Valid arguments: on, off 3750 Default: off 3751 mem_encrypt=on: Activate SME 3752 mem_encrypt=off: Do not activate SME 3753 3754 Refer to Documentation/virt/kvm/x86/amd-memory-encryption.rst 3755 for details on when memory encryption can be activated. 3756 3757 mem_sleep_default= [SUSPEND] Default system suspend mode: 3758 s2idle - Suspend-To-Idle 3759 shallow - Power-On Suspend or equivalent (if supported) 3760 deep - Suspend-To-RAM or equivalent (if supported) 3761 See Documentation/admin-guide/pm/sleep-states.rst. 3762 3763 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when 3764 the BIOS has incorrectly applied a workaround. TinyBIOS 3765 version 0.98 is known to be affected, 0.99 fixes the 3766 problem by letting the user disable the workaround. 3767 3768 mga= [HW,DRM] 3769 3770 microcode.force_minrev= [X86] 3771 Format: <bool> 3772 Enable or disable the microcode minimal revision 3773 enforcement for the runtime microcode loader. 3774 3775 mini2440= [ARM,HW,KNL] 3776 Format:[0..2][b][c][t] 3777 Default: "0tb" 3778 MINI2440 configuration specification: 3779 0 - The attached screen is the 3.5" TFT 3780 1 - The attached screen is the 7" TFT 3781 2 - The VGA Shield is attached (1024x768) 3782 Leaving out the screen size parameter will not load 3783 the TFT driver, and the framebuffer will be left 3784 unconfigured. 3785 b - Enable backlight. The TFT backlight pin will be 3786 linked to the kernel VESA blanking code and a GPIO 3787 LED. This parameter is not necessary when using the 3788 VGA shield. 3789 c - Enable the s3c camera interface. 3790 t - Reserved for enabling touchscreen support. The 3791 touchscreen support is not enabled in the mainstream 3792 kernel as of 2.6.30, a preliminary port can be found 3793 in the "bleeding edge" mini2440 support kernel at 3794 https://repo.or.cz/w/linux-2.6/mini2440.git 3795 3796 mitigations= 3797 [X86,PPC,S390,ARM64,EARLY] Control optional mitigations for 3798 CPU vulnerabilities. This is a set of curated, 3799 arch-independent options, each of which is an 3800 aggregation of existing arch-specific options. 3801 3802 Note, "mitigations" is supported if and only if the 3803 kernel was built with CPU_MITIGATIONS=y. 3804 3805 off 3806 Disable all optional CPU mitigations. This 3807 improves system performance, but it may also 3808 expose users to several CPU vulnerabilities. 3809 Equivalent to: if nokaslr then kpti=0 [ARM64] 3810 gather_data_sampling=off [X86] 3811 indirect_target_selection=off [X86] 3812 kvm.nx_huge_pages=off [X86] 3813 l1tf=off [X86] 3814 mds=off [X86] 3815 mmio_stale_data=off [X86] 3816 no_entry_flush [PPC] 3817 no_uaccess_flush [PPC] 3818 nobp=0 [S390] 3819 nopti [X86,PPC] 3820 nospectre_bhb [ARM64] 3821 nospectre_v1 [X86,PPC] 3822 nospectre_v2 [X86,PPC,S390,ARM64] 3823 reg_file_data_sampling=off [X86] 3824 retbleed=off [X86] 3825 spec_rstack_overflow=off [X86] 3826 spec_store_bypass_disable=off [X86,PPC] 3827 spectre_bhi=off [X86] 3828 spectre_v2_user=off [X86] 3829 srbds=off [X86,INTEL] 3830 ssbd=force-off [ARM64] 3831 tsx_async_abort=off [X86] 3832 3833 Exceptions: 3834 This does not have any effect on 3835 kvm.nx_huge_pages when 3836 kvm.nx_huge_pages=force. 3837 3838 auto (default) 3839 Mitigate all CPU vulnerabilities, but leave SMT 3840 enabled, even if it's vulnerable. This is for 3841 users who don't want to be surprised by SMT 3842 getting disabled across kernel upgrades, or who 3843 have other ways of avoiding SMT-based attacks. 3844 Equivalent to: (default behavior) 3845 3846 auto,nosmt 3847 Mitigate all CPU vulnerabilities, disabling SMT 3848 if needed. This is for users who always want to 3849 be fully mitigated, even if it means losing SMT. 3850 Equivalent to: l1tf=flush,nosmt [X86] 3851 mds=full,nosmt [X86] 3852 tsx_async_abort=full,nosmt [X86] 3853 mmio_stale_data=full,nosmt [X86] 3854 retbleed=auto,nosmt [X86] 3855 3856 [X86] After one of the above options, additionally 3857 supports attack-vector based controls as documented in 3858 Documentation/admin-guide/hw-vuln/attack_vector_controls.rst 3859 3860 mminit_loglevel= 3861 [KNL,EARLY] When CONFIG_DEBUG_MEMORY_INIT is set, this 3862 parameter allows control of the logging verbosity for 3863 the additional memory initialisation checks. A value 3864 of 0 disables mminit logging and a level of 4 will 3865 log everything. Information is printed at KERN_DEBUG 3866 so loglevel=8 may also need to be specified. 3867 3868 mmio_stale_data= 3869 [X86,INTEL,EARLY] Control mitigation for the Processor 3870 MMIO Stale Data vulnerabilities. 3871 3872 Processor MMIO Stale Data is a class of 3873 vulnerabilities that may expose data after an MMIO 3874 operation. Exposed data could originate or end in 3875 the same CPU buffers as affected by MDS and TAA. 3876 Therefore, similar to MDS and TAA, the mitigation 3877 is to clear the affected CPU buffers. 3878 3879 This parameter controls the mitigation. The 3880 options are: 3881 3882 full - Enable mitigation on vulnerable CPUs 3883 3884 full,nosmt - Enable mitigation and disable SMT on 3885 vulnerable CPUs. 3886 3887 off - Unconditionally disable mitigation 3888 3889 On MDS or TAA affected machines, 3890 mmio_stale_data=off can be prevented by an active 3891 MDS or TAA mitigation as these vulnerabilities are 3892 mitigated with the same mechanism so in order to 3893 disable this mitigation, you need to specify 3894 mds=off and tsx_async_abort=off too. 3895 3896 Not specifying this option is equivalent to 3897 mmio_stale_data=full. 3898 3899 For details see: 3900 Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst 3901 3902 <module>.async_probe[=<bool>] [KNL] 3903 If no <bool> value is specified or if the value 3904 specified is not a valid <bool>, enable asynchronous 3905 probe on this module. Otherwise, enable/disable 3906 asynchronous probe on this module as indicated by the 3907 <bool> value. See also: module.async_probe 3908 3909 module.async_probe=<bool> 3910 [KNL] When set to true, modules will use async probing 3911 by default. To enable/disable async probing for a 3912 specific module, use the module specific control that 3913 is documented under <module>.async_probe. When both 3914 module.async_probe and <module>.async_probe are 3915 specified, <module>.async_probe takes precedence for 3916 the specific module. 3917 3918 module.enable_dups_trace 3919 [KNL] When CONFIG_MODULE_DEBUG_AUTOLOAD_DUPS is set, 3920 this means that duplicate request_module() calls will 3921 trigger a WARN_ON() instead of a pr_warn(). Note that 3922 if MODULE_DEBUG_AUTOLOAD_DUPS_TRACE is set, WARN_ON()s 3923 will always be issued and this option does nothing. 3924 module.sig_enforce 3925 [KNL] When CONFIG_MODULE_SIG is set, this means that 3926 modules without (valid) signatures will fail to load. 3927 Note that if CONFIG_MODULE_SIG_FORCE is set, that 3928 is always true, so this option does nothing. 3929 3930 module_blacklist= [KNL] Do not load a comma-separated list of 3931 modules. Useful for debugging problem modules. 3932 3933 mousedev.tap_time= 3934 [MOUSE] Maximum time between finger touching and 3935 leaving touchpad surface for touch to be considered 3936 a tap and be reported as a left button click (for 3937 touchpads working in absolute mode only). 3938 Format: <msecs> 3939 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices 3940 reporting absolute coordinates, such as tablets 3941 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices 3942 reporting absolute coordinates, such as tablets 3943 3944 movablecore= [KNL,X86,PPC,EARLY] 3945 Format: nn[KMGTPE] | nn% 3946 This parameter is the complement to kernelcore=, it 3947 specifies the amount of memory used for migratable 3948 allocations. If both kernelcore and movablecore is 3949 specified, then kernelcore will be at *least* the 3950 specified value but may be more. If movablecore on its 3951 own is specified, the administrator must be careful 3952 that the amount of memory usable for all allocations 3953 is not too small. 3954 3955 movable_node [KNL,EARLY] Boot-time switch to make hotplugable memory 3956 NUMA nodes to be movable. This means that the memory 3957 of such nodes will be usable only for movable 3958 allocations which rules out almost all kernel 3959 allocations. Use with caution! 3960 3961 MTD_Partition= [MTD] 3962 Format: <name>,<region-number>,<size>,<offset> 3963 3964 MTD_Region= [MTD] Format: 3965 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>] 3966 3967 mtdparts= [MTD] 3968 See drivers/mtd/parsers/cmdlinepart.c 3969 3970 mtouchusb.raw_coordinates= 3971 [HW] Make the MicroTouch USB driver use raw coordinates 3972 ('y', default) or cooked coordinates ('n') 3973 3974 mtrr=debug [X86,EARLY] 3975 Enable printing debug information related to MTRR 3976 registers at boot time. 3977 3978 mtrr_chunk_size=nn[KMG,X86,EARLY] 3979 used for mtrr cleanup. It is largest continuous chunk 3980 that could hold holes aka. UC entries. 3981 3982 mtrr_gran_size=nn[KMG,X86,EARLY] 3983 Used for mtrr cleanup. It is granularity of mtrr block. 3984 Default is 1. 3985 Large value could prevent small alignment from 3986 using up MTRRs. 3987 3988 mtrr_spare_reg_nr=n [X86,EARLY] 3989 Format: <integer> 3990 Range: 0,7 : spare reg number 3991 Default : 1 3992 Used for mtrr cleanup. It is spare mtrr entries number. 3993 Set to 2 or more if your graphical card needs more. 3994 3995 multitce=off [PPC] This parameter disables the use of the pSeries 3996 firmware feature for updating multiple TCE entries 3997 at a time. 3998 3999 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card 4000 4001 netdev= [NET] Network devices parameters 4002 Format: <irq>,<io>,<mem_start>,<mem_end>,<name> 4003 Note that mem_start is often overloaded to mean 4004 something different and driver-specific. 4005 This usage is only documented in each driver source 4006 file if at all. 4007 4008 netpoll.carrier_timeout= 4009 [NET] Specifies amount of time (in seconds) that 4010 netpoll should wait for a carrier. By default netpoll 4011 waits 4 seconds. 4012 4013 nf_conntrack.acct= 4014 [NETFILTER] Enable connection tracking flow accounting 4015 0 to disable accounting 4016 1 to enable accounting 4017 Default value is 0. 4018 4019 nfs.cache_getent= 4020 [NFS] sets the pathname to the program which is used 4021 to update the NFS client cache entries. 4022 4023 nfs.cache_getent_timeout= 4024 [NFS] sets the timeout after which an attempt to 4025 update a cache entry is deemed to have failed. 4026 4027 nfs.callback_nr_threads= 4028 [NFSv4] set the total number of threads that the 4029 NFS client will assign to service NFSv4 callback 4030 requests. 4031 4032 nfs.callback_tcpport= 4033 [NFS] set the TCP port on which the NFSv4 callback 4034 channel should listen. 4035 4036 nfs.delay_retrans= 4037 [NFS] specifies the number of times the NFSv4 client 4038 retries the request before returning an EAGAIN error, 4039 after a reply of NFS4ERR_DELAY from the server. 4040 Only applies if the softerr mount option is enabled, 4041 and the specified value is >= 0. 4042 4043 nfs.enable_ino64= 4044 [NFS] enable 64-bit inode numbers. 4045 If zero, the NFS client will fake up a 32-bit inode 4046 number for the readdir() and stat() syscalls instead 4047 of returning the full 64-bit number. 4048 The default is to return 64-bit inode numbers. 4049 4050 nfs.idmap_cache_timeout= 4051 [NFS] set the maximum lifetime for idmapper cache 4052 entries. 4053 4054 nfs.max_session_cb_slots= 4055 [NFSv4.1] Sets the maximum number of session 4056 slots the client will assign to the callback 4057 channel. This determines the maximum number of 4058 callbacks the client will process in parallel for 4059 a particular server. 4060 4061 nfs.max_session_slots= 4062 [NFSv4.1] Sets the maximum number of session slots 4063 the client will attempt to negotiate with the server. 4064 This limits the number of simultaneous RPC requests 4065 that the client can send to the NFSv4.1 server. 4066 Note that there is little point in setting this 4067 value higher than the max_tcp_slot_table_limit. 4068 4069 nfs.nfs4_disable_idmapping= 4070 [NFSv4] When set to the default of '1', this option 4071 ensures that both the RPC level authentication 4072 scheme and the NFS level operations agree to use 4073 numeric uids/gids if the mount is using the 4074 'sec=sys' security flavour. In effect it is 4075 disabling idmapping, which can make migration from 4076 legacy NFSv2/v3 systems to NFSv4 easier. 4077 Servers that do not support this mode of operation 4078 will be autodetected by the client, and it will fall 4079 back to using the idmapper. 4080 To turn off this behaviour, set the value to '0'. 4081 4082 nfs.nfs4_unique_id= 4083 [NFS4] Specify an additional fixed unique ident- 4084 ification string that NFSv4 clients can insert into 4085 their nfs_client_id4 string. This is typically a 4086 UUID that is generated at system install time. 4087 4088 nfs.recover_lost_locks= 4089 [NFSv4] Attempt to recover locks that were lost due 4090 to a lease timeout on the server. Please note that 4091 doing this risks data corruption, since there are 4092 no guarantees that the file will remain unchanged 4093 after the locks are lost. 4094 If you want to enable the kernel legacy behaviour of 4095 attempting to recover these locks, then set this 4096 parameter to '1'. 4097 The default parameter value of '0' causes the kernel 4098 not to attempt recovery of lost locks. 4099 4100 nfs.send_implementation_id= 4101 [NFSv4.1] Send client implementation identification 4102 information in exchange_id requests. 4103 If zero, no implementation identification information 4104 will be sent. 4105 The default is to send the implementation identification 4106 information. 4107 4108 nfs4.layoutstats_timer= 4109 [NFSv4.2] Change the rate at which the kernel sends 4110 layoutstats to the pNFS metadata server. 4111 4112 Setting this to value to 0 causes the kernel to use 4113 whatever value is the default set by the layout 4114 driver. A non-zero value sets the minimum interval 4115 in seconds between layoutstats transmissions. 4116 4117 nfsd.inter_copy_offload_enable= 4118 [NFSv4.2] When set to 1, the server will support 4119 server-to-server copies for which this server is 4120 the destination of the copy. 4121 4122 nfsd.nfs4_disable_idmapping= 4123 [NFSv4] When set to the default of '1', the NFSv4 4124 server will return only numeric uids and gids to 4125 clients using auth_sys, and will accept numeric uids 4126 and gids from such clients. This is intended to ease 4127 migration from NFSv2/v3. 4128 4129 nfsd.nfsd4_ssc_umount_timeout= 4130 [NFSv4.2] When used as the destination of a 4131 server-to-server copy, knfsd temporarily mounts 4132 the source server. It caches the mount in case 4133 it will be needed again, and discards it if not 4134 used for the number of milliseconds specified by 4135 this parameter. 4136 4137 nfsaddrs= [NFS] Deprecated. Use ip= instead. 4138 See Documentation/admin-guide/nfs/nfsroot.rst. 4139 4140 nfsroot= [NFS] nfs root filesystem for disk-less boxes. 4141 See Documentation/admin-guide/nfs/nfsroot.rst. 4142 4143 nfsrootdebug [NFS] enable nfsroot debugging messages. 4144 See Documentation/admin-guide/nfs/nfsroot.rst. 4145 4146 nmi_backtrace.backtrace_idle [KNL] 4147 Dump stacks even of idle CPUs in response to an 4148 NMI stack-backtrace request. 4149 4150 nmi_debug= [KNL,SH] Specify one or more actions to take 4151 when a NMI is triggered. 4152 Format: [state][,regs][,debounce][,die] 4153 4154 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels 4155 Format: [panic,][nopanic,][rNNN,][num] 4156 Valid num: 0 or 1 4157 0 - turn hardlockup detector in nmi_watchdog off 4158 1 - turn hardlockup detector in nmi_watchdog on 4159 rNNN - configure the watchdog with raw perf event 0xNNN 4160 4161 When panic is specified, panic when an NMI watchdog 4162 timeout occurs (or 'nopanic' to not panic on an NMI 4163 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set) 4164 To disable both hard and soft lockup detectors, 4165 please see 'nowatchdog'. 4166 This is useful when you use a panic=... timeout and 4167 need the box quickly up again. 4168 4169 These settings can be accessed at runtime via 4170 the nmi_watchdog and hardlockup_panic sysctls. 4171 4172 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths 4173 emulation library even if a 387 maths coprocessor 4174 is present. 4175 4176 no4lvl [RISCV,EARLY] Disable 4-level and 5-level paging modes. 4177 Forces kernel to use 3-level paging instead. 4178 4179 no5lvl [X86-64,RISCV,EARLY] Disable 5-level paging mode. Forces 4180 kernel to use 4-level paging instead. 4181 4182 noalign [KNL,ARM] 4183 4184 noapic [SMP,APIC,EARLY] Tells the kernel to not make use of any 4185 IOAPICs that may be present in the system. 4186 4187 noapictimer [APIC,X86] Don't set up the APIC timer 4188 4189 noautogroup Disable scheduler automatic task group creation. 4190 4191 nocache [ARM,EARLY] 4192 4193 no_console_suspend 4194 [HW] Never suspend the console 4195 Disable suspending of consoles during suspend and 4196 hibernate operations. Once disabled, debugging 4197 messages can reach various consoles while the rest 4198 of the system is being put to sleep (ie, while 4199 debugging driver suspend/resume hooks). This may 4200 not work reliably with all consoles, but is known 4201 to work with serial and VGA consoles. 4202 To facilitate more flexible debugging, we also add 4203 console_suspend, a printk module parameter to control 4204 it. Users could use console_suspend (usually 4205 /sys/module/printk/parameters/console_suspend) to 4206 turn on/off it dynamically. 4207 4208 no_debug_objects 4209 [KNL,EARLY] Disable object debugging 4210 4211 nodsp [SH] Disable hardware DSP at boot time. 4212 4213 noefi [EFI,EARLY] Disable EFI runtime services support. 4214 4215 no_entry_flush [PPC,EARLY] Don't flush the L1-D cache when entering the kernel. 4216 4217 noexec32 [X86-64] 4218 This affects only 32-bit executables. 4219 noexec32=on: enable non-executable mappings (default) 4220 read doesn't imply executable mappings 4221 noexec32=off: disable non-executable mappings 4222 read implies executable mappings 4223 4224 no_file_caps Tells the kernel not to honor file capabilities. The 4225 only way then for a file to be executed with privilege 4226 is to be setuid root or executed by root. 4227 4228 nofpu [MIPS,SH] Disable hardware FPU at boot time. 4229 4230 nofsgsbase [X86] Disables FSGSBASE instructions. 4231 4232 nofxsr [BUGS=X86-32] Disables x86 floating point extended 4233 register save and restore. The kernel will only save 4234 legacy floating-point registers on task switch. 4235 4236 nogbpages [X86] Do not use GB pages for kernel direct mappings. 4237 4238 no_hash_pointers 4239 [KNL,EARLY] 4240 Alias for "hash_pointers=never". 4241 4242 nohibernate [HIBERNATION] Disable hibernation and resume. 4243 4244 nohlt [ARM,ARM64,MICROBLAZE,MIPS,PPC,RISCV,SH] Forces the kernel to 4245 busy wait in do_idle() and not use the arch_cpu_idle() 4246 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP 4247 to be effective. This is useful on platforms where the 4248 sleep(SH) or wfi(ARM,ARM64) instructions do not work 4249 correctly or when doing power measurements to evaluate 4250 the impact of the sleep instructions. This is also 4251 useful when using JTAG debugger. 4252 4253 nohpet [X86] Don't use the HPET timer. 4254 4255 nohugeiomap [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge I/O mappings. 4256 4257 nohugevmalloc [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge vmalloc mappings. 4258 4259 nohz= [KNL] Boottime enable/disable dynamic ticks 4260 Valid arguments: on, off 4261 Default: on 4262 4263 nohz_full= [KNL,BOOT,SMP,ISOL] 4264 The argument is a cpu list, as described above. 4265 In kernels built with CONFIG_NO_HZ_FULL=y, set 4266 the specified list of CPUs whose tick will be stopped 4267 whenever possible. The boot CPU will be forced outside 4268 the range to maintain the timekeeping. Any CPUs 4269 in this list will have their RCU callbacks offloaded, 4270 just as if they had also been called out in the 4271 rcu_nocbs= boot parameter. 4272 4273 Note that this argument takes precedence over 4274 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option. 4275 4276 noinitrd [RAM] Tells the kernel not to load any configured 4277 initial RAM disk. 4278 4279 nointremap [X86-64,Intel-IOMMU,EARLY] Do not enable interrupt 4280 remapping. 4281 [Deprecated - use intremap=off] 4282 4283 noinvpcid [X86,EARLY] Disable the INVPCID cpu feature. 4284 4285 noiotrap [SH] Disables trapped I/O port accesses. 4286 4287 noirqdebug [X86-32] Disables the code which attempts to detect and 4288 disable unhandled interrupt sources. 4289 4290 noisapnp [ISAPNP] Disables ISA PnP code. 4291 4292 nokaslr [KNL,EARLY] 4293 When CONFIG_RANDOMIZE_BASE is set, this disables 4294 kernel and module base offset ASLR (Address Space 4295 Layout Randomization). 4296 4297 no-kvmapf [X86,KVM,EARLY] Disable paravirtualized asynchronous page 4298 fault handling. 4299 4300 no-kvmclock [X86,KVM,EARLY] Disable paravirtualized KVM clock driver 4301 4302 nolapic [X86-32,APIC,EARLY] Do not enable or use the local APIC. 4303 4304 nolapic_timer [X86-32,APIC,EARLY] Do not use the local APIC timer. 4305 4306 nomce [X86-32] Disable Machine Check Exception 4307 4308 nomfgpt [X86-32] Disable Multi-Function General Purpose 4309 Timer usage (for AMD Geode machines). 4310 4311 nomodeset Disable kernel modesetting. Most systems' firmware 4312 sets up a display mode and provides framebuffer memory 4313 for output. With nomodeset, DRM and fbdev drivers will 4314 not load if they could possibly displace the pre- 4315 initialized output. Only the system framebuffer will 4316 be available for use. The respective drivers will not 4317 perform display-mode changes or accelerated rendering. 4318 4319 Useful as error fallback, or for testing and debugging. 4320 4321 nomodule Disable module load 4322 4323 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to 4324 shutdown the other cpus. Instead use the REBOOT_VECTOR 4325 irq. 4326 4327 nopat [X86,EARLY] Disable PAT (page attribute table extension of 4328 pagetables) support. 4329 4330 nopcid [X86-64,EARLY] Disable the PCID cpu feature. 4331 4332 nopku [X86] Disable Memory Protection Keys CPU feature found 4333 in some Intel CPUs. 4334 4335 nopti [X86-64,EARLY] 4336 Equivalent to pti=off 4337 4338 nopv= [X86,XEN,KVM,HYPER_V,VMWARE,EARLY] 4339 Disables the PV optimizations forcing the guest to run 4340 as generic guest with no PV drivers. Currently support 4341 XEN HVM, KVM, HYPER_V and VMWARE guest. 4342 4343 nopvspin [X86,XEN,KVM,EARLY] 4344 Disables the qspinlock slow path using PV optimizations 4345 which allow the hypervisor to 'idle' the guest on lock 4346 contention. 4347 4348 norandmaps Don't use address space randomization. Equivalent to 4349 echo 0 > /proc/sys/kernel/randomize_va_space 4350 4351 noreplace-smp [X86-32,SMP] Don't replace SMP instructions 4352 with UP alternatives 4353 4354 noresume [SWSUSP] Disables resume and restores original swap 4355 space. 4356 4357 no-scroll [VGA] Disables scrollback. 4358 This is required for the Braillex ib80-piezo Braille 4359 reader made by F.H. Papenmeier (Germany). 4360 4361 nosgx [X86-64,SGX,EARLY] Disables Intel SGX kernel support. 4362 4363 nosmap [PPC,EARLY] 4364 Disable SMAP (Supervisor Mode Access Prevention) 4365 even if it is supported by processor. 4366 4367 nosmep [PPC64s,EARLY] 4368 Disable SMEP (Supervisor Mode Execution Prevention) 4369 even if it is supported by processor. 4370 4371 nosmp [SMP,EARLY] Tells an SMP kernel to act as a UP kernel, 4372 and disable the IO APIC. legacy for "maxcpus=0". 4373 4374 nosmt [KNL,MIPS,PPC,EARLY] Disable symmetric multithreading (SMT). 4375 Equivalent to smt=1. 4376 4377 [KNL,X86,PPC,S390] Disable symmetric multithreading (SMT). 4378 nosmt=force: Force disable SMT, cannot be undone 4379 via the sysfs control file. 4380 4381 nosoftlockup [KNL] Disable the soft-lockup detector. 4382 4383 nospec_store_bypass_disable 4384 [HW,EARLY] Disable all mitigations for the Speculative 4385 Store Bypass vulnerability 4386 4387 nospectre_bhb [ARM64,EARLY] Disable all mitigations for Spectre-BHB (branch 4388 history injection) vulnerability. System may allow data leaks 4389 with this option. 4390 4391 nospectre_v1 [X86,PPC,EARLY] Disable mitigations for Spectre Variant 1 4392 (bounds check bypass). With this option data leaks are 4393 possible in the system. 4394 4395 nospectre_v2 [X86,PPC_E500,ARM64,EARLY] Disable all mitigations 4396 for the Spectre variant 2 (indirect branch 4397 prediction) vulnerability. System may allow data 4398 leaks with this option. 4399 4400 no-steal-acc [X86,PV_OPS,ARM64,PPC/PSERIES,RISCV,LOONGARCH,EARLY] 4401 Disable paravirtualized steal time accounting. steal time 4402 is computed, but won't influence scheduler behaviour 4403 4404 nosync [HW,M68K] Disables sync negotiation for all devices. 4405 4406 no_timer_check [X86,APIC] Disables the code which tests for broken 4407 timer IRQ sources, i.e., the IO-APIC timer. This can 4408 work around problems with incorrect timer 4409 initialization on some boards. 4410 4411 no_uaccess_flush 4412 [PPC,EARLY] Don't flush the L1-D cache after accessing user data. 4413 4414 novmcoredd [KNL,KDUMP] 4415 Disable device dump. Device dump allows drivers to 4416 append dump data to vmcore so you can collect driver 4417 specified debug info. Drivers can append the data 4418 without any limit and this data is stored in memory, 4419 so this may cause significant memory stress. Disabling 4420 device dump can help save memory but the driver debug 4421 data will be no longer available. This parameter 4422 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP 4423 is set. 4424 4425 no-vmw-sched-clock 4426 [X86,PV_OPS,EARLY] Disable paravirtualized VMware 4427 scheduler clock and use the default one. 4428 4429 nowatchdog [KNL] Disable both lockup detectors, i.e. 4430 soft-lockup and NMI watchdog (hard-lockup). 4431 4432 nowb [ARM,EARLY] 4433 4434 nox2apic [X86-64,APIC,EARLY] Do not enable x2APIC mode. 4435 4436 NOTE: this parameter will be ignored on systems with the 4437 LEGACY_XAPIC_DISABLED bit set in the 4438 IA32_XAPIC_DISABLE_STATUS MSR. 4439 4440 noxsave [BUGS=X86] Disables x86 extended register state save 4441 and restore using xsave. The kernel will fallback to 4442 enabling legacy floating-point and sse state. 4443 4444 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended 4445 register states. The kernel will fall back to use 4446 xsave to save the states. By using this parameter, 4447 performance of saving the states is degraded because 4448 xsave doesn't support modified optimization while 4449 xsaveopt supports it on xsaveopt enabled systems. 4450 4451 noxsaves [X86] Disables xsaves and xrstors used in saving and 4452 restoring x86 extended register state in compacted 4453 form of xsave area. The kernel will fall back to use 4454 xsaveopt and xrstor to save and restore the states 4455 in standard form of xsave area. By using this 4456 parameter, xsave area per process might occupy more 4457 memory on xsaves enabled systems. 4458 4459 nr_cpus= [SMP,EARLY] Maximum number of processors that an SMP kernel 4460 could support. nr_cpus=n : n >= 1 limits the kernel to 4461 support 'n' processors. It could be larger than the 4462 number of already plugged CPU during bootup, later in 4463 runtime you can physically add extra cpu until it reaches 4464 n. So during boot up some boot time memory for per-cpu 4465 variables need be pre-allocated for later physical cpu 4466 hot plugging. 4467 4468 nr_uarts= [SERIAL] maximum number of UARTs to be registered. 4469 4470 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86, EARLY] 4471 Disable NUMA, Only set up a single NUMA node 4472 spanning all memory. 4473 4474 numa=fake=<size>[MG] 4475 [KNL, ARM64, RISCV, X86, EARLY] 4476 If given as a memory unit, fills all system RAM with 4477 nodes of size interleaved over physical nodes. 4478 4479 numa=fake=<N> 4480 [KNL, ARM64, RISCV, X86, EARLY] 4481 If given as an integer, fills all system RAM with N 4482 fake nodes interleaved over physical nodes. 4483 4484 numa=fake=<N>U 4485 [KNL, ARM64, RISCV, X86, EARLY] 4486 If given as an integer followed by 'U', it will 4487 divide each physical node into N emulated nodes. 4488 4489 numa=noacpi [X86] Don't parse the SRAT table for NUMA setup 4490 4491 numa=nohmat [X86] Don't parse the HMAT table for NUMA setup, or 4492 soft-reserved memory partitioning. 4493 4494 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic 4495 NUMA balancing. 4496 Allowed values are enable and disable 4497 4498 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA. 4499 'node', 'default' can be specified 4500 This can be set from sysctl after boot. 4501 See Documentation/admin-guide/sysctl/vm.rst for details. 4502 4503 ohci1394_dma=early [HW,EARLY] enable debugging via the ohci1394 driver. 4504 See Documentation/core-api/debugging-via-ohci1394.rst for more 4505 info. 4506 4507 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands 4508 Rather than timing out after 20 ms if an EC 4509 command is not properly ACKed, override the length 4510 of the timeout. We have interrupts disabled while 4511 waiting for the ACK, so if this is set too high 4512 interrupts *may* be lost! 4513 4514 omap_mux= [OMAP] Override bootloader pin multiplexing. 4515 Format: <mux_mode0.mode_name=value>... 4516 For example, to override I2C bus2: 4517 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100 4518 4519 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration 4520 4521 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock] 4522 4523 boundary - index of last SLC block on Flex-OneNAND. 4524 The remaining blocks are configured as MLC blocks. 4525 lock - Configure if Flex-OneNAND boundary should be locked. 4526 Once locked, the boundary cannot be changed. 4527 1 indicates lock status, 0 indicates unlock status. 4528 4529 oops=panic [KNL,EARLY] 4530 Always panic on oopses. Default is to just kill the 4531 process, but there is a small probability of 4532 deadlocking the machine. 4533 This will also cause panics on machine check exceptions. 4534 Useful together with panic=30 to trigger a reboot. 4535 4536 page_alloc.shuffle= 4537 [KNL] Boolean flag to control whether the page allocator 4538 should randomize its free lists. This parameter can be 4539 used to enable/disable page randomization. The state of 4540 the flag can be read from sysfs at: 4541 /sys/module/page_alloc/parameters/shuffle. 4542 This parameter is only available if CONFIG_SHUFFLE_PAGE_ALLOCATOR=y. 4543 4544 page_owner= [KNL,EARLY] Boot-time page_owner enabling option. 4545 Storage of the information about who allocated 4546 each page is disabled in default. With this switch, 4547 we can turn it on. 4548 on: enable the feature 4549 4550 page_poison= [KNL,EARLY] Boot-time parameter changing the state of 4551 poisoning on the buddy allocator, available with 4552 CONFIG_PAGE_POISONING=y. 4553 off: turn off poisoning (default) 4554 on: turn on poisoning 4555 4556 page_reporting.page_reporting_order= 4557 [KNL] Minimal page reporting order 4558 Format: <integer> 4559 Adjust the minimal page reporting order. The page 4560 reporting is disabled when it exceeds MAX_PAGE_ORDER. 4561 4562 panic= [KNL] Kernel behaviour on panic: delay <timeout> 4563 timeout > 0: seconds before rebooting 4564 timeout = 0: wait forever 4565 timeout < 0: reboot immediately 4566 Format: <timeout> 4567 4568 panic_on_taint= [KNL,EARLY] 4569 Bitmask for conditionally calling panic() in add_taint() 4570 Format: <hex>[,nousertaint] 4571 Hexadecimal bitmask representing the set of TAINT flags 4572 that will cause the kernel to panic when add_taint() is 4573 called with any of the flags in this set. 4574 The optional switch "nousertaint" can be utilized to 4575 prevent userspace forced crashes by writing to sysctl 4576 /proc/sys/kernel/tainted any flagset matching with the 4577 bitmask set on panic_on_taint. 4578 See Documentation/admin-guide/tainted-kernels.rst for 4579 extra details on the taint flags that users can pick 4580 to compose the bitmask to assign to panic_on_taint. 4581 4582 panic_on_warn=1 panic() instead of WARN(). Useful to cause kdump 4583 on a WARN(). 4584 4585 panic_print= Bitmask for printing system info when panic happens. 4586 User can chose combination of the following bits: 4587 bit 0: print all tasks info 4588 bit 1: print system memory info 4589 bit 2: print timer info 4590 bit 3: print locks info if CONFIG_LOCKDEP is on 4591 bit 4: print ftrace buffer 4592 bit 5: replay all messages on consoles at the end of panic 4593 bit 6: print all CPUs backtrace (if available in the arch) 4594 bit 7: print only tasks in uninterruptible (blocked) state 4595 *Be aware* that this option may print a _lot_ of lines, 4596 so there are risks of losing older messages in the log. 4597 Use this option carefully, maybe worth to setup a 4598 bigger log buffer with "log_buf_len" along with this. 4599 4600 panic_sys_info= A comma separated list of extra information to be dumped 4601 on panic. 4602 Format: val[,val...] 4603 Where @val can be any of the following: 4604 4605 tasks: print all tasks info 4606 mem: print system memory info 4607 timers: print timers info 4608 locks: print locks info if CONFIG_LOCKDEP is on 4609 ftrace: print ftrace buffer 4610 all_bt: print all CPUs backtrace (if available in the arch) 4611 blocked_tasks: print only tasks in uninterruptible (blocked) state 4612 4613 This is a human readable alternative to the 'panic_print' option. 4614 4615 panic_console_replay 4616 When panic happens, replay all kernel messages on 4617 consoles at the end of panic. 4618 4619 parkbd.port= [HW] Parallel port number the keyboard adapter is 4620 connected to, default is 0. 4621 Format: <parport#> 4622 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation, 4623 0 for XT, 1 for AT (default is AT). 4624 Format: <mode> 4625 4626 parport= [HW,PPT] Specify parallel ports. 0 disables. 4627 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] } 4628 Use 'auto' to force the driver to use any 4629 IRQ/DMA settings detected (the default is to 4630 ignore detected IRQ/DMA settings because of 4631 possible conflicts). You can specify the base 4632 address, IRQ, and DMA settings; IRQ and DMA 4633 should be numbers, or 'auto' (for using detected 4634 settings on that particular port), or 'nofifo' 4635 (to avoid using a FIFO even if it is detected). 4636 Parallel ports are assigned in the order they 4637 are specified on the command line, starting 4638 with parport0. 4639 4640 parport_init_mode= [HW,PPT] 4641 Configure VIA parallel port to operate in 4642 a specific mode. This is necessary on Pegasos 4643 computer where firmware has no options for setting 4644 up parallel port mode and sets it to spp. 4645 Currently this function knows 686a and 8231 chips. 4646 Format: [spp|ps2|epp|ecp|ecpepp] 4647 4648 pata_legacy.all= [HW,LIBATA] 4649 Format: <int> 4650 Set to non-zero to probe primary and secondary ISA 4651 port ranges on PCI systems where no PCI PATA device 4652 has been found at either range. Disabled by default. 4653 4654 pata_legacy.autospeed= [HW,LIBATA] 4655 Format: <int> 4656 Set to non-zero if a chip is present that snoops speed 4657 changes. Disabled by default. 4658 4659 pata_legacy.ht6560a= [HW,LIBATA] 4660 Format: <int> 4661 Set to 1, 2, or 3 for HT 6560A on the primary channel, 4662 the secondary channel, or both channels respectively. 4663 Disabled by default. 4664 4665 pata_legacy.ht6560b= [HW,LIBATA] 4666 Format: <int> 4667 Set to 1, 2, or 3 for HT 6560B on the primary channel, 4668 the secondary channel, or both channels respectively. 4669 Disabled by default. 4670 4671 pata_legacy.iordy_mask= [HW,LIBATA] 4672 Format: <int> 4673 IORDY enable mask. Set individual bits to allow IORDY 4674 for the respective channel. Bit 0 is for the first 4675 legacy channel handled by this driver, bit 1 is for 4676 the second channel, and so on. The sequence will often 4677 correspond to the primary legacy channel, the secondary 4678 legacy channel, and so on, but the handling of a PCI 4679 bus and the use of other driver options may interfere 4680 with the sequence. By default IORDY is allowed across 4681 all channels. 4682 4683 pata_legacy.opti82c46x= [HW,LIBATA] 4684 Format: <int> 4685 Set to 1, 2, or 3 for Opti 82c611A on the primary 4686 channel, the secondary channel, or both channels 4687 respectively. Disabled by default. 4688 4689 pata_legacy.opti82c611a= [HW,LIBATA] 4690 Format: <int> 4691 Set to 1, 2, or 3 for Opti 82c465MV on the primary 4692 channel, the secondary channel, or both channels 4693 respectively. Disabled by default. 4694 4695 pata_legacy.pio_mask= [HW,LIBATA] 4696 Format: <int> 4697 PIO mode mask for autospeed devices. Set individual 4698 bits to allow the use of the respective PIO modes. 4699 Bit 0 is for mode 0, bit 1 is for mode 1, and so on. 4700 All modes allowed by default. 4701 4702 pata_legacy.probe_all= [HW,LIBATA] 4703 Format: <int> 4704 Set to non-zero to probe tertiary and further ISA 4705 port ranges on PCI systems. Disabled by default. 4706 4707 pata_legacy.probe_mask= [HW,LIBATA] 4708 Format: <int> 4709 Probe mask for legacy ISA PATA ports. Depending on 4710 platform configuration and the use of other driver 4711 options up to 6 legacy ports are supported: 0x1f0, 4712 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing 4713 of individual ports can be disabled by setting the 4714 corresponding bits in the mask to 1. Bit 0 is for 4715 the first port in the list above (0x1f0), and so on. 4716 By default all supported ports are probed. 4717 4718 pata_legacy.qdi= [HW,LIBATA] 4719 Format: <int> 4720 Set to non-zero to probe QDI controllers. By default 4721 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise. 4722 4723 pata_legacy.winbond= [HW,LIBATA] 4724 Format: <int> 4725 Set to non-zero to probe Winbond controllers. Use 4726 the standard I/O port (0x130) if 1, otherwise the 4727 value given is the I/O port to use (typically 0x1b0). 4728 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE, 4729 0 otherwise. 4730 4731 pata_platform.pio_mask= [HW,LIBATA] 4732 Format: <int> 4733 Supported PIO mode mask. Set individual bits to allow 4734 the use of the respective PIO modes. Bit 0 is for 4735 mode 0, bit 1 is for mode 1, and so on. Mode 0 only 4736 allowed by default. 4737 4738 pause_on_oops=<int> 4739 Halt all CPUs after the first oops has been printed for 4740 the specified number of seconds. This is to be used if 4741 your oopses keep scrolling off the screen. 4742 4743 pcbit= [HW,ISDN] 4744 4745 pci=option[,option...] [PCI,EARLY] various PCI subsystem options. 4746 4747 Some options herein operate on a specific device 4748 or a set of devices (<pci_dev>). These are 4749 specified in one of the following formats: 4750 4751 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]* 4752 pci:<vendor>:<device>[:<subvendor>:<subdevice>] 4753 4754 Note: the first format specifies a PCI 4755 bus/device/function address which may change 4756 if new hardware is inserted, if motherboard 4757 firmware changes, or due to changes caused 4758 by other kernel parameters. If the 4759 domain is left unspecified, it is 4760 taken to be zero. Optionally, a path 4761 to a device through multiple device/function 4762 addresses can be specified after the base 4763 address (this is more robust against 4764 renumbering issues). The second format 4765 selects devices using IDs from the 4766 configuration space which may match multiple 4767 devices in the system. 4768 4769 earlydump dump PCI config space before the kernel 4770 changes anything 4771 off [X86] don't probe for the PCI bus 4772 bios [X86-32] force use of PCI BIOS, don't access 4773 the hardware directly. Use this if your machine 4774 has a non-standard PCI host bridge. 4775 nobios [X86-32] disallow use of PCI BIOS, only direct 4776 hardware access methods are allowed. Use this 4777 if you experience crashes upon bootup and you 4778 suspect they are caused by the BIOS. 4779 conf1 [X86] Force use of PCI Configuration Access 4780 Mechanism 1 (config address in IO port 0xCF8, 4781 data in IO port 0xCFC, both 32-bit). 4782 conf2 [X86] Force use of PCI Configuration Access 4783 Mechanism 2 (IO port 0xCF8 is an 8-bit port for 4784 the function, IO port 0xCFA, also 8-bit, sets 4785 bus number. The config space is then accessed 4786 through ports 0xC000-0xCFFF). 4787 See http://wiki.osdev.org/PCI for more info 4788 on the configuration access mechanisms. 4789 noaer [PCIE] If the PCIEAER kernel config parameter is 4790 enabled, this kernel boot option can be used to 4791 disable the use of PCIE advanced error reporting. 4792 nodomains [PCI] Disable support for multiple PCI 4793 root domains (aka PCI segments, in ACPI-speak). 4794 nommconf [X86] Disable use of MMCONFIG for PCI 4795 Configuration 4796 check_enable_amd_mmconf [X86] check for and enable 4797 properly configured MMIO access to PCI 4798 config space on AMD family 10h CPU 4799 nomsi [MSI] If the PCI_MSI kernel config parameter is 4800 enabled, this kernel boot option can be used to 4801 disable the use of MSI interrupts system-wide. 4802 noioapicquirk [APIC] Disable all boot interrupt quirks. 4803 Safety option to keep boot IRQs enabled. This 4804 should never be necessary. 4805 ioapicreroute [APIC] Enable rerouting of boot IRQs to the 4806 primary IO-APIC for bridges that cannot disable 4807 boot IRQs. This fixes a source of spurious IRQs 4808 when the system masks IRQs. 4809 noioapicreroute [APIC] Disable workaround that uses the 4810 boot IRQ equivalent of an IRQ that connects to 4811 a chipset where boot IRQs cannot be disabled. 4812 The opposite of ioapicreroute. 4813 biosirq [X86-32] Use PCI BIOS calls to get the interrupt 4814 routing table. These calls are known to be buggy 4815 on several machines and they hang the machine 4816 when used, but on other computers it's the only 4817 way to get the interrupt routing table. Try 4818 this option if the kernel is unable to allocate 4819 IRQs or discover secondary PCI buses on your 4820 motherboard. 4821 rom [X86] Assign address space to expansion ROMs. 4822 Use with caution as certain devices share 4823 address decoders between ROMs and other 4824 resources. 4825 norom [X86] Do not assign address space to 4826 expansion ROMs that do not already have 4827 BIOS assigned address ranges. 4828 nobar [X86] Do not assign address space to the 4829 BARs that weren't assigned by the BIOS. 4830 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be 4831 assigned automatically to PCI devices. You can 4832 make the kernel exclude IRQs of your ISA cards 4833 this way. 4834 pirqaddr=0xAAAAA [X86] Specify the physical address 4835 of the PIRQ table (normally generated 4836 by the BIOS) if it is outside the 4837 F0000h-100000h range. 4838 lastbus=N [X86] Scan all buses thru bus #N. Can be 4839 useful if the kernel is unable to find your 4840 secondary buses and you want to tell it 4841 explicitly which ones they are. 4842 assign-busses [X86] Always assign all PCI bus 4843 numbers ourselves, overriding 4844 whatever the firmware may have done. 4845 usepirqmask [X86] Honor the possible IRQ mask stored 4846 in the BIOS $PIR table. This is needed on 4847 some systems with broken BIOSes, notably 4848 some HP Pavilion N5400 and Omnibook XE3 4849 notebooks. This will have no effect if ACPI 4850 IRQ routing is enabled. 4851 noacpi [X86] Do not use ACPI for IRQ routing 4852 or for PCI scanning. 4853 use_crs [X86] Use PCI host bridge window information 4854 from ACPI. On BIOSes from 2008 or later, this 4855 is enabled by default. If you need to use this, 4856 please report a bug. 4857 nocrs [X86] Ignore PCI host bridge windows from ACPI. 4858 If you need to use this, please report a bug. 4859 use_e820 [X86] Use E820 reservations to exclude parts of 4860 PCI host bridge windows. This is a workaround 4861 for BIOS defects in host bridge _CRS methods. 4862 If you need to use this, please report a bug to 4863 <linux-pci@vger.kernel.org>. 4864 no_e820 [X86] Ignore E820 reservations for PCI host 4865 bridge windows. This is the default on modern 4866 hardware. If you need to use this, please report 4867 a bug to <linux-pci@vger.kernel.org>. 4868 routeirq Do IRQ routing for all PCI devices. 4869 This is normally done in pci_enable_device(), 4870 so this option is a temporary workaround 4871 for broken drivers that don't call it. 4872 skip_isa_align [X86] do not align io start addr, so can 4873 handle more pci cards 4874 noearly [X86] Don't do any early type 1 scanning. 4875 This might help on some broken boards which 4876 machine check when some devices' config space 4877 is read. But various workarounds are disabled 4878 and some IOMMU drivers will not work. 4879 bfsort Sort PCI devices into breadth-first order. 4880 This sorting is done to get a device 4881 order compatible with older (<= 2.4) kernels. 4882 nobfsort Don't sort PCI devices into breadth-first order. 4883 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size) 4884 tuning and use the BIOS-configured MPS defaults. 4885 pcie_bus_safe Set every device's MPS to the largest value 4886 supported by all devices below the root complex. 4887 pcie_bus_perf Set device MPS to the largest allowable MPS 4888 based on its parent bus. Also set MRRS (Max 4889 Read Request Size) to the largest supported 4890 value (no larger than the MPS that the device 4891 or bus can support) for best performance. 4892 pcie_bus_peer2peer Set every device's MPS to 128B, which 4893 every device is guaranteed to support. This 4894 configuration allows peer-to-peer DMA between 4895 any pair of devices, possibly at the cost of 4896 reduced performance. This also guarantees 4897 that hot-added devices will work. 4898 cbiosize=nn[KMG] The fixed amount of bus space which is 4899 reserved for the CardBus bridge's IO window. 4900 The default value is 256 bytes. 4901 cbmemsize=nn[KMG] The fixed amount of bus space which is 4902 reserved for the CardBus bridge's memory 4903 window. The default value is 64 megabytes. 4904 resource_alignment= 4905 Format: 4906 [<order of align>@]<pci_dev>[; ...] 4907 Specifies alignment and device to reassign 4908 aligned memory resources. How to 4909 specify the device is described above. 4910 If <order of align> is not specified, 4911 PAGE_SIZE is used as alignment. 4912 A PCI-PCI bridge can be specified if resource 4913 windows need to be expanded. 4914 To specify the alignment for several 4915 instances of a device, the PCI vendor, 4916 device, subvendor, and subdevice may be 4917 specified, e.g., 12@pci:8086:9c22:103c:198f 4918 for 4096-byte alignment. 4919 ecrc= Enable/disable PCIe ECRC (transaction layer 4920 end-to-end CRC checking). Only effective if 4921 OS has native AER control (either granted by 4922 ACPI _OSC or forced via "pcie_ports=native") 4923 bios: Use BIOS/firmware settings. This is the 4924 the default. 4925 off: Turn ECRC off 4926 on: Turn ECRC on. 4927 hpiosize=nn[KMG] The fixed amount of bus space which is 4928 reserved for hotplug bridge's IO window. 4929 Default size is 256 bytes. 4930 hpmmiosize=nn[KMG] The fixed amount of bus space which is 4931 reserved for hotplug bridge's MMIO window. 4932 Default size is 2 megabytes. 4933 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is 4934 reserved for hotplug bridge's MMIO_PREF window. 4935 Default size is 2 megabytes. 4936 hpmemsize=nn[KMG] The fixed amount of bus space which is 4937 reserved for hotplug bridge's MMIO and 4938 MMIO_PREF window. 4939 Default size is 2 megabytes. 4940 hpbussize=nn The minimum amount of additional bus numbers 4941 reserved for buses below a hotplug bridge. 4942 Default is 1. 4943 realloc= Enable/disable reallocating PCI bridge resources 4944 if allocations done by BIOS are too small to 4945 accommodate resources required by all child 4946 devices. 4947 off: Turn realloc off 4948 on: Turn realloc on 4949 realloc same as realloc=on 4950 noari do not use PCIe ARI. 4951 noats [PCIE, Intel-IOMMU, AMD-IOMMU] 4952 do not use PCIe ATS (and IOMMU device IOTLB). 4953 pcie_scan_all Scan all possible PCIe devices. Otherwise we 4954 only look for one device below a PCIe downstream 4955 port. 4956 big_root_window Try to add a big 64bit memory window to the PCIe 4957 root complex on AMD CPUs. Some GFX hardware 4958 can resize a BAR to allow access to all VRAM. 4959 Adding the window is slightly risky (it may 4960 conflict with unreported devices), so this 4961 taints the kernel. 4962 disable_acs_redir=<pci_dev>[; ...] 4963 Specify one or more PCI devices (in the format 4964 specified above) separated by semicolons. 4965 Each device specified will have the PCI ACS 4966 redirect capabilities forced off which will 4967 allow P2P traffic between devices through 4968 bridges without forcing it upstream. Note: 4969 this removes isolation between devices and 4970 may put more devices in an IOMMU group. 4971 config_acs= 4972 Format: 4973 <ACS flags>@<pci_dev>[; ...] 4974 Specify one or more PCI devices (in the format 4975 specified above) optionally prepended with flags 4976 and separated by semicolons. The respective 4977 capabilities will be enabled, disabled or 4978 unchanged based on what is specified in 4979 flags. 4980 4981 ACS Flags is defined as follows: 4982 bit-0 : ACS Source Validation 4983 bit-1 : ACS Translation Blocking 4984 bit-2 : ACS P2P Request Redirect 4985 bit-3 : ACS P2P Completion Redirect 4986 bit-4 : ACS Upstream Forwarding 4987 bit-5 : ACS P2P Egress Control 4988 bit-6 : ACS Direct Translated P2P 4989 Each bit can be marked as: 4990 '0' – force disabled 4991 '1' – force enabled 4992 'x' – unchanged 4993 For example, 4994 pci=config_acs=10x@pci:0:0 4995 would configure all devices that support 4996 ACS to enable P2P Request Redirect, disable 4997 Translation Blocking, and leave Source 4998 Validation unchanged from whatever power-up 4999 or firmware set it to. 5000 5001 Note: this may remove isolation between devices 5002 and may put more devices in an IOMMU group. 5003 force_floating [S390] Force usage of floating interrupts. 5004 nomio [S390] Do not use MIO instructions. 5005 norid [S390] ignore the RID field and force use of 5006 one PCI domain per PCI function 5007 notph [PCIE] If the PCIE_TPH kernel config parameter 5008 is enabled, this kernel boot option can be used 5009 to disable PCIe TLP Processing Hints support 5010 system-wide. 5011 5012 pcie_aspm= [PCIE] Forcibly enable or ignore PCIe Active State Power 5013 Management. 5014 off Don't touch ASPM configuration at all. Leave any 5015 configuration done by firmware unchanged. 5016 force Enable ASPM even on devices that claim not to support it. 5017 WARNING: Forcing ASPM on may cause system lockups. 5018 5019 pcie_ports= [PCIE] PCIe port services handling: 5020 native Use native PCIe services (PME, AER, DPC, PCIe hotplug) 5021 even if the platform doesn't give the OS permission to 5022 use them. This may cause conflicts if the platform 5023 also tries to use these services. 5024 dpc-native Use native PCIe service for DPC only. May 5025 cause conflicts if firmware uses AER or DPC. 5026 compat Disable native PCIe services (PME, AER, DPC, PCIe 5027 hotplug). 5028 5029 pcie_port_pm= [PCIE] PCIe port power management handling: 5030 off Disable power management of all PCIe ports 5031 force Forcibly enable power management of all PCIe ports 5032 5033 pcie_pme= [PCIE,PM] Native PCIe PME signaling options: 5034 nomsi Do not use MSI for native PCIe PME signaling (this makes 5035 all PCIe root ports use INTx for all services). 5036 5037 pcmv= [HW,PCMCIA] BadgePAD 4 5038 5039 pd_ignore_unused 5040 [PM] 5041 Keep all power-domains already enabled by bootloader on, 5042 even if no driver has claimed them. This is useful 5043 for debug and development, but should not be 5044 needed on a platform with proper driver support. 5045 5046 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at 5047 boot time. 5048 Format: { 0 | 1 } 5049 See arch/parisc/kernel/pdc_chassis.c 5050 5051 percpu_alloc= [MM,EARLY] 5052 Select which percpu first chunk allocator to use. 5053 Currently supported values are "embed" and "page". 5054 Archs may support subset or none of the selections. 5055 See comments in mm/percpu.c for details on each 5056 allocator. This parameter is primarily for debugging 5057 and performance comparison. 5058 5059 pirq= [SMP,APIC] Manual mp-table setup 5060 See Documentation/arch/x86/i386/IO-APIC.rst. 5061 5062 plip= [PPT,NET] Parallel port network link 5063 Format: { parport<nr> | timid | 0 } 5064 See also Documentation/admin-guide/parport.rst. 5065 5066 pmtmr= [X86] Manual setup of pmtmr I/O Port. 5067 Override pmtimer IOPort with a hex value. 5068 e.g. pmtmr=0x508 5069 5070 pmu_override= [PPC] Override the PMU. 5071 This option takes over the PMU facility, so it is no 5072 longer usable by perf. Setting this option starts the 5073 PMU counters by setting MMCR0 to 0 (the FC bit is 5074 cleared). If a number is given, then MMCR1 is set to 5075 that number, otherwise (e.g., 'pmu_override=on'), MMCR1 5076 remains 0. 5077 5078 pm_async= [PM] 5079 Format: off 5080 This parameter sets the initial value of the 5081 /sys/power/pm_async sysfs knob at boot time. 5082 If set to "off", disables asynchronous suspend and 5083 resume of devices during system-wide power transitions. 5084 This can be useful on platforms where device 5085 dependencies are not well-defined, or for debugging 5086 power management issues. Asynchronous operations are 5087 enabled by default. 5088 5089 5090 pm_debug_messages [SUSPEND,KNL] 5091 Enable suspend/resume debug messages during boot up. 5092 5093 pnp.debug=1 [PNP] 5094 Enable PNP debug messages (depends on the 5095 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time 5096 via /sys/module/pnp/parameters/debug. We always show 5097 current resource usage; turning this on also shows 5098 possible settings and some assignment information. 5099 5100 pnpacpi= [ACPI] 5101 { off } 5102 5103 pnpbios= [ISAPNP] 5104 { on | off | curr | res | no-curr | no-res } 5105 5106 pnp_reserve_irq= 5107 [ISAPNP] Exclude IRQs for the autoconfiguration 5108 5109 pnp_reserve_dma= 5110 [ISAPNP] Exclude DMAs for the autoconfiguration 5111 5112 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration 5113 Ranges are in pairs (I/O port base and size). 5114 5115 pnp_reserve_mem= 5116 [ISAPNP] Exclude memory regions for the 5117 autoconfiguration. 5118 Ranges are in pairs (memory base and size). 5119 5120 ports= [IP_VS_FTP] IPVS ftp helper module 5121 Default is 21. 5122 Up to 8 (IP_VS_APP_MAX_PORTS) ports 5123 may be specified. 5124 Format: <port>,<port>.... 5125 5126 possible_cpus= [SMP,S390,X86] 5127 Format: <unsigned int> 5128 Set the number of possible CPUs, overriding the 5129 regular discovery mechanisms (such as ACPI/FW, etc). 5130 5131 powersave=off [PPC] This option disables power saving features. 5132 It specifically disables cpuidle and sets the 5133 platform machine description specific power_save 5134 function to NULL. On Idle the CPU just reduces 5135 execution priority. 5136 5137 ppc_strict_facility_enable 5138 [PPC,ENABLE] This option catches any kernel floating point, 5139 Altivec, VSX and SPE outside of regions specifically 5140 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()). 5141 There is some performance impact when enabling this. 5142 5143 ppc_tm= [PPC,EARLY] 5144 Format: {"off"} 5145 Disable Hardware Transactional Memory 5146 5147 preempt= [KNL] 5148 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC 5149 none - Limited to cond_resched() calls 5150 voluntary - Limited to cond_resched() and might_sleep() calls 5151 full - Any section that isn't explicitly preempt disabled 5152 can be preempted anytime. Tasks will also yield 5153 contended spinlocks (if the critical section isn't 5154 explicitly preempt disabled beyond the lock itself). 5155 lazy - Scheduler controlled. Similar to full but instead 5156 of preempting the task immediately, the task gets 5157 one HZ tick time to yield itself before the 5158 preemption will be forced. One preemption is when the 5159 task returns to user space. 5160 5161 print-fatal-signals= 5162 [KNL] debug: print fatal signals 5163 5164 If enabled, warn about various signal handling 5165 related application anomalies: too many signals, 5166 too many POSIX.1 timers, fatal signals causing a 5167 coredump - etc. 5168 5169 If you hit the warning due to signal overflow, 5170 you might want to try "ulimit -i unlimited". 5171 5172 default: off. 5173 5174 printk.always_kmsg_dump= 5175 Trigger kmsg_dump for cases other than kernel oops or 5176 panics 5177 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5178 default: disabled 5179 5180 printk.console_no_auto_verbose= 5181 Disable console loglevel raise on oops, panic 5182 or lockdep-detected issues (only if lock debug is on). 5183 With an exception to setups with low baudrate on 5184 serial console, keeping this 0 is a good choice 5185 in order to provide more debug information. 5186 Format: <bool> 5187 default: 0 (auto_verbose is enabled) 5188 5189 printk.debug_non_panic_cpus= 5190 Allows storing messages from non-panic CPUs into 5191 the printk log buffer during panic(). They are 5192 flushed to consoles by the panic-CPU on 5193 a best-effort basis. 5194 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5195 Default: disabled 5196 5197 printk.devkmsg={on,off,ratelimit} 5198 Control writing to /dev/kmsg. 5199 on - unlimited logging to /dev/kmsg from userspace 5200 off - logging to /dev/kmsg disabled 5201 ratelimit - ratelimit the logging 5202 Default: ratelimit 5203 5204 printk.time= Show timing data prefixed to each printk message line 5205 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5206 5207 proc_mem.force_override= [KNL] 5208 Format: {always | ptrace | never} 5209 Traditionally /proc/pid/mem allows memory permissions to be 5210 overridden without restrictions. This option may be set to 5211 restrict that. Can be one of: 5212 - 'always': traditional behavior always allows mem overrides. 5213 - 'ptrace': only allow mem overrides for active ptracers. 5214 - 'never': never allow mem overrides. 5215 If not specified, default is the CONFIG_PROC_MEM_* choice. 5216 5217 processor.max_cstate= [HW,ACPI] 5218 Limit processor to maximum C-state 5219 max_cstate=9 overrides any DMI blacklist limit. 5220 5221 processor.nocst [HW,ACPI] 5222 Ignore the _CST method to determine C-states, 5223 instead using the legacy FADT method 5224 5225 profile= [KNL] Enable kernel profiling via /proc/profile 5226 Format: [<profiletype>,]<number> 5227 Param: <profiletype>: "schedule" or "kvm" 5228 [defaults to kernel profiling] 5229 Param: "schedule" - profile schedule points. 5230 Param: "kvm" - profile VM exits. 5231 Param: <number> - step/bucket size as a power of 2 for 5232 statistical time based profiling. 5233 5234 prompt_ramdisk= [RAM] [Deprecated] 5235 5236 prot_virt= [S390] enable hosting protected virtual machines 5237 isolated from the hypervisor (if hardware supports 5238 that). If enabled, the default kernel base address 5239 might be overridden even when Kernel Address Space 5240 Layout Randomization is disabled. 5241 Format: <bool> 5242 5243 psi= [KNL] Enable or disable pressure stall information 5244 tracking. 5245 Format: <bool> 5246 5247 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to 5248 probe for; one of (bare|imps|exps|lifebook|any). 5249 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports 5250 per second. 5251 psmouse.resetafter= [HW,MOUSE] 5252 Try to reset the device after so many bad packets 5253 (0 = never). 5254 psmouse.resolution= 5255 [HW,MOUSE] Set desired mouse resolution, in dpi. 5256 psmouse.smartscroll= 5257 [HW,MOUSE] Controls Logitech smartscroll autorepeat. 5258 0 = disabled, 1 = enabled (default). 5259 5260 pstore.backend= Specify the name of the pstore backend to use 5261 5262 pti= [X86-64] Control Page Table Isolation of user and 5263 kernel address spaces. Disabling this feature 5264 removes hardening, but improves performance of 5265 system calls and interrupts. 5266 5267 on - unconditionally enable 5268 off - unconditionally disable 5269 auto - kernel detects whether your CPU model is 5270 vulnerable to issues that PTI mitigates 5271 5272 Not specifying this option is equivalent to pti=auto. 5273 5274 pty.legacy_count= 5275 [KNL] Number of legacy pty's. Overwrites compiled-in 5276 default number. 5277 5278 quiet [KNL,EARLY] Disable most log messages 5279 5280 r128= [HW,DRM] 5281 5282 radix_hcall_invalidate=on [PPC/PSERIES] 5283 Disable RADIX GTSE feature and use hcall for TLB 5284 invalidate. 5285 5286 raid= [HW,RAID] 5287 See Documentation/admin-guide/md.rst. 5288 5289 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes 5290 See Documentation/admin-guide/blockdev/ramdisk.rst. 5291 5292 ramdisk_start= [RAM] RAM disk image start address 5293 5294 random.trust_cpu=off 5295 [KNL,EARLY] Disable trusting the use of the CPU's 5296 random number generator (if available) to 5297 initialize the kernel's RNG. 5298 5299 random.trust_bootloader=off 5300 [KNL,EARLY] Disable trusting the use of the a seed 5301 passed by the bootloader (if available) to 5302 initialize the kernel's RNG. 5303 5304 randomize_kstack_offset= 5305 [KNL,EARLY] Enable or disable kernel stack offset 5306 randomization, which provides roughly 5 bits of 5307 entropy, frustrating memory corruption attacks 5308 that depend on stack address determinism or 5309 cross-syscall address exposures. This is only 5310 available on architectures that have defined 5311 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET. 5312 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5313 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT. 5314 5315 ras=option[,option,...] [KNL] RAS-specific options 5316 5317 cec_disable [X86] 5318 Disable the Correctable Errors Collector, 5319 see CONFIG_RAS_CEC help text. 5320 5321 rcu_nocbs[=cpu-list] 5322 [KNL] The optional argument is a cpu list, 5323 as described above. 5324 5325 In kernels built with CONFIG_RCU_NOCB_CPU=y, 5326 enable the no-callback CPU mode, which prevents 5327 such CPUs' callbacks from being invoked in 5328 softirq context. Invocation of such CPUs' RCU 5329 callbacks will instead be offloaded to "rcuox/N" 5330 kthreads created for that purpose, where "x" is 5331 "p" for RCU-preempt, "s" for RCU-sched, and "g" 5332 for the kthreads that mediate grace periods; and 5333 "N" is the CPU number. This reduces OS jitter on 5334 the offloaded CPUs, which can be useful for HPC 5335 and real-time workloads. It can also improve 5336 energy efficiency for asymmetric multiprocessors. 5337 5338 If a cpulist is passed as an argument, the specified 5339 list of CPUs is set to no-callback mode from boot. 5340 5341 Otherwise, if the '=' sign and the cpulist 5342 arguments are omitted, no CPU will be set to 5343 no-callback mode from boot but the mode may be 5344 toggled at runtime via cpusets. 5345 5346 Note that this argument takes precedence over 5347 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option. 5348 5349 rcu_nocb_poll [KNL] 5350 Rather than requiring that offloaded CPUs 5351 (specified by rcu_nocbs= above) explicitly 5352 awaken the corresponding "rcuoN" kthreads, 5353 make these kthreads poll for callbacks. 5354 This improves the real-time response for the 5355 offloaded CPUs by relieving them of the need to 5356 wake up the corresponding kthread, but degrades 5357 energy efficiency by requiring that the kthreads 5358 periodically wake up to do the polling. 5359 5360 rcutree.blimit= [KNL] 5361 Set maximum number of finished RCU callbacks to 5362 process in one batch. 5363 5364 rcutree.csd_lock_suppress_rcu_stall= [KNL] 5365 Do only a one-line RCU CPU stall warning when 5366 there is an ongoing too-long CSD-lock wait. 5367 5368 rcutree.do_rcu_barrier= [KNL] 5369 Request a call to rcu_barrier(). This is 5370 throttled so that userspace tests can safely 5371 hammer on the sysfs variable if they so choose. 5372 If triggered before the RCU grace-period machinery 5373 is fully active, this will error out with EAGAIN. 5374 5375 rcutree.dump_tree= [KNL] 5376 Dump the structure of the rcu_node combining tree 5377 out at early boot. This is used for diagnostic 5378 purposes, to verify correct tree setup. 5379 5380 rcutree.gp_cleanup_delay= [KNL] 5381 Set the number of jiffies to delay each step of 5382 RCU grace-period cleanup. 5383 5384 rcutree.gp_init_delay= [KNL] 5385 Set the number of jiffies to delay each step of 5386 RCU grace-period initialization. 5387 5388 rcutree.gp_preinit_delay= [KNL] 5389 Set the number of jiffies to delay each step of 5390 RCU grace-period pre-initialization, that is, 5391 the propagation of recent CPU-hotplug changes up 5392 the rcu_node combining tree. 5393 5394 rcutree.jiffies_till_first_fqs= [KNL] 5395 Set delay from grace-period initialization to 5396 first attempt to force quiescent states. 5397 Units are jiffies, minimum value is zero, 5398 and maximum value is HZ. 5399 5400 rcutree.jiffies_till_next_fqs= [KNL] 5401 Set delay between subsequent attempts to force 5402 quiescent states. Units are jiffies, minimum 5403 value is one, and maximum value is HZ. 5404 5405 rcutree.jiffies_till_sched_qs= [KNL] 5406 Set required age in jiffies for a 5407 given grace period before RCU starts 5408 soliciting quiescent-state help from 5409 rcu_note_context_switch() and cond_resched(). 5410 If not specified, the kernel will calculate 5411 a value based on the most recent settings 5412 of rcutree.jiffies_till_first_fqs 5413 and rcutree.jiffies_till_next_fqs. 5414 This calculated value may be viewed in 5415 rcutree.jiffies_to_sched_qs. Any attempt to set 5416 rcutree.jiffies_to_sched_qs will be cheerfully 5417 overwritten. 5418 5419 rcutree.kthread_prio= [KNL,BOOT] 5420 Set the SCHED_FIFO priority of the RCU per-CPU 5421 kthreads (rcuc/N). This value is also used for 5422 the priority of the RCU boost threads (rcub/N) 5423 and for the RCU grace-period kthreads (rcu_bh, 5424 rcu_preempt, and rcu_sched). If RCU_BOOST is 5425 set, valid values are 1-99 and the default is 1 5426 (the least-favored priority). Otherwise, when 5427 RCU_BOOST is not set, valid values are 0-99 and 5428 the default is zero (non-realtime operation). 5429 When RCU_NOCB_CPU is set, also adjust the 5430 priority of NOCB callback kthreads. 5431 5432 rcutree.nocb_nobypass_lim_per_jiffy= [KNL] 5433 On callback-offloaded (rcu_nocbs) CPUs, 5434 RCU reduces the lock contention that would 5435 otherwise be caused by callback floods through 5436 use of the ->nocb_bypass list. However, in the 5437 common non-flooded case, RCU queues directly to 5438 the main ->cblist in order to avoid the extra 5439 overhead of the ->nocb_bypass list and its lock. 5440 But if there are too many callbacks queued during 5441 a single jiffy, RCU pre-queues the callbacks into 5442 the ->nocb_bypass queue. The definition of "too 5443 many" is supplied by this kernel boot parameter. 5444 5445 rcutree.nohz_full_patience_delay= [KNL] 5446 On callback-offloaded (rcu_nocbs) CPUs, avoid 5447 disturbing RCU unless the grace period has 5448 reached the specified age in milliseconds. 5449 Defaults to zero. Large values will be capped 5450 at five seconds. All values will be rounded down 5451 to the nearest value representable by jiffies. 5452 5453 rcutree.qhimark= [KNL] 5454 Set threshold of queued RCU callbacks beyond which 5455 batch limiting is disabled. 5456 5457 rcutree.qlowmark= [KNL] 5458 Set threshold of queued RCU callbacks below which 5459 batch limiting is re-enabled. 5460 5461 rcutree.qovld= [KNL] 5462 Set threshold of queued RCU callbacks beyond which 5463 RCU's force-quiescent-state scan will aggressively 5464 enlist help from cond_resched() and sched IPIs to 5465 help CPUs more quickly reach quiescent states. 5466 Set to less than zero to make this be set based 5467 on rcutree.qhimark at boot time and to zero to 5468 disable more aggressive help enlistment. 5469 5470 rcutree.rcu_delay_page_cache_fill_msec= [KNL] 5471 Set the page-cache refill delay (in milliseconds) 5472 in response to low-memory conditions. The range 5473 of permitted values is in the range 0:100000. 5474 5475 rcutree.rcu_divisor= [KNL] 5476 Set the shift-right count to use to compute 5477 the callback-invocation batch limit bl from 5478 the number of callbacks queued on this CPU. 5479 The result will be bounded below by the value of 5480 the rcutree.blimit kernel parameter. Every bl 5481 callbacks, the softirq handler will exit in 5482 order to allow the CPU to do other work. 5483 5484 Please note that this callback-invocation batch 5485 limit applies only to non-offloaded callback 5486 invocation. Offloaded callbacks are instead 5487 invoked in the context of an rcuoc kthread, which 5488 scheduler will preempt as it does any other task. 5489 5490 rcutree.rcu_fanout_exact= [KNL] 5491 Disable autobalancing of the rcu_node combining 5492 tree. This is used by rcutorture, and might 5493 possibly be useful for architectures having high 5494 cache-to-cache transfer latencies. 5495 5496 rcutree.rcu_fanout_leaf= [KNL] 5497 Change the number of CPUs assigned to each 5498 leaf rcu_node structure. Useful for very 5499 large systems, which will choose the value 64, 5500 and for NUMA systems with large remote-access 5501 latencies, which will choose a value aligned 5502 with the appropriate hardware boundaries. 5503 5504 rcutree.rcu_min_cached_objs= [KNL] 5505 Minimum number of objects which are cached and 5506 maintained per one CPU. Object size is equal 5507 to PAGE_SIZE. The cache allows to reduce the 5508 pressure to page allocator, also it makes the 5509 whole algorithm to behave better in low memory 5510 condition. 5511 5512 rcutree.rcu_nocb_gp_stride= [KNL] 5513 Set the number of NOCB callback kthreads in 5514 each group, which defaults to the square root 5515 of the number of CPUs. Larger numbers reduce 5516 the wakeup overhead on the global grace-period 5517 kthread, but increases that same overhead on 5518 each group's NOCB grace-period kthread. 5519 5520 rcutree.rcu_kick_kthreads= [KNL] 5521 Cause the grace-period kthread to get an extra 5522 wake_up() if it sleeps three times longer than 5523 it should at force-quiescent-state time. 5524 This wake_up() will be accompanied by a 5525 WARN_ONCE() splat and an ftrace_dump(). 5526 5527 rcutree.rcu_resched_ns= [KNL] 5528 Limit the time spend invoking a batch of RCU 5529 callbacks to the specified number of nanoseconds. 5530 By default, this limit is checked only once 5531 every 32 callbacks in order to limit the pain 5532 inflicted by local_clock() overhead. 5533 5534 rcutree.rcu_unlock_delay= [KNL] 5535 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels, 5536 this specifies an rcu_read_unlock()-time delay 5537 in microseconds. This defaults to zero. 5538 Larger delays increase the probability of 5539 catching RCU pointer leaks, that is, buggy use 5540 of RCU-protected pointers after the relevant 5541 rcu_read_unlock() has completed. 5542 5543 rcutree.sysrq_rcu= [KNL] 5544 Commandeer a sysrq key to dump out Tree RCU's 5545 rcu_node tree with an eye towards determining 5546 why a new grace period has not yet started. 5547 5548 rcutree.use_softirq= [KNL] 5549 If set to zero, move all RCU_SOFTIRQ processing to 5550 per-CPU rcuc kthreads. Defaults to a non-zero 5551 value, meaning that RCU_SOFTIRQ is used by default. 5552 Specify rcutree.use_softirq=0 to use rcuc kthreads. 5553 5554 But note that CONFIG_PREEMPT_RT=y kernels disable 5555 this kernel boot parameter, forcibly setting it 5556 to zero. 5557 5558 rcutree.enable_rcu_lazy= [KNL] 5559 To save power, batch RCU callbacks and flush after 5560 delay, memory pressure or callback list growing too 5561 big. 5562 5563 rcutree.rcu_normal_wake_from_gp= [KNL] 5564 Reduces a latency of synchronize_rcu() call. This approach 5565 maintains its own track of synchronize_rcu() callers, so it 5566 does not interact with regular callbacks because it does not 5567 use a call_rcu[_hurry]() path. Please note, this is for a 5568 normal grace period. 5569 5570 How to enable it: 5571 5572 echo 1 > /sys/module/rcutree/parameters/rcu_normal_wake_from_gp 5573 or pass a boot parameter "rcutree.rcu_normal_wake_from_gp=1" 5574 5575 Default is 1 if num_possible_cpus() <= 16 and it is not explicitly 5576 disabled by the boot parameter passing 0. 5577 5578 rcuscale.gp_async= [KNL] 5579 Measure performance of asynchronous 5580 grace-period primitives such as call_rcu(). 5581 5582 rcuscale.gp_async_max= [KNL] 5583 Specify the maximum number of outstanding 5584 callbacks per writer thread. When a writer 5585 thread exceeds this limit, it invokes the 5586 corresponding flavor of rcu_barrier() to allow 5587 previously posted callbacks to drain. 5588 5589 rcuscale.gp_exp= [KNL] 5590 Measure performance of expedited synchronous 5591 grace-period primitives. 5592 5593 rcuscale.holdoff= [KNL] 5594 Set test-start holdoff period. The purpose of 5595 this parameter is to delay the start of the 5596 test until boot completes in order to avoid 5597 interference. 5598 5599 rcuscale.kfree_by_call_rcu= [KNL] 5600 In kernels built with CONFIG_RCU_LAZY=y, test 5601 call_rcu() instead of kfree_rcu(). 5602 5603 rcuscale.kfree_mult= [KNL] 5604 Instead of allocating an object of size kfree_obj, 5605 allocate one of kfree_mult * sizeof(kfree_obj). 5606 Defaults to 1. 5607 5608 rcuscale.kfree_rcu_test= [KNL] 5609 Set to measure performance of kfree_rcu() flooding. 5610 5611 rcuscale.kfree_rcu_test_double= [KNL] 5612 Test the double-argument variant of kfree_rcu(). 5613 If this parameter has the same value as 5614 rcuscale.kfree_rcu_test_single, both the single- 5615 and double-argument variants are tested. 5616 5617 rcuscale.kfree_rcu_test_single= [KNL] 5618 Test the single-argument variant of kfree_rcu(). 5619 If this parameter has the same value as 5620 rcuscale.kfree_rcu_test_double, both the single- 5621 and double-argument variants are tested. 5622 5623 rcuscale.kfree_nthreads= [KNL] 5624 The number of threads running loops of kfree_rcu(). 5625 5626 rcuscale.kfree_alloc_num= [KNL] 5627 Number of allocations and frees done in an iteration. 5628 5629 rcuscale.kfree_loops= [KNL] 5630 Number of loops doing rcuscale.kfree_alloc_num number 5631 of allocations and frees. 5632 5633 rcuscale.minruntime= [KNL] 5634 Set the minimum test run time in seconds. This 5635 does not affect the data-collection interval, 5636 but instead allows better measurement of things 5637 like CPU consumption. 5638 5639 rcuscale.nreaders= [KNL] 5640 Set number of RCU readers. The value -1 selects 5641 N, where N is the number of CPUs. A value 5642 "n" less than -1 selects N-n+1, where N is again 5643 the number of CPUs. For example, -2 selects N 5644 (the number of CPUs), -3 selects N+1, and so on. 5645 A value of "n" less than or equal to -N selects 5646 a single reader. 5647 5648 rcuscale.nwriters= [KNL] 5649 Set number of RCU writers. The values operate 5650 the same as for rcuscale.nreaders. 5651 N, where N is the number of CPUs 5652 5653 rcuscale.scale_type= [KNL] 5654 Specify the RCU implementation to test. 5655 5656 rcuscale.shutdown= [KNL] 5657 Shut the system down after performance tests 5658 complete. This is useful for hands-off automated 5659 testing. 5660 5661 rcuscale.verbose= [KNL] 5662 Enable additional printk() statements. 5663 5664 rcuscale.writer_holdoff= [KNL] 5665 Write-side holdoff between grace periods, 5666 in microseconds. The default of zero says 5667 no holdoff. 5668 5669 rcuscale.writer_holdoff_jiffies= [KNL] 5670 Additional write-side holdoff between grace 5671 periods, but in jiffies. The default of zero 5672 says no holdoff. 5673 5674 rcutorture.fqs_duration= [KNL] 5675 Set duration of force_quiescent_state bursts 5676 in microseconds. 5677 5678 rcutorture.fqs_holdoff= [KNL] 5679 Set holdoff time within force_quiescent_state bursts 5680 in microseconds. 5681 5682 rcutorture.fqs_stutter= [KNL] 5683 Set wait time between force_quiescent_state bursts 5684 in seconds. 5685 5686 rcutorture.fwd_progress= [KNL] 5687 Specifies the number of kthreads to be used 5688 for RCU grace-period forward-progress testing 5689 for the types of RCU supporting this notion. 5690 Defaults to 1 kthread, values less than zero or 5691 greater than the number of CPUs cause the number 5692 of CPUs to be used. 5693 5694 rcutorture.fwd_progress_div= [KNL] 5695 Specify the fraction of a CPU-stall-warning 5696 period to do tight-loop forward-progress testing. 5697 5698 rcutorture.fwd_progress_holdoff= [KNL] 5699 Number of seconds to wait between successive 5700 forward-progress tests. 5701 5702 rcutorture.fwd_progress_need_resched= [KNL] 5703 Enclose cond_resched() calls within checks for 5704 need_resched() during tight-loop forward-progress 5705 testing. 5706 5707 rcutorture.gp_cond= [KNL] 5708 Use conditional/asynchronous update-side 5709 normal-grace-period primitives, if available. 5710 5711 rcutorture.gp_cond_exp= [KNL] 5712 Use conditional/asynchronous update-side 5713 expedited-grace-period primitives, if available. 5714 5715 rcutorture.gp_cond_full= [KNL] 5716 Use conditional/asynchronous update-side 5717 normal-grace-period primitives that also take 5718 concurrent expedited grace periods into account, 5719 if available. 5720 5721 rcutorture.gp_cond_exp_full= [KNL] 5722 Use conditional/asynchronous update-side 5723 expedited-grace-period primitives that also take 5724 concurrent normal grace periods into account, 5725 if available. 5726 5727 rcutorture.gp_cond_wi= [KNL] 5728 Nominal wait interval for normal conditional 5729 grace periods (specified by rcutorture's 5730 gp_cond and gp_cond_full module parameters), 5731 in microseconds. The actual wait interval will 5732 be randomly selected to nanosecond granularity up 5733 to this wait interval. Defaults to 16 jiffies, 5734 for example, 16,000 microseconds on a system 5735 with HZ=1000. 5736 5737 rcutorture.gp_cond_wi_exp= [KNL] 5738 Nominal wait interval for expedited conditional 5739 grace periods (specified by rcutorture's 5740 gp_cond_exp and gp_cond_exp_full module 5741 parameters), in microseconds. The actual wait 5742 interval will be randomly selected to nanosecond 5743 granularity up to this wait interval. Defaults to 5744 128 microseconds. 5745 5746 rcutorture.gp_exp= [KNL] 5747 Use expedited update-side primitives, if available. 5748 5749 rcutorture.gp_normal= [KNL] 5750 Use normal (non-expedited) asynchronous 5751 update-side primitives, if available. 5752 5753 rcutorture.gp_poll= [KNL] 5754 Use polled update-side normal-grace-period 5755 primitives, if available. 5756 5757 rcutorture.gp_poll_exp= [KNL] 5758 Use polled update-side expedited-grace-period 5759 primitives, if available. 5760 5761 rcutorture.gp_poll_full= [KNL] 5762 Use polled update-side normal-grace-period 5763 primitives that also take concurrent expedited 5764 grace periods into account, if available. 5765 5766 rcutorture.gp_poll_exp_full= [KNL] 5767 Use polled update-side expedited-grace-period 5768 primitives that also take concurrent normal 5769 grace periods into account, if available. 5770 5771 rcutorture.gp_poll_wi= [KNL] 5772 Nominal wait interval for normal conditional 5773 grace periods (specified by rcutorture's 5774 gp_poll and gp_poll_full module parameters), 5775 in microseconds. The actual wait interval will 5776 be randomly selected to nanosecond granularity up 5777 to this wait interval. Defaults to 16 jiffies, 5778 for example, 16,000 microseconds on a system 5779 with HZ=1000. 5780 5781 rcutorture.gp_poll_wi_exp= [KNL] 5782 Nominal wait interval for expedited conditional 5783 grace periods (specified by rcutorture's 5784 gp_poll_exp and gp_poll_exp_full module 5785 parameters), in microseconds. The actual wait 5786 interval will be randomly selected to nanosecond 5787 granularity up to this wait interval. Defaults to 5788 128 microseconds. 5789 5790 rcutorture.gp_sync= [KNL] 5791 Use normal (non-expedited) synchronous 5792 update-side primitives, if available. If all 5793 of rcutorture.gp_cond=, rcutorture.gp_exp=, 5794 rcutorture.gp_normal=, and rcutorture.gp_sync= 5795 are zero, rcutorture acts as if is interpreted 5796 they are all non-zero. 5797 5798 rcutorture.gpwrap_lag= [KNL] 5799 Enable grace-period wrap lag testing. Setting 5800 to false prevents the gpwrap lag test from 5801 running. Default is true. 5802 5803 rcutorture.gpwrap_lag_gps= [KNL] 5804 Set the value for grace-period wrap lag during 5805 active lag testing periods. This controls how many 5806 grace periods differences we tolerate between 5807 rdp and rnp's gp_seq before setting overflow flag. 5808 The default is always set to 8. 5809 5810 rcutorture.gpwrap_lag_cycle_mins= [KNL] 5811 Set the total cycle duration for gpwrap lag 5812 testing in minutes. This is the total time for 5813 one complete cycle of active and inactive 5814 testing periods. Default is 30 minutes. 5815 5816 rcutorture.gpwrap_lag_active_mins= [KNL] 5817 Set the duration for which gpwrap lag is active 5818 within each cycle, in minutes. During this time, 5819 the grace-period wrap lag will be set to the 5820 value specified by gpwrap_lag_gps. Default is 5821 5 minutes. 5822 5823 rcutorture.irqreader= [KNL] 5824 Run RCU readers from irq handlers, or, more 5825 accurately, from a timer handler. Not all RCU 5826 flavors take kindly to this sort of thing. 5827 5828 rcutorture.leakpointer= [KNL] 5829 Leak an RCU-protected pointer out of the reader. 5830 This can of course result in splats, and is 5831 intended to test the ability of things like 5832 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect 5833 such leaks. 5834 5835 rcutorture.n_barrier_cbs= [KNL] 5836 Set callbacks/threads for rcu_barrier() testing. 5837 5838 rcutorture.nfakewriters= [KNL] 5839 Set number of concurrent RCU writers. These just 5840 stress RCU, they don't participate in the actual 5841 test, hence the "fake". 5842 5843 rcutorture.nocbs_nthreads= [KNL] 5844 Set number of RCU callback-offload togglers. 5845 Zero (the default) disables toggling. 5846 5847 rcutorture.nocbs_toggle= [KNL] 5848 Set the delay in milliseconds between successive 5849 callback-offload toggling attempts. 5850 5851 rcutorture.nreaders= [KNL] 5852 Set number of RCU readers. The value -1 selects 5853 N-1, where N is the number of CPUs. A value 5854 "n" less than -1 selects N-n-2, where N is again 5855 the number of CPUs. For example, -2 selects N 5856 (the number of CPUs), -3 selects N+1, and so on. 5857 5858 rcutorture.object_debug= [KNL] 5859 Enable debug-object double-call_rcu() testing. 5860 5861 rcutorture.onoff_holdoff= [KNL] 5862 Set time (s) after boot for CPU-hotplug testing. 5863 5864 rcutorture.onoff_interval= [KNL] 5865 Set time (jiffies) between CPU-hotplug operations, 5866 or zero to disable CPU-hotplug testing. 5867 5868 rcutorture.preempt_duration= [KNL] 5869 Set duration (in milliseconds) of preemptions 5870 by a high-priority FIFO real-time task. Set to 5871 zero (the default) to disable. The CPUs to 5872 preempt are selected randomly from the set that 5873 are online at a given point in time. Races with 5874 CPUs going offline are ignored, with that attempt 5875 at preemption skipped. 5876 5877 rcutorture.preempt_interval= [KNL] 5878 Set interval (in milliseconds, defaulting to one 5879 second) between preemptions by a high-priority 5880 FIFO real-time task. This delay is mediated 5881 by an hrtimer and is further fuzzed to avoid 5882 inadvertent synchronizations. 5883 5884 rcutorture.read_exit_burst= [KNL] 5885 The number of times in a given read-then-exit 5886 episode that a set of read-then-exit kthreads 5887 is spawned. 5888 5889 rcutorture.read_exit_delay= [KNL] 5890 The delay, in seconds, between successive 5891 read-then-exit testing episodes. 5892 5893 rcutorture.reader_flavor= [KNL] 5894 A bit mask indicating which readers to use. 5895 If there is more than one bit set, the readers 5896 are entered from low-order bit up, and are 5897 exited in the opposite order. For SRCU, the 5898 0x1 bit is normal readers, 0x2 NMI-safe readers, 5899 and 0x4 light-weight readers. 5900 5901 rcutorture.shuffle_interval= [KNL] 5902 Set task-shuffle interval (s). Shuffling tasks 5903 allows some CPUs to go into dyntick-idle mode 5904 during the rcutorture test. 5905 5906 rcutorture.shutdown_secs= [KNL] 5907 Set time (s) after boot system shutdown. This 5908 is useful for hands-off automated testing. 5909 5910 rcutorture.stall_cpu= [KNL] 5911 Duration of CPU stall (s) to test RCU CPU stall 5912 warnings, zero to disable. 5913 5914 rcutorture.stall_cpu_block= [KNL] 5915 Sleep while stalling if set. This will result 5916 in warnings from preemptible RCU in addition to 5917 any other stall-related activity. Note that 5918 in kernels built with CONFIG_PREEMPTION=n and 5919 CONFIG_PREEMPT_COUNT=y, this parameter will 5920 cause the CPU to pass through a quiescent state. 5921 Given CONFIG_PREEMPTION=n, this will suppress 5922 RCU CPU stall warnings, but will instead result 5923 in scheduling-while-atomic splats. 5924 5925 Use of this module parameter results in splats. 5926 5927 5928 rcutorture.stall_cpu_holdoff= [KNL] 5929 Time to wait (s) after boot before inducing stall. 5930 5931 rcutorture.stall_cpu_irqsoff= [KNL] 5932 Disable interrupts while stalling if set, but only 5933 on the first stall in the set. 5934 5935 rcutorture.stall_cpu_repeat= [KNL] 5936 Number of times to repeat the stall sequence, 5937 so that rcutorture.stall_cpu_repeat=3 will result 5938 in four stall sequences. 5939 5940 rcutorture.stall_gp_kthread= [KNL] 5941 Duration (s) of forced sleep within RCU 5942 grace-period kthread to test RCU CPU stall 5943 warnings, zero to disable. If both stall_cpu 5944 and stall_gp_kthread are specified, the 5945 kthread is starved first, then the CPU. 5946 5947 rcutorture.stat_interval= [KNL] 5948 Time (s) between statistics printk()s. 5949 5950 rcutorture.stutter= [KNL] 5951 Time (s) to stutter testing, for example, specifying 5952 five seconds causes the test to run for five seconds, 5953 wait for five seconds, and so on. This tests RCU's 5954 ability to transition abruptly to and from idle. 5955 5956 rcutorture.test_boost= [KNL] 5957 Test RCU priority boosting? 0=no, 1=maybe, 2=yes. 5958 "Maybe" means test if the RCU implementation 5959 under test support RCU priority boosting. 5960 5961 rcutorture.test_boost_duration= [KNL] 5962 Duration (s) of each individual boost test. 5963 5964 rcutorture.test_boost_holdoff= [KNL] 5965 Holdoff time (s) from start of test to the start 5966 of RCU priority-boost testing. Defaults to zero, 5967 that is, no holdoff. 5968 5969 rcutorture.test_boost_interval= [KNL] 5970 Interval (s) between each boost test. 5971 5972 rcutorture.test_no_idle_hz= [KNL] 5973 Test RCU's dyntick-idle handling. See also the 5974 rcutorture.shuffle_interval parameter. 5975 5976 rcutorture.torture_type= [KNL] 5977 Specify the RCU implementation to test. 5978 5979 rcutorture.verbose= [KNL] 5980 Enable additional printk() statements. 5981 5982 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL] 5983 Dump ftrace buffer after reporting RCU CPU 5984 stall warning. 5985 5986 rcupdate.rcu_cpu_stall_notifiers= [KNL] 5987 Provide RCU CPU stall notifiers, but see the 5988 warnings in the RCU_CPU_STALL_NOTIFIER Kconfig 5989 option's help text. TL;DR: You almost certainly 5990 do not want rcupdate.rcu_cpu_stall_notifiers. 5991 5992 rcupdate.rcu_cpu_stall_suppress= [KNL] 5993 Suppress RCU CPU stall warning messages. 5994 5995 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL] 5996 Suppress RCU CPU stall warning messages and 5997 rcutorture writer stall warnings that occur 5998 during early boot, that is, during the time 5999 before the init task is spawned. 6000 6001 rcupdate.rcu_cpu_stall_timeout= [KNL] 6002 Set timeout for RCU CPU stall warning messages. 6003 The value is in seconds and the maximum allowed 6004 value is 300 seconds. 6005 6006 rcupdate.rcu_exp_cpu_stall_timeout= [KNL] 6007 Set timeout for expedited RCU CPU stall warning 6008 messages. The value is in milliseconds 6009 and the maximum allowed value is 21000 6010 milliseconds. Please note that this value is 6011 adjusted to an arch timer tick resolution. 6012 Setting this to zero causes the value from 6013 rcupdate.rcu_cpu_stall_timeout to be used (after 6014 conversion from seconds to milliseconds). 6015 6016 rcupdate.rcu_cpu_stall_cputime= [KNL] 6017 Provide statistics on the cputime and count of 6018 interrupts and tasks during the sampling period. For 6019 multiple continuous RCU stalls, all sampling periods 6020 begin at half of the first RCU stall timeout. 6021 6022 rcupdate.rcu_exp_stall_task_details= [KNL] 6023 Print stack dumps of any tasks blocking the 6024 current expedited RCU grace period during an 6025 expedited RCU CPU stall warning. 6026 6027 rcupdate.rcu_expedited= [KNL] 6028 Use expedited grace-period primitives, for 6029 example, synchronize_rcu_expedited() instead 6030 of synchronize_rcu(). This reduces latency, 6031 but can increase CPU utilization, degrade 6032 real-time latency, and degrade energy efficiency. 6033 No effect on CONFIG_TINY_RCU kernels. 6034 6035 rcupdate.rcu_normal= [KNL] 6036 Use only normal grace-period primitives, 6037 for example, synchronize_rcu() instead of 6038 synchronize_rcu_expedited(). This improves 6039 real-time latency, CPU utilization, and 6040 energy efficiency, but can expose users to 6041 increased grace-period latency. This parameter 6042 overrides rcupdate.rcu_expedited. No effect on 6043 CONFIG_TINY_RCU kernels. 6044 6045 rcupdate.rcu_normal_after_boot= [KNL] 6046 Once boot has completed (that is, after 6047 rcu_end_inkernel_boot() has been invoked), use 6048 only normal grace-period primitives. No effect 6049 on CONFIG_TINY_RCU kernels. 6050 6051 But note that CONFIG_PREEMPT_RT=y kernels enables 6052 this kernel boot parameter, forcibly setting 6053 it to the value one, that is, converting any 6054 post-boot attempt at an expedited RCU grace 6055 period to instead use normal non-expedited 6056 grace-period processing. 6057 6058 rcupdate.rcu_task_collapse_lim= [KNL] 6059 Set the maximum number of callbacks present 6060 at the beginning of a grace period that allows 6061 the RCU Tasks flavors to collapse back to using 6062 a single callback queue. This switching only 6063 occurs when rcupdate.rcu_task_enqueue_lim is 6064 set to the default value of -1. 6065 6066 rcupdate.rcu_task_contend_lim= [KNL] 6067 Set the minimum number of callback-queuing-time 6068 lock-contention events per jiffy required to 6069 cause the RCU Tasks flavors to switch to per-CPU 6070 callback queuing. This switching only occurs 6071 when rcupdate.rcu_task_enqueue_lim is set to 6072 the default value of -1. 6073 6074 rcupdate.rcu_task_enqueue_lim= [KNL] 6075 Set the number of callback queues to use for the 6076 RCU Tasks family of RCU flavors. The default 6077 of -1 allows this to be automatically (and 6078 dynamically) adjusted. This parameter is intended 6079 for use in testing. 6080 6081 rcupdate.rcu_task_ipi_delay= [KNL] 6082 Set time in jiffies during which RCU tasks will 6083 avoid sending IPIs, starting with the beginning 6084 of a given grace period. Setting a large 6085 number avoids disturbing real-time workloads, 6086 but lengthens grace periods. 6087 6088 rcupdate.rcu_task_lazy_lim= [KNL] 6089 Number of callbacks on a given CPU that will 6090 cancel laziness on that CPU. Use -1 to disable 6091 cancellation of laziness, but be advised that 6092 doing so increases the danger of OOM due to 6093 callback flooding. 6094 6095 rcupdate.rcu_task_stall_info= [KNL] 6096 Set initial timeout in jiffies for RCU task stall 6097 informational messages, which give some indication 6098 of the problem for those not patient enough to 6099 wait for ten minutes. Informational messages are 6100 only printed prior to the stall-warning message 6101 for a given grace period. Disable with a value 6102 less than or equal to zero. Defaults to ten 6103 seconds. A change in value does not take effect 6104 until the beginning of the next grace period. 6105 6106 rcupdate.rcu_task_stall_info_mult= [KNL] 6107 Multiplier for time interval between successive 6108 RCU task stall informational messages for a given 6109 RCU tasks grace period. This value is clamped 6110 to one through ten, inclusive. It defaults to 6111 the value three, so that the first informational 6112 message is printed 10 seconds into the grace 6113 period, the second at 40 seconds, the third at 6114 160 seconds, and then the stall warning at 600 6115 seconds would prevent a fourth at 640 seconds. 6116 6117 rcupdate.rcu_task_stall_timeout= [KNL] 6118 Set timeout in jiffies for RCU task stall 6119 warning messages. Disable with a value less 6120 than or equal to zero. Defaults to ten minutes. 6121 A change in value does not take effect until 6122 the beginning of the next grace period. 6123 6124 rcupdate.rcu_tasks_lazy_ms= [KNL] 6125 Set timeout in milliseconds RCU Tasks asynchronous 6126 callback batching for call_rcu_tasks(). 6127 A negative value will take the default. A value 6128 of zero will disable batching. Batching is 6129 always disabled for synchronize_rcu_tasks(). 6130 6131 rcupdate.rcu_tasks_trace_lazy_ms= [KNL] 6132 Set timeout in milliseconds RCU Tasks 6133 Trace asynchronous callback batching for 6134 call_rcu_tasks_trace(). A negative value 6135 will take the default. A value of zero will 6136 disable batching. Batching is always disabled 6137 for synchronize_rcu_tasks_trace(). 6138 6139 rcupdate.rcu_self_test= [KNL] 6140 Run the RCU early boot self tests 6141 6142 rdinit= [KNL] 6143 Format: <full_path> 6144 Run specified binary instead of /init from the ramdisk, 6145 used for early userspace startup. See initrd. 6146 6147 rdrand= [X86,EARLY] 6148 force - Override the decision by the kernel to hide the 6149 advertisement of RDRAND support (this affects 6150 certain AMD processors because of buggy BIOS 6151 support, specifically around the suspend/resume 6152 path). 6153 6154 rdt= [HW,X86,RDT] 6155 Turn on/off individual RDT features. List is: 6156 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp, 6157 mba, smba, bmec. 6158 E.g. to turn on cmt and turn off mba use: 6159 rdt=cmt,!mba 6160 6161 reboot= [KNL] 6162 Format (x86 or x86_64): 6163 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] | d[efault] \ 6164 [[,]s[mp]#### \ 6165 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \ 6166 [[,]f[orce] 6167 Where reboot_mode is one of warm (soft) or cold (hard) or gpio 6168 (prefix with 'panic_' to set mode for panic 6169 reboot only), 6170 reboot_type is one of bios, acpi, kbd, triple, efi, or pci, 6171 reboot_force is either force or not specified, 6172 reboot_cpu is s[mp]#### with #### being the processor 6173 to be used for rebooting. 6174 6175 acpi 6176 Use the ACPI RESET_REG in the FADT. If ACPI is not 6177 configured or the ACPI reset does not work, the reboot 6178 path attempts the reset using the keyboard controller. 6179 6180 bios 6181 Use the CPU reboot vector for warm reset 6182 6183 cold 6184 Set the cold reboot flag 6185 6186 default 6187 There are some built-in platform specific "quirks" 6188 - you may see: "reboot: <name> series board detected. 6189 Selecting <type> for reboots." In the case where you 6190 think the quirk is in error (e.g. you have newer BIOS, 6191 or newer board) using this option will ignore the 6192 built-in quirk table, and use the generic default 6193 reboot actions. 6194 6195 efi 6196 Use efi reset_system runtime service. If EFI is not 6197 configured or the EFI reset does not work, the reboot 6198 path attempts the reset using the keyboard controller. 6199 6200 force 6201 Don't stop other CPUs on reboot. This can make reboot 6202 more reliable in some cases. 6203 6204 kbd 6205 Use the keyboard controller. cold reset (default) 6206 6207 pci 6208 Use a write to the PCI config space register 0xcf9 to 6209 trigger reboot. 6210 6211 triple 6212 Force a triple fault (init) 6213 6214 warm 6215 Don't set the cold reboot flag 6216 6217 Using warm reset will be much faster especially on big 6218 memory systems because the BIOS will not go through 6219 the memory check. Disadvantage is that not all 6220 hardware will be completely reinitialized on reboot so 6221 there may be boot problems on some systems. 6222 6223 6224 refscale.holdoff= [KNL] 6225 Set test-start holdoff period. The purpose of 6226 this parameter is to delay the start of the 6227 test until boot completes in order to avoid 6228 interference. 6229 6230 refscale.lookup_instances= [KNL] 6231 Number of data elements to use for the forms of 6232 SLAB_TYPESAFE_BY_RCU testing. A negative number 6233 is negated and multiplied by nr_cpu_ids, while 6234 zero specifies nr_cpu_ids. 6235 6236 refscale.loops= [KNL] 6237 Set the number of loops over the synchronization 6238 primitive under test. Increasing this number 6239 reduces noise due to loop start/end overhead, 6240 but the default has already reduced the per-pass 6241 noise to a handful of picoseconds on ca. 2020 6242 x86 laptops. 6243 6244 refscale.nreaders= [KNL] 6245 Set number of readers. The default value of -1 6246 selects N, where N is roughly 75% of the number 6247 of CPUs. A value of zero is an interesting choice. 6248 6249 refscale.nruns= [KNL] 6250 Set number of runs, each of which is dumped onto 6251 the console log. 6252 6253 refscale.readdelay= [KNL] 6254 Set the read-side critical-section duration, 6255 measured in microseconds. 6256 6257 refscale.scale_type= [KNL] 6258 Specify the read-protection implementation to test. 6259 6260 refscale.shutdown= [KNL] 6261 Shut down the system at the end of the performance 6262 test. This defaults to 1 (shut it down) when 6263 refscale is built into the kernel and to 0 (leave 6264 it running) when refscale is built as a module. 6265 6266 refscale.verbose= [KNL] 6267 Enable additional printk() statements. 6268 6269 refscale.verbose_batched= [KNL] 6270 Batch the additional printk() statements. If zero 6271 (the default) or negative, print everything. Otherwise, 6272 print every Nth verbose statement, where N is the value 6273 specified. 6274 6275 regulator_ignore_unused 6276 [REGULATOR] 6277 Prevents regulator framework from disabling regulators 6278 that are unused, due no driver claiming them. This may 6279 be useful for debug and development, but should not be 6280 needed on a platform with proper driver support. 6281 6282 relax_domain_level= 6283 [KNL, SMP] Set scheduler's default relax_domain_level. 6284 See Documentation/admin-guide/cgroup-v1/cpusets.rst. 6285 6286 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory 6287 Format: <base1>,<size1>[,<base2>,<size2>,...] 6288 Reserve I/O ports or memory so the kernel won't use 6289 them. If <base> is less than 0x10000, the region 6290 is assumed to be I/O ports; otherwise it is memory. 6291 6292 reserve_mem= [RAM] 6293 Format: nn[KMG]:<align>:<label> 6294 Reserve physical memory and label it with a name that 6295 other subsystems can use to access it. This is typically 6296 used for systems that do not wipe the RAM, and this command 6297 line will try to reserve the same physical memory on 6298 soft reboots. Note, it is not guaranteed to be the same 6299 location. For example, if anything about the system changes 6300 or if booting a different kernel. It can also fail if KASLR 6301 places the kernel at the location of where the RAM reservation 6302 was from a previous boot, the new reservation will be at a 6303 different location. 6304 Any subsystem using this feature must add a way to verify 6305 that the contents of the physical memory is from a previous 6306 boot, as there may be cases where the memory will not be 6307 located at the same location. 6308 6309 The format is size:align:label for example, to request 6310 12 megabytes of 4096 alignment for ramoops: 6311 6312 reserve_mem=12M:4096:oops ramoops.mem_name=oops 6313 6314 reservetop= [X86-32,EARLY] 6315 Format: nn[KMG] 6316 Reserves a hole at the top of the kernel virtual 6317 address space. 6318 6319 reset_devices [KNL] Force drivers to reset the underlying device 6320 during initialization. 6321 6322 resume= [SWSUSP] 6323 Specify the partition device for software suspend 6324 Format: 6325 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>} 6326 6327 resume_offset= [SWSUSP] 6328 Specify the offset from the beginning of the partition 6329 given by "resume=" at which the swap header is located, 6330 in <PAGE_SIZE> units (needed only for swap files). 6331 See Documentation/power/swsusp-and-swap-files.rst 6332 6333 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to 6334 read the resume files 6335 6336 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up. 6337 Useful for devices that are detected asynchronously 6338 (e.g. USB and MMC devices). 6339 6340 retain_initrd [RAM] Keep initrd memory after extraction. After boot, it will 6341 be accessible via /sys/firmware/initrd. 6342 6343 retbleed= [X86] Control mitigation of RETBleed (Arbitrary 6344 Speculative Code Execution with Return Instructions) 6345 vulnerability. 6346 6347 AMD-based UNRET and IBPB mitigations alone do not stop 6348 sibling threads from influencing the predictions of other 6349 sibling threads. For that reason, STIBP is used on pro- 6350 cessors that support it, and mitigate SMT on processors 6351 that don't. 6352 6353 off - no mitigation 6354 auto - automatically select a migitation 6355 auto,nosmt - automatically select a mitigation, 6356 disabling SMT if necessary for 6357 the full mitigation (only on Zen1 6358 and older without STIBP). 6359 ibpb - On AMD, mitigate short speculation 6360 windows on basic block boundaries too. 6361 Safe, highest perf impact. It also 6362 enables STIBP if present. Not suitable 6363 on Intel. 6364 ibpb,nosmt - Like "ibpb" above but will disable SMT 6365 when STIBP is not available. This is 6366 the alternative for systems which do not 6367 have STIBP. 6368 unret - Force enable untrained return thunks, 6369 only effective on AMD f15h-f17h based 6370 systems. 6371 unret,nosmt - Like unret, but will disable SMT when STIBP 6372 is not available. This is the alternative for 6373 systems which do not have STIBP. 6374 6375 Selecting 'auto' will choose a mitigation method at run 6376 time according to the CPU. 6377 6378 Not specifying this option is equivalent to retbleed=auto. 6379 6380 rfkill.default_state= 6381 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm, 6382 etc. communication is blocked by default. 6383 1 Unblocked. 6384 6385 rfkill.master_switch_mode= 6386 0 The "airplane mode" button does nothing. 6387 1 The "airplane mode" button toggles between everything 6388 blocked and the previous configuration. 6389 2 The "airplane mode" button toggles between everything 6390 blocked and everything unblocked. 6391 6392 ring3mwait=disable 6393 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported 6394 CPUs. 6395 6396 riscv_isa_fallback [RISCV,EARLY] 6397 When CONFIG_RISCV_ISA_FALLBACK is not enabled, permit 6398 falling back to detecting extension support by parsing 6399 "riscv,isa" property on devicetree systems when the 6400 replacement properties are not found. See the Kconfig 6401 entry for RISCV_ISA_FALLBACK. 6402 6403 ro [KNL] Mount root device read-only on boot 6404 6405 rodata= [KNL,EARLY] 6406 on Mark read-only kernel memory as read-only (default). 6407 off Leave read-only kernel memory writable for debugging. 6408 full Mark read-only kernel memory and aliases as read-only 6409 [arm64] 6410 6411 rockchip.usb_uart 6412 [EARLY] 6413 Enable the uart passthrough on the designated usb port 6414 on Rockchip SoCs. When active, the signals of the 6415 debug-uart get routed to the D+ and D- pins of the usb 6416 port and the regular usb controller gets disabled. 6417 6418 root= [KNL] Root filesystem 6419 Usually this is a block device specifier of some kind, 6420 see the early_lookup_bdev comment in 6421 block/early-lookup.c for details. 6422 Alternatively this can be "ram" for the legacy initial 6423 ramdisk, "nfs" and "cifs" for root on a network file 6424 system, or "mtd" and "ubi" for mounting from raw flash. 6425 6426 rootdelay= [KNL] Delay (in seconds) to pause before attempting to 6427 mount the root filesystem 6428 6429 rootflags= [KNL] Set root filesystem mount option string 6430 6431 rootfstype= [KNL] Set root filesystem type 6432 6433 rootwait [KNL] Wait (indefinitely) for root device to show up. 6434 Useful for devices that are detected asynchronously 6435 (e.g. USB and MMC devices). 6436 6437 rootwait= [KNL] Maximum time (in seconds) to wait for root device 6438 to show up before attempting to mount the root 6439 filesystem. 6440 6441 rproc_mem=nn[KMG][@address] 6442 [KNL,ARM,CMA] Remoteproc physical memory block. 6443 Memory area to be used by remote processor image, 6444 managed by CMA. 6445 6446 rt_group_sched= [KNL] Enable or disable SCHED_RR/FIFO group scheduling 6447 when CONFIG_RT_GROUP_SCHED=y. Defaults to 6448 !CONFIG_RT_GROUP_SCHED_DEFAULT_DISABLED. 6449 Format: <bool> 6450 6451 rw [KNL] Mount root device read-write on boot 6452 6453 S [KNL] Run init in single mode 6454 6455 s390_iommu= [HW,S390] 6456 Set s390 IOTLB flushing mode 6457 strict 6458 With strict flushing every unmap operation will result 6459 in an IOTLB flush. Default is lazy flushing before 6460 reuse, which is faster. Deprecated, equivalent to 6461 iommu.strict=1. 6462 6463 s390_iommu_aperture= [KNL,S390] 6464 Specifies the size of the per device DMA address space 6465 accessible through the DMA and IOMMU APIs as a decimal 6466 factor of the size of main memory. 6467 The default is 1 meaning that one can concurrently use 6468 as many DMA addresses as physical memory is installed, 6469 if supported by hardware, and thus map all of memory 6470 once. With a value of 2 one can map all of memory twice 6471 and so on. As a special case a factor of 0 imposes no 6472 restrictions other than those given by hardware at the 6473 cost of significant additional memory use for tables. 6474 6475 sa1100ir [NET] 6476 See drivers/net/irda/sa1100_ir.c. 6477 6478 sched_proxy_exec= [KNL] 6479 Enables or disables "proxy execution" style 6480 solution to mutex-based priority inversion. 6481 Format: <bool> 6482 6483 sched_verbose [KNL,EARLY] Enables verbose scheduler debug messages. 6484 6485 schedstats= [KNL,X86] Enable or disable scheduled statistics. 6486 Allowed values are enable and disable. This feature 6487 incurs a small amount of overhead in the scheduler 6488 but is useful for debugging and performance tuning. 6489 6490 sched_thermal_decay_shift= 6491 [Deprecated] 6492 [KNL, SMP] Set a decay shift for scheduler thermal 6493 pressure signal. Thermal pressure signal follows the 6494 default decay period of other scheduler pelt 6495 signals(usually 32 ms but configurable). Setting 6496 sched_thermal_decay_shift will left shift the decay 6497 period for the thermal pressure signal by the shift 6498 value. 6499 i.e. with the default pelt decay period of 32 ms 6500 sched_thermal_decay_shift thermal pressure decay pr 6501 1 64 ms 6502 2 128 ms 6503 and so on. 6504 Format: integer between 0 and 10 6505 Default is 0. 6506 6507 scftorture.holdoff= [KNL] 6508 Number of seconds to hold off before starting 6509 test. Defaults to zero for module insertion and 6510 to 10 seconds for built-in smp_call_function() 6511 tests. 6512 6513 scftorture.longwait= [KNL] 6514 Request ridiculously long waits randomly selected 6515 up to the chosen limit in seconds. Zero (the 6516 default) disables this feature. Please note 6517 that requesting even small non-zero numbers of 6518 seconds can result in RCU CPU stall warnings, 6519 softlockup complaints, and so on. 6520 6521 scftorture.nthreads= [KNL] 6522 Number of kthreads to spawn to invoke the 6523 smp_call_function() family of functions. 6524 The default of -1 specifies a number of kthreads 6525 equal to the number of CPUs. 6526 6527 scftorture.onoff_holdoff= [KNL] 6528 Number seconds to wait after the start of the 6529 test before initiating CPU-hotplug operations. 6530 6531 scftorture.onoff_interval= [KNL] 6532 Number seconds to wait between successive 6533 CPU-hotplug operations. Specifying zero (which 6534 is the default) disables CPU-hotplug operations. 6535 6536 scftorture.shutdown_secs= [KNL] 6537 The number of seconds following the start of the 6538 test after which to shut down the system. The 6539 default of zero avoids shutting down the system. 6540 Non-zero values are useful for automated tests. 6541 6542 scftorture.stat_interval= [KNL] 6543 The number of seconds between outputting the 6544 current test statistics to the console. A value 6545 of zero disables statistics output. 6546 6547 scftorture.stutter_cpus= [KNL] 6548 The number of jiffies to wait between each change 6549 to the set of CPUs under test. 6550 6551 scftorture.use_cpus_read_lock= [KNL] 6552 Use use_cpus_read_lock() instead of the default 6553 preempt_disable() to disable CPU hotplug 6554 while invoking one of the smp_call_function*() 6555 functions. 6556 6557 scftorture.verbose= [KNL] 6558 Enable additional printk() statements. 6559 6560 scftorture.weight_single= [KNL] 6561 The probability weighting to use for the 6562 smp_call_function_single() function with a zero 6563 "wait" parameter. A value of -1 selects the 6564 default if all other weights are -1. However, 6565 if at least one weight has some other value, a 6566 value of -1 will instead select a weight of zero. 6567 6568 scftorture.weight_single_wait= [KNL] 6569 The probability weighting to use for the 6570 smp_call_function_single() function with a 6571 non-zero "wait" parameter. See weight_single. 6572 6573 scftorture.weight_many= [KNL] 6574 The probability weighting to use for the 6575 smp_call_function_many() function with a zero 6576 "wait" parameter. See weight_single. 6577 Note well that setting a high probability for 6578 this weighting can place serious IPI load 6579 on the system. 6580 6581 scftorture.weight_many_wait= [KNL] 6582 The probability weighting to use for the 6583 smp_call_function_many() function with a 6584 non-zero "wait" parameter. See weight_single 6585 and weight_many. 6586 6587 scftorture.weight_all= [KNL] 6588 The probability weighting to use for the 6589 smp_call_function_all() function with a zero 6590 "wait" parameter. See weight_single and 6591 weight_many. 6592 6593 scftorture.weight_all_wait= [KNL] 6594 The probability weighting to use for the 6595 smp_call_function_all() function with a 6596 non-zero "wait" parameter. See weight_single 6597 and weight_many. 6598 6599 sdw_mclk_divider=[SDW] 6600 Specify the MCLK divider for Intel SoundWire buses in 6601 case the BIOS does not provide the clock rate properly. 6602 6603 skew_tick= [KNL,EARLY] Offset the periodic timer tick per cpu to mitigate 6604 xtime_lock contention on larger systems, and/or RCU lock 6605 contention on all systems with CONFIG_MAXSMP set. 6606 Format: { "0" | "1" } 6607 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1" 6608 1 -- enable. 6609 Note: increases power consumption, thus should only be 6610 enabled if running jitter sensitive (HPC/RT) workloads. 6611 6612 security= [SECURITY] Choose a legacy "major" security module to 6613 enable at boot. This has been deprecated by the 6614 "lsm=" parameter. 6615 6616 selinux= [SELINUX] Disable or enable SELinux at boot time. 6617 Format: { "0" | "1" } 6618 See security/selinux/Kconfig help text. 6619 0 -- disable. 6620 1 -- enable. 6621 Default value is 1. 6622 6623 serialnumber [BUGS=X86-32] 6624 6625 sev=option[,option...] [X86-64] 6626 6627 debug 6628 Enable debug messages. 6629 6630 nosnp 6631 Do not enable SEV-SNP (applies to host/hypervisor 6632 only). Setting 'nosnp' avoids the RMP check overhead 6633 in memory accesses when users do not want to run 6634 SEV-SNP guests. 6635 6636 shapers= [NET] 6637 Maximal number of shapers. 6638 6639 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller 6640 Limit apic dumping. The parameter defines the maximal 6641 number of local apics being dumped. Also it is possible 6642 to set it to "all" by meaning -- no limit here. 6643 Format: { 1 (default) | 2 | ... | all }. 6644 The parameter valid if only apic=debug or 6645 apic=verbose is specified. 6646 Example: apic=debug show_lapic=all 6647 6648 slab_debug[=options[,slabs][;[options[,slabs]]...] [MM] 6649 Enabling slab_debug allows one to determine the 6650 culprit if slab objects become corrupted. Enabling 6651 slab_debug can create guard zones around objects and 6652 may poison objects when not in use. Also tracks the 6653 last alloc / free. For more information see 6654 Documentation/admin-guide/mm/slab.rst. 6655 (slub_debug legacy name also accepted for now) 6656 6657 Using this option implies the "no_hash_pointers" 6658 option which can be undone by adding the 6659 "hash_pointers=always" option. 6660 6661 slab_max_order= [MM] 6662 Determines the maximum allowed order for slabs. 6663 A high setting may cause OOMs due to memory 6664 fragmentation. For more information see 6665 Documentation/admin-guide/mm/slab.rst. 6666 (slub_max_order legacy name also accepted for now) 6667 6668 slab_merge [MM] 6669 Enable merging of slabs with similar size when the 6670 kernel is built without CONFIG_SLAB_MERGE_DEFAULT. 6671 (slub_merge legacy name also accepted for now) 6672 6673 slab_min_objects= [MM] 6674 The minimum number of objects per slab. SLUB will 6675 increase the slab order up to slab_max_order to 6676 generate a sufficiently large slab able to contain 6677 the number of objects indicated. The higher the number 6678 of objects the smaller the overhead of tracking slabs 6679 and the less frequently locks need to be acquired. 6680 For more information see 6681 Documentation/admin-guide/mm/slab.rst. 6682 (slub_min_objects legacy name also accepted for now) 6683 6684 slab_min_order= [MM] 6685 Determines the minimum page order for slabs. Must be 6686 lower or equal to slab_max_order. For more information see 6687 Documentation/admin-guide/mm/slab.rst. 6688 (slub_min_order legacy name also accepted for now) 6689 6690 slab_nomerge [MM] 6691 Disable merging of slabs with similar size. May be 6692 necessary if there is some reason to distinguish 6693 allocs to different slabs, especially in hardened 6694 environments where the risk of heap overflows and 6695 layout control by attackers can usually be 6696 frustrated by disabling merging. This will reduce 6697 most of the exposure of a heap attack to a single 6698 cache (risks via metadata attacks are mostly 6699 unchanged). Debug options disable merging on their 6700 own. 6701 For more information see 6702 Documentation/admin-guide/mm/slab.rst. 6703 (slub_nomerge legacy name also accepted for now) 6704 6705 slab_strict_numa [MM] 6706 Support memory policies on a per object level 6707 in the slab allocator. The default is for memory 6708 policies to be applied at the folio level when 6709 a new folio is needed or a partial folio is 6710 retrieved from the lists. Increases overhead 6711 in the slab fastpaths but gains more accurate 6712 NUMA kernel object placement which helps with slow 6713 interconnects in NUMA systems. 6714 6715 slram= [HW,MTD] 6716 6717 smart2= [HW] 6718 Format: <io1>[,<io2>[,...,<io8>]] 6719 6720 smp.csd_lock_timeout= [KNL] 6721 Specify the period of time in milliseconds 6722 that smp_call_function() and friends will wait 6723 for a CPU to release the CSD lock. This is 6724 useful when diagnosing bugs involving CPUs 6725 disabling interrupts for extended periods 6726 of time. Defaults to 5,000 milliseconds, and 6727 setting a value of zero disables this feature. 6728 This feature may be more efficiently disabled 6729 using the csdlock_debug- kernel parameter. 6730 6731 smp.panic_on_ipistall= [KNL] 6732 If a csd_lock_timeout extends for more than 6733 the specified number of milliseconds, panic the 6734 system. By default, let CSD-lock acquisition 6735 take as long as they take. Specifying 300,000 6736 for this value provides a 5-minute timeout. 6737 6738 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices 6739 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port 6740 smsc-ircc2.ircc_sir= [HW] SIR base I/O port 6741 smsc-ircc2.ircc_fir= [HW] FIR base I/O port 6742 smsc-ircc2.ircc_irq= [HW] IRQ line 6743 smsc-ircc2.ircc_dma= [HW] DMA channel 6744 smsc-ircc2.ircc_transceiver= [HW] Transceiver type: 6745 0: Toshiba Satellite 1800 (GP data pin select) 6746 1: Fast pin select (default) 6747 2: ATC IRMode 6748 6749 smt= [KNL,MIPS,S390,EARLY] Set the maximum number of threads 6750 (logical CPUs) to use per physical CPU on systems 6751 capable of symmetric multithreading (SMT). Will 6752 be capped to the actual hardware limit. 6753 Format: <integer> 6754 Default: -1 (no limit) 6755 6756 softlockup_panic= 6757 [KNL] Should the soft-lockup detector generate panics. 6758 Format: 0 | 1 6759 6760 A value of 1 instructs the soft-lockup detector 6761 to panic the machine when a soft-lockup occurs. It is 6762 also controlled by the kernel.softlockup_panic sysctl 6763 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the 6764 respective build-time switch to that functionality. 6765 6766 softlockup_all_cpu_backtrace= 6767 [KNL] Should the soft-lockup detector generate 6768 backtraces on all cpus. 6769 Format: 0 | 1 6770 6771 sonypi.*= [HW] Sony Programmable I/O Control Device driver 6772 See Documentation/admin-guide/laptops/sonypi.rst 6773 6774 spectre_bhi= [X86] Control mitigation of Branch History Injection 6775 (BHI) vulnerability. This setting affects the 6776 deployment of the HW BHI control and the SW BHB 6777 clearing sequence. 6778 6779 on - (default) Enable the HW or SW mitigation as 6780 needed. This protects the kernel from 6781 both syscalls and VMs. 6782 vmexit - On systems which don't have the HW mitigation 6783 available, enable the SW mitigation on vmexit 6784 ONLY. On such systems, the host kernel is 6785 protected from VM-originated BHI attacks, but 6786 may still be vulnerable to syscall attacks. 6787 off - Disable the mitigation. 6788 6789 spectre_v2= [X86,EARLY] Control mitigation of Spectre variant 2 6790 (indirect branch speculation) vulnerability. 6791 The default operation protects the kernel from 6792 user space attacks. 6793 6794 on - unconditionally enable, implies 6795 spectre_v2_user=on 6796 off - unconditionally disable, implies 6797 spectre_v2_user=off 6798 auto - kernel detects whether your CPU model is 6799 vulnerable 6800 6801 Selecting 'on' will, and 'auto' may, choose a 6802 mitigation method at run time according to the 6803 CPU, the available microcode, the setting of the 6804 CONFIG_MITIGATION_RETPOLINE configuration option, 6805 and the compiler with which the kernel was built. 6806 6807 Selecting 'on' will also enable the mitigation 6808 against user space to user space task attacks. 6809 Selecting specific mitigation does not force enable 6810 user mitigations. 6811 6812 Selecting 'off' will disable both the kernel and 6813 the user space protections. 6814 6815 Specific mitigations can also be selected manually: 6816 6817 retpoline - replace indirect branches 6818 retpoline,generic - Retpolines 6819 retpoline,lfence - LFENCE; indirect branch 6820 retpoline,amd - alias for retpoline,lfence 6821 eibrs - Enhanced/Auto IBRS 6822 eibrs,retpoline - Enhanced/Auto IBRS + Retpolines 6823 eibrs,lfence - Enhanced/Auto IBRS + LFENCE 6824 ibrs - use IBRS to protect kernel 6825 6826 Not specifying this option is equivalent to 6827 spectre_v2=auto. 6828 6829 spectre_v2_user= 6830 [X86] Control mitigation of Spectre variant 2 6831 (indirect branch speculation) vulnerability between 6832 user space tasks 6833 6834 on - Unconditionally enable mitigations. Is 6835 enforced by spectre_v2=on 6836 6837 off - Unconditionally disable mitigations. Is 6838 enforced by spectre_v2=off 6839 6840 prctl - Indirect branch speculation is enabled, 6841 but mitigation can be enabled via prctl 6842 per thread. The mitigation control state 6843 is inherited on fork. 6844 6845 prctl,ibpb 6846 - Like "prctl" above, but only STIBP is 6847 controlled per thread. IBPB is issued 6848 always when switching between different user 6849 space processes. 6850 6851 seccomp 6852 - Same as "prctl" above, but all seccomp 6853 threads will enable the mitigation unless 6854 they explicitly opt out. 6855 6856 seccomp,ibpb 6857 - Like "seccomp" above, but only STIBP is 6858 controlled per thread. IBPB is issued 6859 always when switching between different 6860 user space processes. 6861 6862 auto - Kernel selects the mitigation depending on 6863 the available CPU features and vulnerability. 6864 6865 Default mitigation: "prctl" 6866 6867 Not specifying this option is equivalent to 6868 spectre_v2_user=auto. 6869 6870 spec_rstack_overflow= 6871 [X86,EARLY] Control RAS overflow mitigation on AMD Zen CPUs 6872 6873 off - Disable mitigation 6874 microcode - Enable microcode mitigation only 6875 safe-ret - Enable sw-only safe RET mitigation (default) 6876 ibpb - Enable mitigation by issuing IBPB on 6877 kernel entry 6878 ibpb-vmexit - Issue IBPB only on VMEXIT 6879 (cloud-specific mitigation) 6880 6881 spec_store_bypass_disable= 6882 [HW,EARLY] Control Speculative Store Bypass (SSB) Disable mitigation 6883 (Speculative Store Bypass vulnerability) 6884 6885 Certain CPUs are vulnerable to an exploit against a 6886 a common industry wide performance optimization known 6887 as "Speculative Store Bypass" in which recent stores 6888 to the same memory location may not be observed by 6889 later loads during speculative execution. The idea 6890 is that such stores are unlikely and that they can 6891 be detected prior to instruction retirement at the 6892 end of a particular speculation execution window. 6893 6894 In vulnerable processors, the speculatively forwarded 6895 store can be used in a cache side channel attack, for 6896 example to read memory to which the attacker does not 6897 directly have access (e.g. inside sandboxed code). 6898 6899 This parameter controls whether the Speculative Store 6900 Bypass optimization is used. 6901 6902 On x86 the options are: 6903 6904 on - Unconditionally disable Speculative Store Bypass 6905 off - Unconditionally enable Speculative Store Bypass 6906 auto - Kernel detects whether the CPU model contains an 6907 implementation of Speculative Store Bypass and 6908 picks the most appropriate mitigation. If the 6909 CPU is not vulnerable, "off" is selected. If the 6910 CPU is vulnerable the default mitigation is 6911 architecture and Kconfig dependent. See below. 6912 prctl - Control Speculative Store Bypass per thread 6913 via prctl. Speculative Store Bypass is enabled 6914 for a process by default. The state of the control 6915 is inherited on fork. 6916 seccomp - Same as "prctl" above, but all seccomp threads 6917 will disable SSB unless they explicitly opt out. 6918 6919 Default mitigations: 6920 X86: "prctl" 6921 6922 On powerpc the options are: 6923 6924 on,auto - On Power8 and Power9 insert a store-forwarding 6925 barrier on kernel entry and exit. On Power7 6926 perform a software flush on kernel entry and 6927 exit. 6928 off - No action. 6929 6930 Not specifying this option is equivalent to 6931 spec_store_bypass_disable=auto. 6932 6933 split_lock_detect= 6934 [X86] Enable split lock detection or bus lock detection 6935 6936 When enabled (and if hardware support is present), atomic 6937 instructions that access data across cache line 6938 boundaries will result in an alignment check exception 6939 for split lock detection or a debug exception for 6940 bus lock detection. 6941 6942 off - not enabled 6943 6944 warn - the kernel will emit rate-limited warnings 6945 about applications triggering the #AC 6946 exception or the #DB exception. This mode is 6947 the default on CPUs that support split lock 6948 detection or bus lock detection. Default 6949 behavior is by #AC if both features are 6950 enabled in hardware. 6951 6952 fatal - the kernel will send SIGBUS to applications 6953 that trigger the #AC exception or the #DB 6954 exception. Default behavior is by #AC if 6955 both features are enabled in hardware. 6956 6957 ratelimit:N - 6958 Set system wide rate limit to N bus locks 6959 per second for bus lock detection. 6960 0 < N <= 1000. 6961 6962 N/A for split lock detection. 6963 6964 6965 If an #AC exception is hit in the kernel or in 6966 firmware (i.e. not while executing in user mode) 6967 the kernel will oops in either "warn" or "fatal" 6968 mode. 6969 6970 #DB exception for bus lock is triggered only when 6971 CPL > 0. 6972 6973 srbds= [X86,INTEL,EARLY] 6974 Control the Special Register Buffer Data Sampling 6975 (SRBDS) mitigation. 6976 6977 Certain CPUs are vulnerable to an MDS-like 6978 exploit which can leak bits from the random 6979 number generator. 6980 6981 By default, this issue is mitigated by 6982 microcode. However, the microcode fix can cause 6983 the RDRAND and RDSEED instructions to become 6984 much slower. Among other effects, this will 6985 result in reduced throughput from /dev/urandom. 6986 6987 The microcode mitigation can be disabled with 6988 the following option: 6989 6990 off: Disable mitigation and remove 6991 performance impact to RDRAND and RDSEED 6992 6993 srcutree.big_cpu_lim [KNL] 6994 Specifies the number of CPUs constituting a 6995 large system, such that srcu_struct structures 6996 should immediately allocate an srcu_node array. 6997 This kernel-boot parameter defaults to 128, 6998 but takes effect only when the low-order four 6999 bits of srcutree.convert_to_big is equal to 3 7000 (decide at boot). 7001 7002 srcutree.convert_to_big [KNL] 7003 Specifies under what conditions an SRCU tree 7004 srcu_struct structure will be converted to big 7005 form, that is, with an rcu_node tree: 7006 7007 0: Never. 7008 1: At init_srcu_struct() time. 7009 2: When rcutorture decides to. 7010 3: Decide at boot time (default). 7011 0x1X: Above plus if high contention. 7012 7013 Either way, the srcu_node tree will be sized based 7014 on the actual runtime number of CPUs (nr_cpu_ids) 7015 instead of the compile-time CONFIG_NR_CPUS. 7016 7017 srcutree.counter_wrap_check [KNL] 7018 Specifies how frequently to check for 7019 grace-period sequence counter wrap for the 7020 srcu_data structure's ->srcu_gp_seq_needed field. 7021 The greater the number of bits set in this kernel 7022 parameter, the less frequently counter wrap will 7023 be checked for. Note that the bottom two bits 7024 are ignored. 7025 7026 srcutree.exp_holdoff [KNL] 7027 Specifies how many nanoseconds must elapse 7028 since the end of the last SRCU grace period for 7029 a given srcu_struct until the next normal SRCU 7030 grace period will be considered for automatic 7031 expediting. Set to zero to disable automatic 7032 expediting. 7033 7034 srcutree.srcu_max_nodelay [KNL] 7035 Specifies the number of no-delay instances 7036 per jiffy for which the SRCU grace period 7037 worker thread will be rescheduled with zero 7038 delay. Beyond this limit, worker thread will 7039 be rescheduled with a sleep delay of one jiffy. 7040 7041 srcutree.srcu_max_nodelay_phase [KNL] 7042 Specifies the per-grace-period phase, number of 7043 non-sleeping polls of readers. Beyond this limit, 7044 grace period worker thread will be rescheduled 7045 with a sleep delay of one jiffy, between each 7046 rescan of the readers, for a grace period phase. 7047 7048 srcutree.srcu_retry_check_delay [KNL] 7049 Specifies number of microseconds of non-sleeping 7050 delay between each non-sleeping poll of readers. 7051 7052 srcutree.small_contention_lim [KNL] 7053 Specifies the number of update-side contention 7054 events per jiffy will be tolerated before 7055 initiating a conversion of an srcu_struct 7056 structure to big form. Note that the value of 7057 srcutree.convert_to_big must have the 0x10 bit 7058 set for contention-based conversions to occur. 7059 7060 ssbd= [ARM64,HW,EARLY] 7061 Speculative Store Bypass Disable control 7062 7063 On CPUs that are vulnerable to the Speculative 7064 Store Bypass vulnerability and offer a 7065 firmware based mitigation, this parameter 7066 indicates how the mitigation should be used: 7067 7068 force-on: Unconditionally enable mitigation for 7069 for both kernel and userspace 7070 force-off: Unconditionally disable mitigation for 7071 for both kernel and userspace 7072 kernel: Always enable mitigation in the 7073 kernel, and offer a prctl interface 7074 to allow userspace to register its 7075 interest in being mitigated too. 7076 7077 stack_guard_gap= [MM] 7078 override the default stack gap protection. The value 7079 is in page units and it defines how many pages prior 7080 to (for stacks growing down) resp. after (for stacks 7081 growing up) the main stack are reserved for no other 7082 mapping. Default value is 256 pages. 7083 7084 stack_depot_disable= [KNL,EARLY] 7085 Setting this to true through kernel command line will 7086 disable the stack depot thereby saving the static memory 7087 consumed by the stack hash table. By default this is set 7088 to false. 7089 7090 stack_depot_max_pools= [KNL,EARLY] 7091 Specify the maximum number of pools to use for storing 7092 stack traces. Pools are allocated on-demand up to this 7093 limit. Default value is 8191 pools. 7094 7095 stacktrace [FTRACE] 7096 Enabled the stack tracer on boot up. 7097 7098 stacktrace_filter=[function-list] 7099 [FTRACE] Limit the functions that the stack tracer 7100 will trace at boot up. function-list is a comma-separated 7101 list of functions. This list can be changed at run 7102 time by the stack_trace_filter file in the debugfs 7103 tracing directory. Note, this enables stack tracing 7104 and the stacktrace above is not needed. 7105 7106 sti= [PARISC,HW] 7107 Format: <num> 7108 Set the STI (builtin display/keyboard on the HP-PARISC 7109 machines) console (graphic card) which should be used 7110 as the initial boot-console. 7111 See also comment in drivers/video/console/sticore.c. 7112 7113 sti_font= [HW] 7114 See comment in drivers/video/console/sticore.c. 7115 7116 stifb= [HW] 7117 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]] 7118 7119 strict_sas_size= 7120 [X86] 7121 Format: <bool> 7122 Enable or disable strict sigaltstack size checks 7123 against the required signal frame size which 7124 depends on the supported FPU features. This can 7125 be used to filter out binaries which have 7126 not yet been made aware of AT_MINSIGSTKSZ. 7127 7128 stress_hpt [PPC,EARLY] 7129 Limits the number of kernel HPT entries in the hash 7130 page table to increase the rate of hash page table 7131 faults on kernel addresses. 7132 7133 stress_slb [PPC,EARLY] 7134 Limits the number of kernel SLB entries, and flushes 7135 them frequently to increase the rate of SLB faults 7136 on kernel addresses. 7137 7138 sunrpc.min_resvport= 7139 sunrpc.max_resvport= 7140 [NFS,SUNRPC] 7141 SunRPC servers often require that client requests 7142 originate from a privileged port (i.e. a port in the 7143 range 0 < portnr < 1024). 7144 An administrator who wishes to reserve some of these 7145 ports for other uses may adjust the range that the 7146 kernel's sunrpc client considers to be privileged 7147 using these two parameters to set the minimum and 7148 maximum port values. 7149 7150 sunrpc.svc_rpc_per_connection_limit= 7151 [NFS,SUNRPC] 7152 Limit the number of requests that the server will 7153 process in parallel from a single connection. 7154 The default value is 0 (no limit). 7155 7156 sunrpc.pool_mode= 7157 [NFS] 7158 Control how the NFS server code allocates CPUs to 7159 service thread pools. Depending on how many NICs 7160 you have and where their interrupts are bound, this 7161 option will affect which CPUs will do NFS serving. 7162 Note: this parameter cannot be changed while the 7163 NFS server is running. 7164 7165 auto the server chooses an appropriate mode 7166 automatically using heuristics 7167 global a single global pool contains all CPUs 7168 percpu one pool for each CPU 7169 pernode one pool for each NUMA node (equivalent 7170 to global on non-NUMA machines) 7171 7172 sunrpc.tcp_slot_table_entries= 7173 sunrpc.udp_slot_table_entries= 7174 [NFS,SUNRPC] 7175 Sets the upper limit on the number of simultaneous 7176 RPC calls that can be sent from the client to a 7177 server. Increasing these values may allow you to 7178 improve throughput, but will also increase the 7179 amount of memory reserved for use by the client. 7180 7181 suspend.pm_test_delay= 7182 [SUSPEND] 7183 Sets the number of seconds to remain in a suspend test 7184 mode before resuming the system (see 7185 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG 7186 is set. Default value is 5. 7187 7188 svm= [PPC] 7189 Format: { on | off | y | n | 1 | 0 } 7190 This parameter controls use of the Protected 7191 Execution Facility on pSeries. 7192 7193 swiotlb= [ARM,PPC,MIPS,X86,S390,EARLY] 7194 Format: { <int> [,<int>] | force | noforce } 7195 <int> -- Number of I/O TLB slabs 7196 <int> -- Second integer after comma. Number of swiotlb 7197 areas with their own lock. Will be rounded up 7198 to a power of 2. 7199 force -- force using of bounce buffers even if they 7200 wouldn't be automatically used by the kernel 7201 noforce -- Never use bounce buffers (for debugging) 7202 7203 switches= [HW,M68k,EARLY] 7204 7205 sysctl.*= [KNL] 7206 Set a sysctl parameter, right before loading the init 7207 process, as if the value was written to the respective 7208 /proc/sys/... file. Both '.' and '/' are recognized as 7209 separators. Unrecognized parameters and invalid values 7210 are reported in the kernel log. Sysctls registered 7211 later by a loaded module cannot be set this way. 7212 Example: sysctl.vm.swappiness=40 7213 7214 sysrq_always_enabled 7215 [KNL] 7216 Ignore sysrq setting - this boot parameter will 7217 neutralize any effect of /proc/sys/kernel/sysrq. 7218 Useful for debugging. 7219 7220 tcpmhash_entries= [KNL,NET] 7221 Set the number of tcp_metrics_hash slots. 7222 Default value is 8192 or 16384 depending on total 7223 ram pages. This is used to specify the TCP metrics 7224 cache size. See Documentation/networking/ip-sysctl.rst 7225 "tcp_no_metrics_save" section for more details. 7226 7227 tdfx= [HW,DRM] 7228 7229 test_suspend= [SUSPEND] 7230 Format: { "mem" | "standby" | "freeze" }[,N] 7231 Specify "mem" (for Suspend-to-RAM) or "standby" (for 7232 standby suspend) or "freeze" (for suspend type freeze) 7233 as the system sleep state during system startup with 7234 the optional capability to repeat N number of times. 7235 The system is woken from this state using a 7236 wakeup-capable RTC alarm. 7237 7238 thash_entries= [KNL,NET] 7239 Set number of hash buckets for TCP connection 7240 7241 thermal.act= [HW,ACPI] 7242 -1: disable all active trip points in all thermal zones 7243 <degrees C>: override all lowest active trip points 7244 7245 thermal.crt= [HW,ACPI] 7246 -1: disable all critical trip points in all thermal zones 7247 <degrees C>: override all critical trip points 7248 7249 thermal.off= [HW,ACPI] 7250 1: disable ACPI thermal control 7251 7252 thermal.psv= [HW,ACPI] 7253 -1: disable all passive trip points 7254 <degrees C>: override all passive trip points to this 7255 value 7256 7257 thermal.tzp= [HW,ACPI] 7258 Specify global default ACPI thermal zone polling rate 7259 <deci-seconds>: poll all this frequency 7260 0: no polling (default) 7261 7262 thp_anon= [KNL] 7263 Format: <size>[KMG],<size>[KMG]:<state>;<size>[KMG]-<size>[KMG]:<state> 7264 state is one of "always", "madvise", "never" or "inherit". 7265 Control the default behavior of the system with respect 7266 to anonymous transparent hugepages. 7267 Can be used multiple times for multiple anon THP sizes. 7268 See Documentation/admin-guide/mm/transhuge.rst for more 7269 details. 7270 7271 threadirqs [KNL,EARLY] 7272 Force threading of all interrupt handlers except those 7273 marked explicitly IRQF_NO_THREAD. 7274 7275 thp_shmem= [KNL] 7276 Format: <size>[KMG],<size>[KMG]:<policy>;<size>[KMG]-<size>[KMG]:<policy> 7277 Control the default policy of each hugepage size for the 7278 internal shmem mount. <policy> is one of policies available 7279 for the shmem mount ("always", "inherit", "never", "within_size", 7280 and "advise"). 7281 It can be used multiple times for multiple shmem THP sizes. 7282 See Documentation/admin-guide/mm/transhuge.rst for more 7283 details. 7284 7285 topology= [S390,EARLY] 7286 Format: {off | on} 7287 Specify if the kernel should make use of the cpu 7288 topology information if the hardware supports this. 7289 The scheduler will make use of this information and 7290 e.g. base its process migration decisions on it. 7291 Default is on. 7292 7293 torture.disable_onoff_at_boot= [KNL] 7294 Prevent the CPU-hotplug component of torturing 7295 until after init has spawned. 7296 7297 torture.ftrace_dump_at_shutdown= [KNL] 7298 Dump the ftrace buffer at torture-test shutdown, 7299 even if there were no errors. This can be a 7300 very costly operation when many torture tests 7301 are running concurrently, especially on systems 7302 with rotating-rust storage. 7303 7304 torture.verbose_sleep_frequency= [KNL] 7305 Specifies how many verbose printk()s should be 7306 emitted between each sleep. The default of zero 7307 disables verbose-printk() sleeping. 7308 7309 torture.verbose_sleep_duration= [KNL] 7310 Duration of each verbose-printk() sleep in jiffies. 7311 7312 tpm.disable_pcr_integrity= [HW,TPM] 7313 Do not protect PCR registers from unintended physical 7314 access, or interposers in the bus by the means of 7315 having an integrity protected session wrapped around 7316 TPM2_PCR_Extend command. Consider this in a situation 7317 where TPM is heavily utilized by IMA, thus protection 7318 causing a major performance hit, and the space where 7319 machines are deployed is by other means guarded. 7320 7321 tpm_crb_ffa.busy_timeout_ms= [ARM64,TPM] 7322 Maximum time in milliseconds to retry sending a message 7323 to the TPM service before giving up. This parameter controls 7324 how long the system will continue retrying when the TPM 7325 service is busy. 7326 Format: <unsigned int> 7327 Default: 2000 (2 seconds) 7328 7329 tpm_suspend_pcr=[HW,TPM] 7330 Format: integer pcr id 7331 Specify that at suspend time, the tpm driver 7332 should extend the specified pcr with zeros, 7333 as a workaround for some chips which fail to 7334 flush the last written pcr on TPM_SaveState. 7335 This will guarantee that all the other pcrs 7336 are saved. 7337 7338 tpm_tis.interrupts= [HW,TPM] 7339 Enable interrupts for the MMIO based physical layer 7340 for the FIFO interface. By default it is set to false 7341 (0). For more information about TPM hardware interfaces 7342 defined by Trusted Computing Group (TCG) see 7343 https://trustedcomputinggroup.org/resource/pc-client-platform-tpm-profile-ptp-specification/ 7344 7345 tp_printk [FTRACE] 7346 Have the tracepoints sent to printk as well as the 7347 tracing ring buffer. This is useful for early boot up 7348 where the system hangs or reboots and does not give the 7349 option for reading the tracing buffer or performing a 7350 ftrace_dump_on_oops. 7351 7352 To turn off having tracepoints sent to printk, 7353 echo 0 > /proc/sys/kernel/tracepoint_printk 7354 Note, echoing 1 into this file without the 7355 tp_printk kernel cmdline option has no effect. 7356 7357 The tp_printk_stop_on_boot (see below) can also be used 7358 to stop the printing of events to console at 7359 late_initcall_sync. 7360 7361 ** CAUTION ** 7362 7363 Having tracepoints sent to printk() and activating high 7364 frequency tracepoints such as irq or sched, can cause 7365 the system to live lock. 7366 7367 tp_printk_stop_on_boot [FTRACE] 7368 When tp_printk (above) is set, it can cause a lot of noise 7369 on the console. It may be useful to only include the 7370 printing of events during boot up, as user space may 7371 make the system inoperable. 7372 7373 This command line option will stop the printing of events 7374 to console at the late_initcall_sync() time frame. 7375 7376 trace_buf_size=nn[KMG] 7377 [FTRACE] will set tracing buffer size on each cpu. 7378 7379 trace_clock= [FTRACE] Set the clock used for tracing events 7380 at boot up. 7381 local - Use the per CPU time stamp counter 7382 (converted into nanoseconds). Fast, but 7383 depending on the architecture, may not be 7384 in sync between CPUs. 7385 global - Event time stamps are synchronize across 7386 CPUs. May be slower than the local clock, 7387 but better for some race conditions. 7388 counter - Simple counting of events (1, 2, ..) 7389 note, some counts may be skipped due to the 7390 infrastructure grabbing the clock more than 7391 once per event. 7392 uptime - Use jiffies as the time stamp. 7393 perf - Use the same clock that perf uses. 7394 mono - Use ktime_get_mono_fast_ns() for time stamps. 7395 mono_raw - Use ktime_get_raw_fast_ns() for time 7396 stamps. 7397 boot - Use ktime_get_boot_fast_ns() for time stamps. 7398 Architectures may add more clocks. See 7399 Documentation/trace/ftrace.rst for more details. 7400 7401 trace_event=[event-list] 7402 [FTRACE] Set and start specified trace events in order 7403 to facilitate early boot debugging. The event-list is a 7404 comma-separated list of trace events to enable. See 7405 also Documentation/trace/events.rst 7406 7407 To enable modules, use :mod: keyword: 7408 7409 trace_event=:mod:<module> 7410 7411 The value before :mod: will only enable specific events 7412 that are part of the module. See the above mentioned 7413 document for more information. 7414 7415 trace_instance=[instance-info] 7416 [FTRACE] Create a ring buffer instance early in boot up. 7417 This will be listed in: 7418 7419 /sys/kernel/tracing/instances 7420 7421 Events can be enabled at the time the instance is created 7422 via: 7423 7424 trace_instance=<name>,<system1>:<event1>,<system2>:<event2> 7425 7426 Note, the "<system*>:" portion is optional if the event is 7427 unique. 7428 7429 trace_instance=foo,sched:sched_switch,irq_handler_entry,initcall 7430 7431 will enable the "sched_switch" event (note, the "sched:" is optional, and 7432 the same thing would happen if it was left off). The irq_handler_entry 7433 event, and all events under the "initcall" system. 7434 7435 Flags can be added to the instance to modify its behavior when it is 7436 created. The flags are separated by '^'. 7437 7438 The available flags are: 7439 7440 traceoff - Have the tracing instance tracing disabled after it is created. 7441 traceprintk - Have trace_printk() write into this trace instance 7442 (note, "printk" and "trace_printk" can also be used) 7443 7444 trace_instance=foo^traceoff^traceprintk,sched,irq 7445 7446 The flags must come before the defined events. 7447 7448 If memory has been reserved (see memmap for x86), the instance 7449 can use that memory: 7450 7451 memmap=12M$0x284500000 trace_instance=boot_map@0x284500000:12M 7452 7453 The above will create a "boot_map" instance that uses the physical 7454 memory at 0x284500000 that is 12Megs. The per CPU buffers of that 7455 instance will be split up accordingly. 7456 7457 Alternatively, the memory can be reserved by the reserve_mem option: 7458 7459 reserve_mem=12M:4096:trace trace_instance=boot_map@trace 7460 7461 This will reserve 12 megabytes at boot up with a 4096 byte alignment 7462 and place the ring buffer in this memory. Note that due to KASLR, the 7463 memory may not be the same location each time, which will not preserve 7464 the buffer content. 7465 7466 Also note that the layout of the ring buffer data may change between 7467 kernel versions where the validator will fail and reset the ring buffer 7468 if the layout is not the same as the previous kernel. 7469 7470 If the ring buffer is used for persistent bootups and has events enabled, 7471 it is recommend to disable tracing so that events from a previous boot do not 7472 mix with events of the current boot (unless you are debugging a random crash 7473 at boot up). 7474 7475 reserve_mem=12M:4096:trace trace_instance=boot_map^traceoff^traceprintk@trace,sched,irq 7476 7477 Note, saving the trace buffer across reboots does require that the system 7478 is set up to not wipe memory. For instance, CONFIG_RESET_ATTACK_MITIGATION 7479 can force a memory reset on boot which will clear any trace that was stored. 7480 This is just one of many ways that can clear memory. Make sure your system 7481 keeps the content of memory across reboots before relying on this option. 7482 7483 NB: Both the mapped address and size must be page aligned for the architecture. 7484 7485 See also Documentation/trace/debugging.rst 7486 7487 7488 trace_options=[option-list] 7489 [FTRACE] Enable or disable tracer options at boot. 7490 The option-list is a comma delimited list of options 7491 that can be enabled or disabled just as if you were 7492 to echo the option name into 7493 7494 /sys/kernel/tracing/trace_options 7495 7496 For example, to enable stacktrace option (to dump the 7497 stack trace of each event), add to the command line: 7498 7499 trace_options=stacktrace 7500 7501 See also Documentation/trace/ftrace.rst "trace options" 7502 section. 7503 7504 trace_trigger=[trigger-list] 7505 [FTRACE] Add a event trigger on specific events. 7506 Set a trigger on top of a specific event, with an optional 7507 filter. 7508 7509 The format is is "trace_trigger=<event>.<trigger>[ if <filter>],..." 7510 Where more than one trigger may be specified that are comma deliminated. 7511 7512 For example: 7513 7514 trace_trigger="sched_switch.stacktrace if prev_state == 2" 7515 7516 The above will enable the "stacktrace" trigger on the "sched_switch" 7517 event but only trigger it if the "prev_state" of the "sched_switch" 7518 event is "2" (TASK_UNINTERUPTIBLE). 7519 7520 See also "Event triggers" in Documentation/trace/events.rst 7521 7522 7523 traceoff_after_boot 7524 [FTRACE] Sometimes tracing is used to debug issues 7525 during the boot process. Since the trace buffer has a 7526 limited amount of storage, it may be prudent to 7527 disable tracing after the boot is finished, otherwise 7528 the critical information may be overwritten. With this 7529 option, the main tracing buffer will be turned off at 7530 the end of the boot process. 7531 7532 traceoff_on_warning 7533 [FTRACE] enable this option to disable tracing when a 7534 warning is hit. This turns off "tracing_on". Tracing can 7535 be enabled again by echoing '1' into the "tracing_on" 7536 file located in /sys/kernel/tracing/ 7537 7538 This option is useful, as it disables the trace before 7539 the WARNING dump is called, which prevents the trace to 7540 be filled with content caused by the warning output. 7541 7542 This option can also be set at run time via the sysctl 7543 option: kernel/traceoff_on_warning 7544 7545 transparent_hugepage= 7546 [KNL] 7547 Format: [always|madvise|never] 7548 Can be used to control the default behavior of the system 7549 with respect to transparent hugepages. 7550 See Documentation/admin-guide/mm/transhuge.rst 7551 for more details. 7552 7553 transparent_hugepage_shmem= [KNL] 7554 Format: [always|within_size|advise|never|deny|force] 7555 Can be used to control the hugepage allocation policy for 7556 the internal shmem mount. 7557 See Documentation/admin-guide/mm/transhuge.rst 7558 for more details. 7559 7560 transparent_hugepage_tmpfs= [KNL] 7561 Format: [always|within_size|advise|never] 7562 Can be used to control the default hugepage allocation policy 7563 for the tmpfs mount. 7564 See Documentation/admin-guide/mm/transhuge.rst 7565 for more details. 7566 7567 trusted.source= [KEYS] 7568 Format: <string> 7569 This parameter identifies the trust source as a backend 7570 for trusted keys implementation. Supported trust 7571 sources: 7572 - "tpm" 7573 - "tee" 7574 - "caam" 7575 - "dcp" 7576 If not specified then it defaults to iterating through 7577 the trust source list starting with TPM and assigns the 7578 first trust source as a backend which is initialized 7579 successfully during iteration. 7580 7581 trusted.rng= [KEYS] 7582 Format: <string> 7583 The RNG used to generate key material for trusted keys. 7584 Can be one of: 7585 - "kernel" 7586 - the same value as trusted.source: "tpm" or "tee" 7587 - "default" 7588 If not specified, "default" is used. In this case, 7589 the RNG's choice is left to each individual trust source. 7590 7591 trusted.dcp_use_otp_key 7592 This is intended to be used in combination with 7593 trusted.source=dcp and will select the DCP OTP key 7594 instead of the DCP UNIQUE key blob encryption. 7595 7596 trusted.dcp_skip_zk_test 7597 This is intended to be used in combination with 7598 trusted.source=dcp and will disable the check if the 7599 blob key is all zeros. This is helpful for situations where 7600 having this key zero'ed is acceptable. E.g. in testing 7601 scenarios. 7602 7603 tsa= [X86] Control mitigation for Transient Scheduler 7604 Attacks on AMD CPUs. Search the following in your 7605 favourite search engine for more details: 7606 7607 "Technical guidance for mitigating transient scheduler 7608 attacks". 7609 7610 off - disable the mitigation 7611 on - enable the mitigation (default) 7612 user - mitigate only user/kernel transitions 7613 vm - mitigate only guest/host transitions 7614 7615 7616 tsc= Disable clocksource stability checks for TSC. 7617 Format: <string> 7618 [x86] reliable: mark tsc clocksource as reliable, this 7619 disables clocksource verification at runtime, as well 7620 as the stability checks done at bootup. Used to enable 7621 high-resolution timer mode on older hardware, and in 7622 virtualized environment. 7623 [x86] noirqtime: Do not use TSC to do irq accounting. 7624 Used to run time disable IRQ_TIME_ACCOUNTING on any 7625 platforms where RDTSC is slow and this accounting 7626 can add overhead. 7627 [x86] unstable: mark the TSC clocksource as unstable, this 7628 marks the TSC unconditionally unstable at bootup and 7629 avoids any further wobbles once the TSC watchdog notices. 7630 [x86] nowatchdog: disable clocksource watchdog. Used 7631 in situations with strict latency requirements (where 7632 interruptions from clocksource watchdog are not 7633 acceptable). 7634 [x86] recalibrate: force recalibration against a HW timer 7635 (HPET or PM timer) on systems whose TSC frequency was 7636 obtained from HW or FW using either an MSR or CPUID(0x15). 7637 Warn if the difference is more than 500 ppm. 7638 [x86] watchdog: Use TSC as the watchdog clocksource with 7639 which to check other HW timers (HPET or PM timer), but 7640 only on systems where TSC has been deemed trustworthy. 7641 This will be suppressed by an earlier tsc=nowatchdog and 7642 can be overridden by a later tsc=nowatchdog. A console 7643 message will flag any such suppression or overriding. 7644 7645 tsc_early_khz= [X86,EARLY] Skip early TSC calibration and use the given 7646 value instead. Useful when the early TSC frequency discovery 7647 procedure is not reliable, such as on overclocked systems 7648 with CPUID.16h support and partial CPUID.15h support. 7649 Format: <unsigned int> 7650 7651 tsx= [X86] Control Transactional Synchronization 7652 Extensions (TSX) feature in Intel processors that 7653 support TSX control. 7654 7655 This parameter controls the TSX feature. The options are: 7656 7657 on - Enable TSX on the system. Although there are 7658 mitigations for all known security vulnerabilities, 7659 TSX has been known to be an accelerator for 7660 several previous speculation-related CVEs, and 7661 so there may be unknown security risks associated 7662 with leaving it enabled. 7663 7664 off - Disable TSX on the system. (Note that this 7665 option takes effect only on newer CPUs which are 7666 not vulnerable to MDS, i.e., have 7667 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get 7668 the new IA32_TSX_CTRL MSR through a microcode 7669 update. This new MSR allows for the reliable 7670 deactivation of the TSX functionality.) 7671 7672 auto - Disable TSX if X86_BUG_TAA is present, 7673 otherwise enable TSX on the system. 7674 7675 Not specifying this option is equivalent to tsx=off. 7676 7677 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst 7678 for more details. 7679 7680 tsx_async_abort= [X86,INTEL,EARLY] Control mitigation for the TSX Async 7681 Abort (TAA) vulnerability. 7682 7683 Similar to Micro-architectural Data Sampling (MDS) 7684 certain CPUs that support Transactional 7685 Synchronization Extensions (TSX) are vulnerable to an 7686 exploit against CPU internal buffers which can forward 7687 information to a disclosure gadget under certain 7688 conditions. 7689 7690 In vulnerable processors, the speculatively forwarded 7691 data can be used in a cache side channel attack, to 7692 access data to which the attacker does not have direct 7693 access. 7694 7695 This parameter controls the TAA mitigation. The 7696 options are: 7697 7698 full - Enable TAA mitigation on vulnerable CPUs 7699 if TSX is enabled. 7700 7701 full,nosmt - Enable TAA mitigation and disable SMT on 7702 vulnerable CPUs. If TSX is disabled, SMT 7703 is not disabled because CPU is not 7704 vulnerable to cross-thread TAA attacks. 7705 off - Unconditionally disable TAA mitigation 7706 7707 On MDS-affected machines, tsx_async_abort=off can be 7708 prevented by an active MDS mitigation as both vulnerabilities 7709 are mitigated with the same mechanism so in order to disable 7710 this mitigation, you need to specify mds=off too. 7711 7712 Not specifying this option is equivalent to 7713 tsx_async_abort=full. On CPUs which are MDS affected 7714 and deploy MDS mitigation, TAA mitigation is not 7715 required and doesn't provide any additional 7716 mitigation. 7717 7718 For details see: 7719 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst 7720 7721 turbografx.map[2|3]= [HW,JOY] 7722 TurboGraFX parallel port interface 7723 Format: 7724 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7> 7725 See also Documentation/input/devices/joystick-parport.rst 7726 7727 udbg-immortal [PPC] When debugging early kernel crashes that 7728 happen after console_init() and before a proper 7729 console driver takes over, this boot options might 7730 help "seeing" what's going on. 7731 7732 uhash_entries= [KNL,NET] 7733 Set number of hash buckets for UDP/UDP-Lite connections 7734 7735 uhci-hcd.ignore_oc= 7736 [USB] Ignore overcurrent events (default N). 7737 Some badly-designed motherboards generate lots of 7738 bogus events, for ports that aren't wired to 7739 anything. Set this parameter to avoid log spamming. 7740 Note that genuine overcurrent events won't be 7741 reported either. 7742 7743 unaligned_scalar_speed= 7744 [RISCV] 7745 Format: {slow | fast | unsupported} 7746 Allow skipping scalar unaligned access speed tests. This 7747 is useful for testing alternative code paths and to skip 7748 the tests in environments where they run too slowly. All 7749 CPUs must have the same scalar unaligned access speed. 7750 7751 unaligned_vector_speed= 7752 [RISCV] 7753 Format: {slow | fast | unsupported} 7754 Allow skipping vector unaligned access speed tests. This 7755 is useful for testing alternative code paths and to skip 7756 the tests in environments where they run too slowly. All 7757 CPUs must have the same vector unaligned access speed. 7758 7759 unknown_nmi_panic 7760 [X86] Cause panic on unknown NMI. 7761 7762 unwind_debug [X86-64,EARLY] 7763 Enable unwinder debug output. This can be 7764 useful for debugging certain unwinder error 7765 conditions, including corrupt stacks and 7766 bad/missing unwinder metadata. 7767 7768 usbcore.authorized_default= 7769 [USB] Default USB device authorization: 7770 (default -1 = authorized (same as 1), 7771 0 = not authorized, 1 = authorized, 2 = authorized 7772 if device connected to internal port) 7773 7774 usbcore.autosuspend= 7775 [USB] The autosuspend time delay (in seconds) used 7776 for newly-detected USB devices (default 2). This 7777 is the time required before an idle device will be 7778 autosuspended. Devices for which the delay is set 7779 to a negative value won't be autosuspended at all. 7780 7781 usbcore.usbfs_snoop= 7782 [USB] Set to log all usbfs traffic (default 0 = off). 7783 7784 usbcore.usbfs_snoop_max= 7785 [USB] Maximum number of bytes to snoop in each URB 7786 (default = 65536). 7787 7788 usbcore.blinkenlights= 7789 [USB] Set to cycle leds on hubs (default 0 = off). 7790 7791 usbcore.old_scheme_first= 7792 [USB] Start with the old device initialization 7793 scheme (default 0 = off). 7794 7795 usbcore.usbfs_memory_mb= 7796 [USB] Memory limit (in MB) for buffers allocated by 7797 usbfs (default = 16, 0 = max = 2047). 7798 7799 usbcore.use_both_schemes= 7800 [USB] Try the other device initialization scheme 7801 if the first one fails (default 1 = enabled). 7802 7803 usbcore.initial_descriptor_timeout= 7804 [USB] Specifies timeout for the initial 64-byte 7805 USB_REQ_GET_DESCRIPTOR request in milliseconds 7806 (default 5000 = 5.0 seconds). 7807 7808 usbcore.nousb [USB] Disable the USB subsystem 7809 7810 usbcore.quirks= 7811 [USB] A list of quirk entries to augment the built-in 7812 usb core quirk list. List entries are separated by 7813 commas. Each entry has the form 7814 VendorID:ProductID:Flags. The IDs are 4-digit hex 7815 numbers and Flags is a set of letters. Each letter 7816 will change the built-in quirk; setting it if it is 7817 clear and clearing it if it is set. The letters have 7818 the following meanings: 7819 a = USB_QUIRK_STRING_FETCH_255 (string 7820 descriptors must not be fetched using 7821 a 255-byte read); 7822 b = USB_QUIRK_RESET_RESUME (device can't resume 7823 correctly so reset it instead); 7824 c = USB_QUIRK_NO_SET_INTF (device can't handle 7825 Set-Interface requests); 7826 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't 7827 handle its Configuration or Interface 7828 strings); 7829 e = USB_QUIRK_RESET (device can't be reset 7830 (e.g morph devices), don't use reset); 7831 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has 7832 more interface descriptions than the 7833 bNumInterfaces count, and can't handle 7834 talking to these interfaces); 7835 g = USB_QUIRK_DELAY_INIT (device needs a pause 7836 during initialization, after we read 7837 the device descriptor); 7838 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For 7839 high speed and super speed interrupt 7840 endpoints, the USB 2.0 and USB 3.0 spec 7841 require the interval in microframes (1 7842 microframe = 125 microseconds) to be 7843 calculated as interval = 2 ^ 7844 (bInterval-1). 7845 Devices with this quirk report their 7846 bInterval as the result of this 7847 calculation instead of the exponent 7848 variable used in the calculation); 7849 i = USB_QUIRK_DEVICE_QUALIFIER (device can't 7850 handle device_qualifier descriptor 7851 requests); 7852 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device 7853 generates spurious wakeup, ignore 7854 remote wakeup capability); 7855 k = USB_QUIRK_NO_LPM (device can't handle Link 7856 Power Management); 7857 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL 7858 (Device reports its bInterval as linear 7859 frames instead of the USB 2.0 7860 calculation); 7861 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs 7862 to be disconnected before suspend to 7863 prevent spurious wakeup); 7864 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a 7865 pause after every control message); 7866 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra 7867 delay after resetting its port); 7868 p = USB_QUIRK_SHORT_SET_ADDRESS_REQ_TIMEOUT 7869 (Reduce timeout of the SET_ADDRESS 7870 request from 5000 ms to 500 ms); 7871 Example: quirks=0781:5580:bk,0a5c:5834:gij 7872 7873 usbhid.mousepoll= 7874 [USBHID] The interval which mice are to be polled at. 7875 7876 usbhid.jspoll= 7877 [USBHID] The interval which joysticks are to be polled at. 7878 7879 usbhid.kbpoll= 7880 [USBHID] The interval which keyboards are to be polled at. 7881 7882 usb-storage.delay_use= 7883 [UMS] The delay in seconds before a new device is 7884 scanned for Logical Units (default 1). 7885 Optionally the delay in milliseconds if the value has 7886 suffix with "ms". 7887 Example: delay_use=2567ms 7888 7889 usb-storage.quirks= 7890 [UMS] A list of quirks entries to supplement or 7891 override the built-in unusual_devs list. List 7892 entries are separated by commas. Each entry has 7893 the form VID:PID:Flags where VID and PID are Vendor 7894 and Product ID values (4-digit hex numbers) and 7895 Flags is a set of characters, each corresponding 7896 to a common usb-storage quirk flag as follows: 7897 a = SANE_SENSE (collect more than 18 bytes 7898 of sense data, not on uas); 7899 b = BAD_SENSE (don't collect more than 18 7900 bytes of sense data, not on uas); 7901 c = FIX_CAPACITY (decrease the reported 7902 device capacity by one sector); 7903 d = NO_READ_DISC_INFO (don't use 7904 READ_DISC_INFO command, not on uas); 7905 e = NO_READ_CAPACITY_16 (don't use 7906 READ_CAPACITY_16 command); 7907 f = NO_REPORT_OPCODES (don't use report opcodes 7908 command, uas only); 7909 g = MAX_SECTORS_240 (don't transfer more than 7910 240 sectors at a time, uas only); 7911 h = CAPACITY_HEURISTICS (decrease the 7912 reported device capacity by one 7913 sector if the number is odd); 7914 i = IGNORE_DEVICE (don't bind to this 7915 device); 7916 j = NO_REPORT_LUNS (don't use report luns 7917 command, uas only); 7918 k = NO_SAME (do not use WRITE_SAME, uas only) 7919 l = NOT_LOCKABLE (don't try to lock and 7920 unlock ejectable media, not on uas); 7921 m = MAX_SECTORS_64 (don't transfer more 7922 than 64 sectors = 32 KB at a time, 7923 not on uas); 7924 n = INITIAL_READ10 (force a retry of the 7925 initial READ(10) command, not on uas); 7926 o = CAPACITY_OK (accept the capacity 7927 reported by the device, not on uas); 7928 p = WRITE_CACHE (the device cache is ON 7929 by default, not on uas); 7930 r = IGNORE_RESIDUE (the device reports 7931 bogus residue values, not on uas); 7932 s = SINGLE_LUN (the device has only one 7933 Logical Unit); 7934 t = NO_ATA_1X (don't allow ATA(12) and ATA(16) 7935 commands, uas only); 7936 u = IGNORE_UAS (don't bind to the uas driver); 7937 w = NO_WP_DETECT (don't test whether the 7938 medium is write-protected). 7939 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE 7940 even if the device claims no cache, 7941 not on uas) 7942 Example: quirks=0419:aaf5:rl,0421:0433:rc 7943 7944 user_debug= [KNL,ARM] 7945 Format: <int> 7946 See arch/arm/Kconfig.debug help text. 7947 1 - undefined instruction events 7948 2 - system calls 7949 4 - invalid data aborts 7950 8 - SIGSEGV faults 7951 16 - SIGBUS faults 7952 Example: user_debug=31 7953 7954 vdso= [X86,SH,SPARC] 7955 On X86_32, this is an alias for vdso32=. Otherwise: 7956 7957 vdso=1: enable VDSO (the default) 7958 vdso=0: disable VDSO mapping 7959 7960 vdso32= [X86] Control the 32-bit vDSO 7961 vdso32=1: enable 32-bit VDSO 7962 vdso32=0 or vdso32=2: disable 32-bit VDSO 7963 7964 See the help text for CONFIG_COMPAT_VDSO for more 7965 details. If CONFIG_COMPAT_VDSO is set, the default is 7966 vdso32=0; otherwise, the default is vdso32=1. 7967 7968 For compatibility with older kernels, vdso32=2 is an 7969 alias for vdso32=0. 7970 7971 Try vdso32=0 if you encounter an error that says: 7972 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed! 7973 7974 video= [FB,EARLY] Frame buffer configuration 7975 See Documentation/fb/modedb.rst. 7976 7977 video.brightness_switch_enabled= [ACPI] 7978 Format: [0|1] 7979 If set to 1, on receiving an ACPI notify event 7980 generated by hotkey, video driver will adjust brightness 7981 level and then send out the event to user space through 7982 the allocated input device. If set to 0, video driver 7983 will only send out the event without touching backlight 7984 brightness level. 7985 default: 1 7986 7987 virtio_mmio.device= 7988 [VMMIO] Memory mapped virtio (platform) device. 7989 7990 <size>@<baseaddr>:<irq>[:<id>] 7991 where: 7992 <size> := size (can use standard suffixes 7993 like K, M and G) 7994 <baseaddr> := physical base address 7995 <irq> := interrupt number (as passed to 7996 request_irq()) 7997 <id> := (optional) platform device id 7998 example: 7999 virtio_mmio.device=1K@0x100b0000:48:7 8000 8001 Can be used multiple times for multiple devices. 8002 8003 vga= [BOOT,X86-32] Select a particular video mode 8004 See Documentation/arch/x86/boot.rst and 8005 Documentation/admin-guide/svga.rst. 8006 Use vga=ask for menu. 8007 This is actually a boot loader parameter; the value is 8008 passed to the kernel using a special protocol. 8009 8010 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y. 8011 May slow down system boot speed, especially when 8012 enabled on systems with a large amount of memory. 8013 All options are enabled by default, and this 8014 interface is meant to allow for selectively 8015 enabling or disabling specific virtual memory 8016 debugging features. 8017 8018 Available options are: 8019 P Enable page structure init time poisoning 8020 - Disable all of the above options 8021 8022 vmalloc=nn[KMG] [KNL,BOOT,EARLY] Forces the vmalloc area to have an 8023 exact size of <nn>. This can be used to increase 8024 the minimum size (128MB on x86, arm32 platforms). 8025 It can also be used to decrease the size and leave more room 8026 for directly mapped kernel RAM. Note that this parameter does 8027 not exist on many other platforms (including arm64, alpha, 8028 loongarch, arc, csky, hexagon, microblaze, mips, nios2, openrisc, 8029 parisc, m64k, powerpc, riscv, sh, um, xtensa, s390, sparc). 8030 8031 vmcp_cma=nn[MG] [KNL,S390,EARLY] 8032 Sets the memory size reserved for contiguous memory 8033 allocations for the vmcp device driver. 8034 8035 vmhalt= [KNL,S390] Perform z/VM CP command after system halt. 8036 Format: <command> 8037 8038 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic. 8039 Format: <command> 8040 8041 vmpoff= [KNL,S390] Perform z/VM CP command after power off. 8042 Format: <command> 8043 8044 vsyscall= [X86-64,EARLY] 8045 Controls the behavior of vsyscalls (i.e. calls to 8046 fixed addresses of 0xffffffffff600x00 from legacy 8047 code). Most statically-linked binaries and older 8048 versions of glibc use these calls. Because these 8049 functions are at fixed addresses, they make nice 8050 targets for exploits that can control RIP. 8051 8052 emulate Vsyscalls turn into traps and are emulated 8053 reasonably safely. The vsyscall page is 8054 readable. 8055 8056 xonly [default] Vsyscalls turn into traps and are 8057 emulated reasonably safely. The vsyscall 8058 page is not readable. 8059 8060 none Vsyscalls don't work at all. This makes 8061 them quite hard to use for exploits but 8062 might break your system. 8063 8064 vt.color= [VT] Default text color. 8065 Format: 0xYX, X = foreground, Y = background. 8066 Default: 0x07 = light gray on black. 8067 8068 vt.cur_default= [VT] Default cursor shape. 8069 Format: 0xCCBBAA, where AA, BB, and CC are the same as 8070 the parameters of the <Esc>[?A;B;Cc escape sequence; 8071 see vga-softcursor.rst. Default: 2 = underline. 8072 8073 vt.default_blu= [VT] 8074 Format: <blue0>,<blue1>,<blue2>,...,<blue15> 8075 Change the default blue palette of the console. 8076 This is a 16-member array composed of values 8077 ranging from 0-255. 8078 8079 vt.default_grn= [VT] 8080 Format: <green0>,<green1>,<green2>,...,<green15> 8081 Change the default green palette of the console. 8082 This is a 16-member array composed of values 8083 ranging from 0-255. 8084 8085 vt.default_red= [VT] 8086 Format: <red0>,<red1>,<red2>,...,<red15> 8087 Change the default red palette of the console. 8088 This is a 16-member array composed of values 8089 ranging from 0-255. 8090 8091 vt.default_utf8= 8092 [VT] 8093 Format=<0|1> 8094 Set system-wide default UTF-8 mode for all tty's. 8095 Default is 1, i.e. UTF-8 mode is enabled for all 8096 newly opened terminals. 8097 8098 vt.global_cursor_default= 8099 [VT] 8100 Format=<-1|0|1> 8101 Set system-wide default for whether a cursor 8102 is shown on new VTs. Default is -1, 8103 i.e. cursors will be created by default unless 8104 overridden by individual drivers. 0 will hide 8105 cursors, 1 will display them. 8106 8107 vt.italic= [VT] Default color for italic text; 0-15. 8108 Default: 2 = green. 8109 8110 vt.underline= [VT] Default color for underlined text; 0-15. 8111 Default: 3 = cyan. 8112 8113 watchdog timers [HW,WDT] For information on watchdog timers, 8114 see Documentation/watchdog/watchdog-parameters.rst 8115 or other driver-specific files in the 8116 Documentation/watchdog/ directory. 8117 8118 watchdog_thresh= 8119 [KNL] 8120 Set the hard lockup detector stall duration 8121 threshold in seconds. The soft lockup detector 8122 threshold is set to twice the value. A value of 0 8123 disables both lockup detectors. Default is 10 8124 seconds. 8125 8126 workqueue.unbound_cpus= 8127 [KNL,SMP] Specify to constrain one or some CPUs 8128 to use in unbound workqueues. 8129 Format: <cpu-list> 8130 By default, all online CPUs are available for 8131 unbound workqueues. 8132 8133 workqueue.watchdog_thresh= 8134 If CONFIG_WQ_WATCHDOG is configured, workqueue can 8135 warn stall conditions and dump internal state to 8136 help debugging. 0 disables workqueue stall 8137 detection; otherwise, it's the stall threshold 8138 duration in seconds. The default value is 30 and 8139 it can be updated at runtime by writing to the 8140 corresponding sysfs file. 8141 8142 workqueue.panic_on_stall=<uint> 8143 Panic when workqueue stall is detected by 8144 CONFIG_WQ_WATCHDOG. It sets the number times of the 8145 stall to trigger panic. 8146 8147 The default is 0, which disables the panic on stall. 8148 8149 workqueue.cpu_intensive_thresh_us= 8150 Per-cpu work items which run for longer than this 8151 threshold are automatically considered CPU intensive 8152 and excluded from concurrency management to prevent 8153 them from noticeably delaying other per-cpu work 8154 items. Default is 10000 (10ms). 8155 8156 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel 8157 will report the work functions which violate this 8158 threshold repeatedly. They are likely good 8159 candidates for using WQ_UNBOUND workqueues instead. 8160 8161 workqueue.cpu_intensive_warning_thresh=<uint> 8162 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel 8163 will report the work functions which violate the 8164 intensive_threshold_us repeatedly. In order to prevent 8165 spurious warnings, start printing only after a work 8166 function has violated this threshold number of times. 8167 8168 The default is 4 times. 0 disables the warning. 8169 8170 workqueue.power_efficient 8171 Per-cpu workqueues are generally preferred because 8172 they show better performance thanks to cache 8173 locality; unfortunately, per-cpu workqueues tend to 8174 be more power hungry than unbound workqueues. 8175 8176 Enabling this makes the per-cpu workqueues which 8177 were observed to contribute significantly to power 8178 consumption unbound, leading to measurably lower 8179 power usage at the cost of small performance 8180 overhead. 8181 8182 The default value of this parameter is determined by 8183 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT. 8184 8185 workqueue.default_affinity_scope= 8186 Select the default affinity scope to use for unbound 8187 workqueues. Can be one of "cpu", "smt", "cache", 8188 "numa" and "system". Default is "cache". For more 8189 information, see the Affinity Scopes section in 8190 Documentation/core-api/workqueue.rst. 8191 8192 This can be changed after boot by writing to the 8193 matching /sys/module/workqueue/parameters file. All 8194 workqueues with the "default" affinity scope will be 8195 updated accordingly. 8196 8197 workqueue.debug_force_rr_cpu 8198 Workqueue used to implicitly guarantee that work 8199 items queued without explicit CPU specified are put 8200 on the local CPU. This guarantee is no longer true 8201 and while local CPU is still preferred work items 8202 may be put on foreign CPUs. This debug option 8203 forces round-robin CPU selection to flush out 8204 usages which depend on the now broken guarantee. 8205 When enabled, memory and cache locality will be 8206 impacted. 8207 8208 writecombine= [LOONGARCH,EARLY] Control the MAT (Memory Access 8209 Type) of ioremap_wc(). 8210 8211 on - Enable writecombine, use WUC for ioremap_wc() 8212 off - Disable writecombine, use SUC for ioremap_wc() 8213 8214 x2apic_phys [X86-64,APIC,EARLY] Use x2apic physical mode instead of 8215 default x2apic cluster mode on platforms 8216 supporting x2apic. 8217 8218 xen_512gb_limit [KNL,X86-64,XEN] 8219 Restricts the kernel running paravirtualized under Xen 8220 to use only up to 512 GB of RAM. The reason to do so is 8221 crash analysis tools and Xen tools for doing domain 8222 save/restore/migration must be enabled to handle larger 8223 domains. 8224 8225 xen_emul_unplug= [HW,X86,XEN,EARLY] 8226 Unplug Xen emulated devices 8227 Format: [unplug0,][unplug1] 8228 ide-disks -- unplug primary master IDE devices 8229 aux-ide-disks -- unplug non-primary-master IDE devices 8230 nics -- unplug network devices 8231 all -- unplug all emulated devices (NICs and IDE disks) 8232 unnecessary -- unplugging emulated devices is 8233 unnecessary even if the host did not respond to 8234 the unplug protocol 8235 never -- do not unplug even if version check succeeds 8236 8237 xen_legacy_crash [X86,XEN,EARLY] 8238 Crash from Xen panic notifier, without executing late 8239 panic() code such as dumping handler. 8240 8241 xen_mc_debug [X86,XEN,EARLY] 8242 Enable multicall debugging when running as a Xen PV guest. 8243 Enabling this feature will reduce performance a little 8244 bit, so it should only be enabled for obtaining extended 8245 debug data in case of multicall errors. 8246 8247 xen_msr_safe= [X86,XEN,EARLY] 8248 Format: <bool> 8249 Select whether to always use non-faulting (safe) MSR 8250 access functions when running as Xen PV guest. The 8251 default value is controlled by CONFIG_XEN_PV_MSR_SAFE. 8252 8253 xen_nopv [X86] 8254 Disables the PV optimizations forcing the HVM guest to 8255 run as generic HVM guest with no PV drivers. 8256 This option is obsoleted by the "nopv" option, which 8257 has equivalent effect for XEN platform. 8258 8259 xen_no_vector_callback 8260 [KNL,X86,XEN,EARLY] Disable the vector callback for Xen 8261 event channel interrupts. 8262 8263 xen_scrub_pages= [XEN] 8264 Boolean option to control scrubbing pages before giving them back 8265 to Xen, for use by other domains. Can be also changed at runtime 8266 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages. 8267 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT. 8268 8269 xen_timer_slop= [X86-64,XEN,EARLY] 8270 Set the timer slop (in nanoseconds) for the virtual Xen 8271 timers (default is 100000). This adjusts the minimum 8272 delta of virtualized Xen timers, where lower values 8273 improve timer resolution at the expense of processing 8274 more timer interrupts. 8275 8276 xen.balloon_boot_timeout= [XEN] 8277 The time (in seconds) to wait before giving up to boot 8278 in case initial ballooning fails to free enough memory. 8279 Applies only when running as HVM or PVH guest and 8280 started with less memory configured than allowed at 8281 max. Default is 180. 8282 8283 xen.event_eoi_delay= [XEN] 8284 How long to delay EOI handling in case of event 8285 storms (jiffies). Default is 10. 8286 8287 xen.event_loop_timeout= [XEN] 8288 After which time (jiffies) the event handling loop 8289 should start to delay EOI handling. Default is 2. 8290 8291 xen.fifo_events= [XEN] 8292 Boolean parameter to disable using fifo event handling 8293 even if available. Normally fifo event handling is 8294 preferred over the 2-level event handling, as it is 8295 fairer and the number of possible event channels is 8296 much higher. Default is on (use fifo events). 8297 8298 xirc2ps_cs= [NET,PCMCIA] 8299 Format: 8300 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]] 8301 8302 xive= [PPC] 8303 By default on POWER9 and above, the kernel will 8304 natively use the XIVE interrupt controller. This option 8305 allows the fallback firmware mode to be used: 8306 8307 off Fallback to firmware control of XIVE interrupt 8308 controller on both pseries and powernv 8309 platforms. Only useful on POWER9 and above. 8310 8311 xive.store-eoi=off [PPC] 8312 By default on POWER10 and above, the kernel will use 8313 stores for EOI handling when the XIVE interrupt mode 8314 is active. This option allows the XIVE driver to use 8315 loads instead, as on POWER9. 8316 8317 xhci-hcd.quirks [USB,KNL] 8318 A hex value specifying bitmask with supplemental xhci 8319 host controller quirks. Meaning of each bit can be 8320 consulted in header drivers/usb/host/xhci.h. 8321 8322 xmon [PPC,EARLY] 8323 Format: { early | on | rw | ro | off } 8324 Controls if xmon debugger is enabled. Default is off. 8325 Passing only "xmon" is equivalent to "xmon=early". 8326 early Call xmon as early as possible on boot; xmon 8327 debugger is called from setup_arch(). 8328 on xmon debugger hooks will be installed so xmon 8329 is only called on a kernel crash. Default mode, 8330 i.e. either "ro" or "rw" mode, is controlled 8331 with CONFIG_XMON_DEFAULT_RO_MODE. 8332 rw xmon debugger hooks will be installed so xmon 8333 is called only on a kernel crash, mode is write, 8334 meaning SPR registers, memory and, other data 8335 can be written using xmon commands. 8336 ro same as "rw" option above but SPR registers, 8337 memory, and other data can't be written using 8338 xmon commands. 8339 off xmon is disabled. 8340