1 ACPI ACPI support is enabled. 2 AGP AGP (Accelerated Graphics Port) is enabled. 3 ALSA ALSA sound support is enabled. 4 APIC APIC support is enabled. 5 APM Advanced Power Management support is enabled. 6 APPARMOR AppArmor support is enabled. 7 ARM ARM architecture is enabled. 8 ARM64 ARM64 architecture is enabled. 9 CLK Common clock infrastructure is enabled. 10 CMA Contiguous Memory Area support is enabled. 11 CPU_FREQ CPU frequency scaling is enabled. 12 CPU_IDLE CPU idle time management is enabled. 13 DRM Direct Rendering Management support is enabled. 14 DYNAMIC_DEBUG Build in debug messages and enable them at runtime 15 EARLY Parameter processed too early to be embedded in initrd. 16 EDD BIOS Enhanced Disk Drive Services (EDD) is enabled 17 EFI EFI Partitioning (GPT) is enabled 18 EVM Extended Verification Module 19 FB The frame buffer device is enabled. 20 FTRACE Function tracing enabled. 21 GCOV GCOV profiling is enabled. 22 HIBERNATION HIBERNATION is enabled. 23 HW Appropriate hardware is enabled. 24 HYPER_V HYPERV support is enabled. 25 IMA Integrity measurement architecture is enabled. 26 IP_PNP IP DHCP, BOOTP, or RARP is enabled. 27 IPV6 IPv6 support is enabled. 28 ISAPNP ISA PnP code is enabled. 29 ISOL CPU Isolation is enabled. 30 JOY Appropriate joystick support is enabled. 31 KGDB Kernel debugger support is enabled. 32 KVM Kernel Virtual Machine support is enabled. 33 LIBATA Libata driver is enabled 34 LOONGARCH LoongArch architecture is enabled. 35 LOOP Loopback device support is enabled. 36 LP Printer support is enabled. 37 M68k M68k architecture is enabled. 38 These options have more detailed description inside of 39 Documentation/arch/m68k/kernel-options.rst. 40 MIPS MIPS architecture is enabled. 41 MOUSE Appropriate mouse support is enabled. 42 MSI Message Signaled Interrupts (PCI). 43 MTD MTD (Memory Technology Device) support is enabled. 44 NET Appropriate network support is enabled. 45 NFS Appropriate NFS support is enabled. 46 NUMA NUMA support is enabled. 47 OF Devicetree is enabled. 48 PARISC The PA-RISC architecture is enabled. 49 PCI PCI bus support is enabled. 50 PCIE PCI Express support is enabled. 51 PCMCIA The PCMCIA subsystem is enabled. 52 PM Power Management support is enabled. 53 PNP Plug & Play support is enabled. 54 PPC PowerPC architecture is enabled. 55 PPT Parallel port support is enabled. 56 PS2 Appropriate PS/2 support is enabled. 57 PV_OPS A paravirtualized kernel is enabled. 58 RAM RAM disk support is enabled. 59 RDT Intel Resource Director Technology. 60 RISCV RISCV architecture is enabled. 61 S390 S390 architecture is enabled. 62 SCSI Appropriate SCSI support is enabled. 63 A lot of drivers have their options described inside 64 the Documentation/scsi/ sub-directory. 65 SDW SoundWire support is enabled. 66 SECURITY Different security models are enabled. 67 SELINUX SELinux support is enabled. 68 SERIAL Serial support is enabled. 69 SH SuperH architecture is enabled. 70 SMP The kernel is an SMP kernel. 71 SPARC Sparc architecture is enabled. 72 SUSPEND System suspend states are enabled. 73 SWSUSP Software suspend (hibernation) is enabled. 74 TPM TPM drivers are enabled. 75 UMS USB Mass Storage support is enabled. 76 USB USB support is enabled. 77 NVME NVMe support is enabled 78 USBHID USB Human Interface Device support is enabled. 79 V4L Video For Linux support is enabled. 80 VGA The VGA console has been enabled. 81 VMMIO Driver for memory mapped virtio devices is enabled. 82 VT Virtual terminal support is enabled. 83 WDT Watchdog support is enabled. 84 X86-32 X86-32, aka i386 architecture is enabled. 85 X86-64 X86-64 architecture is enabled. 86 X86 Either 32-bit or 64-bit x86 (same as X86-32+X86-64) 87 X86_UV SGI UV support is enabled. 88 XEN Xen support is enabled 89 XTENSA xtensa architecture is enabled. 90 91In addition, the following text indicates that the option 92 93 BOOT Is a boot loader parameter. 94 BUGS= Relates to possible processor bugs on the said processor. 95 KNL Is a kernel start-up parameter. 96 97 98Kernel parameters 99 100 accept_memory= [MM] 101 Format: { eager | lazy } 102 default: lazy 103 By default, unaccepted memory is accepted lazily to 104 avoid prolonged boot times. The lazy option will add 105 some runtime overhead until all memory is eventually 106 accepted. In most cases the overhead is negligible. 107 For some workloads or for debugging purposes 108 accept_memory=eager can be used to accept all memory 109 at once during boot. 110 111 acpi= [HW,ACPI,X86,ARM64,RISCV64,EARLY] 112 Advanced Configuration and Power Interface 113 Format: { force | on | off | strict | noirq | rsdt | 114 copy_dsdt | nospcr } 115 force -- enable ACPI if default was off 116 on -- enable ACPI but allow fallback to DT [arm64,riscv64] 117 off -- disable ACPI if default was on 118 noirq -- do not use ACPI for IRQ routing 119 strict -- Be less tolerant of platforms that are not 120 strictly ACPI specification compliant. 121 rsdt -- prefer RSDT over (default) XSDT 122 copy_dsdt -- copy DSDT to memory 123 nocmcff -- Disable firmware first mode for corrected 124 errors. This disables parsing the HEST CMC error 125 source to check if firmware has set the FF flag. This 126 may result in duplicate corrected error reports. 127 nospcr -- disable console in ACPI SPCR table as 128 default _serial_ console on ARM64 129 spcr -- enable console in ACPI SPCR table as 130 default _serial_ console on x86 131 For ARM64, ONLY "acpi=off", "acpi=on", "acpi=force" or 132 "acpi=nospcr" are available 133 For RISCV64, ONLY "acpi=off", "acpi=on" or "acpi=force" 134 are available 135 136 See also Documentation/power/runtime_pm.rst, pci=noacpi 137 138 acpi_apic_instance= [ACPI,IOAPIC,EARLY] 139 Format: <int> 140 2: use 2nd APIC table, if available 141 1,0: use 1st APIC table 142 default: 0 143 144 acpi_backlight= [HW,ACPI] 145 { vendor | video | native | none } 146 If set to vendor, prefer vendor-specific driver 147 (e.g. thinkpad_acpi, sony_acpi, etc.) instead 148 of the ACPI video.ko driver. 149 If set to video, use the ACPI video.ko driver. 150 If set to native, use the device's native backlight mode. 151 If set to none, disable the ACPI backlight interface. 152 153 acpi_force_32bit_fadt_addr [ACPI,EARLY] 154 force FADT to use 32 bit addresses rather than the 155 64 bit X_* addresses. Some firmware have broken 64 156 bit addresses for force ACPI ignore these and use 157 the older legacy 32 bit addresses. 158 159 acpica_no_return_repair [HW, ACPI] 160 Disable AML predefined validation mechanism 161 This mechanism can repair the evaluation result to make 162 the return objects more ACPI specification compliant. 163 This option is useful for developers to identify the 164 root cause of an AML interpreter issue when the issue 165 has something to do with the repair mechanism. 166 167 acpi.debug_layer= [HW,ACPI,ACPI_DEBUG] 168 acpi.debug_level= [HW,ACPI,ACPI_DEBUG] 169 Format: <int> 170 CONFIG_ACPI_DEBUG must be enabled to produce any ACPI 171 debug output. Bits in debug_layer correspond to a 172 _COMPONENT in an ACPI source file, e.g., 173 #define _COMPONENT ACPI_EVENTS 174 Bits in debug_level correspond to a level in 175 ACPI_DEBUG_PRINT statements, e.g., 176 ACPI_DEBUG_PRINT((ACPI_DB_INFO, ... 177 The debug_level mask defaults to "info". See 178 Documentation/firmware-guide/acpi/debug.rst for more information about 179 debug layers and levels. 180 181 Enable processor driver info messages: 182 acpi.debug_layer=0x20000000 183 Enable AML "Debug" output, i.e., stores to the Debug 184 object while interpreting AML: 185 acpi.debug_layer=0xffffffff acpi.debug_level=0x2 186 Enable all messages related to ACPI hardware: 187 acpi.debug_layer=0x2 acpi.debug_level=0xffffffff 188 189 Some values produce so much output that the system is 190 unusable. The "log_buf_len" parameter may be useful 191 if you need to capture more output. 192 193 acpi.poweroff_on_fatal= [ACPI] 194 {0 | 1} 195 Causes the system to poweroff when the ACPI bytecode signals 196 a fatal error. The default value of this setting is 1. 197 Overriding this value should only be done for diagnosing 198 ACPI firmware problems, as the system might behave erratically 199 after having encountered a fatal ACPI error. 200 201 acpi_enforce_resources= [ACPI] 202 { strict | lax | no } 203 Check for resource conflicts between native drivers 204 and ACPI OperationRegions (SystemIO and SystemMemory 205 only). IO ports and memory declared in ACPI might be 206 used by the ACPI subsystem in arbitrary AML code and 207 can interfere with legacy drivers. 208 strict (default): access to resources claimed by ACPI 209 is denied; legacy drivers trying to access reserved 210 resources will fail to bind to device using them. 211 lax: access to resources claimed by ACPI is allowed; 212 legacy drivers trying to access reserved resources 213 will bind successfully but a warning message is logged. 214 no: ACPI OperationRegions are not marked as reserved, 215 no further checks are performed. 216 217 acpi_force_table_verification [HW,ACPI,EARLY] 218 Enable table checksum verification during early stage. 219 By default, this is disabled due to x86 early mapping 220 size limitation. 221 222 acpi_irq_balance [HW,ACPI] 223 ACPI will balance active IRQs 224 default in APIC mode 225 226 acpi_irq_nobalance [HW,ACPI] 227 ACPI will not move active IRQs (default) 228 default in PIC mode 229 230 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA 231 Format: <irq>,<irq>... 232 233 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for 234 use by PCI 235 Format: <irq>,<irq>... 236 237 acpi_mask_gpe= [HW,ACPI] 238 Due to the existence of _Lxx/_Exx, some GPEs triggered 239 by unsupported hardware/firmware features can result in 240 GPE floodings that cannot be automatically disabled by 241 the GPE dispatcher. 242 This facility can be used to prevent such uncontrolled 243 GPE floodings. 244 Format: <byte> or <bitmap-list> 245 246 acpi_no_auto_serialize [HW,ACPI] 247 Disable auto-serialization of AML methods 248 AML control methods that contain the opcodes to create 249 named objects will be marked as "Serialized" by the 250 auto-serialization feature. 251 This feature is enabled by default. 252 This option allows to turn off the feature. 253 254 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump 255 kernels. 256 257 acpi_no_static_ssdt [HW,ACPI,EARLY] 258 Disable installation of static SSDTs at early boot time 259 By default, SSDTs contained in the RSDT/XSDT will be 260 installed automatically and they will appear under 261 /sys/firmware/acpi/tables. 262 This option turns off this feature. 263 Note that specifying this option does not affect 264 dynamic table installation which will install SSDT 265 tables to /sys/firmware/acpi/tables/dynamic. 266 267 acpi_no_watchdog [HW,ACPI,WDT] 268 Ignore the ACPI-based watchdog interface (WDAT) and let 269 a native driver control the watchdog device instead. 270 271 acpi_rsdp= [ACPI,EFI,KEXEC,EARLY] 272 Pass the RSDP address to the kernel, mostly used 273 on machines running EFI runtime service to boot the 274 second kernel for kdump. 275 276 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS 277 Format: To spoof as Windows 98: ="Microsoft Windows" 278 279 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead 280 of 2 which is mandated by ACPI 6) as the supported ACPI 281 specification revision (when using this switch, it may 282 be necessary to carry out a cold reboot _twice_ in a 283 row to make it take effect on the platform firmware). 284 285 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings 286 acpi_osi="string1" # add string1 287 acpi_osi="!string2" # remove string2 288 acpi_osi=!* # remove all strings 289 acpi_osi=! # disable all built-in OS vendor 290 strings 291 acpi_osi=!! # enable all built-in OS vendor 292 strings 293 acpi_osi= # disable all strings 294 295 'acpi_osi=!' can be used in combination with single or 296 multiple 'acpi_osi="string1"' to support specific OS 297 vendor string(s). Note that such command can only 298 affect the default state of the OS vendor strings, thus 299 it cannot affect the default state of the feature group 300 strings and the current state of the OS vendor strings, 301 specifying it multiple times through kernel command line 302 is meaningless. This command is useful when one do not 303 care about the state of the feature group strings which 304 should be controlled by the OSPM. 305 Examples: 306 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent 307 to 'acpi_osi="Windows 2000" acpi_osi=!', they all 308 can make '_OSI("Windows 2000")' TRUE. 309 310 'acpi_osi=' cannot be used in combination with other 311 'acpi_osi=' command lines, the _OSI method will not 312 exist in the ACPI namespace. NOTE that such command can 313 only affect the _OSI support state, thus specifying it 314 multiple times through kernel command line is also 315 meaningless. 316 Examples: 317 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)' 318 FALSE. 319 320 'acpi_osi=!*' can be used in combination with single or 321 multiple 'acpi_osi="string1"' to support specific 322 string(s). Note that such command can affect the 323 current state of both the OS vendor strings and the 324 feature group strings, thus specifying it multiple times 325 through kernel command line is meaningful. But it may 326 still not able to affect the final state of a string if 327 there are quirks related to this string. This command 328 is useful when one want to control the state of the 329 feature group strings to debug BIOS issues related to 330 the OSPM features. 331 Examples: 332 1. 'acpi_osi="Module Device" acpi_osi=!*' can make 333 '_OSI("Module Device")' FALSE. 334 2. 'acpi_osi=!* acpi_osi="Module Device"' can make 335 '_OSI("Module Device")' TRUE. 336 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is 337 equivalent to 338 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"' 339 and 340 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!', 341 they all will make '_OSI("Windows 2000")' TRUE. 342 343 acpi_pm_good [X86] 344 Override the pmtimer bug detection: force the kernel 345 to assume that this machine's pmtimer latches its value 346 and always returns good values. 347 348 acpi_sci= [HW,ACPI,EARLY] ACPI System Control Interrupt trigger mode 349 Format: { level | edge | high | low } 350 351 acpi_skip_timer_override [HW,ACPI,EARLY] 352 Recognize and ignore IRQ0/pin2 Interrupt Override. 353 For broken nForce2 BIOS resulting in XT-PIC timer. 354 355 acpi_sleep= [HW,ACPI] Sleep options 356 Format: { s3_bios, s3_mode, s3_beep, s4_hwsig, 357 s4_nohwsig, old_ordering, nonvs, 358 sci_force_enable, nobl } 359 See Documentation/power/video.rst for information on 360 s3_bios and s3_mode. 361 s3_beep is for debugging; it makes the PC's speaker beep 362 as soon as the kernel's real-mode entry point is called. 363 s4_hwsig causes the kernel to check the ACPI hardware 364 signature during resume from hibernation, and gracefully 365 refuse to resume if it has changed. This complies with 366 the ACPI specification but not with reality, since 367 Windows does not do this and many laptops do change it 368 on docking. So the default behaviour is to allow resume 369 and simply warn when the signature changes, unless the 370 s4_hwsig option is enabled. 371 s4_nohwsig prevents ACPI hardware signature from being 372 used (or even warned about) during resume. 373 old_ordering causes the ACPI 1.0 ordering of the _PTS 374 control method, with respect to putting devices into 375 low power states, to be enforced (the ACPI 2.0 ordering 376 of _PTS is used by default). 377 nonvs prevents the kernel from saving/restoring the 378 ACPI NVS memory during suspend/hibernation and resume. 379 sci_force_enable causes the kernel to set SCI_EN directly 380 on resume from S1/S3 (which is against the ACPI spec, 381 but some broken systems don't work without it). 382 nobl causes the internal blacklist of systems known to 383 behave incorrectly in some ways with respect to system 384 suspend and resume to be ignored (use wisely). 385 386 acpi_use_timer_override [HW,ACPI,EARLY] 387 Use timer override. For some broken Nvidia NF5 boards 388 that require a timer override, but don't have HPET 389 390 add_efi_memmap [EFI,X86,EARLY] Include EFI memory map in 391 kernel's map of available physical RAM. 392 393 agp= [AGP] 394 { off | try_unsupported } 395 off: disable AGP support 396 try_unsupported: try to drive unsupported chipsets 397 (may crash computer or cause data corruption) 398 399 ALSA [HW,ALSA] 400 See Documentation/sound/alsa-configuration.rst 401 402 alignment= [KNL,ARM] 403 Allow the default userspace alignment fault handler 404 behaviour to be specified. Bit 0 enables warnings, 405 bit 1 enables fixups, and bit 2 sends a segfault. 406 407 align_va_addr= [X86-64] 408 Align virtual addresses by clearing slice [14:12] when 409 allocating a VMA at process creation time. This option 410 gives you up to 3% performance improvement on AMD F15h 411 machines (where it is enabled by default) for a 412 CPU-intensive style benchmark, and it can vary highly in 413 a microbenchmark depending on workload and compiler. 414 415 32: only for 32-bit processes 416 64: only for 64-bit processes 417 on: enable for both 32- and 64-bit processes 418 off: disable for both 32- and 64-bit processes 419 420 alloc_snapshot [FTRACE] 421 Allocate the ftrace snapshot buffer on boot up when the 422 main buffer is allocated. This is handy if debugging 423 and you need to use tracing_snapshot() on boot up, and 424 do not want to use tracing_snapshot_alloc() as it needs 425 to be done where GFP_KERNEL allocations are allowed. 426 427 allow_mismatched_32bit_el0 [ARM64,EARLY] 428 Allow execve() of 32-bit applications and setting of the 429 PER_LINUX32 personality on systems where only a strict 430 subset of the CPUs support 32-bit EL0. When this 431 parameter is present, the set of CPUs supporting 32-bit 432 EL0 is indicated by /sys/devices/system/cpu/aarch32_el0 433 and hot-unplug operations may be restricted. 434 435 See Documentation/arch/arm64/asymmetric-32bit.rst for more 436 information. 437 438 amd_iommu= [HW,X86-64] 439 Pass parameters to the AMD IOMMU driver in the system. 440 Possible values are: 441 fullflush - Deprecated, equivalent to iommu.strict=1 442 off - do not initialize any AMD IOMMU found in 443 the system 444 force_isolation - Force device isolation for all 445 devices. The IOMMU driver is not 446 allowed anymore to lift isolation 447 requirements as needed. This option 448 does not override iommu=pt 449 force_enable - Force enable the IOMMU on platforms known 450 to be buggy with IOMMU enabled. Use this 451 option with care. 452 pgtbl_v1 - Use v1 page table for DMA-API (Default). 453 pgtbl_v2 - Use v2 page table for DMA-API. 454 irtcachedis - Disable Interrupt Remapping Table (IRT) caching. 455 nohugepages - Limit page-sizes used for v1 page-tables 456 to 4 KiB. 457 v2_pgsizes_only - Limit page-sizes used for v1 page-tables 458 to 4KiB/2Mib/1GiB. 459 460 461 amd_iommu_dump= [HW,X86-64] 462 Enable AMD IOMMU driver option to dump the ACPI table 463 for AMD IOMMU. With this option enabled, AMD IOMMU 464 driver will print ACPI tables for AMD IOMMU during 465 IOMMU initialization. 466 467 amd_iommu_intr= [HW,X86-64] 468 Specifies one of the following AMD IOMMU interrupt 469 remapping modes: 470 legacy - Use legacy interrupt remapping mode. 471 vapic - Use virtual APIC mode, which allows IOMMU 472 to inject interrupts directly into guest. 473 This mode requires kvm-amd.avic=1. 474 (Default when IOMMU HW support is present.) 475 476 amd_pstate= [X86,EARLY] 477 disable 478 Do not enable amd_pstate as the default 479 scaling driver for the supported processors 480 passive 481 Use amd_pstate with passive mode as a scaling driver. 482 In this mode autonomous selection is disabled. 483 Driver requests a desired performance level and platform 484 tries to match the same performance level if it is 485 satisfied by guaranteed performance level. 486 active 487 Use amd_pstate_epp driver instance as the scaling driver, 488 driver provides a hint to the hardware if software wants 489 to bias toward performance (0x0) or energy efficiency (0xff) 490 to the CPPC firmware. then CPPC power algorithm will 491 calculate the runtime workload and adjust the realtime cores 492 frequency. 493 guided 494 Activate guided autonomous mode. Driver requests minimum and 495 maximum performance level and the platform autonomously 496 selects a performance level in this range and appropriate 497 to the current workload. 498 499 amd_prefcore= 500 [X86] 501 disable 502 Disable amd-pstate preferred core. 503 504 amijoy.map= [HW,JOY] Amiga joystick support 505 Map of devices attached to JOY0DAT and JOY1DAT 506 Format: <a>,<b> 507 See also Documentation/input/joydev/joystick.rst 508 509 analog.map= [HW,JOY] Analog joystick and gamepad support 510 Specifies type or capabilities of an analog joystick 511 connected to one of 16 gameports 512 Format: <type1>,<type2>,..<type16> 513 514 apc= [HW,SPARC] 515 Power management functions (SPARCstation-4/5 + deriv.) 516 Format: noidle 517 Disable APC CPU standby support. SPARCstation-Fox does 518 not play well with APC CPU idle - disable it if you have 519 APC and your system crashes randomly. 520 521 apic [APIC,X86-64] Use IO-APIC. Default. 522 523 apic= [APIC,X86,EARLY] Advanced Programmable Interrupt Controller 524 Change the output verbosity while booting 525 Format: { quiet (default) | verbose | debug } 526 Change the amount of debugging information output 527 when initialising the APIC and IO-APIC components. 528 529 apic_extnmi= [APIC,X86,EARLY] External NMI delivery setting 530 Format: { bsp (default) | all | none } 531 bsp: External NMI is delivered only to CPU 0 532 all: External NMIs are broadcast to all CPUs as a 533 backup of CPU 0 534 none: External NMI is masked for all CPUs. This is 535 useful so that a dump capture kernel won't be 536 shot down by NMI 537 538 apicpmtimer Do APIC timer calibration using the pmtimer. Implies 539 apicmaintimer. Useful when your PIT timer is totally 540 broken. 541 542 autoconf= [IPV6] 543 See Documentation/networking/ipv6.rst. 544 545 apm= [APM] Advanced Power Management 546 See header of arch/x86/kernel/apm_32.c. 547 548 apparmor= [APPARMOR] Disable or enable AppArmor at boot time 549 Format: { "0" | "1" } 550 See security/apparmor/Kconfig help text 551 0 -- disable. 552 1 -- enable. 553 Default value is set via kernel config option. 554 555 arm64.no32bit_el0 [ARM64] Unconditionally disable the execution of 556 32 bit applications. 557 558 arm64.nobti [ARM64] Unconditionally disable Branch Target 559 Identification support 560 561 arm64.nogcs [ARM64] Unconditionally disable Guarded Control Stack 562 support 563 564 arm64.nohaft [ARM64] Unconditionally disable Hardware managed Access 565 Flag for Table descriptors support 566 567 arm64.nomops [ARM64] Unconditionally disable Memory Copy and Memory 568 Set instructions support 569 570 arm64.nompam [ARM64] Unconditionally disable Memory Partitioning And 571 Monitoring support. Only for a machine that does not 572 boot without it. See Documentation/arch/arm64/mpam.rst 573 574 arm64.nomte [ARM64] Unconditionally disable Memory Tagging Extension 575 support 576 577 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication 578 support 579 580 arm64.nosme [ARM64] Unconditionally disable Scalable Matrix 581 Extension support 582 583 arm64.nosve [ARM64] Unconditionally disable Scalable Vector 584 Extension support 585 586 arm64.nowfxt [ARM64] Unconditionally disable Wait For Event with 587 Timeout and Wait For Interrupt with Timeout support 588 589 ataflop= [HW,M68k] 590 591 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess, 592 EzKey and similar keyboards 593 594 atkbd.reset= [HW] Reset keyboard during initialization 595 596 atkbd.set= [HW] Select keyboard code set 597 Format: <int> (2 = AT (default), 3 = PS/2) 598 599 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar 600 keyboards 601 602 atkbd.softraw= [HW] Choose between synthetic and real raw mode 603 Format: <bool> (0 = real, 1 = synthetic (default)) 604 605 atkbd.softrepeat= [HW] 606 Use software keyboard repeat 607 608 audit= [KNL] Enable the audit sub-system 609 Format: { "0" | "1" | "off" | "on" } 610 0 | off - kernel audit is disabled and can not be 611 enabled until the next reboot 612 unset - kernel audit is initialized but disabled and 613 will be fully enabled by the userspace auditd. 614 1 | on - kernel audit is initialized and partially 615 enabled, storing at most audit_backlog_limit 616 messages in RAM until it is fully enabled by the 617 userspace auditd. 618 Default: unset 619 620 audit_backlog_limit= [KNL] Set the audit queue size limit. 621 Format: <int> (must be >=0) 622 Default: 64 623 624 bdev_allow_write_mounted= 625 Format: <bool> 626 Control the ability to open a mounted block device 627 for writing, i.e., allow / disallow writes that bypass 628 the FS. This was implemented as a means to prevent 629 fuzzers from crashing the kernel by overwriting the 630 metadata underneath a mounted FS without its awareness. 631 This also prevents destructive formatting of mounted 632 filesystems by naive storage tooling that don't use 633 O_EXCL. Default is Y and can be changed through the 634 Kconfig option CONFIG_BLK_DEV_WRITE_MOUNTED. 635 636 bert_disable [ACPI] 637 Disable BERT OS support on buggy BIOSes. 638 639 bgrt_disable [ACPI,X86,EARLY] 640 Disable BGRT to avoid flickering OEM logo. 641 642 blkdevparts= Manual partition parsing of block device(s) for 643 embedded devices based on command line input. 644 See Documentation/block/cmdline-partition.rst 645 646 boot_delay= [KNL,EARLY] 647 Milliseconds to delay each printk during boot. 648 Only works if CONFIG_BOOT_PRINTK_DELAY is enabled, 649 and you may also have to specify "lpj=". Boot_delay 650 values larger than 10 seconds (10000) are assumed 651 erroneous and ignored. 652 Format: integer 653 654 bootconfig [KNL,EARLY] 655 Extended command line options can be added to an initrd 656 and this will cause the kernel to look for it. 657 658 See Documentation/admin-guide/bootconfig.rst 659 660 bpf.keyring_unsealed= 661 [BPF] When set to 1, leave the bpf keyring unsealed 662 after initialization so that userspace can provision 663 keys. Once the keyring is restricted it becomes active 664 and can be used for BPF program signature verification. 665 666 Setting this also means that the bpf keyring is the 667 only non-system keyring a loader may select for the 668 rest of the boot: caller-supplied user/session 669 keyrings are refused with -EPERM, whether or not 670 provisioning actually completed. The system keyrings 671 stay selectable. Leaving it unset keeps the prior 672 behaviour, where a caller-supplied keyring is allowed. 673 674 See Documentation/bpf/signing.rst 675 676 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards) 677 bttv.radio= Most important insmod options are available as 678 kernel args too. 679 bttv.pll= See Documentation/admin-guide/media/bttv.rst 680 bttv.tuner= 681 682 bulk_remove=off [PPC] This parameter disables the use of the pSeries 683 firmware feature for flushing multiple hpte entries 684 at a time. 685 686 c101= [NET] Moxa C101 synchronous serial card 687 688 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection. 689 Sometimes CPU hardware bugs make them report the cache 690 size incorrectly. The kernel will attempt work arounds 691 to fix known problems, but for some CPUs it is not 692 possible to determine what the correct size should be. 693 This option provides an override for these situations. 694 695 carrier_timeout= 696 [NET] Specifies amount of time (in seconds) that 697 the kernel should wait for a network carrier. By default 698 it waits 120 seconds. 699 700 ca_keys= [KEYS] This parameter identifies a specific key(s) on 701 the system trusted keyring to be used for certificate 702 trust validation. 703 format: { id:<keyid> | builtin } 704 705 cca= [MIPS,EARLY] Override the kernel pages' cache coherency 706 algorithm. Accepted values range from 0 to 7 707 inclusive. See arch/mips/include/asm/pgtable-bits.h 708 for platform specific values (SB1, Loongson3 and 709 others). 710 711 ccw_timeout_log [S390] 712 See Documentation/arch/s390/common_io.rst for details. 713 714 cfi= [X86-64] Set Control Flow Integrity checking features 715 when CONFIG_FINEIBT is enabled. 716 Format: feature[,feature...] 717 Default: auto 718 719 auto: Use FineIBT if IBT available, otherwise kCFI. 720 Under FineIBT, enable "paranoid" mode when 721 FRED is not available. 722 off: Turn off CFI checking. 723 kcfi: Use kCFI (disable FineIBT). 724 fineibt: Use FineIBT (even if IBT not available). 725 norand: Do not re-randomize CFI hashes. 726 paranoid: Add caller hash checking under FineIBT. 727 bhi: Enable register poisoning to stop speculation 728 across FineIBT. (Disabled by default.) 729 warn: Do not enforce CFI checking: warn only. 730 debug: Report CFI initialization details. 731 732 cgroup_disable= [KNL] Disable a particular controller or optional feature 733 Format: {name of the controller(s) or feature(s) to disable} 734 The effects of cgroup_disable=foo are: 735 - foo isn't auto-mounted if you mount all cgroups in 736 a single hierarchy 737 - foo isn't visible as an individually mountable 738 subsystem 739 - if foo is an optional feature then the feature is 740 disabled and corresponding cgroup files are not 741 created 742 {Currently only "memory" controller deal with this and 743 cut the overhead, others just disable the usage. So 744 only cgroup_disable=memory is actually worthy} 745 Specifying "pressure" disables per-cgroup pressure 746 stall information accounting feature 747 748 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1 749 Format: { { controller | "all" | "named" } 750 [,{ controller | "all" | "named" }...] } 751 Like cgroup_disable, but only applies to cgroup v1; 752 the blacklisted controllers remain available in cgroup2. 753 "all" blacklists all controllers and "named" disables 754 named mounts. Specifying both "all" and "named" disables 755 all v1 hierarchies. 756 757 cgroup_v1_proc= [KNL] Show also missing controllers in /proc/cgroups 758 Format: { "true" | "false" } 759 /proc/cgroups lists only v1 controllers by default. 760 This compatibility option enables listing also v2 761 controllers (whose v1 code is not compiled!), so that 762 semi-legacy software can check this file to decide 763 about usage of v2 (sic) controllers. 764 765 cgroup_favordynmods= [KNL] Enable or Disable favordynmods. 766 Format: { "true" | "false" } 767 Defaults to the value of CONFIG_CGROUP_FAVOR_DYNMODS. 768 769 cgroup.memory= [KNL] Pass options to the cgroup memory controller. 770 Format: <string> 771 nosocket -- Disable socket memory accounting. 772 nokmem -- Disable kernel memory accounting. 773 nobpf -- Disable BPF memory accounting. 774 775 check_pages= [MM,EARLY] Enable sanity checking of pages after 776 allocations / before freeing. This adds checks to catch 777 double-frees, use-after-frees, and other sources of 778 page corruption by inspecting page internals (flags, 779 mapcount/refcount, memcg_data, etc.). 780 Format: { "0" | "1" } 781 Default: 0 (1 if CONFIG_DEBUG_VM is set) 782 783 checkreqprot= [SELINUX] Set initial checkreqprot flag value. 784 Format: { "0" | "1" } 785 See security/selinux/Kconfig help text. 786 0 -- check protection applied by kernel (includes 787 any implied execute protection). 788 1 -- check protection requested by application. 789 Default value is set via a kernel config option. 790 Value can be changed at runtime via 791 /sys/fs/selinux/checkreqprot. 792 Setting checkreqprot to 1 is deprecated. 793 794 cio_ignore= [S390] 795 See Documentation/arch/s390/common_io.rst for details. 796 797 clk_ignore_unused 798 [CLK] 799 Prevents the clock framework from automatically gating 800 clocks that have not been explicitly enabled by a Linux 801 device driver but are enabled in hardware at reset or 802 by the bootloader/firmware. Note that this does not 803 force such clocks to be always-on nor does it reserve 804 those clocks in any way. This parameter is useful for 805 debug and development, but should not be needed on a 806 platform with proper driver support. For more 807 information, see Documentation/driver-api/clk.rst. 808 809 clock= [BUGS=X86-32, HW] gettimeofday clocksource override. 810 [Deprecated] 811 Forces specified clocksource (if available) to be used 812 when calculating gettimeofday(). If specified 813 clocksource is not available, it defaults to PIT. 814 Format: { pit | tsc | cyclone | pmtmr } 815 816 clocksource= Override the default clocksource 817 Format: <string> 818 Override the default clocksource and use the clocksource 819 with the name specified. 820 Some clocksource names to choose from, depending on 821 the platform: 822 [all] jiffies (this is the base, fallback clocksource) 823 [ACPI] acpi_pm 824 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2, 825 pxa_timer,timer3,32k_counter,timer0_1 826 [X86-32] pit,hpet,tsc; 827 scx200_hrt on Geode; cyclone on IBM x440 828 [MIPS] MIPS 829 [PARISC] cr16 830 [S390] tod 831 [SH] SuperH 832 [SPARC64] tick 833 [X86-64] hpet,tsc 834 835 clocksource.arm_arch_timer.evtstrm= 836 [ARM,ARM64,EARLY] 837 Format: <bool> 838 Enable/disable the eventstream feature of the ARM 839 architected timer so that code using WFE-based polling 840 loops can be debugged more effectively on production 841 systems. 842 843 clocksource.verify_n_cpus= [KNL] 844 Limit the number of CPUs checked for clocksources 845 marked with CLOCK_SOURCE_VERIFY_PERCPU that 846 are marked unstable due to excessive skew. 847 A negative value says to check all CPUs, while 848 zero says not to check any. Values larger than 849 nr_cpu_ids are silently truncated to nr_cpu_ids. 850 The actual CPUs are chosen randomly, with 851 no replacement if the same CPU is chosen twice. 852 853 clocksource-wdtest.holdoff= [KNL] 854 Set the time in seconds that the clocksource 855 watchdog test waits before commencing its tests. 856 Defaults to zero when built as a module and to 857 10 seconds when built into the kernel. 858 859 cma=nn[MG]@[start[MG][-end[MG]]] 860 [KNL,CMA,EARLY] 861 Sets the size of kernel global memory area for 862 contiguous memory allocations and optionally the 863 placement constraint by the physical address range of 864 memory allocations. A value of 0 disables CMA 865 altogether. For more information, see 866 kernel/dma/contiguous.c 867 868 cma_pernuma=nn[MG] 869 [KNL,CMA,EARLY] 870 Sets the size of kernel per-numa memory area for 871 contiguous memory allocations. A value of 0 disables 872 per-numa CMA altogether. And If this option is not 873 specified, the default value is 0. 874 With per-numa CMA enabled, DMA users on node nid will 875 first try to allocate buffer from the pernuma area 876 which is located in node nid, if the allocation fails, 877 they will fallback to the global default memory area. 878 879 numa_cma=<node>:nn[MG][,<node>:nn[MG]] 880 [KNL,CMA,EARLY] 881 Sets the size of kernel numa memory area for 882 contiguous memory allocations. It will reserve CMA 883 area for the specified node. 884 885 If it is setup together with upper 'cmd_pernuma=' 886 (unlikely), its size setting takes priority for the 887 specified numa nodes. 888 889 With numa CMA enabled, DMA users on node nid will 890 first try to allocate buffer from the numa area 891 which is located in node nid, if the allocation fails, 892 they will fallback to the global default memory area. 893 894 cmo_free_hint= [PPC] Format: { yes | no } 895 Specify whether pages are marked as being inactive 896 when they are freed. This is used in CMO environments 897 to determine OS memory pressure for page stealing by 898 a hypervisor. 899 Default: yes 900 901 coherent_pool=nn[KMG] [ARM,KNL,EARLY] 902 Sets the size of memory pool for coherent, atomic dma 903 allocations, by default set to 256K. 904 905 condev= [HW,S390] console device 906 conmode= 907 908 con3215_drop= [S390,EARLY] 3215 console drop mode. 909 Format: y|n|Y|N|1|0 910 When set to true, drop data on the 3215 console when 911 the console buffer is full. In this case the 912 operator using a 3270 terminal emulator (for example 913 x3270) does not have to enter the clear key for the 914 console output to advance and the kernel to continue. 915 This leads to a much faster boot time when a 3270 916 terminal emulator is active. If no 3270 terminal 917 emulator is used, this parameter has no effect. 918 919 console= [KNL] Output console device and options. 920 921 tty<n> Use the virtual console device <n>. 922 923 ttyS<n>[,options] 924 ttyUSB0[,options] 925 Use the specified serial port. The options are of 926 the form "bbbbpnf", where "bbbb" is the baud rate, 927 "p" is parity ("n", "o", or "e"), "n" is number of 928 bits, and "f" is flow control ("r" for RTS or 929 omit it). Default is "9600n8". 930 931 See Documentation/admin-guide/serial-console.rst for more 932 information. See 933 Documentation/networking/netconsole.rst for an 934 alternative. 935 936 <DEVNAME>:<n>.<n>[,options] 937 Use the specified serial port on the serial core bus. 938 The addressing uses DEVNAME of the physical serial port 939 device, followed by the serial core controller instance, 940 and the serial port instance. The options are the same 941 as documented for the ttyS addressing above. 942 943 The mapping of the serial ports to the tty instances 944 can be viewed with: 945 946 $ ls -d /sys/bus/serial-base/devices/*:*.*/tty/* 947 /sys/bus/serial-base/devices/00:04:0.0/tty/ttyS0 948 949 In the above example, the console can be addressed with 950 console=00:04:0.0. Note that a console addressed this 951 way will only get added when the related device driver 952 is ready. The use of an earlycon parameter in addition to 953 the console may be desired for console output early on. 954 955 uart[8250],io,<addr>[,options] 956 uart[8250],mmio,<addr>[,options] 957 uart[8250],mmio16,<addr>[,options] 958 uart[8250],mmio32,<addr>[,options] 959 uart[8250],0x<addr>[,options] 960 Start an early, polled-mode console on the 8250/16550 961 UART at the specified I/O port or MMIO address, 962 switching to the matching ttyS device later. 963 MMIO inter-register address stride is either 8-bit 964 (mmio), 16-bit (mmio16), or 32-bit (mmio32). 965 If none of [io|mmio|mmio16|mmio32], <addr> is assumed 966 to be equivalent to 'mmio'. 'options' are specified in 967 the same format described for ttyS above; if unspecified, 968 the h/w is not re-initialized. 969 970 hvc<n> Use the hypervisor console device <n>. This is for 971 both Xen and PowerPC hypervisors. 972 973 { null | "" } 974 Use to disable console output, i.e., to have kernel 975 console messages discarded. 976 This must be the only console= parameter used on the 977 kernel command line. 978 979 If the device connected to the port is not a TTY but a braille 980 device, prepend "brl," before the device type, for instance 981 console=brl,ttyS0 982 For now, only VisioBraille is supported. 983 984 console_msg_format= 985 [KNL] Change console messages format 986 default 987 By default we print messages on consoles in 988 "[time stamp] text\n" format (time stamp may not be 989 printed, depending on CONFIG_PRINTK_TIME or 990 `printk_time' param). 991 syslog 992 Switch to syslog format: "<%u>[time stamp] text\n" 993 IOW, each message will have a facility and loglevel 994 prefix. The format is similar to one used by syslog() 995 syscall, or to executing "dmesg -S --raw" or to reading 996 from /proc/kmsg. 997 998 consoleblank= [KNL] The console blank (screen saver) timeout in 999 seconds. A value of 0 disables the blank timer. 1000 Defaults to 0. 1001 1002 coredump_filter= 1003 [KNL] Change the default value for 1004 /proc/<pid>/coredump_filter. 1005 See also Documentation/filesystems/proc.rst. 1006 1007 coresight_cpu_debug.enable 1008 [ARM,ARM64] 1009 Format: <bool> 1010 Enable/disable the CPU sampling based debugging. 1011 0: default value, disable debugging 1012 1: enable debugging at boot time 1013 1014 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver 1015 Format: 1016 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>] 1017 1018 cpuidle.off=1 [CPU_IDLE] 1019 disable the cpuidle sub-system 1020 1021 cpuidle.governor= 1022 [CPU_IDLE] Name of the cpuidle governor to use. 1023 1024 cpufreq.off=1 [CPU_FREQ] 1025 disable the cpufreq sub-system 1026 1027 cpufreq.default_governor= 1028 [CPU_FREQ] Name of the default cpufreq governor or 1029 policy to use. This governor must be registered in the 1030 kernel before the cpufreq driver probes. 1031 1032 cpu_init_udelay=N 1033 [X86,EARLY] Delay for N microsec between assert and de-assert 1034 of APIC INIT to start processors. This delay occurs 1035 on every CPU online, such as boot, and resume from suspend. 1036 Default: 10000 1037 1038 cpuhp.parallel= 1039 [SMP] Enable/disable parallel bringup of secondary CPUs 1040 Format: <bool> 1041 Default is enabled if CONFIG_HOTPLUG_PARALLEL=y. Otherwise 1042 the parameter has no effect. 1043 1044 crash_kexec_post_notifiers 1045 Only jump to kdump kernel after running the panic 1046 notifiers and dumping kmsg. This option increases 1047 the risks of a kdump failure, since some panic 1048 notifiers can make the crashed kernel more unstable. 1049 In configurations where kdump may not be reliable, 1050 running the panic notifiers could allow collecting 1051 more data on dmesg, like stack traces from other CPUS 1052 or extra data dumped by panic_print. Note that some 1053 configurations enable this option unconditionally, 1054 like Hyper-V, PowerPC (fadump) and AMD SEV-SNP. 1055 1056 crashkernel=size[KMG][@offset[KMG]] 1057 [KNL,EARLY] Using kexec, Linux can switch to a 'crash kernel' 1058 upon panic. This parameter reserves the physical 1059 memory region [offset, offset + size] for that kernel 1060 image. If '@offset' is omitted, then a suitable offset 1061 is selected automatically. 1062 [KNL, X86-64, ARM64, RISCV, LoongArch] Select a region 1063 under 4G first, and fall back to reserve region above 1064 4G when '@offset' hasn't been specified. 1065 See Documentation/admin-guide/kdump/kdump.rst for further details. 1066 1067 crashkernel=range1:size1[,range2:size2,...][@offset] 1068 [KNL] Same as above, but depends on the memory 1069 in the running system. The syntax of range is 1070 start-[end] where start and end are both 1071 a memory unit (amount[KMG]). See also 1072 Documentation/admin-guide/kdump/kdump.rst for an example. 1073 1074 crashkernel=size[KMG],high 1075 [KNL, X86-64, ARM64, RISCV, LoongArch] range could be 1076 above 4G. 1077 Allow kernel to allocate physical memory region from top, 1078 so could be above 4G if system have more than 4G ram 1079 installed. Otherwise memory region will be allocated 1080 below 4G, if available. 1081 It will be ignored if crashkernel=X is specified. 1082 crashkernel=size[KMG],low 1083 [KNL, X86-64, ARM64, RISCV, LoongArch] range under 4G. 1084 When crashkernel=X,high is passed, kernel could allocate 1085 physical memory region above 4G, that cause second kernel 1086 crash on system that require some amount of low memory, 1087 e.g. swiotlb requires at least 64M+32K low memory, also 1088 enough extra low memory is needed to make sure DMA buffers 1089 for 32-bit devices won't run out. Kernel would try to allocate 1090 default size of memory below 4G automatically. The default 1091 size is platform dependent. 1092 --> x86: max(swiotlb_size_or_default() + 8MiB, 256MiB) 1093 --> arm64: 128MiB 1094 --> riscv: 128MiB 1095 --> loongarch: 128MiB 1096 This one lets the user specify own low range under 4G 1097 for second kernel instead. 1098 0: to disable low allocation. 1099 It will be ignored when crashkernel=X,high is not used 1100 or memory reserved is below 4G. 1101 crashkernel=size[KMG],cma 1102 [KNL, X86, ARM64, RISCV, PPC] Reserve additional crash 1103 kernel memory from CMA. This reservation is usable by 1104 the first system's userspace memory and kernel movable 1105 allocations (memory balloon, zswap). Pages allocated 1106 from this memory range will not be included in the vmcore 1107 so this should not be used if dumping of userspace memory 1108 is intended and it has to be expected that some movable 1109 kernel pages may be missing from the dump. 1110 1111 A standard crashkernel reservation, as described above, 1112 is still needed to hold the crash kernel and initrd. 1113 1114 This option increases the risk of a kdump failure: DMA 1115 transfers configured by the first kernel may end up 1116 corrupting the second kernel's memory. 1117 1118 This reservation method is intended for systems that 1119 can't afford to sacrifice enough memory for standard 1120 crashkernel reservation and where less reliable and 1121 possibly incomplete kdump is preferable to no kdump at 1122 all. 1123 1124 cryptomgr.notests 1125 [KNL] Disable crypto self-tests 1126 1127 cs89x0_dma= [HW,NET] 1128 Format: <dma> 1129 1130 cs89x0_media= [HW,NET] 1131 Format: { rj45 | aui | bnc } 1132 1133 csdlock_debug= [KNL] Enable or disable debug add-ons of cross-CPU 1134 function call handling. When switched on, 1135 additional debug data is printed to the console 1136 in case a hanging CPU is detected, and that 1137 CPU is pinged again in order to try to resolve 1138 the hang situation. The default value of this 1139 option depends on the CSD_LOCK_WAIT_DEBUG_DEFAULT 1140 Kconfig option. 1141 1142 dasd= [HW,NET] 1143 See header of drivers/s390/block/dasd_devmap.c. 1144 1145 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port 1146 (one device per port) 1147 Format: <port#>,<type> 1148 See also Documentation/input/devices/joystick-parport.rst 1149 1150 debug [KNL,EARLY] Enable kernel debugging (events log level). 1151 1152 debug_boot_weak_hash 1153 [KNL,EARLY] Enable printing [hashed] pointers early in the 1154 boot sequence. If enabled, we use a weak hash instead 1155 of siphash to hash pointers. Use this option if you are 1156 seeing instances of '(___ptrval___)') and need to see a 1157 value (hashed pointer) instead. Cryptographically 1158 insecure, please do not use on production kernels. 1159 1160 debug_locks_verbose= 1161 [KNL] verbose locking self-tests 1162 Format: <int> 1163 Print debugging info while doing the locking API 1164 self-tests. 1165 Bitmask for the various LOCKTYPE_ tests. Defaults to 0 1166 (no extra messages), setting it to -1 (all bits set) 1167 will print _a_lot_ more information - normally only 1168 useful to lockdep developers. 1169 1170 debug_objects [KNL,EARLY] Enable object debugging 1171 1172 debug_guardpage_minorder= 1173 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this 1174 parameter allows control of the order of pages that will 1175 be intentionally kept free (and hence protected) by the 1176 buddy allocator. Bigger value increase the probability 1177 of catching random memory corruption, but reduce the 1178 amount of memory for normal system use. The maximum 1179 possible value is MAX_PAGE_ORDER/2. Setting this 1180 parameter to 1 or 2 should be enough to identify most 1181 random memory corruption problems caused by bugs in 1182 kernel or driver code when a CPU writes to (or reads 1183 from) a random memory location. Note that there exists 1184 a class of memory corruptions problems caused by buggy 1185 H/W or F/W or by drivers badly programming DMA 1186 (basically when memory is written at bus level and the 1187 CPU MMU is bypassed) which are not detectable by 1188 CONFIG_DEBUG_PAGEALLOC, hence this option will not 1189 help tracking down these problems. 1190 1191 debug_pagealloc= 1192 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this parameter 1193 enables the feature at boot time. By default, it is 1194 disabled and the system will work mostly the same as a 1195 kernel built without CONFIG_DEBUG_PAGEALLOC. 1196 Note: to get most of debug_pagealloc error reports, it's 1197 useful to also enable the page_owner functionality. 1198 on: enable the feature 1199 1200 debugfs= [KNL,EARLY] This parameter enables what is exposed to 1201 userspace and debugfs internal clients. 1202 Format: { on, off } 1203 on: All functions are enabled. 1204 off: Filesystem is not registered and clients 1205 get a -EPERM as result when trying to register files 1206 or directories within debugfs. 1207 This is equivalent of the runtime functionality if 1208 debugfs was not enabled in the kernel at all. 1209 Default value is set in build-time with a kernel configuration. 1210 1211 debugpat [X86] Enable PAT debugging 1212 1213 default_hugepagesz= 1214 [HW] The size of the default HugeTLB page. This is 1215 the size represented by the legacy /proc/ hugepages 1216 APIs. In addition, this is the default hugetlb size 1217 used for shmget(), mmap() and mounting hugetlbfs 1218 filesystems. If not specified, defaults to the 1219 architecture's default huge page size. Huge page 1220 sizes are architecture dependent. See also 1221 Documentation/admin-guide/mm/hugetlbpage.rst. 1222 Format: size[KMG] 1223 1224 deferred_probe_timeout= 1225 [KNL] Debugging option to set a timeout in seconds for 1226 deferred probe to give up waiting on dependencies to 1227 probe. Only specific dependencies (subsystems or 1228 drivers) that have opted in will be ignored. A timeout 1229 of 0 will timeout at the end of initcalls; a negative 1230 value is treated as an infinite timeout value. If the 1231 timeout hasn't expired, it'll be restarted by each 1232 successful driver registration. This option will also 1233 dump out devices still on the deferred probe list after 1234 retrying. 1235 1236 delayacct [KNL] Enable per-task delay accounting 1237 1238 dell_smm_hwmon.ignore_dmi= 1239 [HW] Continue probing hardware even if DMI data 1240 indicates that the driver is running on unsupported 1241 hardware. 1242 1243 dell_smm_hwmon.force= 1244 [HW] Activate driver even if SMM BIOS signature does 1245 not match list of supported models and enable otherwise 1246 blacklisted features. 1247 1248 dell_smm_hwmon.power_status= 1249 [HW] Report power status in /proc/i8k 1250 (disabled by default). 1251 1252 dell_smm_hwmon.restricted= 1253 [HW] Allow controlling fans only if SYS_ADMIN 1254 capability is set. 1255 1256 dell_smm_hwmon.fan_mult= 1257 [HW] Factor to multiply fan speed with. 1258 1259 dell_smm_hwmon.fan_max= 1260 [HW] Maximum configurable fan speed. 1261 1262 dfltcc= [HW,S390] 1263 Format: { on | off | def_only | inf_only | always } 1264 on: s390 zlib hardware support for compression on 1265 level 1 and decompression (default) 1266 off: No s390 zlib hardware support 1267 def_only: s390 zlib hardware support for deflate 1268 only (compression on level 1) 1269 inf_only: s390 zlib hardware support for inflate 1270 only (decompression) 1271 always: Same as 'on' but ignores the selected compression 1272 level always using hardware support (used for debugging) 1273 1274 dhash_entries= [KNL] 1275 Set number of hash buckets for dentry cache. 1276 1277 disable_1tb_segments [PPC,EARLY] 1278 Disables the use of 1TB hash page table segments. This 1279 causes the kernel to fall back to 256MB segments which 1280 can be useful when debugging issues that require an SLB 1281 miss to occur. 1282 1283 disable= [IPV6] 1284 See Documentation/networking/ipv6.rst. 1285 1286 disable_radix [PPC,EARLY] 1287 Disable RADIX MMU mode on POWER9 1288 1289 disable_tlbie [PPC] 1290 Disable TLBIE instruction. Currently does not work 1291 with KVM, with HASH MMU, or with coherent accelerators. 1292 1293 disable_ddw [PPC/PSERIES,EARLY] 1294 Disable Dynamic DMA Window support. Use this 1295 to workaround buggy firmware. 1296 1297 disable_ipv6= [IPV6] 1298 See Documentation/networking/ipv6.rst. 1299 1300 disable_mtrr_cleanup [X86,EARLY] 1301 The kernel tries to adjust MTRR layout from continuous 1302 to discrete, to make X server driver able to add WB 1303 entry later. This parameter disables that. 1304 1305 disable_mtrr_trim [X86, Intel and AMD only,EARLY] 1306 By default the kernel will trim any uncacheable 1307 memory out of your available memory pool based on 1308 MTRR settings. This parameter disables that behavior, 1309 possibly causing your machine to run very slowly. 1310 1311 disable_timer_pin_1 [X86,EARLY] 1312 Disable PIN 1 of APIC timer 1313 Can be useful to work around chipset bugs. 1314 1315 dis_ucode_ldr [X86] Disable the microcode loader. 1316 1317 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support, 1318 this option disables the debugging code at boot. 1319 1320 dma_debug_entries=<number> 1321 This option allows to tune the number of preallocated 1322 entries for DMA-API debugging code. One entry is 1323 required per DMA-API allocation. Use this if the 1324 DMA-API debugging code disables itself because the 1325 architectural default is too low. 1326 1327 dma_debug_driver=<driver_name> 1328 With this option the DMA-API debugging driver 1329 filter feature can be enabled at boot time. Just 1330 pass the driver to filter for as the parameter. 1331 The filter can be disabled or changed to another 1332 driver later using sysfs. 1333 1334 reg_file_data_sampling= 1335 [X86] Controls mitigation for Register File Data 1336 Sampling (RFDS) vulnerability. RFDS is a CPU 1337 vulnerability which may allow userspace to infer 1338 kernel data values previously stored in floating point 1339 registers, vector registers, or integer registers. 1340 RFDS only affects Intel Atom processors. 1341 1342 on: Turns ON the mitigation. 1343 off: Turns OFF the mitigation. 1344 1345 This parameter overrides the compile time default set 1346 by CONFIG_MITIGATION_RFDS. Mitigation cannot be 1347 disabled when other VERW based mitigations (like MDS) 1348 are enabled. In order to disable RFDS mitigation all 1349 VERW based mitigations need to be disabled. 1350 1351 For details see: 1352 Documentation/admin-guide/hw-vuln/reg-file-data-sampling.rst 1353 1354 dm_verity.keyring_unsealed= 1355 [KNL] When set to 1, leave the dm-verity keyring 1356 unsealed after initialization so userspace can 1357 provision keys. Once the keyring is restricted 1358 it becomes active and is searched during signature 1359 verification. 1360 1361 driver_async_probe= [KNL] 1362 List of driver names to be probed asynchronously. * 1363 matches with all driver names. If * is specified, the 1364 rest of the listed driver names are those that will NOT 1365 match the *. 1366 Format: <driver_name1>,<driver_name2>... 1367 1368 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>] 1369 Broken monitors, graphic adapters, KVMs and EDIDless 1370 panels may send no or incorrect EDID data sets. 1371 This parameter allows to specify an EDID data sets 1372 in the /lib/firmware directory that are used instead. 1373 An EDID data set will only be used for a particular 1374 connector, if its name and a colon are prepended to 1375 the EDID name. Each connector may use a unique EDID 1376 data set by separating the files with a comma. An EDID 1377 data set with no connector name will be used for 1378 any connectors not explicitly specified. 1379 1380 dscc4.setup= [NET] 1381 1382 dt_cpu_ftrs= [PPC,EARLY] 1383 Format: {"off" | "known"} 1384 Control how the dt_cpu_ftrs device-tree binding is 1385 used for CPU feature discovery and setup (if it 1386 exists). 1387 off: Do not use it, fall back to legacy cpu table. 1388 known: Do not pass through unknown features to guests 1389 or userspace, only those that the kernel is aware of. 1390 1391 dump_apple_properties [X86] 1392 Dump name and content of EFI device properties on 1393 x86 Macs. Useful for driver authors to determine 1394 what data is available or for reverse-engineering. 1395 1396 dyndbg[="val"] [KNL,DYNAMIC_DEBUG] 1397 <module>.dyndbg[="val"] 1398 Enable debug messages at boot time. See 1399 Documentation/admin-guide/dynamic-debug-howto.rst 1400 for details. 1401 1402 early_ioremap_debug [KNL,EARLY] 1403 Enable debug messages in early_ioremap support. This 1404 is useful for tracking down temporary early mappings 1405 which are not unmapped. 1406 1407 earlycon= [KNL,EARLY] Output early console device and options. 1408 1409 When used with no options, the early console is 1410 determined by stdout-path property in device tree's 1411 chosen node or the ACPI SPCR table if supported by 1412 the platform. 1413 1414 cdns,<addr>[,options] 1415 Start an early, polled-mode console on a Cadence 1416 (xuartps) serial port at the specified address. Only 1417 supported option is baud rate. If baud rate is not 1418 specified, the serial port must already be setup and 1419 configured. 1420 1421 uart[8250],io,<addr>[,options[,uartclk]] 1422 uart[8250],mmio,<addr>[,options[,uartclk]] 1423 uart[8250],mmio32,<addr>[,options[,uartclk]] 1424 uart[8250],mmio32be,<addr>[,options[,uartclk]] 1425 uart[8250],0x<addr>[,options] 1426 Start an early, polled-mode console on the 8250/16550 1427 UART at the specified I/O port or MMIO address. 1428 MMIO inter-register address stride is either 8-bit 1429 (mmio) or 32-bit (mmio32 or mmio32be). 1430 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed 1431 to be equivalent to 'mmio'. 'options' are specified 1432 in the same format described for "console=ttyS<n>"; if 1433 unspecified, the h/w is not initialized. 'uartclk' is 1434 the uart clock frequency; if unspecified, it is set 1435 to 'BASE_BAUD' * 16. 1436 1437 pl011,<addr> 1438 pl011,mmio32,<addr> 1439 Start an early, polled-mode console on a pl011 serial 1440 port at the specified address. The pl011 serial port 1441 must already be setup and configured. Options are not 1442 yet supported. If 'mmio32' is specified, then only 1443 the driver will use only 32-bit accessors to read/write 1444 the device registers. 1445 1446 liteuart,<addr> 1447 Start an early console on a litex serial port at the 1448 specified address. The serial port must already be 1449 setup and configured. Options are not yet supported. 1450 1451 meson,<addr> 1452 Start an early, polled-mode console on a meson serial 1453 port at the specified address. The serial port must 1454 already be setup and configured. Options are not yet 1455 supported. 1456 1457 msm_serial,<addr> 1458 Start an early, polled-mode console on an msm serial 1459 port at the specified address. The serial port 1460 must already be setup and configured. Options are not 1461 yet supported. 1462 1463 msm_serial_dm,<addr> 1464 Start an early, polled-mode console on an msm serial 1465 dm port at the specified address. The serial port 1466 must already be setup and configured. Options are not 1467 yet supported. 1468 1469 owl,<addr> 1470 Start an early, polled-mode console on a serial port 1471 of an Actions Semi SoC, such as S500 or S900, at the 1472 specified address. The serial port must already be 1473 setup and configured. Options are not yet supported. 1474 1475 rda,<addr> 1476 Start an early, polled-mode console on a serial port 1477 of an RDA Micro SoC, such as RDA8810PL, at the 1478 specified address. The serial port must already be 1479 setup and configured. Options are not yet supported. 1480 1481 sbi 1482 Use RISC-V SBI (Supervisor Binary Interface) for early 1483 console. 1484 1485 smh Use ARM semihosting calls for early console. 1486 1487 s3c2410,<addr> 1488 s3c2412,<addr> 1489 s3c2440,<addr> 1490 s3c6400,<addr> 1491 s5pv210,<addr> 1492 exynos4210,<addr> 1493 Use early console provided by serial driver available 1494 on Samsung SoCs, requires selecting proper type and 1495 a correct base address of the selected UART port. The 1496 serial port must already be setup and configured. 1497 Options are not yet supported. 1498 1499 lantiq,<addr> 1500 Start an early, polled-mode console on a lantiq serial 1501 (lqasc) port at the specified address. The serial port 1502 must already be setup and configured. Options are not 1503 yet supported. 1504 1505 lpuart,<addr> 1506 lpuart32,<addr> 1507 Use early console provided by Freescale LP UART driver 1508 found on Freescale Vybrid and QorIQ LS1021A processors. 1509 A valid base address must be provided, and the serial 1510 port must already be setup and configured. 1511 1512 ec_imx21,<addr> 1513 ec_imx6q,<addr> 1514 Start an early, polled-mode, output-only console on the 1515 Freescale i.MX UART at the specified address. The UART 1516 must already be setup and configured. 1517 1518 ar3700_uart,<addr> 1519 Start an early, polled-mode console on the 1520 Armada 3700 serial port at the specified 1521 address. The serial port must already be setup 1522 and configured. Options are not yet supported. 1523 1524 qcom_geni,<addr> 1525 Start an early, polled-mode console on a Qualcomm 1526 Generic Interface (GENI) based serial port at the 1527 specified address. The serial port must already be 1528 setup and configured. Options are not yet supported. 1529 1530 efifb,[options] 1531 Start an early, unaccelerated console on the EFI 1532 memory mapped framebuffer (if available). On cache 1533 coherent non-x86 systems that use system memory for 1534 the framebuffer, pass the 'ram' option so that it is 1535 mapped with the correct attributes. 1536 1537 linflex,<addr> 1538 Use early console provided by Freescale LINFlexD UART 1539 serial driver for NXP S32V234 SoCs. A valid base 1540 address must be provided, and the serial port must 1541 already be setup and configured. 1542 1543 earlyprintk= [X86,SH,ARM,M68k,S390,UM,EARLY] 1544 earlyprintk=vga 1545 earlyprintk=sclp 1546 earlyprintk=xen 1547 earlyprintk=serial[,ttySn[,baudrate]] 1548 earlyprintk=serial[,0x...[,baudrate]] 1549 earlyprintk=ttySn[,baudrate] 1550 earlyprintk=dbgp[debugController#] 1551 earlyprintk=mmio32,membase[,{nocfg|baudrate}] 1552 earlyprintk=pciserial[,force],bus:device.function[,{nocfg|baudrate}] 1553 earlyprintk=xdbc[xhciController#] 1554 earlyprintk=bios 1555 1556 earlyprintk is useful when the kernel crashes before 1557 the normal console is initialized. It is not enabled by 1558 default because it has some cosmetic problems. 1559 1560 Use "nocfg" to skip UART configuration, assume 1561 BIOS/firmware has configured UART correctly. 1562 1563 Append ",keep" to not disable it when the real console 1564 takes over. 1565 1566 Only one of vga, serial, or usb debug port can 1567 be used at a time. 1568 1569 Currently only ttyS0 and ttyS1 may be specified by 1570 name. Other I/O ports may be explicitly specified 1571 on some architectures (x86 and arm at least) by 1572 replacing ttySn with an I/O port address, like this: 1573 earlyprintk=serial,0x1008,115200 1574 You can find the port for a given device in 1575 /proc/tty/driver/serial: 1576 2: uart:ST16650V2 port:00001008 irq:18 ... 1577 1578 Interaction with the standard serial driver is not 1579 very good. 1580 1581 The VGA output is eventually overwritten by 1582 the real console. 1583 1584 The xen option can only be used in Xen domains. 1585 1586 The sclp output can only be used on s390. 1587 1588 The bios output can only be used on SuperH. 1589 1590 The optional "force" to "pciserial" enables use of a 1591 PCI device even when its classcode is not of the 1592 UART class. 1593 1594 edd= [EDD] 1595 Format: {"off" | "on" | "skip[mbr]"} 1596 1597 efi= [EFI,EARLY] 1598 Format: { "debug", "disable_early_pci_dma", 1599 "nochunk", "noruntime", "nosoftreserve", 1600 "novamap", "no_disable_early_pci_dma" } 1601 debug: enable misc debug output. 1602 disable_early_pci_dma: disable the busmaster bit on all 1603 PCI bridges while in the EFI boot stub. 1604 nochunk: disable reading files in "chunks" in the EFI 1605 boot stub, as chunking can cause problems with some 1606 firmware implementations. 1607 noruntime : disable EFI runtime services support 1608 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose) 1609 attribute may cause the kernel to reserve the 1610 memory range for a memory mapping driver to 1611 claim. Specify efi=nosoftreserve to disable this 1612 reservation and treat the memory by its base type 1613 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM"). 1614 novamap: do not call SetVirtualAddressMap(). 1615 no_disable_early_pci_dma: Leave the busmaster bit set 1616 on all PCI bridges while in the EFI boot stub 1617 1618 efi_no_storage_paranoia [EFI,X86,EARLY] 1619 The kernel reserves 5KB of EFI variable storage for 1620 safety, because some UEFI implementation may fail to 1621 boot if there's insufficient space in the EFI variable 1622 storage. 1623 1624 Using this parameter, you can use the 5KB reservation 1625 in the EFI variable storage. 1626 1627 However, Use this parameter only if you are really 1628 sure that your UEFI does sane gc and fulfills the spec 1629 otherwise your board may brick. 1630 1631 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT 1632 that is to be dynamically loaded by Linux. If there are 1633 multiple variables with the same name but with different 1634 vendor GUIDs, all of them will be loaded. See 1635 Documentation/admin-guide/acpi/ssdt-overlays.rst for details. 1636 1637 1638 eisa_irq_edge= [PARISC,HW] 1639 See header of drivers/parisc/eisa.c. 1640 1641 ekgdboc= [X86,KGDB,EARLY] Allow early kernel console debugging 1642 Format: ekgdboc=kbd 1643 1644 This is designed to be used in conjunction with 1645 the boot argument: earlyprintk=vga 1646 1647 This parameter works in place of the kgdboc parameter 1648 but can only be used if the backing tty is available 1649 very early in the boot process. For early debugging 1650 via a serial port see kgdboc_earlycon instead. 1651 1652 elfcorehdr=[size[KMG]@]offset[KMG] [PPC,SH,X86,S390,EARLY] 1653 Specifies physical address of start of kernel core 1654 image elf header and optionally the size. Generally 1655 kexec loader will pass this option to capture kernel. 1656 See Documentation/admin-guide/kdump/kdump.rst for details. 1657 1658 enable_mtrr_cleanup [X86,EARLY] 1659 The kernel tries to adjust MTRR layout from continuous 1660 to discrete, to make X server driver able to add WB 1661 entry later. This parameter enables that. 1662 1663 enforcing= [SELINUX] Set initial enforcing status. 1664 Format: {"0" | "1"} 1665 See security/selinux/Kconfig help text. 1666 0 -- permissive (log only, no denials). 1667 1 -- enforcing (deny and log). 1668 Default value is 0. 1669 Value can be changed at runtime via 1670 /sys/fs/selinux/enforce. 1671 1672 erst_disable [ACPI] 1673 Disable Error Record Serialization Table (ERST) 1674 support. 1675 1676 ether= [HW,NET] Ethernet cards parameters 1677 This option is obsoleted by the "netdev=" option, which 1678 has equivalent usage. See its documentation for details. 1679 1680 evm= [EVM] 1681 Format: { "fix" } 1682 Permit 'security.evm' to be updated regardless of 1683 current integrity status. 1684 1685 early_page_ext [KNL,EARLY] Enforces page_ext initialization to earlier 1686 stages so cover more early boot allocations. 1687 Please note that as side effect some optimizations 1688 might be disabled to achieve that (e.g. parallelized 1689 memory initialization is disabled) so the boot process 1690 might take longer, especially on systems with a lot of 1691 memory. Available with CONFIG_PAGE_EXTENSION=y. 1692 1693 failslab= 1694 fail_usercopy= 1695 fail_page_alloc= 1696 fail_skb_realloc= 1697 fail_make_request=[KNL] 1698 General fault injection mechanism. 1699 Format: <interval>,<probability>,<space>,<times> 1700 See also Documentation/fault-injection/. 1701 1702 fb_tunnels= [NET] 1703 Format: { initns | none } 1704 See Documentation/admin-guide/sysctl/net.rst for 1705 fb_tunnels_only_for_init_ns 1706 1707 floppy= [HW] 1708 See Documentation/admin-guide/blockdev/floppy.rst. 1709 1710 forcepae [X86-32] 1711 Forcefully enable Physical Address Extension (PAE). 1712 Many Pentium M systems disable PAE but may have a 1713 functionally usable PAE implementation. 1714 Warning: use of this parameter will taint the kernel 1715 and may cause unknown problems. 1716 1717 fred= [X86-64] 1718 Enable/disable Flexible Return and Event Delivery. 1719 Format: { on | off } 1720 on: enable FRED when it's present, the default setting. 1721 off: disable FRED. 1722 1723 ftrace=[tracer] 1724 [FTRACE] will set and start the specified tracer 1725 as early as possible in order to facilitate early 1726 boot debugging. 1727 1728 ftrace_boot_snapshot 1729 [FTRACE] On boot up, a snapshot will be taken of the 1730 ftrace ring buffer that can be read at: 1731 /sys/kernel/tracing/snapshot. 1732 This is useful if you need tracing information from kernel 1733 boot up that is likely to be overridden by user space 1734 start up functionality. 1735 1736 Optionally, the snapshot can also be defined for a tracing 1737 instance that was created by the trace_instance= command 1738 line parameter. 1739 1740 trace_instance=foo,sched_switch ftrace_boot_snapshot=foo 1741 1742 The above will cause the "foo" tracing instance to trigger 1743 a snapshot at the end of boot up. 1744 1745 ftrace_dump_on_oops[=2(orig_cpu) | =<instance>][,<instance> | 1746 ,<instance>=2(orig_cpu)] 1747 [FTRACE] will dump the trace buffers on oops. 1748 If no parameter is passed, ftrace will dump global 1749 buffers of all CPUs, if you pass 2 or orig_cpu, it 1750 will dump only the buffer of the CPU that triggered 1751 the oops, or the specific instance will be dumped if 1752 its name is passed. Multiple instance dump is also 1753 supported, and instances are separated by commas. Each 1754 instance supports only dump on CPU that triggered the 1755 oops by passing 2 or orig_cpu to it. 1756 1757 ftrace_dump_on_oops=foo=orig_cpu 1758 1759 The above will dump only the buffer of "foo" instance 1760 on CPU that triggered the oops. 1761 1762 ftrace_dump_on_oops,foo,bar=orig_cpu 1763 1764 The above will dump global buffer on all CPUs, the 1765 buffer of "foo" instance on all CPUs and the buffer 1766 of "bar" instance on CPU that triggered the oops. 1767 1768 ftrace_filter=[function-list] 1769 [FTRACE] Limit the functions traced by the function 1770 tracer at boot up. function-list is a comma-separated 1771 list of functions. This list can be changed at run 1772 time by the set_ftrace_filter file in the debugfs 1773 tracing directory. 1774 1775 ftrace_notrace=[function-list] 1776 [FTRACE] Do not trace the functions specified in 1777 function-list. This list can be changed at run time 1778 by the set_ftrace_notrace file in the debugfs 1779 tracing directory. 1780 1781 ftrace_graph_filter=[function-list] 1782 [FTRACE] Limit the top level callers functions traced 1783 by the function graph tracer at boot up. 1784 function-list is a comma-separated list of functions 1785 that can be changed at run time by the 1786 set_graph_function file in the debugfs tracing directory. 1787 1788 ftrace_graph_notrace=[function-list] 1789 [FTRACE] Do not trace from the functions specified in 1790 function-list. This list is a comma-separated list of 1791 functions that can be changed at run time by the 1792 set_graph_notrace file in the debugfs tracing directory. 1793 1794 ftrace_graph_max_depth=<uint> 1795 [FTRACE] Used with the function graph tracer. This is 1796 the max depth it will trace into a function. This value 1797 can be changed at run time by the max_graph_depth file 1798 in the tracefs tracing directory. default: 0 (no limit) 1799 1800 fw_devlink= [KNL,EARLY] Create device links between consumer and supplier 1801 devices by scanning the firmware to infer the 1802 consumer/supplier relationships. This feature is 1803 especially useful when drivers are loaded as modules as 1804 it ensures proper ordering of tasks like device probing 1805 (suppliers first, then consumers), supplier boot state 1806 clean up (only after all consumers have probed), 1807 suspend/resume & runtime PM (consumers first, then 1808 suppliers). 1809 Format: { off | permissive | on | rpm } 1810 off -- Don't create device links from firmware info. 1811 permissive -- Create device links from firmware info 1812 but use it only for ordering boot state clean 1813 up (sync_state() calls). 1814 on -- Create device links from firmware info and use it 1815 to enforce probe and suspend/resume ordering. 1816 rpm -- Like "on", but also use to order runtime PM. 1817 1818 fw_devlink.strict=<bool> 1819 [KNL,EARLY] Treat all inferred dependencies as mandatory 1820 dependencies. This only applies for fw_devlink=on|rpm. 1821 Format: <bool> 1822 1823 fw_devlink.sync_state = 1824 [KNL,EARLY] When all devices that could probe have finished 1825 probing, this parameter controls what to do with 1826 devices that haven't yet received their sync_state() 1827 calls. 1828 Format: { strict | timeout } 1829 strict -- Default. Continue waiting on consumers to 1830 probe successfully. 1831 timeout -- Give up waiting on consumers and call 1832 sync_state() on any devices that haven't yet 1833 received their sync_state() calls after 1834 deferred_probe_timeout has expired or by 1835 late_initcall() if !CONFIG_MODULES. 1836 1837 gamecon.map[2|3]= 1838 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad 1839 support via parallel port (up to 5 devices per port) 1840 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5> 1841 See also Documentation/input/devices/joystick-parport.rst 1842 1843 gamma= [HW,DRM] 1844 1845 gart_fix_e820= [X86-64,EARLY] disable the fix e820 for K8 GART 1846 Format: off | on 1847 default: on 1848 1849 gather_data_sampling= 1850 [X86,INTEL,EARLY] Control the Gather Data Sampling (GDS) 1851 mitigation. 1852 1853 Gather Data Sampling is a hardware vulnerability which 1854 allows unprivileged speculative access to data which was 1855 previously stored in vector registers. 1856 1857 This issue is mitigated by default in updated microcode. 1858 The mitigation may have a performance impact but can be 1859 disabled. On systems without the microcode mitigation 1860 disabling AVX serves as a mitigation. 1861 1862 force: Disable AVX to mitigate systems without 1863 microcode mitigation. No effect if the microcode 1864 mitigation is present. Known to cause crashes in 1865 userspace with buggy AVX enumeration. 1866 1867 off: Disable GDS mitigation. 1868 1869 gbpages [X86] Use GB pages for kernel direct mappings. 1870 1871 gcov_persist= [GCOV] When non-zero (default), profiling data for 1872 kernel modules is saved and remains accessible via 1873 debugfs, even when the module is unloaded/reloaded. 1874 When zero, profiling data is discarded and associated 1875 debugfs files are removed at module unload time. 1876 1877 goldfish [X86] Enable the goldfish android emulator platform. 1878 Don't use this when you are not running on the 1879 android emulator 1880 1881 gpio-mockup.gpio_mockup_ranges 1882 [HW] Sets the ranges of gpiochip of for this device. 1883 Format: <start1>,<end1>,<start2>,<end2>... 1884 gpio-mockup.gpio_mockup_named_lines 1885 [HW] Let the driver know GPIO lines should be named. 1886 1887 gpt [EFI] Forces disk with valid GPT signature but 1888 invalid Protective MBR to be treated as GPT. If the 1889 primary GPT is corrupted, it enables the backup/alternate 1890 GPT to be used instead. 1891 1892 grcan.enable0= [HW] Configuration of physical interface 0. Determines 1893 the "Enable 0" bit of the configuration register. 1894 Format: 0 | 1 1895 Default: 0 1896 grcan.enable1= [HW] Configuration of physical interface 1. Determines 1897 the "Enable 0" bit of the configuration register. 1898 Format: 0 | 1 1899 Default: 0 1900 grcan.select= [HW] Select which physical interface to use. 1901 Format: 0 | 1 1902 Default: 0 1903 grcan.txsize= [HW] Sets the size of the tx buffer. 1904 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0. 1905 Default: 1024 1906 grcan.rxsize= [HW] Sets the size of the rx buffer. 1907 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0. 1908 Default: 1024 1909 1910 hardened_usercopy= 1911 [KNL] Under CONFIG_HARDENED_USERCOPY, whether 1912 hardening is enabled for this boot. Hardened 1913 usercopy checking is used to protect the kernel 1914 from reading or writing beyond known memory 1915 allocation boundaries as a proactive defense 1916 against bounds-checking flaws in the kernel's 1917 copy_to_user()/copy_from_user() interface. 1918 The default is determined by 1919 CONFIG_HARDENED_USERCOPY_DEFAULT_ON. 1920 on Perform hardened usercopy checks. 1921 off Disable hardened usercopy checks. 1922 1923 hardlockup_all_cpu_backtrace= 1924 [KNL] Should the hard-lockup detector generate 1925 backtraces on all cpus. 1926 Format: 0 | 1 1927 1928 hash_pointers= 1929 [KNL,EARLY] 1930 By default, when pointers are printed to the console 1931 or buffers via the %p format string, that pointer is 1932 "hashed", i.e. obscured by hashing the pointer value. 1933 This is a security feature that hides actual kernel 1934 addresses from unprivileged users, but it also makes 1935 debugging the kernel more difficult since unequal 1936 pointers can no longer be compared. The choices are: 1937 Format: { auto | always | never } 1938 Default: auto 1939 1940 auto - Hash pointers unless slab_debug is enabled. 1941 always - Always hash pointers (even if slab_debug is 1942 enabled). 1943 never - Never hash pointers. This option should only 1944 be specified when debugging the kernel. Do 1945 not use on production kernels. The boot 1946 param "no_hash_pointers" is an alias for 1947 this mode. 1948 1949 For controlling hashing dynamically at runtime, 1950 use the "kernel.kptr_restrict" sysctl instead. 1951 1952 hashdist= [KNL,NUMA] Large hashes allocated during boot 1953 are distributed across NUMA nodes. Defaults on 1954 for 64-bit NUMA, off otherwise. 1955 Format: 0 | 1 (for off | on) 1956 1957 hd= [EIDE] (E)IDE hard drive subsystem geometry 1958 Format: <cyl>,<head>,<sect> 1959 1960 hest_disable [ACPI] 1961 Disable Hardware Error Source Table (HEST) support; 1962 corresponding firmware-first mode error processing 1963 logic will be disabled. 1964 1965 hibernate= [HIBERNATION] 1966 noresume Don't check if there's a hibernation image 1967 present during boot. 1968 nocompress Don't compress/decompress hibernation images. 1969 no Disable hibernation and resume. 1970 protect_image Turn on image protection during restoration 1971 (that will set all pages holding image data 1972 during restoration read-only). 1973 1974 hibernate.compressor= [HIBERNATION] Compression algorithm to be 1975 used with hibernation. 1976 Format: { lzo | lz4 } 1977 Default: lzo 1978 1979 lzo: Select LZO compression algorithm to 1980 compress/decompress hibernation image. 1981 1982 lz4: Select LZ4 compression algorithm to 1983 compress/decompress hibernation image. 1984 1985 hibernate.pm_test_delay= 1986 [HIBERNATION] 1987 Sets the number of seconds to remain in a hibernation test 1988 mode before resuming the system (see 1989 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG 1990 is set. Default value is 5. 1991 1992 hibernate_compression_threads= 1993 [HIBERNATION] 1994 Set the number of threads used for compressing or decompressing 1995 hibernation images. 1996 1997 Format: <integer> 1998 Default: 3 1999 Minimum: 1 2000 Example: hibernate_compression_threads=4 2001 2002 highmem=nn[KMG] [KNL,BOOT,EARLY] forces the highmem zone to have an exact 2003 size of <nn>. This works even on boxes that have no 2004 highmem otherwise. This also works to reduce highmem 2005 size on bigger boxes. 2006 2007 highres= [KNL] Enable/disable high resolution timer mode. 2008 Valid parameters: "on", "off" 2009 Default: "on" 2010 2011 hlt [BUGS=ARM,SH] 2012 2013 hostname= [KNL,EARLY] Set the hostname (aka UTS nodename). 2014 Format: <string> 2015 This allows setting the system's hostname during early 2016 startup. This sets the name returned by gethostname. 2017 Using this parameter to set the hostname makes it 2018 possible to ensure the hostname is correctly set before 2019 any userspace processes run, avoiding the possibility 2020 that a process may call gethostname before the hostname 2021 has been explicitly set, resulting in the calling 2022 process getting an incorrect result. The string must 2023 not exceed the maximum allowed hostname length (usually 2024 64 characters) and will be truncated otherwise. 2025 2026 hpet= [X86-32,HPET] option to control HPET usage 2027 Format: { enable (default) | disable | force | 2028 verbose } 2029 disable: disable HPET and use PIT instead 2030 force: allow force enabled of undocumented chips (ICH4, 2031 VIA, nVidia) 2032 verbose: show contents of HPET registers during setup 2033 2034 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET 2035 registers. Default set by CONFIG_HPET_MMAP_DEFAULT. 2036 2037 hugepages= [HW,EARLY] Number of HugeTLB pages to allocate at boot. 2038 If this follows hugepagesz (below), it specifies 2039 the number of pages of hugepagesz to be allocated. 2040 If this is the first HugeTLB parameter on the command 2041 line, it specifies the number of pages to allocate for 2042 the default huge page size. If using node format, the 2043 number of pages to allocate per-node can be specified. 2044 See also Documentation/admin-guide/mm/hugetlbpage.rst. 2045 Format: <integer> or (node format) 2046 <node>:<integer>[,<node>:<integer>] 2047 2048 hugepagesz= 2049 [HW,EARLY] The size of the HugeTLB pages. This is 2050 used in conjunction with hugepages (above) to 2051 allocate huge pages of a specific size at boot. The 2052 pair hugepagesz=X hugepages=Y can be specified once 2053 for each supported huge page size. Huge page sizes 2054 are architecture dependent. See also 2055 Documentation/admin-guide/mm/hugetlbpage.rst. 2056 Format: size[KMG] 2057 2058 hugepage_alloc_threads= 2059 [HW] The number of threads that should be used to 2060 allocate hugepages during boot. This option can be 2061 used to improve system bootup time when allocating 2062 a large amount of huge pages. 2063 The default value is 25% of the available hardware threads. 2064 2065 Note that this parameter only applies to non-gigantic huge pages. 2066 2067 hugetlb_cma= [HW,CMA,EARLY] The size of a CMA area used for allocation 2068 of gigantic hugepages. Or using node format, the size 2069 of a CMA area per node can be specified. 2070 The size can be an absolute value (e.g., 2G) or a 2071 percentage of the total memory or node memory (e.g., 20%). 2072 Percentage-derived sizes are rounded down to a multiple of 2073 the architecture's gigantic hugepage size and may become 2074 zero. 2075 Format: nn[KMGTPE] or nn% or (node format) 2076 <node>:nn[KMGTPE][,<node>:nn[KMGTPE]] or 2077 <node>:nn%[,<node>:nn%] 2078 2079 The size must be a multiple of the gigantic page size. 2080 When using node format, this applies to each per-node size. 2081 Missaligned values are dropped with a warning. 2082 2083 Reserve a CMA area of given size and allocate gigantic 2084 hugepages using the CMA allocator. If enabled, the 2085 boot-time allocation of gigantic hugepages is skipped. 2086 2087 hugetlb_cma_only= 2088 [HW,CMA,EARLY] When allocating new HugeTLB pages, only 2089 try to allocate from the CMA areas. 2090 2091 This option does nothing if hugetlb_cma= is not also 2092 specified. 2093 2094 hugetlb_free_vmemmap= 2095 [KNL] Requires CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP 2096 enabled. 2097 Control if HugeTLB Vmemmap Optimization (HVO) is enabled. 2098 Allows heavy hugetlb users to free up some more 2099 memory (7 * PAGE_SIZE for each 2MB hugetlb page). 2100 Format: { on | off (default) } 2101 2102 on: enable HVO 2103 off: disable HVO 2104 2105 Built with CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP_DEFAULT_ON=y, 2106 the default is on. 2107 2108 Note that the vmemmap pages may be allocated from the added 2109 memory block itself when memory_hotplug.memmap_on_memory is 2110 enabled, those vmemmap pages cannot be optimized even if this 2111 feature is enabled. Other vmemmap pages not allocated from 2112 the added memory block itself do not be affected. 2113 2114 hung_task_panic= 2115 [KNL] Number of hung tasks to trigger kernel panic. 2116 Format: <int> 2117 2118 When set to a non-zero value, a kernel panic will be triggered if 2119 the number of detected hung tasks reaches this value. 2120 2121 0: don't panic 2122 1: panic immediately on first hung task 2123 N: panic after N hung tasks are detected in a single scan 2124 2125 The default value is controlled by the 2126 CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time option. The value 2127 selected by this boot parameter can be changed later by the 2128 kernel.hung_task_panic sysctl. 2129 2130 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC) 2131 terminal devices. Valid values: 0..8 2132 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs. 2133 If specified, z/VM IUCV HVC accepts connections 2134 from listed z/VM user IDs only. 2135 2136 hv_nopvspin [X86,HYPER_V,EARLY] 2137 Disables the paravirt spinlock optimizations 2138 which allow the hypervisor to 'idle' the guest 2139 on lock contention. 2140 2141 hw_protection= [HW] 2142 Format: reboot | shutdown 2143 2144 Hardware protection action taken on critical events like 2145 overtemperature or imminent voltage loss. 2146 2147 i2c_bus= [HW] Override the default board specific I2C bus speed 2148 or register an additional I2C bus that is not 2149 registered from board initialization code. 2150 Format: 2151 <bus_id>,<clkrate> 2152 2153 i2c_touchscreen_props= [HW,ACPI,X86] 2154 Set device-properties for ACPI-enumerated I2C-attached 2155 touchscreen, to e.g. fix coordinates of upside-down 2156 mounted touchscreens. If you need this option please 2157 submit a drivers/platform/x86/touchscreen_dmi.c patch 2158 adding a DMI quirk for this. 2159 2160 Format: 2161 <ACPI_HW_ID>:<prop_name>=<val>[:prop_name=val][:...] 2162 Where <val> is one of: 2163 Omit "=<val>" entirely Set a boolean device-property 2164 Unsigned number Set a u32 device-property 2165 Anything else Set a string device-property 2166 2167 Examples (split over multiple lines): 2168 i2c_touchscreen_props=GDIX1001:touchscreen-inverted-x: 2169 touchscreen-inverted-y 2170 2171 i2c_touchscreen_props=MSSL1680:touchscreen-size-x=1920: 2172 touchscreen-size-y=1080:touchscreen-inverted-y: 2173 firmware-name=gsl1680-vendor-model.fw:silead,home-button 2174 2175 i8042.debug [HW] Toggle i8042 debug mode 2176 i8042.unmask_kbd_data 2177 [HW] Enable printing of interrupt data from the KBD port 2178 (disabled by default, and as a pre-condition 2179 requires that i8042.debug=1 be enabled) 2180 i8042.direct [HW] Put keyboard port into non-translated mode 2181 i8042.dumbkbd [HW] Pretend that controller can only read data from 2182 keyboard and cannot control its state 2183 (Don't attempt to blink the leds) 2184 i8042.noaux [HW] Don't check for auxiliary (== mouse) port 2185 i8042.nokbd [HW] Don't check/create keyboard port 2186 i8042.noloop [HW] Disable the AUX Loopback command while probing 2187 for the AUX port 2188 i8042.nomux [HW] Don't check presence of an active multiplexing 2189 controller 2190 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX 2191 controllers 2192 i8042.notimeout [HW] Ignore timeout condition signalled by controller 2193 i8042.reset [HW] Reset the controller during init, cleanup and 2194 suspend-to-ram transitions, only during s2r 2195 transitions, or never reset 2196 Format: { 1 | Y | y | 0 | N | n } 2197 1, Y, y: always reset controller 2198 0, N, n: don't ever reset controller 2199 Default: only on s2r transitions on x86; most other 2200 architectures force reset to be always executed 2201 i8042.unlock [HW] Unlock (ignore) the keylock 2202 i8042.kbdreset [HW] Reset device connected to KBD port 2203 i8042.probe_defer 2204 [HW] Allow deferred probing upon i8042 probe errors 2205 2206 i810= [HW,DRM] 2207 2208 i915.invert_brightness= 2209 [DRM] Invert the sense of the variable that is used to 2210 set the brightness of the panel backlight. Normally a 2211 brightness value of 0 indicates backlight switched off, 2212 and the maximum of the brightness value sets the backlight 2213 to maximum brightness. If this parameter is set to 0 2214 (default) and the machine requires it, or this parameter 2215 is set to 1, a brightness value of 0 sets the backlight 2216 to maximum brightness, and the maximum of the brightness 2217 value switches the backlight off. 2218 -1 -- never invert brightness 2219 0 -- machine default 2220 1 -- force brightness inversion 2221 2222 ia32_emulation= [X86-64] 2223 Format: <bool> 2224 When true, allows loading 32-bit programs and executing 32-bit 2225 syscalls, essentially overriding IA32_EMULATION_DEFAULT_DISABLED at 2226 boot time. When false, unconditionally disables IA32 emulation. 2227 2228 2229 idle= [X86,EARLY] 2230 Format: idle=poll, idle=halt, idle=nomwait 2231 2232 idle=poll: Don't do power saving in the idle loop 2233 using HLT, but poll for rescheduling event. This will 2234 make the CPUs eat a lot more power, but may be useful 2235 to get slightly better performance in multiprocessor 2236 benchmarks. It also makes some profiling using 2237 performance counters more accurate. Please note that 2238 on systems with MONITOR/MWAIT support (like Intel 2239 EM64T CPUs) this option has no performance advantage 2240 over the normal idle loop. It may also interact badly 2241 with hyperthreading. 2242 2243 idle=halt: Halt is forced to be used for CPU idle. 2244 In such case C2/C3 won't be used again. 2245 2246 idle=nomwait: Disable mwait for CPU C-states 2247 2248 [ARM64,EARLY] 2249 Format: idle=wfi, idle=yield, idle=nop 2250 2251 idle=wfi: Use the WFI (Wait For Interrupt) hint 2252 instruction in the idle loop. This is the default and 2253 allows the CPU to enter a low-power state until an 2254 interrupt arrives. 2255 2256 idle=yield: Use the YIELD hint instruction instead of 2257 WFI. CPUs supporting simultaneous multi-threading (SMT), 2258 can continue executing another thread when the current 2259 thread reaches the idle loop. This will make the CPUs 2260 eat more power, but may be useful to get slightly better 2261 performance in some applications, since the CPUs will 2262 not enter a low-power state. 2263 2264 idle=nop: Do not execute any idle instruction in the 2265 idle loop. This is useful on platforms where WFI 2266 misbehaves, leading to system instability or loss of CPU 2267 state. This will make the CPUs eat more power, but may 2268 give slightly better performance in some applications, 2269 since the CPUs will not enter a low-power state. 2270 2271 idxd.sva= [HW] 2272 Format: <bool> 2273 Allow force disabling of Shared Virtual Memory (SVA) 2274 support for the idxd driver. By default it is set to 2275 true (1). 2276 2277 idxd.tc_override= [HW] 2278 Format: <bool> 2279 Allow override of default traffic class configuration 2280 for the device. By default it is set to false (0). 2281 2282 ieee754= [MIPS] Select IEEE Std 754 conformance mode 2283 Format: { strict | legacy | 2008 | relaxed | emulated } 2284 Default: strict 2285 2286 Choose which programs will be accepted for execution 2287 based on the IEEE 754 NaN encoding(s) supported by 2288 the FPU and the NaN encoding requested with the value 2289 of an ELF file header flag individually set by each 2290 binary. Hardware implementations are permitted to 2291 support either or both of the legacy and the 2008 NaN 2292 encoding mode. 2293 2294 Available settings are as follows: 2295 strict accept binaries that request a NaN encoding 2296 supported by the FPU 2297 legacy only accept legacy-NaN binaries, if supported 2298 by the FPU 2299 2008 only accept 2008-NaN binaries, if supported 2300 by the FPU 2301 relaxed accept any binaries regardless of whether 2302 supported by the FPU 2303 emulated accept any binaries but enable FPU emulator 2304 if binary mode is unsupported by the FPU. 2305 2306 The FPU emulator is always able to support both NaN 2307 encodings, so if no FPU hardware is present or it has 2308 been disabled with 'nofpu', then the settings of 2309 'legacy' and '2008' strap the emulator accordingly, 2310 'relaxed' straps the emulator for both legacy-NaN and 2311 2008-NaN, whereas 'strict' enables legacy-NaN only on 2312 legacy processors and both NaN encodings on MIPS32 or 2313 MIPS64 CPUs. 2314 2315 The setting for ABS.fmt/NEG.fmt instruction execution 2316 mode generally follows that for the NaN encoding, 2317 except where unsupported by hardware. 2318 2319 ignore_loglevel [KNL,EARLY] 2320 Ignore loglevel setting - this will print /all/ 2321 kernel messages to the console. Useful for debugging. 2322 We also add it as printk module parameter, so users 2323 could change it dynamically, usually by 2324 /sys/module/printk/parameters/ignore_loglevel. 2325 2326 ignore_rlimit_data 2327 Ignore RLIMIT_DATA setting for data mappings, 2328 print warning at first misuse. Can be changed via 2329 /sys/module/kernel/parameters/ignore_rlimit_data. 2330 2331 ihash_entries= [KNL] 2332 Set number of hash buckets for inode cache. 2333 2334 ima_appraise= [IMA] appraise integrity measurements 2335 Format: { "off" | "enforce" | "fix" | "log" } 2336 default: "enforce" 2337 2338 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead. 2339 The builtin appraise policy appraises all files 2340 owned by uid=0. 2341 2342 ima_canonical_fmt [IMA] 2343 Use the canonical format for the binary runtime 2344 measurements, instead of host native format. 2345 2346 ima_flush_htable [IMA] 2347 Flush the IMA hash table when deleting all the 2348 staged measurement records, to achieve maximum 2349 memory saving at the cost of having duplicate 2350 records across the staged measurement lists. 2351 2352 ima_hash= [IMA] 2353 Format: { md5 | sha1 | rmd160 | sha256 | sha384 2354 | sha512 | ... } 2355 default: "sha1" 2356 2357 The list of supported hash algorithms is defined 2358 in crypto/hash_info.h. 2359 2360 ima_policy= [IMA] 2361 The builtin policies to load during IMA setup. 2362 Format: "tcb | appraise_tcb | secure_boot | 2363 fail_securely | critical_data" 2364 2365 The "tcb" policy measures all programs exec'd, files 2366 mmap'd for exec, and all files opened with the read 2367 mode bit set by either the effective uid (euid=0) or 2368 uid=0. 2369 2370 The "appraise_tcb" policy appraises the integrity of 2371 all files owned by root. 2372 2373 The "secure_boot" policy appraises the integrity 2374 of files (eg. kexec kernel image, kernel modules, 2375 firmware, policy, etc) based on file signatures. 2376 2377 The "fail_securely" policy forces file signature 2378 verification failure also on privileged mounted 2379 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE 2380 flag. 2381 2382 The "critical_data" policy measures kernel integrity 2383 critical data. 2384 2385 ima_tcb [IMA] Deprecated. Use ima_policy= instead. 2386 Load a policy which meets the needs of the Trusted 2387 Computing Base. This means IMA will measure all 2388 programs exec'd, files mmap'd for exec, and all files 2389 opened for read by uid=0. 2390 2391 ima_template= [IMA] 2392 Select one of defined IMA measurements template formats. 2393 Formats: { "ima" | "ima-ng" | "ima-ngv2" | "ima-sig" | 2394 "ima-sigv2" } 2395 Default: "ima-ng" 2396 2397 ima_template_fmt= 2398 [IMA] Define a custom template format. 2399 Format: { "field1|...|fieldN" } 2400 2401 ima= [IMA] Enable or disable IMA 2402 Format: { "off" | "on" } 2403 Default: "on" 2404 Note that disabling IMA is limited to kdump kernel. 2405 2406 indirect_target_selection= [X86,Intel] Mitigation control for Indirect 2407 Target Selection(ITS) bug in Intel CPUs. Updated 2408 microcode is also required for a fix in IBPB. 2409 2410 on: Enable mitigation (default). 2411 off: Disable mitigation. 2412 force: Force the ITS bug and deploy default 2413 mitigation. 2414 vmexit: Only deploy mitigation if CPU is affected by 2415 guest/host isolation part of ITS. 2416 stuff: Deploy RSB-fill mitigation when retpoline is 2417 also deployed. Otherwise, deploy the default 2418 mitigation. 2419 2420 For details see: 2421 Documentation/admin-guide/hw-vuln/indirect-target-selection.rst 2422 2423 init= [KNL] 2424 Format: <full_path> 2425 Run specified binary instead of /sbin/init as init 2426 process. 2427 2428 initcall_debug [KNL] Trace initcalls as they are executed. Useful 2429 for working out where the kernel is dying during 2430 startup. 2431 2432 initcall_blacklist= [KNL] Do not execute a comma-separated list of 2433 initcall functions. Useful for debugging built-in 2434 modules and initcalls. 2435 2436 initramfs_async= [KNL] 2437 Format: <bool> 2438 Default: 1 2439 This parameter controls whether the initramfs 2440 image is unpacked asynchronously, concurrently 2441 with devices being probed and 2442 initialized. This should normally just work, 2443 but as a debugging aid, one can get the 2444 historical behaviour of the initramfs 2445 unpacking being completed before device_ and 2446 late_ initcalls. 2447 2448 initrd= [BOOT,EARLY] Specify the location of the initial ramdisk 2449 2450 initrdmem= [KNL,EARLY] Specify a physical address and size from which to 2451 load the initrd. If an initrd is compiled in or 2452 specified in the bootparams, it takes priority over this 2453 setting. 2454 Format: ss[KMG],nn[KMG] 2455 Default is 0, 0 2456 2457 init_on_alloc= [MM,EARLY] Fill newly allocated pages and heap objects with 2458 zeroes. 2459 Format: 0 | 1 2460 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON. 2461 2462 init_on_free= [MM,EARLY] Fill freed pages and heap objects with zeroes. 2463 Format: 0 | 1 2464 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON. 2465 2466 init_pkru= [X86] Specify the default memory protection keys rights 2467 register contents for all processes. 0x55555554 by 2468 default (disallow access to all but pkey 0). Can 2469 override in debugfs after boot. 2470 2471 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver 2472 Format: <irq> 2473 2474 int_pln_enable [X86] Enable power limit notification interrupt 2475 2476 integrity_audit=[IMA] 2477 Format: { "0" | "1" } 2478 0 -- basic integrity auditing messages. (Default) 2479 1 -- additional integrity auditing messages. 2480 2481 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option 2482 on 2483 Enable intel iommu driver. 2484 off 2485 Disable intel iommu driver. 2486 igfx_off [Default Off] 2487 By default, gfx is mapped as normal device. If a gfx 2488 device has a dedicated DMAR unit, the DMAR unit is 2489 bypassed by not enabling DMAR with this option. In 2490 this case, gfx device will use physical address for 2491 DMA. 2492 strict [Default Off] 2493 Deprecated, equivalent to iommu.strict=1. 2494 sp_off [Default Off] 2495 By default, super page will be supported if Intel IOMMU 2496 has the capability. With this option, super page will 2497 not be supported. 2498 sm_on 2499 Enable the Intel IOMMU scalable mode if the hardware 2500 advertises that it has support for the scalable mode 2501 translation. 2502 sm_off 2503 Disallow use of the Intel IOMMU scalable mode. 2504 tboot_noforce [Default Off] 2505 Do not force the Intel IOMMU enabled under tboot. 2506 By default, tboot will force Intel IOMMU on, which 2507 could harm performance of some high-throughput 2508 devices like 40GBit network cards, even if identity 2509 mapping is enabled. 2510 Note that using this option lowers the security 2511 provided by tboot because it makes the system 2512 vulnerable to DMA attacks. 2513 2514 intel_idle.max_cstate= [KNL,HW,ACPI,X86] 2515 0 disables intel_idle and fall back on acpi_idle. 2516 1 to 9 specify maximum depth of C-state. 2517 2518 intel_pstate= [X86,EARLY] 2519 disable 2520 Do not enable intel_pstate as the default 2521 scaling driver for the supported processors 2522 active 2523 Use intel_pstate driver to bypass the scaling 2524 governors layer of cpufreq and provides it own 2525 algorithms for p-state selection. There are two 2526 P-state selection algorithms provided by 2527 intel_pstate in the active mode: powersave and 2528 performance. The way they both operate depends 2529 on whether or not the hardware managed P-states 2530 (HWP) feature has been enabled in the processor 2531 and possibly on the processor model. 2532 passive 2533 Use intel_pstate as a scaling driver, but configure it 2534 to work with generic cpufreq governors (instead of 2535 enabling its internal governor). This mode cannot be 2536 used along with the hardware-managed P-states (HWP) 2537 feature. 2538 force 2539 Enable intel_pstate on systems that prohibit it by default 2540 in favor of acpi-cpufreq. Forcing the intel_pstate driver 2541 instead of acpi-cpufreq may disable platform features, such 2542 as thermal controls and power capping, that rely on ACPI 2543 P-States information being indicated to OSPM and therefore 2544 should be used with caution. This option does not work with 2545 processors that aren't supported by the intel_pstate driver 2546 or on platforms that use pcc-cpufreq instead of acpi-cpufreq. 2547 no_hwp 2548 Do not enable hardware P state control (HWP) 2549 if available. 2550 hwp_only 2551 Only load intel_pstate on systems which support 2552 hardware P state control (HWP) if available. 2553 support_acpi_ppc 2554 Enforce ACPI _PPC performance limits. If the Fixed ACPI 2555 Description Table, specifies preferred power management 2556 profile as "Enterprise Server" or "Performance Server", 2557 then this feature is turned on by default. 2558 per_cpu_perf_limits 2559 Allow per-logical-CPU P-State performance control limits using 2560 cpufreq sysfs interface 2561 no_cas 2562 Do not enable capacity-aware scheduling (CAS) on 2563 hybrid systems 2564 2565 intremap= [X86-64,Intel-IOMMU,EARLY] 2566 on enable Interrupt Remapping (default) 2567 off disable Interrupt Remapping 2568 nosid disable Source ID checking 2569 no_x2apic_optout 2570 BIOS x2APIC opt-out request will be ignored 2571 nopost disable Interrupt Posting 2572 posted_msi 2573 enable MSIs delivered as posted interrupts 2574 2575 iomem= Disable strict checking of access to MMIO memory 2576 strict regions from userspace. 2577 relaxed 2578 2579 iommu= [X86,EARLY] 2580 2581 off 2582 Don't initialize and use any kind of IOMMU. 2583 2584 force 2585 Force the use of the hardware IOMMU even when 2586 it is not actually needed (e.g. because < 3 GB 2587 memory). 2588 2589 noforce 2590 Don't force hardware IOMMU usage when it is not 2591 needed. (default). 2592 2593 merge 2594 Do scatter-gather (SG) merging. Implies "force" 2595 (experimental). 2596 2597 nomerge 2598 Don't do scatter-gather (SG) merging. 2599 2600 biomerge 2601 Do scatter-gather (SG) merging. Implies "force" 2602 (experimental). [same as "merge"] 2603 2604 panic 2605 Always panic when IOMMU overflows. 2606 2607 nopanic 2608 Don't panic on IOMMU overflows. 2609 2610 pt 2611 Use passththrough mode by default 2612 (Equivalent to iommu.passthrough=1) 2613 2614 nopt 2615 Use translated mode for DMA by default 2616 (Equivalent to iommu.passthrough=0) 2617 2618 soft 2619 Use software bounce buffering (SWIOTLB) (default for 2620 Intel machines). This can be used to prevent the usage 2621 of an available hardware IOMMU. 2622 2623 usedac 2624 Use the DAC on VIA PCI bridge 2625 (default: disable the VIA PCI bridge DAC) 2626 2627 AMD Gart HW IOMMU-specific options: (CONFIG_GART_IOMMU) 2628 2629 <size> 2630 Set the size of the remapping area in bytes. 2631 2632 allowed 2633 Overwrite iommu off workarounds for specific chipsets 2634 2635 force 2636 Overwrite iommu off workarounds for specific chipsets 2637 2638 fullflush 2639 Flush IOMMU on each allocation (default). 2640 2641 nofullflush 2642 Don't use IOMMU fullflush. 2643 2644 memaper[=<order>] 2645 Allocate an own aperture over RAM with size 2646 32MB<<order. (default: order=1, i.e. 64MB) 2647 2648 noaperture 2649 Ask the IOMMU not to touch the aperture for AGP. 2650 2651 noagp 2652 Don't initialize the AGP driver and use full aperture. 2653 2654 iommu= [PPC/POWERNV] 2655 2656 nobypass 2657 Disable IOMMU bypass, using IOMMU for PCI devices. 2658 2659 iommu.forcedac= [ARM64,X86,EARLY] Control IOVA allocation for PCI devices. 2660 Format: { "0" | "1" } 2661 0 - Try to allocate a 32-bit DMA address first, before 2662 falling back to the full range if needed. 2663 1 - Allocate directly from the full usable range, 2664 forcing Dual Address Cycle for PCI cards supporting 2665 greater than 32-bit addressing. 2666 2667 iommu.strict= [ARM64,X86,S390,EARLY] Configure TLB invalidation behaviour 2668 Format: { "0" | "1" } 2669 0 - Lazy mode. 2670 Request that DMA unmap operations use deferred 2671 invalidation of hardware TLBs, for increased 2672 throughput at the cost of reduced device isolation. 2673 Will fall back to strict mode if not supported by 2674 the relevant IOMMU driver. 2675 1 - Strict mode. 2676 DMA unmap operations invalidate IOMMU hardware TLBs 2677 synchronously. 2678 unset - Use value of CONFIG_IOMMU_DEFAULT_DMA_{LAZY,STRICT}. 2679 Note: on x86, strict mode specified via one of the 2680 legacy driver-specific options takes precedence. 2681 2682 iommu.passthrough= 2683 [ARM64,X86,EARLY] Configure DMA to bypass the IOMMU by default. 2684 Format: { "0" | "1" } 2685 0 - Use IOMMU translation for DMA. 2686 1 - Bypass the IOMMU for DMA. 2687 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH. 2688 2689 iommu.debug_pagealloc= 2690 [KNL,EARLY] When CONFIG_IOMMU_DEBUG_PAGEALLOC is set, this 2691 parameter enables the feature at boot time. By default, it 2692 is disabled and the system behaves the same way as a kernel 2693 built without CONFIG_IOMMU_DEBUG_PAGEALLOC. 2694 Format: { "0" | "1" } 2695 0 - Sanitizer disabled. 2696 1 - Sanitizer enabled, expect runtime overhead. 2697 2698 io7= [HW] IO7 for Marvel-based Alpha systems 2699 See comment before marvel_specify_io7 in 2700 arch/alpha/kernel/core_marvel.c. 2701 2702 io_delay= [X86,EARLY] I/O delay method 2703 0x80 2704 Standard port 0x80 based delay 2705 0xed 2706 Alternate port 0xed based delay (needed on some systems) 2707 udelay 2708 Simple two microseconds delay 2709 none 2710 No delay 2711 2712 ip= [IP_PNP] 2713 See Documentation/admin-guide/nfs/nfsroot.rst. 2714 2715 ipcmni_extend [KNL,EARLY] Extend the maximum number of unique System V 2716 IPC identifiers from 32,768 to 16,777,216. 2717 2718 ipe.enforce= [IPE] 2719 Format: <bool> 2720 Determine whether IPE starts in permissive (0) or 2721 enforce (1) mode. The default is enforce. 2722 2723 ipe.success_audit= 2724 [IPE] 2725 Format: <bool> 2726 Start IPE with success auditing enabled, emitting 2727 an audit event when a binary is allowed. The default 2728 is 0. 2729 2730 irqaffinity= [SMP] Set the default irq affinity mask 2731 The argument is a cpu list, as described above. 2732 2733 irqchip.gicv2_force_probe= 2734 [ARM,ARM64,EARLY] 2735 Format: <bool> 2736 Force the kernel to look for the second 4kB page 2737 of a GICv2 controller even if the memory range 2738 exposed by the device tree is too small. 2739 2740 irqchip.gicv3_nolpi= 2741 [ARM,ARM64,EARLY] 2742 Force the kernel to ignore the availability of 2743 LPIs (and by consequence ITSs). Intended for system 2744 that use the kernel as a bootloader, and thus want 2745 to let secondary kernels in charge of setting up 2746 LPIs. 2747 2748 irqchip.gicv3_pseudo_nmi= [ARM64,EARLY] 2749 Enables support for pseudo-NMIs in the kernel. This 2750 requires the kernel to be built with 2751 CONFIG_ARM64_PSEUDO_NMI. 2752 2753 irqchip.riscv_imsic_noipi 2754 [RISC-V,EARLY] 2755 Force the kernel to not use IMSIC software injected MSIs 2756 as IPIs. Intended for system where IMSIC is trap-n-emulated, 2757 and thus want to reduce MMIO traps when triggering IPIs 2758 to multiple harts. 2759 2760 irqfixup [HW] 2761 When an interrupt is not handled search all handlers 2762 for it. Intended to get systems with badly broken 2763 firmware running. 2764 2765 irqhandler.duration_warn_us= [KNL] 2766 Warn if an IRQ handler exceeds the specified duration 2767 threshold in microseconds. Useful for identifying 2768 long-running IRQs in the system. 2769 2770 irqpoll [HW] 2771 When an interrupt is not handled search all handlers 2772 for it. Also check all handlers each timer 2773 interrupt. Intended to get systems with badly broken 2774 firmware running. 2775 2776 isapnp= [ISAPNP] 2777 Format: <RDP>,<reset>,<pci_scan>,<verbosity> 2778 2779 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance. 2780 Format: [flag-list,]<cpu-list> 2781 2782 Specify one or more CPUs to isolate from disturbances 2783 specified in the flag list (default: domain): 2784 2785 nohz 2786 Disable the tick when a single task runs as well as 2787 disabling other kernel noises like having RCU callbacks 2788 offloaded. This is equivalent to the nohz_full parameter. 2789 2790 A residual 1Hz tick is offloaded to workqueues, which you 2791 need to affine to housekeeping through the global 2792 workqueue's affinity configured via the 2793 /sys/devices/virtual/workqueue/cpumask sysfs file, or 2794 by using the 'domain' flag described below. 2795 2796 NOTE: by default the global workqueue runs on all CPUs, 2797 so to protect individual CPUs the 'cpumask' file has to 2798 be configured manually after bootup. 2799 2800 domain 2801 Isolate from the general SMP balancing and scheduling 2802 algorithms. Note that performing domain isolation this way 2803 is irreversible: it's not possible to bring back a CPU to 2804 the domains once isolated through this boot time 2805 configuration. Use cpusets for a dynamic configuration 2806 which can be altered at runtime. For details see 2807 Documentation/admin-guide/cpu-isolation.rst. 2808 2809 You can move a process onto or off an "isolated" CPU via 2810 the CPU affinity syscalls or cpuset. 2811 <cpu number> begins at 0 and the maximum value is 2812 "number of CPUs in system - 1". 2813 2814 managed_irq 2815 2816 Isolate from being targeted by managed interrupts 2817 which have an interrupt mask containing isolated 2818 CPUs. The affinity of managed interrupts is 2819 handled by the kernel and cannot be changed via 2820 the /proc/irq/* interfaces. 2821 2822 This isolation is best effort and only effective 2823 if the automatically assigned interrupt mask of a 2824 device queue contains isolated and housekeeping 2825 CPUs. If housekeeping CPUs are online then such 2826 interrupts are directed to the housekeeping CPU 2827 so that IO submitted on the housekeeping CPU 2828 cannot disturb the isolated CPU. 2829 2830 If a queue's affinity mask contains only isolated 2831 CPUs then this parameter has no effect on the 2832 interrupt routing decision, though interrupts are 2833 only delivered when tasks running on those 2834 isolated CPUs submit IO. IO submitted on 2835 housekeeping CPUs has no influence on those 2836 queues. 2837 2838 The format of <cpu-list> is described above. 2839 2840 iucv= [HW,NET] 2841 2842 ivrs_ioapic [HW,X86-64] 2843 Provide an override to the IOAPIC-ID<->DEVICE-ID 2844 mapping provided in the IVRS ACPI table. 2845 By default, PCI segment is 0, and can be omitted. 2846 2847 For example, to map IOAPIC-ID decimal 10 to 2848 PCI segment 0x1 and PCI device 00:14.0, 2849 write the parameter as: 2850 ivrs_ioapic=10@0001:00:14.0 2851 2852 Deprecated formats: 2853 * To map IOAPIC-ID decimal 10 to PCI device 00:14.0 2854 write the parameter as: 2855 ivrs_ioapic[10]=00:14.0 2856 * To map IOAPIC-ID decimal 10 to PCI segment 0x1 and 2857 PCI device 00:14.0 write the parameter as: 2858 ivrs_ioapic[10]=0001:00:14.0 2859 2860 ivrs_hpet [HW,X86-64] 2861 Provide an override to the HPET-ID<->DEVICE-ID 2862 mapping provided in the IVRS ACPI table. 2863 By default, PCI segment is 0, and can be omitted. 2864 2865 For example, to map HPET-ID decimal 10 to 2866 PCI segment 0x1 and PCI device 00:14.0, 2867 write the parameter as: 2868 ivrs_hpet=10@0001:00:14.0 2869 2870 Deprecated formats: 2871 * To map HPET-ID decimal 0 to PCI device 00:14.0 2872 write the parameter as: 2873 ivrs_hpet[0]=00:14.0 2874 * To map HPET-ID decimal 10 to PCI segment 0x1 and 2875 PCI device 00:14.0 write the parameter as: 2876 ivrs_ioapic[10]=0001:00:14.0 2877 2878 ivrs_acpihid [HW,X86-64] 2879 Provide an override to the ACPI-HID:UID<->DEVICE-ID 2880 mapping provided in the IVRS ACPI table. 2881 By default, PCI segment is 0, and can be omitted. 2882 2883 For example, to map UART-HID:UID AMD0020:0 to 2884 PCI segment 0x1 and PCI device ID 00:14.5, 2885 write the parameter as: 2886 ivrs_acpihid=AMD0020:0@0001:00:14.5 2887 2888 Deprecated formats: 2889 * To map UART-HID:UID AMD0020:0 to PCI segment is 0, 2890 PCI device ID 00:14.5, write the parameter as: 2891 ivrs_acpihid[00:14.5]=AMD0020:0 2892 * To map UART-HID:UID AMD0020:0 to PCI segment 0x1 and 2893 PCI device ID 00:14.5, write the parameter as: 2894 ivrs_acpihid[0001:00:14.5]=AMD0020:0 2895 2896 js= [HW,JOY] Analog joystick 2897 See Documentation/input/joydev/joystick.rst. 2898 2899 kasan_multi_shot 2900 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print 2901 report on every invalid memory access. Without this 2902 parameter KASAN will print report only for the first 2903 invalid access. 2904 2905 keep_bootcon [KNL,EARLY] 2906 Do not unregister boot console at start. This is only 2907 useful for debugging when something happens in the window 2908 between unregistering the boot console and initializing 2909 the real console. 2910 2911 keepinitrd [HW,ARM] See retain_initrd. 2912 2913 kernelcore= [KNL,X86,PPC,EARLY] 2914 Format: nn[KMGTPE] | nn% | "mirror" 2915 This parameter specifies the amount of memory usable by 2916 the kernel for non-movable allocations. The requested 2917 amount is spread evenly throughout all nodes in the 2918 system as ZONE_NORMAL. The remaining memory is used for 2919 movable memory in its own zone, ZONE_MOVABLE. In the 2920 event, a node is too small to have both ZONE_NORMAL and 2921 ZONE_MOVABLE, kernelcore memory will take priority and 2922 other nodes will have a larger ZONE_MOVABLE. 2923 2924 ZONE_MOVABLE is used for the allocation of pages that 2925 may be reclaimed or moved by the page migration 2926 subsystem. Note that allocations like PTEs-from-HighMem 2927 still use the HighMem zone if it exists, and the Normal 2928 zone if it does not. 2929 2930 It is possible to specify the exact amount of memory in 2931 the form of "nn[KMGTPE]", a percentage of total system 2932 memory in the form of "nn%", or "mirror". If "mirror" 2933 option is specified, mirrored (reliable) memory is used 2934 for non-movable allocations and remaining memory is used 2935 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror" 2936 are exclusive, so you cannot specify multiple forms. 2937 2938 kfence.burst= [MM,KFENCE] The number of additional successive 2939 allocations to be attempted through KFENCE for each 2940 sample interval. 2941 Format: <unsigned integer> 2942 Default: 0 2943 2944 kfence.check_on_panic= 2945 [MM,KFENCE] Whether to check all KFENCE-managed objects' 2946 canaries on panic. 2947 Format: <bool> 2948 Default: false 2949 2950 kfence.deferrable= 2951 [MM,KFENCE] Whether to use a deferrable timer to trigger 2952 allocations. This avoids forcing CPU wake-ups if the 2953 system is idle, at the risk of a less predictable 2954 sample interval. 2955 Format: <bool> 2956 Default: CONFIG_KFENCE_DEFERRABLE 2957 2958 kfence.fault= [MM,KFENCE] Controls the behavior when a KFENCE 2959 error is detected. 2960 report - print the error report and continue (default). 2961 oops - print the error report and oops. 2962 panic - print the error report and panic. 2963 2964 kfence.sample_interval= 2965 [MM,KFENCE] KFENCE's sample interval in milliseconds. 2966 Format: <unsigned integer> 2967 0 - Disable KFENCE. 2968 >0 - Enabled KFENCE with given sample interval. 2969 Default: CONFIG_KFENCE_SAMPLE_INTERVAL 2970 2971 kfence.skip_covered_thresh= 2972 [MM,KFENCE] If pool utilization reaches this threshold 2973 (pool usage%), KFENCE limits currently covered 2974 allocations of the same source from further filling 2975 up the pool. 2976 Format: <unsigned integer> 2977 Default: 75 2978 2979 kgdbdbgp= [KGDB,HW,EARLY] kgdb over EHCI usb debug port. 2980 Format: <Controller#>[,poll interval] 2981 The controller # is the number of the ehci usb debug 2982 port as it is probed via PCI. The poll interval is 2983 optional and is the number seconds in between 2984 each poll cycle to the debug port in case you need 2985 the functionality for interrupting the kernel with 2986 gdb or control-c on the dbgp connection. When 2987 not using this parameter you use sysrq-g to break into 2988 the kernel debugger. 2989 2990 kgdboc= [KGDB,HW] kgdb over consoles. 2991 Requires a tty driver that supports console polling, 2992 or a supported polling keyboard driver (non-usb). 2993 Serial only format: <serial_device>[,baud] 2994 keyboard only format: kbd 2995 keyboard and serial format: kbd,<serial_device>[,baud] 2996 Optional Kernel mode setting: 2997 kms, kbd format: kms,kbd 2998 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud] 2999 3000 kgdboc_earlycon= [KGDB,HW,EARLY] 3001 If the boot console provides the ability to read 3002 characters and can work in polling mode, you can use 3003 this parameter to tell kgdb to use it as a backend 3004 until the normal console is registered. Intended to 3005 be used together with the kgdboc parameter which 3006 specifies the normal console to transition to. 3007 3008 The name of the early console should be specified 3009 as the value of this parameter. Note that the name of 3010 the early console might be different than the tty 3011 name passed to kgdboc. It's OK to leave the value 3012 blank and the first boot console that implements 3013 read() will be picked. 3014 3015 kgdbwait [KGDB,EARLY] Stop kernel execution and enter the 3016 kernel debugger at the earliest opportunity. 3017 3018 kho= [KEXEC,EARLY] 3019 Format: { "0" | "1" | "off" | "on" | "y" | "n" } 3020 Enables or disables Kexec HandOver. 3021 "0" | "off" | "n" - kexec handover is disabled 3022 "1" | "on" | "y" - kexec handover is enabled 3023 3024 kho_bootmem= [KEXEC,EARLY] 3025 Format: ll[KMG],mm[KMG],nn[KMG] | nn% 3026 3027 Defines the size of the KHO boot memory region. The 3028 KHO boot memory regions are physically contiguous 3029 memory ranges that can only be used for non-kernel 3030 allocations. That way, even when memory is heavily 3031 fragmented with handed over memory, the kexeced 3032 kernel will always have enough contiguous ranges to 3033 bootstrap itself. 3034 3035 It is possible to specify the exact amount of 3036 memory in the form of "ll[KMG],mm[KMG],nn[KMG]" 3037 where the first parameter defines the size of a low 3038 memory bootmem area, the second parameter defines 3039 the size of a global bootmem area and the third 3040 parameter defines the size of additional per-node 3041 bootmem areas. The form "nn%" defines scale factor 3042 (in percents) of memory that was used during boot. 3043 3044 kmac= [MIPS] Korina ethernet MAC address. 3045 Configure the RouterBoard 532 series on-chip 3046 Ethernet adapter MAC address. 3047 3048 kmemleak= [KNL,EARLY] Boot-time kmemleak enable/disable 3049 Valid arguments: on, off 3050 Default: on 3051 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y, 3052 the default is off. 3053 3054 kprobe_event=[probe-list] 3055 [FTRACE] Add kprobe events and enable at boot time. 3056 The probe-list is a semicolon delimited list of probe 3057 definitions. Each definition is same as kprobe_events 3058 interface, but the parameters are comma delimited. 3059 For example, to add a kprobe event on vfs_read with 3060 arg1 and arg2, add to the command line; 3061 3062 kprobe_event=p,vfs_read,$arg1,$arg2 3063 3064 See also Documentation/trace/kprobetrace.rst "Kernel 3065 Boot Parameter" section. 3066 3067 kpti= [ARM64,EARLY] Control page table isolation of 3068 user and kernel address spaces. 3069 Default: enabled on cores which need mitigation. 3070 0: force disabled 3071 1: force enabled 3072 3073 kunit.enable= [KUNIT] Enable executing KUnit tests. Requires 3074 CONFIG_KUNIT to be set to be fully enabled. The 3075 default value can be overridden via 3076 KUNIT_DEFAULT_ENABLED. 3077 Default is 1 (enabled) 3078 3079 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs. 3080 Default is 0 (don't ignore, but inject #GP) 3081 3082 kvm.eager_page_split= 3083 [KVM,X86,RISCV] Controls whether or not KVM will try to 3084 proactively split all huge pages during dirty logging. 3085 Eager page splitting reduces interruptions to vCPU 3086 execution by eliminating the write-protection faults 3087 and MMU lock contention that would otherwise be 3088 required to split huge pages lazily. 3089 3090 VM workloads that rarely perform writes or that write 3091 only to a small region of VM memory may benefit from 3092 disabling eager page splitting to allow huge pages to 3093 still be used for reads. 3094 3095 The behavior of eager page splitting depends on whether 3096 KVM_DIRTY_LOG_INITIALLY_SET is enabled or disabled. If 3097 disabled, all huge pages in a memslot will be eagerly 3098 split when dirty logging is enabled on that memslot. If 3099 enabled, eager page splitting will be performed during 3100 the KVM_CLEAR_DIRTY ioctl, and only for the pages being 3101 cleared. 3102 3103 On x86, eager page splitting is only supported when 3104 kvm.tdp_mmu=Y. 3105 3106 On RISCV, eager page splitting is supported by default. 3107 3108 Default is Y (on). 3109 3110 kvm.enable_pmu=[KVM,X86] 3111 If enabled, KVM will virtualize PMU functionality based 3112 on the virtual CPU model defined by userspace. This 3113 can be overridden on a per-VM basis via 3114 KVM_CAP_PMU_CAPABILITY. 3115 3116 If disabled, KVM will not virtualize PMU functionality, 3117 e.g. MSRs, PMCs, PMIs, etc., even if userspace defines 3118 a virtual CPU model that contains PMU assets. 3119 3120 Note, KVM's vPMU support implicitly requires running 3121 with an in-kernel local APIC, e.g. to deliver PMIs to 3122 the guest. Running without an in-kernel local APIC is 3123 not supported, though KVM will allow such a combination 3124 (with severely degraded functionality). 3125 3126 See also enable_mediated_pmu. 3127 3128 Default is Y (on). 3129 3130 kvm.enable_virt_at_load=[KVM,ARM64,LOONGARCH,MIPS,RISCV,X86] 3131 If enabled, KVM will enable virtualization in hardware 3132 when KVM is loaded, and disable virtualization when KVM 3133 is unloaded (if KVM is built as a module). 3134 3135 If disabled, KVM will dynamically enable and disable 3136 virtualization on-demand when creating and destroying 3137 VMs, i.e. on the 0=>1 and 1=>0 transitions of the 3138 number of VMs. 3139 3140 Enabling virtualization at module load avoids potential 3141 latency for creation of the 0=>1 VM, as KVM serializes 3142 virtualization enabling across all online CPUs. The 3143 "cost" of enabling virtualization when KVM is loaded, 3144 is that doing so may interfere with using out-of-tree 3145 hypervisors that want to "own" virtualization hardware. 3146 3147 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface. 3148 Default is false (don't support). 3149 3150 kvm.gmem_in_place_conversion= 3151 [KVM] Controls whether KVM enables in-place conversion 3152 support for guest_memfd and tracks the private/shared 3153 state of memory per guest_memfd instead of per VM. 3154 3155 If enabled, KVM enables the KVM_SET_MEMORY_ATTRIBUTES2 3156 ioctl on guest_memfd file descriptors and disables the 3157 legacy VM-scoped KVM_SET_MEMORY_ATTRIBUTES ioctl for 3158 private memory state tracking. Only the 3159 KVM_MEMORY_ATTRIBUTE_PRIVATE attribute moves to 3160 per-guest_memfd tracking; other attributes remain 3161 per-VM. 3162 3163 This parameter toggles KVM's in-place conversion 3164 capability support. Whether a VMM uses separate backends 3165 or out-of-place memory management is determined by 3166 userspace VMM design. 3167 3168 Note, this parameter is only available when 3169 CONFIG_KVM_PRIVATE_MEMATTR_VM=y. When 3170 CONFIG_KVM_PRIVATE_MEMATTR_VM is not set, in-place 3171 conversion is enabled (and cannot be disabled) if KVM 3172 is built with private memory support. 3173 3174 Default is N (off). 3175 3176 kvm.nx_huge_pages= 3177 [KVM] Controls the software workaround for the 3178 X86_BUG_ITLB_MULTIHIT bug. 3179 force : Always deploy workaround. 3180 off : Never deploy workaround. 3181 auto : Deploy workaround based on the presence of 3182 X86_BUG_ITLB_MULTIHIT. 3183 3184 Default is 'auto'. 3185 3186 If the software workaround is enabled for the host, 3187 guests do need not to enable it for nested guests. 3188 3189 kvm.nx_huge_pages_recovery_ratio= 3190 [KVM] Controls how many 4KiB pages are periodically zapped 3191 back to huge pages. 0 disables the recovery, otherwise if 3192 the value is N KVM will zap 1/Nth of the 4KiB pages every 3193 period (see below). The default is 60. 3194 3195 kvm.nx_huge_pages_recovery_period_ms= 3196 [KVM] Controls the time period at which KVM zaps 4KiB pages 3197 back to huge pages. If the value is a non-zero N, KVM will 3198 zap a portion (see ratio above) of the pages every N msecs. 3199 If the value is 0 (the default), KVM will pick a period based 3200 on the ratio, such that a page is zapped after 1 hour on average. 3201 3202 kvm-{amd,intel}.enable_mediated_pmu=[KVM,AMD,INTEL] 3203 If enabled, KVM will provide a mediated virtual PMU, 3204 instead of the default perf-based virtual PMU (if 3205 kvm.enable_pmu is true and PMU is enumerated via the 3206 virtual CPU model). 3207 3208 With a perf-based vPMU, KVM operates as a user of perf, 3209 i.e. emulates guest PMU counters using perf events. 3210 KVM-created perf events are managed by perf as regular 3211 (guest-only) events, e.g. are scheduled in/out, contend 3212 for hardware resources, etc. Using a perf-based vPMU 3213 allows guest and host usage of the PMU to co-exist, but 3214 incurs non-trivial overhead and can result in silently 3215 dropped guest events (due to resource contention). 3216 3217 With a mediated vPMU, hardware PMU state is context 3218 switched around the world switch to/from the guest. 3219 KVM mediates which events the guest can utilize, but 3220 gives the guest direct access to all other PMU assets 3221 when possible (KVM may intercept some accesses if the 3222 virtual CPU model provides a subset of hardware PMU 3223 functionality). Using a mediated vPMU significantly 3224 reduces PMU virtualization overhead and eliminates lost 3225 guest events, but is mutually exclusive with using perf 3226 to profile KVM guests and adds latency to most VM-Exits 3227 (to context switch PMU state). 3228 3229 Default is N (off). 3230 3231 kvm-amd.nested= [KVM,AMD] Control nested virtualization feature in 3232 KVM/SVM. Default is 1 (enabled). 3233 3234 kvm-amd.npt= [KVM,AMD] Control KVM's use of Nested Page Tables, 3235 a.k.a. Two-Dimensional Page Tables. Default is 1 3236 (enabled). Disable by KVM if hardware lacks support 3237 for NPT. 3238 3239 kvm-amd.ciphertext_hiding_asids= 3240 [KVM,AMD] Ciphertext hiding prevents disallowed accesses 3241 to SNP private memory from reading ciphertext. Instead, 3242 reads will see constant default values (0xff). 3243 3244 If ciphertext hiding is enabled, the joint SEV-ES and 3245 SEV-SNP ASID space is partitioned into separate SEV-ES 3246 and SEV-SNP ASID ranges, with the SEV-SNP range being 3247 [1..max_snp_asid] and the SEV-ES range being 3248 (max_snp_asid..min_sev_asid), where min_sev_asid is 3249 enumerated by CPUID.0x.8000_001F[EDX]. 3250 3251 A non-zero value enables SEV-SNP ciphertext hiding and 3252 adjusts the ASID ranges for SEV-ES and SEV-SNP guests. 3253 KVM caps the number of SEV-SNP ASIDs at the maximum 3254 possible value, e.g. specifying -1u will assign all 3255 joint SEV-ES and SEV-SNP ASIDs to SEV-SNP. Note, 3256 assigning all joint ASIDs to SEV-SNP, i.e. configuring 3257 max_snp_asid == min_sev_asid-1, will effectively make 3258 SEV-ES unusable. 3259 3260 kvm-arm.mode= 3261 [KVM,ARM,EARLY] Select one of KVM/arm64's modes of 3262 operation. 3263 3264 none: Forcefully disable KVM. 3265 3266 nvhe: Standard nVHE-based mode, without support for 3267 protected guests. 3268 3269 protected: Mode with support for guests whose state is 3270 kept private from the host, using VHE or 3271 nVHE depending on HW support. 3272 3273 nested: VHE-based mode with support for nested 3274 virtualization. Requires at least ARMv8.4 3275 hardware (with FEAT_NV2). 3276 3277 Defaults to VHE/nVHE based on hardware support. Setting 3278 mode to "protected" will disable kexec and hibernation 3279 for the host. To force nVHE on VHE hardware, add 3280 "arm64_sw.hvhe=0 id_aa64mmfr1.vh=0" to the 3281 command-line. 3282 "nested" and "protected" are experimental and should be 3283 used with extreme caution. 3284 3285 kvm-arm.vgic_v3_group0_trap= 3286 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-0 3287 system registers 3288 3289 kvm-arm.vgic_v3_group1_trap= 3290 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-1 3291 system registers 3292 3293 kvm-arm.vgic_v3_common_trap= 3294 [KVM,ARM,EARLY] Trap guest accesses to GICv3 common 3295 system registers 3296 3297 kvm-arm.vgic_v4_enable= 3298 [KVM,ARM,EARLY] Allow use of GICv4 for direct 3299 injection of LPIs. 3300 3301 kvm-arm.wfe_trap_policy= 3302 [KVM,ARM] Control when to set WFE instruction trap for 3303 KVM VMs. Traps are allowed but not guaranteed by the 3304 CPU architecture. 3305 3306 trap: set WFE instruction trap 3307 3308 notrap: clear WFE instruction trap 3309 3310 kvm-arm.wfi_trap_policy= 3311 [KVM,ARM] Control when to set WFI instruction trap for 3312 KVM VMs. Traps are allowed but not guaranteed by the 3313 CPU architecture. 3314 3315 trap: set WFI instruction trap 3316 3317 notrap: clear WFI instruction trap 3318 3319 kvm_cma_resv_ratio=n [PPC,EARLY] 3320 Reserves given percentage from system memory area for 3321 contiguous memory allocation for KVM hash pagetable 3322 allocation. 3323 By default it reserves 5% of total system memory. 3324 Format: <integer> 3325 Default: 5 3326 3327 kvm-intel.ept= [KVM,Intel] Control KVM's use of Extended Page Tables, 3328 a.k.a. Two-Dimensional Page Tables. Default is 1 3329 (enabled). Disable by KVM if hardware lacks support 3330 for EPT. 3331 3332 kvm-intel.emulate_invalid_guest_state= 3333 [KVM,Intel] Control whether to emulate invalid guest 3334 state. Ignored if kvm-intel.enable_unrestricted_guest=1, 3335 as guest state is never invalid for unrestricted 3336 guests. This param doesn't apply to nested guests (L2), 3337 as KVM never emulates invalid L2 guest state. 3338 Default is 1 (enabled). 3339 3340 kvm-intel.flexpriority= 3341 [KVM,Intel] Control KVM's use of FlexPriority feature 3342 (TPR shadow). Default is 1 (enabled). Disable by KVM if 3343 hardware lacks support for it. 3344 3345 kvm-intel.nested= 3346 [KVM,Intel] Control nested virtualization feature in 3347 KVM/VMX. Default is 1 (enabled). 3348 3349 kvm-intel.unrestricted_guest= 3350 [KVM,Intel] Control KVM's use of unrestricted guest 3351 feature (virtualized real and unpaged mode). Default 3352 is 1 (enabled). Disable by KVM if EPT is disabled or 3353 hardware lacks support for it. 3354 3355 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault 3356 CVE-2018-3620. 3357 3358 Valid arguments: never, cond, always 3359 3360 always: L1D cache flush on every VMENTER. 3361 cond: Flush L1D on VMENTER only when the code between 3362 VMEXIT and VMENTER can leak host memory. 3363 never: Disables the mitigation 3364 3365 Default is cond (do L1 cache flush in specific instances) 3366 3367 kvm-intel.vpid= [KVM,Intel] Control KVM's use of Virtual Processor 3368 Identification feature (tagged TLBs). Default is 1 3369 (enabled). Disable by KVM if hardware lacks support 3370 for it. 3371 3372 l1d_flush= [X86,INTEL,EARLY] 3373 Control mitigation for L1D based snooping vulnerability. 3374 3375 Certain CPUs are vulnerable to an exploit against CPU 3376 internal buffers which can forward information to a 3377 disclosure gadget under certain conditions. 3378 3379 In vulnerable processors, the speculatively 3380 forwarded data can be used in a cache side channel 3381 attack, to access data to which the attacker does 3382 not have direct access. 3383 3384 This parameter controls the mitigation. The 3385 options are: 3386 3387 on - enable the interface for the mitigation 3388 3389 l1tf= [X86,EARLY] Control mitigation of the L1TF vulnerability on 3390 affected CPUs 3391 3392 The kernel PTE inversion protection is unconditionally 3393 enabled and cannot be disabled. 3394 3395 full 3396 Provides all available mitigations for the 3397 L1TF vulnerability. Disables SMT and 3398 enables all mitigations in the 3399 hypervisors, i.e. unconditional L1D flush. 3400 3401 SMT control and L1D flush control via the 3402 sysfs interface is still possible after 3403 boot. Hypervisors will issue a warning 3404 when the first VM is started in a 3405 potentially insecure configuration, 3406 i.e. SMT enabled or L1D flush disabled. 3407 3408 full,force 3409 Same as 'full', but disables SMT and L1D 3410 flush runtime control. Implies the 3411 'nosmt=force' command line option. 3412 (i.e. sysfs control of SMT is disabled.) 3413 3414 flush 3415 Leaves SMT enabled and enables the default 3416 hypervisor mitigation, i.e. conditional 3417 L1D flush. 3418 3419 SMT control and L1D flush control via the 3420 sysfs interface is still possible after 3421 boot. Hypervisors will issue a warning 3422 when the first VM is started in a 3423 potentially insecure configuration, 3424 i.e. SMT enabled or L1D flush disabled. 3425 3426 flush,nosmt 3427 3428 Disables SMT and enables the default 3429 hypervisor mitigation. 3430 3431 SMT control and L1D flush control via the 3432 sysfs interface is still possible after 3433 boot. Hypervisors will issue a warning 3434 when the first VM is started in a 3435 potentially insecure configuration, 3436 i.e. SMT enabled or L1D flush disabled. 3437 3438 flush,nowarn 3439 Same as 'flush', but hypervisors will not 3440 warn when a VM is started in a potentially 3441 insecure configuration. 3442 3443 off 3444 Disables hypervisor mitigations and doesn't 3445 emit any warnings. 3446 It also drops the swap size and available 3447 RAM limit restriction on both hypervisor and 3448 bare metal. 3449 3450 Default is 'flush'. 3451 3452 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst 3453 3454 l2cr= [PPC] 3455 3456 l3cr= [PPC] 3457 3458 lapic [X86-32,APIC,EARLY] Enable the local APIC even if BIOS 3459 disabled it. 3460 3461 lapic= [X86,APIC] Do not use TSC deadline 3462 value for LAPIC timer one-shot implementation. Default 3463 back to the programmable timer unit in the LAPIC. 3464 Format: notscdeadline 3465 3466 lapic_timer_c2_ok [X86,APIC,EARLY] trust the local apic timer 3467 in C2 power state. 3468 3469 libata.dma= [LIBATA] DMA control 3470 libata.dma=0 Disable all PATA and SATA DMA 3471 libata.dma=1 PATA and SATA Disk DMA only 3472 libata.dma=2 ATAPI (CDROM) DMA only 3473 libata.dma=4 Compact Flash DMA only 3474 Combinations also work, so libata.dma=3 enables DMA 3475 for disks and CDROMs, but not CFs. 3476 3477 libata.ignore_hpa= [LIBATA] Ignore HPA limit 3478 libata.ignore_hpa=0 keep BIOS limits (default) 3479 libata.ignore_hpa=1 ignore limits, using full disk 3480 3481 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume 3482 when set. 3483 Format: <int> 3484 3485 libata.force= [LIBATA] Force configurations. The format is a comma- 3486 separated list of "[ID:]VAL" where ID is PORT[.DEVICE]. 3487 PORT and DEVICE are decimal numbers matching port, link 3488 or device. Basically, it matches the ATA ID string 3489 printed on console by libata. If the whole ID part is 3490 omitted, the last PORT and DEVICE values are used. If 3491 ID hasn't been specified yet, the configuration applies 3492 to all ports, links and devices. 3493 3494 If only DEVICE is omitted, the parameter applies to 3495 the port and all links and devices behind it. DEVICE 3496 number of 0 either selects the first device or the 3497 first fan-out link behind PMP device. It does not 3498 select the host link. DEVICE number of 15 selects the 3499 host link and device attached to it. 3500 3501 The VAL specifies the configuration to force. As long 3502 as there is no ambiguity, shortcut notation is allowed. 3503 For example, both 1.5 and 1.5G would work for 1.5Gbps. 3504 The following configurations can be forced. 3505 3506 * Cable type: 40c, 80c, short40c, unk, ign or sata. 3507 Any ID with matching PORT is used. 3508 3509 * SATA link speed limit: 1.5Gbps or 3.0Gbps. 3510 3511 * Transfer mode: pio[0-6], mwdma[0-4] and udma[0-7]. 3512 udma[/][16,25,33,44,66,100,133] notation is also 3513 allowed. 3514 3515 * nohrst, nosrst, norst: suppress hard, soft and both 3516 resets. 3517 3518 * rstonce: only attempt one reset during hot-unplug 3519 link recovery. 3520 3521 * [no]dbdelay: Enable or disable the extra 200ms delay 3522 before debouncing a link PHY and device presence 3523 detection. 3524 3525 * [no]ncq: Turn on or off NCQ. 3526 3527 * [no]ncqtrim: Enable or disable queued DSM TRIM. 3528 3529 * [no]ncqati: Enable or disable NCQ on ATI chipsets. 3530 3531 * [no]trim: Enable or disable (unqueued) TRIM. 3532 3533 * trim_zero: Indicate that TRIM command zeroes data. 3534 3535 * max_trim_128m: Set 128M maximum trim size limit. 3536 3537 * [no]dma: Turn on or off DMA transfers. 3538 3539 * atapi_dmadir: Enable ATAPI DMADIR bridge support. 3540 3541 * atapi_mod16_dma: Enable the use of ATAPI DMA for 3542 commands that are not a multiple of 16 bytes. 3543 3544 * [no]dmalog: Enable or disable the use of the 3545 READ LOG DMA EXT command to access logs. 3546 3547 * [no]iddevlog: Enable or disable access to the 3548 identify device data log. 3549 3550 * [no]logdir: Enable or disable access to the general 3551 purpose log directory. 3552 3553 * max_sec=<sectors>: Set the transfer size limit, in 3554 number of 512-byte sectors, to the value specified in 3555 <sectors>. The value specified in <sectors> has to be 3556 a non-zero positive integer. 3557 3558 * max_sec_128: Set transfer size limit to 128 sectors. 3559 3560 * max_sec_1024: Set transfer size limit to 1024 sectors. 3561 3562 * max_sec_lba48: Set transfer size limit to 65535 sectors. 3563 3564 * external: Mark port as external (hotplug-capable). 3565 3566 * [no]lpm: Enable or disable link power management. 3567 3568 * [no]setxfer: Indicate if transfer speed mode setting 3569 should be skipped. 3570 3571 * [no]fua: Disable or enable FUA (Force Unit Access) 3572 support for devices supporting this feature. 3573 3574 * dump_id: Dump IDENTIFY data. 3575 3576 * disable: Disable this device. 3577 3578 If there are multiple matching configurations changing 3579 the same attribute, the last one is used. 3580 3581 liveupdate= [KNL,EARLY] 3582 Format: <bool> 3583 Enable Live Update Orchestrator (LUO). 3584 Default: off. 3585 3586 lockd.nlm_grace_period=P [NFS] Assign grace period. 3587 Format: <integer> 3588 3589 lockd.nlm_tcpport=N [NFS] Assign TCP port. 3590 Format: <integer> 3591 3592 lockd.nlm_timeout=T [NFS] Assign timeout value. 3593 Format: <integer> 3594 3595 lockd.nlm_udpport=M [NFS] Assign UDP port. 3596 Format: <integer> 3597 3598 lockdown= [SECURITY,EARLY] 3599 { integrity | confidentiality } 3600 Enable the kernel lockdown feature. If set to 3601 integrity, kernel features that allow userland to 3602 modify the running kernel are disabled. If set to 3603 confidentiality, kernel features that allow userland 3604 to extract confidential information from the kernel 3605 are also disabled. 3606 3607 locktorture.acq_writer_lim= [KNL] 3608 Set the time limit in jiffies for a lock 3609 acquisition. Acquisitions exceeding this limit 3610 will result in a splat once they do complete. 3611 3612 locktorture.bind_readers= [KNL] 3613 Specify the list of CPUs to which the readers are 3614 to be bound. 3615 3616 locktorture.bind_writers= [KNL] 3617 Specify the list of CPUs to which the writers are 3618 to be bound. 3619 3620 locktorture.call_rcu_chains= [KNL] 3621 Specify the number of self-propagating call_rcu() 3622 chains to set up. These are used to ensure that 3623 there is a high probability of an RCU grace period 3624 in progress at any given time. Defaults to 0, 3625 which disables these call_rcu() chains. 3626 3627 locktorture.long_hold= [KNL] 3628 Specify the duration in milliseconds for the 3629 occasional long-duration lock hold time. Defaults 3630 to 100 milliseconds. Select 0 to disable. 3631 3632 locktorture.nested_locks= [KNL] 3633 Specify the maximum lock nesting depth that 3634 locktorture is to exercise, up to a limit of 8 3635 (MAX_NESTED_LOCKS). Specify zero to disable. 3636 Note that this parameter is ineffective on types 3637 of locks that do not support nested acquisition. 3638 3639 locktorture.nreaders_stress= [KNL] 3640 Set the number of locking read-acquisition kthreads. 3641 Defaults to being automatically set based on the 3642 number of online CPUs. 3643 3644 locktorture.nwriters_stress= [KNL] 3645 Set the number of locking write-acquisition kthreads. 3646 3647 locktorture.onoff_holdoff= [KNL] 3648 Set time (s) after boot for CPU-hotplug testing. 3649 3650 locktorture.onoff_interval= [KNL] 3651 Set time (s) between CPU-hotplug operations, or 3652 zero to disable CPU-hotplug testing. 3653 3654 locktorture.rt_boost= [KNL] 3655 Do periodic testing of real-time lock priority 3656 boosting. Select 0 to disable, 1 to boost 3657 only rt_mutex, and 2 to boost unconditionally. 3658 Defaults to 2, which might seem to be an 3659 odd choice, but which should be harmless for 3660 non-real-time spinlocks, due to their disabling 3661 of preemption. Note that non-realtime mutexes 3662 disable boosting. 3663 3664 locktorture.rt_boost_factor= [KNL] 3665 Number that determines how often and for how 3666 long priority boosting is exercised. This is 3667 scaled down by the number of writers, so that the 3668 number of boosts per unit time remains roughly 3669 constant as the number of writers increases. 3670 On the other hand, the duration of each boost 3671 increases with the number of writers. 3672 3673 locktorture.shuffle_interval= [KNL] 3674 Set task-shuffle interval (jiffies). Shuffling 3675 tasks allows some CPUs to go into dyntick-idle 3676 mode during the locktorture test. 3677 3678 locktorture.shutdown_secs= [KNL] 3679 Set time (s) after boot system shutdown. This 3680 is useful for hands-off automated testing. 3681 3682 locktorture.stat_interval= [KNL] 3683 Time (s) between statistics printk()s. 3684 3685 locktorture.stutter= [KNL] 3686 Time (s) to stutter testing, for example, 3687 specifying five seconds causes the test to run for 3688 five seconds, wait for five seconds, and so on. 3689 This tests the locking primitive's ability to 3690 transition abruptly to and from idle. 3691 3692 locktorture.torture_type= [KNL] 3693 Specify the locking implementation to test. 3694 3695 locktorture.verbose= [KNL] 3696 Enable additional printk() statements. 3697 3698 locktorture.writer_fifo= [KNL] 3699 Run the write-side locktorture kthreads at 3700 sched_set_fifo() real-time priority. 3701 3702 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver 3703 Format: <irq> 3704 3705 loglevel= [KNL,EARLY] 3706 All Kernel Messages with a loglevel smaller than the 3707 console loglevel will be printed to the console. It can 3708 also be changed with klogd or other programs. The 3709 loglevels are defined as follows: 3710 3711 0 (KERN_EMERG) system is unusable 3712 1 (KERN_ALERT) action must be taken immediately 3713 2 (KERN_CRIT) critical conditions 3714 3 (KERN_ERR) error conditions 3715 4 (KERN_WARNING) warning conditions 3716 5 (KERN_NOTICE) normal but significant condition 3717 6 (KERN_INFO) informational 3718 7 (KERN_DEBUG) debug-level messages 3719 3720 log_buf_len=n[KMG] [KNL,EARLY] 3721 Sets the size of the printk ring buffer, in bytes. 3722 n must be a power of two and greater than the 3723 minimal size. The minimal size is defined by 3724 LOG_BUF_SHIFT kernel config parameter. There 3725 is also CONFIG_LOG_CPU_MAX_BUF_SHIFT config 3726 parameter that allows to increase the default size 3727 depending on the number of CPUs. See init/Kconfig 3728 for more details. 3729 3730 logo.nologo [FB] Disables display of the built-in Linux logo. 3731 This may be used to provide more screen space for 3732 kernel log messages and is useful when debugging 3733 kernel boot problems. 3734 3735 lp=0 [LP] Specify parallel ports to use, e.g, 3736 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses 3737 lp=reset first parallel port). 'lp=0' disables the 3738 lp=auto printer driver. 'lp=reset' (which can be 3739 specified in addition to the ports) causes 3740 attached printers to be reset. Using 3741 lp=port1,port2,... specifies the parallel ports 3742 to associate lp devices with, starting with 3743 lp0. A port specification may be 'none' to skip 3744 that lp device, or a parport name such as 3745 'parport0'. Specifying 'lp=auto' instead of a 3746 port specification list means that device IDs 3747 from each port should be examined, to see if 3748 an IEEE 1284-compliant printer is attached; if 3749 so, the driver will manage that printer. 3750 See also header of drivers/char/lp.c. 3751 3752 lpj=n [KNL] 3753 Sets loops_per_jiffy to given constant, thus avoiding 3754 time-consuming boot-time autodetection (up to 250 ms per 3755 CPU). 0 enables autodetection (default). To determine 3756 the correct value for your kernel, boot with normal 3757 autodetection and see what value is printed. Note that 3758 on SMP systems the preset will be applied to all CPUs, 3759 which is likely to cause problems if your CPUs need 3760 significantly divergent settings. An incorrect value 3761 will cause delays in the kernel to be wrong, leading to 3762 unpredictable I/O errors and other breakage. Although 3763 unlikely, in the extreme case this might damage your 3764 hardware. 3765 3766 lsm.debug [SECURITY] Enable LSM initialization debugging output. 3767 3768 lsm=lsm1,...,lsmN 3769 [SECURITY] Choose order of LSM initialization. This 3770 overrides CONFIG_LSM, and the "security=" parameter. 3771 3772 machtype= [Loongson] Share the same kernel image file between 3773 different yeeloong laptops. 3774 Example: machtype=lemote-yeeloong-2f-7inch 3775 3776 maxcpus= [SMP,EARLY] Maximum number of processors that an SMP kernel 3777 will bring up during bootup. maxcpus=n : n >= 0 limits 3778 the kernel to bring up 'n' processors. Surely after 3779 bootup you can bring up the other plugged cpu by executing 3780 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus 3781 only takes effect during system bootup. 3782 While n=0 is a special case, it is equivalent to "nosmp", 3783 which also disables the IO APIC. 3784 3785 max_loop= [LOOP] The number of loop block devices that get 3786 (loop.max_loop) unconditionally pre-created at init time. The default 3787 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead 3788 of statically allocating a predefined number, loop 3789 devices can be requested on-demand with the 3790 /dev/loop-control interface. 3791 3792 mce= [X86-{32,64}] 3793 3794 Please see Documentation/arch/x86/x86_64/machinecheck.rst for sysfs runtime tunables. 3795 3796 off 3797 disable machine check 3798 3799 no_cmci 3800 disable CMCI(Corrected Machine Check Interrupt) that 3801 Intel processor supports. Usually this disablement is 3802 not recommended, but it might be handy if your 3803 hardware is misbehaving. 3804 3805 Note that you'll get more problems without CMCI than 3806 with due to the shared banks, i.e. you might get 3807 duplicated error logs. 3808 3809 dont_log_ce 3810 don't make logs for corrected errors. All events 3811 reported as corrected are silently cleared by OS. This 3812 option will be useful if you have no interest in any 3813 of corrected errors. 3814 3815 ignore_ce 3816 disable features for corrected errors, e.g. 3817 polling timer and CMCI. All events reported as 3818 corrected are not cleared by OS and remained in its 3819 error banks. 3820 3821 Usually this disablement is not recommended, however 3822 if there is an agent checking/clearing corrected 3823 errors (e.g. BIOS or hardware monitoring 3824 applications), conflicting with OS's error handling, 3825 and you cannot deactivate the agent, then this option 3826 will be a help. 3827 3828 no_lmce 3829 do not opt-in to Local MCE delivery. Use legacy method 3830 to broadcast MCEs. 3831 3832 bootlog 3833 enable logging of machine checks left over from 3834 booting. Disabled by default on AMD Fam10h and older 3835 because some BIOS leave bogus ones. 3836 3837 If your BIOS doesn't do that it's a good idea to 3838 enable though to make sure you log even machine check 3839 events that result in a reboot. On Intel systems it is 3840 enabled by default. 3841 3842 nobootlog 3843 disable boot machine check logging. 3844 3845 print_all 3846 print all machine check logs to the console. 3847 3848 monarchtimeout (number) 3849 sets the time in us to wait for other CPUs on machine 3850 checks. 0 to disable. 3851 3852 bios_cmci_threshold 3853 don't overwrite the bios-set CMCI threshold. This boot 3854 option prevents Linux from overwriting the CMCI 3855 threshold set by the bios. Without this option, Linux 3856 always sets the CMCI threshold to 1. Enabling this may 3857 make memory predictive failure analysis less effective 3858 if the bios sets thresholds for memory errors since we 3859 will not see details for all errors. 3860 3861 recovery 3862 force-enable recoverable machine check code paths 3863 3864 Everything else is in sysfs now. 3865 3866 3867 md= [HW] RAID subsystems devices and level 3868 See Documentation/admin-guide/md.rst. 3869 3870 mds= [X86,INTEL,EARLY] 3871 Control mitigation for the Micro-architectural Data 3872 Sampling (MDS) vulnerability. 3873 3874 Certain CPUs are vulnerable to an exploit against CPU 3875 internal buffers which can forward information to a 3876 disclosure gadget under certain conditions. 3877 3878 In vulnerable processors, the speculatively 3879 forwarded data can be used in a cache side channel 3880 attack, to access data to which the attacker does 3881 not have direct access. 3882 3883 This parameter controls the MDS mitigation. The 3884 options are: 3885 3886 full - Enable MDS mitigation on vulnerable CPUs 3887 full,nosmt - Enable MDS mitigation and disable 3888 SMT on vulnerable CPUs 3889 off - Unconditionally disable MDS mitigation 3890 3891 On TAA-affected machines, mds=off can be prevented by 3892 an active TAA mitigation as both vulnerabilities are 3893 mitigated with the same mechanism so in order to disable 3894 this mitigation, you need to specify tsx_async_abort=off 3895 too. 3896 3897 Not specifying this option is equivalent to 3898 mds=full. 3899 3900 For details see: Documentation/admin-guide/hw-vuln/mds.rst 3901 3902 mem=nn[KMG] [HEXAGON,EARLY] Set the memory size. 3903 Must be specified, otherwise memory size will be 0. 3904 3905 mem=nn[KMG] [KNL,BOOT,EARLY] Force usage of a specific amount 3906 of memory Amount of memory to be used in cases 3907 as follows: 3908 3909 1 for test; 3910 2 when the kernel is not able to see the whole system memory; 3911 3 memory that lies after 'mem=' boundary is excluded from 3912 the hypervisor, then assigned to KVM guests. 3913 4 to limit the memory available for kdump kernel. 3914 3915 [ARC,MICROBLAZE] - the limit applies only to low memory, 3916 high memory is not affected. 3917 3918 [ARM64] - only limits memory covered by the linear 3919 mapping. The NOMAP regions are not affected. 3920 3921 [X86] Work as limiting max address. Use together 3922 with memmap= to avoid physical address space collisions. 3923 Without memmap= PCI devices could be placed at addresses 3924 belonging to unused RAM. 3925 3926 Note that this only takes effects during boot time since 3927 in above case 3, memory may need be hot added after boot 3928 if system memory of hypervisor is not sufficient. 3929 3930 mem=nn[KMG]@ss[KMG] 3931 [ARM,MIPS,EARLY] - override the memory layout 3932 reported by firmware. 3933 Define a memory region of size nn[KMG] starting at 3934 ss[KMG]. 3935 Multiple different regions can be specified with 3936 multiple mem= parameters on the command line. 3937 3938 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel 3939 memory. 3940 3941 memblock=debug [KNL,EARLY] Enable memblock debug messages. 3942 3943 memchunk=nn[KMG] 3944 [KNL,SH] Allow user to override the default size for 3945 per-device physically contiguous DMA buffers. 3946 3947 memhp_default_state=online/offline/online_kernel/online_movable 3948 [KNL] Set the initial state for the memory hotplug 3949 onlining policy. If not specified, the default value is 3950 set according to the 3951 CONFIG_MHP_DEFAULT_ONLINE_TYPE kernel config 3952 options. 3953 See Documentation/admin-guide/mm/memory-hotplug.rst. 3954 3955 memmap=exactmap [KNL,X86,EARLY] Enable setting of an exact 3956 E820 memory map, as specified by the user. 3957 Such memmap=exactmap lines can be constructed based on 3958 BIOS output or other requirements. See the memmap=nn@ss 3959 option description. 3960 3961 memmap=nn[KMG]@ss[KMG] 3962 [KNL, X86,MIPS,XTENSA,EARLY] Force usage of a specific region of memory. 3963 Region of memory to be used is from ss to ss+nn. 3964 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG], 3965 which limits max address to nn[KMG]. 3966 Multiple different regions can be specified, 3967 comma delimited. 3968 Example: 3969 memmap=100M@2G,100M#3G,1G!1024G 3970 3971 memmap=nn[KMG]#ss[KMG] 3972 [KNL,ACPI,EARLY] Mark specific memory as ACPI data. 3973 Region of memory to be marked is from ss to ss+nn. 3974 3975 memmap=nn[KMG]$ss[KMG] 3976 [KNL,ACPI,EARLY] Mark specific memory as reserved. 3977 Region of memory to be reserved is from ss to ss+nn. 3978 Example: Exclude memory from 0x18690000-0x1869ffff 3979 memmap=64K$0x18690000 3980 or 3981 memmap=0x10000$0x18690000 3982 Some bootloaders may need an escape character before '$', 3983 like Grub2, otherwise '$' and the following number 3984 will be eaten. 3985 3986 memmap=nn[KMG]!ss[KMG] 3987 [KNL,X86,EARLY] Mark specific memory as protected. 3988 Region of memory to be used, from ss to ss+nn. 3989 The memory region may be marked as e820 type 12 (0xc) 3990 and is NVDIMM or ADR memory. 3991 3992 memmap=<size>%<offset>-<oldtype>+<newtype> 3993 [KNL,ACPI,EARLY] Convert memory within the specified region 3994 from <oldtype> to <newtype>. If "-<oldtype>" is left 3995 out, the whole region will be marked as <newtype>, 3996 even if previously unavailable. If "+<newtype>" is left 3997 out, matching memory will be removed. Types are 3998 specified as e820 types, e.g., 1 = RAM, 2 = reserved, 3999 3 = ACPI, 12 = PRAM. 4000 4001 memory_corruption_check=0/1 [X86,EARLY] 4002 Some BIOSes seem to corrupt the first 64k of 4003 memory when doing things like suspend/resume. 4004 Setting this option will scan the memory 4005 looking for corruption. Enabling this will 4006 both detect corruption and prevent the kernel 4007 from using the memory being corrupted. 4008 However, it's intended as a diagnostic tool; if 4009 repeatable BIOS-originated corruption always 4010 affects the same memory, you can use memmap= 4011 to prevent the kernel from using that memory. 4012 4013 memory_corruption_check_size=size [X86,EARLY] 4014 By default it checks for corruption in the low 4015 64k, making this memory unavailable for normal 4016 use. Use this parameter to scan for 4017 corruption in more or less memory. 4018 4019 memory_corruption_check_period=seconds [X86,EARLY] 4020 By default it checks for corruption every 60 4021 seconds. Use this parameter to check at some 4022 other rate. 0 disables periodic checking. 4023 4024 memory_hotplug.memmap_on_memory 4025 [KNL,X86,ARM] Boolean flag to enable this feature. 4026 Format: {on | off (default)} 4027 When enabled, runtime hotplugged memory will 4028 allocate its internal metadata (struct pages, 4029 those vmemmap pages cannot be optimized even 4030 if hugetlb_free_vmemmap is enabled) from the 4031 hotadded memory which will allow to hotadd a 4032 lot of memory without requiring additional 4033 memory to do so. 4034 This feature is disabled by default because it 4035 has some implication on large (e.g. GB) 4036 allocations in some configurations (e.g. small 4037 memory blocks). 4038 The state of the flag can be read in 4039 /sys/module/memory_hotplug/parameters/memmap_on_memory. 4040 Note that even when enabled, there are a few cases where 4041 the feature is not effective. 4042 4043 mempool_debug [MM] 4044 Enable mempool debugging. This enables element 4045 poison checking when freeing elements back to the 4046 pool. Useful for debugging mempool corruption. 4047 4048 memtest= [KNL,ALPHA,X86,ARM,M68K,PPC,RISCV,EARLY] Enable memtest 4049 Format: <integer> 4050 default : 0 <disable> 4051 Specifies the number of memtest passes to be 4052 performed. Each pass selects another test 4053 pattern from a given set of patterns. Memtest 4054 fills the memory with this pattern, validates 4055 memory contents and reserves bad memory 4056 regions that are detected. 4057 4058 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control 4059 Valid arguments: on, off 4060 Default: off 4061 mem_encrypt=on: Activate SME 4062 mem_encrypt=off: Do not activate SME 4063 4064 Refer to Documentation/virt/kvm/x86/amd-memory-encryption.rst 4065 for details on when memory encryption can be activated. 4066 4067 mem_sleep_default= [SUSPEND] Default system suspend mode: 4068 s2idle - Suspend-To-Idle 4069 shallow - Power-On Suspend or equivalent (if supported) 4070 deep - Suspend-To-RAM or equivalent (if supported) 4071 See Documentation/admin-guide/pm/sleep-states.rst. 4072 4073 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when 4074 the BIOS has incorrectly applied a workaround. TinyBIOS 4075 version 0.98 is known to be affected, 0.99 fixes the 4076 problem by letting the user disable the workaround. 4077 4078 mga= [HW,DRM] 4079 4080 microcode= [X86] Control the behavior of the microcode loader. 4081 Available options, comma separated: 4082 4083 base_rev=X - with <X> with format: <u32> 4084 Set the base microcode revision of each thread when in 4085 debug mode. 4086 4087 dis_ucode_ldr: disable the microcode loader 4088 4089 force_minrev: 4090 Enable or disable the microcode minimal revision 4091 enforcement for the runtime microcode loader. 4092 4093 mitigations= 4094 [X86,PPC,S390,ARM64,EARLY] Control optional mitigations for 4095 CPU vulnerabilities. This is a set of curated, 4096 arch-independent options, each of which is an 4097 aggregation of existing arch-specific options. 4098 4099 Note, "mitigations" is supported if and only if the 4100 kernel was built with CPU_MITIGATIONS=y. 4101 4102 off 4103 Disable all optional CPU mitigations. This 4104 improves system performance, but it may also 4105 expose users to several CPU vulnerabilities. 4106 Equivalent to: if nokaslr then kpti=0 [ARM64] 4107 gather_data_sampling=off [X86] 4108 indirect_target_selection=off [X86] 4109 kvm.nx_huge_pages=off [X86] 4110 l1tf=off [X86] 4111 mds=off [X86] 4112 mmio_stale_data=off [X86] 4113 no_entry_flush [PPC] 4114 no_uaccess_flush [PPC] 4115 nobp=0 [S390] 4116 nopti [X86,PPC] 4117 nospectre_bhb [ARM64] 4118 nospectre_v1 [X86,PPC] 4119 nospectre_v2 [X86,PPC,S390,ARM64] 4120 reg_file_data_sampling=off [X86] 4121 retbleed=off [X86] 4122 spec_rstack_overflow=off [X86] 4123 spec_store_bypass_disable=off [X86,PPC] 4124 spectre_bhi=off [X86] 4125 spectre_v2_user=off [X86] 4126 srbds=off [X86,INTEL] 4127 ssbd=force-off [ARM64] 4128 tsa=off [X86,AMD] 4129 tsx_async_abort=off [X86] 4130 vmscape=off [X86] 4131 4132 Exceptions: 4133 This does not have any effect on 4134 kvm.nx_huge_pages when 4135 kvm.nx_huge_pages=force. 4136 4137 auto (default) 4138 Mitigate all CPU vulnerabilities, but leave SMT 4139 enabled, even if it's vulnerable. This is for 4140 users who don't want to be surprised by SMT 4141 getting disabled across kernel upgrades, or who 4142 have other ways of avoiding SMT-based attacks. 4143 Equivalent to: (default behavior) 4144 4145 auto,nosmt 4146 Mitigate all CPU vulnerabilities, disabling SMT 4147 if needed. This is for users who always want to 4148 be fully mitigated, even if it means losing SMT. 4149 Equivalent to: l1tf=flush,nosmt [X86] 4150 mds=full,nosmt [X86] 4151 tsx_async_abort=full,nosmt [X86] 4152 mmio_stale_data=full,nosmt [X86] 4153 retbleed=auto,nosmt [X86] 4154 4155 [X86] After one of the above options, additionally 4156 supports attack-vector based controls as documented in 4157 Documentation/admin-guide/hw-vuln/attack_vector_controls.rst 4158 4159 mminit_loglevel= 4160 [KNL,EARLY] When CONFIG_DEBUG_MEMORY_INIT is set, this 4161 parameter allows control of the logging verbosity for 4162 the additional memory initialisation checks. A value 4163 of 0 disables mminit logging and a level of 4 will 4164 log everything. Information is printed at KERN_DEBUG 4165 so loglevel=8 may also need to be specified. 4166 4167 mmio_stale_data= 4168 [X86,INTEL,EARLY] Control mitigation for the Processor 4169 MMIO Stale Data vulnerabilities. 4170 4171 Processor MMIO Stale Data is a class of 4172 vulnerabilities that may expose data after an MMIO 4173 operation. Exposed data could originate or end in 4174 the same CPU buffers as affected by MDS and TAA. 4175 Therefore, similar to MDS and TAA, the mitigation 4176 is to clear the affected CPU buffers. 4177 4178 This parameter controls the mitigation. The 4179 options are: 4180 4181 full - Enable mitigation on vulnerable CPUs 4182 4183 full,nosmt - Enable mitigation and disable SMT on 4184 vulnerable CPUs. 4185 4186 off - Unconditionally disable mitigation 4187 4188 On MDS or TAA affected machines, 4189 mmio_stale_data=off can be prevented by an active 4190 MDS or TAA mitigation as these vulnerabilities are 4191 mitigated with the same mechanism so in order to 4192 disable this mitigation, you need to specify 4193 mds=off and tsx_async_abort=off too. 4194 4195 Not specifying this option is equivalent to 4196 mmio_stale_data=full. 4197 4198 For details see: 4199 Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst 4200 4201 <module>.async_probe[=<bool>] [KNL] 4202 If no <bool> value is specified or if the value 4203 specified is not a valid <bool>, enable asynchronous 4204 probe on this module. Otherwise, enable/disable 4205 asynchronous probe on this module as indicated by the 4206 <bool> value. See also: module.async_probe 4207 4208 module.async_probe=<bool> 4209 [KNL] When set to true, modules will use async probing 4210 by default. To enable/disable async probing for a 4211 specific module, use the module specific control that 4212 is documented under <module>.async_probe. When both 4213 module.async_probe and <module>.async_probe are 4214 specified, <module>.async_probe takes precedence for 4215 the specific module. 4216 4217 module.enable_dups_trace 4218 [KNL] When CONFIG_MODULE_DEBUG_AUTOLOAD_DUPS is set, 4219 this means that duplicate request_module() calls will 4220 trigger a WARN_ON() instead of a pr_warn(). Note that 4221 if MODULE_DEBUG_AUTOLOAD_DUPS_TRACE is set, WARN_ON()s 4222 will always be issued and this option does nothing. 4223 module.sig_enforce 4224 [KNL] When CONFIG_MODULE_SIG is set, this means that 4225 modules without (valid) signatures will fail to load. 4226 Note that if CONFIG_MODULE_SIG_FORCE is set, that 4227 is always true, so this option does nothing. 4228 4229 module_blacklist= [KNL] Do not load a comma-separated list of 4230 modules. Useful for debugging problem modules. 4231 4232 mousedev.tap_time= 4233 [MOUSE] Maximum time between finger touching and 4234 leaving touchpad surface for touch to be considered 4235 a tap and be reported as a left button click (for 4236 touchpads working in absolute mode only). 4237 Format: <msecs> 4238 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices 4239 reporting absolute coordinates, such as tablets 4240 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices 4241 reporting absolute coordinates, such as tablets 4242 4243 movablecore= [KNL,X86,PPC,EARLY] 4244 Format: nn[KMGTPE] | nn% 4245 This parameter is the complement to kernelcore=, it 4246 specifies the amount of memory used for migratable 4247 allocations. If both kernelcore and movablecore is 4248 specified, then kernelcore will be at *least* the 4249 specified value but may be more. If movablecore on its 4250 own is specified, the administrator must be careful 4251 that the amount of memory usable for all allocations 4252 is not too small. 4253 4254 movable_node [KNL,EARLY] Boot-time switch to make hotplugable memory 4255 NUMA nodes to be movable. This means that the memory 4256 of such nodes will be usable only for movable 4257 allocations which rules out almost all kernel 4258 allocations. Use with caution! 4259 4260 MTD_Partition= [MTD] 4261 Format: <name>,<region-number>,<size>,<offset> 4262 4263 MTD_Region= [MTD] Format: 4264 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>] 4265 4266 mtdparts= [MTD] 4267 See drivers/mtd/parsers/cmdlinepart.c 4268 4269 mtouchusb.raw_coordinates= 4270 [HW] Make the MicroTouch USB driver use raw coordinates 4271 ('y', default) or cooked coordinates ('n') 4272 4273 mtrr=debug [X86,EARLY] 4274 Enable printing debug information related to MTRR 4275 registers at boot time. 4276 4277 mtrr_chunk_size=nn[KMG] [X86,EARLY] 4278 used for mtrr cleanup. It is largest continuous chunk 4279 that could hold holes aka. UC entries. 4280 4281 mtrr_gran_size=nn[KMG] [X86,EARLY] 4282 Used for mtrr cleanup. It is granularity of mtrr block. 4283 Default is 1. 4284 Large value could prevent small alignment from 4285 using up MTRRs. 4286 4287 mtrr_spare_reg_nr=n [X86,EARLY] 4288 Format: <integer> 4289 Range: 0,7 : spare reg number 4290 Default : 1 4291 Used for mtrr cleanup. It is spare mtrr entries number. 4292 Set to 2 or more if your graphical card needs more. 4293 4294 multitce=off [PPC] This parameter disables the use of the pSeries 4295 firmware feature for updating multiple TCE entries 4296 at a time. 4297 4298 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card 4299 4300 netdev= [NET] NE2000 ISA network devices parameters 4301 Format: <irq>,<io>,<mem_start>,<mem_end>,<name> 4302 4303 netpoll.carrier_timeout= 4304 [NET] Specifies amount of time (in seconds) that 4305 netpoll should wait for a carrier. By default netpoll 4306 waits 4 seconds. 4307 4308 nf_conntrack.acct= 4309 [NETFILTER] Enable connection tracking flow accounting 4310 0 to disable accounting 4311 1 to enable accounting 4312 Default value is 0. 4313 4314 nfs.cache_getent= 4315 [NFS] sets the pathname to the program which is used 4316 to update the NFS client cache entries. 4317 4318 nfs.cache_getent_timeout= 4319 [NFS] sets the timeout after which an attempt to 4320 update a cache entry is deemed to have failed. 4321 4322 nfs.callback_nr_threads= 4323 [NFSv4] set the total number of threads that the 4324 NFS client will assign to service NFSv4 callback 4325 requests. 4326 4327 nfs.callback_tcpport= 4328 [NFS] set the TCP port on which the NFSv4 callback 4329 channel should listen. 4330 4331 nfs.delay_retrans= 4332 [NFS] specifies the number of times the NFSv4 client 4333 retries the request before returning an EAGAIN error, 4334 after a reply of NFS4ERR_DELAY from the server. 4335 Only applies if the softerr mount option is enabled, 4336 and the specified value is >= 0. 4337 4338 nfs.idmap_cache_timeout= 4339 [NFS] set the maximum lifetime for idmapper cache 4340 entries. 4341 4342 nfs.max_session_cb_slots= 4343 [NFSv4.1] Sets the maximum number of session 4344 slots the client will assign to the callback 4345 channel. This determines the maximum number of 4346 callbacks the client will process in parallel for 4347 a particular server. 4348 4349 nfs.max_session_slots= 4350 [NFSv4.1] Sets the maximum number of session slots 4351 the client will attempt to negotiate with the server. 4352 This limits the number of simultaneous RPC requests 4353 that the client can send to the NFSv4.1 server. 4354 Note that there is little point in setting this 4355 value higher than the max_tcp_slot_table_limit. 4356 4357 nfs.nfs4_disable_idmapping= 4358 [NFSv4] When set to the default of '1', this option 4359 ensures that both the RPC level authentication 4360 scheme and the NFS level operations agree to use 4361 numeric uids/gids if the mount is using the 4362 'sec=sys' security flavour. In effect it is 4363 disabling idmapping, which can make migration from 4364 legacy NFSv2/v3 systems to NFSv4 easier. 4365 Servers that do not support this mode of operation 4366 will be autodetected by the client, and it will fall 4367 back to using the idmapper. 4368 To turn off this behaviour, set the value to '0'. 4369 4370 nfs.nfs4_unique_id= 4371 [NFS4] Specify an additional fixed unique ident- 4372 ification string that NFSv4 clients can insert into 4373 their nfs_client_id4 string. This is typically a 4374 UUID that is generated at system install time. 4375 4376 nfs.recover_lost_locks= 4377 [NFSv4] Attempt to recover locks that were lost due 4378 to a lease timeout on the server. Please note that 4379 doing this risks data corruption, since there are 4380 no guarantees that the file will remain unchanged 4381 after the locks are lost. 4382 If you want to enable the kernel legacy behaviour of 4383 attempting to recover these locks, then set this 4384 parameter to '1'. 4385 The default parameter value of '0' causes the kernel 4386 not to attempt recovery of lost locks. 4387 4388 nfs.send_implementation_id= 4389 [NFSv4.1] Send client implementation identification 4390 information in exchange_id requests. 4391 If zero, no implementation identification information 4392 will be sent. 4393 The default is to send the implementation identification 4394 information. 4395 4396 nfs4.layoutstats_timer= 4397 [NFSv4.2] Change the rate at which the kernel sends 4398 layoutstats to the pNFS metadata server. 4399 4400 Setting this to value to 0 causes the kernel to use 4401 whatever value is the default set by the layout 4402 driver. A non-zero value sets the minimum interval 4403 in seconds between layoutstats transmissions. 4404 4405 nfsd.inter_copy_offload_enable= 4406 [NFSv4.2] When set to 1, the server will support 4407 server-to-server copies for which this server is 4408 the destination of the copy. 4409 4410 nfsd.nfs4_disable_idmapping= 4411 [NFSv4] When set to the default of '1', the NFSv4 4412 server will return only numeric uids and gids to 4413 clients using auth_sys, and will accept numeric uids 4414 and gids from such clients. This is intended to ease 4415 migration from NFSv2/v3. 4416 4417 nfsd.nfsd4_ssc_umount_timeout= 4418 [NFSv4.2] When used as the destination of a 4419 server-to-server copy, knfsd temporarily mounts 4420 the source server. It caches the mount in case 4421 it will be needed again, and discards it if not 4422 used for the number of milliseconds specified by 4423 this parameter. 4424 4425 nfsaddrs= [NFS] Deprecated. Use ip= instead. 4426 See Documentation/admin-guide/nfs/nfsroot.rst. 4427 4428 nfsroot= [NFS] nfs root filesystem for disk-less boxes. 4429 See Documentation/admin-guide/nfs/nfsroot.rst. 4430 4431 nfsrootdebug [NFS] enable nfsroot debugging messages. 4432 See Documentation/admin-guide/nfs/nfsroot.rst. 4433 4434 nmi_backtrace.backtrace_idle [KNL] 4435 Dump stacks even of idle CPUs in response to an 4436 NMI stack-backtrace request. 4437 4438 nmi_debug= [KNL,SH] Specify one or more actions to take 4439 when a NMI is triggered. 4440 Format: [state][,regs][,debounce][,die] 4441 4442 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels 4443 Format: [panic,][nopanic,][rNNN,][num] 4444 Valid num: 0 or 1 4445 0 - turn hardlockup detector in nmi_watchdog off 4446 1 - turn hardlockup detector in nmi_watchdog on 4447 rNNN - configure the watchdog with raw perf event 0xNNN 4448 4449 When panic is specified, panic when an NMI watchdog 4450 timeout occurs (or 'nopanic' to not panic on an NMI 4451 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set) 4452 To disable both hard and soft lockup detectors, 4453 please see 'nowatchdog'. 4454 This is useful when you use a panic=... timeout and 4455 need the box quickly up again. 4456 4457 These settings can be accessed at runtime via 4458 the nmi_watchdog and hardlockup_panic sysctls. 4459 4460 no4lvl [RISCV,EARLY] Disable 4-level and 5-level paging modes. 4461 Forces kernel to use 3-level paging instead. 4462 4463 no5lvl [X86-64,RISCV,EARLY] Disable 5-level paging mode. Forces 4464 kernel to use 4-level paging instead. 4465 4466 noalign [KNL,ARM] 4467 4468 noapic [SMP,APIC,EARLY] Tells the kernel to not make use of any 4469 IOAPICs that may be present in the system. 4470 4471 noapictimer [APIC,X86] Don't set up the APIC timer 4472 4473 noautogroup Disable scheduler automatic task group creation. 4474 4475 nocache [ARM,EARLY] 4476 4477 no_console_suspend 4478 [HW] Never suspend the console 4479 Disable suspending of consoles during suspend and 4480 hibernate operations. Once disabled, debugging 4481 messages can reach various consoles while the rest 4482 of the system is being put to sleep (ie, while 4483 debugging driver suspend/resume hooks). This may 4484 not work reliably with all consoles, but is known 4485 to work with serial and VGA consoles. 4486 To facilitate more flexible debugging, we also add 4487 console_suspend, a printk module parameter to control 4488 it. Users could use console_suspend (usually 4489 /sys/module/printk/parameters/console_suspend) to 4490 turn on/off it dynamically. 4491 4492 no_debug_objects 4493 [KNL,EARLY] Disable object debugging 4494 4495 nodsp [SH] Disable hardware DSP at boot time. 4496 4497 noefi [EFI,EARLY] Disable EFI runtime services support. 4498 4499 no_entry_flush [PPC,EARLY] Don't flush the L1-D cache when entering the kernel. 4500 4501 noexec32 [X86-64] 4502 This affects only 32-bit executables. 4503 noexec32=on: enable non-executable mappings (default) 4504 read doesn't imply executable mappings 4505 noexec32=off: disable non-executable mappings 4506 read implies executable mappings 4507 4508 no_file_caps Tells the kernel not to honor file capabilities. The 4509 only way then for a file to be executed with privilege 4510 is to be setuid root or executed by root. 4511 4512 nofpu [MIPS,SH] Disable hardware FPU at boot time. 4513 4514 nofsgsbase [X86] Disables FSGSBASE instructions. 4515 4516 nofxsr [BUGS=X86-32] Disables x86 floating point extended 4517 register save and restore. The kernel will only save 4518 legacy floating-point registers on task switch. 4519 4520 nogbpages [X86] Do not use GB pages for kernel direct mappings. 4521 4522 no_hash_pointers 4523 [KNL,EARLY] 4524 Alias for "hash_pointers=never". 4525 4526 nohibernate [HIBERNATION] Disable hibernation and resume. 4527 4528 nohlt [ARM,ARM64,MICROBLAZE,MIPS,PPC,RISCV,SH] Forces the kernel to 4529 busy wait in do_idle() and not use the arch_cpu_idle() 4530 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP 4531 to be effective. This is useful on platforms where the 4532 sleep(SH) or wfi(ARM,ARM64) instructions do not work 4533 correctly or when doing power measurements to evaluate 4534 the impact of the sleep instructions. This is also 4535 useful when using JTAG debugger. 4536 4537 nohpet [X86] Don't use the HPET timer. 4538 4539 nohugeiomap [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge I/O mappings. 4540 4541 nohugevmalloc [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge vmalloc mappings. 4542 4543 nohz= [KNL] Boottime enable/disable dynamic ticks 4544 Valid arguments: on, off 4545 Default: on 4546 4547 nohz_full= [KNL,BOOT,SMP,ISOL] 4548 The argument is a cpu list, as described above. 4549 In kernels built with CONFIG_NO_HZ_FULL=y, set 4550 the specified list of CPUs whose tick will be stopped 4551 whenever possible. The boot CPU will be forced outside 4552 the range to maintain the timekeeping. Any CPUs 4553 in this list will have their RCU callbacks offloaded, 4554 just as if they had also been called out in the 4555 rcu_nocbs= boot parameter. 4556 4557 Note that this argument takes precedence over 4558 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option. 4559 4560 noinitrd [Deprecated,RAM] Tells the kernel not to load any configured 4561 initial RAM disk. Currently this parameter applies to 4562 initrd only, not to initramfs. But it applies to both 4563 in EFI mode. 4564 4565 nointremap [X86-64,Intel-IOMMU,EARLY] Do not enable interrupt 4566 remapping. 4567 [Deprecated - use intremap=off] 4568 4569 noinvpcid [X86,EARLY] Disable the INVPCID cpu feature. 4570 4571 noiotrap [SH] Disables trapped I/O port accesses. 4572 4573 noirqdebug [X86-32] Disables the code which attempts to detect and 4574 disable unhandled interrupt sources. 4575 4576 noisapnp [ISAPNP] Disables ISA PnP code. 4577 4578 nokaslr [KNL,EARLY] 4579 When CONFIG_RANDOMIZE_BASE is set, this disables 4580 kernel and module base offset ASLR (Address Space 4581 Layout Randomization). 4582 4583 no-kvmapf [X86,KVM,EARLY] Disable paravirtualized asynchronous page 4584 fault handling. 4585 4586 no-kvmclock [X86,KVM,EARLY] Disable paravirtualized KVM clock driver 4587 4588 nolapic [X86-32,APIC,EARLY] Do not enable or use the local APIC. 4589 4590 nolapic_timer [X86-32,APIC,EARLY] Do not use the local APIC timer. 4591 4592 nomce [X86-32] Disable Machine Check Exception 4593 4594 nomodeset Disable kernel modesetting. Most systems' firmware 4595 sets up a display mode and provides framebuffer memory 4596 for output. With nomodeset, DRM and fbdev drivers will 4597 not load if they could possibly displace the pre- 4598 initialized output. Only the system framebuffer will 4599 be available for use. The respective drivers will not 4600 perform display-mode changes or accelerated rendering. 4601 4602 Useful as error fallback, or for testing and debugging. 4603 4604 nomodule Disable module load 4605 4606 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to 4607 shutdown the other cpus. Instead use the REBOOT_VECTOR 4608 irq. 4609 4610 nopat [X86,EARLY] Disable PAT (page attribute table extension of 4611 pagetables) support. 4612 4613 nopcid [X86-64,EARLY] Disable the PCID cpu feature. 4614 4615 nopku [X86] Disable Memory Protection Keys CPU feature found 4616 in some Intel CPUs. 4617 4618 nopti [X86-64,EARLY] 4619 Equivalent to pti=off 4620 4621 nopv= [X86,XEN,KVM,HYPER_V,VMWARE,EARLY] 4622 Disables the PV optimizations forcing the guest to run 4623 as generic guest with no PV drivers. Currently support 4624 XEN HVM, KVM, HYPER_V and VMWARE guest. 4625 4626 nopvspin [X86,XEN,KVM,EARLY] 4627 Disables the qspinlock slow path using PV optimizations 4628 which allow the hypervisor to 'idle' the guest on lock 4629 contention. 4630 4631 norandmaps Don't use address space randomization. Equivalent to 4632 echo 0 > /proc/sys/kernel/randomize_va_space 4633 4634 noresume [SWSUSP] Disables resume and restores original swap 4635 space. 4636 4637 no-scroll [VGA] Disables scrollback. 4638 This is required for the Braillex ib80-piezo Braille 4639 reader made by F.H. Papenmeier (Germany). 4640 4641 nosgx [X86-64,SGX,EARLY] Disables Intel SGX kernel support. 4642 4643 nosmap [PPC,EARLY] 4644 Disable SMAP (Supervisor Mode Access Prevention) 4645 even if it is supported by processor. 4646 4647 nosmep [PPC64s,EARLY] 4648 Disable SMEP (Supervisor Mode Execution Prevention) 4649 even if it is supported by processor. 4650 4651 nosmp [SMP,EARLY] Tells an SMP kernel to act as a UP kernel, 4652 and disable the IO APIC. legacy for "maxcpus=0". 4653 4654 nosmt [KNL,MIPS,PPC,EARLY] Disable symmetric multithreading (SMT). 4655 Equivalent to smt=1. 4656 4657 [KNL,LOONGARCH,X86,ARM64,PPC,S390] Disable symmetric multithreading (SMT). 4658 nosmt=force: Force disable SMT, cannot be undone 4659 via the sysfs control file. 4660 4661 nosoftlockup [KNL] Disable the soft-lockup detector. 4662 4663 nospec_store_bypass_disable 4664 [HW,EARLY] Disable all mitigations for the Speculative 4665 Store Bypass vulnerability 4666 4667 nospectre_bhb [ARM64,EARLY] Disable all mitigations for Spectre-BHB (branch 4668 history injection) vulnerability. System may allow data leaks 4669 with this option. 4670 4671 nospectre_v1 [X86,PPC,EARLY] Disable mitigations for Spectre Variant 1 4672 (bounds check bypass). With this option data leaks are 4673 possible in the system. 4674 4675 nospectre_v2 [X86,PPC_E500,ARM64,EARLY] Disable all mitigations 4676 for the Spectre variant 2 (indirect branch 4677 prediction) vulnerability. System may allow data 4678 leaks with this option. 4679 4680 no-steal-acc [X86,PV_OPS,ARM64,PPC/PSERIES,RISCV,LOONGARCH,EARLY] 4681 Disable paravirtualized steal time accounting. steal time 4682 is computed, but won't influence scheduler behaviour 4683 4684 nosync [HW,M68K] Disables sync negotiation for all devices. 4685 4686 no_timer_check [X86,APIC] Disables the code which tests for broken 4687 timer IRQ sources, i.e., the IO-APIC timer. This can 4688 work around problems with incorrect timer 4689 initialization on some boards. 4690 4691 no_uaccess_flush 4692 [PPC,EARLY] Don't flush the L1-D cache after accessing user data. 4693 4694 novmcoredd [KNL,KDUMP] 4695 Disable device dump. Device dump allows drivers to 4696 append dump data to vmcore so you can collect driver 4697 specified debug info. Drivers can append the data 4698 without any limit and this data is stored in memory, 4699 so this may cause significant memory stress. Disabling 4700 device dump can help save memory but the driver debug 4701 data will be no longer available. This parameter 4702 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP 4703 is set. 4704 4705 no-vmw-sched-clock 4706 [X86,PV_OPS,EARLY] Disable paravirtualized VMware 4707 scheduler clock and use the default one. 4708 4709 nowatchdog [KNL] Disable both lockup detectors, i.e. 4710 soft-lockup and NMI watchdog (hard-lockup). 4711 4712 nowb [ARM,EARLY] 4713 4714 nox2apic [X86-64,APIC,EARLY] Do not enable x2APIC mode. 4715 4716 NOTE: this parameter will be ignored on systems with the 4717 LEGACY_XAPIC_DISABLED bit set in the 4718 IA32_XAPIC_DISABLE_STATUS MSR. 4719 4720 noxsave [BUGS=X86] Disables x86 extended register state save 4721 and restore using xsave. The kernel will fallback to 4722 enabling legacy floating-point and sse state. 4723 4724 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended 4725 register states. The kernel will fall back to use 4726 xsave to save the states. By using this parameter, 4727 performance of saving the states is degraded because 4728 xsave doesn't support modified optimization while 4729 xsaveopt supports it on xsaveopt enabled systems. 4730 4731 noxsaves [X86] Disables xsaves and xrstors used in saving and 4732 restoring x86 extended register state in compacted 4733 form of xsave area. The kernel will fall back to use 4734 xsaveopt and xrstor to save and restore the states 4735 in standard form of xsave area. By using this 4736 parameter, xsave area per process might occupy more 4737 memory on xsaves enabled systems. 4738 4739 nr_cpus= [SMP,EARLY] Maximum number of processors that an SMP kernel 4740 could support. nr_cpus=n : n >= 1 limits the kernel to 4741 support 'n' processors. It could be larger than the 4742 number of already plugged CPU during bootup, later in 4743 runtime you can physically add extra cpu until it reaches 4744 n. So during boot up some boot time memory for per-cpu 4745 variables need be pre-allocated for later physical cpu 4746 hot plugging. 4747 4748 nr_uarts= [SERIAL] maximum number of UARTs to be registered. 4749 4750 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86, EARLY] 4751 Disable NUMA, Only set up a single NUMA node 4752 spanning all memory. 4753 4754 numa=fake=<size>[MG] 4755 [KNL, ARM64, RISCV, X86, EARLY] 4756 If given as a memory unit, fills all system RAM with 4757 nodes of size interleaved over physical nodes. 4758 4759 numa=fake=<N> 4760 [KNL, ARM64, RISCV, X86, EARLY] 4761 If given as an integer, fills all system RAM with N 4762 fake nodes interleaved over physical nodes. 4763 4764 numa=fake=<N>U 4765 [KNL, ARM64, RISCV, X86, EARLY] 4766 If given as an integer followed by 'U', it will 4767 divide each physical node into N emulated nodes. 4768 4769 numa=noacpi [X86] Don't parse the SRAT table for NUMA setup 4770 4771 numa=nohmat [X86] Don't parse the HMAT table for NUMA setup, or 4772 soft-reserved memory partitioning. 4773 4774 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic 4775 NUMA balancing. 4776 Allowed values are enable and disable 4777 4778 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA. 4779 'node', 'default' can be specified 4780 This can be set from sysctl after boot. 4781 See Documentation/admin-guide/sysctl/vm.rst for details. 4782 4783 nvme.quirks= [NVME] A list of quirk entries to augment the built-in 4784 nvme quirk list. List entries are separated by a 4785 '-' character. 4786 Each entry has the form VendorID:ProductID:quirk_names. 4787 The IDs are 4-digits hex numbers and quirk_names is a 4788 list of quirk names separated by commas. A quirk name 4789 can be prefixed by '^', meaning that the specified 4790 quirk must be disabled. 4791 4792 Example: 4793 nvme.quirks=7710:2267:bogus_nid,^identify_cns-9900:7711:broken_msi 4794 4795 nvme_tcp.tls_handshake_timeout= 4796 [NVME] NVMe/TCP host TLS handshake timeout in seconds. 4797 Valid values range from 0 to the smaller of 4798 UINT_MAX / 1000 and MAX_JIFFY_OFFSET / HZ. Default: 10 4799 4800 nvmet_tcp.tls_handshake_timeout= 4801 [NVME] NVMe/TCP target TLS handshake timeout in seconds. 4802 Valid values range from 0 to the smaller of 4803 UINT_MAX / 1000 and MAX_JIFFY_OFFSET / HZ. Default: 10 4804 4805 ohci1394_dma=early [HW,EARLY] enable debugging via the ohci1394 driver. 4806 See Documentation/core-api/debugging-via-ohci1394.rst for more 4807 info. 4808 4809 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands 4810 Rather than timing out after 20 ms if an EC 4811 command is not properly ACKed, override the length 4812 of the timeout. We have interrupts disabled while 4813 waiting for the ACK, so if this is set too high 4814 interrupts *may* be lost! 4815 4816 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration 4817 4818 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock] 4819 4820 boundary - index of last SLC block on Flex-OneNAND. 4821 The remaining blocks are configured as MLC blocks. 4822 lock - Configure if Flex-OneNAND boundary should be locked. 4823 Once locked, the boundary cannot be changed. 4824 1 indicates lock status, 0 indicates unlock status. 4825 4826 oops=panic [KNL,EARLY] 4827 Always panic on oopses. Default is to just kill the 4828 process, but there is a small probability of 4829 deadlocking the machine. 4830 This will also cause panics on machine check exceptions. 4831 Useful together with panic=30 to trigger a reboot. 4832 4833 page_alloc.shuffle= 4834 [KNL] Boolean flag to control whether the page allocator 4835 should randomize its free lists. This parameter can be 4836 used to enable/disable page randomization. The state of 4837 the flag can be read from sysfs at: 4838 /sys/module/page_alloc/parameters/shuffle. 4839 This parameter is only available if CONFIG_SHUFFLE_PAGE_ALLOCATOR=y. 4840 4841 page_owner= [KNL,EARLY] Boot-time page_owner enabling option. 4842 Storage of the information about who allocated 4843 each page is disabled in default. With this switch, 4844 we can turn it on. 4845 on: enable the feature 4846 4847 page_poison= [KNL,EARLY] Boot-time parameter changing the state of 4848 poisoning on the buddy allocator, available with 4849 CONFIG_PAGE_POISONING=y. 4850 off: turn off poisoning (default) 4851 on: turn on poisoning 4852 4853 page_reporting.page_reporting_order= 4854 [KNL] Minimal page reporting order 4855 Format: <integer> 4856 Adjust the minimal page reporting order. The page 4857 reporting is disabled when it exceeds MAX_PAGE_ORDER. 4858 4859 page_reporting.page_reporting_delay_ms= 4860 [KNL] Free page reporting delay in milliseconds 4861 Format: <unsigned integer> 4862 Adjust the delay in milliseconds between free page 4863 reporting intervals. Default is 2000 (2 seconds). 4864 4865 panic= [KNL] Kernel behaviour on panic: delay <timeout> 4866 timeout > 0: seconds before rebooting 4867 timeout = 0: wait forever 4868 timeout < 0: reboot immediately 4869 Format: <timeout> 4870 4871 panic_on_taint= [KNL,EARLY] 4872 Bitmask for conditionally calling panic() in add_taint() 4873 Format: <hex>[,nousertaint] 4874 Hexadecimal bitmask representing the set of TAINT flags 4875 that will cause the kernel to panic when add_taint() is 4876 called with any of the flags in this set. 4877 The optional switch "nousertaint" can be utilized to 4878 prevent userspace forced crashes by writing to sysctl 4879 /proc/sys/kernel/tainted any flagset matching with the 4880 bitmask set on panic_on_taint. 4881 See Documentation/admin-guide/tainted-kernels.rst for 4882 extra details on the taint flags that users can pick 4883 to compose the bitmask to assign to panic_on_taint. 4884 4885 panic_on_warn=1 panic() instead of WARN(). Useful to cause kdump 4886 on a WARN(). 4887 4888 panic_force_cpu= 4889 [KNL,SMP] Force panic handling to execute on a specific CPU. 4890 Format: <cpu number> 4891 Some platforms require panic handling to occur on a 4892 specific CPU for the crash kernel to function correctly. 4893 This can be due to firmware limitations, interrupt routing 4894 constraints, or platform-specific requirements where only 4895 a particular CPU can safely enter the crash kernel. 4896 When set, panic() will redirect execution to the specified 4897 CPU before proceeding with the normal panic and kexec flow. 4898 If the target CPU is offline or unavailable, panic proceeds 4899 on the current CPU. 4900 This option should only be used for systems with the above 4901 constraints as it might cause the panic operation to be less reliable. 4902 4903 panic_print= Bitmask for printing system info when panic happens. 4904 User can chose combination of the following bits: 4905 4906 bit 0: print all tasks info 4907 bit 1: print system memory info 4908 bit 2: print timer info 4909 bit 3: print locks info if CONFIG_LOCKDEP is on 4910 bit 4: print ftrace buffer 4911 bit 5: replay all kernel messages on consoles at the end of panic 4912 bit 6: print all CPUs backtrace (if available in the arch) 4913 bit 7: print only tasks in uninterruptible (blocked) state 4914 4915 *Be aware* that this option may print a _lot_ of lines. 4916 4917 There is an increased risk of losing older messages in 4918 the log. Maybe worth to setup a bigger log buffer with 4919 "log_buf_len" along with this. 4920 4921 Also consider using "printk.console_no_auto_verbose=Y" 4922 if using this along with pstore, to avoid extra delays 4923 due to increased console verbosity during panic. 4924 4925 panic_sys_info= A comma separated list of extra information to be dumped 4926 on panic. 4927 Format: val[,val...] 4928 Where @val can be any of the following: 4929 4930 tasks: print all tasks info 4931 mem: print system memory info 4932 timers: print timers info 4933 locks: print locks info if CONFIG_LOCKDEP is on 4934 ftrace: print ftrace buffer 4935 all_bt: print all CPUs backtrace (if available in the arch) 4936 blocked_tasks: print only tasks in uninterruptible (blocked) state 4937 4938 This is a human readable alternative to the 'panic_print' 4939 option above. Please be aware of the caveats using this, 4940 check the 'panic_print' parameter for more details. 4941 4942 panic_console_replay 4943 When panic happens, replay all kernel messages on 4944 consoles at the end of panic. 4945 4946 parkbd.port= [HW] Parallel port number the keyboard adapter is 4947 connected to, default is 0. 4948 Format: <parport#> 4949 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation, 4950 0 for XT, 1 for AT (default is AT). 4951 Format: <mode> 4952 4953 parport= [HW,PPT] Specify parallel ports. 0 disables. 4954 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] } 4955 Use 'auto' to force the driver to use any 4956 IRQ/DMA settings detected (the default is to 4957 ignore detected IRQ/DMA settings because of 4958 possible conflicts). You can specify the base 4959 address, IRQ, and DMA settings; IRQ and DMA 4960 should be numbers, or 'auto' (for using detected 4961 settings on that particular port), or 'nofifo' 4962 (to avoid using a FIFO even if it is detected). 4963 Parallel ports are assigned in the order they 4964 are specified on the command line, starting 4965 with parport0. 4966 4967 parport_init_mode= [HW,PPT] 4968 Configure VIA parallel port to operate in 4969 a specific mode. This is necessary on Pegasos 4970 computer where firmware has no options for setting 4971 up parallel port mode and sets it to spp. 4972 Currently this function knows 686a and 8231 chips. 4973 Format: [spp|ps2|epp|ecp|ecpepp] 4974 4975 pata_legacy.all= [HW,LIBATA] 4976 Format: <int> 4977 Set to non-zero to probe primary and secondary ISA 4978 port ranges on PCI systems where no PCI PATA device 4979 has been found at either range. Disabled by default. 4980 4981 pata_legacy.autospeed= [HW,LIBATA] 4982 Format: <int> 4983 Set to non-zero if a chip is present that snoops speed 4984 changes. Disabled by default. 4985 4986 pata_legacy.iordy_mask= [HW,LIBATA] 4987 Format: <int> 4988 IORDY enable mask. Set individual bits to allow IORDY 4989 for the respective channel. Bit 0 is for the first 4990 legacy channel handled by this driver, bit 1 is for 4991 the second channel, and so on. The sequence will often 4992 correspond to the primary legacy channel, the secondary 4993 legacy channel, and so on, but the handling of a PCI 4994 bus and the use of other driver options may interfere 4995 with the sequence. By default IORDY is allowed across 4996 all channels. 4997 4998 pata_legacy.pio_mask= [HW,LIBATA] 4999 Format: <int> 5000 PIO mode mask for autospeed devices. Set individual 5001 bits to allow the use of the respective PIO modes. 5002 Bit 0 is for mode 0, bit 1 is for mode 1, and so on. 5003 All modes allowed by default. 5004 5005 pata_legacy.probe_all= [HW,LIBATA] 5006 Format: <int> 5007 Set to non-zero to probe tertiary and further ISA 5008 port ranges on PCI systems. Disabled by default. 5009 5010 pata_legacy.probe_mask= [HW,LIBATA] 5011 Format: <int> 5012 Probe mask for legacy ISA PATA ports. Depending on 5013 platform configuration and the use of other driver 5014 options up to 6 legacy ports are supported: 0x1f0, 5015 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing 5016 of individual ports can be disabled by setting the 5017 corresponding bits in the mask to 1. Bit 0 is for 5018 the first port in the list above (0x1f0), and so on. 5019 By default all supported ports are probed. 5020 5021 pata_platform.pio_mask= [HW,LIBATA] 5022 Format: <int> 5023 Supported PIO mode mask. Set individual bits to allow 5024 the use of the respective PIO modes. Bit 0 is for 5025 mode 0, bit 1 is for mode 1, and so on. Mode 0 only 5026 allowed by default. 5027 5028 pause_on_oops=<int> 5029 Halt all CPUs after the first oops has been printed for 5030 the specified number of seconds. This is to be used if 5031 your oopses keep scrolling off the screen. 5032 5033 pci=option[,option...] [PCI,EARLY] various PCI subsystem options. 5034 5035 Some options herein operate on a specific device 5036 or a set of devices (<pci_dev>). These are 5037 specified in one of the following formats: 5038 5039 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]* 5040 pci:<vendor>:<device>[:<subvendor>:<subdevice>] 5041 5042 Note: the first format specifies a PCI 5043 bus/device/function address which may change 5044 if new hardware is inserted, if motherboard 5045 firmware changes, or due to changes caused 5046 by other kernel parameters. If the 5047 domain is left unspecified, it is 5048 taken to be zero. Optionally, a path 5049 to a device through multiple device/function 5050 addresses can be specified after the base 5051 address (this is more robust against 5052 renumbering issues). The second format 5053 selects devices using IDs from the 5054 configuration space which may match multiple 5055 devices in the system. 5056 5057 earlydump dump PCI config space before the kernel 5058 changes anything 5059 off [X86] don't probe for the PCI bus 5060 bios [X86-32] force use of PCI BIOS, don't access 5061 the hardware directly. Use this if your machine 5062 has a non-standard PCI host bridge. 5063 nobios [X86-32] disallow use of PCI BIOS, only direct 5064 hardware access methods are allowed. Use this 5065 if you experience crashes upon bootup and you 5066 suspect they are caused by the BIOS. 5067 conf1 [X86] Force use of PCI Configuration Access 5068 Mechanism 1 (config address in IO port 0xCF8, 5069 data in IO port 0xCFC, both 32-bit). 5070 conf2 [X86] Force use of PCI Configuration Access 5071 Mechanism 2 (IO port 0xCF8 is an 8-bit port for 5072 the function, IO port 0xCFA, also 8-bit, sets 5073 bus number. The config space is then accessed 5074 through ports 0xC000-0xCFFF). 5075 See http://wiki.osdev.org/PCI for more info 5076 on the configuration access mechanisms. 5077 noaer [PCIE] If the PCIEAER kernel config parameter is 5078 enabled, this kernel boot option can be used to 5079 disable the use of PCIE advanced error reporting. 5080 nodomains [PCI] Disable support for multiple PCI 5081 root domains (aka PCI segments, in ACPI-speak). 5082 nommconf [X86] Disable use of MMCONFIG for PCI 5083 Configuration 5084 check_enable_amd_mmconf [X86] check for and enable 5085 properly configured MMIO access to PCI 5086 config space on AMD family 10h CPU 5087 nomsi [MSI] If the PCI_MSI kernel config parameter is 5088 enabled, this kernel boot option can be used to 5089 disable the use of MSI interrupts system-wide. 5090 noioapicquirk [APIC] Disable all boot interrupt quirks. 5091 Safety option to keep boot IRQs enabled. This 5092 should never be necessary. 5093 ioapicreroute [APIC] Enable rerouting of boot IRQs to the 5094 primary IO-APIC for bridges that cannot disable 5095 boot IRQs. This fixes a source of spurious IRQs 5096 when the system masks IRQs. 5097 noioapicreroute [APIC] Disable workaround that uses the 5098 boot IRQ equivalent of an IRQ that connects to 5099 a chipset where boot IRQs cannot be disabled. 5100 The opposite of ioapicreroute. 5101 biosirq [X86-32] Use PCI BIOS calls to get the interrupt 5102 routing table. These calls are known to be buggy 5103 on several machines and they hang the machine 5104 when used, but on other computers it's the only 5105 way to get the interrupt routing table. Try 5106 this option if the kernel is unable to allocate 5107 IRQs or discover secondary PCI buses on your 5108 motherboard. 5109 rom [X86] Assign address space to expansion ROMs. 5110 Use with caution as certain devices share 5111 address decoders between ROMs and other 5112 resources. 5113 norom [X86] Do not assign address space to 5114 expansion ROMs that do not already have 5115 BIOS assigned address ranges. 5116 nobar [X86] Do not assign address space to the 5117 BARs that weren't assigned by the BIOS. 5118 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be 5119 assigned automatically to PCI devices. You can 5120 make the kernel exclude IRQs of your ISA cards 5121 this way. 5122 pirqaddr=0xAAAAA [X86] Specify the physical address 5123 of the PIRQ table (normally generated 5124 by the BIOS) if it is outside the 5125 F0000h-100000h range. 5126 lastbus=N [X86] Scan all buses thru bus #N. Can be 5127 useful if the kernel is unable to find your 5128 secondary buses and you want to tell it 5129 explicitly which ones they are. 5130 assign-busses [X86] Always assign all PCI bus 5131 numbers ourselves, overriding 5132 whatever the firmware may have done. 5133 usepirqmask [X86] Honor the possible IRQ mask stored 5134 in the BIOS $PIR table. This is needed on 5135 some systems with broken BIOSes, notably 5136 some HP Pavilion N5400 and Omnibook XE3 5137 notebooks. This will have no effect if ACPI 5138 IRQ routing is enabled. 5139 noacpi [X86] Do not use ACPI for IRQ routing 5140 or for PCI scanning. 5141 use_crs [X86] Use PCI host bridge window information 5142 from ACPI. On BIOSes from 2008 or later, this 5143 is enabled by default. If you need to use this, 5144 please report a bug. 5145 nocrs [X86] Ignore PCI host bridge windows from ACPI. 5146 If you need to use this, please report a bug. 5147 use_e820 [X86] Use E820 reservations to exclude parts of 5148 PCI host bridge windows. This is a workaround 5149 for BIOS defects in host bridge _CRS methods. 5150 If you need to use this, please report a bug to 5151 <linux-pci@vger.kernel.org>. 5152 no_e820 [X86] Ignore E820 reservations for PCI host 5153 bridge windows. This is the default on modern 5154 hardware. If you need to use this, please report 5155 a bug to <linux-pci@vger.kernel.org>. 5156 routeirq Do IRQ routing for all PCI devices. 5157 This is normally done in pci_enable_device(), 5158 so this option is a temporary workaround 5159 for broken drivers that don't call it. 5160 skip_isa_align [X86] do not align io start addr, so can 5161 handle more pci cards 5162 noearly [X86] Don't do any early type 1 scanning. 5163 This might help on some broken boards which 5164 machine check when some devices' config space 5165 is read. But various workarounds are disabled 5166 and some IOMMU drivers will not work. 5167 bfsort Sort PCI devices into breadth-first order. 5168 This sorting is done to get a device 5169 order compatible with older (<= 2.4) kernels. 5170 nobfsort Don't sort PCI devices into breadth-first order. 5171 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size) 5172 tuning and use the BIOS-configured MPS defaults. 5173 pcie_bus_safe Set every device's MPS to the largest value 5174 supported by all devices below the root complex. 5175 pcie_bus_perf Set device MPS to the largest allowable MPS 5176 based on its parent bus. Also set MRRS (Max 5177 Read Request Size) to the largest supported 5178 value (no larger than the MPS that the device 5179 or bus can support) for best performance. 5180 pcie_bus_peer2peer Set every device's MPS to 128B, which 5181 every device is guaranteed to support. This 5182 configuration allows peer-to-peer DMA between 5183 any pair of devices, possibly at the cost of 5184 reduced performance. This also guarantees 5185 that hot-added devices will work. 5186 cbiosize=nn[KMG] The fixed amount of bus space which is 5187 reserved for the CardBus bridge's IO window. 5188 The default value is 256 bytes. 5189 cbmemsize=nn[KMG] The fixed amount of bus space which is 5190 reserved for the CardBus bridge's memory 5191 window. The default value is 64 megabytes. 5192 resource_alignment= 5193 Format: 5194 [<order of align>@]<pci_dev>[; ...] 5195 Specifies alignment and device to reassign 5196 aligned memory resources. How to 5197 specify the device is described above. 5198 If <order of align> is not specified, 5199 PAGE_SIZE is used as alignment. 5200 A PCI-PCI bridge can be specified if resource 5201 windows need to be expanded. 5202 To specify the alignment for several 5203 instances of a device, the PCI vendor, 5204 device, subvendor, and subdevice may be 5205 specified, e.g., 12@pci:8086:9c22:103c:198f 5206 for 4096-byte alignment. 5207 ecrc= Enable/disable PCIe ECRC (transaction layer 5208 end-to-end CRC checking). Only effective if 5209 OS has native AER control (either granted by 5210 ACPI _OSC or forced via "pcie_ports=native") 5211 bios: Use BIOS/firmware settings. This is the 5212 the default. 5213 off: Turn ECRC off 5214 on: Turn ECRC on. 5215 hpiosize=nn[KMG] The fixed amount of bus space which is 5216 reserved for hotplug bridge's IO window. 5217 Default size is 256 bytes. 5218 hpmmiosize=nn[KMG] The fixed amount of bus space which is 5219 reserved for hotplug bridge's MMIO window. 5220 Default size is 2 megabytes. 5221 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is 5222 reserved for hotplug bridge's MMIO_PREF window. 5223 Default size is 2 megabytes. 5224 hpmemsize=nn[KMG] The fixed amount of bus space which is 5225 reserved for hotplug bridge's MMIO and 5226 MMIO_PREF window. 5227 Default size is 2 megabytes. 5228 hpbussize=nn The minimum amount of additional bus numbers 5229 reserved for buses below a hotplug bridge. 5230 Default is 1. 5231 realloc= Enable/disable reallocating PCI bridge resources 5232 if allocations done by BIOS are too small to 5233 accommodate resources required by all child 5234 devices. 5235 off: Turn realloc off 5236 on: Turn realloc on 5237 realloc same as realloc=on 5238 noari do not use PCIe ARI. 5239 noats [PCIE, Intel-IOMMU, AMD-IOMMU] 5240 do not use PCIe ATS (and IOMMU device IOTLB). 5241 pcie_scan_all Scan all possible PCIe devices. Otherwise we 5242 only look for one device below a PCIe downstream 5243 port. 5244 big_root_window Try to add a big 64bit memory window to the PCIe 5245 root complex on AMD CPUs. Some GFX hardware 5246 can resize a BAR to allow access to all VRAM. 5247 Adding the window is slightly risky (it may 5248 conflict with unreported devices), so this 5249 taints the kernel. 5250 disable_acs_redir=<pci_dev>[; ...] 5251 Specify one or more PCI devices (in the format 5252 specified above) separated by semicolons. 5253 Each device specified will have the PCI ACS 5254 redirect capabilities forced off which will 5255 allow P2P traffic between devices through 5256 bridges without forcing it upstream. Note: 5257 this removes isolation between devices and 5258 may put more devices in an IOMMU group. 5259 config_acs= 5260 Format: 5261 <ACS flags>@<pci_dev>[; ...] 5262 Specify one or more PCI devices (in the format 5263 specified above) optionally prepended with flags 5264 and separated by semicolons. The respective 5265 capabilities will be enabled, disabled or 5266 unchanged based on what is specified in 5267 flags. 5268 5269 ACS Flags is defined as follows: 5270 bit-0 : ACS Source Validation 5271 bit-1 : ACS Translation Blocking 5272 bit-2 : ACS P2P Request Redirect 5273 bit-3 : ACS P2P Completion Redirect 5274 bit-4 : ACS Upstream Forwarding 5275 bit-5 : ACS P2P Egress Control 5276 bit-6 : ACS Direct Translated P2P 5277 Each bit can be marked as: 5278 '0' – force disabled 5279 '1' – force enabled 5280 'x' – unchanged 5281 For example, 5282 pci=config_acs=10x@pci:0:0 5283 would configure all devices that support 5284 ACS to enable P2P Request Redirect, disable 5285 Translation Blocking, and leave Source 5286 Validation unchanged from whatever power-up 5287 or firmware set it to. 5288 5289 Note: this may remove isolation between devices 5290 and may put more devices in an IOMMU group. 5291 force_floating [S390] Force usage of floating interrupts. 5292 nomio [S390] Do not use MIO instructions. 5293 norid [S390] ignore the RID field and force use of 5294 one PCI domain per PCI function 5295 notph [PCIE] If the PCIE_TPH kernel config parameter 5296 is enabled, this kernel boot option can be used 5297 to disable PCIe TLP Processing Hints support 5298 system-wide. 5299 5300 pcie_aspm= [PCIE] Forcibly enable or ignore PCIe Active State Power 5301 Management. 5302 off Don't touch ASPM configuration at all. Leave any 5303 configuration done by firmware unchanged. 5304 force Enable ASPM even on devices that claim not to support it. 5305 WARNING: Forcing ASPM on may cause system lockups. 5306 5307 pcie_ports= [PCIE] PCIe port services handling: 5308 native Use native PCIe services (PME, AER, DPC, PCIe hotplug) 5309 even if the platform doesn't give the OS permission to 5310 use them. This may cause conflicts if the platform 5311 also tries to use these services. 5312 dpc-native Use native PCIe service for DPC only. May 5313 cause conflicts if firmware uses AER or DPC. 5314 compat Disable native PCIe services (PME, AER, DPC, PCIe 5315 hotplug). 5316 5317 pcie_port_pm= [PCIE] PCIe port power management handling: 5318 off Disable power management of all PCIe ports 5319 force Forcibly enable power management of all PCIe ports 5320 5321 pcie_pme= [PCIE,PM] Native PCIe PME signaling options: 5322 nomsi Do not use MSI for native PCIe PME signaling (this makes 5323 all PCIe root ports use INTx for all services). 5324 5325 pd_ignore_unused 5326 [PM] 5327 Keep all power-domains already enabled by bootloader on, 5328 even if no driver has claimed them. This is useful 5329 for debug and development, but should not be 5330 needed on a platform with proper driver support. 5331 5332 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at 5333 boot time. 5334 Format: { 0 | 1 } 5335 See arch/parisc/kernel/pdc_chassis.c 5336 5337 percpu_alloc= [MM,EARLY] 5338 Select which percpu first chunk allocator to use. 5339 Currently supported values are "embed" and "page". 5340 Archs may support subset or none of the selections. 5341 See comments in mm/percpu.c for details on each 5342 allocator. This parameter is primarily for debugging 5343 and performance comparison. 5344 5345 pirq= [SMP,APIC] Manual mp-table setup 5346 See Documentation/arch/x86/i386/IO-APIC.rst. 5347 5348 plip= [PPT,NET] Parallel port network link 5349 Format: { parport<nr> | timid | 0 } 5350 See also Documentation/admin-guide/parport.rst. 5351 5352 pmtmr= [X86] Manual setup of pmtmr I/O Port. 5353 Override pmtimer IOPort with a hex value. 5354 e.g. pmtmr=0x508 5355 5356 pmu_override= [PPC] Override the PMU. 5357 This option takes over the PMU facility, so it is no 5358 longer usable by perf. Setting this option starts the 5359 PMU counters by setting MMCR0 to 0 (the FC bit is 5360 cleared). If a number is given, then MMCR1 is set to 5361 that number, otherwise (e.g., 'pmu_override=on'), MMCR1 5362 remains 0. 5363 5364 pm_async= [PM] 5365 Format: off 5366 This parameter sets the initial value of the 5367 /sys/power/pm_async sysfs knob at boot time. 5368 If set to "off", disables asynchronous suspend and 5369 resume of devices during system-wide power transitions. 5370 This can be useful on platforms where device 5371 dependencies are not well-defined, or for debugging 5372 power management issues. Asynchronous operations are 5373 enabled by default. 5374 5375 5376 pm_debug_messages [SUSPEND,KNL] 5377 Enable suspend/resume debug messages during boot up. 5378 5379 pm.dpm_watchdog_enabled= 5380 [PM] Enable or disable the DPM watchdog. Requires 5381 CONFIG_PM_SLEEP and CONFIG_DPM_WATCHDOG enabled. 5382 Format: <bool> 5383 Default value is set by CONFIG_DPM_WATCHDOG_ENABLED. 5384 5385 pnp.debug=1 [PNP] 5386 Enable PNP debug messages (depends on the 5387 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time 5388 via /sys/module/pnp/parameters/debug. We always show 5389 current resource usage; turning this on also shows 5390 possible settings and some assignment information. 5391 5392 pnpacpi= [ACPI] 5393 { off } 5394 5395 pnpbios= [ISAPNP] 5396 { on | off | curr | res | no-curr | no-res } 5397 5398 pnp_reserve_irq= 5399 [ISAPNP] Exclude IRQs for the autoconfiguration 5400 5401 pnp_reserve_dma= 5402 [ISAPNP] Exclude DMAs for the autoconfiguration 5403 5404 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration 5405 Ranges are in pairs (I/O port base and size). 5406 5407 pnp_reserve_mem= 5408 [ISAPNP] Exclude memory regions for the 5409 autoconfiguration. 5410 Ranges are in pairs (memory base and size). 5411 5412 ports= [IP_VS_FTP] IPVS ftp helper module 5413 Default is 21. 5414 Up to 8 (IP_VS_APP_MAX_PORTS) ports 5415 may be specified. 5416 Format: <port>,<port>.... 5417 5418 possible_cpus= [SMP,S390,X86] 5419 Format: <unsigned int> 5420 Set the number of possible CPUs, overriding the 5421 regular discovery mechanisms (such as ACPI/FW, etc). 5422 5423 powersave=off [PPC] This option disables power saving features. 5424 It specifically disables cpuidle and sets the 5425 platform machine description specific power_save 5426 function to NULL. On Idle the CPU just reduces 5427 execution priority. 5428 5429 ppc_strict_facility_enable 5430 [PPC,ENABLE] This option catches any kernel floating point, 5431 Altivec, VSX and SPE outside of regions specifically 5432 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()). 5433 There is some performance impact when enabling this. 5434 5435 ppc_tm= [PPC,EARLY] 5436 Format: {"off"} 5437 Disable Hardware Transactional Memory 5438 5439 preempt= [KNL] 5440 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC 5441 none - Limited to cond_resched() calls 5442 voluntary - Limited to cond_resched() and might_sleep() calls 5443 full - Any section that isn't explicitly preempt disabled 5444 can be preempted anytime. Tasks will also yield 5445 contended spinlocks (if the critical section isn't 5446 explicitly preempt disabled beyond the lock itself). 5447 lazy - Scheduler controlled. Similar to full but instead 5448 of preempting the task immediately, the task gets 5449 one HZ tick time to yield itself before the 5450 preemption will be forced. One such preemption 5451 point is when the task returns to user space. 5452 5453 print-fatal-signals= 5454 [KNL] debug: print fatal signals 5455 5456 If enabled, warn about various signal handling 5457 related application anomalies: too many signals, 5458 too many POSIX.1 timers, fatal signals causing a 5459 coredump - etc. 5460 5461 If you hit the warning due to signal overflow, 5462 you might want to try "ulimit -i unlimited". 5463 5464 default: off. 5465 5466 printk.always_kmsg_dump= 5467 Trigger kmsg_dump for cases other than kernel oops or 5468 panics 5469 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5470 default: disabled 5471 5472 printk.console_no_auto_verbose= 5473 Disable console loglevel raise on oops, panic 5474 or lockdep-detected issues (only if lock debug is on). 5475 With an exception to setups with low baudrate on 5476 serial console, keeping this 0 is a good choice 5477 in order to provide more debug information. 5478 Format: <bool> 5479 default: 0 (auto_verbose is enabled) 5480 5481 printk.debug_non_panic_cpus= 5482 Allows storing messages from non-panic CPUs into 5483 the printk log buffer during panic(). They are 5484 flushed to consoles by the panic-CPU on 5485 a best-effort basis. 5486 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5487 Default: disabled 5488 5489 printk.devkmsg={on,off,ratelimit} 5490 Control writing to /dev/kmsg. 5491 on - unlimited logging to /dev/kmsg from userspace 5492 off - logging to /dev/kmsg disabled 5493 ratelimit - ratelimit the logging 5494 Default: ratelimit 5495 5496 printk.time= Show timing data prefixed to each printk message line 5497 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5498 5499 proc_mem.force_override= [KNL] 5500 Format: {always | ptrace | never} 5501 Traditionally /proc/pid/mem allows memory permissions to be 5502 overridden without restrictions. This option may be set to 5503 restrict that. Can be one of: 5504 - 'always': traditional behavior always allows mem overrides. 5505 - 'ptrace': only allow mem overrides for active ptracers. 5506 - 'never': never allow mem overrides. 5507 If not specified, default is the CONFIG_PROC_MEM_* choice. 5508 5509 processor.max_cstate= [HW,ACPI] 5510 Limit processor to maximum C-state 5511 max_cstate=9 overrides any DMI blacklist limit. 5512 5513 processor.nocst [HW,ACPI] 5514 Ignore the _CST method to determine C-states, 5515 instead using the legacy FADT method 5516 5517 profile= [KNL] Enable kernel profiling via /proc/profile 5518 Format: [<profiletype>,]<number> 5519 Param: <profiletype>: "schedule" or "kvm" 5520 [defaults to kernel profiling] 5521 Param: "schedule" - profile schedule points. 5522 Param: "kvm" - profile VM exits. 5523 Param: <number> - step/bucket size as a power of 2 for 5524 statistical time based profiling. 5525 5526 prot_virt= [S390] enable hosting protected virtual machines 5527 isolated from the hypervisor (if hardware supports 5528 that). If enabled, the default kernel base address 5529 might be overridden even when Kernel Address Space 5530 Layout Randomization is disabled. 5531 Format: <bool> 5532 5533 psi= [KNL] Enable or disable pressure stall information 5534 tracking. 5535 Format: <bool> 5536 5537 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to 5538 probe for; one of (bare|imps|exps|lifebook|any). 5539 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports 5540 per second. 5541 psmouse.resetafter= [HW,MOUSE] 5542 Try to reset the device after so many bad packets 5543 (0 = never). 5544 psmouse.resolution= 5545 [HW,MOUSE] Set desired mouse resolution, in dpi. 5546 psmouse.smartscroll= 5547 [HW,MOUSE] Controls Logitech smartscroll autorepeat. 5548 0 = disabled, 1 = enabled (default). 5549 5550 pstore.backend= Specify the name of the pstore backend to use 5551 5552 pti= [X86-64] Control Page Table Isolation of user and 5553 kernel address spaces. Disabling this feature 5554 removes hardening, but improves performance of 5555 system calls and interrupts. 5556 5557 on - unconditionally enable 5558 off - unconditionally disable 5559 auto - kernel detects whether your CPU model is 5560 vulnerable to issues that PTI mitigates 5561 5562 Not specifying this option is equivalent to pti=auto. 5563 5564 pty.legacy_count= 5565 [KNL] Number of legacy pty's. Overwrites compiled-in 5566 default number. 5567 5568 quiet [KNL,EARLY] Disable most log messages 5569 5570 r128= [HW,DRM] 5571 5572 radix_hcall_invalidate=on [PPC/PSERIES] 5573 Disable RADIX GTSE feature and use hcall for TLB 5574 invalidate. 5575 5576 raid= [HW,RAID] 5577 See Documentation/admin-guide/md.rst. 5578 5579 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes 5580 See Documentation/admin-guide/blockdev/ramdisk.rst. 5581 5582 ramdisk_start= [Deprecated,RAM] RAM disk image start address 5583 5584 random.trust_cpu=off 5585 [KNL,EARLY] Disable trusting the use of the CPU's 5586 random number generator (if available) to 5587 initialize the kernel's RNG. 5588 5589 random.trust_bootloader=off 5590 [KNL,EARLY] Disable trusting the use of the a seed 5591 passed by the bootloader (if available) to 5592 initialize the kernel's RNG. 5593 5594 randomize_kstack_offset= 5595 [KNL,EARLY] Enable or disable kernel stack offset 5596 randomization, which provides roughly 5 bits of 5597 entropy, frustrating memory corruption attacks 5598 that depend on stack address determinism or 5599 cross-syscall address exposures. This is only 5600 available on architectures that have defined 5601 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET. 5602 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5603 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT. 5604 5605 ras=option[,option,...] [KNL] RAS-specific options 5606 5607 cec_disable [X86] 5608 Disable the Correctable Errors Collector, 5609 see CONFIG_RAS_CEC help text. 5610 5611 rcu_nocbs[=cpu-list] 5612 [KNL] The optional argument is a cpu list, 5613 as described above. 5614 5615 In kernels built with CONFIG_RCU_NOCB_CPU=y, 5616 enable the no-callback CPU mode, which prevents 5617 such CPUs' callbacks from being invoked in 5618 softirq context. Invocation of such CPUs' RCU 5619 callbacks will instead be offloaded to "rcuox/N" 5620 kthreads created for that purpose, where "x" is 5621 "p" for RCU-preempt, "s" for RCU-sched, and "g" 5622 for the kthreads that mediate grace periods; and 5623 "N" is the CPU number. This reduces OS jitter on 5624 the offloaded CPUs, which can be useful for HPC 5625 and real-time workloads. It can also improve 5626 energy efficiency for asymmetric multiprocessors. 5627 5628 If a cpulist is passed as an argument, the specified 5629 list of CPUs is set to no-callback mode from boot. 5630 5631 Otherwise, if the '=' sign and the cpulist 5632 arguments are omitted, no CPU will be set to 5633 no-callback mode from boot but the mode may be 5634 toggled at runtime via cpusets. 5635 5636 Note that this argument takes precedence over 5637 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option. 5638 5639 rcu_nocb_poll [KNL] 5640 Rather than requiring that offloaded CPUs 5641 (specified by rcu_nocbs= above) explicitly 5642 awaken the corresponding "rcuoN" kthreads, 5643 make these kthreads poll for callbacks. 5644 This improves the real-time response for the 5645 offloaded CPUs by relieving them of the need to 5646 wake up the corresponding kthread, but degrades 5647 energy efficiency by requiring that the kthreads 5648 periodically wake up to do the polling. 5649 5650 rcutree.blimit= [KNL] 5651 Set maximum number of finished RCU callbacks to 5652 process in one batch. 5653 5654 rcutree.csd_lock_suppress_rcu_stall= [KNL] 5655 Do only a one-line RCU CPU stall warning when 5656 there is an ongoing too-long CSD-lock wait. 5657 5658 rcutree.do_rcu_barrier= [KNL] 5659 Wait for deferred kfree_rcu() frees and ordinary 5660 call_rcu() callbacks queued before this request to 5661 complete. This does not prevent new work from being 5662 queued concurrently. Requests are throttled so that 5663 userspace tests can safely hammer on the sysfs 5664 variable if they so choose. 5665 If triggered before the RCU grace-period machinery 5666 is fully active, this will error out with EAGAIN. 5667 5668 rcutree.dump_tree= [KNL] 5669 Dump the structure of the rcu_node combining tree 5670 out at early boot. This is used for diagnostic 5671 purposes, to verify correct tree setup. 5672 5673 rcutree.gp_cleanup_delay= [KNL] 5674 Set the number of jiffies to delay each step of 5675 RCU grace-period cleanup. 5676 5677 rcutree.gp_init_delay= [KNL] 5678 Set the number of jiffies to delay each step of 5679 RCU grace-period initialization. 5680 5681 rcutree.gp_preinit_delay= [KNL] 5682 Set the number of jiffies to delay each step of 5683 RCU grace-period pre-initialization, that is, 5684 the propagation of recent CPU-hotplug changes up 5685 the rcu_node combining tree. 5686 5687 rcutree.jiffies_till_first_fqs= [KNL] 5688 Set delay from grace-period initialization to 5689 first attempt to force quiescent states. 5690 Units are jiffies, minimum value is zero, 5691 and maximum value is HZ. 5692 5693 rcutree.jiffies_till_next_fqs= [KNL] 5694 Set delay between subsequent attempts to force 5695 quiescent states. Units are jiffies, minimum 5696 value is one, and maximum value is HZ. 5697 5698 rcutree.jiffies_till_sched_qs= [KNL] 5699 Set required age in jiffies for a 5700 given grace period before RCU starts 5701 soliciting quiescent-state help from 5702 rcu_note_context_switch() and cond_resched(). 5703 If not specified, the kernel will calculate 5704 a value based on the most recent settings 5705 of rcutree.jiffies_till_first_fqs 5706 and rcutree.jiffies_till_next_fqs. 5707 This calculated value may be viewed in 5708 rcutree.jiffies_to_sched_qs. Any attempt to set 5709 rcutree.jiffies_to_sched_qs will be cheerfully 5710 overwritten. 5711 5712 rcutree.kthread_prio= [KNL,BOOT] 5713 Set the SCHED_FIFO priority of the RCU per-CPU 5714 kthreads (rcuc/N). This value is also used for 5715 the priority of the RCU boost threads (rcub/N) 5716 and for the RCU grace-period kthreads (rcu_bh, 5717 rcu_preempt, and rcu_sched). If RCU_BOOST is 5718 set, valid values are 1-99 and the default is 1 5719 (the least-favored priority). Otherwise, when 5720 RCU_BOOST is not set, valid values are 0-99 and 5721 the default is zero (non-realtime operation). 5722 When RCU_NOCB_CPU is set, also adjust the 5723 priority of NOCB callback kthreads. 5724 5725 rcutree.nocb_nobypass_lim_per_jiffy= [KNL] 5726 On callback-offloaded (rcu_nocbs) CPUs, 5727 RCU reduces the lock contention that would 5728 otherwise be caused by callback floods through 5729 use of the ->nocb_bypass list. However, in the 5730 common non-flooded case, RCU queues directly to 5731 the main ->cblist in order to avoid the extra 5732 overhead of the ->nocb_bypass list and its lock. 5733 But if there are too many callbacks queued during 5734 a single jiffy, RCU pre-queues the callbacks into 5735 the ->nocb_bypass queue. The definition of "too 5736 many" is supplied by this kernel boot parameter. 5737 5738 rcutree.nohz_full_patience_delay= [KNL] 5739 On callback-offloaded (rcu_nocbs) CPUs, avoid 5740 disturbing RCU unless the grace period has 5741 reached the specified age in milliseconds. 5742 Defaults to zero. Large values will be capped 5743 at five seconds. All values will be rounded down 5744 to the nearest value representable by jiffies. 5745 5746 rcutree.qhimark= [KNL] 5747 Set threshold of queued RCU callbacks beyond which 5748 batch limiting is disabled. 5749 5750 rcutree.qlowmark= [KNL] 5751 Set threshold of queued RCU callbacks below which 5752 batch limiting is re-enabled. 5753 5754 rcutree.qovld= [KNL] 5755 Set threshold of queued RCU callbacks beyond which 5756 RCU's force-quiescent-state scan will aggressively 5757 enlist help from cond_resched() and sched IPIs to 5758 help CPUs more quickly reach quiescent states. 5759 Set to less than zero to make this be set based 5760 on rcutree.qhimark at boot time and to zero to 5761 disable more aggressive help enlistment. 5762 5763 rcutree.rcu_delay_page_cache_fill_msec= [KNL] 5764 Set the page-cache refill delay (in milliseconds) 5765 in response to low-memory conditions. The range 5766 of permitted values is in the range 0:100000. 5767 5768 rcutree.rcu_divisor= [KNL] 5769 Set the shift-right count to use to compute 5770 the callback-invocation batch limit bl from 5771 the number of callbacks queued on this CPU. 5772 The result will be bounded below by the value of 5773 the rcutree.blimit kernel parameter. Every bl 5774 callbacks, the softirq handler will exit in 5775 order to allow the CPU to do other work. 5776 5777 Please note that this callback-invocation batch 5778 limit applies only to non-offloaded callback 5779 invocation. Offloaded callbacks are instead 5780 invoked in the context of an rcuoc kthread, which 5781 scheduler will preempt as it does any other task. 5782 5783 rcutree.rcu_fanout_exact= [KNL] 5784 Disable autobalancing of the rcu_node combining 5785 tree. This is used by rcutorture, and might 5786 possibly be useful for architectures having high 5787 cache-to-cache transfer latencies. 5788 5789 rcutree.rcu_fanout_leaf= [KNL] 5790 Change the number of CPUs assigned to each 5791 leaf rcu_node structure. Useful for very 5792 large systems, which will choose the value 64, 5793 and for NUMA systems with large remote-access 5794 latencies, which will choose a value aligned 5795 with the appropriate hardware boundaries. 5796 5797 rcutree.rcu_min_cached_objs= [KNL] 5798 Minimum number of objects which are cached and 5799 maintained per one CPU. Object size is equal 5800 to PAGE_SIZE. The cache allows to reduce the 5801 pressure to page allocator, also it makes the 5802 whole algorithm to behave better in low memory 5803 condition. 5804 5805 rcutree.rcu_nocb_gp_stride= [KNL] 5806 Set the number of NOCB callback kthreads in 5807 each group, which defaults to the square root 5808 of the number of CPUs. Larger numbers reduce 5809 the wakeup overhead on the global grace-period 5810 kthread, but increases that same overhead on 5811 each group's NOCB grace-period kthread. 5812 5813 rcutree.rcu_kick_kthreads= [KNL] 5814 Cause the grace-period kthread to get an extra 5815 wake_up() if it sleeps three times longer than 5816 it should at force-quiescent-state time. 5817 This wake_up() will be accompanied by a 5818 WARN_ONCE() splat and an ftrace_dump(). 5819 5820 rcutree.rcu_resched_ns= [KNL] 5821 Limit the time spend invoking a batch of RCU 5822 callbacks to the specified number of nanoseconds. 5823 By default, this limit is checked only once 5824 every 32 callbacks in order to limit the pain 5825 inflicted by local_clock() overhead. 5826 5827 rcutree.rcu_unlock_delay= [KNL] 5828 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels, 5829 this specifies an rcu_read_unlock()-time delay 5830 in microseconds. This defaults to zero. 5831 Larger delays increase the probability of 5832 catching RCU pointer leaks, that is, buggy use 5833 of RCU-protected pointers after the relevant 5834 rcu_read_unlock() has completed. 5835 5836 rcutree.sysrq_rcu= [KNL] 5837 Commandeer a sysrq key to dump out Tree RCU's 5838 rcu_node tree with an eye towards determining 5839 why a new grace period has not yet started. 5840 5841 rcutree.use_softirq= [KNL] 5842 If set to zero, move all RCU_SOFTIRQ processing to 5843 per-CPU rcuc kthreads. Defaults to a non-zero 5844 value, meaning that RCU_SOFTIRQ is used by default. 5845 Specify rcutree.use_softirq=0 to use rcuc kthreads. 5846 5847 But note that CONFIG_PREEMPT_RT=y kernels disable 5848 this kernel boot parameter, forcibly setting it 5849 to zero. 5850 5851 rcutree.enable_rcu_lazy= [KNL] 5852 To save power, batch RCU callbacks and flush after 5853 delay, memory pressure or callback list growing too 5854 big. 5855 5856 rcutree.rcu_normal_wake_from_gp= [KNL] 5857 Reduces a latency of synchronize_rcu() call. This approach 5858 maintains its own track of synchronize_rcu() callers, so it 5859 does not interact with regular callbacks because it does not 5860 use a call_rcu[_hurry]() path. Please note, this is for a 5861 normal grace period. 5862 5863 How to disable it: 5864 5865 echo 0 > /sys/module/rcutree/parameters/rcu_normal_wake_from_gp 5866 or pass a boot parameter "rcutree.rcu_normal_wake_from_gp=0" 5867 5868 Default is 1 if it is not explicitly disabled by the boot parameter 5869 passing 0. 5870 5871 rcuscale.gp_async= [KNL] 5872 Measure performance of asynchronous 5873 grace-period primitives such as call_rcu(). 5874 5875 rcuscale.gp_async_max= [KNL] 5876 Specify the maximum number of outstanding 5877 callbacks per writer thread. When a writer 5878 thread exceeds this limit, it invokes the 5879 corresponding flavor of rcu_barrier() to allow 5880 previously posted callbacks to drain. 5881 5882 rcuscale.gp_exp= [KNL] 5883 Measure performance of expedited synchronous 5884 grace-period primitives. 5885 5886 rcuscale.holdoff= [KNL] 5887 Set test-start holdoff period. The purpose of 5888 this parameter is to delay the start of the 5889 test until boot completes in order to avoid 5890 interference. 5891 5892 rcuscale.kfree_by_call_rcu= [KNL] 5893 In kernels built with CONFIG_RCU_LAZY=y, test 5894 call_rcu() instead of kfree_rcu(). 5895 5896 rcuscale.kfree_mult= [KNL] 5897 Instead of allocating an object of size kfree_obj, 5898 allocate one of kfree_mult * sizeof(kfree_obj). 5899 Defaults to 1. 5900 5901 rcuscale.kfree_rcu_test= [KNL] 5902 Set to measure performance of kfree_rcu() flooding. 5903 5904 rcuscale.kfree_rcu_test_double= [KNL] 5905 Test the double-argument variant of kfree_rcu(). 5906 If this parameter has the same value as 5907 rcuscale.kfree_rcu_test_single, both the single- 5908 and double-argument variants are tested. 5909 5910 rcuscale.kfree_rcu_test_single= [KNL] 5911 Test the single-argument variant of kfree_rcu(). 5912 If this parameter has the same value as 5913 rcuscale.kfree_rcu_test_double, both the single- 5914 and double-argument variants are tested. 5915 5916 rcuscale.kfree_nthreads= [KNL] 5917 The number of threads running loops of kfree_rcu(). 5918 5919 rcuscale.kfree_alloc_num= [KNL] 5920 Number of allocations and frees done in an iteration. 5921 5922 rcuscale.kfree_loops= [KNL] 5923 Number of loops doing rcuscale.kfree_alloc_num number 5924 of allocations and frees. 5925 5926 rcuscale.minruntime= [KNL] 5927 Set the minimum test run time in seconds. This 5928 does not affect the data-collection interval, 5929 but instead allows better measurement of things 5930 like CPU consumption. 5931 5932 rcuscale.nreaders= [KNL] 5933 Set number of RCU readers. The value -1 selects 5934 N, where N is the number of CPUs. A value 5935 "n" less than -1 selects N-n+1, where N is again 5936 the number of CPUs. For example, -2 selects N 5937 (the number of CPUs), -3 selects N+1, and so on. 5938 A value of "n" less than or equal to -N selects 5939 a single reader. 5940 5941 rcuscale.nwriters= [KNL] 5942 Set number of RCU writers. The values operate 5943 the same as for rcuscale.nreaders. 5944 N, where N is the number of CPUs 5945 5946 rcuscale.scale_type= [KNL] 5947 Specify the RCU implementation to test. 5948 5949 rcuscale.shutdown= [KNL] 5950 Shut the system down after performance tests 5951 complete. This is useful for hands-off automated 5952 testing. 5953 5954 rcuscale.verbose= [KNL] 5955 Enable additional printk() statements. 5956 5957 rcuscale.writer_holdoff= [KNL] 5958 Write-side holdoff between grace periods, 5959 in microseconds. The default of zero says 5960 no holdoff. 5961 5962 rcuscale.writer_holdoff_jiffies= [KNL] 5963 Additional write-side holdoff between grace 5964 periods, but in jiffies. The default of zero 5965 says no holdoff. 5966 5967 rcutorture.fqs_duration= [KNL] 5968 Set duration of force_quiescent_state bursts 5969 in microseconds. 5970 5971 rcutorture.fqs_holdoff= [KNL] 5972 Set holdoff time within force_quiescent_state bursts 5973 in microseconds. 5974 5975 rcutorture.fqs_stutter= [KNL] 5976 Set wait time between force_quiescent_state bursts 5977 in seconds. 5978 5979 rcutorture.fwd_progress= [KNL] 5980 Specifies the number of kthreads to be used 5981 for RCU grace-period forward-progress testing 5982 for the types of RCU supporting this notion. 5983 Defaults to 1 kthread, values less than zero or 5984 greater than the number of CPUs cause the number 5985 of CPUs to be used. 5986 5987 rcutorture.fwd_progress_div= [KNL] 5988 Specify the fraction of a CPU-stall-warning 5989 period to do tight-loop forward-progress testing. 5990 5991 rcutorture.fwd_progress_holdoff= [KNL] 5992 Number of seconds to wait between successive 5993 forward-progress tests. 5994 5995 rcutorture.fwd_progress_need_resched= [KNL] 5996 Enclose cond_resched() calls within checks for 5997 need_resched() during tight-loop forward-progress 5998 testing. 5999 6000 rcutorture.gp_cond= [KNL] 6001 Use conditional/asynchronous update-side 6002 normal-grace-period primitives, if available. 6003 6004 rcutorture.gp_cond_exp= [KNL] 6005 Use conditional/asynchronous update-side 6006 expedited-grace-period primitives, if available. 6007 6008 rcutorture.gp_cond_full= [KNL] 6009 Use conditional/asynchronous update-side 6010 normal-grace-period primitives that also take 6011 concurrent expedited grace periods into account, 6012 if available. 6013 6014 rcutorture.gp_cond_exp_full= [KNL] 6015 Use conditional/asynchronous update-side 6016 expedited-grace-period primitives that also take 6017 concurrent normal grace periods into account, 6018 if available. 6019 6020 rcutorture.gp_cond_wi= [KNL] 6021 Nominal wait interval for normal conditional 6022 grace periods (specified by rcutorture's 6023 gp_cond and gp_cond_full module parameters), 6024 in microseconds. The actual wait interval will 6025 be randomly selected to nanosecond granularity up 6026 to this wait interval. Defaults to 16 jiffies, 6027 for example, 16,000 microseconds on a system 6028 with HZ=1000. 6029 6030 rcutorture.gp_cond_wi_exp= [KNL] 6031 Nominal wait interval for expedited conditional 6032 grace periods (specified by rcutorture's 6033 gp_cond_exp and gp_cond_exp_full module 6034 parameters), in microseconds. The actual wait 6035 interval will be randomly selected to nanosecond 6036 granularity up to this wait interval. Defaults to 6037 128 microseconds. 6038 6039 rcutorture.gp_exp= [KNL] 6040 Use expedited update-side primitives, if available. 6041 6042 rcutorture.gp_normal= [KNL] 6043 Use normal (non-expedited) asynchronous 6044 update-side primitives, if available. 6045 6046 rcutorture.gp_poll= [KNL] 6047 Use polled update-side normal-grace-period 6048 primitives, if available. 6049 6050 rcutorture.gp_poll_exp= [KNL] 6051 Use polled update-side expedited-grace-period 6052 primitives, if available. 6053 6054 rcutorture.gp_poll_full= [KNL] 6055 Use polled update-side normal-grace-period 6056 primitives that also take concurrent expedited 6057 grace periods into account, if available. 6058 6059 rcutorture.gp_poll_exp_full= [KNL] 6060 Use polled update-side expedited-grace-period 6061 primitives that also take concurrent normal 6062 grace periods into account, if available. 6063 6064 rcutorture.gp_poll_wi= [KNL] 6065 Nominal wait interval for normal conditional 6066 grace periods (specified by rcutorture's 6067 gp_poll and gp_poll_full module parameters), 6068 in microseconds. The actual wait interval will 6069 be randomly selected to nanosecond granularity up 6070 to this wait interval. Defaults to 16 jiffies, 6071 for example, 16,000 microseconds on a system 6072 with HZ=1000. 6073 6074 rcutorture.gp_poll_wi_exp= [KNL] 6075 Nominal wait interval for expedited conditional 6076 grace periods (specified by rcutorture's 6077 gp_poll_exp and gp_poll_exp_full module 6078 parameters), in microseconds. The actual wait 6079 interval will be randomly selected to nanosecond 6080 granularity up to this wait interval. Defaults to 6081 128 microseconds. 6082 6083 rcutorture.gp_sync= [KNL] 6084 Use normal (non-expedited) synchronous 6085 update-side primitives, if available. If all 6086 of rcutorture.gp_cond=, rcutorture.gp_exp=, 6087 rcutorture.gp_normal=, and rcutorture.gp_sync= 6088 are zero, rcutorture acts as if is interpreted 6089 they are all non-zero. 6090 6091 rcutorture.gpwrap_lag= [KNL] 6092 Enable grace-period wrap lag testing. Setting 6093 to false prevents the gpwrap lag test from 6094 running. Default is true. 6095 6096 rcutorture.gpwrap_lag_gps= [KNL] 6097 Set the value for grace-period wrap lag during 6098 active lag testing periods. This controls how many 6099 grace periods differences we tolerate between 6100 rdp and rnp's gp_seq before setting overflow flag. 6101 The default is always set to 8. 6102 6103 rcutorture.gpwrap_lag_cycle_mins= [KNL] 6104 Set the total cycle duration for gpwrap lag 6105 testing in minutes. This is the total time for 6106 one complete cycle of active and inactive 6107 testing periods. Default is 30 minutes. 6108 6109 rcutorture.gpwrap_lag_active_mins= [KNL] 6110 Set the duration for which gpwrap lag is active 6111 within each cycle, in minutes. During this time, 6112 the grace-period wrap lag will be set to the 6113 value specified by gpwrap_lag_gps. Default is 6114 5 minutes. 6115 6116 rcutorture.irqreader= [KNL] 6117 Run RCU readers from irq handlers, or, more 6118 accurately, from a timer handler. Not all RCU 6119 flavors take kindly to this sort of thing. 6120 6121 rcutorture.leakpointer= [KNL] 6122 Leak an RCU-protected pointer out of the reader. 6123 This can of course result in splats, and is 6124 intended to test the ability of things like 6125 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect 6126 such leaks. 6127 6128 rcutorture.n_barrier_cbs= [KNL] 6129 Set callbacks/threads for rcu_barrier() testing. 6130 6131 rcutorture.nfakewriters= [KNL] 6132 Set number of concurrent RCU writers. These just 6133 stress RCU, they don't participate in the actual 6134 test, hence the "fake". 6135 6136 rcutorture.nmi_calls= [KNL] 6137 Enable issuing RCU callbacks from an NMI, on the 6138 RCU flavors that support it, to exercise the 6139 any-context callback path. Requires 6140 CONFIG_PERF_EVENTS and a hardware PMU; without 6141 both, nothing is issued. Defaults to enabled. 6142 6143 rcutorture.nocbs_nthreads= [KNL] 6144 Set number of RCU callback-offload togglers. 6145 Zero (the default) disables toggling. 6146 6147 rcutorture.nocbs_toggle= [KNL] 6148 Set the delay in milliseconds between successive 6149 callback-offload toggling attempts. 6150 6151 rcutorture.nreaders= [KNL] 6152 Set number of RCU readers. The value -1 selects 6153 N-1, where N is the number of CPUs. A value 6154 "n" less than -1 selects N-n-2, where N is again 6155 the number of CPUs. For example, -2 selects N 6156 (the number of CPUs), -3 selects N+1, and so on. 6157 6158 rcutorture.nwriters= [KNL] 6159 Set number of RCU writers, which must be either 6160 zero or one. For additional writers, use instead 6161 the rcutorture.nfakewriters parameter. 6162 6163 rcutorture.object_debug= [KNL] 6164 Enable debug-object double-call_rcu() testing. 6165 6166 rcutorture.onoff_holdoff= [KNL] 6167 Set time (s) after boot for CPU-hotplug testing. 6168 6169 rcutorture.onoff_interval= [KNL] 6170 Set time (jiffies) between CPU-hotplug operations, 6171 or zero to disable CPU-hotplug testing. 6172 6173 rcutorture.preempt_duration= [KNL] 6174 Set duration (in milliseconds) of preemptions 6175 by a high-priority FIFO real-time task. Set to 6176 zero (the default) to disable. The CPUs to 6177 preempt are selected randomly from the set that 6178 are online at a given point in time. Races with 6179 CPUs going offline are ignored, with that attempt 6180 at preemption skipped. 6181 6182 rcutorture.preempt_interval= [KNL] 6183 Set interval (in milliseconds, defaulting to one 6184 second) between preemptions by a high-priority 6185 FIFO real-time task. This delay is mediated 6186 by an hrtimer and is further fuzzed to avoid 6187 inadvertent synchronizations. 6188 6189 rcutorture.read_exit_burst= [KNL] 6190 The number of times in a given read-then-exit 6191 episode that a set of read-then-exit kthreads 6192 is spawned. 6193 6194 rcutorture.read_exit_delay= [KNL] 6195 The delay, in seconds, between successive 6196 read-then-exit testing episodes. 6197 6198 rcutorture.reader_flavor= [KNL] 6199 A bit mask indicating which readers to use. 6200 If there is more than one bit set, the readers 6201 are entered from low-order bit up, and are 6202 exited in the opposite order. For SRCU, the 6203 0x1 bit is normal readers, 0x2 NMI-safe readers, 6204 and 0x4 light-weight readers. 6205 6206 rcutorture.shuffle_interval= [KNL] 6207 Set task-shuffle interval (s). Shuffling tasks 6208 allows some CPUs to go into dyntick-idle mode 6209 during the rcutorture test. 6210 6211 rcutorture.shutdown_secs= [KNL] 6212 Set time (s) after boot system shutdown. This 6213 is useful for hands-off automated testing. 6214 6215 rcutorture.stall_cpu= [KNL] 6216 Duration of CPU stall (s) to test RCU CPU stall 6217 warnings, zero to disable. 6218 6219 rcutorture.stall_cpu_block= [KNL] 6220 Sleep while stalling if set. This will result 6221 in warnings from preemptible RCU in addition to 6222 any other stall-related activity. Note that 6223 in kernels built with CONFIG_PREEMPTION=n and 6224 CONFIG_PREEMPT_COUNT=y, this parameter will 6225 cause the CPU to pass through a quiescent state. 6226 Given CONFIG_PREEMPTION=n, this will suppress 6227 RCU CPU stall warnings, but will instead result 6228 in scheduling-while-atomic splats. 6229 6230 Use of this module parameter results in splats. 6231 6232 6233 rcutorture.stall_cpu_holdoff= [KNL] 6234 Time to wait (s) after boot before inducing stall. 6235 6236 rcutorture.stall_cpu_irqsoff= [KNL] 6237 Disable interrupts while stalling if set, but only 6238 on the first stall in the set. 6239 6240 rcutorture.stall_cpu_repeat= [KNL] 6241 Number of times to repeat the stall sequence, 6242 so that rcutorture.stall_cpu_repeat=3 will result 6243 in four stall sequences. 6244 6245 rcutorture.stall_gp_kthread= [KNL] 6246 Duration (s) of forced sleep within RCU 6247 grace-period kthread to test RCU CPU stall 6248 warnings, zero to disable. If both stall_cpu 6249 and stall_gp_kthread are specified, the 6250 kthread is starved first, then the CPU. 6251 6252 rcutorture.stall_only= [KNL] 6253 Shut off all rcutorture kthreads other than the 6254 RCU CPU stall-warning test kthreads. The purpose 6255 of this is to test production applications' 6256 reactions to CPU stalls, and with minimal 6257 additional overhead. Or you can omit the 6258 stall-warning tests as well and get a heavy 6259 no-op, your choice! 6260 6261 rcutorture.stat_interval= [KNL] 6262 Time (s) between statistics printk()s. 6263 6264 rcutorture.stutter= [KNL] 6265 Time (s) to stutter testing, for example, specifying 6266 five seconds causes the test to run for five seconds, 6267 wait for five seconds, and so on. This tests RCU's 6268 ability to transition abruptly to and from idle. 6269 6270 rcutorture.test_boost= [KNL] 6271 Test RCU priority boosting? 0=no, 1=maybe, 2=yes. 6272 "Maybe" means test if the RCU implementation 6273 under test support RCU priority boosting. 6274 6275 rcutorture.test_boost_duration= [KNL] 6276 Duration (s) of each individual boost test. 6277 6278 rcutorture.test_boost_holdoff= [KNL] 6279 Holdoff time (s) from start of test to the start 6280 of RCU priority-boost testing. Defaults to zero, 6281 that is, no holdoff. 6282 6283 rcutorture.test_boost_interval= [KNL] 6284 Interval (s) between each boost test. 6285 6286 rcutorture.test_no_idle_hz= [KNL] 6287 Test RCU's dyntick-idle handling. See also the 6288 rcutorture.shuffle_interval parameter. 6289 6290 rcutorture.torture_type= [KNL] 6291 Specify the RCU implementation to test. 6292 6293 rcutorture.verbose= [KNL] 6294 Enable additional printk() statements. 6295 6296 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL] 6297 Dump ftrace buffer after reporting RCU CPU 6298 stall warning. 6299 6300 rcupdate.rcu_cpu_stall_notifiers= [KNL] 6301 Provide RCU CPU stall notifiers, but see the 6302 warnings in the RCU_CPU_STALL_NOTIFIER Kconfig 6303 option's help text. TL;DR: You almost certainly 6304 do not want rcupdate.rcu_cpu_stall_notifiers. 6305 6306 rcupdate.rcu_cpu_stall_suppress= [KNL] 6307 Suppress RCU CPU stall warning messages. 6308 6309 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL] 6310 Suppress RCU CPU stall warning messages and 6311 rcutorture writer stall warnings that occur 6312 during early boot, that is, during the time 6313 before the init task is spawned. 6314 6315 rcupdate.rcu_cpu_stall_timeout= [KNL] 6316 Set timeout for RCU CPU stall warning messages. 6317 The value is in seconds and the maximum allowed 6318 value is 300 seconds. 6319 6320 rcupdate.rcu_exp_cpu_stall_timeout= [KNL] 6321 Set timeout for expedited RCU CPU stall warning 6322 messages. The value is in milliseconds 6323 and the maximum allowed value is 21000 6324 milliseconds. Please note that this value is 6325 adjusted to an arch timer tick resolution. 6326 Setting this to zero causes the value from 6327 rcupdate.rcu_cpu_stall_timeout to be used (after 6328 conversion from seconds to milliseconds). 6329 6330 rcupdate.rcu_cpu_stall_cputime= [KNL] 6331 Provide statistics on the cputime and count of 6332 interrupts and tasks during the sampling period. For 6333 multiple continuous RCU stalls, all sampling periods 6334 begin at half of the first RCU stall timeout. 6335 6336 rcupdate.rcu_exp_stall_task_details= [KNL] 6337 Print stack dumps of any tasks blocking the 6338 current expedited RCU grace period during an 6339 expedited RCU CPU stall warning. 6340 6341 rcupdate.rcu_expedited= [KNL] 6342 Use expedited grace-period primitives, for 6343 example, synchronize_rcu_expedited() instead 6344 of synchronize_rcu(). This reduces latency, 6345 but can increase CPU utilization, degrade 6346 real-time latency, and degrade energy efficiency. 6347 No effect on CONFIG_TINY_RCU kernels. 6348 6349 rcupdate.rcu_normal= [KNL] 6350 Use only normal grace-period primitives, 6351 for example, synchronize_rcu() instead of 6352 synchronize_rcu_expedited(). This improves 6353 real-time latency, CPU utilization, and 6354 energy efficiency, but can expose users to 6355 increased grace-period latency. This parameter 6356 overrides rcupdate.rcu_expedited. No effect on 6357 CONFIG_TINY_RCU kernels. 6358 6359 rcupdate.rcu_normal_after_boot= [KNL] 6360 Once boot has completed (that is, after 6361 rcu_end_inkernel_boot() has been invoked), use 6362 only normal grace-period primitives. No effect 6363 on CONFIG_TINY_RCU kernels. 6364 6365 But note that CONFIG_PREEMPT_RT=y kernels enables 6366 this kernel boot parameter, forcibly setting 6367 it to the value one, that is, converting any 6368 post-boot attempt at an expedited RCU grace 6369 period to instead use normal non-expedited 6370 grace-period processing. 6371 6372 rcupdate.rcu_task_collapse_debug= [KNL] 6373 Enable debugging prints that record when RCU Tasks 6374 and RCU Tasks Trace expand to per-CPU callback 6375 queuing and collapse back to CPU-0 queuing. 6376 This is default-disabled due to the fact that 6377 some workloads can make it quite noisy. 6378 6379 rcupdate.rcu_task_collapse_lim= [KNL] 6380 Set the maximum number of callbacks present 6381 at the beginning of a grace period that allows 6382 the RCU Tasks flavors to collapse back to using 6383 a single callback queue. This switching only 6384 occurs when rcupdate.rcu_task_enqueue_lim is 6385 set to the default value of -1. 6386 6387 rcupdate.rcu_task_contend_lim= [KNL] 6388 Set the minimum number of callback-queuing-time 6389 lock-contention events per jiffy required to 6390 cause the RCU Tasks flavors to switch to per-CPU 6391 callback queuing. This switching only occurs 6392 when rcupdate.rcu_task_enqueue_lim is set to 6393 the default value of -1. 6394 6395 rcupdate.rcu_task_enqueue_lim= [KNL] 6396 Set the number of callback queues to use for the 6397 RCU Tasks family of RCU flavors. The default 6398 of -1 allows this to be automatically (and 6399 dynamically) adjusted. This parameter is intended 6400 for use in testing. 6401 6402 rcupdate.rcu_task_lazy_lim= [KNL] 6403 Number of callbacks on a given CPU that will 6404 cancel laziness on that CPU. Use -1 to disable 6405 cancellation of laziness, but be advised that 6406 doing so increases the danger of OOM due to 6407 callback flooding. 6408 6409 rcupdate.rcu_task_stall_info= [KNL] 6410 Set initial timeout in jiffies for RCU task stall 6411 informational messages, which give some indication 6412 of the problem for those not patient enough to 6413 wait for ten minutes. Informational messages are 6414 only printed prior to the stall-warning message 6415 for a given grace period. Disable with a value 6416 less than or equal to zero. Defaults to ten 6417 seconds. A change in value does not take effect 6418 until the beginning of the next grace period. 6419 6420 rcupdate.rcu_task_stall_info_mult= [KNL] 6421 Multiplier for time interval between successive 6422 RCU task stall informational messages for a given 6423 RCU tasks grace period. This value is clamped 6424 to one through ten, inclusive. It defaults to 6425 the value three, so that the first informational 6426 message is printed 10 seconds into the grace 6427 period, the second at 40 seconds, the third at 6428 160 seconds, and then the stall warning at 600 6429 seconds would prevent a fourth at 640 seconds. 6430 6431 rcupdate.rcu_task_stall_timeout= [KNL] 6432 Set timeout in jiffies for RCU task stall 6433 warning messages. Disable with a value less 6434 than or equal to zero. Defaults to ten minutes. 6435 A change in value does not take effect until 6436 the beginning of the next grace period. 6437 6438 rcupdate.rcu_tasks_lazy_ms= [KNL] 6439 Set timeout in milliseconds RCU Tasks asynchronous 6440 callback batching for call_rcu_tasks(). 6441 A negative value will take the default. A value 6442 of zero will disable batching. Batching is 6443 always disabled for synchronize_rcu_tasks(). 6444 6445 rcupdate.rcu_self_test= [KNL] 6446 Run the RCU early boot self tests 6447 6448 rdinit= [KNL] 6449 Format: <full_path> 6450 Run specified binary instead of /init from the ramdisk, 6451 used for early userspace startup. See initrd. 6452 6453 rdrand= [X86,EARLY] 6454 force - Override the decision by the kernel to hide the 6455 advertisement of RDRAND support (this affects 6456 certain AMD processors because of buggy BIOS 6457 support, specifically around the suspend/resume 6458 path). 6459 6460 rdt= [HW,X86,RDT] 6461 Turn on/off individual RDT features. List is: 6462 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp, 6463 mba, smba, bmec, abmc, sdciae, energy[:guid], 6464 perf[:guid]. 6465 E.g. to turn on cmt and turn off mba use: 6466 rdt=cmt,!mba 6467 To turn off all energy telemetry monitoring and ensure that 6468 perf telemetry monitoring associated with guid 0x12345 6469 is enabled use: 6470 rdt=!energy,perf:0x12345 6471 6472 reboot= [KNL] 6473 Format (x86 or x86_64): 6474 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] | d[efault] \ 6475 [[,]s[mp]####] \ 6476 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \ 6477 [[,]f[orce]] 6478 Where reboot_mode is one of warm (soft) or cold (hard) or gpio 6479 (prefix with 'panic_' to set mode for panic 6480 reboot only), 6481 reboot_type is one of bios, acpi, kbd, triple, efi, or pci, 6482 reboot_force is either force or not specified, 6483 reboot_cpu is s[mp]#### with #### being the processor 6484 to be used for rebooting. 6485 6486 acpi 6487 Use the ACPI RESET_REG in the FADT. If ACPI is not 6488 configured or the ACPI reset does not work, the reboot 6489 path attempts the reset using the keyboard controller. 6490 6491 bios 6492 Use the CPU reboot vector for warm reset 6493 6494 cold 6495 Set the cold reboot flag 6496 6497 default 6498 There are some built-in platform specific "quirks" 6499 - you may see: "reboot: <name> series board detected. 6500 Selecting <type> for reboots." In the case where you 6501 think the quirk is in error (e.g. you have newer BIOS, 6502 or newer board) using this option will ignore the 6503 built-in quirk table, and use the generic default 6504 reboot actions. 6505 6506 efi 6507 Use efi reset_system runtime service. If EFI is not 6508 configured or the EFI reset does not work, the reboot 6509 path attempts the reset using the keyboard controller. 6510 6511 force 6512 Don't stop other CPUs on reboot. This can make reboot 6513 more reliable in some cases. 6514 6515 kbd 6516 Use the keyboard controller. cold reset (default) 6517 6518 pci 6519 Use a write to the PCI config space register 0xcf9 to 6520 trigger reboot. 6521 6522 triple 6523 Force a triple fault (init) 6524 6525 warm 6526 Don't set the cold reboot flag 6527 6528 Using warm reset will be much faster especially on big 6529 memory systems because the BIOS will not go through 6530 the memory check. Disadvantage is that not all 6531 hardware will be completely reinitialized on reboot so 6532 there may be boot problems on some systems. 6533 6534 6535 refscale.holdoff= [KNL] 6536 Set test-start holdoff period. The purpose of 6537 this parameter is to delay the start of the 6538 test until boot completes in order to avoid 6539 interference. 6540 6541 refscale.lookup_instances= [KNL] 6542 Number of data elements to use for the forms of 6543 SLAB_TYPESAFE_BY_RCU testing. A negative number 6544 is negated and multiplied by nr_cpu_ids, while 6545 zero specifies nr_cpu_ids. 6546 6547 refscale.loops= [KNL] 6548 Set the number of loops over the synchronization 6549 primitive under test. Increasing this number 6550 reduces noise due to loop start/end overhead, 6551 but the default has already reduced the per-pass 6552 noise to a handful of picoseconds on ca. 2020 6553 x86 laptops. 6554 6555 refscale.nreaders= [KNL] 6556 Set number of readers. The default value of -1 6557 selects N, where N is roughly 75% of the number 6558 of CPUs. A value of zero is an interesting choice. 6559 6560 refscale.nruns= [KNL] 6561 Set number of runs, each of which is dumped onto 6562 the console log. 6563 6564 refscale.readdelay= [KNL] 6565 Set the read-side critical-section duration, 6566 measured in microseconds. 6567 6568 refscale.scale_type= [KNL] 6569 Specify the read-protection implementation to test. 6570 6571 refscale.shutdown= [KNL] 6572 Shut down the system at the end of the performance 6573 test. This defaults to 1 (shut it down) when 6574 refscale is built into the kernel and to 0 (leave 6575 it running) when refscale is built as a module. 6576 6577 refscale.verbose= [KNL] 6578 Enable additional printk() statements. 6579 6580 refscale.verbose_batched= [KNL] 6581 Batch the additional printk() statements. If zero 6582 (the default) or negative, print everything. Otherwise, 6583 print every Nth verbose statement, where N is the value 6584 specified. 6585 6586 regulator_ignore_unused 6587 [REGULATOR] 6588 Prevents regulator framework from disabling regulators 6589 that are unused, due no driver claiming them. This may 6590 be useful for debug and development, but should not be 6591 needed on a platform with proper driver support. 6592 6593 relax_domain_level= 6594 [KNL, SMP] Set scheduler's default relax_domain_level. 6595 See Documentation/admin-guide/cgroup-v1/cpusets.rst. 6596 6597 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory 6598 Format: <base1>,<size1>[,<base2>,<size2>,...] 6599 Reserve I/O ports or memory so the kernel won't use 6600 them. If <base> is less than 0x10000, the region 6601 is assumed to be I/O ports; otherwise it is memory. 6602 6603 reserve_mem= [MM] 6604 Format: nn[KMG]:<align>:<label> 6605 Reserve physical memory and label it with a name that 6606 other subsystems can use to access it. This is typically 6607 used for systems that do not wipe the RAM, and this command 6608 line will try to reserve the same physical memory on 6609 soft reboots. Note, it is not guaranteed to be the same 6610 location. For example, if anything about the system changes 6611 or if booting a different kernel. It can also fail if KASLR 6612 places the kernel at the location of where the RAM reservation 6613 was from a previous boot, the new reservation will be at a 6614 different location. 6615 Any subsystem using this feature must add a way to verify 6616 that the contents of the physical memory is from a previous 6617 boot, as there may be cases where the memory will not be 6618 located at the same location. 6619 6620 The format is size:align:label for example, to request 6621 12 megabytes of 4096 alignment for ramoops: 6622 6623 reserve_mem=12M:4096:oops ramoops.mem_name=oops 6624 6625 Both the size and the alignment must be multiples of 6626 the page size. 6627 6628 reservetop= [X86-32,EARLY] 6629 Format: nn[KMG] 6630 Reserves a hole at the top of the kernel virtual 6631 address space. 6632 6633 reset_devices [KNL] Force drivers to reset the underlying device 6634 during initialization. 6635 6636 resume= [SWSUSP] 6637 Specify the partition device for software suspend 6638 Format: 6639 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>} 6640 6641 resume_offset= [SWSUSP] 6642 Specify the offset from the beginning of the partition 6643 given by "resume=" at which the swap header is located, 6644 in <PAGE_SIZE> units (needed only for swap files). 6645 See Documentation/power/swsusp-and-swap-files.rst 6646 6647 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to 6648 read the resume files 6649 6650 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up. 6651 Useful for devices that are detected asynchronously 6652 (e.g. USB and MMC devices). 6653 6654 retain_initrd [RAM] Keep initrd memory after extraction. After boot, it will 6655 be accessible via /sys/firmware/initrd. 6656 6657 retbleed= [X86] Control mitigation of RETBleed (Arbitrary 6658 Speculative Code Execution with Return Instructions) 6659 vulnerability. 6660 6661 AMD-based UNRET and IBPB mitigations alone do not stop 6662 sibling threads from influencing the predictions of other 6663 sibling threads. For that reason, STIBP is used on pro- 6664 cessors that support it, and mitigate SMT on processors 6665 that don't. 6666 6667 off - no mitigation 6668 auto - automatically select a mitigation 6669 auto,nosmt - automatically select a mitigation, 6670 disabling SMT if necessary for 6671 the full mitigation (only on Zen1 6672 and older without STIBP). 6673 ibpb - On AMD, mitigate short speculation 6674 windows on basic block boundaries too. 6675 Safe, highest perf impact. It also 6676 enables STIBP if present. Not suitable 6677 on Intel. 6678 ibpb,nosmt - Like "ibpb" above but will disable SMT 6679 when STIBP is not available. This is 6680 the alternative for systems which do not 6681 have STIBP. 6682 unret - Force enable untrained return thunks, 6683 only effective on AMD f15h-f17h based 6684 systems. 6685 unret,nosmt - Like unret, but will disable SMT when STIBP 6686 is not available. This is the alternative for 6687 systems which do not have STIBP. 6688 6689 Selecting 'auto' will choose a mitigation method at run 6690 time according to the CPU. 6691 6692 Not specifying this option is equivalent to retbleed=auto. 6693 6694 rfkill.default_state= 6695 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm, 6696 etc. communication is blocked by default. 6697 1 Unblocked. 6698 6699 rfkill.master_switch_mode= 6700 0 The "airplane mode" button does nothing. 6701 1 The "airplane mode" button toggles between everything 6702 blocked and the previous configuration. 6703 2 The "airplane mode" button toggles between everything 6704 blocked and everything unblocked. 6705 6706 ring3mwait=disable 6707 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported 6708 CPUs. 6709 6710 riscv_isa_fallback [RISCV,EARLY] 6711 When CONFIG_RISCV_ISA_FALLBACK is not enabled, permit 6712 falling back to detecting extension support by parsing 6713 "riscv,isa" property on devicetree systems when the 6714 replacement properties are not found. See the Kconfig 6715 entry for RISCV_ISA_FALLBACK. 6716 6717 riscv_nousercfi= 6718 all Disable user CFI ABI to userspace even if cpu extension 6719 are available. 6720 bcfi Disable user backward CFI ABI to userspace even if 6721 the shadow stack extension is available. 6722 fcfi Disable user forward CFI ABI to userspace even if the 6723 landing pad extension is available. 6724 6725 ro [KNL] Mount root device read-only on boot 6726 6727 rodata= [KNL,EARLY] 6728 on Mark read-only kernel memory as read-only (default). 6729 off Leave read-only kernel memory writable for debugging. 6730 noalias Mark read-only kernel memory as read-only but retain 6731 writable aliases in the direct map for regions outside 6732 of the kernel image. [arm64] 6733 6734 rockchip.usb_uart 6735 [EARLY] 6736 Enable the uart passthrough on the designated usb port 6737 on Rockchip SoCs. When active, the signals of the 6738 debug-uart get routed to the D+ and D- pins of the usb 6739 port and the regular usb controller gets disabled. 6740 6741 root= [KNL] Root filesystem 6742 Usually this is a block device specifier of some kind, 6743 see the early_lookup_bdev comment in 6744 block/early-lookup.c for details. 6745 Alternatively this can be "ram" for the legacy initial 6746 ramdisk, "nfs" and "cifs" for root on a network file 6747 system, or "mtd" and "ubi" for mounting from raw flash. 6748 6749 rootdelay= [KNL] Delay (in seconds) to pause before attempting to 6750 mount the root filesystem 6751 6752 rootflags= [KNL] Set root filesystem mount option string 6753 6754 rseq_slice_ext= [KNL] RSEQ based time slice extension 6755 Format: boolean 6756 Control enablement of RSEQ based time slice extension. 6757 Default is 'on'. 6758 6759 initramfs_options= [KNL] 6760 Specify mount options for the initramfs mount. 6761 6762 rootfstype= [KNL] Set root filesystem type 6763 6764 rootwait [KNL] Wait (indefinitely) for root device to show up. 6765 Useful for devices that are detected asynchronously 6766 (e.g. USB and MMC devices). 6767 6768 rootwait= [KNL] Maximum time (in seconds) to wait for root device 6769 to show up before attempting to mount the root 6770 filesystem. 6771 6772 rproc_mem=nn[KMG][@address] 6773 [KNL,ARM,CMA] Remoteproc physical memory block. 6774 Memory area to be used by remote processor image, 6775 managed by CMA. 6776 6777 rseq_debug= [KNL] Enable or disable restartable sequence 6778 debug mode. Defaults to CONFIG_RSEQ_DEBUG_DEFAULT_ENABLE. 6779 Format: <bool> 6780 6781 rt_group_sched= [KNL] Enable or disable SCHED_RR/FIFO group scheduling 6782 when CONFIG_RT_GROUP_SCHED=y. Defaults to 6783 !CONFIG_RT_GROUP_SCHED_DEFAULT_DISABLED. 6784 Format: <bool> 6785 6786 rw [KNL] Mount root device read-write on boot 6787 6788 S [KNL] Run init in single mode 6789 6790 s390_iommu= [HW,S390] 6791 Set s390 IOTLB flushing mode 6792 strict 6793 With strict flushing every unmap operation will result 6794 in an IOTLB flush. Default is lazy flushing before 6795 reuse, which is faster. Deprecated, equivalent to 6796 iommu.strict=1. 6797 6798 s390_iommu_aperture= [KNL,S390] 6799 Specifies the size of the per device DMA address space 6800 accessible through the DMA and IOMMU APIs as a decimal 6801 factor of the size of main memory. 6802 The default is 1 meaning that one can concurrently use 6803 as many DMA addresses as physical memory is installed, 6804 if supported by hardware, and thus map all of memory 6805 once. With a value of 2 one can map all of memory twice 6806 and so on. As a special case a factor of 0 imposes no 6807 restrictions other than those given by hardware at the 6808 cost of significant additional memory use for tables. 6809 6810 6811 sched_proxy_exec= [KNL] 6812 Enables or disables "proxy execution" style 6813 solution to mutex-based priority inversion. 6814 Format: <bool> 6815 6816 sched_smt_asym_packing= [KNL,SMP] 6817 Format: on 6818 Force asymmetric packing at the SMT scheduling domain. 6819 Idle CPU selection prefers siblings with a higher 6820 arch_asym_cpu_priority(). The default implementation 6821 prefers lower-numbered logical CPUs, but architecture 6822 overrides remain authoritative. Equal priorities do 6823 not establish a sibling preference. For example, x86 6824 with CONFIG_SCHED_MC_PRIO normally assigns the same 6825 core priority to both SMT siblings, so this option 6826 does not favor either sibling. 6827 6828 sched_verbose [KNL,EARLY] Enables verbose scheduler debug messages. 6829 6830 schedstats= [KNL,X86] Enable or disable scheduled statistics. 6831 Allowed values are enable and disable. This feature 6832 incurs a small amount of overhead in the scheduler 6833 but is useful for debugging and performance tuning. 6834 6835 sched_thermal_decay_shift= 6836 [Deprecated] 6837 [KNL, SMP] Set a decay shift for scheduler thermal 6838 pressure signal. Thermal pressure signal follows the 6839 default decay period of other scheduler pelt 6840 signals(usually 32 ms but configurable). Setting 6841 sched_thermal_decay_shift will left shift the decay 6842 period for the thermal pressure signal by the shift 6843 value. 6844 i.e. with the default pelt decay period of 32 ms 6845 sched_thermal_decay_shift thermal pressure decay pr 6846 1 64 ms 6847 2 128 ms 6848 and so on. 6849 Format: integer between 0 and 10 6850 Default is 0. 6851 6852 scftorture.holdoff= [KNL] 6853 Number of seconds to hold off before starting 6854 test. Defaults to zero for module insertion and 6855 to 10 seconds for built-in smp_call_function() 6856 tests. 6857 6858 scftorture.longwait= [KNL] 6859 Request ridiculously long waits randomly selected 6860 up to the chosen limit in seconds. Zero (the 6861 default) disables this feature. Please note 6862 that requesting even small non-zero numbers of 6863 seconds can result in RCU CPU stall warnings, 6864 softlockup complaints, and so on. 6865 6866 scftorture.nthreads= [KNL] 6867 Number of kthreads to spawn to invoke the 6868 smp_call_function() family of functions. 6869 The default of -1 specifies a number of kthreads 6870 equal to the number of CPUs. 6871 6872 scftorture.onoff_holdoff= [KNL] 6873 Number seconds to wait after the start of the 6874 test before initiating CPU-hotplug operations. 6875 6876 scftorture.onoff_interval= [KNL] 6877 Number seconds to wait between successive 6878 CPU-hotplug operations. Specifying zero (which 6879 is the default) disables CPU-hotplug operations. 6880 6881 scftorture.shutdown_secs= [KNL] 6882 The number of seconds following the start of the 6883 test after which to shut down the system. The 6884 default of zero avoids shutting down the system. 6885 Non-zero values are useful for automated tests. 6886 6887 scftorture.stat_interval= [KNL] 6888 The number of seconds between outputting the 6889 current test statistics to the console. A value 6890 of zero disables statistics output. 6891 6892 scftorture.stutter_cpus= [KNL] 6893 The number of jiffies to wait between each change 6894 to the set of CPUs under test. 6895 6896 scftorture.use_cpus_read_lock= [KNL] 6897 Use use_cpus_read_lock() instead of the default 6898 preempt_disable() to disable CPU hotplug 6899 while invoking one of the smp_call_function*() 6900 functions. 6901 6902 scftorture.verbose= [KNL] 6903 Enable additional printk() statements. 6904 6905 scftorture.weight_single= [KNL] 6906 The probability weighting to use for the 6907 smp_call_function_single() function with a zero 6908 "wait" parameter. A value of -1 selects the 6909 default if all other weights are -1. However, 6910 if at least one weight has some other value, a 6911 value of -1 will instead select a weight of zero. 6912 6913 scftorture.weight_single_wait= [KNL] 6914 The probability weighting to use for the 6915 smp_call_function_single() function with a 6916 non-zero "wait" parameter. See weight_single. 6917 6918 scftorture.weight_many= [KNL] 6919 The probability weighting to use for the 6920 smp_call_function_many() function with a zero 6921 "wait" parameter. See weight_single. 6922 Note well that setting a high probability for 6923 this weighting can place serious IPI load 6924 on the system. 6925 6926 scftorture.weight_many_wait= [KNL] 6927 The probability weighting to use for the 6928 smp_call_function_many() function with a 6929 non-zero "wait" parameter. See weight_single 6930 and weight_many. 6931 6932 scftorture.weight_all= [KNL] 6933 The probability weighting to use for the 6934 smp_call_function_all() function with a zero 6935 "wait" parameter. See weight_single and 6936 weight_many. 6937 6938 scftorture.weight_all_wait= [KNL] 6939 The probability weighting to use for the 6940 smp_call_function_all() function with a 6941 non-zero "wait" parameter. See weight_single 6942 and weight_many. 6943 6944 sdw_mclk_divider=[SDW] 6945 Specify the MCLK divider for Intel SoundWire buses in 6946 case the BIOS does not provide the clock rate properly. 6947 6948 skew_tick= [KNL,EARLY] Offset the periodic timer tick per cpu to mitigate 6949 xtime_lock contention on larger systems, and/or RCU lock 6950 contention on all systems with CONFIG_MAXSMP set. 6951 Format: { "0" | "1" } 6952 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1") 6953 1 -- enable. 6954 Note: increases power consumption, thus should only be 6955 enabled if running jitter sensitive (HPC/RT) workloads. 6956 6957 security= [SECURITY] Choose a legacy "major" security module to 6958 enable at boot. This has been deprecated by the 6959 "lsm=" parameter. 6960 6961 selinux= [SELINUX] Disable or enable SELinux at boot time. 6962 Format: { "0" | "1" } 6963 See security/selinux/Kconfig help text. 6964 0 -- disable. 6965 1 -- enable. 6966 Default value is 1. 6967 6968 serialnumber [BUGS=X86-32] 6969 6970 sev=option[,option...] [X86-64] 6971 6972 debug 6973 Enable debug messages. 6974 6975 nosnp 6976 Do not enable SEV-SNP (applies to host/hypervisor 6977 only). Setting 'nosnp' avoids the RMP check overhead 6978 in memory accesses when users do not want to run 6979 SEV-SNP guests. 6980 6981 shapers= [NET] 6982 Maximal number of shapers. 6983 6984 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller 6985 Limit apic dumping. The parameter defines the maximal 6986 number of local apics being dumped. Also it is possible 6987 to set it to "all" by meaning -- no limit here. 6988 Format: { 1 (default) | 2 | ... | all }. 6989 The parameter valid if only apic=debug or 6990 apic=verbose is specified. 6991 Example: apic=debug show_lapic=all 6992 6993 slab_debug[=options[,slabs][;[options[,slabs]]...]] [MM] 6994 Enabling slab_debug allows one to determine the 6995 culprit if slab objects become corrupted. Enabling 6996 slab_debug can create guard zones around objects and 6997 may poison objects when not in use. Also tracks the 6998 last alloc / free. For more information see 6999 Documentation/admin-guide/mm/slab.rst. 7000 (slub_debug legacy name also accepted for now) 7001 7002 Using this option implies the "no_hash_pointers" 7003 option which can be undone by adding the 7004 "hash_pointers=always" option. 7005 7006 slab_max_order= [MM] 7007 Determines the maximum allowed order for slabs. 7008 A high setting may cause OOMs due to memory 7009 fragmentation. For more information see 7010 Documentation/admin-guide/mm/slab.rst. 7011 (slub_max_order legacy name also accepted for now) 7012 7013 slab_merge [MM] 7014 Enable merging of slabs with similar size when the 7015 kernel is built without CONFIG_SLAB_MERGE_DEFAULT. 7016 (slub_merge legacy name also accepted for now) 7017 7018 slab_min_objects= [MM] 7019 The minimum number of objects per slab. SLUB will 7020 increase the slab order up to slab_max_order to 7021 generate a sufficiently large slab able to contain 7022 the number of objects indicated. The higher the number 7023 of objects the smaller the overhead of tracking slabs 7024 and the less frequently locks need to be acquired. 7025 For more information see 7026 Documentation/admin-guide/mm/slab.rst. 7027 (slub_min_objects legacy name also accepted for now) 7028 7029 slab_min_order= [MM] 7030 Determines the minimum page order for slabs. Must be 7031 lower or equal to slab_max_order. For more information see 7032 Documentation/admin-guide/mm/slab.rst. 7033 (slub_min_order legacy name also accepted for now) 7034 7035 slab_nomerge [MM] 7036 Disable merging of slabs with similar size. May be 7037 necessary if there is some reason to distinguish 7038 allocs to different slabs, especially in hardened 7039 environments where the risk of heap overflows and 7040 layout control by attackers can usually be 7041 frustrated by disabling merging. This will reduce 7042 most of the exposure of a heap attack to a single 7043 cache (risks via metadata attacks are mostly 7044 unchanged). Debug options disable merging on their 7045 own. 7046 For more information see 7047 Documentation/admin-guide/mm/slab.rst. 7048 (slub_nomerge legacy name also accepted for now) 7049 7050 slab_strict_numa [MM] 7051 Support memory policies on a per object level 7052 in the slab allocator. The default is for memory 7053 policies to be applied at the folio level when 7054 a new folio is needed or a partial folio is 7055 retrieved from the lists. Increases overhead 7056 in the slab fastpaths but gains more accurate 7057 NUMA kernel object placement which helps with slow 7058 interconnects in NUMA systems. 7059 7060 slram= [HW,MTD] 7061 7062 smart2= [HW] 7063 Format: <io1>[,<io2>[,...,<io8>]] 7064 7065 smp.csd_lock_timeout= [KNL] 7066 Specify the period of time in milliseconds 7067 that smp_call_function() and friends will wait 7068 for a CPU to release the CSD lock. This is 7069 useful when diagnosing bugs involving CPUs 7070 disabling interrupts for extended periods 7071 of time. Defaults to 5,000 milliseconds, and 7072 setting a value of zero disables this feature. 7073 This feature may be more efficiently disabled 7074 using the csdlock_debug- kernel parameter. 7075 7076 smp.panic_on_ipistall= [KNL] 7077 If a csd_lock_timeout extends for more than 7078 the specified number of milliseconds, panic the 7079 system. By default, let CSD-lock acquisition 7080 take as long as they take. Specifying 300,000 7081 for this value provides a 5-minute timeout. 7082 7083 smt= [KNL,MIPS,S390,EARLY] Set the maximum number of threads 7084 (logical CPUs) to use per physical CPU on systems 7085 capable of symmetric multithreading (SMT). Will 7086 be capped to the actual hardware limit. 7087 Format: <integer> 7088 Default: -1 (no limit) 7089 7090 softlockup_panic= 7091 [KNL] Should the soft-lockup detector generate panics. 7092 Format: <int> 7093 7094 A value of non-zero instructs the soft-lockup detector 7095 to panic the machine when a soft-lockup duration exceeds 7096 N thresholds. It is also controlled by the kernel.softlockup_panic 7097 sysctl and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the 7098 respective build-time switch to that functionality. 7099 7100 softlockup_all_cpu_backtrace= 7101 [KNL] Should the soft-lockup detector generate 7102 backtraces on all cpus. 7103 Format: 0 | 1 7104 7105 sonypi.*= [HW] Sony Programmable I/O Control Device driver 7106 See Documentation/admin-guide/laptops/sonypi.rst 7107 7108 spectre_bhi= [X86] Control mitigation of Branch History Injection 7109 (BHI) vulnerability. This setting affects the 7110 deployment of the HW BHI control and the SW BHB 7111 clearing sequence. 7112 7113 on - (default) Enable the HW or SW mitigation as 7114 needed. This protects the kernel from 7115 both syscalls and VMs. 7116 vmexit - On systems which don't have the HW mitigation 7117 available, enable the SW mitigation on vmexit 7118 ONLY. On such systems, the host kernel is 7119 protected from VM-originated BHI attacks, but 7120 may still be vulnerable to syscall attacks. 7121 off - Disable the mitigation. 7122 7123 spectre_v2= [X86,EARLY] Control mitigation of Spectre variant 2 7124 (indirect branch speculation) vulnerability. 7125 The default operation protects the kernel from 7126 user space attacks. 7127 7128 on - unconditionally enable, implies 7129 spectre_v2_user=on 7130 off - unconditionally disable, implies 7131 spectre_v2_user=off 7132 auto - kernel detects whether your CPU model is 7133 vulnerable 7134 7135 Selecting 'on' will, and 'auto' may, choose a 7136 mitigation method at run time according to the 7137 CPU, the available microcode, the setting of the 7138 CONFIG_MITIGATION_RETPOLINE configuration option, 7139 and the compiler with which the kernel was built. 7140 7141 Selecting 'on' will also enable the mitigation 7142 against user space to user space task attacks. 7143 Selecting specific mitigation does not force enable 7144 user mitigations. 7145 7146 Selecting 'off' will disable both the kernel and 7147 the user space protections. 7148 7149 Specific mitigations can also be selected manually: 7150 7151 retpoline - replace indirect branches 7152 retpoline,generic - Retpolines 7153 retpoline,lfence - LFENCE; indirect branch 7154 retpoline,amd - alias for retpoline,lfence 7155 eibrs - Enhanced/Auto IBRS 7156 eibrs,retpoline - Enhanced/Auto IBRS + Retpolines 7157 eibrs,lfence - Enhanced/Auto IBRS + LFENCE 7158 ibrs - use IBRS to protect kernel 7159 7160 Not specifying this option is equivalent to 7161 spectre_v2=auto. 7162 7163 spectre_v2_user= 7164 [X86] Control mitigation of Spectre variant 2 7165 (indirect branch speculation) vulnerability between 7166 user space tasks 7167 7168 on - Unconditionally enable mitigations. Is 7169 enforced by spectre_v2=on 7170 7171 off - Unconditionally disable mitigations. Is 7172 enforced by spectre_v2=off 7173 7174 prctl - Indirect branch speculation is enabled, 7175 but mitigation can be enabled via prctl 7176 per thread. The mitigation control state 7177 is inherited on fork. 7178 7179 prctl,ibpb 7180 - Like "prctl" above, but only STIBP is 7181 controlled per thread. IBPB is issued 7182 always when switching between different user 7183 space processes. 7184 7185 seccomp 7186 - Same as "prctl" above, but all seccomp 7187 threads will enable the mitigation unless 7188 they explicitly opt out. 7189 7190 seccomp,ibpb 7191 - Like "seccomp" above, but only STIBP is 7192 controlled per thread. IBPB is issued 7193 always when switching between different 7194 user space processes. 7195 7196 auto - Kernel selects the mitigation depending on 7197 the available CPU features and vulnerability. 7198 7199 Default mitigation: "prctl" 7200 7201 Not specifying this option is equivalent to 7202 spectre_v2_user=auto. 7203 7204 spec_rstack_overflow= 7205 [X86,EARLY] Control RAS overflow mitigation on AMD Zen CPUs 7206 7207 off - Disable mitigation 7208 microcode - Enable microcode mitigation only 7209 safe-ret - Enable sw-only safe RET mitigation (default) 7210 ibpb - Enable mitigation by issuing IBPB on 7211 kernel entry 7212 ibpb-vmexit - Issue IBPB only on VMEXIT 7213 (cloud-specific mitigation) 7214 7215 spec_store_bypass_disable= 7216 [HW,EARLY] Control Speculative Store Bypass (SSB) Disable mitigation 7217 (Speculative Store Bypass vulnerability) 7218 7219 Certain CPUs are vulnerable to an exploit against a 7220 a common industry wide performance optimization known 7221 as "Speculative Store Bypass" in which recent stores 7222 to the same memory location may not be observed by 7223 later loads during speculative execution. The idea 7224 is that such stores are unlikely and that they can 7225 be detected prior to instruction retirement at the 7226 end of a particular speculation execution window. 7227 7228 In vulnerable processors, the speculatively forwarded 7229 store can be used in a cache side channel attack, for 7230 example to read memory to which the attacker does not 7231 directly have access (e.g. inside sandboxed code). 7232 7233 This parameter controls whether the Speculative Store 7234 Bypass optimization is used. 7235 7236 On x86 the options are: 7237 7238 on - Unconditionally disable Speculative Store Bypass 7239 off - Unconditionally enable Speculative Store Bypass 7240 auto - Kernel detects whether the CPU model contains an 7241 implementation of Speculative Store Bypass and 7242 picks the most appropriate mitigation. If the 7243 CPU is not vulnerable, "off" is selected. If the 7244 CPU is vulnerable the default mitigation is 7245 architecture and Kconfig dependent. See below. 7246 prctl - Control Speculative Store Bypass per thread 7247 via prctl. Speculative Store Bypass is enabled 7248 for a process by default. The state of the control 7249 is inherited on fork. 7250 seccomp - Same as "prctl" above, but all seccomp threads 7251 will disable SSB unless they explicitly opt out. 7252 7253 Default mitigations: 7254 X86: "prctl" 7255 7256 On powerpc the options are: 7257 7258 on,auto - On Power8 and Power9 insert a store-forwarding 7259 barrier on kernel entry and exit. On Power7 7260 perform a software flush on kernel entry and 7261 exit. 7262 off - No action. 7263 7264 Not specifying this option is equivalent to 7265 spec_store_bypass_disable=auto. 7266 7267 split_llc= 7268 [X86,EARLY] Split the LLC N-ways 7269 7270 When set, the LLC is split this many ways by matching 7271 'core_id % n'. This is setup before SMP bringup and 7272 used during SMP bringup before it knows the full 7273 topology. If your core count doesn't nicely divide by 7274 the number given, you get to keep the pieces. 7275 7276 This is mostly a debug feature to emulate multiple LLCs 7277 on hardware that only have a single LLC. 7278 7279 split_lock_detect= 7280 [X86] Enable split lock detection or bus lock detection 7281 7282 When enabled (and if hardware support is present), atomic 7283 instructions that access data across cache line 7284 boundaries will result in an alignment check exception 7285 for split lock detection or a debug exception for 7286 bus lock detection. 7287 7288 off - not enabled 7289 7290 warn - the kernel will emit rate-limited warnings 7291 about applications triggering the #AC 7292 exception or the #DB exception. This mode is 7293 the default on CPUs that support split lock 7294 detection or bus lock detection. Default 7295 behavior is by #AC if both features are 7296 enabled in hardware. 7297 7298 fatal - the kernel will send SIGBUS to applications 7299 that trigger the #AC exception or the #DB 7300 exception. Default behavior is by #AC if 7301 both features are enabled in hardware. 7302 7303 ratelimit:N - 7304 Set system wide rate limit to N bus locks 7305 per second for bus lock detection. 7306 0 < N <= 1000. 7307 7308 N/A for split lock detection. 7309 7310 7311 If an #AC exception is hit in the kernel or in 7312 firmware (i.e. not while executing in user mode) 7313 the kernel will oops in either "warn" or "fatal" 7314 mode. 7315 7316 #DB exception for bus lock is triggered only when 7317 CPL > 0. 7318 7319 srbds= [X86,INTEL,EARLY] 7320 Control the Special Register Buffer Data Sampling 7321 (SRBDS) mitigation. 7322 7323 Certain CPUs are vulnerable to an MDS-like 7324 exploit which can leak bits from the random 7325 number generator. 7326 7327 By default, this issue is mitigated by 7328 microcode. However, the microcode fix can cause 7329 the RDRAND and RDSEED instructions to become 7330 much slower. Among other effects, this will 7331 result in reduced throughput from /dev/urandom. 7332 7333 The microcode mitigation can be disabled with 7334 the following option: 7335 7336 off: Disable mitigation and remove 7337 performance impact to RDRAND and RDSEED 7338 7339 srcutree.big_cpu_lim [KNL] 7340 Specifies the number of CPUs constituting a 7341 large system, such that srcu_struct structures 7342 should immediately allocate an srcu_node array. 7343 This kernel-boot parameter defaults to 128, 7344 but takes effect only when the low-order four 7345 bits of srcutree.convert_to_big is equal to 3 7346 (decide at boot). 7347 7348 srcutree.convert_to_big [KNL] 7349 Specifies under what conditions an SRCU tree 7350 srcu_struct structure will be converted to big 7351 form, that is, with an rcu_node tree: 7352 7353 0: Never. 7354 1: At init_srcu_struct() time. 7355 2: When rcutorture decides to. 7356 3: Decide at boot time (default). 7357 0x1X: Above plus if high contention. 7358 7359 Either way, the srcu_node tree will be sized based 7360 on the actual runtime number of CPUs (nr_cpu_ids) 7361 instead of the compile-time CONFIG_NR_CPUS. 7362 7363 srcutree.counter_wrap_check [KNL] 7364 Specifies how frequently to check for 7365 grace-period sequence counter wrap for the 7366 srcu_data structure's ->srcu_gp_seq_needed field. 7367 The greater the number of bits set in this kernel 7368 parameter, the less frequently counter wrap will 7369 be checked for. Note that the bottom two bits 7370 are ignored. 7371 7372 srcutree.exp_holdoff [KNL] 7373 Specifies how many nanoseconds must elapse 7374 since the end of the last SRCU grace period for 7375 a given srcu_struct until the next normal SRCU 7376 grace period will be considered for automatic 7377 expediting. Set to zero to disable automatic 7378 expediting. 7379 7380 srcutree.srcu_max_nodelay [KNL] 7381 Specifies the number of no-delay instances 7382 per jiffy for which the SRCU grace period 7383 worker thread will be rescheduled with zero 7384 delay. Beyond this limit, worker thread will 7385 be rescheduled with a sleep delay of one jiffy. 7386 7387 srcutree.srcu_max_nodelay_phase [KNL] 7388 Specifies the per-grace-period phase, number of 7389 non-sleeping polls of readers. Beyond this limit, 7390 grace period worker thread will be rescheduled 7391 with a sleep delay of one jiffy, between each 7392 rescan of the readers, for a grace period phase. 7393 7394 srcutree.srcu_retry_check_delay [KNL] 7395 Specifies number of microseconds of non-sleeping 7396 delay between each non-sleeping poll of readers. 7397 7398 srcutree.small_contention_lim [KNL] 7399 Specifies the number of update-side contention 7400 events per jiffy will be tolerated before 7401 initiating a conversion of an srcu_struct 7402 structure to big form. Note that the value of 7403 srcutree.convert_to_big must have the 0x10 bit 7404 set for contention-based conversions to occur. 7405 7406 ssbd= [ARM64,HW,EARLY] 7407 Speculative Store Bypass Disable control 7408 7409 On CPUs that are vulnerable to the Speculative 7410 Store Bypass vulnerability and offer a 7411 firmware based mitigation, this parameter 7412 indicates how the mitigation should be used: 7413 7414 force-on: Unconditionally enable mitigation for 7415 for both kernel and userspace 7416 force-off: Unconditionally disable mitigation for 7417 for both kernel and userspace 7418 kernel: Always enable mitigation in the 7419 kernel, and offer a prctl interface 7420 to allow userspace to register its 7421 interest in being mitigated too. 7422 7423 stack_guard_gap= [MM] 7424 override the default stack gap protection. The value 7425 is in page units and it defines how many pages prior 7426 to (for stacks growing down) resp. after (for stacks 7427 growing up) the main stack are reserved for no other 7428 mapping. Default value is 256 pages. 7429 7430 stack_depot_disable= [KNL,EARLY] 7431 Setting this to true through kernel command line will 7432 disable the stack depot thereby saving the static memory 7433 consumed by the stack hash table. By default this is set 7434 to false. 7435 7436 stack_depot_max_pools= [KNL,EARLY] 7437 Specify the maximum number of pools to use for storing 7438 stack traces. Pools are allocated on-demand up to this 7439 limit. Default value is 8191 pools. 7440 7441 stacktrace [FTRACE] 7442 Enable the stack tracer on boot up. 7443 7444 stacktrace_filter=[function-list] 7445 [FTRACE] Limit the functions that the stack tracer 7446 will trace at boot up. function-list is a comma-separated 7447 list of functions. This list can be changed at run 7448 time by the stack_trace_filter file in the debugfs 7449 tracing directory. Note, this enables stack tracing 7450 and the stacktrace above is not needed. 7451 7452 sti= [PARISC,HW] 7453 Format: <num> 7454 Set the STI (builtin display/keyboard on the HP-PARISC 7455 machines) console (graphic card) which should be used 7456 as the initial boot-console. 7457 See also comment in drivers/video/sticore.c. 7458 7459 sti_font= [HW] 7460 See comment in drivers/video/sticore.c. 7461 7462 stifb= [HW] 7463 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]] 7464 7465 strict_sas_size= 7466 [X86] 7467 Format: <bool> 7468 Enable or disable strict sigaltstack size checks 7469 against the required signal frame size which 7470 depends on the supported FPU features. This can 7471 be used to filter out binaries which have 7472 not yet been made aware of AT_MINSIGSTKSZ. 7473 7474 stress_hpt [PPC,EARLY] 7475 Limits the number of kernel HPT entries in the hash 7476 page table to increase the rate of hash page table 7477 faults on kernel addresses. 7478 7479 stress_slb [PPC,EARLY] 7480 Limits the number of kernel SLB entries, and flushes 7481 them frequently to increase the rate of SLB faults 7482 on kernel addresses. 7483 7484 no_slb_preload [PPC,EARLY] 7485 Disables slb preloading for userspace. 7486 7487 sunrpc.min_resvport= 7488 sunrpc.max_resvport= 7489 [NFS,SUNRPC] 7490 SunRPC servers often require that client requests 7491 originate from a privileged port (i.e. a port in the 7492 range 0 < portnr < 1024). 7493 An administrator who wishes to reserve some of these 7494 ports for other uses may adjust the range that the 7495 kernel's sunrpc client considers to be privileged 7496 using these two parameters to set the minimum and 7497 maximum port values. 7498 7499 sunrpc.svc_rpc_per_connection_limit= 7500 [NFS,SUNRPC] 7501 Limit the number of requests that the server will 7502 process in parallel from a single connection. 7503 The default value is 0 (no limit). 7504 7505 sunrpc.pool_mode= 7506 [NFS] 7507 Deprecated. The NFS server now always uses one 7508 service thread pool per NUMA node (equivalent to a 7509 single global pool on non-NUMA machines). All of 7510 the previously accepted values (auto, global, 7511 percpu, pernode) are still accepted for backward 7512 compatibility but are ignored: the mode is always 7513 pernode, and reads always return "pernode". 7514 7515 sunrpc.tcp_slot_table_entries= 7516 sunrpc.udp_slot_table_entries= 7517 [NFS,SUNRPC] 7518 Sets the upper limit on the number of simultaneous 7519 RPC calls that can be sent from the client to a 7520 server. Increasing these values may allow you to 7521 improve throughput, but will also increase the 7522 amount of memory reserved for use by the client. 7523 7524 suspend.pm_test_delay= 7525 [SUSPEND] 7526 Sets the number of seconds to remain in a suspend test 7527 mode before resuming the system (see 7528 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG 7529 is set. Default value is 5. 7530 7531 svm= [PPC] 7532 Format: { on | off | y | n | 1 | 0 } 7533 This parameter controls use of the Protected 7534 Execution Facility on pSeries. 7535 7536 swiotlb= [ARM,PPC,MIPS,X86,S390,EARLY] 7537 Format: { <int> [,<int>] | force | noforce | track_hiwater} 7538 <int> -- Number of I/O TLB slabs 7539 <int> -- Second integer after comma. Number of swiotlb 7540 areas with their own lock. Will be rounded up 7541 to a power of 2. 7542 force -- force using of bounce buffers even if they 7543 wouldn't be automatically used by the kernel 7544 noforce -- Never use bounce buffers (for debugging) 7545 track_hiwater -- Track high watermark of swiotlb buffers. 7546 Only available when CONFIG_DEBUG_FS is set. 7547 7548 switches= [HW,M68k,EARLY] 7549 7550 sysctl.*= [KNL] 7551 Set a sysctl parameter, right before loading the init 7552 process, as if the value was written to the respective 7553 /proc/sys/... file. Both '.' and '/' are recognized as 7554 separators. Unrecognized parameters and invalid values 7555 are reported in the kernel log. Sysctls registered 7556 later by a loaded module cannot be set this way. 7557 Example: sysctl.vm.swappiness=40 7558 7559 sysrq_always_enabled 7560 [KNL] 7561 Ignore sysrq setting - this boot parameter will 7562 neutralize any effect of /proc/sys/kernel/sysrq. 7563 Useful for debugging. 7564 7565 tcpmhash_entries= [KNL,NET] 7566 Set the number of tcp_metrics_hash slots. 7567 Default value is 8192 or 16384 depending on total 7568 ram pages. This is used to specify the TCP metrics 7569 cache size. See Documentation/networking/ip-sysctl.rst 7570 "tcp_no_metrics_save" section for more details. 7571 7572 tdfx= [HW,DRM] 7573 7574 test_suspend= [SUSPEND] 7575 Format: { "mem" | "standby" | "freeze" }[,N] 7576 Specify "mem" (for Suspend-to-RAM) or "standby" (for 7577 standby suspend) or "freeze" (for suspend type freeze) 7578 as the system sleep state during system startup with 7579 the optional capability to repeat N number of times. 7580 The system is woken from this state using a 7581 wakeup-capable RTC alarm. 7582 7583 thash_entries= [KNL,NET] 7584 Set number of hash buckets for TCP connection 7585 7586 thermal.act= [HW,ACPI] 7587 -1: disable all active trip points in all thermal zones 7588 <degrees C>: override all lowest active trip points 7589 7590 thermal.crt= [HW,ACPI] 7591 -1: disable all critical trip points in all thermal zones 7592 <degrees C>: override all critical trip points 7593 7594 thermal.off= [HW,ACPI] 7595 1: disable ACPI thermal control 7596 7597 thermal.psv= [HW,ACPI] 7598 -1: disable all passive trip points 7599 <degrees C>: override all passive trip points to this 7600 value 7601 7602 thermal.tzp= [HW,ACPI] 7603 Specify global default ACPI thermal zone polling rate 7604 <deci-seconds>: poll all this frequency 7605 0: no polling (default) 7606 7607 thp_anon= [KNL] 7608 Format: <size>[KMG],<size>[KMG]:<state>;<size>[KMG]-<size>[KMG]:<state> 7609 state is one of "always", "madvise", "never" or "inherit". 7610 Control the default behavior of the system with respect 7611 to anonymous transparent hugepages. 7612 Can be used multiple times for multiple anon THP sizes. 7613 See Documentation/admin-guide/mm/transhuge.rst for more 7614 details. 7615 7616 threadirqs [KNL,EARLY] 7617 Force threading of all interrupt handlers except those 7618 marked explicitly IRQF_NO_THREAD. 7619 7620 thp_shmem= [KNL] 7621 Format: <size>[KMG],<size>[KMG]:<policy>;<size>[KMG]-<size>[KMG]:<policy> 7622 Control the default policy of each hugepage size for the 7623 internal shmem mount. <policy> is one of policies available 7624 for the shmem mount ("always", "inherit", "never", "within_size", 7625 and "advise"). 7626 It can be used multiple times for multiple shmem THP sizes. 7627 See Documentation/admin-guide/mm/transhuge.rst for more 7628 details. 7629 7630 tlbi= [X86-64] 7631 Format: {ipi} 7632 ipi: switch to IPI-based TLB flushing 7633 7634 topology= [S390,EARLY] 7635 Format: {off | on} 7636 Specify if the kernel should make use of the cpu 7637 topology information if the hardware supports this. 7638 The scheduler will make use of this information and 7639 e.g. base its process migration decisions on it. 7640 Default is on. 7641 7642 torture.disable_onoff_at_boot= [KNL] 7643 Prevent the CPU-hotplug component of torturing 7644 until after init has spawned. 7645 7646 torture.ftrace_dump_at_shutdown= [KNL] 7647 Dump the ftrace buffer at torture-test shutdown, 7648 even if there were no errors. This can be a 7649 very costly operation when many torture tests 7650 are running concurrently, especially on systems 7651 with rotating-rust storage. 7652 7653 torture.verbose_sleep_frequency= [KNL] 7654 Specifies how many verbose printk()s should be 7655 emitted between each sleep. The default of zero 7656 disables verbose-printk() sleeping. 7657 7658 torture.verbose_sleep_duration= [KNL] 7659 Duration of each verbose-printk() sleep in jiffies. 7660 7661 tpm.disable_pcr_integrity= [HW,TPM] 7662 Do not protect PCR registers from unintended physical 7663 access, or interposers in the bus by the means of 7664 having an integrity protected session wrapped around 7665 TPM2_PCR_Extend command. Consider this in a situation 7666 where TPM is heavily utilized by IMA, thus protection 7667 causing a major performance hit, and the space where 7668 machines are deployed is by other means guarded. 7669 7670 tpm_crb_ffa.busy_timeout_ms= [ARM64,TPM] 7671 Maximum time in milliseconds to retry sending a message 7672 to the TPM service before giving up. This parameter controls 7673 how long the system will continue retrying when the TPM 7674 service is busy. 7675 Format: <unsigned int> 7676 Default: 2000 (2 seconds) 7677 7678 tpm_suspend_pcr=[HW,TPM] 7679 Format: integer pcr id 7680 Specify that at suspend time, the tpm driver 7681 should extend the specified pcr with zeros, 7682 as a workaround for some chips which fail to 7683 flush the last written pcr on TPM_SaveState. 7684 This will guarantee that all the other pcrs 7685 are saved. 7686 7687 tpm_tis.interrupts= [HW,TPM] 7688 Enable interrupts for the MMIO based physical layer 7689 for the FIFO interface. By default it is set to false 7690 (0). For more information about TPM hardware interfaces 7691 defined by Trusted Computing Group (TCG) see 7692 https://trustedcomputinggroup.org/resource/pc-client-platform-tpm-profile-ptp-specification/ 7693 7694 tp_printk [FTRACE] 7695 Have the tracepoints sent to printk as well as the 7696 tracing ring buffer. This is useful for early boot up 7697 where the system hangs or reboots and does not give the 7698 option for reading the tracing buffer or performing a 7699 ftrace_dump_on_oops. 7700 7701 To turn off having tracepoints sent to printk, 7702 echo 0 > /proc/sys/kernel/tracepoint_printk 7703 Note, echoing 1 into this file without the 7704 tp_printk kernel cmdline option has no effect. 7705 7706 The tp_printk_stop_on_boot (see below) can also be used 7707 to stop the printing of events to console at 7708 late_initcall_sync. 7709 7710 ** CAUTION ** 7711 7712 Having tracepoints sent to printk() and activating high 7713 frequency tracepoints such as irq or sched, can cause 7714 the system to live lock. 7715 7716 tp_printk_stop_on_boot [FTRACE] 7717 When tp_printk (above) is set, it can cause a lot of noise 7718 on the console. It may be useful to only include the 7719 printing of events during boot up, as user space may 7720 make the system inoperable. 7721 7722 This command line option will stop the printing of events 7723 to console at the late_initcall_sync() time frame. 7724 7725 trace_buf_size=nn[KMG] 7726 [FTRACE] will set tracing buffer size on each cpu. 7727 7728 trace_clock= [FTRACE] Set the clock used for tracing events 7729 at boot up. 7730 local - Use the per CPU time stamp counter 7731 (converted into nanoseconds). Fast, but 7732 depending on the architecture, may not be 7733 in sync between CPUs. 7734 global - Event time stamps are synchronized across 7735 CPUs. May be slower than the local clock, 7736 but better for some race conditions. 7737 counter - Simple counting of events (1, 2, ..) 7738 note, some counts may be skipped due to the 7739 infrastructure grabbing the clock more than 7740 once per event. 7741 uptime - Use jiffies as the time stamp. 7742 perf - Use the same clock that perf uses. 7743 mono - Use ktime_get_mono_fast_ns() for time stamps. 7744 mono_raw - Use ktime_get_raw_fast_ns() for time 7745 stamps. 7746 boot - Use ktime_get_boot_fast_ns() for time stamps. 7747 Architectures may add more clocks. See 7748 Documentation/trace/ftrace.rst for more details. 7749 7750 trace_event=[event-list] 7751 [FTRACE] Set and start specified trace events in order 7752 to facilitate early boot debugging. The event-list is a 7753 comma-separated list of trace events to enable. See 7754 also Documentation/trace/events.rst 7755 7756 To enable modules, use :mod: keyword: 7757 7758 trace_event=:mod:<module> 7759 7760 The value before :mod: will only enable specific events 7761 that are part of the module. 7762 7763 To apply a filter, use the :filter: keyword: 7764 7765 trace_event=...:filter:<filter> 7766 7767 See the above mentioned document for more information. 7768 7769 trace_instance=[instance-info] 7770 [FTRACE] Create a ring buffer instance early in boot up. 7771 This will be listed in: 7772 7773 /sys/kernel/tracing/instances 7774 7775 Events can be enabled at the time the instance is created 7776 via: 7777 7778 trace_instance=<name>,<system1>:<event1>,<system2>:<event2> 7779 7780 Note, the "<system*>:" portion is optional if the event is 7781 unique. 7782 7783 trace_instance=foo,sched:sched_switch,irq_handler_entry,initcall 7784 7785 will enable the "sched_switch" event (note, the "sched:" is optional, and 7786 the same thing would happen if it was left off). The irq_handler_entry 7787 event, and all events under the "initcall" system. 7788 7789 Flags can be added to the instance to modify its behavior when it is 7790 created. The flags are separated by '^'. 7791 7792 The available flags are: 7793 7794 traceoff - Have the tracing instance tracing disabled after it is created. 7795 traceprintk - Have trace_printk() write into this trace instance 7796 (note, "printk" and "trace_printk" can also be used) 7797 7798 trace_instance=foo^traceoff^traceprintk,sched,irq 7799 7800 The flags must come before the defined events. 7801 7802 If memory has been reserved (see memmap for x86), the instance 7803 can use that memory: 7804 7805 memmap=12M$0x284500000 trace_instance=boot_map@0x284500000:12M 7806 7807 The above will create a "boot_map" instance that uses the physical 7808 memory at 0x284500000 that is 12Megs. The per CPU buffers of that 7809 instance will be split up accordingly. 7810 7811 Alternatively, the memory can be reserved by the reserve_mem option: 7812 7813 reserve_mem=12M:4096:trace trace_instance=boot_map@trace 7814 7815 This will reserve 12 megabytes at boot up with a 4096 byte alignment 7816 and place the ring buffer in this memory. Note that due to KASLR, the 7817 memory may not be the same location each time, which will not preserve 7818 the buffer content. 7819 7820 Also note that the layout of the ring buffer data may change between 7821 kernel versions where the validator will fail and reset the ring buffer 7822 if the layout is not the same as the previous kernel. 7823 7824 If the ring buffer is used for persistent bootups and has events enabled, 7825 it is recommend to disable tracing so that events from a previous boot do not 7826 mix with events of the current boot (unless you are debugging a random crash 7827 at boot up). 7828 7829 reserve_mem=12M:4096:trace trace_instance=boot_map^traceoff^traceprintk@trace,sched,irq 7830 7831 Note, saving the trace buffer across reboots does require that the system 7832 is set up to not wipe memory. For instance, CONFIG_RESET_ATTACK_MITIGATION 7833 can force a memory reset on boot which will clear any trace that was stored. 7834 This is just one of many ways that can clear memory. Make sure your system 7835 keeps the content of memory across reboots before relying on this option. 7836 7837 NB: Both the mapped address and size must be page aligned for the architecture. 7838 7839 See also Documentation/trace/debugging.rst 7840 7841 7842 trace_options=[option-list] 7843 [FTRACE] Enable or disable tracer options at boot. 7844 The option-list is a comma delimited list of options 7845 that can be enabled or disabled just as if you were 7846 to echo the option name into 7847 7848 /sys/kernel/tracing/trace_options 7849 7850 For example, to enable stacktrace option (to dump the 7851 stack trace of each event), add to the command line: 7852 7853 trace_options=stacktrace 7854 7855 See also Documentation/trace/ftrace.rst "trace options" 7856 section. 7857 7858 trace_trigger=[trigger-list] 7859 [FTRACE] Add an event trigger on specific events. 7860 Set a trigger on top of a specific event, with an optional 7861 filter. 7862 7863 The format is "trace_trigger=<event>.<trigger>[ if <filter>],..." 7864 Where more than one trigger may be specified that are comma delimited. 7865 7866 For example: 7867 7868 trace_trigger="sched_switch.stacktrace if prev_state == 2" 7869 7870 The above will enable the "stacktrace" trigger on the "sched_switch" 7871 event but only trigger it if the "prev_state" of the "sched_switch" 7872 event is "2" (TASK_UNINTERRUPTIBLE). 7873 7874 See also "Event triggers" in Documentation/trace/events.rst 7875 7876 7877 traceoff_after_boot 7878 [FTRACE] Sometimes tracing is used to debug issues 7879 during the boot process. Since the trace buffer has a 7880 limited amount of storage, it may be prudent to 7881 disable tracing after the boot is finished, otherwise 7882 the critical information may be overwritten. With this 7883 option, the main tracing buffer will be turned off at 7884 the end of the boot process. 7885 7886 traceoff_on_warning 7887 [FTRACE] enable this option to disable tracing when a 7888 warning is hit. This turns off "tracing_on". Tracing can 7889 be enabled again by echoing '1' into the "tracing_on" 7890 file located in /sys/kernel/tracing/ 7891 7892 This option is useful, as it disables the trace before 7893 the WARNING dump is called, which prevents the trace to 7894 be filled with content caused by the warning output. 7895 7896 This option can also be set at run time via the sysctl 7897 option: kernel/traceoff_on_warning 7898 7899 transparent_hugepage= 7900 [KNL] 7901 Format: [always|madvise|never] 7902 Can be used to control the default behavior of the system 7903 with respect to transparent hugepages. 7904 See Documentation/admin-guide/mm/transhuge.rst 7905 for more details. 7906 7907 transparent_hugepage_shmem= [KNL] 7908 Format: [always|within_size|advise|never|deny|force] 7909 Can be used to control the hugepage allocation policy for 7910 the internal shmem mount. 7911 See Documentation/admin-guide/mm/transhuge.rst 7912 for more details. 7913 7914 transparent_hugepage_tmpfs= [KNL] 7915 Format: [always|within_size|advise|never] 7916 Can be used to control the default hugepage allocation policy 7917 for the tmpfs mount. 7918 See Documentation/admin-guide/mm/transhuge.rst 7919 for more details. 7920 7921 trusted.source= [KEYS] 7922 Format: <string> 7923 This parameter identifies the trust source as a backend 7924 for trusted keys implementation. Supported trust 7925 sources: 7926 - "tpm" 7927 - "tee" 7928 - "caam" 7929 - "dcp" 7930 - "pkwm" 7931 If not specified then it defaults to iterating through 7932 the trust source list starting with TPM and assigns the 7933 first trust source as a backend which is initialized 7934 successfully during iteration. 7935 7936 trusted.debug= [KEYS] 7937 Format: <bool> 7938 Enable trusted keys debug traces at runtime when 7939 CONFIG_TRUSTED_KEYS_DEBUG=y. 7940 7941 To make the traces visible after enabling the option, 7942 use trusted.dyndbg='+p' as needed. By convention, 7943 the subsystem uses pr_debug() for these traces. 7944 7945 SAFETY: The traces can leak sensitive data, so be 7946 cautious before enabling this. They remain inactive 7947 unless this parameter is set this option to a true 7948 value. 7949 7950 Default: false 7951 7952 trusted.rng= [KEYS] 7953 Format: <string> 7954 The RNG used to generate key material for trusted keys. 7955 Can be one of: 7956 - "kernel" 7957 - the same value as trusted.source: "tpm" or "tee" 7958 - "default" 7959 If not specified, "default" is used. In this case, 7960 the RNG's choice is left to each individual trust source. 7961 7962 trusted.dcp_use_otp_key 7963 This is intended to be used in combination with 7964 trusted.source=dcp and will select the DCP OTP key 7965 instead of the DCP UNIQUE key blob encryption. 7966 7967 trusted.dcp_skip_zk_test 7968 This is intended to be used in combination with 7969 trusted.source=dcp and will disable the check if the 7970 blob key is all zeros. This is helpful for situations where 7971 having this key zero'ed is acceptable. E.g. in testing 7972 scenarios. 7973 7974 tsa= [X86] Control mitigation for Transient Scheduler 7975 Attacks on AMD CPUs. Search the following in your 7976 favourite search engine for more details: 7977 7978 "Technical guidance for mitigating transient scheduler 7979 attacks". 7980 7981 off - disable the mitigation 7982 on - enable the mitigation (default) 7983 user - mitigate only user/kernel transitions 7984 vm - mitigate only guest/host transitions 7985 7986 7987 tsc= Disable clocksource stability checks for TSC. 7988 Format: <string> 7989 [x86] reliable: mark tsc clocksource as reliable, this 7990 disables clocksource verification at runtime, as well 7991 as the stability checks done at bootup. Used to enable 7992 high-resolution timer mode on older hardware, and in 7993 virtualized environment. 7994 [x86] noirqtime: Do not use TSC to do irq accounting. 7995 Used to run time disable IRQ_TIME_ACCOUNTING on any 7996 platforms where RDTSC is slow and this accounting 7997 can add overhead. 7998 [x86] unstable: mark the TSC clocksource as unstable, this 7999 marks the TSC unconditionally unstable at bootup and 8000 avoids any further wobbles once the TSC watchdog notices. 8001 [x86] nowatchdog: disable clocksource watchdog. Used 8002 in situations with strict latency requirements (where 8003 interruptions from clocksource watchdog are not 8004 acceptable). 8005 [x86] recalibrate: force recalibration against a HW timer 8006 (HPET or PM timer) on systems whose TSC frequency was 8007 obtained from HW or FW using either an MSR or CPUID(0x15). 8008 Warn if the difference is more than 500 ppm. 8009 [x86] watchdog: Enforce the clocksource watchdog on TSC 8010 8011 tsc_early_khz= [X86,EARLY] Skip early TSC calibration and use the given 8012 value instead. Useful when the early TSC frequency discovery 8013 procedure is not reliable, such as on overclocked systems 8014 with CPUID.16h support and partial CPUID.15h support. 8015 Format: <unsigned int> 8016 8017 tsx= [X86] Control Transactional Synchronization 8018 Extensions (TSX) feature in Intel processors that 8019 support TSX control. 8020 8021 This parameter controls the TSX feature. The options are: 8022 8023 on - Enable TSX on the system. Although there are 8024 mitigations for all known security vulnerabilities, 8025 TSX has been known to be an accelerator for 8026 several previous speculation-related CVEs, and 8027 so there may be unknown security risks associated 8028 with leaving it enabled. 8029 8030 off - Disable TSX on the system. (Note that this 8031 option takes effect only on newer CPUs which are 8032 not vulnerable to MDS, i.e., have 8033 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get 8034 the new IA32_TSX_CTRL MSR through a microcode 8035 update. This new MSR allows for the reliable 8036 deactivation of the TSX functionality.) 8037 8038 auto - Disable TSX if X86_BUG_TAA is present, 8039 otherwise enable TSX on the system. 8040 8041 Not specifying this option is equivalent to tsx=off. 8042 8043 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst 8044 for more details. 8045 8046 tsx_async_abort= [X86,INTEL,EARLY] Control mitigation for the TSX Async 8047 Abort (TAA) vulnerability. 8048 8049 Similar to Micro-architectural Data Sampling (MDS) 8050 certain CPUs that support Transactional 8051 Synchronization Extensions (TSX) are vulnerable to an 8052 exploit against CPU internal buffers which can forward 8053 information to a disclosure gadget under certain 8054 conditions. 8055 8056 In vulnerable processors, the speculatively forwarded 8057 data can be used in a cache side channel attack, to 8058 access data to which the attacker does not have direct 8059 access. 8060 8061 This parameter controls the TAA mitigation. The 8062 options are: 8063 8064 full - Enable TAA mitigation on vulnerable CPUs 8065 if TSX is enabled. 8066 8067 full,nosmt - Enable TAA mitigation and disable SMT on 8068 vulnerable CPUs. If TSX is disabled, SMT 8069 is not disabled because CPU is not 8070 vulnerable to cross-thread TAA attacks. 8071 off - Unconditionally disable TAA mitigation 8072 8073 On MDS-affected machines, tsx_async_abort=off can be 8074 prevented by an active MDS mitigation as both vulnerabilities 8075 are mitigated with the same mechanism so in order to disable 8076 this mitigation, you need to specify mds=off too. 8077 8078 Not specifying this option is equivalent to 8079 tsx_async_abort=full. On CPUs which are MDS affected 8080 and deploy MDS mitigation, TAA mitigation is not 8081 required and doesn't provide any additional 8082 mitigation. 8083 8084 For details see: 8085 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst 8086 8087 turbografx.map[2|3]= [HW,JOY] 8088 TurboGraFX parallel port interface 8089 Format: 8090 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7> 8091 See also Documentation/input/devices/joystick-parport.rst 8092 8093 udbg-immortal [PPC] When debugging early kernel crashes that 8094 happen after console_init() and before a proper 8095 console driver takes over, this boot options might 8096 help "seeing" what's going on. 8097 8098 uhash_entries= [KNL,NET] 8099 Set number of hash buckets for UDP/UDP-Lite connections 8100 8101 uhci-hcd.ignore_oc= 8102 [USB] Ignore overcurrent events (default N). 8103 Some badly-designed motherboards generate lots of 8104 bogus events, for ports that aren't wired to 8105 anything. Set this parameter to avoid log spamming. 8106 Note that genuine overcurrent events won't be 8107 reported either. 8108 8109 unaligned_scalar_speed= 8110 [RISCV] 8111 Format: {slow | fast | unsupported} 8112 Allow skipping scalar unaligned access speed tests. This 8113 is useful for testing alternative code paths and to skip 8114 the tests in environments where they run too slowly. All 8115 CPUs must have the same scalar unaligned access speed. 8116 8117 unaligned_vector_speed= 8118 [RISCV] 8119 Format: {slow | fast | unsupported} 8120 Allow skipping vector unaligned access speed tests. This 8121 is useful for testing alternative code paths and to skip 8122 the tests in environments where they run too slowly. All 8123 CPUs must have the same vector unaligned access speed. 8124 8125 unknown_nmi_panic 8126 [X86] Cause panic on unknown NMI. 8127 8128 unwind_debug [X86-64,EARLY] 8129 Enable unwinder debug output. This can be 8130 useful for debugging certain unwinder error 8131 conditions, including corrupt stacks and 8132 bad/missing unwinder metadata. 8133 8134 usbcore.authorized_default= 8135 [USB] Default USB device authorization: 8136 (default -1 = authorized (same as 1), 8137 0 = not authorized, 1 = authorized, 2 = authorized 8138 if device connected to internal port) 8139 8140 usbcore.autosuspend= 8141 [USB] The autosuspend time delay (in seconds) used 8142 for newly-detected USB devices (default 2). This 8143 is the time required before an idle device will be 8144 autosuspended. Devices for which the delay is set 8145 to a negative value won't be autosuspended at all. 8146 8147 usbcore.usbfs_snoop= 8148 [USB] Set to log all usbfs traffic (default 0 = off). 8149 8150 usbcore.usbfs_snoop_max= 8151 [USB] Maximum number of bytes to snoop in each URB 8152 (default = 65536). 8153 8154 usbcore.blinkenlights= 8155 [USB] Set to cycle leds on hubs (default 0 = off). 8156 8157 usbcore.old_scheme_first= 8158 [USB] Start with the old device initialization 8159 scheme (default 0 = off). 8160 8161 usbcore.usbfs_memory_mb= 8162 [USB] Memory limit (in MB) for buffers allocated by 8163 usbfs (default = 16, 0 = max = 2047). 8164 8165 usbcore.use_both_schemes= 8166 [USB] Try the other device initialization scheme 8167 if the first one fails (default 1 = enabled). 8168 8169 usbcore.initial_descriptor_timeout= 8170 [USB] Specifies timeout for the initial 64-byte 8171 USB_REQ_GET_DESCRIPTOR request in milliseconds 8172 (default 5000 = 5.0 seconds). 8173 8174 usbcore.nousb [USB] Disable the USB subsystem 8175 8176 usbcore.quirks= 8177 [USB] A list of quirk entries to augment the built-in 8178 usb core quirk list. List entries are separated by 8179 commas. Each entry has the form 8180 VendorID:ProductID:Flags. The IDs are 4-digit hex 8181 numbers and Flags is a set of letters. Each letter 8182 will change the built-in quirk; setting it if it is 8183 clear and clearing it if it is set. The letters have 8184 the following meanings: 8185 a = USB_QUIRK_STRING_FETCH_255 (string 8186 descriptors must not be fetched using 8187 a 255-byte read); 8188 b = USB_QUIRK_RESET_RESUME (device can't resume 8189 correctly so reset it instead); 8190 c = USB_QUIRK_NO_SET_INTF (device can't handle 8191 Set-Interface requests); 8192 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't 8193 handle its Configuration or Interface 8194 strings); 8195 e = USB_QUIRK_RESET (device can't be reset 8196 (e.g morph devices), don't use reset); 8197 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has 8198 more interface descriptions than the 8199 bNumInterfaces count, and can't handle 8200 talking to these interfaces); 8201 g = USB_QUIRK_DELAY_INIT (device needs a pause 8202 during initialization, after we read 8203 the device descriptor); 8204 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For 8205 high speed and super speed interrupt 8206 endpoints, the USB 2.0 and USB 3.0 spec 8207 require the interval in microframes (1 8208 microframe = 125 microseconds) to be 8209 calculated as interval = 2 ^ 8210 (bInterval-1). 8211 Devices with this quirk report their 8212 bInterval as the result of this 8213 calculation instead of the exponent 8214 variable used in the calculation); 8215 i = USB_QUIRK_DEVICE_QUALIFIER (device can't 8216 handle device_qualifier descriptor 8217 requests); 8218 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device 8219 generates spurious wakeup, ignore 8220 remote wakeup capability); 8221 k = USB_QUIRK_NO_LPM (device can't handle Link 8222 Power Management); 8223 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL 8224 (Device reports its bInterval as linear 8225 frames instead of the USB 2.0 8226 calculation); 8227 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs 8228 to be disconnected before suspend to 8229 prevent spurious wakeup); 8230 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a 8231 pause after every control message); 8232 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra 8233 delay after resetting its port); 8234 p = USB_QUIRK_SHORT_SET_ADDRESS_REQ_TIMEOUT 8235 (Reduce timeout of the SET_ADDRESS 8236 request from 5000 ms to 500 ms); 8237 q = USB_QUIRK_FORCE_ONE_CONFIG (Device 8238 claims zero configurations, 8239 forcing to 1); 8240 r = USB_QUIRK_WINDOWS_CONFIG_REQ_SIZE (Device 8241 fails during initialization when asked for 8242 9-bytes configuration descriptor request. 8243 Ask for 255-bytes request instead to mirror 8244 Windows' behavior); 8245 Example: quirks=0781:5580:bk,0a5c:5834:gij 8246 8247 usbhid.mousepoll= 8248 [USBHID] The interval which mice are to be polled at. 8249 8250 usbhid.jspoll= 8251 [USBHID] The interval which joysticks are to be polled at. 8252 8253 usbhid.kbpoll= 8254 [USBHID] The interval which keyboards are to be polled at. 8255 8256 usb-storage.delay_use= 8257 [UMS] The delay in seconds before a new device is 8258 scanned for Logical Units (default 1). 8259 Optionally the delay in milliseconds if the value has 8260 suffix with "ms". 8261 Example: delay_use=2567ms 8262 8263 usb-storage.quirks= 8264 [UMS] A list of quirks entries to supplement or 8265 override the built-in unusual_devs list. List 8266 entries are separated by commas. Each entry has 8267 the form VID:PID:Flags where VID and PID are Vendor 8268 and Product ID values (4-digit hex numbers) and 8269 Flags is a set of characters, each corresponding 8270 to a common usb-storage quirk flag as follows: 8271 a = SANE_SENSE (collect more than 18 bytes 8272 of sense data, not on uas); 8273 b = BAD_SENSE (don't collect more than 18 8274 bytes of sense data, not on uas); 8275 c = FIX_CAPACITY (decrease the reported 8276 device capacity by one sector); 8277 d = NO_READ_DISC_INFO (don't use 8278 READ_DISC_INFO command, not on uas); 8279 e = NO_READ_CAPACITY_16 (don't use 8280 READ_CAPACITY_16 command); 8281 f = NO_REPORT_OPCODES (don't use report opcodes 8282 command, uas only); 8283 g = MAX_SECTORS_240 (don't transfer more than 8284 240 sectors at a time, uas only); 8285 h = CAPACITY_HEURISTICS (decrease the 8286 reported device capacity by one 8287 sector if the number is odd); 8288 i = IGNORE_DEVICE (don't bind to this 8289 device); 8290 j = NO_REPORT_LUNS (don't use report luns 8291 command, uas only); 8292 k = NO_SAME (do not use WRITE_SAME, uas only) 8293 l = NOT_LOCKABLE (don't try to lock and 8294 unlock ejectable media, not on uas); 8295 m = MAX_SECTORS_64 (don't transfer more 8296 than 64 sectors = 32 KB at a time, 8297 not on uas); 8298 n = INITIAL_READ10 (force a retry of the 8299 initial READ(10) command, not on uas); 8300 o = CAPACITY_OK (accept the capacity 8301 reported by the device, not on uas); 8302 p = WRITE_CACHE (the device cache is ON 8303 by default, not on uas); 8304 r = IGNORE_RESIDUE (the device reports 8305 bogus residue values, not on uas); 8306 s = SINGLE_LUN (the device has only one 8307 Logical Unit); 8308 t = NO_ATA_1X (don't allow ATA(12) and ATA(16) 8309 commands, uas only); 8310 u = IGNORE_UAS (don't bind to the uas driver); 8311 w = NO_WP_DETECT (don't test whether the 8312 medium is write-protected). 8313 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE 8314 even if the device claims no cache, 8315 not on uas) 8316 Example: quirks=0419:aaf5:rl,0421:0433:rc 8317 8318 user_debug= [KNL,ARM] 8319 Format: <int> 8320 See arch/arm/Kconfig.debug help text. 8321 1 - undefined instruction events 8322 2 - system calls 8323 4 - invalid data aborts 8324 8 - SIGSEGV faults 8325 16 - SIGBUS faults 8326 Example: user_debug=31 8327 8328 vdso= [X86,SH,SPARC] 8329 On X86_32, this is an alias for vdso32=. Otherwise: 8330 8331 vdso=1: enable VDSO (the default) 8332 vdso=0: disable VDSO mapping 8333 8334 vdso32= [X86] Control the 32-bit vDSO 8335 vdso32=1: enable 32-bit VDSO 8336 vdso32=0 or vdso32=2: disable 32-bit VDSO 8337 8338 See the help text for CONFIG_COMPAT_VDSO for more 8339 details. If CONFIG_COMPAT_VDSO is set, the default is 8340 vdso32=0; otherwise, the default is vdso32=1. 8341 8342 For compatibility with older kernels, vdso32=2 is an 8343 alias for vdso32=0. 8344 8345 Try vdso32=0 if you encounter an error that says: 8346 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed! 8347 8348 video= [FB,EARLY] Frame buffer configuration 8349 See Documentation/fb/modedb.rst. 8350 8351 video.brightness_switch_enabled= [ACPI] 8352 Format: [0|1] 8353 If set to 1, on receiving an ACPI notify event 8354 generated by hotkey, video driver will adjust brightness 8355 level and then send out the event to user space through 8356 the allocated input device. If set to 0, video driver 8357 will only send out the event without touching backlight 8358 brightness level. 8359 default: 1 8360 8361 virtio_mmio.device= 8362 [VMMIO] Memory mapped virtio (platform) device. 8363 8364 <size>@<baseaddr>:<irq>[:<id>] 8365 where: 8366 <size> := size (can use standard suffixes 8367 like K, M and G) 8368 <baseaddr> := physical base address 8369 <irq> := interrupt number (as passed to 8370 request_irq()) 8371 <id> := (optional) platform device id 8372 example: 8373 virtio_mmio.device=1K@0x100b0000:48:7 8374 8375 Can be used multiple times for multiple devices. 8376 8377 vga= [BOOT,X86-32] Select a particular video mode 8378 See Documentation/arch/x86/boot.rst and 8379 Documentation/admin-guide/svga.rst. 8380 Use vga=ask for menu. 8381 This is actually a boot loader parameter; the value is 8382 passed to the kernel using a special protocol. 8383 8384 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y. 8385 May slow down system boot speed, especially when 8386 enabled on systems with a large amount of memory. 8387 All options are enabled by default, and this 8388 interface is meant to allow for selectively 8389 enabling or disabling specific virtual memory 8390 debugging features. 8391 8392 Available options are: 8393 P Enable page structure init time poisoning 8394 - Disable all of the above options 8395 8396 vmalloc=nn[KMG] [KNL,BOOT,EARLY] Forces the vmalloc area to have an 8397 exact size of <nn>. This can be used to increase 8398 the minimum size (128MB on x86, arm32 platforms). 8399 It can also be used to decrease the size and leave more room 8400 for directly mapped kernel RAM. Note that this parameter does 8401 not exist on many other platforms (including arm64, alpha, 8402 loongarch, arc, csky, hexagon, microblaze, mips, nios2, openrisc, 8403 parisc, m64k, powerpc, riscv, sh, um, xtensa, s390, sparc). 8404 8405 vmcp_cma=nn[MG] [KNL,S390,EARLY] 8406 Sets the memory size reserved for contiguous memory 8407 allocations for the vmcp device driver. 8408 8409 vmhalt= [KNL,S390] Perform z/VM CP command after system halt. 8410 Format: <command> 8411 8412 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic. 8413 Format: <command> 8414 8415 vmpoff= [KNL,S390] Perform z/VM CP command after power off. 8416 Format: <command> 8417 8418 vmscape= [X86] Controls mitigation for VMscape attacks. 8419 VMscape attacks can leak information from a userspace 8420 hypervisor to a guest via speculative side-channels. 8421 8422 off - disable the mitigation 8423 ibpb - use Indirect Branch Prediction Barrier 8424 (IBPB) mitigation (default) 8425 force - force vulnerability detection even on 8426 unaffected processors 8427 8428 vsyscall= [X86-64,EARLY] 8429 Controls the behavior of vsyscalls (i.e. calls to 8430 fixed addresses of 0xffffffffff600x00 from legacy 8431 code). Most statically-linked binaries and older 8432 versions of glibc use these calls. Because these 8433 functions are at fixed addresses, they make nice 8434 targets for exploits that can control RIP. 8435 8436 emulate Vsyscalls turn into traps and are emulated 8437 reasonably safely. The vsyscall page is 8438 readable. This disables the Linear 8439 Address Space Separation (LASS) security 8440 feature and makes the system less secure. 8441 8442 xonly [default] Vsyscalls turn into traps and are 8443 emulated reasonably safely. The vsyscall 8444 page is not readable. 8445 8446 none Vsyscalls don't work at all. This makes 8447 them quite hard to use for exploits but 8448 might break your system. 8449 8450 vt.color= [VT] Default text color. 8451 Format: 0xYX, X = foreground, Y = background. 8452 Default: 0x07 = light gray on black. 8453 8454 vt.cur_default= [VT] Default cursor shape. 8455 Format: 0xCCBBAA, where AA, BB, and CC are the same as 8456 the parameters of the <Esc>[?A;B;Cc escape sequence; 8457 see vga-softcursor.rst. Default: 2 = underline. 8458 8459 vt.default_blu= [VT] 8460 Format: <blue0>,<blue1>,<blue2>,...,<blue15> 8461 Change the default blue palette of the console. 8462 This is a 16-member array composed of values 8463 ranging from 0-255. 8464 8465 vt.default_grn= [VT] 8466 Format: <green0>,<green1>,<green2>,...,<green15> 8467 Change the default green palette of the console. 8468 This is a 16-member array composed of values 8469 ranging from 0-255. 8470 8471 vt.default_red= [VT] 8472 Format: <red0>,<red1>,<red2>,...,<red15> 8473 Change the default red palette of the console. 8474 This is a 16-member array composed of values 8475 ranging from 0-255. 8476 8477 vt.default_utf8= 8478 [VT] 8479 Format=<0|1> 8480 Set system-wide default UTF-8 mode for all tty's. 8481 Default is 1, i.e. UTF-8 mode is enabled for all 8482 newly opened terminals. 8483 8484 vt.global_cursor_default= 8485 [VT] 8486 Format=<-1|0|1> 8487 Set system-wide default for whether a cursor 8488 is shown on new VTs. Default is -1, 8489 i.e. cursors will be created by default unless 8490 overridden by individual drivers. 0 will hide 8491 cursors, 1 will display them. 8492 8493 vt.italic= [VT] Default color for italic text; 0-15. 8494 Default: 2 = green. 8495 8496 vt.underline= [VT] Default color for underlined text; 0-15. 8497 Default: 3 = cyan. 8498 8499 watchdog timers [HW,WDT] For information on watchdog timers, 8500 see Documentation/watchdog/watchdog-parameters.rst 8501 or other driver-specific files in the 8502 Documentation/watchdog/ directory. 8503 8504 watchdog_thresh= 8505 [KNL] 8506 Set the hard lockup detector stall duration 8507 threshold in seconds. The soft lockup detector 8508 threshold is set to twice the value. A value of 0 8509 disables both lockup detectors. Default is 10 8510 seconds. 8511 8512 workqueue.unbound_cpus= 8513 [KNL,SMP] Specify to constrain one or some CPUs 8514 to use in unbound workqueues. 8515 Format: <cpu-list> 8516 By default, all online CPUs are available for 8517 unbound workqueues. 8518 8519 workqueue.watchdog_thresh= 8520 If CONFIG_WQ_WATCHDOG is configured, workqueue can 8521 warn stall conditions and dump internal state to 8522 help debugging. 0 disables workqueue stall 8523 detection; otherwise, it's the stall threshold 8524 duration in seconds. The default value is 30 and 8525 it can be updated at runtime by writing to the 8526 corresponding sysfs file. 8527 8528 workqueue.panic_on_stall=<uint> 8529 Panic when workqueue stall is detected by 8530 CONFIG_WQ_WATCHDOG. It sets the number times of the 8531 stall to trigger panic. 8532 8533 The default is set by CONFIG_BOOTPARAM_WQ_STALL_PANIC, 8534 which is 0 (disabled) if not configured. 8535 8536 workqueue.panic_on_stall_time=<uint> 8537 Panic when a workqueue stall has been continuous for 8538 the specified number of seconds. Unlike panic_on_stall 8539 which counts accumulated stall events, this triggers 8540 based on the duration of a single continuous stall. 8541 8542 The default is 0, which disables the time-based panic. 8543 8544 workqueue.cpu_intensive_thresh_us= 8545 Per-cpu work items which run for longer than this 8546 threshold are automatically considered CPU intensive 8547 and excluded from concurrency management to prevent 8548 them from noticeably delaying other per-cpu work 8549 items. Default is 10000 (10ms). 8550 8551 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel 8552 will report the work functions which violate this 8553 threshold repeatedly. They are likely good 8554 candidates for using WQ_UNBOUND workqueues instead. 8555 8556 workqueue.cpu_intensive_warning_thresh=<uint> 8557 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel 8558 will report the work functions which violate the 8559 intensive_threshold_us repeatedly. In order to prevent 8560 spurious warnings, start printing only after a work 8561 function has violated this threshold number of times. 8562 8563 The default is 4 times. 0 disables the warning. 8564 8565 workqueue.power_efficient 8566 Per-cpu workqueues are generally preferred because 8567 they show better performance thanks to cache 8568 locality; unfortunately, per-cpu workqueues tend to 8569 be more power hungry than unbound workqueues. 8570 8571 Enabling this makes the per-cpu workqueues which 8572 were observed to contribute significantly to power 8573 consumption unbound, leading to measurably lower 8574 power usage at the cost of small performance 8575 overhead. 8576 8577 The default value of this parameter is determined by 8578 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT. 8579 8580 workqueue.default_affinity_scope= 8581 Select the default affinity scope to use for unbound 8582 workqueues. Can be one of "cpu", "smt", "cache", 8583 "cache_shard", "numa" and "system". Default is 8584 "cache_shard". For more 8585 information, see the Affinity Scopes section in 8586 Documentation/core-api/workqueue.rst. 8587 8588 This can be changed after boot by writing to the 8589 matching /sys/module/workqueue/parameters file. All 8590 workqueues with the "default" affinity scope will be 8591 updated accordingly. 8592 8593 workqueue.debug_force_rr_cpu 8594 Workqueue used to implicitly guarantee that work 8595 items queued without explicit CPU specified are put 8596 on the local CPU. This guarantee is no longer true 8597 and while local CPU is still preferred work items 8598 may be put on foreign CPUs. This debug option 8599 forces round-robin CPU selection to flush out 8600 usages which depend on the now broken guarantee. 8601 When enabled, memory and cache locality will be 8602 impacted. 8603 8604 writecombine= [LOONGARCH,EARLY] Control the MAT (Memory Access 8605 Type) of ioremap_wc(). 8606 8607 on - Enable writecombine, use WUC for ioremap_wc() 8608 off - Disable writecombine, use SUC for ioremap_wc() 8609 8610 x2apic_phys [X86-64,APIC,EARLY] Use x2apic physical mode instead of 8611 default x2apic cluster mode on platforms 8612 supporting x2apic. 8613 8614 xen_512gb_limit [KNL,X86-64,XEN] 8615 Restricts the kernel running paravirtualized under Xen 8616 to use only up to 512 GB of RAM. The reason to do so is 8617 crash analysis tools and Xen tools for doing domain 8618 save/restore/migration must be enabled to handle larger 8619 domains. 8620 8621 xen_console_io [XEN,EARLY] 8622 Boolean option to enable/disable the usage of the Xen 8623 console_io hypercalls to read and write to the console. 8624 Mostly useful for debugging and development. 8625 8626 xen_emul_unplug= [HW,X86,XEN,EARLY] 8627 Unplug Xen emulated devices 8628 Format: [unplug0,][unplug1] 8629 ide-disks -- unplug primary master IDE devices 8630 aux-ide-disks -- unplug non-primary-master IDE devices 8631 nics -- unplug network devices 8632 all -- unplug all emulated devices (NICs and IDE disks) 8633 unnecessary -- unplugging emulated devices is 8634 unnecessary even if the host did not respond to 8635 the unplug protocol 8636 never -- do not unplug even if version check succeeds 8637 8638 xen_legacy_crash [X86,XEN,EARLY] 8639 Crash from Xen panic notifier, without executing late 8640 panic() code such as dumping handler. 8641 8642 xen_mc_debug [X86,XEN,EARLY] 8643 Enable multicall debugging when running as a Xen PV guest. 8644 Enabling this feature will reduce performance a little 8645 bit, so it should only be enabled for obtaining extended 8646 debug data in case of multicall errors. 8647 8648 xen_msr_safe= [X86,XEN,EARLY] 8649 Format: <bool> 8650 Select whether to always use non-faulting (safe) MSR 8651 access functions when running as Xen PV guest. The 8652 default value is controlled by CONFIG_XEN_PV_MSR_SAFE. 8653 8654 xen_nopv [X86] 8655 Disables the PV optimizations forcing the HVM guest to 8656 run as generic HVM guest with no PV drivers. 8657 This option is obsoleted by the "nopv" option, which 8658 has equivalent effect for XEN platform. 8659 8660 xen_no_vector_callback 8661 [KNL,X86,XEN,EARLY] Disable the vector callback for Xen 8662 event channel interrupts. 8663 8664 xen_scrub_pages= [XEN] 8665 Boolean option to control scrubbing pages before giving them back 8666 to Xen, for use by other domains. Can be also changed at runtime 8667 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages. 8668 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT. 8669 8670 xen_timer_slop= [X86-64,XEN,EARLY] 8671 Set the timer slop (in nanoseconds) for the virtual Xen 8672 timers (default is 100000). This adjusts the minimum 8673 delta of virtualized Xen timers, where lower values 8674 improve timer resolution at the expense of processing 8675 more timer interrupts. 8676 8677 xen.balloon_boot_timeout= [XEN] 8678 The time (in seconds) to wait before giving up to boot 8679 in case initial ballooning fails to free enough memory. 8680 Applies only when running as HVM or PVH guest and 8681 started with less memory configured than allowed at 8682 max. Default is 180. 8683 8684 xen.event_eoi_delay= [XEN] 8685 How long to delay EOI handling in case of event 8686 storms (jiffies). Default is 10. 8687 8688 xen.event_loop_timeout= [XEN] 8689 After which time (jiffies) the event handling loop 8690 should start to delay EOI handling. Default is 2. 8691 8692 xen.fifo_events= [XEN] 8693 Boolean parameter to disable using fifo event handling 8694 even if available. Normally fifo event handling is 8695 preferred over the 2-level event handling, as it is 8696 fairer and the number of possible event channels is 8697 much higher. Default is on (use fifo events). 8698 8699 xirc2ps_cs= [NET,PCMCIA] 8700 Format: 8701 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]] 8702 8703 xive= [PPC] 8704 By default on POWER9 and above, the kernel will 8705 natively use the XIVE interrupt controller. This option 8706 allows the fallback firmware mode to be used: 8707 8708 off Fallback to firmware control of XIVE interrupt 8709 controller on both pseries and powernv 8710 platforms. Only useful on POWER9 and above. 8711 8712 xive.store-eoi=off [PPC] 8713 By default on POWER10 and above, the kernel will use 8714 stores for EOI handling when the XIVE interrupt mode 8715 is active. This option allows the XIVE driver to use 8716 loads instead, as on POWER9. 8717 8718 xhci-hcd.quirks [USB,KNL] 8719 A hex value specifying bitmask with supplemental xhci 8720 host controller quirks. Meaning of each bit can be 8721 consulted in header drivers/usb/host/xhci.h. 8722 8723 xmon [PPC,EARLY] 8724 Format: { early | on | rw | ro | off } 8725 Controls if xmon debugger is enabled. Default is off. 8726 Passing only "xmon" is equivalent to "xmon=early". 8727 early Call xmon as early as possible on boot; xmon 8728 debugger is called from setup_arch(). 8729 on xmon debugger hooks will be installed so xmon 8730 is only called on a kernel crash. Default mode, 8731 i.e. either "ro" or "rw" mode, is controlled 8732 with CONFIG_XMON_DEFAULT_RO_MODE. 8733 rw xmon debugger hooks will be installed so xmon 8734 is called only on a kernel crash, mode is write, 8735 meaning SPR registers, memory and, other data 8736 can be written using xmon commands. 8737 ro same as "rw" option above but SPR registers, 8738 memory, and other data can't be written using 8739 xmon commands. 8740 off xmon is disabled. 8741