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 DRM Direct Rendering Management support is enabled. 12 DYNAMIC_DEBUG Build in debug messages and enable them at runtime 13 EARLY Parameter processed too early to be embedded in initrd. 14 EDD BIOS Enhanced Disk Drive Services (EDD) is enabled 15 EFI EFI Partitioning (GPT) is enabled 16 EVM Extended Verification Module 17 FB The frame buffer device is enabled. 18 FTRACE Function tracing enabled. 19 GCOV GCOV profiling is enabled. 20 HIBERNATION HIBERNATION is enabled. 21 HW Appropriate hardware is enabled. 22 HYPER_V HYPERV support is enabled. 23 IMA Integrity measurement architecture is enabled. 24 IP_PNP IP DHCP, BOOTP, or RARP is enabled. 25 IPV6 IPv6 support is enabled. 26 ISAPNP ISA PnP code is enabled. 27 ISOL CPU Isolation is enabled. 28 JOY Appropriate joystick support is enabled. 29 KGDB Kernel debugger support is enabled. 30 KVM Kernel Virtual Machine support is enabled. 31 LIBATA Libata driver is enabled 32 LOONGARCH LoongArch architecture is enabled. 33 LOOP Loopback device support is enabled. 34 LP Printer support is enabled. 35 M68k M68k architecture is enabled. 36 These options have more detailed description inside of 37 Documentation/arch/m68k/kernel-options.rst. 38 MIPS MIPS architecture is enabled. 39 MOUSE Appropriate mouse support is enabled. 40 MSI Message Signaled Interrupts (PCI). 41 MTD MTD (Memory Technology Device) support is enabled. 42 NET Appropriate network support is enabled. 43 NFS Appropriate NFS support is enabled. 44 NUMA NUMA support is enabled. 45 OF Devicetree is enabled. 46 PARISC The PA-RISC architecture is enabled. 47 PCI PCI bus support is enabled. 48 PCIE PCI Express support is enabled. 49 PCMCIA The PCMCIA subsystem is enabled. 50 PNP Plug & Play support is enabled. 51 PPC PowerPC architecture is enabled. 52 PPT Parallel port support is enabled. 53 PS2 Appropriate PS/2 support is enabled. 54 PV_OPS A paravirtualized kernel is enabled. 55 RAM RAM disk support is enabled. 56 RDT Intel Resource Director Technology. 57 RISCV RISCV architecture is enabled. 58 S390 S390 architecture is enabled. 59 SCSI Appropriate SCSI support is enabled. 60 A lot of drivers have their options described inside 61 the Documentation/scsi/ sub-directory. 62 SDW SoundWire support is enabled. 63 SECURITY Different security models are enabled. 64 SELINUX SELinux support is enabled. 65 SERIAL Serial support is enabled. 66 SH SuperH architecture is enabled. 67 SMP The kernel is an SMP kernel. 68 SPARC Sparc architecture is enabled. 69 SUSPEND System suspend states are enabled. 70 SWSUSP Software suspend (hibernation) is enabled. 71 TPM TPM drivers are enabled. 72 UMS USB Mass Storage support is enabled. 73 USB USB support is enabled. 74 NVME NVMe support is enabled 75 USBHID USB Human Interface Device support is enabled. 76 V4L Video For Linux support is enabled. 77 VGA The VGA console has been enabled. 78 VMMIO Driver for memory mapped virtio devices is enabled. 79 VT Virtual terminal support is enabled. 80 WDT Watchdog support is enabled. 81 X86-32 X86-32, aka i386 architecture is enabled. 82 X86-64 X86-64 architecture is enabled. 83 X86 Either 32-bit or 64-bit x86 (same as X86-32+X86-64) 84 X86_UV SGI UV support is enabled. 85 XEN Xen support is enabled 86 XTENSA xtensa architecture is enabled. 87 88In addition, the following text indicates that the option 89 90 BOOT Is a boot loader parameter. 91 BUGS= Relates to possible processor bugs on the said processor. 92 KNL Is a kernel start-up parameter. 93 94 95Kernel parameters 96 97 accept_memory= [MM] 98 Format: { eager | lazy } 99 default: lazy 100 By default, unaccepted memory is accepted lazily to 101 avoid prolonged boot times. The lazy option will add 102 some runtime overhead until all memory is eventually 103 accepted. In most cases the overhead is negligible. 104 For some workloads or for debugging purposes 105 accept_memory=eager can be used to accept all memory 106 at once during boot. 107 108 acpi= [HW,ACPI,X86,ARM64,RISCV64,EARLY] 109 Advanced Configuration and Power Interface 110 Format: { force | on | off | strict | noirq | rsdt | 111 copy_dsdt | nospcr } 112 force -- enable ACPI if default was off 113 on -- enable ACPI but allow fallback to DT [arm64,riscv64] 114 off -- disable ACPI if default was on 115 noirq -- do not use ACPI for IRQ routing 116 strict -- Be less tolerant of platforms that are not 117 strictly ACPI specification compliant. 118 rsdt -- prefer RSDT over (default) XSDT 119 copy_dsdt -- copy DSDT to memory 120 nocmcff -- Disable firmware first mode for corrected 121 errors. This disables parsing the HEST CMC error 122 source to check if firmware has set the FF flag. This 123 may result in duplicate corrected error reports. 124 nospcr -- disable console in ACPI SPCR table as 125 default _serial_ console on ARM64 126 spcr -- enable console in ACPI SPCR table as 127 default _serial_ console on x86 128 For ARM64, ONLY "acpi=off", "acpi=on", "acpi=force" or 129 "acpi=nospcr" are available 130 For RISCV64, ONLY "acpi=off", "acpi=on" or "acpi=force" 131 are available 132 133 See also Documentation/power/runtime_pm.rst, pci=noacpi 134 135 acpi_apic_instance= [ACPI,IOAPIC,EARLY] 136 Format: <int> 137 2: use 2nd APIC table, if available 138 1,0: use 1st APIC table 139 default: 0 140 141 acpi_backlight= [HW,ACPI] 142 { vendor | video | native | none } 143 If set to vendor, prefer vendor-specific driver 144 (e.g. thinkpad_acpi, sony_acpi, etc.) instead 145 of the ACPI video.ko driver. 146 If set to video, use the ACPI video.ko driver. 147 If set to native, use the device's native backlight mode. 148 If set to none, disable the ACPI backlight interface. 149 150 acpi_force_32bit_fadt_addr [ACPI,EARLY] 151 force FADT to use 32 bit addresses rather than the 152 64 bit X_* addresses. Some firmware have broken 64 153 bit addresses for force ACPI ignore these and use 154 the older legacy 32 bit addresses. 155 156 acpica_no_return_repair [HW, ACPI] 157 Disable AML predefined validation mechanism 158 This mechanism can repair the evaluation result to make 159 the return objects more ACPI specification compliant. 160 This option is useful for developers to identify the 161 root cause of an AML interpreter issue when the issue 162 has something to do with the repair mechanism. 163 164 acpi.debug_layer= [HW,ACPI,ACPI_DEBUG] 165 acpi.debug_level= [HW,ACPI,ACPI_DEBUG] 166 Format: <int> 167 CONFIG_ACPI_DEBUG must be enabled to produce any ACPI 168 debug output. Bits in debug_layer correspond to a 169 _COMPONENT in an ACPI source file, e.g., 170 #define _COMPONENT ACPI_EVENTS 171 Bits in debug_level correspond to a level in 172 ACPI_DEBUG_PRINT statements, e.g., 173 ACPI_DEBUG_PRINT((ACPI_DB_INFO, ... 174 The debug_level mask defaults to "info". See 175 Documentation/firmware-guide/acpi/debug.rst for more information about 176 debug layers and levels. 177 178 Enable processor driver info messages: 179 acpi.debug_layer=0x20000000 180 Enable AML "Debug" output, i.e., stores to the Debug 181 object while interpreting AML: 182 acpi.debug_layer=0xffffffff acpi.debug_level=0x2 183 Enable all messages related to ACPI hardware: 184 acpi.debug_layer=0x2 acpi.debug_level=0xffffffff 185 186 Some values produce so much output that the system is 187 unusable. The "log_buf_len" parameter may be useful 188 if you need to capture more output. 189 190 acpi.poweroff_on_fatal= [ACPI] 191 {0 | 1} 192 Causes the system to poweroff when the ACPI bytecode signals 193 a fatal error. The default value of this setting is 1. 194 Overriding this value should only be done for diagnosing 195 ACPI firmware problems, as the system might behave erratically 196 after having encountered a fatal ACPI error. 197 198 acpi_enforce_resources= [ACPI] 199 { strict | lax | no } 200 Check for resource conflicts between native drivers 201 and ACPI OperationRegions (SystemIO and SystemMemory 202 only). IO ports and memory declared in ACPI might be 203 used by the ACPI subsystem in arbitrary AML code and 204 can interfere with legacy drivers. 205 strict (default): access to resources claimed by ACPI 206 is denied; legacy drivers trying to access reserved 207 resources will fail to bind to device using them. 208 lax: access to resources claimed by ACPI is allowed; 209 legacy drivers trying to access reserved resources 210 will bind successfully but a warning message is logged. 211 no: ACPI OperationRegions are not marked as reserved, 212 no further checks are performed. 213 214 acpi_force_table_verification [HW,ACPI,EARLY] 215 Enable table checksum verification during early stage. 216 By default, this is disabled due to x86 early mapping 217 size limitation. 218 219 acpi_irq_balance [HW,ACPI] 220 ACPI will balance active IRQs 221 default in APIC mode 222 223 acpi_irq_nobalance [HW,ACPI] 224 ACPI will not move active IRQs (default) 225 default in PIC mode 226 227 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA 228 Format: <irq>,<irq>... 229 230 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for 231 use by PCI 232 Format: <irq>,<irq>... 233 234 acpi_mask_gpe= [HW,ACPI] 235 Due to the existence of _Lxx/_Exx, some GPEs triggered 236 by unsupported hardware/firmware features can result in 237 GPE floodings that cannot be automatically disabled by 238 the GPE dispatcher. 239 This facility can be used to prevent such uncontrolled 240 GPE floodings. 241 Format: <byte> or <bitmap-list> 242 243 acpi_no_auto_serialize [HW,ACPI] 244 Disable auto-serialization of AML methods 245 AML control methods that contain the opcodes to create 246 named objects will be marked as "Serialized" by the 247 auto-serialization feature. 248 This feature is enabled by default. 249 This option allows to turn off the feature. 250 251 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump 252 kernels. 253 254 acpi_no_static_ssdt [HW,ACPI,EARLY] 255 Disable installation of static SSDTs at early boot time 256 By default, SSDTs contained in the RSDT/XSDT will be 257 installed automatically and they will appear under 258 /sys/firmware/acpi/tables. 259 This option turns off this feature. 260 Note that specifying this option does not affect 261 dynamic table installation which will install SSDT 262 tables to /sys/firmware/acpi/tables/dynamic. 263 264 acpi_no_watchdog [HW,ACPI,WDT] 265 Ignore the ACPI-based watchdog interface (WDAT) and let 266 a native driver control the watchdog device instead. 267 268 acpi_rsdp= [ACPI,EFI,KEXEC,EARLY] 269 Pass the RSDP address to the kernel, mostly used 270 on machines running EFI runtime service to boot the 271 second kernel for kdump. 272 273 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS 274 Format: To spoof as Windows 98: ="Microsoft Windows" 275 276 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead 277 of 2 which is mandated by ACPI 6) as the supported ACPI 278 specification revision (when using this switch, it may 279 be necessary to carry out a cold reboot _twice_ in a 280 row to make it take effect on the platform firmware). 281 282 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings 283 acpi_osi="string1" # add string1 284 acpi_osi="!string2" # remove string2 285 acpi_osi=!* # remove all strings 286 acpi_osi=! # disable all built-in OS vendor 287 strings 288 acpi_osi=!! # enable all built-in OS vendor 289 strings 290 acpi_osi= # disable all strings 291 292 'acpi_osi=!' can be used in combination with single or 293 multiple 'acpi_osi="string1"' to support specific OS 294 vendor string(s). Note that such command can only 295 affect the default state of the OS vendor strings, thus 296 it cannot affect the default state of the feature group 297 strings and the current state of the OS vendor strings, 298 specifying it multiple times through kernel command line 299 is meaningless. This command is useful when one do not 300 care about the state of the feature group strings which 301 should be controlled by the OSPM. 302 Examples: 303 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent 304 to 'acpi_osi="Windows 2000" acpi_osi=!', they all 305 can make '_OSI("Windows 2000")' TRUE. 306 307 'acpi_osi=' cannot be used in combination with other 308 'acpi_osi=' command lines, the _OSI method will not 309 exist in the ACPI namespace. NOTE that such command can 310 only affect the _OSI support state, thus specifying it 311 multiple times through kernel command line is also 312 meaningless. 313 Examples: 314 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)' 315 FALSE. 316 317 'acpi_osi=!*' can be used in combination with single or 318 multiple 'acpi_osi="string1"' to support specific 319 string(s). Note that such command can affect the 320 current state of both the OS vendor strings and the 321 feature group strings, thus specifying it multiple times 322 through kernel command line is meaningful. But it may 323 still not able to affect the final state of a string if 324 there are quirks related to this string. This command 325 is useful when one want to control the state of the 326 feature group strings to debug BIOS issues related to 327 the OSPM features. 328 Examples: 329 1. 'acpi_osi="Module Device" acpi_osi=!*' can make 330 '_OSI("Module Device")' FALSE. 331 2. 'acpi_osi=!* acpi_osi="Module Device"' can make 332 '_OSI("Module Device")' TRUE. 333 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is 334 equivalent to 335 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"' 336 and 337 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!', 338 they all will make '_OSI("Windows 2000")' TRUE. 339 340 acpi_pm_good [X86] 341 Override the pmtimer bug detection: force the kernel 342 to assume that this machine's pmtimer latches its value 343 and always returns good values. 344 345 acpi_sci= [HW,ACPI,EARLY] ACPI System Control Interrupt trigger mode 346 Format: { level | edge | high | low } 347 348 acpi_skip_timer_override [HW,ACPI,EARLY] 349 Recognize and ignore IRQ0/pin2 Interrupt Override. 350 For broken nForce2 BIOS resulting in XT-PIC timer. 351 352 acpi_sleep= [HW,ACPI] Sleep options 353 Format: { s3_bios, s3_mode, s3_beep, s4_hwsig, 354 s4_nohwsig, old_ordering, nonvs, 355 sci_force_enable, nobl } 356 See Documentation/power/video.rst for information on 357 s3_bios and s3_mode. 358 s3_beep is for debugging; it makes the PC's speaker beep 359 as soon as the kernel's real-mode entry point is called. 360 s4_hwsig causes the kernel to check the ACPI hardware 361 signature during resume from hibernation, and gracefully 362 refuse to resume if it has changed. This complies with 363 the ACPI specification but not with reality, since 364 Windows does not do this and many laptops do change it 365 on docking. So the default behaviour is to allow resume 366 and simply warn when the signature changes, unless the 367 s4_hwsig option is enabled. 368 s4_nohwsig prevents ACPI hardware signature from being 369 used (or even warned about) during resume. 370 old_ordering causes the ACPI 1.0 ordering of the _PTS 371 control method, with respect to putting devices into 372 low power states, to be enforced (the ACPI 2.0 ordering 373 of _PTS is used by default). 374 nonvs prevents the kernel from saving/restoring the 375 ACPI NVS memory during suspend/hibernation and resume. 376 sci_force_enable causes the kernel to set SCI_EN directly 377 on resume from S1/S3 (which is against the ACPI spec, 378 but some broken systems don't work without it). 379 nobl causes the internal blacklist of systems known to 380 behave incorrectly in some ways with respect to system 381 suspend and resume to be ignored (use wisely). 382 383 acpi_use_timer_override [HW,ACPI,EARLY] 384 Use timer override. For some broken Nvidia NF5 boards 385 that require a timer override, but don't have HPET 386 387 add_efi_memmap [EFI,X86,EARLY] Include EFI memory map in 388 kernel's map of available physical RAM. 389 390 agp= [AGP] 391 { off | try_unsupported } 392 off: disable AGP support 393 try_unsupported: try to drive unsupported chipsets 394 (may crash computer or cause data corruption) 395 396 ALSA [HW,ALSA] 397 See Documentation/sound/alsa-configuration.rst 398 399 alignment= [KNL,ARM] 400 Allow the default userspace alignment fault handler 401 behaviour to be specified. Bit 0 enables warnings, 402 bit 1 enables fixups, and bit 2 sends a segfault. 403 404 align_va_addr= [X86-64] 405 Align virtual addresses by clearing slice [14:12] when 406 allocating a VMA at process creation time. This option 407 gives you up to 3% performance improvement on AMD F15h 408 machines (where it is enabled by default) for a 409 CPU-intensive style benchmark, and it can vary highly in 410 a microbenchmark depending on workload and compiler. 411 412 32: only for 32-bit processes 413 64: only for 64-bit processes 414 on: enable for both 32- and 64-bit processes 415 off: disable for both 32- and 64-bit processes 416 417 alloc_snapshot [FTRACE] 418 Allocate the ftrace snapshot buffer on boot up when the 419 main buffer is allocated. This is handy if debugging 420 and you need to use tracing_snapshot() on boot up, and 421 do not want to use tracing_snapshot_alloc() as it needs 422 to be done where GFP_KERNEL allocations are allowed. 423 424 allow_mismatched_32bit_el0 [ARM64,EARLY] 425 Allow execve() of 32-bit applications and setting of the 426 PER_LINUX32 personality on systems where only a strict 427 subset of the CPUs support 32-bit EL0. When this 428 parameter is present, the set of CPUs supporting 32-bit 429 EL0 is indicated by /sys/devices/system/cpu/aarch32_el0 430 and hot-unplug operations may be restricted. 431 432 See Documentation/arch/arm64/asymmetric-32bit.rst for more 433 information. 434 435 amd_iommu= [HW,X86-64] 436 Pass parameters to the AMD IOMMU driver in the system. 437 Possible values are: 438 fullflush - Deprecated, equivalent to iommu.strict=1 439 off - do not initialize any AMD IOMMU found in 440 the system 441 force_isolation - Force device isolation for all 442 devices. The IOMMU driver is not 443 allowed anymore to lift isolation 444 requirements as needed. This option 445 does not override iommu=pt 446 force_enable - Force enable the IOMMU on platforms known 447 to be buggy with IOMMU enabled. Use this 448 option with care. 449 pgtbl_v1 - Use v1 page table for DMA-API (Default). 450 pgtbl_v2 - Use v2 page table for DMA-API. 451 irtcachedis - Disable Interrupt Remapping Table (IRT) caching. 452 nohugepages - Limit page-sizes used for v1 page-tables 453 to 4 KiB. 454 v2_pgsizes_only - Limit page-sizes used for v1 page-tables 455 to 4KiB/2Mib/1GiB. 456 457 458 amd_iommu_dump= [HW,X86-64] 459 Enable AMD IOMMU driver option to dump the ACPI table 460 for AMD IOMMU. With this option enabled, AMD IOMMU 461 driver will print ACPI tables for AMD IOMMU during 462 IOMMU initialization. 463 464 amd_iommu_intr= [HW,X86-64] 465 Specifies one of the following AMD IOMMU interrupt 466 remapping modes: 467 legacy - Use legacy interrupt remapping mode. 468 vapic - Use virtual APIC mode, which allows IOMMU 469 to inject interrupts directly into guest. 470 This mode requires kvm-amd.avic=1. 471 (Default when IOMMU HW support is present.) 472 473 amd_pstate= [X86,EARLY] 474 disable 475 Do not enable amd_pstate as the default 476 scaling driver for the supported processors 477 passive 478 Use amd_pstate with passive mode as a scaling driver. 479 In this mode autonomous selection is disabled. 480 Driver requests a desired performance level and platform 481 tries to match the same performance level if it is 482 satisfied by guaranteed performance level. 483 active 484 Use amd_pstate_epp driver instance as the scaling driver, 485 driver provides a hint to the hardware if software wants 486 to bias toward performance (0x0) or energy efficiency (0xff) 487 to the CPPC firmware. then CPPC power algorithm will 488 calculate the runtime workload and adjust the realtime cores 489 frequency. 490 guided 491 Activate guided autonomous mode. Driver requests minimum and 492 maximum performance level and the platform autonomously 493 selects a performance level in this range and appropriate 494 to the current workload. 495 496 amd_prefcore= 497 [X86] 498 disable 499 Disable amd-pstate preferred core. 500 501 amd_dynamic_epp= 502 [X86] 503 disable 504 Disable amd-pstate dynamic EPP. 505 enable 506 Enable amd-pstate dynamic EPP. 507 508 amijoy.map= [HW,JOY] Amiga joystick support 509 Map of devices attached to JOY0DAT and JOY1DAT 510 Format: <a>,<b> 511 See also Documentation/input/joydev/joystick.rst 512 513 analog.map= [HW,JOY] Analog joystick and gamepad support 514 Specifies type or capabilities of an analog joystick 515 connected to one of 16 gameports 516 Format: <type1>,<type2>,..<type16> 517 518 apc= [HW,SPARC] 519 Power management functions (SPARCstation-4/5 + deriv.) 520 Format: noidle 521 Disable APC CPU standby support. SPARCstation-Fox does 522 not play well with APC CPU idle - disable it if you have 523 APC and your system crashes randomly. 524 525 apic [APIC,X86-64] Use IO-APIC. Default. 526 527 apic= [APIC,X86,EARLY] Advanced Programmable Interrupt Controller 528 Change the output verbosity while booting 529 Format: { quiet (default) | verbose | debug } 530 Change the amount of debugging information output 531 when initialising the APIC and IO-APIC components. 532 533 apic_extnmi= [APIC,X86,EARLY] External NMI delivery setting 534 Format: { bsp (default) | all | none } 535 bsp: External NMI is delivered only to CPU 0 536 all: External NMIs are broadcast to all CPUs as a 537 backup of CPU 0 538 none: External NMI is masked for all CPUs. This is 539 useful so that a dump capture kernel won't be 540 shot down by NMI 541 542 apicpmtimer Do APIC timer calibration using the pmtimer. Implies 543 apicmaintimer. Useful when your PIT timer is totally 544 broken. 545 546 autoconf= [IPV6] 547 See Documentation/networking/ipv6.rst. 548 549 apm= [APM] Advanced Power Management 550 See header of arch/x86/kernel/apm_32.c. 551 552 apparmor= [APPARMOR] Disable or enable AppArmor at boot time 553 Format: { "0" | "1" } 554 See security/apparmor/Kconfig help text 555 0 -- disable. 556 1 -- enable. 557 Default value is set via kernel config option. 558 559 arm64.no32bit_el0 [ARM64] Unconditionally disable the execution of 560 32 bit applications. 561 562 arm64.nobti [ARM64] Unconditionally disable Branch Target 563 Identification support 564 565 arm64.nogcs [ARM64] Unconditionally disable Guarded Control Stack 566 support 567 568 arm64.nomops [ARM64] Unconditionally disable Memory Copy and Memory 569 Set instructions support 570 571 arm64.nompam [ARM64] Unconditionally disable Memory Partitioning And 572 Monitoring support 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 ataflop= [HW,M68k] 587 588 atarimouse= [HW,MOUSE] Atari Mouse 589 590 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess, 591 EzKey and similar keyboards 592 593 atkbd.reset= [HW] Reset keyboard during initialization 594 595 atkbd.set= [HW] Select keyboard code set 596 Format: <int> (2 = AT (default), 3 = PS/2) 597 598 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar 599 keyboards 600 601 atkbd.softraw= [HW] Choose between synthetic and real raw mode 602 Format: <bool> (0 = real, 1 = synthetic (default)) 603 604 atkbd.softrepeat= [HW] 605 Use software keyboard repeat 606 607 audit= [KNL] Enable the audit sub-system 608 Format: { "0" | "1" | "off" | "on" } 609 0 | off - kernel audit is disabled and can not be 610 enabled until the next reboot 611 unset - kernel audit is initialized but disabled and 612 will be fully enabled by the userspace auditd. 613 1 | on - kernel audit is initialized and partially 614 enabled, storing at most audit_backlog_limit 615 messages in RAM until it is fully enabled by the 616 userspace auditd. 617 Default: unset 618 619 audit_backlog_limit= [KNL] Set the audit queue size limit. 620 Format: <int> (must be >=0) 621 Default: 64 622 623 bau= [X86_UV] Enable the BAU on SGI UV. The default 624 behavior is to disable the BAU (i.e. bau=0). 625 Format: { "0" | "1" } 626 0 - Disable the BAU. 627 1 - Enable the BAU. 628 unset - Disable the BAU. 629 630 bdev_allow_write_mounted= 631 Format: <bool> 632 Control the ability to open a mounted block device 633 for writing, i.e., allow / disallow writes that bypass 634 the FS. This was implemented as a means to prevent 635 fuzzers from crashing the kernel by overwriting the 636 metadata underneath a mounted FS without its awareness. 637 This also prevents destructive formatting of mounted 638 filesystems by naive storage tooling that don't use 639 O_EXCL. Default is Y and can be changed through the 640 Kconfig option CONFIG_BLK_DEV_WRITE_MOUNTED. 641 642 bert_disable [ACPI] 643 Disable BERT OS support on buggy BIOSes. 644 645 bgrt_disable [ACPI,X86,EARLY] 646 Disable BGRT to avoid flickering OEM logo. 647 648 blkdevparts= Manual partition parsing of block device(s) for 649 embedded devices based on command line input. 650 See Documentation/block/cmdline-partition.rst 651 652 boot_delay= [KNL,EARLY] 653 Milliseconds to delay each printk during boot. 654 Only works if CONFIG_BOOT_PRINTK_DELAY is enabled, 655 and you may also have to specify "lpj=". Boot_delay 656 values larger than 10 seconds (10000) are assumed 657 erroneous and ignored. 658 Format: integer 659 660 bootconfig [KNL,EARLY] 661 Extended command line options can be added to an initrd 662 and this will cause the kernel to look for it. 663 664 See Documentation/admin-guide/bootconfig.rst 665 666 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards) 667 bttv.radio= Most important insmod options are available as 668 kernel args too. 669 bttv.pll= See Documentation/admin-guide/media/bttv.rst 670 bttv.tuner= 671 672 bulk_remove=off [PPC] This parameter disables the use of the pSeries 673 firmware feature for flushing multiple hpte entries 674 at a time. 675 676 c101= [NET] Moxa C101 synchronous serial card 677 678 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection. 679 Sometimes CPU hardware bugs make them report the cache 680 size incorrectly. The kernel will attempt work arounds 681 to fix known problems, but for some CPUs it is not 682 possible to determine what the correct size should be. 683 This option provides an override for these situations. 684 685 carrier_timeout= 686 [NET] Specifies amount of time (in seconds) that 687 the kernel should wait for a network carrier. By default 688 it waits 120 seconds. 689 690 ca_keys= [KEYS] This parameter identifies a specific key(s) on 691 the system trusted keyring to be used for certificate 692 trust validation. 693 format: { id:<keyid> | builtin } 694 695 cca= [MIPS,EARLY] Override the kernel pages' cache coherency 696 algorithm. Accepted values range from 0 to 7 697 inclusive. See arch/mips/include/asm/pgtable-bits.h 698 for platform specific values (SB1, Loongson3 and 699 others). 700 701 ccw_timeout_log [S390] 702 See Documentation/arch/s390/common_io.rst for details. 703 704 cfi= [X86-64] Set Control Flow Integrity checking features 705 when CONFIG_FINEIBT is enabled. 706 Format: feature[,feature...] 707 Default: auto 708 709 auto: Use FineIBT if IBT available, otherwise kCFI. 710 Under FineIBT, enable "paranoid" mode when 711 FRED is not available. 712 off: Turn off CFI checking. 713 kcfi: Use kCFI (disable FineIBT). 714 fineibt: Use FineIBT (even if IBT not available). 715 norand: Do not re-randomize CFI hashes. 716 paranoid: Add caller hash checking under FineIBT. 717 bhi: Enable register poisoning to stop speculation 718 across FineIBT. (Disabled by default.) 719 warn: Do not enforce CFI checking: warn only. 720 debug: Report CFI initialization details. 721 722 cgroup_disable= [KNL] Disable a particular controller or optional feature 723 Format: {name of the controller(s) or feature(s) to disable} 724 The effects of cgroup_disable=foo are: 725 - foo isn't auto-mounted if you mount all cgroups in 726 a single hierarchy 727 - foo isn't visible as an individually mountable 728 subsystem 729 - if foo is an optional feature then the feature is 730 disabled and corresponding cgroup files are not 731 created 732 {Currently only "memory" controller deal with this and 733 cut the overhead, others just disable the usage. So 734 only cgroup_disable=memory is actually worthy} 735 Specifying "pressure" disables per-cgroup pressure 736 stall information accounting feature 737 738 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1 739 Format: { { controller | "all" | "named" } 740 [,{ controller | "all" | "named" }...] } 741 Like cgroup_disable, but only applies to cgroup v1; 742 the blacklisted controllers remain available in cgroup2. 743 "all" blacklists all controllers and "named" disables 744 named mounts. Specifying both "all" and "named" disables 745 all v1 hierarchies. 746 747 cgroup_v1_proc= [KNL] Show also missing controllers in /proc/cgroups 748 Format: { "true" | "false" } 749 /proc/cgroups lists only v1 controllers by default. 750 This compatibility option enables listing also v2 751 controllers (whose v1 code is not compiled!), so that 752 semi-legacy software can check this file to decide 753 about usage of v2 (sic) controllers. 754 755 cgroup_favordynmods= [KNL] Enable or Disable favordynmods. 756 Format: { "true" | "false" } 757 Defaults to the value of CONFIG_CGROUP_FAVOR_DYNMODS. 758 759 cgroup.memory= [KNL] Pass options to the cgroup memory controller. 760 Format: <string> 761 nosocket -- Disable socket memory accounting. 762 nokmem -- Disable kernel memory accounting. 763 nobpf -- Disable BPF memory accounting. 764 765 check_pages= [MM,EARLY] Enable sanity checking of pages after 766 allocations / before freeing. This adds checks to catch 767 double-frees, use-after-frees, and other sources of 768 page corruption by inspecting page internals (flags, 769 mapcount/refcount, memcg_data, etc.). 770 Format: { "0" | "1" } 771 Default: 0 (1 if CONFIG_DEBUG_VM is set) 772 773 checkreqprot= [SELINUX] Set initial checkreqprot flag value. 774 Format: { "0" | "1" } 775 See security/selinux/Kconfig help text. 776 0 -- check protection applied by kernel (includes 777 any implied execute protection). 778 1 -- check protection requested by application. 779 Default value is set via a kernel config option. 780 Value can be changed at runtime via 781 /sys/fs/selinux/checkreqprot. 782 Setting checkreqprot to 1 is deprecated. 783 784 cio_ignore= [S390] 785 See Documentation/arch/s390/common_io.rst for details. 786 787 clk_ignore_unused 788 [CLK] 789 Prevents the clock framework from automatically gating 790 clocks that have not been explicitly enabled by a Linux 791 device driver but are enabled in hardware at reset or 792 by the bootloader/firmware. Note that this does not 793 force such clocks to be always-on nor does it reserve 794 those clocks in any way. This parameter is useful for 795 debug and development, but should not be needed on a 796 platform with proper driver support. For more 797 information, see Documentation/driver-api/clk.rst. 798 799 clock= [BUGS=X86-32, HW] gettimeofday clocksource override. 800 [Deprecated] 801 Forces specified clocksource (if available) to be used 802 when calculating gettimeofday(). If specified 803 clocksource is not available, it defaults to PIT. 804 Format: { pit | tsc | cyclone | pmtmr } 805 806 clocksource= Override the default clocksource 807 Format: <string> 808 Override the default clocksource and use the clocksource 809 with the name specified. 810 Some clocksource names to choose from, depending on 811 the platform: 812 [all] jiffies (this is the base, fallback clocksource) 813 [ACPI] acpi_pm 814 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2, 815 pxa_timer,timer3,32k_counter,timer0_1 816 [X86-32] pit,hpet,tsc; 817 scx200_hrt on Geode; cyclone on IBM x440 818 [MIPS] MIPS 819 [PARISC] cr16 820 [S390] tod 821 [SH] SuperH 822 [SPARC64] tick 823 [X86-64] hpet,tsc 824 825 clocksource.arm_arch_timer.evtstrm= 826 [ARM,ARM64,EARLY] 827 Format: <bool> 828 Enable/disable the eventstream feature of the ARM 829 architected timer so that code using WFE-based polling 830 loops can be debugged more effectively on production 831 systems. 832 833 clocksource.verify_n_cpus= [KNL] 834 Limit the number of CPUs checked for clocksources 835 marked with CLOCK_SOURCE_VERIFY_PERCPU that 836 are marked unstable due to excessive skew. 837 A negative value says to check all CPUs, while 838 zero says not to check any. Values larger than 839 nr_cpu_ids are silently truncated to nr_cpu_ids. 840 The actual CPUs are chosen randomly, with 841 no replacement if the same CPU is chosen twice. 842 843 clocksource-wdtest.holdoff= [KNL] 844 Set the time in seconds that the clocksource 845 watchdog test waits before commencing its tests. 846 Defaults to zero when built as a module and to 847 10 seconds when built into the kernel. 848 849 cma=nn[MG]@[start[MG][-end[MG]]] 850 [KNL,CMA,EARLY] 851 Sets the size of kernel global memory area for 852 contiguous memory allocations and optionally the 853 placement constraint by the physical address range of 854 memory allocations. A value of 0 disables CMA 855 altogether. For more information, see 856 kernel/dma/contiguous.c 857 858 cma_pernuma=nn[MG] 859 [KNL,CMA,EARLY] 860 Sets the size of kernel per-numa memory area for 861 contiguous memory allocations. A value of 0 disables 862 per-numa CMA altogether. And If this option is not 863 specified, the default value is 0. 864 With per-numa CMA enabled, DMA users on node nid will 865 first try to allocate buffer from the pernuma area 866 which is located in node nid, if the allocation fails, 867 they will fallback to the global default memory area. 868 869 numa_cma=<node>:nn[MG][,<node>:nn[MG]] 870 [KNL,CMA,EARLY] 871 Sets the size of kernel numa memory area for 872 contiguous memory allocations. It will reserve CMA 873 area for the specified node. 874 875 If it is setup together with upper 'cmd_pernuma=' 876 (unlikely), its size setting takes priority for the 877 specified numa nodes. 878 879 With numa CMA enabled, DMA users on node nid will 880 first try to allocate buffer from the numa area 881 which is located in node nid, if the allocation fails, 882 they will fallback to the global default memory area. 883 884 cmo_free_hint= [PPC] Format: { yes | no } 885 Specify whether pages are marked as being inactive 886 when they are freed. This is used in CMO environments 887 to determine OS memory pressure for page stealing by 888 a hypervisor. 889 Default: yes 890 891 coherent_pool=nn[KMG] [ARM,KNL,EARLY] 892 Sets the size of memory pool for coherent, atomic dma 893 allocations, by default set to 256K. 894 895 condev= [HW,S390] console device 896 conmode= 897 898 con3215_drop= [S390,EARLY] 3215 console drop mode. 899 Format: y|n|Y|N|1|0 900 When set to true, drop data on the 3215 console when 901 the console buffer is full. In this case the 902 operator using a 3270 terminal emulator (for example 903 x3270) does not have to enter the clear key for the 904 console output to advance and the kernel to continue. 905 This leads to a much faster boot time when a 3270 906 terminal emulator is active. If no 3270 terminal 907 emulator is used, this parameter has no effect. 908 909 console= [KNL] Output console device and options. 910 911 tty<n> Use the virtual console device <n>. 912 913 ttyS<n>[,options] 914 ttyUSB0[,options] 915 Use the specified serial port. The options are of 916 the form "bbbbpnf", where "bbbb" is the baud rate, 917 "p" is parity ("n", "o", or "e"), "n" is number of 918 bits, and "f" is flow control ("r" for RTS or 919 omit it). Default is "9600n8". 920 921 See Documentation/admin-guide/serial-console.rst for more 922 information. See 923 Documentation/networking/netconsole.rst for an 924 alternative. 925 926 <DEVNAME>:<n>.<n>[,options] 927 Use the specified serial port on the serial core bus. 928 The addressing uses DEVNAME of the physical serial port 929 device, followed by the serial core controller instance, 930 and the serial port instance. The options are the same 931 as documented for the ttyS addressing above. 932 933 The mapping of the serial ports to the tty instances 934 can be viewed with: 935 936 $ ls -d /sys/bus/serial-base/devices/*:*.*/tty/* 937 /sys/bus/serial-base/devices/00:04:0.0/tty/ttyS0 938 939 In the above example, the console can be addressed with 940 console=00:04:0.0. Note that a console addressed this 941 way will only get added when the related device driver 942 is ready. The use of an earlycon parameter in addition to 943 the console may be desired for console output early on. 944 945 uart[8250],io,<addr>[,options] 946 uart[8250],mmio,<addr>[,options] 947 uart[8250],mmio16,<addr>[,options] 948 uart[8250],mmio32,<addr>[,options] 949 uart[8250],0x<addr>[,options] 950 Start an early, polled-mode console on the 8250/16550 951 UART at the specified I/O port or MMIO address, 952 switching to the matching ttyS device later. 953 MMIO inter-register address stride is either 8-bit 954 (mmio), 16-bit (mmio16), or 32-bit (mmio32). 955 If none of [io|mmio|mmio16|mmio32], <addr> is assumed 956 to be equivalent to 'mmio'. 'options' are specified in 957 the same format described for ttyS above; if unspecified, 958 the h/w is not re-initialized. 959 960 hvc<n> Use the hypervisor console device <n>. This is for 961 both Xen and PowerPC hypervisors. 962 963 { null | "" } 964 Use to disable console output, i.e., to have kernel 965 console messages discarded. 966 This must be the only console= parameter used on the 967 kernel command line. 968 969 If the device connected to the port is not a TTY but a braille 970 device, prepend "brl," before the device type, for instance 971 console=brl,ttyS0 972 For now, only VisioBraille is supported. 973 974 console_msg_format= 975 [KNL] Change console messages format 976 default 977 By default we print messages on consoles in 978 "[time stamp] text\n" format (time stamp may not be 979 printed, depending on CONFIG_PRINTK_TIME or 980 `printk_time' param). 981 syslog 982 Switch to syslog format: "<%u>[time stamp] text\n" 983 IOW, each message will have a facility and loglevel 984 prefix. The format is similar to one used by syslog() 985 syscall, or to executing "dmesg -S --raw" or to reading 986 from /proc/kmsg. 987 988 consoleblank= [KNL] The console blank (screen saver) timeout in 989 seconds. A value of 0 disables the blank timer. 990 Defaults to 0. 991 992 coredump_filter= 993 [KNL] Change the default value for 994 /proc/<pid>/coredump_filter. 995 See also Documentation/filesystems/proc.rst. 996 997 coresight_cpu_debug.enable 998 [ARM,ARM64] 999 Format: <bool> 1000 Enable/disable the CPU sampling based debugging. 1001 0: default value, disable debugging 1002 1: enable debugging at boot time 1003 1004 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver 1005 Format: 1006 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>] 1007 1008 cpuidle.off=1 [CPU_IDLE] 1009 disable the cpuidle sub-system 1010 1011 cpuidle.governor= 1012 [CPU_IDLE] Name of the cpuidle governor to use. 1013 1014 cpufreq.off=1 [CPU_FREQ] 1015 disable the cpufreq sub-system 1016 1017 cpufreq.default_governor= 1018 [CPU_FREQ] Name of the default cpufreq governor or 1019 policy to use. This governor must be registered in the 1020 kernel before the cpufreq driver probes. 1021 1022 cpu_init_udelay=N 1023 [X86,EARLY] Delay for N microsec between assert and de-assert 1024 of APIC INIT to start processors. This delay occurs 1025 on every CPU online, such as boot, and resume from suspend. 1026 Default: 10000 1027 1028 cpuhp.parallel= 1029 [SMP] Enable/disable parallel bringup of secondary CPUs 1030 Format: <bool> 1031 Default is enabled if CONFIG_HOTPLUG_PARALLEL=y. Otherwise 1032 the parameter has no effect. 1033 1034 crash_kexec_post_notifiers 1035 Only jump to kdump kernel after running the panic 1036 notifiers and dumping kmsg. This option increases 1037 the risks of a kdump failure, since some panic 1038 notifiers can make the crashed kernel more unstable. 1039 In configurations where kdump may not be reliable, 1040 running the panic notifiers could allow collecting 1041 more data on dmesg, like stack traces from other CPUS 1042 or extra data dumped by panic_print. Note that some 1043 configurations enable this option unconditionally, 1044 like Hyper-V, PowerPC (fadump) and AMD SEV-SNP. 1045 1046 crashkernel=size[KMG][@offset[KMG]] 1047 [KNL,EARLY] Using kexec, Linux can switch to a 'crash kernel' 1048 upon panic. This parameter reserves the physical 1049 memory region [offset, offset + size] for that kernel 1050 image. If '@offset' is omitted, then a suitable offset 1051 is selected automatically. 1052 [KNL, X86-64, ARM64, RISCV, LoongArch] Select a region 1053 under 4G first, and fall back to reserve region above 1054 4G when '@offset' hasn't been specified. 1055 See Documentation/admin-guide/kdump/kdump.rst for further details. 1056 1057 crashkernel=range1:size1[,range2:size2,...][@offset] 1058 [KNL] Same as above, but depends on the memory 1059 in the running system. The syntax of range is 1060 start-[end] where start and end are both 1061 a memory unit (amount[KMG]). See also 1062 Documentation/admin-guide/kdump/kdump.rst for an example. 1063 1064 crashkernel=size[KMG],high 1065 [KNL, X86-64, ARM64, RISCV, LoongArch] range could be 1066 above 4G. 1067 Allow kernel to allocate physical memory region from top, 1068 so could be above 4G if system have more than 4G ram 1069 installed. Otherwise memory region will be allocated 1070 below 4G, if available. 1071 It will be ignored if crashkernel=X is specified. 1072 crashkernel=size[KMG],low 1073 [KNL, X86-64, ARM64, RISCV, LoongArch] range under 4G. 1074 When crashkernel=X,high is passed, kernel could allocate 1075 physical memory region above 4G, that cause second kernel 1076 crash on system that require some amount of low memory, 1077 e.g. swiotlb requires at least 64M+32K low memory, also 1078 enough extra low memory is needed to make sure DMA buffers 1079 for 32-bit devices won't run out. Kernel would try to allocate 1080 default size of memory below 4G automatically. The default 1081 size is platform dependent. 1082 --> x86: max(swiotlb_size_or_default() + 8MiB, 256MiB) 1083 --> arm64: 128MiB 1084 --> riscv: 128MiB 1085 --> loongarch: 128MiB 1086 This one lets the user specify own low range under 4G 1087 for second kernel instead. 1088 0: to disable low allocation. 1089 It will be ignored when crashkernel=X,high is not used 1090 or memory reserved is below 4G. 1091 crashkernel=size[KMG],cma 1092 [KNL, X86, ppc] Reserve additional crash kernel memory from 1093 CMA. This reservation is usable by the first system's 1094 userspace memory and kernel movable allocations (memory 1095 balloon, zswap). Pages allocated from this memory range 1096 will not be included in the vmcore so this should not 1097 be used if dumping of userspace memory is intended and 1098 it has to be expected that some movable kernel pages 1099 may be missing from the dump. 1100 1101 A standard crashkernel reservation, as described above, 1102 is still needed to hold the crash kernel and initrd. 1103 1104 This option increases the risk of a kdump failure: DMA 1105 transfers configured by the first kernel may end up 1106 corrupting the second kernel's memory. 1107 1108 This reservation method is intended for systems that 1109 can't afford to sacrifice enough memory for standard 1110 crashkernel reservation and where less reliable and 1111 possibly incomplete kdump is preferable to no kdump at 1112 all. 1113 1114 cryptomgr.notests 1115 [KNL] Disable crypto self-tests 1116 1117 cs89x0_dma= [HW,NET] 1118 Format: <dma> 1119 1120 cs89x0_media= [HW,NET] 1121 Format: { rj45 | aui | bnc } 1122 1123 csdlock_debug= [KNL] Enable or disable debug add-ons of cross-CPU 1124 function call handling. When switched on, 1125 additional debug data is printed to the console 1126 in case a hanging CPU is detected, and that 1127 CPU is pinged again in order to try to resolve 1128 the hang situation. The default value of this 1129 option depends on the CSD_LOCK_WAIT_DEBUG_DEFAULT 1130 Kconfig option. 1131 1132 dasd= [HW,NET] 1133 See header of drivers/s390/block/dasd_devmap.c. 1134 1135 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port 1136 (one device per port) 1137 Format: <port#>,<type> 1138 See also Documentation/input/devices/joystick-parport.rst 1139 1140 debug [KNL,EARLY] Enable kernel debugging (events log level). 1141 1142 debug_boot_weak_hash 1143 [KNL,EARLY] Enable printing [hashed] pointers early in the 1144 boot sequence. If enabled, we use a weak hash instead 1145 of siphash to hash pointers. Use this option if you are 1146 seeing instances of '(___ptrval___)') and need to see a 1147 value (hashed pointer) instead. Cryptographically 1148 insecure, please do not use on production kernels. 1149 1150 debug_locks_verbose= 1151 [KNL] verbose locking self-tests 1152 Format: <int> 1153 Print debugging info while doing the locking API 1154 self-tests. 1155 Bitmask for the various LOCKTYPE_ tests. Defaults to 0 1156 (no extra messages), setting it to -1 (all bits set) 1157 will print _a_lot_ more information - normally only 1158 useful to lockdep developers. 1159 1160 debug_objects [KNL,EARLY] Enable object debugging 1161 1162 debug_guardpage_minorder= 1163 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this 1164 parameter allows control of the order of pages that will 1165 be intentionally kept free (and hence protected) by the 1166 buddy allocator. Bigger value increase the probability 1167 of catching random memory corruption, but reduce the 1168 amount of memory for normal system use. The maximum 1169 possible value is MAX_PAGE_ORDER/2. Setting this 1170 parameter to 1 or 2 should be enough to identify most 1171 random memory corruption problems caused by bugs in 1172 kernel or driver code when a CPU writes to (or reads 1173 from) a random memory location. Note that there exists 1174 a class of memory corruptions problems caused by buggy 1175 H/W or F/W or by drivers badly programming DMA 1176 (basically when memory is written at bus level and the 1177 CPU MMU is bypassed) which are not detectable by 1178 CONFIG_DEBUG_PAGEALLOC, hence this option will not 1179 help tracking down these problems. 1180 1181 debug_pagealloc= 1182 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this parameter 1183 enables the feature at boot time. By default, it is 1184 disabled and the system will work mostly the same as a 1185 kernel built without CONFIG_DEBUG_PAGEALLOC. 1186 Note: to get most of debug_pagealloc error reports, it's 1187 useful to also enable the page_owner functionality. 1188 on: enable the feature 1189 1190 debugfs= [KNL,EARLY] This parameter enables what is exposed to 1191 userspace and debugfs internal clients. 1192 Format: { on, off } 1193 on: All functions are enabled. 1194 off: Filesystem is not registered and clients 1195 get a -EPERM as result when trying to register files 1196 or directories within debugfs. 1197 This is equivalent of the runtime functionality if 1198 debugfs was not enabled in the kernel at all. 1199 Default value is set in build-time with a kernel configuration. 1200 1201 debugpat [X86] Enable PAT debugging 1202 1203 default_hugepagesz= 1204 [HW] The size of the default HugeTLB page. This is 1205 the size represented by the legacy /proc/ hugepages 1206 APIs. In addition, this is the default hugetlb size 1207 used for shmget(), mmap() and mounting hugetlbfs 1208 filesystems. If not specified, defaults to the 1209 architecture's default huge page size. Huge page 1210 sizes are architecture dependent. See also 1211 Documentation/admin-guide/mm/hugetlbpage.rst. 1212 Format: size[KMG] 1213 1214 deferred_probe_timeout= 1215 [KNL] Debugging option to set a timeout in seconds for 1216 deferred probe to give up waiting on dependencies to 1217 probe. Only specific dependencies (subsystems or 1218 drivers) that have opted in will be ignored. A timeout 1219 of 0 will timeout at the end of initcalls; a negative 1220 value is treated as an infinite timeout value. If the 1221 timeout hasn't expired, it'll be restarted by each 1222 successful driver registration. This option will also 1223 dump out devices still on the deferred probe list after 1224 retrying. 1225 1226 delayacct [KNL] Enable per-task delay accounting 1227 1228 dell_smm_hwmon.ignore_dmi= 1229 [HW] Continue probing hardware even if DMI data 1230 indicates that the driver is running on unsupported 1231 hardware. 1232 1233 dell_smm_hwmon.force= 1234 [HW] Activate driver even if SMM BIOS signature does 1235 not match list of supported models and enable otherwise 1236 blacklisted features. 1237 1238 dell_smm_hwmon.power_status= 1239 [HW] Report power status in /proc/i8k 1240 (disabled by default). 1241 1242 dell_smm_hwmon.restricted= 1243 [HW] Allow controlling fans only if SYS_ADMIN 1244 capability is set. 1245 1246 dell_smm_hwmon.fan_mult= 1247 [HW] Factor to multiply fan speed with. 1248 1249 dell_smm_hwmon.fan_max= 1250 [HW] Maximum configurable fan speed. 1251 1252 dfltcc= [HW,S390] 1253 Format: { on | off | def_only | inf_only | always } 1254 on: s390 zlib hardware support for compression on 1255 level 1 and decompression (default) 1256 off: No s390 zlib hardware support 1257 def_only: s390 zlib hardware support for deflate 1258 only (compression on level 1) 1259 inf_only: s390 zlib hardware support for inflate 1260 only (decompression) 1261 always: Same as 'on' but ignores the selected compression 1262 level always using hardware support (used for debugging) 1263 1264 dhash_entries= [KNL] 1265 Set number of hash buckets for dentry cache. 1266 1267 disable_1tb_segments [PPC,EARLY] 1268 Disables the use of 1TB hash page table segments. This 1269 causes the kernel to fall back to 256MB segments which 1270 can be useful when debugging issues that require an SLB 1271 miss to occur. 1272 1273 disable= [IPV6] 1274 See Documentation/networking/ipv6.rst. 1275 1276 disable_radix [PPC,EARLY] 1277 Disable RADIX MMU mode on POWER9 1278 1279 disable_tlbie [PPC] 1280 Disable TLBIE instruction. Currently does not work 1281 with KVM, with HASH MMU, or with coherent accelerators. 1282 1283 disable_ddw [PPC/PSERIES,EARLY] 1284 Disable Dynamic DMA Window support. Use this 1285 to workaround buggy firmware. 1286 1287 disable_ipv6= [IPV6] 1288 See Documentation/networking/ipv6.rst. 1289 1290 disable_mtrr_cleanup [X86,EARLY] 1291 The kernel tries to adjust MTRR layout from continuous 1292 to discrete, to make X server driver able to add WB 1293 entry later. This parameter disables that. 1294 1295 disable_mtrr_trim [X86, Intel and AMD only,EARLY] 1296 By default the kernel will trim any uncacheable 1297 memory out of your available memory pool based on 1298 MTRR settings. This parameter disables that behavior, 1299 possibly causing your machine to run very slowly. 1300 1301 disable_timer_pin_1 [X86,EARLY] 1302 Disable PIN 1 of APIC timer 1303 Can be useful to work around chipset bugs. 1304 1305 dis_ucode_ldr [X86] Disable the microcode loader. 1306 1307 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support, 1308 this option disables the debugging code at boot. 1309 1310 dma_debug_entries=<number> 1311 This option allows to tune the number of preallocated 1312 entries for DMA-API debugging code. One entry is 1313 required per DMA-API allocation. Use this if the 1314 DMA-API debugging code disables itself because the 1315 architectural default is too low. 1316 1317 dma_debug_driver=<driver_name> 1318 With this option the DMA-API debugging driver 1319 filter feature can be enabled at boot time. Just 1320 pass the driver to filter for as the parameter. 1321 The filter can be disabled or changed to another 1322 driver later using sysfs. 1323 1324 reg_file_data_sampling= 1325 [X86] Controls mitigation for Register File Data 1326 Sampling (RFDS) vulnerability. RFDS is a CPU 1327 vulnerability which may allow userspace to infer 1328 kernel data values previously stored in floating point 1329 registers, vector registers, or integer registers. 1330 RFDS only affects Intel Atom processors. 1331 1332 on: Turns ON the mitigation. 1333 off: Turns OFF the mitigation. 1334 1335 This parameter overrides the compile time default set 1336 by CONFIG_MITIGATION_RFDS. Mitigation cannot be 1337 disabled when other VERW based mitigations (like MDS) 1338 are enabled. In order to disable RFDS mitigation all 1339 VERW based mitigations need to be disabled. 1340 1341 For details see: 1342 Documentation/admin-guide/hw-vuln/reg-file-data-sampling.rst 1343 1344 dm_verity.keyring_unsealed= 1345 [KNL] When set to 1, leave the dm-verity keyring 1346 unsealed after initialization so userspace can 1347 provision keys. Once the keyring is restricted 1348 it becomes active and is searched during signature 1349 verification. 1350 1351 driver_async_probe= [KNL] 1352 List of driver names to be probed asynchronously. * 1353 matches with all driver names. If * is specified, the 1354 rest of the listed driver names are those that will NOT 1355 match the *. 1356 Format: <driver_name1>,<driver_name2>... 1357 1358 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>] 1359 Broken monitors, graphic adapters, KVMs and EDIDless 1360 panels may send no or incorrect EDID data sets. 1361 This parameter allows to specify an EDID data sets 1362 in the /lib/firmware directory that are used instead. 1363 An EDID data set will only be used for a particular 1364 connector, if its name and a colon are prepended to 1365 the EDID name. Each connector may use a unique EDID 1366 data set by separating the files with a comma. An EDID 1367 data set with no connector name will be used for 1368 any connectors not explicitly specified. 1369 1370 dscc4.setup= [NET] 1371 1372 dt_cpu_ftrs= [PPC,EARLY] 1373 Format: {"off" | "known"} 1374 Control how the dt_cpu_ftrs device-tree binding is 1375 used for CPU feature discovery and setup (if it 1376 exists). 1377 off: Do not use it, fall back to legacy cpu table. 1378 known: Do not pass through unknown features to guests 1379 or userspace, only those that the kernel is aware of. 1380 1381 dump_apple_properties [X86] 1382 Dump name and content of EFI device properties on 1383 x86 Macs. Useful for driver authors to determine 1384 what data is available or for reverse-engineering. 1385 1386 dyndbg[="val"] [KNL,DYNAMIC_DEBUG] 1387 <module>.dyndbg[="val"] 1388 Enable debug messages at boot time. See 1389 Documentation/admin-guide/dynamic-debug-howto.rst 1390 for details. 1391 1392 early_ioremap_debug [KNL,EARLY] 1393 Enable debug messages in early_ioremap support. This 1394 is useful for tracking down temporary early mappings 1395 which are not unmapped. 1396 1397 earlycon= [KNL,EARLY] Output early console device and options. 1398 1399 When used with no options, the early console is 1400 determined by stdout-path property in device tree's 1401 chosen node or the ACPI SPCR table if supported by 1402 the platform. 1403 1404 cdns,<addr>[,options] 1405 Start an early, polled-mode console on a Cadence 1406 (xuartps) serial port at the specified address. Only 1407 supported option is baud rate. If baud rate is not 1408 specified, the serial port must already be setup and 1409 configured. 1410 1411 uart[8250],io,<addr>[,options[,uartclk]] 1412 uart[8250],mmio,<addr>[,options[,uartclk]] 1413 uart[8250],mmio32,<addr>[,options[,uartclk]] 1414 uart[8250],mmio32be,<addr>[,options[,uartclk]] 1415 uart[8250],0x<addr>[,options] 1416 Start an early, polled-mode console on the 8250/16550 1417 UART at the specified I/O port or MMIO address. 1418 MMIO inter-register address stride is either 8-bit 1419 (mmio) or 32-bit (mmio32 or mmio32be). 1420 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed 1421 to be equivalent to 'mmio'. 'options' are specified 1422 in the same format described for "console=ttyS<n>"; if 1423 unspecified, the h/w is not initialized. 'uartclk' is 1424 the uart clock frequency; if unspecified, it is set 1425 to 'BASE_BAUD' * 16. 1426 1427 pl011,<addr> 1428 pl011,mmio32,<addr> 1429 Start an early, polled-mode console on a pl011 serial 1430 port at the specified address. The pl011 serial port 1431 must already be setup and configured. Options are not 1432 yet supported. If 'mmio32' is specified, then only 1433 the driver will use only 32-bit accessors to read/write 1434 the device registers. 1435 1436 liteuart,<addr> 1437 Start an early console on a litex serial port at the 1438 specified address. The serial port must already be 1439 setup and configured. Options are not yet supported. 1440 1441 meson,<addr> 1442 Start an early, polled-mode console on a meson serial 1443 port at the specified address. The serial port must 1444 already be setup and configured. Options are not yet 1445 supported. 1446 1447 msm_serial,<addr> 1448 Start an early, polled-mode console on an msm serial 1449 port at the specified address. The serial port 1450 must already be setup and configured. Options are not 1451 yet supported. 1452 1453 msm_serial_dm,<addr> 1454 Start an early, polled-mode console on an msm serial 1455 dm port at the specified address. The serial port 1456 must already be setup and configured. Options are not 1457 yet supported. 1458 1459 owl,<addr> 1460 Start an early, polled-mode console on a serial port 1461 of an Actions Semi SoC, such as S500 or S900, at the 1462 specified address. The serial port must already be 1463 setup and configured. Options are not yet supported. 1464 1465 rda,<addr> 1466 Start an early, polled-mode console on a serial port 1467 of an RDA Micro SoC, such as RDA8810PL, at the 1468 specified address. The serial port must already be 1469 setup and configured. Options are not yet supported. 1470 1471 sbi 1472 Use RISC-V SBI (Supervisor Binary Interface) for early 1473 console. 1474 1475 smh Use ARM semihosting calls for early console. 1476 1477 s3c2410,<addr> 1478 s3c2412,<addr> 1479 s3c2440,<addr> 1480 s3c6400,<addr> 1481 s5pv210,<addr> 1482 exynos4210,<addr> 1483 Use early console provided by serial driver available 1484 on Samsung SoCs, requires selecting proper type and 1485 a correct base address of the selected UART port. The 1486 serial port must already be setup and configured. 1487 Options are not yet supported. 1488 1489 lantiq,<addr> 1490 Start an early, polled-mode console on a lantiq serial 1491 (lqasc) port at the specified address. The serial port 1492 must already be setup and configured. Options are not 1493 yet supported. 1494 1495 lpuart,<addr> 1496 lpuart32,<addr> 1497 Use early console provided by Freescale LP UART driver 1498 found on Freescale Vybrid and QorIQ LS1021A processors. 1499 A valid base address must be provided, and the serial 1500 port must already be setup and configured. 1501 1502 ec_imx21,<addr> 1503 ec_imx6q,<addr> 1504 Start an early, polled-mode, output-only console on the 1505 Freescale i.MX UART at the specified address. The UART 1506 must already be setup and configured. 1507 1508 ar3700_uart,<addr> 1509 Start an early, polled-mode console on the 1510 Armada 3700 serial port at the specified 1511 address. The serial port must already be setup 1512 and configured. Options are not yet supported. 1513 1514 qcom_geni,<addr> 1515 Start an early, polled-mode console on a Qualcomm 1516 Generic Interface (GENI) based serial port at the 1517 specified address. The serial port must already be 1518 setup and configured. Options are not yet supported. 1519 1520 efifb,[options] 1521 Start an early, unaccelerated console on the EFI 1522 memory mapped framebuffer (if available). On cache 1523 coherent non-x86 systems that use system memory for 1524 the framebuffer, pass the 'ram' option so that it is 1525 mapped with the correct attributes. 1526 1527 linflex,<addr> 1528 Use early console provided by Freescale LINFlexD UART 1529 serial driver for NXP S32V234 SoCs. A valid base 1530 address must be provided, and the serial port must 1531 already be setup and configured. 1532 1533 earlyprintk= [X86,SH,ARM,M68k,S390,UM,EARLY] 1534 earlyprintk=vga 1535 earlyprintk=sclp 1536 earlyprintk=xen 1537 earlyprintk=serial[,ttySn[,baudrate]] 1538 earlyprintk=serial[,0x...[,baudrate]] 1539 earlyprintk=ttySn[,baudrate] 1540 earlyprintk=dbgp[debugController#] 1541 earlyprintk=mmio32,membase[,{nocfg|baudrate}] 1542 earlyprintk=pciserial[,force],bus:device.function[,{nocfg|baudrate}] 1543 earlyprintk=xdbc[xhciController#] 1544 earlyprintk=bios 1545 1546 earlyprintk is useful when the kernel crashes before 1547 the normal console is initialized. It is not enabled by 1548 default because it has some cosmetic problems. 1549 1550 Use "nocfg" to skip UART configuration, assume 1551 BIOS/firmware has configured UART correctly. 1552 1553 Append ",keep" to not disable it when the real console 1554 takes over. 1555 1556 Only one of vga, serial, or usb debug port can 1557 be used at a time. 1558 1559 Currently only ttyS0 and ttyS1 may be specified by 1560 name. Other I/O ports may be explicitly specified 1561 on some architectures (x86 and arm at least) by 1562 replacing ttySn with an I/O port address, like this: 1563 earlyprintk=serial,0x1008,115200 1564 You can find the port for a given device in 1565 /proc/tty/driver/serial: 1566 2: uart:ST16650V2 port:00001008 irq:18 ... 1567 1568 Interaction with the standard serial driver is not 1569 very good. 1570 1571 The VGA output is eventually overwritten by 1572 the real console. 1573 1574 The xen option can only be used in Xen domains. 1575 1576 The sclp output can only be used on s390. 1577 1578 The bios output can only be used on SuperH. 1579 1580 The optional "force" to "pciserial" enables use of a 1581 PCI device even when its classcode is not of the 1582 UART class. 1583 1584 edac_report= [HW,EDAC] Control how to report EDAC event 1585 Format: {"on" | "off" | "force"} 1586 on: enable EDAC to report H/W event. May be overridden 1587 by other higher priority error reporting module. 1588 off: disable H/W event reporting through EDAC. 1589 force: enforce the use of EDAC to report H/W event. 1590 default: on. 1591 1592 edd= [EDD] 1593 Format: {"off" | "on" | "skip[mbr]"} 1594 1595 efi= [EFI,EARLY] 1596 Format: { "debug", "disable_early_pci_dma", 1597 "nochunk", "noruntime", "nosoftreserve", 1598 "novamap", "no_disable_early_pci_dma" } 1599 debug: enable misc debug output. 1600 disable_early_pci_dma: disable the busmaster bit on all 1601 PCI bridges while in the EFI boot stub. 1602 nochunk: disable reading files in "chunks" in the EFI 1603 boot stub, as chunking can cause problems with some 1604 firmware implementations. 1605 noruntime : disable EFI runtime services support 1606 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose) 1607 attribute may cause the kernel to reserve the 1608 memory range for a memory mapping driver to 1609 claim. Specify efi=nosoftreserve to disable this 1610 reservation and treat the memory by its base type 1611 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM"). 1612 novamap: do not call SetVirtualAddressMap(). 1613 no_disable_early_pci_dma: Leave the busmaster bit set 1614 on all PCI bridges while in the EFI boot stub 1615 1616 efi_no_storage_paranoia [EFI,X86,EARLY] 1617 Using this parameter you can use more than 50% of 1618 your efi variable storage. Use this parameter only if 1619 you are really sure that your UEFI does sane gc and 1620 fulfills the spec otherwise your board may brick. 1621 1622 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT 1623 that is to be dynamically loaded by Linux. If there are 1624 multiple variables with the same name but with different 1625 vendor GUIDs, all of them will be loaded. See 1626 Documentation/admin-guide/acpi/ssdt-overlays.rst for details. 1627 1628 1629 eisa_irq_edge= [PARISC,HW] 1630 See header of drivers/parisc/eisa.c. 1631 1632 ekgdboc= [X86,KGDB,EARLY] Allow early kernel console debugging 1633 Format: ekgdboc=kbd 1634 1635 This is designed to be used in conjunction with 1636 the boot argument: earlyprintk=vga 1637 1638 This parameter works in place of the kgdboc parameter 1639 but can only be used if the backing tty is available 1640 very early in the boot process. For early debugging 1641 via a serial port see kgdboc_earlycon instead. 1642 1643 elfcorehdr=[size[KMG]@]offset[KMG] [PPC,SH,X86,S390,EARLY] 1644 Specifies physical address of start of kernel core 1645 image elf header and optionally the size. Generally 1646 kexec loader will pass this option to capture kernel. 1647 See Documentation/admin-guide/kdump/kdump.rst for details. 1648 1649 enable_mtrr_cleanup [X86,EARLY] 1650 The kernel tries to adjust MTRR layout from continuous 1651 to discrete, to make X server driver able to add WB 1652 entry later. This parameter enables that. 1653 1654 enable_timer_pin_1 [X86] 1655 Enable PIN 1 of APIC timer 1656 Can be useful to work around chipset bugs 1657 (in particular on some ATI chipsets). 1658 The kernel tries to set a reasonable default. 1659 1660 enforcing= [SELINUX] Set initial enforcing status. 1661 Format: {"0" | "1"} 1662 See security/selinux/Kconfig help text. 1663 0 -- permissive (log only, no denials). 1664 1 -- enforcing (deny and log). 1665 Default value is 0. 1666 Value can be changed at runtime via 1667 /sys/fs/selinux/enforce. 1668 1669 erst_disable [ACPI] 1670 Disable Error Record Serialization Table (ERST) 1671 support. 1672 1673 ether= [HW,NET] Ethernet cards parameters 1674 This option is obsoleted by the "netdev=" option, which 1675 has equivalent usage. See its documentation for details. 1676 1677 evm= [EVM] 1678 Format: { "fix" } 1679 Permit 'security.evm' to be updated regardless of 1680 current integrity status. 1681 1682 early_page_ext [KNL,EARLY] Enforces page_ext initialization to earlier 1683 stages so cover more early boot allocations. 1684 Please note that as side effect some optimizations 1685 might be disabled to achieve that (e.g. parallelized 1686 memory initialization is disabled) so the boot process 1687 might take longer, especially on systems with a lot of 1688 memory. Available with CONFIG_PAGE_EXTENSION=y. 1689 1690 failslab= 1691 fail_usercopy= 1692 fail_page_alloc= 1693 fail_skb_realloc= 1694 fail_make_request=[KNL] 1695 General fault injection mechanism. 1696 Format: <interval>,<probability>,<space>,<times> 1697 See also Documentation/fault-injection/. 1698 1699 fb_tunnels= [NET] 1700 Format: { initns | none } 1701 See Documentation/admin-guide/sysctl/net.rst for 1702 fb_tunnels_only_for_init_ns 1703 1704 floppy= [HW] 1705 See Documentation/admin-guide/blockdev/floppy.rst. 1706 1707 forcepae [X86-32] 1708 Forcefully enable Physical Address Extension (PAE). 1709 Many Pentium M systems disable PAE but may have a 1710 functionally usable PAE implementation. 1711 Warning: use of this parameter will taint the kernel 1712 and may cause unknown problems. 1713 1714 fred= [X86-64] 1715 Enable/disable Flexible Return and Event Delivery. 1716 Format: { on | off } 1717 on: enable FRED when it's present, the default setting. 1718 off: disable FRED. 1719 1720 ftrace=[tracer] 1721 [FTRACE] will set and start the specified tracer 1722 as early as possible in order to facilitate early 1723 boot debugging. 1724 1725 ftrace_boot_snapshot 1726 [FTRACE] On boot up, a snapshot will be taken of the 1727 ftrace ring buffer that can be read at: 1728 /sys/kernel/tracing/snapshot. 1729 This is useful if you need tracing information from kernel 1730 boot up that is likely to be overridden by user space 1731 start up functionality. 1732 1733 Optionally, the snapshot can also be defined for a tracing 1734 instance that was created by the trace_instance= command 1735 line parameter. 1736 1737 trace_instance=foo,sched_switch ftrace_boot_snapshot=foo 1738 1739 The above will cause the "foo" tracing instance to trigger 1740 a snapshot at the end of boot up. 1741 1742 ftrace_dump_on_oops[=2(orig_cpu) | =<instance>][,<instance> | 1743 ,<instance>=2(orig_cpu)] 1744 [FTRACE] will dump the trace buffers on oops. 1745 If no parameter is passed, ftrace will dump global 1746 buffers of all CPUs, if you pass 2 or orig_cpu, it 1747 will dump only the buffer of the CPU that triggered 1748 the oops, or the specific instance will be dumped if 1749 its name is passed. Multiple instance dump is also 1750 supported, and instances are separated by commas. Each 1751 instance supports only dump on CPU that triggered the 1752 oops by passing 2 or orig_cpu to it. 1753 1754 ftrace_dump_on_oops=foo=orig_cpu 1755 1756 The above will dump only the buffer of "foo" instance 1757 on CPU that triggered the oops. 1758 1759 ftrace_dump_on_oops,foo,bar=orig_cpu 1760 1761 The above will dump global buffer on all CPUs, the 1762 buffer of "foo" instance on all CPUs and the buffer 1763 of "bar" instance on CPU that triggered the oops. 1764 1765 ftrace_filter=[function-list] 1766 [FTRACE] Limit the functions traced by the function 1767 tracer at boot up. function-list is a comma-separated 1768 list of functions. This list can be changed at run 1769 time by the set_ftrace_filter file in the debugfs 1770 tracing directory. 1771 1772 ftrace_notrace=[function-list] 1773 [FTRACE] Do not trace the functions specified in 1774 function-list. This list can be changed at run time 1775 by the set_ftrace_notrace file in the debugfs 1776 tracing directory. 1777 1778 ftrace_graph_filter=[function-list] 1779 [FTRACE] Limit the top level callers functions traced 1780 by the function graph tracer at boot up. 1781 function-list is a comma-separated list of functions 1782 that can be changed at run time by the 1783 set_graph_function file in the debugfs tracing directory. 1784 1785 ftrace_graph_notrace=[function-list] 1786 [FTRACE] Do not trace from the functions specified in 1787 function-list. This list is a comma-separated list of 1788 functions that can be changed at run time by the 1789 set_graph_notrace file in the debugfs tracing directory. 1790 1791 ftrace_graph_max_depth=<uint> 1792 [FTRACE] Used with the function graph tracer. This is 1793 the max depth it will trace into a function. This value 1794 can be changed at run time by the max_graph_depth file 1795 in the tracefs tracing directory. default: 0 (no limit) 1796 1797 fw_devlink= [KNL,EARLY] Create device links between consumer and supplier 1798 devices by scanning the firmware to infer the 1799 consumer/supplier relationships. This feature is 1800 especially useful when drivers are loaded as modules as 1801 it ensures proper ordering of tasks like device probing 1802 (suppliers first, then consumers), supplier boot state 1803 clean up (only after all consumers have probed), 1804 suspend/resume & runtime PM (consumers first, then 1805 suppliers). 1806 Format: { off | permissive | on | rpm } 1807 off -- Don't create device links from firmware info. 1808 permissive -- Create device links from firmware info 1809 but use it only for ordering boot state clean 1810 up (sync_state() calls). 1811 on -- Create device links from firmware info and use it 1812 to enforce probe and suspend/resume ordering. 1813 rpm -- Like "on", but also use to order runtime PM. 1814 1815 fw_devlink.strict=<bool> 1816 [KNL,EARLY] Treat all inferred dependencies as mandatory 1817 dependencies. This only applies for fw_devlink=on|rpm. 1818 Format: <bool> 1819 1820 fw_devlink.sync_state = 1821 [KNL,EARLY] When all devices that could probe have finished 1822 probing, this parameter controls what to do with 1823 devices that haven't yet received their sync_state() 1824 calls. 1825 Format: { strict | timeout } 1826 strict -- Default. Continue waiting on consumers to 1827 probe successfully. 1828 timeout -- Give up waiting on consumers and call 1829 sync_state() on any devices that haven't yet 1830 received their sync_state() calls after 1831 deferred_probe_timeout has expired or by 1832 late_initcall() if !CONFIG_MODULES. 1833 1834 gamecon.map[2|3]= 1835 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad 1836 support via parallel port (up to 5 devices per port) 1837 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5> 1838 See also Documentation/input/devices/joystick-parport.rst 1839 1840 gamma= [HW,DRM] 1841 1842 gart_fix_e820= [X86-64,EARLY] disable the fix e820 for K8 GART 1843 Format: off | on 1844 default: on 1845 1846 gather_data_sampling= 1847 [X86,INTEL,EARLY] Control the Gather Data Sampling (GDS) 1848 mitigation. 1849 1850 Gather Data Sampling is a hardware vulnerability which 1851 allows unprivileged speculative access to data which was 1852 previously stored in vector registers. 1853 1854 This issue is mitigated by default in updated microcode. 1855 The mitigation may have a performance impact but can be 1856 disabled. On systems without the microcode mitigation 1857 disabling AVX serves as a mitigation. 1858 1859 force: Disable AVX to mitigate systems without 1860 microcode mitigation. No effect if the microcode 1861 mitigation is present. Known to cause crashes in 1862 userspace with buggy AVX enumeration. 1863 1864 off: Disable GDS mitigation. 1865 1866 gbpages [X86] Use GB pages for kernel direct mappings. 1867 1868 gcov_persist= [GCOV] When non-zero (default), profiling data for 1869 kernel modules is saved and remains accessible via 1870 debugfs, even when the module is unloaded/reloaded. 1871 When zero, profiling data is discarded and associated 1872 debugfs files are removed at module unload time. 1873 1874 goldfish [X86] Enable the goldfish android emulator platform. 1875 Don't use this when you are not running on the 1876 android emulator 1877 1878 gpio-mockup.gpio_mockup_ranges 1879 [HW] Sets the ranges of gpiochip of for this device. 1880 Format: <start1>,<end1>,<start2>,<end2>... 1881 gpio-mockup.gpio_mockup_named_lines 1882 [HW] Let the driver know GPIO lines should be named. 1883 1884 gpt [EFI] Forces disk with valid GPT signature but 1885 invalid Protective MBR to be treated as GPT. If the 1886 primary GPT is corrupted, it enables the backup/alternate 1887 GPT to be used instead. 1888 1889 grcan.enable0= [HW] Configuration of physical interface 0. Determines 1890 the "Enable 0" bit of the configuration register. 1891 Format: 0 | 1 1892 Default: 0 1893 grcan.enable1= [HW] Configuration of physical interface 1. Determines 1894 the "Enable 0" bit of the configuration register. 1895 Format: 0 | 1 1896 Default: 0 1897 grcan.select= [HW] Select which physical interface to use. 1898 Format: 0 | 1 1899 Default: 0 1900 grcan.txsize= [HW] Sets the size of the tx buffer. 1901 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0. 1902 Default: 1024 1903 grcan.rxsize= [HW] Sets the size of the rx buffer. 1904 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0. 1905 Default: 1024 1906 1907 hardened_usercopy= 1908 [KNL] Under CONFIG_HARDENED_USERCOPY, whether 1909 hardening is enabled for this boot. Hardened 1910 usercopy checking is used to protect the kernel 1911 from reading or writing beyond known memory 1912 allocation boundaries as a proactive defense 1913 against bounds-checking flaws in the kernel's 1914 copy_to_user()/copy_from_user() interface. 1915 The default is determined by 1916 CONFIG_HARDENED_USERCOPY_DEFAULT_ON. 1917 on Perform hardened usercopy checks. 1918 off Disable hardened usercopy checks. 1919 1920 hardlockup_all_cpu_backtrace= 1921 [KNL] Should the hard-lockup detector generate 1922 backtraces on all cpus. 1923 Format: 0 | 1 1924 1925 hash_pointers= 1926 [KNL,EARLY] 1927 By default, when pointers are printed to the console 1928 or buffers via the %p format string, that pointer is 1929 "hashed", i.e. obscured by hashing the pointer value. 1930 This is a security feature that hides actual kernel 1931 addresses from unprivileged users, but it also makes 1932 debugging the kernel more difficult since unequal 1933 pointers can no longer be compared. The choices are: 1934 Format: { auto | always | never } 1935 Default: auto 1936 1937 auto - Hash pointers unless slab_debug is enabled. 1938 always - Always hash pointers (even if slab_debug is 1939 enabled). 1940 never - Never hash pointers. This option should only 1941 be specified when debugging the kernel. Do 1942 not use on production kernels. The boot 1943 param "no_hash_pointers" is an alias for 1944 this mode. 1945 1946 For controlling hashing dynamically at runtime, 1947 use the "kernel.kptr_restrict" sysctl instead. 1948 1949 hashdist= [KNL,NUMA] Large hashes allocated during boot 1950 are distributed across NUMA nodes. Defaults on 1951 for 64-bit NUMA, off otherwise. 1952 Format: 0 | 1 (for off | on) 1953 1954 hd= [EIDE] (E)IDE hard drive subsystem geometry 1955 Format: <cyl>,<head>,<sect> 1956 1957 hest_disable [ACPI] 1958 Disable Hardware Error Source Table (HEST) support; 1959 corresponding firmware-first mode error processing 1960 logic will be disabled. 1961 1962 hibernate= [HIBERNATION] 1963 noresume Don't check if there's a hibernation image 1964 present during boot. 1965 nocompress Don't compress/decompress hibernation images. 1966 no Disable hibernation and resume. 1967 protect_image Turn on image protection during restoration 1968 (that will set all pages holding image data 1969 during restoration read-only). 1970 1971 hibernate.compressor= [HIBERNATION] Compression algorithm to be 1972 used with hibernation. 1973 Format: { lzo | lz4 } 1974 Default: lzo 1975 1976 lzo: Select LZO compression algorithm to 1977 compress/decompress hibernation image. 1978 1979 lz4: Select LZ4 compression algorithm to 1980 compress/decompress hibernation image. 1981 1982 hibernate.pm_test_delay= 1983 [HIBERNATION] 1984 Sets the number of seconds to remain in a hibernation test 1985 mode before resuming the system (see 1986 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG 1987 is set. Default value is 5. 1988 1989 hibernate_compression_threads= 1990 [HIBERNATION] 1991 Set the number of threads used for compressing or decompressing 1992 hibernation images. 1993 1994 Format: <integer> 1995 Default: 3 1996 Minimum: 1 1997 Example: hibernate_compression_threads=4 1998 1999 highmem=nn[KMG] [KNL,BOOT,EARLY] forces the highmem zone to have an exact 2000 size of <nn>. This works even on boxes that have no 2001 highmem otherwise. This also works to reduce highmem 2002 size on bigger boxes. 2003 2004 highres= [KNL] Enable/disable high resolution timer mode. 2005 Valid parameters: "on", "off" 2006 Default: "on" 2007 2008 hlt [BUGS=ARM,SH] 2009 2010 hostname= [KNL,EARLY] Set the hostname (aka UTS nodename). 2011 Format: <string> 2012 This allows setting the system's hostname during early 2013 startup. This sets the name returned by gethostname. 2014 Using this parameter to set the hostname makes it 2015 possible to ensure the hostname is correctly set before 2016 any userspace processes run, avoiding the possibility 2017 that a process may call gethostname before the hostname 2018 has been explicitly set, resulting in the calling 2019 process getting an incorrect result. The string must 2020 not exceed the maximum allowed hostname length (usually 2021 64 characters) and will be truncated otherwise. 2022 2023 hpet= [X86-32,HPET] option to control HPET usage 2024 Format: { enable (default) | disable | force | 2025 verbose } 2026 disable: disable HPET and use PIT instead 2027 force: allow force enabled of undocumented chips (ICH4, 2028 VIA, nVidia) 2029 verbose: show contents of HPET registers during setup 2030 2031 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET 2032 registers. Default set by CONFIG_HPET_MMAP_DEFAULT. 2033 2034 hugepages= [HW,EARLY] Number of HugeTLB pages to allocate at boot. 2035 If this follows hugepagesz (below), it specifies 2036 the number of pages of hugepagesz to be allocated. 2037 If this is the first HugeTLB parameter on the command 2038 line, it specifies the number of pages to allocate for 2039 the default huge page size. If using node format, the 2040 number of pages to allocate per-node can be specified. 2041 See also Documentation/admin-guide/mm/hugetlbpage.rst. 2042 Format: <integer> or (node format) 2043 <node>:<integer>[,<node>:<integer>] 2044 2045 hugepagesz= 2046 [HW,EARLY] The size of the HugeTLB pages. This is 2047 used in conjunction with hugepages (above) to 2048 allocate huge pages of a specific size at boot. The 2049 pair hugepagesz=X hugepages=Y can be specified once 2050 for each supported huge page size. Huge page sizes 2051 are architecture dependent. See also 2052 Documentation/admin-guide/mm/hugetlbpage.rst. 2053 Format: size[KMG] 2054 2055 hugepage_alloc_threads= 2056 [HW] The number of threads that should be used to 2057 allocate hugepages during boot. This option can be 2058 used to improve system bootup time when allocating 2059 a large amount of huge pages. 2060 The default value is 25% of the available hardware threads. 2061 2062 Note that this parameter only applies to non-gigantic huge pages. 2063 2064 hugetlb_cma= [HW,CMA,EARLY] The size of a CMA area used for allocation 2065 of gigantic hugepages. Or using node format, the size 2066 of a CMA area per node can be specified. 2067 Format: nn[KMGTPE] or (node format) 2068 <node>:nn[KMGTPE][,<node>:nn[KMGTPE]] 2069 2070 Reserve a CMA area of given size and allocate gigantic 2071 hugepages using the CMA allocator. If enabled, the 2072 boot-time allocation of gigantic hugepages is skipped. 2073 2074 hugetlb_cma_only= 2075 [HW,CMA,EARLY] When allocating new HugeTLB pages, only 2076 try to allocate from the CMA areas. 2077 2078 This option does nothing if hugetlb_cma= is not also 2079 specified. 2080 2081 hugetlb_free_vmemmap= 2082 [KNL] Requires CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP 2083 enabled. 2084 Control if HugeTLB Vmemmap Optimization (HVO) is enabled. 2085 Allows heavy hugetlb users to free up some more 2086 memory (7 * PAGE_SIZE for each 2MB hugetlb page). 2087 Format: { on | off (default) } 2088 2089 on: enable HVO 2090 off: disable HVO 2091 2092 Built with CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP_DEFAULT_ON=y, 2093 the default is on. 2094 2095 Note that the vmemmap pages may be allocated from the added 2096 memory block itself when memory_hotplug.memmap_on_memory is 2097 enabled, those vmemmap pages cannot be optimized even if this 2098 feature is enabled. Other vmemmap pages not allocated from 2099 the added memory block itself do not be affected. 2100 2101 hung_task_panic= 2102 [KNL] Number of hung tasks to trigger kernel panic. 2103 Format: <int> 2104 2105 When set to a non-zero value, a kernel panic will be triggered if 2106 the number of detected hung tasks reaches this value. 2107 2108 0: don't panic 2109 1: panic immediately on first hung task 2110 N: panic after N hung tasks are detected in a single scan 2111 2112 The default value is controlled by the 2113 CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time option. The value 2114 selected by this boot parameter can be changed later by the 2115 kernel.hung_task_panic sysctl. 2116 2117 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC) 2118 terminal devices. Valid values: 0..8 2119 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs. 2120 If specified, z/VM IUCV HVC accepts connections 2121 from listed z/VM user IDs only. 2122 2123 hv_nopvspin [X86,HYPER_V,EARLY] 2124 Disables the paravirt spinlock optimizations 2125 which allow the hypervisor to 'idle' the guest 2126 on lock contention. 2127 2128 hw_protection= [HW] 2129 Format: reboot | shutdown 2130 2131 Hardware protection action taken on critical events like 2132 overtemperature or imminent voltage loss. 2133 2134 i2c_bus= [HW] Override the default board specific I2C bus speed 2135 or register an additional I2C bus that is not 2136 registered from board initialization code. 2137 Format: 2138 <bus_id>,<clkrate> 2139 2140 i2c_touchscreen_props= [HW,ACPI,X86] 2141 Set device-properties for ACPI-enumerated I2C-attached 2142 touchscreen, to e.g. fix coordinates of upside-down 2143 mounted touchscreens. If you need this option please 2144 submit a drivers/platform/x86/touchscreen_dmi.c patch 2145 adding a DMI quirk for this. 2146 2147 Format: 2148 <ACPI_HW_ID>:<prop_name>=<val>[:prop_name=val][:...] 2149 Where <val> is one of: 2150 Omit "=<val>" entirely Set a boolean device-property 2151 Unsigned number Set a u32 device-property 2152 Anything else Set a string device-property 2153 2154 Examples (split over multiple lines): 2155 i2c_touchscreen_props=GDIX1001:touchscreen-inverted-x: 2156 touchscreen-inverted-y 2157 2158 i2c_touchscreen_props=MSSL1680:touchscreen-size-x=1920: 2159 touchscreen-size-y=1080:touchscreen-inverted-y: 2160 firmware-name=gsl1680-vendor-model.fw:silead,home-button 2161 2162 i8042.debug [HW] Toggle i8042 debug mode 2163 i8042.unmask_kbd_data 2164 [HW] Enable printing of interrupt data from the KBD port 2165 (disabled by default, and as a pre-condition 2166 requires that i8042.debug=1 be enabled) 2167 i8042.direct [HW] Put keyboard port into non-translated mode 2168 i8042.dumbkbd [HW] Pretend that controller can only read data from 2169 keyboard and cannot control its state 2170 (Don't attempt to blink the leds) 2171 i8042.noaux [HW] Don't check for auxiliary (== mouse) port 2172 i8042.nokbd [HW] Don't check/create keyboard port 2173 i8042.noloop [HW] Disable the AUX Loopback command while probing 2174 for the AUX port 2175 i8042.nomux [HW] Don't check presence of an active multiplexing 2176 controller 2177 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX 2178 controllers 2179 i8042.notimeout [HW] Ignore timeout condition signalled by controller 2180 i8042.reset [HW] Reset the controller during init, cleanup and 2181 suspend-to-ram transitions, only during s2r 2182 transitions, or never reset 2183 Format: { 1 | Y | y | 0 | N | n } 2184 1, Y, y: always reset controller 2185 0, N, n: don't ever reset controller 2186 Default: only on s2r transitions on x86; most other 2187 architectures force reset to be always executed 2188 i8042.unlock [HW] Unlock (ignore) the keylock 2189 i8042.kbdreset [HW] Reset device connected to KBD port 2190 i8042.probe_defer 2191 [HW] Allow deferred probing upon i8042 probe errors 2192 2193 i810= [HW,DRM] 2194 2195 i915.invert_brightness= 2196 [DRM] Invert the sense of the variable that is used to 2197 set the brightness of the panel backlight. Normally a 2198 brightness value of 0 indicates backlight switched off, 2199 and the maximum of the brightness value sets the backlight 2200 to maximum brightness. If this parameter is set to 0 2201 (default) and the machine requires it, or this parameter 2202 is set to 1, a brightness value of 0 sets the backlight 2203 to maximum brightness, and the maximum of the brightness 2204 value switches the backlight off. 2205 -1 -- never invert brightness 2206 0 -- machine default 2207 1 -- force brightness inversion 2208 2209 ia32_emulation= [X86-64] 2210 Format: <bool> 2211 When true, allows loading 32-bit programs and executing 32-bit 2212 syscalls, essentially overriding IA32_EMULATION_DEFAULT_DISABLED at 2213 boot time. When false, unconditionally disables IA32 emulation. 2214 2215 2216 idle= [X86,EARLY] 2217 Format: idle=poll, idle=halt, idle=nomwait 2218 2219 idle=poll: Don't do power saving in the idle loop 2220 using HLT, but poll for rescheduling event. This will 2221 make the CPUs eat a lot more power, but may be useful 2222 to get slightly better performance in multiprocessor 2223 benchmarks. It also makes some profiling using 2224 performance counters more accurate. Please note that 2225 on systems with MONITOR/MWAIT support (like Intel 2226 EM64T CPUs) this option has no performance advantage 2227 over the normal idle loop. It may also interact badly 2228 with hyperthreading. 2229 2230 idle=halt: Halt is forced to be used for CPU idle. 2231 In such case C2/C3 won't be used again. 2232 2233 idle=nomwait: Disable mwait for CPU C-states 2234 2235 idxd.sva= [HW] 2236 Format: <bool> 2237 Allow force disabling of Shared Virtual Memory (SVA) 2238 support for the idxd driver. By default it is set to 2239 true (1). 2240 2241 idxd.tc_override= [HW] 2242 Format: <bool> 2243 Allow override of default traffic class configuration 2244 for the device. By default it is set to false (0). 2245 2246 ieee754= [MIPS] Select IEEE Std 754 conformance mode 2247 Format: { strict | legacy | 2008 | relaxed | emulated } 2248 Default: strict 2249 2250 Choose which programs will be accepted for execution 2251 based on the IEEE 754 NaN encoding(s) supported by 2252 the FPU and the NaN encoding requested with the value 2253 of an ELF file header flag individually set by each 2254 binary. Hardware implementations are permitted to 2255 support either or both of the legacy and the 2008 NaN 2256 encoding mode. 2257 2258 Available settings are as follows: 2259 strict accept binaries that request a NaN encoding 2260 supported by the FPU 2261 legacy only accept legacy-NaN binaries, if supported 2262 by the FPU 2263 2008 only accept 2008-NaN binaries, if supported 2264 by the FPU 2265 relaxed accept any binaries regardless of whether 2266 supported by the FPU 2267 emulated accept any binaries but enable FPU emulator 2268 if binary mode is unsupported by the FPU. 2269 2270 The FPU emulator is always able to support both NaN 2271 encodings, so if no FPU hardware is present or it has 2272 been disabled with 'nofpu', then the settings of 2273 'legacy' and '2008' strap the emulator accordingly, 2274 'relaxed' straps the emulator for both legacy-NaN and 2275 2008-NaN, whereas 'strict' enables legacy-NaN only on 2276 legacy processors and both NaN encodings on MIPS32 or 2277 MIPS64 CPUs. 2278 2279 The setting for ABS.fmt/NEG.fmt instruction execution 2280 mode generally follows that for the NaN encoding, 2281 except where unsupported by hardware. 2282 2283 ignore_loglevel [KNL,EARLY] 2284 Ignore loglevel setting - this will print /all/ 2285 kernel messages to the console. Useful for debugging. 2286 We also add it as printk module parameter, so users 2287 could change it dynamically, usually by 2288 /sys/module/printk/parameters/ignore_loglevel. 2289 2290 ignore_rlimit_data 2291 Ignore RLIMIT_DATA setting for data mappings, 2292 print warning at first misuse. Can be changed via 2293 /sys/module/kernel/parameters/ignore_rlimit_data. 2294 2295 ihash_entries= [KNL] 2296 Set number of hash buckets for inode cache. 2297 2298 ima_appraise= [IMA] appraise integrity measurements 2299 Format: { "off" | "enforce" | "fix" | "log" } 2300 default: "enforce" 2301 2302 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead. 2303 The builtin appraise policy appraises all files 2304 owned by uid=0. 2305 2306 ima_canonical_fmt [IMA] 2307 Use the canonical format for the binary runtime 2308 measurements, instead of host native format. 2309 2310 ima_hash= [IMA] 2311 Format: { md5 | sha1 | rmd160 | sha256 | sha384 2312 | sha512 | ... } 2313 default: "sha1" 2314 2315 The list of supported hash algorithms is defined 2316 in crypto/hash_info.h. 2317 2318 ima_policy= [IMA] 2319 The builtin policies to load during IMA setup. 2320 Format: "tcb | appraise_tcb | secure_boot | 2321 fail_securely | critical_data" 2322 2323 The "tcb" policy measures all programs exec'd, files 2324 mmap'd for exec, and all files opened with the read 2325 mode bit set by either the effective uid (euid=0) or 2326 uid=0. 2327 2328 The "appraise_tcb" policy appraises the integrity of 2329 all files owned by root. 2330 2331 The "secure_boot" policy appraises the integrity 2332 of files (eg. kexec kernel image, kernel modules, 2333 firmware, policy, etc) based on file signatures. 2334 2335 The "fail_securely" policy forces file signature 2336 verification failure also on privileged mounted 2337 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE 2338 flag. 2339 2340 The "critical_data" policy measures kernel integrity 2341 critical data. 2342 2343 ima_tcb [IMA] Deprecated. Use ima_policy= instead. 2344 Load a policy which meets the needs of the Trusted 2345 Computing Base. This means IMA will measure all 2346 programs exec'd, files mmap'd for exec, and all files 2347 opened for read by uid=0. 2348 2349 ima_template= [IMA] 2350 Select one of defined IMA measurements template formats. 2351 Formats: { "ima" | "ima-ng" | "ima-ngv2" | "ima-sig" | 2352 "ima-sigv2" } 2353 Default: "ima-ng" 2354 2355 ima_template_fmt= 2356 [IMA] Define a custom template format. 2357 Format: { "field1|...|fieldN" } 2358 2359 ima= [IMA] Enable or disable IMA 2360 Format: { "off" | "on" } 2361 Default: "on" 2362 Note that disabling IMA is limited to kdump kernel. 2363 2364 indirect_target_selection= [X86,Intel] Mitigation control for Indirect 2365 Target Selection(ITS) bug in Intel CPUs. Updated 2366 microcode is also required for a fix in IBPB. 2367 2368 on: Enable mitigation (default). 2369 off: Disable mitigation. 2370 force: Force the ITS bug and deploy default 2371 mitigation. 2372 vmexit: Only deploy mitigation if CPU is affected by 2373 guest/host isolation part of ITS. 2374 stuff: Deploy RSB-fill mitigation when retpoline is 2375 also deployed. Otherwise, deploy the default 2376 mitigation. 2377 2378 For details see: 2379 Documentation/admin-guide/hw-vuln/indirect-target-selection.rst 2380 2381 init= [KNL] 2382 Format: <full_path> 2383 Run specified binary instead of /sbin/init as init 2384 process. 2385 2386 initcall_debug [KNL] Trace initcalls as they are executed. Useful 2387 for working out where the kernel is dying during 2388 startup. 2389 2390 initcall_blacklist= [KNL] Do not execute a comma-separated list of 2391 initcall functions. Useful for debugging built-in 2392 modules and initcalls. 2393 2394 initramfs_async= [KNL] 2395 Format: <bool> 2396 Default: 1 2397 This parameter controls whether the initramfs 2398 image is unpacked asynchronously, concurrently 2399 with devices being probed and 2400 initialized. This should normally just work, 2401 but as a debugging aid, one can get the 2402 historical behaviour of the initramfs 2403 unpacking being completed before device_ and 2404 late_ initcalls. 2405 2406 initrd= [BOOT,EARLY] Specify the location of the initial ramdisk 2407 2408 initrdmem= [KNL,EARLY] Specify a physical address and size from which to 2409 load the initrd. If an initrd is compiled in or 2410 specified in the bootparams, it takes priority over this 2411 setting. 2412 Format: ss[KMG],nn[KMG] 2413 Default is 0, 0 2414 2415 init_on_alloc= [MM,EARLY] Fill newly allocated pages and heap objects with 2416 zeroes. 2417 Format: 0 | 1 2418 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON. 2419 2420 init_on_free= [MM,EARLY] Fill freed pages and heap objects with zeroes. 2421 Format: 0 | 1 2422 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON. 2423 2424 init_pkru= [X86] Specify the default memory protection keys rights 2425 register contents for all processes. 0x55555554 by 2426 default (disallow access to all but pkey 0). Can 2427 override in debugfs after boot. 2428 2429 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver 2430 Format: <irq> 2431 2432 int_pln_enable [X86] Enable power limit notification interrupt 2433 2434 integrity_audit=[IMA] 2435 Format: { "0" | "1" } 2436 0 -- basic integrity auditing messages. (Default) 2437 1 -- additional integrity auditing messages. 2438 2439 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option 2440 on 2441 Enable intel iommu driver. 2442 off 2443 Disable intel iommu driver. 2444 igfx_off [Default Off] 2445 By default, gfx is mapped as normal device. If a gfx 2446 device has a dedicated DMAR unit, the DMAR unit is 2447 bypassed by not enabling DMAR with this option. In 2448 this case, gfx device will use physical address for 2449 DMA. 2450 strict [Default Off] 2451 Deprecated, equivalent to iommu.strict=1. 2452 sp_off [Default Off] 2453 By default, super page will be supported if Intel IOMMU 2454 has the capability. With this option, super page will 2455 not be supported. 2456 sm_on 2457 Enable the Intel IOMMU scalable mode if the hardware 2458 advertises that it has support for the scalable mode 2459 translation. 2460 sm_off 2461 Disallow use of the Intel IOMMU scalable mode. 2462 tboot_noforce [Default Off] 2463 Do not force the Intel IOMMU enabled under tboot. 2464 By default, tboot will force Intel IOMMU on, which 2465 could harm performance of some high-throughput 2466 devices like 40GBit network cards, even if identity 2467 mapping is enabled. 2468 Note that using this option lowers the security 2469 provided by tboot because it makes the system 2470 vulnerable to DMA attacks. 2471 2472 intel_idle.max_cstate= [KNL,HW,ACPI,X86] 2473 0 disables intel_idle and fall back on acpi_idle. 2474 1 to 9 specify maximum depth of C-state. 2475 2476 intel_pstate= [X86,EARLY] 2477 disable 2478 Do not enable intel_pstate as the default 2479 scaling driver for the supported processors 2480 active 2481 Use intel_pstate driver to bypass the scaling 2482 governors layer of cpufreq and provides it own 2483 algorithms for p-state selection. There are two 2484 P-state selection algorithms provided by 2485 intel_pstate in the active mode: powersave and 2486 performance. The way they both operate depends 2487 on whether or not the hardware managed P-states 2488 (HWP) feature has been enabled in the processor 2489 and possibly on the processor model. 2490 passive 2491 Use intel_pstate as a scaling driver, but configure it 2492 to work with generic cpufreq governors (instead of 2493 enabling its internal governor). This mode cannot be 2494 used along with the hardware-managed P-states (HWP) 2495 feature. 2496 force 2497 Enable intel_pstate on systems that prohibit it by default 2498 in favor of acpi-cpufreq. Forcing the intel_pstate driver 2499 instead of acpi-cpufreq may disable platform features, such 2500 as thermal controls and power capping, that rely on ACPI 2501 P-States information being indicated to OSPM and therefore 2502 should be used with caution. This option does not work with 2503 processors that aren't supported by the intel_pstate driver 2504 or on platforms that use pcc-cpufreq instead of acpi-cpufreq. 2505 no_hwp 2506 Do not enable hardware P state control (HWP) 2507 if available. 2508 hwp_only 2509 Only load intel_pstate on systems which support 2510 hardware P state control (HWP) if available. 2511 support_acpi_ppc 2512 Enforce ACPI _PPC performance limits. If the Fixed ACPI 2513 Description Table, specifies preferred power management 2514 profile as "Enterprise Server" or "Performance Server", 2515 then this feature is turned on by default. 2516 per_cpu_perf_limits 2517 Allow per-logical-CPU P-State performance control limits using 2518 cpufreq sysfs interface 2519 no_cas 2520 Do not enable capacity-aware scheduling (CAS) on 2521 hybrid systems 2522 2523 intremap= [X86-64,Intel-IOMMU,EARLY] 2524 on enable Interrupt Remapping (default) 2525 off disable Interrupt Remapping 2526 nosid disable Source ID checking 2527 no_x2apic_optout 2528 BIOS x2APIC opt-out request will be ignored 2529 nopost disable Interrupt Posting 2530 posted_msi 2531 enable MSIs delivered as posted interrupts 2532 2533 iomem= Disable strict checking of access to MMIO memory 2534 strict regions from userspace. 2535 relaxed 2536 2537 iommu= [X86,EARLY] 2538 2539 off 2540 Don't initialize and use any kind of IOMMU. 2541 2542 force 2543 Force the use of the hardware IOMMU even when 2544 it is not actually needed (e.g. because < 3 GB 2545 memory). 2546 2547 noforce 2548 Don't force hardware IOMMU usage when it is not 2549 needed. (default). 2550 2551 merge 2552 Do scatter-gather (SG) merging. Implies "force" 2553 (experimental). 2554 2555 nomerge 2556 Don't do scatter-gather (SG) merging. 2557 2558 biomerge 2559 Do scatter-gather (SG) merging. Implies "force" 2560 (experimental). [same as "merge"] 2561 2562 panic 2563 Always panic when IOMMU overflows. 2564 2565 nopanic 2566 Don't panic on IOMMU overflows. 2567 2568 pt 2569 Use passththrough mode by default 2570 (Equivalent to iommu.passthrough=1) 2571 2572 nopt 2573 Use translated mode for DMA by default 2574 (Equivalent to iommu.passthrough=0) 2575 2576 soft 2577 Use software bounce buffering (SWIOTLB) (default for 2578 Intel machines). This can be used to prevent the usage 2579 of an available hardware IOMMU. 2580 2581 usedac 2582 Use the DAC on VIA PCI bridge 2583 (default: disable the VIA PCI bridge DAC) 2584 2585 AMD Gart HW IOMMU-specific options: (CONFIG_GART_IOMMU) 2586 2587 <size> 2588 Set the size of the remapping area in bytes. 2589 2590 allowed 2591 Overwrite iommu off workarounds for specific chipsets 2592 2593 force 2594 Overwrite iommu off workarounds for specific chipsets 2595 2596 fullflush 2597 Flush IOMMU on each allocation (default). 2598 2599 nofullflush 2600 Don't use IOMMU fullflush. 2601 2602 memaper[=<order>] 2603 Allocate an own aperture over RAM with size 2604 32MB<<order. (default: order=1, i.e. 64MB) 2605 2606 noaperture 2607 Ask the IOMMU not to touch the aperture for AGP. 2608 2609 noagp 2610 Don't initialize the AGP driver and use full aperture. 2611 2612 iommu= [PPC/POWERNV] 2613 2614 nobypass 2615 Disable IOMMU bypass, using IOMMU for PCI devices. 2616 2617 iommu.forcedac= [ARM64,X86,EARLY] Control IOVA allocation for PCI devices. 2618 Format: { "0" | "1" } 2619 0 - Try to allocate a 32-bit DMA address first, before 2620 falling back to the full range if needed. 2621 1 - Allocate directly from the full usable range, 2622 forcing Dual Address Cycle for PCI cards supporting 2623 greater than 32-bit addressing. 2624 2625 iommu.strict= [ARM64,X86,S390,EARLY] Configure TLB invalidation behaviour 2626 Format: { "0" | "1" } 2627 0 - Lazy mode. 2628 Request that DMA unmap operations use deferred 2629 invalidation of hardware TLBs, for increased 2630 throughput at the cost of reduced device isolation. 2631 Will fall back to strict mode if not supported by 2632 the relevant IOMMU driver. 2633 1 - Strict mode. 2634 DMA unmap operations invalidate IOMMU hardware TLBs 2635 synchronously. 2636 unset - Use value of CONFIG_IOMMU_DEFAULT_DMA_{LAZY,STRICT}. 2637 Note: on x86, strict mode specified via one of the 2638 legacy driver-specific options takes precedence. 2639 2640 iommu.passthrough= 2641 [ARM64,X86,EARLY] Configure DMA to bypass the IOMMU by default. 2642 Format: { "0" | "1" } 2643 0 - Use IOMMU translation for DMA. 2644 1 - Bypass the IOMMU for DMA. 2645 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH. 2646 2647 iommu.debug_pagealloc= 2648 [KNL,EARLY] When CONFIG_IOMMU_DEBUG_PAGEALLOC is set, this 2649 parameter enables the feature at boot time. By default, it 2650 is disabled and the system behaves the same way as a kernel 2651 built without CONFIG_IOMMU_DEBUG_PAGEALLOC. 2652 Format: { "0" | "1" } 2653 0 - Sanitizer disabled. 2654 1 - Sanitizer enabled, expect runtime overhead. 2655 2656 io7= [HW] IO7 for Marvel-based Alpha systems 2657 See comment before marvel_specify_io7 in 2658 arch/alpha/kernel/core_marvel.c. 2659 2660 io_delay= [X86,EARLY] I/O delay method 2661 0x80 2662 Standard port 0x80 based delay 2663 0xed 2664 Alternate port 0xed based delay (needed on some systems) 2665 udelay 2666 Simple two microseconds delay 2667 none 2668 No delay 2669 2670 ip= [IP_PNP] 2671 See Documentation/admin-guide/nfs/nfsroot.rst. 2672 2673 ipcmni_extend [KNL,EARLY] Extend the maximum number of unique System V 2674 IPC identifiers from 32,768 to 16,777,216. 2675 2676 ipe.enforce= [IPE] 2677 Format: <bool> 2678 Determine whether IPE starts in permissive (0) or 2679 enforce (1) mode. The default is enforce. 2680 2681 ipe.success_audit= 2682 [IPE] 2683 Format: <bool> 2684 Start IPE with success auditing enabled, emitting 2685 an audit event when a binary is allowed. The default 2686 is 0. 2687 2688 irqaffinity= [SMP] Set the default irq affinity mask 2689 The argument is a cpu list, as described above. 2690 2691 irqchip.gicv2_force_probe= 2692 [ARM,ARM64,EARLY] 2693 Format: <bool> 2694 Force the kernel to look for the second 4kB page 2695 of a GICv2 controller even if the memory range 2696 exposed by the device tree is too small. 2697 2698 irqchip.gicv3_nolpi= 2699 [ARM,ARM64,EARLY] 2700 Force the kernel to ignore the availability of 2701 LPIs (and by consequence ITSs). Intended for system 2702 that use the kernel as a bootloader, and thus want 2703 to let secondary kernels in charge of setting up 2704 LPIs. 2705 2706 irqchip.gicv3_pseudo_nmi= [ARM64,EARLY] 2707 Enables support for pseudo-NMIs in the kernel. This 2708 requires the kernel to be built with 2709 CONFIG_ARM64_PSEUDO_NMI. 2710 2711 irqchip.riscv_imsic_noipi 2712 [RISC-V,EARLY] 2713 Force the kernel to not use IMSIC software injected MSIs 2714 as IPIs. Intended for system where IMSIC is trap-n-emulated, 2715 and thus want to reduce MMIO traps when triggering IPIs 2716 to multiple harts. 2717 2718 irqfixup [HW] 2719 When an interrupt is not handled search all handlers 2720 for it. Intended to get systems with badly broken 2721 firmware running. 2722 2723 irqhandler.duration_warn_us= [KNL] 2724 Warn if an IRQ handler exceeds the specified duration 2725 threshold in microseconds. Useful for identifying 2726 long-running IRQs in the system. 2727 2728 irqpoll [HW] 2729 When an interrupt is not handled search all handlers 2730 for it. Also check all handlers each timer 2731 interrupt. Intended to get systems with badly broken 2732 firmware running. 2733 2734 isapnp= [ISAPNP] 2735 Format: <RDP>,<reset>,<pci_scan>,<verbosity> 2736 2737 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance. 2738 Format: [flag-list,]<cpu-list> 2739 2740 Specify one or more CPUs to isolate from disturbances 2741 specified in the flag list (default: domain): 2742 2743 nohz 2744 Disable the tick when a single task runs as well as 2745 disabling other kernel noises like having RCU callbacks 2746 offloaded. This is equivalent to the nohz_full parameter. 2747 2748 A residual 1Hz tick is offloaded to workqueues, which you 2749 need to affine to housekeeping through the global 2750 workqueue's affinity configured via the 2751 /sys/devices/virtual/workqueue/cpumask sysfs file, or 2752 by using the 'domain' flag described below. 2753 2754 NOTE: by default the global workqueue runs on all CPUs, 2755 so to protect individual CPUs the 'cpumask' file has to 2756 be configured manually after bootup. 2757 2758 domain 2759 Isolate from the general SMP balancing and scheduling 2760 algorithms. Note that performing domain isolation this way 2761 is irreversible: it's not possible to bring back a CPU to 2762 the domains once isolated through this boot time 2763 configuration. Use cpusets for a dynamic configuration 2764 which can be altered at runtime. For details see 2765 Documentation/admin-guide/cpu-isolation.rst. 2766 2767 You can move a process onto or off an "isolated" CPU via 2768 the CPU affinity syscalls or cpuset. 2769 <cpu number> begins at 0 and the maximum value is 2770 "number of CPUs in system - 1". 2771 2772 managed_irq 2773 2774 Isolate from being targeted by managed interrupts 2775 which have an interrupt mask containing isolated 2776 CPUs. The affinity of managed interrupts is 2777 handled by the kernel and cannot be changed via 2778 the /proc/irq/* interfaces. 2779 2780 This isolation is best effort and only effective 2781 if the automatically assigned interrupt mask of a 2782 device queue contains isolated and housekeeping 2783 CPUs. If housekeeping CPUs are online then such 2784 interrupts are directed to the housekeeping CPU 2785 so that IO submitted on the housekeeping CPU 2786 cannot disturb the isolated CPU. 2787 2788 If a queue's affinity mask contains only isolated 2789 CPUs then this parameter has no effect on the 2790 interrupt routing decision, though interrupts are 2791 only delivered when tasks running on those 2792 isolated CPUs submit IO. IO submitted on 2793 housekeeping CPUs has no influence on those 2794 queues. 2795 2796 The format of <cpu-list> is described above. 2797 2798 iucv= [HW,NET] 2799 2800 ivrs_ioapic [HW,X86-64] 2801 Provide an override to the IOAPIC-ID<->DEVICE-ID 2802 mapping provided in the IVRS ACPI table. 2803 By default, PCI segment is 0, and can be omitted. 2804 2805 For example, to map IOAPIC-ID decimal 10 to 2806 PCI segment 0x1 and PCI device 00:14.0, 2807 write the parameter as: 2808 ivrs_ioapic=10@0001:00:14.0 2809 2810 Deprecated formats: 2811 * To map IOAPIC-ID decimal 10 to PCI device 00:14.0 2812 write the parameter as: 2813 ivrs_ioapic[10]=00:14.0 2814 * To map IOAPIC-ID decimal 10 to PCI segment 0x1 and 2815 PCI device 00:14.0 write the parameter as: 2816 ivrs_ioapic[10]=0001:00:14.0 2817 2818 ivrs_hpet [HW,X86-64] 2819 Provide an override to the HPET-ID<->DEVICE-ID 2820 mapping provided in the IVRS ACPI table. 2821 By default, PCI segment is 0, and can be omitted. 2822 2823 For example, to map HPET-ID decimal 10 to 2824 PCI segment 0x1 and PCI device 00:14.0, 2825 write the parameter as: 2826 ivrs_hpet=10@0001:00:14.0 2827 2828 Deprecated formats: 2829 * To map HPET-ID decimal 0 to PCI device 00:14.0 2830 write the parameter as: 2831 ivrs_hpet[0]=00:14.0 2832 * To map HPET-ID decimal 10 to PCI segment 0x1 and 2833 PCI device 00:14.0 write the parameter as: 2834 ivrs_ioapic[10]=0001:00:14.0 2835 2836 ivrs_acpihid [HW,X86-64] 2837 Provide an override to the ACPI-HID:UID<->DEVICE-ID 2838 mapping provided in the IVRS ACPI table. 2839 By default, PCI segment is 0, and can be omitted. 2840 2841 For example, to map UART-HID:UID AMD0020:0 to 2842 PCI segment 0x1 and PCI device ID 00:14.5, 2843 write the parameter as: 2844 ivrs_acpihid=AMD0020:0@0001:00:14.5 2845 2846 Deprecated formats: 2847 * To map UART-HID:UID AMD0020:0 to PCI segment is 0, 2848 PCI device ID 00:14.5, write the parameter as: 2849 ivrs_acpihid[00:14.5]=AMD0020:0 2850 * To map UART-HID:UID AMD0020:0 to PCI segment 0x1 and 2851 PCI device ID 00:14.5, write the parameter as: 2852 ivrs_acpihid[0001:00:14.5]=AMD0020:0 2853 2854 js= [HW,JOY] Analog joystick 2855 See Documentation/input/joydev/joystick.rst. 2856 2857 kasan_multi_shot 2858 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print 2859 report on every invalid memory access. Without this 2860 parameter KASAN will print report only for the first 2861 invalid access. 2862 2863 keep_bootcon [KNL,EARLY] 2864 Do not unregister boot console at start. This is only 2865 useful for debugging when something happens in the window 2866 between unregistering the boot console and initializing 2867 the real console. 2868 2869 keepinitrd [HW,ARM] See retain_initrd. 2870 2871 kernelcore= [KNL,X86,PPC,EARLY] 2872 Format: nn[KMGTPE] | nn% | "mirror" 2873 This parameter specifies the amount of memory usable by 2874 the kernel for non-movable allocations. The requested 2875 amount is spread evenly throughout all nodes in the 2876 system as ZONE_NORMAL. The remaining memory is used for 2877 movable memory in its own zone, ZONE_MOVABLE. In the 2878 event, a node is too small to have both ZONE_NORMAL and 2879 ZONE_MOVABLE, kernelcore memory will take priority and 2880 other nodes will have a larger ZONE_MOVABLE. 2881 2882 ZONE_MOVABLE is used for the allocation of pages that 2883 may be reclaimed or moved by the page migration 2884 subsystem. Note that allocations like PTEs-from-HighMem 2885 still use the HighMem zone if it exists, and the Normal 2886 zone if it does not. 2887 2888 It is possible to specify the exact amount of memory in 2889 the form of "nn[KMGTPE]", a percentage of total system 2890 memory in the form of "nn%", or "mirror". If "mirror" 2891 option is specified, mirrored (reliable) memory is used 2892 for non-movable allocations and remaining memory is used 2893 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror" 2894 are exclusive, so you cannot specify multiple forms. 2895 2896 kfence.burst= [MM,KFENCE] The number of additional successive 2897 allocations to be attempted through KFENCE for each 2898 sample interval. 2899 Format: <unsigned integer> 2900 Default: 0 2901 2902 kfence.check_on_panic= 2903 [MM,KFENCE] Whether to check all KFENCE-managed objects' 2904 canaries on panic. 2905 Format: <bool> 2906 Default: false 2907 2908 kfence.deferrable= 2909 [MM,KFENCE] Whether to use a deferrable timer to trigger 2910 allocations. This avoids forcing CPU wake-ups if the 2911 system is idle, at the risk of a less predictable 2912 sample interval. 2913 Format: <bool> 2914 Default: CONFIG_KFENCE_DEFERRABLE 2915 2916 kfence.fault= [MM,KFENCE] Controls the behavior when a KFENCE 2917 error is detected. 2918 report - print the error report and continue (default). 2919 oops - print the error report and oops. 2920 panic - print the error report and panic. 2921 2922 kfence.sample_interval= 2923 [MM,KFENCE] KFENCE's sample interval in milliseconds. 2924 Format: <unsigned integer> 2925 0 - Disable KFENCE. 2926 >0 - Enabled KFENCE with given sample interval. 2927 Default: CONFIG_KFENCE_SAMPLE_INTERVAL 2928 2929 kfence.skip_covered_thresh= 2930 [MM,KFENCE] If pool utilization reaches this threshold 2931 (pool usage%), KFENCE limits currently covered 2932 allocations of the same source from further filling 2933 up the pool. 2934 Format: <unsigned integer> 2935 Default: 75 2936 2937 kgdbdbgp= [KGDB,HW,EARLY] kgdb over EHCI usb debug port. 2938 Format: <Controller#>[,poll interval] 2939 The controller # is the number of the ehci usb debug 2940 port as it is probed via PCI. The poll interval is 2941 optional and is the number seconds in between 2942 each poll cycle to the debug port in case you need 2943 the functionality for interrupting the kernel with 2944 gdb or control-c on the dbgp connection. When 2945 not using this parameter you use sysrq-g to break into 2946 the kernel debugger. 2947 2948 kgdboc= [KGDB,HW] kgdb over consoles. 2949 Requires a tty driver that supports console polling, 2950 or a supported polling keyboard driver (non-usb). 2951 Serial only format: <serial_device>[,baud] 2952 keyboard only format: kbd 2953 keyboard and serial format: kbd,<serial_device>[,baud] 2954 Optional Kernel mode setting: 2955 kms, kbd format: kms,kbd 2956 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud] 2957 2958 kgdboc_earlycon= [KGDB,HW,EARLY] 2959 If the boot console provides the ability to read 2960 characters and can work in polling mode, you can use 2961 this parameter to tell kgdb to use it as a backend 2962 until the normal console is registered. Intended to 2963 be used together with the kgdboc parameter which 2964 specifies the normal console to transition to. 2965 2966 The name of the early console should be specified 2967 as the value of this parameter. Note that the name of 2968 the early console might be different than the tty 2969 name passed to kgdboc. It's OK to leave the value 2970 blank and the first boot console that implements 2971 read() will be picked. 2972 2973 kgdbwait [KGDB,EARLY] Stop kernel execution and enter the 2974 kernel debugger at the earliest opportunity. 2975 2976 kho= [KEXEC,EARLY] 2977 Format: { "0" | "1" | "off" | "on" | "y" | "n" } 2978 Enables or disables Kexec HandOver. 2979 "0" | "off" | "n" - kexec handover is disabled 2980 "1" | "on" | "y" - kexec handover is enabled 2981 2982 kho_scratch= [KEXEC,EARLY] 2983 Format: ll[KMG],mm[KMG],nn[KMG] | nn% 2984 Defines the size of the KHO scratch region. The KHO 2985 scratch regions are physically contiguous memory 2986 ranges that can only be used for non-kernel 2987 allocations. That way, even when memory is heavily 2988 fragmented with handed over memory, the kexeced 2989 kernel will always have enough contiguous ranges to 2990 bootstrap itself. 2991 2992 It is possible to specify the exact amount of 2993 memory in the form of "ll[KMG],mm[KMG],nn[KMG]" 2994 where the first parameter defines the size of a low 2995 memory scratch area, the second parameter defines 2996 the size of a global scratch area and the third 2997 parameter defines the size of additional per-node 2998 scratch areas. The form "nn%" defines scale factor 2999 (in percents) of memory that was used during boot. 3000 3001 kmac= [MIPS] Korina ethernet MAC address. 3002 Configure the RouterBoard 532 series on-chip 3003 Ethernet adapter MAC address. 3004 3005 kmemleak= [KNL,EARLY] Boot-time kmemleak enable/disable 3006 Valid arguments: on, off 3007 Default: on 3008 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y, 3009 the default is off. 3010 3011 kprobe_event=[probe-list] 3012 [FTRACE] Add kprobe events and enable at boot time. 3013 The probe-list is a semicolon delimited list of probe 3014 definitions. Each definition is same as kprobe_events 3015 interface, but the parameters are comma delimited. 3016 For example, to add a kprobe event on vfs_read with 3017 arg1 and arg2, add to the command line; 3018 3019 kprobe_event=p,vfs_read,$arg1,$arg2 3020 3021 See also Documentation/trace/kprobetrace.rst "Kernel 3022 Boot Parameter" section. 3023 3024 kpti= [ARM64,EARLY] Control page table isolation of 3025 user and kernel address spaces. 3026 Default: enabled on cores which need mitigation. 3027 0: force disabled 3028 1: force enabled 3029 3030 kunit.enable= [KUNIT] Enable executing KUnit tests. Requires 3031 CONFIG_KUNIT to be set to be fully enabled. The 3032 default value can be overridden via 3033 KUNIT_DEFAULT_ENABLED. 3034 Default is 1 (enabled) 3035 3036 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs. 3037 Default is 0 (don't ignore, but inject #GP) 3038 3039 kvm.eager_page_split= 3040 [KVM,X86] Controls whether or not KVM will try to 3041 proactively split all huge pages during dirty logging. 3042 Eager page splitting reduces interruptions to vCPU 3043 execution by eliminating the write-protection faults 3044 and MMU lock contention that would otherwise be 3045 required to split huge pages lazily. 3046 3047 VM workloads that rarely perform writes or that write 3048 only to a small region of VM memory may benefit from 3049 disabling eager page splitting to allow huge pages to 3050 still be used for reads. 3051 3052 The behavior of eager page splitting depends on whether 3053 KVM_DIRTY_LOG_INITIALLY_SET is enabled or disabled. If 3054 disabled, all huge pages in a memslot will be eagerly 3055 split when dirty logging is enabled on that memslot. If 3056 enabled, eager page splitting will be performed during 3057 the KVM_CLEAR_DIRTY ioctl, and only for the pages being 3058 cleared. 3059 3060 Eager page splitting is only supported when kvm.tdp_mmu=Y. 3061 3062 Default is Y (on). 3063 3064 kvm.enable_pmu=[KVM,X86] 3065 If enabled, KVM will virtualize PMU functionality based 3066 on the virtual CPU model defined by userspace. This 3067 can be overridden on a per-VM basis via 3068 KVM_CAP_PMU_CAPABILITY. 3069 3070 If disabled, KVM will not virtualize PMU functionality, 3071 e.g. MSRs, PMCs, PMIs, etc., even if userspace defines 3072 a virtual CPU model that contains PMU assets. 3073 3074 Note, KVM's vPMU support implicitly requires running 3075 with an in-kernel local APIC, e.g. to deliver PMIs to 3076 the guest. Running without an in-kernel local APIC is 3077 not supported, though KVM will allow such a combination 3078 (with severely degraded functionality). 3079 3080 See also enable_mediated_pmu. 3081 3082 Default is Y (on). 3083 3084 kvm.enable_virt_at_load=[KVM,ARM64,LOONGARCH,MIPS,RISCV,X86] 3085 If enabled, KVM will enable virtualization in hardware 3086 when KVM is loaded, and disable virtualization when KVM 3087 is unloaded (if KVM is built as a module). 3088 3089 If disabled, KVM will dynamically enable and disable 3090 virtualization on-demand when creating and destroying 3091 VMs, i.e. on the 0=>1 and 1=>0 transitions of the 3092 number of VMs. 3093 3094 Enabling virtualization at module load avoids potential 3095 latency for creation of the 0=>1 VM, as KVM serializes 3096 virtualization enabling across all online CPUs. The 3097 "cost" of enabling virtualization when KVM is loaded, 3098 is that doing so may interfere with using out-of-tree 3099 hypervisors that want to "own" virtualization hardware. 3100 3101 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface. 3102 Default is false (don't support). 3103 3104 kvm.nx_huge_pages= 3105 [KVM] Controls the software workaround for the 3106 X86_BUG_ITLB_MULTIHIT bug. 3107 force : Always deploy workaround. 3108 off : Never deploy workaround. 3109 auto : Deploy workaround based on the presence of 3110 X86_BUG_ITLB_MULTIHIT. 3111 3112 Default is 'auto'. 3113 3114 If the software workaround is enabled for the host, 3115 guests do need not to enable it for nested guests. 3116 3117 kvm.nx_huge_pages_recovery_ratio= 3118 [KVM] Controls how many 4KiB pages are periodically zapped 3119 back to huge pages. 0 disables the recovery, otherwise if 3120 the value is N KVM will zap 1/Nth of the 4KiB pages every 3121 period (see below). The default is 60. 3122 3123 kvm.nx_huge_pages_recovery_period_ms= 3124 [KVM] Controls the time period at which KVM zaps 4KiB pages 3125 back to huge pages. If the value is a non-zero N, KVM will 3126 zap a portion (see ratio above) of the pages every N msecs. 3127 If the value is 0 (the default), KVM will pick a period based 3128 on the ratio, such that a page is zapped after 1 hour on average. 3129 3130 kvm-{amd,intel}.enable_mediated_pmu=[KVM,AMD,INTEL] 3131 If enabled, KVM will provide a mediated virtual PMU, 3132 instead of the default perf-based virtual PMU (if 3133 kvm.enable_pmu is true and PMU is enumerated via the 3134 virtual CPU model). 3135 3136 With a perf-based vPMU, KVM operates as a user of perf, 3137 i.e. emulates guest PMU counters using perf events. 3138 KVM-created perf events are managed by perf as regular 3139 (guest-only) events, e.g. are scheduled in/out, contend 3140 for hardware resources, etc. Using a perf-based vPMU 3141 allows guest and host usage of the PMU to co-exist, but 3142 incurs non-trivial overhead and can result in silently 3143 dropped guest events (due to resource contention). 3144 3145 With a mediated vPMU, hardware PMU state is context 3146 switched around the world switch to/from the guest. 3147 KVM mediates which events the guest can utilize, but 3148 gives the guest direct access to all other PMU assets 3149 when possible (KVM may intercept some accesses if the 3150 virtual CPU model provides a subset of hardware PMU 3151 functionality). Using a mediated vPMU significantly 3152 reduces PMU virtualization overhead and eliminates lost 3153 guest events, but is mutually exclusive with using perf 3154 to profile KVM guests and adds latency to most VM-Exits 3155 (to context switch PMU state). 3156 3157 Default is N (off). 3158 3159 kvm-amd.nested= [KVM,AMD] Control nested virtualization feature in 3160 KVM/SVM. Default is 1 (enabled). 3161 3162 kvm-amd.npt= [KVM,AMD] Control KVM's use of Nested Page Tables, 3163 a.k.a. Two-Dimensional Page Tables. Default is 1 3164 (enabled). Disable by KVM if hardware lacks support 3165 for NPT. 3166 3167 kvm-amd.ciphertext_hiding_asids= 3168 [KVM,AMD] Ciphertext hiding prevents disallowed accesses 3169 to SNP private memory from reading ciphertext. Instead, 3170 reads will see constant default values (0xff). 3171 3172 If ciphertext hiding is enabled, the joint SEV-ES and 3173 SEV-SNP ASID space is partitioned into separate SEV-ES 3174 and SEV-SNP ASID ranges, with the SEV-SNP range being 3175 [1..max_snp_asid] and the SEV-ES range being 3176 (max_snp_asid..min_sev_asid), where min_sev_asid is 3177 enumerated by CPUID.0x.8000_001F[EDX]. 3178 3179 A non-zero value enables SEV-SNP ciphertext hiding and 3180 adjusts the ASID ranges for SEV-ES and SEV-SNP guests. 3181 KVM caps the number of SEV-SNP ASIDs at the maximum 3182 possible value, e.g. specifying -1u will assign all 3183 joint SEV-ES and SEV-SNP ASIDs to SEV-SNP. Note, 3184 assigning all joint ASIDs to SEV-SNP, i.e. configuring 3185 max_snp_asid == min_sev_asid-1, will effectively make 3186 SEV-ES unusable. 3187 3188 kvm-arm.mode= 3189 [KVM,ARM,EARLY] Select one of KVM/arm64's modes of 3190 operation. 3191 3192 none: Forcefully disable KVM. 3193 3194 nvhe: Standard nVHE-based mode, without support for 3195 protected guests. 3196 3197 protected: Mode with support for guests whose state is 3198 kept private from the host, using VHE or 3199 nVHE depending on HW support. 3200 3201 nested: VHE-based mode with support for nested 3202 virtualization. Requires at least ARMv8.4 3203 hardware (with FEAT_NV2). 3204 3205 Defaults to VHE/nVHE based on hardware support. Setting 3206 mode to "protected" will disable kexec and hibernation 3207 for the host. To force nVHE on VHE hardware, add 3208 "arm64_sw.hvhe=0 id_aa64mmfr1.vh=0" to the 3209 command-line. 3210 "nested" and "protected" are experimental and should be 3211 used with extreme caution. 3212 3213 kvm-arm.vgic_v3_group0_trap= 3214 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-0 3215 system registers 3216 3217 kvm-arm.vgic_v3_group1_trap= 3218 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-1 3219 system registers 3220 3221 kvm-arm.vgic_v3_common_trap= 3222 [KVM,ARM,EARLY] Trap guest accesses to GICv3 common 3223 system registers 3224 3225 kvm-arm.vgic_v4_enable= 3226 [KVM,ARM,EARLY] Allow use of GICv4 for direct 3227 injection of LPIs. 3228 3229 kvm-arm.wfe_trap_policy= 3230 [KVM,ARM] Control when to set WFE instruction trap for 3231 KVM VMs. Traps are allowed but not guaranteed by the 3232 CPU architecture. 3233 3234 trap: set WFE instruction trap 3235 3236 notrap: clear WFE instruction trap 3237 3238 kvm-arm.wfi_trap_policy= 3239 [KVM,ARM] Control when to set WFI instruction trap for 3240 KVM VMs. Traps are allowed but not guaranteed by the 3241 CPU architecture. 3242 3243 trap: set WFI instruction trap 3244 3245 notrap: clear WFI instruction trap 3246 3247 kvm_cma_resv_ratio=n [PPC,EARLY] 3248 Reserves given percentage from system memory area for 3249 contiguous memory allocation for KVM hash pagetable 3250 allocation. 3251 By default it reserves 5% of total system memory. 3252 Format: <integer> 3253 Default: 5 3254 3255 kvm-intel.ept= [KVM,Intel] Control KVM's use of Extended Page Tables, 3256 a.k.a. Two-Dimensional Page Tables. Default is 1 3257 (enabled). Disable by KVM if hardware lacks support 3258 for EPT. 3259 3260 kvm-intel.emulate_invalid_guest_state= 3261 [KVM,Intel] Control whether to emulate invalid guest 3262 state. Ignored if kvm-intel.enable_unrestricted_guest=1, 3263 as guest state is never invalid for unrestricted 3264 guests. This param doesn't apply to nested guests (L2), 3265 as KVM never emulates invalid L2 guest state. 3266 Default is 1 (enabled). 3267 3268 kvm-intel.flexpriority= 3269 [KVM,Intel] Control KVM's use of FlexPriority feature 3270 (TPR shadow). Default is 1 (enabled). Disable by KVM if 3271 hardware lacks support for it. 3272 3273 kvm-intel.nested= 3274 [KVM,Intel] Control nested virtualization feature in 3275 KVM/VMX. Default is 1 (enabled). 3276 3277 kvm-intel.unrestricted_guest= 3278 [KVM,Intel] Control KVM's use of unrestricted guest 3279 feature (virtualized real and unpaged mode). Default 3280 is 1 (enabled). Disable by KVM if EPT is disabled or 3281 hardware lacks support for it. 3282 3283 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault 3284 CVE-2018-3620. 3285 3286 Valid arguments: never, cond, always 3287 3288 always: L1D cache flush on every VMENTER. 3289 cond: Flush L1D on VMENTER only when the code between 3290 VMEXIT and VMENTER can leak host memory. 3291 never: Disables the mitigation 3292 3293 Default is cond (do L1 cache flush in specific instances) 3294 3295 kvm-intel.vpid= [KVM,Intel] Control KVM's use of Virtual Processor 3296 Identification feature (tagged TLBs). Default is 1 3297 (enabled). Disable by KVM if hardware lacks support 3298 for it. 3299 3300 l1d_flush= [X86,INTEL,EARLY] 3301 Control mitigation for L1D based snooping vulnerability. 3302 3303 Certain CPUs are vulnerable to an exploit against CPU 3304 internal buffers which can forward information to a 3305 disclosure gadget under certain conditions. 3306 3307 In vulnerable processors, the speculatively 3308 forwarded data can be used in a cache side channel 3309 attack, to access data to which the attacker does 3310 not have direct access. 3311 3312 This parameter controls the mitigation. The 3313 options are: 3314 3315 on - enable the interface for the mitigation 3316 3317 l1tf= [X86,EARLY] Control mitigation of the L1TF vulnerability on 3318 affected CPUs 3319 3320 The kernel PTE inversion protection is unconditionally 3321 enabled and cannot be disabled. 3322 3323 full 3324 Provides all available mitigations for the 3325 L1TF vulnerability. Disables SMT and 3326 enables all mitigations in the 3327 hypervisors, i.e. unconditional L1D flush. 3328 3329 SMT control and L1D flush control via the 3330 sysfs interface is still possible after 3331 boot. Hypervisors will issue a warning 3332 when the first VM is started in a 3333 potentially insecure configuration, 3334 i.e. SMT enabled or L1D flush disabled. 3335 3336 full,force 3337 Same as 'full', but disables SMT and L1D 3338 flush runtime control. Implies the 3339 'nosmt=force' command line option. 3340 (i.e. sysfs control of SMT is disabled.) 3341 3342 flush 3343 Leaves SMT enabled and enables the default 3344 hypervisor mitigation, i.e. conditional 3345 L1D flush. 3346 3347 SMT control and L1D flush control via the 3348 sysfs interface is still possible after 3349 boot. Hypervisors will issue a warning 3350 when the first VM is started in a 3351 potentially insecure configuration, 3352 i.e. SMT enabled or L1D flush disabled. 3353 3354 flush,nosmt 3355 3356 Disables SMT and enables the default 3357 hypervisor mitigation. 3358 3359 SMT control and L1D flush control via the 3360 sysfs interface is still possible after 3361 boot. Hypervisors will issue a warning 3362 when the first VM is started in a 3363 potentially insecure configuration, 3364 i.e. SMT enabled or L1D flush disabled. 3365 3366 flush,nowarn 3367 Same as 'flush', but hypervisors will not 3368 warn when a VM is started in a potentially 3369 insecure configuration. 3370 3371 off 3372 Disables hypervisor mitigations and doesn't 3373 emit any warnings. 3374 It also drops the swap size and available 3375 RAM limit restriction on both hypervisor and 3376 bare metal. 3377 3378 Default is 'flush'. 3379 3380 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst 3381 3382 l2cr= [PPC] 3383 3384 l3cr= [PPC] 3385 3386 lapic [X86-32,APIC,EARLY] Enable the local APIC even if BIOS 3387 disabled it. 3388 3389 lapic= [X86,APIC] Do not use TSC deadline 3390 value for LAPIC timer one-shot implementation. Default 3391 back to the programmable timer unit in the LAPIC. 3392 Format: notscdeadline 3393 3394 lapic_timer_c2_ok [X86,APIC,EARLY] trust the local apic timer 3395 in C2 power state. 3396 3397 libata.dma= [LIBATA] DMA control 3398 libata.dma=0 Disable all PATA and SATA DMA 3399 libata.dma=1 PATA and SATA Disk DMA only 3400 libata.dma=2 ATAPI (CDROM) DMA only 3401 libata.dma=4 Compact Flash DMA only 3402 Combinations also work, so libata.dma=3 enables DMA 3403 for disks and CDROMs, but not CFs. 3404 3405 libata.ignore_hpa= [LIBATA] Ignore HPA limit 3406 libata.ignore_hpa=0 keep BIOS limits (default) 3407 libata.ignore_hpa=1 ignore limits, using full disk 3408 3409 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume 3410 when set. 3411 Format: <int> 3412 3413 libata.force= [LIBATA] Force configurations. The format is a comma- 3414 separated list of "[ID:]VAL" where ID is PORT[.DEVICE]. 3415 PORT and DEVICE are decimal numbers matching port, link 3416 or device. Basically, it matches the ATA ID string 3417 printed on console by libata. If the whole ID part is 3418 omitted, the last PORT and DEVICE values are used. If 3419 ID hasn't been specified yet, the configuration applies 3420 to all ports, links and devices. 3421 3422 If only DEVICE is omitted, the parameter applies to 3423 the port and all links and devices behind it. DEVICE 3424 number of 0 either selects the first device or the 3425 first fan-out link behind PMP device. It does not 3426 select the host link. DEVICE number of 15 selects the 3427 host link and device attached to it. 3428 3429 The VAL specifies the configuration to force. As long 3430 as there is no ambiguity, shortcut notation is allowed. 3431 For example, both 1.5 and 1.5G would work for 1.5Gbps. 3432 The following configurations can be forced. 3433 3434 * Cable type: 40c, 80c, short40c, unk, ign or sata. 3435 Any ID with matching PORT is used. 3436 3437 * SATA link speed limit: 1.5Gbps or 3.0Gbps. 3438 3439 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7]. 3440 udma[/][16,25,33,44,66,100,133] notation is also 3441 allowed. 3442 3443 * nohrst, nosrst, norst: suppress hard, soft and both 3444 resets. 3445 3446 * rstonce: only attempt one reset during hot-unplug 3447 link recovery. 3448 3449 * [no]dbdelay: Enable or disable the extra 200ms delay 3450 before debouncing a link PHY and device presence 3451 detection. 3452 3453 * [no]ncq: Turn on or off NCQ. 3454 3455 * [no]ncqtrim: Enable or disable queued DSM TRIM. 3456 3457 * [no]ncqati: Enable or disable NCQ trim on ATI chipset. 3458 3459 * [no]trim: Enable or disable (unqueued) TRIM. 3460 3461 * trim_zero: Indicate that TRIM command zeroes data. 3462 3463 * max_trim_128m: Set 128M maximum trim size limit. 3464 3465 * [no]dma: Turn on or off DMA transfers. 3466 3467 * atapi_dmadir: Enable ATAPI DMADIR bridge support. 3468 3469 * atapi_mod16_dma: Enable the use of ATAPI DMA for 3470 commands that are not a multiple of 16 bytes. 3471 3472 * [no]dmalog: Enable or disable the use of the 3473 READ LOG DMA EXT command to access logs. 3474 3475 * [no]iddevlog: Enable or disable access to the 3476 identify device data log. 3477 3478 * [no]logdir: Enable or disable access to the general 3479 purpose log directory. 3480 3481 * max_sec=<sectors>: Set the transfer size limit, in 3482 number of 512-byte sectors, to the value specified in 3483 <sectors>. The value specified in <sectors> has to be 3484 a non-zero positive integer. 3485 3486 * max_sec_128: Set transfer size limit to 128 sectors. 3487 3488 * max_sec_1024: Set or clear transfer size limit to 3489 1024 sectors. 3490 3491 * max_sec_lba48: Set or clear transfer size limit to 3492 65535 sectors. 3493 3494 * external: Mark port as external (hotplug-capable). 3495 3496 * [no]lpm: Enable or disable link power management. 3497 3498 * [no]setxfer: Indicate if transfer speed mode setting 3499 should be skipped. 3500 3501 * [no]fua: Disable or enable FUA (Force Unit Access) 3502 support for devices supporting this feature. 3503 3504 * dump_id: Dump IDENTIFY data. 3505 3506 * disable: Disable this device. 3507 3508 If there are multiple matching configurations changing 3509 the same attribute, the last one is used. 3510 3511 liveupdate= [KNL,EARLY] 3512 Format: <bool> 3513 Enable Live Update Orchestrator (LUO). 3514 Default: off. 3515 3516 lockd.nlm_grace_period=P [NFS] Assign grace period. 3517 Format: <integer> 3518 3519 lockd.nlm_tcpport=N [NFS] Assign TCP port. 3520 Format: <integer> 3521 3522 lockd.nlm_timeout=T [NFS] Assign timeout value. 3523 Format: <integer> 3524 3525 lockd.nlm_udpport=M [NFS] Assign UDP port. 3526 Format: <integer> 3527 3528 lockdown= [SECURITY,EARLY] 3529 { integrity | confidentiality } 3530 Enable the kernel lockdown feature. If set to 3531 integrity, kernel features that allow userland to 3532 modify the running kernel are disabled. If set to 3533 confidentiality, kernel features that allow userland 3534 to extract confidential information from the kernel 3535 are also disabled. 3536 3537 locktorture.acq_writer_lim= [KNL] 3538 Set the time limit in jiffies for a lock 3539 acquisition. Acquisitions exceeding this limit 3540 will result in a splat once they do complete. 3541 3542 locktorture.bind_readers= [KNL] 3543 Specify the list of CPUs to which the readers are 3544 to be bound. 3545 3546 locktorture.bind_writers= [KNL] 3547 Specify the list of CPUs to which the writers are 3548 to be bound. 3549 3550 locktorture.call_rcu_chains= [KNL] 3551 Specify the number of self-propagating call_rcu() 3552 chains to set up. These are used to ensure that 3553 there is a high probability of an RCU grace period 3554 in progress at any given time. Defaults to 0, 3555 which disables these call_rcu() chains. 3556 3557 locktorture.long_hold= [KNL] 3558 Specify the duration in milliseconds for the 3559 occasional long-duration lock hold time. Defaults 3560 to 100 milliseconds. Select 0 to disable. 3561 3562 locktorture.nested_locks= [KNL] 3563 Specify the maximum lock nesting depth that 3564 locktorture is to exercise, up to a limit of 8 3565 (MAX_NESTED_LOCKS). Specify zero to disable. 3566 Note that this parameter is ineffective on types 3567 of locks that do not support nested acquisition. 3568 3569 locktorture.nreaders_stress= [KNL] 3570 Set the number of locking read-acquisition kthreads. 3571 Defaults to being automatically set based on the 3572 number of online CPUs. 3573 3574 locktorture.nwriters_stress= [KNL] 3575 Set the number of locking write-acquisition kthreads. 3576 3577 locktorture.onoff_holdoff= [KNL] 3578 Set time (s) after boot for CPU-hotplug testing. 3579 3580 locktorture.onoff_interval= [KNL] 3581 Set time (s) between CPU-hotplug operations, or 3582 zero to disable CPU-hotplug testing. 3583 3584 locktorture.rt_boost= [KNL] 3585 Do periodic testing of real-time lock priority 3586 boosting. Select 0 to disable, 1 to boost 3587 only rt_mutex, and 2 to boost unconditionally. 3588 Defaults to 2, which might seem to be an 3589 odd choice, but which should be harmless for 3590 non-real-time spinlocks, due to their disabling 3591 of preemption. Note that non-realtime mutexes 3592 disable boosting. 3593 3594 locktorture.rt_boost_factor= [KNL] 3595 Number that determines how often and for how 3596 long priority boosting is exercised. This is 3597 scaled down by the number of writers, so that the 3598 number of boosts per unit time remains roughly 3599 constant as the number of writers increases. 3600 On the other hand, the duration of each boost 3601 increases with the number of writers. 3602 3603 locktorture.shuffle_interval= [KNL] 3604 Set task-shuffle interval (jiffies). Shuffling 3605 tasks allows some CPUs to go into dyntick-idle 3606 mode during the locktorture test. 3607 3608 locktorture.shutdown_secs= [KNL] 3609 Set time (s) after boot system shutdown. This 3610 is useful for hands-off automated testing. 3611 3612 locktorture.stat_interval= [KNL] 3613 Time (s) between statistics printk()s. 3614 3615 locktorture.stutter= [KNL] 3616 Time (s) to stutter testing, for example, 3617 specifying five seconds causes the test to run for 3618 five seconds, wait for five seconds, and so on. 3619 This tests the locking primitive's ability to 3620 transition abruptly to and from idle. 3621 3622 locktorture.torture_type= [KNL] 3623 Specify the locking implementation to test. 3624 3625 locktorture.verbose= [KNL] 3626 Enable additional printk() statements. 3627 3628 locktorture.writer_fifo= [KNL] 3629 Run the write-side locktorture kthreads at 3630 sched_set_fifo() real-time priority. 3631 3632 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver 3633 Format: <irq> 3634 3635 loglevel= [KNL,EARLY] 3636 All Kernel Messages with a loglevel smaller than the 3637 console loglevel will be printed to the console. It can 3638 also be changed with klogd or other programs. The 3639 loglevels are defined as follows: 3640 3641 0 (KERN_EMERG) system is unusable 3642 1 (KERN_ALERT) action must be taken immediately 3643 2 (KERN_CRIT) critical conditions 3644 3 (KERN_ERR) error conditions 3645 4 (KERN_WARNING) warning conditions 3646 5 (KERN_NOTICE) normal but significant condition 3647 6 (KERN_INFO) informational 3648 7 (KERN_DEBUG) debug-level messages 3649 3650 log_buf_len=n[KMG] [KNL,EARLY] 3651 Sets the size of the printk ring buffer, in bytes. 3652 n must be a power of two and greater than the 3653 minimal size. The minimal size is defined by 3654 LOG_BUF_SHIFT kernel config parameter. There 3655 is also CONFIG_LOG_CPU_MAX_BUF_SHIFT config 3656 parameter that allows to increase the default size 3657 depending on the number of CPUs. See init/Kconfig 3658 for more details. 3659 3660 logo.nologo [FB] Disables display of the built-in Linux logo. 3661 This may be used to provide more screen space for 3662 kernel log messages and is useful when debugging 3663 kernel boot problems. 3664 3665 lp=0 [LP] Specify parallel ports to use, e.g, 3666 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses 3667 lp=reset first parallel port). 'lp=0' disables the 3668 lp=auto printer driver. 'lp=reset' (which can be 3669 specified in addition to the ports) causes 3670 attached printers to be reset. Using 3671 lp=port1,port2,... specifies the parallel ports 3672 to associate lp devices with, starting with 3673 lp0. A port specification may be 'none' to skip 3674 that lp device, or a parport name such as 3675 'parport0'. Specifying 'lp=auto' instead of a 3676 port specification list means that device IDs 3677 from each port should be examined, to see if 3678 an IEEE 1284-compliant printer is attached; if 3679 so, the driver will manage that printer. 3680 See also header of drivers/char/lp.c. 3681 3682 lpj=n [KNL] 3683 Sets loops_per_jiffy to given constant, thus avoiding 3684 time-consuming boot-time autodetection (up to 250 ms per 3685 CPU). 0 enables autodetection (default). To determine 3686 the correct value for your kernel, boot with normal 3687 autodetection and see what value is printed. Note that 3688 on SMP systems the preset will be applied to all CPUs, 3689 which is likely to cause problems if your CPUs need 3690 significantly divergent settings. An incorrect value 3691 will cause delays in the kernel to be wrong, leading to 3692 unpredictable I/O errors and other breakage. Although 3693 unlikely, in the extreme case this might damage your 3694 hardware. 3695 3696 lsm.debug [SECURITY] Enable LSM initialization debugging output. 3697 3698 lsm=lsm1,...,lsmN 3699 [SECURITY] Choose order of LSM initialization. This 3700 overrides CONFIG_LSM, and the "security=" parameter. 3701 3702 machtype= [Loongson] Share the same kernel image file between 3703 different yeeloong laptops. 3704 Example: machtype=lemote-yeeloong-2f-7inch 3705 3706 maxcpus= [SMP,EARLY] Maximum number of processors that an SMP kernel 3707 will bring up during bootup. maxcpus=n : n >= 0 limits 3708 the kernel to bring up 'n' processors. Surely after 3709 bootup you can bring up the other plugged cpu by executing 3710 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus 3711 only takes effect during system bootup. 3712 While n=0 is a special case, it is equivalent to "nosmp", 3713 which also disables the IO APIC. 3714 3715 max_loop= [LOOP] The number of loop block devices that get 3716 (loop.max_loop) unconditionally pre-created at init time. The default 3717 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead 3718 of statically allocating a predefined number, loop 3719 devices can be requested on-demand with the 3720 /dev/loop-control interface. 3721 3722 mce= [X86-{32,64}] 3723 3724 Please see Documentation/arch/x86/x86_64/machinecheck.rst for sysfs runtime tunables. 3725 3726 off 3727 disable machine check 3728 3729 no_cmci 3730 disable CMCI(Corrected Machine Check Interrupt) that 3731 Intel processor supports. Usually this disablement is 3732 not recommended, but it might be handy if your 3733 hardware is misbehaving. 3734 3735 Note that you'll get more problems without CMCI than 3736 with due to the shared banks, i.e. you might get 3737 duplicated error logs. 3738 3739 dont_log_ce 3740 don't make logs for corrected errors. All events 3741 reported as corrected are silently cleared by OS. This 3742 option will be useful if you have no interest in any 3743 of corrected errors. 3744 3745 ignore_ce 3746 disable features for corrected errors, e.g. 3747 polling timer and CMCI. All events reported as 3748 corrected are not cleared by OS and remained in its 3749 error banks. 3750 3751 Usually this disablement is not recommended, however 3752 if there is an agent checking/clearing corrected 3753 errors (e.g. BIOS or hardware monitoring 3754 applications), conflicting with OS's error handling, 3755 and you cannot deactivate the agent, then this option 3756 will be a help. 3757 3758 no_lmce 3759 do not opt-in to Local MCE delivery. Use legacy method 3760 to broadcast MCEs. 3761 3762 bootlog 3763 enable logging of machine checks left over from 3764 booting. Disabled by default on AMD Fam10h and older 3765 because some BIOS leave bogus ones. 3766 3767 If your BIOS doesn't do that it's a good idea to 3768 enable though to make sure you log even machine check 3769 events that result in a reboot. On Intel systems it is 3770 enabled by default. 3771 3772 nobootlog 3773 disable boot machine check logging. 3774 3775 monarchtimeout (number) 3776 sets the time in us to wait for other CPUs on machine 3777 checks. 0 to disable. 3778 3779 bios_cmci_threshold 3780 don't overwrite the bios-set CMCI threshold. This boot 3781 option prevents Linux from overwriting the CMCI 3782 threshold set by the bios. Without this option, Linux 3783 always sets the CMCI threshold to 1. Enabling this may 3784 make memory predictive failure analysis less effective 3785 if the bios sets thresholds for memory errors since we 3786 will not see details for all errors. 3787 3788 recovery 3789 force-enable recoverable machine check code paths 3790 3791 Everything else is in sysfs now. 3792 3793 3794 md= [HW] RAID subsystems devices and level 3795 See Documentation/admin-guide/md.rst. 3796 3797 mds= [X86,INTEL,EARLY] 3798 Control mitigation for the Micro-architectural Data 3799 Sampling (MDS) vulnerability. 3800 3801 Certain CPUs are vulnerable to an exploit against CPU 3802 internal buffers which can forward information to a 3803 disclosure gadget under certain conditions. 3804 3805 In vulnerable processors, the speculatively 3806 forwarded data can be used in a cache side channel 3807 attack, to access data to which the attacker does 3808 not have direct access. 3809 3810 This parameter controls the MDS mitigation. The 3811 options are: 3812 3813 full - Enable MDS mitigation on vulnerable CPUs 3814 full,nosmt - Enable MDS mitigation and disable 3815 SMT on vulnerable CPUs 3816 off - Unconditionally disable MDS mitigation 3817 3818 On TAA-affected machines, mds=off can be prevented by 3819 an active TAA mitigation as both vulnerabilities are 3820 mitigated with the same mechanism so in order to disable 3821 this mitigation, you need to specify tsx_async_abort=off 3822 too. 3823 3824 Not specifying this option is equivalent to 3825 mds=full. 3826 3827 For details see: Documentation/admin-guide/hw-vuln/mds.rst 3828 3829 mem=nn[KMG] [HEXAGON,EARLY] Set the memory size. 3830 Must be specified, otherwise memory size will be 0. 3831 3832 mem=nn[KMG] [KNL,BOOT,EARLY] Force usage of a specific amount 3833 of memory Amount of memory to be used in cases 3834 as follows: 3835 3836 1 for test; 3837 2 when the kernel is not able to see the whole system memory; 3838 3 memory that lies after 'mem=' boundary is excluded from 3839 the hypervisor, then assigned to KVM guests. 3840 4 to limit the memory available for kdump kernel. 3841 3842 [ARC,MICROBLAZE] - the limit applies only to low memory, 3843 high memory is not affected. 3844 3845 [ARM64] - only limits memory covered by the linear 3846 mapping. The NOMAP regions are not affected. 3847 3848 [X86] Work as limiting max address. Use together 3849 with memmap= to avoid physical address space collisions. 3850 Without memmap= PCI devices could be placed at addresses 3851 belonging to unused RAM. 3852 3853 Note that this only takes effects during boot time since 3854 in above case 3, memory may need be hot added after boot 3855 if system memory of hypervisor is not sufficient. 3856 3857 mem=nn[KMG]@ss[KMG] 3858 [ARM,MIPS,EARLY] - override the memory layout 3859 reported by firmware. 3860 Define a memory region of size nn[KMG] starting at 3861 ss[KMG]. 3862 Multiple different regions can be specified with 3863 multiple mem= parameters on the command line. 3864 3865 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel 3866 memory. 3867 3868 memblock=debug [KNL,EARLY] Enable memblock debug messages. 3869 3870 memchunk=nn[KMG] 3871 [KNL,SH] Allow user to override the default size for 3872 per-device physically contiguous DMA buffers. 3873 3874 memhp_default_state=online/offline/online_kernel/online_movable 3875 [KNL] Set the initial state for the memory hotplug 3876 onlining policy. If not specified, the default value is 3877 set according to the 3878 CONFIG_MHP_DEFAULT_ONLINE_TYPE kernel config 3879 options. 3880 See Documentation/admin-guide/mm/memory-hotplug.rst. 3881 3882 memmap=exactmap [KNL,X86,EARLY] Enable setting of an exact 3883 E820 memory map, as specified by the user. 3884 Such memmap=exactmap lines can be constructed based on 3885 BIOS output or other requirements. See the memmap=nn@ss 3886 option description. 3887 3888 memmap=nn[KMG]@ss[KMG] 3889 [KNL, X86,MIPS,XTENSA,EARLY] Force usage of a specific region of memory. 3890 Region of memory to be used is from ss to ss+nn. 3891 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG], 3892 which limits max address to nn[KMG]. 3893 Multiple different regions can be specified, 3894 comma delimited. 3895 Example: 3896 memmap=100M@2G,100M#3G,1G!1024G 3897 3898 memmap=nn[KMG]#ss[KMG] 3899 [KNL,ACPI,EARLY] Mark specific memory as ACPI data. 3900 Region of memory to be marked is from ss to ss+nn. 3901 3902 memmap=nn[KMG]$ss[KMG] 3903 [KNL,ACPI,EARLY] Mark specific memory as reserved. 3904 Region of memory to be reserved is from ss to ss+nn. 3905 Example: Exclude memory from 0x18690000-0x1869ffff 3906 memmap=64K$0x18690000 3907 or 3908 memmap=0x10000$0x18690000 3909 Some bootloaders may need an escape character before '$', 3910 like Grub2, otherwise '$' and the following number 3911 will be eaten. 3912 3913 memmap=nn[KMG]!ss[KMG,EARLY] 3914 [KNL,X86] Mark specific memory as protected. 3915 Region of memory to be used, from ss to ss+nn. 3916 The memory region may be marked as e820 type 12 (0xc) 3917 and is NVDIMM or ADR memory. 3918 3919 memmap=<size>%<offset>-<oldtype>+<newtype> 3920 [KNL,ACPI,EARLY] Convert memory within the specified region 3921 from <oldtype> to <newtype>. If "-<oldtype>" is left 3922 out, the whole region will be marked as <newtype>, 3923 even if previously unavailable. If "+<newtype>" is left 3924 out, matching memory will be removed. Types are 3925 specified as e820 types, e.g., 1 = RAM, 2 = reserved, 3926 3 = ACPI, 12 = PRAM. 3927 3928 memory_corruption_check=0/1 [X86,EARLY] 3929 Some BIOSes seem to corrupt the first 64k of 3930 memory when doing things like suspend/resume. 3931 Setting this option will scan the memory 3932 looking for corruption. Enabling this will 3933 both detect corruption and prevent the kernel 3934 from using the memory being corrupted. 3935 However, it's intended as a diagnostic tool; if 3936 repeatable BIOS-originated corruption always 3937 affects the same memory, you can use memmap= 3938 to prevent the kernel from using that memory. 3939 3940 memory_corruption_check_size=size [X86,EARLY] 3941 By default it checks for corruption in the low 3942 64k, making this memory unavailable for normal 3943 use. Use this parameter to scan for 3944 corruption in more or less memory. 3945 3946 memory_corruption_check_period=seconds [X86,EARLY] 3947 By default it checks for corruption every 60 3948 seconds. Use this parameter to check at some 3949 other rate. 0 disables periodic checking. 3950 3951 memory_hotplug.memmap_on_memory 3952 [KNL,X86,ARM] Boolean flag to enable this feature. 3953 Format: {on | off (default)} 3954 When enabled, runtime hotplugged memory will 3955 allocate its internal metadata (struct pages, 3956 those vmemmap pages cannot be optimized even 3957 if hugetlb_free_vmemmap is enabled) from the 3958 hotadded memory which will allow to hotadd a 3959 lot of memory without requiring additional 3960 memory to do so. 3961 This feature is disabled by default because it 3962 has some implication on large (e.g. GB) 3963 allocations in some configurations (e.g. small 3964 memory blocks). 3965 The state of the flag can be read in 3966 /sys/module/memory_hotplug/parameters/memmap_on_memory. 3967 Note that even when enabled, there are a few cases where 3968 the feature is not effective. 3969 3970 memtest= [KNL,X86,ARM,M68K,PPC,RISCV,EARLY] Enable memtest 3971 Format: <integer> 3972 default : 0 <disable> 3973 Specifies the number of memtest passes to be 3974 performed. Each pass selects another test 3975 pattern from a given set of patterns. Memtest 3976 fills the memory with this pattern, validates 3977 memory contents and reserves bad memory 3978 regions that are detected. 3979 3980 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control 3981 Valid arguments: on, off 3982 Default: off 3983 mem_encrypt=on: Activate SME 3984 mem_encrypt=off: Do not activate SME 3985 3986 Refer to Documentation/virt/kvm/x86/amd-memory-encryption.rst 3987 for details on when memory encryption can be activated. 3988 3989 mem_sleep_default= [SUSPEND] Default system suspend mode: 3990 s2idle - Suspend-To-Idle 3991 shallow - Power-On Suspend or equivalent (if supported) 3992 deep - Suspend-To-RAM or equivalent (if supported) 3993 See Documentation/admin-guide/pm/sleep-states.rst. 3994 3995 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when 3996 the BIOS has incorrectly applied a workaround. TinyBIOS 3997 version 0.98 is known to be affected, 0.99 fixes the 3998 problem by letting the user disable the workaround. 3999 4000 mga= [HW,DRM] 4001 4002 microcode= [X86] Control the behavior of the microcode loader. 4003 Available options, comma separated: 4004 4005 base_rev=X - with <X> with format: <u32> 4006 Set the base microcode revision of each thread when in 4007 debug mode. 4008 4009 dis_ucode_ldr: disable the microcode loader 4010 4011 force_minrev: 4012 Enable or disable the microcode minimal revision 4013 enforcement for the runtime microcode loader. 4014 4015 mini2440= [ARM,HW,KNL] 4016 Format:[0..2][b][c][t] 4017 Default: "0tb" 4018 MINI2440 configuration specification: 4019 0 - The attached screen is the 3.5" TFT 4020 1 - The attached screen is the 7" TFT 4021 2 - The VGA Shield is attached (1024x768) 4022 Leaving out the screen size parameter will not load 4023 the TFT driver, and the framebuffer will be left 4024 unconfigured. 4025 b - Enable backlight. The TFT backlight pin will be 4026 linked to the kernel VESA blanking code and a GPIO 4027 LED. This parameter is not necessary when using the 4028 VGA shield. 4029 c - Enable the s3c camera interface. 4030 t - Reserved for enabling touchscreen support. The 4031 touchscreen support is not enabled in the mainstream 4032 kernel as of 2.6.30, a preliminary port can be found 4033 in the "bleeding edge" mini2440 support kernel at 4034 https://repo.or.cz/w/linux-2.6/mini2440.git 4035 4036 mitigations= 4037 [X86,PPC,S390,ARM64,EARLY] Control optional mitigations for 4038 CPU vulnerabilities. This is a set of curated, 4039 arch-independent options, each of which is an 4040 aggregation of existing arch-specific options. 4041 4042 Note, "mitigations" is supported if and only if the 4043 kernel was built with CPU_MITIGATIONS=y. 4044 4045 off 4046 Disable all optional CPU mitigations. This 4047 improves system performance, but it may also 4048 expose users to several CPU vulnerabilities. 4049 Equivalent to: if nokaslr then kpti=0 [ARM64] 4050 gather_data_sampling=off [X86] 4051 indirect_target_selection=off [X86] 4052 kvm.nx_huge_pages=off [X86] 4053 l1tf=off [X86] 4054 mds=off [X86] 4055 mmio_stale_data=off [X86] 4056 no_entry_flush [PPC] 4057 no_uaccess_flush [PPC] 4058 nobp=0 [S390] 4059 nopti [X86,PPC] 4060 nospectre_bhb [ARM64] 4061 nospectre_v1 [X86,PPC] 4062 nospectre_v2 [X86,PPC,S390,ARM64] 4063 reg_file_data_sampling=off [X86] 4064 retbleed=off [X86] 4065 spec_rstack_overflow=off [X86] 4066 spec_store_bypass_disable=off [X86,PPC] 4067 spectre_bhi=off [X86] 4068 spectre_v2_user=off [X86] 4069 srbds=off [X86,INTEL] 4070 ssbd=force-off [ARM64] 4071 tsa=off [X86,AMD] 4072 tsx_async_abort=off [X86] 4073 vmscape=off [X86] 4074 4075 Exceptions: 4076 This does not have any effect on 4077 kvm.nx_huge_pages when 4078 kvm.nx_huge_pages=force. 4079 4080 auto (default) 4081 Mitigate all CPU vulnerabilities, but leave SMT 4082 enabled, even if it's vulnerable. This is for 4083 users who don't want to be surprised by SMT 4084 getting disabled across kernel upgrades, or who 4085 have other ways of avoiding SMT-based attacks. 4086 Equivalent to: (default behavior) 4087 4088 auto,nosmt 4089 Mitigate all CPU vulnerabilities, disabling SMT 4090 if needed. This is for users who always want to 4091 be fully mitigated, even if it means losing SMT. 4092 Equivalent to: l1tf=flush,nosmt [X86] 4093 mds=full,nosmt [X86] 4094 tsx_async_abort=full,nosmt [X86] 4095 mmio_stale_data=full,nosmt [X86] 4096 retbleed=auto,nosmt [X86] 4097 4098 [X86] After one of the above options, additionally 4099 supports attack-vector based controls as documented in 4100 Documentation/admin-guide/hw-vuln/attack_vector_controls.rst 4101 4102 mminit_loglevel= 4103 [KNL,EARLY] When CONFIG_DEBUG_MEMORY_INIT is set, this 4104 parameter allows control of the logging verbosity for 4105 the additional memory initialisation checks. A value 4106 of 0 disables mminit logging and a level of 4 will 4107 log everything. Information is printed at KERN_DEBUG 4108 so loglevel=8 may also need to be specified. 4109 4110 mmio_stale_data= 4111 [X86,INTEL,EARLY] Control mitigation for the Processor 4112 MMIO Stale Data vulnerabilities. 4113 4114 Processor MMIO Stale Data is a class of 4115 vulnerabilities that may expose data after an MMIO 4116 operation. Exposed data could originate or end in 4117 the same CPU buffers as affected by MDS and TAA. 4118 Therefore, similar to MDS and TAA, the mitigation 4119 is to clear the affected CPU buffers. 4120 4121 This parameter controls the mitigation. The 4122 options are: 4123 4124 full - Enable mitigation on vulnerable CPUs 4125 4126 full,nosmt - Enable mitigation and disable SMT on 4127 vulnerable CPUs. 4128 4129 off - Unconditionally disable mitigation 4130 4131 On MDS or TAA affected machines, 4132 mmio_stale_data=off can be prevented by an active 4133 MDS or TAA mitigation as these vulnerabilities are 4134 mitigated with the same mechanism so in order to 4135 disable this mitigation, you need to specify 4136 mds=off and tsx_async_abort=off too. 4137 4138 Not specifying this option is equivalent to 4139 mmio_stale_data=full. 4140 4141 For details see: 4142 Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst 4143 4144 <module>.async_probe[=<bool>] [KNL] 4145 If no <bool> value is specified or if the value 4146 specified is not a valid <bool>, enable asynchronous 4147 probe on this module. Otherwise, enable/disable 4148 asynchronous probe on this module as indicated by the 4149 <bool> value. See also: module.async_probe 4150 4151 module.async_probe=<bool> 4152 [KNL] When set to true, modules will use async probing 4153 by default. To enable/disable async probing for a 4154 specific module, use the module specific control that 4155 is documented under <module>.async_probe. When both 4156 module.async_probe and <module>.async_probe are 4157 specified, <module>.async_probe takes precedence for 4158 the specific module. 4159 4160 module.enable_dups_trace 4161 [KNL] When CONFIG_MODULE_DEBUG_AUTOLOAD_DUPS is set, 4162 this means that duplicate request_module() calls will 4163 trigger a WARN_ON() instead of a pr_warn(). Note that 4164 if MODULE_DEBUG_AUTOLOAD_DUPS_TRACE is set, WARN_ON()s 4165 will always be issued and this option does nothing. 4166 module.sig_enforce 4167 [KNL] When CONFIG_MODULE_SIG is set, this means that 4168 modules without (valid) signatures will fail to load. 4169 Note that if CONFIG_MODULE_SIG_FORCE is set, that 4170 is always true, so this option does nothing. 4171 4172 module_blacklist= [KNL] Do not load a comma-separated list of 4173 modules. Useful for debugging problem modules. 4174 4175 mousedev.tap_time= 4176 [MOUSE] Maximum time between finger touching and 4177 leaving touchpad surface for touch to be considered 4178 a tap and be reported as a left button click (for 4179 touchpads working in absolute mode only). 4180 Format: <msecs> 4181 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices 4182 reporting absolute coordinates, such as tablets 4183 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices 4184 reporting absolute coordinates, such as tablets 4185 4186 movablecore= [KNL,X86,PPC,EARLY] 4187 Format: nn[KMGTPE] | nn% 4188 This parameter is the complement to kernelcore=, it 4189 specifies the amount of memory used for migratable 4190 allocations. If both kernelcore and movablecore is 4191 specified, then kernelcore will be at *least* the 4192 specified value but may be more. If movablecore on its 4193 own is specified, the administrator must be careful 4194 that the amount of memory usable for all allocations 4195 is not too small. 4196 4197 movable_node [KNL,EARLY] Boot-time switch to make hotplugable memory 4198 NUMA nodes to be movable. This means that the memory 4199 of such nodes will be usable only for movable 4200 allocations which rules out almost all kernel 4201 allocations. Use with caution! 4202 4203 MTD_Partition= [MTD] 4204 Format: <name>,<region-number>,<size>,<offset> 4205 4206 MTD_Region= [MTD] Format: 4207 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>] 4208 4209 mtdparts= [MTD] 4210 See drivers/mtd/parsers/cmdlinepart.c 4211 4212 mtouchusb.raw_coordinates= 4213 [HW] Make the MicroTouch USB driver use raw coordinates 4214 ('y', default) or cooked coordinates ('n') 4215 4216 mtrr=debug [X86,EARLY] 4217 Enable printing debug information related to MTRR 4218 registers at boot time. 4219 4220 mtrr_chunk_size=nn[KMG,X86,EARLY] 4221 used for mtrr cleanup. It is largest continuous chunk 4222 that could hold holes aka. UC entries. 4223 4224 mtrr_gran_size=nn[KMG,X86,EARLY] 4225 Used for mtrr cleanup. It is granularity of mtrr block. 4226 Default is 1. 4227 Large value could prevent small alignment from 4228 using up MTRRs. 4229 4230 mtrr_spare_reg_nr=n [X86,EARLY] 4231 Format: <integer> 4232 Range: 0,7 : spare reg number 4233 Default : 1 4234 Used for mtrr cleanup. It is spare mtrr entries number. 4235 Set to 2 or more if your graphical card needs more. 4236 4237 multitce=off [PPC] This parameter disables the use of the pSeries 4238 firmware feature for updating multiple TCE entries 4239 at a time. 4240 4241 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card 4242 4243 netdev= [NET] NE2000 ISA network devices parameters 4244 Format: <irq>,<io>,<mem_start>,<mem_end>,<name> 4245 4246 netpoll.carrier_timeout= 4247 [NET] Specifies amount of time (in seconds) that 4248 netpoll should wait for a carrier. By default netpoll 4249 waits 4 seconds. 4250 4251 nf_conntrack.acct= 4252 [NETFILTER] Enable connection tracking flow accounting 4253 0 to disable accounting 4254 1 to enable accounting 4255 Default value is 0. 4256 4257 nfs.cache_getent= 4258 [NFS] sets the pathname to the program which is used 4259 to update the NFS client cache entries. 4260 4261 nfs.cache_getent_timeout= 4262 [NFS] sets the timeout after which an attempt to 4263 update a cache entry is deemed to have failed. 4264 4265 nfs.callback_nr_threads= 4266 [NFSv4] set the total number of threads that the 4267 NFS client will assign to service NFSv4 callback 4268 requests. 4269 4270 nfs.callback_tcpport= 4271 [NFS] set the TCP port on which the NFSv4 callback 4272 channel should listen. 4273 4274 nfs.delay_retrans= 4275 [NFS] specifies the number of times the NFSv4 client 4276 retries the request before returning an EAGAIN error, 4277 after a reply of NFS4ERR_DELAY from the server. 4278 Only applies if the softerr mount option is enabled, 4279 and the specified value is >= 0. 4280 4281 nfs.enable_ino64= 4282 [NFS] enable 64-bit inode numbers. 4283 If zero, the NFS client will fake up a 32-bit inode 4284 number for the readdir() and stat() syscalls instead 4285 of returning the full 64-bit number. 4286 The default is to return 64-bit inode numbers. 4287 4288 nfs.idmap_cache_timeout= 4289 [NFS] set the maximum lifetime for idmapper cache 4290 entries. 4291 4292 nfs.max_session_cb_slots= 4293 [NFSv4.1] Sets the maximum number of session 4294 slots the client will assign to the callback 4295 channel. This determines the maximum number of 4296 callbacks the client will process in parallel for 4297 a particular server. 4298 4299 nfs.max_session_slots= 4300 [NFSv4.1] Sets the maximum number of session slots 4301 the client will attempt to negotiate with the server. 4302 This limits the number of simultaneous RPC requests 4303 that the client can send to the NFSv4.1 server. 4304 Note that there is little point in setting this 4305 value higher than the max_tcp_slot_table_limit. 4306 4307 nfs.nfs4_disable_idmapping= 4308 [NFSv4] When set to the default of '1', this option 4309 ensures that both the RPC level authentication 4310 scheme and the NFS level operations agree to use 4311 numeric uids/gids if the mount is using the 4312 'sec=sys' security flavour. In effect it is 4313 disabling idmapping, which can make migration from 4314 legacy NFSv2/v3 systems to NFSv4 easier. 4315 Servers that do not support this mode of operation 4316 will be autodetected by the client, and it will fall 4317 back to using the idmapper. 4318 To turn off this behaviour, set the value to '0'. 4319 4320 nfs.nfs4_unique_id= 4321 [NFS4] Specify an additional fixed unique ident- 4322 ification string that NFSv4 clients can insert into 4323 their nfs_client_id4 string. This is typically a 4324 UUID that is generated at system install time. 4325 4326 nfs.recover_lost_locks= 4327 [NFSv4] Attempt to recover locks that were lost due 4328 to a lease timeout on the server. Please note that 4329 doing this risks data corruption, since there are 4330 no guarantees that the file will remain unchanged 4331 after the locks are lost. 4332 If you want to enable the kernel legacy behaviour of 4333 attempting to recover these locks, then set this 4334 parameter to '1'. 4335 The default parameter value of '0' causes the kernel 4336 not to attempt recovery of lost locks. 4337 4338 nfs.send_implementation_id= 4339 [NFSv4.1] Send client implementation identification 4340 information in exchange_id requests. 4341 If zero, no implementation identification information 4342 will be sent. 4343 The default is to send the implementation identification 4344 information. 4345 4346 nfs4.layoutstats_timer= 4347 [NFSv4.2] Change the rate at which the kernel sends 4348 layoutstats to the pNFS metadata server. 4349 4350 Setting this to value to 0 causes the kernel to use 4351 whatever value is the default set by the layout 4352 driver. A non-zero value sets the minimum interval 4353 in seconds between layoutstats transmissions. 4354 4355 nfsd.inter_copy_offload_enable= 4356 [NFSv4.2] When set to 1, the server will support 4357 server-to-server copies for which this server is 4358 the destination of the copy. 4359 4360 nfsd.nfs4_disable_idmapping= 4361 [NFSv4] When set to the default of '1', the NFSv4 4362 server will return only numeric uids and gids to 4363 clients using auth_sys, and will accept numeric uids 4364 and gids from such clients. This is intended to ease 4365 migration from NFSv2/v3. 4366 4367 nfsd.nfsd4_ssc_umount_timeout= 4368 [NFSv4.2] When used as the destination of a 4369 server-to-server copy, knfsd temporarily mounts 4370 the source server. It caches the mount in case 4371 it will be needed again, and discards it if not 4372 used for the number of milliseconds specified by 4373 this parameter. 4374 4375 nfsaddrs= [NFS] Deprecated. Use ip= instead. 4376 See Documentation/admin-guide/nfs/nfsroot.rst. 4377 4378 nfsroot= [NFS] nfs root filesystem for disk-less boxes. 4379 See Documentation/admin-guide/nfs/nfsroot.rst. 4380 4381 nfsrootdebug [NFS] enable nfsroot debugging messages. 4382 See Documentation/admin-guide/nfs/nfsroot.rst. 4383 4384 nmi_backtrace.backtrace_idle [KNL] 4385 Dump stacks even of idle CPUs in response to an 4386 NMI stack-backtrace request. 4387 4388 nmi_debug= [KNL,SH] Specify one or more actions to take 4389 when a NMI is triggered. 4390 Format: [state][,regs][,debounce][,die] 4391 4392 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels 4393 Format: [panic,][nopanic,][rNNN,][num] 4394 Valid num: 0 or 1 4395 0 - turn hardlockup detector in nmi_watchdog off 4396 1 - turn hardlockup detector in nmi_watchdog on 4397 rNNN - configure the watchdog with raw perf event 0xNNN 4398 4399 When panic is specified, panic when an NMI watchdog 4400 timeout occurs (or 'nopanic' to not panic on an NMI 4401 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set) 4402 To disable both hard and soft lockup detectors, 4403 please see 'nowatchdog'. 4404 This is useful when you use a panic=... timeout and 4405 need the box quickly up again. 4406 4407 These settings can be accessed at runtime via 4408 the nmi_watchdog and hardlockup_panic sysctls. 4409 4410 no4lvl [RISCV,EARLY] Disable 4-level and 5-level paging modes. 4411 Forces kernel to use 3-level paging instead. 4412 4413 no5lvl [X86-64,RISCV,EARLY] Disable 5-level paging mode. Forces 4414 kernel to use 4-level paging instead. 4415 4416 noalign [KNL,ARM] 4417 4418 noapic [SMP,APIC,EARLY] Tells the kernel to not make use of any 4419 IOAPICs that may be present in the system. 4420 4421 noapictimer [APIC,X86] Don't set up the APIC timer 4422 4423 noautogroup Disable scheduler automatic task group creation. 4424 4425 nocache [ARM,EARLY] 4426 4427 no_console_suspend 4428 [HW] Never suspend the console 4429 Disable suspending of consoles during suspend and 4430 hibernate operations. Once disabled, debugging 4431 messages can reach various consoles while the rest 4432 of the system is being put to sleep (ie, while 4433 debugging driver suspend/resume hooks). This may 4434 not work reliably with all consoles, but is known 4435 to work with serial and VGA consoles. 4436 To facilitate more flexible debugging, we also add 4437 console_suspend, a printk module parameter to control 4438 it. Users could use console_suspend (usually 4439 /sys/module/printk/parameters/console_suspend) to 4440 turn on/off it dynamically. 4441 4442 no_debug_objects 4443 [KNL,EARLY] Disable object debugging 4444 4445 nodsp [SH] Disable hardware DSP at boot time. 4446 4447 noefi [EFI,EARLY] Disable EFI runtime services support. 4448 4449 no_entry_flush [PPC,EARLY] Don't flush the L1-D cache when entering the kernel. 4450 4451 noexec32 [X86-64] 4452 This affects only 32-bit executables. 4453 noexec32=on: enable non-executable mappings (default) 4454 read doesn't imply executable mappings 4455 noexec32=off: disable non-executable mappings 4456 read implies executable mappings 4457 4458 no_file_caps Tells the kernel not to honor file capabilities. The 4459 only way then for a file to be executed with privilege 4460 is to be setuid root or executed by root. 4461 4462 nofpu [MIPS,SH] Disable hardware FPU at boot time. 4463 4464 nofsgsbase [X86] Disables FSGSBASE instructions. 4465 4466 nofxsr [BUGS=X86-32] Disables x86 floating point extended 4467 register save and restore. The kernel will only save 4468 legacy floating-point registers on task switch. 4469 4470 nogbpages [X86] Do not use GB pages for kernel direct mappings. 4471 4472 no_hash_pointers 4473 [KNL,EARLY] 4474 Alias for "hash_pointers=never". 4475 4476 nohibernate [HIBERNATION] Disable hibernation and resume. 4477 4478 nohlt [ARM,ARM64,MICROBLAZE,MIPS,PPC,RISCV,SH] Forces the kernel to 4479 busy wait in do_idle() and not use the arch_cpu_idle() 4480 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP 4481 to be effective. This is useful on platforms where the 4482 sleep(SH) or wfi(ARM,ARM64) instructions do not work 4483 correctly or when doing power measurements to evaluate 4484 the impact of the sleep instructions. This is also 4485 useful when using JTAG debugger. 4486 4487 nohpet [X86] Don't use the HPET timer. 4488 4489 nohugeiomap [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge I/O mappings. 4490 4491 nohugevmalloc [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge vmalloc mappings. 4492 4493 nohz= [KNL] Boottime enable/disable dynamic ticks 4494 Valid arguments: on, off 4495 Default: on 4496 4497 nohz_full= [KNL,BOOT,SMP,ISOL] 4498 The argument is a cpu list, as described above. 4499 In kernels built with CONFIG_NO_HZ_FULL=y, set 4500 the specified list of CPUs whose tick will be stopped 4501 whenever possible. The boot CPU will be forced outside 4502 the range to maintain the timekeeping. Any CPUs 4503 in this list will have their RCU callbacks offloaded, 4504 just as if they had also been called out in the 4505 rcu_nocbs= boot parameter. 4506 4507 Note that this argument takes precedence over 4508 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option. 4509 4510 noinitrd [Deprecated,RAM] Tells the kernel not to load any configured 4511 initial RAM disk. Currently this parameter applies to 4512 initrd only, not to initramfs. But it applies to both 4513 in EFI mode. 4514 4515 nointremap [X86-64,Intel-IOMMU,EARLY] Do not enable interrupt 4516 remapping. 4517 [Deprecated - use intremap=off] 4518 4519 noinvpcid [X86,EARLY] Disable the INVPCID cpu feature. 4520 4521 noiotrap [SH] Disables trapped I/O port accesses. 4522 4523 noirqdebug [X86-32] Disables the code which attempts to detect and 4524 disable unhandled interrupt sources. 4525 4526 noisapnp [ISAPNP] Disables ISA PnP code. 4527 4528 nokaslr [KNL,EARLY] 4529 When CONFIG_RANDOMIZE_BASE is set, this disables 4530 kernel and module base offset ASLR (Address Space 4531 Layout Randomization). 4532 4533 no-kvmapf [X86,KVM,EARLY] Disable paravirtualized asynchronous page 4534 fault handling. 4535 4536 no-kvmclock [X86,KVM,EARLY] Disable paravirtualized KVM clock driver 4537 4538 nolapic [X86-32,APIC,EARLY] Do not enable or use the local APIC. 4539 4540 nolapic_timer [X86-32,APIC,EARLY] Do not use the local APIC timer. 4541 4542 nomce [X86-32] Disable Machine Check Exception 4543 4544 nomfgpt [X86-32] Disable Multi-Function General Purpose 4545 Timer usage (for AMD Geode machines). 4546 4547 nomodeset Disable kernel modesetting. Most systems' firmware 4548 sets up a display mode and provides framebuffer memory 4549 for output. With nomodeset, DRM and fbdev drivers will 4550 not load if they could possibly displace the pre- 4551 initialized output. Only the system framebuffer will 4552 be available for use. The respective drivers will not 4553 perform display-mode changes or accelerated rendering. 4554 4555 Useful as error fallback, or for testing and debugging. 4556 4557 nomodule Disable module load 4558 4559 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to 4560 shutdown the other cpus. Instead use the REBOOT_VECTOR 4561 irq. 4562 4563 nopat [X86,EARLY] Disable PAT (page attribute table extension of 4564 pagetables) support. 4565 4566 nopcid [X86-64,EARLY] Disable the PCID cpu feature. 4567 4568 nopku [X86] Disable Memory Protection Keys CPU feature found 4569 in some Intel CPUs. 4570 4571 nopti [X86-64,EARLY] 4572 Equivalent to pti=off 4573 4574 nopv= [X86,XEN,KVM,HYPER_V,VMWARE,EARLY] 4575 Disables the PV optimizations forcing the guest to run 4576 as generic guest with no PV drivers. Currently support 4577 XEN HVM, KVM, HYPER_V and VMWARE guest. 4578 4579 nopvspin [X86,XEN,KVM,EARLY] 4580 Disables the qspinlock slow path using PV optimizations 4581 which allow the hypervisor to 'idle' the guest on lock 4582 contention. 4583 4584 norandmaps Don't use address space randomization. Equivalent to 4585 echo 0 > /proc/sys/kernel/randomize_va_space 4586 4587 noreplace-smp [X86-32,SMP] Don't replace SMP instructions 4588 with UP alternatives 4589 4590 noresume [SWSUSP] Disables resume and restores original swap 4591 space. 4592 4593 no-scroll [VGA] Disables scrollback. 4594 This is required for the Braillex ib80-piezo Braille 4595 reader made by F.H. Papenmeier (Germany). 4596 4597 nosgx [X86-64,SGX,EARLY] Disables Intel SGX kernel support. 4598 4599 nosmap [PPC,EARLY] 4600 Disable SMAP (Supervisor Mode Access Prevention) 4601 even if it is supported by processor. 4602 4603 nosmep [PPC64s,EARLY] 4604 Disable SMEP (Supervisor Mode Execution Prevention) 4605 even if it is supported by processor. 4606 4607 nosmp [SMP,EARLY] Tells an SMP kernel to act as a UP kernel, 4608 and disable the IO APIC. legacy for "maxcpus=0". 4609 4610 nosmt [KNL,MIPS,PPC,EARLY] Disable symmetric multithreading (SMT). 4611 Equivalent to smt=1. 4612 4613 [KNL,LOONGARCH,X86,ARM64,PPC,S390] Disable symmetric multithreading (SMT). 4614 nosmt=force: Force disable SMT, cannot be undone 4615 via the sysfs control file. 4616 4617 nosoftlockup [KNL] Disable the soft-lockup detector. 4618 4619 nospec_store_bypass_disable 4620 [HW,EARLY] Disable all mitigations for the Speculative 4621 Store Bypass vulnerability 4622 4623 nospectre_bhb [ARM64,EARLY] Disable all mitigations for Spectre-BHB (branch 4624 history injection) vulnerability. System may allow data leaks 4625 with this option. 4626 4627 nospectre_v1 [X86,PPC,EARLY] Disable mitigations for Spectre Variant 1 4628 (bounds check bypass). With this option data leaks are 4629 possible in the system. 4630 4631 nospectre_v2 [X86,PPC_E500,ARM64,EARLY] Disable all mitigations 4632 for the Spectre variant 2 (indirect branch 4633 prediction) vulnerability. System may allow data 4634 leaks with this option. 4635 4636 no-steal-acc [X86,PV_OPS,ARM64,PPC/PSERIES,RISCV,LOONGARCH,EARLY] 4637 Disable paravirtualized steal time accounting. steal time 4638 is computed, but won't influence scheduler behaviour 4639 4640 nosync [HW,M68K] Disables sync negotiation for all devices. 4641 4642 no_timer_check [X86,APIC] Disables the code which tests for broken 4643 timer IRQ sources, i.e., the IO-APIC timer. This can 4644 work around problems with incorrect timer 4645 initialization on some boards. 4646 4647 no_uaccess_flush 4648 [PPC,EARLY] Don't flush the L1-D cache after accessing user data. 4649 4650 novmcoredd [KNL,KDUMP] 4651 Disable device dump. Device dump allows drivers to 4652 append dump data to vmcore so you can collect driver 4653 specified debug info. Drivers can append the data 4654 without any limit and this data is stored in memory, 4655 so this may cause significant memory stress. Disabling 4656 device dump can help save memory but the driver debug 4657 data will be no longer available. This parameter 4658 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP 4659 is set. 4660 4661 no-vmw-sched-clock 4662 [X86,PV_OPS,EARLY] Disable paravirtualized VMware 4663 scheduler clock and use the default one. 4664 4665 nowatchdog [KNL] Disable both lockup detectors, i.e. 4666 soft-lockup and NMI watchdog (hard-lockup). 4667 4668 nowb [ARM,EARLY] 4669 4670 nox2apic [X86-64,APIC,EARLY] Do not enable x2APIC mode. 4671 4672 NOTE: this parameter will be ignored on systems with the 4673 LEGACY_XAPIC_DISABLED bit set in the 4674 IA32_XAPIC_DISABLE_STATUS MSR. 4675 4676 noxsave [BUGS=X86] Disables x86 extended register state save 4677 and restore using xsave. The kernel will fallback to 4678 enabling legacy floating-point and sse state. 4679 4680 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended 4681 register states. The kernel will fall back to use 4682 xsave to save the states. By using this parameter, 4683 performance of saving the states is degraded because 4684 xsave doesn't support modified optimization while 4685 xsaveopt supports it on xsaveopt enabled systems. 4686 4687 noxsaves [X86] Disables xsaves and xrstors used in saving and 4688 restoring x86 extended register state in compacted 4689 form of xsave area. The kernel will fall back to use 4690 xsaveopt and xrstor to save and restore the states 4691 in standard form of xsave area. By using this 4692 parameter, xsave area per process might occupy more 4693 memory on xsaves enabled systems. 4694 4695 nr_cpus= [SMP,EARLY] Maximum number of processors that an SMP kernel 4696 could support. nr_cpus=n : n >= 1 limits the kernel to 4697 support 'n' processors. It could be larger than the 4698 number of already plugged CPU during bootup, later in 4699 runtime you can physically add extra cpu until it reaches 4700 n. So during boot up some boot time memory for per-cpu 4701 variables need be pre-allocated for later physical cpu 4702 hot plugging. 4703 4704 nr_uarts= [SERIAL] maximum number of UARTs to be registered. 4705 4706 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86, EARLY] 4707 Disable NUMA, Only set up a single NUMA node 4708 spanning all memory. 4709 4710 numa=fake=<size>[MG] 4711 [KNL, ARM64, RISCV, X86, EARLY] 4712 If given as a memory unit, fills all system RAM with 4713 nodes of size interleaved over physical nodes. 4714 4715 numa=fake=<N> 4716 [KNL, ARM64, RISCV, X86, EARLY] 4717 If given as an integer, fills all system RAM with N 4718 fake nodes interleaved over physical nodes. 4719 4720 numa=fake=<N>U 4721 [KNL, ARM64, RISCV, X86, EARLY] 4722 If given as an integer followed by 'U', it will 4723 divide each physical node into N emulated nodes. 4724 4725 numa=noacpi [X86] Don't parse the SRAT table for NUMA setup 4726 4727 numa=nohmat [X86] Don't parse the HMAT table for NUMA setup, or 4728 soft-reserved memory partitioning. 4729 4730 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic 4731 NUMA balancing. 4732 Allowed values are enable and disable 4733 4734 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA. 4735 'node', 'default' can be specified 4736 This can be set from sysctl after boot. 4737 See Documentation/admin-guide/sysctl/vm.rst for details. 4738 4739 nvme.quirks= [NVME] A list of quirk entries to augment the built-in 4740 nvme quirk list. List entries are separated by a 4741 '-' character. 4742 Each entry has the form VendorID:ProductID:quirk_names. 4743 The IDs are 4-digits hex numbers and quirk_names is a 4744 list of quirk names separated by commas. A quirk name 4745 can be prefixed by '^', meaning that the specified 4746 quirk must be disabled. 4747 4748 Example: 4749 nvme.quirks=7710:2267:bogus_nid,^identify_cns-9900:7711:broken_msi 4750 4751 ohci1394_dma=early [HW,EARLY] enable debugging via the ohci1394 driver. 4752 See Documentation/core-api/debugging-via-ohci1394.rst for more 4753 info. 4754 4755 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands 4756 Rather than timing out after 20 ms if an EC 4757 command is not properly ACKed, override the length 4758 of the timeout. We have interrupts disabled while 4759 waiting for the ACK, so if this is set too high 4760 interrupts *may* be lost! 4761 4762 omap_mux= [OMAP] Override bootloader pin multiplexing. 4763 Format: <mux_mode0.mode_name=value>... 4764 For example, to override I2C bus2: 4765 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100 4766 4767 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration 4768 4769 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock] 4770 4771 boundary - index of last SLC block on Flex-OneNAND. 4772 The remaining blocks are configured as MLC blocks. 4773 lock - Configure if Flex-OneNAND boundary should be locked. 4774 Once locked, the boundary cannot be changed. 4775 1 indicates lock status, 0 indicates unlock status. 4776 4777 oops=panic [KNL,EARLY] 4778 Always panic on oopses. Default is to just kill the 4779 process, but there is a small probability of 4780 deadlocking the machine. 4781 This will also cause panics on machine check exceptions. 4782 Useful together with panic=30 to trigger a reboot. 4783 4784 page_alloc.shuffle= 4785 [KNL] Boolean flag to control whether the page allocator 4786 should randomize its free lists. This parameter can be 4787 used to enable/disable page randomization. The state of 4788 the flag can be read from sysfs at: 4789 /sys/module/page_alloc/parameters/shuffle. 4790 This parameter is only available if CONFIG_SHUFFLE_PAGE_ALLOCATOR=y. 4791 4792 page_owner= [KNL,EARLY] Boot-time page_owner enabling option. 4793 Storage of the information about who allocated 4794 each page is disabled in default. With this switch, 4795 we can turn it on. 4796 on: enable the feature 4797 4798 page_poison= [KNL,EARLY] Boot-time parameter changing the state of 4799 poisoning on the buddy allocator, available with 4800 CONFIG_PAGE_POISONING=y. 4801 off: turn off poisoning (default) 4802 on: turn on poisoning 4803 4804 page_reporting.page_reporting_order= 4805 [KNL] Minimal page reporting order 4806 Format: <integer> 4807 Adjust the minimal page reporting order. The page 4808 reporting is disabled when it exceeds MAX_PAGE_ORDER. 4809 4810 panic= [KNL] Kernel behaviour on panic: delay <timeout> 4811 timeout > 0: seconds before rebooting 4812 timeout = 0: wait forever 4813 timeout < 0: reboot immediately 4814 Format: <timeout> 4815 4816 panic_on_taint= [KNL,EARLY] 4817 Bitmask for conditionally calling panic() in add_taint() 4818 Format: <hex>[,nousertaint] 4819 Hexadecimal bitmask representing the set of TAINT flags 4820 that will cause the kernel to panic when add_taint() is 4821 called with any of the flags in this set. 4822 The optional switch "nousertaint" can be utilized to 4823 prevent userspace forced crashes by writing to sysctl 4824 /proc/sys/kernel/tainted any flagset matching with the 4825 bitmask set on panic_on_taint. 4826 See Documentation/admin-guide/tainted-kernels.rst for 4827 extra details on the taint flags that users can pick 4828 to compose the bitmask to assign to panic_on_taint. 4829 4830 panic_on_warn=1 panic() instead of WARN(). Useful to cause kdump 4831 on a WARN(). 4832 4833 panic_force_cpu= 4834 [KNL,SMP] Force panic handling to execute on a specific CPU. 4835 Format: <cpu number> 4836 Some platforms require panic handling to occur on a 4837 specific CPU for the crash kernel to function correctly. 4838 This can be due to firmware limitations, interrupt routing 4839 constraints, or platform-specific requirements where only 4840 a particular CPU can safely enter the crash kernel. 4841 When set, panic() will redirect execution to the specified 4842 CPU before proceeding with the normal panic and kexec flow. 4843 If the target CPU is offline or unavailable, panic proceeds 4844 on the current CPU. 4845 This option should only be used for systems with the above 4846 constraints as it might cause the panic operation to be less reliable. 4847 4848 panic_print= Bitmask for printing system info when panic happens. 4849 User can chose combination of the following bits: 4850 bit 0: print all tasks info 4851 bit 1: print system memory info 4852 bit 2: print timer info 4853 bit 3: print locks info if CONFIG_LOCKDEP is on 4854 bit 4: print ftrace buffer 4855 bit 5: replay all kernel messages on consoles at the end of panic 4856 bit 6: print all CPUs backtrace (if available in the arch) 4857 bit 7: print only tasks in uninterruptible (blocked) state 4858 *Be aware* that this option may print a _lot_ of lines, 4859 so there are risks of losing older messages in the log. 4860 Use this option carefully, maybe worth to setup a 4861 bigger log buffer with "log_buf_len" along with this. 4862 4863 panic_sys_info= A comma separated list of extra information to be dumped 4864 on panic. 4865 Format: val[,val...] 4866 Where @val can be any of the following: 4867 4868 tasks: print all tasks info 4869 mem: print system memory info 4870 timers: print timers info 4871 locks: print locks info if CONFIG_LOCKDEP is on 4872 ftrace: print ftrace buffer 4873 all_bt: print all CPUs backtrace (if available in the arch) 4874 blocked_tasks: print only tasks in uninterruptible (blocked) state 4875 4876 This is a human readable alternative to the 'panic_print' option. 4877 4878 panic_console_replay 4879 When panic happens, replay all kernel messages on 4880 consoles at the end of panic. 4881 4882 parkbd.port= [HW] Parallel port number the keyboard adapter is 4883 connected to, default is 0. 4884 Format: <parport#> 4885 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation, 4886 0 for XT, 1 for AT (default is AT). 4887 Format: <mode> 4888 4889 parport= [HW,PPT] Specify parallel ports. 0 disables. 4890 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] } 4891 Use 'auto' to force the driver to use any 4892 IRQ/DMA settings detected (the default is to 4893 ignore detected IRQ/DMA settings because of 4894 possible conflicts). You can specify the base 4895 address, IRQ, and DMA settings; IRQ and DMA 4896 should be numbers, or 'auto' (for using detected 4897 settings on that particular port), or 'nofifo' 4898 (to avoid using a FIFO even if it is detected). 4899 Parallel ports are assigned in the order they 4900 are specified on the command line, starting 4901 with parport0. 4902 4903 parport_init_mode= [HW,PPT] 4904 Configure VIA parallel port to operate in 4905 a specific mode. This is necessary on Pegasos 4906 computer where firmware has no options for setting 4907 up parallel port mode and sets it to spp. 4908 Currently this function knows 686a and 8231 chips. 4909 Format: [spp|ps2|epp|ecp|ecpepp] 4910 4911 pata_legacy.all= [HW,LIBATA] 4912 Format: <int> 4913 Set to non-zero to probe primary and secondary ISA 4914 port ranges on PCI systems where no PCI PATA device 4915 has been found at either range. Disabled by default. 4916 4917 pata_legacy.autospeed= [HW,LIBATA] 4918 Format: <int> 4919 Set to non-zero if a chip is present that snoops speed 4920 changes. Disabled by default. 4921 4922 pata_legacy.ht6560a= [HW,LIBATA] 4923 Format: <int> 4924 Set to 1, 2, or 3 for HT 6560A on the primary channel, 4925 the secondary channel, or both channels respectively. 4926 Disabled by default. 4927 4928 pata_legacy.ht6560b= [HW,LIBATA] 4929 Format: <int> 4930 Set to 1, 2, or 3 for HT 6560B on the primary channel, 4931 the secondary channel, or both channels respectively. 4932 Disabled by default. 4933 4934 pata_legacy.iordy_mask= [HW,LIBATA] 4935 Format: <int> 4936 IORDY enable mask. Set individual bits to allow IORDY 4937 for the respective channel. Bit 0 is for the first 4938 legacy channel handled by this driver, bit 1 is for 4939 the second channel, and so on. The sequence will often 4940 correspond to the primary legacy channel, the secondary 4941 legacy channel, and so on, but the handling of a PCI 4942 bus and the use of other driver options may interfere 4943 with the sequence. By default IORDY is allowed across 4944 all channels. 4945 4946 pata_legacy.opti82c46x= [HW,LIBATA] 4947 Format: <int> 4948 Set to 1, 2, or 3 for Opti 82c611A on the primary 4949 channel, the secondary channel, or both channels 4950 respectively. Disabled by default. 4951 4952 pata_legacy.opti82c611a= [HW,LIBATA] 4953 Format: <int> 4954 Set to 1, 2, or 3 for Opti 82c465MV on the primary 4955 channel, the secondary channel, or both channels 4956 respectively. Disabled by default. 4957 4958 pata_legacy.pio_mask= [HW,LIBATA] 4959 Format: <int> 4960 PIO mode mask for autospeed devices. Set individual 4961 bits to allow the use of the respective PIO modes. 4962 Bit 0 is for mode 0, bit 1 is for mode 1, and so on. 4963 All modes allowed by default. 4964 4965 pata_legacy.probe_all= [HW,LIBATA] 4966 Format: <int> 4967 Set to non-zero to probe tertiary and further ISA 4968 port ranges on PCI systems. Disabled by default. 4969 4970 pata_legacy.probe_mask= [HW,LIBATA] 4971 Format: <int> 4972 Probe mask for legacy ISA PATA ports. Depending on 4973 platform configuration and the use of other driver 4974 options up to 6 legacy ports are supported: 0x1f0, 4975 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing 4976 of individual ports can be disabled by setting the 4977 corresponding bits in the mask to 1. Bit 0 is for 4978 the first port in the list above (0x1f0), and so on. 4979 By default all supported ports are probed. 4980 4981 pata_legacy.qdi= [HW,LIBATA] 4982 Format: <int> 4983 Set to non-zero to probe QDI controllers. By default 4984 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise. 4985 4986 pata_legacy.winbond= [HW,LIBATA] 4987 Format: <int> 4988 Set to non-zero to probe Winbond controllers. Use 4989 the standard I/O port (0x130) if 1, otherwise the 4990 value given is the I/O port to use (typically 0x1b0). 4991 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE, 4992 0 otherwise. 4993 4994 pata_platform.pio_mask= [HW,LIBATA] 4995 Format: <int> 4996 Supported PIO mode mask. Set individual bits to allow 4997 the use of the respective PIO modes. Bit 0 is for 4998 mode 0, bit 1 is for mode 1, and so on. Mode 0 only 4999 allowed by default. 5000 5001 pause_on_oops=<int> 5002 Halt all CPUs after the first oops has been printed for 5003 the specified number of seconds. This is to be used if 5004 your oopses keep scrolling off the screen. 5005 5006 pci=option[,option...] [PCI,EARLY] various PCI subsystem options. 5007 5008 Some options herein operate on a specific device 5009 or a set of devices (<pci_dev>). These are 5010 specified in one of the following formats: 5011 5012 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]* 5013 pci:<vendor>:<device>[:<subvendor>:<subdevice>] 5014 5015 Note: the first format specifies a PCI 5016 bus/device/function address which may change 5017 if new hardware is inserted, if motherboard 5018 firmware changes, or due to changes caused 5019 by other kernel parameters. If the 5020 domain is left unspecified, it is 5021 taken to be zero. Optionally, a path 5022 to a device through multiple device/function 5023 addresses can be specified after the base 5024 address (this is more robust against 5025 renumbering issues). The second format 5026 selects devices using IDs from the 5027 configuration space which may match multiple 5028 devices in the system. 5029 5030 earlydump dump PCI config space before the kernel 5031 changes anything 5032 off [X86] don't probe for the PCI bus 5033 bios [X86-32] force use of PCI BIOS, don't access 5034 the hardware directly. Use this if your machine 5035 has a non-standard PCI host bridge. 5036 nobios [X86-32] disallow use of PCI BIOS, only direct 5037 hardware access methods are allowed. Use this 5038 if you experience crashes upon bootup and you 5039 suspect they are caused by the BIOS. 5040 conf1 [X86] Force use of PCI Configuration Access 5041 Mechanism 1 (config address in IO port 0xCF8, 5042 data in IO port 0xCFC, both 32-bit). 5043 conf2 [X86] Force use of PCI Configuration Access 5044 Mechanism 2 (IO port 0xCF8 is an 8-bit port for 5045 the function, IO port 0xCFA, also 8-bit, sets 5046 bus number. The config space is then accessed 5047 through ports 0xC000-0xCFFF). 5048 See http://wiki.osdev.org/PCI for more info 5049 on the configuration access mechanisms. 5050 noaer [PCIE] If the PCIEAER kernel config parameter is 5051 enabled, this kernel boot option can be used to 5052 disable the use of PCIE advanced error reporting. 5053 nodomains [PCI] Disable support for multiple PCI 5054 root domains (aka PCI segments, in ACPI-speak). 5055 nommconf [X86] Disable use of MMCONFIG for PCI 5056 Configuration 5057 check_enable_amd_mmconf [X86] check for and enable 5058 properly configured MMIO access to PCI 5059 config space on AMD family 10h CPU 5060 nomsi [MSI] If the PCI_MSI kernel config parameter is 5061 enabled, this kernel boot option can be used to 5062 disable the use of MSI interrupts system-wide. 5063 noioapicquirk [APIC] Disable all boot interrupt quirks. 5064 Safety option to keep boot IRQs enabled. This 5065 should never be necessary. 5066 ioapicreroute [APIC] Enable rerouting of boot IRQs to the 5067 primary IO-APIC for bridges that cannot disable 5068 boot IRQs. This fixes a source of spurious IRQs 5069 when the system masks IRQs. 5070 noioapicreroute [APIC] Disable workaround that uses the 5071 boot IRQ equivalent of an IRQ that connects to 5072 a chipset where boot IRQs cannot be disabled. 5073 The opposite of ioapicreroute. 5074 biosirq [X86-32] Use PCI BIOS calls to get the interrupt 5075 routing table. These calls are known to be buggy 5076 on several machines and they hang the machine 5077 when used, but on other computers it's the only 5078 way to get the interrupt routing table. Try 5079 this option if the kernel is unable to allocate 5080 IRQs or discover secondary PCI buses on your 5081 motherboard. 5082 rom [X86] Assign address space to expansion ROMs. 5083 Use with caution as certain devices share 5084 address decoders between ROMs and other 5085 resources. 5086 norom [X86] Do not assign address space to 5087 expansion ROMs that do not already have 5088 BIOS assigned address ranges. 5089 nobar [X86] Do not assign address space to the 5090 BARs that weren't assigned by the BIOS. 5091 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be 5092 assigned automatically to PCI devices. You can 5093 make the kernel exclude IRQs of your ISA cards 5094 this way. 5095 pirqaddr=0xAAAAA [X86] Specify the physical address 5096 of the PIRQ table (normally generated 5097 by the BIOS) if it is outside the 5098 F0000h-100000h range. 5099 lastbus=N [X86] Scan all buses thru bus #N. Can be 5100 useful if the kernel is unable to find your 5101 secondary buses and you want to tell it 5102 explicitly which ones they are. 5103 assign-busses [X86] Always assign all PCI bus 5104 numbers ourselves, overriding 5105 whatever the firmware may have done. 5106 usepirqmask [X86] Honor the possible IRQ mask stored 5107 in the BIOS $PIR table. This is needed on 5108 some systems with broken BIOSes, notably 5109 some HP Pavilion N5400 and Omnibook XE3 5110 notebooks. This will have no effect if ACPI 5111 IRQ routing is enabled. 5112 noacpi [X86] Do not use ACPI for IRQ routing 5113 or for PCI scanning. 5114 use_crs [X86] Use PCI host bridge window information 5115 from ACPI. On BIOSes from 2008 or later, this 5116 is enabled by default. If you need to use this, 5117 please report a bug. 5118 nocrs [X86] Ignore PCI host bridge windows from ACPI. 5119 If you need to use this, please report a bug. 5120 use_e820 [X86] Use E820 reservations to exclude parts of 5121 PCI host bridge windows. This is a workaround 5122 for BIOS defects in host bridge _CRS methods. 5123 If you need to use this, please report a bug to 5124 <linux-pci@vger.kernel.org>. 5125 no_e820 [X86] Ignore E820 reservations for PCI host 5126 bridge windows. This is the default on modern 5127 hardware. If you need to use this, please report 5128 a bug to <linux-pci@vger.kernel.org>. 5129 routeirq Do IRQ routing for all PCI devices. 5130 This is normally done in pci_enable_device(), 5131 so this option is a temporary workaround 5132 for broken drivers that don't call it. 5133 skip_isa_align [X86] do not align io start addr, so can 5134 handle more pci cards 5135 noearly [X86] Don't do any early type 1 scanning. 5136 This might help on some broken boards which 5137 machine check when some devices' config space 5138 is read. But various workarounds are disabled 5139 and some IOMMU drivers will not work. 5140 bfsort Sort PCI devices into breadth-first order. 5141 This sorting is done to get a device 5142 order compatible with older (<= 2.4) kernels. 5143 nobfsort Don't sort PCI devices into breadth-first order. 5144 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size) 5145 tuning and use the BIOS-configured MPS defaults. 5146 pcie_bus_safe Set every device's MPS to the largest value 5147 supported by all devices below the root complex. 5148 pcie_bus_perf Set device MPS to the largest allowable MPS 5149 based on its parent bus. Also set MRRS (Max 5150 Read Request Size) to the largest supported 5151 value (no larger than the MPS that the device 5152 or bus can support) for best performance. 5153 pcie_bus_peer2peer Set every device's MPS to 128B, which 5154 every device is guaranteed to support. This 5155 configuration allows peer-to-peer DMA between 5156 any pair of devices, possibly at the cost of 5157 reduced performance. This also guarantees 5158 that hot-added devices will work. 5159 cbiosize=nn[KMG] The fixed amount of bus space which is 5160 reserved for the CardBus bridge's IO window. 5161 The default value is 256 bytes. 5162 cbmemsize=nn[KMG] The fixed amount of bus space which is 5163 reserved for the CardBus bridge's memory 5164 window. The default value is 64 megabytes. 5165 resource_alignment= 5166 Format: 5167 [<order of align>@]<pci_dev>[; ...] 5168 Specifies alignment and device to reassign 5169 aligned memory resources. How to 5170 specify the device is described above. 5171 If <order of align> is not specified, 5172 PAGE_SIZE is used as alignment. 5173 A PCI-PCI bridge can be specified if resource 5174 windows need to be expanded. 5175 To specify the alignment for several 5176 instances of a device, the PCI vendor, 5177 device, subvendor, and subdevice may be 5178 specified, e.g., 12@pci:8086:9c22:103c:198f 5179 for 4096-byte alignment. 5180 ecrc= Enable/disable PCIe ECRC (transaction layer 5181 end-to-end CRC checking). Only effective if 5182 OS has native AER control (either granted by 5183 ACPI _OSC or forced via "pcie_ports=native") 5184 bios: Use BIOS/firmware settings. This is the 5185 the default. 5186 off: Turn ECRC off 5187 on: Turn ECRC on. 5188 hpiosize=nn[KMG] The fixed amount of bus space which is 5189 reserved for hotplug bridge's IO window. 5190 Default size is 256 bytes. 5191 hpmmiosize=nn[KMG] The fixed amount of bus space which is 5192 reserved for hotplug bridge's MMIO window. 5193 Default size is 2 megabytes. 5194 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is 5195 reserved for hotplug bridge's MMIO_PREF window. 5196 Default size is 2 megabytes. 5197 hpmemsize=nn[KMG] The fixed amount of bus space which is 5198 reserved for hotplug bridge's MMIO and 5199 MMIO_PREF window. 5200 Default size is 2 megabytes. 5201 hpbussize=nn The minimum amount of additional bus numbers 5202 reserved for buses below a hotplug bridge. 5203 Default is 1. 5204 realloc= Enable/disable reallocating PCI bridge resources 5205 if allocations done by BIOS are too small to 5206 accommodate resources required by all child 5207 devices. 5208 off: Turn realloc off 5209 on: Turn realloc on 5210 realloc same as realloc=on 5211 noari do not use PCIe ARI. 5212 noats [PCIE, Intel-IOMMU, AMD-IOMMU] 5213 do not use PCIe ATS (and IOMMU device IOTLB). 5214 pcie_scan_all Scan all possible PCIe devices. Otherwise we 5215 only look for one device below a PCIe downstream 5216 port. 5217 big_root_window Try to add a big 64bit memory window to the PCIe 5218 root complex on AMD CPUs. Some GFX hardware 5219 can resize a BAR to allow access to all VRAM. 5220 Adding the window is slightly risky (it may 5221 conflict with unreported devices), so this 5222 taints the kernel. 5223 disable_acs_redir=<pci_dev>[; ...] 5224 Specify one or more PCI devices (in the format 5225 specified above) separated by semicolons. 5226 Each device specified will have the PCI ACS 5227 redirect capabilities forced off which will 5228 allow P2P traffic between devices through 5229 bridges without forcing it upstream. Note: 5230 this removes isolation between devices and 5231 may put more devices in an IOMMU group. 5232 config_acs= 5233 Format: 5234 <ACS flags>@<pci_dev>[; ...] 5235 Specify one or more PCI devices (in the format 5236 specified above) optionally prepended with flags 5237 and separated by semicolons. The respective 5238 capabilities will be enabled, disabled or 5239 unchanged based on what is specified in 5240 flags. 5241 5242 ACS Flags is defined as follows: 5243 bit-0 : ACS Source Validation 5244 bit-1 : ACS Translation Blocking 5245 bit-2 : ACS P2P Request Redirect 5246 bit-3 : ACS P2P Completion Redirect 5247 bit-4 : ACS Upstream Forwarding 5248 bit-5 : ACS P2P Egress Control 5249 bit-6 : ACS Direct Translated P2P 5250 Each bit can be marked as: 5251 '0' – force disabled 5252 '1' – force enabled 5253 'x' – unchanged 5254 For example, 5255 pci=config_acs=10x@pci:0:0 5256 would configure all devices that support 5257 ACS to enable P2P Request Redirect, disable 5258 Translation Blocking, and leave Source 5259 Validation unchanged from whatever power-up 5260 or firmware set it to. 5261 5262 Note: this may remove isolation between devices 5263 and may put more devices in an IOMMU group. 5264 force_floating [S390] Force usage of floating interrupts. 5265 nomio [S390] Do not use MIO instructions. 5266 norid [S390] ignore the RID field and force use of 5267 one PCI domain per PCI function 5268 notph [PCIE] If the PCIE_TPH kernel config parameter 5269 is enabled, this kernel boot option can be used 5270 to disable PCIe TLP Processing Hints support 5271 system-wide. 5272 5273 pcie_aspm= [PCIE] Forcibly enable or ignore PCIe Active State Power 5274 Management. 5275 off Don't touch ASPM configuration at all. Leave any 5276 configuration done by firmware unchanged. 5277 force Enable ASPM even on devices that claim not to support it. 5278 WARNING: Forcing ASPM on may cause system lockups. 5279 5280 pcie_ports= [PCIE] PCIe port services handling: 5281 native Use native PCIe services (PME, AER, DPC, PCIe hotplug) 5282 even if the platform doesn't give the OS permission to 5283 use them. This may cause conflicts if the platform 5284 also tries to use these services. 5285 dpc-native Use native PCIe service for DPC only. May 5286 cause conflicts if firmware uses AER or DPC. 5287 compat Disable native PCIe services (PME, AER, DPC, PCIe 5288 hotplug). 5289 5290 pcie_port_pm= [PCIE] PCIe port power management handling: 5291 off Disable power management of all PCIe ports 5292 force Forcibly enable power management of all PCIe ports 5293 5294 pcie_pme= [PCIE,PM] Native PCIe PME signaling options: 5295 nomsi Do not use MSI for native PCIe PME signaling (this makes 5296 all PCIe root ports use INTx for all services). 5297 5298 pcmv= [HW,PCMCIA] BadgePAD 4 5299 5300 pd_ignore_unused 5301 [PM] 5302 Keep all power-domains already enabled by bootloader on, 5303 even if no driver has claimed them. This is useful 5304 for debug and development, but should not be 5305 needed on a platform with proper driver support. 5306 5307 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at 5308 boot time. 5309 Format: { 0 | 1 } 5310 See arch/parisc/kernel/pdc_chassis.c 5311 5312 percpu_alloc= [MM,EARLY] 5313 Select which percpu first chunk allocator to use. 5314 Currently supported values are "embed" and "page". 5315 Archs may support subset or none of the selections. 5316 See comments in mm/percpu.c for details on each 5317 allocator. This parameter is primarily for debugging 5318 and performance comparison. 5319 5320 pirq= [SMP,APIC] Manual mp-table setup 5321 See Documentation/arch/x86/i386/IO-APIC.rst. 5322 5323 plip= [PPT,NET] Parallel port network link 5324 Format: { parport<nr> | timid | 0 } 5325 See also Documentation/admin-guide/parport.rst. 5326 5327 pmtmr= [X86] Manual setup of pmtmr I/O Port. 5328 Override pmtimer IOPort with a hex value. 5329 e.g. pmtmr=0x508 5330 5331 pmu_override= [PPC] Override the PMU. 5332 This option takes over the PMU facility, so it is no 5333 longer usable by perf. Setting this option starts the 5334 PMU counters by setting MMCR0 to 0 (the FC bit is 5335 cleared). If a number is given, then MMCR1 is set to 5336 that number, otherwise (e.g., 'pmu_override=on'), MMCR1 5337 remains 0. 5338 5339 pm_async= [PM] 5340 Format: off 5341 This parameter sets the initial value of the 5342 /sys/power/pm_async sysfs knob at boot time. 5343 If set to "off", disables asynchronous suspend and 5344 resume of devices during system-wide power transitions. 5345 This can be useful on platforms where device 5346 dependencies are not well-defined, or for debugging 5347 power management issues. Asynchronous operations are 5348 enabled by default. 5349 5350 5351 pm_debug_messages [SUSPEND,KNL] 5352 Enable suspend/resume debug messages during boot up. 5353 5354 pnp.debug=1 [PNP] 5355 Enable PNP debug messages (depends on the 5356 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time 5357 via /sys/module/pnp/parameters/debug. We always show 5358 current resource usage; turning this on also shows 5359 possible settings and some assignment information. 5360 5361 pnpacpi= [ACPI] 5362 { off } 5363 5364 pnpbios= [ISAPNP] 5365 { on | off | curr | res | no-curr | no-res } 5366 5367 pnp_reserve_irq= 5368 [ISAPNP] Exclude IRQs for the autoconfiguration 5369 5370 pnp_reserve_dma= 5371 [ISAPNP] Exclude DMAs for the autoconfiguration 5372 5373 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration 5374 Ranges are in pairs (I/O port base and size). 5375 5376 pnp_reserve_mem= 5377 [ISAPNP] Exclude memory regions for the 5378 autoconfiguration. 5379 Ranges are in pairs (memory base and size). 5380 5381 ports= [IP_VS_FTP] IPVS ftp helper module 5382 Default is 21. 5383 Up to 8 (IP_VS_APP_MAX_PORTS) ports 5384 may be specified. 5385 Format: <port>,<port>.... 5386 5387 possible_cpus= [SMP,S390,X86] 5388 Format: <unsigned int> 5389 Set the number of possible CPUs, overriding the 5390 regular discovery mechanisms (such as ACPI/FW, etc). 5391 5392 powersave=off [PPC] This option disables power saving features. 5393 It specifically disables cpuidle and sets the 5394 platform machine description specific power_save 5395 function to NULL. On Idle the CPU just reduces 5396 execution priority. 5397 5398 ppc_strict_facility_enable 5399 [PPC,ENABLE] This option catches any kernel floating point, 5400 Altivec, VSX and SPE outside of regions specifically 5401 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()). 5402 There is some performance impact when enabling this. 5403 5404 ppc_tm= [PPC,EARLY] 5405 Format: {"off"} 5406 Disable Hardware Transactional Memory 5407 5408 preempt= [KNL] 5409 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC 5410 none - Limited to cond_resched() calls 5411 voluntary - Limited to cond_resched() and might_sleep() calls 5412 full - Any section that isn't explicitly preempt disabled 5413 can be preempted anytime. Tasks will also yield 5414 contended spinlocks (if the critical section isn't 5415 explicitly preempt disabled beyond the lock itself). 5416 lazy - Scheduler controlled. Similar to full but instead 5417 of preempting the task immediately, the task gets 5418 one HZ tick time to yield itself before the 5419 preemption will be forced. One preemption is when the 5420 task returns to user space. 5421 5422 print-fatal-signals= 5423 [KNL] debug: print fatal signals 5424 5425 If enabled, warn about various signal handling 5426 related application anomalies: too many signals, 5427 too many POSIX.1 timers, fatal signals causing a 5428 coredump - etc. 5429 5430 If you hit the warning due to signal overflow, 5431 you might want to try "ulimit -i unlimited". 5432 5433 default: off. 5434 5435 printk.always_kmsg_dump= 5436 Trigger kmsg_dump for cases other than kernel oops or 5437 panics 5438 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5439 default: disabled 5440 5441 printk.console_no_auto_verbose= 5442 Disable console loglevel raise on oops, panic 5443 or lockdep-detected issues (only if lock debug is on). 5444 With an exception to setups with low baudrate on 5445 serial console, keeping this 0 is a good choice 5446 in order to provide more debug information. 5447 Format: <bool> 5448 default: 0 (auto_verbose is enabled) 5449 5450 printk.debug_non_panic_cpus= 5451 Allows storing messages from non-panic CPUs into 5452 the printk log buffer during panic(). They are 5453 flushed to consoles by the panic-CPU on 5454 a best-effort basis. 5455 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5456 Default: disabled 5457 5458 printk.devkmsg={on,off,ratelimit} 5459 Control writing to /dev/kmsg. 5460 on - unlimited logging to /dev/kmsg from userspace 5461 off - logging to /dev/kmsg disabled 5462 ratelimit - ratelimit the logging 5463 Default: ratelimit 5464 5465 printk.time= Show timing data prefixed to each printk message line 5466 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5467 5468 proc_mem.force_override= [KNL] 5469 Format: {always | ptrace | never} 5470 Traditionally /proc/pid/mem allows memory permissions to be 5471 overridden without restrictions. This option may be set to 5472 restrict that. Can be one of: 5473 - 'always': traditional behavior always allows mem overrides. 5474 - 'ptrace': only allow mem overrides for active ptracers. 5475 - 'never': never allow mem overrides. 5476 If not specified, default is the CONFIG_PROC_MEM_* choice. 5477 5478 processor.max_cstate= [HW,ACPI] 5479 Limit processor to maximum C-state 5480 max_cstate=9 overrides any DMI blacklist limit. 5481 5482 processor.nocst [HW,ACPI] 5483 Ignore the _CST method to determine C-states, 5484 instead using the legacy FADT method 5485 5486 profile= [KNL] Enable kernel profiling via /proc/profile 5487 Format: [<profiletype>,]<number> 5488 Param: <profiletype>: "schedule" or "kvm" 5489 [defaults to kernel profiling] 5490 Param: "schedule" - profile schedule points. 5491 Param: "kvm" - profile VM exits. 5492 Param: <number> - step/bucket size as a power of 2 for 5493 statistical time based profiling. 5494 5495 prot_virt= [S390] enable hosting protected virtual machines 5496 isolated from the hypervisor (if hardware supports 5497 that). If enabled, the default kernel base address 5498 might be overridden even when Kernel Address Space 5499 Layout Randomization is disabled. 5500 Format: <bool> 5501 5502 psi= [KNL] Enable or disable pressure stall information 5503 tracking. 5504 Format: <bool> 5505 5506 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to 5507 probe for; one of (bare|imps|exps|lifebook|any). 5508 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports 5509 per second. 5510 psmouse.resetafter= [HW,MOUSE] 5511 Try to reset the device after so many bad packets 5512 (0 = never). 5513 psmouse.resolution= 5514 [HW,MOUSE] Set desired mouse resolution, in dpi. 5515 psmouse.smartscroll= 5516 [HW,MOUSE] Controls Logitech smartscroll autorepeat. 5517 0 = disabled, 1 = enabled (default). 5518 5519 pstore.backend= Specify the name of the pstore backend to use 5520 5521 pti= [X86-64] Control Page Table Isolation of user and 5522 kernel address spaces. Disabling this feature 5523 removes hardening, but improves performance of 5524 system calls and interrupts. 5525 5526 on - unconditionally enable 5527 off - unconditionally disable 5528 auto - kernel detects whether your CPU model is 5529 vulnerable to issues that PTI mitigates 5530 5531 Not specifying this option is equivalent to pti=auto. 5532 5533 pty.legacy_count= 5534 [KNL] Number of legacy pty's. Overwrites compiled-in 5535 default number. 5536 5537 quiet [KNL,EARLY] Disable most log messages 5538 5539 r128= [HW,DRM] 5540 5541 radix_hcall_invalidate=on [PPC/PSERIES] 5542 Disable RADIX GTSE feature and use hcall for TLB 5543 invalidate. 5544 5545 raid= [HW,RAID] 5546 See Documentation/admin-guide/md.rst. 5547 5548 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes 5549 See Documentation/admin-guide/blockdev/ramdisk.rst. 5550 5551 ramdisk_start= [Deprecated,RAM] RAM disk image start address 5552 5553 random.trust_cpu=off 5554 [KNL,EARLY] Disable trusting the use of the CPU's 5555 random number generator (if available) to 5556 initialize the kernel's RNG. 5557 5558 random.trust_bootloader=off 5559 [KNL,EARLY] Disable trusting the use of the a seed 5560 passed by the bootloader (if available) to 5561 initialize the kernel's RNG. 5562 5563 randomize_kstack_offset= 5564 [KNL,EARLY] Enable or disable kernel stack offset 5565 randomization, which provides roughly 5 bits of 5566 entropy, frustrating memory corruption attacks 5567 that depend on stack address determinism or 5568 cross-syscall address exposures. This is only 5569 available on architectures that have defined 5570 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET. 5571 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5572 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT. 5573 5574 ras=option[,option,...] [KNL] RAS-specific options 5575 5576 cec_disable [X86] 5577 Disable the Correctable Errors Collector, 5578 see CONFIG_RAS_CEC help text. 5579 5580 rcu_nocbs[=cpu-list] 5581 [KNL] The optional argument is a cpu list, 5582 as described above. 5583 5584 In kernels built with CONFIG_RCU_NOCB_CPU=y, 5585 enable the no-callback CPU mode, which prevents 5586 such CPUs' callbacks from being invoked in 5587 softirq context. Invocation of such CPUs' RCU 5588 callbacks will instead be offloaded to "rcuox/N" 5589 kthreads created for that purpose, where "x" is 5590 "p" for RCU-preempt, "s" for RCU-sched, and "g" 5591 for the kthreads that mediate grace periods; and 5592 "N" is the CPU number. This reduces OS jitter on 5593 the offloaded CPUs, which can be useful for HPC 5594 and real-time workloads. It can also improve 5595 energy efficiency for asymmetric multiprocessors. 5596 5597 If a cpulist is passed as an argument, the specified 5598 list of CPUs is set to no-callback mode from boot. 5599 5600 Otherwise, if the '=' sign and the cpulist 5601 arguments are omitted, no CPU will be set to 5602 no-callback mode from boot but the mode may be 5603 toggled at runtime via cpusets. 5604 5605 Note that this argument takes precedence over 5606 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option. 5607 5608 rcu_nocb_poll [KNL] 5609 Rather than requiring that offloaded CPUs 5610 (specified by rcu_nocbs= above) explicitly 5611 awaken the corresponding "rcuoN" kthreads, 5612 make these kthreads poll for callbacks. 5613 This improves the real-time response for the 5614 offloaded CPUs by relieving them of the need to 5615 wake up the corresponding kthread, but degrades 5616 energy efficiency by requiring that the kthreads 5617 periodically wake up to do the polling. 5618 5619 rcutree.blimit= [KNL] 5620 Set maximum number of finished RCU callbacks to 5621 process in one batch. 5622 5623 rcutree.csd_lock_suppress_rcu_stall= [KNL] 5624 Do only a one-line RCU CPU stall warning when 5625 there is an ongoing too-long CSD-lock wait. 5626 5627 rcutree.do_rcu_barrier= [KNL] 5628 Request a call to rcu_barrier(). This is 5629 throttled so that userspace tests can safely 5630 hammer on the sysfs variable if they so choose. 5631 If triggered before the RCU grace-period machinery 5632 is fully active, this will error out with EAGAIN. 5633 5634 rcutree.dump_tree= [KNL] 5635 Dump the structure of the rcu_node combining tree 5636 out at early boot. This is used for diagnostic 5637 purposes, to verify correct tree setup. 5638 5639 rcutree.gp_cleanup_delay= [KNL] 5640 Set the number of jiffies to delay each step of 5641 RCU grace-period cleanup. 5642 5643 rcutree.gp_init_delay= [KNL] 5644 Set the number of jiffies to delay each step of 5645 RCU grace-period initialization. 5646 5647 rcutree.gp_preinit_delay= [KNL] 5648 Set the number of jiffies to delay each step of 5649 RCU grace-period pre-initialization, that is, 5650 the propagation of recent CPU-hotplug changes up 5651 the rcu_node combining tree. 5652 5653 rcutree.jiffies_till_first_fqs= [KNL] 5654 Set delay from grace-period initialization to 5655 first attempt to force quiescent states. 5656 Units are jiffies, minimum value is zero, 5657 and maximum value is HZ. 5658 5659 rcutree.jiffies_till_next_fqs= [KNL] 5660 Set delay between subsequent attempts to force 5661 quiescent states. Units are jiffies, minimum 5662 value is one, and maximum value is HZ. 5663 5664 rcutree.jiffies_till_sched_qs= [KNL] 5665 Set required age in jiffies for a 5666 given grace period before RCU starts 5667 soliciting quiescent-state help from 5668 rcu_note_context_switch() and cond_resched(). 5669 If not specified, the kernel will calculate 5670 a value based on the most recent settings 5671 of rcutree.jiffies_till_first_fqs 5672 and rcutree.jiffies_till_next_fqs. 5673 This calculated value may be viewed in 5674 rcutree.jiffies_to_sched_qs. Any attempt to set 5675 rcutree.jiffies_to_sched_qs will be cheerfully 5676 overwritten. 5677 5678 rcutree.kthread_prio= [KNL,BOOT] 5679 Set the SCHED_FIFO priority of the RCU per-CPU 5680 kthreads (rcuc/N). This value is also used for 5681 the priority of the RCU boost threads (rcub/N) 5682 and for the RCU grace-period kthreads (rcu_bh, 5683 rcu_preempt, and rcu_sched). If RCU_BOOST is 5684 set, valid values are 1-99 and the default is 1 5685 (the least-favored priority). Otherwise, when 5686 RCU_BOOST is not set, valid values are 0-99 and 5687 the default is zero (non-realtime operation). 5688 When RCU_NOCB_CPU is set, also adjust the 5689 priority of NOCB callback kthreads. 5690 5691 rcutree.nocb_nobypass_lim_per_jiffy= [KNL] 5692 On callback-offloaded (rcu_nocbs) CPUs, 5693 RCU reduces the lock contention that would 5694 otherwise be caused by callback floods through 5695 use of the ->nocb_bypass list. However, in the 5696 common non-flooded case, RCU queues directly to 5697 the main ->cblist in order to avoid the extra 5698 overhead of the ->nocb_bypass list and its lock. 5699 But if there are too many callbacks queued during 5700 a single jiffy, RCU pre-queues the callbacks into 5701 the ->nocb_bypass queue. The definition of "too 5702 many" is supplied by this kernel boot parameter. 5703 5704 rcutree.nohz_full_patience_delay= [KNL] 5705 On callback-offloaded (rcu_nocbs) CPUs, avoid 5706 disturbing RCU unless the grace period has 5707 reached the specified age in milliseconds. 5708 Defaults to zero. Large values will be capped 5709 at five seconds. All values will be rounded down 5710 to the nearest value representable by jiffies. 5711 5712 rcutree.qhimark= [KNL] 5713 Set threshold of queued RCU callbacks beyond which 5714 batch limiting is disabled. 5715 5716 rcutree.qlowmark= [KNL] 5717 Set threshold of queued RCU callbacks below which 5718 batch limiting is re-enabled. 5719 5720 rcutree.qovld= [KNL] 5721 Set threshold of queued RCU callbacks beyond which 5722 RCU's force-quiescent-state scan will aggressively 5723 enlist help from cond_resched() and sched IPIs to 5724 help CPUs more quickly reach quiescent states. 5725 Set to less than zero to make this be set based 5726 on rcutree.qhimark at boot time and to zero to 5727 disable more aggressive help enlistment. 5728 5729 rcutree.rcu_delay_page_cache_fill_msec= [KNL] 5730 Set the page-cache refill delay (in milliseconds) 5731 in response to low-memory conditions. The range 5732 of permitted values is in the range 0:100000. 5733 5734 rcutree.rcu_divisor= [KNL] 5735 Set the shift-right count to use to compute 5736 the callback-invocation batch limit bl from 5737 the number of callbacks queued on this CPU. 5738 The result will be bounded below by the value of 5739 the rcutree.blimit kernel parameter. Every bl 5740 callbacks, the softirq handler will exit in 5741 order to allow the CPU to do other work. 5742 5743 Please note that this callback-invocation batch 5744 limit applies only to non-offloaded callback 5745 invocation. Offloaded callbacks are instead 5746 invoked in the context of an rcuoc kthread, which 5747 scheduler will preempt as it does any other task. 5748 5749 rcutree.rcu_fanout_exact= [KNL] 5750 Disable autobalancing of the rcu_node combining 5751 tree. This is used by rcutorture, and might 5752 possibly be useful for architectures having high 5753 cache-to-cache transfer latencies. 5754 5755 rcutree.rcu_fanout_leaf= [KNL] 5756 Change the number of CPUs assigned to each 5757 leaf rcu_node structure. Useful for very 5758 large systems, which will choose the value 64, 5759 and for NUMA systems with large remote-access 5760 latencies, which will choose a value aligned 5761 with the appropriate hardware boundaries. 5762 5763 rcutree.rcu_min_cached_objs= [KNL] 5764 Minimum number of objects which are cached and 5765 maintained per one CPU. Object size is equal 5766 to PAGE_SIZE. The cache allows to reduce the 5767 pressure to page allocator, also it makes the 5768 whole algorithm to behave better in low memory 5769 condition. 5770 5771 rcutree.rcu_nocb_gp_stride= [KNL] 5772 Set the number of NOCB callback kthreads in 5773 each group, which defaults to the square root 5774 of the number of CPUs. Larger numbers reduce 5775 the wakeup overhead on the global grace-period 5776 kthread, but increases that same overhead on 5777 each group's NOCB grace-period kthread. 5778 5779 rcutree.rcu_kick_kthreads= [KNL] 5780 Cause the grace-period kthread to get an extra 5781 wake_up() if it sleeps three times longer than 5782 it should at force-quiescent-state time. 5783 This wake_up() will be accompanied by a 5784 WARN_ONCE() splat and an ftrace_dump(). 5785 5786 rcutree.rcu_resched_ns= [KNL] 5787 Limit the time spend invoking a batch of RCU 5788 callbacks to the specified number of nanoseconds. 5789 By default, this limit is checked only once 5790 every 32 callbacks in order to limit the pain 5791 inflicted by local_clock() overhead. 5792 5793 rcutree.rcu_unlock_delay= [KNL] 5794 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels, 5795 this specifies an rcu_read_unlock()-time delay 5796 in microseconds. This defaults to zero. 5797 Larger delays increase the probability of 5798 catching RCU pointer leaks, that is, buggy use 5799 of RCU-protected pointers after the relevant 5800 rcu_read_unlock() has completed. 5801 5802 rcutree.sysrq_rcu= [KNL] 5803 Commandeer a sysrq key to dump out Tree RCU's 5804 rcu_node tree with an eye towards determining 5805 why a new grace period has not yet started. 5806 5807 rcutree.use_softirq= [KNL] 5808 If set to zero, move all RCU_SOFTIRQ processing to 5809 per-CPU rcuc kthreads. Defaults to a non-zero 5810 value, meaning that RCU_SOFTIRQ is used by default. 5811 Specify rcutree.use_softirq=0 to use rcuc kthreads. 5812 5813 But note that CONFIG_PREEMPT_RT=y kernels disable 5814 this kernel boot parameter, forcibly setting it 5815 to zero. 5816 5817 rcutree.enable_rcu_lazy= [KNL] 5818 To save power, batch RCU callbacks and flush after 5819 delay, memory pressure or callback list growing too 5820 big. 5821 5822 rcutree.rcu_normal_wake_from_gp= [KNL] 5823 Reduces a latency of synchronize_rcu() call. This approach 5824 maintains its own track of synchronize_rcu() callers, so it 5825 does not interact with regular callbacks because it does not 5826 use a call_rcu[_hurry]() path. Please note, this is for a 5827 normal grace period. 5828 5829 How to disable it: 5830 5831 echo 0 > /sys/module/rcutree/parameters/rcu_normal_wake_from_gp 5832 or pass a boot parameter "rcutree.rcu_normal_wake_from_gp=0" 5833 5834 Default is 1 if it is not explicitly disabled by the boot parameter 5835 passing 0. 5836 5837 rcuscale.gp_async= [KNL] 5838 Measure performance of asynchronous 5839 grace-period primitives such as call_rcu(). 5840 5841 rcuscale.gp_async_max= [KNL] 5842 Specify the maximum number of outstanding 5843 callbacks per writer thread. When a writer 5844 thread exceeds this limit, it invokes the 5845 corresponding flavor of rcu_barrier() to allow 5846 previously posted callbacks to drain. 5847 5848 rcuscale.gp_exp= [KNL] 5849 Measure performance of expedited synchronous 5850 grace-period primitives. 5851 5852 rcuscale.holdoff= [KNL] 5853 Set test-start holdoff period. The purpose of 5854 this parameter is to delay the start of the 5855 test until boot completes in order to avoid 5856 interference. 5857 5858 rcuscale.kfree_by_call_rcu= [KNL] 5859 In kernels built with CONFIG_RCU_LAZY=y, test 5860 call_rcu() instead of kfree_rcu(). 5861 5862 rcuscale.kfree_mult= [KNL] 5863 Instead of allocating an object of size kfree_obj, 5864 allocate one of kfree_mult * sizeof(kfree_obj). 5865 Defaults to 1. 5866 5867 rcuscale.kfree_rcu_test= [KNL] 5868 Set to measure performance of kfree_rcu() flooding. 5869 5870 rcuscale.kfree_rcu_test_double= [KNL] 5871 Test the double-argument variant of kfree_rcu(). 5872 If this parameter has the same value as 5873 rcuscale.kfree_rcu_test_single, both the single- 5874 and double-argument variants are tested. 5875 5876 rcuscale.kfree_rcu_test_single= [KNL] 5877 Test the single-argument variant of kfree_rcu(). 5878 If this parameter has the same value as 5879 rcuscale.kfree_rcu_test_double, both the single- 5880 and double-argument variants are tested. 5881 5882 rcuscale.kfree_nthreads= [KNL] 5883 The number of threads running loops of kfree_rcu(). 5884 5885 rcuscale.kfree_alloc_num= [KNL] 5886 Number of allocations and frees done in an iteration. 5887 5888 rcuscale.kfree_loops= [KNL] 5889 Number of loops doing rcuscale.kfree_alloc_num number 5890 of allocations and frees. 5891 5892 rcuscale.minruntime= [KNL] 5893 Set the minimum test run time in seconds. This 5894 does not affect the data-collection interval, 5895 but instead allows better measurement of things 5896 like CPU consumption. 5897 5898 rcuscale.nreaders= [KNL] 5899 Set number of RCU readers. The value -1 selects 5900 N, where N is the number of CPUs. A value 5901 "n" less than -1 selects N-n+1, where N is again 5902 the number of CPUs. For example, -2 selects N 5903 (the number of CPUs), -3 selects N+1, and so on. 5904 A value of "n" less than or equal to -N selects 5905 a single reader. 5906 5907 rcuscale.nwriters= [KNL] 5908 Set number of RCU writers. The values operate 5909 the same as for rcuscale.nreaders. 5910 N, where N is the number of CPUs 5911 5912 rcuscale.scale_type= [KNL] 5913 Specify the RCU implementation to test. 5914 5915 rcuscale.shutdown= [KNL] 5916 Shut the system down after performance tests 5917 complete. This is useful for hands-off automated 5918 testing. 5919 5920 rcuscale.verbose= [KNL] 5921 Enable additional printk() statements. 5922 5923 rcuscale.writer_holdoff= [KNL] 5924 Write-side holdoff between grace periods, 5925 in microseconds. The default of zero says 5926 no holdoff. 5927 5928 rcuscale.writer_holdoff_jiffies= [KNL] 5929 Additional write-side holdoff between grace 5930 periods, but in jiffies. The default of zero 5931 says no holdoff. 5932 5933 rcutorture.fqs_duration= [KNL] 5934 Set duration of force_quiescent_state bursts 5935 in microseconds. 5936 5937 rcutorture.fqs_holdoff= [KNL] 5938 Set holdoff time within force_quiescent_state bursts 5939 in microseconds. 5940 5941 rcutorture.fqs_stutter= [KNL] 5942 Set wait time between force_quiescent_state bursts 5943 in seconds. 5944 5945 rcutorture.fwd_progress= [KNL] 5946 Specifies the number of kthreads to be used 5947 for RCU grace-period forward-progress testing 5948 for the types of RCU supporting this notion. 5949 Defaults to 1 kthread, values less than zero or 5950 greater than the number of CPUs cause the number 5951 of CPUs to be used. 5952 5953 rcutorture.fwd_progress_div= [KNL] 5954 Specify the fraction of a CPU-stall-warning 5955 period to do tight-loop forward-progress testing. 5956 5957 rcutorture.fwd_progress_holdoff= [KNL] 5958 Number of seconds to wait between successive 5959 forward-progress tests. 5960 5961 rcutorture.fwd_progress_need_resched= [KNL] 5962 Enclose cond_resched() calls within checks for 5963 need_resched() during tight-loop forward-progress 5964 testing. 5965 5966 rcutorture.gp_cond= [KNL] 5967 Use conditional/asynchronous update-side 5968 normal-grace-period primitives, if available. 5969 5970 rcutorture.gp_cond_exp= [KNL] 5971 Use conditional/asynchronous update-side 5972 expedited-grace-period primitives, if available. 5973 5974 rcutorture.gp_cond_full= [KNL] 5975 Use conditional/asynchronous update-side 5976 normal-grace-period primitives that also take 5977 concurrent expedited grace periods into account, 5978 if available. 5979 5980 rcutorture.gp_cond_exp_full= [KNL] 5981 Use conditional/asynchronous update-side 5982 expedited-grace-period primitives that also take 5983 concurrent normal grace periods into account, 5984 if available. 5985 5986 rcutorture.gp_cond_wi= [KNL] 5987 Nominal wait interval for normal conditional 5988 grace periods (specified by rcutorture's 5989 gp_cond and gp_cond_full module parameters), 5990 in microseconds. The actual wait interval will 5991 be randomly selected to nanosecond granularity up 5992 to this wait interval. Defaults to 16 jiffies, 5993 for example, 16,000 microseconds on a system 5994 with HZ=1000. 5995 5996 rcutorture.gp_cond_wi_exp= [KNL] 5997 Nominal wait interval for expedited conditional 5998 grace periods (specified by rcutorture's 5999 gp_cond_exp and gp_cond_exp_full module 6000 parameters), in microseconds. The actual wait 6001 interval will be randomly selected to nanosecond 6002 granularity up to this wait interval. Defaults to 6003 128 microseconds. 6004 6005 rcutorture.gp_exp= [KNL] 6006 Use expedited update-side primitives, if available. 6007 6008 rcutorture.gp_normal= [KNL] 6009 Use normal (non-expedited) asynchronous 6010 update-side primitives, if available. 6011 6012 rcutorture.gp_poll= [KNL] 6013 Use polled update-side normal-grace-period 6014 primitives, if available. 6015 6016 rcutorture.gp_poll_exp= [KNL] 6017 Use polled update-side expedited-grace-period 6018 primitives, if available. 6019 6020 rcutorture.gp_poll_full= [KNL] 6021 Use polled update-side normal-grace-period 6022 primitives that also take concurrent expedited 6023 grace periods into account, if available. 6024 6025 rcutorture.gp_poll_exp_full= [KNL] 6026 Use polled update-side expedited-grace-period 6027 primitives that also take concurrent normal 6028 grace periods into account, if available. 6029 6030 rcutorture.gp_poll_wi= [KNL] 6031 Nominal wait interval for normal conditional 6032 grace periods (specified by rcutorture's 6033 gp_poll and gp_poll_full module parameters), 6034 in microseconds. The actual wait interval will 6035 be randomly selected to nanosecond granularity up 6036 to this wait interval. Defaults to 16 jiffies, 6037 for example, 16,000 microseconds on a system 6038 with HZ=1000. 6039 6040 rcutorture.gp_poll_wi_exp= [KNL] 6041 Nominal wait interval for expedited conditional 6042 grace periods (specified by rcutorture's 6043 gp_poll_exp and gp_poll_exp_full module 6044 parameters), in microseconds. The actual wait 6045 interval will be randomly selected to nanosecond 6046 granularity up to this wait interval. Defaults to 6047 128 microseconds. 6048 6049 rcutorture.gp_sync= [KNL] 6050 Use normal (non-expedited) synchronous 6051 update-side primitives, if available. If all 6052 of rcutorture.gp_cond=, rcutorture.gp_exp=, 6053 rcutorture.gp_normal=, and rcutorture.gp_sync= 6054 are zero, rcutorture acts as if is interpreted 6055 they are all non-zero. 6056 6057 rcutorture.gpwrap_lag= [KNL] 6058 Enable grace-period wrap lag testing. Setting 6059 to false prevents the gpwrap lag test from 6060 running. Default is true. 6061 6062 rcutorture.gpwrap_lag_gps= [KNL] 6063 Set the value for grace-period wrap lag during 6064 active lag testing periods. This controls how many 6065 grace periods differences we tolerate between 6066 rdp and rnp's gp_seq before setting overflow flag. 6067 The default is always set to 8. 6068 6069 rcutorture.gpwrap_lag_cycle_mins= [KNL] 6070 Set the total cycle duration for gpwrap lag 6071 testing in minutes. This is the total time for 6072 one complete cycle of active and inactive 6073 testing periods. Default is 30 minutes. 6074 6075 rcutorture.gpwrap_lag_active_mins= [KNL] 6076 Set the duration for which gpwrap lag is active 6077 within each cycle, in minutes. During this time, 6078 the grace-period wrap lag will be set to the 6079 value specified by gpwrap_lag_gps. Default is 6080 5 minutes. 6081 6082 rcutorture.irqreader= [KNL] 6083 Run RCU readers from irq handlers, or, more 6084 accurately, from a timer handler. Not all RCU 6085 flavors take kindly to this sort of thing. 6086 6087 rcutorture.leakpointer= [KNL] 6088 Leak an RCU-protected pointer out of the reader. 6089 This can of course result in splats, and is 6090 intended to test the ability of things like 6091 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect 6092 such leaks. 6093 6094 rcutorture.n_barrier_cbs= [KNL] 6095 Set callbacks/threads for rcu_barrier() testing. 6096 6097 rcutorture.nfakewriters= [KNL] 6098 Set number of concurrent RCU writers. These just 6099 stress RCU, they don't participate in the actual 6100 test, hence the "fake". 6101 6102 rcutorture.nocbs_nthreads= [KNL] 6103 Set number of RCU callback-offload togglers. 6104 Zero (the default) disables toggling. 6105 6106 rcutorture.nocbs_toggle= [KNL] 6107 Set the delay in milliseconds between successive 6108 callback-offload toggling attempts. 6109 6110 rcutorture.nreaders= [KNL] 6111 Set number of RCU readers. The value -1 selects 6112 N-1, where N is the number of CPUs. A value 6113 "n" less than -1 selects N-n-2, where N is again 6114 the number of CPUs. For example, -2 selects N 6115 (the number of CPUs), -3 selects N+1, and so on. 6116 6117 rcutorture.object_debug= [KNL] 6118 Enable debug-object double-call_rcu() testing. 6119 6120 rcutorture.onoff_holdoff= [KNL] 6121 Set time (s) after boot for CPU-hotplug testing. 6122 6123 rcutorture.onoff_interval= [KNL] 6124 Set time (jiffies) between CPU-hotplug operations, 6125 or zero to disable CPU-hotplug testing. 6126 6127 rcutorture.preempt_duration= [KNL] 6128 Set duration (in milliseconds) of preemptions 6129 by a high-priority FIFO real-time task. Set to 6130 zero (the default) to disable. The CPUs to 6131 preempt are selected randomly from the set that 6132 are online at a given point in time. Races with 6133 CPUs going offline are ignored, with that attempt 6134 at preemption skipped. 6135 6136 rcutorture.preempt_interval= [KNL] 6137 Set interval (in milliseconds, defaulting to one 6138 second) between preemptions by a high-priority 6139 FIFO real-time task. This delay is mediated 6140 by an hrtimer and is further fuzzed to avoid 6141 inadvertent synchronizations. 6142 6143 rcutorture.read_exit_burst= [KNL] 6144 The number of times in a given read-then-exit 6145 episode that a set of read-then-exit kthreads 6146 is spawned. 6147 6148 rcutorture.read_exit_delay= [KNL] 6149 The delay, in seconds, between successive 6150 read-then-exit testing episodes. 6151 6152 rcutorture.reader_flavor= [KNL] 6153 A bit mask indicating which readers to use. 6154 If there is more than one bit set, the readers 6155 are entered from low-order bit up, and are 6156 exited in the opposite order. For SRCU, the 6157 0x1 bit is normal readers, 0x2 NMI-safe readers, 6158 and 0x4 light-weight readers. 6159 6160 rcutorture.shuffle_interval= [KNL] 6161 Set task-shuffle interval (s). Shuffling tasks 6162 allows some CPUs to go into dyntick-idle mode 6163 during the rcutorture test. 6164 6165 rcutorture.shutdown_secs= [KNL] 6166 Set time (s) after boot system shutdown. This 6167 is useful for hands-off automated testing. 6168 6169 rcutorture.stall_cpu= [KNL] 6170 Duration of CPU stall (s) to test RCU CPU stall 6171 warnings, zero to disable. 6172 6173 rcutorture.stall_cpu_block= [KNL] 6174 Sleep while stalling if set. This will result 6175 in warnings from preemptible RCU in addition to 6176 any other stall-related activity. Note that 6177 in kernels built with CONFIG_PREEMPTION=n and 6178 CONFIG_PREEMPT_COUNT=y, this parameter will 6179 cause the CPU to pass through a quiescent state. 6180 Given CONFIG_PREEMPTION=n, this will suppress 6181 RCU CPU stall warnings, but will instead result 6182 in scheduling-while-atomic splats. 6183 6184 Use of this module parameter results in splats. 6185 6186 6187 rcutorture.stall_cpu_holdoff= [KNL] 6188 Time to wait (s) after boot before inducing stall. 6189 6190 rcutorture.stall_cpu_irqsoff= [KNL] 6191 Disable interrupts while stalling if set, but only 6192 on the first stall in the set. 6193 6194 rcutorture.stall_cpu_repeat= [KNL] 6195 Number of times to repeat the stall sequence, 6196 so that rcutorture.stall_cpu_repeat=3 will result 6197 in four stall sequences. 6198 6199 rcutorture.stall_gp_kthread= [KNL] 6200 Duration (s) of forced sleep within RCU 6201 grace-period kthread to test RCU CPU stall 6202 warnings, zero to disable. If both stall_cpu 6203 and stall_gp_kthread are specified, the 6204 kthread is starved first, then the CPU. 6205 6206 rcutorture.stat_interval= [KNL] 6207 Time (s) between statistics printk()s. 6208 6209 rcutorture.stutter= [KNL] 6210 Time (s) to stutter testing, for example, specifying 6211 five seconds causes the test to run for five seconds, 6212 wait for five seconds, and so on. This tests RCU's 6213 ability to transition abruptly to and from idle. 6214 6215 rcutorture.test_boost= [KNL] 6216 Test RCU priority boosting? 0=no, 1=maybe, 2=yes. 6217 "Maybe" means test if the RCU implementation 6218 under test support RCU priority boosting. 6219 6220 rcutorture.test_boost_duration= [KNL] 6221 Duration (s) of each individual boost test. 6222 6223 rcutorture.test_boost_holdoff= [KNL] 6224 Holdoff time (s) from start of test to the start 6225 of RCU priority-boost testing. Defaults to zero, 6226 that is, no holdoff. 6227 6228 rcutorture.test_boost_interval= [KNL] 6229 Interval (s) between each boost test. 6230 6231 rcutorture.test_no_idle_hz= [KNL] 6232 Test RCU's dyntick-idle handling. See also the 6233 rcutorture.shuffle_interval parameter. 6234 6235 rcutorture.torture_type= [KNL] 6236 Specify the RCU implementation to test. 6237 6238 rcutorture.verbose= [KNL] 6239 Enable additional printk() statements. 6240 6241 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL] 6242 Dump ftrace buffer after reporting RCU CPU 6243 stall warning. 6244 6245 rcupdate.rcu_cpu_stall_notifiers= [KNL] 6246 Provide RCU CPU stall notifiers, but see the 6247 warnings in the RCU_CPU_STALL_NOTIFIER Kconfig 6248 option's help text. TL;DR: You almost certainly 6249 do not want rcupdate.rcu_cpu_stall_notifiers. 6250 6251 rcupdate.rcu_cpu_stall_suppress= [KNL] 6252 Suppress RCU CPU stall warning messages. 6253 6254 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL] 6255 Suppress RCU CPU stall warning messages and 6256 rcutorture writer stall warnings that occur 6257 during early boot, that is, during the time 6258 before the init task is spawned. 6259 6260 rcupdate.rcu_cpu_stall_timeout= [KNL] 6261 Set timeout for RCU CPU stall warning messages. 6262 The value is in seconds and the maximum allowed 6263 value is 300 seconds. 6264 6265 rcupdate.rcu_exp_cpu_stall_timeout= [KNL] 6266 Set timeout for expedited RCU CPU stall warning 6267 messages. The value is in milliseconds 6268 and the maximum allowed value is 21000 6269 milliseconds. Please note that this value is 6270 adjusted to an arch timer tick resolution. 6271 Setting this to zero causes the value from 6272 rcupdate.rcu_cpu_stall_timeout to be used (after 6273 conversion from seconds to milliseconds). 6274 6275 rcupdate.rcu_cpu_stall_cputime= [KNL] 6276 Provide statistics on the cputime and count of 6277 interrupts and tasks during the sampling period. For 6278 multiple continuous RCU stalls, all sampling periods 6279 begin at half of the first RCU stall timeout. 6280 6281 rcupdate.rcu_exp_stall_task_details= [KNL] 6282 Print stack dumps of any tasks blocking the 6283 current expedited RCU grace period during an 6284 expedited RCU CPU stall warning. 6285 6286 rcupdate.rcu_expedited= [KNL] 6287 Use expedited grace-period primitives, for 6288 example, synchronize_rcu_expedited() instead 6289 of synchronize_rcu(). This reduces latency, 6290 but can increase CPU utilization, degrade 6291 real-time latency, and degrade energy efficiency. 6292 No effect on CONFIG_TINY_RCU kernels. 6293 6294 rcupdate.rcu_normal= [KNL] 6295 Use only normal grace-period primitives, 6296 for example, synchronize_rcu() instead of 6297 synchronize_rcu_expedited(). This improves 6298 real-time latency, CPU utilization, and 6299 energy efficiency, but can expose users to 6300 increased grace-period latency. This parameter 6301 overrides rcupdate.rcu_expedited. No effect on 6302 CONFIG_TINY_RCU kernels. 6303 6304 rcupdate.rcu_normal_after_boot= [KNL] 6305 Once boot has completed (that is, after 6306 rcu_end_inkernel_boot() has been invoked), use 6307 only normal grace-period primitives. No effect 6308 on CONFIG_TINY_RCU kernels. 6309 6310 But note that CONFIG_PREEMPT_RT=y kernels enables 6311 this kernel boot parameter, forcibly setting 6312 it to the value one, that is, converting any 6313 post-boot attempt at an expedited RCU grace 6314 period to instead use normal non-expedited 6315 grace-period processing. 6316 6317 rcupdate.rcu_task_collapse_lim= [KNL] 6318 Set the maximum number of callbacks present 6319 at the beginning of a grace period that allows 6320 the RCU Tasks flavors to collapse back to using 6321 a single callback queue. This switching only 6322 occurs when rcupdate.rcu_task_enqueue_lim is 6323 set to the default value of -1. 6324 6325 rcupdate.rcu_task_contend_lim= [KNL] 6326 Set the minimum number of callback-queuing-time 6327 lock-contention events per jiffy required to 6328 cause the RCU Tasks flavors to switch to per-CPU 6329 callback queuing. This switching only occurs 6330 when rcupdate.rcu_task_enqueue_lim is set to 6331 the default value of -1. 6332 6333 rcupdate.rcu_task_enqueue_lim= [KNL] 6334 Set the number of callback queues to use for the 6335 RCU Tasks family of RCU flavors. The default 6336 of -1 allows this to be automatically (and 6337 dynamically) adjusted. This parameter is intended 6338 for use in testing. 6339 6340 rcupdate.rcu_task_lazy_lim= [KNL] 6341 Number of callbacks on a given CPU that will 6342 cancel laziness on that CPU. Use -1 to disable 6343 cancellation of laziness, but be advised that 6344 doing so increases the danger of OOM due to 6345 callback flooding. 6346 6347 rcupdate.rcu_task_stall_info= [KNL] 6348 Set initial timeout in jiffies for RCU task stall 6349 informational messages, which give some indication 6350 of the problem for those not patient enough to 6351 wait for ten minutes. Informational messages are 6352 only printed prior to the stall-warning message 6353 for a given grace period. Disable with a value 6354 less than or equal to zero. Defaults to ten 6355 seconds. A change in value does not take effect 6356 until the beginning of the next grace period. 6357 6358 rcupdate.rcu_task_stall_info_mult= [KNL] 6359 Multiplier for time interval between successive 6360 RCU task stall informational messages for a given 6361 RCU tasks grace period. This value is clamped 6362 to one through ten, inclusive. It defaults to 6363 the value three, so that the first informational 6364 message is printed 10 seconds into the grace 6365 period, the second at 40 seconds, the third at 6366 160 seconds, and then the stall warning at 600 6367 seconds would prevent a fourth at 640 seconds. 6368 6369 rcupdate.rcu_task_stall_timeout= [KNL] 6370 Set timeout in jiffies for RCU task stall 6371 warning messages. Disable with a value less 6372 than or equal to zero. Defaults to ten minutes. 6373 A change in value does not take effect until 6374 the beginning of the next grace period. 6375 6376 rcupdate.rcu_tasks_lazy_ms= [KNL] 6377 Set timeout in milliseconds RCU Tasks asynchronous 6378 callback batching for call_rcu_tasks(). 6379 A negative value will take the default. A value 6380 of zero will disable batching. Batching is 6381 always disabled for synchronize_rcu_tasks(). 6382 6383 rcupdate.rcu_self_test= [KNL] 6384 Run the RCU early boot self tests 6385 6386 rdinit= [KNL] 6387 Format: <full_path> 6388 Run specified binary instead of /init from the ramdisk, 6389 used for early userspace startup. See initrd. 6390 6391 rdrand= [X86,EARLY] 6392 force - Override the decision by the kernel to hide the 6393 advertisement of RDRAND support (this affects 6394 certain AMD processors because of buggy BIOS 6395 support, specifically around the suspend/resume 6396 path). 6397 6398 rdt= [HW,X86,RDT] 6399 Turn on/off individual RDT features. List is: 6400 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp, 6401 mba, smba, bmec, abmc, sdciae, energy[:guid], 6402 perf[:guid]. 6403 E.g. to turn on cmt and turn off mba use: 6404 rdt=cmt,!mba 6405 To turn off all energy telemetry monitoring and ensure that 6406 perf telemetry monitoring associated with guid 0x12345 6407 is enabled use: 6408 rdt=!energy,perf:0x12345 6409 6410 reboot= [KNL] 6411 Format (x86 or x86_64): 6412 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] | d[efault] \ 6413 [[,]s[mp]#### \ 6414 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \ 6415 [[,]f[orce] 6416 Where reboot_mode is one of warm (soft) or cold (hard) or gpio 6417 (prefix with 'panic_' to set mode for panic 6418 reboot only), 6419 reboot_type is one of bios, acpi, kbd, triple, efi, or pci, 6420 reboot_force is either force or not specified, 6421 reboot_cpu is s[mp]#### with #### being the processor 6422 to be used for rebooting. 6423 6424 acpi 6425 Use the ACPI RESET_REG in the FADT. If ACPI is not 6426 configured or the ACPI reset does not work, the reboot 6427 path attempts the reset using the keyboard controller. 6428 6429 bios 6430 Use the CPU reboot vector for warm reset 6431 6432 cold 6433 Set the cold reboot flag 6434 6435 default 6436 There are some built-in platform specific "quirks" 6437 - you may see: "reboot: <name> series board detected. 6438 Selecting <type> for reboots." In the case where you 6439 think the quirk is in error (e.g. you have newer BIOS, 6440 or newer board) using this option will ignore the 6441 built-in quirk table, and use the generic default 6442 reboot actions. 6443 6444 efi 6445 Use efi reset_system runtime service. If EFI is not 6446 configured or the EFI reset does not work, the reboot 6447 path attempts the reset using the keyboard controller. 6448 6449 force 6450 Don't stop other CPUs on reboot. This can make reboot 6451 more reliable in some cases. 6452 6453 kbd 6454 Use the keyboard controller. cold reset (default) 6455 6456 pci 6457 Use a write to the PCI config space register 0xcf9 to 6458 trigger reboot. 6459 6460 triple 6461 Force a triple fault (init) 6462 6463 warm 6464 Don't set the cold reboot flag 6465 6466 Using warm reset will be much faster especially on big 6467 memory systems because the BIOS will not go through 6468 the memory check. Disadvantage is that not all 6469 hardware will be completely reinitialized on reboot so 6470 there may be boot problems on some systems. 6471 6472 6473 refscale.holdoff= [KNL] 6474 Set test-start holdoff period. The purpose of 6475 this parameter is to delay the start of the 6476 test until boot completes in order to avoid 6477 interference. 6478 6479 refscale.lookup_instances= [KNL] 6480 Number of data elements to use for the forms of 6481 SLAB_TYPESAFE_BY_RCU testing. A negative number 6482 is negated and multiplied by nr_cpu_ids, while 6483 zero specifies nr_cpu_ids. 6484 6485 refscale.loops= [KNL] 6486 Set the number of loops over the synchronization 6487 primitive under test. Increasing this number 6488 reduces noise due to loop start/end overhead, 6489 but the default has already reduced the per-pass 6490 noise to a handful of picoseconds on ca. 2020 6491 x86 laptops. 6492 6493 refscale.nreaders= [KNL] 6494 Set number of readers. The default value of -1 6495 selects N, where N is roughly 75% of the number 6496 of CPUs. A value of zero is an interesting choice. 6497 6498 refscale.nruns= [KNL] 6499 Set number of runs, each of which is dumped onto 6500 the console log. 6501 6502 refscale.readdelay= [KNL] 6503 Set the read-side critical-section duration, 6504 measured in microseconds. 6505 6506 refscale.scale_type= [KNL] 6507 Specify the read-protection implementation to test. 6508 6509 refscale.shutdown= [KNL] 6510 Shut down the system at the end of the performance 6511 test. This defaults to 1 (shut it down) when 6512 refscale is built into the kernel and to 0 (leave 6513 it running) when refscale is built as a module. 6514 6515 refscale.verbose= [KNL] 6516 Enable additional printk() statements. 6517 6518 refscale.verbose_batched= [KNL] 6519 Batch the additional printk() statements. If zero 6520 (the default) or negative, print everything. Otherwise, 6521 print every Nth verbose statement, where N is the value 6522 specified. 6523 6524 regulator_ignore_unused 6525 [REGULATOR] 6526 Prevents regulator framework from disabling regulators 6527 that are unused, due no driver claiming them. This may 6528 be useful for debug and development, but should not be 6529 needed on a platform with proper driver support. 6530 6531 relax_domain_level= 6532 [KNL, SMP] Set scheduler's default relax_domain_level. 6533 See Documentation/admin-guide/cgroup-v1/cpusets.rst. 6534 6535 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory 6536 Format: <base1>,<size1>[,<base2>,<size2>,...] 6537 Reserve I/O ports or memory so the kernel won't use 6538 them. If <base> is less than 0x10000, the region 6539 is assumed to be I/O ports; otherwise it is memory. 6540 6541 reserve_mem= [RAM] 6542 Format: nn[KMG]:<align>:<label> 6543 Reserve physical memory and label it with a name that 6544 other subsystems can use to access it. This is typically 6545 used for systems that do not wipe the RAM, and this command 6546 line will try to reserve the same physical memory on 6547 soft reboots. Note, it is not guaranteed to be the same 6548 location. For example, if anything about the system changes 6549 or if booting a different kernel. It can also fail if KASLR 6550 places the kernel at the location of where the RAM reservation 6551 was from a previous boot, the new reservation will be at a 6552 different location. 6553 Any subsystem using this feature must add a way to verify 6554 that the contents of the physical memory is from a previous 6555 boot, as there may be cases where the memory will not be 6556 located at the same location. 6557 6558 The format is size:align:label for example, to request 6559 12 megabytes of 4096 alignment for ramoops: 6560 6561 reserve_mem=12M:4096:oops ramoops.mem_name=oops 6562 6563 reservetop= [X86-32,EARLY] 6564 Format: nn[KMG] 6565 Reserves a hole at the top of the kernel virtual 6566 address space. 6567 6568 reset_devices [KNL] Force drivers to reset the underlying device 6569 during initialization. 6570 6571 resume= [SWSUSP] 6572 Specify the partition device for software suspend 6573 Format: 6574 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>} 6575 6576 resume_offset= [SWSUSP] 6577 Specify the offset from the beginning of the partition 6578 given by "resume=" at which the swap header is located, 6579 in <PAGE_SIZE> units (needed only for swap files). 6580 See Documentation/power/swsusp-and-swap-files.rst 6581 6582 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to 6583 read the resume files 6584 6585 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up. 6586 Useful for devices that are detected asynchronously 6587 (e.g. USB and MMC devices). 6588 6589 retain_initrd [RAM] Keep initrd memory after extraction. After boot, it will 6590 be accessible via /sys/firmware/initrd. 6591 6592 retbleed= [X86] Control mitigation of RETBleed (Arbitrary 6593 Speculative Code Execution with Return Instructions) 6594 vulnerability. 6595 6596 AMD-based UNRET and IBPB mitigations alone do not stop 6597 sibling threads from influencing the predictions of other 6598 sibling threads. For that reason, STIBP is used on pro- 6599 cessors that support it, and mitigate SMT on processors 6600 that don't. 6601 6602 off - no mitigation 6603 auto - automatically select a mitigation 6604 auto,nosmt - automatically select a mitigation, 6605 disabling SMT if necessary for 6606 the full mitigation (only on Zen1 6607 and older without STIBP). 6608 ibpb - On AMD, mitigate short speculation 6609 windows on basic block boundaries too. 6610 Safe, highest perf impact. It also 6611 enables STIBP if present. Not suitable 6612 on Intel. 6613 ibpb,nosmt - Like "ibpb" above but will disable SMT 6614 when STIBP is not available. This is 6615 the alternative for systems which do not 6616 have STIBP. 6617 unret - Force enable untrained return thunks, 6618 only effective on AMD f15h-f17h based 6619 systems. 6620 unret,nosmt - Like unret, but will disable SMT when STIBP 6621 is not available. This is the alternative for 6622 systems which do not have STIBP. 6623 6624 Selecting 'auto' will choose a mitigation method at run 6625 time according to the CPU. 6626 6627 Not specifying this option is equivalent to retbleed=auto. 6628 6629 rfkill.default_state= 6630 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm, 6631 etc. communication is blocked by default. 6632 1 Unblocked. 6633 6634 rfkill.master_switch_mode= 6635 0 The "airplane mode" button does nothing. 6636 1 The "airplane mode" button toggles between everything 6637 blocked and the previous configuration. 6638 2 The "airplane mode" button toggles between everything 6639 blocked and everything unblocked. 6640 6641 ring3mwait=disable 6642 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported 6643 CPUs. 6644 6645 riscv_isa_fallback [RISCV,EARLY] 6646 When CONFIG_RISCV_ISA_FALLBACK is not enabled, permit 6647 falling back to detecting extension support by parsing 6648 "riscv,isa" property on devicetree systems when the 6649 replacement properties are not found. See the Kconfig 6650 entry for RISCV_ISA_FALLBACK. 6651 6652 riscv_nousercfi= 6653 all Disable user CFI ABI to userspace even if cpu extension 6654 are available. 6655 bcfi Disable user backward CFI ABI to userspace even if 6656 the shadow stack extension is available. 6657 fcfi Disable user forward CFI ABI to userspace even if the 6658 landing pad extension is available. 6659 6660 ro [KNL] Mount root device read-only on boot 6661 6662 rodata= [KNL,EARLY] 6663 on Mark read-only kernel memory as read-only (default). 6664 off Leave read-only kernel memory writable for debugging. 6665 noalias Mark read-only kernel memory as read-only but retain 6666 writable aliases in the direct map for regions outside 6667 of the kernel image. [arm64] 6668 6669 rockchip.usb_uart 6670 [EARLY] 6671 Enable the uart passthrough on the designated usb port 6672 on Rockchip SoCs. When active, the signals of the 6673 debug-uart get routed to the D+ and D- pins of the usb 6674 port and the regular usb controller gets disabled. 6675 6676 root= [KNL] Root filesystem 6677 Usually this is a block device specifier of some kind, 6678 see the early_lookup_bdev comment in 6679 block/early-lookup.c for details. 6680 Alternatively this can be "ram" for the legacy initial 6681 ramdisk, "nfs" and "cifs" for root on a network file 6682 system, or "mtd" and "ubi" for mounting from raw flash. 6683 6684 rootdelay= [KNL] Delay (in seconds) to pause before attempting to 6685 mount the root filesystem 6686 6687 rootflags= [KNL] Set root filesystem mount option string 6688 6689 rseq_slice_ext= [KNL] RSEQ based time slice extension 6690 Format: boolean 6691 Control enablement of RSEQ based time slice extension. 6692 Default is 'on'. 6693 6694 initramfs_options= [KNL] 6695 Specify mount options for the initramfs mount. 6696 6697 rootfstype= [KNL] Set root filesystem type 6698 6699 rootwait [KNL] Wait (indefinitely) for root device to show up. 6700 Useful for devices that are detected asynchronously 6701 (e.g. USB and MMC devices). 6702 6703 rootwait= [KNL] Maximum time (in seconds) to wait for root device 6704 to show up before attempting to mount the root 6705 filesystem. 6706 6707 rproc_mem=nn[KMG][@address] 6708 [KNL,ARM,CMA] Remoteproc physical memory block. 6709 Memory area to be used by remote processor image, 6710 managed by CMA. 6711 6712 rseq_debug= [KNL] Enable or disable restartable sequence 6713 debug mode. Defaults to CONFIG_RSEQ_DEBUG_DEFAULT_ENABLE. 6714 Format: <bool> 6715 6716 rt_group_sched= [KNL] Enable or disable SCHED_RR/FIFO group scheduling 6717 when CONFIG_RT_GROUP_SCHED=y. Defaults to 6718 !CONFIG_RT_GROUP_SCHED_DEFAULT_DISABLED. 6719 Format: <bool> 6720 6721 rw [KNL] Mount root device read-write on boot 6722 6723 S [KNL] Run init in single mode 6724 6725 s390_iommu= [HW,S390] 6726 Set s390 IOTLB flushing mode 6727 strict 6728 With strict flushing every unmap operation will result 6729 in an IOTLB flush. Default is lazy flushing before 6730 reuse, which is faster. Deprecated, equivalent to 6731 iommu.strict=1. 6732 6733 s390_iommu_aperture= [KNL,S390] 6734 Specifies the size of the per device DMA address space 6735 accessible through the DMA and IOMMU APIs as a decimal 6736 factor of the size of main memory. 6737 The default is 1 meaning that one can concurrently use 6738 as many DMA addresses as physical memory is installed, 6739 if supported by hardware, and thus map all of memory 6740 once. With a value of 2 one can map all of memory twice 6741 and so on. As a special case a factor of 0 imposes no 6742 restrictions other than those given by hardware at the 6743 cost of significant additional memory use for tables. 6744 6745 6746 sched_proxy_exec= [KNL] 6747 Enables or disables "proxy execution" style 6748 solution to mutex-based priority inversion. 6749 Format: <bool> 6750 6751 sched_verbose [KNL,EARLY] Enables verbose scheduler debug messages. 6752 6753 schedstats= [KNL,X86] Enable or disable scheduled statistics. 6754 Allowed values are enable and disable. This feature 6755 incurs a small amount of overhead in the scheduler 6756 but is useful for debugging and performance tuning. 6757 6758 sched_thermal_decay_shift= 6759 [Deprecated] 6760 [KNL, SMP] Set a decay shift for scheduler thermal 6761 pressure signal. Thermal pressure signal follows the 6762 default decay period of other scheduler pelt 6763 signals(usually 32 ms but configurable). Setting 6764 sched_thermal_decay_shift will left shift the decay 6765 period for the thermal pressure signal by the shift 6766 value. 6767 i.e. with the default pelt decay period of 32 ms 6768 sched_thermal_decay_shift thermal pressure decay pr 6769 1 64 ms 6770 2 128 ms 6771 and so on. 6772 Format: integer between 0 and 10 6773 Default is 0. 6774 6775 scftorture.holdoff= [KNL] 6776 Number of seconds to hold off before starting 6777 test. Defaults to zero for module insertion and 6778 to 10 seconds for built-in smp_call_function() 6779 tests. 6780 6781 scftorture.longwait= [KNL] 6782 Request ridiculously long waits randomly selected 6783 up to the chosen limit in seconds. Zero (the 6784 default) disables this feature. Please note 6785 that requesting even small non-zero numbers of 6786 seconds can result in RCU CPU stall warnings, 6787 softlockup complaints, and so on. 6788 6789 scftorture.nthreads= [KNL] 6790 Number of kthreads to spawn to invoke the 6791 smp_call_function() family of functions. 6792 The default of -1 specifies a number of kthreads 6793 equal to the number of CPUs. 6794 6795 scftorture.onoff_holdoff= [KNL] 6796 Number seconds to wait after the start of the 6797 test before initiating CPU-hotplug operations. 6798 6799 scftorture.onoff_interval= [KNL] 6800 Number seconds to wait between successive 6801 CPU-hotplug operations. Specifying zero (which 6802 is the default) disables CPU-hotplug operations. 6803 6804 scftorture.shutdown_secs= [KNL] 6805 The number of seconds following the start of the 6806 test after which to shut down the system. The 6807 default of zero avoids shutting down the system. 6808 Non-zero values are useful for automated tests. 6809 6810 scftorture.stat_interval= [KNL] 6811 The number of seconds between outputting the 6812 current test statistics to the console. A value 6813 of zero disables statistics output. 6814 6815 scftorture.stutter_cpus= [KNL] 6816 The number of jiffies to wait between each change 6817 to the set of CPUs under test. 6818 6819 scftorture.use_cpus_read_lock= [KNL] 6820 Use use_cpus_read_lock() instead of the default 6821 preempt_disable() to disable CPU hotplug 6822 while invoking one of the smp_call_function*() 6823 functions. 6824 6825 scftorture.verbose= [KNL] 6826 Enable additional printk() statements. 6827 6828 scftorture.weight_single= [KNL] 6829 The probability weighting to use for the 6830 smp_call_function_single() function with a zero 6831 "wait" parameter. A value of -1 selects the 6832 default if all other weights are -1. However, 6833 if at least one weight has some other value, a 6834 value of -1 will instead select a weight of zero. 6835 6836 scftorture.weight_single_wait= [KNL] 6837 The probability weighting to use for the 6838 smp_call_function_single() function with a 6839 non-zero "wait" parameter. See weight_single. 6840 6841 scftorture.weight_many= [KNL] 6842 The probability weighting to use for the 6843 smp_call_function_many() function with a zero 6844 "wait" parameter. See weight_single. 6845 Note well that setting a high probability for 6846 this weighting can place serious IPI load 6847 on the system. 6848 6849 scftorture.weight_many_wait= [KNL] 6850 The probability weighting to use for the 6851 smp_call_function_many() function with a 6852 non-zero "wait" parameter. See weight_single 6853 and weight_many. 6854 6855 scftorture.weight_all= [KNL] 6856 The probability weighting to use for the 6857 smp_call_function_all() function with a zero 6858 "wait" parameter. See weight_single and 6859 weight_many. 6860 6861 scftorture.weight_all_wait= [KNL] 6862 The probability weighting to use for the 6863 smp_call_function_all() function with a 6864 non-zero "wait" parameter. See weight_single 6865 and weight_many. 6866 6867 sdw_mclk_divider=[SDW] 6868 Specify the MCLK divider for Intel SoundWire buses in 6869 case the BIOS does not provide the clock rate properly. 6870 6871 skew_tick= [KNL,EARLY] Offset the periodic timer tick per cpu to mitigate 6872 xtime_lock contention on larger systems, and/or RCU lock 6873 contention on all systems with CONFIG_MAXSMP set. 6874 Format: { "0" | "1" } 6875 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1" 6876 1 -- enable. 6877 Note: increases power consumption, thus should only be 6878 enabled if running jitter sensitive (HPC/RT) workloads. 6879 6880 security= [SECURITY] Choose a legacy "major" security module to 6881 enable at boot. This has been deprecated by the 6882 "lsm=" parameter. 6883 6884 selinux= [SELINUX] Disable or enable SELinux at boot time. 6885 Format: { "0" | "1" } 6886 See security/selinux/Kconfig help text. 6887 0 -- disable. 6888 1 -- enable. 6889 Default value is 1. 6890 6891 serialnumber [BUGS=X86-32] 6892 6893 sev=option[,option...] [X86-64] 6894 6895 debug 6896 Enable debug messages. 6897 6898 nosnp 6899 Do not enable SEV-SNP (applies to host/hypervisor 6900 only). Setting 'nosnp' avoids the RMP check overhead 6901 in memory accesses when users do not want to run 6902 SEV-SNP guests. 6903 6904 shapers= [NET] 6905 Maximal number of shapers. 6906 6907 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller 6908 Limit apic dumping. The parameter defines the maximal 6909 number of local apics being dumped. Also it is possible 6910 to set it to "all" by meaning -- no limit here. 6911 Format: { 1 (default) | 2 | ... | all }. 6912 The parameter valid if only apic=debug or 6913 apic=verbose is specified. 6914 Example: apic=debug show_lapic=all 6915 6916 slab_debug[=options[,slabs][;[options[,slabs]]...] [MM] 6917 Enabling slab_debug allows one to determine the 6918 culprit if slab objects become corrupted. Enabling 6919 slab_debug can create guard zones around objects and 6920 may poison objects when not in use. Also tracks the 6921 last alloc / free. For more information see 6922 Documentation/admin-guide/mm/slab.rst. 6923 (slub_debug legacy name also accepted for now) 6924 6925 Using this option implies the "no_hash_pointers" 6926 option which can be undone by adding the 6927 "hash_pointers=always" option. 6928 6929 slab_max_order= [MM] 6930 Determines the maximum allowed order for slabs. 6931 A high setting may cause OOMs due to memory 6932 fragmentation. For more information see 6933 Documentation/admin-guide/mm/slab.rst. 6934 (slub_max_order legacy name also accepted for now) 6935 6936 slab_merge [MM] 6937 Enable merging of slabs with similar size when the 6938 kernel is built without CONFIG_SLAB_MERGE_DEFAULT. 6939 (slub_merge legacy name also accepted for now) 6940 6941 slab_min_objects= [MM] 6942 The minimum number of objects per slab. SLUB will 6943 increase the slab order up to slab_max_order to 6944 generate a sufficiently large slab able to contain 6945 the number of objects indicated. The higher the number 6946 of objects the smaller the overhead of tracking slabs 6947 and the less frequently locks need to be acquired. 6948 For more information see 6949 Documentation/admin-guide/mm/slab.rst. 6950 (slub_min_objects legacy name also accepted for now) 6951 6952 slab_min_order= [MM] 6953 Determines the minimum page order for slabs. Must be 6954 lower or equal to slab_max_order. For more information see 6955 Documentation/admin-guide/mm/slab.rst. 6956 (slub_min_order legacy name also accepted for now) 6957 6958 slab_nomerge [MM] 6959 Disable merging of slabs with similar size. May be 6960 necessary if there is some reason to distinguish 6961 allocs to different slabs, especially in hardened 6962 environments where the risk of heap overflows and 6963 layout control by attackers can usually be 6964 frustrated by disabling merging. This will reduce 6965 most of the exposure of a heap attack to a single 6966 cache (risks via metadata attacks are mostly 6967 unchanged). Debug options disable merging on their 6968 own. 6969 For more information see 6970 Documentation/admin-guide/mm/slab.rst. 6971 (slub_nomerge legacy name also accepted for now) 6972 6973 slab_strict_numa [MM] 6974 Support memory policies on a per object level 6975 in the slab allocator. The default is for memory 6976 policies to be applied at the folio level when 6977 a new folio is needed or a partial folio is 6978 retrieved from the lists. Increases overhead 6979 in the slab fastpaths but gains more accurate 6980 NUMA kernel object placement which helps with slow 6981 interconnects in NUMA systems. 6982 6983 slram= [HW,MTD] 6984 6985 smart2= [HW] 6986 Format: <io1>[,<io2>[,...,<io8>]] 6987 6988 smp.csd_lock_timeout= [KNL] 6989 Specify the period of time in milliseconds 6990 that smp_call_function() and friends will wait 6991 for a CPU to release the CSD lock. This is 6992 useful when diagnosing bugs involving CPUs 6993 disabling interrupts for extended periods 6994 of time. Defaults to 5,000 milliseconds, and 6995 setting a value of zero disables this feature. 6996 This feature may be more efficiently disabled 6997 using the csdlock_debug- kernel parameter. 6998 6999 smp.panic_on_ipistall= [KNL] 7000 If a csd_lock_timeout extends for more than 7001 the specified number of milliseconds, panic the 7002 system. By default, let CSD-lock acquisition 7003 take as long as they take. Specifying 300,000 7004 for this value provides a 5-minute timeout. 7005 7006 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices 7007 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port 7008 smsc-ircc2.ircc_sir= [HW] SIR base I/O port 7009 smsc-ircc2.ircc_fir= [HW] FIR base I/O port 7010 smsc-ircc2.ircc_irq= [HW] IRQ line 7011 smsc-ircc2.ircc_dma= [HW] DMA channel 7012 smsc-ircc2.ircc_transceiver= [HW] Transceiver type: 7013 0: Toshiba Satellite 1800 (GP data pin select) 7014 1: Fast pin select (default) 7015 2: ATC IRMode 7016 7017 smt= [KNL,MIPS,S390,EARLY] Set the maximum number of threads 7018 (logical CPUs) to use per physical CPU on systems 7019 capable of symmetric multithreading (SMT). Will 7020 be capped to the actual hardware limit. 7021 Format: <integer> 7022 Default: -1 (no limit) 7023 7024 softlockup_panic= 7025 [KNL] Should the soft-lockup detector generate panics. 7026 Format: <int> 7027 7028 A value of non-zero instructs the soft-lockup detector 7029 to panic the machine when a soft-lockup duration exceeds 7030 N thresholds. It is also controlled by the kernel.softlockup_panic 7031 sysctl and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the 7032 respective build-time switch to that functionality. 7033 7034 softlockup_all_cpu_backtrace= 7035 [KNL] Should the soft-lockup detector generate 7036 backtraces on all cpus. 7037 Format: 0 | 1 7038 7039 sonypi.*= [HW] Sony Programmable I/O Control Device driver 7040 See Documentation/admin-guide/laptops/sonypi.rst 7041 7042 spectre_bhi= [X86] Control mitigation of Branch History Injection 7043 (BHI) vulnerability. This setting affects the 7044 deployment of the HW BHI control and the SW BHB 7045 clearing sequence. 7046 7047 on - (default) Enable the HW or SW mitigation as 7048 needed. This protects the kernel from 7049 both syscalls and VMs. 7050 vmexit - On systems which don't have the HW mitigation 7051 available, enable the SW mitigation on vmexit 7052 ONLY. On such systems, the host kernel is 7053 protected from VM-originated BHI attacks, but 7054 may still be vulnerable to syscall attacks. 7055 off - Disable the mitigation. 7056 7057 spectre_v2= [X86,EARLY] Control mitigation of Spectre variant 2 7058 (indirect branch speculation) vulnerability. 7059 The default operation protects the kernel from 7060 user space attacks. 7061 7062 on - unconditionally enable, implies 7063 spectre_v2_user=on 7064 off - unconditionally disable, implies 7065 spectre_v2_user=off 7066 auto - kernel detects whether your CPU model is 7067 vulnerable 7068 7069 Selecting 'on' will, and 'auto' may, choose a 7070 mitigation method at run time according to the 7071 CPU, the available microcode, the setting of the 7072 CONFIG_MITIGATION_RETPOLINE configuration option, 7073 and the compiler with which the kernel was built. 7074 7075 Selecting 'on' will also enable the mitigation 7076 against user space to user space task attacks. 7077 Selecting specific mitigation does not force enable 7078 user mitigations. 7079 7080 Selecting 'off' will disable both the kernel and 7081 the user space protections. 7082 7083 Specific mitigations can also be selected manually: 7084 7085 retpoline - replace indirect branches 7086 retpoline,generic - Retpolines 7087 retpoline,lfence - LFENCE; indirect branch 7088 retpoline,amd - alias for retpoline,lfence 7089 eibrs - Enhanced/Auto IBRS 7090 eibrs,retpoline - Enhanced/Auto IBRS + Retpolines 7091 eibrs,lfence - Enhanced/Auto IBRS + LFENCE 7092 ibrs - use IBRS to protect kernel 7093 7094 Not specifying this option is equivalent to 7095 spectre_v2=auto. 7096 7097 spectre_v2_user= 7098 [X86] Control mitigation of Spectre variant 2 7099 (indirect branch speculation) vulnerability between 7100 user space tasks 7101 7102 on - Unconditionally enable mitigations. Is 7103 enforced by spectre_v2=on 7104 7105 off - Unconditionally disable mitigations. Is 7106 enforced by spectre_v2=off 7107 7108 prctl - Indirect branch speculation is enabled, 7109 but mitigation can be enabled via prctl 7110 per thread. The mitigation control state 7111 is inherited on fork. 7112 7113 prctl,ibpb 7114 - Like "prctl" above, but only STIBP is 7115 controlled per thread. IBPB is issued 7116 always when switching between different user 7117 space processes. 7118 7119 seccomp 7120 - Same as "prctl" above, but all seccomp 7121 threads will enable the mitigation unless 7122 they explicitly opt out. 7123 7124 seccomp,ibpb 7125 - Like "seccomp" above, but only STIBP is 7126 controlled per thread. IBPB is issued 7127 always when switching between different 7128 user space processes. 7129 7130 auto - Kernel selects the mitigation depending on 7131 the available CPU features and vulnerability. 7132 7133 Default mitigation: "prctl" 7134 7135 Not specifying this option is equivalent to 7136 spectre_v2_user=auto. 7137 7138 spec_rstack_overflow= 7139 [X86,EARLY] Control RAS overflow mitigation on AMD Zen CPUs 7140 7141 off - Disable mitigation 7142 microcode - Enable microcode mitigation only 7143 safe-ret - Enable sw-only safe RET mitigation (default) 7144 ibpb - Enable mitigation by issuing IBPB on 7145 kernel entry 7146 ibpb-vmexit - Issue IBPB only on VMEXIT 7147 (cloud-specific mitigation) 7148 7149 spec_store_bypass_disable= 7150 [HW,EARLY] Control Speculative Store Bypass (SSB) Disable mitigation 7151 (Speculative Store Bypass vulnerability) 7152 7153 Certain CPUs are vulnerable to an exploit against a 7154 a common industry wide performance optimization known 7155 as "Speculative Store Bypass" in which recent stores 7156 to the same memory location may not be observed by 7157 later loads during speculative execution. The idea 7158 is that such stores are unlikely and that they can 7159 be detected prior to instruction retirement at the 7160 end of a particular speculation execution window. 7161 7162 In vulnerable processors, the speculatively forwarded 7163 store can be used in a cache side channel attack, for 7164 example to read memory to which the attacker does not 7165 directly have access (e.g. inside sandboxed code). 7166 7167 This parameter controls whether the Speculative Store 7168 Bypass optimization is used. 7169 7170 On x86 the options are: 7171 7172 on - Unconditionally disable Speculative Store Bypass 7173 off - Unconditionally enable Speculative Store Bypass 7174 auto - Kernel detects whether the CPU model contains an 7175 implementation of Speculative Store Bypass and 7176 picks the most appropriate mitigation. If the 7177 CPU is not vulnerable, "off" is selected. If the 7178 CPU is vulnerable the default mitigation is 7179 architecture and Kconfig dependent. See below. 7180 prctl - Control Speculative Store Bypass per thread 7181 via prctl. Speculative Store Bypass is enabled 7182 for a process by default. The state of the control 7183 is inherited on fork. 7184 seccomp - Same as "prctl" above, but all seccomp threads 7185 will disable SSB unless they explicitly opt out. 7186 7187 Default mitigations: 7188 X86: "prctl" 7189 7190 On powerpc the options are: 7191 7192 on,auto - On Power8 and Power9 insert a store-forwarding 7193 barrier on kernel entry and exit. On Power7 7194 perform a software flush on kernel entry and 7195 exit. 7196 off - No action. 7197 7198 Not specifying this option is equivalent to 7199 spec_store_bypass_disable=auto. 7200 7201 split_llc= 7202 [X86,EARLY] Split the LLC N-ways 7203 7204 When set, the LLC is split this many ways by matching 7205 'core_id % n'. This is setup before SMP bringup and 7206 used during SMP bringup before it knows the full 7207 topology. If your core count doesn't nicely divide by 7208 the number given, you get to keep the pieces. 7209 7210 This is mostly a debug feature to emulate multiple LLCs 7211 on hardware that only have a single LLC. 7212 7213 split_lock_detect= 7214 [X86] Enable split lock detection or bus lock detection 7215 7216 When enabled (and if hardware support is present), atomic 7217 instructions that access data across cache line 7218 boundaries will result in an alignment check exception 7219 for split lock detection or a debug exception for 7220 bus lock detection. 7221 7222 off - not enabled 7223 7224 warn - the kernel will emit rate-limited warnings 7225 about applications triggering the #AC 7226 exception or the #DB exception. This mode is 7227 the default on CPUs that support split lock 7228 detection or bus lock detection. Default 7229 behavior is by #AC if both features are 7230 enabled in hardware. 7231 7232 fatal - the kernel will send SIGBUS to applications 7233 that trigger the #AC exception or the #DB 7234 exception. Default behavior is by #AC if 7235 both features are enabled in hardware. 7236 7237 ratelimit:N - 7238 Set system wide rate limit to N bus locks 7239 per second for bus lock detection. 7240 0 < N <= 1000. 7241 7242 N/A for split lock detection. 7243 7244 7245 If an #AC exception is hit in the kernel or in 7246 firmware (i.e. not while executing in user mode) 7247 the kernel will oops in either "warn" or "fatal" 7248 mode. 7249 7250 #DB exception for bus lock is triggered only when 7251 CPL > 0. 7252 7253 srbds= [X86,INTEL,EARLY] 7254 Control the Special Register Buffer Data Sampling 7255 (SRBDS) mitigation. 7256 7257 Certain CPUs are vulnerable to an MDS-like 7258 exploit which can leak bits from the random 7259 number generator. 7260 7261 By default, this issue is mitigated by 7262 microcode. However, the microcode fix can cause 7263 the RDRAND and RDSEED instructions to become 7264 much slower. Among other effects, this will 7265 result in reduced throughput from /dev/urandom. 7266 7267 The microcode mitigation can be disabled with 7268 the following option: 7269 7270 off: Disable mitigation and remove 7271 performance impact to RDRAND and RDSEED 7272 7273 srcutree.big_cpu_lim [KNL] 7274 Specifies the number of CPUs constituting a 7275 large system, such that srcu_struct structures 7276 should immediately allocate an srcu_node array. 7277 This kernel-boot parameter defaults to 128, 7278 but takes effect only when the low-order four 7279 bits of srcutree.convert_to_big is equal to 3 7280 (decide at boot). 7281 7282 srcutree.convert_to_big [KNL] 7283 Specifies under what conditions an SRCU tree 7284 srcu_struct structure will be converted to big 7285 form, that is, with an rcu_node tree: 7286 7287 0: Never. 7288 1: At init_srcu_struct() time. 7289 2: When rcutorture decides to. 7290 3: Decide at boot time (default). 7291 0x1X: Above plus if high contention. 7292 7293 Either way, the srcu_node tree will be sized based 7294 on the actual runtime number of CPUs (nr_cpu_ids) 7295 instead of the compile-time CONFIG_NR_CPUS. 7296 7297 srcutree.counter_wrap_check [KNL] 7298 Specifies how frequently to check for 7299 grace-period sequence counter wrap for the 7300 srcu_data structure's ->srcu_gp_seq_needed field. 7301 The greater the number of bits set in this kernel 7302 parameter, the less frequently counter wrap will 7303 be checked for. Note that the bottom two bits 7304 are ignored. 7305 7306 srcutree.exp_holdoff [KNL] 7307 Specifies how many nanoseconds must elapse 7308 since the end of the last SRCU grace period for 7309 a given srcu_struct until the next normal SRCU 7310 grace period will be considered for automatic 7311 expediting. Set to zero to disable automatic 7312 expediting. 7313 7314 srcutree.srcu_max_nodelay [KNL] 7315 Specifies the number of no-delay instances 7316 per jiffy for which the SRCU grace period 7317 worker thread will be rescheduled with zero 7318 delay. Beyond this limit, worker thread will 7319 be rescheduled with a sleep delay of one jiffy. 7320 7321 srcutree.srcu_max_nodelay_phase [KNL] 7322 Specifies the per-grace-period phase, number of 7323 non-sleeping polls of readers. Beyond this limit, 7324 grace period worker thread will be rescheduled 7325 with a sleep delay of one jiffy, between each 7326 rescan of the readers, for a grace period phase. 7327 7328 srcutree.srcu_retry_check_delay [KNL] 7329 Specifies number of microseconds of non-sleeping 7330 delay between each non-sleeping poll of readers. 7331 7332 srcutree.small_contention_lim [KNL] 7333 Specifies the number of update-side contention 7334 events per jiffy will be tolerated before 7335 initiating a conversion of an srcu_struct 7336 structure to big form. Note that the value of 7337 srcutree.convert_to_big must have the 0x10 bit 7338 set for contention-based conversions to occur. 7339 7340 ssbd= [ARM64,HW,EARLY] 7341 Speculative Store Bypass Disable control 7342 7343 On CPUs that are vulnerable to the Speculative 7344 Store Bypass vulnerability and offer a 7345 firmware based mitigation, this parameter 7346 indicates how the mitigation should be used: 7347 7348 force-on: Unconditionally enable mitigation for 7349 for both kernel and userspace 7350 force-off: Unconditionally disable mitigation for 7351 for both kernel and userspace 7352 kernel: Always enable mitigation in the 7353 kernel, and offer a prctl interface 7354 to allow userspace to register its 7355 interest in being mitigated too. 7356 7357 stack_guard_gap= [MM] 7358 override the default stack gap protection. The value 7359 is in page units and it defines how many pages prior 7360 to (for stacks growing down) resp. after (for stacks 7361 growing up) the main stack are reserved for no other 7362 mapping. Default value is 256 pages. 7363 7364 stack_depot_disable= [KNL,EARLY] 7365 Setting this to true through kernel command line will 7366 disable the stack depot thereby saving the static memory 7367 consumed by the stack hash table. By default this is set 7368 to false. 7369 7370 stack_depot_max_pools= [KNL,EARLY] 7371 Specify the maximum number of pools to use for storing 7372 stack traces. Pools are allocated on-demand up to this 7373 limit. Default value is 8191 pools. 7374 7375 stacktrace [FTRACE] 7376 Enable the stack tracer on boot up. 7377 7378 stacktrace_filter=[function-list] 7379 [FTRACE] Limit the functions that the stack tracer 7380 will trace at boot up. function-list is a comma-separated 7381 list of functions. This list can be changed at run 7382 time by the stack_trace_filter file in the debugfs 7383 tracing directory. Note, this enables stack tracing 7384 and the stacktrace above is not needed. 7385 7386 sti= [PARISC,HW] 7387 Format: <num> 7388 Set the STI (builtin display/keyboard on the HP-PARISC 7389 machines) console (graphic card) which should be used 7390 as the initial boot-console. 7391 See also comment in drivers/video/sticore.c. 7392 7393 sti_font= [HW] 7394 See comment in drivers/video/sticore.c. 7395 7396 stifb= [HW] 7397 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]] 7398 7399 strict_sas_size= 7400 [X86] 7401 Format: <bool> 7402 Enable or disable strict sigaltstack size checks 7403 against the required signal frame size which 7404 depends on the supported FPU features. This can 7405 be used to filter out binaries which have 7406 not yet been made aware of AT_MINSIGSTKSZ. 7407 7408 stress_hpt [PPC,EARLY] 7409 Limits the number of kernel HPT entries in the hash 7410 page table to increase the rate of hash page table 7411 faults on kernel addresses. 7412 7413 stress_slb [PPC,EARLY] 7414 Limits the number of kernel SLB entries, and flushes 7415 them frequently to increase the rate of SLB faults 7416 on kernel addresses. 7417 7418 no_slb_preload [PPC,EARLY] 7419 Disables slb preloading for userspace. 7420 7421 sunrpc.min_resvport= 7422 sunrpc.max_resvport= 7423 [NFS,SUNRPC] 7424 SunRPC servers often require that client requests 7425 originate from a privileged port (i.e. a port in the 7426 range 0 < portnr < 1024). 7427 An administrator who wishes to reserve some of these 7428 ports for other uses may adjust the range that the 7429 kernel's sunrpc client considers to be privileged 7430 using these two parameters to set the minimum and 7431 maximum port values. 7432 7433 sunrpc.svc_rpc_per_connection_limit= 7434 [NFS,SUNRPC] 7435 Limit the number of requests that the server will 7436 process in parallel from a single connection. 7437 The default value is 0 (no limit). 7438 7439 sunrpc.pool_mode= 7440 [NFS] 7441 Control how the NFS server code allocates CPUs to 7442 service thread pools. Depending on how many NICs 7443 you have and where their interrupts are bound, this 7444 option will affect which CPUs will do NFS serving. 7445 Note: this parameter cannot be changed while the 7446 NFS server is running. 7447 7448 auto the server chooses an appropriate mode 7449 automatically using heuristics 7450 global a single global pool contains all CPUs 7451 percpu one pool for each CPU 7452 pernode one pool for each NUMA node (equivalent 7453 to global on non-NUMA machines) 7454 7455 sunrpc.tcp_slot_table_entries= 7456 sunrpc.udp_slot_table_entries= 7457 [NFS,SUNRPC] 7458 Sets the upper limit on the number of simultaneous 7459 RPC calls that can be sent from the client to a 7460 server. Increasing these values may allow you to 7461 improve throughput, but will also increase the 7462 amount of memory reserved for use by the client. 7463 7464 suspend.pm_test_delay= 7465 [SUSPEND] 7466 Sets the number of seconds to remain in a suspend test 7467 mode before resuming the system (see 7468 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG 7469 is set. Default value is 5. 7470 7471 svm= [PPC] 7472 Format: { on | off | y | n | 1 | 0 } 7473 This parameter controls use of the Protected 7474 Execution Facility on pSeries. 7475 7476 swiotlb= [ARM,PPC,MIPS,X86,S390,EARLY] 7477 Format: { <int> [,<int>] | force | noforce } 7478 <int> -- Number of I/O TLB slabs 7479 <int> -- Second integer after comma. Number of swiotlb 7480 areas with their own lock. Will be rounded up 7481 to a power of 2. 7482 force -- force using of bounce buffers even if they 7483 wouldn't be automatically used by the kernel 7484 noforce -- Never use bounce buffers (for debugging) 7485 7486 switches= [HW,M68k,EARLY] 7487 7488 sysctl.*= [KNL] 7489 Set a sysctl parameter, right before loading the init 7490 process, as if the value was written to the respective 7491 /proc/sys/... file. Both '.' and '/' are recognized as 7492 separators. Unrecognized parameters and invalid values 7493 are reported in the kernel log. Sysctls registered 7494 later by a loaded module cannot be set this way. 7495 Example: sysctl.vm.swappiness=40 7496 7497 sysrq_always_enabled 7498 [KNL] 7499 Ignore sysrq setting - this boot parameter will 7500 neutralize any effect of /proc/sys/kernel/sysrq. 7501 Useful for debugging. 7502 7503 tcpmhash_entries= [KNL,NET] 7504 Set the number of tcp_metrics_hash slots. 7505 Default value is 8192 or 16384 depending on total 7506 ram pages. This is used to specify the TCP metrics 7507 cache size. See Documentation/networking/ip-sysctl.rst 7508 "tcp_no_metrics_save" section for more details. 7509 7510 tdfx= [HW,DRM] 7511 7512 test_suspend= [SUSPEND] 7513 Format: { "mem" | "standby" | "freeze" }[,N] 7514 Specify "mem" (for Suspend-to-RAM) or "standby" (for 7515 standby suspend) or "freeze" (for suspend type freeze) 7516 as the system sleep state during system startup with 7517 the optional capability to repeat N number of times. 7518 The system is woken from this state using a 7519 wakeup-capable RTC alarm. 7520 7521 thash_entries= [KNL,NET] 7522 Set number of hash buckets for TCP connection 7523 7524 thermal.act= [HW,ACPI] 7525 -1: disable all active trip points in all thermal zones 7526 <degrees C>: override all lowest active trip points 7527 7528 thermal.crt= [HW,ACPI] 7529 -1: disable all critical trip points in all thermal zones 7530 <degrees C>: override all critical trip points 7531 7532 thermal.off= [HW,ACPI] 7533 1: disable ACPI thermal control 7534 7535 thermal.psv= [HW,ACPI] 7536 -1: disable all passive trip points 7537 <degrees C>: override all passive trip points to this 7538 value 7539 7540 thermal.tzp= [HW,ACPI] 7541 Specify global default ACPI thermal zone polling rate 7542 <deci-seconds>: poll all this frequency 7543 0: no polling (default) 7544 7545 thp_anon= [KNL] 7546 Format: <size>[KMG],<size>[KMG]:<state>;<size>[KMG]-<size>[KMG]:<state> 7547 state is one of "always", "madvise", "never" or "inherit". 7548 Control the default behavior of the system with respect 7549 to anonymous transparent hugepages. 7550 Can be used multiple times for multiple anon THP sizes. 7551 See Documentation/admin-guide/mm/transhuge.rst for more 7552 details. 7553 7554 threadirqs [KNL,EARLY] 7555 Force threading of all interrupt handlers except those 7556 marked explicitly IRQF_NO_THREAD. 7557 7558 thp_shmem= [KNL] 7559 Format: <size>[KMG],<size>[KMG]:<policy>;<size>[KMG]-<size>[KMG]:<policy> 7560 Control the default policy of each hugepage size for the 7561 internal shmem mount. <policy> is one of policies available 7562 for the shmem mount ("always", "inherit", "never", "within_size", 7563 and "advise"). 7564 It can be used multiple times for multiple shmem THP sizes. 7565 See Documentation/admin-guide/mm/transhuge.rst for more 7566 details. 7567 7568 topology= [S390,EARLY] 7569 Format: {off | on} 7570 Specify if the kernel should make use of the cpu 7571 topology information if the hardware supports this. 7572 The scheduler will make use of this information and 7573 e.g. base its process migration decisions on it. 7574 Default is on. 7575 7576 torture.disable_onoff_at_boot= [KNL] 7577 Prevent the CPU-hotplug component of torturing 7578 until after init has spawned. 7579 7580 torture.ftrace_dump_at_shutdown= [KNL] 7581 Dump the ftrace buffer at torture-test shutdown, 7582 even if there were no errors. This can be a 7583 very costly operation when many torture tests 7584 are running concurrently, especially on systems 7585 with rotating-rust storage. 7586 7587 torture.verbose_sleep_frequency= [KNL] 7588 Specifies how many verbose printk()s should be 7589 emitted between each sleep. The default of zero 7590 disables verbose-printk() sleeping. 7591 7592 torture.verbose_sleep_duration= [KNL] 7593 Duration of each verbose-printk() sleep in jiffies. 7594 7595 tpm.disable_pcr_integrity= [HW,TPM] 7596 Do not protect PCR registers from unintended physical 7597 access, or interposers in the bus by the means of 7598 having an integrity protected session wrapped around 7599 TPM2_PCR_Extend command. Consider this in a situation 7600 where TPM is heavily utilized by IMA, thus protection 7601 causing a major performance hit, and the space where 7602 machines are deployed is by other means guarded. 7603 7604 tpm_crb_ffa.busy_timeout_ms= [ARM64,TPM] 7605 Maximum time in milliseconds to retry sending a message 7606 to the TPM service before giving up. This parameter controls 7607 how long the system will continue retrying when the TPM 7608 service is busy. 7609 Format: <unsigned int> 7610 Default: 2000 (2 seconds) 7611 7612 tpm_suspend_pcr=[HW,TPM] 7613 Format: integer pcr id 7614 Specify that at suspend time, the tpm driver 7615 should extend the specified pcr with zeros, 7616 as a workaround for some chips which fail to 7617 flush the last written pcr on TPM_SaveState. 7618 This will guarantee that all the other pcrs 7619 are saved. 7620 7621 tpm_tis.interrupts= [HW,TPM] 7622 Enable interrupts for the MMIO based physical layer 7623 for the FIFO interface. By default it is set to false 7624 (0). For more information about TPM hardware interfaces 7625 defined by Trusted Computing Group (TCG) see 7626 https://trustedcomputinggroup.org/resource/pc-client-platform-tpm-profile-ptp-specification/ 7627 7628 tp_printk [FTRACE] 7629 Have the tracepoints sent to printk as well as the 7630 tracing ring buffer. This is useful for early boot up 7631 where the system hangs or reboots and does not give the 7632 option for reading the tracing buffer or performing a 7633 ftrace_dump_on_oops. 7634 7635 To turn off having tracepoints sent to printk, 7636 echo 0 > /proc/sys/kernel/tracepoint_printk 7637 Note, echoing 1 into this file without the 7638 tp_printk kernel cmdline option has no effect. 7639 7640 The tp_printk_stop_on_boot (see below) can also be used 7641 to stop the printing of events to console at 7642 late_initcall_sync. 7643 7644 ** CAUTION ** 7645 7646 Having tracepoints sent to printk() and activating high 7647 frequency tracepoints such as irq or sched, can cause 7648 the system to live lock. 7649 7650 tp_printk_stop_on_boot [FTRACE] 7651 When tp_printk (above) is set, it can cause a lot of noise 7652 on the console. It may be useful to only include the 7653 printing of events during boot up, as user space may 7654 make the system inoperable. 7655 7656 This command line option will stop the printing of events 7657 to console at the late_initcall_sync() time frame. 7658 7659 trace_buf_size=nn[KMG] 7660 [FTRACE] will set tracing buffer size on each cpu. 7661 7662 trace_clock= [FTRACE] Set the clock used for tracing events 7663 at boot up. 7664 local - Use the per CPU time stamp counter 7665 (converted into nanoseconds). Fast, but 7666 depending on the architecture, may not be 7667 in sync between CPUs. 7668 global - Event time stamps are synchronized across 7669 CPUs. May be slower than the local clock, 7670 but better for some race conditions. 7671 counter - Simple counting of events (1, 2, ..) 7672 note, some counts may be skipped due to the 7673 infrastructure grabbing the clock more than 7674 once per event. 7675 uptime - Use jiffies as the time stamp. 7676 perf - Use the same clock that perf uses. 7677 mono - Use ktime_get_mono_fast_ns() for time stamps. 7678 mono_raw - Use ktime_get_raw_fast_ns() for time 7679 stamps. 7680 boot - Use ktime_get_boot_fast_ns() for time stamps. 7681 Architectures may add more clocks. See 7682 Documentation/trace/ftrace.rst for more details. 7683 7684 trace_event=[event-list] 7685 [FTRACE] Set and start specified trace events in order 7686 to facilitate early boot debugging. The event-list is a 7687 comma-separated list of trace events to enable. See 7688 also Documentation/trace/events.rst 7689 7690 To enable modules, use :mod: keyword: 7691 7692 trace_event=:mod:<module> 7693 7694 The value before :mod: will only enable specific events 7695 that are part of the module. See the above mentioned 7696 document for more information. 7697 7698 trace_instance=[instance-info] 7699 [FTRACE] Create a ring buffer instance early in boot up. 7700 This will be listed in: 7701 7702 /sys/kernel/tracing/instances 7703 7704 Events can be enabled at the time the instance is created 7705 via: 7706 7707 trace_instance=<name>,<system1>:<event1>,<system2>:<event2> 7708 7709 Note, the "<system*>:" portion is optional if the event is 7710 unique. 7711 7712 trace_instance=foo,sched:sched_switch,irq_handler_entry,initcall 7713 7714 will enable the "sched_switch" event (note, the "sched:" is optional, and 7715 the same thing would happen if it was left off). The irq_handler_entry 7716 event, and all events under the "initcall" system. 7717 7718 Flags can be added to the instance to modify its behavior when it is 7719 created. The flags are separated by '^'. 7720 7721 The available flags are: 7722 7723 traceoff - Have the tracing instance tracing disabled after it is created. 7724 traceprintk - Have trace_printk() write into this trace instance 7725 (note, "printk" and "trace_printk" can also be used) 7726 7727 trace_instance=foo^traceoff^traceprintk,sched,irq 7728 7729 The flags must come before the defined events. 7730 7731 If memory has been reserved (see memmap for x86), the instance 7732 can use that memory: 7733 7734 memmap=12M$0x284500000 trace_instance=boot_map@0x284500000:12M 7735 7736 The above will create a "boot_map" instance that uses the physical 7737 memory at 0x284500000 that is 12Megs. The per CPU buffers of that 7738 instance will be split up accordingly. 7739 7740 Alternatively, the memory can be reserved by the reserve_mem option: 7741 7742 reserve_mem=12M:4096:trace trace_instance=boot_map@trace 7743 7744 This will reserve 12 megabytes at boot up with a 4096 byte alignment 7745 and place the ring buffer in this memory. Note that due to KASLR, the 7746 memory may not be the same location each time, which will not preserve 7747 the buffer content. 7748 7749 Also note that the layout of the ring buffer data may change between 7750 kernel versions where the validator will fail and reset the ring buffer 7751 if the layout is not the same as the previous kernel. 7752 7753 If the ring buffer is used for persistent bootups and has events enabled, 7754 it is recommend to disable tracing so that events from a previous boot do not 7755 mix with events of the current boot (unless you are debugging a random crash 7756 at boot up). 7757 7758 reserve_mem=12M:4096:trace trace_instance=boot_map^traceoff^traceprintk@trace,sched,irq 7759 7760 Note, saving the trace buffer across reboots does require that the system 7761 is set up to not wipe memory. For instance, CONFIG_RESET_ATTACK_MITIGATION 7762 can force a memory reset on boot which will clear any trace that was stored. 7763 This is just one of many ways that can clear memory. Make sure your system 7764 keeps the content of memory across reboots before relying on this option. 7765 7766 NB: Both the mapped address and size must be page aligned for the architecture. 7767 7768 See also Documentation/trace/debugging.rst 7769 7770 7771 trace_options=[option-list] 7772 [FTRACE] Enable or disable tracer options at boot. 7773 The option-list is a comma delimited list of options 7774 that can be enabled or disabled just as if you were 7775 to echo the option name into 7776 7777 /sys/kernel/tracing/trace_options 7778 7779 For example, to enable stacktrace option (to dump the 7780 stack trace of each event), add to the command line: 7781 7782 trace_options=stacktrace 7783 7784 See also Documentation/trace/ftrace.rst "trace options" 7785 section. 7786 7787 trace_trigger=[trigger-list] 7788 [FTRACE] Add an event trigger on specific events. 7789 Set a trigger on top of a specific event, with an optional 7790 filter. 7791 7792 The format is "trace_trigger=<event>.<trigger>[ if <filter>],..." 7793 Where more than one trigger may be specified that are comma delimited. 7794 7795 For example: 7796 7797 trace_trigger="sched_switch.stacktrace if prev_state == 2" 7798 7799 The above will enable the "stacktrace" trigger on the "sched_switch" 7800 event but only trigger it if the "prev_state" of the "sched_switch" 7801 event is "2" (TASK_UNINTERRUPTIBLE). 7802 7803 See also "Event triggers" in Documentation/trace/events.rst 7804 7805 7806 traceoff_after_boot 7807 [FTRACE] Sometimes tracing is used to debug issues 7808 during the boot process. Since the trace buffer has a 7809 limited amount of storage, it may be prudent to 7810 disable tracing after the boot is finished, otherwise 7811 the critical information may be overwritten. With this 7812 option, the main tracing buffer will be turned off at 7813 the end of the boot process. 7814 7815 traceoff_on_warning 7816 [FTRACE] enable this option to disable tracing when a 7817 warning is hit. This turns off "tracing_on". Tracing can 7818 be enabled again by echoing '1' into the "tracing_on" 7819 file located in /sys/kernel/tracing/ 7820 7821 This option is useful, as it disables the trace before 7822 the WARNING dump is called, which prevents the trace to 7823 be filled with content caused by the warning output. 7824 7825 This option can also be set at run time via the sysctl 7826 option: kernel/traceoff_on_warning 7827 7828 transparent_hugepage= 7829 [KNL] 7830 Format: [always|madvise|never] 7831 Can be used to control the default behavior of the system 7832 with respect to transparent hugepages. 7833 See Documentation/admin-guide/mm/transhuge.rst 7834 for more details. 7835 7836 transparent_hugepage_shmem= [KNL] 7837 Format: [always|within_size|advise|never|deny|force] 7838 Can be used to control the hugepage allocation policy for 7839 the internal shmem mount. 7840 See Documentation/admin-guide/mm/transhuge.rst 7841 for more details. 7842 7843 transparent_hugepage_tmpfs= [KNL] 7844 Format: [always|within_size|advise|never] 7845 Can be used to control the default hugepage allocation policy 7846 for the tmpfs mount. 7847 See Documentation/admin-guide/mm/transhuge.rst 7848 for more details. 7849 7850 trusted.source= [KEYS] 7851 Format: <string> 7852 This parameter identifies the trust source as a backend 7853 for trusted keys implementation. Supported trust 7854 sources: 7855 - "tpm" 7856 - "tee" 7857 - "caam" 7858 - "dcp" 7859 - "pkwm" 7860 If not specified then it defaults to iterating through 7861 the trust source list starting with TPM and assigns the 7862 first trust source as a backend which is initialized 7863 successfully during iteration. 7864 7865 trusted.rng= [KEYS] 7866 Format: <string> 7867 The RNG used to generate key material for trusted keys. 7868 Can be one of: 7869 - "kernel" 7870 - the same value as trusted.source: "tpm" or "tee" 7871 - "default" 7872 If not specified, "default" is used. In this case, 7873 the RNG's choice is left to each individual trust source. 7874 7875 trusted.dcp_use_otp_key 7876 This is intended to be used in combination with 7877 trusted.source=dcp and will select the DCP OTP key 7878 instead of the DCP UNIQUE key blob encryption. 7879 7880 trusted.dcp_skip_zk_test 7881 This is intended to be used in combination with 7882 trusted.source=dcp and will disable the check if the 7883 blob key is all zeros. This is helpful for situations where 7884 having this key zero'ed is acceptable. E.g. in testing 7885 scenarios. 7886 7887 tsa= [X86] Control mitigation for Transient Scheduler 7888 Attacks on AMD CPUs. Search the following in your 7889 favourite search engine for more details: 7890 7891 "Technical guidance for mitigating transient scheduler 7892 attacks". 7893 7894 off - disable the mitigation 7895 on - enable the mitigation (default) 7896 user - mitigate only user/kernel transitions 7897 vm - mitigate only guest/host transitions 7898 7899 7900 tsc= Disable clocksource stability checks for TSC. 7901 Format: <string> 7902 [x86] reliable: mark tsc clocksource as reliable, this 7903 disables clocksource verification at runtime, as well 7904 as the stability checks done at bootup. Used to enable 7905 high-resolution timer mode on older hardware, and in 7906 virtualized environment. 7907 [x86] noirqtime: Do not use TSC to do irq accounting. 7908 Used to run time disable IRQ_TIME_ACCOUNTING on any 7909 platforms where RDTSC is slow and this accounting 7910 can add overhead. 7911 [x86] unstable: mark the TSC clocksource as unstable, this 7912 marks the TSC unconditionally unstable at bootup and 7913 avoids any further wobbles once the TSC watchdog notices. 7914 [x86] nowatchdog: disable clocksource watchdog. Used 7915 in situations with strict latency requirements (where 7916 interruptions from clocksource watchdog are not 7917 acceptable). 7918 [x86] recalibrate: force recalibration against a HW timer 7919 (HPET or PM timer) on systems whose TSC frequency was 7920 obtained from HW or FW using either an MSR or CPUID(0x15). 7921 Warn if the difference is more than 500 ppm. 7922 [x86] watchdog: Enforce the clocksource watchdog on TSC 7923 7924 tsc_early_khz= [X86,EARLY] Skip early TSC calibration and use the given 7925 value instead. Useful when the early TSC frequency discovery 7926 procedure is not reliable, such as on overclocked systems 7927 with CPUID.16h support and partial CPUID.15h support. 7928 Format: <unsigned int> 7929 7930 tsx= [X86] Control Transactional Synchronization 7931 Extensions (TSX) feature in Intel processors that 7932 support TSX control. 7933 7934 This parameter controls the TSX feature. The options are: 7935 7936 on - Enable TSX on the system. Although there are 7937 mitigations for all known security vulnerabilities, 7938 TSX has been known to be an accelerator for 7939 several previous speculation-related CVEs, and 7940 so there may be unknown security risks associated 7941 with leaving it enabled. 7942 7943 off - Disable TSX on the system. (Note that this 7944 option takes effect only on newer CPUs which are 7945 not vulnerable to MDS, i.e., have 7946 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get 7947 the new IA32_TSX_CTRL MSR through a microcode 7948 update. This new MSR allows for the reliable 7949 deactivation of the TSX functionality.) 7950 7951 auto - Disable TSX if X86_BUG_TAA is present, 7952 otherwise enable TSX on the system. 7953 7954 Not specifying this option is equivalent to tsx=off. 7955 7956 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst 7957 for more details. 7958 7959 tsx_async_abort= [X86,INTEL,EARLY] Control mitigation for the TSX Async 7960 Abort (TAA) vulnerability. 7961 7962 Similar to Micro-architectural Data Sampling (MDS) 7963 certain CPUs that support Transactional 7964 Synchronization Extensions (TSX) are vulnerable to an 7965 exploit against CPU internal buffers which can forward 7966 information to a disclosure gadget under certain 7967 conditions. 7968 7969 In vulnerable processors, the speculatively forwarded 7970 data can be used in a cache side channel attack, to 7971 access data to which the attacker does not have direct 7972 access. 7973 7974 This parameter controls the TAA mitigation. The 7975 options are: 7976 7977 full - Enable TAA mitigation on vulnerable CPUs 7978 if TSX is enabled. 7979 7980 full,nosmt - Enable TAA mitigation and disable SMT on 7981 vulnerable CPUs. If TSX is disabled, SMT 7982 is not disabled because CPU is not 7983 vulnerable to cross-thread TAA attacks. 7984 off - Unconditionally disable TAA mitigation 7985 7986 On MDS-affected machines, tsx_async_abort=off can be 7987 prevented by an active MDS mitigation as both vulnerabilities 7988 are mitigated with the same mechanism so in order to disable 7989 this mitigation, you need to specify mds=off too. 7990 7991 Not specifying this option is equivalent to 7992 tsx_async_abort=full. On CPUs which are MDS affected 7993 and deploy MDS mitigation, TAA mitigation is not 7994 required and doesn't provide any additional 7995 mitigation. 7996 7997 For details see: 7998 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst 7999 8000 turbografx.map[2|3]= [HW,JOY] 8001 TurboGraFX parallel port interface 8002 Format: 8003 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7> 8004 See also Documentation/input/devices/joystick-parport.rst 8005 8006 udbg-immortal [PPC] When debugging early kernel crashes that 8007 happen after console_init() and before a proper 8008 console driver takes over, this boot options might 8009 help "seeing" what's going on. 8010 8011 uhash_entries= [KNL,NET] 8012 Set number of hash buckets for UDP/UDP-Lite connections 8013 8014 uhci-hcd.ignore_oc= 8015 [USB] Ignore overcurrent events (default N). 8016 Some badly-designed motherboards generate lots of 8017 bogus events, for ports that aren't wired to 8018 anything. Set this parameter to avoid log spamming. 8019 Note that genuine overcurrent events won't be 8020 reported either. 8021 8022 unaligned_scalar_speed= 8023 [RISCV] 8024 Format: {slow | fast | unsupported} 8025 Allow skipping scalar unaligned access speed tests. This 8026 is useful for testing alternative code paths and to skip 8027 the tests in environments where they run too slowly. All 8028 CPUs must have the same scalar unaligned access speed. 8029 8030 unaligned_vector_speed= 8031 [RISCV] 8032 Format: {slow | fast | unsupported} 8033 Allow skipping vector unaligned access speed tests. This 8034 is useful for testing alternative code paths and to skip 8035 the tests in environments where they run too slowly. All 8036 CPUs must have the same vector unaligned access speed. 8037 8038 unknown_nmi_panic 8039 [X86] Cause panic on unknown NMI. 8040 8041 unwind_debug [X86-64,EARLY] 8042 Enable unwinder debug output. This can be 8043 useful for debugging certain unwinder error 8044 conditions, including corrupt stacks and 8045 bad/missing unwinder metadata. 8046 8047 usbcore.authorized_default= 8048 [USB] Default USB device authorization: 8049 (default -1 = authorized (same as 1), 8050 0 = not authorized, 1 = authorized, 2 = authorized 8051 if device connected to internal port) 8052 8053 usbcore.autosuspend= 8054 [USB] The autosuspend time delay (in seconds) used 8055 for newly-detected USB devices (default 2). This 8056 is the time required before an idle device will be 8057 autosuspended. Devices for which the delay is set 8058 to a negative value won't be autosuspended at all. 8059 8060 usbcore.usbfs_snoop= 8061 [USB] Set to log all usbfs traffic (default 0 = off). 8062 8063 usbcore.usbfs_snoop_max= 8064 [USB] Maximum number of bytes to snoop in each URB 8065 (default = 65536). 8066 8067 usbcore.blinkenlights= 8068 [USB] Set to cycle leds on hubs (default 0 = off). 8069 8070 usbcore.old_scheme_first= 8071 [USB] Start with the old device initialization 8072 scheme (default 0 = off). 8073 8074 usbcore.usbfs_memory_mb= 8075 [USB] Memory limit (in MB) for buffers allocated by 8076 usbfs (default = 16, 0 = max = 2047). 8077 8078 usbcore.use_both_schemes= 8079 [USB] Try the other device initialization scheme 8080 if the first one fails (default 1 = enabled). 8081 8082 usbcore.initial_descriptor_timeout= 8083 [USB] Specifies timeout for the initial 64-byte 8084 USB_REQ_GET_DESCRIPTOR request in milliseconds 8085 (default 5000 = 5.0 seconds). 8086 8087 usbcore.nousb [USB] Disable the USB subsystem 8088 8089 usbcore.quirks= 8090 [USB] A list of quirk entries to augment the built-in 8091 usb core quirk list. List entries are separated by 8092 commas. Each entry has the form 8093 VendorID:ProductID:Flags. The IDs are 4-digit hex 8094 numbers and Flags is a set of letters. Each letter 8095 will change the built-in quirk; setting it if it is 8096 clear and clearing it if it is set. The letters have 8097 the following meanings: 8098 a = USB_QUIRK_STRING_FETCH_255 (string 8099 descriptors must not be fetched using 8100 a 255-byte read); 8101 b = USB_QUIRK_RESET_RESUME (device can't resume 8102 correctly so reset it instead); 8103 c = USB_QUIRK_NO_SET_INTF (device can't handle 8104 Set-Interface requests); 8105 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't 8106 handle its Configuration or Interface 8107 strings); 8108 e = USB_QUIRK_RESET (device can't be reset 8109 (e.g morph devices), don't use reset); 8110 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has 8111 more interface descriptions than the 8112 bNumInterfaces count, and can't handle 8113 talking to these interfaces); 8114 g = USB_QUIRK_DELAY_INIT (device needs a pause 8115 during initialization, after we read 8116 the device descriptor); 8117 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For 8118 high speed and super speed interrupt 8119 endpoints, the USB 2.0 and USB 3.0 spec 8120 require the interval in microframes (1 8121 microframe = 125 microseconds) to be 8122 calculated as interval = 2 ^ 8123 (bInterval-1). 8124 Devices with this quirk report their 8125 bInterval as the result of this 8126 calculation instead of the exponent 8127 variable used in the calculation); 8128 i = USB_QUIRK_DEVICE_QUALIFIER (device can't 8129 handle device_qualifier descriptor 8130 requests); 8131 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device 8132 generates spurious wakeup, ignore 8133 remote wakeup capability); 8134 k = USB_QUIRK_NO_LPM (device can't handle Link 8135 Power Management); 8136 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL 8137 (Device reports its bInterval as linear 8138 frames instead of the USB 2.0 8139 calculation); 8140 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs 8141 to be disconnected before suspend to 8142 prevent spurious wakeup); 8143 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a 8144 pause after every control message); 8145 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra 8146 delay after resetting its port); 8147 p = USB_QUIRK_SHORT_SET_ADDRESS_REQ_TIMEOUT 8148 (Reduce timeout of the SET_ADDRESS 8149 request from 5000 ms to 500 ms); 8150 q = USB_QUIRK_FORCE_ONE_CONFIG (Device 8151 claims zero configurations, 8152 forcing to 1); 8153 Example: quirks=0781:5580:bk,0a5c:5834:gij 8154 8155 usbhid.mousepoll= 8156 [USBHID] The interval which mice are to be polled at. 8157 8158 usbhid.jspoll= 8159 [USBHID] The interval which joysticks are to be polled at. 8160 8161 usbhid.kbpoll= 8162 [USBHID] The interval which keyboards are to be polled at. 8163 8164 usb-storage.delay_use= 8165 [UMS] The delay in seconds before a new device is 8166 scanned for Logical Units (default 1). 8167 Optionally the delay in milliseconds if the value has 8168 suffix with "ms". 8169 Example: delay_use=2567ms 8170 8171 usb-storage.quirks= 8172 [UMS] A list of quirks entries to supplement or 8173 override the built-in unusual_devs list. List 8174 entries are separated by commas. Each entry has 8175 the form VID:PID:Flags where VID and PID are Vendor 8176 and Product ID values (4-digit hex numbers) and 8177 Flags is a set of characters, each corresponding 8178 to a common usb-storage quirk flag as follows: 8179 a = SANE_SENSE (collect more than 18 bytes 8180 of sense data, not on uas); 8181 b = BAD_SENSE (don't collect more than 18 8182 bytes of sense data, not on uas); 8183 c = FIX_CAPACITY (decrease the reported 8184 device capacity by one sector); 8185 d = NO_READ_DISC_INFO (don't use 8186 READ_DISC_INFO command, not on uas); 8187 e = NO_READ_CAPACITY_16 (don't use 8188 READ_CAPACITY_16 command); 8189 f = NO_REPORT_OPCODES (don't use report opcodes 8190 command, uas only); 8191 g = MAX_SECTORS_240 (don't transfer more than 8192 240 sectors at a time, uas only); 8193 h = CAPACITY_HEURISTICS (decrease the 8194 reported device capacity by one 8195 sector if the number is odd); 8196 i = IGNORE_DEVICE (don't bind to this 8197 device); 8198 j = NO_REPORT_LUNS (don't use report luns 8199 command, uas only); 8200 k = NO_SAME (do not use WRITE_SAME, uas only) 8201 l = NOT_LOCKABLE (don't try to lock and 8202 unlock ejectable media, not on uas); 8203 m = MAX_SECTORS_64 (don't transfer more 8204 than 64 sectors = 32 KB at a time, 8205 not on uas); 8206 n = INITIAL_READ10 (force a retry of the 8207 initial READ(10) command, not on uas); 8208 o = CAPACITY_OK (accept the capacity 8209 reported by the device, not on uas); 8210 p = WRITE_CACHE (the device cache is ON 8211 by default, not on uas); 8212 r = IGNORE_RESIDUE (the device reports 8213 bogus residue values, not on uas); 8214 s = SINGLE_LUN (the device has only one 8215 Logical Unit); 8216 t = NO_ATA_1X (don't allow ATA(12) and ATA(16) 8217 commands, uas only); 8218 u = IGNORE_UAS (don't bind to the uas driver); 8219 w = NO_WP_DETECT (don't test whether the 8220 medium is write-protected). 8221 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE 8222 even if the device claims no cache, 8223 not on uas) 8224 Example: quirks=0419:aaf5:rl,0421:0433:rc 8225 8226 user_debug= [KNL,ARM] 8227 Format: <int> 8228 See arch/arm/Kconfig.debug help text. 8229 1 - undefined instruction events 8230 2 - system calls 8231 4 - invalid data aborts 8232 8 - SIGSEGV faults 8233 16 - SIGBUS faults 8234 Example: user_debug=31 8235 8236 vdso= [X86,SH,SPARC] 8237 On X86_32, this is an alias for vdso32=. Otherwise: 8238 8239 vdso=1: enable VDSO (the default) 8240 vdso=0: disable VDSO mapping 8241 8242 vdso32= [X86] Control the 32-bit vDSO 8243 vdso32=1: enable 32-bit VDSO 8244 vdso32=0 or vdso32=2: disable 32-bit VDSO 8245 8246 See the help text for CONFIG_COMPAT_VDSO for more 8247 details. If CONFIG_COMPAT_VDSO is set, the default is 8248 vdso32=0; otherwise, the default is vdso32=1. 8249 8250 For compatibility with older kernels, vdso32=2 is an 8251 alias for vdso32=0. 8252 8253 Try vdso32=0 if you encounter an error that says: 8254 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed! 8255 8256 video= [FB,EARLY] Frame buffer configuration 8257 See Documentation/fb/modedb.rst. 8258 8259 video.brightness_switch_enabled= [ACPI] 8260 Format: [0|1] 8261 If set to 1, on receiving an ACPI notify event 8262 generated by hotkey, video driver will adjust brightness 8263 level and then send out the event to user space through 8264 the allocated input device. If set to 0, video driver 8265 will only send out the event without touching backlight 8266 brightness level. 8267 default: 1 8268 8269 virtio_mmio.device= 8270 [VMMIO] Memory mapped virtio (platform) device. 8271 8272 <size>@<baseaddr>:<irq>[:<id>] 8273 where: 8274 <size> := size (can use standard suffixes 8275 like K, M and G) 8276 <baseaddr> := physical base address 8277 <irq> := interrupt number (as passed to 8278 request_irq()) 8279 <id> := (optional) platform device id 8280 example: 8281 virtio_mmio.device=1K@0x100b0000:48:7 8282 8283 Can be used multiple times for multiple devices. 8284 8285 vga= [BOOT,X86-32] Select a particular video mode 8286 See Documentation/arch/x86/boot.rst and 8287 Documentation/admin-guide/svga.rst. 8288 Use vga=ask for menu. 8289 This is actually a boot loader parameter; the value is 8290 passed to the kernel using a special protocol. 8291 8292 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y. 8293 May slow down system boot speed, especially when 8294 enabled on systems with a large amount of memory. 8295 All options are enabled by default, and this 8296 interface is meant to allow for selectively 8297 enabling or disabling specific virtual memory 8298 debugging features. 8299 8300 Available options are: 8301 P Enable page structure init time poisoning 8302 - Disable all of the above options 8303 8304 vmalloc=nn[KMG] [KNL,BOOT,EARLY] Forces the vmalloc area to have an 8305 exact size of <nn>. This can be used to increase 8306 the minimum size (128MB on x86, arm32 platforms). 8307 It can also be used to decrease the size and leave more room 8308 for directly mapped kernel RAM. Note that this parameter does 8309 not exist on many other platforms (including arm64, alpha, 8310 loongarch, arc, csky, hexagon, microblaze, mips, nios2, openrisc, 8311 parisc, m64k, powerpc, riscv, sh, um, xtensa, s390, sparc). 8312 8313 vmcp_cma=nn[MG] [KNL,S390,EARLY] 8314 Sets the memory size reserved for contiguous memory 8315 allocations for the vmcp device driver. 8316 8317 vmhalt= [KNL,S390] Perform z/VM CP command after system halt. 8318 Format: <command> 8319 8320 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic. 8321 Format: <command> 8322 8323 vmpoff= [KNL,S390] Perform z/VM CP command after power off. 8324 Format: <command> 8325 8326 vmscape= [X86] Controls mitigation for VMscape attacks. 8327 VMscape attacks can leak information from a userspace 8328 hypervisor to a guest via speculative side-channels. 8329 8330 off - disable the mitigation 8331 ibpb - use Indirect Branch Prediction Barrier 8332 (IBPB) mitigation (default) 8333 force - force vulnerability detection even on 8334 unaffected processors 8335 8336 vsyscall= [X86-64,EARLY] 8337 Controls the behavior of vsyscalls (i.e. calls to 8338 fixed addresses of 0xffffffffff600x00 from legacy 8339 code). Most statically-linked binaries and older 8340 versions of glibc use these calls. Because these 8341 functions are at fixed addresses, they make nice 8342 targets for exploits that can control RIP. 8343 8344 emulate Vsyscalls turn into traps and are emulated 8345 reasonably safely. The vsyscall page is 8346 readable. This disables the Linear 8347 Address Space Separation (LASS) security 8348 feature and makes the system less secure. 8349 8350 xonly [default] Vsyscalls turn into traps and are 8351 emulated reasonably safely. The vsyscall 8352 page is not readable. 8353 8354 none Vsyscalls don't work at all. This makes 8355 them quite hard to use for exploits but 8356 might break your system. 8357 8358 vt.color= [VT] Default text color. 8359 Format: 0xYX, X = foreground, Y = background. 8360 Default: 0x07 = light gray on black. 8361 8362 vt.cur_default= [VT] Default cursor shape. 8363 Format: 0xCCBBAA, where AA, BB, and CC are the same as 8364 the parameters of the <Esc>[?A;B;Cc escape sequence; 8365 see vga-softcursor.rst. Default: 2 = underline. 8366 8367 vt.default_blu= [VT] 8368 Format: <blue0>,<blue1>,<blue2>,...,<blue15> 8369 Change the default blue palette of the console. 8370 This is a 16-member array composed of values 8371 ranging from 0-255. 8372 8373 vt.default_grn= [VT] 8374 Format: <green0>,<green1>,<green2>,...,<green15> 8375 Change the default green palette of the console. 8376 This is a 16-member array composed of values 8377 ranging from 0-255. 8378 8379 vt.default_red= [VT] 8380 Format: <red0>,<red1>,<red2>,...,<red15> 8381 Change the default red palette of the console. 8382 This is a 16-member array composed of values 8383 ranging from 0-255. 8384 8385 vt.default_utf8= 8386 [VT] 8387 Format=<0|1> 8388 Set system-wide default UTF-8 mode for all tty's. 8389 Default is 1, i.e. UTF-8 mode is enabled for all 8390 newly opened terminals. 8391 8392 vt.global_cursor_default= 8393 [VT] 8394 Format=<-1|0|1> 8395 Set system-wide default for whether a cursor 8396 is shown on new VTs. Default is -1, 8397 i.e. cursors will be created by default unless 8398 overridden by individual drivers. 0 will hide 8399 cursors, 1 will display them. 8400 8401 vt.italic= [VT] Default color for italic text; 0-15. 8402 Default: 2 = green. 8403 8404 vt.underline= [VT] Default color for underlined text; 0-15. 8405 Default: 3 = cyan. 8406 8407 watchdog timers [HW,WDT] For information on watchdog timers, 8408 see Documentation/watchdog/watchdog-parameters.rst 8409 or other driver-specific files in the 8410 Documentation/watchdog/ directory. 8411 8412 watchdog_thresh= 8413 [KNL] 8414 Set the hard lockup detector stall duration 8415 threshold in seconds. The soft lockup detector 8416 threshold is set to twice the value. A value of 0 8417 disables both lockup detectors. Default is 10 8418 seconds. 8419 8420 workqueue.unbound_cpus= 8421 [KNL,SMP] Specify to constrain one or some CPUs 8422 to use in unbound workqueues. 8423 Format: <cpu-list> 8424 By default, all online CPUs are available for 8425 unbound workqueues. 8426 8427 workqueue.watchdog_thresh= 8428 If CONFIG_WQ_WATCHDOG is configured, workqueue can 8429 warn stall conditions and dump internal state to 8430 help debugging. 0 disables workqueue stall 8431 detection; otherwise, it's the stall threshold 8432 duration in seconds. The default value is 30 and 8433 it can be updated at runtime by writing to the 8434 corresponding sysfs file. 8435 8436 workqueue.panic_on_stall=<uint> 8437 Panic when workqueue stall is detected by 8438 CONFIG_WQ_WATCHDOG. It sets the number times of the 8439 stall to trigger panic. 8440 8441 The default is set by CONFIG_BOOTPARAM_WQ_STALL_PANIC, 8442 which is 0 (disabled) if not configured. 8443 8444 workqueue.panic_on_stall_time=<uint> 8445 Panic when a workqueue stall has been continuous for 8446 the specified number of seconds. Unlike panic_on_stall 8447 which counts accumulated stall events, this triggers 8448 based on the duration of a single continuous stall. 8449 8450 The default is 0, which disables the time-based panic. 8451 8452 workqueue.cpu_intensive_thresh_us= 8453 Per-cpu work items which run for longer than this 8454 threshold are automatically considered CPU intensive 8455 and excluded from concurrency management to prevent 8456 them from noticeably delaying other per-cpu work 8457 items. Default is 10000 (10ms). 8458 8459 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel 8460 will report the work functions which violate this 8461 threshold repeatedly. They are likely good 8462 candidates for using WQ_UNBOUND workqueues instead. 8463 8464 workqueue.cpu_intensive_warning_thresh=<uint> 8465 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel 8466 will report the work functions which violate the 8467 intensive_threshold_us repeatedly. In order to prevent 8468 spurious warnings, start printing only after a work 8469 function has violated this threshold number of times. 8470 8471 The default is 4 times. 0 disables the warning. 8472 8473 workqueue.power_efficient 8474 Per-cpu workqueues are generally preferred because 8475 they show better performance thanks to cache 8476 locality; unfortunately, per-cpu workqueues tend to 8477 be more power hungry than unbound workqueues. 8478 8479 Enabling this makes the per-cpu workqueues which 8480 were observed to contribute significantly to power 8481 consumption unbound, leading to measurably lower 8482 power usage at the cost of small performance 8483 overhead. 8484 8485 The default value of this parameter is determined by 8486 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT. 8487 8488 workqueue.default_affinity_scope= 8489 Select the default affinity scope to use for unbound 8490 workqueues. Can be one of "cpu", "smt", "cache", 8491 "cache_shard", "numa" and "system". Default is 8492 "cache_shard". For more 8493 information, see the Affinity Scopes section in 8494 Documentation/core-api/workqueue.rst. 8495 8496 This can be changed after boot by writing to the 8497 matching /sys/module/workqueue/parameters file. All 8498 workqueues with the "default" affinity scope will be 8499 updated accordingly. 8500 8501 workqueue.debug_force_rr_cpu 8502 Workqueue used to implicitly guarantee that work 8503 items queued without explicit CPU specified are put 8504 on the local CPU. This guarantee is no longer true 8505 and while local CPU is still preferred work items 8506 may be put on foreign CPUs. This debug option 8507 forces round-robin CPU selection to flush out 8508 usages which depend on the now broken guarantee. 8509 When enabled, memory and cache locality will be 8510 impacted. 8511 8512 writecombine= [LOONGARCH,EARLY] Control the MAT (Memory Access 8513 Type) of ioremap_wc(). 8514 8515 on - Enable writecombine, use WUC for ioremap_wc() 8516 off - Disable writecombine, use SUC for ioremap_wc() 8517 8518 x2apic_phys [X86-64,APIC,EARLY] Use x2apic physical mode instead of 8519 default x2apic cluster mode on platforms 8520 supporting x2apic. 8521 8522 xen_512gb_limit [KNL,X86-64,XEN] 8523 Restricts the kernel running paravirtualized under Xen 8524 to use only up to 512 GB of RAM. The reason to do so is 8525 crash analysis tools and Xen tools for doing domain 8526 save/restore/migration must be enabled to handle larger 8527 domains. 8528 8529 xen_console_io [XEN,EARLY] 8530 Boolean option to enable/disable the usage of the Xen 8531 console_io hypercalls to read and write to the console. 8532 Mostly useful for debugging and development. 8533 8534 xen_emul_unplug= [HW,X86,XEN,EARLY] 8535 Unplug Xen emulated devices 8536 Format: [unplug0,][unplug1] 8537 ide-disks -- unplug primary master IDE devices 8538 aux-ide-disks -- unplug non-primary-master IDE devices 8539 nics -- unplug network devices 8540 all -- unplug all emulated devices (NICs and IDE disks) 8541 unnecessary -- unplugging emulated devices is 8542 unnecessary even if the host did not respond to 8543 the unplug protocol 8544 never -- do not unplug even if version check succeeds 8545 8546 xen_legacy_crash [X86,XEN,EARLY] 8547 Crash from Xen panic notifier, without executing late 8548 panic() code such as dumping handler. 8549 8550 xen_mc_debug [X86,XEN,EARLY] 8551 Enable multicall debugging when running as a Xen PV guest. 8552 Enabling this feature will reduce performance a little 8553 bit, so it should only be enabled for obtaining extended 8554 debug data in case of multicall errors. 8555 8556 xen_msr_safe= [X86,XEN,EARLY] 8557 Format: <bool> 8558 Select whether to always use non-faulting (safe) MSR 8559 access functions when running as Xen PV guest. The 8560 default value is controlled by CONFIG_XEN_PV_MSR_SAFE. 8561 8562 xen_nopv [X86] 8563 Disables the PV optimizations forcing the HVM guest to 8564 run as generic HVM guest with no PV drivers. 8565 This option is obsoleted by the "nopv" option, which 8566 has equivalent effect for XEN platform. 8567 8568 xen_no_vector_callback 8569 [KNL,X86,XEN,EARLY] Disable the vector callback for Xen 8570 event channel interrupts. 8571 8572 xen_scrub_pages= [XEN] 8573 Boolean option to control scrubbing pages before giving them back 8574 to Xen, for use by other domains. Can be also changed at runtime 8575 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages. 8576 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT. 8577 8578 xen_timer_slop= [X86-64,XEN,EARLY] 8579 Set the timer slop (in nanoseconds) for the virtual Xen 8580 timers (default is 100000). This adjusts the minimum 8581 delta of virtualized Xen timers, where lower values 8582 improve timer resolution at the expense of processing 8583 more timer interrupts. 8584 8585 xen.balloon_boot_timeout= [XEN] 8586 The time (in seconds) to wait before giving up to boot 8587 in case initial ballooning fails to free enough memory. 8588 Applies only when running as HVM or PVH guest and 8589 started with less memory configured than allowed at 8590 max. Default is 180. 8591 8592 xen.event_eoi_delay= [XEN] 8593 How long to delay EOI handling in case of event 8594 storms (jiffies). Default is 10. 8595 8596 xen.event_loop_timeout= [XEN] 8597 After which time (jiffies) the event handling loop 8598 should start to delay EOI handling. Default is 2. 8599 8600 xen.fifo_events= [XEN] 8601 Boolean parameter to disable using fifo event handling 8602 even if available. Normally fifo event handling is 8603 preferred over the 2-level event handling, as it is 8604 fairer and the number of possible event channels is 8605 much higher. Default is on (use fifo events). 8606 8607 xirc2ps_cs= [NET,PCMCIA] 8608 Format: 8609 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]] 8610 8611 xive= [PPC] 8612 By default on POWER9 and above, the kernel will 8613 natively use the XIVE interrupt controller. This option 8614 allows the fallback firmware mode to be used: 8615 8616 off Fallback to firmware control of XIVE interrupt 8617 controller on both pseries and powernv 8618 platforms. Only useful on POWER9 and above. 8619 8620 xive.store-eoi=off [PPC] 8621 By default on POWER10 and above, the kernel will use 8622 stores for EOI handling when the XIVE interrupt mode 8623 is active. This option allows the XIVE driver to use 8624 loads instead, as on POWER9. 8625 8626 xhci-hcd.quirks [USB,KNL] 8627 A hex value specifying bitmask with supplemental xhci 8628 host controller quirks. Meaning of each bit can be 8629 consulted in header drivers/usb/host/xhci.h. 8630 8631 xmon [PPC,EARLY] 8632 Format: { early | on | rw | ro | off } 8633 Controls if xmon debugger is enabled. Default is off. 8634 Passing only "xmon" is equivalent to "xmon=early". 8635 early Call xmon as early as possible on boot; xmon 8636 debugger is called from setup_arch(). 8637 on xmon debugger hooks will be installed so xmon 8638 is only called on a kernel crash. Default mode, 8639 i.e. either "ro" or "rw" mode, is controlled 8640 with CONFIG_XMON_DEFAULT_RO_MODE. 8641 rw xmon debugger hooks will be installed so xmon 8642 is called only on a kernel crash, mode is write, 8643 meaning SPR registers, memory and, other data 8644 can be written using xmon commands. 8645 ro same as "rw" option above but SPR registers, 8646 memory, and other data can't be written using 8647 xmon commands. 8648 off xmon is disabled. 8649