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 AX25 Appropriate AX.25 support is enabled. 10 CLK Common clock infrastructure is enabled. 11 CMA Contiguous Memory Area support is enabled. 12 DRM Direct Rendering Management support is enabled. 13 DYNAMIC_DEBUG Build in debug messages and enable them at runtime 14 EARLY Parameter processed too early to be embedded in initrd. 15 EDD BIOS Enhanced Disk Drive Services (EDD) is enabled 16 EFI EFI Partitioning (GPT) is enabled 17 EVM Extended Verification Module 18 FB The frame buffer device is enabled. 19 FTRACE Function tracing enabled. 20 GCOV GCOV profiling is enabled. 21 HIBERNATION HIBERNATION is enabled. 22 HW Appropriate hardware is enabled. 23 HYPER_V HYPERV support is enabled. 24 IMA Integrity measurement architecture is enabled. 25 IP_PNP IP DHCP, BOOTP, or RARP is enabled. 26 IPV6 IPv6 support is enabled. 27 ISAPNP ISA PnP code is enabled. 28 ISDN Appropriate ISDN support is enabled. 29 ISOL CPU Isolation is enabled. 30 JOY Appropriate joystick support is enabled. 31 KGDB Kernel debugger support is enabled. 32 KVM Kernel Virtual Machine support is enabled. 33 LIBATA Libata driver is enabled 34 LOONGARCH LoongArch architecture is enabled. 35 LOOP Loopback device support is enabled. 36 LP Printer support is enabled. 37 M68k M68k architecture is enabled. 38 These options have more detailed description inside of 39 Documentation/arch/m68k/kernel-options.rst. 40 MDA MDA console support is enabled. 41 MIPS MIPS architecture is enabled. 42 MOUSE Appropriate mouse support is enabled. 43 MSI Message Signaled Interrupts (PCI). 44 MTD MTD (Memory Technology Device) support is enabled. 45 NET Appropriate network support is enabled. 46 NFS Appropriate NFS support is enabled. 47 NUMA NUMA support is enabled. 48 OF Devicetree is enabled. 49 PARISC The PA-RISC architecture is enabled. 50 PCI PCI bus support is enabled. 51 PCIE PCI Express support is enabled. 52 PCMCIA The PCMCIA subsystem is enabled. 53 PNP Plug & Play support is enabled. 54 PPC PowerPC architecture is enabled. 55 PPT Parallel port support is enabled. 56 PS2 Appropriate PS/2 support is enabled. 57 PV_OPS A paravirtualized kernel is enabled. 58 RAM RAM disk support is enabled. 59 RDT Intel Resource Director Technology. 60 RISCV RISCV architecture is enabled. 61 S390 S390 architecture is enabled. 62 SCSI Appropriate SCSI support is enabled. 63 A lot of drivers have their options described inside 64 the Documentation/scsi/ sub-directory. 65 SDW SoundWire support is enabled. 66 SECURITY Different security models are enabled. 67 SELINUX SELinux support is enabled. 68 SERIAL Serial support is enabled. 69 SH SuperH architecture is enabled. 70 SMP The kernel is an SMP kernel. 71 SPARC Sparc architecture is enabled. 72 SUSPEND System suspend states are enabled. 73 SWSUSP Software suspend (hibernation) is enabled. 74 TPM TPM drivers are enabled. 75 UMS USB Mass Storage support is enabled. 76 USB USB support is enabled. 77 USBHID USB Human Interface Device support is enabled. 78 V4L Video For Linux support is enabled. 79 VGA The VGA console has been enabled. 80 VMMIO Driver for memory mapped virtio devices is enabled. 81 VT Virtual terminal support is enabled. 82 WDT Watchdog support is enabled. 83 X86-32 X86-32, aka i386 architecture is enabled. 84 X86-64 X86-64 architecture is enabled. 85 X86 Either 32-bit or 64-bit x86 (same as X86-32+X86-64) 86 X86_UV SGI UV support is enabled. 87 XEN Xen support is enabled 88 XTENSA xtensa architecture is enabled. 89 90In addition, the following text indicates that the option 91 92 BOOT Is a boot loader parameter. 93 BUGS= Relates to possible processor bugs on the said processor. 94 KNL Is a kernel start-up parameter. 95 96 97Kernel parameters 98 99 accept_memory= [MM] 100 Format: { eager | lazy } 101 default: lazy 102 By default, unaccepted memory is accepted lazily to 103 avoid prolonged boot times. The lazy option will add 104 some runtime overhead until all memory is eventually 105 accepted. In most cases the overhead is negligible. 106 For some workloads or for debugging purposes 107 accept_memory=eager can be used to accept all memory 108 at once during boot. 109 110 acpi= [HW,ACPI,X86,ARM64,RISCV64,EARLY] 111 Advanced Configuration and Power Interface 112 Format: { force | on | off | strict | noirq | rsdt | 113 copy_dsdt | nospcr } 114 force -- enable ACPI if default was off 115 on -- enable ACPI but allow fallback to DT [arm64,riscv64] 116 off -- disable ACPI if default was on 117 noirq -- do not use ACPI for IRQ routing 118 strict -- Be less tolerant of platforms that are not 119 strictly ACPI specification compliant. 120 rsdt -- prefer RSDT over (default) XSDT 121 copy_dsdt -- copy DSDT to memory 122 nocmcff -- Disable firmware first mode for corrected 123 errors. This disables parsing the HEST CMC error 124 source to check if firmware has set the FF flag. This 125 may result in duplicate corrected error reports. 126 nospcr -- disable console in ACPI SPCR table as 127 default _serial_ console on ARM64 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_enforce_resources= [ACPI] 191 { strict | lax | no } 192 Check for resource conflicts between native drivers 193 and ACPI OperationRegions (SystemIO and SystemMemory 194 only). IO ports and memory declared in ACPI might be 195 used by the ACPI subsystem in arbitrary AML code and 196 can interfere with legacy drivers. 197 strict (default): access to resources claimed by ACPI 198 is denied; legacy drivers trying to access reserved 199 resources will fail to bind to device using them. 200 lax: access to resources claimed by ACPI is allowed; 201 legacy drivers trying to access reserved resources 202 will bind successfully but a warning message is logged. 203 no: ACPI OperationRegions are not marked as reserved, 204 no further checks are performed. 205 206 acpi_force_table_verification [HW,ACPI,EARLY] 207 Enable table checksum verification during early stage. 208 By default, this is disabled due to x86 early mapping 209 size limitation. 210 211 acpi_irq_balance [HW,ACPI] 212 ACPI will balance active IRQs 213 default in APIC mode 214 215 acpi_irq_nobalance [HW,ACPI] 216 ACPI will not move active IRQs (default) 217 default in PIC mode 218 219 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA 220 Format: <irq>,<irq>... 221 222 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for 223 use by PCI 224 Format: <irq>,<irq>... 225 226 acpi_mask_gpe= [HW,ACPI] 227 Due to the existence of _Lxx/_Exx, some GPEs triggered 228 by unsupported hardware/firmware features can result in 229 GPE floodings that cannot be automatically disabled by 230 the GPE dispatcher. 231 This facility can be used to prevent such uncontrolled 232 GPE floodings. 233 Format: <byte> or <bitmap-list> 234 235 acpi_no_auto_serialize [HW,ACPI] 236 Disable auto-serialization of AML methods 237 AML control methods that contain the opcodes to create 238 named objects will be marked as "Serialized" by the 239 auto-serialization feature. 240 This feature is enabled by default. 241 This option allows to turn off the feature. 242 243 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump 244 kernels. 245 246 acpi_no_static_ssdt [HW,ACPI,EARLY] 247 Disable installation of static SSDTs at early boot time 248 By default, SSDTs contained in the RSDT/XSDT will be 249 installed automatically and they will appear under 250 /sys/firmware/acpi/tables. 251 This option turns off this feature. 252 Note that specifying this option does not affect 253 dynamic table installation which will install SSDT 254 tables to /sys/firmware/acpi/tables/dynamic. 255 256 acpi_no_watchdog [HW,ACPI,WDT] 257 Ignore the ACPI-based watchdog interface (WDAT) and let 258 a native driver control the watchdog device instead. 259 260 acpi_rsdp= [ACPI,EFI,KEXEC,EARLY] 261 Pass the RSDP address to the kernel, mostly used 262 on machines running EFI runtime service to boot the 263 second kernel for kdump. 264 265 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS 266 Format: To spoof as Windows 98: ="Microsoft Windows" 267 268 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead 269 of 2 which is mandated by ACPI 6) as the supported ACPI 270 specification revision (when using this switch, it may 271 be necessary to carry out a cold reboot _twice_ in a 272 row to make it take effect on the platform firmware). 273 274 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings 275 acpi_osi="string1" # add string1 276 acpi_osi="!string2" # remove string2 277 acpi_osi=!* # remove all strings 278 acpi_osi=! # disable all built-in OS vendor 279 strings 280 acpi_osi=!! # enable all built-in OS vendor 281 strings 282 acpi_osi= # disable all strings 283 284 'acpi_osi=!' can be used in combination with single or 285 multiple 'acpi_osi="string1"' to support specific OS 286 vendor string(s). Note that such command can only 287 affect the default state of the OS vendor strings, thus 288 it cannot affect the default state of the feature group 289 strings and the current state of the OS vendor strings, 290 specifying it multiple times through kernel command line 291 is meaningless. This command is useful when one do not 292 care about the state of the feature group strings which 293 should be controlled by the OSPM. 294 Examples: 295 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent 296 to 'acpi_osi="Windows 2000" acpi_osi=!', they all 297 can make '_OSI("Windows 2000")' TRUE. 298 299 'acpi_osi=' cannot be used in combination with other 300 'acpi_osi=' command lines, the _OSI method will not 301 exist in the ACPI namespace. NOTE that such command can 302 only affect the _OSI support state, thus specifying it 303 multiple times through kernel command line is also 304 meaningless. 305 Examples: 306 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)' 307 FALSE. 308 309 'acpi_osi=!*' can be used in combination with single or 310 multiple 'acpi_osi="string1"' to support specific 311 string(s). Note that such command can affect the 312 current state of both the OS vendor strings and the 313 feature group strings, thus specifying it multiple times 314 through kernel command line is meaningful. But it may 315 still not able to affect the final state of a string if 316 there are quirks related to this string. This command 317 is useful when one want to control the state of the 318 feature group strings to debug BIOS issues related to 319 the OSPM features. 320 Examples: 321 1. 'acpi_osi="Module Device" acpi_osi=!*' can make 322 '_OSI("Module Device")' FALSE. 323 2. 'acpi_osi=!* acpi_osi="Module Device"' can make 324 '_OSI("Module Device")' TRUE. 325 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is 326 equivalent to 327 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"' 328 and 329 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!', 330 they all will make '_OSI("Windows 2000")' TRUE. 331 332 acpi_pm_good [X86] 333 Override the pmtimer bug detection: force the kernel 334 to assume that this machine's pmtimer latches its value 335 and always returns good values. 336 337 acpi_sci= [HW,ACPI,EARLY] ACPI System Control Interrupt trigger mode 338 Format: { level | edge | high | low } 339 340 acpi_skip_timer_override [HW,ACPI,EARLY] 341 Recognize and ignore IRQ0/pin2 Interrupt Override. 342 For broken nForce2 BIOS resulting in XT-PIC timer. 343 344 acpi_sleep= [HW,ACPI] Sleep options 345 Format: { s3_bios, s3_mode, s3_beep, s4_hwsig, 346 s4_nohwsig, old_ordering, nonvs, 347 sci_force_enable, nobl } 348 See Documentation/power/video.rst for information on 349 s3_bios and s3_mode. 350 s3_beep is for debugging; it makes the PC's speaker beep 351 as soon as the kernel's real-mode entry point is called. 352 s4_hwsig causes the kernel to check the ACPI hardware 353 signature during resume from hibernation, and gracefully 354 refuse to resume if it has changed. This complies with 355 the ACPI specification but not with reality, since 356 Windows does not do this and many laptops do change it 357 on docking. So the default behaviour is to allow resume 358 and simply warn when the signature changes, unless the 359 s4_hwsig option is enabled. 360 s4_nohwsig prevents ACPI hardware signature from being 361 used (or even warned about) during resume. 362 old_ordering causes the ACPI 1.0 ordering of the _PTS 363 control method, with respect to putting devices into 364 low power states, to be enforced (the ACPI 2.0 ordering 365 of _PTS is used by default). 366 nonvs prevents the kernel from saving/restoring the 367 ACPI NVS memory during suspend/hibernation and resume. 368 sci_force_enable causes the kernel to set SCI_EN directly 369 on resume from S1/S3 (which is against the ACPI spec, 370 but some broken systems don't work without it). 371 nobl causes the internal blacklist of systems known to 372 behave incorrectly in some ways with respect to system 373 suspend and resume to be ignored (use wisely). 374 375 acpi_use_timer_override [HW,ACPI,EARLY] 376 Use timer override. For some broken Nvidia NF5 boards 377 that require a timer override, but don't have HPET 378 379 add_efi_memmap [EFI,X86,EARLY] Include EFI memory map in 380 kernel's map of available physical RAM. 381 382 agp= [AGP] 383 { off | try_unsupported } 384 off: disable AGP support 385 try_unsupported: try to drive unsupported chipsets 386 (may crash computer or cause data corruption) 387 388 ALSA [HW,ALSA] 389 See Documentation/sound/alsa-configuration.rst 390 391 alignment= [KNL,ARM] 392 Allow the default userspace alignment fault handler 393 behaviour to be specified. Bit 0 enables warnings, 394 bit 1 enables fixups, and bit 2 sends a segfault. 395 396 align_va_addr= [X86-64] 397 Align virtual addresses by clearing slice [14:12] when 398 allocating a VMA at process creation time. This option 399 gives you up to 3% performance improvement on AMD F15h 400 machines (where it is enabled by default) for a 401 CPU-intensive style benchmark, and it can vary highly in 402 a microbenchmark depending on workload and compiler. 403 404 32: only for 32-bit processes 405 64: only for 64-bit processes 406 on: enable for both 32- and 64-bit processes 407 off: disable for both 32- and 64-bit processes 408 409 alloc_snapshot [FTRACE] 410 Allocate the ftrace snapshot buffer on boot up when the 411 main buffer is allocated. This is handy if debugging 412 and you need to use tracing_snapshot() on boot up, and 413 do not want to use tracing_snapshot_alloc() as it needs 414 to be done where GFP_KERNEL allocations are allowed. 415 416 allow_mismatched_32bit_el0 [ARM64,EARLY] 417 Allow execve() of 32-bit applications and setting of the 418 PER_LINUX32 personality on systems where only a strict 419 subset of the CPUs support 32-bit EL0. When this 420 parameter is present, the set of CPUs supporting 32-bit 421 EL0 is indicated by /sys/devices/system/cpu/aarch32_el0 422 and hot-unplug operations may be restricted. 423 424 See Documentation/arch/arm64/asymmetric-32bit.rst for more 425 information. 426 427 amd_iommu= [HW,X86-64] 428 Pass parameters to the AMD IOMMU driver in the system. 429 Possible values are: 430 fullflush - Deprecated, equivalent to iommu.strict=1 431 off - do not initialize any AMD IOMMU found in 432 the system 433 force_isolation - Force device isolation for all 434 devices. The IOMMU driver is not 435 allowed anymore to lift isolation 436 requirements as needed. This option 437 does not override iommu=pt 438 force_enable - Force enable the IOMMU on platforms known 439 to be buggy with IOMMU enabled. Use this 440 option with care. 441 pgtbl_v1 - Use v1 page table for DMA-API (Default). 442 pgtbl_v2 - Use v2 page table for DMA-API. 443 irtcachedis - Disable Interrupt Remapping Table (IRT) caching. 444 nohugepages - Limit page-sizes used for v1 page-tables 445 to 4 KiB. 446 v2_pgsizes_only - Limit page-sizes used for v1 page-tables 447 to 4KiB/2Mib/1GiB. 448 449 450 amd_iommu_dump= [HW,X86-64] 451 Enable AMD IOMMU driver option to dump the ACPI table 452 for AMD IOMMU. With this option enabled, AMD IOMMU 453 driver will print ACPI tables for AMD IOMMU during 454 IOMMU initialization. 455 456 amd_iommu_intr= [HW,X86-64] 457 Specifies one of the following AMD IOMMU interrupt 458 remapping modes: 459 legacy - Use legacy interrupt remapping mode. 460 vapic - Use virtual APIC mode, which allows IOMMU 461 to inject interrupts directly into guest. 462 This mode requires kvm-amd.avic=1. 463 (Default when IOMMU HW support is present.) 464 465 amd_pstate= [X86,EARLY] 466 disable 467 Do not enable amd_pstate as the default 468 scaling driver for the supported processors 469 passive 470 Use amd_pstate with passive mode as a scaling driver. 471 In this mode autonomous selection is disabled. 472 Driver requests a desired performance level and platform 473 tries to match the same performance level if it is 474 satisfied by guaranteed performance level. 475 active 476 Use amd_pstate_epp driver instance as the scaling driver, 477 driver provides a hint to the hardware if software wants 478 to bias toward performance (0x0) or energy efficiency (0xff) 479 to the CPPC firmware. then CPPC power algorithm will 480 calculate the runtime workload and adjust the realtime cores 481 frequency. 482 guided 483 Activate guided autonomous mode. Driver requests minimum and 484 maximum performance level and the platform autonomously 485 selects a performance level in this range and appropriate 486 to the current workload. 487 488 amd_prefcore= 489 [X86] 490 disable 491 Disable amd-pstate preferred core. 492 493 amijoy.map= [HW,JOY] Amiga joystick support 494 Map of devices attached to JOY0DAT and JOY1DAT 495 Format: <a>,<b> 496 See also Documentation/input/joydev/joystick.rst 497 498 analog.map= [HW,JOY] Analog joystick and gamepad support 499 Specifies type or capabilities of an analog joystick 500 connected to one of 16 gameports 501 Format: <type1>,<type2>,..<type16> 502 503 apc= [HW,SPARC] 504 Power management functions (SPARCstation-4/5 + deriv.) 505 Format: noidle 506 Disable APC CPU standby support. SPARCstation-Fox does 507 not play well with APC CPU idle - disable it if you have 508 APC and your system crashes randomly. 509 510 apic [APIC,X86-64] Use IO-APIC. Default. 511 512 apic= [APIC,X86,EARLY] Advanced Programmable Interrupt Controller 513 Change the output verbosity while booting 514 Format: { quiet (default) | verbose | debug } 515 Change the amount of debugging information output 516 when initialising the APIC and IO-APIC components. 517 518 apic_extnmi= [APIC,X86,EARLY] External NMI delivery setting 519 Format: { bsp (default) | all | none } 520 bsp: External NMI is delivered only to CPU 0 521 all: External NMIs are broadcast to all CPUs as a 522 backup of CPU 0 523 none: External NMI is masked for all CPUs. This is 524 useful so that a dump capture kernel won't be 525 shot down by NMI 526 527 apicpmtimer Do APIC timer calibration using the pmtimer. Implies 528 apicmaintimer. Useful when your PIT timer is totally 529 broken. 530 531 autoconf= [IPV6] 532 See Documentation/networking/ipv6.rst. 533 534 apm= [APM] Advanced Power Management 535 See header of arch/x86/kernel/apm_32.c. 536 537 apparmor= [APPARMOR] Disable or enable AppArmor at boot time 538 Format: { "0" | "1" } 539 See security/apparmor/Kconfig help text 540 0 -- disable. 541 1 -- enable. 542 Default value is set via kernel config option. 543 544 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards 545 Format: <io>,<irq>,<nodeID> 546 547 arm64.no32bit_el0 [ARM64] Unconditionally disable the execution of 548 32 bit applications. 549 550 arm64.nobti [ARM64] Unconditionally disable Branch Target 551 Identification support 552 553 arm64.nogcs [ARM64] Unconditionally disable Guarded Control Stack 554 support 555 556 arm64.nomops [ARM64] Unconditionally disable Memory Copy and Memory 557 Set instructions support 558 559 arm64.nompam [ARM64] Unconditionally disable Memory Partitioning And 560 Monitoring support 561 562 arm64.nomte [ARM64] Unconditionally disable Memory Tagging Extension 563 support 564 565 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication 566 support 567 568 arm64.nosme [ARM64] Unconditionally disable Scalable Matrix 569 Extension support 570 571 arm64.nosve [ARM64] Unconditionally disable Scalable Vector 572 Extension support 573 574 ataflop= [HW,M68k] 575 576 atarimouse= [HW,MOUSE] Atari Mouse 577 578 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess, 579 EzKey and similar keyboards 580 581 atkbd.reset= [HW] Reset keyboard during initialization 582 583 atkbd.set= [HW] Select keyboard code set 584 Format: <int> (2 = AT (default), 3 = PS/2) 585 586 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar 587 keyboards 588 589 atkbd.softraw= [HW] Choose between synthetic and real raw mode 590 Format: <bool> (0 = real, 1 = synthetic (default)) 591 592 atkbd.softrepeat= [HW] 593 Use software keyboard repeat 594 595 audit= [KNL] Enable the audit sub-system 596 Format: { "0" | "1" | "off" | "on" } 597 0 | off - kernel audit is disabled and can not be 598 enabled until the next reboot 599 unset - kernel audit is initialized but disabled and 600 will be fully enabled by the userspace auditd. 601 1 | on - kernel audit is initialized and partially 602 enabled, storing at most audit_backlog_limit 603 messages in RAM until it is fully enabled by the 604 userspace auditd. 605 Default: unset 606 607 audit_backlog_limit= [KNL] Set the audit queue size limit. 608 Format: <int> (must be >=0) 609 Default: 64 610 611 bau= [X86_UV] Enable the BAU on SGI UV. The default 612 behavior is to disable the BAU (i.e. bau=0). 613 Format: { "0" | "1" } 614 0 - Disable the BAU. 615 1 - Enable the BAU. 616 unset - Disable the BAU. 617 618 baycom_epp= [HW,AX25] 619 Format: <io>,<mode> 620 621 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem 622 Format: <io>,<mode> 623 See header of drivers/net/hamradio/baycom_par.c. 624 625 baycom_ser_fdx= [HW,AX25] 626 BayCom Serial Port AX.25 Modem (Full Duplex Mode) 627 Format: <io>,<irq>,<mode>[,<baud>] 628 See header of drivers/net/hamradio/baycom_ser_fdx.c. 629 630 baycom_ser_hdx= [HW,AX25] 631 BayCom Serial Port AX.25 Modem (Half Duplex Mode) 632 Format: <io>,<irq>,<mode> 633 See header of drivers/net/hamradio/baycom_ser_hdx.c. 634 635 bdev_allow_write_mounted= 636 Format: <bool> 637 Control the ability to open a mounted block device 638 for writing, i.e., allow / disallow writes that bypass 639 the FS. This was implemented as a means to prevent 640 fuzzers from crashing the kernel by overwriting the 641 metadata underneath a mounted FS without its awareness. 642 This also prevents destructive formatting of mounted 643 filesystems by naive storage tooling that don't use 644 O_EXCL. Default is Y and can be changed through the 645 Kconfig option CONFIG_BLK_DEV_WRITE_MOUNTED. 646 647 bert_disable [ACPI] 648 Disable BERT OS support on buggy BIOSes. 649 650 bgrt_disable [ACPI,X86,EARLY] 651 Disable BGRT to avoid flickering OEM logo. 652 653 blkdevparts= Manual partition parsing of block device(s) for 654 embedded devices based on command line input. 655 See Documentation/block/cmdline-partition.rst 656 657 boot_delay= [KNL,EARLY] 658 Milliseconds to delay each printk during boot. 659 Only works if CONFIG_BOOT_PRINTK_DELAY is enabled, 660 and you may also have to specify "lpj=". Boot_delay 661 values larger than 10 seconds (10000) are assumed 662 erroneous and ignored. 663 Format: integer 664 665 bootconfig [KNL,EARLY] 666 Extended command line options can be added to an initrd 667 and this will cause the kernel to look for it. 668 669 See Documentation/admin-guide/bootconfig.rst 670 671 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards) 672 bttv.radio= Most important insmod options are available as 673 kernel args too. 674 bttv.pll= See Documentation/admin-guide/media/bttv.rst 675 bttv.tuner= 676 677 bulk_remove=off [PPC] This parameter disables the use of the pSeries 678 firmware feature for flushing multiple hpte entries 679 at a time. 680 681 c101= [NET] Moxa C101 synchronous serial card 682 683 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection. 684 Sometimes CPU hardware bugs make them report the cache 685 size incorrectly. The kernel will attempt work arounds 686 to fix known problems, but for some CPUs it is not 687 possible to determine what the correct size should be. 688 This option provides an override for these situations. 689 690 carrier_timeout= 691 [NET] Specifies amount of time (in seconds) that 692 the kernel should wait for a network carrier. By default 693 it waits 120 seconds. 694 695 ca_keys= [KEYS] This parameter identifies a specific key(s) on 696 the system trusted keyring to be used for certificate 697 trust validation. 698 format: { id:<keyid> | builtin } 699 700 cca= [MIPS,EARLY] Override the kernel pages' cache coherency 701 algorithm. Accepted values range from 0 to 7 702 inclusive. See arch/mips/include/asm/pgtable-bits.h 703 for platform specific values (SB1, Loongson3 and 704 others). 705 706 ccw_timeout_log [S390] 707 See Documentation/arch/s390/common_io.rst for details. 708 709 cfi= [X86-64] Set Control Flow Integrity checking features 710 when CONFIG_FINEIBT is enabled. 711 Format: feature[,feature...] 712 Default: auto 713 714 auto: Use FineIBT if IBT available, otherwise kCFI. 715 Under FineIBT, enable "paranoid" mode when 716 FRED is not available. 717 off: Turn off CFI checking. 718 kcfi: Use kCFI (disable FineIBT). 719 fineibt: Use FineIBT (even if IBT not available). 720 norand: Do not re-randomize CFI hashes. 721 paranoid: Add caller hash checking under FineIBT. 722 bhi: Enable register poisoning to stop speculation 723 across FineIBT. (Disabled by default.) 724 warn: Do not enforce CFI checking: warn only. 725 debug: Report CFI initialization details. 726 727 cgroup_disable= [KNL] Disable a particular controller or optional feature 728 Format: {name of the controller(s) or feature(s) to disable} 729 The effects of cgroup_disable=foo are: 730 - foo isn't auto-mounted if you mount all cgroups in 731 a single hierarchy 732 - foo isn't visible as an individually mountable 733 subsystem 734 - if foo is an optional feature then the feature is 735 disabled and corresponding cgroup files are not 736 created 737 {Currently only "memory" controller deal with this and 738 cut the overhead, others just disable the usage. So 739 only cgroup_disable=memory is actually worthy} 740 Specifying "pressure" disables per-cgroup pressure 741 stall information accounting feature 742 743 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1 744 Format: { { controller | "all" | "named" } 745 [,{ controller | "all" | "named" }...] } 746 Like cgroup_disable, but only applies to cgroup v1; 747 the blacklisted controllers remain available in cgroup2. 748 "all" blacklists all controllers and "named" disables 749 named mounts. Specifying both "all" and "named" disables 750 all v1 hierarchies. 751 752 cgroup_v1_proc= [KNL] Show also missing controllers in /proc/cgroups 753 Format: { "true" | "false" } 754 /proc/cgroups lists only v1 controllers by default. 755 This compatibility option enables listing also v2 756 controllers (whose v1 code is not compiled!), so that 757 semi-legacy software can check this file to decide 758 about usage of v2 (sic) controllers. 759 760 cgroup_favordynmods= [KNL] Enable or Disable favordynmods. 761 Format: { "true" | "false" } 762 Defaults to the value of CONFIG_CGROUP_FAVOR_DYNMODS. 763 764 cgroup.memory= [KNL] Pass options to the cgroup memory controller. 765 Format: <string> 766 nosocket -- Disable socket memory accounting. 767 nokmem -- Disable kernel memory accounting. 768 nobpf -- Disable BPF memory accounting. 769 770 checkreqprot= [SELINUX] Set initial checkreqprot flag value. 771 Format: { "0" | "1" } 772 See security/selinux/Kconfig help text. 773 0 -- check protection applied by kernel (includes 774 any implied execute protection). 775 1 -- check protection requested by application. 776 Default value is set via a kernel config option. 777 Value can be changed at runtime via 778 /sys/fs/selinux/checkreqprot. 779 Setting checkreqprot to 1 is deprecated. 780 781 cio_ignore= [S390] 782 See Documentation/arch/s390/common_io.rst for details. 783 784 clearcpuid=X[,X...] [X86] 785 Disable CPUID feature X for the kernel. See 786 arch/x86/include/asm/cpufeatures.h for the valid bit 787 numbers X. Note the Linux-specific bits are not necessarily 788 stable over kernel options, but the vendor-specific 789 ones should be. 790 X can also be a string as appearing in the flags: line 791 in /proc/cpuinfo which does not have the above 792 instability issue. However, not all features have names 793 in /proc/cpuinfo. 794 Note that using this option will taint your kernel. 795 Also note that user programs calling CPUID directly 796 or using the feature without checking anything 797 will still see it. This just prevents it from 798 being used by the kernel or shown in /proc/cpuinfo. 799 Also note the kernel might malfunction if you disable 800 some critical bits. 801 802 clk_ignore_unused 803 [CLK] 804 Prevents the clock framework from automatically gating 805 clocks that have not been explicitly enabled by a Linux 806 device driver but are enabled in hardware at reset or 807 by the bootloader/firmware. Note that this does not 808 force such clocks to be always-on nor does it reserve 809 those clocks in any way. This parameter is useful for 810 debug and development, but should not be needed on a 811 platform with proper driver support. For more 812 information, see Documentation/driver-api/clk.rst. 813 814 clock= [BUGS=X86-32, HW] gettimeofday clocksource override. 815 [Deprecated] 816 Forces specified clocksource (if available) to be used 817 when calculating gettimeofday(). If specified 818 clocksource is not available, it defaults to PIT. 819 Format: { pit | tsc | cyclone | pmtmr } 820 821 clocksource= Override the default clocksource 822 Format: <string> 823 Override the default clocksource and use the clocksource 824 with the name specified. 825 Some clocksource names to choose from, depending on 826 the platform: 827 [all] jiffies (this is the base, fallback clocksource) 828 [ACPI] acpi_pm 829 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2, 830 pxa_timer,timer3,32k_counter,timer0_1 831 [X86-32] pit,hpet,tsc; 832 scx200_hrt on Geode; cyclone on IBM x440 833 [MIPS] MIPS 834 [PARISC] cr16 835 [S390] tod 836 [SH] SuperH 837 [SPARC64] tick 838 [X86-64] hpet,tsc 839 840 clocksource.arm_arch_timer.evtstrm= 841 [ARM,ARM64,EARLY] 842 Format: <bool> 843 Enable/disable the eventstream feature of the ARM 844 architected timer so that code using WFE-based polling 845 loops can be debugged more effectively on production 846 systems. 847 848 clocksource.verify_n_cpus= [KNL] 849 Limit the number of CPUs checked for clocksources 850 marked with CLOCK_SOURCE_VERIFY_PERCPU that 851 are marked unstable due to excessive skew. 852 A negative value says to check all CPUs, while 853 zero says not to check any. Values larger than 854 nr_cpu_ids are silently truncated to nr_cpu_ids. 855 The actual CPUs are chosen randomly, with 856 no replacement if the same CPU is chosen twice. 857 858 clocksource-wdtest.holdoff= [KNL] 859 Set the time in seconds that the clocksource 860 watchdog test waits before commencing its tests. 861 Defaults to zero when built as a module and to 862 10 seconds when built into the kernel. 863 864 cma=nn[MG]@[start[MG][-end[MG]]] 865 [KNL,CMA,EARLY] 866 Sets the size of kernel global memory area for 867 contiguous memory allocations and optionally the 868 placement constraint by the physical address range of 869 memory allocations. A value of 0 disables CMA 870 altogether. For more information, see 871 kernel/dma/contiguous.c 872 873 cma_pernuma=nn[MG] 874 [KNL,CMA,EARLY] 875 Sets the size of kernel per-numa memory area for 876 contiguous memory allocations. A value of 0 disables 877 per-numa CMA altogether. And If this option is not 878 specified, the default value is 0. 879 With per-numa CMA enabled, DMA users on node nid will 880 first try to allocate buffer from the pernuma area 881 which is located in node nid, if the allocation fails, 882 they will fallback to the global default memory area. 883 884 numa_cma=<node>:nn[MG][,<node>:nn[MG]] 885 [KNL,CMA,EARLY] 886 Sets the size of kernel numa memory area for 887 contiguous memory allocations. It will reserve CMA 888 area for the specified node. 889 890 With numa CMA enabled, DMA users on node nid will 891 first try to allocate buffer from the numa area 892 which is located in node nid, if the allocation fails, 893 they will fallback to the global default memory area. 894 895 cmo_free_hint= [PPC] Format: { yes | no } 896 Specify whether pages are marked as being inactive 897 when they are freed. This is used in CMO environments 898 to determine OS memory pressure for page stealing by 899 a hypervisor. 900 Default: yes 901 902 coherent_pool=nn[KMG] [ARM,KNL,EARLY] 903 Sets the size of memory pool for coherent, atomic dma 904 allocations, by default set to 256K. 905 906 com20020= [HW,NET] ARCnet - COM20020 chipset 907 Format: 908 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]] 909 910 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers) 911 Format: <io>[,<irq>] 912 913 com90xx= [HW,NET] 914 ARCnet - COM90xx chipset (memory-mapped buffers) 915 Format: <io>[,<irq>[,<memstart>]] 916 917 condev= [HW,S390] console device 918 conmode= 919 920 con3215_drop= [S390,EARLY] 3215 console drop mode. 921 Format: y|n|Y|N|1|0 922 When set to true, drop data on the 3215 console when 923 the console buffer is full. In this case the 924 operator using a 3270 terminal emulator (for example 925 x3270) does not have to enter the clear key for the 926 console output to advance and the kernel to continue. 927 This leads to a much faster boot time when a 3270 928 terminal emulator is active. If no 3270 terminal 929 emulator is used, this parameter has no effect. 930 931 console= [KNL] Output console device and options. 932 933 tty<n> Use the virtual console device <n>. 934 935 ttyS<n>[,options] 936 ttyUSB0[,options] 937 Use the specified serial port. The options are of 938 the form "bbbbpnf", where "bbbb" is the baud rate, 939 "p" is parity ("n", "o", or "e"), "n" is number of 940 bits, and "f" is flow control ("r" for RTS or 941 omit it). Default is "9600n8". 942 943 See Documentation/admin-guide/serial-console.rst for more 944 information. See 945 Documentation/networking/netconsole.rst for an 946 alternative. 947 948 <DEVNAME>:<n>.<n>[,options] 949 Use the specified serial port on the serial core bus. 950 The addressing uses DEVNAME of the physical serial port 951 device, followed by the serial core controller instance, 952 and the serial port instance. The options are the same 953 as documented for the ttyS addressing above. 954 955 The mapping of the serial ports to the tty instances 956 can be viewed with: 957 958 $ ls -d /sys/bus/serial-base/devices/*:*.*/tty/* 959 /sys/bus/serial-base/devices/00:04:0.0/tty/ttyS0 960 961 In the above example, the console can be addressed with 962 console=00:04:0.0. Note that a console addressed this 963 way will only get added when the related device driver 964 is ready. The use of an earlycon parameter in addition to 965 the console may be desired for console output early on. 966 967 uart[8250],io,<addr>[,options] 968 uart[8250],mmio,<addr>[,options] 969 uart[8250],mmio16,<addr>[,options] 970 uart[8250],mmio32,<addr>[,options] 971 uart[8250],0x<addr>[,options] 972 Start an early, polled-mode console on the 8250/16550 973 UART at the specified I/O port or MMIO address, 974 switching to the matching ttyS device later. 975 MMIO inter-register address stride is either 8-bit 976 (mmio), 16-bit (mmio16), or 32-bit (mmio32). 977 If none of [io|mmio|mmio16|mmio32], <addr> is assumed 978 to be equivalent to 'mmio'. 'options' are specified in 979 the same format described for ttyS above; if unspecified, 980 the h/w is not re-initialized. 981 982 hvc<n> Use the hypervisor console device <n>. This is for 983 both Xen and PowerPC hypervisors. 984 985 { null | "" } 986 Use to disable console output, i.e., to have kernel 987 console messages discarded. 988 This must be the only console= parameter used on the 989 kernel command line. 990 991 If the device connected to the port is not a TTY but a braille 992 device, prepend "brl," before the device type, for instance 993 console=brl,ttyS0 994 For now, only VisioBraille is supported. 995 996 console_msg_format= 997 [KNL] Change console messages format 998 default 999 By default we print messages on consoles in 1000 "[time stamp] text\n" format (time stamp may not be 1001 printed, depending on CONFIG_PRINTK_TIME or 1002 `printk_time' param). 1003 syslog 1004 Switch to syslog format: "<%u>[time stamp] text\n" 1005 IOW, each message will have a facility and loglevel 1006 prefix. The format is similar to one used by syslog() 1007 syscall, or to executing "dmesg -S --raw" or to reading 1008 from /proc/kmsg. 1009 1010 consoleblank= [KNL] The console blank (screen saver) timeout in 1011 seconds. A value of 0 disables the blank timer. 1012 Defaults to 0. 1013 1014 coredump_filter= 1015 [KNL] Change the default value for 1016 /proc/<pid>/coredump_filter. 1017 See also Documentation/filesystems/proc.rst. 1018 1019 coresight_cpu_debug.enable 1020 [ARM,ARM64] 1021 Format: <bool> 1022 Enable/disable the CPU sampling based debugging. 1023 0: default value, disable debugging 1024 1: enable debugging at boot time 1025 1026 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver 1027 Format: 1028 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>] 1029 1030 cpuidle.off=1 [CPU_IDLE] 1031 disable the cpuidle sub-system 1032 1033 cpuidle.governor= 1034 [CPU_IDLE] Name of the cpuidle governor to use. 1035 1036 cpufreq.off=1 [CPU_FREQ] 1037 disable the cpufreq sub-system 1038 1039 cpufreq.default_governor= 1040 [CPU_FREQ] Name of the default cpufreq governor or 1041 policy to use. This governor must be registered in the 1042 kernel before the cpufreq driver probes. 1043 1044 cpu_init_udelay=N 1045 [X86,EARLY] Delay for N microsec between assert and de-assert 1046 of APIC INIT to start processors. This delay occurs 1047 on every CPU online, such as boot, and resume from suspend. 1048 Default: 10000 1049 1050 cpuhp.parallel= 1051 [SMP] Enable/disable parallel bringup of secondary CPUs 1052 Format: <bool> 1053 Default is enabled if CONFIG_HOTPLUG_PARALLEL=y. Otherwise 1054 the parameter has no effect. 1055 1056 crash_kexec_post_notifiers 1057 Only jump to kdump kernel after running the panic 1058 notifiers and dumping kmsg. This option increases 1059 the risks of a kdump failure, since some panic 1060 notifiers can make the crashed kernel more unstable. 1061 In configurations where kdump may not be reliable, 1062 running the panic notifiers could allow collecting 1063 more data on dmesg, like stack traces from other CPUS 1064 or extra data dumped by panic_print. Note that some 1065 configurations enable this option unconditionally, 1066 like Hyper-V, PowerPC (fadump) and AMD SEV-SNP. 1067 1068 crashkernel=size[KMG][@offset[KMG]] 1069 [KNL,EARLY] Using kexec, Linux can switch to a 'crash kernel' 1070 upon panic. This parameter reserves the physical 1071 memory region [offset, offset + size] for that kernel 1072 image. If '@offset' is omitted, then a suitable offset 1073 is selected automatically. 1074 [KNL, X86-64, ARM64, RISCV, LoongArch] Select a region 1075 under 4G first, and fall back to reserve region above 1076 4G when '@offset' hasn't been specified. 1077 See Documentation/admin-guide/kdump/kdump.rst for further details. 1078 1079 crashkernel=range1:size1[,range2:size2,...][@offset] 1080 [KNL] Same as above, but depends on the memory 1081 in the running system. The syntax of range is 1082 start-[end] where start and end are both 1083 a memory unit (amount[KMG]). See also 1084 Documentation/admin-guide/kdump/kdump.rst for an example. 1085 1086 crashkernel=size[KMG],high 1087 [KNL, X86-64, ARM64, RISCV, LoongArch] range could be 1088 above 4G. 1089 Allow kernel to allocate physical memory region from top, 1090 so could be above 4G if system have more than 4G ram 1091 installed. Otherwise memory region will be allocated 1092 below 4G, if available. 1093 It will be ignored if crashkernel=X is specified. 1094 crashkernel=size[KMG],low 1095 [KNL, X86-64, ARM64, RISCV, LoongArch] range under 4G. 1096 When crashkernel=X,high is passed, kernel could allocate 1097 physical memory region above 4G, that cause second kernel 1098 crash on system that require some amount of low memory, 1099 e.g. swiotlb requires at least 64M+32K low memory, also 1100 enough extra low memory is needed to make sure DMA buffers 1101 for 32-bit devices won't run out. Kernel would try to allocate 1102 default size of memory below 4G automatically. The default 1103 size is platform dependent. 1104 --> x86: max(swiotlb_size_or_default() + 8MiB, 256MiB) 1105 --> arm64: 128MiB 1106 --> riscv: 128MiB 1107 --> loongarch: 128MiB 1108 This one lets the user specify own low range under 4G 1109 for second kernel instead. 1110 0: to disable low allocation. 1111 It will be ignored when crashkernel=X,high is not used 1112 or memory reserved is below 4G. 1113 crashkernel=size[KMG],cma 1114 [KNL, X86, ppc] Reserve additional crash kernel memory from 1115 CMA. This reservation is usable by the first system's 1116 userspace memory and kernel movable allocations (memory 1117 balloon, zswap). Pages allocated from this memory range 1118 will not be included in the vmcore so this should not 1119 be used if dumping of userspace memory is intended and 1120 it has to be expected that some movable kernel pages 1121 may be missing from the dump. 1122 1123 A standard crashkernel reservation, as described above, 1124 is still needed to hold the crash kernel and initrd. 1125 1126 This option increases the risk of a kdump failure: DMA 1127 transfers configured by the first kernel may end up 1128 corrupting the second kernel's memory. 1129 1130 This reservation method is intended for systems that 1131 can't afford to sacrifice enough memory for standard 1132 crashkernel reservation and where less reliable and 1133 possibly incomplete kdump is preferable to no kdump at 1134 all. 1135 1136 cryptomgr.notests 1137 [KNL] Disable crypto self-tests 1138 1139 cs89x0_dma= [HW,NET] 1140 Format: <dma> 1141 1142 cs89x0_media= [HW,NET] 1143 Format: { rj45 | aui | bnc } 1144 1145 csdlock_debug= [KNL] Enable or disable debug add-ons of cross-CPU 1146 function call handling. When switched on, 1147 additional debug data is printed to the console 1148 in case a hanging CPU is detected, and that 1149 CPU is pinged again in order to try to resolve 1150 the hang situation. The default value of this 1151 option depends on the CSD_LOCK_WAIT_DEBUG_DEFAULT 1152 Kconfig option. 1153 1154 dasd= [HW,NET] 1155 See header of drivers/s390/block/dasd_devmap.c. 1156 1157 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port 1158 (one device per port) 1159 Format: <port#>,<type> 1160 See also Documentation/input/devices/joystick-parport.rst 1161 1162 debug [KNL,EARLY] Enable kernel debugging (events log level). 1163 1164 debug_boot_weak_hash 1165 [KNL,EARLY] Enable printing [hashed] pointers early in the 1166 boot sequence. If enabled, we use a weak hash instead 1167 of siphash to hash pointers. Use this option if you are 1168 seeing instances of '(___ptrval___)') and need to see a 1169 value (hashed pointer) instead. Cryptographically 1170 insecure, please do not use on production kernels. 1171 1172 debug_locks_verbose= 1173 [KNL] verbose locking self-tests 1174 Format: <int> 1175 Print debugging info while doing the locking API 1176 self-tests. 1177 Bitmask for the various LOCKTYPE_ tests. Defaults to 0 1178 (no extra messages), setting it to -1 (all bits set) 1179 will print _a_lot_ more information - normally only 1180 useful to lockdep developers. 1181 1182 debug_objects [KNL,EARLY] Enable object debugging 1183 1184 debug_guardpage_minorder= 1185 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this 1186 parameter allows control of the order of pages that will 1187 be intentionally kept free (and hence protected) by the 1188 buddy allocator. Bigger value increase the probability 1189 of catching random memory corruption, but reduce the 1190 amount of memory for normal system use. The maximum 1191 possible value is MAX_PAGE_ORDER/2. Setting this 1192 parameter to 1 or 2 should be enough to identify most 1193 random memory corruption problems caused by bugs in 1194 kernel or driver code when a CPU writes to (or reads 1195 from) a random memory location. Note that there exists 1196 a class of memory corruptions problems caused by buggy 1197 H/W or F/W or by drivers badly programming DMA 1198 (basically when memory is written at bus level and the 1199 CPU MMU is bypassed) which are not detectable by 1200 CONFIG_DEBUG_PAGEALLOC, hence this option will not 1201 help tracking down these problems. 1202 1203 debug_pagealloc= 1204 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this parameter 1205 enables the feature at boot time. By default, it is 1206 disabled and the system will work mostly the same as a 1207 kernel built without CONFIG_DEBUG_PAGEALLOC. 1208 Note: to get most of debug_pagealloc error reports, it's 1209 useful to also enable the page_owner functionality. 1210 on: enable the feature 1211 1212 debugfs= [KNL,EARLY] This parameter enables what is exposed to 1213 userspace and debugfs internal clients. 1214 Format: { on, off } 1215 on: All functions are enabled. 1216 off: Filesystem is not registered and clients 1217 get a -EPERM as result when trying to register files 1218 or directories within debugfs. 1219 This is equivalent of the runtime functionality if 1220 debugfs was not enabled in the kernel at all. 1221 Default value is set in build-time with a kernel configuration. 1222 1223 debugpat [X86] Enable PAT debugging 1224 1225 default_hugepagesz= 1226 [HW] The size of the default HugeTLB page. This is 1227 the size represented by the legacy /proc/ hugepages 1228 APIs. In addition, this is the default hugetlb size 1229 used for shmget(), mmap() and mounting hugetlbfs 1230 filesystems. If not specified, defaults to the 1231 architecture's default huge page size. Huge page 1232 sizes are architecture dependent. See also 1233 Documentation/admin-guide/mm/hugetlbpage.rst. 1234 Format: size[KMG] 1235 1236 deferred_probe_timeout= 1237 [KNL] Debugging option to set a timeout in seconds for 1238 deferred probe to give up waiting on dependencies to 1239 probe. Only specific dependencies (subsystems or 1240 drivers) that have opted in will be ignored. A timeout 1241 of 0 will timeout at the end of initcalls. If the time 1242 out hasn't expired, it'll be restarted by each 1243 successful driver registration. This option will also 1244 dump out devices still on the deferred probe list after 1245 retrying. 1246 1247 delayacct [KNL] Enable per-task delay accounting 1248 1249 dell_smm_hwmon.ignore_dmi= 1250 [HW] Continue probing hardware even if DMI data 1251 indicates that the driver is running on unsupported 1252 hardware. 1253 1254 dell_smm_hwmon.force= 1255 [HW] Activate driver even if SMM BIOS signature does 1256 not match list of supported models and enable otherwise 1257 blacklisted features. 1258 1259 dell_smm_hwmon.power_status= 1260 [HW] Report power status in /proc/i8k 1261 (disabled by default). 1262 1263 dell_smm_hwmon.restricted= 1264 [HW] Allow controlling fans only if SYS_ADMIN 1265 capability is set. 1266 1267 dell_smm_hwmon.fan_mult= 1268 [HW] Factor to multiply fan speed with. 1269 1270 dell_smm_hwmon.fan_max= 1271 [HW] Maximum configurable fan speed. 1272 1273 dfltcc= [HW,S390] 1274 Format: { on | off | def_only | inf_only | always } 1275 on: s390 zlib hardware support for compression on 1276 level 1 and decompression (default) 1277 off: No s390 zlib hardware support 1278 def_only: s390 zlib hardware support for deflate 1279 only (compression on level 1) 1280 inf_only: s390 zlib hardware support for inflate 1281 only (decompression) 1282 always: Same as 'on' but ignores the selected compression 1283 level always using hardware support (used for debugging) 1284 1285 dhash_entries= [KNL] 1286 Set number of hash buckets for dentry cache. 1287 1288 disable_1tb_segments [PPC,EARLY] 1289 Disables the use of 1TB hash page table segments. This 1290 causes the kernel to fall back to 256MB segments which 1291 can be useful when debugging issues that require an SLB 1292 miss to occur. 1293 1294 disable= [IPV6] 1295 See Documentation/networking/ipv6.rst. 1296 1297 disable_radix [PPC,EARLY] 1298 Disable RADIX MMU mode on POWER9 1299 1300 disable_tlbie [PPC] 1301 Disable TLBIE instruction. Currently does not work 1302 with KVM, with HASH MMU, or with coherent accelerators. 1303 1304 disable_ddw [PPC/PSERIES,EARLY] 1305 Disable Dynamic DMA Window support. Use this 1306 to workaround buggy firmware. 1307 1308 disable_ipv6= [IPV6] 1309 See Documentation/networking/ipv6.rst. 1310 1311 disable_mtrr_cleanup [X86,EARLY] 1312 The kernel tries to adjust MTRR layout from continuous 1313 to discrete, to make X server driver able to add WB 1314 entry later. This parameter disables that. 1315 1316 disable_mtrr_trim [X86, Intel and AMD only,EARLY] 1317 By default the kernel will trim any uncacheable 1318 memory out of your available memory pool based on 1319 MTRR settings. This parameter disables that behavior, 1320 possibly causing your machine to run very slowly. 1321 1322 disable_timer_pin_1 [X86,EARLY] 1323 Disable PIN 1 of APIC timer 1324 Can be useful to work around chipset bugs. 1325 1326 dis_ucode_ldr [X86] Disable the microcode loader. 1327 1328 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support, 1329 this option disables the debugging code at boot. 1330 1331 dma_debug_entries=<number> 1332 This option allows to tune the number of preallocated 1333 entries for DMA-API debugging code. One entry is 1334 required per DMA-API allocation. Use this if the 1335 DMA-API debugging code disables itself because the 1336 architectural default is too low. 1337 1338 dma_debug_driver=<driver_name> 1339 With this option the DMA-API debugging driver 1340 filter feature can be enabled at boot time. Just 1341 pass the driver to filter for as the parameter. 1342 The filter can be disabled or changed to another 1343 driver later using sysfs. 1344 1345 reg_file_data_sampling= 1346 [X86] Controls mitigation for Register File Data 1347 Sampling (RFDS) vulnerability. RFDS is a CPU 1348 vulnerability which may allow userspace to infer 1349 kernel data values previously stored in floating point 1350 registers, vector registers, or integer registers. 1351 RFDS only affects Intel Atom processors. 1352 1353 on: Turns ON the mitigation. 1354 off: Turns OFF the mitigation. 1355 1356 This parameter overrides the compile time default set 1357 by CONFIG_MITIGATION_RFDS. Mitigation cannot be 1358 disabled when other VERW based mitigations (like MDS) 1359 are enabled. In order to disable RFDS mitigation all 1360 VERW based mitigations need to be disabled. 1361 1362 For details see: 1363 Documentation/admin-guide/hw-vuln/reg-file-data-sampling.rst 1364 1365 driver_async_probe= [KNL] 1366 List of driver names to be probed asynchronously. * 1367 matches with all driver names. If * is specified, the 1368 rest of the listed driver names are those that will NOT 1369 match the *. 1370 Format: <driver_name1>,<driver_name2>... 1371 1372 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>] 1373 Broken monitors, graphic adapters, KVMs and EDIDless 1374 panels may send no or incorrect EDID data sets. 1375 This parameter allows to specify an EDID data sets 1376 in the /lib/firmware directory that are used instead. 1377 An EDID data set will only be used for a particular 1378 connector, if its name and a colon are prepended to 1379 the EDID name. Each connector may use a unique EDID 1380 data set by separating the files with a comma. An EDID 1381 data set with no connector name will be used for 1382 any connectors not explicitly specified. 1383 1384 dscc4.setup= [NET] 1385 1386 dt_cpu_ftrs= [PPC,EARLY] 1387 Format: {"off" | "known"} 1388 Control how the dt_cpu_ftrs device-tree binding is 1389 used for CPU feature discovery and setup (if it 1390 exists). 1391 off: Do not use it, fall back to legacy cpu table. 1392 known: Do not pass through unknown features to guests 1393 or userspace, only those that the kernel is aware of. 1394 1395 dump_apple_properties [X86] 1396 Dump name and content of EFI device properties on 1397 x86 Macs. Useful for driver authors to determine 1398 what data is available or for reverse-engineering. 1399 1400 dyndbg[="val"] [KNL,DYNAMIC_DEBUG] 1401 <module>.dyndbg[="val"] 1402 Enable debug messages at boot time. See 1403 Documentation/admin-guide/dynamic-debug-howto.rst 1404 for details. 1405 1406 early_ioremap_debug [KNL,EARLY] 1407 Enable debug messages in early_ioremap support. This 1408 is useful for tracking down temporary early mappings 1409 which are not unmapped. 1410 1411 earlycon= [KNL,EARLY] Output early console device and options. 1412 1413 When used with no options, the early console is 1414 determined by stdout-path property in device tree's 1415 chosen node or the ACPI SPCR table if supported by 1416 the platform. 1417 1418 cdns,<addr>[,options] 1419 Start an early, polled-mode console on a Cadence 1420 (xuartps) serial port at the specified address. Only 1421 supported option is baud rate. If baud rate is not 1422 specified, the serial port must already be setup and 1423 configured. 1424 1425 uart[8250],io,<addr>[,options[,uartclk]] 1426 uart[8250],mmio,<addr>[,options[,uartclk]] 1427 uart[8250],mmio32,<addr>[,options[,uartclk]] 1428 uart[8250],mmio32be,<addr>[,options[,uartclk]] 1429 uart[8250],0x<addr>[,options] 1430 Start an early, polled-mode console on the 8250/16550 1431 UART at the specified I/O port or MMIO address. 1432 MMIO inter-register address stride is either 8-bit 1433 (mmio) or 32-bit (mmio32 or mmio32be). 1434 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed 1435 to be equivalent to 'mmio'. 'options' are specified 1436 in the same format described for "console=ttyS<n>"; if 1437 unspecified, the h/w is not initialized. 'uartclk' is 1438 the uart clock frequency; if unspecified, it is set 1439 to 'BASE_BAUD' * 16. 1440 1441 pl011,<addr> 1442 pl011,mmio32,<addr> 1443 Start an early, polled-mode console on a pl011 serial 1444 port at the specified address. The pl011 serial port 1445 must already be setup and configured. Options are not 1446 yet supported. If 'mmio32' is specified, then only 1447 the driver will use only 32-bit accessors to read/write 1448 the device registers. 1449 1450 liteuart,<addr> 1451 Start an early console on a litex serial port at the 1452 specified address. The serial port must already be 1453 setup and configured. Options are not yet supported. 1454 1455 meson,<addr> 1456 Start an early, polled-mode console on a meson serial 1457 port at the specified address. The serial port must 1458 already be setup and configured. Options are not yet 1459 supported. 1460 1461 msm_serial,<addr> 1462 Start an early, polled-mode console on an msm serial 1463 port at the specified address. The serial port 1464 must already be setup and configured. Options are not 1465 yet supported. 1466 1467 msm_serial_dm,<addr> 1468 Start an early, polled-mode console on an msm serial 1469 dm port at the specified address. The serial port 1470 must already be setup and configured. Options are not 1471 yet supported. 1472 1473 owl,<addr> 1474 Start an early, polled-mode console on a serial port 1475 of an Actions Semi SoC, such as S500 or S900, at the 1476 specified address. The serial port must already be 1477 setup and configured. Options are not yet supported. 1478 1479 rda,<addr> 1480 Start an early, polled-mode console on a serial port 1481 of an RDA Micro SoC, such as RDA8810PL, at the 1482 specified address. The serial port must already be 1483 setup and configured. Options are not yet supported. 1484 1485 sbi 1486 Use RISC-V SBI (Supervisor Binary Interface) for early 1487 console. 1488 1489 smh Use ARM semihosting calls for early console. 1490 1491 s3c2410,<addr> 1492 s3c2412,<addr> 1493 s3c2440,<addr> 1494 s3c6400,<addr> 1495 s5pv210,<addr> 1496 exynos4210,<addr> 1497 Use early console provided by serial driver available 1498 on Samsung SoCs, requires selecting proper type and 1499 a correct base address of the selected UART port. The 1500 serial port must already be setup and configured. 1501 Options are not yet supported. 1502 1503 lantiq,<addr> 1504 Start an early, polled-mode console on a lantiq serial 1505 (lqasc) port at the specified address. The serial port 1506 must already be setup and configured. Options are not 1507 yet supported. 1508 1509 lpuart,<addr> 1510 lpuart32,<addr> 1511 Use early console provided by Freescale LP UART driver 1512 found on Freescale Vybrid and QorIQ LS1021A processors. 1513 A valid base address must be provided, and the serial 1514 port must already be setup and configured. 1515 1516 ec_imx21,<addr> 1517 ec_imx6q,<addr> 1518 Start an early, polled-mode, output-only console on the 1519 Freescale i.MX UART at the specified address. The UART 1520 must already be setup and configured. 1521 1522 ar3700_uart,<addr> 1523 Start an early, polled-mode console on the 1524 Armada 3700 serial port at the specified 1525 address. The serial port must already be setup 1526 and configured. Options are not yet supported. 1527 1528 qcom_geni,<addr> 1529 Start an early, polled-mode console on a Qualcomm 1530 Generic Interface (GENI) based serial port at the 1531 specified address. The serial port must already be 1532 setup and configured. Options are not yet supported. 1533 1534 efifb,[options] 1535 Start an early, unaccelerated console on the EFI 1536 memory mapped framebuffer (if available). On cache 1537 coherent non-x86 systems that use system memory for 1538 the framebuffer, pass the 'ram' option so that it is 1539 mapped with the correct attributes. 1540 1541 linflex,<addr> 1542 Use early console provided by Freescale LINFlexD UART 1543 serial driver for NXP S32V234 SoCs. A valid base 1544 address must be provided, and the serial port must 1545 already be setup and configured. 1546 1547 earlyprintk= [X86,SH,ARM,M68k,S390,UM,EARLY] 1548 earlyprintk=vga 1549 earlyprintk=sclp 1550 earlyprintk=xen 1551 earlyprintk=serial[,ttySn[,baudrate]] 1552 earlyprintk=serial[,0x...[,baudrate]] 1553 earlyprintk=ttySn[,baudrate] 1554 earlyprintk=dbgp[debugController#] 1555 earlyprintk=mmio32,membase[,{nocfg|baudrate}] 1556 earlyprintk=pciserial[,force],bus:device.function[,{nocfg|baudrate}] 1557 earlyprintk=xdbc[xhciController#] 1558 earlyprintk=bios 1559 1560 earlyprintk is useful when the kernel crashes before 1561 the normal console is initialized. It is not enabled by 1562 default because it has some cosmetic problems. 1563 1564 Use "nocfg" to skip UART configuration, assume 1565 BIOS/firmware has configured UART correctly. 1566 1567 Append ",keep" to not disable it when the real console 1568 takes over. 1569 1570 Only one of vga, serial, or usb debug port can 1571 be used at a time. 1572 1573 Currently only ttyS0 and ttyS1 may be specified by 1574 name. Other I/O ports may be explicitly specified 1575 on some architectures (x86 and arm at least) by 1576 replacing ttySn with an I/O port address, like this: 1577 earlyprintk=serial,0x1008,115200 1578 You can find the port for a given device in 1579 /proc/tty/driver/serial: 1580 2: uart:ST16650V2 port:00001008 irq:18 ... 1581 1582 Interaction with the standard serial driver is not 1583 very good. 1584 1585 The VGA output is eventually overwritten by 1586 the real console. 1587 1588 The xen option can only be used in Xen domains. 1589 1590 The sclp output can only be used on s390. 1591 1592 The bios output can only be used on SuperH. 1593 1594 The optional "force" to "pciserial" enables use of a 1595 PCI device even when its classcode is not of the 1596 UART class. 1597 1598 edac_report= [HW,EDAC] Control how to report EDAC event 1599 Format: {"on" | "off" | "force"} 1600 on: enable EDAC to report H/W event. May be overridden 1601 by other higher priority error reporting module. 1602 off: disable H/W event reporting through EDAC. 1603 force: enforce the use of EDAC to report H/W event. 1604 default: on. 1605 1606 edd= [EDD] 1607 Format: {"off" | "on" | "skip[mbr]"} 1608 1609 efi= [EFI,EARLY] 1610 Format: { "debug", "disable_early_pci_dma", 1611 "nochunk", "noruntime", "nosoftreserve", 1612 "novamap", "no_disable_early_pci_dma" } 1613 debug: enable misc debug output. 1614 disable_early_pci_dma: disable the busmaster bit on all 1615 PCI bridges while in the EFI boot stub. 1616 nochunk: disable reading files in "chunks" in the EFI 1617 boot stub, as chunking can cause problems with some 1618 firmware implementations. 1619 noruntime : disable EFI runtime services support 1620 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose) 1621 attribute may cause the kernel to reserve the 1622 memory range for a memory mapping driver to 1623 claim. Specify efi=nosoftreserve to disable this 1624 reservation and treat the memory by its base type 1625 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM"). 1626 novamap: do not call SetVirtualAddressMap(). 1627 no_disable_early_pci_dma: Leave the busmaster bit set 1628 on all PCI bridges while in the EFI boot stub 1629 1630 efi_no_storage_paranoia [EFI,X86,EARLY] 1631 Using this parameter you can use more than 50% of 1632 your efi variable storage. Use this parameter only if 1633 you are really sure that your UEFI does sane gc and 1634 fulfills the spec otherwise your board may brick. 1635 1636 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT 1637 that is to be dynamically loaded by Linux. If there are 1638 multiple variables with the same name but with different 1639 vendor GUIDs, all of them will be loaded. See 1640 Documentation/admin-guide/acpi/ssdt-overlays.rst for details. 1641 1642 1643 eisa_irq_edge= [PARISC,HW] 1644 See header of drivers/parisc/eisa.c. 1645 1646 ekgdboc= [X86,KGDB,EARLY] Allow early kernel console debugging 1647 Format: ekgdboc=kbd 1648 1649 This is designed to be used in conjunction with 1650 the boot argument: earlyprintk=vga 1651 1652 This parameter works in place of the kgdboc parameter 1653 but can only be used if the backing tty is available 1654 very early in the boot process. For early debugging 1655 via a serial port see kgdboc_earlycon instead. 1656 1657 elanfreq= [X86-32] 1658 See comment before function elanfreq_setup() in 1659 arch/x86/kernel/cpu/cpufreq/elanfreq.c. 1660 1661 elfcorehdr=[size[KMG]@]offset[KMG] [PPC,SH,X86,S390,EARLY] 1662 Specifies physical address of start of kernel core 1663 image elf header and optionally the size. Generally 1664 kexec loader will pass this option to capture kernel. 1665 See Documentation/admin-guide/kdump/kdump.rst for details. 1666 1667 enable_mtrr_cleanup [X86,EARLY] 1668 The kernel tries to adjust MTRR layout from continuous 1669 to discrete, to make X server driver able to add WB 1670 entry later. This parameter enables that. 1671 1672 enable_timer_pin_1 [X86] 1673 Enable PIN 1 of APIC timer 1674 Can be useful to work around chipset bugs 1675 (in particular on some ATI chipsets). 1676 The kernel tries to set a reasonable default. 1677 1678 enforcing= [SELINUX] Set initial enforcing status. 1679 Format: {"0" | "1"} 1680 See security/selinux/Kconfig help text. 1681 0 -- permissive (log only, no denials). 1682 1 -- enforcing (deny and log). 1683 Default value is 0. 1684 Value can be changed at runtime via 1685 /sys/fs/selinux/enforce. 1686 1687 erst_disable [ACPI] 1688 Disable Error Record Serialization Table (ERST) 1689 support. 1690 1691 ether= [HW,NET] Ethernet cards parameters 1692 This option is obsoleted by the "netdev=" option, which 1693 has equivalent usage. See its documentation for details. 1694 1695 evm= [EVM] 1696 Format: { "fix" } 1697 Permit 'security.evm' to be updated regardless of 1698 current integrity status. 1699 1700 early_page_ext [KNL,EARLY] Enforces page_ext initialization to earlier 1701 stages so cover more early boot allocations. 1702 Please note that as side effect some optimizations 1703 might be disabled to achieve that (e.g. parallelized 1704 memory initialization is disabled) so the boot process 1705 might take longer, especially on systems with a lot of 1706 memory. Available with CONFIG_PAGE_EXTENSION=y. 1707 1708 failslab= 1709 fail_usercopy= 1710 fail_page_alloc= 1711 fail_skb_realloc= 1712 fail_make_request=[KNL] 1713 General fault injection mechanism. 1714 Format: <interval>,<probability>,<space>,<times> 1715 See also Documentation/fault-injection/. 1716 1717 fb_tunnels= [NET] 1718 Format: { initns | none } 1719 See Documentation/admin-guide/sysctl/net.rst for 1720 fb_tunnels_only_for_init_ns 1721 1722 floppy= [HW] 1723 See Documentation/admin-guide/blockdev/floppy.rst. 1724 1725 forcepae [X86-32] 1726 Forcefully enable Physical Address Extension (PAE). 1727 Many Pentium M systems disable PAE but may have a 1728 functionally usable PAE implementation. 1729 Warning: use of this parameter will taint the kernel 1730 and may cause unknown problems. 1731 1732 fred= [X86-64] 1733 Enable/disable Flexible Return and Event Delivery. 1734 Format: { on | off } 1735 on: enable FRED when it's present. 1736 off: disable FRED, the default setting. 1737 1738 ftrace=[tracer] 1739 [FTRACE] will set and start the specified tracer 1740 as early as possible in order to facilitate early 1741 boot debugging. 1742 1743 ftrace_boot_snapshot 1744 [FTRACE] On boot up, a snapshot will be taken of the 1745 ftrace ring buffer that can be read at: 1746 /sys/kernel/tracing/snapshot. 1747 This is useful if you need tracing information from kernel 1748 boot up that is likely to be overridden by user space 1749 start up functionality. 1750 1751 Optionally, the snapshot can also be defined for a tracing 1752 instance that was created by the trace_instance= command 1753 line parameter. 1754 1755 trace_instance=foo,sched_switch ftrace_boot_snapshot=foo 1756 1757 The above will cause the "foo" tracing instance to trigger 1758 a snapshot at the end of boot up. 1759 1760 ftrace_dump_on_oops[=2(orig_cpu) | =<instance>][,<instance> | 1761 ,<instance>=2(orig_cpu)] 1762 [FTRACE] will dump the trace buffers on oops. 1763 If no parameter is passed, ftrace will dump global 1764 buffers of all CPUs, if you pass 2 or orig_cpu, it 1765 will dump only the buffer of the CPU that triggered 1766 the oops, or the specific instance will be dumped if 1767 its name is passed. Multiple instance dump is also 1768 supported, and instances are separated by commas. Each 1769 instance supports only dump on CPU that triggered the 1770 oops by passing 2 or orig_cpu to it. 1771 1772 ftrace_dump_on_oops=foo=orig_cpu 1773 1774 The above will dump only the buffer of "foo" instance 1775 on CPU that triggered the oops. 1776 1777 ftrace_dump_on_oops,foo,bar=orig_cpu 1778 1779 The above will dump global buffer on all CPUs, the 1780 buffer of "foo" instance on all CPUs and the buffer 1781 of "bar" instance on CPU that triggered the oops. 1782 1783 ftrace_filter=[function-list] 1784 [FTRACE] Limit the functions traced by the function 1785 tracer at boot up. function-list is a comma-separated 1786 list of functions. This list can be changed at run 1787 time by the set_ftrace_filter file in the debugfs 1788 tracing directory. 1789 1790 ftrace_notrace=[function-list] 1791 [FTRACE] Do not trace the functions specified in 1792 function-list. This list can be changed at run time 1793 by the set_ftrace_notrace file in the debugfs 1794 tracing directory. 1795 1796 ftrace_graph_filter=[function-list] 1797 [FTRACE] Limit the top level callers functions traced 1798 by the function graph tracer at boot up. 1799 function-list is a comma-separated list of functions 1800 that can be changed at run time by the 1801 set_graph_function file in the debugfs tracing directory. 1802 1803 ftrace_graph_notrace=[function-list] 1804 [FTRACE] Do not trace from the functions specified in 1805 function-list. This list is a comma-separated list of 1806 functions that can be changed at run time by the 1807 set_graph_notrace file in the debugfs tracing directory. 1808 1809 ftrace_graph_max_depth=<uint> 1810 [FTRACE] Used with the function graph tracer. This is 1811 the max depth it will trace into a function. This value 1812 can be changed at run time by the max_graph_depth file 1813 in the tracefs tracing directory. default: 0 (no limit) 1814 1815 fw_devlink= [KNL,EARLY] Create device links between consumer and supplier 1816 devices by scanning the firmware to infer the 1817 consumer/supplier relationships. This feature is 1818 especially useful when drivers are loaded as modules as 1819 it ensures proper ordering of tasks like device probing 1820 (suppliers first, then consumers), supplier boot state 1821 clean up (only after all consumers have probed), 1822 suspend/resume & runtime PM (consumers first, then 1823 suppliers). 1824 Format: { off | permissive | on | rpm } 1825 off -- Don't create device links from firmware info. 1826 permissive -- Create device links from firmware info 1827 but use it only for ordering boot state clean 1828 up (sync_state() calls). 1829 on -- Create device links from firmware info and use it 1830 to enforce probe and suspend/resume ordering. 1831 rpm -- Like "on", but also use to order runtime PM. 1832 1833 fw_devlink.strict=<bool> 1834 [KNL,EARLY] Treat all inferred dependencies as mandatory 1835 dependencies. This only applies for fw_devlink=on|rpm. 1836 Format: <bool> 1837 1838 fw_devlink.sync_state = 1839 [KNL,EARLY] When all devices that could probe have finished 1840 probing, this parameter controls what to do with 1841 devices that haven't yet received their sync_state() 1842 calls. 1843 Format: { strict | timeout } 1844 strict -- Default. Continue waiting on consumers to 1845 probe successfully. 1846 timeout -- Give up waiting on consumers and call 1847 sync_state() on any devices that haven't yet 1848 received their sync_state() calls after 1849 deferred_probe_timeout has expired or by 1850 late_initcall() if !CONFIG_MODULES. 1851 1852 gamecon.map[2|3]= 1853 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad 1854 support via parallel port (up to 5 devices per port) 1855 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5> 1856 See also Documentation/input/devices/joystick-parport.rst 1857 1858 gamma= [HW,DRM] 1859 1860 gart_fix_e820= [X86-64,EARLY] disable the fix e820 for K8 GART 1861 Format: off | on 1862 default: on 1863 1864 gather_data_sampling= 1865 [X86,INTEL,EARLY] Control the Gather Data Sampling (GDS) 1866 mitigation. 1867 1868 Gather Data Sampling is a hardware vulnerability which 1869 allows unprivileged speculative access to data which was 1870 previously stored in vector registers. 1871 1872 This issue is mitigated by default in updated microcode. 1873 The mitigation may have a performance impact but can be 1874 disabled. On systems without the microcode mitigation 1875 disabling AVX serves as a mitigation. 1876 1877 force: Disable AVX to mitigate systems without 1878 microcode mitigation. No effect if the microcode 1879 mitigation is present. Known to cause crashes in 1880 userspace with buggy AVX enumeration. 1881 1882 off: Disable GDS mitigation. 1883 1884 gbpages [X86] Use GB pages for kernel direct mappings. 1885 1886 gcov_persist= [GCOV] When non-zero (default), profiling data for 1887 kernel modules is saved and remains accessible via 1888 debugfs, even when the module is unloaded/reloaded. 1889 When zero, profiling data is discarded and associated 1890 debugfs files are removed at module unload time. 1891 1892 goldfish [X86] Enable the goldfish android emulator platform. 1893 Don't use this when you are not running on the 1894 android emulator 1895 1896 gpio-mockup.gpio_mockup_ranges 1897 [HW] Sets the ranges of gpiochip of for this device. 1898 Format: <start1>,<end1>,<start2>,<end2>... 1899 gpio-mockup.gpio_mockup_named_lines 1900 [HW] Let the driver know GPIO lines should be named. 1901 1902 gpt [EFI] Forces disk with valid GPT signature but 1903 invalid Protective MBR to be treated as GPT. If the 1904 primary GPT is corrupted, it enables the backup/alternate 1905 GPT to be used instead. 1906 1907 grcan.enable0= [HW] Configuration of physical interface 0. Determines 1908 the "Enable 0" bit of the configuration register. 1909 Format: 0 | 1 1910 Default: 0 1911 grcan.enable1= [HW] Configuration of physical interface 1. Determines 1912 the "Enable 0" bit of the configuration register. 1913 Format: 0 | 1 1914 Default: 0 1915 grcan.select= [HW] Select which physical interface to use. 1916 Format: 0 | 1 1917 Default: 0 1918 grcan.txsize= [HW] Sets the size of the tx buffer. 1919 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0. 1920 Default: 1024 1921 grcan.rxsize= [HW] Sets the size of the rx buffer. 1922 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0. 1923 Default: 1024 1924 1925 hardened_usercopy= 1926 [KNL] Under CONFIG_HARDENED_USERCOPY, whether 1927 hardening is enabled for this boot. Hardened 1928 usercopy checking is used to protect the kernel 1929 from reading or writing beyond known memory 1930 allocation boundaries as a proactive defense 1931 against bounds-checking flaws in the kernel's 1932 copy_to_user()/copy_from_user() interface. 1933 The default is determined by 1934 CONFIG_HARDENED_USERCOPY_DEFAULT_ON. 1935 on Perform hardened usercopy checks. 1936 off Disable hardened usercopy checks. 1937 1938 hardlockup_all_cpu_backtrace= 1939 [KNL] Should the hard-lockup detector generate 1940 backtraces on all cpus. 1941 Format: 0 | 1 1942 1943 hash_pointers= 1944 [KNL,EARLY] 1945 By default, when pointers are printed to the console 1946 or buffers via the %p format string, that pointer is 1947 "hashed", i.e. obscured by hashing the pointer value. 1948 This is a security feature that hides actual kernel 1949 addresses from unprivileged users, but it also makes 1950 debugging the kernel more difficult since unequal 1951 pointers can no longer be compared. The choices are: 1952 Format: { auto | always | never } 1953 Default: auto 1954 1955 auto - Hash pointers unless slab_debug is enabled. 1956 always - Always hash pointers (even if slab_debug is 1957 enabled). 1958 never - Never hash pointers. This option should only 1959 be specified when debugging the kernel. Do 1960 not use on production kernels. The boot 1961 param "no_hash_pointers" is an alias for 1962 this mode. 1963 1964 hashdist= [KNL,NUMA] Large hashes allocated during boot 1965 are distributed across NUMA nodes. Defaults on 1966 for 64-bit NUMA, off otherwise. 1967 Format: 0 | 1 (for off | on) 1968 1969 hd= [EIDE] (E)IDE hard drive subsystem geometry 1970 Format: <cyl>,<head>,<sect> 1971 1972 hest_disable [ACPI] 1973 Disable Hardware Error Source Table (HEST) support; 1974 corresponding firmware-first mode error processing 1975 logic will be disabled. 1976 1977 hibernate= [HIBERNATION] 1978 noresume Don't check if there's a hibernation image 1979 present during boot. 1980 nocompress Don't compress/decompress hibernation images. 1981 no Disable hibernation and resume. 1982 protect_image Turn on image protection during restoration 1983 (that will set all pages holding image data 1984 during restoration read-only). 1985 1986 hibernate.compressor= [HIBERNATION] Compression algorithm to be 1987 used with hibernation. 1988 Format: { lzo | lz4 } 1989 Default: lzo 1990 1991 lzo: Select LZO compression algorithm to 1992 compress/decompress hibernation image. 1993 1994 lz4: Select LZ4 compression algorithm to 1995 compress/decompress hibernation image. 1996 1997 hibernate.pm_test_delay= 1998 [HIBERNATION] 1999 Sets the number of seconds to remain in a hibernation test 2000 mode before resuming the system (see 2001 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG 2002 is set. Default value is 5. 2003 2004 hibernate_compression_threads= 2005 [HIBERNATION] 2006 Set the number of threads used for compressing or decompressing 2007 hibernation images. 2008 2009 Format: <integer> 2010 Default: 3 2011 Minimum: 1 2012 Example: hibernate_compression_threads=4 2013 2014 highmem=nn[KMG] [KNL,BOOT,EARLY] forces the highmem zone to have an exact 2015 size of <nn>. This works even on boxes that have no 2016 highmem otherwise. This also works to reduce highmem 2017 size on bigger boxes. 2018 2019 highres= [KNL] Enable/disable high resolution timer mode. 2020 Valid parameters: "on", "off" 2021 Default: "on" 2022 2023 hlt [BUGS=ARM,SH] 2024 2025 hostname= [KNL,EARLY] Set the hostname (aka UTS nodename). 2026 Format: <string> 2027 This allows setting the system's hostname during early 2028 startup. This sets the name returned by gethostname. 2029 Using this parameter to set the hostname makes it 2030 possible to ensure the hostname is correctly set before 2031 any userspace processes run, avoiding the possibility 2032 that a process may call gethostname before the hostname 2033 has been explicitly set, resulting in the calling 2034 process getting an incorrect result. The string must 2035 not exceed the maximum allowed hostname length (usually 2036 64 characters) and will be truncated otherwise. 2037 2038 hpet= [X86-32,HPET] option to control HPET usage 2039 Format: { enable (default) | disable | force | 2040 verbose } 2041 disable: disable HPET and use PIT instead 2042 force: allow force enabled of undocumented chips (ICH4, 2043 VIA, nVidia) 2044 verbose: show contents of HPET registers during setup 2045 2046 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET 2047 registers. Default set by CONFIG_HPET_MMAP_DEFAULT. 2048 2049 hugepages= [HW,EARLY] Number of HugeTLB pages to allocate at boot. 2050 If this follows hugepagesz (below), it specifies 2051 the number of pages of hugepagesz to be allocated. 2052 If this is the first HugeTLB parameter on the command 2053 line, it specifies the number of pages to allocate for 2054 the default huge page size. If using node format, the 2055 number of pages to allocate per-node can be specified. 2056 See also Documentation/admin-guide/mm/hugetlbpage.rst. 2057 Format: <integer> or (node format) 2058 <node>:<integer>[,<node>:<integer>] 2059 2060 hugepagesz= 2061 [HW,EARLY] The size of the HugeTLB pages. This is 2062 used in conjunction with hugepages (above) to 2063 allocate huge pages of a specific size at boot. The 2064 pair hugepagesz=X hugepages=Y can be specified once 2065 for each supported huge page size. Huge page sizes 2066 are architecture dependent. See also 2067 Documentation/admin-guide/mm/hugetlbpage.rst. 2068 Format: size[KMG] 2069 2070 hugepage_alloc_threads= 2071 [HW] The number of threads that should be used to 2072 allocate hugepages during boot. This option can be 2073 used to improve system bootup time when allocating 2074 a large amount of huge pages. 2075 The default value is 25% of the available hardware threads. 2076 2077 Note that this parameter only applies to non-gigantic huge pages. 2078 2079 hugetlb_cma= [HW,CMA,EARLY] The size of a CMA area used for allocation 2080 of gigantic hugepages. Or using node format, the size 2081 of a CMA area per node can be specified. 2082 Format: nn[KMGTPE] or (node format) 2083 <node>:nn[KMGTPE][,<node>:nn[KMGTPE]] 2084 2085 Reserve a CMA area of given size and allocate gigantic 2086 hugepages using the CMA allocator. If enabled, the 2087 boot-time allocation of gigantic hugepages is skipped. 2088 2089 hugetlb_cma_only= 2090 [HW,CMA,EARLY] When allocating new HugeTLB pages, only 2091 try to allocate from the CMA areas. 2092 2093 This option does nothing if hugetlb_cma= is not also 2094 specified. 2095 2096 hugetlb_free_vmemmap= 2097 [KNL] Requires CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP 2098 enabled. 2099 Control if HugeTLB Vmemmap Optimization (HVO) is enabled. 2100 Allows heavy hugetlb users to free up some more 2101 memory (7 * PAGE_SIZE for each 2MB hugetlb page). 2102 Format: { on | off (default) } 2103 2104 on: enable HVO 2105 off: disable HVO 2106 2107 Built with CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP_DEFAULT_ON=y, 2108 the default is on. 2109 2110 Note that the vmemmap pages may be allocated from the added 2111 memory block itself when memory_hotplug.memmap_on_memory is 2112 enabled, those vmemmap pages cannot be optimized even if this 2113 feature is enabled. Other vmemmap pages not allocated from 2114 the added memory block itself do not be affected. 2115 2116 hung_task_panic= 2117 [KNL] Should the hung task detector generate panics. 2118 Format: 0 | 1 2119 2120 A value of 1 instructs the kernel to panic when a 2121 hung task is detected. The default value is controlled 2122 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time 2123 option. The value selected by this boot parameter can 2124 be changed later by the kernel.hung_task_panic sysctl. 2125 2126 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC) 2127 terminal devices. Valid values: 0..8 2128 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs. 2129 If specified, z/VM IUCV HVC accepts connections 2130 from listed z/VM user IDs only. 2131 2132 hv_nopvspin [X86,HYPER_V,EARLY] 2133 Disables the paravirt spinlock optimizations 2134 which allow the hypervisor to 'idle' the guest 2135 on lock contention. 2136 2137 hw_protection= [HW] 2138 Format: reboot | shutdown 2139 2140 Hardware protection action taken on critical events like 2141 overtemperature or imminent voltage loss. 2142 2143 i2c_bus= [HW] Override the default board specific I2C bus speed 2144 or register an additional I2C bus that is not 2145 registered from board initialization code. 2146 Format: 2147 <bus_id>,<clkrate> 2148 2149 i2c_touchscreen_props= [HW,ACPI,X86] 2150 Set device-properties for ACPI-enumerated I2C-attached 2151 touchscreen, to e.g. fix coordinates of upside-down 2152 mounted touchscreens. If you need this option please 2153 submit a drivers/platform/x86/touchscreen_dmi.c patch 2154 adding a DMI quirk for this. 2155 2156 Format: 2157 <ACPI_HW_ID>:<prop_name>=<val>[:prop_name=val][:...] 2158 Where <val> is one of: 2159 Omit "=<val>" entirely Set a boolean device-property 2160 Unsigned number Set a u32 device-property 2161 Anything else Set a string device-property 2162 2163 Examples (split over multiple lines): 2164 i2c_touchscreen_props=GDIX1001:touchscreen-inverted-x: 2165 touchscreen-inverted-y 2166 2167 i2c_touchscreen_props=MSSL1680:touchscreen-size-x=1920: 2168 touchscreen-size-y=1080:touchscreen-inverted-y: 2169 firmware-name=gsl1680-vendor-model.fw:silead,home-button 2170 2171 i8042.debug [HW] Toggle i8042 debug mode 2172 i8042.unmask_kbd_data 2173 [HW] Enable printing of interrupt data from the KBD port 2174 (disabled by default, and as a pre-condition 2175 requires that i8042.debug=1 be enabled) 2176 i8042.direct [HW] Put keyboard port into non-translated mode 2177 i8042.dumbkbd [HW] Pretend that controller can only read data from 2178 keyboard and cannot control its state 2179 (Don't attempt to blink the leds) 2180 i8042.noaux [HW] Don't check for auxiliary (== mouse) port 2181 i8042.nokbd [HW] Don't check/create keyboard port 2182 i8042.noloop [HW] Disable the AUX Loopback command while probing 2183 for the AUX port 2184 i8042.nomux [HW] Don't check presence of an active multiplexing 2185 controller 2186 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX 2187 controllers 2188 i8042.notimeout [HW] Ignore timeout condition signalled by controller 2189 i8042.reset [HW] Reset the controller during init, cleanup and 2190 suspend-to-ram transitions, only during s2r 2191 transitions, or never reset 2192 Format: { 1 | Y | y | 0 | N | n } 2193 1, Y, y: always reset controller 2194 0, N, n: don't ever reset controller 2195 Default: only on s2r transitions on x86; most other 2196 architectures force reset to be always executed 2197 i8042.unlock [HW] Unlock (ignore) the keylock 2198 i8042.kbdreset [HW] Reset device connected to KBD port 2199 i8042.probe_defer 2200 [HW] Allow deferred probing upon i8042 probe errors 2201 2202 i810= [HW,DRM] 2203 2204 i915.invert_brightness= 2205 [DRM] Invert the sense of the variable that is used to 2206 set the brightness of the panel backlight. Normally a 2207 brightness value of 0 indicates backlight switched off, 2208 and the maximum of the brightness value sets the backlight 2209 to maximum brightness. If this parameter is set to 0 2210 (default) and the machine requires it, or this parameter 2211 is set to 1, a brightness value of 0 sets the backlight 2212 to maximum brightness, and the maximum of the brightness 2213 value switches the backlight off. 2214 -1 -- never invert brightness 2215 0 -- machine default 2216 1 -- force brightness inversion 2217 2218 ia32_emulation= [X86-64] 2219 Format: <bool> 2220 When true, allows loading 32-bit programs and executing 32-bit 2221 syscalls, essentially overriding IA32_EMULATION_DEFAULT_DISABLED at 2222 boot time. When false, unconditionally disables IA32 emulation. 2223 2224 icn= [HW,ISDN] 2225 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]] 2226 2227 2228 idle= [X86,EARLY] 2229 Format: idle=poll, idle=halt, idle=nomwait 2230 2231 idle=poll: Don't do power saving in the idle loop 2232 using HLT, but poll for rescheduling event. This will 2233 make the CPUs eat a lot more power, but may be useful 2234 to get slightly better performance in multiprocessor 2235 benchmarks. It also makes some profiling using 2236 performance counters more accurate. Please note that 2237 on systems with MONITOR/MWAIT support (like Intel 2238 EM64T CPUs) this option has no performance advantage 2239 over the normal idle loop. It may also interact badly 2240 with hyperthreading. 2241 2242 idle=halt: Halt is forced to be used for CPU idle. 2243 In such case C2/C3 won't be used again. 2244 2245 idle=nomwait: Disable mwait for CPU C-states 2246 2247 idxd.sva= [HW] 2248 Format: <bool> 2249 Allow force disabling of Shared Virtual Memory (SVA) 2250 support for the idxd driver. By default it is set to 2251 true (1). 2252 2253 idxd.tc_override= [HW] 2254 Format: <bool> 2255 Allow override of default traffic class configuration 2256 for the device. By default it is set to false (0). 2257 2258 ieee754= [MIPS] Select IEEE Std 754 conformance mode 2259 Format: { strict | legacy | 2008 | relaxed | emulated } 2260 Default: strict 2261 2262 Choose which programs will be accepted for execution 2263 based on the IEEE 754 NaN encoding(s) supported by 2264 the FPU and the NaN encoding requested with the value 2265 of an ELF file header flag individually set by each 2266 binary. Hardware implementations are permitted to 2267 support either or both of the legacy and the 2008 NaN 2268 encoding mode. 2269 2270 Available settings are as follows: 2271 strict accept binaries that request a NaN encoding 2272 supported by the FPU 2273 legacy only accept legacy-NaN binaries, if supported 2274 by the FPU 2275 2008 only accept 2008-NaN binaries, if supported 2276 by the FPU 2277 relaxed accept any binaries regardless of whether 2278 supported by the FPU 2279 emulated accept any binaries but enable FPU emulator 2280 if binary mode is unsupported by the FPU. 2281 2282 The FPU emulator is always able to support both NaN 2283 encodings, so if no FPU hardware is present or it has 2284 been disabled with 'nofpu', then the settings of 2285 'legacy' and '2008' strap the emulator accordingly, 2286 'relaxed' straps the emulator for both legacy-NaN and 2287 2008-NaN, whereas 'strict' enables legacy-NaN only on 2288 legacy processors and both NaN encodings on MIPS32 or 2289 MIPS64 CPUs. 2290 2291 The setting for ABS.fmt/NEG.fmt instruction execution 2292 mode generally follows that for the NaN encoding, 2293 except where unsupported by hardware. 2294 2295 ignore_loglevel [KNL,EARLY] 2296 Ignore loglevel setting - this will print /all/ 2297 kernel messages to the console. Useful for debugging. 2298 We also add it as printk module parameter, so users 2299 could change it dynamically, usually by 2300 /sys/module/printk/parameters/ignore_loglevel. 2301 2302 ignore_rlimit_data 2303 Ignore RLIMIT_DATA setting for data mappings, 2304 print warning at first misuse. Can be changed via 2305 /sys/module/kernel/parameters/ignore_rlimit_data. 2306 2307 ihash_entries= [KNL] 2308 Set number of hash buckets for inode cache. 2309 2310 ima_appraise= [IMA] appraise integrity measurements 2311 Format: { "off" | "enforce" | "fix" | "log" } 2312 default: "enforce" 2313 2314 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead. 2315 The builtin appraise policy appraises all files 2316 owned by uid=0. 2317 2318 ima_canonical_fmt [IMA] 2319 Use the canonical format for the binary runtime 2320 measurements, instead of host native format. 2321 2322 ima_hash= [IMA] 2323 Format: { md5 | sha1 | rmd160 | sha256 | sha384 2324 | sha512 | ... } 2325 default: "sha1" 2326 2327 The list of supported hash algorithms is defined 2328 in crypto/hash_info.h. 2329 2330 ima_policy= [IMA] 2331 The builtin policies to load during IMA setup. 2332 Format: "tcb | appraise_tcb | secure_boot | 2333 fail_securely | critical_data" 2334 2335 The "tcb" policy measures all programs exec'd, files 2336 mmap'd for exec, and all files opened with the read 2337 mode bit set by either the effective uid (euid=0) or 2338 uid=0. 2339 2340 The "appraise_tcb" policy appraises the integrity of 2341 all files owned by root. 2342 2343 The "secure_boot" policy appraises the integrity 2344 of files (eg. kexec kernel image, kernel modules, 2345 firmware, policy, etc) based on file signatures. 2346 2347 The "fail_securely" policy forces file signature 2348 verification failure also on privileged mounted 2349 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE 2350 flag. 2351 2352 The "critical_data" policy measures kernel integrity 2353 critical data. 2354 2355 ima_tcb [IMA] Deprecated. Use ima_policy= instead. 2356 Load a policy which meets the needs of the Trusted 2357 Computing Base. This means IMA will measure all 2358 programs exec'd, files mmap'd for exec, and all files 2359 opened for read by uid=0. 2360 2361 ima_template= [IMA] 2362 Select one of defined IMA measurements template formats. 2363 Formats: { "ima" | "ima-ng" | "ima-ngv2" | "ima-sig" | 2364 "ima-sigv2" } 2365 Default: "ima-ng" 2366 2367 ima_template_fmt= 2368 [IMA] Define a custom template format. 2369 Format: { "field1|...|fieldN" } 2370 2371 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage 2372 Format: <min_file_size> 2373 Set the minimal file size for using asynchronous hash. 2374 If left unspecified, ahash usage is disabled. 2375 2376 ahash performance varies for different data sizes on 2377 different crypto accelerators. This option can be used 2378 to achieve the best performance for a particular HW. 2379 2380 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size 2381 Format: <bufsize> 2382 Set hashing buffer size. Default: 4k. 2383 2384 ahash performance varies for different chunk sizes on 2385 different crypto accelerators. This option can be used 2386 to achieve best performance for particular HW. 2387 2388 ima= [IMA] Enable or disable IMA 2389 Format: { "off" | "on" } 2390 Default: "on" 2391 Note that disabling IMA is limited to kdump kernel. 2392 2393 indirect_target_selection= [X86,Intel] Mitigation control for Indirect 2394 Target Selection(ITS) bug in Intel CPUs. Updated 2395 microcode is also required for a fix in IBPB. 2396 2397 on: Enable mitigation (default). 2398 off: Disable mitigation. 2399 force: Force the ITS bug and deploy default 2400 mitigation. 2401 vmexit: Only deploy mitigation if CPU is affected by 2402 guest/host isolation part of ITS. 2403 stuff: Deploy RSB-fill mitigation when retpoline is 2404 also deployed. Otherwise, deploy the default 2405 mitigation. 2406 2407 For details see: 2408 Documentation/admin-guide/hw-vuln/indirect-target-selection.rst 2409 2410 init= [KNL] 2411 Format: <full_path> 2412 Run specified binary instead of /sbin/init as init 2413 process. 2414 2415 initcall_debug [KNL] Trace initcalls as they are executed. Useful 2416 for working out where the kernel is dying during 2417 startup. 2418 2419 initcall_blacklist= [KNL] Do not execute a comma-separated list of 2420 initcall functions. Useful for debugging built-in 2421 modules and initcalls. 2422 2423 initramfs_async= [KNL] 2424 Format: <bool> 2425 Default: 1 2426 This parameter controls whether the initramfs 2427 image is unpacked asynchronously, concurrently 2428 with devices being probed and 2429 initialized. This should normally just work, 2430 but as a debugging aid, one can get the 2431 historical behaviour of the initramfs 2432 unpacking being completed before device_ and 2433 late_ initcalls. 2434 2435 initrd= [BOOT,EARLY] Specify the location of the initial ramdisk 2436 2437 initrdmem= [KNL,EARLY] Specify a physical address and size from which to 2438 load the initrd. If an initrd is compiled in or 2439 specified in the bootparams, it takes priority over this 2440 setting. 2441 Format: ss[KMG],nn[KMG] 2442 Default is 0, 0 2443 2444 init_on_alloc= [MM,EARLY] Fill newly allocated pages and heap objects with 2445 zeroes. 2446 Format: 0 | 1 2447 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON. 2448 2449 init_on_free= [MM,EARLY] Fill freed pages and heap objects with zeroes. 2450 Format: 0 | 1 2451 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON. 2452 2453 init_pkru= [X86] Specify the default memory protection keys rights 2454 register contents for all processes. 0x55555554 by 2455 default (disallow access to all but pkey 0). Can 2456 override in debugfs after boot. 2457 2458 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver 2459 Format: <irq> 2460 2461 int_pln_enable [X86] Enable power limit notification interrupt 2462 2463 integrity_audit=[IMA] 2464 Format: { "0" | "1" } 2465 0 -- basic integrity auditing messages. (Default) 2466 1 -- additional integrity auditing messages. 2467 2468 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option 2469 on 2470 Enable intel iommu driver. 2471 off 2472 Disable intel iommu driver. 2473 igfx_off [Default Off] 2474 By default, gfx is mapped as normal device. If a gfx 2475 device has a dedicated DMAR unit, the DMAR unit is 2476 bypassed by not enabling DMAR with this option. In 2477 this case, gfx device will use physical address for 2478 DMA. 2479 strict [Default Off] 2480 Deprecated, equivalent to iommu.strict=1. 2481 sp_off [Default Off] 2482 By default, super page will be supported if Intel IOMMU 2483 has the capability. With this option, super page will 2484 not be supported. 2485 sm_on 2486 Enable the Intel IOMMU scalable mode if the hardware 2487 advertises that it has support for the scalable mode 2488 translation. 2489 sm_off 2490 Disallow use of the Intel IOMMU scalable mode. 2491 tboot_noforce [Default Off] 2492 Do not force the Intel IOMMU enabled under tboot. 2493 By default, tboot will force Intel IOMMU on, which 2494 could harm performance of some high-throughput 2495 devices like 40GBit network cards, even if identity 2496 mapping is enabled. 2497 Note that using this option lowers the security 2498 provided by tboot because it makes the system 2499 vulnerable to DMA attacks. 2500 2501 intel_idle.max_cstate= [KNL,HW,ACPI,X86] 2502 0 disables intel_idle and fall back on acpi_idle. 2503 1 to 9 specify maximum depth of C-state. 2504 2505 intel_pstate= [X86,EARLY] 2506 disable 2507 Do not enable intel_pstate as the default 2508 scaling driver for the supported processors 2509 active 2510 Use intel_pstate driver to bypass the scaling 2511 governors layer of cpufreq and provides it own 2512 algorithms for p-state selection. There are two 2513 P-state selection algorithms provided by 2514 intel_pstate in the active mode: powersave and 2515 performance. The way they both operate depends 2516 on whether or not the hardware managed P-states 2517 (HWP) feature has been enabled in the processor 2518 and possibly on the processor model. 2519 passive 2520 Use intel_pstate as a scaling driver, but configure it 2521 to work with generic cpufreq governors (instead of 2522 enabling its internal governor). This mode cannot be 2523 used along with the hardware-managed P-states (HWP) 2524 feature. 2525 force 2526 Enable intel_pstate on systems that prohibit it by default 2527 in favor of acpi-cpufreq. Forcing the intel_pstate driver 2528 instead of acpi-cpufreq may disable platform features, such 2529 as thermal controls and power capping, that rely on ACPI 2530 P-States information being indicated to OSPM and therefore 2531 should be used with caution. This option does not work with 2532 processors that aren't supported by the intel_pstate driver 2533 or on platforms that use pcc-cpufreq instead of acpi-cpufreq. 2534 no_hwp 2535 Do not enable hardware P state control (HWP) 2536 if available. 2537 hwp_only 2538 Only load intel_pstate on systems which support 2539 hardware P state control (HWP) if available. 2540 support_acpi_ppc 2541 Enforce ACPI _PPC performance limits. If the Fixed ACPI 2542 Description Table, specifies preferred power management 2543 profile as "Enterprise Server" or "Performance Server", 2544 then this feature is turned on by default. 2545 per_cpu_perf_limits 2546 Allow per-logical-CPU P-State performance control limits using 2547 cpufreq sysfs interface 2548 no_cas 2549 Do not enable capacity-aware scheduling (CAS) on 2550 hybrid systems 2551 2552 intremap= [X86-64,Intel-IOMMU,EARLY] 2553 on enable Interrupt Remapping (default) 2554 off disable Interrupt Remapping 2555 nosid disable Source ID checking 2556 no_x2apic_optout 2557 BIOS x2APIC opt-out request will be ignored 2558 nopost disable Interrupt Posting 2559 posted_msi 2560 enable MSIs delivered as posted interrupts 2561 2562 iomem= Disable strict checking of access to MMIO memory 2563 strict regions from userspace. 2564 relaxed 2565 2566 iommu= [X86,EARLY] 2567 2568 off 2569 Don't initialize and use any kind of IOMMU. 2570 2571 force 2572 Force the use of the hardware IOMMU even when 2573 it is not actually needed (e.g. because < 3 GB 2574 memory). 2575 2576 noforce 2577 Don't force hardware IOMMU usage when it is not 2578 needed. (default). 2579 2580 biomerge 2581 panic 2582 nopanic 2583 merge 2584 nomerge 2585 2586 soft 2587 Use software bounce buffering (SWIOTLB) (default for 2588 Intel machines). This can be used to prevent the usage 2589 of an available hardware IOMMU. 2590 2591 [X86] 2592 pt 2593 [X86] 2594 nopt 2595 [PPC/POWERNV] 2596 nobypass 2597 Disable IOMMU bypass, using IOMMU for PCI devices. 2598 2599 [X86] 2600 AMD Gart HW IOMMU-specific options: 2601 2602 <size> 2603 Set the size of the remapping area in bytes. 2604 2605 allowed 2606 Overwrite iommu off workarounds for specific chipsets 2607 2608 fullflush 2609 Flush IOMMU on each allocation (default). 2610 2611 nofullflush 2612 Don't use IOMMU fullflush. 2613 2614 memaper[=<order>] 2615 Allocate an own aperture over RAM with size 2616 32MB<<order. (default: order=1, i.e. 64MB) 2617 2618 merge 2619 Do scatter-gather (SG) merging. Implies "force" 2620 (experimental). 2621 2622 nomerge 2623 Don't do scatter-gather (SG) merging. 2624 2625 noaperture 2626 Ask the IOMMU not to touch the aperture for AGP. 2627 2628 noagp 2629 Don't initialize the AGP driver and use full aperture. 2630 2631 panic 2632 Always panic when IOMMU overflows. 2633 2634 iommu.forcedac= [ARM64,X86,EARLY] Control IOVA allocation for PCI devices. 2635 Format: { "0" | "1" } 2636 0 - Try to allocate a 32-bit DMA address first, before 2637 falling back to the full range if needed. 2638 1 - Allocate directly from the full usable range, 2639 forcing Dual Address Cycle for PCI cards supporting 2640 greater than 32-bit addressing. 2641 2642 iommu.strict= [ARM64,X86,S390,EARLY] Configure TLB invalidation behaviour 2643 Format: { "0" | "1" } 2644 0 - Lazy mode. 2645 Request that DMA unmap operations use deferred 2646 invalidation of hardware TLBs, for increased 2647 throughput at the cost of reduced device isolation. 2648 Will fall back to strict mode if not supported by 2649 the relevant IOMMU driver. 2650 1 - Strict mode. 2651 DMA unmap operations invalidate IOMMU hardware TLBs 2652 synchronously. 2653 unset - Use value of CONFIG_IOMMU_DEFAULT_DMA_{LAZY,STRICT}. 2654 Note: on x86, strict mode specified via one of the 2655 legacy driver-specific options takes precedence. 2656 2657 iommu.passthrough= 2658 [ARM64,X86,EARLY] Configure DMA to bypass the IOMMU by default. 2659 Format: { "0" | "1" } 2660 0 - Use IOMMU translation for DMA. 2661 1 - Bypass the IOMMU for DMA. 2662 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH. 2663 2664 io7= [HW] IO7 for Marvel-based Alpha systems 2665 See comment before marvel_specify_io7 in 2666 arch/alpha/kernel/core_marvel.c. 2667 2668 io_delay= [X86,EARLY] I/O delay method 2669 0x80 2670 Standard port 0x80 based delay 2671 0xed 2672 Alternate port 0xed based delay (needed on some systems) 2673 udelay 2674 Simple two microseconds delay 2675 none 2676 No delay 2677 2678 ip= [IP_PNP] 2679 See Documentation/admin-guide/nfs/nfsroot.rst. 2680 2681 ipcmni_extend [KNL,EARLY] Extend the maximum number of unique System V 2682 IPC identifiers from 32,768 to 16,777,216. 2683 2684 ipe.enforce= [IPE] 2685 Format: <bool> 2686 Determine whether IPE starts in permissive (0) or 2687 enforce (1) mode. The default is enforce. 2688 2689 ipe.success_audit= 2690 [IPE] 2691 Format: <bool> 2692 Start IPE with success auditing enabled, emitting 2693 an audit event when a binary is allowed. The default 2694 is 0. 2695 2696 irqaffinity= [SMP] Set the default irq affinity mask 2697 The argument is a cpu list, as described above. 2698 2699 irqchip.gicv2_force_probe= 2700 [ARM,ARM64,EARLY] 2701 Format: <bool> 2702 Force the kernel to look for the second 4kB page 2703 of a GICv2 controller even if the memory range 2704 exposed by the device tree is too small. 2705 2706 irqchip.gicv3_nolpi= 2707 [ARM,ARM64,EARLY] 2708 Force the kernel to ignore the availability of 2709 LPIs (and by consequence ITSs). Intended for system 2710 that use the kernel as a bootloader, and thus want 2711 to let secondary kernels in charge of setting up 2712 LPIs. 2713 2714 irqchip.gicv3_pseudo_nmi= [ARM64,EARLY] 2715 Enables support for pseudo-NMIs in the kernel. This 2716 requires the kernel to be built with 2717 CONFIG_ARM64_PSEUDO_NMI. 2718 2719 irqchip.riscv_imsic_noipi 2720 [RISC-V,EARLY] 2721 Force the kernel to not use IMSIC software injected MSIs 2722 as IPIs. Intended for system where IMSIC is trap-n-emulated, 2723 and thus want to reduce MMIO traps when triggering IPIs 2724 to multiple harts. 2725 2726 irqfixup [HW] 2727 When an interrupt is not handled search all handlers 2728 for it. Intended to get systems with badly broken 2729 firmware running. 2730 2731 irqhandler.duration_warn_us= [KNL] 2732 Warn if an IRQ handler exceeds the specified duration 2733 threshold in microseconds. Useful for identifying 2734 long-running IRQs in the system. 2735 2736 irqpoll [HW] 2737 When an interrupt is not handled search all handlers 2738 for it. Also check all handlers each timer 2739 interrupt. Intended to get systems with badly broken 2740 firmware running. 2741 2742 isapnp= [ISAPNP] 2743 Format: <RDP>,<reset>,<pci_scan>,<verbosity> 2744 2745 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance. 2746 [Deprecated - use cpusets instead] 2747 Format: [flag-list,]<cpu-list> 2748 2749 Specify one or more CPUs to isolate from disturbances 2750 specified in the flag list (default: domain): 2751 2752 nohz 2753 Disable the tick when a single task runs as well as 2754 disabling other kernel noises like having RCU callbacks 2755 offloaded. This is equivalent to the nohz_full parameter. 2756 2757 A residual 1Hz tick is offloaded to workqueues, which you 2758 need to affine to housekeeping through the global 2759 workqueue's affinity configured via the 2760 /sys/devices/virtual/workqueue/cpumask sysfs file, or 2761 by using the 'domain' flag described below. 2762 2763 NOTE: by default the global workqueue runs on all CPUs, 2764 so to protect individual CPUs the 'cpumask' file has to 2765 be configured manually after bootup. 2766 2767 domain 2768 Isolate from the general SMP balancing and scheduling 2769 algorithms. Note that performing domain isolation this way 2770 is irreversible: it's not possible to bring back a CPU to 2771 the domains once isolated through isolcpus. It's strongly 2772 advised to use cpusets instead to disable scheduler load 2773 balancing through the "cpuset.sched_load_balance" file. 2774 It offers a much more flexible interface where CPUs can 2775 move in and out of an isolated set anytime. 2776 2777 You can move a process onto or off an "isolated" CPU via 2778 the CPU affinity syscalls or cpuset. 2779 <cpu number> begins at 0 and the maximum value is 2780 "number of CPUs in system - 1". 2781 2782 managed_irq 2783 2784 Isolate from being targeted by managed interrupts 2785 which have an interrupt mask containing isolated 2786 CPUs. The affinity of managed interrupts is 2787 handled by the kernel and cannot be changed via 2788 the /proc/irq/* interfaces. 2789 2790 This isolation is best effort and only effective 2791 if the automatically assigned interrupt mask of a 2792 device queue contains isolated and housekeeping 2793 CPUs. If housekeeping CPUs are online then such 2794 interrupts are directed to the housekeeping CPU 2795 so that IO submitted on the housekeeping CPU 2796 cannot disturb the isolated CPU. 2797 2798 If a queue's affinity mask contains only isolated 2799 CPUs then this parameter has no effect on the 2800 interrupt routing decision, though interrupts are 2801 only delivered when tasks running on those 2802 isolated CPUs submit IO. IO submitted on 2803 housekeeping CPUs has no influence on those 2804 queues. 2805 2806 The format of <cpu-list> is described above. 2807 2808 iucv= [HW,NET] 2809 2810 ivrs_ioapic [HW,X86-64] 2811 Provide an override to the IOAPIC-ID<->DEVICE-ID 2812 mapping provided in the IVRS ACPI table. 2813 By default, PCI segment is 0, and can be omitted. 2814 2815 For example, to map IOAPIC-ID decimal 10 to 2816 PCI segment 0x1 and PCI device 00:14.0, 2817 write the parameter as: 2818 ivrs_ioapic=10@0001:00:14.0 2819 2820 Deprecated formats: 2821 * To map IOAPIC-ID decimal 10 to PCI device 00:14.0 2822 write the parameter as: 2823 ivrs_ioapic[10]=00:14.0 2824 * To map IOAPIC-ID decimal 10 to PCI segment 0x1 and 2825 PCI device 00:14.0 write the parameter as: 2826 ivrs_ioapic[10]=0001:00:14.0 2827 2828 ivrs_hpet [HW,X86-64] 2829 Provide an override to the HPET-ID<->DEVICE-ID 2830 mapping provided in the IVRS ACPI table. 2831 By default, PCI segment is 0, and can be omitted. 2832 2833 For example, to map HPET-ID decimal 10 to 2834 PCI segment 0x1 and PCI device 00:14.0, 2835 write the parameter as: 2836 ivrs_hpet=10@0001:00:14.0 2837 2838 Deprecated formats: 2839 * To map HPET-ID decimal 0 to PCI device 00:14.0 2840 write the parameter as: 2841 ivrs_hpet[0]=00:14.0 2842 * To map HPET-ID decimal 10 to PCI segment 0x1 and 2843 PCI device 00:14.0 write the parameter as: 2844 ivrs_ioapic[10]=0001:00:14.0 2845 2846 ivrs_acpihid [HW,X86-64] 2847 Provide an override to the ACPI-HID:UID<->DEVICE-ID 2848 mapping provided in the IVRS ACPI table. 2849 By default, PCI segment is 0, and can be omitted. 2850 2851 For example, to map UART-HID:UID AMD0020:0 to 2852 PCI segment 0x1 and PCI device ID 00:14.5, 2853 write the parameter as: 2854 ivrs_acpihid=AMD0020:0@0001:00:14.5 2855 2856 Deprecated formats: 2857 * To map UART-HID:UID AMD0020:0 to PCI segment is 0, 2858 PCI device ID 00:14.5, write the parameter as: 2859 ivrs_acpihid[00:14.5]=AMD0020:0 2860 * To map UART-HID:UID AMD0020:0 to PCI segment 0x1 and 2861 PCI device ID 00:14.5, write the parameter as: 2862 ivrs_acpihid[0001:00:14.5]=AMD0020:0 2863 2864 js= [HW,JOY] Analog joystick 2865 See Documentation/input/joydev/joystick.rst. 2866 2867 kasan_multi_shot 2868 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print 2869 report on every invalid memory access. Without this 2870 parameter KASAN will print report only for the first 2871 invalid access. 2872 2873 keep_bootcon [KNL,EARLY] 2874 Do not unregister boot console at start. This is only 2875 useful for debugging when something happens in the window 2876 between unregistering the boot console and initializing 2877 the real console. 2878 2879 keepinitrd [HW,ARM] See retain_initrd. 2880 2881 kernelcore= [KNL,X86,PPC,EARLY] 2882 Format: nn[KMGTPE] | nn% | "mirror" 2883 This parameter specifies the amount of memory usable by 2884 the kernel for non-movable allocations. The requested 2885 amount is spread evenly throughout all nodes in the 2886 system as ZONE_NORMAL. The remaining memory is used for 2887 movable memory in its own zone, ZONE_MOVABLE. In the 2888 event, a node is too small to have both ZONE_NORMAL and 2889 ZONE_MOVABLE, kernelcore memory will take priority and 2890 other nodes will have a larger ZONE_MOVABLE. 2891 2892 ZONE_MOVABLE is used for the allocation of pages that 2893 may be reclaimed or moved by the page migration 2894 subsystem. Note that allocations like PTEs-from-HighMem 2895 still use the HighMem zone if it exists, and the Normal 2896 zone if it does not. 2897 2898 It is possible to specify the exact amount of memory in 2899 the form of "nn[KMGTPE]", a percentage of total system 2900 memory in the form of "nn%", or "mirror". If "mirror" 2901 option is specified, mirrored (reliable) memory is used 2902 for non-movable allocations and remaining memory is used 2903 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror" 2904 are exclusive, so you cannot specify multiple forms. 2905 2906 kgdbdbgp= [KGDB,HW,EARLY] kgdb over EHCI usb debug port. 2907 Format: <Controller#>[,poll interval] 2908 The controller # is the number of the ehci usb debug 2909 port as it is probed via PCI. The poll interval is 2910 optional and is the number seconds in between 2911 each poll cycle to the debug port in case you need 2912 the functionality for interrupting the kernel with 2913 gdb or control-c on the dbgp connection. When 2914 not using this parameter you use sysrq-g to break into 2915 the kernel debugger. 2916 2917 kgdboc= [KGDB,HW] kgdb over consoles. 2918 Requires a tty driver that supports console polling, 2919 or a supported polling keyboard driver (non-usb). 2920 Serial only format: <serial_device>[,baud] 2921 keyboard only format: kbd 2922 keyboard and serial format: kbd,<serial_device>[,baud] 2923 Optional Kernel mode setting: 2924 kms, kbd format: kms,kbd 2925 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud] 2926 2927 kgdboc_earlycon= [KGDB,HW,EARLY] 2928 If the boot console provides the ability to read 2929 characters and can work in polling mode, you can use 2930 this parameter to tell kgdb to use it as a backend 2931 until the normal console is registered. Intended to 2932 be used together with the kgdboc parameter which 2933 specifies the normal console to transition to. 2934 2935 The name of the early console should be specified 2936 as the value of this parameter. Note that the name of 2937 the early console might be different than the tty 2938 name passed to kgdboc. It's OK to leave the value 2939 blank and the first boot console that implements 2940 read() will be picked. 2941 2942 kgdbwait [KGDB,EARLY] Stop kernel execution and enter the 2943 kernel debugger at the earliest opportunity. 2944 2945 kho= [KEXEC,EARLY] 2946 Format: { "0" | "1" | "off" | "on" | "y" | "n" } 2947 Enables or disables Kexec HandOver. 2948 "0" | "off" | "n" - kexec handover is disabled 2949 "1" | "on" | "y" - kexec handover is enabled 2950 2951 kho_scratch= [KEXEC,EARLY] 2952 Format: ll[KMG],mm[KMG],nn[KMG] | nn% 2953 Defines the size of the KHO scratch region. The KHO 2954 scratch regions are physically contiguous memory 2955 ranges that can only be used for non-kernel 2956 allocations. That way, even when memory is heavily 2957 fragmented with handed over memory, the kexeced 2958 kernel will always have enough contiguous ranges to 2959 bootstrap itself. 2960 2961 It is possible to specify the exact amount of 2962 memory in the form of "ll[KMG],mm[KMG],nn[KMG]" 2963 where the first parameter defines the size of a low 2964 memory scratch area, the second parameter defines 2965 the size of a global scratch area and the third 2966 parameter defines the size of additional per-node 2967 scratch areas. The form "nn%" defines scale factor 2968 (in percents) of memory that was used during boot. 2969 2970 kmac= [MIPS] Korina ethernet MAC address. 2971 Configure the RouterBoard 532 series on-chip 2972 Ethernet adapter MAC address. 2973 2974 kmemleak= [KNL,EARLY] Boot-time kmemleak enable/disable 2975 Valid arguments: on, off 2976 Default: on 2977 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y, 2978 the default is off. 2979 2980 kprobe_event=[probe-list] 2981 [FTRACE] Add kprobe events and enable at boot time. 2982 The probe-list is a semicolon delimited list of probe 2983 definitions. Each definition is same as kprobe_events 2984 interface, but the parameters are comma delimited. 2985 For example, to add a kprobe event on vfs_read with 2986 arg1 and arg2, add to the command line; 2987 2988 kprobe_event=p,vfs_read,$arg1,$arg2 2989 2990 See also Documentation/trace/kprobetrace.rst "Kernel 2991 Boot Parameter" section. 2992 2993 kpti= [ARM64,EARLY] Control page table isolation of 2994 user and kernel address spaces. 2995 Default: enabled on cores which need mitigation. 2996 0: force disabled 2997 1: force enabled 2998 2999 kunit.enable= [KUNIT] Enable executing KUnit tests. Requires 3000 CONFIG_KUNIT to be set to be fully enabled. The 3001 default value can be overridden via 3002 KUNIT_DEFAULT_ENABLED. 3003 Default is 1 (enabled) 3004 3005 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs. 3006 Default is 0 (don't ignore, but inject #GP) 3007 3008 kvm.eager_page_split= 3009 [KVM,X86] Controls whether or not KVM will try to 3010 proactively split all huge pages during dirty logging. 3011 Eager page splitting reduces interruptions to vCPU 3012 execution by eliminating the write-protection faults 3013 and MMU lock contention that would otherwise be 3014 required to split huge pages lazily. 3015 3016 VM workloads that rarely perform writes or that write 3017 only to a small region of VM memory may benefit from 3018 disabling eager page splitting to allow huge pages to 3019 still be used for reads. 3020 3021 The behavior of eager page splitting depends on whether 3022 KVM_DIRTY_LOG_INITIALLY_SET is enabled or disabled. If 3023 disabled, all huge pages in a memslot will be eagerly 3024 split when dirty logging is enabled on that memslot. If 3025 enabled, eager page splitting will be performed during 3026 the KVM_CLEAR_DIRTY ioctl, and only for the pages being 3027 cleared. 3028 3029 Eager page splitting is only supported when kvm.tdp_mmu=Y. 3030 3031 Default is Y (on). 3032 3033 kvm.enable_virt_at_load=[KVM,ARM64,LOONGARCH,MIPS,RISCV,X86] 3034 If enabled, KVM will enable virtualization in hardware 3035 when KVM is loaded, and disable virtualization when KVM 3036 is unloaded (if KVM is built as a module). 3037 3038 If disabled, KVM will dynamically enable and disable 3039 virtualization on-demand when creating and destroying 3040 VMs, i.e. on the 0=>1 and 1=>0 transitions of the 3041 number of VMs. 3042 3043 Enabling virtualization at module load avoids potential 3044 latency for creation of the 0=>1 VM, as KVM serializes 3045 virtualization enabling across all online CPUs. The 3046 "cost" of enabling virtualization when KVM is loaded, 3047 is that doing so may interfere with using out-of-tree 3048 hypervisors that want to "own" virtualization hardware. 3049 3050 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface. 3051 Default is false (don't support). 3052 3053 kvm.nx_huge_pages= 3054 [KVM] Controls the software workaround for the 3055 X86_BUG_ITLB_MULTIHIT bug. 3056 force : Always deploy workaround. 3057 off : Never deploy workaround. 3058 auto : Deploy workaround based on the presence of 3059 X86_BUG_ITLB_MULTIHIT. 3060 3061 Default is 'auto'. 3062 3063 If the software workaround is enabled for the host, 3064 guests do need not to enable it for nested guests. 3065 3066 kvm.nx_huge_pages_recovery_ratio= 3067 [KVM] Controls how many 4KiB pages are periodically zapped 3068 back to huge pages. 0 disables the recovery, otherwise if 3069 the value is N KVM will zap 1/Nth of the 4KiB pages every 3070 period (see below). The default is 60. 3071 3072 kvm.nx_huge_pages_recovery_period_ms= 3073 [KVM] Controls the time period at which KVM zaps 4KiB pages 3074 back to huge pages. If the value is a non-zero N, KVM will 3075 zap a portion (see ratio above) of the pages every N msecs. 3076 If the value is 0 (the default), KVM will pick a period based 3077 on the ratio, such that a page is zapped after 1 hour on average. 3078 3079 kvm-amd.nested= [KVM,AMD] Control nested virtualization feature in 3080 KVM/SVM. Default is 1 (enabled). 3081 3082 kvm-amd.npt= [KVM,AMD] Control KVM's use of Nested Page Tables, 3083 a.k.a. Two-Dimensional Page Tables. Default is 1 3084 (enabled). Disable by KVM if hardware lacks support 3085 for NPT. 3086 3087 kvm-amd.ciphertext_hiding_asids= 3088 [KVM,AMD] Ciphertext hiding prevents disallowed accesses 3089 to SNP private memory from reading ciphertext. Instead, 3090 reads will see constant default values (0xff). 3091 3092 If ciphertext hiding is enabled, the joint SEV-ES and 3093 SEV-SNP ASID space is partitioned into separate SEV-ES 3094 and SEV-SNP ASID ranges, with the SEV-SNP range being 3095 [1..max_snp_asid] and the SEV-ES range being 3096 (max_snp_asid..min_sev_asid), where min_sev_asid is 3097 enumerated by CPUID.0x.8000_001F[EDX]. 3098 3099 A non-zero value enables SEV-SNP ciphertext hiding and 3100 adjusts the ASID ranges for SEV-ES and SEV-SNP guests. 3101 KVM caps the number of SEV-SNP ASIDs at the maximum 3102 possible value, e.g. specifying -1u will assign all 3103 joint SEV-ES and SEV-SNP ASIDs to SEV-SNP. Note, 3104 assigning all joint ASIDs to SEV-SNP, i.e. configuring 3105 max_snp_asid == min_sev_asid-1, will effectively make 3106 SEV-ES unusable. 3107 3108 kvm-arm.mode= 3109 [KVM,ARM,EARLY] Select one of KVM/arm64's modes of 3110 operation. 3111 3112 none: Forcefully disable KVM. 3113 3114 nvhe: Standard nVHE-based mode, without support for 3115 protected guests. 3116 3117 protected: Mode with support for guests whose state is 3118 kept private from the host, using VHE or 3119 nVHE depending on HW support. 3120 3121 nested: VHE-based mode with support for nested 3122 virtualization. Requires at least ARMv8.4 3123 hardware (with FEAT_NV2). 3124 3125 Defaults to VHE/nVHE based on hardware support. Setting 3126 mode to "protected" will disable kexec and hibernation 3127 for the host. To force nVHE on VHE hardware, add 3128 "arm64_sw.hvhe=0 id_aa64mmfr1.vh=0" to the 3129 command-line. 3130 "nested" is experimental and should be used with 3131 extreme caution. 3132 3133 kvm-arm.vgic_v3_group0_trap= 3134 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-0 3135 system registers 3136 3137 kvm-arm.vgic_v3_group1_trap= 3138 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-1 3139 system registers 3140 3141 kvm-arm.vgic_v3_common_trap= 3142 [KVM,ARM,EARLY] Trap guest accesses to GICv3 common 3143 system registers 3144 3145 kvm-arm.vgic_v4_enable= 3146 [KVM,ARM,EARLY] Allow use of GICv4 for direct 3147 injection of LPIs. 3148 3149 kvm-arm.wfe_trap_policy= 3150 [KVM,ARM] Control when to set WFE instruction trap for 3151 KVM VMs. Traps are allowed but not guaranteed by the 3152 CPU architecture. 3153 3154 trap: set WFE instruction trap 3155 3156 notrap: clear WFE instruction trap 3157 3158 kvm-arm.wfi_trap_policy= 3159 [KVM,ARM] Control when to set WFI instruction trap for 3160 KVM VMs. Traps are allowed but not guaranteed by the 3161 CPU architecture. 3162 3163 trap: set WFI instruction trap 3164 3165 notrap: clear WFI instruction trap 3166 3167 kvm_cma_resv_ratio=n [PPC,EARLY] 3168 Reserves given percentage from system memory area for 3169 contiguous memory allocation for KVM hash pagetable 3170 allocation. 3171 By default it reserves 5% of total system memory. 3172 Format: <integer> 3173 Default: 5 3174 3175 kvm-intel.ept= [KVM,Intel] Control KVM's use of Extended Page Tables, 3176 a.k.a. Two-Dimensional Page Tables. Default is 1 3177 (enabled). Disable by KVM if hardware lacks support 3178 for EPT. 3179 3180 kvm-intel.emulate_invalid_guest_state= 3181 [KVM,Intel] Control whether to emulate invalid guest 3182 state. Ignored if kvm-intel.enable_unrestricted_guest=1, 3183 as guest state is never invalid for unrestricted 3184 guests. This param doesn't apply to nested guests (L2), 3185 as KVM never emulates invalid L2 guest state. 3186 Default is 1 (enabled). 3187 3188 kvm-intel.flexpriority= 3189 [KVM,Intel] Control KVM's use of FlexPriority feature 3190 (TPR shadow). Default is 1 (enabled). Disable by KVM if 3191 hardware lacks support for it. 3192 3193 kvm-intel.nested= 3194 [KVM,Intel] Control nested virtualization feature in 3195 KVM/VMX. Default is 1 (enabled). 3196 3197 kvm-intel.unrestricted_guest= 3198 [KVM,Intel] Control KVM's use of unrestricted guest 3199 feature (virtualized real and unpaged mode). Default 3200 is 1 (enabled). Disable by KVM if EPT is disabled or 3201 hardware lacks support for it. 3202 3203 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault 3204 CVE-2018-3620. 3205 3206 Valid arguments: never, cond, always 3207 3208 always: L1D cache flush on every VMENTER. 3209 cond: Flush L1D on VMENTER only when the code between 3210 VMEXIT and VMENTER can leak host memory. 3211 never: Disables the mitigation 3212 3213 Default is cond (do L1 cache flush in specific instances) 3214 3215 kvm-intel.vpid= [KVM,Intel] Control KVM's use of Virtual Processor 3216 Identification feature (tagged TLBs). Default is 1 3217 (enabled). Disable by KVM if hardware lacks support 3218 for it. 3219 3220 l1d_flush= [X86,INTEL,EARLY] 3221 Control mitigation for L1D based snooping vulnerability. 3222 3223 Certain CPUs are vulnerable to an exploit against CPU 3224 internal buffers which can forward information to a 3225 disclosure gadget under certain conditions. 3226 3227 In vulnerable processors, the speculatively 3228 forwarded data can be used in a cache side channel 3229 attack, to access data to which the attacker does 3230 not have direct access. 3231 3232 This parameter controls the mitigation. The 3233 options are: 3234 3235 on - enable the interface for the mitigation 3236 3237 l1tf= [X86,EARLY] Control mitigation of the L1TF vulnerability on 3238 affected CPUs 3239 3240 The kernel PTE inversion protection is unconditionally 3241 enabled and cannot be disabled. 3242 3243 full 3244 Provides all available mitigations for the 3245 L1TF vulnerability. Disables SMT and 3246 enables all mitigations in the 3247 hypervisors, i.e. unconditional L1D flush. 3248 3249 SMT control and L1D flush control via the 3250 sysfs interface is still possible after 3251 boot. Hypervisors will issue a warning 3252 when the first VM is started in a 3253 potentially insecure configuration, 3254 i.e. SMT enabled or L1D flush disabled. 3255 3256 full,force 3257 Same as 'full', but disables SMT and L1D 3258 flush runtime control. Implies the 3259 'nosmt=force' command line option. 3260 (i.e. sysfs control of SMT is disabled.) 3261 3262 flush 3263 Leaves SMT enabled and enables the default 3264 hypervisor mitigation, i.e. conditional 3265 L1D flush. 3266 3267 SMT control and L1D flush control via the 3268 sysfs interface is still possible after 3269 boot. Hypervisors will issue a warning 3270 when the first VM is started in a 3271 potentially insecure configuration, 3272 i.e. SMT enabled or L1D flush disabled. 3273 3274 flush,nosmt 3275 3276 Disables SMT and enables the default 3277 hypervisor mitigation. 3278 3279 SMT control and L1D flush control via the 3280 sysfs interface is still possible after 3281 boot. Hypervisors will issue a warning 3282 when the first VM is started in a 3283 potentially insecure configuration, 3284 i.e. SMT enabled or L1D flush disabled. 3285 3286 flush,nowarn 3287 Same as 'flush', but hypervisors will not 3288 warn when a VM is started in a potentially 3289 insecure configuration. 3290 3291 off 3292 Disables hypervisor mitigations and doesn't 3293 emit any warnings. 3294 It also drops the swap size and available 3295 RAM limit restriction on both hypervisor and 3296 bare metal. 3297 3298 Default is 'flush'. 3299 3300 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst 3301 3302 l2cr= [PPC] 3303 3304 l3cr= [PPC] 3305 3306 lapic [X86-32,APIC,EARLY] Enable the local APIC even if BIOS 3307 disabled it. 3308 3309 lapic= [X86,APIC] Do not use TSC deadline 3310 value for LAPIC timer one-shot implementation. Default 3311 back to the programmable timer unit in the LAPIC. 3312 Format: notscdeadline 3313 3314 lapic_timer_c2_ok [X86,APIC,EARLY] trust the local apic timer 3315 in C2 power state. 3316 3317 libata.dma= [LIBATA] DMA control 3318 libata.dma=0 Disable all PATA and SATA DMA 3319 libata.dma=1 PATA and SATA Disk DMA only 3320 libata.dma=2 ATAPI (CDROM) DMA only 3321 libata.dma=4 Compact Flash DMA only 3322 Combinations also work, so libata.dma=3 enables DMA 3323 for disks and CDROMs, but not CFs. 3324 3325 libata.ignore_hpa= [LIBATA] Ignore HPA limit 3326 libata.ignore_hpa=0 keep BIOS limits (default) 3327 libata.ignore_hpa=1 ignore limits, using full disk 3328 3329 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume 3330 when set. 3331 Format: <int> 3332 3333 libata.force= [LIBATA] Force configurations. The format is a comma- 3334 separated list of "[ID:]VAL" where ID is PORT[.DEVICE]. 3335 PORT and DEVICE are decimal numbers matching port, link 3336 or device. Basically, it matches the ATA ID string 3337 printed on console by libata. If the whole ID part is 3338 omitted, the last PORT and DEVICE values are used. If 3339 ID hasn't been specified yet, the configuration applies 3340 to all ports, links and devices. 3341 3342 If only DEVICE is omitted, the parameter applies to 3343 the port and all links and devices behind it. DEVICE 3344 number of 0 either selects the first device or the 3345 first fan-out link behind PMP device. It does not 3346 select the host link. DEVICE number of 15 selects the 3347 host link and device attached to it. 3348 3349 The VAL specifies the configuration to force. As long 3350 as there is no ambiguity, shortcut notation is allowed. 3351 For example, both 1.5 and 1.5G would work for 1.5Gbps. 3352 The following configurations can be forced. 3353 3354 * Cable type: 40c, 80c, short40c, unk, ign or sata. 3355 Any ID with matching PORT is used. 3356 3357 * SATA link speed limit: 1.5Gbps or 3.0Gbps. 3358 3359 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7]. 3360 udma[/][16,25,33,44,66,100,133] notation is also 3361 allowed. 3362 3363 * nohrst, nosrst, norst: suppress hard, soft and both 3364 resets. 3365 3366 * rstonce: only attempt one reset during hot-unplug 3367 link recovery. 3368 3369 * [no]dbdelay: Enable or disable the extra 200ms delay 3370 before debouncing a link PHY and device presence 3371 detection. 3372 3373 * [no]ncq: Turn on or off NCQ. 3374 3375 * [no]ncqtrim: Enable or disable queued DSM TRIM. 3376 3377 * [no]ncqati: Enable or disable NCQ trim on ATI chipset. 3378 3379 * [no]trim: Enable or disable (unqueued) TRIM. 3380 3381 * trim_zero: Indicate that TRIM command zeroes data. 3382 3383 * max_trim_128m: Set 128M maximum trim size limit. 3384 3385 * [no]dma: Turn on or off DMA transfers. 3386 3387 * atapi_dmadir: Enable ATAPI DMADIR bridge support. 3388 3389 * atapi_mod16_dma: Enable the use of ATAPI DMA for 3390 commands that are not a multiple of 16 bytes. 3391 3392 * [no]dmalog: Enable or disable the use of the 3393 READ LOG DMA EXT command to access logs. 3394 3395 * [no]iddevlog: Enable or disable access to the 3396 identify device data log. 3397 3398 * [no]logdir: Enable or disable access to the general 3399 purpose log directory. 3400 3401 * max_sec_128: Set transfer size limit to 128 sectors. 3402 3403 * max_sec_1024: Set or clear transfer size limit to 3404 1024 sectors. 3405 3406 * max_sec_lba48: Set or clear transfer size limit to 3407 65535 sectors. 3408 3409 * external: Mark port as external (hotplug-capable). 3410 3411 * [no]lpm: Enable or disable link power management. 3412 3413 * [no]setxfer: Indicate if transfer speed mode setting 3414 should be skipped. 3415 3416 * [no]fua: Disable or enable FUA (Force Unit Access) 3417 support for devices supporting this feature. 3418 3419 * dump_id: Dump IDENTIFY data. 3420 3421 * disable: Disable this device. 3422 3423 If there are multiple matching configurations changing 3424 the same attribute, the last one is used. 3425 3426 load_ramdisk= [RAM] [Deprecated] 3427 3428 lockd.nlm_grace_period=P [NFS] Assign grace period. 3429 Format: <integer> 3430 3431 lockd.nlm_tcpport=N [NFS] Assign TCP port. 3432 Format: <integer> 3433 3434 lockd.nlm_timeout=T [NFS] Assign timeout value. 3435 Format: <integer> 3436 3437 lockd.nlm_udpport=M [NFS] Assign UDP port. 3438 Format: <integer> 3439 3440 lockdown= [SECURITY,EARLY] 3441 { integrity | confidentiality } 3442 Enable the kernel lockdown feature. If set to 3443 integrity, kernel features that allow userland to 3444 modify the running kernel are disabled. If set to 3445 confidentiality, kernel features that allow userland 3446 to extract confidential information from the kernel 3447 are also disabled. 3448 3449 locktorture.acq_writer_lim= [KNL] 3450 Set the time limit in jiffies for a lock 3451 acquisition. Acquisitions exceeding this limit 3452 will result in a splat once they do complete. 3453 3454 locktorture.bind_readers= [KNL] 3455 Specify the list of CPUs to which the readers are 3456 to be bound. 3457 3458 locktorture.bind_writers= [KNL] 3459 Specify the list of CPUs to which the writers are 3460 to be bound. 3461 3462 locktorture.call_rcu_chains= [KNL] 3463 Specify the number of self-propagating call_rcu() 3464 chains to set up. These are used to ensure that 3465 there is a high probability of an RCU grace period 3466 in progress at any given time. Defaults to 0, 3467 which disables these call_rcu() chains. 3468 3469 locktorture.long_hold= [KNL] 3470 Specify the duration in milliseconds for the 3471 occasional long-duration lock hold time. Defaults 3472 to 100 milliseconds. Select 0 to disable. 3473 3474 locktorture.nested_locks= [KNL] 3475 Specify the maximum lock nesting depth that 3476 locktorture is to exercise, up to a limit of 8 3477 (MAX_NESTED_LOCKS). Specify zero to disable. 3478 Note that this parameter is ineffective on types 3479 of locks that do not support nested acquisition. 3480 3481 locktorture.nreaders_stress= [KNL] 3482 Set the number of locking read-acquisition kthreads. 3483 Defaults to being automatically set based on the 3484 number of online CPUs. 3485 3486 locktorture.nwriters_stress= [KNL] 3487 Set the number of locking write-acquisition kthreads. 3488 3489 locktorture.onoff_holdoff= [KNL] 3490 Set time (s) after boot for CPU-hotplug testing. 3491 3492 locktorture.onoff_interval= [KNL] 3493 Set time (s) between CPU-hotplug operations, or 3494 zero to disable CPU-hotplug testing. 3495 3496 locktorture.rt_boost= [KNL] 3497 Do periodic testing of real-time lock priority 3498 boosting. Select 0 to disable, 1 to boost 3499 only rt_mutex, and 2 to boost unconditionally. 3500 Defaults to 2, which might seem to be an 3501 odd choice, but which should be harmless for 3502 non-real-time spinlocks, due to their disabling 3503 of preemption. Note that non-realtime mutexes 3504 disable boosting. 3505 3506 locktorture.rt_boost_factor= [KNL] 3507 Number that determines how often and for how 3508 long priority boosting is exercised. This is 3509 scaled down by the number of writers, so that the 3510 number of boosts per unit time remains roughly 3511 constant as the number of writers increases. 3512 On the other hand, the duration of each boost 3513 increases with the number of writers. 3514 3515 locktorture.shuffle_interval= [KNL] 3516 Set task-shuffle interval (jiffies). Shuffling 3517 tasks allows some CPUs to go into dyntick-idle 3518 mode during the locktorture test. 3519 3520 locktorture.shutdown_secs= [KNL] 3521 Set time (s) after boot system shutdown. This 3522 is useful for hands-off automated testing. 3523 3524 locktorture.stat_interval= [KNL] 3525 Time (s) between statistics printk()s. 3526 3527 locktorture.stutter= [KNL] 3528 Time (s) to stutter testing, for example, 3529 specifying five seconds causes the test to run for 3530 five seconds, wait for five seconds, and so on. 3531 This tests the locking primitive's ability to 3532 transition abruptly to and from idle. 3533 3534 locktorture.torture_type= [KNL] 3535 Specify the locking implementation to test. 3536 3537 locktorture.verbose= [KNL] 3538 Enable additional printk() statements. 3539 3540 locktorture.writer_fifo= [KNL] 3541 Run the write-side locktorture kthreads at 3542 sched_set_fifo() real-time priority. 3543 3544 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver 3545 Format: <irq> 3546 3547 loglevel= [KNL,EARLY] 3548 All Kernel Messages with a loglevel smaller than the 3549 console loglevel will be printed to the console. It can 3550 also be changed with klogd or other programs. The 3551 loglevels are defined as follows: 3552 3553 0 (KERN_EMERG) system is unusable 3554 1 (KERN_ALERT) action must be taken immediately 3555 2 (KERN_CRIT) critical conditions 3556 3 (KERN_ERR) error conditions 3557 4 (KERN_WARNING) warning conditions 3558 5 (KERN_NOTICE) normal but significant condition 3559 6 (KERN_INFO) informational 3560 7 (KERN_DEBUG) debug-level messages 3561 3562 log_buf_len=n[KMG] [KNL,EARLY] 3563 Sets the size of the printk ring buffer, in bytes. 3564 n must be a power of two and greater than the 3565 minimal size. The minimal size is defined by 3566 LOG_BUF_SHIFT kernel config parameter. There 3567 is also CONFIG_LOG_CPU_MAX_BUF_SHIFT config 3568 parameter that allows to increase the default size 3569 depending on the number of CPUs. See init/Kconfig 3570 for more details. 3571 3572 logo.nologo [FB] Disables display of the built-in Linux logo. 3573 This may be used to provide more screen space for 3574 kernel log messages and is useful when debugging 3575 kernel boot problems. 3576 3577 lp=0 [LP] Specify parallel ports to use, e.g, 3578 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses 3579 lp=reset first parallel port). 'lp=0' disables the 3580 lp=auto printer driver. 'lp=reset' (which can be 3581 specified in addition to the ports) causes 3582 attached printers to be reset. Using 3583 lp=port1,port2,... specifies the parallel ports 3584 to associate lp devices with, starting with 3585 lp0. A port specification may be 'none' to skip 3586 that lp device, or a parport name such as 3587 'parport0'. Specifying 'lp=auto' instead of a 3588 port specification list means that device IDs 3589 from each port should be examined, to see if 3590 an IEEE 1284-compliant printer is attached; if 3591 so, the driver will manage that printer. 3592 See also header of drivers/char/lp.c. 3593 3594 lpj=n [KNL] 3595 Sets loops_per_jiffy to given constant, thus avoiding 3596 time-consuming boot-time autodetection (up to 250 ms per 3597 CPU). 0 enables autodetection (default). To determine 3598 the correct value for your kernel, boot with normal 3599 autodetection and see what value is printed. Note that 3600 on SMP systems the preset will be applied to all CPUs, 3601 which is likely to cause problems if your CPUs need 3602 significantly divergent settings. An incorrect value 3603 will cause delays in the kernel to be wrong, leading to 3604 unpredictable I/O errors and other breakage. Although 3605 unlikely, in the extreme case this might damage your 3606 hardware. 3607 3608 lsm.debug [SECURITY] Enable LSM initialization debugging output. 3609 3610 lsm=lsm1,...,lsmN 3611 [SECURITY] Choose order of LSM initialization. This 3612 overrides CONFIG_LSM, and the "security=" parameter. 3613 3614 machtype= [Loongson] Share the same kernel image file between 3615 different yeeloong laptops. 3616 Example: machtype=lemote-yeeloong-2f-7inch 3617 3618 maxcpus= [SMP,EARLY] Maximum number of processors that an SMP kernel 3619 will bring up during bootup. maxcpus=n : n >= 0 limits 3620 the kernel to bring up 'n' processors. Surely after 3621 bootup you can bring up the other plugged cpu by executing 3622 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus 3623 only takes effect during system bootup. 3624 While n=0 is a special case, it is equivalent to "nosmp", 3625 which also disables the IO APIC. 3626 3627 max_loop= [LOOP] The number of loop block devices that get 3628 (loop.max_loop) unconditionally pre-created at init time. The default 3629 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead 3630 of statically allocating a predefined number, loop 3631 devices can be requested on-demand with the 3632 /dev/loop-control interface. 3633 3634 mce= [X86-{32,64}] 3635 3636 Please see Documentation/arch/x86/x86_64/machinecheck.rst for sysfs runtime tunables. 3637 3638 off 3639 disable machine check 3640 3641 no_cmci 3642 disable CMCI(Corrected Machine Check Interrupt) that 3643 Intel processor supports. Usually this disablement is 3644 not recommended, but it might be handy if your 3645 hardware is misbehaving. 3646 3647 Note that you'll get more problems without CMCI than 3648 with due to the shared banks, i.e. you might get 3649 duplicated error logs. 3650 3651 dont_log_ce 3652 don't make logs for corrected errors. All events 3653 reported as corrected are silently cleared by OS. This 3654 option will be useful if you have no interest in any 3655 of corrected errors. 3656 3657 ignore_ce 3658 disable features for corrected errors, e.g. 3659 polling timer and CMCI. All events reported as 3660 corrected are not cleared by OS and remained in its 3661 error banks. 3662 3663 Usually this disablement is not recommended, however 3664 if there is an agent checking/clearing corrected 3665 errors (e.g. BIOS or hardware monitoring 3666 applications), conflicting with OS's error handling, 3667 and you cannot deactivate the agent, then this option 3668 will be a help. 3669 3670 no_lmce 3671 do not opt-in to Local MCE delivery. Use legacy method 3672 to broadcast MCEs. 3673 3674 bootlog 3675 enable logging of machine checks left over from 3676 booting. Disabled by default on AMD Fam10h and older 3677 because some BIOS leave bogus ones. 3678 3679 If your BIOS doesn't do that it's a good idea to 3680 enable though to make sure you log even machine check 3681 events that result in a reboot. On Intel systems it is 3682 enabled by default. 3683 3684 nobootlog 3685 disable boot machine check logging. 3686 3687 monarchtimeout (number) 3688 sets the time in us to wait for other CPUs on machine 3689 checks. 0 to disable. 3690 3691 bios_cmci_threshold 3692 don't overwrite the bios-set CMCI threshold. This boot 3693 option prevents Linux from overwriting the CMCI 3694 threshold set by the bios. Without this option, Linux 3695 always sets the CMCI threshold to 1. Enabling this may 3696 make memory predictive failure analysis less effective 3697 if the bios sets thresholds for memory errors since we 3698 will not see details for all errors. 3699 3700 recovery 3701 force-enable recoverable machine check code paths 3702 3703 Everything else is in sysfs now. 3704 3705 3706 md= [HW] RAID subsystems devices and level 3707 See Documentation/admin-guide/md.rst. 3708 3709 mdacon= [MDA] 3710 Format: <first>,<last> 3711 Specifies range of consoles to be captured by the MDA. 3712 3713 mds= [X86,INTEL,EARLY] 3714 Control mitigation for the Micro-architectural Data 3715 Sampling (MDS) vulnerability. 3716 3717 Certain CPUs are vulnerable to an exploit against CPU 3718 internal buffers which can forward information to a 3719 disclosure gadget under certain conditions. 3720 3721 In vulnerable processors, the speculatively 3722 forwarded data can be used in a cache side channel 3723 attack, to access data to which the attacker does 3724 not have direct access. 3725 3726 This parameter controls the MDS mitigation. The 3727 options are: 3728 3729 full - Enable MDS mitigation on vulnerable CPUs 3730 full,nosmt - Enable MDS mitigation and disable 3731 SMT on vulnerable CPUs 3732 off - Unconditionally disable MDS mitigation 3733 3734 On TAA-affected machines, mds=off can be prevented by 3735 an active TAA mitigation as both vulnerabilities are 3736 mitigated with the same mechanism so in order to disable 3737 this mitigation, you need to specify tsx_async_abort=off 3738 too. 3739 3740 Not specifying this option is equivalent to 3741 mds=full. 3742 3743 For details see: Documentation/admin-guide/hw-vuln/mds.rst 3744 3745 mem=nn[KMG] [HEXAGON,EARLY] Set the memory size. 3746 Must be specified, otherwise memory size will be 0. 3747 3748 mem=nn[KMG] [KNL,BOOT,EARLY] Force usage of a specific amount 3749 of memory Amount of memory to be used in cases 3750 as follows: 3751 3752 1 for test; 3753 2 when the kernel is not able to see the whole system memory; 3754 3 memory that lies after 'mem=' boundary is excluded from 3755 the hypervisor, then assigned to KVM guests. 3756 4 to limit the memory available for kdump kernel. 3757 3758 [ARC,MICROBLAZE] - the limit applies only to low memory, 3759 high memory is not affected. 3760 3761 [ARM64] - only limits memory covered by the linear 3762 mapping. The NOMAP regions are not affected. 3763 3764 [X86] Work as limiting max address. Use together 3765 with memmap= to avoid physical address space collisions. 3766 Without memmap= PCI devices could be placed at addresses 3767 belonging to unused RAM. 3768 3769 Note that this only takes effects during boot time since 3770 in above case 3, memory may need be hot added after boot 3771 if system memory of hypervisor is not sufficient. 3772 3773 mem=nn[KMG]@ss[KMG] 3774 [ARM,MIPS,EARLY] - override the memory layout 3775 reported by firmware. 3776 Define a memory region of size nn[KMG] starting at 3777 ss[KMG]. 3778 Multiple different regions can be specified with 3779 multiple mem= parameters on the command line. 3780 3781 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel 3782 memory. 3783 3784 memblock=debug [KNL,EARLY] Enable memblock debug messages. 3785 3786 memchunk=nn[KMG] 3787 [KNL,SH] Allow user to override the default size for 3788 per-device physically contiguous DMA buffers. 3789 3790 memhp_default_state=online/offline/online_kernel/online_movable 3791 [KNL] Set the initial state for the memory hotplug 3792 onlining policy. If not specified, the default value is 3793 set according to the 3794 CONFIG_MHP_DEFAULT_ONLINE_TYPE kernel config 3795 options. 3796 See Documentation/admin-guide/mm/memory-hotplug.rst. 3797 3798 memmap=exactmap [KNL,X86,EARLY] Enable setting of an exact 3799 E820 memory map, as specified by the user. 3800 Such memmap=exactmap lines can be constructed based on 3801 BIOS output or other requirements. See the memmap=nn@ss 3802 option description. 3803 3804 memmap=nn[KMG]@ss[KMG] 3805 [KNL, X86,MIPS,XTENSA,EARLY] Force usage of a specific region of memory. 3806 Region of memory to be used is from ss to ss+nn. 3807 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG], 3808 which limits max address to nn[KMG]. 3809 Multiple different regions can be specified, 3810 comma delimited. 3811 Example: 3812 memmap=100M@2G,100M#3G,1G!1024G 3813 3814 memmap=nn[KMG]#ss[KMG] 3815 [KNL,ACPI,EARLY] Mark specific memory as ACPI data. 3816 Region of memory to be marked is from ss to ss+nn. 3817 3818 memmap=nn[KMG]$ss[KMG] 3819 [KNL,ACPI,EARLY] Mark specific memory as reserved. 3820 Region of memory to be reserved is from ss to ss+nn. 3821 Example: Exclude memory from 0x18690000-0x1869ffff 3822 memmap=64K$0x18690000 3823 or 3824 memmap=0x10000$0x18690000 3825 Some bootloaders may need an escape character before '$', 3826 like Grub2, otherwise '$' and the following number 3827 will be eaten. 3828 3829 memmap=nn[KMG]!ss[KMG,EARLY] 3830 [KNL,X86] Mark specific memory as protected. 3831 Region of memory to be used, from ss to ss+nn. 3832 The memory region may be marked as e820 type 12 (0xc) 3833 and is NVDIMM or ADR memory. 3834 3835 memmap=<size>%<offset>-<oldtype>+<newtype> 3836 [KNL,ACPI,EARLY] Convert memory within the specified region 3837 from <oldtype> to <newtype>. If "-<oldtype>" is left 3838 out, the whole region will be marked as <newtype>, 3839 even if previously unavailable. If "+<newtype>" is left 3840 out, matching memory will be removed. Types are 3841 specified as e820 types, e.g., 1 = RAM, 2 = reserved, 3842 3 = ACPI, 12 = PRAM. 3843 3844 memory_corruption_check=0/1 [X86,EARLY] 3845 Some BIOSes seem to corrupt the first 64k of 3846 memory when doing things like suspend/resume. 3847 Setting this option will scan the memory 3848 looking for corruption. Enabling this will 3849 both detect corruption and prevent the kernel 3850 from using the memory being corrupted. 3851 However, it's intended as a diagnostic tool; if 3852 repeatable BIOS-originated corruption always 3853 affects the same memory, you can use memmap= 3854 to prevent the kernel from using that memory. 3855 3856 memory_corruption_check_size=size [X86,EARLY] 3857 By default it checks for corruption in the low 3858 64k, making this memory unavailable for normal 3859 use. Use this parameter to scan for 3860 corruption in more or less memory. 3861 3862 memory_corruption_check_period=seconds [X86,EARLY] 3863 By default it checks for corruption every 60 3864 seconds. Use this parameter to check at some 3865 other rate. 0 disables periodic checking. 3866 3867 memory_hotplug.memmap_on_memory 3868 [KNL,X86,ARM] Boolean flag to enable this feature. 3869 Format: {on | off (default)} 3870 When enabled, runtime hotplugged memory will 3871 allocate its internal metadata (struct pages, 3872 those vmemmap pages cannot be optimized even 3873 if hugetlb_free_vmemmap is enabled) from the 3874 hotadded memory which will allow to hotadd a 3875 lot of memory without requiring additional 3876 memory to do so. 3877 This feature is disabled by default because it 3878 has some implication on large (e.g. GB) 3879 allocations in some configurations (e.g. small 3880 memory blocks). 3881 The state of the flag can be read in 3882 /sys/module/memory_hotplug/parameters/memmap_on_memory. 3883 Note that even when enabled, there are a few cases where 3884 the feature is not effective. 3885 3886 memtest= [KNL,X86,ARM,M68K,PPC,RISCV,EARLY] Enable memtest 3887 Format: <integer> 3888 default : 0 <disable> 3889 Specifies the number of memtest passes to be 3890 performed. Each pass selects another test 3891 pattern from a given set of patterns. Memtest 3892 fills the memory with this pattern, validates 3893 memory contents and reserves bad memory 3894 regions that are detected. 3895 3896 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control 3897 Valid arguments: on, off 3898 Default: off 3899 mem_encrypt=on: Activate SME 3900 mem_encrypt=off: Do not activate SME 3901 3902 Refer to Documentation/virt/kvm/x86/amd-memory-encryption.rst 3903 for details on when memory encryption can be activated. 3904 3905 mem_sleep_default= [SUSPEND] Default system suspend mode: 3906 s2idle - Suspend-To-Idle 3907 shallow - Power-On Suspend or equivalent (if supported) 3908 deep - Suspend-To-RAM or equivalent (if supported) 3909 See Documentation/admin-guide/pm/sleep-states.rst. 3910 3911 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when 3912 the BIOS has incorrectly applied a workaround. TinyBIOS 3913 version 0.98 is known to be affected, 0.99 fixes the 3914 problem by letting the user disable the workaround. 3915 3916 mga= [HW,DRM] 3917 3918 microcode= [X86] Control the behavior of the microcode loader. 3919 Available options, comma separated: 3920 3921 base_rev=X - with <X> with format: <u32> 3922 Set the base microcode revision of each thread when in 3923 debug mode. 3924 3925 dis_ucode_ldr: disable the microcode loader 3926 3927 force_minrev: 3928 Enable or disable the microcode minimal revision 3929 enforcement for the runtime microcode loader. 3930 3931 mini2440= [ARM,HW,KNL] 3932 Format:[0..2][b][c][t] 3933 Default: "0tb" 3934 MINI2440 configuration specification: 3935 0 - The attached screen is the 3.5" TFT 3936 1 - The attached screen is the 7" TFT 3937 2 - The VGA Shield is attached (1024x768) 3938 Leaving out the screen size parameter will not load 3939 the TFT driver, and the framebuffer will be left 3940 unconfigured. 3941 b - Enable backlight. The TFT backlight pin will be 3942 linked to the kernel VESA blanking code and a GPIO 3943 LED. This parameter is not necessary when using the 3944 VGA shield. 3945 c - Enable the s3c camera interface. 3946 t - Reserved for enabling touchscreen support. The 3947 touchscreen support is not enabled in the mainstream 3948 kernel as of 2.6.30, a preliminary port can be found 3949 in the "bleeding edge" mini2440 support kernel at 3950 https://repo.or.cz/w/linux-2.6/mini2440.git 3951 3952 mitigations= 3953 [X86,PPC,S390,ARM64,EARLY] Control optional mitigations for 3954 CPU vulnerabilities. This is a set of curated, 3955 arch-independent options, each of which is an 3956 aggregation of existing arch-specific options. 3957 3958 Note, "mitigations" is supported if and only if the 3959 kernel was built with CPU_MITIGATIONS=y. 3960 3961 off 3962 Disable all optional CPU mitigations. This 3963 improves system performance, but it may also 3964 expose users to several CPU vulnerabilities. 3965 Equivalent to: if nokaslr then kpti=0 [ARM64] 3966 gather_data_sampling=off [X86] 3967 indirect_target_selection=off [X86] 3968 kvm.nx_huge_pages=off [X86] 3969 l1tf=off [X86] 3970 mds=off [X86] 3971 mmio_stale_data=off [X86] 3972 no_entry_flush [PPC] 3973 no_uaccess_flush [PPC] 3974 nobp=0 [S390] 3975 nopti [X86,PPC] 3976 nospectre_bhb [ARM64] 3977 nospectre_v1 [X86,PPC] 3978 nospectre_v2 [X86,PPC,S390,ARM64] 3979 reg_file_data_sampling=off [X86] 3980 retbleed=off [X86] 3981 spec_rstack_overflow=off [X86] 3982 spec_store_bypass_disable=off [X86,PPC] 3983 spectre_bhi=off [X86] 3984 spectre_v2_user=off [X86] 3985 srbds=off [X86,INTEL] 3986 ssbd=force-off [ARM64] 3987 tsx_async_abort=off [X86] 3988 vmscape=off [X86] 3989 3990 Exceptions: 3991 This does not have any effect on 3992 kvm.nx_huge_pages when 3993 kvm.nx_huge_pages=force. 3994 3995 auto (default) 3996 Mitigate all CPU vulnerabilities, but leave SMT 3997 enabled, even if it's vulnerable. This is for 3998 users who don't want to be surprised by SMT 3999 getting disabled across kernel upgrades, or who 4000 have other ways of avoiding SMT-based attacks. 4001 Equivalent to: (default behavior) 4002 4003 auto,nosmt 4004 Mitigate all CPU vulnerabilities, disabling SMT 4005 if needed. This is for users who always want to 4006 be fully mitigated, even if it means losing SMT. 4007 Equivalent to: l1tf=flush,nosmt [X86] 4008 mds=full,nosmt [X86] 4009 tsx_async_abort=full,nosmt [X86] 4010 mmio_stale_data=full,nosmt [X86] 4011 retbleed=auto,nosmt [X86] 4012 4013 [X86] After one of the above options, additionally 4014 supports attack-vector based controls as documented in 4015 Documentation/admin-guide/hw-vuln/attack_vector_controls.rst 4016 4017 mminit_loglevel= 4018 [KNL,EARLY] When CONFIG_DEBUG_MEMORY_INIT is set, this 4019 parameter allows control of the logging verbosity for 4020 the additional memory initialisation checks. A value 4021 of 0 disables mminit logging and a level of 4 will 4022 log everything. Information is printed at KERN_DEBUG 4023 so loglevel=8 may also need to be specified. 4024 4025 mmio_stale_data= 4026 [X86,INTEL,EARLY] Control mitigation for the Processor 4027 MMIO Stale Data vulnerabilities. 4028 4029 Processor MMIO Stale Data is a class of 4030 vulnerabilities that may expose data after an MMIO 4031 operation. Exposed data could originate or end in 4032 the same CPU buffers as affected by MDS and TAA. 4033 Therefore, similar to MDS and TAA, the mitigation 4034 is to clear the affected CPU buffers. 4035 4036 This parameter controls the mitigation. The 4037 options are: 4038 4039 full - Enable mitigation on vulnerable CPUs 4040 4041 full,nosmt - Enable mitigation and disable SMT on 4042 vulnerable CPUs. 4043 4044 off - Unconditionally disable mitigation 4045 4046 On MDS or TAA affected machines, 4047 mmio_stale_data=off can be prevented by an active 4048 MDS or TAA mitigation as these vulnerabilities are 4049 mitigated with the same mechanism so in order to 4050 disable this mitigation, you need to specify 4051 mds=off and tsx_async_abort=off too. 4052 4053 Not specifying this option is equivalent to 4054 mmio_stale_data=full. 4055 4056 For details see: 4057 Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst 4058 4059 <module>.async_probe[=<bool>] [KNL] 4060 If no <bool> value is specified or if the value 4061 specified is not a valid <bool>, enable asynchronous 4062 probe on this module. Otherwise, enable/disable 4063 asynchronous probe on this module as indicated by the 4064 <bool> value. See also: module.async_probe 4065 4066 module.async_probe=<bool> 4067 [KNL] When set to true, modules will use async probing 4068 by default. To enable/disable async probing for a 4069 specific module, use the module specific control that 4070 is documented under <module>.async_probe. When both 4071 module.async_probe and <module>.async_probe are 4072 specified, <module>.async_probe takes precedence for 4073 the specific module. 4074 4075 module.enable_dups_trace 4076 [KNL] When CONFIG_MODULE_DEBUG_AUTOLOAD_DUPS is set, 4077 this means that duplicate request_module() calls will 4078 trigger a WARN_ON() instead of a pr_warn(). Note that 4079 if MODULE_DEBUG_AUTOLOAD_DUPS_TRACE is set, WARN_ON()s 4080 will always be issued and this option does nothing. 4081 module.sig_enforce 4082 [KNL] When CONFIG_MODULE_SIG is set, this means that 4083 modules without (valid) signatures will fail to load. 4084 Note that if CONFIG_MODULE_SIG_FORCE is set, that 4085 is always true, so this option does nothing. 4086 4087 module_blacklist= [KNL] Do not load a comma-separated list of 4088 modules. Useful for debugging problem modules. 4089 4090 mousedev.tap_time= 4091 [MOUSE] Maximum time between finger touching and 4092 leaving touchpad surface for touch to be considered 4093 a tap and be reported as a left button click (for 4094 touchpads working in absolute mode only). 4095 Format: <msecs> 4096 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices 4097 reporting absolute coordinates, such as tablets 4098 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices 4099 reporting absolute coordinates, such as tablets 4100 4101 movablecore= [KNL,X86,PPC,EARLY] 4102 Format: nn[KMGTPE] | nn% 4103 This parameter is the complement to kernelcore=, it 4104 specifies the amount of memory used for migratable 4105 allocations. If both kernelcore and movablecore is 4106 specified, then kernelcore will be at *least* the 4107 specified value but may be more. If movablecore on its 4108 own is specified, the administrator must be careful 4109 that the amount of memory usable for all allocations 4110 is not too small. 4111 4112 movable_node [KNL,EARLY] Boot-time switch to make hotplugable memory 4113 NUMA nodes to be movable. This means that the memory 4114 of such nodes will be usable only for movable 4115 allocations which rules out almost all kernel 4116 allocations. Use with caution! 4117 4118 MTD_Partition= [MTD] 4119 Format: <name>,<region-number>,<size>,<offset> 4120 4121 MTD_Region= [MTD] Format: 4122 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>] 4123 4124 mtdparts= [MTD] 4125 See drivers/mtd/parsers/cmdlinepart.c 4126 4127 mtouchusb.raw_coordinates= 4128 [HW] Make the MicroTouch USB driver use raw coordinates 4129 ('y', default) or cooked coordinates ('n') 4130 4131 mtrr=debug [X86,EARLY] 4132 Enable printing debug information related to MTRR 4133 registers at boot time. 4134 4135 mtrr_chunk_size=nn[KMG,X86,EARLY] 4136 used for mtrr cleanup. It is largest continuous chunk 4137 that could hold holes aka. UC entries. 4138 4139 mtrr_gran_size=nn[KMG,X86,EARLY] 4140 Used for mtrr cleanup. It is granularity of mtrr block. 4141 Default is 1. 4142 Large value could prevent small alignment from 4143 using up MTRRs. 4144 4145 mtrr_spare_reg_nr=n [X86,EARLY] 4146 Format: <integer> 4147 Range: 0,7 : spare reg number 4148 Default : 1 4149 Used for mtrr cleanup. It is spare mtrr entries number. 4150 Set to 2 or more if your graphical card needs more. 4151 4152 multitce=off [PPC] This parameter disables the use of the pSeries 4153 firmware feature for updating multiple TCE entries 4154 at a time. 4155 4156 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card 4157 4158 netdev= [NET] Network devices parameters 4159 Format: <irq>,<io>,<mem_start>,<mem_end>,<name> 4160 Note that mem_start is often overloaded to mean 4161 something different and driver-specific. 4162 This usage is only documented in each driver source 4163 file if at all. 4164 4165 netpoll.carrier_timeout= 4166 [NET] Specifies amount of time (in seconds) that 4167 netpoll should wait for a carrier. By default netpoll 4168 waits 4 seconds. 4169 4170 nf_conntrack.acct= 4171 [NETFILTER] Enable connection tracking flow accounting 4172 0 to disable accounting 4173 1 to enable accounting 4174 Default value is 0. 4175 4176 nfs.cache_getent= 4177 [NFS] sets the pathname to the program which is used 4178 to update the NFS client cache entries. 4179 4180 nfs.cache_getent_timeout= 4181 [NFS] sets the timeout after which an attempt to 4182 update a cache entry is deemed to have failed. 4183 4184 nfs.callback_nr_threads= 4185 [NFSv4] set the total number of threads that the 4186 NFS client will assign to service NFSv4 callback 4187 requests. 4188 4189 nfs.callback_tcpport= 4190 [NFS] set the TCP port on which the NFSv4 callback 4191 channel should listen. 4192 4193 nfs.delay_retrans= 4194 [NFS] specifies the number of times the NFSv4 client 4195 retries the request before returning an EAGAIN error, 4196 after a reply of NFS4ERR_DELAY from the server. 4197 Only applies if the softerr mount option is enabled, 4198 and the specified value is >= 0. 4199 4200 nfs.enable_ino64= 4201 [NFS] enable 64-bit inode numbers. 4202 If zero, the NFS client will fake up a 32-bit inode 4203 number for the readdir() and stat() syscalls instead 4204 of returning the full 64-bit number. 4205 The default is to return 64-bit inode numbers. 4206 4207 nfs.idmap_cache_timeout= 4208 [NFS] set the maximum lifetime for idmapper cache 4209 entries. 4210 4211 nfs.max_session_cb_slots= 4212 [NFSv4.1] Sets the maximum number of session 4213 slots the client will assign to the callback 4214 channel. This determines the maximum number of 4215 callbacks the client will process in parallel for 4216 a particular server. 4217 4218 nfs.max_session_slots= 4219 [NFSv4.1] Sets the maximum number of session slots 4220 the client will attempt to negotiate with the server. 4221 This limits the number of simultaneous RPC requests 4222 that the client can send to the NFSv4.1 server. 4223 Note that there is little point in setting this 4224 value higher than the max_tcp_slot_table_limit. 4225 4226 nfs.nfs4_disable_idmapping= 4227 [NFSv4] When set to the default of '1', this option 4228 ensures that both the RPC level authentication 4229 scheme and the NFS level operations agree to use 4230 numeric uids/gids if the mount is using the 4231 'sec=sys' security flavour. In effect it is 4232 disabling idmapping, which can make migration from 4233 legacy NFSv2/v3 systems to NFSv4 easier. 4234 Servers that do not support this mode of operation 4235 will be autodetected by the client, and it will fall 4236 back to using the idmapper. 4237 To turn off this behaviour, set the value to '0'. 4238 4239 nfs.nfs4_unique_id= 4240 [NFS4] Specify an additional fixed unique ident- 4241 ification string that NFSv4 clients can insert into 4242 their nfs_client_id4 string. This is typically a 4243 UUID that is generated at system install time. 4244 4245 nfs.recover_lost_locks= 4246 [NFSv4] Attempt to recover locks that were lost due 4247 to a lease timeout on the server. Please note that 4248 doing this risks data corruption, since there are 4249 no guarantees that the file will remain unchanged 4250 after the locks are lost. 4251 If you want to enable the kernel legacy behaviour of 4252 attempting to recover these locks, then set this 4253 parameter to '1'. 4254 The default parameter value of '0' causes the kernel 4255 not to attempt recovery of lost locks. 4256 4257 nfs.send_implementation_id= 4258 [NFSv4.1] Send client implementation identification 4259 information in exchange_id requests. 4260 If zero, no implementation identification information 4261 will be sent. 4262 The default is to send the implementation identification 4263 information. 4264 4265 nfs4.layoutstats_timer= 4266 [NFSv4.2] Change the rate at which the kernel sends 4267 layoutstats to the pNFS metadata server. 4268 4269 Setting this to value to 0 causes the kernel to use 4270 whatever value is the default set by the layout 4271 driver. A non-zero value sets the minimum interval 4272 in seconds between layoutstats transmissions. 4273 4274 nfsd.inter_copy_offload_enable= 4275 [NFSv4.2] When set to 1, the server will support 4276 server-to-server copies for which this server is 4277 the destination of the copy. 4278 4279 nfsd.nfs4_disable_idmapping= 4280 [NFSv4] When set to the default of '1', the NFSv4 4281 server will return only numeric uids and gids to 4282 clients using auth_sys, and will accept numeric uids 4283 and gids from such clients. This is intended to ease 4284 migration from NFSv2/v3. 4285 4286 nfsd.nfsd4_ssc_umount_timeout= 4287 [NFSv4.2] When used as the destination of a 4288 server-to-server copy, knfsd temporarily mounts 4289 the source server. It caches the mount in case 4290 it will be needed again, and discards it if not 4291 used for the number of milliseconds specified by 4292 this parameter. 4293 4294 nfsaddrs= [NFS] Deprecated. Use ip= instead. 4295 See Documentation/admin-guide/nfs/nfsroot.rst. 4296 4297 nfsroot= [NFS] nfs root filesystem for disk-less boxes. 4298 See Documentation/admin-guide/nfs/nfsroot.rst. 4299 4300 nfsrootdebug [NFS] enable nfsroot debugging messages. 4301 See Documentation/admin-guide/nfs/nfsroot.rst. 4302 4303 nmi_backtrace.backtrace_idle [KNL] 4304 Dump stacks even of idle CPUs in response to an 4305 NMI stack-backtrace request. 4306 4307 nmi_debug= [KNL,SH] Specify one or more actions to take 4308 when a NMI is triggered. 4309 Format: [state][,regs][,debounce][,die] 4310 4311 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels 4312 Format: [panic,][nopanic,][rNNN,][num] 4313 Valid num: 0 or 1 4314 0 - turn hardlockup detector in nmi_watchdog off 4315 1 - turn hardlockup detector in nmi_watchdog on 4316 rNNN - configure the watchdog with raw perf event 0xNNN 4317 4318 When panic is specified, panic when an NMI watchdog 4319 timeout occurs (or 'nopanic' to not panic on an NMI 4320 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set) 4321 To disable both hard and soft lockup detectors, 4322 please see 'nowatchdog'. 4323 This is useful when you use a panic=... timeout and 4324 need the box quickly up again. 4325 4326 These settings can be accessed at runtime via 4327 the nmi_watchdog and hardlockup_panic sysctls. 4328 4329 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths 4330 emulation library even if a 387 maths coprocessor 4331 is present. 4332 4333 no4lvl [RISCV,EARLY] Disable 4-level and 5-level paging modes. 4334 Forces kernel to use 3-level paging instead. 4335 4336 no5lvl [X86-64,RISCV,EARLY] Disable 5-level paging mode. Forces 4337 kernel to use 4-level paging instead. 4338 4339 noalign [KNL,ARM] 4340 4341 noapic [SMP,APIC,EARLY] Tells the kernel to not make use of any 4342 IOAPICs that may be present in the system. 4343 4344 noapictimer [APIC,X86] Don't set up the APIC timer 4345 4346 noautogroup Disable scheduler automatic task group creation. 4347 4348 nocache [ARM,EARLY] 4349 4350 no_console_suspend 4351 [HW] Never suspend the console 4352 Disable suspending of consoles during suspend and 4353 hibernate operations. Once disabled, debugging 4354 messages can reach various consoles while the rest 4355 of the system is being put to sleep (ie, while 4356 debugging driver suspend/resume hooks). This may 4357 not work reliably with all consoles, but is known 4358 to work with serial and VGA consoles. 4359 To facilitate more flexible debugging, we also add 4360 console_suspend, a printk module parameter to control 4361 it. Users could use console_suspend (usually 4362 /sys/module/printk/parameters/console_suspend) to 4363 turn on/off it dynamically. 4364 4365 no_debug_objects 4366 [KNL,EARLY] Disable object debugging 4367 4368 nodsp [SH] Disable hardware DSP at boot time. 4369 4370 noefi [EFI,EARLY] Disable EFI runtime services support. 4371 4372 no_entry_flush [PPC,EARLY] Don't flush the L1-D cache when entering the kernel. 4373 4374 noexec32 [X86-64] 4375 This affects only 32-bit executables. 4376 noexec32=on: enable non-executable mappings (default) 4377 read doesn't imply executable mappings 4378 noexec32=off: disable non-executable mappings 4379 read implies executable mappings 4380 4381 no_file_caps Tells the kernel not to honor file capabilities. The 4382 only way then for a file to be executed with privilege 4383 is to be setuid root or executed by root. 4384 4385 nofpu [MIPS,SH] Disable hardware FPU at boot time. 4386 4387 nofsgsbase [X86] Disables FSGSBASE instructions. 4388 4389 nofxsr [BUGS=X86-32] Disables x86 floating point extended 4390 register save and restore. The kernel will only save 4391 legacy floating-point registers on task switch. 4392 4393 nogbpages [X86] Do not use GB pages for kernel direct mappings. 4394 4395 no_hash_pointers 4396 [KNL,EARLY] 4397 Alias for "hash_pointers=never". 4398 4399 nohibernate [HIBERNATION] Disable hibernation and resume. 4400 4401 nohlt [ARM,ARM64,MICROBLAZE,MIPS,PPC,RISCV,SH] Forces the kernel to 4402 busy wait in do_idle() and not use the arch_cpu_idle() 4403 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP 4404 to be effective. This is useful on platforms where the 4405 sleep(SH) or wfi(ARM,ARM64) instructions do not work 4406 correctly or when doing power measurements to evaluate 4407 the impact of the sleep instructions. This is also 4408 useful when using JTAG debugger. 4409 4410 nohpet [X86] Don't use the HPET timer. 4411 4412 nohugeiomap [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge I/O mappings. 4413 4414 nohugevmalloc [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge vmalloc mappings. 4415 4416 nohz= [KNL] Boottime enable/disable dynamic ticks 4417 Valid arguments: on, off 4418 Default: on 4419 4420 nohz_full= [KNL,BOOT,SMP,ISOL] 4421 The argument is a cpu list, as described above. 4422 In kernels built with CONFIG_NO_HZ_FULL=y, set 4423 the specified list of CPUs whose tick will be stopped 4424 whenever possible. The boot CPU will be forced outside 4425 the range to maintain the timekeeping. Any CPUs 4426 in this list will have their RCU callbacks offloaded, 4427 just as if they had also been called out in the 4428 rcu_nocbs= boot parameter. 4429 4430 Note that this argument takes precedence over 4431 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option. 4432 4433 noinitrd [RAM] Tells the kernel not to load any configured 4434 initial RAM disk. 4435 4436 nointremap [X86-64,Intel-IOMMU,EARLY] Do not enable interrupt 4437 remapping. 4438 [Deprecated - use intremap=off] 4439 4440 noinvpcid [X86,EARLY] Disable the INVPCID cpu feature. 4441 4442 noiotrap [SH] Disables trapped I/O port accesses. 4443 4444 noirqdebug [X86-32] Disables the code which attempts to detect and 4445 disable unhandled interrupt sources. 4446 4447 noisapnp [ISAPNP] Disables ISA PnP code. 4448 4449 nokaslr [KNL,EARLY] 4450 When CONFIG_RANDOMIZE_BASE is set, this disables 4451 kernel and module base offset ASLR (Address Space 4452 Layout Randomization). 4453 4454 no-kvmapf [X86,KVM,EARLY] Disable paravirtualized asynchronous page 4455 fault handling. 4456 4457 no-kvmclock [X86,KVM,EARLY] Disable paravirtualized KVM clock driver 4458 4459 nolapic [X86-32,APIC,EARLY] Do not enable or use the local APIC. 4460 4461 nolapic_timer [X86-32,APIC,EARLY] Do not use the local APIC timer. 4462 4463 nomce [X86-32] Disable Machine Check Exception 4464 4465 nomfgpt [X86-32] Disable Multi-Function General Purpose 4466 Timer usage (for AMD Geode machines). 4467 4468 nomodeset Disable kernel modesetting. Most systems' firmware 4469 sets up a display mode and provides framebuffer memory 4470 for output. With nomodeset, DRM and fbdev drivers will 4471 not load if they could possibly displace the pre- 4472 initialized output. Only the system framebuffer will 4473 be available for use. The respective drivers will not 4474 perform display-mode changes or accelerated rendering. 4475 4476 Useful as error fallback, or for testing and debugging. 4477 4478 nomodule Disable module load 4479 4480 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to 4481 shutdown the other cpus. Instead use the REBOOT_VECTOR 4482 irq. 4483 4484 nopat [X86,EARLY] Disable PAT (page attribute table extension of 4485 pagetables) support. 4486 4487 nopcid [X86-64,EARLY] Disable the PCID cpu feature. 4488 4489 nopku [X86] Disable Memory Protection Keys CPU feature found 4490 in some Intel CPUs. 4491 4492 nopti [X86-64,EARLY] 4493 Equivalent to pti=off 4494 4495 nopv= [X86,XEN,KVM,HYPER_V,VMWARE,EARLY] 4496 Disables the PV optimizations forcing the guest to run 4497 as generic guest with no PV drivers. Currently support 4498 XEN HVM, KVM, HYPER_V and VMWARE guest. 4499 4500 nopvspin [X86,XEN,KVM,EARLY] 4501 Disables the qspinlock slow path using PV optimizations 4502 which allow the hypervisor to 'idle' the guest on lock 4503 contention. 4504 4505 norandmaps Don't use address space randomization. Equivalent to 4506 echo 0 > /proc/sys/kernel/randomize_va_space 4507 4508 noreplace-smp [X86-32,SMP] Don't replace SMP instructions 4509 with UP alternatives 4510 4511 noresume [SWSUSP] Disables resume and restores original swap 4512 space. 4513 4514 no-scroll [VGA] Disables scrollback. 4515 This is required for the Braillex ib80-piezo Braille 4516 reader made by F.H. Papenmeier (Germany). 4517 4518 nosgx [X86-64,SGX,EARLY] Disables Intel SGX kernel support. 4519 4520 nosmap [PPC,EARLY] 4521 Disable SMAP (Supervisor Mode Access Prevention) 4522 even if it is supported by processor. 4523 4524 nosmep [PPC64s,EARLY] 4525 Disable SMEP (Supervisor Mode Execution Prevention) 4526 even if it is supported by processor. 4527 4528 nosmp [SMP,EARLY] Tells an SMP kernel to act as a UP kernel, 4529 and disable the IO APIC. legacy for "maxcpus=0". 4530 4531 nosmt [KNL,MIPS,PPC,EARLY] Disable symmetric multithreading (SMT). 4532 Equivalent to smt=1. 4533 4534 [KNL,X86,PPC,S390] Disable symmetric multithreading (SMT). 4535 nosmt=force: Force disable SMT, cannot be undone 4536 via the sysfs control file. 4537 4538 nosoftlockup [KNL] Disable the soft-lockup detector. 4539 4540 nospec_store_bypass_disable 4541 [HW,EARLY] Disable all mitigations for the Speculative 4542 Store Bypass vulnerability 4543 4544 nospectre_bhb [ARM64,EARLY] Disable all mitigations for Spectre-BHB (branch 4545 history injection) vulnerability. System may allow data leaks 4546 with this option. 4547 4548 nospectre_v1 [X86,PPC,EARLY] Disable mitigations for Spectre Variant 1 4549 (bounds check bypass). With this option data leaks are 4550 possible in the system. 4551 4552 nospectre_v2 [X86,PPC_E500,ARM64,EARLY] Disable all mitigations 4553 for the Spectre variant 2 (indirect branch 4554 prediction) vulnerability. System may allow data 4555 leaks with this option. 4556 4557 no-steal-acc [X86,PV_OPS,ARM64,PPC/PSERIES,RISCV,LOONGARCH,EARLY] 4558 Disable paravirtualized steal time accounting. steal time 4559 is computed, but won't influence scheduler behaviour 4560 4561 nosync [HW,M68K] Disables sync negotiation for all devices. 4562 4563 no_timer_check [X86,APIC] Disables the code which tests for broken 4564 timer IRQ sources, i.e., the IO-APIC timer. This can 4565 work around problems with incorrect timer 4566 initialization on some boards. 4567 4568 no_uaccess_flush 4569 [PPC,EARLY] Don't flush the L1-D cache after accessing user data. 4570 4571 novmcoredd [KNL,KDUMP] 4572 Disable device dump. Device dump allows drivers to 4573 append dump data to vmcore so you can collect driver 4574 specified debug info. Drivers can append the data 4575 without any limit and this data is stored in memory, 4576 so this may cause significant memory stress. Disabling 4577 device dump can help save memory but the driver debug 4578 data will be no longer available. This parameter 4579 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP 4580 is set. 4581 4582 no-vmw-sched-clock 4583 [X86,PV_OPS,EARLY] Disable paravirtualized VMware 4584 scheduler clock and use the default one. 4585 4586 nowatchdog [KNL] Disable both lockup detectors, i.e. 4587 soft-lockup and NMI watchdog (hard-lockup). 4588 4589 nowb [ARM,EARLY] 4590 4591 nox2apic [X86-64,APIC,EARLY] Do not enable x2APIC mode. 4592 4593 NOTE: this parameter will be ignored on systems with the 4594 LEGACY_XAPIC_DISABLED bit set in the 4595 IA32_XAPIC_DISABLE_STATUS MSR. 4596 4597 noxsave [BUGS=X86] Disables x86 extended register state save 4598 and restore using xsave. The kernel will fallback to 4599 enabling legacy floating-point and sse state. 4600 4601 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended 4602 register states. The kernel will fall back to use 4603 xsave to save the states. By using this parameter, 4604 performance of saving the states is degraded because 4605 xsave doesn't support modified optimization while 4606 xsaveopt supports it on xsaveopt enabled systems. 4607 4608 noxsaves [X86] Disables xsaves and xrstors used in saving and 4609 restoring x86 extended register state in compacted 4610 form of xsave area. The kernel will fall back to use 4611 xsaveopt and xrstor to save and restore the states 4612 in standard form of xsave area. By using this 4613 parameter, xsave area per process might occupy more 4614 memory on xsaves enabled systems. 4615 4616 nr_cpus= [SMP,EARLY] Maximum number of processors that an SMP kernel 4617 could support. nr_cpus=n : n >= 1 limits the kernel to 4618 support 'n' processors. It could be larger than the 4619 number of already plugged CPU during bootup, later in 4620 runtime you can physically add extra cpu until it reaches 4621 n. So during boot up some boot time memory for per-cpu 4622 variables need be pre-allocated for later physical cpu 4623 hot plugging. 4624 4625 nr_uarts= [SERIAL] maximum number of UARTs to be registered. 4626 4627 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86, EARLY] 4628 Disable NUMA, Only set up a single NUMA node 4629 spanning all memory. 4630 4631 numa=fake=<size>[MG] 4632 [KNL, ARM64, RISCV, X86, EARLY] 4633 If given as a memory unit, fills all system RAM with 4634 nodes of size interleaved over physical nodes. 4635 4636 numa=fake=<N> 4637 [KNL, ARM64, RISCV, X86, EARLY] 4638 If given as an integer, fills all system RAM with N 4639 fake nodes interleaved over physical nodes. 4640 4641 numa=fake=<N>U 4642 [KNL, ARM64, RISCV, X86, EARLY] 4643 If given as an integer followed by 'U', it will 4644 divide each physical node into N emulated nodes. 4645 4646 numa=noacpi [X86] Don't parse the SRAT table for NUMA setup 4647 4648 numa=nohmat [X86] Don't parse the HMAT table for NUMA setup, or 4649 soft-reserved memory partitioning. 4650 4651 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic 4652 NUMA balancing. 4653 Allowed values are enable and disable 4654 4655 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA. 4656 'node', 'default' can be specified 4657 This can be set from sysctl after boot. 4658 See Documentation/admin-guide/sysctl/vm.rst for details. 4659 4660 ohci1394_dma=early [HW,EARLY] enable debugging via the ohci1394 driver. 4661 See Documentation/core-api/debugging-via-ohci1394.rst for more 4662 info. 4663 4664 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands 4665 Rather than timing out after 20 ms if an EC 4666 command is not properly ACKed, override the length 4667 of the timeout. We have interrupts disabled while 4668 waiting for the ACK, so if this is set too high 4669 interrupts *may* be lost! 4670 4671 omap_mux= [OMAP] Override bootloader pin multiplexing. 4672 Format: <mux_mode0.mode_name=value>... 4673 For example, to override I2C bus2: 4674 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100 4675 4676 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration 4677 4678 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock] 4679 4680 boundary - index of last SLC block on Flex-OneNAND. 4681 The remaining blocks are configured as MLC blocks. 4682 lock - Configure if Flex-OneNAND boundary should be locked. 4683 Once locked, the boundary cannot be changed. 4684 1 indicates lock status, 0 indicates unlock status. 4685 4686 oops=panic [KNL,EARLY] 4687 Always panic on oopses. Default is to just kill the 4688 process, but there is a small probability of 4689 deadlocking the machine. 4690 This will also cause panics on machine check exceptions. 4691 Useful together with panic=30 to trigger a reboot. 4692 4693 page_alloc.shuffle= 4694 [KNL] Boolean flag to control whether the page allocator 4695 should randomize its free lists. This parameter can be 4696 used to enable/disable page randomization. The state of 4697 the flag can be read from sysfs at: 4698 /sys/module/page_alloc/parameters/shuffle. 4699 This parameter is only available if CONFIG_SHUFFLE_PAGE_ALLOCATOR=y. 4700 4701 page_owner= [KNL,EARLY] Boot-time page_owner enabling option. 4702 Storage of the information about who allocated 4703 each page is disabled in default. With this switch, 4704 we can turn it on. 4705 on: enable the feature 4706 4707 page_poison= [KNL,EARLY] Boot-time parameter changing the state of 4708 poisoning on the buddy allocator, available with 4709 CONFIG_PAGE_POISONING=y. 4710 off: turn off poisoning (default) 4711 on: turn on poisoning 4712 4713 page_reporting.page_reporting_order= 4714 [KNL] Minimal page reporting order 4715 Format: <integer> 4716 Adjust the minimal page reporting order. The page 4717 reporting is disabled when it exceeds MAX_PAGE_ORDER. 4718 4719 panic= [KNL] Kernel behaviour on panic: delay <timeout> 4720 timeout > 0: seconds before rebooting 4721 timeout = 0: wait forever 4722 timeout < 0: reboot immediately 4723 Format: <timeout> 4724 4725 panic_on_taint= [KNL,EARLY] 4726 Bitmask for conditionally calling panic() in add_taint() 4727 Format: <hex>[,nousertaint] 4728 Hexadecimal bitmask representing the set of TAINT flags 4729 that will cause the kernel to panic when add_taint() is 4730 called with any of the flags in this set. 4731 The optional switch "nousertaint" can be utilized to 4732 prevent userspace forced crashes by writing to sysctl 4733 /proc/sys/kernel/tainted any flagset matching with the 4734 bitmask set on panic_on_taint. 4735 See Documentation/admin-guide/tainted-kernels.rst for 4736 extra details on the taint flags that users can pick 4737 to compose the bitmask to assign to panic_on_taint. 4738 4739 panic_on_warn=1 panic() instead of WARN(). Useful to cause kdump 4740 on a WARN(). 4741 4742 panic_print= Bitmask for printing system info when panic happens. 4743 User can chose combination of the following bits: 4744 bit 0: print all tasks info 4745 bit 1: print system memory info 4746 bit 2: print timer info 4747 bit 3: print locks info if CONFIG_LOCKDEP is on 4748 bit 4: print ftrace buffer 4749 bit 5: replay all kernel messages on consoles at the end of panic 4750 bit 6: print all CPUs backtrace (if available in the arch) 4751 bit 7: print only tasks in uninterruptible (blocked) state 4752 *Be aware* that this option may print a _lot_ of lines, 4753 so there are risks of losing older messages in the log. 4754 Use this option carefully, maybe worth to setup a 4755 bigger log buffer with "log_buf_len" along with this. 4756 4757 panic_sys_info= A comma separated list of extra information to be dumped 4758 on panic. 4759 Format: val[,val...] 4760 Where @val can be any of the following: 4761 4762 tasks: print all tasks info 4763 mem: print system memory info 4764 timers: print timers info 4765 locks: print locks info if CONFIG_LOCKDEP is on 4766 ftrace: print ftrace buffer 4767 all_bt: print all CPUs backtrace (if available in the arch) 4768 blocked_tasks: print only tasks in uninterruptible (blocked) state 4769 4770 This is a human readable alternative to the 'panic_print' option. 4771 4772 panic_console_replay 4773 When panic happens, replay all kernel messages on 4774 consoles at the end of panic. 4775 4776 parkbd.port= [HW] Parallel port number the keyboard adapter is 4777 connected to, default is 0. 4778 Format: <parport#> 4779 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation, 4780 0 for XT, 1 for AT (default is AT). 4781 Format: <mode> 4782 4783 parport= [HW,PPT] Specify parallel ports. 0 disables. 4784 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] } 4785 Use 'auto' to force the driver to use any 4786 IRQ/DMA settings detected (the default is to 4787 ignore detected IRQ/DMA settings because of 4788 possible conflicts). You can specify the base 4789 address, IRQ, and DMA settings; IRQ and DMA 4790 should be numbers, or 'auto' (for using detected 4791 settings on that particular port), or 'nofifo' 4792 (to avoid using a FIFO even if it is detected). 4793 Parallel ports are assigned in the order they 4794 are specified on the command line, starting 4795 with parport0. 4796 4797 parport_init_mode= [HW,PPT] 4798 Configure VIA parallel port to operate in 4799 a specific mode. This is necessary on Pegasos 4800 computer where firmware has no options for setting 4801 up parallel port mode and sets it to spp. 4802 Currently this function knows 686a and 8231 chips. 4803 Format: [spp|ps2|epp|ecp|ecpepp] 4804 4805 pata_legacy.all= [HW,LIBATA] 4806 Format: <int> 4807 Set to non-zero to probe primary and secondary ISA 4808 port ranges on PCI systems where no PCI PATA device 4809 has been found at either range. Disabled by default. 4810 4811 pata_legacy.autospeed= [HW,LIBATA] 4812 Format: <int> 4813 Set to non-zero if a chip is present that snoops speed 4814 changes. Disabled by default. 4815 4816 pata_legacy.ht6560a= [HW,LIBATA] 4817 Format: <int> 4818 Set to 1, 2, or 3 for HT 6560A on the primary channel, 4819 the secondary channel, or both channels respectively. 4820 Disabled by default. 4821 4822 pata_legacy.ht6560b= [HW,LIBATA] 4823 Format: <int> 4824 Set to 1, 2, or 3 for HT 6560B on the primary channel, 4825 the secondary channel, or both channels respectively. 4826 Disabled by default. 4827 4828 pata_legacy.iordy_mask= [HW,LIBATA] 4829 Format: <int> 4830 IORDY enable mask. Set individual bits to allow IORDY 4831 for the respective channel. Bit 0 is for the first 4832 legacy channel handled by this driver, bit 1 is for 4833 the second channel, and so on. The sequence will often 4834 correspond to the primary legacy channel, the secondary 4835 legacy channel, and so on, but the handling of a PCI 4836 bus and the use of other driver options may interfere 4837 with the sequence. By default IORDY is allowed across 4838 all channels. 4839 4840 pata_legacy.opti82c46x= [HW,LIBATA] 4841 Format: <int> 4842 Set to 1, 2, or 3 for Opti 82c611A on the primary 4843 channel, the secondary channel, or both channels 4844 respectively. Disabled by default. 4845 4846 pata_legacy.opti82c611a= [HW,LIBATA] 4847 Format: <int> 4848 Set to 1, 2, or 3 for Opti 82c465MV on the primary 4849 channel, the secondary channel, or both channels 4850 respectively. Disabled by default. 4851 4852 pata_legacy.pio_mask= [HW,LIBATA] 4853 Format: <int> 4854 PIO mode mask for autospeed devices. Set individual 4855 bits to allow the use of the respective PIO modes. 4856 Bit 0 is for mode 0, bit 1 is for mode 1, and so on. 4857 All modes allowed by default. 4858 4859 pata_legacy.probe_all= [HW,LIBATA] 4860 Format: <int> 4861 Set to non-zero to probe tertiary and further ISA 4862 port ranges on PCI systems. Disabled by default. 4863 4864 pata_legacy.probe_mask= [HW,LIBATA] 4865 Format: <int> 4866 Probe mask for legacy ISA PATA ports. Depending on 4867 platform configuration and the use of other driver 4868 options up to 6 legacy ports are supported: 0x1f0, 4869 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing 4870 of individual ports can be disabled by setting the 4871 corresponding bits in the mask to 1. Bit 0 is for 4872 the first port in the list above (0x1f0), and so on. 4873 By default all supported ports are probed. 4874 4875 pata_legacy.qdi= [HW,LIBATA] 4876 Format: <int> 4877 Set to non-zero to probe QDI controllers. By default 4878 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise. 4879 4880 pata_legacy.winbond= [HW,LIBATA] 4881 Format: <int> 4882 Set to non-zero to probe Winbond controllers. Use 4883 the standard I/O port (0x130) if 1, otherwise the 4884 value given is the I/O port to use (typically 0x1b0). 4885 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE, 4886 0 otherwise. 4887 4888 pata_platform.pio_mask= [HW,LIBATA] 4889 Format: <int> 4890 Supported PIO mode mask. Set individual bits to allow 4891 the use of the respective PIO modes. Bit 0 is for 4892 mode 0, bit 1 is for mode 1, and so on. Mode 0 only 4893 allowed by default. 4894 4895 pause_on_oops=<int> 4896 Halt all CPUs after the first oops has been printed for 4897 the specified number of seconds. This is to be used if 4898 your oopses keep scrolling off the screen. 4899 4900 pcbit= [HW,ISDN] 4901 4902 pci=option[,option...] [PCI,EARLY] various PCI subsystem options. 4903 4904 Some options herein operate on a specific device 4905 or a set of devices (<pci_dev>). These are 4906 specified in one of the following formats: 4907 4908 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]* 4909 pci:<vendor>:<device>[:<subvendor>:<subdevice>] 4910 4911 Note: the first format specifies a PCI 4912 bus/device/function address which may change 4913 if new hardware is inserted, if motherboard 4914 firmware changes, or due to changes caused 4915 by other kernel parameters. If the 4916 domain is left unspecified, it is 4917 taken to be zero. Optionally, a path 4918 to a device through multiple device/function 4919 addresses can be specified after the base 4920 address (this is more robust against 4921 renumbering issues). The second format 4922 selects devices using IDs from the 4923 configuration space which may match multiple 4924 devices in the system. 4925 4926 earlydump dump PCI config space before the kernel 4927 changes anything 4928 off [X86] don't probe for the PCI bus 4929 bios [X86-32] force use of PCI BIOS, don't access 4930 the hardware directly. Use this if your machine 4931 has a non-standard PCI host bridge. 4932 nobios [X86-32] disallow use of PCI BIOS, only direct 4933 hardware access methods are allowed. Use this 4934 if you experience crashes upon bootup and you 4935 suspect they are caused by the BIOS. 4936 conf1 [X86] Force use of PCI Configuration Access 4937 Mechanism 1 (config address in IO port 0xCF8, 4938 data in IO port 0xCFC, both 32-bit). 4939 conf2 [X86] Force use of PCI Configuration Access 4940 Mechanism 2 (IO port 0xCF8 is an 8-bit port for 4941 the function, IO port 0xCFA, also 8-bit, sets 4942 bus number. The config space is then accessed 4943 through ports 0xC000-0xCFFF). 4944 See http://wiki.osdev.org/PCI for more info 4945 on the configuration access mechanisms. 4946 noaer [PCIE] If the PCIEAER kernel config parameter is 4947 enabled, this kernel boot option can be used to 4948 disable the use of PCIE advanced error reporting. 4949 nodomains [PCI] Disable support for multiple PCI 4950 root domains (aka PCI segments, in ACPI-speak). 4951 nommconf [X86] Disable use of MMCONFIG for PCI 4952 Configuration 4953 check_enable_amd_mmconf [X86] check for and enable 4954 properly configured MMIO access to PCI 4955 config space on AMD family 10h CPU 4956 nomsi [MSI] If the PCI_MSI kernel config parameter is 4957 enabled, this kernel boot option can be used to 4958 disable the use of MSI interrupts system-wide. 4959 noioapicquirk [APIC] Disable all boot interrupt quirks. 4960 Safety option to keep boot IRQs enabled. This 4961 should never be necessary. 4962 ioapicreroute [APIC] Enable rerouting of boot IRQs to the 4963 primary IO-APIC for bridges that cannot disable 4964 boot IRQs. This fixes a source of spurious IRQs 4965 when the system masks IRQs. 4966 noioapicreroute [APIC] Disable workaround that uses the 4967 boot IRQ equivalent of an IRQ that connects to 4968 a chipset where boot IRQs cannot be disabled. 4969 The opposite of ioapicreroute. 4970 biosirq [X86-32] Use PCI BIOS calls to get the interrupt 4971 routing table. These calls are known to be buggy 4972 on several machines and they hang the machine 4973 when used, but on other computers it's the only 4974 way to get the interrupt routing table. Try 4975 this option if the kernel is unable to allocate 4976 IRQs or discover secondary PCI buses on your 4977 motherboard. 4978 rom [X86] Assign address space to expansion ROMs. 4979 Use with caution as certain devices share 4980 address decoders between ROMs and other 4981 resources. 4982 norom [X86] Do not assign address space to 4983 expansion ROMs that do not already have 4984 BIOS assigned address ranges. 4985 nobar [X86] Do not assign address space to the 4986 BARs that weren't assigned by the BIOS. 4987 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be 4988 assigned automatically to PCI devices. You can 4989 make the kernel exclude IRQs of your ISA cards 4990 this way. 4991 pirqaddr=0xAAAAA [X86] Specify the physical address 4992 of the PIRQ table (normally generated 4993 by the BIOS) if it is outside the 4994 F0000h-100000h range. 4995 lastbus=N [X86] Scan all buses thru bus #N. Can be 4996 useful if the kernel is unable to find your 4997 secondary buses and you want to tell it 4998 explicitly which ones they are. 4999 assign-busses [X86] Always assign all PCI bus 5000 numbers ourselves, overriding 5001 whatever the firmware may have done. 5002 usepirqmask [X86] Honor the possible IRQ mask stored 5003 in the BIOS $PIR table. This is needed on 5004 some systems with broken BIOSes, notably 5005 some HP Pavilion N5400 and Omnibook XE3 5006 notebooks. This will have no effect if ACPI 5007 IRQ routing is enabled. 5008 noacpi [X86] Do not use ACPI for IRQ routing 5009 or for PCI scanning. 5010 use_crs [X86] Use PCI host bridge window information 5011 from ACPI. On BIOSes from 2008 or later, this 5012 is enabled by default. If you need to use this, 5013 please report a bug. 5014 nocrs [X86] Ignore PCI host bridge windows from ACPI. 5015 If you need to use this, please report a bug. 5016 use_e820 [X86] Use E820 reservations to exclude parts of 5017 PCI host bridge windows. This is a workaround 5018 for BIOS defects in host bridge _CRS methods. 5019 If you need to use this, please report a bug to 5020 <linux-pci@vger.kernel.org>. 5021 no_e820 [X86] Ignore E820 reservations for PCI host 5022 bridge windows. This is the default on modern 5023 hardware. If you need to use this, please report 5024 a bug to <linux-pci@vger.kernel.org>. 5025 routeirq Do IRQ routing for all PCI devices. 5026 This is normally done in pci_enable_device(), 5027 so this option is a temporary workaround 5028 for broken drivers that don't call it. 5029 skip_isa_align [X86] do not align io start addr, so can 5030 handle more pci cards 5031 noearly [X86] Don't do any early type 1 scanning. 5032 This might help on some broken boards which 5033 machine check when some devices' config space 5034 is read. But various workarounds are disabled 5035 and some IOMMU drivers will not work. 5036 bfsort Sort PCI devices into breadth-first order. 5037 This sorting is done to get a device 5038 order compatible with older (<= 2.4) kernels. 5039 nobfsort Don't sort PCI devices into breadth-first order. 5040 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size) 5041 tuning and use the BIOS-configured MPS defaults. 5042 pcie_bus_safe Set every device's MPS to the largest value 5043 supported by all devices below the root complex. 5044 pcie_bus_perf Set device MPS to the largest allowable MPS 5045 based on its parent bus. Also set MRRS (Max 5046 Read Request Size) to the largest supported 5047 value (no larger than the MPS that the device 5048 or bus can support) for best performance. 5049 pcie_bus_peer2peer Set every device's MPS to 128B, which 5050 every device is guaranteed to support. This 5051 configuration allows peer-to-peer DMA between 5052 any pair of devices, possibly at the cost of 5053 reduced performance. This also guarantees 5054 that hot-added devices will work. 5055 cbiosize=nn[KMG] The fixed amount of bus space which is 5056 reserved for the CardBus bridge's IO window. 5057 The default value is 256 bytes. 5058 cbmemsize=nn[KMG] The fixed amount of bus space which is 5059 reserved for the CardBus bridge's memory 5060 window. The default value is 64 megabytes. 5061 resource_alignment= 5062 Format: 5063 [<order of align>@]<pci_dev>[; ...] 5064 Specifies alignment and device to reassign 5065 aligned memory resources. How to 5066 specify the device is described above. 5067 If <order of align> is not specified, 5068 PAGE_SIZE is used as alignment. 5069 A PCI-PCI bridge can be specified if resource 5070 windows need to be expanded. 5071 To specify the alignment for several 5072 instances of a device, the PCI vendor, 5073 device, subvendor, and subdevice may be 5074 specified, e.g., 12@pci:8086:9c22:103c:198f 5075 for 4096-byte alignment. 5076 ecrc= Enable/disable PCIe ECRC (transaction layer 5077 end-to-end CRC checking). Only effective if 5078 OS has native AER control (either granted by 5079 ACPI _OSC or forced via "pcie_ports=native") 5080 bios: Use BIOS/firmware settings. This is the 5081 the default. 5082 off: Turn ECRC off 5083 on: Turn ECRC on. 5084 hpiosize=nn[KMG] The fixed amount of bus space which is 5085 reserved for hotplug bridge's IO window. 5086 Default size is 256 bytes. 5087 hpmmiosize=nn[KMG] The fixed amount of bus space which is 5088 reserved for hotplug bridge's MMIO window. 5089 Default size is 2 megabytes. 5090 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is 5091 reserved for hotplug bridge's MMIO_PREF window. 5092 Default size is 2 megabytes. 5093 hpmemsize=nn[KMG] The fixed amount of bus space which is 5094 reserved for hotplug bridge's MMIO and 5095 MMIO_PREF window. 5096 Default size is 2 megabytes. 5097 hpbussize=nn The minimum amount of additional bus numbers 5098 reserved for buses below a hotplug bridge. 5099 Default is 1. 5100 realloc= Enable/disable reallocating PCI bridge resources 5101 if allocations done by BIOS are too small to 5102 accommodate resources required by all child 5103 devices. 5104 off: Turn realloc off 5105 on: Turn realloc on 5106 realloc same as realloc=on 5107 noari do not use PCIe ARI. 5108 noats [PCIE, Intel-IOMMU, AMD-IOMMU] 5109 do not use PCIe ATS (and IOMMU device IOTLB). 5110 pcie_scan_all Scan all possible PCIe devices. Otherwise we 5111 only look for one device below a PCIe downstream 5112 port. 5113 big_root_window Try to add a big 64bit memory window to the PCIe 5114 root complex on AMD CPUs. Some GFX hardware 5115 can resize a BAR to allow access to all VRAM. 5116 Adding the window is slightly risky (it may 5117 conflict with unreported devices), so this 5118 taints the kernel. 5119 disable_acs_redir=<pci_dev>[; ...] 5120 Specify one or more PCI devices (in the format 5121 specified above) separated by semicolons. 5122 Each device specified will have the PCI ACS 5123 redirect capabilities forced off which will 5124 allow P2P traffic between devices through 5125 bridges without forcing it upstream. Note: 5126 this removes isolation between devices and 5127 may put more devices in an IOMMU group. 5128 config_acs= 5129 Format: 5130 <ACS flags>@<pci_dev>[; ...] 5131 Specify one or more PCI devices (in the format 5132 specified above) optionally prepended with flags 5133 and separated by semicolons. The respective 5134 capabilities will be enabled, disabled or 5135 unchanged based on what is specified in 5136 flags. 5137 5138 ACS Flags is defined as follows: 5139 bit-0 : ACS Source Validation 5140 bit-1 : ACS Translation Blocking 5141 bit-2 : ACS P2P Request Redirect 5142 bit-3 : ACS P2P Completion Redirect 5143 bit-4 : ACS Upstream Forwarding 5144 bit-5 : ACS P2P Egress Control 5145 bit-6 : ACS Direct Translated P2P 5146 Each bit can be marked as: 5147 '0' – force disabled 5148 '1' – force enabled 5149 'x' – unchanged 5150 For example, 5151 pci=config_acs=10x@pci:0:0 5152 would configure all devices that support 5153 ACS to enable P2P Request Redirect, disable 5154 Translation Blocking, and leave Source 5155 Validation unchanged from whatever power-up 5156 or firmware set it to. 5157 5158 Note: this may remove isolation between devices 5159 and may put more devices in an IOMMU group. 5160 force_floating [S390] Force usage of floating interrupts. 5161 nomio [S390] Do not use MIO instructions. 5162 norid [S390] ignore the RID field and force use of 5163 one PCI domain per PCI function 5164 notph [PCIE] If the PCIE_TPH kernel config parameter 5165 is enabled, this kernel boot option can be used 5166 to disable PCIe TLP Processing Hints support 5167 system-wide. 5168 5169 pcie_aspm= [PCIE] Forcibly enable or ignore PCIe Active State Power 5170 Management. 5171 off Don't touch ASPM configuration at all. Leave any 5172 configuration done by firmware unchanged. 5173 force Enable ASPM even on devices that claim not to support it. 5174 WARNING: Forcing ASPM on may cause system lockups. 5175 5176 pcie_ports= [PCIE] PCIe port services handling: 5177 native Use native PCIe services (PME, AER, DPC, PCIe hotplug) 5178 even if the platform doesn't give the OS permission to 5179 use them. This may cause conflicts if the platform 5180 also tries to use these services. 5181 dpc-native Use native PCIe service for DPC only. May 5182 cause conflicts if firmware uses AER or DPC. 5183 compat Disable native PCIe services (PME, AER, DPC, PCIe 5184 hotplug). 5185 5186 pcie_port_pm= [PCIE] PCIe port power management handling: 5187 off Disable power management of all PCIe ports 5188 force Forcibly enable power management of all PCIe ports 5189 5190 pcie_pme= [PCIE,PM] Native PCIe PME signaling options: 5191 nomsi Do not use MSI for native PCIe PME signaling (this makes 5192 all PCIe root ports use INTx for all services). 5193 5194 pcmv= [HW,PCMCIA] BadgePAD 4 5195 5196 pd_ignore_unused 5197 [PM] 5198 Keep all power-domains already enabled by bootloader on, 5199 even if no driver has claimed them. This is useful 5200 for debug and development, but should not be 5201 needed on a platform with proper driver support. 5202 5203 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at 5204 boot time. 5205 Format: { 0 | 1 } 5206 See arch/parisc/kernel/pdc_chassis.c 5207 5208 percpu_alloc= [MM,EARLY] 5209 Select which percpu first chunk allocator to use. 5210 Currently supported values are "embed" and "page". 5211 Archs may support subset or none of the selections. 5212 See comments in mm/percpu.c for details on each 5213 allocator. This parameter is primarily for debugging 5214 and performance comparison. 5215 5216 pirq= [SMP,APIC] Manual mp-table setup 5217 See Documentation/arch/x86/i386/IO-APIC.rst. 5218 5219 plip= [PPT,NET] Parallel port network link 5220 Format: { parport<nr> | timid | 0 } 5221 See also Documentation/admin-guide/parport.rst. 5222 5223 pmtmr= [X86] Manual setup of pmtmr I/O Port. 5224 Override pmtimer IOPort with a hex value. 5225 e.g. pmtmr=0x508 5226 5227 pmu_override= [PPC] Override the PMU. 5228 This option takes over the PMU facility, so it is no 5229 longer usable by perf. Setting this option starts the 5230 PMU counters by setting MMCR0 to 0 (the FC bit is 5231 cleared). If a number is given, then MMCR1 is set to 5232 that number, otherwise (e.g., 'pmu_override=on'), MMCR1 5233 remains 0. 5234 5235 pm_async= [PM] 5236 Format: off 5237 This parameter sets the initial value of the 5238 /sys/power/pm_async sysfs knob at boot time. 5239 If set to "off", disables asynchronous suspend and 5240 resume of devices during system-wide power transitions. 5241 This can be useful on platforms where device 5242 dependencies are not well-defined, or for debugging 5243 power management issues. Asynchronous operations are 5244 enabled by default. 5245 5246 5247 pm_debug_messages [SUSPEND,KNL] 5248 Enable suspend/resume debug messages during boot up. 5249 5250 pnp.debug=1 [PNP] 5251 Enable PNP debug messages (depends on the 5252 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time 5253 via /sys/module/pnp/parameters/debug. We always show 5254 current resource usage; turning this on also shows 5255 possible settings and some assignment information. 5256 5257 pnpacpi= [ACPI] 5258 { off } 5259 5260 pnpbios= [ISAPNP] 5261 { on | off | curr | res | no-curr | no-res } 5262 5263 pnp_reserve_irq= 5264 [ISAPNP] Exclude IRQs for the autoconfiguration 5265 5266 pnp_reserve_dma= 5267 [ISAPNP] Exclude DMAs for the autoconfiguration 5268 5269 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration 5270 Ranges are in pairs (I/O port base and size). 5271 5272 pnp_reserve_mem= 5273 [ISAPNP] Exclude memory regions for the 5274 autoconfiguration. 5275 Ranges are in pairs (memory base and size). 5276 5277 ports= [IP_VS_FTP] IPVS ftp helper module 5278 Default is 21. 5279 Up to 8 (IP_VS_APP_MAX_PORTS) ports 5280 may be specified. 5281 Format: <port>,<port>.... 5282 5283 possible_cpus= [SMP,S390,X86] 5284 Format: <unsigned int> 5285 Set the number of possible CPUs, overriding the 5286 regular discovery mechanisms (such as ACPI/FW, etc). 5287 5288 powersave=off [PPC] This option disables power saving features. 5289 It specifically disables cpuidle and sets the 5290 platform machine description specific power_save 5291 function to NULL. On Idle the CPU just reduces 5292 execution priority. 5293 5294 ppc_strict_facility_enable 5295 [PPC,ENABLE] This option catches any kernel floating point, 5296 Altivec, VSX and SPE outside of regions specifically 5297 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()). 5298 There is some performance impact when enabling this. 5299 5300 ppc_tm= [PPC,EARLY] 5301 Format: {"off"} 5302 Disable Hardware Transactional Memory 5303 5304 preempt= [KNL] 5305 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC 5306 none - Limited to cond_resched() calls 5307 voluntary - Limited to cond_resched() and might_sleep() calls 5308 full - Any section that isn't explicitly preempt disabled 5309 can be preempted anytime. Tasks will also yield 5310 contended spinlocks (if the critical section isn't 5311 explicitly preempt disabled beyond the lock itself). 5312 lazy - Scheduler controlled. Similar to full but instead 5313 of preempting the task immediately, the task gets 5314 one HZ tick time to yield itself before the 5315 preemption will be forced. One preemption is when the 5316 task returns to user space. 5317 5318 print-fatal-signals= 5319 [KNL] debug: print fatal signals 5320 5321 If enabled, warn about various signal handling 5322 related application anomalies: too many signals, 5323 too many POSIX.1 timers, fatal signals causing a 5324 coredump - etc. 5325 5326 If you hit the warning due to signal overflow, 5327 you might want to try "ulimit -i unlimited". 5328 5329 default: off. 5330 5331 printk.always_kmsg_dump= 5332 Trigger kmsg_dump for cases other than kernel oops or 5333 panics 5334 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5335 default: disabled 5336 5337 printk.console_no_auto_verbose= 5338 Disable console loglevel raise on oops, panic 5339 or lockdep-detected issues (only if lock debug is on). 5340 With an exception to setups with low baudrate on 5341 serial console, keeping this 0 is a good choice 5342 in order to provide more debug information. 5343 Format: <bool> 5344 default: 0 (auto_verbose is enabled) 5345 5346 printk.debug_non_panic_cpus= 5347 Allows storing messages from non-panic CPUs into 5348 the printk log buffer during panic(). They are 5349 flushed to consoles by the panic-CPU on 5350 a best-effort basis. 5351 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5352 Default: disabled 5353 5354 printk.devkmsg={on,off,ratelimit} 5355 Control writing to /dev/kmsg. 5356 on - unlimited logging to /dev/kmsg from userspace 5357 off - logging to /dev/kmsg disabled 5358 ratelimit - ratelimit the logging 5359 Default: ratelimit 5360 5361 printk.time= Show timing data prefixed to each printk message line 5362 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5363 5364 proc_mem.force_override= [KNL] 5365 Format: {always | ptrace | never} 5366 Traditionally /proc/pid/mem allows memory permissions to be 5367 overridden without restrictions. This option may be set to 5368 restrict that. Can be one of: 5369 - 'always': traditional behavior always allows mem overrides. 5370 - 'ptrace': only allow mem overrides for active ptracers. 5371 - 'never': never allow mem overrides. 5372 If not specified, default is the CONFIG_PROC_MEM_* choice. 5373 5374 processor.max_cstate= [HW,ACPI] 5375 Limit processor to maximum C-state 5376 max_cstate=9 overrides any DMI blacklist limit. 5377 5378 processor.nocst [HW,ACPI] 5379 Ignore the _CST method to determine C-states, 5380 instead using the legacy FADT method 5381 5382 profile= [KNL] Enable kernel profiling via /proc/profile 5383 Format: [<profiletype>,]<number> 5384 Param: <profiletype>: "schedule" or "kvm" 5385 [defaults to kernel profiling] 5386 Param: "schedule" - profile schedule points. 5387 Param: "kvm" - profile VM exits. 5388 Param: <number> - step/bucket size as a power of 2 for 5389 statistical time based profiling. 5390 5391 prompt_ramdisk= [RAM] [Deprecated] 5392 5393 prot_virt= [S390] enable hosting protected virtual machines 5394 isolated from the hypervisor (if hardware supports 5395 that). If enabled, the default kernel base address 5396 might be overridden even when Kernel Address Space 5397 Layout Randomization is disabled. 5398 Format: <bool> 5399 5400 psi= [KNL] Enable or disable pressure stall information 5401 tracking. 5402 Format: <bool> 5403 5404 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to 5405 probe for; one of (bare|imps|exps|lifebook|any). 5406 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports 5407 per second. 5408 psmouse.resetafter= [HW,MOUSE] 5409 Try to reset the device after so many bad packets 5410 (0 = never). 5411 psmouse.resolution= 5412 [HW,MOUSE] Set desired mouse resolution, in dpi. 5413 psmouse.smartscroll= 5414 [HW,MOUSE] Controls Logitech smartscroll autorepeat. 5415 0 = disabled, 1 = enabled (default). 5416 5417 pstore.backend= Specify the name of the pstore backend to use 5418 5419 pti= [X86-64] Control Page Table Isolation of user and 5420 kernel address spaces. Disabling this feature 5421 removes hardening, but improves performance of 5422 system calls and interrupts. 5423 5424 on - unconditionally enable 5425 off - unconditionally disable 5426 auto - kernel detects whether your CPU model is 5427 vulnerable to issues that PTI mitigates 5428 5429 Not specifying this option is equivalent to pti=auto. 5430 5431 pty.legacy_count= 5432 [KNL] Number of legacy pty's. Overwrites compiled-in 5433 default number. 5434 5435 quiet [KNL,EARLY] Disable most log messages 5436 5437 r128= [HW,DRM] 5438 5439 radix_hcall_invalidate=on [PPC/PSERIES] 5440 Disable RADIX GTSE feature and use hcall for TLB 5441 invalidate. 5442 5443 raid= [HW,RAID] 5444 See Documentation/admin-guide/md.rst. 5445 5446 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes 5447 See Documentation/admin-guide/blockdev/ramdisk.rst. 5448 5449 ramdisk_start= [RAM] RAM disk image start address 5450 5451 random.trust_cpu=off 5452 [KNL,EARLY] Disable trusting the use of the CPU's 5453 random number generator (if available) to 5454 initialize the kernel's RNG. 5455 5456 random.trust_bootloader=off 5457 [KNL,EARLY] Disable trusting the use of the a seed 5458 passed by the bootloader (if available) to 5459 initialize the kernel's RNG. 5460 5461 randomize_kstack_offset= 5462 [KNL,EARLY] Enable or disable kernel stack offset 5463 randomization, which provides roughly 5 bits of 5464 entropy, frustrating memory corruption attacks 5465 that depend on stack address determinism or 5466 cross-syscall address exposures. This is only 5467 available on architectures that have defined 5468 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET. 5469 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5470 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT. 5471 5472 ras=option[,option,...] [KNL] RAS-specific options 5473 5474 cec_disable [X86] 5475 Disable the Correctable Errors Collector, 5476 see CONFIG_RAS_CEC help text. 5477 5478 rcu_nocbs[=cpu-list] 5479 [KNL] The optional argument is a cpu list, 5480 as described above. 5481 5482 In kernels built with CONFIG_RCU_NOCB_CPU=y, 5483 enable the no-callback CPU mode, which prevents 5484 such CPUs' callbacks from being invoked in 5485 softirq context. Invocation of such CPUs' RCU 5486 callbacks will instead be offloaded to "rcuox/N" 5487 kthreads created for that purpose, where "x" is 5488 "p" for RCU-preempt, "s" for RCU-sched, and "g" 5489 for the kthreads that mediate grace periods; and 5490 "N" is the CPU number. This reduces OS jitter on 5491 the offloaded CPUs, which can be useful for HPC 5492 and real-time workloads. It can also improve 5493 energy efficiency for asymmetric multiprocessors. 5494 5495 If a cpulist is passed as an argument, the specified 5496 list of CPUs is set to no-callback mode from boot. 5497 5498 Otherwise, if the '=' sign and the cpulist 5499 arguments are omitted, no CPU will be set to 5500 no-callback mode from boot but the mode may be 5501 toggled at runtime via cpusets. 5502 5503 Note that this argument takes precedence over 5504 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option. 5505 5506 rcu_nocb_poll [KNL] 5507 Rather than requiring that offloaded CPUs 5508 (specified by rcu_nocbs= above) explicitly 5509 awaken the corresponding "rcuoN" kthreads, 5510 make these kthreads poll for callbacks. 5511 This improves the real-time response for the 5512 offloaded CPUs by relieving them of the need to 5513 wake up the corresponding kthread, but degrades 5514 energy efficiency by requiring that the kthreads 5515 periodically wake up to do the polling. 5516 5517 rcutree.blimit= [KNL] 5518 Set maximum number of finished RCU callbacks to 5519 process in one batch. 5520 5521 rcutree.csd_lock_suppress_rcu_stall= [KNL] 5522 Do only a one-line RCU CPU stall warning when 5523 there is an ongoing too-long CSD-lock wait. 5524 5525 rcutree.do_rcu_barrier= [KNL] 5526 Request a call to rcu_barrier(). This is 5527 throttled so that userspace tests can safely 5528 hammer on the sysfs variable if they so choose. 5529 If triggered before the RCU grace-period machinery 5530 is fully active, this will error out with EAGAIN. 5531 5532 rcutree.dump_tree= [KNL] 5533 Dump the structure of the rcu_node combining tree 5534 out at early boot. This is used for diagnostic 5535 purposes, to verify correct tree setup. 5536 5537 rcutree.gp_cleanup_delay= [KNL] 5538 Set the number of jiffies to delay each step of 5539 RCU grace-period cleanup. 5540 5541 rcutree.gp_init_delay= [KNL] 5542 Set the number of jiffies to delay each step of 5543 RCU grace-period initialization. 5544 5545 rcutree.gp_preinit_delay= [KNL] 5546 Set the number of jiffies to delay each step of 5547 RCU grace-period pre-initialization, that is, 5548 the propagation of recent CPU-hotplug changes up 5549 the rcu_node combining tree. 5550 5551 rcutree.jiffies_till_first_fqs= [KNL] 5552 Set delay from grace-period initialization to 5553 first attempt to force quiescent states. 5554 Units are jiffies, minimum value is zero, 5555 and maximum value is HZ. 5556 5557 rcutree.jiffies_till_next_fqs= [KNL] 5558 Set delay between subsequent attempts to force 5559 quiescent states. Units are jiffies, minimum 5560 value is one, and maximum value is HZ. 5561 5562 rcutree.jiffies_till_sched_qs= [KNL] 5563 Set required age in jiffies for a 5564 given grace period before RCU starts 5565 soliciting quiescent-state help from 5566 rcu_note_context_switch() and cond_resched(). 5567 If not specified, the kernel will calculate 5568 a value based on the most recent settings 5569 of rcutree.jiffies_till_first_fqs 5570 and rcutree.jiffies_till_next_fqs. 5571 This calculated value may be viewed in 5572 rcutree.jiffies_to_sched_qs. Any attempt to set 5573 rcutree.jiffies_to_sched_qs will be cheerfully 5574 overwritten. 5575 5576 rcutree.kthread_prio= [KNL,BOOT] 5577 Set the SCHED_FIFO priority of the RCU per-CPU 5578 kthreads (rcuc/N). This value is also used for 5579 the priority of the RCU boost threads (rcub/N) 5580 and for the RCU grace-period kthreads (rcu_bh, 5581 rcu_preempt, and rcu_sched). If RCU_BOOST is 5582 set, valid values are 1-99 and the default is 1 5583 (the least-favored priority). Otherwise, when 5584 RCU_BOOST is not set, valid values are 0-99 and 5585 the default is zero (non-realtime operation). 5586 When RCU_NOCB_CPU is set, also adjust the 5587 priority of NOCB callback kthreads. 5588 5589 rcutree.nocb_nobypass_lim_per_jiffy= [KNL] 5590 On callback-offloaded (rcu_nocbs) CPUs, 5591 RCU reduces the lock contention that would 5592 otherwise be caused by callback floods through 5593 use of the ->nocb_bypass list. However, in the 5594 common non-flooded case, RCU queues directly to 5595 the main ->cblist in order to avoid the extra 5596 overhead of the ->nocb_bypass list and its lock. 5597 But if there are too many callbacks queued during 5598 a single jiffy, RCU pre-queues the callbacks into 5599 the ->nocb_bypass queue. The definition of "too 5600 many" is supplied by this kernel boot parameter. 5601 5602 rcutree.nohz_full_patience_delay= [KNL] 5603 On callback-offloaded (rcu_nocbs) CPUs, avoid 5604 disturbing RCU unless the grace period has 5605 reached the specified age in milliseconds. 5606 Defaults to zero. Large values will be capped 5607 at five seconds. All values will be rounded down 5608 to the nearest value representable by jiffies. 5609 5610 rcutree.qhimark= [KNL] 5611 Set threshold of queued RCU callbacks beyond which 5612 batch limiting is disabled. 5613 5614 rcutree.qlowmark= [KNL] 5615 Set threshold of queued RCU callbacks below which 5616 batch limiting is re-enabled. 5617 5618 rcutree.qovld= [KNL] 5619 Set threshold of queued RCU callbacks beyond which 5620 RCU's force-quiescent-state scan will aggressively 5621 enlist help from cond_resched() and sched IPIs to 5622 help CPUs more quickly reach quiescent states. 5623 Set to less than zero to make this be set based 5624 on rcutree.qhimark at boot time and to zero to 5625 disable more aggressive help enlistment. 5626 5627 rcutree.rcu_delay_page_cache_fill_msec= [KNL] 5628 Set the page-cache refill delay (in milliseconds) 5629 in response to low-memory conditions. The range 5630 of permitted values is in the range 0:100000. 5631 5632 rcutree.rcu_divisor= [KNL] 5633 Set the shift-right count to use to compute 5634 the callback-invocation batch limit bl from 5635 the number of callbacks queued on this CPU. 5636 The result will be bounded below by the value of 5637 the rcutree.blimit kernel parameter. Every bl 5638 callbacks, the softirq handler will exit in 5639 order to allow the CPU to do other work. 5640 5641 Please note that this callback-invocation batch 5642 limit applies only to non-offloaded callback 5643 invocation. Offloaded callbacks are instead 5644 invoked in the context of an rcuoc kthread, which 5645 scheduler will preempt as it does any other task. 5646 5647 rcutree.rcu_fanout_exact= [KNL] 5648 Disable autobalancing of the rcu_node combining 5649 tree. This is used by rcutorture, and might 5650 possibly be useful for architectures having high 5651 cache-to-cache transfer latencies. 5652 5653 rcutree.rcu_fanout_leaf= [KNL] 5654 Change the number of CPUs assigned to each 5655 leaf rcu_node structure. Useful for very 5656 large systems, which will choose the value 64, 5657 and for NUMA systems with large remote-access 5658 latencies, which will choose a value aligned 5659 with the appropriate hardware boundaries. 5660 5661 rcutree.rcu_min_cached_objs= [KNL] 5662 Minimum number of objects which are cached and 5663 maintained per one CPU. Object size is equal 5664 to PAGE_SIZE. The cache allows to reduce the 5665 pressure to page allocator, also it makes the 5666 whole algorithm to behave better in low memory 5667 condition. 5668 5669 rcutree.rcu_nocb_gp_stride= [KNL] 5670 Set the number of NOCB callback kthreads in 5671 each group, which defaults to the square root 5672 of the number of CPUs. Larger numbers reduce 5673 the wakeup overhead on the global grace-period 5674 kthread, but increases that same overhead on 5675 each group's NOCB grace-period kthread. 5676 5677 rcutree.rcu_kick_kthreads= [KNL] 5678 Cause the grace-period kthread to get an extra 5679 wake_up() if it sleeps three times longer than 5680 it should at force-quiescent-state time. 5681 This wake_up() will be accompanied by a 5682 WARN_ONCE() splat and an ftrace_dump(). 5683 5684 rcutree.rcu_resched_ns= [KNL] 5685 Limit the time spend invoking a batch of RCU 5686 callbacks to the specified number of nanoseconds. 5687 By default, this limit is checked only once 5688 every 32 callbacks in order to limit the pain 5689 inflicted by local_clock() overhead. 5690 5691 rcutree.rcu_unlock_delay= [KNL] 5692 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels, 5693 this specifies an rcu_read_unlock()-time delay 5694 in microseconds. This defaults to zero. 5695 Larger delays increase the probability of 5696 catching RCU pointer leaks, that is, buggy use 5697 of RCU-protected pointers after the relevant 5698 rcu_read_unlock() has completed. 5699 5700 rcutree.sysrq_rcu= [KNL] 5701 Commandeer a sysrq key to dump out Tree RCU's 5702 rcu_node tree with an eye towards determining 5703 why a new grace period has not yet started. 5704 5705 rcutree.use_softirq= [KNL] 5706 If set to zero, move all RCU_SOFTIRQ processing to 5707 per-CPU rcuc kthreads. Defaults to a non-zero 5708 value, meaning that RCU_SOFTIRQ is used by default. 5709 Specify rcutree.use_softirq=0 to use rcuc kthreads. 5710 5711 But note that CONFIG_PREEMPT_RT=y kernels disable 5712 this kernel boot parameter, forcibly setting it 5713 to zero. 5714 5715 rcutree.enable_rcu_lazy= [KNL] 5716 To save power, batch RCU callbacks and flush after 5717 delay, memory pressure or callback list growing too 5718 big. 5719 5720 rcutree.rcu_normal_wake_from_gp= [KNL] 5721 Reduces a latency of synchronize_rcu() call. This approach 5722 maintains its own track of synchronize_rcu() callers, so it 5723 does not interact with regular callbacks because it does not 5724 use a call_rcu[_hurry]() path. Please note, this is for a 5725 normal grace period. 5726 5727 How to enable it: 5728 5729 echo 1 > /sys/module/rcutree/parameters/rcu_normal_wake_from_gp 5730 or pass a boot parameter "rcutree.rcu_normal_wake_from_gp=1" 5731 5732 Default is 1 if num_possible_cpus() <= 16 and it is not explicitly 5733 disabled by the boot parameter passing 0. 5734 5735 rcuscale.gp_async= [KNL] 5736 Measure performance of asynchronous 5737 grace-period primitives such as call_rcu(). 5738 5739 rcuscale.gp_async_max= [KNL] 5740 Specify the maximum number of outstanding 5741 callbacks per writer thread. When a writer 5742 thread exceeds this limit, it invokes the 5743 corresponding flavor of rcu_barrier() to allow 5744 previously posted callbacks to drain. 5745 5746 rcuscale.gp_exp= [KNL] 5747 Measure performance of expedited synchronous 5748 grace-period primitives. 5749 5750 rcuscale.holdoff= [KNL] 5751 Set test-start holdoff period. The purpose of 5752 this parameter is to delay the start of the 5753 test until boot completes in order to avoid 5754 interference. 5755 5756 rcuscale.kfree_by_call_rcu= [KNL] 5757 In kernels built with CONFIG_RCU_LAZY=y, test 5758 call_rcu() instead of kfree_rcu(). 5759 5760 rcuscale.kfree_mult= [KNL] 5761 Instead of allocating an object of size kfree_obj, 5762 allocate one of kfree_mult * sizeof(kfree_obj). 5763 Defaults to 1. 5764 5765 rcuscale.kfree_rcu_test= [KNL] 5766 Set to measure performance of kfree_rcu() flooding. 5767 5768 rcuscale.kfree_rcu_test_double= [KNL] 5769 Test the double-argument variant of kfree_rcu(). 5770 If this parameter has the same value as 5771 rcuscale.kfree_rcu_test_single, both the single- 5772 and double-argument variants are tested. 5773 5774 rcuscale.kfree_rcu_test_single= [KNL] 5775 Test the single-argument variant of kfree_rcu(). 5776 If this parameter has the same value as 5777 rcuscale.kfree_rcu_test_double, both the single- 5778 and double-argument variants are tested. 5779 5780 rcuscale.kfree_nthreads= [KNL] 5781 The number of threads running loops of kfree_rcu(). 5782 5783 rcuscale.kfree_alloc_num= [KNL] 5784 Number of allocations and frees done in an iteration. 5785 5786 rcuscale.kfree_loops= [KNL] 5787 Number of loops doing rcuscale.kfree_alloc_num number 5788 of allocations and frees. 5789 5790 rcuscale.minruntime= [KNL] 5791 Set the minimum test run time in seconds. This 5792 does not affect the data-collection interval, 5793 but instead allows better measurement of things 5794 like CPU consumption. 5795 5796 rcuscale.nreaders= [KNL] 5797 Set number of RCU readers. The value -1 selects 5798 N, where N is the number of CPUs. A value 5799 "n" less than -1 selects N-n+1, where N is again 5800 the number of CPUs. For example, -2 selects N 5801 (the number of CPUs), -3 selects N+1, and so on. 5802 A value of "n" less than or equal to -N selects 5803 a single reader. 5804 5805 rcuscale.nwriters= [KNL] 5806 Set number of RCU writers. The values operate 5807 the same as for rcuscale.nreaders. 5808 N, where N is the number of CPUs 5809 5810 rcuscale.scale_type= [KNL] 5811 Specify the RCU implementation to test. 5812 5813 rcuscale.shutdown= [KNL] 5814 Shut the system down after performance tests 5815 complete. This is useful for hands-off automated 5816 testing. 5817 5818 rcuscale.verbose= [KNL] 5819 Enable additional printk() statements. 5820 5821 rcuscale.writer_holdoff= [KNL] 5822 Write-side holdoff between grace periods, 5823 in microseconds. The default of zero says 5824 no holdoff. 5825 5826 rcuscale.writer_holdoff_jiffies= [KNL] 5827 Additional write-side holdoff between grace 5828 periods, but in jiffies. The default of zero 5829 says no holdoff. 5830 5831 rcutorture.fqs_duration= [KNL] 5832 Set duration of force_quiescent_state bursts 5833 in microseconds. 5834 5835 rcutorture.fqs_holdoff= [KNL] 5836 Set holdoff time within force_quiescent_state bursts 5837 in microseconds. 5838 5839 rcutorture.fqs_stutter= [KNL] 5840 Set wait time between force_quiescent_state bursts 5841 in seconds. 5842 5843 rcutorture.fwd_progress= [KNL] 5844 Specifies the number of kthreads to be used 5845 for RCU grace-period forward-progress testing 5846 for the types of RCU supporting this notion. 5847 Defaults to 1 kthread, values less than zero or 5848 greater than the number of CPUs cause the number 5849 of CPUs to be used. 5850 5851 rcutorture.fwd_progress_div= [KNL] 5852 Specify the fraction of a CPU-stall-warning 5853 period to do tight-loop forward-progress testing. 5854 5855 rcutorture.fwd_progress_holdoff= [KNL] 5856 Number of seconds to wait between successive 5857 forward-progress tests. 5858 5859 rcutorture.fwd_progress_need_resched= [KNL] 5860 Enclose cond_resched() calls within checks for 5861 need_resched() during tight-loop forward-progress 5862 testing. 5863 5864 rcutorture.gp_cond= [KNL] 5865 Use conditional/asynchronous update-side 5866 normal-grace-period primitives, if available. 5867 5868 rcutorture.gp_cond_exp= [KNL] 5869 Use conditional/asynchronous update-side 5870 expedited-grace-period primitives, if available. 5871 5872 rcutorture.gp_cond_full= [KNL] 5873 Use conditional/asynchronous update-side 5874 normal-grace-period primitives that also take 5875 concurrent expedited grace periods into account, 5876 if available. 5877 5878 rcutorture.gp_cond_exp_full= [KNL] 5879 Use conditional/asynchronous update-side 5880 expedited-grace-period primitives that also take 5881 concurrent normal grace periods into account, 5882 if available. 5883 5884 rcutorture.gp_cond_wi= [KNL] 5885 Nominal wait interval for normal conditional 5886 grace periods (specified by rcutorture's 5887 gp_cond and gp_cond_full module parameters), 5888 in microseconds. The actual wait interval will 5889 be randomly selected to nanosecond granularity up 5890 to this wait interval. Defaults to 16 jiffies, 5891 for example, 16,000 microseconds on a system 5892 with HZ=1000. 5893 5894 rcutorture.gp_cond_wi_exp= [KNL] 5895 Nominal wait interval for expedited conditional 5896 grace periods (specified by rcutorture's 5897 gp_cond_exp and gp_cond_exp_full module 5898 parameters), in microseconds. The actual wait 5899 interval will be randomly selected to nanosecond 5900 granularity up to this wait interval. Defaults to 5901 128 microseconds. 5902 5903 rcutorture.gp_exp= [KNL] 5904 Use expedited update-side primitives, if available. 5905 5906 rcutorture.gp_normal= [KNL] 5907 Use normal (non-expedited) asynchronous 5908 update-side primitives, if available. 5909 5910 rcutorture.gp_poll= [KNL] 5911 Use polled update-side normal-grace-period 5912 primitives, if available. 5913 5914 rcutorture.gp_poll_exp= [KNL] 5915 Use polled update-side expedited-grace-period 5916 primitives, if available. 5917 5918 rcutorture.gp_poll_full= [KNL] 5919 Use polled update-side normal-grace-period 5920 primitives that also take concurrent expedited 5921 grace periods into account, if available. 5922 5923 rcutorture.gp_poll_exp_full= [KNL] 5924 Use polled update-side expedited-grace-period 5925 primitives that also take concurrent normal 5926 grace periods into account, if available. 5927 5928 rcutorture.gp_poll_wi= [KNL] 5929 Nominal wait interval for normal conditional 5930 grace periods (specified by rcutorture's 5931 gp_poll and gp_poll_full module parameters), 5932 in microseconds. The actual wait interval will 5933 be randomly selected to nanosecond granularity up 5934 to this wait interval. Defaults to 16 jiffies, 5935 for example, 16,000 microseconds on a system 5936 with HZ=1000. 5937 5938 rcutorture.gp_poll_wi_exp= [KNL] 5939 Nominal wait interval for expedited conditional 5940 grace periods (specified by rcutorture's 5941 gp_poll_exp and gp_poll_exp_full module 5942 parameters), in microseconds. The actual wait 5943 interval will be randomly selected to nanosecond 5944 granularity up to this wait interval. Defaults to 5945 128 microseconds. 5946 5947 rcutorture.gp_sync= [KNL] 5948 Use normal (non-expedited) synchronous 5949 update-side primitives, if available. If all 5950 of rcutorture.gp_cond=, rcutorture.gp_exp=, 5951 rcutorture.gp_normal=, and rcutorture.gp_sync= 5952 are zero, rcutorture acts as if is interpreted 5953 they are all non-zero. 5954 5955 rcutorture.gpwrap_lag= [KNL] 5956 Enable grace-period wrap lag testing. Setting 5957 to false prevents the gpwrap lag test from 5958 running. Default is true. 5959 5960 rcutorture.gpwrap_lag_gps= [KNL] 5961 Set the value for grace-period wrap lag during 5962 active lag testing periods. This controls how many 5963 grace periods differences we tolerate between 5964 rdp and rnp's gp_seq before setting overflow flag. 5965 The default is always set to 8. 5966 5967 rcutorture.gpwrap_lag_cycle_mins= [KNL] 5968 Set the total cycle duration for gpwrap lag 5969 testing in minutes. This is the total time for 5970 one complete cycle of active and inactive 5971 testing periods. Default is 30 minutes. 5972 5973 rcutorture.gpwrap_lag_active_mins= [KNL] 5974 Set the duration for which gpwrap lag is active 5975 within each cycle, in minutes. During this time, 5976 the grace-period wrap lag will be set to the 5977 value specified by gpwrap_lag_gps. Default is 5978 5 minutes. 5979 5980 rcutorture.irqreader= [KNL] 5981 Run RCU readers from irq handlers, or, more 5982 accurately, from a timer handler. Not all RCU 5983 flavors take kindly to this sort of thing. 5984 5985 rcutorture.leakpointer= [KNL] 5986 Leak an RCU-protected pointer out of the reader. 5987 This can of course result in splats, and is 5988 intended to test the ability of things like 5989 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect 5990 such leaks. 5991 5992 rcutorture.n_barrier_cbs= [KNL] 5993 Set callbacks/threads for rcu_barrier() testing. 5994 5995 rcutorture.nfakewriters= [KNL] 5996 Set number of concurrent RCU writers. These just 5997 stress RCU, they don't participate in the actual 5998 test, hence the "fake". 5999 6000 rcutorture.nocbs_nthreads= [KNL] 6001 Set number of RCU callback-offload togglers. 6002 Zero (the default) disables toggling. 6003 6004 rcutorture.nocbs_toggle= [KNL] 6005 Set the delay in milliseconds between successive 6006 callback-offload toggling attempts. 6007 6008 rcutorture.nreaders= [KNL] 6009 Set number of RCU readers. The value -1 selects 6010 N-1, where N is the number of CPUs. A value 6011 "n" less than -1 selects N-n-2, where N is again 6012 the number of CPUs. For example, -2 selects N 6013 (the number of CPUs), -3 selects N+1, and so on. 6014 6015 rcutorture.object_debug= [KNL] 6016 Enable debug-object double-call_rcu() testing. 6017 6018 rcutorture.onoff_holdoff= [KNL] 6019 Set time (s) after boot for CPU-hotplug testing. 6020 6021 rcutorture.onoff_interval= [KNL] 6022 Set time (jiffies) between CPU-hotplug operations, 6023 or zero to disable CPU-hotplug testing. 6024 6025 rcutorture.preempt_duration= [KNL] 6026 Set duration (in milliseconds) of preemptions 6027 by a high-priority FIFO real-time task. Set to 6028 zero (the default) to disable. The CPUs to 6029 preempt are selected randomly from the set that 6030 are online at a given point in time. Races with 6031 CPUs going offline are ignored, with that attempt 6032 at preemption skipped. 6033 6034 rcutorture.preempt_interval= [KNL] 6035 Set interval (in milliseconds, defaulting to one 6036 second) between preemptions by a high-priority 6037 FIFO real-time task. This delay is mediated 6038 by an hrtimer and is further fuzzed to avoid 6039 inadvertent synchronizations. 6040 6041 rcutorture.read_exit_burst= [KNL] 6042 The number of times in a given read-then-exit 6043 episode that a set of read-then-exit kthreads 6044 is spawned. 6045 6046 rcutorture.read_exit_delay= [KNL] 6047 The delay, in seconds, between successive 6048 read-then-exit testing episodes. 6049 6050 rcutorture.reader_flavor= [KNL] 6051 A bit mask indicating which readers to use. 6052 If there is more than one bit set, the readers 6053 are entered from low-order bit up, and are 6054 exited in the opposite order. For SRCU, the 6055 0x1 bit is normal readers, 0x2 NMI-safe readers, 6056 and 0x4 light-weight readers. 6057 6058 rcutorture.shuffle_interval= [KNL] 6059 Set task-shuffle interval (s). Shuffling tasks 6060 allows some CPUs to go into dyntick-idle mode 6061 during the rcutorture test. 6062 6063 rcutorture.shutdown_secs= [KNL] 6064 Set time (s) after boot system shutdown. This 6065 is useful for hands-off automated testing. 6066 6067 rcutorture.stall_cpu= [KNL] 6068 Duration of CPU stall (s) to test RCU CPU stall 6069 warnings, zero to disable. 6070 6071 rcutorture.stall_cpu_block= [KNL] 6072 Sleep while stalling if set. This will result 6073 in warnings from preemptible RCU in addition to 6074 any other stall-related activity. Note that 6075 in kernels built with CONFIG_PREEMPTION=n and 6076 CONFIG_PREEMPT_COUNT=y, this parameter will 6077 cause the CPU to pass through a quiescent state. 6078 Given CONFIG_PREEMPTION=n, this will suppress 6079 RCU CPU stall warnings, but will instead result 6080 in scheduling-while-atomic splats. 6081 6082 Use of this module parameter results in splats. 6083 6084 6085 rcutorture.stall_cpu_holdoff= [KNL] 6086 Time to wait (s) after boot before inducing stall. 6087 6088 rcutorture.stall_cpu_irqsoff= [KNL] 6089 Disable interrupts while stalling if set, but only 6090 on the first stall in the set. 6091 6092 rcutorture.stall_cpu_repeat= [KNL] 6093 Number of times to repeat the stall sequence, 6094 so that rcutorture.stall_cpu_repeat=3 will result 6095 in four stall sequences. 6096 6097 rcutorture.stall_gp_kthread= [KNL] 6098 Duration (s) of forced sleep within RCU 6099 grace-period kthread to test RCU CPU stall 6100 warnings, zero to disable. If both stall_cpu 6101 and stall_gp_kthread are specified, the 6102 kthread is starved first, then the CPU. 6103 6104 rcutorture.stat_interval= [KNL] 6105 Time (s) between statistics printk()s. 6106 6107 rcutorture.stutter= [KNL] 6108 Time (s) to stutter testing, for example, specifying 6109 five seconds causes the test to run for five seconds, 6110 wait for five seconds, and so on. This tests RCU's 6111 ability to transition abruptly to and from idle. 6112 6113 rcutorture.test_boost= [KNL] 6114 Test RCU priority boosting? 0=no, 1=maybe, 2=yes. 6115 "Maybe" means test if the RCU implementation 6116 under test support RCU priority boosting. 6117 6118 rcutorture.test_boost_duration= [KNL] 6119 Duration (s) of each individual boost test. 6120 6121 rcutorture.test_boost_holdoff= [KNL] 6122 Holdoff time (s) from start of test to the start 6123 of RCU priority-boost testing. Defaults to zero, 6124 that is, no holdoff. 6125 6126 rcutorture.test_boost_interval= [KNL] 6127 Interval (s) between each boost test. 6128 6129 rcutorture.test_no_idle_hz= [KNL] 6130 Test RCU's dyntick-idle handling. See also the 6131 rcutorture.shuffle_interval parameter. 6132 6133 rcutorture.torture_type= [KNL] 6134 Specify the RCU implementation to test. 6135 6136 rcutorture.verbose= [KNL] 6137 Enable additional printk() statements. 6138 6139 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL] 6140 Dump ftrace buffer after reporting RCU CPU 6141 stall warning. 6142 6143 rcupdate.rcu_cpu_stall_notifiers= [KNL] 6144 Provide RCU CPU stall notifiers, but see the 6145 warnings in the RCU_CPU_STALL_NOTIFIER Kconfig 6146 option's help text. TL;DR: You almost certainly 6147 do not want rcupdate.rcu_cpu_stall_notifiers. 6148 6149 rcupdate.rcu_cpu_stall_suppress= [KNL] 6150 Suppress RCU CPU stall warning messages. 6151 6152 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL] 6153 Suppress RCU CPU stall warning messages and 6154 rcutorture writer stall warnings that occur 6155 during early boot, that is, during the time 6156 before the init task is spawned. 6157 6158 rcupdate.rcu_cpu_stall_timeout= [KNL] 6159 Set timeout for RCU CPU stall warning messages. 6160 The value is in seconds and the maximum allowed 6161 value is 300 seconds. 6162 6163 rcupdate.rcu_exp_cpu_stall_timeout= [KNL] 6164 Set timeout for expedited RCU CPU stall warning 6165 messages. The value is in milliseconds 6166 and the maximum allowed value is 21000 6167 milliseconds. Please note that this value is 6168 adjusted to an arch timer tick resolution. 6169 Setting this to zero causes the value from 6170 rcupdate.rcu_cpu_stall_timeout to be used (after 6171 conversion from seconds to milliseconds). 6172 6173 rcupdate.rcu_cpu_stall_cputime= [KNL] 6174 Provide statistics on the cputime and count of 6175 interrupts and tasks during the sampling period. For 6176 multiple continuous RCU stalls, all sampling periods 6177 begin at half of the first RCU stall timeout. 6178 6179 rcupdate.rcu_exp_stall_task_details= [KNL] 6180 Print stack dumps of any tasks blocking the 6181 current expedited RCU grace period during an 6182 expedited RCU CPU stall warning. 6183 6184 rcupdate.rcu_expedited= [KNL] 6185 Use expedited grace-period primitives, for 6186 example, synchronize_rcu_expedited() instead 6187 of synchronize_rcu(). This reduces latency, 6188 but can increase CPU utilization, degrade 6189 real-time latency, and degrade energy efficiency. 6190 No effect on CONFIG_TINY_RCU kernels. 6191 6192 rcupdate.rcu_normal= [KNL] 6193 Use only normal grace-period primitives, 6194 for example, synchronize_rcu() instead of 6195 synchronize_rcu_expedited(). This improves 6196 real-time latency, CPU utilization, and 6197 energy efficiency, but can expose users to 6198 increased grace-period latency. This parameter 6199 overrides rcupdate.rcu_expedited. No effect on 6200 CONFIG_TINY_RCU kernels. 6201 6202 rcupdate.rcu_normal_after_boot= [KNL] 6203 Once boot has completed (that is, after 6204 rcu_end_inkernel_boot() has been invoked), use 6205 only normal grace-period primitives. No effect 6206 on CONFIG_TINY_RCU kernels. 6207 6208 But note that CONFIG_PREEMPT_RT=y kernels enables 6209 this kernel boot parameter, forcibly setting 6210 it to the value one, that is, converting any 6211 post-boot attempt at an expedited RCU grace 6212 period to instead use normal non-expedited 6213 grace-period processing. 6214 6215 rcupdate.rcu_task_collapse_lim= [KNL] 6216 Set the maximum number of callbacks present 6217 at the beginning of a grace period that allows 6218 the RCU Tasks flavors to collapse back to using 6219 a single callback queue. This switching only 6220 occurs when rcupdate.rcu_task_enqueue_lim is 6221 set to the default value of -1. 6222 6223 rcupdate.rcu_task_contend_lim= [KNL] 6224 Set the minimum number of callback-queuing-time 6225 lock-contention events per jiffy required to 6226 cause the RCU Tasks flavors to switch to per-CPU 6227 callback queuing. This switching only occurs 6228 when rcupdate.rcu_task_enqueue_lim is set to 6229 the default value of -1. 6230 6231 rcupdate.rcu_task_enqueue_lim= [KNL] 6232 Set the number of callback queues to use for the 6233 RCU Tasks family of RCU flavors. The default 6234 of -1 allows this to be automatically (and 6235 dynamically) adjusted. This parameter is intended 6236 for use in testing. 6237 6238 rcupdate.rcu_task_ipi_delay= [KNL] 6239 Set time in jiffies during which RCU tasks will 6240 avoid sending IPIs, starting with the beginning 6241 of a given grace period. Setting a large 6242 number avoids disturbing real-time workloads, 6243 but lengthens grace periods. 6244 6245 rcupdate.rcu_task_lazy_lim= [KNL] 6246 Number of callbacks on a given CPU that will 6247 cancel laziness on that CPU. Use -1 to disable 6248 cancellation of laziness, but be advised that 6249 doing so increases the danger of OOM due to 6250 callback flooding. 6251 6252 rcupdate.rcu_task_stall_info= [KNL] 6253 Set initial timeout in jiffies for RCU task stall 6254 informational messages, which give some indication 6255 of the problem for those not patient enough to 6256 wait for ten minutes. Informational messages are 6257 only printed prior to the stall-warning message 6258 for a given grace period. Disable with a value 6259 less than or equal to zero. Defaults to ten 6260 seconds. A change in value does not take effect 6261 until the beginning of the next grace period. 6262 6263 rcupdate.rcu_task_stall_info_mult= [KNL] 6264 Multiplier for time interval between successive 6265 RCU task stall informational messages for a given 6266 RCU tasks grace period. This value is clamped 6267 to one through ten, inclusive. It defaults to 6268 the value three, so that the first informational 6269 message is printed 10 seconds into the grace 6270 period, the second at 40 seconds, the third at 6271 160 seconds, and then the stall warning at 600 6272 seconds would prevent a fourth at 640 seconds. 6273 6274 rcupdate.rcu_task_stall_timeout= [KNL] 6275 Set timeout in jiffies for RCU task stall 6276 warning messages. Disable with a value less 6277 than or equal to zero. Defaults to ten minutes. 6278 A change in value does not take effect until 6279 the beginning of the next grace period. 6280 6281 rcupdate.rcu_tasks_lazy_ms= [KNL] 6282 Set timeout in milliseconds RCU Tasks asynchronous 6283 callback batching for call_rcu_tasks(). 6284 A negative value will take the default. A value 6285 of zero will disable batching. Batching is 6286 always disabled for synchronize_rcu_tasks(). 6287 6288 rcupdate.rcu_tasks_trace_lazy_ms= [KNL] 6289 Set timeout in milliseconds RCU Tasks 6290 Trace asynchronous callback batching for 6291 call_rcu_tasks_trace(). A negative value 6292 will take the default. A value of zero will 6293 disable batching. Batching is always disabled 6294 for synchronize_rcu_tasks_trace(). 6295 6296 rcupdate.rcu_self_test= [KNL] 6297 Run the RCU early boot self tests 6298 6299 rdinit= [KNL] 6300 Format: <full_path> 6301 Run specified binary instead of /init from the ramdisk, 6302 used for early userspace startup. See initrd. 6303 6304 rdrand= [X86,EARLY] 6305 force - Override the decision by the kernel to hide the 6306 advertisement of RDRAND support (this affects 6307 certain AMD processors because of buggy BIOS 6308 support, specifically around the suspend/resume 6309 path). 6310 6311 rdt= [HW,X86,RDT] 6312 Turn on/off individual RDT features. List is: 6313 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp, 6314 mba, smba, bmec, abmc, sdciae. 6315 E.g. to turn on cmt and turn off mba use: 6316 rdt=cmt,!mba 6317 6318 reboot= [KNL] 6319 Format (x86 or x86_64): 6320 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] | d[efault] \ 6321 [[,]s[mp]#### \ 6322 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \ 6323 [[,]f[orce] 6324 Where reboot_mode is one of warm (soft) or cold (hard) or gpio 6325 (prefix with 'panic_' to set mode for panic 6326 reboot only), 6327 reboot_type is one of bios, acpi, kbd, triple, efi, or pci, 6328 reboot_force is either force or not specified, 6329 reboot_cpu is s[mp]#### with #### being the processor 6330 to be used for rebooting. 6331 6332 acpi 6333 Use the ACPI RESET_REG in the FADT. If ACPI is not 6334 configured or the ACPI reset does not work, the reboot 6335 path attempts the reset using the keyboard controller. 6336 6337 bios 6338 Use the CPU reboot vector for warm reset 6339 6340 cold 6341 Set the cold reboot flag 6342 6343 default 6344 There are some built-in platform specific "quirks" 6345 - you may see: "reboot: <name> series board detected. 6346 Selecting <type> for reboots." In the case where you 6347 think the quirk is in error (e.g. you have newer BIOS, 6348 or newer board) using this option will ignore the 6349 built-in quirk table, and use the generic default 6350 reboot actions. 6351 6352 efi 6353 Use efi reset_system runtime service. If EFI is not 6354 configured or the EFI reset does not work, the reboot 6355 path attempts the reset using the keyboard controller. 6356 6357 force 6358 Don't stop other CPUs on reboot. This can make reboot 6359 more reliable in some cases. 6360 6361 kbd 6362 Use the keyboard controller. cold reset (default) 6363 6364 pci 6365 Use a write to the PCI config space register 0xcf9 to 6366 trigger reboot. 6367 6368 triple 6369 Force a triple fault (init) 6370 6371 warm 6372 Don't set the cold reboot flag 6373 6374 Using warm reset will be much faster especially on big 6375 memory systems because the BIOS will not go through 6376 the memory check. Disadvantage is that not all 6377 hardware will be completely reinitialized on reboot so 6378 there may be boot problems on some systems. 6379 6380 6381 refscale.holdoff= [KNL] 6382 Set test-start holdoff period. The purpose of 6383 this parameter is to delay the start of the 6384 test until boot completes in order to avoid 6385 interference. 6386 6387 refscale.lookup_instances= [KNL] 6388 Number of data elements to use for the forms of 6389 SLAB_TYPESAFE_BY_RCU testing. A negative number 6390 is negated and multiplied by nr_cpu_ids, while 6391 zero specifies nr_cpu_ids. 6392 6393 refscale.loops= [KNL] 6394 Set the number of loops over the synchronization 6395 primitive under test. Increasing this number 6396 reduces noise due to loop start/end overhead, 6397 but the default has already reduced the per-pass 6398 noise to a handful of picoseconds on ca. 2020 6399 x86 laptops. 6400 6401 refscale.nreaders= [KNL] 6402 Set number of readers. The default value of -1 6403 selects N, where N is roughly 75% of the number 6404 of CPUs. A value of zero is an interesting choice. 6405 6406 refscale.nruns= [KNL] 6407 Set number of runs, each of which is dumped onto 6408 the console log. 6409 6410 refscale.readdelay= [KNL] 6411 Set the read-side critical-section duration, 6412 measured in microseconds. 6413 6414 refscale.scale_type= [KNL] 6415 Specify the read-protection implementation to test. 6416 6417 refscale.shutdown= [KNL] 6418 Shut down the system at the end of the performance 6419 test. This defaults to 1 (shut it down) when 6420 refscale is built into the kernel and to 0 (leave 6421 it running) when refscale is built as a module. 6422 6423 refscale.verbose= [KNL] 6424 Enable additional printk() statements. 6425 6426 refscale.verbose_batched= [KNL] 6427 Batch the additional printk() statements. If zero 6428 (the default) or negative, print everything. Otherwise, 6429 print every Nth verbose statement, where N is the value 6430 specified. 6431 6432 regulator_ignore_unused 6433 [REGULATOR] 6434 Prevents regulator framework from disabling regulators 6435 that are unused, due no driver claiming them. This may 6436 be useful for debug and development, but should not be 6437 needed on a platform with proper driver support. 6438 6439 relax_domain_level= 6440 [KNL, SMP] Set scheduler's default relax_domain_level. 6441 See Documentation/admin-guide/cgroup-v1/cpusets.rst. 6442 6443 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory 6444 Format: <base1>,<size1>[,<base2>,<size2>,...] 6445 Reserve I/O ports or memory so the kernel won't use 6446 them. If <base> is less than 0x10000, the region 6447 is assumed to be I/O ports; otherwise it is memory. 6448 6449 reserve_mem= [RAM] 6450 Format: nn[KMG]:<align>:<label> 6451 Reserve physical memory and label it with a name that 6452 other subsystems can use to access it. This is typically 6453 used for systems that do not wipe the RAM, and this command 6454 line will try to reserve the same physical memory on 6455 soft reboots. Note, it is not guaranteed to be the same 6456 location. For example, if anything about the system changes 6457 or if booting a different kernel. It can also fail if KASLR 6458 places the kernel at the location of where the RAM reservation 6459 was from a previous boot, the new reservation will be at a 6460 different location. 6461 Any subsystem using this feature must add a way to verify 6462 that the contents of the physical memory is from a previous 6463 boot, as there may be cases where the memory will not be 6464 located at the same location. 6465 6466 The format is size:align:label for example, to request 6467 12 megabytes of 4096 alignment for ramoops: 6468 6469 reserve_mem=12M:4096:oops ramoops.mem_name=oops 6470 6471 reservetop= [X86-32,EARLY] 6472 Format: nn[KMG] 6473 Reserves a hole at the top of the kernel virtual 6474 address space. 6475 6476 reset_devices [KNL] Force drivers to reset the underlying device 6477 during initialization. 6478 6479 resume= [SWSUSP] 6480 Specify the partition device for software suspend 6481 Format: 6482 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>} 6483 6484 resume_offset= [SWSUSP] 6485 Specify the offset from the beginning of the partition 6486 given by "resume=" at which the swap header is located, 6487 in <PAGE_SIZE> units (needed only for swap files). 6488 See Documentation/power/swsusp-and-swap-files.rst 6489 6490 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to 6491 read the resume files 6492 6493 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up. 6494 Useful for devices that are detected asynchronously 6495 (e.g. USB and MMC devices). 6496 6497 retain_initrd [RAM] Keep initrd memory after extraction. After boot, it will 6498 be accessible via /sys/firmware/initrd. 6499 6500 retbleed= [X86] Control mitigation of RETBleed (Arbitrary 6501 Speculative Code Execution with Return Instructions) 6502 vulnerability. 6503 6504 AMD-based UNRET and IBPB mitigations alone do not stop 6505 sibling threads from influencing the predictions of other 6506 sibling threads. For that reason, STIBP is used on pro- 6507 cessors that support it, and mitigate SMT on processors 6508 that don't. 6509 6510 off - no mitigation 6511 auto - automatically select a mitigation 6512 auto,nosmt - automatically select a mitigation, 6513 disabling SMT if necessary for 6514 the full mitigation (only on Zen1 6515 and older without STIBP). 6516 ibpb - On AMD, mitigate short speculation 6517 windows on basic block boundaries too. 6518 Safe, highest perf impact. It also 6519 enables STIBP if present. Not suitable 6520 on Intel. 6521 ibpb,nosmt - Like "ibpb" above but will disable SMT 6522 when STIBP is not available. This is 6523 the alternative for systems which do not 6524 have STIBP. 6525 unret - Force enable untrained return thunks, 6526 only effective on AMD f15h-f17h based 6527 systems. 6528 unret,nosmt - Like unret, but will disable SMT when STIBP 6529 is not available. This is the alternative for 6530 systems which do not have STIBP. 6531 6532 Selecting 'auto' will choose a mitigation method at run 6533 time according to the CPU. 6534 6535 Not specifying this option is equivalent to retbleed=auto. 6536 6537 rfkill.default_state= 6538 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm, 6539 etc. communication is blocked by default. 6540 1 Unblocked. 6541 6542 rfkill.master_switch_mode= 6543 0 The "airplane mode" button does nothing. 6544 1 The "airplane mode" button toggles between everything 6545 blocked and the previous configuration. 6546 2 The "airplane mode" button toggles between everything 6547 blocked and everything unblocked. 6548 6549 ring3mwait=disable 6550 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported 6551 CPUs. 6552 6553 riscv_isa_fallback [RISCV,EARLY] 6554 When CONFIG_RISCV_ISA_FALLBACK is not enabled, permit 6555 falling back to detecting extension support by parsing 6556 "riscv,isa" property on devicetree systems when the 6557 replacement properties are not found. See the Kconfig 6558 entry for RISCV_ISA_FALLBACK. 6559 6560 ro [KNL] Mount root device read-only on boot 6561 6562 rodata= [KNL,EARLY] 6563 on Mark read-only kernel memory as read-only (default). 6564 off Leave read-only kernel memory writable for debugging. 6565 noalias Mark read-only kernel memory as read-only but retain 6566 writable aliases in the direct map for regions outside 6567 of the kernel image. [arm64] 6568 6569 rockchip.usb_uart 6570 [EARLY] 6571 Enable the uart passthrough on the designated usb port 6572 on Rockchip SoCs. When active, the signals of the 6573 debug-uart get routed to the D+ and D- pins of the usb 6574 port and the regular usb controller gets disabled. 6575 6576 root= [KNL] Root filesystem 6577 Usually this is a block device specifier of some kind, 6578 see the early_lookup_bdev comment in 6579 block/early-lookup.c for details. 6580 Alternatively this can be "ram" for the legacy initial 6581 ramdisk, "nfs" and "cifs" for root on a network file 6582 system, or "mtd" and "ubi" for mounting from raw flash. 6583 6584 rootdelay= [KNL] Delay (in seconds) to pause before attempting to 6585 mount the root filesystem 6586 6587 rootflags= [KNL] Set root filesystem mount option string 6588 6589 initramfs_options= [KNL] 6590 Specify mount options for for the initramfs mount. 6591 6592 rootfstype= [KNL] Set root filesystem type 6593 6594 rootwait [KNL] Wait (indefinitely) for root device to show up. 6595 Useful for devices that are detected asynchronously 6596 (e.g. USB and MMC devices). 6597 6598 rootwait= [KNL] Maximum time (in seconds) to wait for root device 6599 to show up before attempting to mount the root 6600 filesystem. 6601 6602 rproc_mem=nn[KMG][@address] 6603 [KNL,ARM,CMA] Remoteproc physical memory block. 6604 Memory area to be used by remote processor image, 6605 managed by CMA. 6606 6607 rseq_debug= [KNL] Enable or disable restartable sequence 6608 debug mode. Defaults to CONFIG_RSEQ_DEBUG_DEFAULT_ENABLE. 6609 Format: <bool> 6610 6611 rt_group_sched= [KNL] Enable or disable SCHED_RR/FIFO group scheduling 6612 when CONFIG_RT_GROUP_SCHED=y. Defaults to 6613 !CONFIG_RT_GROUP_SCHED_DEFAULT_DISABLED. 6614 Format: <bool> 6615 6616 rw [KNL] Mount root device read-write on boot 6617 6618 S [KNL] Run init in single mode 6619 6620 s390_iommu= [HW,S390] 6621 Set s390 IOTLB flushing mode 6622 strict 6623 With strict flushing every unmap operation will result 6624 in an IOTLB flush. Default is lazy flushing before 6625 reuse, which is faster. Deprecated, equivalent to 6626 iommu.strict=1. 6627 6628 s390_iommu_aperture= [KNL,S390] 6629 Specifies the size of the per device DMA address space 6630 accessible through the DMA and IOMMU APIs as a decimal 6631 factor of the size of main memory. 6632 The default is 1 meaning that one can concurrently use 6633 as many DMA addresses as physical memory is installed, 6634 if supported by hardware, and thus map all of memory 6635 once. With a value of 2 one can map all of memory twice 6636 and so on. As a special case a factor of 0 imposes no 6637 restrictions other than those given by hardware at the 6638 cost of significant additional memory use for tables. 6639 6640 sa1100ir [NET] 6641 See drivers/net/irda/sa1100_ir.c. 6642 6643 sched_proxy_exec= [KNL] 6644 Enables or disables "proxy execution" style 6645 solution to mutex-based priority inversion. 6646 Format: <bool> 6647 6648 sched_verbose [KNL,EARLY] Enables verbose scheduler debug messages. 6649 6650 schedstats= [KNL,X86] Enable or disable scheduled statistics. 6651 Allowed values are enable and disable. This feature 6652 incurs a small amount of overhead in the scheduler 6653 but is useful for debugging and performance tuning. 6654 6655 sched_thermal_decay_shift= 6656 [Deprecated] 6657 [KNL, SMP] Set a decay shift for scheduler thermal 6658 pressure signal. Thermal pressure signal follows the 6659 default decay period of other scheduler pelt 6660 signals(usually 32 ms but configurable). Setting 6661 sched_thermal_decay_shift will left shift the decay 6662 period for the thermal pressure signal by the shift 6663 value. 6664 i.e. with the default pelt decay period of 32 ms 6665 sched_thermal_decay_shift thermal pressure decay pr 6666 1 64 ms 6667 2 128 ms 6668 and so on. 6669 Format: integer between 0 and 10 6670 Default is 0. 6671 6672 scftorture.holdoff= [KNL] 6673 Number of seconds to hold off before starting 6674 test. Defaults to zero for module insertion and 6675 to 10 seconds for built-in smp_call_function() 6676 tests. 6677 6678 scftorture.longwait= [KNL] 6679 Request ridiculously long waits randomly selected 6680 up to the chosen limit in seconds. Zero (the 6681 default) disables this feature. Please note 6682 that requesting even small non-zero numbers of 6683 seconds can result in RCU CPU stall warnings, 6684 softlockup complaints, and so on. 6685 6686 scftorture.nthreads= [KNL] 6687 Number of kthreads to spawn to invoke the 6688 smp_call_function() family of functions. 6689 The default of -1 specifies a number of kthreads 6690 equal to the number of CPUs. 6691 6692 scftorture.onoff_holdoff= [KNL] 6693 Number seconds to wait after the start of the 6694 test before initiating CPU-hotplug operations. 6695 6696 scftorture.onoff_interval= [KNL] 6697 Number seconds to wait between successive 6698 CPU-hotplug operations. Specifying zero (which 6699 is the default) disables CPU-hotplug operations. 6700 6701 scftorture.shutdown_secs= [KNL] 6702 The number of seconds following the start of the 6703 test after which to shut down the system. The 6704 default of zero avoids shutting down the system. 6705 Non-zero values are useful for automated tests. 6706 6707 scftorture.stat_interval= [KNL] 6708 The number of seconds between outputting the 6709 current test statistics to the console. A value 6710 of zero disables statistics output. 6711 6712 scftorture.stutter_cpus= [KNL] 6713 The number of jiffies to wait between each change 6714 to the set of CPUs under test. 6715 6716 scftorture.use_cpus_read_lock= [KNL] 6717 Use use_cpus_read_lock() instead of the default 6718 preempt_disable() to disable CPU hotplug 6719 while invoking one of the smp_call_function*() 6720 functions. 6721 6722 scftorture.verbose= [KNL] 6723 Enable additional printk() statements. 6724 6725 scftorture.weight_single= [KNL] 6726 The probability weighting to use for the 6727 smp_call_function_single() function with a zero 6728 "wait" parameter. A value of -1 selects the 6729 default if all other weights are -1. However, 6730 if at least one weight has some other value, a 6731 value of -1 will instead select a weight of zero. 6732 6733 scftorture.weight_single_wait= [KNL] 6734 The probability weighting to use for the 6735 smp_call_function_single() function with a 6736 non-zero "wait" parameter. See weight_single. 6737 6738 scftorture.weight_many= [KNL] 6739 The probability weighting to use for the 6740 smp_call_function_many() function with a zero 6741 "wait" parameter. See weight_single. 6742 Note well that setting a high probability for 6743 this weighting can place serious IPI load 6744 on the system. 6745 6746 scftorture.weight_many_wait= [KNL] 6747 The probability weighting to use for the 6748 smp_call_function_many() function with a 6749 non-zero "wait" parameter. See weight_single 6750 and weight_many. 6751 6752 scftorture.weight_all= [KNL] 6753 The probability weighting to use for the 6754 smp_call_function_all() function with a zero 6755 "wait" parameter. See weight_single and 6756 weight_many. 6757 6758 scftorture.weight_all_wait= [KNL] 6759 The probability weighting to use for the 6760 smp_call_function_all() function with a 6761 non-zero "wait" parameter. See weight_single 6762 and weight_many. 6763 6764 sdw_mclk_divider=[SDW] 6765 Specify the MCLK divider for Intel SoundWire buses in 6766 case the BIOS does not provide the clock rate properly. 6767 6768 skew_tick= [KNL,EARLY] Offset the periodic timer tick per cpu to mitigate 6769 xtime_lock contention on larger systems, and/or RCU lock 6770 contention on all systems with CONFIG_MAXSMP set. 6771 Format: { "0" | "1" } 6772 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1" 6773 1 -- enable. 6774 Note: increases power consumption, thus should only be 6775 enabled if running jitter sensitive (HPC/RT) workloads. 6776 6777 security= [SECURITY] Choose a legacy "major" security module to 6778 enable at boot. This has been deprecated by the 6779 "lsm=" parameter. 6780 6781 selinux= [SELINUX] Disable or enable SELinux at boot time. 6782 Format: { "0" | "1" } 6783 See security/selinux/Kconfig help text. 6784 0 -- disable. 6785 1 -- enable. 6786 Default value is 1. 6787 6788 serialnumber [BUGS=X86-32] 6789 6790 sev=option[,option...] [X86-64] 6791 6792 debug 6793 Enable debug messages. 6794 6795 nosnp 6796 Do not enable SEV-SNP (applies to host/hypervisor 6797 only). Setting 'nosnp' avoids the RMP check overhead 6798 in memory accesses when users do not want to run 6799 SEV-SNP guests. 6800 6801 shapers= [NET] 6802 Maximal number of shapers. 6803 6804 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller 6805 Limit apic dumping. The parameter defines the maximal 6806 number of local apics being dumped. Also it is possible 6807 to set it to "all" by meaning -- no limit here. 6808 Format: { 1 (default) | 2 | ... | all }. 6809 The parameter valid if only apic=debug or 6810 apic=verbose is specified. 6811 Example: apic=debug show_lapic=all 6812 6813 slab_debug[=options[,slabs][;[options[,slabs]]...] [MM] 6814 Enabling slab_debug allows one to determine the 6815 culprit if slab objects become corrupted. Enabling 6816 slab_debug can create guard zones around objects and 6817 may poison objects when not in use. Also tracks the 6818 last alloc / free. For more information see 6819 Documentation/admin-guide/mm/slab.rst. 6820 (slub_debug legacy name also accepted for now) 6821 6822 Using this option implies the "no_hash_pointers" 6823 option which can be undone by adding the 6824 "hash_pointers=always" option. 6825 6826 slab_max_order= [MM] 6827 Determines the maximum allowed order for slabs. 6828 A high setting may cause OOMs due to memory 6829 fragmentation. For more information see 6830 Documentation/admin-guide/mm/slab.rst. 6831 (slub_max_order legacy name also accepted for now) 6832 6833 slab_merge [MM] 6834 Enable merging of slabs with similar size when the 6835 kernel is built without CONFIG_SLAB_MERGE_DEFAULT. 6836 (slub_merge legacy name also accepted for now) 6837 6838 slab_min_objects= [MM] 6839 The minimum number of objects per slab. SLUB will 6840 increase the slab order up to slab_max_order to 6841 generate a sufficiently large slab able to contain 6842 the number of objects indicated. The higher the number 6843 of objects the smaller the overhead of tracking slabs 6844 and the less frequently locks need to be acquired. 6845 For more information see 6846 Documentation/admin-guide/mm/slab.rst. 6847 (slub_min_objects legacy name also accepted for now) 6848 6849 slab_min_order= [MM] 6850 Determines the minimum page order for slabs. Must be 6851 lower or equal to slab_max_order. For more information see 6852 Documentation/admin-guide/mm/slab.rst. 6853 (slub_min_order legacy name also accepted for now) 6854 6855 slab_nomerge [MM] 6856 Disable merging of slabs with similar size. May be 6857 necessary if there is some reason to distinguish 6858 allocs to different slabs, especially in hardened 6859 environments where the risk of heap overflows and 6860 layout control by attackers can usually be 6861 frustrated by disabling merging. This will reduce 6862 most of the exposure of a heap attack to a single 6863 cache (risks via metadata attacks are mostly 6864 unchanged). Debug options disable merging on their 6865 own. 6866 For more information see 6867 Documentation/admin-guide/mm/slab.rst. 6868 (slub_nomerge legacy name also accepted for now) 6869 6870 slab_strict_numa [MM] 6871 Support memory policies on a per object level 6872 in the slab allocator. The default is for memory 6873 policies to be applied at the folio level when 6874 a new folio is needed or a partial folio is 6875 retrieved from the lists. Increases overhead 6876 in the slab fastpaths but gains more accurate 6877 NUMA kernel object placement which helps with slow 6878 interconnects in NUMA systems. 6879 6880 slram= [HW,MTD] 6881 6882 smart2= [HW] 6883 Format: <io1>[,<io2>[,...,<io8>]] 6884 6885 smp.csd_lock_timeout= [KNL] 6886 Specify the period of time in milliseconds 6887 that smp_call_function() and friends will wait 6888 for a CPU to release the CSD lock. This is 6889 useful when diagnosing bugs involving CPUs 6890 disabling interrupts for extended periods 6891 of time. Defaults to 5,000 milliseconds, and 6892 setting a value of zero disables this feature. 6893 This feature may be more efficiently disabled 6894 using the csdlock_debug- kernel parameter. 6895 6896 smp.panic_on_ipistall= [KNL] 6897 If a csd_lock_timeout extends for more than 6898 the specified number of milliseconds, panic the 6899 system. By default, let CSD-lock acquisition 6900 take as long as they take. Specifying 300,000 6901 for this value provides a 5-minute timeout. 6902 6903 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices 6904 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port 6905 smsc-ircc2.ircc_sir= [HW] SIR base I/O port 6906 smsc-ircc2.ircc_fir= [HW] FIR base I/O port 6907 smsc-ircc2.ircc_irq= [HW] IRQ line 6908 smsc-ircc2.ircc_dma= [HW] DMA channel 6909 smsc-ircc2.ircc_transceiver= [HW] Transceiver type: 6910 0: Toshiba Satellite 1800 (GP data pin select) 6911 1: Fast pin select (default) 6912 2: ATC IRMode 6913 6914 smt= [KNL,MIPS,S390,EARLY] Set the maximum number of threads 6915 (logical CPUs) to use per physical CPU on systems 6916 capable of symmetric multithreading (SMT). Will 6917 be capped to the actual hardware limit. 6918 Format: <integer> 6919 Default: -1 (no limit) 6920 6921 softlockup_panic= 6922 [KNL] Should the soft-lockup detector generate panics. 6923 Format: 0 | 1 6924 6925 A value of 1 instructs the soft-lockup detector 6926 to panic the machine when a soft-lockup occurs. It is 6927 also controlled by the kernel.softlockup_panic sysctl 6928 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the 6929 respective build-time switch to that functionality. 6930 6931 softlockup_all_cpu_backtrace= 6932 [KNL] Should the soft-lockup detector generate 6933 backtraces on all cpus. 6934 Format: 0 | 1 6935 6936 sonypi.*= [HW] Sony Programmable I/O Control Device driver 6937 See Documentation/admin-guide/laptops/sonypi.rst 6938 6939 spectre_bhi= [X86] Control mitigation of Branch History Injection 6940 (BHI) vulnerability. This setting affects the 6941 deployment of the HW BHI control and the SW BHB 6942 clearing sequence. 6943 6944 on - (default) Enable the HW or SW mitigation as 6945 needed. This protects the kernel from 6946 both syscalls and VMs. 6947 vmexit - On systems which don't have the HW mitigation 6948 available, enable the SW mitigation on vmexit 6949 ONLY. On such systems, the host kernel is 6950 protected from VM-originated BHI attacks, but 6951 may still be vulnerable to syscall attacks. 6952 off - Disable the mitigation. 6953 6954 spectre_v2= [X86,EARLY] Control mitigation of Spectre variant 2 6955 (indirect branch speculation) vulnerability. 6956 The default operation protects the kernel from 6957 user space attacks. 6958 6959 on - unconditionally enable, implies 6960 spectre_v2_user=on 6961 off - unconditionally disable, implies 6962 spectre_v2_user=off 6963 auto - kernel detects whether your CPU model is 6964 vulnerable 6965 6966 Selecting 'on' will, and 'auto' may, choose a 6967 mitigation method at run time according to the 6968 CPU, the available microcode, the setting of the 6969 CONFIG_MITIGATION_RETPOLINE configuration option, 6970 and the compiler with which the kernel was built. 6971 6972 Selecting 'on' will also enable the mitigation 6973 against user space to user space task attacks. 6974 Selecting specific mitigation does not force enable 6975 user mitigations. 6976 6977 Selecting 'off' will disable both the kernel and 6978 the user space protections. 6979 6980 Specific mitigations can also be selected manually: 6981 6982 retpoline - replace indirect branches 6983 retpoline,generic - Retpolines 6984 retpoline,lfence - LFENCE; indirect branch 6985 retpoline,amd - alias for retpoline,lfence 6986 eibrs - Enhanced/Auto IBRS 6987 eibrs,retpoline - Enhanced/Auto IBRS + Retpolines 6988 eibrs,lfence - Enhanced/Auto IBRS + LFENCE 6989 ibrs - use IBRS to protect kernel 6990 6991 Not specifying this option is equivalent to 6992 spectre_v2=auto. 6993 6994 spectre_v2_user= 6995 [X86] Control mitigation of Spectre variant 2 6996 (indirect branch speculation) vulnerability between 6997 user space tasks 6998 6999 on - Unconditionally enable mitigations. Is 7000 enforced by spectre_v2=on 7001 7002 off - Unconditionally disable mitigations. Is 7003 enforced by spectre_v2=off 7004 7005 prctl - Indirect branch speculation is enabled, 7006 but mitigation can be enabled via prctl 7007 per thread. The mitigation control state 7008 is inherited on fork. 7009 7010 prctl,ibpb 7011 - Like "prctl" above, but only STIBP is 7012 controlled per thread. IBPB is issued 7013 always when switching between different user 7014 space processes. 7015 7016 seccomp 7017 - Same as "prctl" above, but all seccomp 7018 threads will enable the mitigation unless 7019 they explicitly opt out. 7020 7021 seccomp,ibpb 7022 - Like "seccomp" above, but only STIBP is 7023 controlled per thread. IBPB is issued 7024 always when switching between different 7025 user space processes. 7026 7027 auto - Kernel selects the mitigation depending on 7028 the available CPU features and vulnerability. 7029 7030 Default mitigation: "prctl" 7031 7032 Not specifying this option is equivalent to 7033 spectre_v2_user=auto. 7034 7035 spec_rstack_overflow= 7036 [X86,EARLY] Control RAS overflow mitigation on AMD Zen CPUs 7037 7038 off - Disable mitigation 7039 microcode - Enable microcode mitigation only 7040 safe-ret - Enable sw-only safe RET mitigation (default) 7041 ibpb - Enable mitigation by issuing IBPB on 7042 kernel entry 7043 ibpb-vmexit - Issue IBPB only on VMEXIT 7044 (cloud-specific mitigation) 7045 7046 spec_store_bypass_disable= 7047 [HW,EARLY] Control Speculative Store Bypass (SSB) Disable mitigation 7048 (Speculative Store Bypass vulnerability) 7049 7050 Certain CPUs are vulnerable to an exploit against a 7051 a common industry wide performance optimization known 7052 as "Speculative Store Bypass" in which recent stores 7053 to the same memory location may not be observed by 7054 later loads during speculative execution. The idea 7055 is that such stores are unlikely and that they can 7056 be detected prior to instruction retirement at the 7057 end of a particular speculation execution window. 7058 7059 In vulnerable processors, the speculatively forwarded 7060 store can be used in a cache side channel attack, for 7061 example to read memory to which the attacker does not 7062 directly have access (e.g. inside sandboxed code). 7063 7064 This parameter controls whether the Speculative Store 7065 Bypass optimization is used. 7066 7067 On x86 the options are: 7068 7069 on - Unconditionally disable Speculative Store Bypass 7070 off - Unconditionally enable Speculative Store Bypass 7071 auto - Kernel detects whether the CPU model contains an 7072 implementation of Speculative Store Bypass and 7073 picks the most appropriate mitigation. If the 7074 CPU is not vulnerable, "off" is selected. If the 7075 CPU is vulnerable the default mitigation is 7076 architecture and Kconfig dependent. See below. 7077 prctl - Control Speculative Store Bypass per thread 7078 via prctl. Speculative Store Bypass is enabled 7079 for a process by default. The state of the control 7080 is inherited on fork. 7081 seccomp - Same as "prctl" above, but all seccomp threads 7082 will disable SSB unless they explicitly opt out. 7083 7084 Default mitigations: 7085 X86: "prctl" 7086 7087 On powerpc the options are: 7088 7089 on,auto - On Power8 and Power9 insert a store-forwarding 7090 barrier on kernel entry and exit. On Power7 7091 perform a software flush on kernel entry and 7092 exit. 7093 off - No action. 7094 7095 Not specifying this option is equivalent to 7096 spec_store_bypass_disable=auto. 7097 7098 split_lock_detect= 7099 [X86] Enable split lock detection or bus lock detection 7100 7101 When enabled (and if hardware support is present), atomic 7102 instructions that access data across cache line 7103 boundaries will result in an alignment check exception 7104 for split lock detection or a debug exception for 7105 bus lock detection. 7106 7107 off - not enabled 7108 7109 warn - the kernel will emit rate-limited warnings 7110 about applications triggering the #AC 7111 exception or the #DB exception. This mode is 7112 the default on CPUs that support split lock 7113 detection or bus lock detection. Default 7114 behavior is by #AC if both features are 7115 enabled in hardware. 7116 7117 fatal - the kernel will send SIGBUS to applications 7118 that trigger the #AC exception or the #DB 7119 exception. Default behavior is by #AC if 7120 both features are enabled in hardware. 7121 7122 ratelimit:N - 7123 Set system wide rate limit to N bus locks 7124 per second for bus lock detection. 7125 0 < N <= 1000. 7126 7127 N/A for split lock detection. 7128 7129 7130 If an #AC exception is hit in the kernel or in 7131 firmware (i.e. not while executing in user mode) 7132 the kernel will oops in either "warn" or "fatal" 7133 mode. 7134 7135 #DB exception for bus lock is triggered only when 7136 CPL > 0. 7137 7138 srbds= [X86,INTEL,EARLY] 7139 Control the Special Register Buffer Data Sampling 7140 (SRBDS) mitigation. 7141 7142 Certain CPUs are vulnerable to an MDS-like 7143 exploit which can leak bits from the random 7144 number generator. 7145 7146 By default, this issue is mitigated by 7147 microcode. However, the microcode fix can cause 7148 the RDRAND and RDSEED instructions to become 7149 much slower. Among other effects, this will 7150 result in reduced throughput from /dev/urandom. 7151 7152 The microcode mitigation can be disabled with 7153 the following option: 7154 7155 off: Disable mitigation and remove 7156 performance impact to RDRAND and RDSEED 7157 7158 srcutree.big_cpu_lim [KNL] 7159 Specifies the number of CPUs constituting a 7160 large system, such that srcu_struct structures 7161 should immediately allocate an srcu_node array. 7162 This kernel-boot parameter defaults to 128, 7163 but takes effect only when the low-order four 7164 bits of srcutree.convert_to_big is equal to 3 7165 (decide at boot). 7166 7167 srcutree.convert_to_big [KNL] 7168 Specifies under what conditions an SRCU tree 7169 srcu_struct structure will be converted to big 7170 form, that is, with an rcu_node tree: 7171 7172 0: Never. 7173 1: At init_srcu_struct() time. 7174 2: When rcutorture decides to. 7175 3: Decide at boot time (default). 7176 0x1X: Above plus if high contention. 7177 7178 Either way, the srcu_node tree will be sized based 7179 on the actual runtime number of CPUs (nr_cpu_ids) 7180 instead of the compile-time CONFIG_NR_CPUS. 7181 7182 srcutree.counter_wrap_check [KNL] 7183 Specifies how frequently to check for 7184 grace-period sequence counter wrap for the 7185 srcu_data structure's ->srcu_gp_seq_needed field. 7186 The greater the number of bits set in this kernel 7187 parameter, the less frequently counter wrap will 7188 be checked for. Note that the bottom two bits 7189 are ignored. 7190 7191 srcutree.exp_holdoff [KNL] 7192 Specifies how many nanoseconds must elapse 7193 since the end of the last SRCU grace period for 7194 a given srcu_struct until the next normal SRCU 7195 grace period will be considered for automatic 7196 expediting. Set to zero to disable automatic 7197 expediting. 7198 7199 srcutree.srcu_max_nodelay [KNL] 7200 Specifies the number of no-delay instances 7201 per jiffy for which the SRCU grace period 7202 worker thread will be rescheduled with zero 7203 delay. Beyond this limit, worker thread will 7204 be rescheduled with a sleep delay of one jiffy. 7205 7206 srcutree.srcu_max_nodelay_phase [KNL] 7207 Specifies the per-grace-period phase, number of 7208 non-sleeping polls of readers. Beyond this limit, 7209 grace period worker thread will be rescheduled 7210 with a sleep delay of one jiffy, between each 7211 rescan of the readers, for a grace period phase. 7212 7213 srcutree.srcu_retry_check_delay [KNL] 7214 Specifies number of microseconds of non-sleeping 7215 delay between each non-sleeping poll of readers. 7216 7217 srcutree.small_contention_lim [KNL] 7218 Specifies the number of update-side contention 7219 events per jiffy will be tolerated before 7220 initiating a conversion of an srcu_struct 7221 structure to big form. Note that the value of 7222 srcutree.convert_to_big must have the 0x10 bit 7223 set for contention-based conversions to occur. 7224 7225 ssbd= [ARM64,HW,EARLY] 7226 Speculative Store Bypass Disable control 7227 7228 On CPUs that are vulnerable to the Speculative 7229 Store Bypass vulnerability and offer a 7230 firmware based mitigation, this parameter 7231 indicates how the mitigation should be used: 7232 7233 force-on: Unconditionally enable mitigation for 7234 for both kernel and userspace 7235 force-off: Unconditionally disable mitigation for 7236 for both kernel and userspace 7237 kernel: Always enable mitigation in the 7238 kernel, and offer a prctl interface 7239 to allow userspace to register its 7240 interest in being mitigated too. 7241 7242 stack_guard_gap= [MM] 7243 override the default stack gap protection. The value 7244 is in page units and it defines how many pages prior 7245 to (for stacks growing down) resp. after (for stacks 7246 growing up) the main stack are reserved for no other 7247 mapping. Default value is 256 pages. 7248 7249 stack_depot_disable= [KNL,EARLY] 7250 Setting this to true through kernel command line will 7251 disable the stack depot thereby saving the static memory 7252 consumed by the stack hash table. By default this is set 7253 to false. 7254 7255 stack_depot_max_pools= [KNL,EARLY] 7256 Specify the maximum number of pools to use for storing 7257 stack traces. Pools are allocated on-demand up to this 7258 limit. Default value is 8191 pools. 7259 7260 stacktrace [FTRACE] 7261 Enable the stack tracer on boot up. 7262 7263 stacktrace_filter=[function-list] 7264 [FTRACE] Limit the functions that the stack tracer 7265 will trace at boot up. function-list is a comma-separated 7266 list of functions. This list can be changed at run 7267 time by the stack_trace_filter file in the debugfs 7268 tracing directory. Note, this enables stack tracing 7269 and the stacktrace above is not needed. 7270 7271 sti= [PARISC,HW] 7272 Format: <num> 7273 Set the STI (builtin display/keyboard on the HP-PARISC 7274 machines) console (graphic card) which should be used 7275 as the initial boot-console. 7276 See also comment in drivers/video/console/sticore.c. 7277 7278 sti_font= [HW] 7279 See comment in drivers/video/console/sticore.c. 7280 7281 stifb= [HW] 7282 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]] 7283 7284 strict_sas_size= 7285 [X86] 7286 Format: <bool> 7287 Enable or disable strict sigaltstack size checks 7288 against the required signal frame size which 7289 depends on the supported FPU features. This can 7290 be used to filter out binaries which have 7291 not yet been made aware of AT_MINSIGSTKSZ. 7292 7293 stress_hpt [PPC,EARLY] 7294 Limits the number of kernel HPT entries in the hash 7295 page table to increase the rate of hash page table 7296 faults on kernel addresses. 7297 7298 stress_slb [PPC,EARLY] 7299 Limits the number of kernel SLB entries, and flushes 7300 them frequently to increase the rate of SLB faults 7301 on kernel addresses. 7302 7303 no_slb_preload [PPC,EARLY] 7304 Disables slb preloading for userspace. 7305 7306 sunrpc.min_resvport= 7307 sunrpc.max_resvport= 7308 [NFS,SUNRPC] 7309 SunRPC servers often require that client requests 7310 originate from a privileged port (i.e. a port in the 7311 range 0 < portnr < 1024). 7312 An administrator who wishes to reserve some of these 7313 ports for other uses may adjust the range that the 7314 kernel's sunrpc client considers to be privileged 7315 using these two parameters to set the minimum and 7316 maximum port values. 7317 7318 sunrpc.svc_rpc_per_connection_limit= 7319 [NFS,SUNRPC] 7320 Limit the number of requests that the server will 7321 process in parallel from a single connection. 7322 The default value is 0 (no limit). 7323 7324 sunrpc.pool_mode= 7325 [NFS] 7326 Control how the NFS server code allocates CPUs to 7327 service thread pools. Depending on how many NICs 7328 you have and where their interrupts are bound, this 7329 option will affect which CPUs will do NFS serving. 7330 Note: this parameter cannot be changed while the 7331 NFS server is running. 7332 7333 auto the server chooses an appropriate mode 7334 automatically using heuristics 7335 global a single global pool contains all CPUs 7336 percpu one pool for each CPU 7337 pernode one pool for each NUMA node (equivalent 7338 to global on non-NUMA machines) 7339 7340 sunrpc.tcp_slot_table_entries= 7341 sunrpc.udp_slot_table_entries= 7342 [NFS,SUNRPC] 7343 Sets the upper limit on the number of simultaneous 7344 RPC calls that can be sent from the client to a 7345 server. Increasing these values may allow you to 7346 improve throughput, but will also increase the 7347 amount of memory reserved for use by the client. 7348 7349 suspend.pm_test_delay= 7350 [SUSPEND] 7351 Sets the number of seconds to remain in a suspend test 7352 mode before resuming the system (see 7353 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG 7354 is set. Default value is 5. 7355 7356 svm= [PPC] 7357 Format: { on | off | y | n | 1 | 0 } 7358 This parameter controls use of the Protected 7359 Execution Facility on pSeries. 7360 7361 swiotlb= [ARM,PPC,MIPS,X86,S390,EARLY] 7362 Format: { <int> [,<int>] | force | noforce } 7363 <int> -- Number of I/O TLB slabs 7364 <int> -- Second integer after comma. Number of swiotlb 7365 areas with their own lock. Will be rounded up 7366 to a power of 2. 7367 force -- force using of bounce buffers even if they 7368 wouldn't be automatically used by the kernel 7369 noforce -- Never use bounce buffers (for debugging) 7370 7371 switches= [HW,M68k,EARLY] 7372 7373 sysctl.*= [KNL] 7374 Set a sysctl parameter, right before loading the init 7375 process, as if the value was written to the respective 7376 /proc/sys/... file. Both '.' and '/' are recognized as 7377 separators. Unrecognized parameters and invalid values 7378 are reported in the kernel log. Sysctls registered 7379 later by a loaded module cannot be set this way. 7380 Example: sysctl.vm.swappiness=40 7381 7382 sysrq_always_enabled 7383 [KNL] 7384 Ignore sysrq setting - this boot parameter will 7385 neutralize any effect of /proc/sys/kernel/sysrq. 7386 Useful for debugging. 7387 7388 tcpmhash_entries= [KNL,NET] 7389 Set the number of tcp_metrics_hash slots. 7390 Default value is 8192 or 16384 depending on total 7391 ram pages. This is used to specify the TCP metrics 7392 cache size. See Documentation/networking/ip-sysctl.rst 7393 "tcp_no_metrics_save" section for more details. 7394 7395 tdfx= [HW,DRM] 7396 7397 test_suspend= [SUSPEND] 7398 Format: { "mem" | "standby" | "freeze" }[,N] 7399 Specify "mem" (for Suspend-to-RAM) or "standby" (for 7400 standby suspend) or "freeze" (for suspend type freeze) 7401 as the system sleep state during system startup with 7402 the optional capability to repeat N number of times. 7403 The system is woken from this state using a 7404 wakeup-capable RTC alarm. 7405 7406 thash_entries= [KNL,NET] 7407 Set number of hash buckets for TCP connection 7408 7409 thermal.act= [HW,ACPI] 7410 -1: disable all active trip points in all thermal zones 7411 <degrees C>: override all lowest active trip points 7412 7413 thermal.crt= [HW,ACPI] 7414 -1: disable all critical trip points in all thermal zones 7415 <degrees C>: override all critical trip points 7416 7417 thermal.off= [HW,ACPI] 7418 1: disable ACPI thermal control 7419 7420 thermal.psv= [HW,ACPI] 7421 -1: disable all passive trip points 7422 <degrees C>: override all passive trip points to this 7423 value 7424 7425 thermal.tzp= [HW,ACPI] 7426 Specify global default ACPI thermal zone polling rate 7427 <deci-seconds>: poll all this frequency 7428 0: no polling (default) 7429 7430 thp_anon= [KNL] 7431 Format: <size>[KMG],<size>[KMG]:<state>;<size>[KMG]-<size>[KMG]:<state> 7432 state is one of "always", "madvise", "never" or "inherit". 7433 Control the default behavior of the system with respect 7434 to anonymous transparent hugepages. 7435 Can be used multiple times for multiple anon THP sizes. 7436 See Documentation/admin-guide/mm/transhuge.rst for more 7437 details. 7438 7439 threadirqs [KNL,EARLY] 7440 Force threading of all interrupt handlers except those 7441 marked explicitly IRQF_NO_THREAD. 7442 7443 thp_shmem= [KNL] 7444 Format: <size>[KMG],<size>[KMG]:<policy>;<size>[KMG]-<size>[KMG]:<policy> 7445 Control the default policy of each hugepage size for the 7446 internal shmem mount. <policy> is one of policies available 7447 for the shmem mount ("always", "inherit", "never", "within_size", 7448 and "advise"). 7449 It can be used multiple times for multiple shmem THP sizes. 7450 See Documentation/admin-guide/mm/transhuge.rst for more 7451 details. 7452 7453 topology= [S390,EARLY] 7454 Format: {off | on} 7455 Specify if the kernel should make use of the cpu 7456 topology information if the hardware supports this. 7457 The scheduler will make use of this information and 7458 e.g. base its process migration decisions on it. 7459 Default is on. 7460 7461 torture.disable_onoff_at_boot= [KNL] 7462 Prevent the CPU-hotplug component of torturing 7463 until after init has spawned. 7464 7465 torture.ftrace_dump_at_shutdown= [KNL] 7466 Dump the ftrace buffer at torture-test shutdown, 7467 even if there were no errors. This can be a 7468 very costly operation when many torture tests 7469 are running concurrently, especially on systems 7470 with rotating-rust storage. 7471 7472 torture.verbose_sleep_frequency= [KNL] 7473 Specifies how many verbose printk()s should be 7474 emitted between each sleep. The default of zero 7475 disables verbose-printk() sleeping. 7476 7477 torture.verbose_sleep_duration= [KNL] 7478 Duration of each verbose-printk() sleep in jiffies. 7479 7480 tpm.disable_pcr_integrity= [HW,TPM] 7481 Do not protect PCR registers from unintended physical 7482 access, or interposers in the bus by the means of 7483 having an integrity protected session wrapped around 7484 TPM2_PCR_Extend command. Consider this in a situation 7485 where TPM is heavily utilized by IMA, thus protection 7486 causing a major performance hit, and the space where 7487 machines are deployed is by other means guarded. 7488 7489 tpm_crb_ffa.busy_timeout_ms= [ARM64,TPM] 7490 Maximum time in milliseconds to retry sending a message 7491 to the TPM service before giving up. This parameter controls 7492 how long the system will continue retrying when the TPM 7493 service is busy. 7494 Format: <unsigned int> 7495 Default: 2000 (2 seconds) 7496 7497 tpm_suspend_pcr=[HW,TPM] 7498 Format: integer pcr id 7499 Specify that at suspend time, the tpm driver 7500 should extend the specified pcr with zeros, 7501 as a workaround for some chips which fail to 7502 flush the last written pcr on TPM_SaveState. 7503 This will guarantee that all the other pcrs 7504 are saved. 7505 7506 tpm_tis.interrupts= [HW,TPM] 7507 Enable interrupts for the MMIO based physical layer 7508 for the FIFO interface. By default it is set to false 7509 (0). For more information about TPM hardware interfaces 7510 defined by Trusted Computing Group (TCG) see 7511 https://trustedcomputinggroup.org/resource/pc-client-platform-tpm-profile-ptp-specification/ 7512 7513 tp_printk [FTRACE] 7514 Have the tracepoints sent to printk as well as the 7515 tracing ring buffer. This is useful for early boot up 7516 where the system hangs or reboots and does not give the 7517 option for reading the tracing buffer or performing a 7518 ftrace_dump_on_oops. 7519 7520 To turn off having tracepoints sent to printk, 7521 echo 0 > /proc/sys/kernel/tracepoint_printk 7522 Note, echoing 1 into this file without the 7523 tp_printk kernel cmdline option has no effect. 7524 7525 The tp_printk_stop_on_boot (see below) can also be used 7526 to stop the printing of events to console at 7527 late_initcall_sync. 7528 7529 ** CAUTION ** 7530 7531 Having tracepoints sent to printk() and activating high 7532 frequency tracepoints such as irq or sched, can cause 7533 the system to live lock. 7534 7535 tp_printk_stop_on_boot [FTRACE] 7536 When tp_printk (above) is set, it can cause a lot of noise 7537 on the console. It may be useful to only include the 7538 printing of events during boot up, as user space may 7539 make the system inoperable. 7540 7541 This command line option will stop the printing of events 7542 to console at the late_initcall_sync() time frame. 7543 7544 trace_buf_size=nn[KMG] 7545 [FTRACE] will set tracing buffer size on each cpu. 7546 7547 trace_clock= [FTRACE] Set the clock used for tracing events 7548 at boot up. 7549 local - Use the per CPU time stamp counter 7550 (converted into nanoseconds). Fast, but 7551 depending on the architecture, may not be 7552 in sync between CPUs. 7553 global - Event time stamps are synchronized across 7554 CPUs. May be slower than the local clock, 7555 but better for some race conditions. 7556 counter - Simple counting of events (1, 2, ..) 7557 note, some counts may be skipped due to the 7558 infrastructure grabbing the clock more than 7559 once per event. 7560 uptime - Use jiffies as the time stamp. 7561 perf - Use the same clock that perf uses. 7562 mono - Use ktime_get_mono_fast_ns() for time stamps. 7563 mono_raw - Use ktime_get_raw_fast_ns() for time 7564 stamps. 7565 boot - Use ktime_get_boot_fast_ns() for time stamps. 7566 Architectures may add more clocks. See 7567 Documentation/trace/ftrace.rst for more details. 7568 7569 trace_event=[event-list] 7570 [FTRACE] Set and start specified trace events in order 7571 to facilitate early boot debugging. The event-list is a 7572 comma-separated list of trace events to enable. See 7573 also Documentation/trace/events.rst 7574 7575 To enable modules, use :mod: keyword: 7576 7577 trace_event=:mod:<module> 7578 7579 The value before :mod: will only enable specific events 7580 that are part of the module. See the above mentioned 7581 document for more information. 7582 7583 trace_instance=[instance-info] 7584 [FTRACE] Create a ring buffer instance early in boot up. 7585 This will be listed in: 7586 7587 /sys/kernel/tracing/instances 7588 7589 Events can be enabled at the time the instance is created 7590 via: 7591 7592 trace_instance=<name>,<system1>:<event1>,<system2>:<event2> 7593 7594 Note, the "<system*>:" portion is optional if the event is 7595 unique. 7596 7597 trace_instance=foo,sched:sched_switch,irq_handler_entry,initcall 7598 7599 will enable the "sched_switch" event (note, the "sched:" is optional, and 7600 the same thing would happen if it was left off). The irq_handler_entry 7601 event, and all events under the "initcall" system. 7602 7603 Flags can be added to the instance to modify its behavior when it is 7604 created. The flags are separated by '^'. 7605 7606 The available flags are: 7607 7608 traceoff - Have the tracing instance tracing disabled after it is created. 7609 traceprintk - Have trace_printk() write into this trace instance 7610 (note, "printk" and "trace_printk" can also be used) 7611 7612 trace_instance=foo^traceoff^traceprintk,sched,irq 7613 7614 The flags must come before the defined events. 7615 7616 If memory has been reserved (see memmap for x86), the instance 7617 can use that memory: 7618 7619 memmap=12M$0x284500000 trace_instance=boot_map@0x284500000:12M 7620 7621 The above will create a "boot_map" instance that uses the physical 7622 memory at 0x284500000 that is 12Megs. The per CPU buffers of that 7623 instance will be split up accordingly. 7624 7625 Alternatively, the memory can be reserved by the reserve_mem option: 7626 7627 reserve_mem=12M:4096:trace trace_instance=boot_map@trace 7628 7629 This will reserve 12 megabytes at boot up with a 4096 byte alignment 7630 and place the ring buffer in this memory. Note that due to KASLR, the 7631 memory may not be the same location each time, which will not preserve 7632 the buffer content. 7633 7634 Also note that the layout of the ring buffer data may change between 7635 kernel versions where the validator will fail and reset the ring buffer 7636 if the layout is not the same as the previous kernel. 7637 7638 If the ring buffer is used for persistent bootups and has events enabled, 7639 it is recommend to disable tracing so that events from a previous boot do not 7640 mix with events of the current boot (unless you are debugging a random crash 7641 at boot up). 7642 7643 reserve_mem=12M:4096:trace trace_instance=boot_map^traceoff^traceprintk@trace,sched,irq 7644 7645 Note, saving the trace buffer across reboots does require that the system 7646 is set up to not wipe memory. For instance, CONFIG_RESET_ATTACK_MITIGATION 7647 can force a memory reset on boot which will clear any trace that was stored. 7648 This is just one of many ways that can clear memory. Make sure your system 7649 keeps the content of memory across reboots before relying on this option. 7650 7651 NB: Both the mapped address and size must be page aligned for the architecture. 7652 7653 See also Documentation/trace/debugging.rst 7654 7655 7656 trace_options=[option-list] 7657 [FTRACE] Enable or disable tracer options at boot. 7658 The option-list is a comma delimited list of options 7659 that can be enabled or disabled just as if you were 7660 to echo the option name into 7661 7662 /sys/kernel/tracing/trace_options 7663 7664 For example, to enable stacktrace option (to dump the 7665 stack trace of each event), add to the command line: 7666 7667 trace_options=stacktrace 7668 7669 See also Documentation/trace/ftrace.rst "trace options" 7670 section. 7671 7672 trace_trigger=[trigger-list] 7673 [FTRACE] Add an event trigger on specific events. 7674 Set a trigger on top of a specific event, with an optional 7675 filter. 7676 7677 The format is "trace_trigger=<event>.<trigger>[ if <filter>],..." 7678 Where more than one trigger may be specified that are comma delimited. 7679 7680 For example: 7681 7682 trace_trigger="sched_switch.stacktrace if prev_state == 2" 7683 7684 The above will enable the "stacktrace" trigger on the "sched_switch" 7685 event but only trigger it if the "prev_state" of the "sched_switch" 7686 event is "2" (TASK_UNINTERRUPTIBLE). 7687 7688 See also "Event triggers" in Documentation/trace/events.rst 7689 7690 7691 traceoff_after_boot 7692 [FTRACE] Sometimes tracing is used to debug issues 7693 during the boot process. Since the trace buffer has a 7694 limited amount of storage, it may be prudent to 7695 disable tracing after the boot is finished, otherwise 7696 the critical information may be overwritten. With this 7697 option, the main tracing buffer will be turned off at 7698 the end of the boot process. 7699 7700 traceoff_on_warning 7701 [FTRACE] enable this option to disable tracing when a 7702 warning is hit. This turns off "tracing_on". Tracing can 7703 be enabled again by echoing '1' into the "tracing_on" 7704 file located in /sys/kernel/tracing/ 7705 7706 This option is useful, as it disables the trace before 7707 the WARNING dump is called, which prevents the trace to 7708 be filled with content caused by the warning output. 7709 7710 This option can also be set at run time via the sysctl 7711 option: kernel/traceoff_on_warning 7712 7713 transparent_hugepage= 7714 [KNL] 7715 Format: [always|madvise|never] 7716 Can be used to control the default behavior of the system 7717 with respect to transparent hugepages. 7718 See Documentation/admin-guide/mm/transhuge.rst 7719 for more details. 7720 7721 transparent_hugepage_shmem= [KNL] 7722 Format: [always|within_size|advise|never|deny|force] 7723 Can be used to control the hugepage allocation policy for 7724 the internal shmem mount. 7725 See Documentation/admin-guide/mm/transhuge.rst 7726 for more details. 7727 7728 transparent_hugepage_tmpfs= [KNL] 7729 Format: [always|within_size|advise|never] 7730 Can be used to control the default hugepage allocation policy 7731 for the tmpfs mount. 7732 See Documentation/admin-guide/mm/transhuge.rst 7733 for more details. 7734 7735 trusted.source= [KEYS] 7736 Format: <string> 7737 This parameter identifies the trust source as a backend 7738 for trusted keys implementation. Supported trust 7739 sources: 7740 - "tpm" 7741 - "tee" 7742 - "caam" 7743 - "dcp" 7744 If not specified then it defaults to iterating through 7745 the trust source list starting with TPM and assigns the 7746 first trust source as a backend which is initialized 7747 successfully during iteration. 7748 7749 trusted.rng= [KEYS] 7750 Format: <string> 7751 The RNG used to generate key material for trusted keys. 7752 Can be one of: 7753 - "kernel" 7754 - the same value as trusted.source: "tpm" or "tee" 7755 - "default" 7756 If not specified, "default" is used. In this case, 7757 the RNG's choice is left to each individual trust source. 7758 7759 trusted.dcp_use_otp_key 7760 This is intended to be used in combination with 7761 trusted.source=dcp and will select the DCP OTP key 7762 instead of the DCP UNIQUE key blob encryption. 7763 7764 trusted.dcp_skip_zk_test 7765 This is intended to be used in combination with 7766 trusted.source=dcp and will disable the check if the 7767 blob key is all zeros. This is helpful for situations where 7768 having this key zero'ed is acceptable. E.g. in testing 7769 scenarios. 7770 7771 tsa= [X86] Control mitigation for Transient Scheduler 7772 Attacks on AMD CPUs. Search the following in your 7773 favourite search engine for more details: 7774 7775 "Technical guidance for mitigating transient scheduler 7776 attacks". 7777 7778 off - disable the mitigation 7779 on - enable the mitigation (default) 7780 user - mitigate only user/kernel transitions 7781 vm - mitigate only guest/host transitions 7782 7783 7784 tsc= Disable clocksource stability checks for TSC. 7785 Format: <string> 7786 [x86] reliable: mark tsc clocksource as reliable, this 7787 disables clocksource verification at runtime, as well 7788 as the stability checks done at bootup. Used to enable 7789 high-resolution timer mode on older hardware, and in 7790 virtualized environment. 7791 [x86] noirqtime: Do not use TSC to do irq accounting. 7792 Used to run time disable IRQ_TIME_ACCOUNTING on any 7793 platforms where RDTSC is slow and this accounting 7794 can add overhead. 7795 [x86] unstable: mark the TSC clocksource as unstable, this 7796 marks the TSC unconditionally unstable at bootup and 7797 avoids any further wobbles once the TSC watchdog notices. 7798 [x86] nowatchdog: disable clocksource watchdog. Used 7799 in situations with strict latency requirements (where 7800 interruptions from clocksource watchdog are not 7801 acceptable). 7802 [x86] recalibrate: force recalibration against a HW timer 7803 (HPET or PM timer) on systems whose TSC frequency was 7804 obtained from HW or FW using either an MSR or CPUID(0x15). 7805 Warn if the difference is more than 500 ppm. 7806 [x86] watchdog: Use TSC as the watchdog clocksource with 7807 which to check other HW timers (HPET or PM timer), but 7808 only on systems where TSC has been deemed trustworthy. 7809 This will be suppressed by an earlier tsc=nowatchdog and 7810 can be overridden by a later tsc=nowatchdog. A console 7811 message will flag any such suppression or overriding. 7812 7813 tsc_early_khz= [X86,EARLY] Skip early TSC calibration and use the given 7814 value instead. Useful when the early TSC frequency discovery 7815 procedure is not reliable, such as on overclocked systems 7816 with CPUID.16h support and partial CPUID.15h support. 7817 Format: <unsigned int> 7818 7819 tsx= [X86] Control Transactional Synchronization 7820 Extensions (TSX) feature in Intel processors that 7821 support TSX control. 7822 7823 This parameter controls the TSX feature. The options are: 7824 7825 on - Enable TSX on the system. Although there are 7826 mitigations for all known security vulnerabilities, 7827 TSX has been known to be an accelerator for 7828 several previous speculation-related CVEs, and 7829 so there may be unknown security risks associated 7830 with leaving it enabled. 7831 7832 off - Disable TSX on the system. (Note that this 7833 option takes effect only on newer CPUs which are 7834 not vulnerable to MDS, i.e., have 7835 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get 7836 the new IA32_TSX_CTRL MSR through a microcode 7837 update. This new MSR allows for the reliable 7838 deactivation of the TSX functionality.) 7839 7840 auto - Disable TSX if X86_BUG_TAA is present, 7841 otherwise enable TSX on the system. 7842 7843 Not specifying this option is equivalent to tsx=off. 7844 7845 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst 7846 for more details. 7847 7848 tsx_async_abort= [X86,INTEL,EARLY] Control mitigation for the TSX Async 7849 Abort (TAA) vulnerability. 7850 7851 Similar to Micro-architectural Data Sampling (MDS) 7852 certain CPUs that support Transactional 7853 Synchronization Extensions (TSX) are vulnerable to an 7854 exploit against CPU internal buffers which can forward 7855 information to a disclosure gadget under certain 7856 conditions. 7857 7858 In vulnerable processors, the speculatively forwarded 7859 data can be used in a cache side channel attack, to 7860 access data to which the attacker does not have direct 7861 access. 7862 7863 This parameter controls the TAA mitigation. The 7864 options are: 7865 7866 full - Enable TAA mitigation on vulnerable CPUs 7867 if TSX is enabled. 7868 7869 full,nosmt - Enable TAA mitigation and disable SMT on 7870 vulnerable CPUs. If TSX is disabled, SMT 7871 is not disabled because CPU is not 7872 vulnerable to cross-thread TAA attacks. 7873 off - Unconditionally disable TAA mitigation 7874 7875 On MDS-affected machines, tsx_async_abort=off can be 7876 prevented by an active MDS mitigation as both vulnerabilities 7877 are mitigated with the same mechanism so in order to disable 7878 this mitigation, you need to specify mds=off too. 7879 7880 Not specifying this option is equivalent to 7881 tsx_async_abort=full. On CPUs which are MDS affected 7882 and deploy MDS mitigation, TAA mitigation is not 7883 required and doesn't provide any additional 7884 mitigation. 7885 7886 For details see: 7887 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst 7888 7889 turbografx.map[2|3]= [HW,JOY] 7890 TurboGraFX parallel port interface 7891 Format: 7892 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7> 7893 See also Documentation/input/devices/joystick-parport.rst 7894 7895 udbg-immortal [PPC] When debugging early kernel crashes that 7896 happen after console_init() and before a proper 7897 console driver takes over, this boot options might 7898 help "seeing" what's going on. 7899 7900 uhash_entries= [KNL,NET] 7901 Set number of hash buckets for UDP/UDP-Lite connections 7902 7903 uhci-hcd.ignore_oc= 7904 [USB] Ignore overcurrent events (default N). 7905 Some badly-designed motherboards generate lots of 7906 bogus events, for ports that aren't wired to 7907 anything. Set this parameter to avoid log spamming. 7908 Note that genuine overcurrent events won't be 7909 reported either. 7910 7911 unaligned_scalar_speed= 7912 [RISCV] 7913 Format: {slow | fast | unsupported} 7914 Allow skipping scalar unaligned access speed tests. This 7915 is useful for testing alternative code paths and to skip 7916 the tests in environments where they run too slowly. All 7917 CPUs must have the same scalar unaligned access speed. 7918 7919 unaligned_vector_speed= 7920 [RISCV] 7921 Format: {slow | fast | unsupported} 7922 Allow skipping vector unaligned access speed tests. This 7923 is useful for testing alternative code paths and to skip 7924 the tests in environments where they run too slowly. All 7925 CPUs must have the same vector unaligned access speed. 7926 7927 unknown_nmi_panic 7928 [X86] Cause panic on unknown NMI. 7929 7930 unwind_debug [X86-64,EARLY] 7931 Enable unwinder debug output. This can be 7932 useful for debugging certain unwinder error 7933 conditions, including corrupt stacks and 7934 bad/missing unwinder metadata. 7935 7936 usbcore.authorized_default= 7937 [USB] Default USB device authorization: 7938 (default -1 = authorized (same as 1), 7939 0 = not authorized, 1 = authorized, 2 = authorized 7940 if device connected to internal port) 7941 7942 usbcore.autosuspend= 7943 [USB] The autosuspend time delay (in seconds) used 7944 for newly-detected USB devices (default 2). This 7945 is the time required before an idle device will be 7946 autosuspended. Devices for which the delay is set 7947 to a negative value won't be autosuspended at all. 7948 7949 usbcore.usbfs_snoop= 7950 [USB] Set to log all usbfs traffic (default 0 = off). 7951 7952 usbcore.usbfs_snoop_max= 7953 [USB] Maximum number of bytes to snoop in each URB 7954 (default = 65536). 7955 7956 usbcore.blinkenlights= 7957 [USB] Set to cycle leds on hubs (default 0 = off). 7958 7959 usbcore.old_scheme_first= 7960 [USB] Start with the old device initialization 7961 scheme (default 0 = off). 7962 7963 usbcore.usbfs_memory_mb= 7964 [USB] Memory limit (in MB) for buffers allocated by 7965 usbfs (default = 16, 0 = max = 2047). 7966 7967 usbcore.use_both_schemes= 7968 [USB] Try the other device initialization scheme 7969 if the first one fails (default 1 = enabled). 7970 7971 usbcore.initial_descriptor_timeout= 7972 [USB] Specifies timeout for the initial 64-byte 7973 USB_REQ_GET_DESCRIPTOR request in milliseconds 7974 (default 5000 = 5.0 seconds). 7975 7976 usbcore.nousb [USB] Disable the USB subsystem 7977 7978 usbcore.quirks= 7979 [USB] A list of quirk entries to augment the built-in 7980 usb core quirk list. List entries are separated by 7981 commas. Each entry has the form 7982 VendorID:ProductID:Flags. The IDs are 4-digit hex 7983 numbers and Flags is a set of letters. Each letter 7984 will change the built-in quirk; setting it if it is 7985 clear and clearing it if it is set. The letters have 7986 the following meanings: 7987 a = USB_QUIRK_STRING_FETCH_255 (string 7988 descriptors must not be fetched using 7989 a 255-byte read); 7990 b = USB_QUIRK_RESET_RESUME (device can't resume 7991 correctly so reset it instead); 7992 c = USB_QUIRK_NO_SET_INTF (device can't handle 7993 Set-Interface requests); 7994 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't 7995 handle its Configuration or Interface 7996 strings); 7997 e = USB_QUIRK_RESET (device can't be reset 7998 (e.g morph devices), don't use reset); 7999 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has 8000 more interface descriptions than the 8001 bNumInterfaces count, and can't handle 8002 talking to these interfaces); 8003 g = USB_QUIRK_DELAY_INIT (device needs a pause 8004 during initialization, after we read 8005 the device descriptor); 8006 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For 8007 high speed and super speed interrupt 8008 endpoints, the USB 2.0 and USB 3.0 spec 8009 require the interval in microframes (1 8010 microframe = 125 microseconds) to be 8011 calculated as interval = 2 ^ 8012 (bInterval-1). 8013 Devices with this quirk report their 8014 bInterval as the result of this 8015 calculation instead of the exponent 8016 variable used in the calculation); 8017 i = USB_QUIRK_DEVICE_QUALIFIER (device can't 8018 handle device_qualifier descriptor 8019 requests); 8020 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device 8021 generates spurious wakeup, ignore 8022 remote wakeup capability); 8023 k = USB_QUIRK_NO_LPM (device can't handle Link 8024 Power Management); 8025 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL 8026 (Device reports its bInterval as linear 8027 frames instead of the USB 2.0 8028 calculation); 8029 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs 8030 to be disconnected before suspend to 8031 prevent spurious wakeup); 8032 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a 8033 pause after every control message); 8034 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra 8035 delay after resetting its port); 8036 p = USB_QUIRK_SHORT_SET_ADDRESS_REQ_TIMEOUT 8037 (Reduce timeout of the SET_ADDRESS 8038 request from 5000 ms to 500 ms); 8039 Example: quirks=0781:5580:bk,0a5c:5834:gij 8040 8041 usbhid.mousepoll= 8042 [USBHID] The interval which mice are to be polled at. 8043 8044 usbhid.jspoll= 8045 [USBHID] The interval which joysticks are to be polled at. 8046 8047 usbhid.kbpoll= 8048 [USBHID] The interval which keyboards are to be polled at. 8049 8050 usb-storage.delay_use= 8051 [UMS] The delay in seconds before a new device is 8052 scanned for Logical Units (default 1). 8053 Optionally the delay in milliseconds if the value has 8054 suffix with "ms". 8055 Example: delay_use=2567ms 8056 8057 usb-storage.quirks= 8058 [UMS] A list of quirks entries to supplement or 8059 override the built-in unusual_devs list. List 8060 entries are separated by commas. Each entry has 8061 the form VID:PID:Flags where VID and PID are Vendor 8062 and Product ID values (4-digit hex numbers) and 8063 Flags is a set of characters, each corresponding 8064 to a common usb-storage quirk flag as follows: 8065 a = SANE_SENSE (collect more than 18 bytes 8066 of sense data, not on uas); 8067 b = BAD_SENSE (don't collect more than 18 8068 bytes of sense data, not on uas); 8069 c = FIX_CAPACITY (decrease the reported 8070 device capacity by one sector); 8071 d = NO_READ_DISC_INFO (don't use 8072 READ_DISC_INFO command, not on uas); 8073 e = NO_READ_CAPACITY_16 (don't use 8074 READ_CAPACITY_16 command); 8075 f = NO_REPORT_OPCODES (don't use report opcodes 8076 command, uas only); 8077 g = MAX_SECTORS_240 (don't transfer more than 8078 240 sectors at a time, uas only); 8079 h = CAPACITY_HEURISTICS (decrease the 8080 reported device capacity by one 8081 sector if the number is odd); 8082 i = IGNORE_DEVICE (don't bind to this 8083 device); 8084 j = NO_REPORT_LUNS (don't use report luns 8085 command, uas only); 8086 k = NO_SAME (do not use WRITE_SAME, uas only) 8087 l = NOT_LOCKABLE (don't try to lock and 8088 unlock ejectable media, not on uas); 8089 m = MAX_SECTORS_64 (don't transfer more 8090 than 64 sectors = 32 KB at a time, 8091 not on uas); 8092 n = INITIAL_READ10 (force a retry of the 8093 initial READ(10) command, not on uas); 8094 o = CAPACITY_OK (accept the capacity 8095 reported by the device, not on uas); 8096 p = WRITE_CACHE (the device cache is ON 8097 by default, not on uas); 8098 r = IGNORE_RESIDUE (the device reports 8099 bogus residue values, not on uas); 8100 s = SINGLE_LUN (the device has only one 8101 Logical Unit); 8102 t = NO_ATA_1X (don't allow ATA(12) and ATA(16) 8103 commands, uas only); 8104 u = IGNORE_UAS (don't bind to the uas driver); 8105 w = NO_WP_DETECT (don't test whether the 8106 medium is write-protected). 8107 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE 8108 even if the device claims no cache, 8109 not on uas) 8110 Example: quirks=0419:aaf5:rl,0421:0433:rc 8111 8112 user_debug= [KNL,ARM] 8113 Format: <int> 8114 See arch/arm/Kconfig.debug help text. 8115 1 - undefined instruction events 8116 2 - system calls 8117 4 - invalid data aborts 8118 8 - SIGSEGV faults 8119 16 - SIGBUS faults 8120 Example: user_debug=31 8121 8122 vdso= [X86,SH,SPARC] 8123 On X86_32, this is an alias for vdso32=. Otherwise: 8124 8125 vdso=1: enable VDSO (the default) 8126 vdso=0: disable VDSO mapping 8127 8128 vdso32= [X86] Control the 32-bit vDSO 8129 vdso32=1: enable 32-bit VDSO 8130 vdso32=0 or vdso32=2: disable 32-bit VDSO 8131 8132 See the help text for CONFIG_COMPAT_VDSO for more 8133 details. If CONFIG_COMPAT_VDSO is set, the default is 8134 vdso32=0; otherwise, the default is vdso32=1. 8135 8136 For compatibility with older kernels, vdso32=2 is an 8137 alias for vdso32=0. 8138 8139 Try vdso32=0 if you encounter an error that says: 8140 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed! 8141 8142 video= [FB,EARLY] Frame buffer configuration 8143 See Documentation/fb/modedb.rst. 8144 8145 video.brightness_switch_enabled= [ACPI] 8146 Format: [0|1] 8147 If set to 1, on receiving an ACPI notify event 8148 generated by hotkey, video driver will adjust brightness 8149 level and then send out the event to user space through 8150 the allocated input device. If set to 0, video driver 8151 will only send out the event without touching backlight 8152 brightness level. 8153 default: 1 8154 8155 virtio_mmio.device= 8156 [VMMIO] Memory mapped virtio (platform) device. 8157 8158 <size>@<baseaddr>:<irq>[:<id>] 8159 where: 8160 <size> := size (can use standard suffixes 8161 like K, M and G) 8162 <baseaddr> := physical base address 8163 <irq> := interrupt number (as passed to 8164 request_irq()) 8165 <id> := (optional) platform device id 8166 example: 8167 virtio_mmio.device=1K@0x100b0000:48:7 8168 8169 Can be used multiple times for multiple devices. 8170 8171 vga= [BOOT,X86-32] Select a particular video mode 8172 See Documentation/arch/x86/boot.rst and 8173 Documentation/admin-guide/svga.rst. 8174 Use vga=ask for menu. 8175 This is actually a boot loader parameter; the value is 8176 passed to the kernel using a special protocol. 8177 8178 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y. 8179 May slow down system boot speed, especially when 8180 enabled on systems with a large amount of memory. 8181 All options are enabled by default, and this 8182 interface is meant to allow for selectively 8183 enabling or disabling specific virtual memory 8184 debugging features. 8185 8186 Available options are: 8187 P Enable page structure init time poisoning 8188 - Disable all of the above options 8189 8190 vmalloc=nn[KMG] [KNL,BOOT,EARLY] Forces the vmalloc area to have an 8191 exact size of <nn>. This can be used to increase 8192 the minimum size (128MB on x86, arm32 platforms). 8193 It can also be used to decrease the size and leave more room 8194 for directly mapped kernel RAM. Note that this parameter does 8195 not exist on many other platforms (including arm64, alpha, 8196 loongarch, arc, csky, hexagon, microblaze, mips, nios2, openrisc, 8197 parisc, m64k, powerpc, riscv, sh, um, xtensa, s390, sparc). 8198 8199 vmcp_cma=nn[MG] [KNL,S390,EARLY] 8200 Sets the memory size reserved for contiguous memory 8201 allocations for the vmcp device driver. 8202 8203 vmhalt= [KNL,S390] Perform z/VM CP command after system halt. 8204 Format: <command> 8205 8206 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic. 8207 Format: <command> 8208 8209 vmpoff= [KNL,S390] Perform z/VM CP command after power off. 8210 Format: <command> 8211 8212 vmscape= [X86] Controls mitigation for VMscape attacks. 8213 VMscape attacks can leak information from a userspace 8214 hypervisor to a guest via speculative side-channels. 8215 8216 off - disable the mitigation 8217 ibpb - use Indirect Branch Prediction Barrier 8218 (IBPB) mitigation (default) 8219 force - force vulnerability detection even on 8220 unaffected processors 8221 8222 vsyscall= [X86-64,EARLY] 8223 Controls the behavior of vsyscalls (i.e. calls to 8224 fixed addresses of 0xffffffffff600x00 from legacy 8225 code). Most statically-linked binaries and older 8226 versions of glibc use these calls. Because these 8227 functions are at fixed addresses, they make nice 8228 targets for exploits that can control RIP. 8229 8230 emulate Vsyscalls turn into traps and are emulated 8231 reasonably safely. The vsyscall page is 8232 readable. 8233 8234 xonly [default] Vsyscalls turn into traps and are 8235 emulated reasonably safely. The vsyscall 8236 page is not readable. 8237 8238 none Vsyscalls don't work at all. This makes 8239 them quite hard to use for exploits but 8240 might break your system. 8241 8242 vt.color= [VT] Default text color. 8243 Format: 0xYX, X = foreground, Y = background. 8244 Default: 0x07 = light gray on black. 8245 8246 vt.cur_default= [VT] Default cursor shape. 8247 Format: 0xCCBBAA, where AA, BB, and CC are the same as 8248 the parameters of the <Esc>[?A;B;Cc escape sequence; 8249 see vga-softcursor.rst. Default: 2 = underline. 8250 8251 vt.default_blu= [VT] 8252 Format: <blue0>,<blue1>,<blue2>,...,<blue15> 8253 Change the default blue palette of the console. 8254 This is a 16-member array composed of values 8255 ranging from 0-255. 8256 8257 vt.default_grn= [VT] 8258 Format: <green0>,<green1>,<green2>,...,<green15> 8259 Change the default green palette of the console. 8260 This is a 16-member array composed of values 8261 ranging from 0-255. 8262 8263 vt.default_red= [VT] 8264 Format: <red0>,<red1>,<red2>,...,<red15> 8265 Change the default red palette of the console. 8266 This is a 16-member array composed of values 8267 ranging from 0-255. 8268 8269 vt.default_utf8= 8270 [VT] 8271 Format=<0|1> 8272 Set system-wide default UTF-8 mode for all tty's. 8273 Default is 1, i.e. UTF-8 mode is enabled for all 8274 newly opened terminals. 8275 8276 vt.global_cursor_default= 8277 [VT] 8278 Format=<-1|0|1> 8279 Set system-wide default for whether a cursor 8280 is shown on new VTs. Default is -1, 8281 i.e. cursors will be created by default unless 8282 overridden by individual drivers. 0 will hide 8283 cursors, 1 will display them. 8284 8285 vt.italic= [VT] Default color for italic text; 0-15. 8286 Default: 2 = green. 8287 8288 vt.underline= [VT] Default color for underlined text; 0-15. 8289 Default: 3 = cyan. 8290 8291 watchdog timers [HW,WDT] For information on watchdog timers, 8292 see Documentation/watchdog/watchdog-parameters.rst 8293 or other driver-specific files in the 8294 Documentation/watchdog/ directory. 8295 8296 watchdog_thresh= 8297 [KNL] 8298 Set the hard lockup detector stall duration 8299 threshold in seconds. The soft lockup detector 8300 threshold is set to twice the value. A value of 0 8301 disables both lockup detectors. Default is 10 8302 seconds. 8303 8304 workqueue.unbound_cpus= 8305 [KNL,SMP] Specify to constrain one or some CPUs 8306 to use in unbound workqueues. 8307 Format: <cpu-list> 8308 By default, all online CPUs are available for 8309 unbound workqueues. 8310 8311 workqueue.watchdog_thresh= 8312 If CONFIG_WQ_WATCHDOG is configured, workqueue can 8313 warn stall conditions and dump internal state to 8314 help debugging. 0 disables workqueue stall 8315 detection; otherwise, it's the stall threshold 8316 duration in seconds. The default value is 30 and 8317 it can be updated at runtime by writing to the 8318 corresponding sysfs file. 8319 8320 workqueue.panic_on_stall=<uint> 8321 Panic when workqueue stall is detected by 8322 CONFIG_WQ_WATCHDOG. It sets the number times of the 8323 stall to trigger panic. 8324 8325 The default is 0, which disables the panic on stall. 8326 8327 workqueue.cpu_intensive_thresh_us= 8328 Per-cpu work items which run for longer than this 8329 threshold are automatically considered CPU intensive 8330 and excluded from concurrency management to prevent 8331 them from noticeably delaying other per-cpu work 8332 items. Default is 10000 (10ms). 8333 8334 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel 8335 will report the work functions which violate this 8336 threshold repeatedly. They are likely good 8337 candidates for using WQ_UNBOUND workqueues instead. 8338 8339 workqueue.cpu_intensive_warning_thresh=<uint> 8340 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel 8341 will report the work functions which violate the 8342 intensive_threshold_us repeatedly. In order to prevent 8343 spurious warnings, start printing only after a work 8344 function has violated this threshold number of times. 8345 8346 The default is 4 times. 0 disables the warning. 8347 8348 workqueue.power_efficient 8349 Per-cpu workqueues are generally preferred because 8350 they show better performance thanks to cache 8351 locality; unfortunately, per-cpu workqueues tend to 8352 be more power hungry than unbound workqueues. 8353 8354 Enabling this makes the per-cpu workqueues which 8355 were observed to contribute significantly to power 8356 consumption unbound, leading to measurably lower 8357 power usage at the cost of small performance 8358 overhead. 8359 8360 The default value of this parameter is determined by 8361 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT. 8362 8363 workqueue.default_affinity_scope= 8364 Select the default affinity scope to use for unbound 8365 workqueues. Can be one of "cpu", "smt", "cache", 8366 "numa" and "system". Default is "cache". For more 8367 information, see the Affinity Scopes section in 8368 Documentation/core-api/workqueue.rst. 8369 8370 This can be changed after boot by writing to the 8371 matching /sys/module/workqueue/parameters file. All 8372 workqueues with the "default" affinity scope will be 8373 updated accordingly. 8374 8375 workqueue.debug_force_rr_cpu 8376 Workqueue used to implicitly guarantee that work 8377 items queued without explicit CPU specified are put 8378 on the local CPU. This guarantee is no longer true 8379 and while local CPU is still preferred work items 8380 may be put on foreign CPUs. This debug option 8381 forces round-robin CPU selection to flush out 8382 usages which depend on the now broken guarantee. 8383 When enabled, memory and cache locality will be 8384 impacted. 8385 8386 writecombine= [LOONGARCH,EARLY] Control the MAT (Memory Access 8387 Type) of ioremap_wc(). 8388 8389 on - Enable writecombine, use WUC for ioremap_wc() 8390 off - Disable writecombine, use SUC for ioremap_wc() 8391 8392 x2apic_phys [X86-64,APIC,EARLY] Use x2apic physical mode instead of 8393 default x2apic cluster mode on platforms 8394 supporting x2apic. 8395 8396 xen_512gb_limit [KNL,X86-64,XEN] 8397 Restricts the kernel running paravirtualized under Xen 8398 to use only up to 512 GB of RAM. The reason to do so is 8399 crash analysis tools and Xen tools for doing domain 8400 save/restore/migration must be enabled to handle larger 8401 domains. 8402 8403 xen_emul_unplug= [HW,X86,XEN,EARLY] 8404 Unplug Xen emulated devices 8405 Format: [unplug0,][unplug1] 8406 ide-disks -- unplug primary master IDE devices 8407 aux-ide-disks -- unplug non-primary-master IDE devices 8408 nics -- unplug network devices 8409 all -- unplug all emulated devices (NICs and IDE disks) 8410 unnecessary -- unplugging emulated devices is 8411 unnecessary even if the host did not respond to 8412 the unplug protocol 8413 never -- do not unplug even if version check succeeds 8414 8415 xen_legacy_crash [X86,XEN,EARLY] 8416 Crash from Xen panic notifier, without executing late 8417 panic() code such as dumping handler. 8418 8419 xen_mc_debug [X86,XEN,EARLY] 8420 Enable multicall debugging when running as a Xen PV guest. 8421 Enabling this feature will reduce performance a little 8422 bit, so it should only be enabled for obtaining extended 8423 debug data in case of multicall errors. 8424 8425 xen_msr_safe= [X86,XEN,EARLY] 8426 Format: <bool> 8427 Select whether to always use non-faulting (safe) MSR 8428 access functions when running as Xen PV guest. The 8429 default value is controlled by CONFIG_XEN_PV_MSR_SAFE. 8430 8431 xen_nopv [X86] 8432 Disables the PV optimizations forcing the HVM guest to 8433 run as generic HVM guest with no PV drivers. 8434 This option is obsoleted by the "nopv" option, which 8435 has equivalent effect for XEN platform. 8436 8437 xen_no_vector_callback 8438 [KNL,X86,XEN,EARLY] Disable the vector callback for Xen 8439 event channel interrupts. 8440 8441 xen_scrub_pages= [XEN] 8442 Boolean option to control scrubbing pages before giving them back 8443 to Xen, for use by other domains. Can be also changed at runtime 8444 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages. 8445 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT. 8446 8447 xen_timer_slop= [X86-64,XEN,EARLY] 8448 Set the timer slop (in nanoseconds) for the virtual Xen 8449 timers (default is 100000). This adjusts the minimum 8450 delta of virtualized Xen timers, where lower values 8451 improve timer resolution at the expense of processing 8452 more timer interrupts. 8453 8454 xen.balloon_boot_timeout= [XEN] 8455 The time (in seconds) to wait before giving up to boot 8456 in case initial ballooning fails to free enough memory. 8457 Applies only when running as HVM or PVH guest and 8458 started with less memory configured than allowed at 8459 max. Default is 180. 8460 8461 xen.event_eoi_delay= [XEN] 8462 How long to delay EOI handling in case of event 8463 storms (jiffies). Default is 10. 8464 8465 xen.event_loop_timeout= [XEN] 8466 After which time (jiffies) the event handling loop 8467 should start to delay EOI handling. Default is 2. 8468 8469 xen.fifo_events= [XEN] 8470 Boolean parameter to disable using fifo event handling 8471 even if available. Normally fifo event handling is 8472 preferred over the 2-level event handling, as it is 8473 fairer and the number of possible event channels is 8474 much higher. Default is on (use fifo events). 8475 8476 xirc2ps_cs= [NET,PCMCIA] 8477 Format: 8478 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]] 8479 8480 xive= [PPC] 8481 By default on POWER9 and above, the kernel will 8482 natively use the XIVE interrupt controller. This option 8483 allows the fallback firmware mode to be used: 8484 8485 off Fallback to firmware control of XIVE interrupt 8486 controller on both pseries and powernv 8487 platforms. Only useful on POWER9 and above. 8488 8489 xive.store-eoi=off [PPC] 8490 By default on POWER10 and above, the kernel will use 8491 stores for EOI handling when the XIVE interrupt mode 8492 is active. This option allows the XIVE driver to use 8493 loads instead, as on POWER9. 8494 8495 xhci-hcd.quirks [USB,KNL] 8496 A hex value specifying bitmask with supplemental xhci 8497 host controller quirks. Meaning of each bit can be 8498 consulted in header drivers/usb/host/xhci.h. 8499 8500 xmon [PPC,EARLY] 8501 Format: { early | on | rw | ro | off } 8502 Controls if xmon debugger is enabled. Default is off. 8503 Passing only "xmon" is equivalent to "xmon=early". 8504 early Call xmon as early as possible on boot; xmon 8505 debugger is called from setup_arch(). 8506 on xmon debugger hooks will be installed so xmon 8507 is only called on a kernel crash. Default mode, 8508 i.e. either "ro" or "rw" mode, is controlled 8509 with CONFIG_XMON_DEFAULT_RO_MODE. 8510 rw xmon debugger hooks will be installed so xmon 8511 is called only on a kernel crash, mode is write, 8512 meaning SPR registers, memory and, other data 8513 can be written using xmon commands. 8514 ro same as "rw" option above but SPR registers, 8515 memory, and other data can't be written using 8516 xmon commands. 8517 off xmon is disabled. 8518