1 accept_memory= [MM] 2 Format: { eager | lazy } 3 default: lazy 4 By default, unaccepted memory is accepted lazily to 5 avoid prolonged boot times. The lazy option will add 6 some runtime overhead until all memory is eventually 7 accepted. In most cases the overhead is negligible. 8 For some workloads or for debugging purposes 9 accept_memory=eager can be used to accept all memory 10 at once during boot. 11 12 acpi= [HW,ACPI,X86,ARM64,RISCV64,EARLY] 13 Advanced Configuration and Power Interface 14 Format: { force | on | off | strict | noirq | rsdt | 15 copy_dsdt | nospcr } 16 force -- enable ACPI if default was off 17 on -- enable ACPI but allow fallback to DT [arm64,riscv64] 18 off -- disable ACPI if default was on 19 noirq -- do not use ACPI for IRQ routing 20 strict -- Be less tolerant of platforms that are not 21 strictly ACPI specification compliant. 22 rsdt -- prefer RSDT over (default) XSDT 23 copy_dsdt -- copy DSDT to memory 24 nocmcff -- Disable firmware first mode for corrected 25 errors. This disables parsing the HEST CMC error 26 source to check if firmware has set the FF flag. This 27 may result in duplicate corrected error reports. 28 nospcr -- disable console in ACPI SPCR table as 29 default _serial_ console on ARM64 30 For ARM64, ONLY "acpi=off", "acpi=on", "acpi=force" or 31 "acpi=nospcr" are available 32 For RISCV64, ONLY "acpi=off", "acpi=on" or "acpi=force" 33 are available 34 35 See also Documentation/power/runtime_pm.rst, pci=noacpi 36 37 acpi_apic_instance= [ACPI,IOAPIC,EARLY] 38 Format: <int> 39 2: use 2nd APIC table, if available 40 1,0: use 1st APIC table 41 default: 0 42 43 acpi_backlight= [HW,ACPI] 44 { vendor | video | native | none } 45 If set to vendor, prefer vendor-specific driver 46 (e.g. thinkpad_acpi, sony_acpi, etc.) instead 47 of the ACPI video.ko driver. 48 If set to video, use the ACPI video.ko driver. 49 If set to native, use the device's native backlight mode. 50 If set to none, disable the ACPI backlight interface. 51 52 acpi_force_32bit_fadt_addr [ACPI,EARLY] 53 force FADT to use 32 bit addresses rather than the 54 64 bit X_* addresses. Some firmware have broken 64 55 bit addresses for force ACPI ignore these and use 56 the older legacy 32 bit addresses. 57 58 acpica_no_return_repair [HW, ACPI] 59 Disable AML predefined validation mechanism 60 This mechanism can repair the evaluation result to make 61 the return objects more ACPI specification compliant. 62 This option is useful for developers to identify the 63 root cause of an AML interpreter issue when the issue 64 has something to do with the repair mechanism. 65 66 acpi.debug_layer= [HW,ACPI,ACPI_DEBUG] 67 acpi.debug_level= [HW,ACPI,ACPI_DEBUG] 68 Format: <int> 69 CONFIG_ACPI_DEBUG must be enabled to produce any ACPI 70 debug output. Bits in debug_layer correspond to a 71 _COMPONENT in an ACPI source file, e.g., 72 #define _COMPONENT ACPI_EVENTS 73 Bits in debug_level correspond to a level in 74 ACPI_DEBUG_PRINT statements, e.g., 75 ACPI_DEBUG_PRINT((ACPI_DB_INFO, ... 76 The debug_level mask defaults to "info". See 77 Documentation/firmware-guide/acpi/debug.rst for more information about 78 debug layers and levels. 79 80 Enable processor driver info messages: 81 acpi.debug_layer=0x20000000 82 Enable AML "Debug" output, i.e., stores to the Debug 83 object while interpreting AML: 84 acpi.debug_layer=0xffffffff acpi.debug_level=0x2 85 Enable all messages related to ACPI hardware: 86 acpi.debug_layer=0x2 acpi.debug_level=0xffffffff 87 88 Some values produce so much output that the system is 89 unusable. The "log_buf_len" parameter may be useful 90 if you need to capture more output. 91 92 acpi_enforce_resources= [ACPI] 93 { strict | lax | no } 94 Check for resource conflicts between native drivers 95 and ACPI OperationRegions (SystemIO and SystemMemory 96 only). IO ports and memory declared in ACPI might be 97 used by the ACPI subsystem in arbitrary AML code and 98 can interfere with legacy drivers. 99 strict (default): access to resources claimed by ACPI 100 is denied; legacy drivers trying to access reserved 101 resources will fail to bind to device using them. 102 lax: access to resources claimed by ACPI is allowed; 103 legacy drivers trying to access reserved resources 104 will bind successfully but a warning message is logged. 105 no: ACPI OperationRegions are not marked as reserved, 106 no further checks are performed. 107 108 acpi_force_table_verification [HW,ACPI,EARLY] 109 Enable table checksum verification during early stage. 110 By default, this is disabled due to x86 early mapping 111 size limitation. 112 113 acpi_irq_balance [HW,ACPI] 114 ACPI will balance active IRQs 115 default in APIC mode 116 117 acpi_irq_nobalance [HW,ACPI] 118 ACPI will not move active IRQs (default) 119 default in PIC mode 120 121 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA 122 Format: <irq>,<irq>... 123 124 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for 125 use by PCI 126 Format: <irq>,<irq>... 127 128 acpi_mask_gpe= [HW,ACPI] 129 Due to the existence of _Lxx/_Exx, some GPEs triggered 130 by unsupported hardware/firmware features can result in 131 GPE floodings that cannot be automatically disabled by 132 the GPE dispatcher. 133 This facility can be used to prevent such uncontrolled 134 GPE floodings. 135 Format: <byte> or <bitmap-list> 136 137 acpi_no_auto_serialize [HW,ACPI] 138 Disable auto-serialization of AML methods 139 AML control methods that contain the opcodes to create 140 named objects will be marked as "Serialized" by the 141 auto-serialization feature. 142 This feature is enabled by default. 143 This option allows to turn off the feature. 144 145 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump 146 kernels. 147 148 acpi_no_static_ssdt [HW,ACPI,EARLY] 149 Disable installation of static SSDTs at early boot time 150 By default, SSDTs contained in the RSDT/XSDT will be 151 installed automatically and they will appear under 152 /sys/firmware/acpi/tables. 153 This option turns off this feature. 154 Note that specifying this option does not affect 155 dynamic table installation which will install SSDT 156 tables to /sys/firmware/acpi/tables/dynamic. 157 158 acpi_no_watchdog [HW,ACPI,WDT] 159 Ignore the ACPI-based watchdog interface (WDAT) and let 160 a native driver control the watchdog device instead. 161 162 acpi_rsdp= [ACPI,EFI,KEXEC,EARLY] 163 Pass the RSDP address to the kernel, mostly used 164 on machines running EFI runtime service to boot the 165 second kernel for kdump. 166 167 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS 168 Format: To spoof as Windows 98: ="Microsoft Windows" 169 170 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead 171 of 2 which is mandated by ACPI 6) as the supported ACPI 172 specification revision (when using this switch, it may 173 be necessary to carry out a cold reboot _twice_ in a 174 row to make it take effect on the platform firmware). 175 176 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings 177 acpi_osi="string1" # add string1 178 acpi_osi="!string2" # remove string2 179 acpi_osi=!* # remove all strings 180 acpi_osi=! # disable all built-in OS vendor 181 strings 182 acpi_osi=!! # enable all built-in OS vendor 183 strings 184 acpi_osi= # disable all strings 185 186 'acpi_osi=!' can be used in combination with single or 187 multiple 'acpi_osi="string1"' to support specific OS 188 vendor string(s). Note that such command can only 189 affect the default state of the OS vendor strings, thus 190 it cannot affect the default state of the feature group 191 strings and the current state of the OS vendor strings, 192 specifying it multiple times through kernel command line 193 is meaningless. This command is useful when one do not 194 care about the state of the feature group strings which 195 should be controlled by the OSPM. 196 Examples: 197 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent 198 to 'acpi_osi="Windows 2000" acpi_osi=!', they all 199 can make '_OSI("Windows 2000")' TRUE. 200 201 'acpi_osi=' cannot be used in combination with other 202 'acpi_osi=' command lines, the _OSI method will not 203 exist in the ACPI namespace. NOTE that such command can 204 only affect the _OSI support state, thus specifying it 205 multiple times through kernel command line is also 206 meaningless. 207 Examples: 208 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)' 209 FALSE. 210 211 'acpi_osi=!*' can be used in combination with single or 212 multiple 'acpi_osi="string1"' to support specific 213 string(s). Note that such command can affect the 214 current state of both the OS vendor strings and the 215 feature group strings, thus specifying it multiple times 216 through kernel command line is meaningful. But it may 217 still not able to affect the final state of a string if 218 there are quirks related to this string. This command 219 is useful when one want to control the state of the 220 feature group strings to debug BIOS issues related to 221 the OSPM features. 222 Examples: 223 1. 'acpi_osi="Module Device" acpi_osi=!*' can make 224 '_OSI("Module Device")' FALSE. 225 2. 'acpi_osi=!* acpi_osi="Module Device"' can make 226 '_OSI("Module Device")' TRUE. 227 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is 228 equivalent to 229 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"' 230 and 231 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!', 232 they all will make '_OSI("Windows 2000")' TRUE. 233 234 acpi_pm_good [X86] 235 Override the pmtimer bug detection: force the kernel 236 to assume that this machine's pmtimer latches its value 237 and always returns good values. 238 239 acpi_sci= [HW,ACPI,EARLY] ACPI System Control Interrupt trigger mode 240 Format: { level | edge | high | low } 241 242 acpi_skip_timer_override [HW,ACPI,EARLY] 243 Recognize and ignore IRQ0/pin2 Interrupt Override. 244 For broken nForce2 BIOS resulting in XT-PIC timer. 245 246 acpi_sleep= [HW,ACPI] Sleep options 247 Format: { s3_bios, s3_mode, s3_beep, s4_hwsig, 248 s4_nohwsig, old_ordering, nonvs, 249 sci_force_enable, nobl } 250 See Documentation/power/video.rst for information on 251 s3_bios and s3_mode. 252 s3_beep is for debugging; it makes the PC's speaker beep 253 as soon as the kernel's real-mode entry point is called. 254 s4_hwsig causes the kernel to check the ACPI hardware 255 signature during resume from hibernation, and gracefully 256 refuse to resume if it has changed. This complies with 257 the ACPI specification but not with reality, since 258 Windows does not do this and many laptops do change it 259 on docking. So the default behaviour is to allow resume 260 and simply warn when the signature changes, unless the 261 s4_hwsig option is enabled. 262 s4_nohwsig prevents ACPI hardware signature from being 263 used (or even warned about) during resume. 264 old_ordering causes the ACPI 1.0 ordering of the _PTS 265 control method, with respect to putting devices into 266 low power states, to be enforced (the ACPI 2.0 ordering 267 of _PTS is used by default). 268 nonvs prevents the kernel from saving/restoring the 269 ACPI NVS memory during suspend/hibernation and resume. 270 sci_force_enable causes the kernel to set SCI_EN directly 271 on resume from S1/S3 (which is against the ACPI spec, 272 but some broken systems don't work without it). 273 nobl causes the internal blacklist of systems known to 274 behave incorrectly in some ways with respect to system 275 suspend and resume to be ignored (use wisely). 276 277 acpi_use_timer_override [HW,ACPI,EARLY] 278 Use timer override. For some broken Nvidia NF5 boards 279 that require a timer override, but don't have HPET 280 281 add_efi_memmap [EFI,X86,EARLY] Include EFI memory map in 282 kernel's map of available physical RAM. 283 284 agp= [AGP] 285 { off | try_unsupported } 286 off: disable AGP support 287 try_unsupported: try to drive unsupported chipsets 288 (may crash computer or cause data corruption) 289 290 ALSA [HW,ALSA] 291 See Documentation/sound/alsa-configuration.rst 292 293 alignment= [KNL,ARM] 294 Allow the default userspace alignment fault handler 295 behaviour to be specified. Bit 0 enables warnings, 296 bit 1 enables fixups, and bit 2 sends a segfault. 297 298 align_va_addr= [X86-64] 299 Align virtual addresses by clearing slice [14:12] when 300 allocating a VMA at process creation time. This option 301 gives you up to 3% performance improvement on AMD F15h 302 machines (where it is enabled by default) for a 303 CPU-intensive style benchmark, and it can vary highly in 304 a microbenchmark depending on workload and compiler. 305 306 32: only for 32-bit processes 307 64: only for 64-bit processes 308 on: enable for both 32- and 64-bit processes 309 off: disable for both 32- and 64-bit processes 310 311 alloc_snapshot [FTRACE] 312 Allocate the ftrace snapshot buffer on boot up when the 313 main buffer is allocated. This is handy if debugging 314 and you need to use tracing_snapshot() on boot up, and 315 do not want to use tracing_snapshot_alloc() as it needs 316 to be done where GFP_KERNEL allocations are allowed. 317 318 allow_mismatched_32bit_el0 [ARM64,EARLY] 319 Allow execve() of 32-bit applications and setting of the 320 PER_LINUX32 personality on systems where only a strict 321 subset of the CPUs support 32-bit EL0. When this 322 parameter is present, the set of CPUs supporting 32-bit 323 EL0 is indicated by /sys/devices/system/cpu/aarch32_el0 324 and hot-unplug operations may be restricted. 325 326 See Documentation/arch/arm64/asymmetric-32bit.rst for more 327 information. 328 329 amd_iommu= [HW,X86-64] 330 Pass parameters to the AMD IOMMU driver in the system. 331 Possible values are: 332 fullflush - Deprecated, equivalent to iommu.strict=1 333 off - do not initialize any AMD IOMMU found in 334 the system 335 force_isolation - Force device isolation for all 336 devices. The IOMMU driver is not 337 allowed anymore to lift isolation 338 requirements as needed. This option 339 does not override iommu=pt 340 force_enable - Force enable the IOMMU on platforms known 341 to be buggy with IOMMU enabled. Use this 342 option with care. 343 pgtbl_v1 - Use v1 page table for DMA-API (Default). 344 pgtbl_v2 - Use v2 page table for DMA-API. 345 irtcachedis - Disable Interrupt Remapping Table (IRT) caching. 346 nohugepages - Limit page-sizes used for v1 page-tables 347 to 4 KiB. 348 v2_pgsizes_only - Limit page-sizes used for v1 page-tables 349 to 4KiB/2Mib/1GiB. 350 351 352 amd_iommu_dump= [HW,X86-64] 353 Enable AMD IOMMU driver option to dump the ACPI table 354 for AMD IOMMU. With this option enabled, AMD IOMMU 355 driver will print ACPI tables for AMD IOMMU during 356 IOMMU initialization. 357 358 amd_iommu_intr= [HW,X86-64] 359 Specifies one of the following AMD IOMMU interrupt 360 remapping modes: 361 legacy - Use legacy interrupt remapping mode. 362 vapic - Use virtual APIC mode, which allows IOMMU 363 to inject interrupts directly into guest. 364 This mode requires kvm-amd.avic=1. 365 (Default when IOMMU HW support is present.) 366 367 amd_pstate= [X86,EARLY] 368 disable 369 Do not enable amd_pstate as the default 370 scaling driver for the supported processors 371 passive 372 Use amd_pstate with passive mode as a scaling driver. 373 In this mode autonomous selection is disabled. 374 Driver requests a desired performance level and platform 375 tries to match the same performance level if it is 376 satisfied by guaranteed performance level. 377 active 378 Use amd_pstate_epp driver instance as the scaling driver, 379 driver provides a hint to the hardware if software wants 380 to bias toward performance (0x0) or energy efficiency (0xff) 381 to the CPPC firmware. then CPPC power algorithm will 382 calculate the runtime workload and adjust the realtime cores 383 frequency. 384 guided 385 Activate guided autonomous mode. Driver requests minimum and 386 maximum performance level and the platform autonomously 387 selects a performance level in this range and appropriate 388 to the current workload. 389 390 amd_prefcore= 391 [X86] 392 disable 393 Disable amd-pstate preferred core. 394 395 amijoy.map= [HW,JOY] Amiga joystick support 396 Map of devices attached to JOY0DAT and JOY1DAT 397 Format: <a>,<b> 398 See also Documentation/input/joydev/joystick.rst 399 400 analog.map= [HW,JOY] Analog joystick and gamepad support 401 Specifies type or capabilities of an analog joystick 402 connected to one of 16 gameports 403 Format: <type1>,<type2>,..<type16> 404 405 apc= [HW,SPARC] 406 Power management functions (SPARCstation-4/5 + deriv.) 407 Format: noidle 408 Disable APC CPU standby support. SPARCstation-Fox does 409 not play well with APC CPU idle - disable it if you have 410 APC and your system crashes randomly. 411 412 apic [APIC,X86-64] Use IO-APIC. Default. 413 414 apic= [APIC,X86,EARLY] Advanced Programmable Interrupt Controller 415 Change the output verbosity while booting 416 Format: { quiet (default) | verbose | debug } 417 Change the amount of debugging information output 418 when initialising the APIC and IO-APIC components. 419 420 apic_extnmi= [APIC,X86,EARLY] External NMI delivery setting 421 Format: { bsp (default) | all | none } 422 bsp: External NMI is delivered only to CPU 0 423 all: External NMIs are broadcast to all CPUs as a 424 backup of CPU 0 425 none: External NMI is masked for all CPUs. This is 426 useful so that a dump capture kernel won't be 427 shot down by NMI 428 429 apicpmtimer Do APIC timer calibration using the pmtimer. Implies 430 apicmaintimer. Useful when your PIT timer is totally 431 broken. 432 433 autoconf= [IPV6] 434 See Documentation/networking/ipv6.rst. 435 436 apm= [APM] Advanced Power Management 437 See header of arch/x86/kernel/apm_32.c. 438 439 apparmor= [APPARMOR] Disable or enable AppArmor at boot time 440 Format: { "0" | "1" } 441 See security/apparmor/Kconfig help text 442 0 -- disable. 443 1 -- enable. 444 Default value is set via kernel config option. 445 446 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards 447 Format: <io>,<irq>,<nodeID> 448 449 arm64.no32bit_el0 [ARM64] Unconditionally disable the execution of 450 32 bit applications. 451 452 arm64.nobti [ARM64] Unconditionally disable Branch Target 453 Identification support 454 455 arm64.nogcs [ARM64] Unconditionally disable Guarded Control Stack 456 support 457 458 arm64.nomops [ARM64] Unconditionally disable Memory Copy and Memory 459 Set instructions support 460 461 arm64.nomte [ARM64] Unconditionally disable Memory Tagging Extension 462 support 463 464 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication 465 support 466 467 arm64.nosme [ARM64] Unconditionally disable Scalable Matrix 468 Extension support 469 470 arm64.nosve [ARM64] Unconditionally disable Scalable Vector 471 Extension support 472 473 ataflop= [HW,M68k] 474 475 atarimouse= [HW,MOUSE] Atari Mouse 476 477 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess, 478 EzKey and similar keyboards 479 480 atkbd.reset= [HW] Reset keyboard during initialization 481 482 atkbd.set= [HW] Select keyboard code set 483 Format: <int> (2 = AT (default), 3 = PS/2) 484 485 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar 486 keyboards 487 488 atkbd.softraw= [HW] Choose between synthetic and real raw mode 489 Format: <bool> (0 = real, 1 = synthetic (default)) 490 491 atkbd.softrepeat= [HW] 492 Use software keyboard repeat 493 494 audit= [KNL] Enable the audit sub-system 495 Format: { "0" | "1" | "off" | "on" } 496 0 | off - kernel audit is disabled and can not be 497 enabled until the next reboot 498 unset - kernel audit is initialized but disabled and 499 will be fully enabled by the userspace auditd. 500 1 | on - kernel audit is initialized and partially 501 enabled, storing at most audit_backlog_limit 502 messages in RAM until it is fully enabled by the 503 userspace auditd. 504 Default: unset 505 506 audit_backlog_limit= [KNL] Set the audit queue size limit. 507 Format: <int> (must be >=0) 508 Default: 64 509 510 bau= [X86_UV] Enable the BAU on SGI UV. The default 511 behavior is to disable the BAU (i.e. bau=0). 512 Format: { "0" | "1" } 513 0 - Disable the BAU. 514 1 - Enable the BAU. 515 unset - Disable the BAU. 516 517 baycom_epp= [HW,AX25] 518 Format: <io>,<mode> 519 520 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem 521 Format: <io>,<mode> 522 See header of drivers/net/hamradio/baycom_par.c. 523 524 baycom_ser_fdx= [HW,AX25] 525 BayCom Serial Port AX.25 Modem (Full Duplex Mode) 526 Format: <io>,<irq>,<mode>[,<baud>] 527 See header of drivers/net/hamradio/baycom_ser_fdx.c. 528 529 baycom_ser_hdx= [HW,AX25] 530 BayCom Serial Port AX.25 Modem (Half Duplex Mode) 531 Format: <io>,<irq>,<mode> 532 See header of drivers/net/hamradio/baycom_ser_hdx.c. 533 534 bdev_allow_write_mounted= 535 Format: <bool> 536 Control the ability to open a mounted block device 537 for writing, i.e., allow / disallow writes that bypass 538 the FS. This was implemented as a means to prevent 539 fuzzers from crashing the kernel by overwriting the 540 metadata underneath a mounted FS without its awareness. 541 This also prevents destructive formatting of mounted 542 filesystems by naive storage tooling that don't use 543 O_EXCL. Default is Y and can be changed through the 544 Kconfig option CONFIG_BLK_DEV_WRITE_MOUNTED. 545 546 bert_disable [ACPI] 547 Disable BERT OS support on buggy BIOSes. 548 549 bgrt_disable [ACPI,X86,EARLY] 550 Disable BGRT to avoid flickering OEM logo. 551 552 blkdevparts= Manual partition parsing of block device(s) for 553 embedded devices based on command line input. 554 See Documentation/block/cmdline-partition.rst 555 556 boot_delay= [KNL,EARLY] 557 Milliseconds to delay each printk during boot. 558 Only works if CONFIG_BOOT_PRINTK_DELAY is enabled, 559 and you may also have to specify "lpj=". Boot_delay 560 values larger than 10 seconds (10000) are assumed 561 erroneous and ignored. 562 Format: integer 563 564 bootconfig [KNL,EARLY] 565 Extended command line options can be added to an initrd 566 and this will cause the kernel to look for it. 567 568 See Documentation/admin-guide/bootconfig.rst 569 570 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards) 571 bttv.radio= Most important insmod options are available as 572 kernel args too. 573 bttv.pll= See Documentation/admin-guide/media/bttv.rst 574 bttv.tuner= 575 576 bulk_remove=off [PPC] This parameter disables the use of the pSeries 577 firmware feature for flushing multiple hpte entries 578 at a time. 579 580 c101= [NET] Moxa C101 synchronous serial card 581 582 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection. 583 Sometimes CPU hardware bugs make them report the cache 584 size incorrectly. The kernel will attempt work arounds 585 to fix known problems, but for some CPUs it is not 586 possible to determine what the correct size should be. 587 This option provides an override for these situations. 588 589 carrier_timeout= 590 [NET] Specifies amount of time (in seconds) that 591 the kernel should wait for a network carrier. By default 592 it waits 120 seconds. 593 594 ca_keys= [KEYS] This parameter identifies a specific key(s) on 595 the system trusted keyring to be used for certificate 596 trust validation. 597 format: { id:<keyid> | builtin } 598 599 cca= [MIPS,EARLY] Override the kernel pages' cache coherency 600 algorithm. Accepted values range from 0 to 7 601 inclusive. See arch/mips/include/asm/pgtable-bits.h 602 for platform specific values (SB1, Loongson3 and 603 others). 604 605 ccw_timeout_log [S390] 606 See Documentation/arch/s390/common_io.rst for details. 607 608 cgroup_disable= [KNL] Disable a particular controller or optional feature 609 Format: {name of the controller(s) or feature(s) to disable} 610 The effects of cgroup_disable=foo are: 611 - foo isn't auto-mounted if you mount all cgroups in 612 a single hierarchy 613 - foo isn't visible as an individually mountable 614 subsystem 615 - if foo is an optional feature then the feature is 616 disabled and corresponding cgroup files are not 617 created 618 {Currently only "memory" controller deal with this and 619 cut the overhead, others just disable the usage. So 620 only cgroup_disable=memory is actually worthy} 621 Specifying "pressure" disables per-cgroup pressure 622 stall information accounting feature 623 624 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1 625 Format: { { controller | "all" | "named" } 626 [,{ controller | "all" | "named" }...] } 627 Like cgroup_disable, but only applies to cgroup v1; 628 the blacklisted controllers remain available in cgroup2. 629 "all" blacklists all controllers and "named" disables 630 named mounts. Specifying both "all" and "named" disables 631 all v1 hierarchies. 632 633 cgroup_favordynmods= [KNL] Enable or Disable favordynmods. 634 Format: { "true" | "false" } 635 Defaults to the value of CONFIG_CGROUP_FAVOR_DYNMODS. 636 637 cgroup.memory= [KNL] Pass options to the cgroup memory controller. 638 Format: <string> 639 nosocket -- Disable socket memory accounting. 640 nokmem -- Disable kernel memory accounting. 641 nobpf -- Disable BPF memory accounting. 642 643 checkreqprot= [SELINUX] Set initial checkreqprot flag value. 644 Format: { "0" | "1" } 645 See security/selinux/Kconfig help text. 646 0 -- check protection applied by kernel (includes 647 any implied execute protection). 648 1 -- check protection requested by application. 649 Default value is set via a kernel config option. 650 Value can be changed at runtime via 651 /sys/fs/selinux/checkreqprot. 652 Setting checkreqprot to 1 is deprecated. 653 654 cio_ignore= [S390] 655 See Documentation/arch/s390/common_io.rst for details. 656 657 clearcpuid=X[,X...] [X86] 658 Disable CPUID feature X for the kernel. See 659 arch/x86/include/asm/cpufeatures.h for the valid bit 660 numbers X. Note the Linux-specific bits are not necessarily 661 stable over kernel options, but the vendor-specific 662 ones should be. 663 X can also be a string as appearing in the flags: line 664 in /proc/cpuinfo which does not have the above 665 instability issue. However, not all features have names 666 in /proc/cpuinfo. 667 Note that using this option will taint your kernel. 668 Also note that user programs calling CPUID directly 669 or using the feature without checking anything 670 will still see it. This just prevents it from 671 being used by the kernel or shown in /proc/cpuinfo. 672 Also note the kernel might malfunction if you disable 673 some critical bits. 674 675 clk_ignore_unused 676 [CLK] 677 Prevents the clock framework from automatically gating 678 clocks that have not been explicitly enabled by a Linux 679 device driver but are enabled in hardware at reset or 680 by the bootloader/firmware. Note that this does not 681 force such clocks to be always-on nor does it reserve 682 those clocks in any way. This parameter is useful for 683 debug and development, but should not be needed on a 684 platform with proper driver support. For more 685 information, see Documentation/driver-api/clk.rst. 686 687 clock= [BUGS=X86-32, HW] gettimeofday clocksource override. 688 [Deprecated] 689 Forces specified clocksource (if available) to be used 690 when calculating gettimeofday(). If specified 691 clocksource is not available, it defaults to PIT. 692 Format: { pit | tsc | cyclone | pmtmr } 693 694 clocksource= Override the default clocksource 695 Format: <string> 696 Override the default clocksource and use the clocksource 697 with the name specified. 698 Some clocksource names to choose from, depending on 699 the platform: 700 [all] jiffies (this is the base, fallback clocksource) 701 [ACPI] acpi_pm 702 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2, 703 pxa_timer,timer3,32k_counter,timer0_1 704 [X86-32] pit,hpet,tsc; 705 scx200_hrt on Geode; cyclone on IBM x440 706 [MIPS] MIPS 707 [PARISC] cr16 708 [S390] tod 709 [SH] SuperH 710 [SPARC64] tick 711 [X86-64] hpet,tsc 712 713 clocksource.arm_arch_timer.evtstrm= 714 [ARM,ARM64,EARLY] 715 Format: <bool> 716 Enable/disable the eventstream feature of the ARM 717 architected timer so that code using WFE-based polling 718 loops can be debugged more effectively on production 719 systems. 720 721 clocksource.verify_n_cpus= [KNL] 722 Limit the number of CPUs checked for clocksources 723 marked with CLOCK_SOURCE_VERIFY_PERCPU that 724 are marked unstable due to excessive skew. 725 A negative value says to check all CPUs, while 726 zero says not to check any. Values larger than 727 nr_cpu_ids are silently truncated to nr_cpu_ids. 728 The actual CPUs are chosen randomly, with 729 no replacement if the same CPU is chosen twice. 730 731 clocksource-wdtest.holdoff= [KNL] 732 Set the time in seconds that the clocksource 733 watchdog test waits before commencing its tests. 734 Defaults to zero when built as a module and to 735 10 seconds when built into the kernel. 736 737 cma=nn[MG]@[start[MG][-end[MG]]] 738 [KNL,CMA,EARLY] 739 Sets the size of kernel global memory area for 740 contiguous memory allocations and optionally the 741 placement constraint by the physical address range of 742 memory allocations. A value of 0 disables CMA 743 altogether. For more information, see 744 kernel/dma/contiguous.c 745 746 cma_pernuma=nn[MG] 747 [KNL,CMA,EARLY] 748 Sets the size of kernel per-numa memory area for 749 contiguous memory allocations. A value of 0 disables 750 per-numa CMA altogether. And If this option is not 751 specified, the default value is 0. 752 With per-numa CMA enabled, DMA users on node nid will 753 first try to allocate buffer from the pernuma area 754 which is located in node nid, if the allocation fails, 755 they will fallback to the global default memory area. 756 757 numa_cma=<node>:nn[MG][,<node>:nn[MG]] 758 [KNL,CMA,EARLY] 759 Sets the size of kernel numa memory area for 760 contiguous memory allocations. It will reserve CMA 761 area for the specified node. 762 763 With numa CMA enabled, DMA users on node nid will 764 first try to allocate buffer from the numa area 765 which is located in node nid, if the allocation fails, 766 they will fallback to the global default memory area. 767 768 cmo_free_hint= [PPC] Format: { yes | no } 769 Specify whether pages are marked as being inactive 770 when they are freed. This is used in CMO environments 771 to determine OS memory pressure for page stealing by 772 a hypervisor. 773 Default: yes 774 775 coherent_pool=nn[KMG] [ARM,KNL,EARLY] 776 Sets the size of memory pool for coherent, atomic dma 777 allocations, by default set to 256K. 778 779 com20020= [HW,NET] ARCnet - COM20020 chipset 780 Format: 781 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]] 782 783 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers) 784 Format: <io>[,<irq>] 785 786 com90xx= [HW,NET] 787 ARCnet - COM90xx chipset (memory-mapped buffers) 788 Format: <io>[,<irq>[,<memstart>]] 789 790 condev= [HW,S390] console device 791 conmode= 792 793 con3215_drop= [S390,EARLY] 3215 console drop mode. 794 Format: y|n|Y|N|1|0 795 When set to true, drop data on the 3215 console when 796 the console buffer is full. In this case the 797 operator using a 3270 terminal emulator (for example 798 x3270) does not have to enter the clear key for the 799 console output to advance and the kernel to continue. 800 This leads to a much faster boot time when a 3270 801 terminal emulator is active. If no 3270 terminal 802 emulator is used, this parameter has no effect. 803 804 console= [KNL] Output console device and options. 805 806 tty<n> Use the virtual console device <n>. 807 808 ttyS<n>[,options] 809 ttyUSB0[,options] 810 Use the specified serial port. The options are of 811 the form "bbbbpnf", where "bbbb" is the baud rate, 812 "p" is parity ("n", "o", or "e"), "n" is number of 813 bits, and "f" is flow control ("r" for RTS or 814 omit it). Default is "9600n8". 815 816 See Documentation/admin-guide/serial-console.rst for more 817 information. See 818 Documentation/networking/netconsole.rst for an 819 alternative. 820 821 <DEVNAME>:<n>.<n>[,options] 822 Use the specified serial port on the serial core bus. 823 The addressing uses DEVNAME of the physical serial port 824 device, followed by the serial core controller instance, 825 and the serial port instance. The options are the same 826 as documented for the ttyS addressing above. 827 828 The mapping of the serial ports to the tty instances 829 can be viewed with: 830 831 $ ls -d /sys/bus/serial-base/devices/*:*.*/tty/* 832 /sys/bus/serial-base/devices/00:04:0.0/tty/ttyS0 833 834 In the above example, the console can be addressed with 835 console=00:04:0.0. Note that a console addressed this 836 way will only get added when the related device driver 837 is ready. The use of an earlycon parameter in addition to 838 the console may be desired for console output early on. 839 840 uart[8250],io,<addr>[,options] 841 uart[8250],mmio,<addr>[,options] 842 uart[8250],mmio16,<addr>[,options] 843 uart[8250],mmio32,<addr>[,options] 844 uart[8250],0x<addr>[,options] 845 Start an early, polled-mode console on the 8250/16550 846 UART at the specified I/O port or MMIO address, 847 switching to the matching ttyS device later. 848 MMIO inter-register address stride is either 8-bit 849 (mmio), 16-bit (mmio16), or 32-bit (mmio32). 850 If none of [io|mmio|mmio16|mmio32], <addr> is assumed 851 to be equivalent to 'mmio'. 'options' are specified in 852 the same format described for ttyS above; if unspecified, 853 the h/w is not re-initialized. 854 855 hvc<n> Use the hypervisor console device <n>. This is for 856 both Xen and PowerPC hypervisors. 857 858 { null | "" } 859 Use to disable console output, i.e., to have kernel 860 console messages discarded. 861 This must be the only console= parameter used on the 862 kernel command line. 863 864 If the device connected to the port is not a TTY but a braille 865 device, prepend "brl," before the device type, for instance 866 console=brl,ttyS0 867 For now, only VisioBraille is supported. 868 869 console_msg_format= 870 [KNL] Change console messages format 871 default 872 By default we print messages on consoles in 873 "[time stamp] text\n" format (time stamp may not be 874 printed, depending on CONFIG_PRINTK_TIME or 875 `printk_time' param). 876 syslog 877 Switch to syslog format: "<%u>[time stamp] text\n" 878 IOW, each message will have a facility and loglevel 879 prefix. The format is similar to one used by syslog() 880 syscall, or to executing "dmesg -S --raw" or to reading 881 from /proc/kmsg. 882 883 consoleblank= [KNL] The console blank (screen saver) timeout in 884 seconds. A value of 0 disables the blank timer. 885 Defaults to 0. 886 887 coredump_filter= 888 [KNL] Change the default value for 889 /proc/<pid>/coredump_filter. 890 See also Documentation/filesystems/proc.rst. 891 892 coresight_cpu_debug.enable 893 [ARM,ARM64] 894 Format: <bool> 895 Enable/disable the CPU sampling based debugging. 896 0: default value, disable debugging 897 1: enable debugging at boot time 898 899 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver 900 Format: 901 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>] 902 903 cpuidle.off=1 [CPU_IDLE] 904 disable the cpuidle sub-system 905 906 cpuidle.governor= 907 [CPU_IDLE] Name of the cpuidle governor to use. 908 909 cpufreq.off=1 [CPU_FREQ] 910 disable the cpufreq sub-system 911 912 cpufreq.default_governor= 913 [CPU_FREQ] Name of the default cpufreq governor or 914 policy to use. This governor must be registered in the 915 kernel before the cpufreq driver probes. 916 917 cpu_init_udelay=N 918 [X86,EARLY] Delay for N microsec between assert and de-assert 919 of APIC INIT to start processors. This delay occurs 920 on every CPU online, such as boot, and resume from suspend. 921 Default: 10000 922 923 cpuhp.parallel= 924 [SMP] Enable/disable parallel bringup of secondary CPUs 925 Format: <bool> 926 Default is enabled if CONFIG_HOTPLUG_PARALLEL=y. Otherwise 927 the parameter has no effect. 928 929 crash_kexec_post_notifiers 930 Only jump to kdump kernel after running the panic 931 notifiers and dumping kmsg. This option increases 932 the risks of a kdump failure, since some panic 933 notifiers can make the crashed kernel more unstable. 934 In configurations where kdump may not be reliable, 935 running the panic notifiers could allow collecting 936 more data on dmesg, like stack traces from other CPUS 937 or extra data dumped by panic_print. Note that some 938 configurations enable this option unconditionally, 939 like Hyper-V, PowerPC (fadump) and AMD SEV-SNP. 940 941 crashkernel=size[KMG][@offset[KMG]] 942 [KNL,EARLY] Using kexec, Linux can switch to a 'crash kernel' 943 upon panic. This parameter reserves the physical 944 memory region [offset, offset + size] for that kernel 945 image. If '@offset' is omitted, then a suitable offset 946 is selected automatically. 947 [KNL, X86-64, ARM64, RISCV, LoongArch] Select a region 948 under 4G first, and fall back to reserve region above 949 4G when '@offset' hasn't been specified. 950 See Documentation/admin-guide/kdump/kdump.rst for further details. 951 952 crashkernel=range1:size1[,range2:size2,...][@offset] 953 [KNL] Same as above, but depends on the memory 954 in the running system. The syntax of range is 955 start-[end] where start and end are both 956 a memory unit (amount[KMG]). See also 957 Documentation/admin-guide/kdump/kdump.rst for an example. 958 959 crashkernel=size[KMG],high 960 [KNL, X86-64, ARM64, RISCV, LoongArch] range could be 961 above 4G. 962 Allow kernel to allocate physical memory region from top, 963 so could be above 4G if system have more than 4G ram 964 installed. Otherwise memory region will be allocated 965 below 4G, if available. 966 It will be ignored if crashkernel=X is specified. 967 crashkernel=size[KMG],low 968 [KNL, X86-64, ARM64, RISCV, LoongArch] range under 4G. 969 When crashkernel=X,high is passed, kernel could allocate 970 physical memory region above 4G, that cause second kernel 971 crash on system that require some amount of low memory, 972 e.g. swiotlb requires at least 64M+32K low memory, also 973 enough extra low memory is needed to make sure DMA buffers 974 for 32-bit devices won't run out. Kernel would try to allocate 975 default size of memory below 4G automatically. The default 976 size is platform dependent. 977 --> x86: max(swiotlb_size_or_default() + 8MiB, 256MiB) 978 --> arm64: 128MiB 979 --> riscv: 128MiB 980 --> loongarch: 128MiB 981 This one lets the user specify own low range under 4G 982 for second kernel instead. 983 0: to disable low allocation. 984 It will be ignored when crashkernel=X,high is not used 985 or memory reserved is below 4G. 986 987 cryptomgr.notests 988 [KNL] Disable crypto self-tests 989 990 cs89x0_dma= [HW,NET] 991 Format: <dma> 992 993 cs89x0_media= [HW,NET] 994 Format: { rj45 | aui | bnc } 995 996 csdlock_debug= [KNL] Enable or disable debug add-ons of cross-CPU 997 function call handling. When switched on, 998 additional debug data is printed to the console 999 in case a hanging CPU is detected, and that 1000 CPU is pinged again in order to try to resolve 1001 the hang situation. The default value of this 1002 option depends on the CSD_LOCK_WAIT_DEBUG_DEFAULT 1003 Kconfig option. 1004 1005 dasd= [HW,NET] 1006 See header of drivers/s390/block/dasd_devmap.c. 1007 1008 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port 1009 (one device per port) 1010 Format: <port#>,<type> 1011 See also Documentation/input/devices/joystick-parport.rst 1012 1013 debug [KNL,EARLY] Enable kernel debugging (events log level). 1014 1015 debug_boot_weak_hash 1016 [KNL,EARLY] Enable printing [hashed] pointers early in the 1017 boot sequence. If enabled, we use a weak hash instead 1018 of siphash to hash pointers. Use this option if you are 1019 seeing instances of '(___ptrval___)') and need to see a 1020 value (hashed pointer) instead. Cryptographically 1021 insecure, please do not use on production kernels. 1022 1023 debug_locks_verbose= 1024 [KNL] verbose locking self-tests 1025 Format: <int> 1026 Print debugging info while doing the locking API 1027 self-tests. 1028 Bitmask for the various LOCKTYPE_ tests. Defaults to 0 1029 (no extra messages), setting it to -1 (all bits set) 1030 will print _a_lot_ more information - normally only 1031 useful to lockdep developers. 1032 1033 debug_objects [KNL,EARLY] Enable object debugging 1034 1035 debug_guardpage_minorder= 1036 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this 1037 parameter allows control of the order of pages that will 1038 be intentionally kept free (and hence protected) by the 1039 buddy allocator. Bigger value increase the probability 1040 of catching random memory corruption, but reduce the 1041 amount of memory for normal system use. The maximum 1042 possible value is MAX_PAGE_ORDER/2. Setting this 1043 parameter to 1 or 2 should be enough to identify most 1044 random memory corruption problems caused by bugs in 1045 kernel or driver code when a CPU writes to (or reads 1046 from) a random memory location. Note that there exists 1047 a class of memory corruptions problems caused by buggy 1048 H/W or F/W or by drivers badly programming DMA 1049 (basically when memory is written at bus level and the 1050 CPU MMU is bypassed) which are not detectable by 1051 CONFIG_DEBUG_PAGEALLOC, hence this option will not 1052 help tracking down these problems. 1053 1054 debug_pagealloc= 1055 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this parameter 1056 enables the feature at boot time. By default, it is 1057 disabled and the system will work mostly the same as a 1058 kernel built without CONFIG_DEBUG_PAGEALLOC. 1059 Note: to get most of debug_pagealloc error reports, it's 1060 useful to also enable the page_owner functionality. 1061 on: enable the feature 1062 1063 debugfs= [KNL,EARLY] This parameter enables what is exposed to 1064 userspace and debugfs internal clients. 1065 Format: { on, no-mount, off } 1066 on: All functions are enabled. 1067 no-mount: 1068 Filesystem is not registered but kernel clients can 1069 access APIs and a crashkernel can be used to read 1070 its content. There is nothing to mount. 1071 off: Filesystem is not registered and clients 1072 get a -EPERM as result when trying to register files 1073 or directories within debugfs. 1074 This is equivalent of the runtime functionality if 1075 debugfs was not enabled in the kernel at all. 1076 Default value is set in build-time with a kernel configuration. 1077 1078 debugpat [X86] Enable PAT debugging 1079 1080 default_hugepagesz= 1081 [HW] The size of the default HugeTLB page. This is 1082 the size represented by the legacy /proc/ hugepages 1083 APIs. In addition, this is the default hugetlb size 1084 used for shmget(), mmap() and mounting hugetlbfs 1085 filesystems. If not specified, defaults to the 1086 architecture's default huge page size. Huge page 1087 sizes are architecture dependent. See also 1088 Documentation/admin-guide/mm/hugetlbpage.rst. 1089 Format: size[KMG] 1090 1091 deferred_probe_timeout= 1092 [KNL] Debugging option to set a timeout in seconds for 1093 deferred probe to give up waiting on dependencies to 1094 probe. Only specific dependencies (subsystems or 1095 drivers) that have opted in will be ignored. A timeout 1096 of 0 will timeout at the end of initcalls. If the time 1097 out hasn't expired, it'll be restarted by each 1098 successful driver registration. This option will also 1099 dump out devices still on the deferred probe list after 1100 retrying. 1101 1102 delayacct [KNL] Enable per-task delay accounting 1103 1104 dell_smm_hwmon.ignore_dmi= 1105 [HW] Continue probing hardware even if DMI data 1106 indicates that the driver is running on unsupported 1107 hardware. 1108 1109 dell_smm_hwmon.force= 1110 [HW] Activate driver even if SMM BIOS signature does 1111 not match list of supported models and enable otherwise 1112 blacklisted features. 1113 1114 dell_smm_hwmon.power_status= 1115 [HW] Report power status in /proc/i8k 1116 (disabled by default). 1117 1118 dell_smm_hwmon.restricted= 1119 [HW] Allow controlling fans only if SYS_ADMIN 1120 capability is set. 1121 1122 dell_smm_hwmon.fan_mult= 1123 [HW] Factor to multiply fan speed with. 1124 1125 dell_smm_hwmon.fan_max= 1126 [HW] Maximum configurable fan speed. 1127 1128 dfltcc= [HW,S390] 1129 Format: { on | off | def_only | inf_only | always } 1130 on: s390 zlib hardware support for compression on 1131 level 1 and decompression (default) 1132 off: No s390 zlib hardware support 1133 def_only: s390 zlib hardware support for deflate 1134 only (compression on level 1) 1135 inf_only: s390 zlib hardware support for inflate 1136 only (decompression) 1137 always: Same as 'on' but ignores the selected compression 1138 level always using hardware support (used for debugging) 1139 1140 dhash_entries= [KNL] 1141 Set number of hash buckets for dentry cache. 1142 1143 disable_1tb_segments [PPC,EARLY] 1144 Disables the use of 1TB hash page table segments. This 1145 causes the kernel to fall back to 256MB segments which 1146 can be useful when debugging issues that require an SLB 1147 miss to occur. 1148 1149 disable= [IPV6] 1150 See Documentation/networking/ipv6.rst. 1151 1152 disable_radix [PPC,EARLY] 1153 Disable RADIX MMU mode on POWER9 1154 1155 disable_tlbie [PPC] 1156 Disable TLBIE instruction. Currently does not work 1157 with KVM, with HASH MMU, or with coherent accelerators. 1158 1159 disable_ddw [PPC/PSERIES,EARLY] 1160 Disable Dynamic DMA Window support. Use this 1161 to workaround buggy firmware. 1162 1163 disable_ipv6= [IPV6] 1164 See Documentation/networking/ipv6.rst. 1165 1166 disable_mtrr_cleanup [X86,EARLY] 1167 The kernel tries to adjust MTRR layout from continuous 1168 to discrete, to make X server driver able to add WB 1169 entry later. This parameter disables that. 1170 1171 disable_mtrr_trim [X86, Intel and AMD only,EARLY] 1172 By default the kernel will trim any uncacheable 1173 memory out of your available memory pool based on 1174 MTRR settings. This parameter disables that behavior, 1175 possibly causing your machine to run very slowly. 1176 1177 disable_timer_pin_1 [X86,EARLY] 1178 Disable PIN 1 of APIC timer 1179 Can be useful to work around chipset bugs. 1180 1181 dis_ucode_ldr [X86] Disable the microcode loader. 1182 1183 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support, 1184 this option disables the debugging code at boot. 1185 1186 dma_debug_entries=<number> 1187 This option allows to tune the number of preallocated 1188 entries for DMA-API debugging code. One entry is 1189 required per DMA-API allocation. Use this if the 1190 DMA-API debugging code disables itself because the 1191 architectural default is too low. 1192 1193 dma_debug_driver=<driver_name> 1194 With this option the DMA-API debugging driver 1195 filter feature can be enabled at boot time. Just 1196 pass the driver to filter for as the parameter. 1197 The filter can be disabled or changed to another 1198 driver later using sysfs. 1199 1200 reg_file_data_sampling= 1201 [X86] Controls mitigation for Register File Data 1202 Sampling (RFDS) vulnerability. RFDS is a CPU 1203 vulnerability which may allow userspace to infer 1204 kernel data values previously stored in floating point 1205 registers, vector registers, or integer registers. 1206 RFDS only affects Intel Atom processors. 1207 1208 on: Turns ON the mitigation. 1209 off: Turns OFF the mitigation. 1210 1211 This parameter overrides the compile time default set 1212 by CONFIG_MITIGATION_RFDS. Mitigation cannot be 1213 disabled when other VERW based mitigations (like MDS) 1214 are enabled. In order to disable RFDS mitigation all 1215 VERW based mitigations need to be disabled. 1216 1217 For details see: 1218 Documentation/admin-guide/hw-vuln/reg-file-data-sampling.rst 1219 1220 driver_async_probe= [KNL] 1221 List of driver names to be probed asynchronously. * 1222 matches with all driver names. If * is specified, the 1223 rest of the listed driver names are those that will NOT 1224 match the *. 1225 Format: <driver_name1>,<driver_name2>... 1226 1227 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>] 1228 Broken monitors, graphic adapters, KVMs and EDIDless 1229 panels may send no or incorrect EDID data sets. 1230 This parameter allows to specify an EDID data sets 1231 in the /lib/firmware directory that are used instead. 1232 An EDID data set will only be used for a particular 1233 connector, if its name and a colon are prepended to 1234 the EDID name. Each connector may use a unique EDID 1235 data set by separating the files with a comma. An EDID 1236 data set with no connector name will be used for 1237 any connectors not explicitly specified. 1238 1239 dscc4.setup= [NET] 1240 1241 dt_cpu_ftrs= [PPC,EARLY] 1242 Format: {"off" | "known"} 1243 Control how the dt_cpu_ftrs device-tree binding is 1244 used for CPU feature discovery and setup (if it 1245 exists). 1246 off: Do not use it, fall back to legacy cpu table. 1247 known: Do not pass through unknown features to guests 1248 or userspace, only those that the kernel is aware of. 1249 1250 dump_apple_properties [X86] 1251 Dump name and content of EFI device properties on 1252 x86 Macs. Useful for driver authors to determine 1253 what data is available or for reverse-engineering. 1254 1255 dyndbg[="val"] [KNL,DYNAMIC_DEBUG] 1256 <module>.dyndbg[="val"] 1257 Enable debug messages at boot time. See 1258 Documentation/admin-guide/dynamic-debug-howto.rst 1259 for details. 1260 1261 early_ioremap_debug [KNL,EARLY] 1262 Enable debug messages in early_ioremap support. This 1263 is useful for tracking down temporary early mappings 1264 which are not unmapped. 1265 1266 earlycon= [KNL,EARLY] Output early console device and options. 1267 1268 When used with no options, the early console is 1269 determined by stdout-path property in device tree's 1270 chosen node or the ACPI SPCR table if supported by 1271 the platform. 1272 1273 cdns,<addr>[,options] 1274 Start an early, polled-mode console on a Cadence 1275 (xuartps) serial port at the specified address. Only 1276 supported option is baud rate. If baud rate is not 1277 specified, the serial port must already be setup and 1278 configured. 1279 1280 uart[8250],io,<addr>[,options[,uartclk]] 1281 uart[8250],mmio,<addr>[,options[,uartclk]] 1282 uart[8250],mmio32,<addr>[,options[,uartclk]] 1283 uart[8250],mmio32be,<addr>[,options[,uartclk]] 1284 uart[8250],0x<addr>[,options] 1285 Start an early, polled-mode console on the 8250/16550 1286 UART at the specified I/O port or MMIO address. 1287 MMIO inter-register address stride is either 8-bit 1288 (mmio) or 32-bit (mmio32 or mmio32be). 1289 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed 1290 to be equivalent to 'mmio'. 'options' are specified 1291 in the same format described for "console=ttyS<n>"; if 1292 unspecified, the h/w is not initialized. 'uartclk' is 1293 the uart clock frequency; if unspecified, it is set 1294 to 'BASE_BAUD' * 16. 1295 1296 pl011,<addr> 1297 pl011,mmio32,<addr> 1298 Start an early, polled-mode console on a pl011 serial 1299 port at the specified address. The pl011 serial port 1300 must already be setup and configured. Options are not 1301 yet supported. If 'mmio32' is specified, then only 1302 the driver will use only 32-bit accessors to read/write 1303 the device registers. 1304 1305 liteuart,<addr> 1306 Start an early console on a litex serial port at the 1307 specified address. The serial port must already be 1308 setup and configured. Options are not yet supported. 1309 1310 meson,<addr> 1311 Start an early, polled-mode console on a meson serial 1312 port at the specified address. The serial port must 1313 already be setup and configured. Options are not yet 1314 supported. 1315 1316 msm_serial,<addr> 1317 Start an early, polled-mode console on an msm serial 1318 port at the specified address. The serial port 1319 must already be setup and configured. Options are not 1320 yet supported. 1321 1322 msm_serial_dm,<addr> 1323 Start an early, polled-mode console on an msm serial 1324 dm port at the specified address. The serial port 1325 must already be setup and configured. Options are not 1326 yet supported. 1327 1328 owl,<addr> 1329 Start an early, polled-mode console on a serial port 1330 of an Actions Semi SoC, such as S500 or S900, at the 1331 specified address. The serial port must already be 1332 setup and configured. Options are not yet supported. 1333 1334 rda,<addr> 1335 Start an early, polled-mode console on a serial port 1336 of an RDA Micro SoC, such as RDA8810PL, at the 1337 specified address. The serial port must already be 1338 setup and configured. Options are not yet supported. 1339 1340 sbi 1341 Use RISC-V SBI (Supervisor Binary Interface) for early 1342 console. 1343 1344 smh Use ARM semihosting calls for early console. 1345 1346 s3c2410,<addr> 1347 s3c2412,<addr> 1348 s3c2440,<addr> 1349 s3c6400,<addr> 1350 s5pv210,<addr> 1351 exynos4210,<addr> 1352 Use early console provided by serial driver available 1353 on Samsung SoCs, requires selecting proper type and 1354 a correct base address of the selected UART port. The 1355 serial port must already be setup and configured. 1356 Options are not yet supported. 1357 1358 lantiq,<addr> 1359 Start an early, polled-mode console on a lantiq serial 1360 (lqasc) port at the specified address. The serial port 1361 must already be setup and configured. Options are not 1362 yet supported. 1363 1364 lpuart,<addr> 1365 lpuart32,<addr> 1366 Use early console provided by Freescale LP UART driver 1367 found on Freescale Vybrid and QorIQ LS1021A processors. 1368 A valid base address must be provided, and the serial 1369 port must already be setup and configured. 1370 1371 ec_imx21,<addr> 1372 ec_imx6q,<addr> 1373 Start an early, polled-mode, output-only console on the 1374 Freescale i.MX UART at the specified address. The UART 1375 must already be setup and configured. 1376 1377 ar3700_uart,<addr> 1378 Start an early, polled-mode console on the 1379 Armada 3700 serial port at the specified 1380 address. The serial port must already be setup 1381 and configured. Options are not yet supported. 1382 1383 qcom_geni,<addr> 1384 Start an early, polled-mode console on a Qualcomm 1385 Generic Interface (GENI) based serial port at the 1386 specified address. The serial port must already be 1387 setup and configured. Options are not yet supported. 1388 1389 efifb,[options] 1390 Start an early, unaccelerated console on the EFI 1391 memory mapped framebuffer (if available). On cache 1392 coherent non-x86 systems that use system memory for 1393 the framebuffer, pass the 'ram' option so that it is 1394 mapped with the correct attributes. 1395 1396 linflex,<addr> 1397 Use early console provided by Freescale LINFlexD UART 1398 serial driver for NXP S32V234 SoCs. A valid base 1399 address must be provided, and the serial port must 1400 already be setup and configured. 1401 1402 earlyprintk= [X86,SH,ARM,M68k,S390,UM,EARLY] 1403 earlyprintk=vga 1404 earlyprintk=sclp 1405 earlyprintk=xen 1406 earlyprintk=serial[,ttySn[,baudrate]] 1407 earlyprintk=serial[,0x...[,baudrate]] 1408 earlyprintk=ttySn[,baudrate] 1409 earlyprintk=dbgp[debugController#] 1410 earlyprintk=mmio32,membase[,{nocfg|baudrate}] 1411 earlyprintk=pciserial[,force],bus:device.function[,{nocfg|baudrate}] 1412 earlyprintk=xdbc[xhciController#] 1413 earlyprintk=bios 1414 1415 earlyprintk is useful when the kernel crashes before 1416 the normal console is initialized. It is not enabled by 1417 default because it has some cosmetic problems. 1418 1419 Use "nocfg" to skip UART configuration, assume 1420 BIOS/firmware has configured UART correctly. 1421 1422 Append ",keep" to not disable it when the real console 1423 takes over. 1424 1425 Only one of vga, serial, or usb debug port can 1426 be used at a time. 1427 1428 Currently only ttyS0 and ttyS1 may be specified by 1429 name. Other I/O ports may be explicitly specified 1430 on some architectures (x86 and arm at least) by 1431 replacing ttySn with an I/O port address, like this: 1432 earlyprintk=serial,0x1008,115200 1433 You can find the port for a given device in 1434 /proc/tty/driver/serial: 1435 2: uart:ST16650V2 port:00001008 irq:18 ... 1436 1437 Interaction with the standard serial driver is not 1438 very good. 1439 1440 The VGA output is eventually overwritten by 1441 the real console. 1442 1443 The xen option can only be used in Xen domains. 1444 1445 The sclp output can only be used on s390. 1446 1447 The bios output can only be used on SuperH. 1448 1449 The optional "force" to "pciserial" enables use of a 1450 PCI device even when its classcode is not of the 1451 UART class. 1452 1453 edac_report= [HW,EDAC] Control how to report EDAC event 1454 Format: {"on" | "off" | "force"} 1455 on: enable EDAC to report H/W event. May be overridden 1456 by other higher priority error reporting module. 1457 off: disable H/W event reporting through EDAC. 1458 force: enforce the use of EDAC to report H/W event. 1459 default: on. 1460 1461 edd= [EDD] 1462 Format: {"off" | "on" | "skip[mbr]"} 1463 1464 efi= [EFI,EARLY] 1465 Format: { "debug", "disable_early_pci_dma", 1466 "nochunk", "noruntime", "nosoftreserve", 1467 "novamap", "no_disable_early_pci_dma" } 1468 debug: enable misc debug output. 1469 disable_early_pci_dma: disable the busmaster bit on all 1470 PCI bridges while in the EFI boot stub. 1471 nochunk: disable reading files in "chunks" in the EFI 1472 boot stub, as chunking can cause problems with some 1473 firmware implementations. 1474 noruntime : disable EFI runtime services support 1475 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose) 1476 attribute may cause the kernel to reserve the 1477 memory range for a memory mapping driver to 1478 claim. Specify efi=nosoftreserve to disable this 1479 reservation and treat the memory by its base type 1480 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM"). 1481 novamap: do not call SetVirtualAddressMap(). 1482 no_disable_early_pci_dma: Leave the busmaster bit set 1483 on all PCI bridges while in the EFI boot stub 1484 1485 efi_no_storage_paranoia [EFI,X86,EARLY] 1486 Using this parameter you can use more than 50% of 1487 your efi variable storage. Use this parameter only if 1488 you are really sure that your UEFI does sane gc and 1489 fulfills the spec otherwise your board may brick. 1490 1491 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT 1492 that is to be dynamically loaded by Linux. If there are 1493 multiple variables with the same name but with different 1494 vendor GUIDs, all of them will be loaded. See 1495 Documentation/admin-guide/acpi/ssdt-overlays.rst for details. 1496 1497 1498 eisa_irq_edge= [PARISC,HW] 1499 See header of drivers/parisc/eisa.c. 1500 1501 ekgdboc= [X86,KGDB,EARLY] Allow early kernel console debugging 1502 Format: ekgdboc=kbd 1503 1504 This is designed to be used in conjunction with 1505 the boot argument: earlyprintk=vga 1506 1507 This parameter works in place of the kgdboc parameter 1508 but can only be used if the backing tty is available 1509 very early in the boot process. For early debugging 1510 via a serial port see kgdboc_earlycon instead. 1511 1512 elanfreq= [X86-32] 1513 See comment before function elanfreq_setup() in 1514 arch/x86/kernel/cpu/cpufreq/elanfreq.c. 1515 1516 elfcorehdr=[size[KMG]@]offset[KMG] [PPC,SH,X86,S390,EARLY] 1517 Specifies physical address of start of kernel core 1518 image elf header and optionally the size. Generally 1519 kexec loader will pass this option to capture kernel. 1520 See Documentation/admin-guide/kdump/kdump.rst for details. 1521 1522 enable_mtrr_cleanup [X86,EARLY] 1523 The kernel tries to adjust MTRR layout from continuous 1524 to discrete, to make X server driver able to add WB 1525 entry later. This parameter enables that. 1526 1527 enable_timer_pin_1 [X86] 1528 Enable PIN 1 of APIC timer 1529 Can be useful to work around chipset bugs 1530 (in particular on some ATI chipsets). 1531 The kernel tries to set a reasonable default. 1532 1533 enforcing= [SELINUX] Set initial enforcing status. 1534 Format: {"0" | "1"} 1535 See security/selinux/Kconfig help text. 1536 0 -- permissive (log only, no denials). 1537 1 -- enforcing (deny and log). 1538 Default value is 0. 1539 Value can be changed at runtime via 1540 /sys/fs/selinux/enforce. 1541 1542 erst_disable [ACPI] 1543 Disable Error Record Serialization Table (ERST) 1544 support. 1545 1546 ether= [HW,NET] Ethernet cards parameters 1547 This option is obsoleted by the "netdev=" option, which 1548 has equivalent usage. See its documentation for details. 1549 1550 evm= [EVM] 1551 Format: { "fix" } 1552 Permit 'security.evm' to be updated regardless of 1553 current integrity status. 1554 1555 early_page_ext [KNL,EARLY] Enforces page_ext initialization to earlier 1556 stages so cover more early boot allocations. 1557 Please note that as side effect some optimizations 1558 might be disabled to achieve that (e.g. parallelized 1559 memory initialization is disabled) so the boot process 1560 might take longer, especially on systems with a lot of 1561 memory. Available with CONFIG_PAGE_EXTENSION=y. 1562 1563 failslab= 1564 fail_usercopy= 1565 fail_page_alloc= 1566 fail_skb_realloc= 1567 fail_make_request=[KNL] 1568 General fault injection mechanism. 1569 Format: <interval>,<probability>,<space>,<times> 1570 See also Documentation/fault-injection/. 1571 1572 fb_tunnels= [NET] 1573 Format: { initns | none } 1574 See Documentation/admin-guide/sysctl/net.rst for 1575 fb_tunnels_only_for_init_ns 1576 1577 floppy= [HW] 1578 See Documentation/admin-guide/blockdev/floppy.rst. 1579 1580 forcepae [X86-32] 1581 Forcefully enable Physical Address Extension (PAE). 1582 Many Pentium M systems disable PAE but may have a 1583 functionally usable PAE implementation. 1584 Warning: use of this parameter will taint the kernel 1585 and may cause unknown problems. 1586 1587 fred= [X86-64] 1588 Enable/disable Flexible Return and Event Delivery. 1589 Format: { on | off } 1590 on: enable FRED when it's present. 1591 off: disable FRED, the default setting. 1592 1593 ftrace=[tracer] 1594 [FTRACE] will set and start the specified tracer 1595 as early as possible in order to facilitate early 1596 boot debugging. 1597 1598 ftrace_boot_snapshot 1599 [FTRACE] On boot up, a snapshot will be taken of the 1600 ftrace ring buffer that can be read at: 1601 /sys/kernel/tracing/snapshot. 1602 This is useful if you need tracing information from kernel 1603 boot up that is likely to be overridden by user space 1604 start up functionality. 1605 1606 Optionally, the snapshot can also be defined for a tracing 1607 instance that was created by the trace_instance= command 1608 line parameter. 1609 1610 trace_instance=foo,sched_switch ftrace_boot_snapshot=foo 1611 1612 The above will cause the "foo" tracing instance to trigger 1613 a snapshot at the end of boot up. 1614 1615 ftrace_dump_on_oops[=2(orig_cpu) | =<instance>][,<instance> | 1616 ,<instance>=2(orig_cpu)] 1617 [FTRACE] will dump the trace buffers on oops. 1618 If no parameter is passed, ftrace will dump global 1619 buffers of all CPUs, if you pass 2 or orig_cpu, it 1620 will dump only the buffer of the CPU that triggered 1621 the oops, or the specific instance will be dumped if 1622 its name is passed. Multiple instance dump is also 1623 supported, and instances are separated by commas. Each 1624 instance supports only dump on CPU that triggered the 1625 oops by passing 2 or orig_cpu to it. 1626 1627 ftrace_dump_on_oops=foo=orig_cpu 1628 1629 The above will dump only the buffer of "foo" instance 1630 on CPU that triggered the oops. 1631 1632 ftrace_dump_on_oops,foo,bar=orig_cpu 1633 1634 The above will dump global buffer on all CPUs, the 1635 buffer of "foo" instance on all CPUs and the buffer 1636 of "bar" instance on CPU that triggered the oops. 1637 1638 ftrace_filter=[function-list] 1639 [FTRACE] Limit the functions traced by the function 1640 tracer at boot up. function-list is a comma-separated 1641 list of functions. This list can be changed at run 1642 time by the set_ftrace_filter file in the debugfs 1643 tracing directory. 1644 1645 ftrace_notrace=[function-list] 1646 [FTRACE] Do not trace the functions specified in 1647 function-list. This list can be changed at run time 1648 by the set_ftrace_notrace file in the debugfs 1649 tracing directory. 1650 1651 ftrace_graph_filter=[function-list] 1652 [FTRACE] Limit the top level callers functions traced 1653 by the function graph tracer at boot up. 1654 function-list is a comma-separated list of functions 1655 that can be changed at run time by the 1656 set_graph_function file in the debugfs tracing directory. 1657 1658 ftrace_graph_notrace=[function-list] 1659 [FTRACE] Do not trace from the functions specified in 1660 function-list. This list is a comma-separated list of 1661 functions that can be changed at run time by the 1662 set_graph_notrace file in the debugfs tracing directory. 1663 1664 ftrace_graph_max_depth=<uint> 1665 [FTRACE] Used with the function graph tracer. This is 1666 the max depth it will trace into a function. This value 1667 can be changed at run time by the max_graph_depth file 1668 in the tracefs tracing directory. default: 0 (no limit) 1669 1670 fw_devlink= [KNL,EARLY] Create device links between consumer and supplier 1671 devices by scanning the firmware to infer the 1672 consumer/supplier relationships. This feature is 1673 especially useful when drivers are loaded as modules as 1674 it ensures proper ordering of tasks like device probing 1675 (suppliers first, then consumers), supplier boot state 1676 clean up (only after all consumers have probed), 1677 suspend/resume & runtime PM (consumers first, then 1678 suppliers). 1679 Format: { off | permissive | on | rpm } 1680 off -- Don't create device links from firmware info. 1681 permissive -- Create device links from firmware info 1682 but use it only for ordering boot state clean 1683 up (sync_state() calls). 1684 on -- Create device links from firmware info and use it 1685 to enforce probe and suspend/resume ordering. 1686 rpm -- Like "on", but also use to order runtime PM. 1687 1688 fw_devlink.strict=<bool> 1689 [KNL,EARLY] Treat all inferred dependencies as mandatory 1690 dependencies. This only applies for fw_devlink=on|rpm. 1691 Format: <bool> 1692 1693 fw_devlink.sync_state = 1694 [KNL,EARLY] When all devices that could probe have finished 1695 probing, this parameter controls what to do with 1696 devices that haven't yet received their sync_state() 1697 calls. 1698 Format: { strict | timeout } 1699 strict -- Default. Continue waiting on consumers to 1700 probe successfully. 1701 timeout -- Give up waiting on consumers and call 1702 sync_state() on any devices that haven't yet 1703 received their sync_state() calls after 1704 deferred_probe_timeout has expired or by 1705 late_initcall() if !CONFIG_MODULES. 1706 1707 gamecon.map[2|3]= 1708 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad 1709 support via parallel port (up to 5 devices per port) 1710 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5> 1711 See also Documentation/input/devices/joystick-parport.rst 1712 1713 gamma= [HW,DRM] 1714 1715 gart_fix_e820= [X86-64,EARLY] disable the fix e820 for K8 GART 1716 Format: off | on 1717 default: on 1718 1719 gather_data_sampling= 1720 [X86,INTEL,EARLY] Control the Gather Data Sampling (GDS) 1721 mitigation. 1722 1723 Gather Data Sampling is a hardware vulnerability which 1724 allows unprivileged speculative access to data which was 1725 previously stored in vector registers. 1726 1727 This issue is mitigated by default in updated microcode. 1728 The mitigation may have a performance impact but can be 1729 disabled. On systems without the microcode mitigation 1730 disabling AVX serves as a mitigation. 1731 1732 force: Disable AVX to mitigate systems without 1733 microcode mitigation. No effect if the microcode 1734 mitigation is present. Known to cause crashes in 1735 userspace with buggy AVX enumeration. 1736 1737 off: Disable GDS mitigation. 1738 1739 gbpages [X86] Use GB pages for kernel direct mappings. 1740 1741 gcov_persist= [GCOV] When non-zero (default), profiling data for 1742 kernel modules is saved and remains accessible via 1743 debugfs, even when the module is unloaded/reloaded. 1744 When zero, profiling data is discarded and associated 1745 debugfs files are removed at module unload time. 1746 1747 goldfish [X86] Enable the goldfish android emulator platform. 1748 Don't use this when you are not running on the 1749 android emulator 1750 1751 gpio-mockup.gpio_mockup_ranges 1752 [HW] Sets the ranges of gpiochip of for this device. 1753 Format: <start1>,<end1>,<start2>,<end2>... 1754 gpio-mockup.gpio_mockup_named_lines 1755 [HW] Let the driver know GPIO lines should be named. 1756 1757 gpt [EFI] Forces disk with valid GPT signature but 1758 invalid Protective MBR to be treated as GPT. If the 1759 primary GPT is corrupted, it enables the backup/alternate 1760 GPT to be used instead. 1761 1762 grcan.enable0= [HW] Configuration of physical interface 0. Determines 1763 the "Enable 0" bit of the configuration register. 1764 Format: 0 | 1 1765 Default: 0 1766 grcan.enable1= [HW] Configuration of physical interface 1. Determines 1767 the "Enable 0" bit of the configuration register. 1768 Format: 0 | 1 1769 Default: 0 1770 grcan.select= [HW] Select which physical interface to use. 1771 Format: 0 | 1 1772 Default: 0 1773 grcan.txsize= [HW] Sets the size of the tx buffer. 1774 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0. 1775 Default: 1024 1776 grcan.rxsize= [HW] Sets the size of the rx buffer. 1777 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0. 1778 Default: 1024 1779 1780 hardened_usercopy= 1781 [KNL] Under CONFIG_HARDENED_USERCOPY, whether 1782 hardening is enabled for this boot. Hardened 1783 usercopy checking is used to protect the kernel 1784 from reading or writing beyond known memory 1785 allocation boundaries as a proactive defense 1786 against bounds-checking flaws in the kernel's 1787 copy_to_user()/copy_from_user() interface. 1788 The default is determined by 1789 CONFIG_HARDENED_USERCOPY_DEFAULT_ON. 1790 on Perform hardened usercopy checks. 1791 off Disable hardened usercopy checks. 1792 1793 hardlockup_all_cpu_backtrace= 1794 [KNL] Should the hard-lockup detector generate 1795 backtraces on all cpus. 1796 Format: 0 | 1 1797 1798 hashdist= [KNL,NUMA] Large hashes allocated during boot 1799 are distributed across NUMA nodes. Defaults on 1800 for 64-bit NUMA, off otherwise. 1801 Format: 0 | 1 (for off | on) 1802 1803 hd= [EIDE] (E)IDE hard drive subsystem geometry 1804 Format: <cyl>,<head>,<sect> 1805 1806 hest_disable [ACPI] 1807 Disable Hardware Error Source Table (HEST) support; 1808 corresponding firmware-first mode error processing 1809 logic will be disabled. 1810 1811 hibernate= [HIBERNATION] 1812 noresume Don't check if there's a hibernation image 1813 present during boot. 1814 nocompress Don't compress/decompress hibernation images. 1815 no Disable hibernation and resume. 1816 protect_image Turn on image protection during restoration 1817 (that will set all pages holding image data 1818 during restoration read-only). 1819 1820 hibernate.compressor= [HIBERNATION] Compression algorithm to be 1821 used with hibernation. 1822 Format: { lzo | lz4 } 1823 Default: lzo 1824 1825 lzo: Select LZO compression algorithm to 1826 compress/decompress hibernation image. 1827 1828 lz4: Select LZ4 compression algorithm to 1829 compress/decompress hibernation image. 1830 1831 highmem=nn[KMG] [KNL,BOOT,EARLY] forces the highmem zone to have an exact 1832 size of <nn>. This works even on boxes that have no 1833 highmem otherwise. This also works to reduce highmem 1834 size on bigger boxes. 1835 1836 highres= [KNL] Enable/disable high resolution timer mode. 1837 Valid parameters: "on", "off" 1838 Default: "on" 1839 1840 hlt [BUGS=ARM,SH] 1841 1842 hostname= [KNL,EARLY] Set the hostname (aka UTS nodename). 1843 Format: <string> 1844 This allows setting the system's hostname during early 1845 startup. This sets the name returned by gethostname. 1846 Using this parameter to set the hostname makes it 1847 possible to ensure the hostname is correctly set before 1848 any userspace processes run, avoiding the possibility 1849 that a process may call gethostname before the hostname 1850 has been explicitly set, resulting in the calling 1851 process getting an incorrect result. The string must 1852 not exceed the maximum allowed hostname length (usually 1853 64 characters) and will be truncated otherwise. 1854 1855 hpet= [X86-32,HPET] option to control HPET usage 1856 Format: { enable (default) | disable | force | 1857 verbose } 1858 disable: disable HPET and use PIT instead 1859 force: allow force enabled of undocumented chips (ICH4, 1860 VIA, nVidia) 1861 verbose: show contents of HPET registers during setup 1862 1863 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET 1864 registers. Default set by CONFIG_HPET_MMAP_DEFAULT. 1865 1866 hugepages= [HW,EARLY] Number of HugeTLB pages to allocate at boot. 1867 If this follows hugepagesz (below), it specifies 1868 the number of pages of hugepagesz to be allocated. 1869 If this is the first HugeTLB parameter on the command 1870 line, it specifies the number of pages to allocate for 1871 the default huge page size. If using node format, the 1872 number of pages to allocate per-node can be specified. 1873 See also Documentation/admin-guide/mm/hugetlbpage.rst. 1874 Format: <integer> or (node format) 1875 <node>:<integer>[,<node>:<integer>] 1876 1877 hugepagesz= 1878 [HW,EARLY] The size of the HugeTLB pages. This is 1879 used in conjunction with hugepages (above) to 1880 allocate huge pages of a specific size at boot. The 1881 pair hugepagesz=X hugepages=Y can be specified once 1882 for each supported huge page size. Huge page sizes 1883 are architecture dependent. See also 1884 Documentation/admin-guide/mm/hugetlbpage.rst. 1885 Format: size[KMG] 1886 1887 hugepage_alloc_threads= 1888 [HW] The number of threads that should be used to 1889 allocate hugepages during boot. This option can be 1890 used to improve system bootup time when allocating 1891 a large amount of huge pages. 1892 The default value is 25% of the available hardware threads. 1893 1894 Note that this parameter only applies to non-gigantic huge pages. 1895 1896 hugetlb_cma= [HW,CMA,EARLY] The size of a CMA area used for allocation 1897 of gigantic hugepages. Or using node format, the size 1898 of a CMA area per node can be specified. 1899 Format: nn[KMGTPE] or (node format) 1900 <node>:nn[KMGTPE][,<node>:nn[KMGTPE]] 1901 1902 Reserve a CMA area of given size and allocate gigantic 1903 hugepages using the CMA allocator. If enabled, the 1904 boot-time allocation of gigantic hugepages is skipped. 1905 1906 hugetlb_cma_only= 1907 [HW,CMA,EARLY] When allocating new HugeTLB pages, only 1908 try to allocate from the CMA areas. 1909 1910 This option does nothing if hugetlb_cma= is not also 1911 specified. 1912 1913 hugetlb_free_vmemmap= 1914 [KNL] Requires CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP 1915 enabled. 1916 Control if HugeTLB Vmemmap Optimization (HVO) is enabled. 1917 Allows heavy hugetlb users to free up some more 1918 memory (7 * PAGE_SIZE for each 2MB hugetlb page). 1919 Format: { on | off (default) } 1920 1921 on: enable HVO 1922 off: disable HVO 1923 1924 Built with CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP_DEFAULT_ON=y, 1925 the default is on. 1926 1927 Note that the vmemmap pages may be allocated from the added 1928 memory block itself when memory_hotplug.memmap_on_memory is 1929 enabled, those vmemmap pages cannot be optimized even if this 1930 feature is enabled. Other vmemmap pages not allocated from 1931 the added memory block itself do not be affected. 1932 1933 hung_task_panic= 1934 [KNL] Should the hung task detector generate panics. 1935 Format: 0 | 1 1936 1937 A value of 1 instructs the kernel to panic when a 1938 hung task is detected. The default value is controlled 1939 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time 1940 option. The value selected by this boot parameter can 1941 be changed later by the kernel.hung_task_panic sysctl. 1942 1943 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC) 1944 terminal devices. Valid values: 0..8 1945 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs. 1946 If specified, z/VM IUCV HVC accepts connections 1947 from listed z/VM user IDs only. 1948 1949 hv_nopvspin [X86,HYPER_V,EARLY] 1950 Disables the paravirt spinlock optimizations 1951 which allow the hypervisor to 'idle' the guest 1952 on lock contention. 1953 1954 hw_protection= [HW] 1955 Format: reboot | shutdown 1956 1957 Hardware protection action taken on critical events like 1958 overtemperature or imminent voltage loss. 1959 1960 i2c_bus= [HW] Override the default board specific I2C bus speed 1961 or register an additional I2C bus that is not 1962 registered from board initialization code. 1963 Format: 1964 <bus_id>,<clkrate> 1965 1966 i2c_touchscreen_props= [HW,ACPI,X86] 1967 Set device-properties for ACPI-enumerated I2C-attached 1968 touchscreen, to e.g. fix coordinates of upside-down 1969 mounted touchscreens. If you need this option please 1970 submit a drivers/platform/x86/touchscreen_dmi.c patch 1971 adding a DMI quirk for this. 1972 1973 Format: 1974 <ACPI_HW_ID>:<prop_name>=<val>[:prop_name=val][:...] 1975 Where <val> is one of: 1976 Omit "=<val>" entirely Set a boolean device-property 1977 Unsigned number Set a u32 device-property 1978 Anything else Set a string device-property 1979 1980 Examples (split over multiple lines): 1981 i2c_touchscreen_props=GDIX1001:touchscreen-inverted-x: 1982 touchscreen-inverted-y 1983 1984 i2c_touchscreen_props=MSSL1680:touchscreen-size-x=1920: 1985 touchscreen-size-y=1080:touchscreen-inverted-y: 1986 firmware-name=gsl1680-vendor-model.fw:silead,home-button 1987 1988 i8042.debug [HW] Toggle i8042 debug mode 1989 i8042.unmask_kbd_data 1990 [HW] Enable printing of interrupt data from the KBD port 1991 (disabled by default, and as a pre-condition 1992 requires that i8042.debug=1 be enabled) 1993 i8042.direct [HW] Put keyboard port into non-translated mode 1994 i8042.dumbkbd [HW] Pretend that controller can only read data from 1995 keyboard and cannot control its state 1996 (Don't attempt to blink the leds) 1997 i8042.noaux [HW] Don't check for auxiliary (== mouse) port 1998 i8042.nokbd [HW] Don't check/create keyboard port 1999 i8042.noloop [HW] Disable the AUX Loopback command while probing 2000 for the AUX port 2001 i8042.nomux [HW] Don't check presence of an active multiplexing 2002 controller 2003 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX 2004 controllers 2005 i8042.notimeout [HW] Ignore timeout condition signalled by controller 2006 i8042.reset [HW] Reset the controller during init, cleanup and 2007 suspend-to-ram transitions, only during s2r 2008 transitions, or never reset 2009 Format: { 1 | Y | y | 0 | N | n } 2010 1, Y, y: always reset controller 2011 0, N, n: don't ever reset controller 2012 Default: only on s2r transitions on x86; most other 2013 architectures force reset to be always executed 2014 i8042.unlock [HW] Unlock (ignore) the keylock 2015 i8042.kbdreset [HW] Reset device connected to KBD port 2016 i8042.probe_defer 2017 [HW] Allow deferred probing upon i8042 probe errors 2018 2019 i810= [HW,DRM] 2020 2021 i915.invert_brightness= 2022 [DRM] Invert the sense of the variable that is used to 2023 set the brightness of the panel backlight. Normally a 2024 brightness value of 0 indicates backlight switched off, 2025 and the maximum of the brightness value sets the backlight 2026 to maximum brightness. If this parameter is set to 0 2027 (default) and the machine requires it, or this parameter 2028 is set to 1, a brightness value of 0 sets the backlight 2029 to maximum brightness, and the maximum of the brightness 2030 value switches the backlight off. 2031 -1 -- never invert brightness 2032 0 -- machine default 2033 1 -- force brightness inversion 2034 2035 ia32_emulation= [X86-64] 2036 Format: <bool> 2037 When true, allows loading 32-bit programs and executing 32-bit 2038 syscalls, essentially overriding IA32_EMULATION_DEFAULT_DISABLED at 2039 boot time. When false, unconditionally disables IA32 emulation. 2040 2041 icn= [HW,ISDN] 2042 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]] 2043 2044 2045 idle= [X86,EARLY] 2046 Format: idle=poll, idle=halt, idle=nomwait 2047 2048 idle=poll: Don't do power saving in the idle loop 2049 using HLT, but poll for rescheduling event. This will 2050 make the CPUs eat a lot more power, but may be useful 2051 to get slightly better performance in multiprocessor 2052 benchmarks. It also makes some profiling using 2053 performance counters more accurate. Please note that 2054 on systems with MONITOR/MWAIT support (like Intel 2055 EM64T CPUs) this option has no performance advantage 2056 over the normal idle loop. It may also interact badly 2057 with hyperthreading. 2058 2059 idle=halt: Halt is forced to be used for CPU idle. 2060 In such case C2/C3 won't be used again. 2061 2062 idle=nomwait: Disable mwait for CPU C-states 2063 2064 idxd.sva= [HW] 2065 Format: <bool> 2066 Allow force disabling of Shared Virtual Memory (SVA) 2067 support for the idxd driver. By default it is set to 2068 true (1). 2069 2070 idxd.tc_override= [HW] 2071 Format: <bool> 2072 Allow override of default traffic class configuration 2073 for the device. By default it is set to false (0). 2074 2075 ieee754= [MIPS] Select IEEE Std 754 conformance mode 2076 Format: { strict | legacy | 2008 | relaxed | emulated } 2077 Default: strict 2078 2079 Choose which programs will be accepted for execution 2080 based on the IEEE 754 NaN encoding(s) supported by 2081 the FPU and the NaN encoding requested with the value 2082 of an ELF file header flag individually set by each 2083 binary. Hardware implementations are permitted to 2084 support either or both of the legacy and the 2008 NaN 2085 encoding mode. 2086 2087 Available settings are as follows: 2088 strict accept binaries that request a NaN encoding 2089 supported by the FPU 2090 legacy only accept legacy-NaN binaries, if supported 2091 by the FPU 2092 2008 only accept 2008-NaN binaries, if supported 2093 by the FPU 2094 relaxed accept any binaries regardless of whether 2095 supported by the FPU 2096 emulated accept any binaries but enable FPU emulator 2097 if binary mode is unsupported by the FPU. 2098 2099 The FPU emulator is always able to support both NaN 2100 encodings, so if no FPU hardware is present or it has 2101 been disabled with 'nofpu', then the settings of 2102 'legacy' and '2008' strap the emulator accordingly, 2103 'relaxed' straps the emulator for both legacy-NaN and 2104 2008-NaN, whereas 'strict' enables legacy-NaN only on 2105 legacy processors and both NaN encodings on MIPS32 or 2106 MIPS64 CPUs. 2107 2108 The setting for ABS.fmt/NEG.fmt instruction execution 2109 mode generally follows that for the NaN encoding, 2110 except where unsupported by hardware. 2111 2112 ignore_loglevel [KNL,EARLY] 2113 Ignore loglevel setting - this will print /all/ 2114 kernel messages to the console. Useful for debugging. 2115 We also add it as printk module parameter, so users 2116 could change it dynamically, usually by 2117 /sys/module/printk/parameters/ignore_loglevel. 2118 2119 ignore_rlimit_data 2120 Ignore RLIMIT_DATA setting for data mappings, 2121 print warning at first misuse. Can be changed via 2122 /sys/module/kernel/parameters/ignore_rlimit_data. 2123 2124 ihash_entries= [KNL] 2125 Set number of hash buckets for inode cache. 2126 2127 ima_appraise= [IMA] appraise integrity measurements 2128 Format: { "off" | "enforce" | "fix" | "log" } 2129 default: "enforce" 2130 2131 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead. 2132 The builtin appraise policy appraises all files 2133 owned by uid=0. 2134 2135 ima_canonical_fmt [IMA] 2136 Use the canonical format for the binary runtime 2137 measurements, instead of host native format. 2138 2139 ima_hash= [IMA] 2140 Format: { md5 | sha1 | rmd160 | sha256 | sha384 2141 | sha512 | ... } 2142 default: "sha1" 2143 2144 The list of supported hash algorithms is defined 2145 in crypto/hash_info.h. 2146 2147 ima_policy= [IMA] 2148 The builtin policies to load during IMA setup. 2149 Format: "tcb | appraise_tcb | secure_boot | 2150 fail_securely | critical_data" 2151 2152 The "tcb" policy measures all programs exec'd, files 2153 mmap'd for exec, and all files opened with the read 2154 mode bit set by either the effective uid (euid=0) or 2155 uid=0. 2156 2157 The "appraise_tcb" policy appraises the integrity of 2158 all files owned by root. 2159 2160 The "secure_boot" policy appraises the integrity 2161 of files (eg. kexec kernel image, kernel modules, 2162 firmware, policy, etc) based on file signatures. 2163 2164 The "fail_securely" policy forces file signature 2165 verification failure also on privileged mounted 2166 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE 2167 flag. 2168 2169 The "critical_data" policy measures kernel integrity 2170 critical data. 2171 2172 ima_tcb [IMA] Deprecated. Use ima_policy= instead. 2173 Load a policy which meets the needs of the Trusted 2174 Computing Base. This means IMA will measure all 2175 programs exec'd, files mmap'd for exec, and all files 2176 opened for read by uid=0. 2177 2178 ima_template= [IMA] 2179 Select one of defined IMA measurements template formats. 2180 Formats: { "ima" | "ima-ng" | "ima-ngv2" | "ima-sig" | 2181 "ima-sigv2" } 2182 Default: "ima-ng" 2183 2184 ima_template_fmt= 2185 [IMA] Define a custom template format. 2186 Format: { "field1|...|fieldN" } 2187 2188 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage 2189 Format: <min_file_size> 2190 Set the minimal file size for using asynchronous hash. 2191 If left unspecified, ahash usage is disabled. 2192 2193 ahash performance varies for different data sizes on 2194 different crypto accelerators. This option can be used 2195 to achieve the best performance for a particular HW. 2196 2197 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size 2198 Format: <bufsize> 2199 Set hashing buffer size. Default: 4k. 2200 2201 ahash performance varies for different chunk sizes on 2202 different crypto accelerators. This option can be used 2203 to achieve best performance for particular HW. 2204 2205 indirect_target_selection= [X86,Intel] Mitigation control for Indirect 2206 Target Selection(ITS) bug in Intel CPUs. Updated 2207 microcode is also required for a fix in IBPB. 2208 2209 on: Enable mitigation (default). 2210 off: Disable mitigation. 2211 force: Force the ITS bug and deploy default 2212 mitigation. 2213 vmexit: Only deploy mitigation if CPU is affected by 2214 guest/host isolation part of ITS. 2215 stuff: Deploy RSB-fill mitigation when retpoline is 2216 also deployed. Otherwise, deploy the default 2217 mitigation. 2218 2219 For details see: 2220 Documentation/admin-guide/hw-vuln/indirect-target-selection.rst 2221 2222 init= [KNL] 2223 Format: <full_path> 2224 Run specified binary instead of /sbin/init as init 2225 process. 2226 2227 initcall_debug [KNL] Trace initcalls as they are executed. Useful 2228 for working out where the kernel is dying during 2229 startup. 2230 2231 initcall_blacklist= [KNL] Do not execute a comma-separated list of 2232 initcall functions. Useful for debugging built-in 2233 modules and initcalls. 2234 2235 initramfs_async= [KNL] 2236 Format: <bool> 2237 Default: 1 2238 This parameter controls whether the initramfs 2239 image is unpacked asynchronously, concurrently 2240 with devices being probed and 2241 initialized. This should normally just work, 2242 but as a debugging aid, one can get the 2243 historical behaviour of the initramfs 2244 unpacking being completed before device_ and 2245 late_ initcalls. 2246 2247 initrd= [BOOT,EARLY] Specify the location of the initial ramdisk 2248 2249 initrdmem= [KNL,EARLY] Specify a physical address and size from which to 2250 load the initrd. If an initrd is compiled in or 2251 specified in the bootparams, it takes priority over this 2252 setting. 2253 Format: ss[KMG],nn[KMG] 2254 Default is 0, 0 2255 2256 init_on_alloc= [MM,EARLY] Fill newly allocated pages and heap objects with 2257 zeroes. 2258 Format: 0 | 1 2259 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON. 2260 2261 init_on_free= [MM,EARLY] Fill freed pages and heap objects with zeroes. 2262 Format: 0 | 1 2263 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON. 2264 2265 init_pkru= [X86] Specify the default memory protection keys rights 2266 register contents for all processes. 0x55555554 by 2267 default (disallow access to all but pkey 0). Can 2268 override in debugfs after boot. 2269 2270 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver 2271 Format: <irq> 2272 2273 int_pln_enable [X86] Enable power limit notification interrupt 2274 2275 integrity_audit=[IMA] 2276 Format: { "0" | "1" } 2277 0 -- basic integrity auditing messages. (Default) 2278 1 -- additional integrity auditing messages. 2279 2280 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option 2281 on 2282 Enable intel iommu driver. 2283 off 2284 Disable intel iommu driver. 2285 igfx_off [Default Off] 2286 By default, gfx is mapped as normal device. If a gfx 2287 device has a dedicated DMAR unit, the DMAR unit is 2288 bypassed by not enabling DMAR with this option. In 2289 this case, gfx device will use physical address for 2290 DMA. 2291 strict [Default Off] 2292 Deprecated, equivalent to iommu.strict=1. 2293 sp_off [Default Off] 2294 By default, super page will be supported if Intel IOMMU 2295 has the capability. With this option, super page will 2296 not be supported. 2297 sm_on 2298 Enable the Intel IOMMU scalable mode if the hardware 2299 advertises that it has support for the scalable mode 2300 translation. 2301 sm_off 2302 Disallow use of the Intel IOMMU scalable mode. 2303 tboot_noforce [Default Off] 2304 Do not force the Intel IOMMU enabled under tboot. 2305 By default, tboot will force Intel IOMMU on, which 2306 could harm performance of some high-throughput 2307 devices like 40GBit network cards, even if identity 2308 mapping is enabled. 2309 Note that using this option lowers the security 2310 provided by tboot because it makes the system 2311 vulnerable to DMA attacks. 2312 2313 intel_idle.max_cstate= [KNL,HW,ACPI,X86] 2314 0 disables intel_idle and fall back on acpi_idle. 2315 1 to 9 specify maximum depth of C-state. 2316 2317 intel_pstate= [X86,EARLY] 2318 disable 2319 Do not enable intel_pstate as the default 2320 scaling driver for the supported processors 2321 active 2322 Use intel_pstate driver to bypass the scaling 2323 governors layer of cpufreq and provides it own 2324 algorithms for p-state selection. There are two 2325 P-state selection algorithms provided by 2326 intel_pstate in the active mode: powersave and 2327 performance. The way they both operate depends 2328 on whether or not the hardware managed P-states 2329 (HWP) feature has been enabled in the processor 2330 and possibly on the processor model. 2331 passive 2332 Use intel_pstate as a scaling driver, but configure it 2333 to work with generic cpufreq governors (instead of 2334 enabling its internal governor). This mode cannot be 2335 used along with the hardware-managed P-states (HWP) 2336 feature. 2337 force 2338 Enable intel_pstate on systems that prohibit it by default 2339 in favor of acpi-cpufreq. Forcing the intel_pstate driver 2340 instead of acpi-cpufreq may disable platform features, such 2341 as thermal controls and power capping, that rely on ACPI 2342 P-States information being indicated to OSPM and therefore 2343 should be used with caution. This option does not work with 2344 processors that aren't supported by the intel_pstate driver 2345 or on platforms that use pcc-cpufreq instead of acpi-cpufreq. 2346 no_hwp 2347 Do not enable hardware P state control (HWP) 2348 if available. 2349 hwp_only 2350 Only load intel_pstate on systems which support 2351 hardware P state control (HWP) if available. 2352 support_acpi_ppc 2353 Enforce ACPI _PPC performance limits. If the Fixed ACPI 2354 Description Table, specifies preferred power management 2355 profile as "Enterprise Server" or "Performance Server", 2356 then this feature is turned on by default. 2357 per_cpu_perf_limits 2358 Allow per-logical-CPU P-State performance control limits using 2359 cpufreq sysfs interface 2360 no_cas 2361 Do not enable capacity-aware scheduling (CAS) on 2362 hybrid systems 2363 2364 intremap= [X86-64,Intel-IOMMU,EARLY] 2365 on enable Interrupt Remapping (default) 2366 off disable Interrupt Remapping 2367 nosid disable Source ID checking 2368 no_x2apic_optout 2369 BIOS x2APIC opt-out request will be ignored 2370 nopost disable Interrupt Posting 2371 posted_msi 2372 enable MSIs delivered as posted interrupts 2373 2374 iomem= Disable strict checking of access to MMIO memory 2375 strict regions from userspace. 2376 relaxed 2377 2378 iommu= [X86,EARLY] 2379 2380 off 2381 Don't initialize and use any kind of IOMMU. 2382 2383 force 2384 Force the use of the hardware IOMMU even when 2385 it is not actually needed (e.g. because < 3 GB 2386 memory). 2387 2388 noforce 2389 Don't force hardware IOMMU usage when it is not 2390 needed. (default). 2391 2392 biomerge 2393 panic 2394 nopanic 2395 merge 2396 nomerge 2397 2398 soft 2399 Use software bounce buffering (SWIOTLB) (default for 2400 Intel machines). This can be used to prevent the usage 2401 of an available hardware IOMMU. 2402 2403 [X86] 2404 pt 2405 [X86] 2406 nopt 2407 [PPC/POWERNV] 2408 nobypass 2409 Disable IOMMU bypass, using IOMMU for PCI devices. 2410 2411 [X86] 2412 AMD Gart HW IOMMU-specific options: 2413 2414 <size> 2415 Set the size of the remapping area in bytes. 2416 2417 allowed 2418 Overwrite iommu off workarounds for specific chipsets 2419 2420 fullflush 2421 Flush IOMMU on each allocation (default). 2422 2423 nofullflush 2424 Don't use IOMMU fullflush. 2425 2426 memaper[=<order>] 2427 Allocate an own aperture over RAM with size 2428 32MB<<order. (default: order=1, i.e. 64MB) 2429 2430 merge 2431 Do scatter-gather (SG) merging. Implies "force" 2432 (experimental). 2433 2434 nomerge 2435 Don't do scatter-gather (SG) merging. 2436 2437 noaperture 2438 Ask the IOMMU not to touch the aperture for AGP. 2439 2440 noagp 2441 Don't initialize the AGP driver and use full aperture. 2442 2443 panic 2444 Always panic when IOMMU overflows. 2445 2446 iommu.forcedac= [ARM64,X86,EARLY] Control IOVA allocation for PCI devices. 2447 Format: { "0" | "1" } 2448 0 - Try to allocate a 32-bit DMA address first, before 2449 falling back to the full range if needed. 2450 1 - Allocate directly from the full usable range, 2451 forcing Dual Address Cycle for PCI cards supporting 2452 greater than 32-bit addressing. 2453 2454 iommu.strict= [ARM64,X86,S390,EARLY] Configure TLB invalidation behaviour 2455 Format: { "0" | "1" } 2456 0 - Lazy mode. 2457 Request that DMA unmap operations use deferred 2458 invalidation of hardware TLBs, for increased 2459 throughput at the cost of reduced device isolation. 2460 Will fall back to strict mode if not supported by 2461 the relevant IOMMU driver. 2462 1 - Strict mode. 2463 DMA unmap operations invalidate IOMMU hardware TLBs 2464 synchronously. 2465 unset - Use value of CONFIG_IOMMU_DEFAULT_DMA_{LAZY,STRICT}. 2466 Note: on x86, strict mode specified via one of the 2467 legacy driver-specific options takes precedence. 2468 2469 iommu.passthrough= 2470 [ARM64,X86,EARLY] Configure DMA to bypass the IOMMU by default. 2471 Format: { "0" | "1" } 2472 0 - Use IOMMU translation for DMA. 2473 1 - Bypass the IOMMU for DMA. 2474 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH. 2475 2476 io7= [HW] IO7 for Marvel-based Alpha systems 2477 See comment before marvel_specify_io7 in 2478 arch/alpha/kernel/core_marvel.c. 2479 2480 io_delay= [X86,EARLY] I/O delay method 2481 0x80 2482 Standard port 0x80 based delay 2483 0xed 2484 Alternate port 0xed based delay (needed on some systems) 2485 udelay 2486 Simple two microseconds delay 2487 none 2488 No delay 2489 2490 ip= [IP_PNP] 2491 See Documentation/admin-guide/nfs/nfsroot.rst. 2492 2493 ipcmni_extend [KNL,EARLY] Extend the maximum number of unique System V 2494 IPC identifiers from 32,768 to 16,777,216. 2495 2496 ipe.enforce= [IPE] 2497 Format: <bool> 2498 Determine whether IPE starts in permissive (0) or 2499 enforce (1) mode. The default is enforce. 2500 2501 ipe.success_audit= 2502 [IPE] 2503 Format: <bool> 2504 Start IPE with success auditing enabled, emitting 2505 an audit event when a binary is allowed. The default 2506 is 0. 2507 2508 irqaffinity= [SMP] Set the default irq affinity mask 2509 The argument is a cpu list, as described above. 2510 2511 irqchip.gicv2_force_probe= 2512 [ARM,ARM64,EARLY] 2513 Format: <bool> 2514 Force the kernel to look for the second 4kB page 2515 of a GICv2 controller even if the memory range 2516 exposed by the device tree is too small. 2517 2518 irqchip.gicv3_nolpi= 2519 [ARM,ARM64,EARLY] 2520 Force the kernel to ignore the availability of 2521 LPIs (and by consequence ITSs). Intended for system 2522 that use the kernel as a bootloader, and thus want 2523 to let secondary kernels in charge of setting up 2524 LPIs. 2525 2526 irqchip.gicv3_pseudo_nmi= [ARM64,EARLY] 2527 Enables support for pseudo-NMIs in the kernel. This 2528 requires the kernel to be built with 2529 CONFIG_ARM64_PSEUDO_NMI. 2530 2531 irqfixup [HW] 2532 When an interrupt is not handled search all handlers 2533 for it. Intended to get systems with badly broken 2534 firmware running. 2535 2536 irqpoll [HW] 2537 When an interrupt is not handled search all handlers 2538 for it. Also check all handlers each timer 2539 interrupt. Intended to get systems with badly broken 2540 firmware running. 2541 2542 isapnp= [ISAPNP] 2543 Format: <RDP>,<reset>,<pci_scan>,<verbosity> 2544 2545 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance. 2546 [Deprecated - use cpusets instead] 2547 Format: [flag-list,]<cpu-list> 2548 2549 Specify one or more CPUs to isolate from disturbances 2550 specified in the flag list (default: domain): 2551 2552 nohz 2553 Disable the tick when a single task runs as well as 2554 disabling other kernel noises like having RCU callbacks 2555 offloaded. This is equivalent to the nohz_full parameter. 2556 2557 A residual 1Hz tick is offloaded to workqueues, which you 2558 need to affine to housekeeping through the global 2559 workqueue's affinity configured via the 2560 /sys/devices/virtual/workqueue/cpumask sysfs file, or 2561 by using the 'domain' flag described below. 2562 2563 NOTE: by default the global workqueue runs on all CPUs, 2564 so to protect individual CPUs the 'cpumask' file has to 2565 be configured manually after bootup. 2566 2567 domain 2568 Isolate from the general SMP balancing and scheduling 2569 algorithms. Note that performing domain isolation this way 2570 is irreversible: it's not possible to bring back a CPU to 2571 the domains once isolated through isolcpus. It's strongly 2572 advised to use cpusets instead to disable scheduler load 2573 balancing through the "cpuset.sched_load_balance" file. 2574 It offers a much more flexible interface where CPUs can 2575 move in and out of an isolated set anytime. 2576 2577 You can move a process onto or off an "isolated" CPU via 2578 the CPU affinity syscalls or cpuset. 2579 <cpu number> begins at 0 and the maximum value is 2580 "number of CPUs in system - 1". 2581 2582 managed_irq 2583 2584 Isolate from being targeted by managed interrupts 2585 which have an interrupt mask containing isolated 2586 CPUs. The affinity of managed interrupts is 2587 handled by the kernel and cannot be changed via 2588 the /proc/irq/* interfaces. 2589 2590 This isolation is best effort and only effective 2591 if the automatically assigned interrupt mask of a 2592 device queue contains isolated and housekeeping 2593 CPUs. If housekeeping CPUs are online then such 2594 interrupts are directed to the housekeeping CPU 2595 so that IO submitted on the housekeeping CPU 2596 cannot disturb the isolated CPU. 2597 2598 If a queue's affinity mask contains only isolated 2599 CPUs then this parameter has no effect on the 2600 interrupt routing decision, though interrupts are 2601 only delivered when tasks running on those 2602 isolated CPUs submit IO. IO submitted on 2603 housekeeping CPUs has no influence on those 2604 queues. 2605 2606 The format of <cpu-list> is described above. 2607 2608 iucv= [HW,NET] 2609 2610 ivrs_ioapic [HW,X86-64] 2611 Provide an override to the IOAPIC-ID<->DEVICE-ID 2612 mapping provided in the IVRS ACPI table. 2613 By default, PCI segment is 0, and can be omitted. 2614 2615 For example, to map IOAPIC-ID decimal 10 to 2616 PCI segment 0x1 and PCI device 00:14.0, 2617 write the parameter as: 2618 ivrs_ioapic=10@0001:00:14.0 2619 2620 Deprecated formats: 2621 * To map IOAPIC-ID decimal 10 to PCI device 00:14.0 2622 write the parameter as: 2623 ivrs_ioapic[10]=00:14.0 2624 * To map IOAPIC-ID decimal 10 to PCI segment 0x1 and 2625 PCI device 00:14.0 write the parameter as: 2626 ivrs_ioapic[10]=0001:00:14.0 2627 2628 ivrs_hpet [HW,X86-64] 2629 Provide an override to the HPET-ID<->DEVICE-ID 2630 mapping provided in the IVRS ACPI table. 2631 By default, PCI segment is 0, and can be omitted. 2632 2633 For example, to map HPET-ID decimal 10 to 2634 PCI segment 0x1 and PCI device 00:14.0, 2635 write the parameter as: 2636 ivrs_hpet=10@0001:00:14.0 2637 2638 Deprecated formats: 2639 * To map HPET-ID decimal 0 to PCI device 00:14.0 2640 write the parameter as: 2641 ivrs_hpet[0]=00:14.0 2642 * To map HPET-ID decimal 10 to PCI segment 0x1 and 2643 PCI device 00:14.0 write the parameter as: 2644 ivrs_ioapic[10]=0001:00:14.0 2645 2646 ivrs_acpihid [HW,X86-64] 2647 Provide an override to the ACPI-HID:UID<->DEVICE-ID 2648 mapping provided in the IVRS ACPI table. 2649 By default, PCI segment is 0, and can be omitted. 2650 2651 For example, to map UART-HID:UID AMD0020:0 to 2652 PCI segment 0x1 and PCI device ID 00:14.5, 2653 write the parameter as: 2654 ivrs_acpihid=AMD0020:0@0001:00:14.5 2655 2656 Deprecated formats: 2657 * To map UART-HID:UID AMD0020:0 to PCI segment is 0, 2658 PCI device ID 00:14.5, write the parameter as: 2659 ivrs_acpihid[00:14.5]=AMD0020:0 2660 * To map UART-HID:UID AMD0020:0 to PCI segment 0x1 and 2661 PCI device ID 00:14.5, write the parameter as: 2662 ivrs_acpihid[0001:00:14.5]=AMD0020:0 2663 2664 js= [HW,JOY] Analog joystick 2665 See Documentation/input/joydev/joystick.rst. 2666 2667 kasan_multi_shot 2668 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print 2669 report on every invalid memory access. Without this 2670 parameter KASAN will print report only for the first 2671 invalid access. 2672 2673 keep_bootcon [KNL,EARLY] 2674 Do not unregister boot console at start. This is only 2675 useful for debugging when something happens in the window 2676 between unregistering the boot console and initializing 2677 the real console. 2678 2679 keepinitrd [HW,ARM] See retain_initrd. 2680 2681 kernelcore= [KNL,X86,PPC,EARLY] 2682 Format: nn[KMGTPE] | nn% | "mirror" 2683 This parameter specifies the amount of memory usable by 2684 the kernel for non-movable allocations. The requested 2685 amount is spread evenly throughout all nodes in the 2686 system as ZONE_NORMAL. The remaining memory is used for 2687 movable memory in its own zone, ZONE_MOVABLE. In the 2688 event, a node is too small to have both ZONE_NORMAL and 2689 ZONE_MOVABLE, kernelcore memory will take priority and 2690 other nodes will have a larger ZONE_MOVABLE. 2691 2692 ZONE_MOVABLE is used for the allocation of pages that 2693 may be reclaimed or moved by the page migration 2694 subsystem. Note that allocations like PTEs-from-HighMem 2695 still use the HighMem zone if it exists, and the Normal 2696 zone if it does not. 2697 2698 It is possible to specify the exact amount of memory in 2699 the form of "nn[KMGTPE]", a percentage of total system 2700 memory in the form of "nn%", or "mirror". If "mirror" 2701 option is specified, mirrored (reliable) memory is used 2702 for non-movable allocations and remaining memory is used 2703 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror" 2704 are exclusive, so you cannot specify multiple forms. 2705 2706 kgdbdbgp= [KGDB,HW,EARLY] kgdb over EHCI usb debug port. 2707 Format: <Controller#>[,poll interval] 2708 The controller # is the number of the ehci usb debug 2709 port as it is probed via PCI. The poll interval is 2710 optional and is the number seconds in between 2711 each poll cycle to the debug port in case you need 2712 the functionality for interrupting the kernel with 2713 gdb or control-c on the dbgp connection. When 2714 not using this parameter you use sysrq-g to break into 2715 the kernel debugger. 2716 2717 kgdboc= [KGDB,HW] kgdb over consoles. 2718 Requires a tty driver that supports console polling, 2719 or a supported polling keyboard driver (non-usb). 2720 Serial only format: <serial_device>[,baud] 2721 keyboard only format: kbd 2722 keyboard and serial format: kbd,<serial_device>[,baud] 2723 Optional Kernel mode setting: 2724 kms, kbd format: kms,kbd 2725 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud] 2726 2727 kgdboc_earlycon= [KGDB,HW,EARLY] 2728 If the boot console provides the ability to read 2729 characters and can work in polling mode, you can use 2730 this parameter to tell kgdb to use it as a backend 2731 until the normal console is registered. Intended to 2732 be used together with the kgdboc parameter which 2733 specifies the normal console to transition to. 2734 2735 The name of the early console should be specified 2736 as the value of this parameter. Note that the name of 2737 the early console might be different than the tty 2738 name passed to kgdboc. It's OK to leave the value 2739 blank and the first boot console that implements 2740 read() will be picked. 2741 2742 kgdbwait [KGDB,EARLY] Stop kernel execution and enter the 2743 kernel debugger at the earliest opportunity. 2744 2745 kmac= [MIPS] Korina ethernet MAC address. 2746 Configure the RouterBoard 532 series on-chip 2747 Ethernet adapter MAC address. 2748 2749 kmemleak= [KNL,EARLY] Boot-time kmemleak enable/disable 2750 Valid arguments: on, off 2751 Default: on 2752 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y, 2753 the default is off. 2754 2755 kprobe_event=[probe-list] 2756 [FTRACE] Add kprobe events and enable at boot time. 2757 The probe-list is a semicolon delimited list of probe 2758 definitions. Each definition is same as kprobe_events 2759 interface, but the parameters are comma delimited. 2760 For example, to add a kprobe event on vfs_read with 2761 arg1 and arg2, add to the command line; 2762 2763 kprobe_event=p,vfs_read,$arg1,$arg2 2764 2765 See also Documentation/trace/kprobetrace.rst "Kernel 2766 Boot Parameter" section. 2767 2768 kpti= [ARM64,EARLY] Control page table isolation of 2769 user and kernel address spaces. 2770 Default: enabled on cores which need mitigation. 2771 0: force disabled 2772 1: force enabled 2773 2774 kunit.enable= [KUNIT] Enable executing KUnit tests. Requires 2775 CONFIG_KUNIT to be set to be fully enabled. The 2776 default value can be overridden via 2777 KUNIT_DEFAULT_ENABLED. 2778 Default is 1 (enabled) 2779 2780 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs. 2781 Default is 0 (don't ignore, but inject #GP) 2782 2783 kvm.eager_page_split= 2784 [KVM,X86] Controls whether or not KVM will try to 2785 proactively split all huge pages during dirty logging. 2786 Eager page splitting reduces interruptions to vCPU 2787 execution by eliminating the write-protection faults 2788 and MMU lock contention that would otherwise be 2789 required to split huge pages lazily. 2790 2791 VM workloads that rarely perform writes or that write 2792 only to a small region of VM memory may benefit from 2793 disabling eager page splitting to allow huge pages to 2794 still be used for reads. 2795 2796 The behavior of eager page splitting depends on whether 2797 KVM_DIRTY_LOG_INITIALLY_SET is enabled or disabled. If 2798 disabled, all huge pages in a memslot will be eagerly 2799 split when dirty logging is enabled on that memslot. If 2800 enabled, eager page splitting will be performed during 2801 the KVM_CLEAR_DIRTY ioctl, and only for the pages being 2802 cleared. 2803 2804 Eager page splitting is only supported when kvm.tdp_mmu=Y. 2805 2806 Default is Y (on). 2807 2808 kvm.enable_virt_at_load=[KVM,ARM64,LOONGARCH,MIPS,RISCV,X86] 2809 If enabled, KVM will enable virtualization in hardware 2810 when KVM is loaded, and disable virtualization when KVM 2811 is unloaded (if KVM is built as a module). 2812 2813 If disabled, KVM will dynamically enable and disable 2814 virtualization on-demand when creating and destroying 2815 VMs, i.e. on the 0=>1 and 1=>0 transitions of the 2816 number of VMs. 2817 2818 Enabling virtualization at module load avoids potential 2819 latency for creation of the 0=>1 VM, as KVM serializes 2820 virtualization enabling across all online CPUs. The 2821 "cost" of enabling virtualization when KVM is loaded, 2822 is that doing so may interfere with using out-of-tree 2823 hypervisors that want to "own" virtualization hardware. 2824 2825 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface. 2826 Default is false (don't support). 2827 2828 kvm.nx_huge_pages= 2829 [KVM] Controls the software workaround for the 2830 X86_BUG_ITLB_MULTIHIT bug. 2831 force : Always deploy workaround. 2832 off : Never deploy workaround. 2833 auto : Deploy workaround based on the presence of 2834 X86_BUG_ITLB_MULTIHIT. 2835 2836 Default is 'auto'. 2837 2838 If the software workaround is enabled for the host, 2839 guests do need not to enable it for nested guests. 2840 2841 kvm.nx_huge_pages_recovery_ratio= 2842 [KVM] Controls how many 4KiB pages are periodically zapped 2843 back to huge pages. 0 disables the recovery, otherwise if 2844 the value is N KVM will zap 1/Nth of the 4KiB pages every 2845 period (see below). The default is 60. 2846 2847 kvm.nx_huge_pages_recovery_period_ms= 2848 [KVM] Controls the time period at which KVM zaps 4KiB pages 2849 back to huge pages. If the value is a non-zero N, KVM will 2850 zap a portion (see ratio above) of the pages every N msecs. 2851 If the value is 0 (the default), KVM will pick a period based 2852 on the ratio, such that a page is zapped after 1 hour on average. 2853 2854 kvm-amd.nested= [KVM,AMD] Control nested virtualization feature in 2855 KVM/SVM. Default is 1 (enabled). 2856 2857 kvm-amd.npt= [KVM,AMD] Control KVM's use of Nested Page Tables, 2858 a.k.a. Two-Dimensional Page Tables. Default is 1 2859 (enabled). Disable by KVM if hardware lacks support 2860 for NPT. 2861 2862 kvm-arm.mode= 2863 [KVM,ARM,EARLY] Select one of KVM/arm64's modes of 2864 operation. 2865 2866 none: Forcefully disable KVM. 2867 2868 nvhe: Standard nVHE-based mode, without support for 2869 protected guests. 2870 2871 protected: Mode with support for guests whose state is 2872 kept private from the host, using VHE or 2873 nVHE depending on HW support. 2874 2875 nested: VHE-based mode with support for nested 2876 virtualization. Requires at least ARMv8.4 2877 hardware (with FEAT_NV2). 2878 2879 Defaults to VHE/nVHE based on hardware support. Setting 2880 mode to "protected" will disable kexec and hibernation 2881 for the host. To force nVHE on VHE hardware, add 2882 "arm64_sw.hvhe=0 id_aa64mmfr1.vh=0" to the 2883 command-line. 2884 "nested" is experimental and should be used with 2885 extreme caution. 2886 2887 kvm-arm.vgic_v3_group0_trap= 2888 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-0 2889 system registers 2890 2891 kvm-arm.vgic_v3_group1_trap= 2892 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-1 2893 system registers 2894 2895 kvm-arm.vgic_v3_common_trap= 2896 [KVM,ARM,EARLY] Trap guest accesses to GICv3 common 2897 system registers 2898 2899 kvm-arm.vgic_v4_enable= 2900 [KVM,ARM,EARLY] Allow use of GICv4 for direct 2901 injection of LPIs. 2902 2903 kvm-arm.wfe_trap_policy= 2904 [KVM,ARM] Control when to set WFE instruction trap for 2905 KVM VMs. Traps are allowed but not guaranteed by the 2906 CPU architecture. 2907 2908 trap: set WFE instruction trap 2909 2910 notrap: clear WFE instruction trap 2911 2912 kvm-arm.wfi_trap_policy= 2913 [KVM,ARM] Control when to set WFI instruction trap for 2914 KVM VMs. Traps are allowed but not guaranteed by the 2915 CPU architecture. 2916 2917 trap: set WFI instruction trap 2918 2919 notrap: clear WFI instruction trap 2920 2921 kvm_cma_resv_ratio=n [PPC,EARLY] 2922 Reserves given percentage from system memory area for 2923 contiguous memory allocation for KVM hash pagetable 2924 allocation. 2925 By default it reserves 5% of total system memory. 2926 Format: <integer> 2927 Default: 5 2928 2929 kvm-intel.ept= [KVM,Intel] Control KVM's use of Extended Page Tables, 2930 a.k.a. Two-Dimensional Page Tables. Default is 1 2931 (enabled). Disable by KVM if hardware lacks support 2932 for EPT. 2933 2934 kvm-intel.emulate_invalid_guest_state= 2935 [KVM,Intel] Control whether to emulate invalid guest 2936 state. Ignored if kvm-intel.enable_unrestricted_guest=1, 2937 as guest state is never invalid for unrestricted 2938 guests. This param doesn't apply to nested guests (L2), 2939 as KVM never emulates invalid L2 guest state. 2940 Default is 1 (enabled). 2941 2942 kvm-intel.flexpriority= 2943 [KVM,Intel] Control KVM's use of FlexPriority feature 2944 (TPR shadow). Default is 1 (enabled). Disable by KVM if 2945 hardware lacks support for it. 2946 2947 kvm-intel.nested= 2948 [KVM,Intel] Control nested virtualization feature in 2949 KVM/VMX. Default is 1 (enabled). 2950 2951 kvm-intel.unrestricted_guest= 2952 [KVM,Intel] Control KVM's use of unrestricted guest 2953 feature (virtualized real and unpaged mode). Default 2954 is 1 (enabled). Disable by KVM if EPT is disabled or 2955 hardware lacks support for it. 2956 2957 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault 2958 CVE-2018-3620. 2959 2960 Valid arguments: never, cond, always 2961 2962 always: L1D cache flush on every VMENTER. 2963 cond: Flush L1D on VMENTER only when the code between 2964 VMEXIT and VMENTER can leak host memory. 2965 never: Disables the mitigation 2966 2967 Default is cond (do L1 cache flush in specific instances) 2968 2969 kvm-intel.vpid= [KVM,Intel] Control KVM's use of Virtual Processor 2970 Identification feature (tagged TLBs). Default is 1 2971 (enabled). Disable by KVM if hardware lacks support 2972 for it. 2973 2974 l1d_flush= [X86,INTEL,EARLY] 2975 Control mitigation for L1D based snooping vulnerability. 2976 2977 Certain CPUs are vulnerable to an exploit against CPU 2978 internal buffers which can forward information to a 2979 disclosure gadget under certain conditions. 2980 2981 In vulnerable processors, the speculatively 2982 forwarded data can be used in a cache side channel 2983 attack, to access data to which the attacker does 2984 not have direct access. 2985 2986 This parameter controls the mitigation. The 2987 options are: 2988 2989 on - enable the interface for the mitigation 2990 2991 l1tf= [X86,EARLY] Control mitigation of the L1TF vulnerability on 2992 affected CPUs 2993 2994 The kernel PTE inversion protection is unconditionally 2995 enabled and cannot be disabled. 2996 2997 full 2998 Provides all available mitigations for the 2999 L1TF vulnerability. Disables SMT and 3000 enables all mitigations in the 3001 hypervisors, i.e. unconditional L1D flush. 3002 3003 SMT control and L1D flush control via the 3004 sysfs interface is still possible after 3005 boot. Hypervisors will issue a warning 3006 when the first VM is started in a 3007 potentially insecure configuration, 3008 i.e. SMT enabled or L1D flush disabled. 3009 3010 full,force 3011 Same as 'full', but disables SMT and L1D 3012 flush runtime control. Implies the 3013 'nosmt=force' command line option. 3014 (i.e. sysfs control of SMT is disabled.) 3015 3016 flush 3017 Leaves SMT enabled and enables the default 3018 hypervisor mitigation, i.e. conditional 3019 L1D flush. 3020 3021 SMT control and L1D flush control via the 3022 sysfs interface is still possible after 3023 boot. Hypervisors will issue a warning 3024 when the first VM is started in a 3025 potentially insecure configuration, 3026 i.e. SMT enabled or L1D flush disabled. 3027 3028 flush,nosmt 3029 3030 Disables SMT and enables the default 3031 hypervisor mitigation. 3032 3033 SMT control and L1D flush control via the 3034 sysfs interface is still possible after 3035 boot. Hypervisors will issue a warning 3036 when the first VM is started in a 3037 potentially insecure configuration, 3038 i.e. SMT enabled or L1D flush disabled. 3039 3040 flush,nowarn 3041 Same as 'flush', but hypervisors will not 3042 warn when a VM is started in a potentially 3043 insecure configuration. 3044 3045 off 3046 Disables hypervisor mitigations and doesn't 3047 emit any warnings. 3048 It also drops the swap size and available 3049 RAM limit restriction on both hypervisor and 3050 bare metal. 3051 3052 Default is 'flush'. 3053 3054 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst 3055 3056 l2cr= [PPC] 3057 3058 l3cr= [PPC] 3059 3060 lapic [X86-32,APIC,EARLY] Enable the local APIC even if BIOS 3061 disabled it. 3062 3063 lapic= [X86,APIC] Do not use TSC deadline 3064 value for LAPIC timer one-shot implementation. Default 3065 back to the programmable timer unit in the LAPIC. 3066 Format: notscdeadline 3067 3068 lapic_timer_c2_ok [X86,APIC,EARLY] trust the local apic timer 3069 in C2 power state. 3070 3071 libata.dma= [LIBATA] DMA control 3072 libata.dma=0 Disable all PATA and SATA DMA 3073 libata.dma=1 PATA and SATA Disk DMA only 3074 libata.dma=2 ATAPI (CDROM) DMA only 3075 libata.dma=4 Compact Flash DMA only 3076 Combinations also work, so libata.dma=3 enables DMA 3077 for disks and CDROMs, but not CFs. 3078 3079 libata.ignore_hpa= [LIBATA] Ignore HPA limit 3080 libata.ignore_hpa=0 keep BIOS limits (default) 3081 libata.ignore_hpa=1 ignore limits, using full disk 3082 3083 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume 3084 when set. 3085 Format: <int> 3086 3087 libata.force= [LIBATA] Force configurations. The format is a comma- 3088 separated list of "[ID:]VAL" where ID is PORT[.DEVICE]. 3089 PORT and DEVICE are decimal numbers matching port, link 3090 or device. Basically, it matches the ATA ID string 3091 printed on console by libata. If the whole ID part is 3092 omitted, the last PORT and DEVICE values are used. If 3093 ID hasn't been specified yet, the configuration applies 3094 to all ports, links and devices. 3095 3096 If only DEVICE is omitted, the parameter applies to 3097 the port and all links and devices behind it. DEVICE 3098 number of 0 either selects the first device or the 3099 first fan-out link behind PMP device. It does not 3100 select the host link. DEVICE number of 15 selects the 3101 host link and device attached to it. 3102 3103 The VAL specifies the configuration to force. As long 3104 as there is no ambiguity, shortcut notation is allowed. 3105 For example, both 1.5 and 1.5G would work for 1.5Gbps. 3106 The following configurations can be forced. 3107 3108 * Cable type: 40c, 80c, short40c, unk, ign or sata. 3109 Any ID with matching PORT is used. 3110 3111 * SATA link speed limit: 1.5Gbps or 3.0Gbps. 3112 3113 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7]. 3114 udma[/][16,25,33,44,66,100,133] notation is also 3115 allowed. 3116 3117 * nohrst, nosrst, norst: suppress hard, soft and both 3118 resets. 3119 3120 * rstonce: only attempt one reset during hot-unplug 3121 link recovery. 3122 3123 * [no]dbdelay: Enable or disable the extra 200ms delay 3124 before debouncing a link PHY and device presence 3125 detection. 3126 3127 * [no]ncq: Turn on or off NCQ. 3128 3129 * [no]ncqtrim: Enable or disable queued DSM TRIM. 3130 3131 * [no]ncqati: Enable or disable NCQ trim on ATI chipset. 3132 3133 * [no]trim: Enable or disable (unqueued) TRIM. 3134 3135 * trim_zero: Indicate that TRIM command zeroes data. 3136 3137 * max_trim_128m: Set 128M maximum trim size limit. 3138 3139 * [no]dma: Turn on or off DMA transfers. 3140 3141 * atapi_dmadir: Enable ATAPI DMADIR bridge support. 3142 3143 * atapi_mod16_dma: Enable the use of ATAPI DMA for 3144 commands that are not a multiple of 16 bytes. 3145 3146 * [no]dmalog: Enable or disable the use of the 3147 READ LOG DMA EXT command to access logs. 3148 3149 * [no]iddevlog: Enable or disable access to the 3150 identify device data log. 3151 3152 * [no]logdir: Enable or disable access to the general 3153 purpose log directory. 3154 3155 * max_sec_128: Set transfer size limit to 128 sectors. 3156 3157 * max_sec_1024: Set or clear transfer size limit to 3158 1024 sectors. 3159 3160 * max_sec_lba48: Set or clear transfer size limit to 3161 65535 sectors. 3162 3163 * external: Mark port as external (hotplug-capable). 3164 3165 * [no]lpm: Enable or disable link power management. 3166 3167 * [no]setxfer: Indicate if transfer speed mode setting 3168 should be skipped. 3169 3170 * [no]fua: Disable or enable FUA (Force Unit Access) 3171 support for devices supporting this feature. 3172 3173 * dump_id: Dump IDENTIFY data. 3174 3175 * disable: Disable this device. 3176 3177 If there are multiple matching configurations changing 3178 the same attribute, the last one is used. 3179 3180 load_ramdisk= [RAM] [Deprecated] 3181 3182 lockd.nlm_grace_period=P [NFS] Assign grace period. 3183 Format: <integer> 3184 3185 lockd.nlm_tcpport=N [NFS] Assign TCP port. 3186 Format: <integer> 3187 3188 lockd.nlm_timeout=T [NFS] Assign timeout value. 3189 Format: <integer> 3190 3191 lockd.nlm_udpport=M [NFS] Assign UDP port. 3192 Format: <integer> 3193 3194 lockdown= [SECURITY,EARLY] 3195 { integrity | confidentiality } 3196 Enable the kernel lockdown feature. If set to 3197 integrity, kernel features that allow userland to 3198 modify the running kernel are disabled. If set to 3199 confidentiality, kernel features that allow userland 3200 to extract confidential information from the kernel 3201 are also disabled. 3202 3203 locktorture.acq_writer_lim= [KNL] 3204 Set the time limit in jiffies for a lock 3205 acquisition. Acquisitions exceeding this limit 3206 will result in a splat once they do complete. 3207 3208 locktorture.bind_readers= [KNL] 3209 Specify the list of CPUs to which the readers are 3210 to be bound. 3211 3212 locktorture.bind_writers= [KNL] 3213 Specify the list of CPUs to which the writers are 3214 to be bound. 3215 3216 locktorture.call_rcu_chains= [KNL] 3217 Specify the number of self-propagating call_rcu() 3218 chains to set up. These are used to ensure that 3219 there is a high probability of an RCU grace period 3220 in progress at any given time. Defaults to 0, 3221 which disables these call_rcu() chains. 3222 3223 locktorture.long_hold= [KNL] 3224 Specify the duration in milliseconds for the 3225 occasional long-duration lock hold time. Defaults 3226 to 100 milliseconds. Select 0 to disable. 3227 3228 locktorture.nested_locks= [KNL] 3229 Specify the maximum lock nesting depth that 3230 locktorture is to exercise, up to a limit of 8 3231 (MAX_NESTED_LOCKS). Specify zero to disable. 3232 Note that this parameter is ineffective on types 3233 of locks that do not support nested acquisition. 3234 3235 locktorture.nreaders_stress= [KNL] 3236 Set the number of locking read-acquisition kthreads. 3237 Defaults to being automatically set based on the 3238 number of online CPUs. 3239 3240 locktorture.nwriters_stress= [KNL] 3241 Set the number of locking write-acquisition kthreads. 3242 3243 locktorture.onoff_holdoff= [KNL] 3244 Set time (s) after boot for CPU-hotplug testing. 3245 3246 locktorture.onoff_interval= [KNL] 3247 Set time (s) between CPU-hotplug operations, or 3248 zero to disable CPU-hotplug testing. 3249 3250 locktorture.rt_boost= [KNL] 3251 Do periodic testing of real-time lock priority 3252 boosting. Select 0 to disable, 1 to boost 3253 only rt_mutex, and 2 to boost unconditionally. 3254 Defaults to 2, which might seem to be an 3255 odd choice, but which should be harmless for 3256 non-real-time spinlocks, due to their disabling 3257 of preemption. Note that non-realtime mutexes 3258 disable boosting. 3259 3260 locktorture.rt_boost_factor= [KNL] 3261 Number that determines how often and for how 3262 long priority boosting is exercised. This is 3263 scaled down by the number of writers, so that the 3264 number of boosts per unit time remains roughly 3265 constant as the number of writers increases. 3266 On the other hand, the duration of each boost 3267 increases with the number of writers. 3268 3269 locktorture.shuffle_interval= [KNL] 3270 Set task-shuffle interval (jiffies). Shuffling 3271 tasks allows some CPUs to go into dyntick-idle 3272 mode during the locktorture test. 3273 3274 locktorture.shutdown_secs= [KNL] 3275 Set time (s) after boot system shutdown. This 3276 is useful for hands-off automated testing. 3277 3278 locktorture.stat_interval= [KNL] 3279 Time (s) between statistics printk()s. 3280 3281 locktorture.stutter= [KNL] 3282 Time (s) to stutter testing, for example, 3283 specifying five seconds causes the test to run for 3284 five seconds, wait for five seconds, and so on. 3285 This tests the locking primitive's ability to 3286 transition abruptly to and from idle. 3287 3288 locktorture.torture_type= [KNL] 3289 Specify the locking implementation to test. 3290 3291 locktorture.verbose= [KNL] 3292 Enable additional printk() statements. 3293 3294 locktorture.writer_fifo= [KNL] 3295 Run the write-side locktorture kthreads at 3296 sched_set_fifo() real-time priority. 3297 3298 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver 3299 Format: <irq> 3300 3301 loglevel= [KNL,EARLY] 3302 All Kernel Messages with a loglevel smaller than the 3303 console loglevel will be printed to the console. It can 3304 also be changed with klogd or other programs. The 3305 loglevels are defined as follows: 3306 3307 0 (KERN_EMERG) system is unusable 3308 1 (KERN_ALERT) action must be taken immediately 3309 2 (KERN_CRIT) critical conditions 3310 3 (KERN_ERR) error conditions 3311 4 (KERN_WARNING) warning conditions 3312 5 (KERN_NOTICE) normal but significant condition 3313 6 (KERN_INFO) informational 3314 7 (KERN_DEBUG) debug-level messages 3315 3316 log_buf_len=n[KMG] [KNL,EARLY] 3317 Sets the size of the printk ring buffer, in bytes. 3318 n must be a power of two and greater than the 3319 minimal size. The minimal size is defined by 3320 LOG_BUF_SHIFT kernel config parameter. There 3321 is also CONFIG_LOG_CPU_MAX_BUF_SHIFT config 3322 parameter that allows to increase the default size 3323 depending on the number of CPUs. See init/Kconfig 3324 for more details. 3325 3326 logo.nologo [FB] Disables display of the built-in Linux logo. 3327 This may be used to provide more screen space for 3328 kernel log messages and is useful when debugging 3329 kernel boot problems. 3330 3331 lp=0 [LP] Specify parallel ports to use, e.g, 3332 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses 3333 lp=reset first parallel port). 'lp=0' disables the 3334 lp=auto printer driver. 'lp=reset' (which can be 3335 specified in addition to the ports) causes 3336 attached printers to be reset. Using 3337 lp=port1,port2,... specifies the parallel ports 3338 to associate lp devices with, starting with 3339 lp0. A port specification may be 'none' to skip 3340 that lp device, or a parport name such as 3341 'parport0'. Specifying 'lp=auto' instead of a 3342 port specification list means that device IDs 3343 from each port should be examined, to see if 3344 an IEEE 1284-compliant printer is attached; if 3345 so, the driver will manage that printer. 3346 See also header of drivers/char/lp.c. 3347 3348 lpj=n [KNL] 3349 Sets loops_per_jiffy to given constant, thus avoiding 3350 time-consuming boot-time autodetection (up to 250 ms per 3351 CPU). 0 enables autodetection (default). To determine 3352 the correct value for your kernel, boot with normal 3353 autodetection and see what value is printed. Note that 3354 on SMP systems the preset will be applied to all CPUs, 3355 which is likely to cause problems if your CPUs need 3356 significantly divergent settings. An incorrect value 3357 will cause delays in the kernel to be wrong, leading to 3358 unpredictable I/O errors and other breakage. Although 3359 unlikely, in the extreme case this might damage your 3360 hardware. 3361 3362 lsm.debug [SECURITY] Enable LSM initialization debugging output. 3363 3364 lsm=lsm1,...,lsmN 3365 [SECURITY] Choose order of LSM initialization. This 3366 overrides CONFIG_LSM, and the "security=" parameter. 3367 3368 machtype= [Loongson] Share the same kernel image file between 3369 different yeeloong laptops. 3370 Example: machtype=lemote-yeeloong-2f-7inch 3371 3372 maxcpus= [SMP,EARLY] Maximum number of processors that an SMP kernel 3373 will bring up during bootup. maxcpus=n : n >= 0 limits 3374 the kernel to bring up 'n' processors. Surely after 3375 bootup you can bring up the other plugged cpu by executing 3376 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus 3377 only takes effect during system bootup. 3378 While n=0 is a special case, it is equivalent to "nosmp", 3379 which also disables the IO APIC. 3380 3381 max_loop= [LOOP] The number of loop block devices that get 3382 (loop.max_loop) unconditionally pre-created at init time. The default 3383 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead 3384 of statically allocating a predefined number, loop 3385 devices can be requested on-demand with the 3386 /dev/loop-control interface. 3387 3388 mce= [X86-{32,64}] 3389 3390 Please see Documentation/arch/x86/x86_64/machinecheck.rst for sysfs runtime tunables. 3391 3392 off 3393 disable machine check 3394 3395 no_cmci 3396 disable CMCI(Corrected Machine Check Interrupt) that 3397 Intel processor supports. Usually this disablement is 3398 not recommended, but it might be handy if your 3399 hardware is misbehaving. 3400 3401 Note that you'll get more problems without CMCI than 3402 with due to the shared banks, i.e. you might get 3403 duplicated error logs. 3404 3405 dont_log_ce 3406 don't make logs for corrected errors. All events 3407 reported as corrected are silently cleared by OS. This 3408 option will be useful if you have no interest in any 3409 of corrected errors. 3410 3411 ignore_ce 3412 disable features for corrected errors, e.g. 3413 polling timer and CMCI. All events reported as 3414 corrected are not cleared by OS and remained in its 3415 error banks. 3416 3417 Usually this disablement is not recommended, however 3418 if there is an agent checking/clearing corrected 3419 errors (e.g. BIOS or hardware monitoring 3420 applications), conflicting with OS's error handling, 3421 and you cannot deactivate the agent, then this option 3422 will be a help. 3423 3424 no_lmce 3425 do not opt-in to Local MCE delivery. Use legacy method 3426 to broadcast MCEs. 3427 3428 bootlog 3429 enable logging of machine checks left over from 3430 booting. Disabled by default on AMD Fam10h and older 3431 because some BIOS leave bogus ones. 3432 3433 If your BIOS doesn't do that it's a good idea to 3434 enable though to make sure you log even machine check 3435 events that result in a reboot. On Intel systems it is 3436 enabled by default. 3437 3438 nobootlog 3439 disable boot machine check logging. 3440 3441 monarchtimeout (number) 3442 sets the time in us to wait for other CPUs on machine 3443 checks. 0 to disable. 3444 3445 bios_cmci_threshold 3446 don't overwrite the bios-set CMCI threshold. This boot 3447 option prevents Linux from overwriting the CMCI 3448 threshold set by the bios. Without this option, Linux 3449 always sets the CMCI threshold to 1. Enabling this may 3450 make memory predictive failure analysis less effective 3451 if the bios sets thresholds for memory errors since we 3452 will not see details for all errors. 3453 3454 recovery 3455 force-enable recoverable machine check code paths 3456 3457 Everything else is in sysfs now. 3458 3459 3460 md= [HW] RAID subsystems devices and level 3461 See Documentation/admin-guide/md.rst. 3462 3463 mdacon= [MDA] 3464 Format: <first>,<last> 3465 Specifies range of consoles to be captured by the MDA. 3466 3467 mds= [X86,INTEL,EARLY] 3468 Control mitigation for the Micro-architectural Data 3469 Sampling (MDS) vulnerability. 3470 3471 Certain CPUs are vulnerable to an exploit against CPU 3472 internal buffers which can forward information to a 3473 disclosure gadget under certain conditions. 3474 3475 In vulnerable processors, the speculatively 3476 forwarded data can be used in a cache side channel 3477 attack, to access data to which the attacker does 3478 not have direct access. 3479 3480 This parameter controls the MDS mitigation. The 3481 options are: 3482 3483 full - Enable MDS mitigation on vulnerable CPUs 3484 full,nosmt - Enable MDS mitigation and disable 3485 SMT on vulnerable CPUs 3486 off - Unconditionally disable MDS mitigation 3487 3488 On TAA-affected machines, mds=off can be prevented by 3489 an active TAA mitigation as both vulnerabilities are 3490 mitigated with the same mechanism so in order to disable 3491 this mitigation, you need to specify tsx_async_abort=off 3492 too. 3493 3494 Not specifying this option is equivalent to 3495 mds=full. 3496 3497 For details see: Documentation/admin-guide/hw-vuln/mds.rst 3498 3499 mem=nn[KMG] [HEXAGON,EARLY] Set the memory size. 3500 Must be specified, otherwise memory size will be 0. 3501 3502 mem=nn[KMG] [KNL,BOOT,EARLY] Force usage of a specific amount 3503 of memory Amount of memory to be used in cases 3504 as follows: 3505 3506 1 for test; 3507 2 when the kernel is not able to see the whole system memory; 3508 3 memory that lies after 'mem=' boundary is excluded from 3509 the hypervisor, then assigned to KVM guests. 3510 4 to limit the memory available for kdump kernel. 3511 3512 [ARC,MICROBLAZE] - the limit applies only to low memory, 3513 high memory is not affected. 3514 3515 [ARM64] - only limits memory covered by the linear 3516 mapping. The NOMAP regions are not affected. 3517 3518 [X86] Work as limiting max address. Use together 3519 with memmap= to avoid physical address space collisions. 3520 Without memmap= PCI devices could be placed at addresses 3521 belonging to unused RAM. 3522 3523 Note that this only takes effects during boot time since 3524 in above case 3, memory may need be hot added after boot 3525 if system memory of hypervisor is not sufficient. 3526 3527 mem=nn[KMG]@ss[KMG] 3528 [ARM,MIPS,EARLY] - override the memory layout 3529 reported by firmware. 3530 Define a memory region of size nn[KMG] starting at 3531 ss[KMG]. 3532 Multiple different regions can be specified with 3533 multiple mem= parameters on the command line. 3534 3535 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel 3536 memory. 3537 3538 memblock=debug [KNL,EARLY] Enable memblock debug messages. 3539 3540 memchunk=nn[KMG] 3541 [KNL,SH] Allow user to override the default size for 3542 per-device physically contiguous DMA buffers. 3543 3544 memhp_default_state=online/offline/online_kernel/online_movable 3545 [KNL] Set the initial state for the memory hotplug 3546 onlining policy. If not specified, the default value is 3547 set according to the 3548 CONFIG_MHP_DEFAULT_ONLINE_TYPE kernel config 3549 options. 3550 See Documentation/admin-guide/mm/memory-hotplug.rst. 3551 3552 memmap=exactmap [KNL,X86,EARLY] Enable setting of an exact 3553 E820 memory map, as specified by the user. 3554 Such memmap=exactmap lines can be constructed based on 3555 BIOS output or other requirements. See the memmap=nn@ss 3556 option description. 3557 3558 memmap=nn[KMG]@ss[KMG] 3559 [KNL, X86,MIPS,XTENSA,EARLY] Force usage of a specific region of memory. 3560 Region of memory to be used is from ss to ss+nn. 3561 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG], 3562 which limits max address to nn[KMG]. 3563 Multiple different regions can be specified, 3564 comma delimited. 3565 Example: 3566 memmap=100M@2G,100M#3G,1G!1024G 3567 3568 memmap=nn[KMG]#ss[KMG] 3569 [KNL,ACPI,EARLY] Mark specific memory as ACPI data. 3570 Region of memory to be marked is from ss to ss+nn. 3571 3572 memmap=nn[KMG]$ss[KMG] 3573 [KNL,ACPI,EARLY] Mark specific memory as reserved. 3574 Region of memory to be reserved is from ss to ss+nn. 3575 Example: Exclude memory from 0x18690000-0x1869ffff 3576 memmap=64K$0x18690000 3577 or 3578 memmap=0x10000$0x18690000 3579 Some bootloaders may need an escape character before '$', 3580 like Grub2, otherwise '$' and the following number 3581 will be eaten. 3582 3583 memmap=nn[KMG]!ss[KMG,EARLY] 3584 [KNL,X86] Mark specific memory as protected. 3585 Region of memory to be used, from ss to ss+nn. 3586 The memory region may be marked as e820 type 12 (0xc) 3587 and is NVDIMM or ADR memory. 3588 3589 memmap=<size>%<offset>-<oldtype>+<newtype> 3590 [KNL,ACPI,EARLY] Convert memory within the specified region 3591 from <oldtype> to <newtype>. If "-<oldtype>" is left 3592 out, the whole region will be marked as <newtype>, 3593 even if previously unavailable. If "+<newtype>" is left 3594 out, matching memory will be removed. Types are 3595 specified as e820 types, e.g., 1 = RAM, 2 = reserved, 3596 3 = ACPI, 12 = PRAM. 3597 3598 memory_corruption_check=0/1 [X86,EARLY] 3599 Some BIOSes seem to corrupt the first 64k of 3600 memory when doing things like suspend/resume. 3601 Setting this option will scan the memory 3602 looking for corruption. Enabling this will 3603 both detect corruption and prevent the kernel 3604 from using the memory being corrupted. 3605 However, its intended as a diagnostic tool; if 3606 repeatable BIOS-originated corruption always 3607 affects the same memory, you can use memmap= 3608 to prevent the kernel from using that memory. 3609 3610 memory_corruption_check_size=size [X86,EARLY] 3611 By default it checks for corruption in the low 3612 64k, making this memory unavailable for normal 3613 use. Use this parameter to scan for 3614 corruption in more or less memory. 3615 3616 memory_corruption_check_period=seconds [X86,EARLY] 3617 By default it checks for corruption every 60 3618 seconds. Use this parameter to check at some 3619 other rate. 0 disables periodic checking. 3620 3621 memory_hotplug.memmap_on_memory 3622 [KNL,X86,ARM] Boolean flag to enable this feature. 3623 Format: {on | off (default)} 3624 When enabled, runtime hotplugged memory will 3625 allocate its internal metadata (struct pages, 3626 those vmemmap pages cannot be optimized even 3627 if hugetlb_free_vmemmap is enabled) from the 3628 hotadded memory which will allow to hotadd a 3629 lot of memory without requiring additional 3630 memory to do so. 3631 This feature is disabled by default because it 3632 has some implication on large (e.g. GB) 3633 allocations in some configurations (e.g. small 3634 memory blocks). 3635 The state of the flag can be read in 3636 /sys/module/memory_hotplug/parameters/memmap_on_memory. 3637 Note that even when enabled, there are a few cases where 3638 the feature is not effective. 3639 3640 memtest= [KNL,X86,ARM,M68K,PPC,RISCV,EARLY] Enable memtest 3641 Format: <integer> 3642 default : 0 <disable> 3643 Specifies the number of memtest passes to be 3644 performed. Each pass selects another test 3645 pattern from a given set of patterns. Memtest 3646 fills the memory with this pattern, validates 3647 memory contents and reserves bad memory 3648 regions that are detected. 3649 3650 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control 3651 Valid arguments: on, off 3652 Default: off 3653 mem_encrypt=on: Activate SME 3654 mem_encrypt=off: Do not activate SME 3655 3656 Refer to Documentation/virt/kvm/x86/amd-memory-encryption.rst 3657 for details on when memory encryption can be activated. 3658 3659 mem_sleep_default= [SUSPEND] Default system suspend mode: 3660 s2idle - Suspend-To-Idle 3661 shallow - Power-On Suspend or equivalent (if supported) 3662 deep - Suspend-To-RAM or equivalent (if supported) 3663 See Documentation/admin-guide/pm/sleep-states.rst. 3664 3665 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when 3666 the BIOS has incorrectly applied a workaround. TinyBIOS 3667 version 0.98 is known to be affected, 0.99 fixes the 3668 problem by letting the user disable the workaround. 3669 3670 mga= [HW,DRM] 3671 3672 microcode.force_minrev= [X86] 3673 Format: <bool> 3674 Enable or disable the microcode minimal revision 3675 enforcement for the runtime microcode loader. 3676 3677 mini2440= [ARM,HW,KNL] 3678 Format:[0..2][b][c][t] 3679 Default: "0tb" 3680 MINI2440 configuration specification: 3681 0 - The attached screen is the 3.5" TFT 3682 1 - The attached screen is the 7" TFT 3683 2 - The VGA Shield is attached (1024x768) 3684 Leaving out the screen size parameter will not load 3685 the TFT driver, and the framebuffer will be left 3686 unconfigured. 3687 b - Enable backlight. The TFT backlight pin will be 3688 linked to the kernel VESA blanking code and a GPIO 3689 LED. This parameter is not necessary when using the 3690 VGA shield. 3691 c - Enable the s3c camera interface. 3692 t - Reserved for enabling touchscreen support. The 3693 touchscreen support is not enabled in the mainstream 3694 kernel as of 2.6.30, a preliminary port can be found 3695 in the "bleeding edge" mini2440 support kernel at 3696 https://repo.or.cz/w/linux-2.6/mini2440.git 3697 3698 mitigations= 3699 [X86,PPC,S390,ARM64,EARLY] Control optional mitigations for 3700 CPU vulnerabilities. This is a set of curated, 3701 arch-independent options, each of which is an 3702 aggregation of existing arch-specific options. 3703 3704 Note, "mitigations" is supported if and only if the 3705 kernel was built with CPU_MITIGATIONS=y. 3706 3707 off 3708 Disable all optional CPU mitigations. This 3709 improves system performance, but it may also 3710 expose users to several CPU vulnerabilities. 3711 Equivalent to: if nokaslr then kpti=0 [ARM64] 3712 gather_data_sampling=off [X86] 3713 indirect_target_selection=off [X86] 3714 kvm.nx_huge_pages=off [X86] 3715 l1tf=off [X86] 3716 mds=off [X86] 3717 mmio_stale_data=off [X86] 3718 no_entry_flush [PPC] 3719 no_uaccess_flush [PPC] 3720 nobp=0 [S390] 3721 nopti [X86,PPC] 3722 nospectre_bhb [ARM64] 3723 nospectre_v1 [X86,PPC] 3724 nospectre_v2 [X86,PPC,S390,ARM64] 3725 reg_file_data_sampling=off [X86] 3726 retbleed=off [X86] 3727 spec_rstack_overflow=off [X86] 3728 spec_store_bypass_disable=off [X86,PPC] 3729 spectre_bhi=off [X86] 3730 spectre_v2_user=off [X86] 3731 srbds=off [X86,INTEL] 3732 ssbd=force-off [ARM64] 3733 tsx_async_abort=off [X86] 3734 3735 Exceptions: 3736 This does not have any effect on 3737 kvm.nx_huge_pages when 3738 kvm.nx_huge_pages=force. 3739 3740 auto (default) 3741 Mitigate all CPU vulnerabilities, but leave SMT 3742 enabled, even if it's vulnerable. This is for 3743 users who don't want to be surprised by SMT 3744 getting disabled across kernel upgrades, or who 3745 have other ways of avoiding SMT-based attacks. 3746 Equivalent to: (default behavior) 3747 3748 auto,nosmt 3749 Mitigate all CPU vulnerabilities, disabling SMT 3750 if needed. This is for users who always want to 3751 be fully mitigated, even if it means losing SMT. 3752 Equivalent to: l1tf=flush,nosmt [X86] 3753 mds=full,nosmt [X86] 3754 tsx_async_abort=full,nosmt [X86] 3755 mmio_stale_data=full,nosmt [X86] 3756 retbleed=auto,nosmt [X86] 3757 3758 mminit_loglevel= 3759 [KNL,EARLY] When CONFIG_DEBUG_MEMORY_INIT is set, this 3760 parameter allows control of the logging verbosity for 3761 the additional memory initialisation checks. A value 3762 of 0 disables mminit logging and a level of 4 will 3763 log everything. Information is printed at KERN_DEBUG 3764 so loglevel=8 may also need to be specified. 3765 3766 mmio_stale_data= 3767 [X86,INTEL,EARLY] Control mitigation for the Processor 3768 MMIO Stale Data vulnerabilities. 3769 3770 Processor MMIO Stale Data is a class of 3771 vulnerabilities that may expose data after an MMIO 3772 operation. Exposed data could originate or end in 3773 the same CPU buffers as affected by MDS and TAA. 3774 Therefore, similar to MDS and TAA, the mitigation 3775 is to clear the affected CPU buffers. 3776 3777 This parameter controls the mitigation. The 3778 options are: 3779 3780 full - Enable mitigation on vulnerable CPUs 3781 3782 full,nosmt - Enable mitigation and disable SMT on 3783 vulnerable CPUs. 3784 3785 off - Unconditionally disable mitigation 3786 3787 On MDS or TAA affected machines, 3788 mmio_stale_data=off can be prevented by an active 3789 MDS or TAA mitigation as these vulnerabilities are 3790 mitigated with the same mechanism so in order to 3791 disable this mitigation, you need to specify 3792 mds=off and tsx_async_abort=off too. 3793 3794 Not specifying this option is equivalent to 3795 mmio_stale_data=full. 3796 3797 For details see: 3798 Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst 3799 3800 <module>.async_probe[=<bool>] [KNL] 3801 If no <bool> value is specified or if the value 3802 specified is not a valid <bool>, enable asynchronous 3803 probe on this module. Otherwise, enable/disable 3804 asynchronous probe on this module as indicated by the 3805 <bool> value. See also: module.async_probe 3806 3807 module.async_probe=<bool> 3808 [KNL] When set to true, modules will use async probing 3809 by default. To enable/disable async probing for a 3810 specific module, use the module specific control that 3811 is documented under <module>.async_probe. When both 3812 module.async_probe and <module>.async_probe are 3813 specified, <module>.async_probe takes precedence for 3814 the specific module. 3815 3816 module.enable_dups_trace 3817 [KNL] When CONFIG_MODULE_DEBUG_AUTOLOAD_DUPS is set, 3818 this means that duplicate request_module() calls will 3819 trigger a WARN_ON() instead of a pr_warn(). Note that 3820 if MODULE_DEBUG_AUTOLOAD_DUPS_TRACE is set, WARN_ON()s 3821 will always be issued and this option does nothing. 3822 module.sig_enforce 3823 [KNL] When CONFIG_MODULE_SIG is set, this means that 3824 modules without (valid) signatures will fail to load. 3825 Note that if CONFIG_MODULE_SIG_FORCE is set, that 3826 is always true, so this option does nothing. 3827 3828 module_blacklist= [KNL] Do not load a comma-separated list of 3829 modules. Useful for debugging problem modules. 3830 3831 mousedev.tap_time= 3832 [MOUSE] Maximum time between finger touching and 3833 leaving touchpad surface for touch to be considered 3834 a tap and be reported as a left button click (for 3835 touchpads working in absolute mode only). 3836 Format: <msecs> 3837 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices 3838 reporting absolute coordinates, such as tablets 3839 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices 3840 reporting absolute coordinates, such as tablets 3841 3842 movablecore= [KNL,X86,PPC,EARLY] 3843 Format: nn[KMGTPE] | nn% 3844 This parameter is the complement to kernelcore=, it 3845 specifies the amount of memory used for migratable 3846 allocations. If both kernelcore and movablecore is 3847 specified, then kernelcore will be at *least* the 3848 specified value but may be more. If movablecore on its 3849 own is specified, the administrator must be careful 3850 that the amount of memory usable for all allocations 3851 is not too small. 3852 3853 movable_node [KNL,EARLY] Boot-time switch to make hotplugable memory 3854 NUMA nodes to be movable. This means that the memory 3855 of such nodes will be usable only for movable 3856 allocations which rules out almost all kernel 3857 allocations. Use with caution! 3858 3859 MTD_Partition= [MTD] 3860 Format: <name>,<region-number>,<size>,<offset> 3861 3862 MTD_Region= [MTD] Format: 3863 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>] 3864 3865 mtdparts= [MTD] 3866 See drivers/mtd/parsers/cmdlinepart.c 3867 3868 mtouchusb.raw_coordinates= 3869 [HW] Make the MicroTouch USB driver use raw coordinates 3870 ('y', default) or cooked coordinates ('n') 3871 3872 mtrr=debug [X86,EARLY] 3873 Enable printing debug information related to MTRR 3874 registers at boot time. 3875 3876 mtrr_chunk_size=nn[KMG,X86,EARLY] 3877 used for mtrr cleanup. It is largest continuous chunk 3878 that could hold holes aka. UC entries. 3879 3880 mtrr_gran_size=nn[KMG,X86,EARLY] 3881 Used for mtrr cleanup. It is granularity of mtrr block. 3882 Default is 1. 3883 Large value could prevent small alignment from 3884 using up MTRRs. 3885 3886 mtrr_spare_reg_nr=n [X86,EARLY] 3887 Format: <integer> 3888 Range: 0,7 : spare reg number 3889 Default : 1 3890 Used for mtrr cleanup. It is spare mtrr entries number. 3891 Set to 2 or more if your graphical card needs more. 3892 3893 multitce=off [PPC] This parameter disables the use of the pSeries 3894 firmware feature for updating multiple TCE entries 3895 at a time. 3896 3897 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card 3898 3899 netdev= [NET] Network devices parameters 3900 Format: <irq>,<io>,<mem_start>,<mem_end>,<name> 3901 Note that mem_start is often overloaded to mean 3902 something different and driver-specific. 3903 This usage is only documented in each driver source 3904 file if at all. 3905 3906 netpoll.carrier_timeout= 3907 [NET] Specifies amount of time (in seconds) that 3908 netpoll should wait for a carrier. By default netpoll 3909 waits 4 seconds. 3910 3911 nf_conntrack.acct= 3912 [NETFILTER] Enable connection tracking flow accounting 3913 0 to disable accounting 3914 1 to enable accounting 3915 Default value is 0. 3916 3917 nfs.cache_getent= 3918 [NFS] sets the pathname to the program which is used 3919 to update the NFS client cache entries. 3920 3921 nfs.cache_getent_timeout= 3922 [NFS] sets the timeout after which an attempt to 3923 update a cache entry is deemed to have failed. 3924 3925 nfs.callback_nr_threads= 3926 [NFSv4] set the total number of threads that the 3927 NFS client will assign to service NFSv4 callback 3928 requests. 3929 3930 nfs.callback_tcpport= 3931 [NFS] set the TCP port on which the NFSv4 callback 3932 channel should listen. 3933 3934 nfs.delay_retrans= 3935 [NFS] specifies the number of times the NFSv4 client 3936 retries the request before returning an EAGAIN error, 3937 after a reply of NFS4ERR_DELAY from the server. 3938 Only applies if the softerr mount option is enabled, 3939 and the specified value is >= 0. 3940 3941 nfs.enable_ino64= 3942 [NFS] enable 64-bit inode numbers. 3943 If zero, the NFS client will fake up a 32-bit inode 3944 number for the readdir() and stat() syscalls instead 3945 of returning the full 64-bit number. 3946 The default is to return 64-bit inode numbers. 3947 3948 nfs.idmap_cache_timeout= 3949 [NFS] set the maximum lifetime for idmapper cache 3950 entries. 3951 3952 nfs.max_session_cb_slots= 3953 [NFSv4.1] Sets the maximum number of session 3954 slots the client will assign to the callback 3955 channel. This determines the maximum number of 3956 callbacks the client will process in parallel for 3957 a particular server. 3958 3959 nfs.max_session_slots= 3960 [NFSv4.1] Sets the maximum number of session slots 3961 the client will attempt to negotiate with the server. 3962 This limits the number of simultaneous RPC requests 3963 that the client can send to the NFSv4.1 server. 3964 Note that there is little point in setting this 3965 value higher than the max_tcp_slot_table_limit. 3966 3967 nfs.nfs4_disable_idmapping= 3968 [NFSv4] When set to the default of '1', this option 3969 ensures that both the RPC level authentication 3970 scheme and the NFS level operations agree to use 3971 numeric uids/gids if the mount is using the 3972 'sec=sys' security flavour. In effect it is 3973 disabling idmapping, which can make migration from 3974 legacy NFSv2/v3 systems to NFSv4 easier. 3975 Servers that do not support this mode of operation 3976 will be autodetected by the client, and it will fall 3977 back to using the idmapper. 3978 To turn off this behaviour, set the value to '0'. 3979 3980 nfs.nfs4_unique_id= 3981 [NFS4] Specify an additional fixed unique ident- 3982 ification string that NFSv4 clients can insert into 3983 their nfs_client_id4 string. This is typically a 3984 UUID that is generated at system install time. 3985 3986 nfs.recover_lost_locks= 3987 [NFSv4] Attempt to recover locks that were lost due 3988 to a lease timeout on the server. Please note that 3989 doing this risks data corruption, since there are 3990 no guarantees that the file will remain unchanged 3991 after the locks are lost. 3992 If you want to enable the kernel legacy behaviour of 3993 attempting to recover these locks, then set this 3994 parameter to '1'. 3995 The default parameter value of '0' causes the kernel 3996 not to attempt recovery of lost locks. 3997 3998 nfs.send_implementation_id= 3999 [NFSv4.1] Send client implementation identification 4000 information in exchange_id requests. 4001 If zero, no implementation identification information 4002 will be sent. 4003 The default is to send the implementation identification 4004 information. 4005 4006 nfs4.layoutstats_timer= 4007 [NFSv4.2] Change the rate at which the kernel sends 4008 layoutstats to the pNFS metadata server. 4009 4010 Setting this to value to 0 causes the kernel to use 4011 whatever value is the default set by the layout 4012 driver. A non-zero value sets the minimum interval 4013 in seconds between layoutstats transmissions. 4014 4015 nfsd.inter_copy_offload_enable= 4016 [NFSv4.2] When set to 1, the server will support 4017 server-to-server copies for which this server is 4018 the destination of the copy. 4019 4020 nfsd.nfs4_disable_idmapping= 4021 [NFSv4] When set to the default of '1', the NFSv4 4022 server will return only numeric uids and gids to 4023 clients using auth_sys, and will accept numeric uids 4024 and gids from such clients. This is intended to ease 4025 migration from NFSv2/v3. 4026 4027 nfsd.nfsd4_ssc_umount_timeout= 4028 [NFSv4.2] When used as the destination of a 4029 server-to-server copy, knfsd temporarily mounts 4030 the source server. It caches the mount in case 4031 it will be needed again, and discards it if not 4032 used for the number of milliseconds specified by 4033 this parameter. 4034 4035 nfsaddrs= [NFS] Deprecated. Use ip= instead. 4036 See Documentation/admin-guide/nfs/nfsroot.rst. 4037 4038 nfsroot= [NFS] nfs root filesystem for disk-less boxes. 4039 See Documentation/admin-guide/nfs/nfsroot.rst. 4040 4041 nfsrootdebug [NFS] enable nfsroot debugging messages. 4042 See Documentation/admin-guide/nfs/nfsroot.rst. 4043 4044 nmi_backtrace.backtrace_idle [KNL] 4045 Dump stacks even of idle CPUs in response to an 4046 NMI stack-backtrace request. 4047 4048 nmi_debug= [KNL,SH] Specify one or more actions to take 4049 when a NMI is triggered. 4050 Format: [state][,regs][,debounce][,die] 4051 4052 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels 4053 Format: [panic,][nopanic,][rNNN,][num] 4054 Valid num: 0 or 1 4055 0 - turn hardlockup detector in nmi_watchdog off 4056 1 - turn hardlockup detector in nmi_watchdog on 4057 rNNN - configure the watchdog with raw perf event 0xNNN 4058 4059 When panic is specified, panic when an NMI watchdog 4060 timeout occurs (or 'nopanic' to not panic on an NMI 4061 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set) 4062 To disable both hard and soft lockup detectors, 4063 please see 'nowatchdog'. 4064 This is useful when you use a panic=... timeout and 4065 need the box quickly up again. 4066 4067 These settings can be accessed at runtime via 4068 the nmi_watchdog and hardlockup_panic sysctls. 4069 4070 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths 4071 emulation library even if a 387 maths coprocessor 4072 is present. 4073 4074 no4lvl [RISCV,EARLY] Disable 4-level and 5-level paging modes. 4075 Forces kernel to use 3-level paging instead. 4076 4077 no5lvl [X86-64,RISCV,EARLY] Disable 5-level paging mode. Forces 4078 kernel to use 4-level paging instead. 4079 4080 noalign [KNL,ARM] 4081 4082 noapic [SMP,APIC,EARLY] Tells the kernel to not make use of any 4083 IOAPICs that may be present in the system. 4084 4085 noapictimer [APIC,X86] Don't set up the APIC timer 4086 4087 noautogroup Disable scheduler automatic task group creation. 4088 4089 nocache [ARM,EARLY] 4090 4091 no_console_suspend 4092 [HW] Never suspend the console 4093 Disable suspending of consoles during suspend and 4094 hibernate operations. Once disabled, debugging 4095 messages can reach various consoles while the rest 4096 of the system is being put to sleep (ie, while 4097 debugging driver suspend/resume hooks). This may 4098 not work reliably with all consoles, but is known 4099 to work with serial and VGA consoles. 4100 To facilitate more flexible debugging, we also add 4101 console_suspend, a printk module parameter to control 4102 it. Users could use console_suspend (usually 4103 /sys/module/printk/parameters/console_suspend) to 4104 turn on/off it dynamically. 4105 4106 no_debug_objects 4107 [KNL,EARLY] Disable object debugging 4108 4109 nodsp [SH] Disable hardware DSP at boot time. 4110 4111 noefi [EFI,EARLY] Disable EFI runtime services support. 4112 4113 no_entry_flush [PPC,EARLY] Don't flush the L1-D cache when entering the kernel. 4114 4115 noexec32 [X86-64] 4116 This affects only 32-bit executables. 4117 noexec32=on: enable non-executable mappings (default) 4118 read doesn't imply executable mappings 4119 noexec32=off: disable non-executable mappings 4120 read implies executable mappings 4121 4122 no_file_caps Tells the kernel not to honor file capabilities. The 4123 only way then for a file to be executed with privilege 4124 is to be setuid root or executed by root. 4125 4126 nofpu [MIPS,SH] Disable hardware FPU at boot time. 4127 4128 nofsgsbase [X86] Disables FSGSBASE instructions. 4129 4130 nofxsr [BUGS=X86-32] Disables x86 floating point extended 4131 register save and restore. The kernel will only save 4132 legacy floating-point registers on task switch. 4133 4134 nogbpages [X86] Do not use GB pages for kernel direct mappings. 4135 4136 no_hash_pointers 4137 [KNL,EARLY] 4138 Force pointers printed to the console or buffers to be 4139 unhashed. By default, when a pointer is printed via %p 4140 format string, that pointer is "hashed", i.e. obscured 4141 by hashing the pointer value. This is a security feature 4142 that hides actual kernel addresses from unprivileged 4143 users, but it also makes debugging the kernel more 4144 difficult since unequal pointers can no longer be 4145 compared. However, if this command-line option is 4146 specified, then all normal pointers will have their true 4147 value printed. This option should only be specified when 4148 debugging the kernel. Please do not use on production 4149 kernels. 4150 4151 nohibernate [HIBERNATION] Disable hibernation and resume. 4152 4153 nohlt [ARM,ARM64,MICROBLAZE,MIPS,PPC,RISCV,SH] Forces the kernel to 4154 busy wait in do_idle() and not use the arch_cpu_idle() 4155 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP 4156 to be effective. This is useful on platforms where the 4157 sleep(SH) or wfi(ARM,ARM64) instructions do not work 4158 correctly or when doing power measurements to evaluate 4159 the impact of the sleep instructions. This is also 4160 useful when using JTAG debugger. 4161 4162 nohpet [X86] Don't use the HPET timer. 4163 4164 nohugeiomap [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge I/O mappings. 4165 4166 nohugevmalloc [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge vmalloc mappings. 4167 4168 nohz= [KNL] Boottime enable/disable dynamic ticks 4169 Valid arguments: on, off 4170 Default: on 4171 4172 nohz_full= [KNL,BOOT,SMP,ISOL] 4173 The argument is a cpu list, as described above. 4174 In kernels built with CONFIG_NO_HZ_FULL=y, set 4175 the specified list of CPUs whose tick will be stopped 4176 whenever possible. The boot CPU will be forced outside 4177 the range to maintain the timekeeping. Any CPUs 4178 in this list will have their RCU callbacks offloaded, 4179 just as if they had also been called out in the 4180 rcu_nocbs= boot parameter. 4181 4182 Note that this argument takes precedence over 4183 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option. 4184 4185 noinitrd [RAM] Tells the kernel not to load any configured 4186 initial RAM disk. 4187 4188 nointremap [X86-64,Intel-IOMMU,EARLY] Do not enable interrupt 4189 remapping. 4190 [Deprecated - use intremap=off] 4191 4192 noinvpcid [X86,EARLY] Disable the INVPCID cpu feature. 4193 4194 noiotrap [SH] Disables trapped I/O port accesses. 4195 4196 noirqdebug [X86-32] Disables the code which attempts to detect and 4197 disable unhandled interrupt sources. 4198 4199 noisapnp [ISAPNP] Disables ISA PnP code. 4200 4201 nokaslr [KNL,EARLY] 4202 When CONFIG_RANDOMIZE_BASE is set, this disables 4203 kernel and module base offset ASLR (Address Space 4204 Layout Randomization). 4205 4206 no-kvmapf [X86,KVM,EARLY] Disable paravirtualized asynchronous page 4207 fault handling. 4208 4209 no-kvmclock [X86,KVM,EARLY] Disable paravirtualized KVM clock driver 4210 4211 nolapic [X86-32,APIC,EARLY] Do not enable or use the local APIC. 4212 4213 nolapic_timer [X86-32,APIC,EARLY] Do not use the local APIC timer. 4214 4215 nomce [X86-32] Disable Machine Check Exception 4216 4217 nomfgpt [X86-32] Disable Multi-Function General Purpose 4218 Timer usage (for AMD Geode machines). 4219 4220 nomodeset Disable kernel modesetting. Most systems' firmware 4221 sets up a display mode and provides framebuffer memory 4222 for output. With nomodeset, DRM and fbdev drivers will 4223 not load if they could possibly displace the pre- 4224 initialized output. Only the system framebuffer will 4225 be available for use. The respective drivers will not 4226 perform display-mode changes or accelerated rendering. 4227 4228 Useful as error fallback, or for testing and debugging. 4229 4230 nomodule Disable module load 4231 4232 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to 4233 shutdown the other cpus. Instead use the REBOOT_VECTOR 4234 irq. 4235 4236 nopat [X86,EARLY] Disable PAT (page attribute table extension of 4237 pagetables) support. 4238 4239 nopcid [X86-64,EARLY] Disable the PCID cpu feature. 4240 4241 nopku [X86] Disable Memory Protection Keys CPU feature found 4242 in some Intel CPUs. 4243 4244 nopti [X86-64,EARLY] 4245 Equivalent to pti=off 4246 4247 nopv= [X86,XEN,KVM,HYPER_V,VMWARE,EARLY] 4248 Disables the PV optimizations forcing the guest to run 4249 as generic guest with no PV drivers. Currently support 4250 XEN HVM, KVM, HYPER_V and VMWARE guest. 4251 4252 nopvspin [X86,XEN,KVM,EARLY] 4253 Disables the qspinlock slow path using PV optimizations 4254 which allow the hypervisor to 'idle' the guest on lock 4255 contention. 4256 4257 norandmaps Don't use address space randomization. Equivalent to 4258 echo 0 > /proc/sys/kernel/randomize_va_space 4259 4260 noreplace-smp [X86-32,SMP] Don't replace SMP instructions 4261 with UP alternatives 4262 4263 noresume [SWSUSP] Disables resume and restores original swap 4264 space. 4265 4266 no-scroll [VGA] Disables scrollback. 4267 This is required for the Braillex ib80-piezo Braille 4268 reader made by F.H. Papenmeier (Germany). 4269 4270 nosgx [X86-64,SGX,EARLY] Disables Intel SGX kernel support. 4271 4272 nosmap [PPC,EARLY] 4273 Disable SMAP (Supervisor Mode Access Prevention) 4274 even if it is supported by processor. 4275 4276 nosmep [PPC64s,EARLY] 4277 Disable SMEP (Supervisor Mode Execution Prevention) 4278 even if it is supported by processor. 4279 4280 nosmp [SMP,EARLY] Tells an SMP kernel to act as a UP kernel, 4281 and disable the IO APIC. legacy for "maxcpus=0". 4282 4283 nosmt [KNL,MIPS,PPC,EARLY] Disable symmetric multithreading (SMT). 4284 Equivalent to smt=1. 4285 4286 [KNL,X86,PPC,S390] Disable symmetric multithreading (SMT). 4287 nosmt=force: Force disable SMT, cannot be undone 4288 via the sysfs control file. 4289 4290 nosoftlockup [KNL] Disable the soft-lockup detector. 4291 4292 nospec_store_bypass_disable 4293 [HW,EARLY] Disable all mitigations for the Speculative 4294 Store Bypass vulnerability 4295 4296 nospectre_bhb [ARM64,EARLY] Disable all mitigations for Spectre-BHB (branch 4297 history injection) vulnerability. System may allow data leaks 4298 with this option. 4299 4300 nospectre_v1 [X86,PPC,EARLY] Disable mitigations for Spectre Variant 1 4301 (bounds check bypass). With this option data leaks are 4302 possible in the system. 4303 4304 nospectre_v2 [X86,PPC_E500,ARM64,EARLY] Disable all mitigations 4305 for the Spectre variant 2 (indirect branch 4306 prediction) vulnerability. System may allow data 4307 leaks with this option. 4308 4309 no-steal-acc [X86,PV_OPS,ARM64,PPC/PSERIES,RISCV,LOONGARCH,EARLY] 4310 Disable paravirtualized steal time accounting. steal time 4311 is computed, but won't influence scheduler behaviour 4312 4313 nosync [HW,M68K] Disables sync negotiation for all devices. 4314 4315 no_timer_check [X86,APIC] Disables the code which tests for broken 4316 timer IRQ sources, i.e., the IO-APIC timer. This can 4317 work around problems with incorrect timer 4318 initialization on some boards. 4319 4320 no_uaccess_flush 4321 [PPC,EARLY] Don't flush the L1-D cache after accessing user data. 4322 4323 novmcoredd [KNL,KDUMP] 4324 Disable device dump. Device dump allows drivers to 4325 append dump data to vmcore so you can collect driver 4326 specified debug info. Drivers can append the data 4327 without any limit and this data is stored in memory, 4328 so this may cause significant memory stress. Disabling 4329 device dump can help save memory but the driver debug 4330 data will be no longer available. This parameter 4331 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP 4332 is set. 4333 4334 no-vmw-sched-clock 4335 [X86,PV_OPS,EARLY] Disable paravirtualized VMware 4336 scheduler clock and use the default one. 4337 4338 nowatchdog [KNL] Disable both lockup detectors, i.e. 4339 soft-lockup and NMI watchdog (hard-lockup). 4340 4341 nowb [ARM,EARLY] 4342 4343 nox2apic [X86-64,APIC,EARLY] Do not enable x2APIC mode. 4344 4345 NOTE: this parameter will be ignored on systems with the 4346 LEGACY_XAPIC_DISABLED bit set in the 4347 IA32_XAPIC_DISABLE_STATUS MSR. 4348 4349 noxsave [BUGS=X86] Disables x86 extended register state save 4350 and restore using xsave. The kernel will fallback to 4351 enabling legacy floating-point and sse state. 4352 4353 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended 4354 register states. The kernel will fall back to use 4355 xsave to save the states. By using this parameter, 4356 performance of saving the states is degraded because 4357 xsave doesn't support modified optimization while 4358 xsaveopt supports it on xsaveopt enabled systems. 4359 4360 noxsaves [X86] Disables xsaves and xrstors used in saving and 4361 restoring x86 extended register state in compacted 4362 form of xsave area. The kernel will fall back to use 4363 xsaveopt and xrstor to save and restore the states 4364 in standard form of xsave area. By using this 4365 parameter, xsave area per process might occupy more 4366 memory on xsaves enabled systems. 4367 4368 nr_cpus= [SMP,EARLY] Maximum number of processors that an SMP kernel 4369 could support. nr_cpus=n : n >= 1 limits the kernel to 4370 support 'n' processors. It could be larger than the 4371 number of already plugged CPU during bootup, later in 4372 runtime you can physically add extra cpu until it reaches 4373 n. So during boot up some boot time memory for per-cpu 4374 variables need be pre-allocated for later physical cpu 4375 hot plugging. 4376 4377 nr_uarts= [SERIAL] maximum number of UARTs to be registered. 4378 4379 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86, EARLY] 4380 Disable NUMA, Only set up a single NUMA node 4381 spanning all memory. 4382 4383 numa=fake=<size>[MG] 4384 [KNL, ARM64, RISCV, X86, EARLY] 4385 If given as a memory unit, fills all system RAM with 4386 nodes of size interleaved over physical nodes. 4387 4388 numa=fake=<N> 4389 [KNL, ARM64, RISCV, X86, EARLY] 4390 If given as an integer, fills all system RAM with N 4391 fake nodes interleaved over physical nodes. 4392 4393 numa=fake=<N>U 4394 [KNL, ARM64, RISCV, X86, EARLY] 4395 If given as an integer followed by 'U', it will 4396 divide each physical node into N emulated nodes. 4397 4398 numa=noacpi [X86] Don't parse the SRAT table for NUMA setup 4399 4400 numa=nohmat [X86] Don't parse the HMAT table for NUMA setup, or 4401 soft-reserved memory partitioning. 4402 4403 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic 4404 NUMA balancing. 4405 Allowed values are enable and disable 4406 4407 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA. 4408 'node', 'default' can be specified 4409 This can be set from sysctl after boot. 4410 See Documentation/admin-guide/sysctl/vm.rst for details. 4411 4412 ohci1394_dma=early [HW,EARLY] enable debugging via the ohci1394 driver. 4413 See Documentation/core-api/debugging-via-ohci1394.rst for more 4414 info. 4415 4416 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands 4417 Rather than timing out after 20 ms if an EC 4418 command is not properly ACKed, override the length 4419 of the timeout. We have interrupts disabled while 4420 waiting for the ACK, so if this is set too high 4421 interrupts *may* be lost! 4422 4423 omap_mux= [OMAP] Override bootloader pin multiplexing. 4424 Format: <mux_mode0.mode_name=value>... 4425 For example, to override I2C bus2: 4426 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100 4427 4428 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration 4429 4430 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock] 4431 4432 boundary - index of last SLC block on Flex-OneNAND. 4433 The remaining blocks are configured as MLC blocks. 4434 lock - Configure if Flex-OneNAND boundary should be locked. 4435 Once locked, the boundary cannot be changed. 4436 1 indicates lock status, 0 indicates unlock status. 4437 4438 oops=panic [KNL,EARLY] 4439 Always panic on oopses. Default is to just kill the 4440 process, but there is a small probability of 4441 deadlocking the machine. 4442 This will also cause panics on machine check exceptions. 4443 Useful together with panic=30 to trigger a reboot. 4444 4445 page_alloc.shuffle= 4446 [KNL] Boolean flag to control whether the page allocator 4447 should randomize its free lists. This parameter can be 4448 used to enable/disable page randomization. The state of 4449 the flag can be read from sysfs at: 4450 /sys/module/page_alloc/parameters/shuffle. 4451 This parameter is only available if CONFIG_SHUFFLE_PAGE_ALLOCATOR=y. 4452 4453 page_owner= [KNL,EARLY] Boot-time page_owner enabling option. 4454 Storage of the information about who allocated 4455 each page is disabled in default. With this switch, 4456 we can turn it on. 4457 on: enable the feature 4458 4459 page_poison= [KNL,EARLY] Boot-time parameter changing the state of 4460 poisoning on the buddy allocator, available with 4461 CONFIG_PAGE_POISONING=y. 4462 off: turn off poisoning (default) 4463 on: turn on poisoning 4464 4465 page_reporting.page_reporting_order= 4466 [KNL] Minimal page reporting order 4467 Format: <integer> 4468 Adjust the minimal page reporting order. The page 4469 reporting is disabled when it exceeds MAX_PAGE_ORDER. 4470 4471 panic= [KNL] Kernel behaviour on panic: delay <timeout> 4472 timeout > 0: seconds before rebooting 4473 timeout = 0: wait forever 4474 timeout < 0: reboot immediately 4475 Format: <timeout> 4476 4477 panic_on_taint= [KNL,EARLY] 4478 Bitmask for conditionally calling panic() in add_taint() 4479 Format: <hex>[,nousertaint] 4480 Hexadecimal bitmask representing the set of TAINT flags 4481 that will cause the kernel to panic when add_taint() is 4482 called with any of the flags in this set. 4483 The optional switch "nousertaint" can be utilized to 4484 prevent userspace forced crashes by writing to sysctl 4485 /proc/sys/kernel/tainted any flagset matching with the 4486 bitmask set on panic_on_taint. 4487 See Documentation/admin-guide/tainted-kernels.rst for 4488 extra details on the taint flags that users can pick 4489 to compose the bitmask to assign to panic_on_taint. 4490 4491 panic_on_warn=1 panic() instead of WARN(). Useful to cause kdump 4492 on a WARN(). 4493 4494 panic_print= Bitmask for printing system info when panic happens. 4495 User can chose combination of the following bits: 4496 bit 0: print all tasks info 4497 bit 1: print system memory info 4498 bit 2: print timer info 4499 bit 3: print locks info if CONFIG_LOCKDEP is on 4500 bit 4: print ftrace buffer 4501 bit 5: print all printk messages in buffer 4502 bit 6: print all CPUs backtrace (if available in the arch) 4503 bit 7: print only tasks in uninterruptible (blocked) state 4504 *Be aware* that this option may print a _lot_ of lines, 4505 so there are risks of losing older messages in the log. 4506 Use this option carefully, maybe worth to setup a 4507 bigger log buffer with "log_buf_len" along with this. 4508 4509 parkbd.port= [HW] Parallel port number the keyboard adapter is 4510 connected to, default is 0. 4511 Format: <parport#> 4512 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation, 4513 0 for XT, 1 for AT (default is AT). 4514 Format: <mode> 4515 4516 parport= [HW,PPT] Specify parallel ports. 0 disables. 4517 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] } 4518 Use 'auto' to force the driver to use any 4519 IRQ/DMA settings detected (the default is to 4520 ignore detected IRQ/DMA settings because of 4521 possible conflicts). You can specify the base 4522 address, IRQ, and DMA settings; IRQ and DMA 4523 should be numbers, or 'auto' (for using detected 4524 settings on that particular port), or 'nofifo' 4525 (to avoid using a FIFO even if it is detected). 4526 Parallel ports are assigned in the order they 4527 are specified on the command line, starting 4528 with parport0. 4529 4530 parport_init_mode= [HW,PPT] 4531 Configure VIA parallel port to operate in 4532 a specific mode. This is necessary on Pegasos 4533 computer where firmware has no options for setting 4534 up parallel port mode and sets it to spp. 4535 Currently this function knows 686a and 8231 chips. 4536 Format: [spp|ps2|epp|ecp|ecpepp] 4537 4538 pata_legacy.all= [HW,LIBATA] 4539 Format: <int> 4540 Set to non-zero to probe primary and secondary ISA 4541 port ranges on PCI systems where no PCI PATA device 4542 has been found at either range. Disabled by default. 4543 4544 pata_legacy.autospeed= [HW,LIBATA] 4545 Format: <int> 4546 Set to non-zero if a chip is present that snoops speed 4547 changes. Disabled by default. 4548 4549 pata_legacy.ht6560a= [HW,LIBATA] 4550 Format: <int> 4551 Set to 1, 2, or 3 for HT 6560A on the primary channel, 4552 the secondary channel, or both channels respectively. 4553 Disabled by default. 4554 4555 pata_legacy.ht6560b= [HW,LIBATA] 4556 Format: <int> 4557 Set to 1, 2, or 3 for HT 6560B on the primary channel, 4558 the secondary channel, or both channels respectively. 4559 Disabled by default. 4560 4561 pata_legacy.iordy_mask= [HW,LIBATA] 4562 Format: <int> 4563 IORDY enable mask. Set individual bits to allow IORDY 4564 for the respective channel. Bit 0 is for the first 4565 legacy channel handled by this driver, bit 1 is for 4566 the second channel, and so on. The sequence will often 4567 correspond to the primary legacy channel, the secondary 4568 legacy channel, and so on, but the handling of a PCI 4569 bus and the use of other driver options may interfere 4570 with the sequence. By default IORDY is allowed across 4571 all channels. 4572 4573 pata_legacy.opti82c46x= [HW,LIBATA] 4574 Format: <int> 4575 Set to 1, 2, or 3 for Opti 82c611A on the primary 4576 channel, the secondary channel, or both channels 4577 respectively. Disabled by default. 4578 4579 pata_legacy.opti82c611a= [HW,LIBATA] 4580 Format: <int> 4581 Set to 1, 2, or 3 for Opti 82c465MV on the primary 4582 channel, the secondary channel, or both channels 4583 respectively. Disabled by default. 4584 4585 pata_legacy.pio_mask= [HW,LIBATA] 4586 Format: <int> 4587 PIO mode mask for autospeed devices. Set individual 4588 bits to allow the use of the respective PIO modes. 4589 Bit 0 is for mode 0, bit 1 is for mode 1, and so on. 4590 All modes allowed by default. 4591 4592 pata_legacy.probe_all= [HW,LIBATA] 4593 Format: <int> 4594 Set to non-zero to probe tertiary and further ISA 4595 port ranges on PCI systems. Disabled by default. 4596 4597 pata_legacy.probe_mask= [HW,LIBATA] 4598 Format: <int> 4599 Probe mask for legacy ISA PATA ports. Depending on 4600 platform configuration and the use of other driver 4601 options up to 6 legacy ports are supported: 0x1f0, 4602 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing 4603 of individual ports can be disabled by setting the 4604 corresponding bits in the mask to 1. Bit 0 is for 4605 the first port in the list above (0x1f0), and so on. 4606 By default all supported ports are probed. 4607 4608 pata_legacy.qdi= [HW,LIBATA] 4609 Format: <int> 4610 Set to non-zero to probe QDI controllers. By default 4611 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise. 4612 4613 pata_legacy.winbond= [HW,LIBATA] 4614 Format: <int> 4615 Set to non-zero to probe Winbond controllers. Use 4616 the standard I/O port (0x130) if 1, otherwise the 4617 value given is the I/O port to use (typically 0x1b0). 4618 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE, 4619 0 otherwise. 4620 4621 pata_platform.pio_mask= [HW,LIBATA] 4622 Format: <int> 4623 Supported PIO mode mask. Set individual bits to allow 4624 the use of the respective PIO modes. Bit 0 is for 4625 mode 0, bit 1 is for mode 1, and so on. Mode 0 only 4626 allowed by default. 4627 4628 pause_on_oops=<int> 4629 Halt all CPUs after the first oops has been printed for 4630 the specified number of seconds. This is to be used if 4631 your oopses keep scrolling off the screen. 4632 4633 pcbit= [HW,ISDN] 4634 4635 pci=option[,option...] [PCI,EARLY] various PCI subsystem options. 4636 4637 Some options herein operate on a specific device 4638 or a set of devices (<pci_dev>). These are 4639 specified in one of the following formats: 4640 4641 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]* 4642 pci:<vendor>:<device>[:<subvendor>:<subdevice>] 4643 4644 Note: the first format specifies a PCI 4645 bus/device/function address which may change 4646 if new hardware is inserted, if motherboard 4647 firmware changes, or due to changes caused 4648 by other kernel parameters. If the 4649 domain is left unspecified, it is 4650 taken to be zero. Optionally, a path 4651 to a device through multiple device/function 4652 addresses can be specified after the base 4653 address (this is more robust against 4654 renumbering issues). The second format 4655 selects devices using IDs from the 4656 configuration space which may match multiple 4657 devices in the system. 4658 4659 earlydump dump PCI config space before the kernel 4660 changes anything 4661 off [X86] don't probe for the PCI bus 4662 bios [X86-32] force use of PCI BIOS, don't access 4663 the hardware directly. Use this if your machine 4664 has a non-standard PCI host bridge. 4665 nobios [X86-32] disallow use of PCI BIOS, only direct 4666 hardware access methods are allowed. Use this 4667 if you experience crashes upon bootup and you 4668 suspect they are caused by the BIOS. 4669 conf1 [X86] Force use of PCI Configuration Access 4670 Mechanism 1 (config address in IO port 0xCF8, 4671 data in IO port 0xCFC, both 32-bit). 4672 conf2 [X86] Force use of PCI Configuration Access 4673 Mechanism 2 (IO port 0xCF8 is an 8-bit port for 4674 the function, IO port 0xCFA, also 8-bit, sets 4675 bus number. The config space is then accessed 4676 through ports 0xC000-0xCFFF). 4677 See http://wiki.osdev.org/PCI for more info 4678 on the configuration access mechanisms. 4679 noaer [PCIE] If the PCIEAER kernel config parameter is 4680 enabled, this kernel boot option can be used to 4681 disable the use of PCIE advanced error reporting. 4682 nodomains [PCI] Disable support for multiple PCI 4683 root domains (aka PCI segments, in ACPI-speak). 4684 nommconf [X86] Disable use of MMCONFIG for PCI 4685 Configuration 4686 check_enable_amd_mmconf [X86] check for and enable 4687 properly configured MMIO access to PCI 4688 config space on AMD family 10h CPU 4689 nomsi [MSI] If the PCI_MSI kernel config parameter is 4690 enabled, this kernel boot option can be used to 4691 disable the use of MSI interrupts system-wide. 4692 noioapicquirk [APIC] Disable all boot interrupt quirks. 4693 Safety option to keep boot IRQs enabled. This 4694 should never be necessary. 4695 ioapicreroute [APIC] Enable rerouting of boot IRQs to the 4696 primary IO-APIC for bridges that cannot disable 4697 boot IRQs. This fixes a source of spurious IRQs 4698 when the system masks IRQs. 4699 noioapicreroute [APIC] Disable workaround that uses the 4700 boot IRQ equivalent of an IRQ that connects to 4701 a chipset where boot IRQs cannot be disabled. 4702 The opposite of ioapicreroute. 4703 biosirq [X86-32] Use PCI BIOS calls to get the interrupt 4704 routing table. These calls are known to be buggy 4705 on several machines and they hang the machine 4706 when used, but on other computers it's the only 4707 way to get the interrupt routing table. Try 4708 this option if the kernel is unable to allocate 4709 IRQs or discover secondary PCI buses on your 4710 motherboard. 4711 rom [X86] Assign address space to expansion ROMs. 4712 Use with caution as certain devices share 4713 address decoders between ROMs and other 4714 resources. 4715 norom [X86] Do not assign address space to 4716 expansion ROMs that do not already have 4717 BIOS assigned address ranges. 4718 nobar [X86] Do not assign address space to the 4719 BARs that weren't assigned by the BIOS. 4720 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be 4721 assigned automatically to PCI devices. You can 4722 make the kernel exclude IRQs of your ISA cards 4723 this way. 4724 pirqaddr=0xAAAAA [X86] Specify the physical address 4725 of the PIRQ table (normally generated 4726 by the BIOS) if it is outside the 4727 F0000h-100000h range. 4728 lastbus=N [X86] Scan all buses thru bus #N. Can be 4729 useful if the kernel is unable to find your 4730 secondary buses and you want to tell it 4731 explicitly which ones they are. 4732 assign-busses [X86] Always assign all PCI bus 4733 numbers ourselves, overriding 4734 whatever the firmware may have done. 4735 usepirqmask [X86] Honor the possible IRQ mask stored 4736 in the BIOS $PIR table. This is needed on 4737 some systems with broken BIOSes, notably 4738 some HP Pavilion N5400 and Omnibook XE3 4739 notebooks. This will have no effect if ACPI 4740 IRQ routing is enabled. 4741 noacpi [X86] Do not use ACPI for IRQ routing 4742 or for PCI scanning. 4743 use_crs [X86] Use PCI host bridge window information 4744 from ACPI. On BIOSes from 2008 or later, this 4745 is enabled by default. If you need to use this, 4746 please report a bug. 4747 nocrs [X86] Ignore PCI host bridge windows from ACPI. 4748 If you need to use this, please report a bug. 4749 use_e820 [X86] Use E820 reservations to exclude parts of 4750 PCI host bridge windows. This is a workaround 4751 for BIOS defects in host bridge _CRS methods. 4752 If you need to use this, please report a bug to 4753 <linux-pci@vger.kernel.org>. 4754 no_e820 [X86] Ignore E820 reservations for PCI host 4755 bridge windows. This is the default on modern 4756 hardware. If you need to use this, please report 4757 a bug to <linux-pci@vger.kernel.org>. 4758 routeirq Do IRQ routing for all PCI devices. 4759 This is normally done in pci_enable_device(), 4760 so this option is a temporary workaround 4761 for broken drivers that don't call it. 4762 skip_isa_align [X86] do not align io start addr, so can 4763 handle more pci cards 4764 noearly [X86] Don't do any early type 1 scanning. 4765 This might help on some broken boards which 4766 machine check when some devices' config space 4767 is read. But various workarounds are disabled 4768 and some IOMMU drivers will not work. 4769 bfsort Sort PCI devices into breadth-first order. 4770 This sorting is done to get a device 4771 order compatible with older (<= 2.4) kernels. 4772 nobfsort Don't sort PCI devices into breadth-first order. 4773 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size) 4774 tuning and use the BIOS-configured MPS defaults. 4775 pcie_bus_safe Set every device's MPS to the largest value 4776 supported by all devices below the root complex. 4777 pcie_bus_perf Set device MPS to the largest allowable MPS 4778 based on its parent bus. Also set MRRS (Max 4779 Read Request Size) to the largest supported 4780 value (no larger than the MPS that the device 4781 or bus can support) for best performance. 4782 pcie_bus_peer2peer Set every device's MPS to 128B, which 4783 every device is guaranteed to support. This 4784 configuration allows peer-to-peer DMA between 4785 any pair of devices, possibly at the cost of 4786 reduced performance. This also guarantees 4787 that hot-added devices will work. 4788 cbiosize=nn[KMG] The fixed amount of bus space which is 4789 reserved for the CardBus bridge's IO window. 4790 The default value is 256 bytes. 4791 cbmemsize=nn[KMG] The fixed amount of bus space which is 4792 reserved for the CardBus bridge's memory 4793 window. The default value is 64 megabytes. 4794 resource_alignment= 4795 Format: 4796 [<order of align>@]<pci_dev>[; ...] 4797 Specifies alignment and device to reassign 4798 aligned memory resources. How to 4799 specify the device is described above. 4800 If <order of align> is not specified, 4801 PAGE_SIZE is used as alignment. 4802 A PCI-PCI bridge can be specified if resource 4803 windows need to be expanded. 4804 To specify the alignment for several 4805 instances of a device, the PCI vendor, 4806 device, subvendor, and subdevice may be 4807 specified, e.g., 12@pci:8086:9c22:103c:198f 4808 for 4096-byte alignment. 4809 ecrc= Enable/disable PCIe ECRC (transaction layer 4810 end-to-end CRC checking). Only effective if 4811 OS has native AER control (either granted by 4812 ACPI _OSC or forced via "pcie_ports=native") 4813 bios: Use BIOS/firmware settings. This is the 4814 the default. 4815 off: Turn ECRC off 4816 on: Turn ECRC on. 4817 hpiosize=nn[KMG] The fixed amount of bus space which is 4818 reserved for hotplug bridge's IO window. 4819 Default size is 256 bytes. 4820 hpmmiosize=nn[KMG] The fixed amount of bus space which is 4821 reserved for hotplug bridge's MMIO window. 4822 Default size is 2 megabytes. 4823 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is 4824 reserved for hotplug bridge's MMIO_PREF window. 4825 Default size is 2 megabytes. 4826 hpmemsize=nn[KMG] The fixed amount of bus space which is 4827 reserved for hotplug bridge's MMIO and 4828 MMIO_PREF window. 4829 Default size is 2 megabytes. 4830 hpbussize=nn The minimum amount of additional bus numbers 4831 reserved for buses below a hotplug bridge. 4832 Default is 1. 4833 realloc= Enable/disable reallocating PCI bridge resources 4834 if allocations done by BIOS are too small to 4835 accommodate resources required by all child 4836 devices. 4837 off: Turn realloc off 4838 on: Turn realloc on 4839 realloc same as realloc=on 4840 noari do not use PCIe ARI. 4841 noats [PCIE, Intel-IOMMU, AMD-IOMMU] 4842 do not use PCIe ATS (and IOMMU device IOTLB). 4843 pcie_scan_all Scan all possible PCIe devices. Otherwise we 4844 only look for one device below a PCIe downstream 4845 port. 4846 big_root_window Try to add a big 64bit memory window to the PCIe 4847 root complex on AMD CPUs. Some GFX hardware 4848 can resize a BAR to allow access to all VRAM. 4849 Adding the window is slightly risky (it may 4850 conflict with unreported devices), so this 4851 taints the kernel. 4852 disable_acs_redir=<pci_dev>[; ...] 4853 Specify one or more PCI devices (in the format 4854 specified above) separated by semicolons. 4855 Each device specified will have the PCI ACS 4856 redirect capabilities forced off which will 4857 allow P2P traffic between devices through 4858 bridges without forcing it upstream. Note: 4859 this removes isolation between devices and 4860 may put more devices in an IOMMU group. 4861 config_acs= 4862 Format: 4863 <ACS flags>@<pci_dev>[; ...] 4864 Specify one or more PCI devices (in the format 4865 specified above) optionally prepended with flags 4866 and separated by semicolons. The respective 4867 capabilities will be enabled, disabled or 4868 unchanged based on what is specified in 4869 flags. 4870 4871 ACS Flags is defined as follows: 4872 bit-0 : ACS Source Validation 4873 bit-1 : ACS Translation Blocking 4874 bit-2 : ACS P2P Request Redirect 4875 bit-3 : ACS P2P Completion Redirect 4876 bit-4 : ACS Upstream Forwarding 4877 bit-5 : ACS P2P Egress Control 4878 bit-6 : ACS Direct Translated P2P 4879 Each bit can be marked as: 4880 '0' – force disabled 4881 '1' – force enabled 4882 'x' – unchanged 4883 For example, 4884 pci=config_acs=10x@pci:0:0 4885 would configure all devices that support 4886 ACS to enable P2P Request Redirect, disable 4887 Translation Blocking, and leave Source 4888 Validation unchanged from whatever power-up 4889 or firmware set it to. 4890 4891 Note: this may remove isolation between devices 4892 and may put more devices in an IOMMU group. 4893 force_floating [S390] Force usage of floating interrupts. 4894 nomio [S390] Do not use MIO instructions. 4895 norid [S390] ignore the RID field and force use of 4896 one PCI domain per PCI function 4897 notph [PCIE] If the PCIE_TPH kernel config parameter 4898 is enabled, this kernel boot option can be used 4899 to disable PCIe TLP Processing Hints support 4900 system-wide. 4901 4902 pcie_aspm= [PCIE] Forcibly enable or ignore PCIe Active State Power 4903 Management. 4904 off Don't touch ASPM configuration at all. Leave any 4905 configuration done by firmware unchanged. 4906 force Enable ASPM even on devices that claim not to support it. 4907 WARNING: Forcing ASPM on may cause system lockups. 4908 4909 pcie_ports= [PCIE] PCIe port services handling: 4910 native Use native PCIe services (PME, AER, DPC, PCIe hotplug) 4911 even if the platform doesn't give the OS permission to 4912 use them. This may cause conflicts if the platform 4913 also tries to use these services. 4914 dpc-native Use native PCIe service for DPC only. May 4915 cause conflicts if firmware uses AER or DPC. 4916 compat Disable native PCIe services (PME, AER, DPC, PCIe 4917 hotplug). 4918 4919 pcie_port_pm= [PCIE] PCIe port power management handling: 4920 off Disable power management of all PCIe ports 4921 force Forcibly enable power management of all PCIe ports 4922 4923 pcie_pme= [PCIE,PM] Native PCIe PME signaling options: 4924 nomsi Do not use MSI for native PCIe PME signaling (this makes 4925 all PCIe root ports use INTx for all services). 4926 4927 pcmv= [HW,PCMCIA] BadgePAD 4 4928 4929 pd_ignore_unused 4930 [PM] 4931 Keep all power-domains already enabled by bootloader on, 4932 even if no driver has claimed them. This is useful 4933 for debug and development, but should not be 4934 needed on a platform with proper driver support. 4935 4936 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at 4937 boot time. 4938 Format: { 0 | 1 } 4939 See arch/parisc/kernel/pdc_chassis.c 4940 4941 percpu_alloc= [MM,EARLY] 4942 Select which percpu first chunk allocator to use. 4943 Currently supported values are "embed" and "page". 4944 Archs may support subset or none of the selections. 4945 See comments in mm/percpu.c for details on each 4946 allocator. This parameter is primarily for debugging 4947 and performance comparison. 4948 4949 pirq= [SMP,APIC] Manual mp-table setup 4950 See Documentation/arch/x86/i386/IO-APIC.rst. 4951 4952 plip= [PPT,NET] Parallel port network link 4953 Format: { parport<nr> | timid | 0 } 4954 See also Documentation/admin-guide/parport.rst. 4955 4956 pmtmr= [X86] Manual setup of pmtmr I/O Port. 4957 Override pmtimer IOPort with a hex value. 4958 e.g. pmtmr=0x508 4959 4960 pmu_override= [PPC] Override the PMU. 4961 This option takes over the PMU facility, so it is no 4962 longer usable by perf. Setting this option starts the 4963 PMU counters by setting MMCR0 to 0 (the FC bit is 4964 cleared). If a number is given, then MMCR1 is set to 4965 that number, otherwise (e.g., 'pmu_override=on'), MMCR1 4966 remains 0. 4967 4968 pm_debug_messages [SUSPEND,KNL] 4969 Enable suspend/resume debug messages during boot up. 4970 4971 pnp.debug=1 [PNP] 4972 Enable PNP debug messages (depends on the 4973 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time 4974 via /sys/module/pnp/parameters/debug. We always show 4975 current resource usage; turning this on also shows 4976 possible settings and some assignment information. 4977 4978 pnpacpi= [ACPI] 4979 { off } 4980 4981 pnpbios= [ISAPNP] 4982 { on | off | curr | res | no-curr | no-res } 4983 4984 pnp_reserve_irq= 4985 [ISAPNP] Exclude IRQs for the autoconfiguration 4986 4987 pnp_reserve_dma= 4988 [ISAPNP] Exclude DMAs for the autoconfiguration 4989 4990 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration 4991 Ranges are in pairs (I/O port base and size). 4992 4993 pnp_reserve_mem= 4994 [ISAPNP] Exclude memory regions for the 4995 autoconfiguration. 4996 Ranges are in pairs (memory base and size). 4997 4998 ports= [IP_VS_FTP] IPVS ftp helper module 4999 Default is 21. 5000 Up to 8 (IP_VS_APP_MAX_PORTS) ports 5001 may be specified. 5002 Format: <port>,<port>.... 5003 5004 possible_cpus= [SMP,S390,X86] 5005 Format: <unsigned int> 5006 Set the number of possible CPUs, overriding the 5007 regular discovery mechanisms (such as ACPI/FW, etc). 5008 5009 powersave=off [PPC] This option disables power saving features. 5010 It specifically disables cpuidle and sets the 5011 platform machine description specific power_save 5012 function to NULL. On Idle the CPU just reduces 5013 execution priority. 5014 5015 ppc_strict_facility_enable 5016 [PPC,ENABLE] This option catches any kernel floating point, 5017 Altivec, VSX and SPE outside of regions specifically 5018 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()). 5019 There is some performance impact when enabling this. 5020 5021 ppc_tm= [PPC,EARLY] 5022 Format: {"off"} 5023 Disable Hardware Transactional Memory 5024 5025 preempt= [KNL] 5026 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC 5027 none - Limited to cond_resched() calls 5028 voluntary - Limited to cond_resched() and might_sleep() calls 5029 full - Any section that isn't explicitly preempt disabled 5030 can be preempted anytime. Tasks will also yield 5031 contended spinlocks (if the critical section isn't 5032 explicitly preempt disabled beyond the lock itself). 5033 lazy - Scheduler controlled. Similar to full but instead 5034 of preempting the task immediately, the task gets 5035 one HZ tick time to yield itself before the 5036 preemption will be forced. One preemption is when the 5037 task returns to user space. 5038 5039 print-fatal-signals= 5040 [KNL] debug: print fatal signals 5041 5042 If enabled, warn about various signal handling 5043 related application anomalies: too many signals, 5044 too many POSIX.1 timers, fatal signals causing a 5045 coredump - etc. 5046 5047 If you hit the warning due to signal overflow, 5048 you might want to try "ulimit -i unlimited". 5049 5050 default: off. 5051 5052 printk.always_kmsg_dump= 5053 Trigger kmsg_dump for cases other than kernel oops or 5054 panics 5055 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5056 default: disabled 5057 5058 printk.console_no_auto_verbose= 5059 Disable console loglevel raise on oops, panic 5060 or lockdep-detected issues (only if lock debug is on). 5061 With an exception to setups with low baudrate on 5062 serial console, keeping this 0 is a good choice 5063 in order to provide more debug information. 5064 Format: <bool> 5065 default: 0 (auto_verbose is enabled) 5066 5067 printk.debug_non_panic_cpus= 5068 Allows storing messages from non-panic CPUs into 5069 the printk log buffer during panic(). They are 5070 flushed to consoles by the panic-CPU on 5071 a best-effort basis. 5072 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5073 Default: disabled 5074 5075 printk.devkmsg={on,off,ratelimit} 5076 Control writing to /dev/kmsg. 5077 on - unlimited logging to /dev/kmsg from userspace 5078 off - logging to /dev/kmsg disabled 5079 ratelimit - ratelimit the logging 5080 Default: ratelimit 5081 5082 printk.time= Show timing data prefixed to each printk message line 5083 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5084 5085 proc_mem.force_override= [KNL] 5086 Format: {always | ptrace | never} 5087 Traditionally /proc/pid/mem allows memory permissions to be 5088 overridden without restrictions. This option may be set to 5089 restrict that. Can be one of: 5090 - 'always': traditional behavior always allows mem overrides. 5091 - 'ptrace': only allow mem overrides for active ptracers. 5092 - 'never': never allow mem overrides. 5093 If not specified, default is the CONFIG_PROC_MEM_* choice. 5094 5095 processor.max_cstate= [HW,ACPI] 5096 Limit processor to maximum C-state 5097 max_cstate=9 overrides any DMI blacklist limit. 5098 5099 processor.nocst [HW,ACPI] 5100 Ignore the _CST method to determine C-states, 5101 instead using the legacy FADT method 5102 5103 profile= [KNL] Enable kernel profiling via /proc/profile 5104 Format: [<profiletype>,]<number> 5105 Param: <profiletype>: "schedule" or "kvm" 5106 [defaults to kernel profiling] 5107 Param: "schedule" - profile schedule points. 5108 Param: "kvm" - profile VM exits. 5109 Param: <number> - step/bucket size as a power of 2 for 5110 statistical time based profiling. 5111 5112 prompt_ramdisk= [RAM] [Deprecated] 5113 5114 prot_virt= [S390] enable hosting protected virtual machines 5115 isolated from the hypervisor (if hardware supports 5116 that). If enabled, the default kernel base address 5117 might be overridden even when Kernel Address Space 5118 Layout Randomization is disabled. 5119 Format: <bool> 5120 5121 psi= [KNL] Enable or disable pressure stall information 5122 tracking. 5123 Format: <bool> 5124 5125 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to 5126 probe for; one of (bare|imps|exps|lifebook|any). 5127 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports 5128 per second. 5129 psmouse.resetafter= [HW,MOUSE] 5130 Try to reset the device after so many bad packets 5131 (0 = never). 5132 psmouse.resolution= 5133 [HW,MOUSE] Set desired mouse resolution, in dpi. 5134 psmouse.smartscroll= 5135 [HW,MOUSE] Controls Logitech smartscroll autorepeat. 5136 0 = disabled, 1 = enabled (default). 5137 5138 pstore.backend= Specify the name of the pstore backend to use 5139 5140 pti= [X86-64] Control Page Table Isolation of user and 5141 kernel address spaces. Disabling this feature 5142 removes hardening, but improves performance of 5143 system calls and interrupts. 5144 5145 on - unconditionally enable 5146 off - unconditionally disable 5147 auto - kernel detects whether your CPU model is 5148 vulnerable to issues that PTI mitigates 5149 5150 Not specifying this option is equivalent to pti=auto. 5151 5152 pty.legacy_count= 5153 [KNL] Number of legacy pty's. Overwrites compiled-in 5154 default number. 5155 5156 quiet [KNL,EARLY] Disable most log messages 5157 5158 r128= [HW,DRM] 5159 5160 radix_hcall_invalidate=on [PPC/PSERIES] 5161 Disable RADIX GTSE feature and use hcall for TLB 5162 invalidate. 5163 5164 raid= [HW,RAID] 5165 See Documentation/admin-guide/md.rst. 5166 5167 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes 5168 See Documentation/admin-guide/blockdev/ramdisk.rst. 5169 5170 ramdisk_start= [RAM] RAM disk image start address 5171 5172 random.trust_cpu=off 5173 [KNL,EARLY] Disable trusting the use of the CPU's 5174 random number generator (if available) to 5175 initialize the kernel's RNG. 5176 5177 random.trust_bootloader=off 5178 [KNL,EARLY] Disable trusting the use of the a seed 5179 passed by the bootloader (if available) to 5180 initialize the kernel's RNG. 5181 5182 randomize_kstack_offset= 5183 [KNL,EARLY] Enable or disable kernel stack offset 5184 randomization, which provides roughly 5 bits of 5185 entropy, frustrating memory corruption attacks 5186 that depend on stack address determinism or 5187 cross-syscall address exposures. This is only 5188 available on architectures that have defined 5189 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET. 5190 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 5191 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT. 5192 5193 ras=option[,option,...] [KNL] RAS-specific options 5194 5195 cec_disable [X86] 5196 Disable the Correctable Errors Collector, 5197 see CONFIG_RAS_CEC help text. 5198 5199 rcu_nocbs[=cpu-list] 5200 [KNL] The optional argument is a cpu list, 5201 as described above. 5202 5203 In kernels built with CONFIG_RCU_NOCB_CPU=y, 5204 enable the no-callback CPU mode, which prevents 5205 such CPUs' callbacks from being invoked in 5206 softirq context. Invocation of such CPUs' RCU 5207 callbacks will instead be offloaded to "rcuox/N" 5208 kthreads created for that purpose, where "x" is 5209 "p" for RCU-preempt, "s" for RCU-sched, and "g" 5210 for the kthreads that mediate grace periods; and 5211 "N" is the CPU number. This reduces OS jitter on 5212 the offloaded CPUs, which can be useful for HPC 5213 and real-time workloads. It can also improve 5214 energy efficiency for asymmetric multiprocessors. 5215 5216 If a cpulist is passed as an argument, the specified 5217 list of CPUs is set to no-callback mode from boot. 5218 5219 Otherwise, if the '=' sign and the cpulist 5220 arguments are omitted, no CPU will be set to 5221 no-callback mode from boot but the mode may be 5222 toggled at runtime via cpusets. 5223 5224 Note that this argument takes precedence over 5225 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option. 5226 5227 rcu_nocb_poll [KNL] 5228 Rather than requiring that offloaded CPUs 5229 (specified by rcu_nocbs= above) explicitly 5230 awaken the corresponding "rcuoN" kthreads, 5231 make these kthreads poll for callbacks. 5232 This improves the real-time response for the 5233 offloaded CPUs by relieving them of the need to 5234 wake up the corresponding kthread, but degrades 5235 energy efficiency by requiring that the kthreads 5236 periodically wake up to do the polling. 5237 5238 rcutree.blimit= [KNL] 5239 Set maximum number of finished RCU callbacks to 5240 process in one batch. 5241 5242 rcutree.csd_lock_suppress_rcu_stall= [KNL] 5243 Do only a one-line RCU CPU stall warning when 5244 there is an ongoing too-long CSD-lock wait. 5245 5246 rcutree.do_rcu_barrier= [KNL] 5247 Request a call to rcu_barrier(). This is 5248 throttled so that userspace tests can safely 5249 hammer on the sysfs variable if they so choose. 5250 If triggered before the RCU grace-period machinery 5251 is fully active, this will error out with EAGAIN. 5252 5253 rcutree.dump_tree= [KNL] 5254 Dump the structure of the rcu_node combining tree 5255 out at early boot. This is used for diagnostic 5256 purposes, to verify correct tree setup. 5257 5258 rcutree.gp_cleanup_delay= [KNL] 5259 Set the number of jiffies to delay each step of 5260 RCU grace-period cleanup. 5261 5262 rcutree.gp_init_delay= [KNL] 5263 Set the number of jiffies to delay each step of 5264 RCU grace-period initialization. 5265 5266 rcutree.gp_preinit_delay= [KNL] 5267 Set the number of jiffies to delay each step of 5268 RCU grace-period pre-initialization, that is, 5269 the propagation of recent CPU-hotplug changes up 5270 the rcu_node combining tree. 5271 5272 rcutree.jiffies_till_first_fqs= [KNL] 5273 Set delay from grace-period initialization to 5274 first attempt to force quiescent states. 5275 Units are jiffies, minimum value is zero, 5276 and maximum value is HZ. 5277 5278 rcutree.jiffies_till_next_fqs= [KNL] 5279 Set delay between subsequent attempts to force 5280 quiescent states. Units are jiffies, minimum 5281 value is one, and maximum value is HZ. 5282 5283 rcutree.jiffies_till_sched_qs= [KNL] 5284 Set required age in jiffies for a 5285 given grace period before RCU starts 5286 soliciting quiescent-state help from 5287 rcu_note_context_switch() and cond_resched(). 5288 If not specified, the kernel will calculate 5289 a value based on the most recent settings 5290 of rcutree.jiffies_till_first_fqs 5291 and rcutree.jiffies_till_next_fqs. 5292 This calculated value may be viewed in 5293 rcutree.jiffies_to_sched_qs. Any attempt to set 5294 rcutree.jiffies_to_sched_qs will be cheerfully 5295 overwritten. 5296 5297 rcutree.kthread_prio= [KNL,BOOT] 5298 Set the SCHED_FIFO priority of the RCU per-CPU 5299 kthreads (rcuc/N). This value is also used for 5300 the priority of the RCU boost threads (rcub/N) 5301 and for the RCU grace-period kthreads (rcu_bh, 5302 rcu_preempt, and rcu_sched). If RCU_BOOST is 5303 set, valid values are 1-99 and the default is 1 5304 (the least-favored priority). Otherwise, when 5305 RCU_BOOST is not set, valid values are 0-99 and 5306 the default is zero (non-realtime operation). 5307 When RCU_NOCB_CPU is set, also adjust the 5308 priority of NOCB callback kthreads. 5309 5310 rcutree.nocb_nobypass_lim_per_jiffy= [KNL] 5311 On callback-offloaded (rcu_nocbs) CPUs, 5312 RCU reduces the lock contention that would 5313 otherwise be caused by callback floods through 5314 use of the ->nocb_bypass list. However, in the 5315 common non-flooded case, RCU queues directly to 5316 the main ->cblist in order to avoid the extra 5317 overhead of the ->nocb_bypass list and its lock. 5318 But if there are too many callbacks queued during 5319 a single jiffy, RCU pre-queues the callbacks into 5320 the ->nocb_bypass queue. The definition of "too 5321 many" is supplied by this kernel boot parameter. 5322 5323 rcutree.nohz_full_patience_delay= [KNL] 5324 On callback-offloaded (rcu_nocbs) CPUs, avoid 5325 disturbing RCU unless the grace period has 5326 reached the specified age in milliseconds. 5327 Defaults to zero. Large values will be capped 5328 at five seconds. All values will be rounded down 5329 to the nearest value representable by jiffies. 5330 5331 rcutree.qhimark= [KNL] 5332 Set threshold of queued RCU callbacks beyond which 5333 batch limiting is disabled. 5334 5335 rcutree.qlowmark= [KNL] 5336 Set threshold of queued RCU callbacks below which 5337 batch limiting is re-enabled. 5338 5339 rcutree.qovld= [KNL] 5340 Set threshold of queued RCU callbacks beyond which 5341 RCU's force-quiescent-state scan will aggressively 5342 enlist help from cond_resched() and sched IPIs to 5343 help CPUs more quickly reach quiescent states. 5344 Set to less than zero to make this be set based 5345 on rcutree.qhimark at boot time and to zero to 5346 disable more aggressive help enlistment. 5347 5348 rcutree.rcu_delay_page_cache_fill_msec= [KNL] 5349 Set the page-cache refill delay (in milliseconds) 5350 in response to low-memory conditions. The range 5351 of permitted values is in the range 0:100000. 5352 5353 rcutree.rcu_divisor= [KNL] 5354 Set the shift-right count to use to compute 5355 the callback-invocation batch limit bl from 5356 the number of callbacks queued on this CPU. 5357 The result will be bounded below by the value of 5358 the rcutree.blimit kernel parameter. Every bl 5359 callbacks, the softirq handler will exit in 5360 order to allow the CPU to do other work. 5361 5362 Please note that this callback-invocation batch 5363 limit applies only to non-offloaded callback 5364 invocation. Offloaded callbacks are instead 5365 invoked in the context of an rcuoc kthread, which 5366 scheduler will preempt as it does any other task. 5367 5368 rcutree.rcu_fanout_exact= [KNL] 5369 Disable autobalancing of the rcu_node combining 5370 tree. This is used by rcutorture, and might 5371 possibly be useful for architectures having high 5372 cache-to-cache transfer latencies. 5373 5374 rcutree.rcu_fanout_leaf= [KNL] 5375 Change the number of CPUs assigned to each 5376 leaf rcu_node structure. Useful for very 5377 large systems, which will choose the value 64, 5378 and for NUMA systems with large remote-access 5379 latencies, which will choose a value aligned 5380 with the appropriate hardware boundaries. 5381 5382 rcutree.rcu_min_cached_objs= [KNL] 5383 Minimum number of objects which are cached and 5384 maintained per one CPU. Object size is equal 5385 to PAGE_SIZE. The cache allows to reduce the 5386 pressure to page allocator, also it makes the 5387 whole algorithm to behave better in low memory 5388 condition. 5389 5390 rcutree.rcu_nocb_gp_stride= [KNL] 5391 Set the number of NOCB callback kthreads in 5392 each group, which defaults to the square root 5393 of the number of CPUs. Larger numbers reduce 5394 the wakeup overhead on the global grace-period 5395 kthread, but increases that same overhead on 5396 each group's NOCB grace-period kthread. 5397 5398 rcutree.rcu_kick_kthreads= [KNL] 5399 Cause the grace-period kthread to get an extra 5400 wake_up() if it sleeps three times longer than 5401 it should at force-quiescent-state time. 5402 This wake_up() will be accompanied by a 5403 WARN_ONCE() splat and an ftrace_dump(). 5404 5405 rcutree.rcu_resched_ns= [KNL] 5406 Limit the time spend invoking a batch of RCU 5407 callbacks to the specified number of nanoseconds. 5408 By default, this limit is checked only once 5409 every 32 callbacks in order to limit the pain 5410 inflicted by local_clock() overhead. 5411 5412 rcutree.rcu_unlock_delay= [KNL] 5413 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels, 5414 this specifies an rcu_read_unlock()-time delay 5415 in microseconds. This defaults to zero. 5416 Larger delays increase the probability of 5417 catching RCU pointer leaks, that is, buggy use 5418 of RCU-protected pointers after the relevant 5419 rcu_read_unlock() has completed. 5420 5421 rcutree.sysrq_rcu= [KNL] 5422 Commandeer a sysrq key to dump out Tree RCU's 5423 rcu_node tree with an eye towards determining 5424 why a new grace period has not yet started. 5425 5426 rcutree.use_softirq= [KNL] 5427 If set to zero, move all RCU_SOFTIRQ processing to 5428 per-CPU rcuc kthreads. Defaults to a non-zero 5429 value, meaning that RCU_SOFTIRQ is used by default. 5430 Specify rcutree.use_softirq=0 to use rcuc kthreads. 5431 5432 But note that CONFIG_PREEMPT_RT=y kernels disable 5433 this kernel boot parameter, forcibly setting it 5434 to zero. 5435 5436 rcutree.enable_rcu_lazy= [KNL] 5437 To save power, batch RCU callbacks and flush after 5438 delay, memory pressure or callback list growing too 5439 big. 5440 5441 rcutree.rcu_normal_wake_from_gp= [KNL] 5442 Reduces a latency of synchronize_rcu() call. This approach 5443 maintains its own track of synchronize_rcu() callers, so it 5444 does not interact with regular callbacks because it does not 5445 use a call_rcu[_hurry]() path. Please note, this is for a 5446 normal grace period. 5447 5448 How to enable it: 5449 5450 echo 1 > /sys/module/rcutree/parameters/rcu_normal_wake_from_gp 5451 or pass a boot parameter "rcutree.rcu_normal_wake_from_gp=1" 5452 5453 Default is 0. 5454 5455 rcuscale.gp_async= [KNL] 5456 Measure performance of asynchronous 5457 grace-period primitives such as call_rcu(). 5458 5459 rcuscale.gp_async_max= [KNL] 5460 Specify the maximum number of outstanding 5461 callbacks per writer thread. When a writer 5462 thread exceeds this limit, it invokes the 5463 corresponding flavor of rcu_barrier() to allow 5464 previously posted callbacks to drain. 5465 5466 rcuscale.gp_exp= [KNL] 5467 Measure performance of expedited synchronous 5468 grace-period primitives. 5469 5470 rcuscale.holdoff= [KNL] 5471 Set test-start holdoff period. The purpose of 5472 this parameter is to delay the start of the 5473 test until boot completes in order to avoid 5474 interference. 5475 5476 rcuscale.kfree_by_call_rcu= [KNL] 5477 In kernels built with CONFIG_RCU_LAZY=y, test 5478 call_rcu() instead of kfree_rcu(). 5479 5480 rcuscale.kfree_mult= [KNL] 5481 Instead of allocating an object of size kfree_obj, 5482 allocate one of kfree_mult * sizeof(kfree_obj). 5483 Defaults to 1. 5484 5485 rcuscale.kfree_rcu_test= [KNL] 5486 Set to measure performance of kfree_rcu() flooding. 5487 5488 rcuscale.kfree_rcu_test_double= [KNL] 5489 Test the double-argument variant of kfree_rcu(). 5490 If this parameter has the same value as 5491 rcuscale.kfree_rcu_test_single, both the single- 5492 and double-argument variants are tested. 5493 5494 rcuscale.kfree_rcu_test_single= [KNL] 5495 Test the single-argument variant of kfree_rcu(). 5496 If this parameter has the same value as 5497 rcuscale.kfree_rcu_test_double, both the single- 5498 and double-argument variants are tested. 5499 5500 rcuscale.kfree_nthreads= [KNL] 5501 The number of threads running loops of kfree_rcu(). 5502 5503 rcuscale.kfree_alloc_num= [KNL] 5504 Number of allocations and frees done in an iteration. 5505 5506 rcuscale.kfree_loops= [KNL] 5507 Number of loops doing rcuscale.kfree_alloc_num number 5508 of allocations and frees. 5509 5510 rcuscale.minruntime= [KNL] 5511 Set the minimum test run time in seconds. This 5512 does not affect the data-collection interval, 5513 but instead allows better measurement of things 5514 like CPU consumption. 5515 5516 rcuscale.nreaders= [KNL] 5517 Set number of RCU readers. The value -1 selects 5518 N, where N is the number of CPUs. A value 5519 "n" less than -1 selects N-n+1, where N is again 5520 the number of CPUs. For example, -2 selects N 5521 (the number of CPUs), -3 selects N+1, and so on. 5522 A value of "n" less than or equal to -N selects 5523 a single reader. 5524 5525 rcuscale.nwriters= [KNL] 5526 Set number of RCU writers. The values operate 5527 the same as for rcuscale.nreaders. 5528 N, where N is the number of CPUs 5529 5530 rcuscale.scale_type= [KNL] 5531 Specify the RCU implementation to test. 5532 5533 rcuscale.shutdown= [KNL] 5534 Shut the system down after performance tests 5535 complete. This is useful for hands-off automated 5536 testing. 5537 5538 rcuscale.verbose= [KNL] 5539 Enable additional printk() statements. 5540 5541 rcuscale.writer_holdoff= [KNL] 5542 Write-side holdoff between grace periods, 5543 in microseconds. The default of zero says 5544 no holdoff. 5545 5546 rcuscale.writer_holdoff_jiffies= [KNL] 5547 Additional write-side holdoff between grace 5548 periods, but in jiffies. The default of zero 5549 says no holdoff. 5550 5551 rcutorture.fqs_duration= [KNL] 5552 Set duration of force_quiescent_state bursts 5553 in microseconds. 5554 5555 rcutorture.fqs_holdoff= [KNL] 5556 Set holdoff time within force_quiescent_state bursts 5557 in microseconds. 5558 5559 rcutorture.fqs_stutter= [KNL] 5560 Set wait time between force_quiescent_state bursts 5561 in seconds. 5562 5563 rcutorture.fwd_progress= [KNL] 5564 Specifies the number of kthreads to be used 5565 for RCU grace-period forward-progress testing 5566 for the types of RCU supporting this notion. 5567 Defaults to 1 kthread, values less than zero or 5568 greater than the number of CPUs cause the number 5569 of CPUs to be used. 5570 5571 rcutorture.fwd_progress_div= [KNL] 5572 Specify the fraction of a CPU-stall-warning 5573 period to do tight-loop forward-progress testing. 5574 5575 rcutorture.fwd_progress_holdoff= [KNL] 5576 Number of seconds to wait between successive 5577 forward-progress tests. 5578 5579 rcutorture.fwd_progress_need_resched= [KNL] 5580 Enclose cond_resched() calls within checks for 5581 need_resched() during tight-loop forward-progress 5582 testing. 5583 5584 rcutorture.gp_cond= [KNL] 5585 Use conditional/asynchronous update-side 5586 normal-grace-period primitives, if available. 5587 5588 rcutorture.gp_cond_exp= [KNL] 5589 Use conditional/asynchronous update-side 5590 expedited-grace-period primitives, if available. 5591 5592 rcutorture.gp_cond_full= [KNL] 5593 Use conditional/asynchronous update-side 5594 normal-grace-period primitives that also take 5595 concurrent expedited grace periods into account, 5596 if available. 5597 5598 rcutorture.gp_cond_exp_full= [KNL] 5599 Use conditional/asynchronous update-side 5600 expedited-grace-period primitives that also take 5601 concurrent normal grace periods into account, 5602 if available. 5603 5604 rcutorture.gp_cond_wi= [KNL] 5605 Nominal wait interval for normal conditional 5606 grace periods (specified by rcutorture's 5607 gp_cond and gp_cond_full module parameters), 5608 in microseconds. The actual wait interval will 5609 be randomly selected to nanosecond granularity up 5610 to this wait interval. Defaults to 16 jiffies, 5611 for example, 16,000 microseconds on a system 5612 with HZ=1000. 5613 5614 rcutorture.gp_cond_wi_exp= [KNL] 5615 Nominal wait interval for expedited conditional 5616 grace periods (specified by rcutorture's 5617 gp_cond_exp and gp_cond_exp_full module 5618 parameters), in microseconds. The actual wait 5619 interval will be randomly selected to nanosecond 5620 granularity up to this wait interval. Defaults to 5621 128 microseconds. 5622 5623 rcutorture.gp_exp= [KNL] 5624 Use expedited update-side primitives, if available. 5625 5626 rcutorture.gp_normal= [KNL] 5627 Use normal (non-expedited) asynchronous 5628 update-side primitives, if available. 5629 5630 rcutorture.gp_poll= [KNL] 5631 Use polled update-side normal-grace-period 5632 primitives, if available. 5633 5634 rcutorture.gp_poll_exp= [KNL] 5635 Use polled update-side expedited-grace-period 5636 primitives, if available. 5637 5638 rcutorture.gp_poll_full= [KNL] 5639 Use polled update-side normal-grace-period 5640 primitives that also take concurrent expedited 5641 grace periods into account, if available. 5642 5643 rcutorture.gp_poll_exp_full= [KNL] 5644 Use polled update-side expedited-grace-period 5645 primitives that also take concurrent normal 5646 grace periods into account, if available. 5647 5648 rcutorture.gp_poll_wi= [KNL] 5649 Nominal wait interval for normal conditional 5650 grace periods (specified by rcutorture's 5651 gp_poll and gp_poll_full module parameters), 5652 in microseconds. The actual wait interval will 5653 be randomly selected to nanosecond granularity up 5654 to this wait interval. Defaults to 16 jiffies, 5655 for example, 16,000 microseconds on a system 5656 with HZ=1000. 5657 5658 rcutorture.gp_poll_wi_exp= [KNL] 5659 Nominal wait interval for expedited conditional 5660 grace periods (specified by rcutorture's 5661 gp_poll_exp and gp_poll_exp_full module 5662 parameters), in microseconds. The actual wait 5663 interval will be randomly selected to nanosecond 5664 granularity up to this wait interval. Defaults to 5665 128 microseconds. 5666 5667 rcutorture.gp_sync= [KNL] 5668 Use normal (non-expedited) synchronous 5669 update-side primitives, if available. If all 5670 of rcutorture.gp_cond=, rcutorture.gp_exp=, 5671 rcutorture.gp_normal=, and rcutorture.gp_sync= 5672 are zero, rcutorture acts as if is interpreted 5673 they are all non-zero. 5674 5675 rcutorture.irqreader= [KNL] 5676 Run RCU readers from irq handlers, or, more 5677 accurately, from a timer handler. Not all RCU 5678 flavors take kindly to this sort of thing. 5679 5680 rcutorture.leakpointer= [KNL] 5681 Leak an RCU-protected pointer out of the reader. 5682 This can of course result in splats, and is 5683 intended to test the ability of things like 5684 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect 5685 such leaks. 5686 5687 rcutorture.n_barrier_cbs= [KNL] 5688 Set callbacks/threads for rcu_barrier() testing. 5689 5690 rcutorture.nfakewriters= [KNL] 5691 Set number of concurrent RCU writers. These just 5692 stress RCU, they don't participate in the actual 5693 test, hence the "fake". 5694 5695 rcutorture.nocbs_nthreads= [KNL] 5696 Set number of RCU callback-offload togglers. 5697 Zero (the default) disables toggling. 5698 5699 rcutorture.nocbs_toggle= [KNL] 5700 Set the delay in milliseconds between successive 5701 callback-offload toggling attempts. 5702 5703 rcutorture.nreaders= [KNL] 5704 Set number of RCU readers. The value -1 selects 5705 N-1, where N is the number of CPUs. A value 5706 "n" less than -1 selects N-n-2, where N is again 5707 the number of CPUs. For example, -2 selects N 5708 (the number of CPUs), -3 selects N+1, and so on. 5709 5710 rcutorture.object_debug= [KNL] 5711 Enable debug-object double-call_rcu() testing. 5712 5713 rcutorture.onoff_holdoff= [KNL] 5714 Set time (s) after boot for CPU-hotplug testing. 5715 5716 rcutorture.onoff_interval= [KNL] 5717 Set time (jiffies) between CPU-hotplug operations, 5718 or zero to disable CPU-hotplug testing. 5719 5720 rcutorture.preempt_duration= [KNL] 5721 Set duration (in milliseconds) of preemptions 5722 by a high-priority FIFO real-time task. Set to 5723 zero (the default) to disable. The CPUs to 5724 preempt are selected randomly from the set that 5725 are online at a given point in time. Races with 5726 CPUs going offline are ignored, with that attempt 5727 at preemption skipped. 5728 5729 rcutorture.preempt_interval= [KNL] 5730 Set interval (in milliseconds, defaulting to one 5731 second) between preemptions by a high-priority 5732 FIFO real-time task. This delay is mediated 5733 by an hrtimer and is further fuzzed to avoid 5734 inadvertent synchronizations. 5735 5736 rcutorture.read_exit_burst= [KNL] 5737 The number of times in a given read-then-exit 5738 episode that a set of read-then-exit kthreads 5739 is spawned. 5740 5741 rcutorture.read_exit_delay= [KNL] 5742 The delay, in seconds, between successive 5743 read-then-exit testing episodes. 5744 5745 rcutorture.reader_flavor= [KNL] 5746 A bit mask indicating which readers to use. 5747 If there is more than one bit set, the readers 5748 are entered from low-order bit up, and are 5749 exited in the opposite order. For SRCU, the 5750 0x1 bit is normal readers, 0x2 NMI-safe readers, 5751 and 0x4 light-weight readers. 5752 5753 rcutorture.shuffle_interval= [KNL] 5754 Set task-shuffle interval (s). Shuffling tasks 5755 allows some CPUs to go into dyntick-idle mode 5756 during the rcutorture test. 5757 5758 rcutorture.shutdown_secs= [KNL] 5759 Set time (s) after boot system shutdown. This 5760 is useful for hands-off automated testing. 5761 5762 rcutorture.stall_cpu= [KNL] 5763 Duration of CPU stall (s) to test RCU CPU stall 5764 warnings, zero to disable. 5765 5766 rcutorture.stall_cpu_block= [KNL] 5767 Sleep while stalling if set. This will result 5768 in warnings from preemptible RCU in addition to 5769 any other stall-related activity. Note that 5770 in kernels built with CONFIG_PREEMPTION=n and 5771 CONFIG_PREEMPT_COUNT=y, this parameter will 5772 cause the CPU to pass through a quiescent state. 5773 Given CONFIG_PREEMPTION=n, this will suppress 5774 RCU CPU stall warnings, but will instead result 5775 in scheduling-while-atomic splats. 5776 5777 Use of this module parameter results in splats. 5778 5779 5780 rcutorture.stall_cpu_holdoff= [KNL] 5781 Time to wait (s) after boot before inducing stall. 5782 5783 rcutorture.stall_cpu_irqsoff= [KNL] 5784 Disable interrupts while stalling if set, but only 5785 on the first stall in the set. 5786 5787 rcutorture.stall_cpu_repeat= [KNL] 5788 Number of times to repeat the stall sequence, 5789 so that rcutorture.stall_cpu_repeat=3 will result 5790 in four stall sequences. 5791 5792 rcutorture.stall_gp_kthread= [KNL] 5793 Duration (s) of forced sleep within RCU 5794 grace-period kthread to test RCU CPU stall 5795 warnings, zero to disable. If both stall_cpu 5796 and stall_gp_kthread are specified, the 5797 kthread is starved first, then the CPU. 5798 5799 rcutorture.stat_interval= [KNL] 5800 Time (s) between statistics printk()s. 5801 5802 rcutorture.stutter= [KNL] 5803 Time (s) to stutter testing, for example, specifying 5804 five seconds causes the test to run for five seconds, 5805 wait for five seconds, and so on. This tests RCU's 5806 ability to transition abruptly to and from idle. 5807 5808 rcutorture.test_boost= [KNL] 5809 Test RCU priority boosting? 0=no, 1=maybe, 2=yes. 5810 "Maybe" means test if the RCU implementation 5811 under test support RCU priority boosting. 5812 5813 rcutorture.test_boost_duration= [KNL] 5814 Duration (s) of each individual boost test. 5815 5816 rcutorture.test_boost_holdoff= [KNL] 5817 Holdoff time (s) from start of test to the start 5818 of RCU priority-boost testing. Defaults to zero, 5819 that is, no holdoff. 5820 5821 rcutorture.test_boost_interval= [KNL] 5822 Interval (s) between each boost test. 5823 5824 rcutorture.test_no_idle_hz= [KNL] 5825 Test RCU's dyntick-idle handling. See also the 5826 rcutorture.shuffle_interval parameter. 5827 5828 rcutorture.torture_type= [KNL] 5829 Specify the RCU implementation to test. 5830 5831 rcutorture.verbose= [KNL] 5832 Enable additional printk() statements. 5833 5834 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL] 5835 Dump ftrace buffer after reporting RCU CPU 5836 stall warning. 5837 5838 rcupdate.rcu_cpu_stall_notifiers= [KNL] 5839 Provide RCU CPU stall notifiers, but see the 5840 warnings in the RCU_CPU_STALL_NOTIFIER Kconfig 5841 option's help text. TL;DR: You almost certainly 5842 do not want rcupdate.rcu_cpu_stall_notifiers. 5843 5844 rcupdate.rcu_cpu_stall_suppress= [KNL] 5845 Suppress RCU CPU stall warning messages. 5846 5847 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL] 5848 Suppress RCU CPU stall warning messages and 5849 rcutorture writer stall warnings that occur 5850 during early boot, that is, during the time 5851 before the init task is spawned. 5852 5853 rcupdate.rcu_cpu_stall_timeout= [KNL] 5854 Set timeout for RCU CPU stall warning messages. 5855 The value is in seconds and the maximum allowed 5856 value is 300 seconds. 5857 5858 rcupdate.rcu_exp_cpu_stall_timeout= [KNL] 5859 Set timeout for expedited RCU CPU stall warning 5860 messages. The value is in milliseconds 5861 and the maximum allowed value is 21000 5862 milliseconds. Please note that this value is 5863 adjusted to an arch timer tick resolution. 5864 Setting this to zero causes the value from 5865 rcupdate.rcu_cpu_stall_timeout to be used (after 5866 conversion from seconds to milliseconds). 5867 5868 rcupdate.rcu_cpu_stall_cputime= [KNL] 5869 Provide statistics on the cputime and count of 5870 interrupts and tasks during the sampling period. For 5871 multiple continuous RCU stalls, all sampling periods 5872 begin at half of the first RCU stall timeout. 5873 5874 rcupdate.rcu_exp_stall_task_details= [KNL] 5875 Print stack dumps of any tasks blocking the 5876 current expedited RCU grace period during an 5877 expedited RCU CPU stall warning. 5878 5879 rcupdate.rcu_expedited= [KNL] 5880 Use expedited grace-period primitives, for 5881 example, synchronize_rcu_expedited() instead 5882 of synchronize_rcu(). This reduces latency, 5883 but can increase CPU utilization, degrade 5884 real-time latency, and degrade energy efficiency. 5885 No effect on CONFIG_TINY_RCU kernels. 5886 5887 rcupdate.rcu_normal= [KNL] 5888 Use only normal grace-period primitives, 5889 for example, synchronize_rcu() instead of 5890 synchronize_rcu_expedited(). This improves 5891 real-time latency, CPU utilization, and 5892 energy efficiency, but can expose users to 5893 increased grace-period latency. This parameter 5894 overrides rcupdate.rcu_expedited. No effect on 5895 CONFIG_TINY_RCU kernels. 5896 5897 rcupdate.rcu_normal_after_boot= [KNL] 5898 Once boot has completed (that is, after 5899 rcu_end_inkernel_boot() has been invoked), use 5900 only normal grace-period primitives. No effect 5901 on CONFIG_TINY_RCU kernels. 5902 5903 But note that CONFIG_PREEMPT_RT=y kernels enables 5904 this kernel boot parameter, forcibly setting 5905 it to the value one, that is, converting any 5906 post-boot attempt at an expedited RCU grace 5907 period to instead use normal non-expedited 5908 grace-period processing. 5909 5910 rcupdate.rcu_task_collapse_lim= [KNL] 5911 Set the maximum number of callbacks present 5912 at the beginning of a grace period that allows 5913 the RCU Tasks flavors to collapse back to using 5914 a single callback queue. This switching only 5915 occurs when rcupdate.rcu_task_enqueue_lim is 5916 set to the default value of -1. 5917 5918 rcupdate.rcu_task_contend_lim= [KNL] 5919 Set the minimum number of callback-queuing-time 5920 lock-contention events per jiffy required to 5921 cause the RCU Tasks flavors to switch to per-CPU 5922 callback queuing. This switching only occurs 5923 when rcupdate.rcu_task_enqueue_lim is set to 5924 the default value of -1. 5925 5926 rcupdate.rcu_task_enqueue_lim= [KNL] 5927 Set the number of callback queues to use for the 5928 RCU Tasks family of RCU flavors. The default 5929 of -1 allows this to be automatically (and 5930 dynamically) adjusted. This parameter is intended 5931 for use in testing. 5932 5933 rcupdate.rcu_task_ipi_delay= [KNL] 5934 Set time in jiffies during which RCU tasks will 5935 avoid sending IPIs, starting with the beginning 5936 of a given grace period. Setting a large 5937 number avoids disturbing real-time workloads, 5938 but lengthens grace periods. 5939 5940 rcupdate.rcu_task_lazy_lim= [KNL] 5941 Number of callbacks on a given CPU that will 5942 cancel laziness on that CPU. Use -1 to disable 5943 cancellation of laziness, but be advised that 5944 doing so increases the danger of OOM due to 5945 callback flooding. 5946 5947 rcupdate.rcu_task_stall_info= [KNL] 5948 Set initial timeout in jiffies for RCU task stall 5949 informational messages, which give some indication 5950 of the problem for those not patient enough to 5951 wait for ten minutes. Informational messages are 5952 only printed prior to the stall-warning message 5953 for a given grace period. Disable with a value 5954 less than or equal to zero. Defaults to ten 5955 seconds. A change in value does not take effect 5956 until the beginning of the next grace period. 5957 5958 rcupdate.rcu_task_stall_info_mult= [KNL] 5959 Multiplier for time interval between successive 5960 RCU task stall informational messages for a given 5961 RCU tasks grace period. This value is clamped 5962 to one through ten, inclusive. It defaults to 5963 the value three, so that the first informational 5964 message is printed 10 seconds into the grace 5965 period, the second at 40 seconds, the third at 5966 160 seconds, and then the stall warning at 600 5967 seconds would prevent a fourth at 640 seconds. 5968 5969 rcupdate.rcu_task_stall_timeout= [KNL] 5970 Set timeout in jiffies for RCU task stall 5971 warning messages. Disable with a value less 5972 than or equal to zero. Defaults to ten minutes. 5973 A change in value does not take effect until 5974 the beginning of the next grace period. 5975 5976 rcupdate.rcu_tasks_lazy_ms= [KNL] 5977 Set timeout in milliseconds RCU Tasks asynchronous 5978 callback batching for call_rcu_tasks(). 5979 A negative value will take the default. A value 5980 of zero will disable batching. Batching is 5981 always disabled for synchronize_rcu_tasks(). 5982 5983 rcupdate.rcu_tasks_trace_lazy_ms= [KNL] 5984 Set timeout in milliseconds RCU Tasks 5985 Trace asynchronous callback batching for 5986 call_rcu_tasks_trace(). A negative value 5987 will take the default. A value of zero will 5988 disable batching. Batching is always disabled 5989 for synchronize_rcu_tasks_trace(). 5990 5991 rcupdate.rcu_self_test= [KNL] 5992 Run the RCU early boot self tests 5993 5994 rdinit= [KNL] 5995 Format: <full_path> 5996 Run specified binary instead of /init from the ramdisk, 5997 used for early userspace startup. See initrd. 5998 5999 rdrand= [X86,EARLY] 6000 force - Override the decision by the kernel to hide the 6001 advertisement of RDRAND support (this affects 6002 certain AMD processors because of buggy BIOS 6003 support, specifically around the suspend/resume 6004 path). 6005 6006 rdt= [HW,X86,RDT] 6007 Turn on/off individual RDT features. List is: 6008 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp, 6009 mba, smba, bmec. 6010 E.g. to turn on cmt and turn off mba use: 6011 rdt=cmt,!mba 6012 6013 reboot= [KNL] 6014 Format (x86 or x86_64): 6015 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] | d[efault] \ 6016 [[,]s[mp]#### \ 6017 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \ 6018 [[,]f[orce] 6019 Where reboot_mode is one of warm (soft) or cold (hard) or gpio 6020 (prefix with 'panic_' to set mode for panic 6021 reboot only), 6022 reboot_type is one of bios, acpi, kbd, triple, efi, or pci, 6023 reboot_force is either force or not specified, 6024 reboot_cpu is s[mp]#### with #### being the processor 6025 to be used for rebooting. 6026 6027 acpi 6028 Use the ACPI RESET_REG in the FADT. If ACPI is not 6029 configured or the ACPI reset does not work, the reboot 6030 path attempts the reset using the keyboard controller. 6031 6032 bios 6033 Use the CPU reboot vector for warm reset 6034 6035 cold 6036 Set the cold reboot flag 6037 6038 default 6039 There are some built-in platform specific "quirks" 6040 - you may see: "reboot: <name> series board detected. 6041 Selecting <type> for reboots." In the case where you 6042 think the quirk is in error (e.g. you have newer BIOS, 6043 or newer board) using this option will ignore the 6044 built-in quirk table, and use the generic default 6045 reboot actions. 6046 6047 efi 6048 Use efi reset_system runtime service. If EFI is not 6049 configured or the EFI reset does not work, the reboot 6050 path attempts the reset using the keyboard controller. 6051 6052 force 6053 Don't stop other CPUs on reboot. This can make reboot 6054 more reliable in some cases. 6055 6056 kbd 6057 Use the keyboard controller. cold reset (default) 6058 6059 pci 6060 Use a write to the PCI config space register 0xcf9 to 6061 trigger reboot. 6062 6063 triple 6064 Force a triple fault (init) 6065 6066 warm 6067 Don't set the cold reboot flag 6068 6069 Using warm reset will be much faster especially on big 6070 memory systems because the BIOS will not go through 6071 the memory check. Disadvantage is that not all 6072 hardware will be completely reinitialized on reboot so 6073 there may be boot problems on some systems. 6074 6075 6076 refscale.holdoff= [KNL] 6077 Set test-start holdoff period. The purpose of 6078 this parameter is to delay the start of the 6079 test until boot completes in order to avoid 6080 interference. 6081 6082 refscale.lookup_instances= [KNL] 6083 Number of data elements to use for the forms of 6084 SLAB_TYPESAFE_BY_RCU testing. A negative number 6085 is negated and multiplied by nr_cpu_ids, while 6086 zero specifies nr_cpu_ids. 6087 6088 refscale.loops= [KNL] 6089 Set the number of loops over the synchronization 6090 primitive under test. Increasing this number 6091 reduces noise due to loop start/end overhead, 6092 but the default has already reduced the per-pass 6093 noise to a handful of picoseconds on ca. 2020 6094 x86 laptops. 6095 6096 refscale.nreaders= [KNL] 6097 Set number of readers. The default value of -1 6098 selects N, where N is roughly 75% of the number 6099 of CPUs. A value of zero is an interesting choice. 6100 6101 refscale.nruns= [KNL] 6102 Set number of runs, each of which is dumped onto 6103 the console log. 6104 6105 refscale.readdelay= [KNL] 6106 Set the read-side critical-section duration, 6107 measured in microseconds. 6108 6109 refscale.scale_type= [KNL] 6110 Specify the read-protection implementation to test. 6111 6112 refscale.shutdown= [KNL] 6113 Shut down the system at the end of the performance 6114 test. This defaults to 1 (shut it down) when 6115 refscale is built into the kernel and to 0 (leave 6116 it running) when refscale is built as a module. 6117 6118 refscale.verbose= [KNL] 6119 Enable additional printk() statements. 6120 6121 refscale.verbose_batched= [KNL] 6122 Batch the additional printk() statements. If zero 6123 (the default) or negative, print everything. Otherwise, 6124 print every Nth verbose statement, where N is the value 6125 specified. 6126 6127 regulator_ignore_unused 6128 [REGULATOR] 6129 Prevents regulator framework from disabling regulators 6130 that are unused, due no driver claiming them. This may 6131 be useful for debug and development, but should not be 6132 needed on a platform with proper driver support. 6133 6134 relax_domain_level= 6135 [KNL, SMP] Set scheduler's default relax_domain_level. 6136 See Documentation/admin-guide/cgroup-v1/cpusets.rst. 6137 6138 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory 6139 Format: <base1>,<size1>[,<base2>,<size2>,...] 6140 Reserve I/O ports or memory so the kernel won't use 6141 them. If <base> is less than 0x10000, the region 6142 is assumed to be I/O ports; otherwise it is memory. 6143 6144 reserve_mem= [RAM] 6145 Format: nn[KMG]:<align>:<label> 6146 Reserve physical memory and label it with a name that 6147 other subsystems can use to access it. This is typically 6148 used for systems that do not wipe the RAM, and this command 6149 line will try to reserve the same physical memory on 6150 soft reboots. Note, it is not guaranteed to be the same 6151 location. For example, if anything about the system changes 6152 or if booting a different kernel. It can also fail if KASLR 6153 places the kernel at the location of where the RAM reservation 6154 was from a previous boot, the new reservation will be at a 6155 different location. 6156 Any subsystem using this feature must add a way to verify 6157 that the contents of the physical memory is from a previous 6158 boot, as there may be cases where the memory will not be 6159 located at the same location. 6160 6161 The format is size:align:label for example, to request 6162 12 megabytes of 4096 alignment for ramoops: 6163 6164 reserve_mem=12M:4096:oops ramoops.mem_name=oops 6165 6166 reservetop= [X86-32,EARLY] 6167 Format: nn[KMG] 6168 Reserves a hole at the top of the kernel virtual 6169 address space. 6170 6171 reset_devices [KNL] Force drivers to reset the underlying device 6172 during initialization. 6173 6174 resume= [SWSUSP] 6175 Specify the partition device for software suspend 6176 Format: 6177 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>} 6178 6179 resume_offset= [SWSUSP] 6180 Specify the offset from the beginning of the partition 6181 given by "resume=" at which the swap header is located, 6182 in <PAGE_SIZE> units (needed only for swap files). 6183 See Documentation/power/swsusp-and-swap-files.rst 6184 6185 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to 6186 read the resume files 6187 6188 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up. 6189 Useful for devices that are detected asynchronously 6190 (e.g. USB and MMC devices). 6191 6192 retain_initrd [RAM] Keep initrd memory after extraction. After boot, it will 6193 be accessible via /sys/firmware/initrd. 6194 6195 retbleed= [X86] Control mitigation of RETBleed (Arbitrary 6196 Speculative Code Execution with Return Instructions) 6197 vulnerability. 6198 6199 AMD-based UNRET and IBPB mitigations alone do not stop 6200 sibling threads from influencing the predictions of other 6201 sibling threads. For that reason, STIBP is used on pro- 6202 cessors that support it, and mitigate SMT on processors 6203 that don't. 6204 6205 off - no mitigation 6206 auto - automatically select a migitation 6207 auto,nosmt - automatically select a mitigation, 6208 disabling SMT if necessary for 6209 the full mitigation (only on Zen1 6210 and older without STIBP). 6211 ibpb - On AMD, mitigate short speculation 6212 windows on basic block boundaries too. 6213 Safe, highest perf impact. It also 6214 enables STIBP if present. Not suitable 6215 on Intel. 6216 ibpb,nosmt - Like "ibpb" above but will disable SMT 6217 when STIBP is not available. This is 6218 the alternative for systems which do not 6219 have STIBP. 6220 unret - Force enable untrained return thunks, 6221 only effective on AMD f15h-f17h based 6222 systems. 6223 unret,nosmt - Like unret, but will disable SMT when STIBP 6224 is not available. This is the alternative for 6225 systems which do not have STIBP. 6226 6227 Selecting 'auto' will choose a mitigation method at run 6228 time according to the CPU. 6229 6230 Not specifying this option is equivalent to retbleed=auto. 6231 6232 rfkill.default_state= 6233 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm, 6234 etc. communication is blocked by default. 6235 1 Unblocked. 6236 6237 rfkill.master_switch_mode= 6238 0 The "airplane mode" button does nothing. 6239 1 The "airplane mode" button toggles between everything 6240 blocked and the previous configuration. 6241 2 The "airplane mode" button toggles between everything 6242 blocked and everything unblocked. 6243 6244 ring3mwait=disable 6245 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported 6246 CPUs. 6247 6248 riscv_isa_fallback [RISCV,EARLY] 6249 When CONFIG_RISCV_ISA_FALLBACK is not enabled, permit 6250 falling back to detecting extension support by parsing 6251 "riscv,isa" property on devicetree systems when the 6252 replacement properties are not found. See the Kconfig 6253 entry for RISCV_ISA_FALLBACK. 6254 6255 ro [KNL] Mount root device read-only on boot 6256 6257 rodata= [KNL,EARLY] 6258 on Mark read-only kernel memory as read-only (default). 6259 off Leave read-only kernel memory writable for debugging. 6260 full Mark read-only kernel memory and aliases as read-only 6261 [arm64] 6262 6263 rockchip.usb_uart 6264 [EARLY] 6265 Enable the uart passthrough on the designated usb port 6266 on Rockchip SoCs. When active, the signals of the 6267 debug-uart get routed to the D+ and D- pins of the usb 6268 port and the regular usb controller gets disabled. 6269 6270 root= [KNL] Root filesystem 6271 Usually this a a block device specifier of some kind, 6272 see the early_lookup_bdev comment in 6273 block/early-lookup.c for details. 6274 Alternatively this can be "ram" for the legacy initial 6275 ramdisk, "nfs" and "cifs" for root on a network file 6276 system, or "mtd" and "ubi" for mounting from raw flash. 6277 6278 rootdelay= [KNL] Delay (in seconds) to pause before attempting to 6279 mount the root filesystem 6280 6281 rootflags= [KNL] Set root filesystem mount option string 6282 6283 rootfstype= [KNL] Set root filesystem type 6284 6285 rootwait [KNL] Wait (indefinitely) for root device to show up. 6286 Useful for devices that are detected asynchronously 6287 (e.g. USB and MMC devices). 6288 6289 rootwait= [KNL] Maximum time (in seconds) to wait for root device 6290 to show up before attempting to mount the root 6291 filesystem. 6292 6293 rproc_mem=nn[KMG][@address] 6294 [KNL,ARM,CMA] Remoteproc physical memory block. 6295 Memory area to be used by remote processor image, 6296 managed by CMA. 6297 6298 rw [KNL] Mount root device read-write on boot 6299 6300 S [KNL] Run init in single mode 6301 6302 s390_iommu= [HW,S390] 6303 Set s390 IOTLB flushing mode 6304 strict 6305 With strict flushing every unmap operation will result 6306 in an IOTLB flush. Default is lazy flushing before 6307 reuse, which is faster. Deprecated, equivalent to 6308 iommu.strict=1. 6309 6310 s390_iommu_aperture= [KNL,S390] 6311 Specifies the size of the per device DMA address space 6312 accessible through the DMA and IOMMU APIs as a decimal 6313 factor of the size of main memory. 6314 The default is 1 meaning that one can concurrently use 6315 as many DMA addresses as physical memory is installed, 6316 if supported by hardware, and thus map all of memory 6317 once. With a value of 2 one can map all of memory twice 6318 and so on. As a special case a factor of 0 imposes no 6319 restrictions other than those given by hardware at the 6320 cost of significant additional memory use for tables. 6321 6322 sa1100ir [NET] 6323 See drivers/net/irda/sa1100_ir.c. 6324 6325 sched_verbose [KNL,EARLY] Enables verbose scheduler debug messages. 6326 6327 schedstats= [KNL,X86] Enable or disable scheduled statistics. 6328 Allowed values are enable and disable. This feature 6329 incurs a small amount of overhead in the scheduler 6330 but is useful for debugging and performance tuning. 6331 6332 sched_thermal_decay_shift= 6333 [Deprecated] 6334 [KNL, SMP] Set a decay shift for scheduler thermal 6335 pressure signal. Thermal pressure signal follows the 6336 default decay period of other scheduler pelt 6337 signals(usually 32 ms but configurable). Setting 6338 sched_thermal_decay_shift will left shift the decay 6339 period for the thermal pressure signal by the shift 6340 value. 6341 i.e. with the default pelt decay period of 32 ms 6342 sched_thermal_decay_shift thermal pressure decay pr 6343 1 64 ms 6344 2 128 ms 6345 and so on. 6346 Format: integer between 0 and 10 6347 Default is 0. 6348 6349 scftorture.holdoff= [KNL] 6350 Number of seconds to hold off before starting 6351 test. Defaults to zero for module insertion and 6352 to 10 seconds for built-in smp_call_function() 6353 tests. 6354 6355 scftorture.longwait= [KNL] 6356 Request ridiculously long waits randomly selected 6357 up to the chosen limit in seconds. Zero (the 6358 default) disables this feature. Please note 6359 that requesting even small non-zero numbers of 6360 seconds can result in RCU CPU stall warnings, 6361 softlockup complaints, and so on. 6362 6363 scftorture.nthreads= [KNL] 6364 Number of kthreads to spawn to invoke the 6365 smp_call_function() family of functions. 6366 The default of -1 specifies a number of kthreads 6367 equal to the number of CPUs. 6368 6369 scftorture.onoff_holdoff= [KNL] 6370 Number seconds to wait after the start of the 6371 test before initiating CPU-hotplug operations. 6372 6373 scftorture.onoff_interval= [KNL] 6374 Number seconds to wait between successive 6375 CPU-hotplug operations. Specifying zero (which 6376 is the default) disables CPU-hotplug operations. 6377 6378 scftorture.shutdown_secs= [KNL] 6379 The number of seconds following the start of the 6380 test after which to shut down the system. The 6381 default of zero avoids shutting down the system. 6382 Non-zero values are useful for automated tests. 6383 6384 scftorture.stat_interval= [KNL] 6385 The number of seconds between outputting the 6386 current test statistics to the console. A value 6387 of zero disables statistics output. 6388 6389 scftorture.stutter_cpus= [KNL] 6390 The number of jiffies to wait between each change 6391 to the set of CPUs under test. 6392 6393 scftorture.use_cpus_read_lock= [KNL] 6394 Use use_cpus_read_lock() instead of the default 6395 preempt_disable() to disable CPU hotplug 6396 while invoking one of the smp_call_function*() 6397 functions. 6398 6399 scftorture.verbose= [KNL] 6400 Enable additional printk() statements. 6401 6402 scftorture.weight_single= [KNL] 6403 The probability weighting to use for the 6404 smp_call_function_single() function with a zero 6405 "wait" parameter. A value of -1 selects the 6406 default if all other weights are -1. However, 6407 if at least one weight has some other value, a 6408 value of -1 will instead select a weight of zero. 6409 6410 scftorture.weight_single_wait= [KNL] 6411 The probability weighting to use for the 6412 smp_call_function_single() function with a 6413 non-zero "wait" parameter. See weight_single. 6414 6415 scftorture.weight_many= [KNL] 6416 The probability weighting to use for the 6417 smp_call_function_many() function with a zero 6418 "wait" parameter. See weight_single. 6419 Note well that setting a high probability for 6420 this weighting can place serious IPI load 6421 on the system. 6422 6423 scftorture.weight_many_wait= [KNL] 6424 The probability weighting to use for the 6425 smp_call_function_many() function with a 6426 non-zero "wait" parameter. See weight_single 6427 and weight_many. 6428 6429 scftorture.weight_all= [KNL] 6430 The probability weighting to use for the 6431 smp_call_function_all() function with a zero 6432 "wait" parameter. See weight_single and 6433 weight_many. 6434 6435 scftorture.weight_all_wait= [KNL] 6436 The probability weighting to use for the 6437 smp_call_function_all() function with a 6438 non-zero "wait" parameter. See weight_single 6439 and weight_many. 6440 6441 sdw_mclk_divider=[SDW] 6442 Specify the MCLK divider for Intel SoundWire buses in 6443 case the BIOS does not provide the clock rate properly. 6444 6445 skew_tick= [KNL,EARLY] Offset the periodic timer tick per cpu to mitigate 6446 xtime_lock contention on larger systems, and/or RCU lock 6447 contention on all systems with CONFIG_MAXSMP set. 6448 Format: { "0" | "1" } 6449 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1" 6450 1 -- enable. 6451 Note: increases power consumption, thus should only be 6452 enabled if running jitter sensitive (HPC/RT) workloads. 6453 6454 security= [SECURITY] Choose a legacy "major" security module to 6455 enable at boot. This has been deprecated by the 6456 "lsm=" parameter. 6457 6458 selinux= [SELINUX] Disable or enable SELinux at boot time. 6459 Format: { "0" | "1" } 6460 See security/selinux/Kconfig help text. 6461 0 -- disable. 6462 1 -- enable. 6463 Default value is 1. 6464 6465 serialnumber [BUGS=X86-32] 6466 6467 sev=option[,option...] [X86-64] 6468 6469 debug 6470 Enable debug messages. 6471 6472 nosnp 6473 Do not enable SEV-SNP (applies to host/hypervisor 6474 only). Setting 'nosnp' avoids the RMP check overhead 6475 in memory accesses when users do not want to run 6476 SEV-SNP guests. 6477 6478 shapers= [NET] 6479 Maximal number of shapers. 6480 6481 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller 6482 Limit apic dumping. The parameter defines the maximal 6483 number of local apics being dumped. Also it is possible 6484 to set it to "all" by meaning -- no limit here. 6485 Format: { 1 (default) | 2 | ... | all }. 6486 The parameter valid if only apic=debug or 6487 apic=verbose is specified. 6488 Example: apic=debug show_lapic=all 6489 6490 slab_debug[=options[,slabs][;[options[,slabs]]...] [MM] 6491 Enabling slab_debug allows one to determine the 6492 culprit if slab objects become corrupted. Enabling 6493 slab_debug can create guard zones around objects and 6494 may poison objects when not in use. Also tracks the 6495 last alloc / free. For more information see 6496 Documentation/mm/slub.rst. 6497 (slub_debug legacy name also accepted for now) 6498 6499 slab_max_order= [MM] 6500 Determines the maximum allowed order for slabs. 6501 A high setting may cause OOMs due to memory 6502 fragmentation. For more information see 6503 Documentation/mm/slub.rst. 6504 (slub_max_order legacy name also accepted for now) 6505 6506 slab_merge [MM] 6507 Enable merging of slabs with similar size when the 6508 kernel is built without CONFIG_SLAB_MERGE_DEFAULT. 6509 (slub_merge legacy name also accepted for now) 6510 6511 slab_min_objects= [MM] 6512 The minimum number of objects per slab. SLUB will 6513 increase the slab order up to slab_max_order to 6514 generate a sufficiently large slab able to contain 6515 the number of objects indicated. The higher the number 6516 of objects the smaller the overhead of tracking slabs 6517 and the less frequently locks need to be acquired. 6518 For more information see Documentation/mm/slub.rst. 6519 (slub_min_objects legacy name also accepted for now) 6520 6521 slab_min_order= [MM] 6522 Determines the minimum page order for slabs. Must be 6523 lower or equal to slab_max_order. For more information see 6524 Documentation/mm/slub.rst. 6525 (slub_min_order legacy name also accepted for now) 6526 6527 slab_nomerge [MM] 6528 Disable merging of slabs with similar size. May be 6529 necessary if there is some reason to distinguish 6530 allocs to different slabs, especially in hardened 6531 environments where the risk of heap overflows and 6532 layout control by attackers can usually be 6533 frustrated by disabling merging. This will reduce 6534 most of the exposure of a heap attack to a single 6535 cache (risks via metadata attacks are mostly 6536 unchanged). Debug options disable merging on their 6537 own. 6538 For more information see Documentation/mm/slub.rst. 6539 (slub_nomerge legacy name also accepted for now) 6540 6541 slab_strict_numa [MM] 6542 Support memory policies on a per object level 6543 in the slab allocator. The default is for memory 6544 policies to be applied at the folio level when 6545 a new folio is needed or a partial folio is 6546 retrieved from the lists. Increases overhead 6547 in the slab fastpaths but gains more accurate 6548 NUMA kernel object placement which helps with slow 6549 interconnects in NUMA systems. 6550 6551 slram= [HW,MTD] 6552 6553 smart2= [HW] 6554 Format: <io1>[,<io2>[,...,<io8>]] 6555 6556 smp.csd_lock_timeout= [KNL] 6557 Specify the period of time in milliseconds 6558 that smp_call_function() and friends will wait 6559 for a CPU to release the CSD lock. This is 6560 useful when diagnosing bugs involving CPUs 6561 disabling interrupts for extended periods 6562 of time. Defaults to 5,000 milliseconds, and 6563 setting a value of zero disables this feature. 6564 This feature may be more efficiently disabled 6565 using the csdlock_debug- kernel parameter. 6566 6567 smp.panic_on_ipistall= [KNL] 6568 If a csd_lock_timeout extends for more than 6569 the specified number of milliseconds, panic the 6570 system. By default, let CSD-lock acquisition 6571 take as long as they take. Specifying 300,000 6572 for this value provides a 5-minute timeout. 6573 6574 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices 6575 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port 6576 smsc-ircc2.ircc_sir= [HW] SIR base I/O port 6577 smsc-ircc2.ircc_fir= [HW] FIR base I/O port 6578 smsc-ircc2.ircc_irq= [HW] IRQ line 6579 smsc-ircc2.ircc_dma= [HW] DMA channel 6580 smsc-ircc2.ircc_transceiver= [HW] Transceiver type: 6581 0: Toshiba Satellite 1800 (GP data pin select) 6582 1: Fast pin select (default) 6583 2: ATC IRMode 6584 6585 smt= [KNL,MIPS,S390,EARLY] Set the maximum number of threads 6586 (logical CPUs) to use per physical CPU on systems 6587 capable of symmetric multithreading (SMT). Will 6588 be capped to the actual hardware limit. 6589 Format: <integer> 6590 Default: -1 (no limit) 6591 6592 softlockup_panic= 6593 [KNL] Should the soft-lockup detector generate panics. 6594 Format: 0 | 1 6595 6596 A value of 1 instructs the soft-lockup detector 6597 to panic the machine when a soft-lockup occurs. It is 6598 also controlled by the kernel.softlockup_panic sysctl 6599 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the 6600 respective build-time switch to that functionality. 6601 6602 softlockup_all_cpu_backtrace= 6603 [KNL] Should the soft-lockup detector generate 6604 backtraces on all cpus. 6605 Format: 0 | 1 6606 6607 sonypi.*= [HW] Sony Programmable I/O Control Device driver 6608 See Documentation/admin-guide/laptops/sonypi.rst 6609 6610 spectre_bhi= [X86] Control mitigation of Branch History Injection 6611 (BHI) vulnerability. This setting affects the 6612 deployment of the HW BHI control and the SW BHB 6613 clearing sequence. 6614 6615 on - (default) Enable the HW or SW mitigation as 6616 needed. This protects the kernel from 6617 both syscalls and VMs. 6618 vmexit - On systems which don't have the HW mitigation 6619 available, enable the SW mitigation on vmexit 6620 ONLY. On such systems, the host kernel is 6621 protected from VM-originated BHI attacks, but 6622 may still be vulnerable to syscall attacks. 6623 off - Disable the mitigation. 6624 6625 spectre_v2= [X86,EARLY] Control mitigation of Spectre variant 2 6626 (indirect branch speculation) vulnerability. 6627 The default operation protects the kernel from 6628 user space attacks. 6629 6630 on - unconditionally enable, implies 6631 spectre_v2_user=on 6632 off - unconditionally disable, implies 6633 spectre_v2_user=off 6634 auto - kernel detects whether your CPU model is 6635 vulnerable 6636 6637 Selecting 'on' will, and 'auto' may, choose a 6638 mitigation method at run time according to the 6639 CPU, the available microcode, the setting of the 6640 CONFIG_MITIGATION_RETPOLINE configuration option, 6641 and the compiler with which the kernel was built. 6642 6643 Selecting 'on' will also enable the mitigation 6644 against user space to user space task attacks. 6645 Selecting specific mitigation does not force enable 6646 user mitigations. 6647 6648 Selecting 'off' will disable both the kernel and 6649 the user space protections. 6650 6651 Specific mitigations can also be selected manually: 6652 6653 retpoline - replace indirect branches 6654 retpoline,generic - Retpolines 6655 retpoline,lfence - LFENCE; indirect branch 6656 retpoline,amd - alias for retpoline,lfence 6657 eibrs - Enhanced/Auto IBRS 6658 eibrs,retpoline - Enhanced/Auto IBRS + Retpolines 6659 eibrs,lfence - Enhanced/Auto IBRS + LFENCE 6660 ibrs - use IBRS to protect kernel 6661 6662 Not specifying this option is equivalent to 6663 spectre_v2=auto. 6664 6665 spectre_v2_user= 6666 [X86] Control mitigation of Spectre variant 2 6667 (indirect branch speculation) vulnerability between 6668 user space tasks 6669 6670 on - Unconditionally enable mitigations. Is 6671 enforced by spectre_v2=on 6672 6673 off - Unconditionally disable mitigations. Is 6674 enforced by spectre_v2=off 6675 6676 prctl - Indirect branch speculation is enabled, 6677 but mitigation can be enabled via prctl 6678 per thread. The mitigation control state 6679 is inherited on fork. 6680 6681 prctl,ibpb 6682 - Like "prctl" above, but only STIBP is 6683 controlled per thread. IBPB is issued 6684 always when switching between different user 6685 space processes. 6686 6687 seccomp 6688 - Same as "prctl" above, but all seccomp 6689 threads will enable the mitigation unless 6690 they explicitly opt out. 6691 6692 seccomp,ibpb 6693 - Like "seccomp" above, but only STIBP is 6694 controlled per thread. IBPB is issued 6695 always when switching between different 6696 user space processes. 6697 6698 auto - Kernel selects the mitigation depending on 6699 the available CPU features and vulnerability. 6700 6701 Default mitigation: "prctl" 6702 6703 Not specifying this option is equivalent to 6704 spectre_v2_user=auto. 6705 6706 spec_rstack_overflow= 6707 [X86,EARLY] Control RAS overflow mitigation on AMD Zen CPUs 6708 6709 off - Disable mitigation 6710 microcode - Enable microcode mitigation only 6711 safe-ret - Enable sw-only safe RET mitigation (default) 6712 ibpb - Enable mitigation by issuing IBPB on 6713 kernel entry 6714 ibpb-vmexit - Issue IBPB only on VMEXIT 6715 (cloud-specific mitigation) 6716 6717 spec_store_bypass_disable= 6718 [HW,EARLY] Control Speculative Store Bypass (SSB) Disable mitigation 6719 (Speculative Store Bypass vulnerability) 6720 6721 Certain CPUs are vulnerable to an exploit against a 6722 a common industry wide performance optimization known 6723 as "Speculative Store Bypass" in which recent stores 6724 to the same memory location may not be observed by 6725 later loads during speculative execution. The idea 6726 is that such stores are unlikely and that they can 6727 be detected prior to instruction retirement at the 6728 end of a particular speculation execution window. 6729 6730 In vulnerable processors, the speculatively forwarded 6731 store can be used in a cache side channel attack, for 6732 example to read memory to which the attacker does not 6733 directly have access (e.g. inside sandboxed code). 6734 6735 This parameter controls whether the Speculative Store 6736 Bypass optimization is used. 6737 6738 On x86 the options are: 6739 6740 on - Unconditionally disable Speculative Store Bypass 6741 off - Unconditionally enable Speculative Store Bypass 6742 auto - Kernel detects whether the CPU model contains an 6743 implementation of Speculative Store Bypass and 6744 picks the most appropriate mitigation. If the 6745 CPU is not vulnerable, "off" is selected. If the 6746 CPU is vulnerable the default mitigation is 6747 architecture and Kconfig dependent. See below. 6748 prctl - Control Speculative Store Bypass per thread 6749 via prctl. Speculative Store Bypass is enabled 6750 for a process by default. The state of the control 6751 is inherited on fork. 6752 seccomp - Same as "prctl" above, but all seccomp threads 6753 will disable SSB unless they explicitly opt out. 6754 6755 Default mitigations: 6756 X86: "prctl" 6757 6758 On powerpc the options are: 6759 6760 on,auto - On Power8 and Power9 insert a store-forwarding 6761 barrier on kernel entry and exit. On Power7 6762 perform a software flush on kernel entry and 6763 exit. 6764 off - No action. 6765 6766 Not specifying this option is equivalent to 6767 spec_store_bypass_disable=auto. 6768 6769 split_lock_detect= 6770 [X86] Enable split lock detection or bus lock detection 6771 6772 When enabled (and if hardware support is present), atomic 6773 instructions that access data across cache line 6774 boundaries will result in an alignment check exception 6775 for split lock detection or a debug exception for 6776 bus lock detection. 6777 6778 off - not enabled 6779 6780 warn - the kernel will emit rate-limited warnings 6781 about applications triggering the #AC 6782 exception or the #DB exception. This mode is 6783 the default on CPUs that support split lock 6784 detection or bus lock detection. Default 6785 behavior is by #AC if both features are 6786 enabled in hardware. 6787 6788 fatal - the kernel will send SIGBUS to applications 6789 that trigger the #AC exception or the #DB 6790 exception. Default behavior is by #AC if 6791 both features are enabled in hardware. 6792 6793 ratelimit:N - 6794 Set system wide rate limit to N bus locks 6795 per second for bus lock detection. 6796 0 < N <= 1000. 6797 6798 N/A for split lock detection. 6799 6800 6801 If an #AC exception is hit in the kernel or in 6802 firmware (i.e. not while executing in user mode) 6803 the kernel will oops in either "warn" or "fatal" 6804 mode. 6805 6806 #DB exception for bus lock is triggered only when 6807 CPL > 0. 6808 6809 srbds= [X86,INTEL,EARLY] 6810 Control the Special Register Buffer Data Sampling 6811 (SRBDS) mitigation. 6812 6813 Certain CPUs are vulnerable to an MDS-like 6814 exploit which can leak bits from the random 6815 number generator. 6816 6817 By default, this issue is mitigated by 6818 microcode. However, the microcode fix can cause 6819 the RDRAND and RDSEED instructions to become 6820 much slower. Among other effects, this will 6821 result in reduced throughput from /dev/urandom. 6822 6823 The microcode mitigation can be disabled with 6824 the following option: 6825 6826 off: Disable mitigation and remove 6827 performance impact to RDRAND and RDSEED 6828 6829 srcutree.big_cpu_lim [KNL] 6830 Specifies the number of CPUs constituting a 6831 large system, such that srcu_struct structures 6832 should immediately allocate an srcu_node array. 6833 This kernel-boot parameter defaults to 128, 6834 but takes effect only when the low-order four 6835 bits of srcutree.convert_to_big is equal to 3 6836 (decide at boot). 6837 6838 srcutree.convert_to_big [KNL] 6839 Specifies under what conditions an SRCU tree 6840 srcu_struct structure will be converted to big 6841 form, that is, with an rcu_node tree: 6842 6843 0: Never. 6844 1: At init_srcu_struct() time. 6845 2: When rcutorture decides to. 6846 3: Decide at boot time (default). 6847 0x1X: Above plus if high contention. 6848 6849 Either way, the srcu_node tree will be sized based 6850 on the actual runtime number of CPUs (nr_cpu_ids) 6851 instead of the compile-time CONFIG_NR_CPUS. 6852 6853 srcutree.counter_wrap_check [KNL] 6854 Specifies how frequently to check for 6855 grace-period sequence counter wrap for the 6856 srcu_data structure's ->srcu_gp_seq_needed field. 6857 The greater the number of bits set in this kernel 6858 parameter, the less frequently counter wrap will 6859 be checked for. Note that the bottom two bits 6860 are ignored. 6861 6862 srcutree.exp_holdoff [KNL] 6863 Specifies how many nanoseconds must elapse 6864 since the end of the last SRCU grace period for 6865 a given srcu_struct until the next normal SRCU 6866 grace period will be considered for automatic 6867 expediting. Set to zero to disable automatic 6868 expediting. 6869 6870 srcutree.srcu_max_nodelay [KNL] 6871 Specifies the number of no-delay instances 6872 per jiffy for which the SRCU grace period 6873 worker thread will be rescheduled with zero 6874 delay. Beyond this limit, worker thread will 6875 be rescheduled with a sleep delay of one jiffy. 6876 6877 srcutree.srcu_max_nodelay_phase [KNL] 6878 Specifies the per-grace-period phase, number of 6879 non-sleeping polls of readers. Beyond this limit, 6880 grace period worker thread will be rescheduled 6881 with a sleep delay of one jiffy, between each 6882 rescan of the readers, for a grace period phase. 6883 6884 srcutree.srcu_retry_check_delay [KNL] 6885 Specifies number of microseconds of non-sleeping 6886 delay between each non-sleeping poll of readers. 6887 6888 srcutree.small_contention_lim [KNL] 6889 Specifies the number of update-side contention 6890 events per jiffy will be tolerated before 6891 initiating a conversion of an srcu_struct 6892 structure to big form. Note that the value of 6893 srcutree.convert_to_big must have the 0x10 bit 6894 set for contention-based conversions to occur. 6895 6896 ssbd= [ARM64,HW,EARLY] 6897 Speculative Store Bypass Disable control 6898 6899 On CPUs that are vulnerable to the Speculative 6900 Store Bypass vulnerability and offer a 6901 firmware based mitigation, this parameter 6902 indicates how the mitigation should be used: 6903 6904 force-on: Unconditionally enable mitigation for 6905 for both kernel and userspace 6906 force-off: Unconditionally disable mitigation for 6907 for both kernel and userspace 6908 kernel: Always enable mitigation in the 6909 kernel, and offer a prctl interface 6910 to allow userspace to register its 6911 interest in being mitigated too. 6912 6913 stack_guard_gap= [MM] 6914 override the default stack gap protection. The value 6915 is in page units and it defines how many pages prior 6916 to (for stacks growing down) resp. after (for stacks 6917 growing up) the main stack are reserved for no other 6918 mapping. Default value is 256 pages. 6919 6920 stack_depot_disable= [KNL,EARLY] 6921 Setting this to true through kernel command line will 6922 disable the stack depot thereby saving the static memory 6923 consumed by the stack hash table. By default this is set 6924 to false. 6925 6926 stacktrace [FTRACE] 6927 Enabled the stack tracer on boot up. 6928 6929 stacktrace_filter=[function-list] 6930 [FTRACE] Limit the functions that the stack tracer 6931 will trace at boot up. function-list is a comma-separated 6932 list of functions. This list can be changed at run 6933 time by the stack_trace_filter file in the debugfs 6934 tracing directory. Note, this enables stack tracing 6935 and the stacktrace above is not needed. 6936 6937 sti= [PARISC,HW] 6938 Format: <num> 6939 Set the STI (builtin display/keyboard on the HP-PARISC 6940 machines) console (graphic card) which should be used 6941 as the initial boot-console. 6942 See also comment in drivers/video/console/sticore.c. 6943 6944 sti_font= [HW] 6945 See comment in drivers/video/console/sticore.c. 6946 6947 stifb= [HW] 6948 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]] 6949 6950 strict_sas_size= 6951 [X86] 6952 Format: <bool> 6953 Enable or disable strict sigaltstack size checks 6954 against the required signal frame size which 6955 depends on the supported FPU features. This can 6956 be used to filter out binaries which have 6957 not yet been made aware of AT_MINSIGSTKSZ. 6958 6959 stress_hpt [PPC,EARLY] 6960 Limits the number of kernel HPT entries in the hash 6961 page table to increase the rate of hash page table 6962 faults on kernel addresses. 6963 6964 stress_slb [PPC,EARLY] 6965 Limits the number of kernel SLB entries, and flushes 6966 them frequently to increase the rate of SLB faults 6967 on kernel addresses. 6968 6969 sunrpc.min_resvport= 6970 sunrpc.max_resvport= 6971 [NFS,SUNRPC] 6972 SunRPC servers often require that client requests 6973 originate from a privileged port (i.e. a port in the 6974 range 0 < portnr < 1024). 6975 An administrator who wishes to reserve some of these 6976 ports for other uses may adjust the range that the 6977 kernel's sunrpc client considers to be privileged 6978 using these two parameters to set the minimum and 6979 maximum port values. 6980 6981 sunrpc.svc_rpc_per_connection_limit= 6982 [NFS,SUNRPC] 6983 Limit the number of requests that the server will 6984 process in parallel from a single connection. 6985 The default value is 0 (no limit). 6986 6987 sunrpc.pool_mode= 6988 [NFS] 6989 Control how the NFS server code allocates CPUs to 6990 service thread pools. Depending on how many NICs 6991 you have and where their interrupts are bound, this 6992 option will affect which CPUs will do NFS serving. 6993 Note: this parameter cannot be changed while the 6994 NFS server is running. 6995 6996 auto the server chooses an appropriate mode 6997 automatically using heuristics 6998 global a single global pool contains all CPUs 6999 percpu one pool for each CPU 7000 pernode one pool for each NUMA node (equivalent 7001 to global on non-NUMA machines) 7002 7003 sunrpc.tcp_slot_table_entries= 7004 sunrpc.udp_slot_table_entries= 7005 [NFS,SUNRPC] 7006 Sets the upper limit on the number of simultaneous 7007 RPC calls that can be sent from the client to a 7008 server. Increasing these values may allow you to 7009 improve throughput, but will also increase the 7010 amount of memory reserved for use by the client. 7011 7012 suspend.pm_test_delay= 7013 [SUSPEND] 7014 Sets the number of seconds to remain in a suspend test 7015 mode before resuming the system (see 7016 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG 7017 is set. Default value is 5. 7018 7019 svm= [PPC] 7020 Format: { on | off | y | n | 1 | 0 } 7021 This parameter controls use of the Protected 7022 Execution Facility on pSeries. 7023 7024 swiotlb= [ARM,PPC,MIPS,X86,S390,EARLY] 7025 Format: { <int> [,<int>] | force | noforce } 7026 <int> -- Number of I/O TLB slabs 7027 <int> -- Second integer after comma. Number of swiotlb 7028 areas with their own lock. Will be rounded up 7029 to a power of 2. 7030 force -- force using of bounce buffers even if they 7031 wouldn't be automatically used by the kernel 7032 noforce -- Never use bounce buffers (for debugging) 7033 7034 switches= [HW,M68k,EARLY] 7035 7036 sysctl.*= [KNL] 7037 Set a sysctl parameter, right before loading the init 7038 process, as if the value was written to the respective 7039 /proc/sys/... file. Both '.' and '/' are recognized as 7040 separators. Unrecognized parameters and invalid values 7041 are reported in the kernel log. Sysctls registered 7042 later by a loaded module cannot be set this way. 7043 Example: sysctl.vm.swappiness=40 7044 7045 sysrq_always_enabled 7046 [KNL] 7047 Ignore sysrq setting - this boot parameter will 7048 neutralize any effect of /proc/sys/kernel/sysrq. 7049 Useful for debugging. 7050 7051 tcpmhash_entries= [KNL,NET] 7052 Set the number of tcp_metrics_hash slots. 7053 Default value is 8192 or 16384 depending on total 7054 ram pages. This is used to specify the TCP metrics 7055 cache size. See Documentation/networking/ip-sysctl.rst 7056 "tcp_no_metrics_save" section for more details. 7057 7058 tdfx= [HW,DRM] 7059 7060 test_suspend= [SUSPEND] 7061 Format: { "mem" | "standby" | "freeze" }[,N] 7062 Specify "mem" (for Suspend-to-RAM) or "standby" (for 7063 standby suspend) or "freeze" (for suspend type freeze) 7064 as the system sleep state during system startup with 7065 the optional capability to repeat N number of times. 7066 The system is woken from this state using a 7067 wakeup-capable RTC alarm. 7068 7069 thash_entries= [KNL,NET] 7070 Set number of hash buckets for TCP connection 7071 7072 thermal.act= [HW,ACPI] 7073 -1: disable all active trip points in all thermal zones 7074 <degrees C>: override all lowest active trip points 7075 7076 thermal.crt= [HW,ACPI] 7077 -1: disable all critical trip points in all thermal zones 7078 <degrees C>: override all critical trip points 7079 7080 thermal.off= [HW,ACPI] 7081 1: disable ACPI thermal control 7082 7083 thermal.psv= [HW,ACPI] 7084 -1: disable all passive trip points 7085 <degrees C>: override all passive trip points to this 7086 value 7087 7088 thermal.tzp= [HW,ACPI] 7089 Specify global default ACPI thermal zone polling rate 7090 <deci-seconds>: poll all this frequency 7091 0: no polling (default) 7092 7093 thp_anon= [KNL] 7094 Format: <size>[KMG],<size>[KMG]:<state>;<size>[KMG]-<size>[KMG]:<state> 7095 state is one of "always", "madvise", "never" or "inherit". 7096 Control the default behavior of the system with respect 7097 to anonymous transparent hugepages. 7098 Can be used multiple times for multiple anon THP sizes. 7099 See Documentation/admin-guide/mm/transhuge.rst for more 7100 details. 7101 7102 threadirqs [KNL,EARLY] 7103 Force threading of all interrupt handlers except those 7104 marked explicitly IRQF_NO_THREAD. 7105 7106 thp_shmem= [KNL] 7107 Format: <size>[KMG],<size>[KMG]:<policy>;<size>[KMG]-<size>[KMG]:<policy> 7108 Control the default policy of each hugepage size for the 7109 internal shmem mount. <policy> is one of policies available 7110 for the shmem mount ("always", "inherit", "never", "within_size", 7111 and "advise"). 7112 It can be used multiple times for multiple shmem THP sizes. 7113 See Documentation/admin-guide/mm/transhuge.rst for more 7114 details. 7115 7116 topology= [S390,EARLY] 7117 Format: {off | on} 7118 Specify if the kernel should make use of the cpu 7119 topology information if the hardware supports this. 7120 The scheduler will make use of this information and 7121 e.g. base its process migration decisions on it. 7122 Default is on. 7123 7124 torture.disable_onoff_at_boot= [KNL] 7125 Prevent the CPU-hotplug component of torturing 7126 until after init has spawned. 7127 7128 torture.ftrace_dump_at_shutdown= [KNL] 7129 Dump the ftrace buffer at torture-test shutdown, 7130 even if there were no errors. This can be a 7131 very costly operation when many torture tests 7132 are running concurrently, especially on systems 7133 with rotating-rust storage. 7134 7135 torture.verbose_sleep_frequency= [KNL] 7136 Specifies how many verbose printk()s should be 7137 emitted between each sleep. The default of zero 7138 disables verbose-printk() sleeping. 7139 7140 torture.verbose_sleep_duration= [KNL] 7141 Duration of each verbose-printk() sleep in jiffies. 7142 7143 tpm.disable_pcr_integrity= [HW,TPM] 7144 Do not protect PCR registers from unintended physical 7145 access, or interposers in the bus by the means of 7146 having an integrity protected session wrapped around 7147 TPM2_PCR_Extend command. Consider this in a situation 7148 where TPM is heavily utilized by IMA, thus protection 7149 causing a major performance hit, and the space where 7150 machines are deployed is by other means guarded. 7151 7152 tpm_suspend_pcr=[HW,TPM] 7153 Format: integer pcr id 7154 Specify that at suspend time, the tpm driver 7155 should extend the specified pcr with zeros, 7156 as a workaround for some chips which fail to 7157 flush the last written pcr on TPM_SaveState. 7158 This will guarantee that all the other pcrs 7159 are saved. 7160 7161 tpm_tis.interrupts= [HW,TPM] 7162 Enable interrupts for the MMIO based physical layer 7163 for the FIFO interface. By default it is set to false 7164 (0). For more information about TPM hardware interfaces 7165 defined by Trusted Computing Group (TCG) see 7166 https://trustedcomputinggroup.org/resource/pc-client-platform-tpm-profile-ptp-specification/ 7167 7168 tp_printk [FTRACE] 7169 Have the tracepoints sent to printk as well as the 7170 tracing ring buffer. This is useful for early boot up 7171 where the system hangs or reboots and does not give the 7172 option for reading the tracing buffer or performing a 7173 ftrace_dump_on_oops. 7174 7175 To turn off having tracepoints sent to printk, 7176 echo 0 > /proc/sys/kernel/tracepoint_printk 7177 Note, echoing 1 into this file without the 7178 tp_printk kernel cmdline option has no effect. 7179 7180 The tp_printk_stop_on_boot (see below) can also be used 7181 to stop the printing of events to console at 7182 late_initcall_sync. 7183 7184 ** CAUTION ** 7185 7186 Having tracepoints sent to printk() and activating high 7187 frequency tracepoints such as irq or sched, can cause 7188 the system to live lock. 7189 7190 tp_printk_stop_on_boot [FTRACE] 7191 When tp_printk (above) is set, it can cause a lot of noise 7192 on the console. It may be useful to only include the 7193 printing of events during boot up, as user space may 7194 make the system inoperable. 7195 7196 This command line option will stop the printing of events 7197 to console at the late_initcall_sync() time frame. 7198 7199 trace_buf_size=nn[KMG] 7200 [FTRACE] will set tracing buffer size on each cpu. 7201 7202 trace_clock= [FTRACE] Set the clock used for tracing events 7203 at boot up. 7204 local - Use the per CPU time stamp counter 7205 (converted into nanoseconds). Fast, but 7206 depending on the architecture, may not be 7207 in sync between CPUs. 7208 global - Event time stamps are synchronize across 7209 CPUs. May be slower than the local clock, 7210 but better for some race conditions. 7211 counter - Simple counting of events (1, 2, ..) 7212 note, some counts may be skipped due to the 7213 infrastructure grabbing the clock more than 7214 once per event. 7215 uptime - Use jiffies as the time stamp. 7216 perf - Use the same clock that perf uses. 7217 mono - Use ktime_get_mono_fast_ns() for time stamps. 7218 mono_raw - Use ktime_get_raw_fast_ns() for time 7219 stamps. 7220 boot - Use ktime_get_boot_fast_ns() for time stamps. 7221 Architectures may add more clocks. See 7222 Documentation/trace/ftrace.rst for more details. 7223 7224 trace_event=[event-list] 7225 [FTRACE] Set and start specified trace events in order 7226 to facilitate early boot debugging. The event-list is a 7227 comma-separated list of trace events to enable. See 7228 also Documentation/trace/events.rst 7229 7230 To enable modules, use :mod: keyword: 7231 7232 trace_event=:mod:<module> 7233 7234 The value before :mod: will only enable specific events 7235 that are part of the module. See the above mentioned 7236 document for more information. 7237 7238 trace_instance=[instance-info] 7239 [FTRACE] Create a ring buffer instance early in boot up. 7240 This will be listed in: 7241 7242 /sys/kernel/tracing/instances 7243 7244 Events can be enabled at the time the instance is created 7245 via: 7246 7247 trace_instance=<name>,<system1>:<event1>,<system2>:<event2> 7248 7249 Note, the "<system*>:" portion is optional if the event is 7250 unique. 7251 7252 trace_instance=foo,sched:sched_switch,irq_handler_entry,initcall 7253 7254 will enable the "sched_switch" event (note, the "sched:" is optional, and 7255 the same thing would happen if it was left off). The irq_handler_entry 7256 event, and all events under the "initcall" system. 7257 7258 Flags can be added to the instance to modify its behavior when it is 7259 created. The flags are separated by '^'. 7260 7261 The available flags are: 7262 7263 traceoff - Have the tracing instance tracing disabled after it is created. 7264 traceprintk - Have trace_printk() write into this trace instance 7265 (note, "printk" and "trace_printk" can also be used) 7266 7267 trace_instance=foo^traceoff^traceprintk,sched,irq 7268 7269 The flags must come before the defined events. 7270 7271 If memory has been reserved (see memmap for x86), the instance 7272 can use that memory: 7273 7274 memmap=12M$0x284500000 trace_instance=boot_map@0x284500000:12M 7275 7276 The above will create a "boot_map" instance that uses the physical 7277 memory at 0x284500000 that is 12Megs. The per CPU buffers of that 7278 instance will be split up accordingly. 7279 7280 Alternatively, the memory can be reserved by the reserve_mem option: 7281 7282 reserve_mem=12M:4096:trace trace_instance=boot_map@trace 7283 7284 This will reserve 12 megabytes at boot up with a 4096 byte alignment 7285 and place the ring buffer in this memory. Note that due to KASLR, the 7286 memory may not be the same location each time, which will not preserve 7287 the buffer content. 7288 7289 Also note that the layout of the ring buffer data may change between 7290 kernel versions where the validator will fail and reset the ring buffer 7291 if the layout is not the same as the previous kernel. 7292 7293 If the ring buffer is used for persistent bootups and has events enabled, 7294 it is recommend to disable tracing so that events from a previous boot do not 7295 mix with events of the current boot (unless you are debugging a random crash 7296 at boot up). 7297 7298 reserve_mem=12M:4096:trace trace_instance=boot_map^traceoff^traceprintk@trace,sched,irq 7299 7300 Note, saving the trace buffer across reboots does require that the system 7301 is set up to not wipe memory. For instance, CONFIG_RESET_ATTACK_MITIGATION 7302 can force a memory reset on boot which will clear any trace that was stored. 7303 This is just one of many ways that can clear memory. Make sure your system 7304 keeps the content of memory across reboots before relying on this option. 7305 7306 NB: Both the mapped address and size must be page aligned for the architecture. 7307 7308 See also Documentation/trace/debugging.rst 7309 7310 7311 trace_options=[option-list] 7312 [FTRACE] Enable or disable tracer options at boot. 7313 The option-list is a comma delimited list of options 7314 that can be enabled or disabled just as if you were 7315 to echo the option name into 7316 7317 /sys/kernel/tracing/trace_options 7318 7319 For example, to enable stacktrace option (to dump the 7320 stack trace of each event), add to the command line: 7321 7322 trace_options=stacktrace 7323 7324 See also Documentation/trace/ftrace.rst "trace options" 7325 section. 7326 7327 trace_trigger=[trigger-list] 7328 [FTRACE] Add a event trigger on specific events. 7329 Set a trigger on top of a specific event, with an optional 7330 filter. 7331 7332 The format is is "trace_trigger=<event>.<trigger>[ if <filter>],..." 7333 Where more than one trigger may be specified that are comma deliminated. 7334 7335 For example: 7336 7337 trace_trigger="sched_switch.stacktrace if prev_state == 2" 7338 7339 The above will enable the "stacktrace" trigger on the "sched_switch" 7340 event but only trigger it if the "prev_state" of the "sched_switch" 7341 event is "2" (TASK_UNINTERUPTIBLE). 7342 7343 See also "Event triggers" in Documentation/trace/events.rst 7344 7345 7346 traceoff_after_boot 7347 [FTRACE] Sometimes tracing is used to debug issues 7348 during the boot process. Since the trace buffer has a 7349 limited amount of storage, it may be prudent to 7350 disable tracing after the boot is finished, otherwise 7351 the critical information may be overwritten. With this 7352 option, the main tracing buffer will be turned off at 7353 the end of the boot process. 7354 7355 traceoff_on_warning 7356 [FTRACE] enable this option to disable tracing when a 7357 warning is hit. This turns off "tracing_on". Tracing can 7358 be enabled again by echoing '1' into the "tracing_on" 7359 file located in /sys/kernel/tracing/ 7360 7361 This option is useful, as it disables the trace before 7362 the WARNING dump is called, which prevents the trace to 7363 be filled with content caused by the warning output. 7364 7365 This option can also be set at run time via the sysctl 7366 option: kernel/traceoff_on_warning 7367 7368 transparent_hugepage= 7369 [KNL] 7370 Format: [always|madvise|never] 7371 Can be used to control the default behavior of the system 7372 with respect to transparent hugepages. 7373 See Documentation/admin-guide/mm/transhuge.rst 7374 for more details. 7375 7376 transparent_hugepage_shmem= [KNL] 7377 Format: [always|within_size|advise|never|deny|force] 7378 Can be used to control the hugepage allocation policy for 7379 the internal shmem mount. 7380 See Documentation/admin-guide/mm/transhuge.rst 7381 for more details. 7382 7383 transparent_hugepage_tmpfs= [KNL] 7384 Format: [always|within_size|advise|never] 7385 Can be used to control the default hugepage allocation policy 7386 for the tmpfs mount. 7387 See Documentation/admin-guide/mm/transhuge.rst 7388 for more details. 7389 7390 trusted.source= [KEYS] 7391 Format: <string> 7392 This parameter identifies the trust source as a backend 7393 for trusted keys implementation. Supported trust 7394 sources: 7395 - "tpm" 7396 - "tee" 7397 - "caam" 7398 - "dcp" 7399 If not specified then it defaults to iterating through 7400 the trust source list starting with TPM and assigns the 7401 first trust source as a backend which is initialized 7402 successfully during iteration. 7403 7404 trusted.rng= [KEYS] 7405 Format: <string> 7406 The RNG used to generate key material for trusted keys. 7407 Can be one of: 7408 - "kernel" 7409 - the same value as trusted.source: "tpm" or "tee" 7410 - "default" 7411 If not specified, "default" is used. In this case, 7412 the RNG's choice is left to each individual trust source. 7413 7414 trusted.dcp_use_otp_key 7415 This is intended to be used in combination with 7416 trusted.source=dcp and will select the DCP OTP key 7417 instead of the DCP UNIQUE key blob encryption. 7418 7419 trusted.dcp_skip_zk_test 7420 This is intended to be used in combination with 7421 trusted.source=dcp and will disable the check if the 7422 blob key is all zeros. This is helpful for situations where 7423 having this key zero'ed is acceptable. E.g. in testing 7424 scenarios. 7425 7426 tsc= Disable clocksource stability checks for TSC. 7427 Format: <string> 7428 [x86] reliable: mark tsc clocksource as reliable, this 7429 disables clocksource verification at runtime, as well 7430 as the stability checks done at bootup. Used to enable 7431 high-resolution timer mode on older hardware, and in 7432 virtualized environment. 7433 [x86] noirqtime: Do not use TSC to do irq accounting. 7434 Used to run time disable IRQ_TIME_ACCOUNTING on any 7435 platforms where RDTSC is slow and this accounting 7436 can add overhead. 7437 [x86] unstable: mark the TSC clocksource as unstable, this 7438 marks the TSC unconditionally unstable at bootup and 7439 avoids any further wobbles once the TSC watchdog notices. 7440 [x86] nowatchdog: disable clocksource watchdog. Used 7441 in situations with strict latency requirements (where 7442 interruptions from clocksource watchdog are not 7443 acceptable). 7444 [x86] recalibrate: force recalibration against a HW timer 7445 (HPET or PM timer) on systems whose TSC frequency was 7446 obtained from HW or FW using either an MSR or CPUID(0x15). 7447 Warn if the difference is more than 500 ppm. 7448 [x86] watchdog: Use TSC as the watchdog clocksource with 7449 which to check other HW timers (HPET or PM timer), but 7450 only on systems where TSC has been deemed trustworthy. 7451 This will be suppressed by an earlier tsc=nowatchdog and 7452 can be overridden by a later tsc=nowatchdog. A console 7453 message will flag any such suppression or overriding. 7454 7455 tsc_early_khz= [X86,EARLY] Skip early TSC calibration and use the given 7456 value instead. Useful when the early TSC frequency discovery 7457 procedure is not reliable, such as on overclocked systems 7458 with CPUID.16h support and partial CPUID.15h support. 7459 Format: <unsigned int> 7460 7461 tsx= [X86] Control Transactional Synchronization 7462 Extensions (TSX) feature in Intel processors that 7463 support TSX control. 7464 7465 This parameter controls the TSX feature. The options are: 7466 7467 on - Enable TSX on the system. Although there are 7468 mitigations for all known security vulnerabilities, 7469 TSX has been known to be an accelerator for 7470 several previous speculation-related CVEs, and 7471 so there may be unknown security risks associated 7472 with leaving it enabled. 7473 7474 off - Disable TSX on the system. (Note that this 7475 option takes effect only on newer CPUs which are 7476 not vulnerable to MDS, i.e., have 7477 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get 7478 the new IA32_TSX_CTRL MSR through a microcode 7479 update. This new MSR allows for the reliable 7480 deactivation of the TSX functionality.) 7481 7482 auto - Disable TSX if X86_BUG_TAA is present, 7483 otherwise enable TSX on the system. 7484 7485 Not specifying this option is equivalent to tsx=off. 7486 7487 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst 7488 for more details. 7489 7490 tsx_async_abort= [X86,INTEL,EARLY] Control mitigation for the TSX Async 7491 Abort (TAA) vulnerability. 7492 7493 Similar to Micro-architectural Data Sampling (MDS) 7494 certain CPUs that support Transactional 7495 Synchronization Extensions (TSX) are vulnerable to an 7496 exploit against CPU internal buffers which can forward 7497 information to a disclosure gadget under certain 7498 conditions. 7499 7500 In vulnerable processors, the speculatively forwarded 7501 data can be used in a cache side channel attack, to 7502 access data to which the attacker does not have direct 7503 access. 7504 7505 This parameter controls the TAA mitigation. The 7506 options are: 7507 7508 full - Enable TAA mitigation on vulnerable CPUs 7509 if TSX is enabled. 7510 7511 full,nosmt - Enable TAA mitigation and disable SMT on 7512 vulnerable CPUs. If TSX is disabled, SMT 7513 is not disabled because CPU is not 7514 vulnerable to cross-thread TAA attacks. 7515 off - Unconditionally disable TAA mitigation 7516 7517 On MDS-affected machines, tsx_async_abort=off can be 7518 prevented by an active MDS mitigation as both vulnerabilities 7519 are mitigated with the same mechanism so in order to disable 7520 this mitigation, you need to specify mds=off too. 7521 7522 Not specifying this option is equivalent to 7523 tsx_async_abort=full. On CPUs which are MDS affected 7524 and deploy MDS mitigation, TAA mitigation is not 7525 required and doesn't provide any additional 7526 mitigation. 7527 7528 For details see: 7529 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst 7530 7531 turbografx.map[2|3]= [HW,JOY] 7532 TurboGraFX parallel port interface 7533 Format: 7534 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7> 7535 See also Documentation/input/devices/joystick-parport.rst 7536 7537 udbg-immortal [PPC] When debugging early kernel crashes that 7538 happen after console_init() and before a proper 7539 console driver takes over, this boot options might 7540 help "seeing" what's going on. 7541 7542 uhash_entries= [KNL,NET] 7543 Set number of hash buckets for UDP/UDP-Lite connections 7544 7545 uhci-hcd.ignore_oc= 7546 [USB] Ignore overcurrent events (default N). 7547 Some badly-designed motherboards generate lots of 7548 bogus events, for ports that aren't wired to 7549 anything. Set this parameter to avoid log spamming. 7550 Note that genuine overcurrent events won't be 7551 reported either. 7552 7553 unaligned_scalar_speed= 7554 [RISCV] 7555 Format: {slow | fast | unsupported} 7556 Allow skipping scalar unaligned access speed tests. This 7557 is useful for testing alternative code paths and to skip 7558 the tests in environments where they run too slowly. All 7559 CPUs must have the same scalar unaligned access speed. 7560 7561 unaligned_vector_speed= 7562 [RISCV] 7563 Format: {slow | fast | unsupported} 7564 Allow skipping vector unaligned access speed tests. This 7565 is useful for testing alternative code paths and to skip 7566 the tests in environments where they run too slowly. All 7567 CPUs must have the same vector unaligned access speed. 7568 7569 unknown_nmi_panic 7570 [X86] Cause panic on unknown NMI. 7571 7572 unwind_debug [X86-64,EARLY] 7573 Enable unwinder debug output. This can be 7574 useful for debugging certain unwinder error 7575 conditions, including corrupt stacks and 7576 bad/missing unwinder metadata. 7577 7578 usbcore.authorized_default= 7579 [USB] Default USB device authorization: 7580 (default -1 = authorized (same as 1), 7581 0 = not authorized, 1 = authorized, 2 = authorized 7582 if device connected to internal port) 7583 7584 usbcore.autosuspend= 7585 [USB] The autosuspend time delay (in seconds) used 7586 for newly-detected USB devices (default 2). This 7587 is the time required before an idle device will be 7588 autosuspended. Devices for which the delay is set 7589 to a negative value won't be autosuspended at all. 7590 7591 usbcore.usbfs_snoop= 7592 [USB] Set to log all usbfs traffic (default 0 = off). 7593 7594 usbcore.usbfs_snoop_max= 7595 [USB] Maximum number of bytes to snoop in each URB 7596 (default = 65536). 7597 7598 usbcore.blinkenlights= 7599 [USB] Set to cycle leds on hubs (default 0 = off). 7600 7601 usbcore.old_scheme_first= 7602 [USB] Start with the old device initialization 7603 scheme (default 0 = off). 7604 7605 usbcore.usbfs_memory_mb= 7606 [USB] Memory limit (in MB) for buffers allocated by 7607 usbfs (default = 16, 0 = max = 2047). 7608 7609 usbcore.use_both_schemes= 7610 [USB] Try the other device initialization scheme 7611 if the first one fails (default 1 = enabled). 7612 7613 usbcore.initial_descriptor_timeout= 7614 [USB] Specifies timeout for the initial 64-byte 7615 USB_REQ_GET_DESCRIPTOR request in milliseconds 7616 (default 5000 = 5.0 seconds). 7617 7618 usbcore.nousb [USB] Disable the USB subsystem 7619 7620 usbcore.quirks= 7621 [USB] A list of quirk entries to augment the built-in 7622 usb core quirk list. List entries are separated by 7623 commas. Each entry has the form 7624 VendorID:ProductID:Flags. The IDs are 4-digit hex 7625 numbers and Flags is a set of letters. Each letter 7626 will change the built-in quirk; setting it if it is 7627 clear and clearing it if it is set. The letters have 7628 the following meanings: 7629 a = USB_QUIRK_STRING_FETCH_255 (string 7630 descriptors must not be fetched using 7631 a 255-byte read); 7632 b = USB_QUIRK_RESET_RESUME (device can't resume 7633 correctly so reset it instead); 7634 c = USB_QUIRK_NO_SET_INTF (device can't handle 7635 Set-Interface requests); 7636 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't 7637 handle its Configuration or Interface 7638 strings); 7639 e = USB_QUIRK_RESET (device can't be reset 7640 (e.g morph devices), don't use reset); 7641 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has 7642 more interface descriptions than the 7643 bNumInterfaces count, and can't handle 7644 talking to these interfaces); 7645 g = USB_QUIRK_DELAY_INIT (device needs a pause 7646 during initialization, after we read 7647 the device descriptor); 7648 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For 7649 high speed and super speed interrupt 7650 endpoints, the USB 2.0 and USB 3.0 spec 7651 require the interval in microframes (1 7652 microframe = 125 microseconds) to be 7653 calculated as interval = 2 ^ 7654 (bInterval-1). 7655 Devices with this quirk report their 7656 bInterval as the result of this 7657 calculation instead of the exponent 7658 variable used in the calculation); 7659 i = USB_QUIRK_DEVICE_QUALIFIER (device can't 7660 handle device_qualifier descriptor 7661 requests); 7662 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device 7663 generates spurious wakeup, ignore 7664 remote wakeup capability); 7665 k = USB_QUIRK_NO_LPM (device can't handle Link 7666 Power Management); 7667 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL 7668 (Device reports its bInterval as linear 7669 frames instead of the USB 2.0 7670 calculation); 7671 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs 7672 to be disconnected before suspend to 7673 prevent spurious wakeup); 7674 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a 7675 pause after every control message); 7676 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra 7677 delay after resetting its port); 7678 p = USB_QUIRK_SHORT_SET_ADDRESS_REQ_TIMEOUT 7679 (Reduce timeout of the SET_ADDRESS 7680 request from 5000 ms to 500 ms); 7681 Example: quirks=0781:5580:bk,0a5c:5834:gij 7682 7683 usbhid.mousepoll= 7684 [USBHID] The interval which mice are to be polled at. 7685 7686 usbhid.jspoll= 7687 [USBHID] The interval which joysticks are to be polled at. 7688 7689 usbhid.kbpoll= 7690 [USBHID] The interval which keyboards are to be polled at. 7691 7692 usb-storage.delay_use= 7693 [UMS] The delay in seconds before a new device is 7694 scanned for Logical Units (default 1). 7695 Optionally the delay in milliseconds if the value has 7696 suffix with "ms". 7697 Example: delay_use=2567ms 7698 7699 usb-storage.quirks= 7700 [UMS] A list of quirks entries to supplement or 7701 override the built-in unusual_devs list. List 7702 entries are separated by commas. Each entry has 7703 the form VID:PID:Flags where VID and PID are Vendor 7704 and Product ID values (4-digit hex numbers) and 7705 Flags is a set of characters, each corresponding 7706 to a common usb-storage quirk flag as follows: 7707 a = SANE_SENSE (collect more than 18 bytes 7708 of sense data, not on uas); 7709 b = BAD_SENSE (don't collect more than 18 7710 bytes of sense data, not on uas); 7711 c = FIX_CAPACITY (decrease the reported 7712 device capacity by one sector); 7713 d = NO_READ_DISC_INFO (don't use 7714 READ_DISC_INFO command, not on uas); 7715 e = NO_READ_CAPACITY_16 (don't use 7716 READ_CAPACITY_16 command); 7717 f = NO_REPORT_OPCODES (don't use report opcodes 7718 command, uas only); 7719 g = MAX_SECTORS_240 (don't transfer more than 7720 240 sectors at a time, uas only); 7721 h = CAPACITY_HEURISTICS (decrease the 7722 reported device capacity by one 7723 sector if the number is odd); 7724 i = IGNORE_DEVICE (don't bind to this 7725 device); 7726 j = NO_REPORT_LUNS (don't use report luns 7727 command, uas only); 7728 k = NO_SAME (do not use WRITE_SAME, uas only) 7729 l = NOT_LOCKABLE (don't try to lock and 7730 unlock ejectable media, not on uas); 7731 m = MAX_SECTORS_64 (don't transfer more 7732 than 64 sectors = 32 KB at a time, 7733 not on uas); 7734 n = INITIAL_READ10 (force a retry of the 7735 initial READ(10) command, not on uas); 7736 o = CAPACITY_OK (accept the capacity 7737 reported by the device, not on uas); 7738 p = WRITE_CACHE (the device cache is ON 7739 by default, not on uas); 7740 r = IGNORE_RESIDUE (the device reports 7741 bogus residue values, not on uas); 7742 s = SINGLE_LUN (the device has only one 7743 Logical Unit); 7744 t = NO_ATA_1X (don't allow ATA(12) and ATA(16) 7745 commands, uas only); 7746 u = IGNORE_UAS (don't bind to the uas driver); 7747 w = NO_WP_DETECT (don't test whether the 7748 medium is write-protected). 7749 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE 7750 even if the device claims no cache, 7751 not on uas) 7752 Example: quirks=0419:aaf5:rl,0421:0433:rc 7753 7754 user_debug= [KNL,ARM] 7755 Format: <int> 7756 See arch/arm/Kconfig.debug help text. 7757 1 - undefined instruction events 7758 2 - system calls 7759 4 - invalid data aborts 7760 8 - SIGSEGV faults 7761 16 - SIGBUS faults 7762 Example: user_debug=31 7763 7764 vdso= [X86,SH,SPARC] 7765 On X86_32, this is an alias for vdso32=. Otherwise: 7766 7767 vdso=1: enable VDSO (the default) 7768 vdso=0: disable VDSO mapping 7769 7770 vdso32= [X86] Control the 32-bit vDSO 7771 vdso32=1: enable 32-bit VDSO 7772 vdso32=0 or vdso32=2: disable 32-bit VDSO 7773 7774 See the help text for CONFIG_COMPAT_VDSO for more 7775 details. If CONFIG_COMPAT_VDSO is set, the default is 7776 vdso32=0; otherwise, the default is vdso32=1. 7777 7778 For compatibility with older kernels, vdso32=2 is an 7779 alias for vdso32=0. 7780 7781 Try vdso32=0 if you encounter an error that says: 7782 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed! 7783 7784 video= [FB,EARLY] Frame buffer configuration 7785 See Documentation/fb/modedb.rst. 7786 7787 video.brightness_switch_enabled= [ACPI] 7788 Format: [0|1] 7789 If set to 1, on receiving an ACPI notify event 7790 generated by hotkey, video driver will adjust brightness 7791 level and then send out the event to user space through 7792 the allocated input device. If set to 0, video driver 7793 will only send out the event without touching backlight 7794 brightness level. 7795 default: 1 7796 7797 virtio_mmio.device= 7798 [VMMIO] Memory mapped virtio (platform) device. 7799 7800 <size>@<baseaddr>:<irq>[:<id>] 7801 where: 7802 <size> := size (can use standard suffixes 7803 like K, M and G) 7804 <baseaddr> := physical base address 7805 <irq> := interrupt number (as passed to 7806 request_irq()) 7807 <id> := (optional) platform device id 7808 example: 7809 virtio_mmio.device=1K@0x100b0000:48:7 7810 7811 Can be used multiple times for multiple devices. 7812 7813 vga= [BOOT,X86-32] Select a particular video mode 7814 See Documentation/arch/x86/boot.rst and 7815 Documentation/admin-guide/svga.rst. 7816 Use vga=ask for menu. 7817 This is actually a boot loader parameter; the value is 7818 passed to the kernel using a special protocol. 7819 7820 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y. 7821 May slow down system boot speed, especially when 7822 enabled on systems with a large amount of memory. 7823 All options are enabled by default, and this 7824 interface is meant to allow for selectively 7825 enabling or disabling specific virtual memory 7826 debugging features. 7827 7828 Available options are: 7829 P Enable page structure init time poisoning 7830 - Disable all of the above options 7831 7832 vmalloc=nn[KMG] [KNL,BOOT,EARLY] Forces the vmalloc area to have an 7833 exact size of <nn>. This can be used to increase 7834 the minimum size (128MB on x86, arm32 platforms). 7835 It can also be used to decrease the size and leave more room 7836 for directly mapped kernel RAM. Note that this parameter does 7837 not exist on many other platforms (including arm64, alpha, 7838 loongarch, arc, csky, hexagon, microblaze, mips, nios2, openrisc, 7839 parisc, m64k, powerpc, riscv, sh, um, xtensa, s390, sparc). 7840 7841 vmcp_cma=nn[MG] [KNL,S390,EARLY] 7842 Sets the memory size reserved for contiguous memory 7843 allocations for the vmcp device driver. 7844 7845 vmhalt= [KNL,S390] Perform z/VM CP command after system halt. 7846 Format: <command> 7847 7848 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic. 7849 Format: <command> 7850 7851 vmpoff= [KNL,S390] Perform z/VM CP command after power off. 7852 Format: <command> 7853 7854 vsyscall= [X86-64,EARLY] 7855 Controls the behavior of vsyscalls (i.e. calls to 7856 fixed addresses of 0xffffffffff600x00 from legacy 7857 code). Most statically-linked binaries and older 7858 versions of glibc use these calls. Because these 7859 functions are at fixed addresses, they make nice 7860 targets for exploits that can control RIP. 7861 7862 emulate Vsyscalls turn into traps and are emulated 7863 reasonably safely. The vsyscall page is 7864 readable. 7865 7866 xonly [default] Vsyscalls turn into traps and are 7867 emulated reasonably safely. The vsyscall 7868 page is not readable. 7869 7870 none Vsyscalls don't work at all. This makes 7871 them quite hard to use for exploits but 7872 might break your system. 7873 7874 vt.color= [VT] Default text color. 7875 Format: 0xYX, X = foreground, Y = background. 7876 Default: 0x07 = light gray on black. 7877 7878 vt.cur_default= [VT] Default cursor shape. 7879 Format: 0xCCBBAA, where AA, BB, and CC are the same as 7880 the parameters of the <Esc>[?A;B;Cc escape sequence; 7881 see vga-softcursor.rst. Default: 2 = underline. 7882 7883 vt.default_blu= [VT] 7884 Format: <blue0>,<blue1>,<blue2>,...,<blue15> 7885 Change the default blue palette of the console. 7886 This is a 16-member array composed of values 7887 ranging from 0-255. 7888 7889 vt.default_grn= [VT] 7890 Format: <green0>,<green1>,<green2>,...,<green15> 7891 Change the default green palette of the console. 7892 This is a 16-member array composed of values 7893 ranging from 0-255. 7894 7895 vt.default_red= [VT] 7896 Format: <red0>,<red1>,<red2>,...,<red15> 7897 Change the default red palette of the console. 7898 This is a 16-member array composed of values 7899 ranging from 0-255. 7900 7901 vt.default_utf8= 7902 [VT] 7903 Format=<0|1> 7904 Set system-wide default UTF-8 mode for all tty's. 7905 Default is 1, i.e. UTF-8 mode is enabled for all 7906 newly opened terminals. 7907 7908 vt.global_cursor_default= 7909 [VT] 7910 Format=<-1|0|1> 7911 Set system-wide default for whether a cursor 7912 is shown on new VTs. Default is -1, 7913 i.e. cursors will be created by default unless 7914 overridden by individual drivers. 0 will hide 7915 cursors, 1 will display them. 7916 7917 vt.italic= [VT] Default color for italic text; 0-15. 7918 Default: 2 = green. 7919 7920 vt.underline= [VT] Default color for underlined text; 0-15. 7921 Default: 3 = cyan. 7922 7923 watchdog timers [HW,WDT] For information on watchdog timers, 7924 see Documentation/watchdog/watchdog-parameters.rst 7925 or other driver-specific files in the 7926 Documentation/watchdog/ directory. 7927 7928 watchdog_thresh= 7929 [KNL] 7930 Set the hard lockup detector stall duration 7931 threshold in seconds. The soft lockup detector 7932 threshold is set to twice the value. A value of 0 7933 disables both lockup detectors. Default is 10 7934 seconds. 7935 7936 workqueue.unbound_cpus= 7937 [KNL,SMP] Specify to constrain one or some CPUs 7938 to use in unbound workqueues. 7939 Format: <cpu-list> 7940 By default, all online CPUs are available for 7941 unbound workqueues. 7942 7943 workqueue.watchdog_thresh= 7944 If CONFIG_WQ_WATCHDOG is configured, workqueue can 7945 warn stall conditions and dump internal state to 7946 help debugging. 0 disables workqueue stall 7947 detection; otherwise, it's the stall threshold 7948 duration in seconds. The default value is 30 and 7949 it can be updated at runtime by writing to the 7950 corresponding sysfs file. 7951 7952 workqueue.panic_on_stall=<uint> 7953 Panic when workqueue stall is detected by 7954 CONFIG_WQ_WATCHDOG. It sets the number times of the 7955 stall to trigger panic. 7956 7957 The default is 0, which disables the panic on stall. 7958 7959 workqueue.cpu_intensive_thresh_us= 7960 Per-cpu work items which run for longer than this 7961 threshold are automatically considered CPU intensive 7962 and excluded from concurrency management to prevent 7963 them from noticeably delaying other per-cpu work 7964 items. Default is 10000 (10ms). 7965 7966 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel 7967 will report the work functions which violate this 7968 threshold repeatedly. They are likely good 7969 candidates for using WQ_UNBOUND workqueues instead. 7970 7971 workqueue.cpu_intensive_warning_thresh=<uint> 7972 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel 7973 will report the work functions which violate the 7974 intensive_threshold_us repeatedly. In order to prevent 7975 spurious warnings, start printing only after a work 7976 function has violated this threshold number of times. 7977 7978 The default is 4 times. 0 disables the warning. 7979 7980 workqueue.power_efficient 7981 Per-cpu workqueues are generally preferred because 7982 they show better performance thanks to cache 7983 locality; unfortunately, per-cpu workqueues tend to 7984 be more power hungry than unbound workqueues. 7985 7986 Enabling this makes the per-cpu workqueues which 7987 were observed to contribute significantly to power 7988 consumption unbound, leading to measurably lower 7989 power usage at the cost of small performance 7990 overhead. 7991 7992 The default value of this parameter is determined by 7993 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT. 7994 7995 workqueue.default_affinity_scope= 7996 Select the default affinity scope to use for unbound 7997 workqueues. Can be one of "cpu", "smt", "cache", 7998 "numa" and "system". Default is "cache". For more 7999 information, see the Affinity Scopes section in 8000 Documentation/core-api/workqueue.rst. 8001 8002 This can be changed after boot by writing to the 8003 matching /sys/module/workqueue/parameters file. All 8004 workqueues with the "default" affinity scope will be 8005 updated accordingly. 8006 8007 workqueue.debug_force_rr_cpu 8008 Workqueue used to implicitly guarantee that work 8009 items queued without explicit CPU specified are put 8010 on the local CPU. This guarantee is no longer true 8011 and while local CPU is still preferred work items 8012 may be put on foreign CPUs. This debug option 8013 forces round-robin CPU selection to flush out 8014 usages which depend on the now broken guarantee. 8015 When enabled, memory and cache locality will be 8016 impacted. 8017 8018 writecombine= [LOONGARCH,EARLY] Control the MAT (Memory Access 8019 Type) of ioremap_wc(). 8020 8021 on - Enable writecombine, use WUC for ioremap_wc() 8022 off - Disable writecombine, use SUC for ioremap_wc() 8023 8024 x2apic_phys [X86-64,APIC,EARLY] Use x2apic physical mode instead of 8025 default x2apic cluster mode on platforms 8026 supporting x2apic. 8027 8028 xen_512gb_limit [KNL,X86-64,XEN] 8029 Restricts the kernel running paravirtualized under Xen 8030 to use only up to 512 GB of RAM. The reason to do so is 8031 crash analysis tools and Xen tools for doing domain 8032 save/restore/migration must be enabled to handle larger 8033 domains. 8034 8035 xen_emul_unplug= [HW,X86,XEN,EARLY] 8036 Unplug Xen emulated devices 8037 Format: [unplug0,][unplug1] 8038 ide-disks -- unplug primary master IDE devices 8039 aux-ide-disks -- unplug non-primary-master IDE devices 8040 nics -- unplug network devices 8041 all -- unplug all emulated devices (NICs and IDE disks) 8042 unnecessary -- unplugging emulated devices is 8043 unnecessary even if the host did not respond to 8044 the unplug protocol 8045 never -- do not unplug even if version check succeeds 8046 8047 xen_legacy_crash [X86,XEN,EARLY] 8048 Crash from Xen panic notifier, without executing late 8049 panic() code such as dumping handler. 8050 8051 xen_mc_debug [X86,XEN,EARLY] 8052 Enable multicall debugging when running as a Xen PV guest. 8053 Enabling this feature will reduce performance a little 8054 bit, so it should only be enabled for obtaining extended 8055 debug data in case of multicall errors. 8056 8057 xen_msr_safe= [X86,XEN,EARLY] 8058 Format: <bool> 8059 Select whether to always use non-faulting (safe) MSR 8060 access functions when running as Xen PV guest. The 8061 default value is controlled by CONFIG_XEN_PV_MSR_SAFE. 8062 8063 xen_nopv [X86] 8064 Disables the PV optimizations forcing the HVM guest to 8065 run as generic HVM guest with no PV drivers. 8066 This option is obsoleted by the "nopv" option, which 8067 has equivalent effect for XEN platform. 8068 8069 xen_no_vector_callback 8070 [KNL,X86,XEN,EARLY] Disable the vector callback for Xen 8071 event channel interrupts. 8072 8073 xen_scrub_pages= [XEN] 8074 Boolean option to control scrubbing pages before giving them back 8075 to Xen, for use by other domains. Can be also changed at runtime 8076 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages. 8077 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT. 8078 8079 xen_timer_slop= [X86-64,XEN,EARLY] 8080 Set the timer slop (in nanoseconds) for the virtual Xen 8081 timers (default is 100000). This adjusts the minimum 8082 delta of virtualized Xen timers, where lower values 8083 improve timer resolution at the expense of processing 8084 more timer interrupts. 8085 8086 xen.balloon_boot_timeout= [XEN] 8087 The time (in seconds) to wait before giving up to boot 8088 in case initial ballooning fails to free enough memory. 8089 Applies only when running as HVM or PVH guest and 8090 started with less memory configured than allowed at 8091 max. Default is 180. 8092 8093 xen.event_eoi_delay= [XEN] 8094 How long to delay EOI handling in case of event 8095 storms (jiffies). Default is 10. 8096 8097 xen.event_loop_timeout= [XEN] 8098 After which time (jiffies) the event handling loop 8099 should start to delay EOI handling. Default is 2. 8100 8101 xen.fifo_events= [XEN] 8102 Boolean parameter to disable using fifo event handling 8103 even if available. Normally fifo event handling is 8104 preferred over the 2-level event handling, as it is 8105 fairer and the number of possible event channels is 8106 much higher. Default is on (use fifo events). 8107 8108 xirc2ps_cs= [NET,PCMCIA] 8109 Format: 8110 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]] 8111 8112 xive= [PPC] 8113 By default on POWER9 and above, the kernel will 8114 natively use the XIVE interrupt controller. This option 8115 allows the fallback firmware mode to be used: 8116 8117 off Fallback to firmware control of XIVE interrupt 8118 controller on both pseries and powernv 8119 platforms. Only useful on POWER9 and above. 8120 8121 xive.store-eoi=off [PPC] 8122 By default on POWER10 and above, the kernel will use 8123 stores for EOI handling when the XIVE interrupt mode 8124 is active. This option allows the XIVE driver to use 8125 loads instead, as on POWER9. 8126 8127 xhci-hcd.quirks [USB,KNL] 8128 A hex value specifying bitmask with supplemental xhci 8129 host controller quirks. Meaning of each bit can be 8130 consulted in header drivers/usb/host/xhci.h. 8131 8132 xmon [PPC,EARLY] 8133 Format: { early | on | rw | ro | off } 8134 Controls if xmon debugger is enabled. Default is off. 8135 Passing only "xmon" is equivalent to "xmon=early". 8136 early Call xmon as early as possible on boot; xmon 8137 debugger is called from setup_arch(). 8138 on xmon debugger hooks will be installed so xmon 8139 is only called on a kernel crash. Default mode, 8140 i.e. either "ro" or "rw" mode, is controlled 8141 with CONFIG_XMON_DEFAULT_RO_MODE. 8142 rw xmon debugger hooks will be installed so xmon 8143 is called only on a kernel crash, mode is write, 8144 meaning SPR registers, memory and, other data 8145 can be written using xmon commands. 8146 ro same as "rw" option above but SPR registers, 8147 memory, and other data can't be written using 8148 xmon commands. 8149 off xmon is disabled. 8150