1 acpi= [HW,ACPI,X86,ARM64] 2 Advanced Configuration and Power Interface 3 Format: { force | on | off | strict | noirq | rsdt | 4 copy_dsdt } 5 force -- enable ACPI if default was off 6 on -- enable ACPI but allow fallback to DT [arm64] 7 off -- disable ACPI if default was on 8 noirq -- do not use ACPI for IRQ routing 9 strict -- Be less tolerant of platforms that are not 10 strictly ACPI specification compliant. 11 rsdt -- prefer RSDT over (default) XSDT 12 copy_dsdt -- copy DSDT to memory 13 For ARM64, ONLY "acpi=off", "acpi=on" or "acpi=force" 14 are available 15 16 See also Documentation/power/runtime_pm.rst, pci=noacpi 17 18 acpi_apic_instance= [ACPI, IOAPIC] 19 Format: <int> 20 2: use 2nd APIC table, if available 21 1,0: use 1st APIC table 22 default: 0 23 24 acpi_backlight= [HW,ACPI] 25 { vendor | video | native | none } 26 If set to vendor, prefer vendor-specific driver 27 (e.g. thinkpad_acpi, sony_acpi, etc.) instead 28 of the ACPI video.ko driver. 29 If set to video, use the ACPI video.ko driver. 30 If set to native, use the device's native backlight mode. 31 If set to none, disable the ACPI backlight interface. 32 33 acpi_force_32bit_fadt_addr 34 force FADT to use 32 bit addresses rather than the 35 64 bit X_* addresses. Some firmware have broken 64 36 bit addresses for force ACPI ignore these and use 37 the older legacy 32 bit addresses. 38 39 acpica_no_return_repair [HW, ACPI] 40 Disable AML predefined validation mechanism 41 This mechanism can repair the evaluation result to make 42 the return objects more ACPI specification compliant. 43 This option is useful for developers to identify the 44 root cause of an AML interpreter issue when the issue 45 has something to do with the repair mechanism. 46 47 acpi.debug_layer= [HW,ACPI,ACPI_DEBUG] 48 acpi.debug_level= [HW,ACPI,ACPI_DEBUG] 49 Format: <int> 50 CONFIG_ACPI_DEBUG must be enabled to produce any ACPI 51 debug output. Bits in debug_layer correspond to a 52 _COMPONENT in an ACPI source file, e.g., 53 #define _COMPONENT ACPI_PCI_COMPONENT 54 Bits in debug_level correspond to a level in 55 ACPI_DEBUG_PRINT statements, e.g., 56 ACPI_DEBUG_PRINT((ACPI_DB_INFO, ... 57 The debug_level mask defaults to "info". See 58 Documentation/firmware-guide/acpi/debug.rst for more information about 59 debug layers and levels. 60 61 Enable processor driver info messages: 62 acpi.debug_layer=0x20000000 63 Enable PCI/PCI interrupt routing info messages: 64 acpi.debug_layer=0x400000 65 Enable AML "Debug" output, i.e., stores to the Debug 66 object while interpreting AML: 67 acpi.debug_layer=0xffffffff acpi.debug_level=0x2 68 Enable all messages related to ACPI hardware: 69 acpi.debug_layer=0x2 acpi.debug_level=0xffffffff 70 71 Some values produce so much output that the system is 72 unusable. The "log_buf_len" parameter may be useful 73 if you need to capture more output. 74 75 acpi_enforce_resources= [ACPI] 76 { strict | lax | no } 77 Check for resource conflicts between native drivers 78 and ACPI OperationRegions (SystemIO and SystemMemory 79 only). IO ports and memory declared in ACPI might be 80 used by the ACPI subsystem in arbitrary AML code and 81 can interfere with legacy drivers. 82 strict (default): access to resources claimed by ACPI 83 is denied; legacy drivers trying to access reserved 84 resources will fail to bind to device using them. 85 lax: access to resources claimed by ACPI is allowed; 86 legacy drivers trying to access reserved resources 87 will bind successfully but a warning message is logged. 88 no: ACPI OperationRegions are not marked as reserved, 89 no further checks are performed. 90 91 acpi_force_table_verification [HW,ACPI] 92 Enable table checksum verification during early stage. 93 By default, this is disabled due to x86 early mapping 94 size limitation. 95 96 acpi_irq_balance [HW,ACPI] 97 ACPI will balance active IRQs 98 default in APIC mode 99 100 acpi_irq_nobalance [HW,ACPI] 101 ACPI will not move active IRQs (default) 102 default in PIC mode 103 104 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA 105 Format: <irq>,<irq>... 106 107 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for 108 use by PCI 109 Format: <irq>,<irq>... 110 111 acpi_mask_gpe= [HW,ACPI] 112 Due to the existence of _Lxx/_Exx, some GPEs triggered 113 by unsupported hardware/firmware features can result in 114 GPE floodings that cannot be automatically disabled by 115 the GPE dispatcher. 116 This facility can be used to prevent such uncontrolled 117 GPE floodings. 118 Format: <byte> 119 120 acpi_no_auto_serialize [HW,ACPI] 121 Disable auto-serialization of AML methods 122 AML control methods that contain the opcodes to create 123 named objects will be marked as "Serialized" by the 124 auto-serialization feature. 125 This feature is enabled by default. 126 This option allows to turn off the feature. 127 128 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump 129 kernels. 130 131 acpi_no_static_ssdt [HW,ACPI] 132 Disable installation of static SSDTs at early boot time 133 By default, SSDTs contained in the RSDT/XSDT will be 134 installed automatically and they will appear under 135 /sys/firmware/acpi/tables. 136 This option turns off this feature. 137 Note that specifying this option does not affect 138 dynamic table installation which will install SSDT 139 tables to /sys/firmware/acpi/tables/dynamic. 140 141 acpi_no_watchdog [HW,ACPI,WDT] 142 Ignore the ACPI-based watchdog interface (WDAT) and let 143 a native driver control the watchdog device instead. 144 145 acpi_rsdp= [ACPI,EFI,KEXEC] 146 Pass the RSDP address to the kernel, mostly used 147 on machines running EFI runtime service to boot the 148 second kernel for kdump. 149 150 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS 151 Format: To spoof as Windows 98: ="Microsoft Windows" 152 153 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead 154 of 2 which is mandated by ACPI 6) as the supported ACPI 155 specification revision (when using this switch, it may 156 be necessary to carry out a cold reboot _twice_ in a 157 row to make it take effect on the platform firmware). 158 159 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings 160 acpi_osi="string1" # add string1 161 acpi_osi="!string2" # remove string2 162 acpi_osi=!* # remove all strings 163 acpi_osi=! # disable all built-in OS vendor 164 strings 165 acpi_osi=!! # enable all built-in OS vendor 166 strings 167 acpi_osi= # disable all strings 168 169 'acpi_osi=!' can be used in combination with single or 170 multiple 'acpi_osi="string1"' to support specific OS 171 vendor string(s). Note that such command can only 172 affect the default state of the OS vendor strings, thus 173 it cannot affect the default state of the feature group 174 strings and the current state of the OS vendor strings, 175 specifying it multiple times through kernel command line 176 is meaningless. This command is useful when one do not 177 care about the state of the feature group strings which 178 should be controlled by the OSPM. 179 Examples: 180 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent 181 to 'acpi_osi="Windows 2000" acpi_osi=!', they all 182 can make '_OSI("Windows 2000")' TRUE. 183 184 'acpi_osi=' cannot be used in combination with other 185 'acpi_osi=' command lines, the _OSI method will not 186 exist in the ACPI namespace. NOTE that such command can 187 only affect the _OSI support state, thus specifying it 188 multiple times through kernel command line is also 189 meaningless. 190 Examples: 191 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)' 192 FALSE. 193 194 'acpi_osi=!*' can be used in combination with single or 195 multiple 'acpi_osi="string1"' to support specific 196 string(s). Note that such command can affect the 197 current state of both the OS vendor strings and the 198 feature group strings, thus specifying it multiple times 199 through kernel command line is meaningful. But it may 200 still not able to affect the final state of a string if 201 there are quirks related to this string. This command 202 is useful when one want to control the state of the 203 feature group strings to debug BIOS issues related to 204 the OSPM features. 205 Examples: 206 1. 'acpi_osi="Module Device" acpi_osi=!*' can make 207 '_OSI("Module Device")' FALSE. 208 2. 'acpi_osi=!* acpi_osi="Module Device"' can make 209 '_OSI("Module Device")' TRUE. 210 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is 211 equivalent to 212 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"' 213 and 214 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!', 215 they all will make '_OSI("Windows 2000")' TRUE. 216 217 acpi_pm_good [X86] 218 Override the pmtimer bug detection: force the kernel 219 to assume that this machine's pmtimer latches its value 220 and always returns good values. 221 222 acpi_sci= [HW,ACPI] ACPI System Control Interrupt trigger mode 223 Format: { level | edge | high | low } 224 225 acpi_skip_timer_override [HW,ACPI] 226 Recognize and ignore IRQ0/pin2 Interrupt Override. 227 For broken nForce2 BIOS resulting in XT-PIC timer. 228 229 acpi_sleep= [HW,ACPI] Sleep options 230 Format: { s3_bios, s3_mode, s3_beep, s4_nohwsig, 231 old_ordering, nonvs, sci_force_enable, nobl } 232 See Documentation/power/video.rst for information on 233 s3_bios and s3_mode. 234 s3_beep is for debugging; it makes the PC's speaker beep 235 as soon as the kernel's real-mode entry point is called. 236 s4_nohwsig prevents ACPI hardware signature from being 237 used during resume from hibernation. 238 old_ordering causes the ACPI 1.0 ordering of the _PTS 239 control method, with respect to putting devices into 240 low power states, to be enforced (the ACPI 2.0 ordering 241 of _PTS is used by default). 242 nonvs prevents the kernel from saving/restoring the 243 ACPI NVS memory during suspend/hibernation and resume. 244 sci_force_enable causes the kernel to set SCI_EN directly 245 on resume from S1/S3 (which is against the ACPI spec, 246 but some broken systems don't work without it). 247 nobl causes the internal blacklist of systems known to 248 behave incorrectly in some ways with respect to system 249 suspend and resume to be ignored (use wisely). 250 251 acpi_use_timer_override [HW,ACPI] 252 Use timer override. For some broken Nvidia NF5 boards 253 that require a timer override, but don't have HPET 254 255 add_efi_memmap [EFI; X86] Include EFI memory map in 256 kernel's map of available physical RAM. 257 258 agp= [AGP] 259 { off | try_unsupported } 260 off: disable AGP support 261 try_unsupported: try to drive unsupported chipsets 262 (may crash computer or cause data corruption) 263 264 ALSA [HW,ALSA] 265 See Documentation/sound/alsa-configuration.rst 266 267 alignment= [KNL,ARM] 268 Allow the default userspace alignment fault handler 269 behaviour to be specified. Bit 0 enables warnings, 270 bit 1 enables fixups, and bit 2 sends a segfault. 271 272 align_va_addr= [X86-64] 273 Align virtual addresses by clearing slice [14:12] when 274 allocating a VMA at process creation time. This option 275 gives you up to 3% performance improvement on AMD F15h 276 machines (where it is enabled by default) for a 277 CPU-intensive style benchmark, and it can vary highly in 278 a microbenchmark depending on workload and compiler. 279 280 32: only for 32-bit processes 281 64: only for 64-bit processes 282 on: enable for both 32- and 64-bit processes 283 off: disable for both 32- and 64-bit processes 284 285 alloc_snapshot [FTRACE] 286 Allocate the ftrace snapshot buffer on boot up when the 287 main buffer is allocated. This is handy if debugging 288 and you need to use tracing_snapshot() on boot up, and 289 do not want to use tracing_snapshot_alloc() as it needs 290 to be done where GFP_KERNEL allocations are allowed. 291 292 amd_iommu= [HW,X86-64] 293 Pass parameters to the AMD IOMMU driver in the system. 294 Possible values are: 295 fullflush - enable flushing of IO/TLB entries when 296 they are unmapped. Otherwise they are 297 flushed before they will be reused, which 298 is a lot of faster 299 off - do not initialize any AMD IOMMU found in 300 the system 301 force_isolation - Force device isolation for all 302 devices. The IOMMU driver is not 303 allowed anymore to lift isolation 304 requirements as needed. This option 305 does not override iommu=pt 306 307 amd_iommu_dump= [HW,X86-64] 308 Enable AMD IOMMU driver option to dump the ACPI table 309 for AMD IOMMU. With this option enabled, AMD IOMMU 310 driver will print ACPI tables for AMD IOMMU during 311 IOMMU initialization. 312 313 amd_iommu_intr= [HW,X86-64] 314 Specifies one of the following AMD IOMMU interrupt 315 remapping modes: 316 legacy - Use legacy interrupt remapping mode. 317 vapic - Use virtual APIC mode, which allows IOMMU 318 to inject interrupts directly into guest. 319 This mode requires kvm-amd.avic=1. 320 (Default when IOMMU HW support is present.) 321 322 amijoy.map= [HW,JOY] Amiga joystick support 323 Map of devices attached to JOY0DAT and JOY1DAT 324 Format: <a>,<b> 325 See also Documentation/input/joydev/joystick.rst 326 327 analog.map= [HW,JOY] Analog joystick and gamepad support 328 Specifies type or capabilities of an analog joystick 329 connected to one of 16 gameports 330 Format: <type1>,<type2>,..<type16> 331 332 apc= [HW,SPARC] 333 Power management functions (SPARCstation-4/5 + deriv.) 334 Format: noidle 335 Disable APC CPU standby support. SPARCstation-Fox does 336 not play well with APC CPU idle - disable it if you have 337 APC and your system crashes randomly. 338 339 apic= [APIC,X86] Advanced Programmable Interrupt Controller 340 Change the output verbosity while booting 341 Format: { quiet (default) | verbose | debug } 342 Change the amount of debugging information output 343 when initialising the APIC and IO-APIC components. 344 For X86-32, this can also be used to specify an APIC 345 driver name. 346 Format: apic=driver_name 347 Examples: apic=bigsmp 348 349 apic_extnmi= [APIC,X86] External NMI delivery setting 350 Format: { bsp (default) | all | none } 351 bsp: External NMI is delivered only to CPU 0 352 all: External NMIs are broadcast to all CPUs as a 353 backup of CPU 0 354 none: External NMI is masked for all CPUs. This is 355 useful so that a dump capture kernel won't be 356 shot down by NMI 357 358 autoconf= [IPV6] 359 See Documentation/networking/ipv6.rst. 360 361 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller 362 Limit apic dumping. The parameter defines the maximal 363 number of local apics being dumped. Also it is possible 364 to set it to "all" by meaning -- no limit here. 365 Format: { 1 (default) | 2 | ... | all }. 366 The parameter valid if only apic=debug or 367 apic=verbose is specified. 368 Example: apic=debug show_lapic=all 369 370 apm= [APM] Advanced Power Management 371 See header of arch/x86/kernel/apm_32.c. 372 373 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards 374 Format: <io>,<irq>,<nodeID> 375 376 ataflop= [HW,M68k] 377 378 atarimouse= [HW,MOUSE] Atari Mouse 379 380 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess, 381 EzKey and similar keyboards 382 383 atkbd.reset= [HW] Reset keyboard during initialization 384 385 atkbd.set= [HW] Select keyboard code set 386 Format: <int> (2 = AT (default), 3 = PS/2) 387 388 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar 389 keyboards 390 391 atkbd.softraw= [HW] Choose between synthetic and real raw mode 392 Format: <bool> (0 = real, 1 = synthetic (default)) 393 394 atkbd.softrepeat= [HW] 395 Use software keyboard repeat 396 397 audit= [KNL] Enable the audit sub-system 398 Format: { "0" | "1" | "off" | "on" } 399 0 | off - kernel audit is disabled and can not be 400 enabled until the next reboot 401 unset - kernel audit is initialized but disabled and 402 will be fully enabled by the userspace auditd. 403 1 | on - kernel audit is initialized and partially 404 enabled, storing at most audit_backlog_limit 405 messages in RAM until it is fully enabled by the 406 userspace auditd. 407 Default: unset 408 409 audit_backlog_limit= [KNL] Set the audit queue size limit. 410 Format: <int> (must be >=0) 411 Default: 64 412 413 bau= [X86_UV] Enable the BAU on SGI UV. The default 414 behavior is to disable the BAU (i.e. bau=0). 415 Format: { "0" | "1" } 416 0 - Disable the BAU. 417 1 - Enable the BAU. 418 unset - Disable the BAU. 419 420 baycom_epp= [HW,AX25] 421 Format: <io>,<mode> 422 423 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem 424 Format: <io>,<mode> 425 See header of drivers/net/hamradio/baycom_par.c. 426 427 baycom_ser_fdx= [HW,AX25] 428 BayCom Serial Port AX.25 Modem (Full Duplex Mode) 429 Format: <io>,<irq>,<mode>[,<baud>] 430 See header of drivers/net/hamradio/baycom_ser_fdx.c. 431 432 baycom_ser_hdx= [HW,AX25] 433 BayCom Serial Port AX.25 Modem (Half Duplex Mode) 434 Format: <io>,<irq>,<mode> 435 See header of drivers/net/hamradio/baycom_ser_hdx.c. 436 437 blkdevparts= Manual partition parsing of block device(s) for 438 embedded devices based on command line input. 439 See Documentation/block/cmdline-partition.rst 440 441 boot_delay= Milliseconds to delay each printk during boot. 442 Values larger than 10 seconds (10000) are changed to 443 no delay (0). 444 Format: integer 445 446 bootconfig [KNL] 447 Extended command line options can be added to an initrd 448 and this will cause the kernel to look for it. 449 450 See Documentation/admin-guide/bootconfig.rst 451 452 bert_disable [ACPI] 453 Disable BERT OS support on buggy BIOSes. 454 455 bgrt_disable [ACPI][X86] 456 Disable BGRT to avoid flickering OEM logo. 457 458 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards) 459 bttv.radio= Most important insmod options are available as 460 kernel args too. 461 bttv.pll= See Documentation/admin-guide/media/bttv.rst 462 bttv.tuner= 463 464 bulk_remove=off [PPC] This parameter disables the use of the pSeries 465 firmware feature for flushing multiple hpte entries 466 at a time. 467 468 c101= [NET] Moxa C101 synchronous serial card 469 470 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection. 471 Sometimes CPU hardware bugs make them report the cache 472 size incorrectly. The kernel will attempt work arounds 473 to fix known problems, but for some CPUs it is not 474 possible to determine what the correct size should be. 475 This option provides an override for these situations. 476 477 carrier_timeout= 478 [NET] Specifies amount of time (in seconds) that 479 the kernel should wait for a network carrier. By default 480 it waits 120 seconds. 481 482 ca_keys= [KEYS] This parameter identifies a specific key(s) on 483 the system trusted keyring to be used for certificate 484 trust validation. 485 format: { id:<keyid> | builtin } 486 487 cca= [MIPS] Override the kernel pages' cache coherency 488 algorithm. Accepted values range from 0 to 7 489 inclusive. See arch/mips/include/asm/pgtable-bits.h 490 for platform specific values (SB1, Loongson3 and 491 others). 492 493 ccw_timeout_log [S390] 494 See Documentation/s390/common_io.rst for details. 495 496 cgroup_disable= [KNL] Disable a particular controller 497 Format: {name of the controller(s) to disable} 498 The effects of cgroup_disable=foo are: 499 - foo isn't auto-mounted if you mount all cgroups in 500 a single hierarchy 501 - foo isn't visible as an individually mountable 502 subsystem 503 {Currently only "memory" controller deal with this and 504 cut the overhead, others just disable the usage. So 505 only cgroup_disable=memory is actually worthy} 506 507 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1 508 Format: { { controller | "all" | "named" } 509 [,{ controller | "all" | "named" }...] } 510 Like cgroup_disable, but only applies to cgroup v1; 511 the blacklisted controllers remain available in cgroup2. 512 "all" blacklists all controllers and "named" disables 513 named mounts. Specifying both "all" and "named" disables 514 all v1 hierarchies. 515 516 cgroup.memory= [KNL] Pass options to the cgroup memory controller. 517 Format: <string> 518 nosocket -- Disable socket memory accounting. 519 nokmem -- Disable kernel memory accounting. 520 521 checkreqprot [SELINUX] Set initial checkreqprot flag value. 522 Format: { "0" | "1" } 523 See security/selinux/Kconfig help text. 524 0 -- check protection applied by kernel (includes 525 any implied execute protection). 526 1 -- check protection requested by application. 527 Default value is set via a kernel config option. 528 Value can be changed at runtime via 529 /sys/fs/selinux/checkreqprot. 530 Setting checkreqprot to 1 is deprecated. 531 532 cio_ignore= [S390] 533 See Documentation/s390/common_io.rst for details. 534 clk_ignore_unused 535 [CLK] 536 Prevents the clock framework from automatically gating 537 clocks that have not been explicitly enabled by a Linux 538 device driver but are enabled in hardware at reset or 539 by the bootloader/firmware. Note that this does not 540 force such clocks to be always-on nor does it reserve 541 those clocks in any way. This parameter is useful for 542 debug and development, but should not be needed on a 543 platform with proper driver support. For more 544 information, see Documentation/driver-api/clk.rst. 545 546 clock= [BUGS=X86-32, HW] gettimeofday clocksource override. 547 [Deprecated] 548 Forces specified clocksource (if available) to be used 549 when calculating gettimeofday(). If specified 550 clocksource is not available, it defaults to PIT. 551 Format: { pit | tsc | cyclone | pmtmr } 552 553 clocksource= Override the default clocksource 554 Format: <string> 555 Override the default clocksource and use the clocksource 556 with the name specified. 557 Some clocksource names to choose from, depending on 558 the platform: 559 [all] jiffies (this is the base, fallback clocksource) 560 [ACPI] acpi_pm 561 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2, 562 pxa_timer,timer3,32k_counter,timer0_1 563 [X86-32] pit,hpet,tsc; 564 scx200_hrt on Geode; cyclone on IBM x440 565 [MIPS] MIPS 566 [PARISC] cr16 567 [S390] tod 568 [SH] SuperH 569 [SPARC64] tick 570 [X86-64] hpet,tsc 571 572 clocksource.arm_arch_timer.evtstrm= 573 [ARM,ARM64] 574 Format: <bool> 575 Enable/disable the eventstream feature of the ARM 576 architected timer so that code using WFE-based polling 577 loops can be debugged more effectively on production 578 systems. 579 580 clearcpuid=BITNUM[,BITNUM...] [X86] 581 Disable CPUID feature X for the kernel. See 582 arch/x86/include/asm/cpufeatures.h for the valid bit 583 numbers. Note the Linux specific bits are not necessarily 584 stable over kernel options, but the vendor specific 585 ones should be. 586 Also note that user programs calling CPUID directly 587 or using the feature without checking anything 588 will still see it. This just prevents it from 589 being used by the kernel or shown in /proc/cpuinfo. 590 Also note the kernel might malfunction if you disable 591 some critical bits. 592 593 cma=nn[MG]@[start[MG][-end[MG]]] 594 [KNL,CMA] 595 Sets the size of kernel global memory area for 596 contiguous memory allocations and optionally the 597 placement constraint by the physical address range of 598 memory allocations. A value of 0 disables CMA 599 altogether. For more information, see 600 kernel/dma/contiguous.c 601 602 cma_pernuma=nn[MG] 603 [ARM64,KNL] 604 Sets the size of kernel per-numa memory area for 605 contiguous memory allocations. A value of 0 disables 606 per-numa CMA altogether. And If this option is not 607 specificed, the default value is 0. 608 With per-numa CMA enabled, DMA users on node nid will 609 first try to allocate buffer from the pernuma area 610 which is located in node nid, if the allocation fails, 611 they will fallback to the global default memory area. 612 613 cmo_free_hint= [PPC] Format: { yes | no } 614 Specify whether pages are marked as being inactive 615 when they are freed. This is used in CMO environments 616 to determine OS memory pressure for page stealing by 617 a hypervisor. 618 Default: yes 619 620 coherent_pool=nn[KMG] [ARM,KNL] 621 Sets the size of memory pool for coherent, atomic dma 622 allocations, by default set to 256K. 623 624 com20020= [HW,NET] ARCnet - COM20020 chipset 625 Format: 626 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]] 627 628 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers) 629 Format: <io>[,<irq>] 630 631 com90xx= [HW,NET] 632 ARCnet - COM90xx chipset (memory-mapped buffers) 633 Format: <io>[,<irq>[,<memstart>]] 634 635 condev= [HW,S390] console device 636 conmode= 637 638 console= [KNL] Output console device and options. 639 640 tty<n> Use the virtual console device <n>. 641 642 ttyS<n>[,options] 643 ttyUSB0[,options] 644 Use the specified serial port. The options are of 645 the form "bbbbpnf", where "bbbb" is the baud rate, 646 "p" is parity ("n", "o", or "e"), "n" is number of 647 bits, and "f" is flow control ("r" for RTS or 648 omit it). Default is "9600n8". 649 650 See Documentation/admin-guide/serial-console.rst for more 651 information. See 652 Documentation/networking/netconsole.rst for an 653 alternative. 654 655 uart[8250],io,<addr>[,options] 656 uart[8250],mmio,<addr>[,options] 657 uart[8250],mmio16,<addr>[,options] 658 uart[8250],mmio32,<addr>[,options] 659 uart[8250],0x<addr>[,options] 660 Start an early, polled-mode console on the 8250/16550 661 UART at the specified I/O port or MMIO address, 662 switching to the matching ttyS device later. 663 MMIO inter-register address stride is either 8-bit 664 (mmio), 16-bit (mmio16), or 32-bit (mmio32). 665 If none of [io|mmio|mmio16|mmio32], <addr> is assumed 666 to be equivalent to 'mmio'. 'options' are specified in 667 the same format described for ttyS above; if unspecified, 668 the h/w is not re-initialized. 669 670 hvc<n> Use the hypervisor console device <n>. This is for 671 both Xen and PowerPC hypervisors. 672 673 If the device connected to the port is not a TTY but a braille 674 device, prepend "brl," before the device type, for instance 675 console=brl,ttyS0 676 For now, only VisioBraille is supported. 677 678 console_msg_format= 679 [KNL] Change console messages format 680 default 681 By default we print messages on consoles in 682 "[time stamp] text\n" format (time stamp may not be 683 printed, depending on CONFIG_PRINTK_TIME or 684 `printk_time' param). 685 syslog 686 Switch to syslog format: "<%u>[time stamp] text\n" 687 IOW, each message will have a facility and loglevel 688 prefix. The format is similar to one used by syslog() 689 syscall, or to executing "dmesg -S --raw" or to reading 690 from /proc/kmsg. 691 692 consoleblank= [KNL] The console blank (screen saver) timeout in 693 seconds. A value of 0 disables the blank timer. 694 Defaults to 0. 695 696 coredump_filter= 697 [KNL] Change the default value for 698 /proc/<pid>/coredump_filter. 699 See also Documentation/filesystems/proc.rst. 700 701 coresight_cpu_debug.enable 702 [ARM,ARM64] 703 Format: <bool> 704 Enable/disable the CPU sampling based debugging. 705 0: default value, disable debugging 706 1: enable debugging at boot time 707 708 cpuidle.off=1 [CPU_IDLE] 709 disable the cpuidle sub-system 710 711 cpuidle.governor= 712 [CPU_IDLE] Name of the cpuidle governor to use. 713 714 cpufreq.off=1 [CPU_FREQ] 715 disable the cpufreq sub-system 716 717 cpufreq.default_governor= 718 [CPU_FREQ] Name of the default cpufreq governor or 719 policy to use. This governor must be registered in the 720 kernel before the cpufreq driver probes. 721 722 cpu_init_udelay=N 723 [X86] Delay for N microsec between assert and de-assert 724 of APIC INIT to start processors. This delay occurs 725 on every CPU online, such as boot, and resume from suspend. 726 Default: 10000 727 728 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver 729 Format: 730 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>] 731 732 crashkernel=size[KMG][@offset[KMG]] 733 [KNL] Using kexec, Linux can switch to a 'crash kernel' 734 upon panic. This parameter reserves the physical 735 memory region [offset, offset + size] for that kernel 736 image. If '@offset' is omitted, then a suitable offset 737 is selected automatically. 738 [KNL, X86-64] Select a region under 4G first, and 739 fall back to reserve region above 4G when '@offset' 740 hasn't been specified. 741 See Documentation/admin-guide/kdump/kdump.rst for further details. 742 743 crashkernel=range1:size1[,range2:size2,...][@offset] 744 [KNL] Same as above, but depends on the memory 745 in the running system. The syntax of range is 746 start-[end] where start and end are both 747 a memory unit (amount[KMG]). See also 748 Documentation/admin-guide/kdump/kdump.rst for an example. 749 750 crashkernel=size[KMG],high 751 [KNL, X86-64] range could be above 4G. Allow kernel 752 to allocate physical memory region from top, so could 753 be above 4G if system have more than 4G ram installed. 754 Otherwise memory region will be allocated below 4G, if 755 available. 756 It will be ignored if crashkernel=X is specified. 757 crashkernel=size[KMG],low 758 [KNL, X86-64] range under 4G. When crashkernel=X,high 759 is passed, kernel could allocate physical memory region 760 above 4G, that cause second kernel crash on system 761 that require some amount of low memory, e.g. swiotlb 762 requires at least 64M+32K low memory, also enough extra 763 low memory is needed to make sure DMA buffers for 32-bit 764 devices won't run out. Kernel would try to allocate at 765 at least 256M below 4G automatically. 766 This one let user to specify own low range under 4G 767 for second kernel instead. 768 0: to disable low allocation. 769 It will be ignored when crashkernel=X,high is not used 770 or memory reserved is below 4G. 771 772 cryptomgr.notests 773 [KNL] Disable crypto self-tests 774 775 cs89x0_dma= [HW,NET] 776 Format: <dma> 777 778 cs89x0_media= [HW,NET] 779 Format: { rj45 | aui | bnc } 780 781 dasd= [HW,NET] 782 See header of drivers/s390/block/dasd_devmap.c. 783 784 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port 785 (one device per port) 786 Format: <port#>,<type> 787 See also Documentation/input/devices/joystick-parport.rst 788 789 ddebug_query= [KNL,DYNAMIC_DEBUG] Enable debug messages at early boot 790 time. See 791 Documentation/admin-guide/dynamic-debug-howto.rst for 792 details. Deprecated, see dyndbg. 793 794 debug [KNL] Enable kernel debugging (events log level). 795 796 debug_boot_weak_hash 797 [KNL] Enable printing [hashed] pointers early in the 798 boot sequence. If enabled, we use a weak hash instead 799 of siphash to hash pointers. Use this option if you are 800 seeing instances of '(___ptrval___)') and need to see a 801 value (hashed pointer) instead. Cryptographically 802 insecure, please do not use on production kernels. 803 804 debug_locks_verbose= 805 [KNL] verbose locking self-tests 806 Format: <int> 807 Print debugging info while doing the locking API 808 self-tests. 809 Bitmask for the various LOCKTYPE_ tests. Defaults to 0 810 (no extra messages), setting it to -1 (all bits set) 811 will print _a_lot_ more information - normally only 812 useful to lockdep developers. 813 814 debug_objects [KNL] Enable object debugging 815 816 no_debug_objects 817 [KNL] Disable object debugging 818 819 debug_guardpage_minorder= 820 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this 821 parameter allows control of the order of pages that will 822 be intentionally kept free (and hence protected) by the 823 buddy allocator. Bigger value increase the probability 824 of catching random memory corruption, but reduce the 825 amount of memory for normal system use. The maximum 826 possible value is MAX_ORDER/2. Setting this parameter 827 to 1 or 2 should be enough to identify most random 828 memory corruption problems caused by bugs in kernel or 829 driver code when a CPU writes to (or reads from) a 830 random memory location. Note that there exists a class 831 of memory corruptions problems caused by buggy H/W or 832 F/W or by drivers badly programing DMA (basically when 833 memory is written at bus level and the CPU MMU is 834 bypassed) which are not detectable by 835 CONFIG_DEBUG_PAGEALLOC, hence this option will not help 836 tracking down these problems. 837 838 debug_pagealloc= 839 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this parameter 840 enables the feature at boot time. By default, it is 841 disabled and the system will work mostly the same as a 842 kernel built without CONFIG_DEBUG_PAGEALLOC. 843 Note: to get most of debug_pagealloc error reports, it's 844 useful to also enable the page_owner functionality. 845 on: enable the feature 846 847 debugfs= [KNL] This parameter enables what is exposed to userspace 848 and debugfs internal clients. 849 Format: { on, no-mount, off } 850 on: All functions are enabled. 851 no-mount: 852 Filesystem is not registered but kernel clients can 853 access APIs and a crashkernel can be used to read 854 its content. There is nothing to mount. 855 off: Filesystem is not registered and clients 856 get a -EPERM as result when trying to register files 857 or directories within debugfs. 858 This is equivalent of the runtime functionality if 859 debugfs was not enabled in the kernel at all. 860 Default value is set in build-time with a kernel configuration. 861 862 debugpat [X86] Enable PAT debugging 863 864 decnet.addr= [HW,NET] 865 Format: <area>[,<node>] 866 See also Documentation/networking/decnet.rst. 867 868 default_hugepagesz= 869 [HW] The size of the default HugeTLB page. This is 870 the size represented by the legacy /proc/ hugepages 871 APIs. In addition, this is the default hugetlb size 872 used for shmget(), mmap() and mounting hugetlbfs 873 filesystems. If not specified, defaults to the 874 architecture's default huge page size. Huge page 875 sizes are architecture dependent. See also 876 Documentation/admin-guide/mm/hugetlbpage.rst. 877 Format: size[KMG] 878 879 deferred_probe_timeout= 880 [KNL] Debugging option to set a timeout in seconds for 881 deferred probe to give up waiting on dependencies to 882 probe. Only specific dependencies (subsystems or 883 drivers) that have opted in will be ignored. A timeout of 0 884 will timeout at the end of initcalls. This option will also 885 dump out devices still on the deferred probe list after 886 retrying. 887 888 dfltcc= [HW,S390] 889 Format: { on | off | def_only | inf_only | always } 890 on: s390 zlib hardware support for compression on 891 level 1 and decompression (default) 892 off: No s390 zlib hardware support 893 def_only: s390 zlib hardware support for deflate 894 only (compression on level 1) 895 inf_only: s390 zlib hardware support for inflate 896 only (decompression) 897 always: Same as 'on' but ignores the selected compression 898 level always using hardware support (used for debugging) 899 900 dhash_entries= [KNL] 901 Set number of hash buckets for dentry cache. 902 903 disable_1tb_segments [PPC] 904 Disables the use of 1TB hash page table segments. This 905 causes the kernel to fall back to 256MB segments which 906 can be useful when debugging issues that require an SLB 907 miss to occur. 908 909 stress_slb [PPC] 910 Limits the number of kernel SLB entries, and flushes 911 them frequently to increase the rate of SLB faults 912 on kernel addresses. 913 914 disable= [IPV6] 915 See Documentation/networking/ipv6.rst. 916 917 hardened_usercopy= 918 [KNL] Under CONFIG_HARDENED_USERCOPY, whether 919 hardening is enabled for this boot. Hardened 920 usercopy checking is used to protect the kernel 921 from reading or writing beyond known memory 922 allocation boundaries as a proactive defense 923 against bounds-checking flaws in the kernel's 924 copy_to_user()/copy_from_user() interface. 925 on Perform hardened usercopy checks (default). 926 off Disable hardened usercopy checks. 927 928 disable_radix [PPC] 929 Disable RADIX MMU mode on POWER9 930 931 radix_hcall_invalidate=on [PPC/PSERIES] 932 Disable RADIX GTSE feature and use hcall for TLB 933 invalidate. 934 935 disable_tlbie [PPC] 936 Disable TLBIE instruction. Currently does not work 937 with KVM, with HASH MMU, or with coherent accelerators. 938 939 disable_cpu_apicid= [X86,APIC,SMP] 940 Format: <int> 941 The number of initial APIC ID for the 942 corresponding CPU to be disabled at boot, 943 mostly used for the kdump 2nd kernel to 944 disable BSP to wake up multiple CPUs without 945 causing system reset or hang due to sending 946 INIT from AP to BSP. 947 948 disable_ddw [PPC/PSERIES] 949 Disable Dynamic DMA Window support. Use this 950 to workaround buggy firmware. 951 952 disable_ipv6= [IPV6] 953 See Documentation/networking/ipv6.rst. 954 955 disable_mtrr_cleanup [X86] 956 The kernel tries to adjust MTRR layout from continuous 957 to discrete, to make X server driver able to add WB 958 entry later. This parameter disables that. 959 960 disable_mtrr_trim [X86, Intel and AMD only] 961 By default the kernel will trim any uncacheable 962 memory out of your available memory pool based on 963 MTRR settings. This parameter disables that behavior, 964 possibly causing your machine to run very slowly. 965 966 disable_timer_pin_1 [X86] 967 Disable PIN 1 of APIC timer 968 Can be useful to work around chipset bugs. 969 970 dis_ucode_ldr [X86] Disable the microcode loader. 971 972 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support, 973 this option disables the debugging code at boot. 974 975 dma_debug_entries=<number> 976 This option allows to tune the number of preallocated 977 entries for DMA-API debugging code. One entry is 978 required per DMA-API allocation. Use this if the 979 DMA-API debugging code disables itself because the 980 architectural default is too low. 981 982 dma_debug_driver=<driver_name> 983 With this option the DMA-API debugging driver 984 filter feature can be enabled at boot time. Just 985 pass the driver to filter for as the parameter. 986 The filter can be disabled or changed to another 987 driver later using sysfs. 988 989 driver_async_probe= [KNL] 990 List of driver names to be probed asynchronously. 991 Format: <driver_name1>,<driver_name2>... 992 993 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>] 994 Broken monitors, graphic adapters, KVMs and EDIDless 995 panels may send no or incorrect EDID data sets. 996 This parameter allows to specify an EDID data sets 997 in the /lib/firmware directory that are used instead. 998 Generic built-in EDID data sets are used, if one of 999 edid/1024x768.bin, edid/1280x1024.bin, 1000 edid/1680x1050.bin, or edid/1920x1080.bin is given 1001 and no file with the same name exists. Details and 1002 instructions how to build your own EDID data are 1003 available in Documentation/admin-guide/edid.rst. An EDID 1004 data set will only be used for a particular connector, 1005 if its name and a colon are prepended to the EDID 1006 name. Each connector may use a unique EDID data 1007 set by separating the files with a comma. An EDID 1008 data set with no connector name will be used for 1009 any connectors not explicitly specified. 1010 1011 dscc4.setup= [NET] 1012 1013 dt_cpu_ftrs= [PPC] 1014 Format: {"off" | "known"} 1015 Control how the dt_cpu_ftrs device-tree binding is 1016 used for CPU feature discovery and setup (if it 1017 exists). 1018 off: Do not use it, fall back to legacy cpu table. 1019 known: Do not pass through unknown features to guests 1020 or userspace, only those that the kernel is aware of. 1021 1022 dump_apple_properties [X86] 1023 Dump name and content of EFI device properties on 1024 x86 Macs. Useful for driver authors to determine 1025 what data is available or for reverse-engineering. 1026 1027 dyndbg[="val"] [KNL,DYNAMIC_DEBUG] 1028 <module>.dyndbg[="val"] 1029 Enable debug messages at boot time. See 1030 Documentation/admin-guide/dynamic-debug-howto.rst 1031 for details. 1032 1033 nopku [X86] Disable Memory Protection Keys CPU feature found 1034 in some Intel CPUs. 1035 1036 <module>.async_probe [KNL] 1037 Enable asynchronous probe on this module. 1038 1039 early_ioremap_debug [KNL] 1040 Enable debug messages in early_ioremap support. This 1041 is useful for tracking down temporary early mappings 1042 which are not unmapped. 1043 1044 earlycon= [KNL] Output early console device and options. 1045 1046 When used with no options, the early console is 1047 determined by stdout-path property in device tree's 1048 chosen node or the ACPI SPCR table if supported by 1049 the platform. 1050 1051 cdns,<addr>[,options] 1052 Start an early, polled-mode console on a Cadence 1053 (xuartps) serial port at the specified address. Only 1054 supported option is baud rate. If baud rate is not 1055 specified, the serial port must already be setup and 1056 configured. 1057 1058 uart[8250],io,<addr>[,options] 1059 uart[8250],mmio,<addr>[,options] 1060 uart[8250],mmio32,<addr>[,options] 1061 uart[8250],mmio32be,<addr>[,options] 1062 uart[8250],0x<addr>[,options] 1063 Start an early, polled-mode console on the 8250/16550 1064 UART at the specified I/O port or MMIO address. 1065 MMIO inter-register address stride is either 8-bit 1066 (mmio) or 32-bit (mmio32 or mmio32be). 1067 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed 1068 to be equivalent to 'mmio'. 'options' are specified 1069 in the same format described for "console=ttyS<n>"; if 1070 unspecified, the h/w is not initialized. 1071 1072 pl011,<addr> 1073 pl011,mmio32,<addr> 1074 Start an early, polled-mode console on a pl011 serial 1075 port at the specified address. The pl011 serial port 1076 must already be setup and configured. Options are not 1077 yet supported. If 'mmio32' is specified, then only 1078 the driver will use only 32-bit accessors to read/write 1079 the device registers. 1080 1081 meson,<addr> 1082 Start an early, polled-mode console on a meson serial 1083 port at the specified address. The serial port must 1084 already be setup and configured. Options are not yet 1085 supported. 1086 1087 msm_serial,<addr> 1088 Start an early, polled-mode console on an msm serial 1089 port at the specified address. The serial port 1090 must already be setup and configured. Options are not 1091 yet supported. 1092 1093 msm_serial_dm,<addr> 1094 Start an early, polled-mode console on an msm serial 1095 dm port at the specified address. The serial port 1096 must already be setup and configured. Options are not 1097 yet supported. 1098 1099 owl,<addr> 1100 Start an early, polled-mode console on a serial port 1101 of an Actions Semi SoC, such as S500 or S900, at the 1102 specified address. The serial port must already be 1103 setup and configured. Options are not yet supported. 1104 1105 rda,<addr> 1106 Start an early, polled-mode console on a serial port 1107 of an RDA Micro SoC, such as RDA8810PL, at the 1108 specified address. The serial port must already be 1109 setup and configured. Options are not yet supported. 1110 1111 sbi 1112 Use RISC-V SBI (Supervisor Binary Interface) for early 1113 console. 1114 1115 smh Use ARM semihosting calls for early console. 1116 1117 s3c2410,<addr> 1118 s3c2412,<addr> 1119 s3c2440,<addr> 1120 s3c6400,<addr> 1121 s5pv210,<addr> 1122 exynos4210,<addr> 1123 Use early console provided by serial driver available 1124 on Samsung SoCs, requires selecting proper type and 1125 a correct base address of the selected UART port. The 1126 serial port must already be setup and configured. 1127 Options are not yet supported. 1128 1129 lantiq,<addr> 1130 Start an early, polled-mode console on a lantiq serial 1131 (lqasc) port at the specified address. The serial port 1132 must already be setup and configured. Options are not 1133 yet supported. 1134 1135 lpuart,<addr> 1136 lpuart32,<addr> 1137 Use early console provided by Freescale LP UART driver 1138 found on Freescale Vybrid and QorIQ LS1021A processors. 1139 A valid base address must be provided, and the serial 1140 port must already be setup and configured. 1141 1142 ec_imx21,<addr> 1143 ec_imx6q,<addr> 1144 Start an early, polled-mode, output-only console on the 1145 Freescale i.MX UART at the specified address. The UART 1146 must already be setup and configured. 1147 1148 ar3700_uart,<addr> 1149 Start an early, polled-mode console on the 1150 Armada 3700 serial port at the specified 1151 address. The serial port must already be setup 1152 and configured. Options are not yet supported. 1153 1154 qcom_geni,<addr> 1155 Start an early, polled-mode console on a Qualcomm 1156 Generic Interface (GENI) based serial port at the 1157 specified address. The serial port must already be 1158 setup and configured. Options are not yet supported. 1159 1160 efifb,[options] 1161 Start an early, unaccelerated console on the EFI 1162 memory mapped framebuffer (if available). On cache 1163 coherent non-x86 systems that use system memory for 1164 the framebuffer, pass the 'ram' option so that it is 1165 mapped with the correct attributes. 1166 1167 linflex,<addr> 1168 Use early console provided by Freescale LINFlexD UART 1169 serial driver for NXP S32V234 SoCs. A valid base 1170 address must be provided, and the serial port must 1171 already be setup and configured. 1172 1173 earlyprintk= [X86,SH,ARM,M68k,S390] 1174 earlyprintk=vga 1175 earlyprintk=sclp 1176 earlyprintk=xen 1177 earlyprintk=serial[,ttySn[,baudrate]] 1178 earlyprintk=serial[,0x...[,baudrate]] 1179 earlyprintk=ttySn[,baudrate] 1180 earlyprintk=dbgp[debugController#] 1181 earlyprintk=pciserial[,force],bus:device.function[,baudrate] 1182 earlyprintk=xdbc[xhciController#] 1183 1184 earlyprintk is useful when the kernel crashes before 1185 the normal console is initialized. It is not enabled by 1186 default because it has some cosmetic problems. 1187 1188 Append ",keep" to not disable it when the real console 1189 takes over. 1190 1191 Only one of vga, efi, serial, or usb debug port can 1192 be used at a time. 1193 1194 Currently only ttyS0 and ttyS1 may be specified by 1195 name. Other I/O ports may be explicitly specified 1196 on some architectures (x86 and arm at least) by 1197 replacing ttySn with an I/O port address, like this: 1198 earlyprintk=serial,0x1008,115200 1199 You can find the port for a given device in 1200 /proc/tty/driver/serial: 1201 2: uart:ST16650V2 port:00001008 irq:18 ... 1202 1203 Interaction with the standard serial driver is not 1204 very good. 1205 1206 The VGA and EFI output is eventually overwritten by 1207 the real console. 1208 1209 The xen output can only be used by Xen PV guests. 1210 1211 The sclp output can only be used on s390. 1212 1213 The optional "force" to "pciserial" enables use of a 1214 PCI device even when its classcode is not of the 1215 UART class. 1216 1217 edac_report= [HW,EDAC] Control how to report EDAC event 1218 Format: {"on" | "off" | "force"} 1219 on: enable EDAC to report H/W event. May be overridden 1220 by other higher priority error reporting module. 1221 off: disable H/W event reporting through EDAC. 1222 force: enforce the use of EDAC to report H/W event. 1223 default: on. 1224 1225 ekgdboc= [X86,KGDB] Allow early kernel console debugging 1226 ekgdboc=kbd 1227 1228 This is designed to be used in conjunction with 1229 the boot argument: earlyprintk=vga 1230 1231 This parameter works in place of the kgdboc parameter 1232 but can only be used if the backing tty is available 1233 very early in the boot process. For early debugging 1234 via a serial port see kgdboc_earlycon instead. 1235 1236 edd= [EDD] 1237 Format: {"off" | "on" | "skip[mbr]"} 1238 1239 efi= [EFI] 1240 Format: { "debug", "disable_early_pci_dma", 1241 "nochunk", "noruntime", "nosoftreserve", 1242 "novamap", "no_disable_early_pci_dma" } 1243 debug: enable misc debug output. 1244 disable_early_pci_dma: disable the busmaster bit on all 1245 PCI bridges while in the EFI boot stub. 1246 nochunk: disable reading files in "chunks" in the EFI 1247 boot stub, as chunking can cause problems with some 1248 firmware implementations. 1249 noruntime : disable EFI runtime services support 1250 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose) 1251 attribute may cause the kernel to reserve the 1252 memory range for a memory mapping driver to 1253 claim. Specify efi=nosoftreserve to disable this 1254 reservation and treat the memory by its base type 1255 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM"). 1256 novamap: do not call SetVirtualAddressMap(). 1257 no_disable_early_pci_dma: Leave the busmaster bit set 1258 on all PCI bridges while in the EFI boot stub 1259 1260 efi_no_storage_paranoia [EFI; X86] 1261 Using this parameter you can use more than 50% of 1262 your efi variable storage. Use this parameter only if 1263 you are really sure that your UEFI does sane gc and 1264 fulfills the spec otherwise your board may brick. 1265 1266 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86] 1267 Add arbitrary attribute to specific memory range by 1268 updating original EFI memory map. 1269 Region of memory which aa attribute is added to is 1270 from ss to ss+nn. 1271 1272 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000 1273 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000) 1274 attribute is added to range 0x100000000-0x180000000 and 1275 0x10a0000000-0x1120000000. 1276 1277 If efi_fake_mem=8G@9G:0x40000 is specified, the 1278 EFI_MEMORY_SP(0x40000) attribute is added to 1279 range 0x240000000-0x43fffffff. 1280 1281 Using this parameter you can do debugging of EFI memmap 1282 related features. For example, you can do debugging of 1283 Address Range Mirroring feature even if your box 1284 doesn't support it, or mark specific memory as 1285 "soft reserved". 1286 1287 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT 1288 that is to be dynamically loaded by Linux. If there are 1289 multiple variables with the same name but with different 1290 vendor GUIDs, all of them will be loaded. See 1291 Documentation/admin-guide/acpi/ssdt-overlays.rst for details. 1292 1293 1294 eisa_irq_edge= [PARISC,HW] 1295 See header of drivers/parisc/eisa.c. 1296 1297 elanfreq= [X86-32] 1298 See comment before function elanfreq_setup() in 1299 arch/x86/kernel/cpu/cpufreq/elanfreq.c. 1300 1301 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390] 1302 Specifies physical address of start of kernel core 1303 image elf header and optionally the size. Generally 1304 kexec loader will pass this option to capture kernel. 1305 See Documentation/admin-guide/kdump/kdump.rst for details. 1306 1307 enable_mtrr_cleanup [X86] 1308 The kernel tries to adjust MTRR layout from continuous 1309 to discrete, to make X server driver able to add WB 1310 entry later. This parameter enables that. 1311 1312 enable_timer_pin_1 [X86] 1313 Enable PIN 1 of APIC timer 1314 Can be useful to work around chipset bugs 1315 (in particular on some ATI chipsets). 1316 The kernel tries to set a reasonable default. 1317 1318 enforcing [SELINUX] Set initial enforcing status. 1319 Format: {"0" | "1"} 1320 See security/selinux/Kconfig help text. 1321 0 -- permissive (log only, no denials). 1322 1 -- enforcing (deny and log). 1323 Default value is 0. 1324 Value can be changed at runtime via 1325 /sys/fs/selinux/enforce. 1326 1327 erst_disable [ACPI] 1328 Disable Error Record Serialization Table (ERST) 1329 support. 1330 1331 ether= [HW,NET] Ethernet cards parameters 1332 This option is obsoleted by the "netdev=" option, which 1333 has equivalent usage. See its documentation for details. 1334 1335 evm= [EVM] 1336 Format: { "fix" } 1337 Permit 'security.evm' to be updated regardless of 1338 current integrity status. 1339 1340 failslab= 1341 fail_usercopy= 1342 fail_page_alloc= 1343 fail_make_request=[KNL] 1344 General fault injection mechanism. 1345 Format: <interval>,<probability>,<space>,<times> 1346 See also Documentation/fault-injection/. 1347 1348 fb_tunnels= [NET] 1349 Format: { initns | none } 1350 See Documentation/admin-guide/sysctl/net.rst for 1351 fb_tunnels_only_for_init_ns 1352 1353 floppy= [HW] 1354 See Documentation/admin-guide/blockdev/floppy.rst. 1355 1356 force_pal_cache_flush 1357 [IA-64] Avoid check_sal_cache_flush which may hang on 1358 buggy SAL_CACHE_FLUSH implementations. Using this 1359 parameter will force ia64_sal_cache_flush to call 1360 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH. 1361 1362 forcepae [X86-32] 1363 Forcefully enable Physical Address Extension (PAE). 1364 Many Pentium M systems disable PAE but may have a 1365 functionally usable PAE implementation. 1366 Warning: use of this parameter will taint the kernel 1367 and may cause unknown problems. 1368 1369 ftrace=[tracer] 1370 [FTRACE] will set and start the specified tracer 1371 as early as possible in order to facilitate early 1372 boot debugging. 1373 1374 ftrace_dump_on_oops[=orig_cpu] 1375 [FTRACE] will dump the trace buffers on oops. 1376 If no parameter is passed, ftrace will dump 1377 buffers of all CPUs, but if you pass orig_cpu, it will 1378 dump only the buffer of the CPU that triggered the 1379 oops. 1380 1381 ftrace_filter=[function-list] 1382 [FTRACE] Limit the functions traced by the function 1383 tracer at boot up. function-list is a comma-separated 1384 list of functions. This list can be changed at run 1385 time by the set_ftrace_filter file in the debugfs 1386 tracing directory. 1387 1388 ftrace_notrace=[function-list] 1389 [FTRACE] Do not trace the functions specified in 1390 function-list. This list can be changed at run time 1391 by the set_ftrace_notrace file in the debugfs 1392 tracing directory. 1393 1394 ftrace_graph_filter=[function-list] 1395 [FTRACE] Limit the top level callers functions traced 1396 by the function graph tracer at boot up. 1397 function-list is a comma-separated list of functions 1398 that can be changed at run time by the 1399 set_graph_function file in the debugfs tracing directory. 1400 1401 ftrace_graph_notrace=[function-list] 1402 [FTRACE] Do not trace from the functions specified in 1403 function-list. This list is a comma-separated list of 1404 functions that can be changed at run time by the 1405 set_graph_notrace file in the debugfs tracing directory. 1406 1407 ftrace_graph_max_depth=<uint> 1408 [FTRACE] Used with the function graph tracer. This is 1409 the max depth it will trace into a function. This value 1410 can be changed at run time by the max_graph_depth file 1411 in the tracefs tracing directory. default: 0 (no limit) 1412 1413 fw_devlink= [KNL] Create device links between consumer and supplier 1414 devices by scanning the firmware to infer the 1415 consumer/supplier relationships. This feature is 1416 especially useful when drivers are loaded as modules as 1417 it ensures proper ordering of tasks like device probing 1418 (suppliers first, then consumers), supplier boot state 1419 clean up (only after all consumers have probed), 1420 suspend/resume & runtime PM (consumers first, then 1421 suppliers). 1422 Format: { off | permissive | on | rpm } 1423 off -- Don't create device links from firmware info. 1424 permissive -- Create device links from firmware info 1425 but use it only for ordering boot state clean 1426 up (sync_state() calls). 1427 on -- Create device links from firmware info and use it 1428 to enforce probe and suspend/resume ordering. 1429 rpm -- Like "on", but also use to order runtime PM. 1430 1431 gamecon.map[2|3]= 1432 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad 1433 support via parallel port (up to 5 devices per port) 1434 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5> 1435 See also Documentation/input/devices/joystick-parport.rst 1436 1437 gamma= [HW,DRM] 1438 1439 gart_fix_e820= [X86-64] disable the fix e820 for K8 GART 1440 Format: off | on 1441 default: on 1442 1443 gcov_persist= [GCOV] When non-zero (default), profiling data for 1444 kernel modules is saved and remains accessible via 1445 debugfs, even when the module is unloaded/reloaded. 1446 When zero, profiling data is discarded and associated 1447 debugfs files are removed at module unload time. 1448 1449 goldfish [X86] Enable the goldfish android emulator platform. 1450 Don't use this when you are not running on the 1451 android emulator 1452 1453 gpt [EFI] Forces disk with valid GPT signature but 1454 invalid Protective MBR to be treated as GPT. If the 1455 primary GPT is corrupted, it enables the backup/alternate 1456 GPT to be used instead. 1457 1458 grcan.enable0= [HW] Configuration of physical interface 0. Determines 1459 the "Enable 0" bit of the configuration register. 1460 Format: 0 | 1 1461 Default: 0 1462 grcan.enable1= [HW] Configuration of physical interface 1. Determines 1463 the "Enable 0" bit of the configuration register. 1464 Format: 0 | 1 1465 Default: 0 1466 grcan.select= [HW] Select which physical interface to use. 1467 Format: 0 | 1 1468 Default: 0 1469 grcan.txsize= [HW] Sets the size of the tx buffer. 1470 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0. 1471 Default: 1024 1472 grcan.rxsize= [HW] Sets the size of the rx buffer. 1473 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0. 1474 Default: 1024 1475 1476 gpio-mockup.gpio_mockup_ranges 1477 [HW] Sets the ranges of gpiochip of for this device. 1478 Format: <start1>,<end1>,<start2>,<end2>... 1479 1480 hardlockup_all_cpu_backtrace= 1481 [KNL] Should the hard-lockup detector generate 1482 backtraces on all cpus. 1483 Format: 0 | 1 1484 1485 hashdist= [KNL,NUMA] Large hashes allocated during boot 1486 are distributed across NUMA nodes. Defaults on 1487 for 64-bit NUMA, off otherwise. 1488 Format: 0 | 1 (for off | on) 1489 1490 hcl= [IA-64] SGI's Hardware Graph compatibility layer 1491 1492 hd= [EIDE] (E)IDE hard drive subsystem geometry 1493 Format: <cyl>,<head>,<sect> 1494 1495 hest_disable [ACPI] 1496 Disable Hardware Error Source Table (HEST) support; 1497 corresponding firmware-first mode error processing 1498 logic will be disabled. 1499 1500 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact 1501 size of <nn>. This works even on boxes that have no 1502 highmem otherwise. This also works to reduce highmem 1503 size on bigger boxes. 1504 1505 highres= [KNL] Enable/disable high resolution timer mode. 1506 Valid parameters: "on", "off" 1507 Default: "on" 1508 1509 hlt [BUGS=ARM,SH] 1510 1511 hpet= [X86-32,HPET] option to control HPET usage 1512 Format: { enable (default) | disable | force | 1513 verbose } 1514 disable: disable HPET and use PIT instead 1515 force: allow force enabled of undocumented chips (ICH4, 1516 VIA, nVidia) 1517 verbose: show contents of HPET registers during setup 1518 1519 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET 1520 registers. Default set by CONFIG_HPET_MMAP_DEFAULT. 1521 1522 hugetlb_cma= [HW] The size of a cma area used for allocation 1523 of gigantic hugepages. 1524 Format: nn[KMGTPE] 1525 1526 Reserve a cma area of given size and allocate gigantic 1527 hugepages using the cma allocator. If enabled, the 1528 boot-time allocation of gigantic hugepages is skipped. 1529 1530 hugepages= [HW] Number of HugeTLB pages to allocate at boot. 1531 If this follows hugepagesz (below), it specifies 1532 the number of pages of hugepagesz to be allocated. 1533 If this is the first HugeTLB parameter on the command 1534 line, it specifies the number of pages to allocate for 1535 the default huge page size. See also 1536 Documentation/admin-guide/mm/hugetlbpage.rst. 1537 Format: <integer> 1538 1539 hugepagesz= 1540 [HW] The size of the HugeTLB pages. This is used in 1541 conjunction with hugepages (above) to allocate huge 1542 pages of a specific size at boot. The pair 1543 hugepagesz=X hugepages=Y can be specified once for 1544 each supported huge page size. Huge page sizes are 1545 architecture dependent. See also 1546 Documentation/admin-guide/mm/hugetlbpage.rst. 1547 Format: size[KMG] 1548 1549 hung_task_panic= 1550 [KNL] Should the hung task detector generate panics. 1551 Format: 0 | 1 1552 1553 A value of 1 instructs the kernel to panic when a 1554 hung task is detected. The default value is controlled 1555 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time 1556 option. The value selected by this boot parameter can 1557 be changed later by the kernel.hung_task_panic sysctl. 1558 1559 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC) 1560 terminal devices. Valid values: 0..8 1561 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs. 1562 If specified, z/VM IUCV HVC accepts connections 1563 from listed z/VM user IDs only. 1564 1565 hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations 1566 which allow the hypervisor to 'idle' the 1567 guest on lock contention. 1568 1569 keep_bootcon [KNL] 1570 Do not unregister boot console at start. This is only 1571 useful for debugging when something happens in the window 1572 between unregistering the boot console and initializing 1573 the real console. 1574 1575 i2c_bus= [HW] Override the default board specific I2C bus speed 1576 or register an additional I2C bus that is not 1577 registered from board initialization code. 1578 Format: 1579 <bus_id>,<clkrate> 1580 1581 i8042.debug [HW] Toggle i8042 debug mode 1582 i8042.unmask_kbd_data 1583 [HW] Enable printing of interrupt data from the KBD port 1584 (disabled by default, and as a pre-condition 1585 requires that i8042.debug=1 be enabled) 1586 i8042.direct [HW] Put keyboard port into non-translated mode 1587 i8042.dumbkbd [HW] Pretend that controller can only read data from 1588 keyboard and cannot control its state 1589 (Don't attempt to blink the leds) 1590 i8042.noaux [HW] Don't check for auxiliary (== mouse) port 1591 i8042.nokbd [HW] Don't check/create keyboard port 1592 i8042.noloop [HW] Disable the AUX Loopback command while probing 1593 for the AUX port 1594 i8042.nomux [HW] Don't check presence of an active multiplexing 1595 controller 1596 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX 1597 controllers 1598 i8042.notimeout [HW] Ignore timeout condition signalled by controller 1599 i8042.reset [HW] Reset the controller during init, cleanup and 1600 suspend-to-ram transitions, only during s2r 1601 transitions, or never reset 1602 Format: { 1 | Y | y | 0 | N | n } 1603 1, Y, y: always reset controller 1604 0, N, n: don't ever reset controller 1605 Default: only on s2r transitions on x86; most other 1606 architectures force reset to be always executed 1607 i8042.unlock [HW] Unlock (ignore) the keylock 1608 i8042.kbdreset [HW] Reset device connected to KBD port 1609 1610 i810= [HW,DRM] 1611 1612 i8k.ignore_dmi [HW] Continue probing hardware even if DMI data 1613 indicates that the driver is running on unsupported 1614 hardware. 1615 i8k.force [HW] Activate i8k driver even if SMM BIOS signature 1616 does not match list of supported models. 1617 i8k.power_status 1618 [HW] Report power status in /proc/i8k 1619 (disabled by default) 1620 i8k.restricted [HW] Allow controlling fans only if SYS_ADMIN 1621 capability is set. 1622 1623 i915.invert_brightness= 1624 [DRM] Invert the sense of the variable that is used to 1625 set the brightness of the panel backlight. Normally a 1626 brightness value of 0 indicates backlight switched off, 1627 and the maximum of the brightness value sets the backlight 1628 to maximum brightness. If this parameter is set to 0 1629 (default) and the machine requires it, or this parameter 1630 is set to 1, a brightness value of 0 sets the backlight 1631 to maximum brightness, and the maximum of the brightness 1632 value switches the backlight off. 1633 -1 -- never invert brightness 1634 0 -- machine default 1635 1 -- force brightness inversion 1636 1637 icn= [HW,ISDN] 1638 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]] 1639 1640 ide-core.nodma= [HW] (E)IDE subsystem 1641 Format: =0.0 to prevent dma on hda, =0.1 hdb =1.0 hdc 1642 .vlb_clock .pci_clock .noflush .nohpa .noprobe .nowerr 1643 .cdrom .chs .ignore_cable are additional options 1644 See Documentation/ide/ide.rst. 1645 1646 ide-generic.probe-mask= [HW] (E)IDE subsystem 1647 Format: <int> 1648 Probe mask for legacy ISA IDE ports. Depending on 1649 platform up to 6 ports are supported, enabled by 1650 setting corresponding bits in the mask to 1. The 1651 default value is 0x0, which has a special meaning. 1652 On systems that have PCI, it triggers scanning the 1653 PCI bus for the first and the second port, which 1654 are then probed. On systems without PCI the value 1655 of 0x0 enables probing the two first ports as if it 1656 was 0x3. 1657 1658 ide-pci-generic.all-generic-ide [HW] (E)IDE subsystem 1659 Claim all unknown PCI IDE storage controllers. 1660 1661 idle= [X86] 1662 Format: idle=poll, idle=halt, idle=nomwait 1663 Poll forces a polling idle loop that can slightly 1664 improve the performance of waking up a idle CPU, but 1665 will use a lot of power and make the system run hot. 1666 Not recommended. 1667 idle=halt: Halt is forced to be used for CPU idle. 1668 In such case C2/C3 won't be used again. 1669 idle=nomwait: Disable mwait for CPU C-states 1670 1671 ieee754= [MIPS] Select IEEE Std 754 conformance mode 1672 Format: { strict | legacy | 2008 | relaxed } 1673 Default: strict 1674 1675 Choose which programs will be accepted for execution 1676 based on the IEEE 754 NaN encoding(s) supported by 1677 the FPU and the NaN encoding requested with the value 1678 of an ELF file header flag individually set by each 1679 binary. Hardware implementations are permitted to 1680 support either or both of the legacy and the 2008 NaN 1681 encoding mode. 1682 1683 Available settings are as follows: 1684 strict accept binaries that request a NaN encoding 1685 supported by the FPU 1686 legacy only accept legacy-NaN binaries, if supported 1687 by the FPU 1688 2008 only accept 2008-NaN binaries, if supported 1689 by the FPU 1690 relaxed accept any binaries regardless of whether 1691 supported by the FPU 1692 1693 The FPU emulator is always able to support both NaN 1694 encodings, so if no FPU hardware is present or it has 1695 been disabled with 'nofpu', then the settings of 1696 'legacy' and '2008' strap the emulator accordingly, 1697 'relaxed' straps the emulator for both legacy-NaN and 1698 2008-NaN, whereas 'strict' enables legacy-NaN only on 1699 legacy processors and both NaN encodings on MIPS32 or 1700 MIPS64 CPUs. 1701 1702 The setting for ABS.fmt/NEG.fmt instruction execution 1703 mode generally follows that for the NaN encoding, 1704 except where unsupported by hardware. 1705 1706 ignore_loglevel [KNL] 1707 Ignore loglevel setting - this will print /all/ 1708 kernel messages to the console. Useful for debugging. 1709 We also add it as printk module parameter, so users 1710 could change it dynamically, usually by 1711 /sys/module/printk/parameters/ignore_loglevel. 1712 1713 ignore_rlimit_data 1714 Ignore RLIMIT_DATA setting for data mappings, 1715 print warning at first misuse. Can be changed via 1716 /sys/module/kernel/parameters/ignore_rlimit_data. 1717 1718 ihash_entries= [KNL] 1719 Set number of hash buckets for inode cache. 1720 1721 ima_appraise= [IMA] appraise integrity measurements 1722 Format: { "off" | "enforce" | "fix" | "log" } 1723 default: "enforce" 1724 1725 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead. 1726 The builtin appraise policy appraises all files 1727 owned by uid=0. 1728 1729 ima_canonical_fmt [IMA] 1730 Use the canonical format for the binary runtime 1731 measurements, instead of host native format. 1732 1733 ima_hash= [IMA] 1734 Format: { md5 | sha1 | rmd160 | sha256 | sha384 1735 | sha512 | ... } 1736 default: "sha1" 1737 1738 The list of supported hash algorithms is defined 1739 in crypto/hash_info.h. 1740 1741 ima_policy= [IMA] 1742 The builtin policies to load during IMA setup. 1743 Format: "tcb | appraise_tcb | secure_boot | 1744 fail_securely" 1745 1746 The "tcb" policy measures all programs exec'd, files 1747 mmap'd for exec, and all files opened with the read 1748 mode bit set by either the effective uid (euid=0) or 1749 uid=0. 1750 1751 The "appraise_tcb" policy appraises the integrity of 1752 all files owned by root. 1753 1754 The "secure_boot" policy appraises the integrity 1755 of files (eg. kexec kernel image, kernel modules, 1756 firmware, policy, etc) based on file signatures. 1757 1758 The "fail_securely" policy forces file signature 1759 verification failure also on privileged mounted 1760 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE 1761 flag. 1762 1763 ima_tcb [IMA] Deprecated. Use ima_policy= instead. 1764 Load a policy which meets the needs of the Trusted 1765 Computing Base. This means IMA will measure all 1766 programs exec'd, files mmap'd for exec, and all files 1767 opened for read by uid=0. 1768 1769 ima_template= [IMA] 1770 Select one of defined IMA measurements template formats. 1771 Formats: { "ima" | "ima-ng" | "ima-sig" } 1772 Default: "ima-ng" 1773 1774 ima_template_fmt= 1775 [IMA] Define a custom template format. 1776 Format: { "field1|...|fieldN" } 1777 1778 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage 1779 Format: <min_file_size> 1780 Set the minimal file size for using asynchronous hash. 1781 If left unspecified, ahash usage is disabled. 1782 1783 ahash performance varies for different data sizes on 1784 different crypto accelerators. This option can be used 1785 to achieve the best performance for a particular HW. 1786 1787 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size 1788 Format: <bufsize> 1789 Set hashing buffer size. Default: 4k. 1790 1791 ahash performance varies for different chunk sizes on 1792 different crypto accelerators. This option can be used 1793 to achieve best performance for particular HW. 1794 1795 init= [KNL] 1796 Format: <full_path> 1797 Run specified binary instead of /sbin/init as init 1798 process. 1799 1800 initcall_debug [KNL] Trace initcalls as they are executed. Useful 1801 for working out where the kernel is dying during 1802 startup. 1803 1804 initcall_blacklist= [KNL] Do not execute a comma-separated list of 1805 initcall functions. Useful for debugging built-in 1806 modules and initcalls. 1807 1808 initrd= [BOOT] Specify the location of the initial ramdisk 1809 1810 initrdmem= [KNL] Specify a physical address and size from which to 1811 load the initrd. If an initrd is compiled in or 1812 specified in the bootparams, it takes priority over this 1813 setting. 1814 Format: ss[KMG],nn[KMG] 1815 Default is 0, 0 1816 1817 init_on_alloc= [MM] Fill newly allocated pages and heap objects with 1818 zeroes. 1819 Format: 0 | 1 1820 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON. 1821 1822 init_on_free= [MM] Fill freed pages and heap objects with zeroes. 1823 Format: 0 | 1 1824 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON. 1825 1826 init_pkru= [X86] Specify the default memory protection keys rights 1827 register contents for all processes. 0x55555554 by 1828 default (disallow access to all but pkey 0). Can 1829 override in debugfs after boot. 1830 1831 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver 1832 Format: <irq> 1833 1834 int_pln_enable [X86] Enable power limit notification interrupt 1835 1836 integrity_audit=[IMA] 1837 Format: { "0" | "1" } 1838 0 -- basic integrity auditing messages. (Default) 1839 1 -- additional integrity auditing messages. 1840 1841 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option 1842 on 1843 Enable intel iommu driver. 1844 off 1845 Disable intel iommu driver. 1846 igfx_off [Default Off] 1847 By default, gfx is mapped as normal device. If a gfx 1848 device has a dedicated DMAR unit, the DMAR unit is 1849 bypassed by not enabling DMAR with this option. In 1850 this case, gfx device will use physical address for 1851 DMA. 1852 forcedac [X86-64] 1853 With this option iommu will not optimize to look 1854 for io virtual address below 32-bit forcing dual 1855 address cycle on pci bus for cards supporting greater 1856 than 32-bit addressing. The default is to look 1857 for translation below 32-bit and if not available 1858 then look in the higher range. 1859 strict [Default Off] 1860 With this option on every unmap_single operation will 1861 result in a hardware IOTLB flush operation as opposed 1862 to batching them for performance. 1863 sp_off [Default Off] 1864 By default, super page will be supported if Intel IOMMU 1865 has the capability. With this option, super page will 1866 not be supported. 1867 sm_on [Default Off] 1868 By default, scalable mode will be disabled even if the 1869 hardware advertises that it has support for the scalable 1870 mode translation. With this option set, scalable mode 1871 will be used on hardware which claims to support it. 1872 tboot_noforce [Default Off] 1873 Do not force the Intel IOMMU enabled under tboot. 1874 By default, tboot will force Intel IOMMU on, which 1875 could harm performance of some high-throughput 1876 devices like 40GBit network cards, even if identity 1877 mapping is enabled. 1878 Note that using this option lowers the security 1879 provided by tboot because it makes the system 1880 vulnerable to DMA attacks. 1881 1882 intel_idle.max_cstate= [KNL,HW,ACPI,X86] 1883 0 disables intel_idle and fall back on acpi_idle. 1884 1 to 9 specify maximum depth of C-state. 1885 1886 intel_pstate= [X86] 1887 disable 1888 Do not enable intel_pstate as the default 1889 scaling driver for the supported processors 1890 passive 1891 Use intel_pstate as a scaling driver, but configure it 1892 to work with generic cpufreq governors (instead of 1893 enabling its internal governor). This mode cannot be 1894 used along with the hardware-managed P-states (HWP) 1895 feature. 1896 force 1897 Enable intel_pstate on systems that prohibit it by default 1898 in favor of acpi-cpufreq. Forcing the intel_pstate driver 1899 instead of acpi-cpufreq may disable platform features, such 1900 as thermal controls and power capping, that rely on ACPI 1901 P-States information being indicated to OSPM and therefore 1902 should be used with caution. This option does not work with 1903 processors that aren't supported by the intel_pstate driver 1904 or on platforms that use pcc-cpufreq instead of acpi-cpufreq. 1905 no_hwp 1906 Do not enable hardware P state control (HWP) 1907 if available. 1908 hwp_only 1909 Only load intel_pstate on systems which support 1910 hardware P state control (HWP) if available. 1911 support_acpi_ppc 1912 Enforce ACPI _PPC performance limits. If the Fixed ACPI 1913 Description Table, specifies preferred power management 1914 profile as "Enterprise Server" or "Performance Server", 1915 then this feature is turned on by default. 1916 per_cpu_perf_limits 1917 Allow per-logical-CPU P-State performance control limits using 1918 cpufreq sysfs interface 1919 1920 intremap= [X86-64, Intel-IOMMU] 1921 on enable Interrupt Remapping (default) 1922 off disable Interrupt Remapping 1923 nosid disable Source ID checking 1924 no_x2apic_optout 1925 BIOS x2APIC opt-out request will be ignored 1926 nopost disable Interrupt Posting 1927 1928 iomem= Disable strict checking of access to MMIO memory 1929 strict regions from userspace. 1930 relaxed 1931 1932 iommu= [X86] 1933 off 1934 force 1935 noforce 1936 biomerge 1937 panic 1938 nopanic 1939 merge 1940 nomerge 1941 soft 1942 pt [X86] 1943 nopt [X86] 1944 nobypass [PPC/POWERNV] 1945 Disable IOMMU bypass, using IOMMU for PCI devices. 1946 1947 iommu.strict= [ARM64] Configure TLB invalidation behaviour 1948 Format: { "0" | "1" } 1949 0 - Lazy mode. 1950 Request that DMA unmap operations use deferred 1951 invalidation of hardware TLBs, for increased 1952 throughput at the cost of reduced device isolation. 1953 Will fall back to strict mode if not supported by 1954 the relevant IOMMU driver. 1955 1 - Strict mode (default). 1956 DMA unmap operations invalidate IOMMU hardware TLBs 1957 synchronously. 1958 1959 iommu.passthrough= 1960 [ARM64, X86] Configure DMA to bypass the IOMMU by default. 1961 Format: { "0" | "1" } 1962 0 - Use IOMMU translation for DMA. 1963 1 - Bypass the IOMMU for DMA. 1964 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH. 1965 1966 io7= [HW] IO7 for Marvel-based Alpha systems 1967 See comment before marvel_specify_io7 in 1968 arch/alpha/kernel/core_marvel.c. 1969 1970 io_delay= [X86] I/O delay method 1971 0x80 1972 Standard port 0x80 based delay 1973 0xed 1974 Alternate port 0xed based delay (needed on some systems) 1975 udelay 1976 Simple two microseconds delay 1977 none 1978 No delay 1979 1980 ip= [IP_PNP] 1981 See Documentation/admin-guide/nfs/nfsroot.rst. 1982 1983 ipcmni_extend [KNL] Extend the maximum number of unique System V 1984 IPC identifiers from 32,768 to 16,777,216. 1985 1986 irqaffinity= [SMP] Set the default irq affinity mask 1987 The argument is a cpu list, as described above. 1988 1989 irqchip.gicv2_force_probe= 1990 [ARM, ARM64] 1991 Format: <bool> 1992 Force the kernel to look for the second 4kB page 1993 of a GICv2 controller even if the memory range 1994 exposed by the device tree is too small. 1995 1996 irqchip.gicv3_nolpi= 1997 [ARM, ARM64] 1998 Force the kernel to ignore the availability of 1999 LPIs (and by consequence ITSs). Intended for system 2000 that use the kernel as a bootloader, and thus want 2001 to let secondary kernels in charge of setting up 2002 LPIs. 2003 2004 irqchip.gicv3_pseudo_nmi= [ARM64] 2005 Enables support for pseudo-NMIs in the kernel. This 2006 requires the kernel to be built with 2007 CONFIG_ARM64_PSEUDO_NMI. 2008 2009 irqfixup [HW] 2010 When an interrupt is not handled search all handlers 2011 for it. Intended to get systems with badly broken 2012 firmware running. 2013 2014 irqpoll [HW] 2015 When an interrupt is not handled search all handlers 2016 for it. Also check all handlers each timer 2017 interrupt. Intended to get systems with badly broken 2018 firmware running. 2019 2020 isapnp= [ISAPNP] 2021 Format: <RDP>,<reset>,<pci_scan>,<verbosity> 2022 2023 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance. 2024 [Deprecated - use cpusets instead] 2025 Format: [flag-list,]<cpu-list> 2026 2027 Specify one or more CPUs to isolate from disturbances 2028 specified in the flag list (default: domain): 2029 2030 nohz 2031 Disable the tick when a single task runs. 2032 2033 A residual 1Hz tick is offloaded to workqueues, which you 2034 need to affine to housekeeping through the global 2035 workqueue's affinity configured via the 2036 /sys/devices/virtual/workqueue/cpumask sysfs file, or 2037 by using the 'domain' flag described below. 2038 2039 NOTE: by default the global workqueue runs on all CPUs, 2040 so to protect individual CPUs the 'cpumask' file has to 2041 be configured manually after bootup. 2042 2043 domain 2044 Isolate from the general SMP balancing and scheduling 2045 algorithms. Note that performing domain isolation this way 2046 is irreversible: it's not possible to bring back a CPU to 2047 the domains once isolated through isolcpus. It's strongly 2048 advised to use cpusets instead to disable scheduler load 2049 balancing through the "cpuset.sched_load_balance" file. 2050 It offers a much more flexible interface where CPUs can 2051 move in and out of an isolated set anytime. 2052 2053 You can move a process onto or off an "isolated" CPU via 2054 the CPU affinity syscalls or cpuset. 2055 <cpu number> begins at 0 and the maximum value is 2056 "number of CPUs in system - 1". 2057 2058 managed_irq 2059 2060 Isolate from being targeted by managed interrupts 2061 which have an interrupt mask containing isolated 2062 CPUs. The affinity of managed interrupts is 2063 handled by the kernel and cannot be changed via 2064 the /proc/irq/* interfaces. 2065 2066 This isolation is best effort and only effective 2067 if the automatically assigned interrupt mask of a 2068 device queue contains isolated and housekeeping 2069 CPUs. If housekeeping CPUs are online then such 2070 interrupts are directed to the housekeeping CPU 2071 so that IO submitted on the housekeeping CPU 2072 cannot disturb the isolated CPU. 2073 2074 If a queue's affinity mask contains only isolated 2075 CPUs then this parameter has no effect on the 2076 interrupt routing decision, though interrupts are 2077 only delivered when tasks running on those 2078 isolated CPUs submit IO. IO submitted on 2079 housekeeping CPUs has no influence on those 2080 queues. 2081 2082 The format of <cpu-list> is described above. 2083 2084 iucv= [HW,NET] 2085 2086 ivrs_ioapic [HW,X86-64] 2087 Provide an override to the IOAPIC-ID<->DEVICE-ID 2088 mapping provided in the IVRS ACPI table. For 2089 example, to map IOAPIC-ID decimal 10 to 2090 PCI device 00:14.0 write the parameter as: 2091 ivrs_ioapic[10]=00:14.0 2092 2093 ivrs_hpet [HW,X86-64] 2094 Provide an override to the HPET-ID<->DEVICE-ID 2095 mapping provided in the IVRS ACPI table. For 2096 example, to map HPET-ID decimal 0 to 2097 PCI device 00:14.0 write the parameter as: 2098 ivrs_hpet[0]=00:14.0 2099 2100 ivrs_acpihid [HW,X86-64] 2101 Provide an override to the ACPI-HID:UID<->DEVICE-ID 2102 mapping provided in the IVRS ACPI table. For 2103 example, to map UART-HID:UID AMD0020:0 to 2104 PCI device 00:14.5 write the parameter as: 2105 ivrs_acpihid[00:14.5]=AMD0020:0 2106 2107 js= [HW,JOY] Analog joystick 2108 See Documentation/input/joydev/joystick.rst. 2109 2110 nokaslr [KNL] 2111 When CONFIG_RANDOMIZE_BASE is set, this disables 2112 kernel and module base offset ASLR (Address Space 2113 Layout Randomization). 2114 2115 kasan_multi_shot 2116 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print 2117 report on every invalid memory access. Without this 2118 parameter KASAN will print report only for the first 2119 invalid access. 2120 2121 keepinitrd [HW,ARM] 2122 2123 kernelcore= [KNL,X86,IA-64,PPC] 2124 Format: nn[KMGTPE] | nn% | "mirror" 2125 This parameter specifies the amount of memory usable by 2126 the kernel for non-movable allocations. The requested 2127 amount is spread evenly throughout all nodes in the 2128 system as ZONE_NORMAL. The remaining memory is used for 2129 movable memory in its own zone, ZONE_MOVABLE. In the 2130 event, a node is too small to have both ZONE_NORMAL and 2131 ZONE_MOVABLE, kernelcore memory will take priority and 2132 other nodes will have a larger ZONE_MOVABLE. 2133 2134 ZONE_MOVABLE is used for the allocation of pages that 2135 may be reclaimed or moved by the page migration 2136 subsystem. Note that allocations like PTEs-from-HighMem 2137 still use the HighMem zone if it exists, and the Normal 2138 zone if it does not. 2139 2140 It is possible to specify the exact amount of memory in 2141 the form of "nn[KMGTPE]", a percentage of total system 2142 memory in the form of "nn%", or "mirror". If "mirror" 2143 option is specified, mirrored (reliable) memory is used 2144 for non-movable allocations and remaining memory is used 2145 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror" 2146 are exclusive, so you cannot specify multiple forms. 2147 2148 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port. 2149 Format: <Controller#>[,poll interval] 2150 The controller # is the number of the ehci usb debug 2151 port as it is probed via PCI. The poll interval is 2152 optional and is the number seconds in between 2153 each poll cycle to the debug port in case you need 2154 the functionality for interrupting the kernel with 2155 gdb or control-c on the dbgp connection. When 2156 not using this parameter you use sysrq-g to break into 2157 the kernel debugger. 2158 2159 kgdboc= [KGDB,HW] kgdb over consoles. 2160 Requires a tty driver that supports console polling, 2161 or a supported polling keyboard driver (non-usb). 2162 Serial only format: <serial_device>[,baud] 2163 keyboard only format: kbd 2164 keyboard and serial format: kbd,<serial_device>[,baud] 2165 Optional Kernel mode setting: 2166 kms, kbd format: kms,kbd 2167 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud] 2168 2169 kgdboc_earlycon= [KGDB,HW] 2170 If the boot console provides the ability to read 2171 characters and can work in polling mode, you can use 2172 this parameter to tell kgdb to use it as a backend 2173 until the normal console is registered. Intended to 2174 be used together with the kgdboc parameter which 2175 specifies the normal console to transition to. 2176 2177 The name of the early console should be specified 2178 as the value of this parameter. Note that the name of 2179 the early console might be different than the tty 2180 name passed to kgdboc. It's OK to leave the value 2181 blank and the first boot console that implements 2182 read() will be picked. 2183 2184 kgdbwait [KGDB] Stop kernel execution and enter the 2185 kernel debugger at the earliest opportunity. 2186 2187 kmac= [MIPS] Korina ethernet MAC address. 2188 Configure the RouterBoard 532 series on-chip 2189 Ethernet adapter MAC address. 2190 2191 kmemleak= [KNL] Boot-time kmemleak enable/disable 2192 Valid arguments: on, off 2193 Default: on 2194 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y, 2195 the default is off. 2196 2197 kprobe_event=[probe-list] 2198 [FTRACE] Add kprobe events and enable at boot time. 2199 The probe-list is a semicolon delimited list of probe 2200 definitions. Each definition is same as kprobe_events 2201 interface, but the parameters are comma delimited. 2202 For example, to add a kprobe event on vfs_read with 2203 arg1 and arg2, add to the command line; 2204 2205 kprobe_event=p,vfs_read,$arg1,$arg2 2206 2207 See also Documentation/trace/kprobetrace.rst "Kernel 2208 Boot Parameter" section. 2209 2210 kpti= [ARM64] Control page table isolation of user 2211 and kernel address spaces. 2212 Default: enabled on cores which need mitigation. 2213 0: force disabled 2214 1: force enabled 2215 2216 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs. 2217 Default is 0 (don't ignore, but inject #GP) 2218 2219 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface. 2220 Default is false (don't support). 2221 2222 kvm.mmu_audit= [KVM] This is a R/W parameter which allows audit 2223 KVM MMU at runtime. 2224 Default is 0 (off) 2225 2226 kvm.nx_huge_pages= 2227 [KVM] Controls the software workaround for the 2228 X86_BUG_ITLB_MULTIHIT bug. 2229 force : Always deploy workaround. 2230 off : Never deploy workaround. 2231 auto : Deploy workaround based on the presence of 2232 X86_BUG_ITLB_MULTIHIT. 2233 2234 Default is 'auto'. 2235 2236 If the software workaround is enabled for the host, 2237 guests do need not to enable it for nested guests. 2238 2239 kvm.nx_huge_pages_recovery_ratio= 2240 [KVM] Controls how many 4KiB pages are periodically zapped 2241 back to huge pages. 0 disables the recovery, otherwise if 2242 the value is N KVM will zap 1/Nth of the 4KiB pages every 2243 minute. The default is 60. 2244 2245 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM. 2246 Default is 1 (enabled) 2247 2248 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU) 2249 for all guests. 2250 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode. 2251 2252 kvm-arm.mode= 2253 [KVM,ARM] Select one of KVM/arm64's modes of operation. 2254 2255 protected: nVHE-based mode with support for guests whose 2256 state is kept private from the host. 2257 Not valid if the kernel is running in EL2. 2258 2259 Defaults to VHE/nVHE based on hardware support and 2260 the value of CONFIG_ARM64_VHE. 2261 2262 kvm-arm.vgic_v3_group0_trap= 2263 [KVM,ARM] Trap guest accesses to GICv3 group-0 2264 system registers 2265 2266 kvm-arm.vgic_v3_group1_trap= 2267 [KVM,ARM] Trap guest accesses to GICv3 group-1 2268 system registers 2269 2270 kvm-arm.vgic_v3_common_trap= 2271 [KVM,ARM] Trap guest accesses to GICv3 common 2272 system registers 2273 2274 kvm-arm.vgic_v4_enable= 2275 [KVM,ARM] Allow use of GICv4 for direct injection of 2276 LPIs. 2277 2278 kvm_cma_resv_ratio=n [PPC] 2279 Reserves given percentage from system memory area for 2280 contiguous memory allocation for KVM hash pagetable 2281 allocation. 2282 By default it reserves 5% of total system memory. 2283 Format: <integer> 2284 Default: 5 2285 2286 kvm-intel.ept= [KVM,Intel] Disable extended page tables 2287 (virtualized MMU) support on capable Intel chips. 2288 Default is 1 (enabled) 2289 2290 kvm-intel.emulate_invalid_guest_state= 2291 [KVM,Intel] Enable emulation of invalid guest states 2292 Default is 0 (disabled) 2293 2294 kvm-intel.flexpriority= 2295 [KVM,Intel] Disable FlexPriority feature (TPR shadow). 2296 Default is 1 (enabled) 2297 2298 kvm-intel.nested= 2299 [KVM,Intel] Enable VMX nesting (nVMX). 2300 Default is 0 (disabled) 2301 2302 kvm-intel.unrestricted_guest= 2303 [KVM,Intel] Disable unrestricted guest feature 2304 (virtualized real and unpaged mode) on capable 2305 Intel chips. Default is 1 (enabled) 2306 2307 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault 2308 CVE-2018-3620. 2309 2310 Valid arguments: never, cond, always 2311 2312 always: L1D cache flush on every VMENTER. 2313 cond: Flush L1D on VMENTER only when the code between 2314 VMEXIT and VMENTER can leak host memory. 2315 never: Disables the mitigation 2316 2317 Default is cond (do L1 cache flush in specific instances) 2318 2319 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification 2320 feature (tagged TLBs) on capable Intel chips. 2321 Default is 1 (enabled) 2322 2323 l1tf= [X86] Control mitigation of the L1TF vulnerability on 2324 affected CPUs 2325 2326 The kernel PTE inversion protection is unconditionally 2327 enabled and cannot be disabled. 2328 2329 full 2330 Provides all available mitigations for the 2331 L1TF vulnerability. Disables SMT and 2332 enables all mitigations in the 2333 hypervisors, i.e. unconditional L1D flush. 2334 2335 SMT control and L1D flush control via the 2336 sysfs interface is still possible after 2337 boot. Hypervisors will issue a warning 2338 when the first VM is started in a 2339 potentially insecure configuration, 2340 i.e. SMT enabled or L1D flush disabled. 2341 2342 full,force 2343 Same as 'full', but disables SMT and L1D 2344 flush runtime control. Implies the 2345 'nosmt=force' command line option. 2346 (i.e. sysfs control of SMT is disabled.) 2347 2348 flush 2349 Leaves SMT enabled and enables the default 2350 hypervisor mitigation, i.e. conditional 2351 L1D flush. 2352 2353 SMT control and L1D flush control via the 2354 sysfs interface is still possible after 2355 boot. Hypervisors will issue a warning 2356 when the first VM is started in a 2357 potentially insecure configuration, 2358 i.e. SMT enabled or L1D flush disabled. 2359 2360 flush,nosmt 2361 2362 Disables SMT and enables the default 2363 hypervisor mitigation. 2364 2365 SMT control and L1D flush control via the 2366 sysfs interface is still possible after 2367 boot. Hypervisors will issue a warning 2368 when the first VM is started in a 2369 potentially insecure configuration, 2370 i.e. SMT enabled or L1D flush disabled. 2371 2372 flush,nowarn 2373 Same as 'flush', but hypervisors will not 2374 warn when a VM is started in a potentially 2375 insecure configuration. 2376 2377 off 2378 Disables hypervisor mitigations and doesn't 2379 emit any warnings. 2380 It also drops the swap size and available 2381 RAM limit restriction on both hypervisor and 2382 bare metal. 2383 2384 Default is 'flush'. 2385 2386 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst 2387 2388 l2cr= [PPC] 2389 2390 l3cr= [PPC] 2391 2392 lapic [X86-32,APIC] Enable the local APIC even if BIOS 2393 disabled it. 2394 2395 lapic= [X86,APIC] Do not use TSC deadline 2396 value for LAPIC timer one-shot implementation. Default 2397 back to the programmable timer unit in the LAPIC. 2398 Format: notscdeadline 2399 2400 lapic_timer_c2_ok [X86,APIC] trust the local apic timer 2401 in C2 power state. 2402 2403 libata.dma= [LIBATA] DMA control 2404 libata.dma=0 Disable all PATA and SATA DMA 2405 libata.dma=1 PATA and SATA Disk DMA only 2406 libata.dma=2 ATAPI (CDROM) DMA only 2407 libata.dma=4 Compact Flash DMA only 2408 Combinations also work, so libata.dma=3 enables DMA 2409 for disks and CDROMs, but not CFs. 2410 2411 libata.ignore_hpa= [LIBATA] Ignore HPA limit 2412 libata.ignore_hpa=0 keep BIOS limits (default) 2413 libata.ignore_hpa=1 ignore limits, using full disk 2414 2415 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume 2416 when set. 2417 Format: <int> 2418 2419 libata.force= [LIBATA] Force configurations. The format is comma- 2420 separated list of "[ID:]VAL" where ID is 2421 PORT[.DEVICE]. PORT and DEVICE are decimal numbers 2422 matching port, link or device. Basically, it matches 2423 the ATA ID string printed on console by libata. If 2424 the whole ID part is omitted, the last PORT and DEVICE 2425 values are used. If ID hasn't been specified yet, the 2426 configuration applies to all ports, links and devices. 2427 2428 If only DEVICE is omitted, the parameter applies to 2429 the port and all links and devices behind it. DEVICE 2430 number of 0 either selects the first device or the 2431 first fan-out link behind PMP device. It does not 2432 select the host link. DEVICE number of 15 selects the 2433 host link and device attached to it. 2434 2435 The VAL specifies the configuration to force. As long 2436 as there's no ambiguity shortcut notation is allowed. 2437 For example, both 1.5 and 1.5G would work for 1.5Gbps. 2438 The following configurations can be forced. 2439 2440 * Cable type: 40c, 80c, short40c, unk, ign or sata. 2441 Any ID with matching PORT is used. 2442 2443 * SATA link speed limit: 1.5Gbps or 3.0Gbps. 2444 2445 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7]. 2446 udma[/][16,25,33,44,66,100,133] notation is also 2447 allowed. 2448 2449 * [no]ncq: Turn on or off NCQ. 2450 2451 * [no]ncqtrim: Turn off queued DSM TRIM. 2452 2453 * nohrst, nosrst, norst: suppress hard, soft 2454 and both resets. 2455 2456 * rstonce: only attempt one reset during 2457 hot-unplug link recovery 2458 2459 * dump_id: dump IDENTIFY data. 2460 2461 * atapi_dmadir: Enable ATAPI DMADIR bridge support 2462 2463 * disable: Disable this device. 2464 2465 If there are multiple matching configurations changing 2466 the same attribute, the last one is used. 2467 2468 memblock=debug [KNL] Enable memblock debug messages. 2469 2470 load_ramdisk= [RAM] [Deprecated] 2471 2472 lockd.nlm_grace_period=P [NFS] Assign grace period. 2473 Format: <integer> 2474 2475 lockd.nlm_tcpport=N [NFS] Assign TCP port. 2476 Format: <integer> 2477 2478 lockd.nlm_timeout=T [NFS] Assign timeout value. 2479 Format: <integer> 2480 2481 lockd.nlm_udpport=M [NFS] Assign UDP port. 2482 Format: <integer> 2483 2484 lockdown= [SECURITY] 2485 { integrity | confidentiality } 2486 Enable the kernel lockdown feature. If set to 2487 integrity, kernel features that allow userland to 2488 modify the running kernel are disabled. If set to 2489 confidentiality, kernel features that allow userland 2490 to extract confidential information from the kernel 2491 are also disabled. 2492 2493 locktorture.nreaders_stress= [KNL] 2494 Set the number of locking read-acquisition kthreads. 2495 Defaults to being automatically set based on the 2496 number of online CPUs. 2497 2498 locktorture.nwriters_stress= [KNL] 2499 Set the number of locking write-acquisition kthreads. 2500 2501 locktorture.onoff_holdoff= [KNL] 2502 Set time (s) after boot for CPU-hotplug testing. 2503 2504 locktorture.onoff_interval= [KNL] 2505 Set time (s) between CPU-hotplug operations, or 2506 zero to disable CPU-hotplug testing. 2507 2508 locktorture.shuffle_interval= [KNL] 2509 Set task-shuffle interval (jiffies). Shuffling 2510 tasks allows some CPUs to go into dyntick-idle 2511 mode during the locktorture test. 2512 2513 locktorture.shutdown_secs= [KNL] 2514 Set time (s) after boot system shutdown. This 2515 is useful for hands-off automated testing. 2516 2517 locktorture.stat_interval= [KNL] 2518 Time (s) between statistics printk()s. 2519 2520 locktorture.stutter= [KNL] 2521 Time (s) to stutter testing, for example, 2522 specifying five seconds causes the test to run for 2523 five seconds, wait for five seconds, and so on. 2524 This tests the locking primitive's ability to 2525 transition abruptly to and from idle. 2526 2527 locktorture.torture_type= [KNL] 2528 Specify the locking implementation to test. 2529 2530 locktorture.verbose= [KNL] 2531 Enable additional printk() statements. 2532 2533 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver 2534 Format: <irq> 2535 2536 loglevel= All Kernel Messages with a loglevel smaller than the 2537 console loglevel will be printed to the console. It can 2538 also be changed with klogd or other programs. The 2539 loglevels are defined as follows: 2540 2541 0 (KERN_EMERG) system is unusable 2542 1 (KERN_ALERT) action must be taken immediately 2543 2 (KERN_CRIT) critical conditions 2544 3 (KERN_ERR) error conditions 2545 4 (KERN_WARNING) warning conditions 2546 5 (KERN_NOTICE) normal but significant condition 2547 6 (KERN_INFO) informational 2548 7 (KERN_DEBUG) debug-level messages 2549 2550 log_buf_len=n[KMG] Sets the size of the printk ring buffer, 2551 in bytes. n must be a power of two and greater 2552 than the minimal size. The minimal size is defined 2553 by LOG_BUF_SHIFT kernel config parameter. There is 2554 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter 2555 that allows to increase the default size depending on 2556 the number of CPUs. See init/Kconfig for more details. 2557 2558 logo.nologo [FB] Disables display of the built-in Linux logo. 2559 This may be used to provide more screen space for 2560 kernel log messages and is useful when debugging 2561 kernel boot problems. 2562 2563 lp=0 [LP] Specify parallel ports to use, e.g, 2564 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses 2565 lp=reset first parallel port). 'lp=0' disables the 2566 lp=auto printer driver. 'lp=reset' (which can be 2567 specified in addition to the ports) causes 2568 attached printers to be reset. Using 2569 lp=port1,port2,... specifies the parallel ports 2570 to associate lp devices with, starting with 2571 lp0. A port specification may be 'none' to skip 2572 that lp device, or a parport name such as 2573 'parport0'. Specifying 'lp=auto' instead of a 2574 port specification list means that device IDs 2575 from each port should be examined, to see if 2576 an IEEE 1284-compliant printer is attached; if 2577 so, the driver will manage that printer. 2578 See also header of drivers/char/lp.c. 2579 2580 lpj=n [KNL] 2581 Sets loops_per_jiffy to given constant, thus avoiding 2582 time-consuming boot-time autodetection (up to 250 ms per 2583 CPU). 0 enables autodetection (default). To determine 2584 the correct value for your kernel, boot with normal 2585 autodetection and see what value is printed. Note that 2586 on SMP systems the preset will be applied to all CPUs, 2587 which is likely to cause problems if your CPUs need 2588 significantly divergent settings. An incorrect value 2589 will cause delays in the kernel to be wrong, leading to 2590 unpredictable I/O errors and other breakage. Although 2591 unlikely, in the extreme case this might damage your 2592 hardware. 2593 2594 ltpc= [NET] 2595 Format: <io>,<irq>,<dma> 2596 2597 lsm.debug [SECURITY] Enable LSM initialization debugging output. 2598 2599 lsm=lsm1,...,lsmN 2600 [SECURITY] Choose order of LSM initialization. This 2601 overrides CONFIG_LSM, and the "security=" parameter. 2602 2603 machvec= [IA-64] Force the use of a particular machine-vector 2604 (machvec) in a generic kernel. 2605 Example: machvec=hpzx1 2606 2607 machtype= [Loongson] Share the same kernel image file between 2608 different yeeloong laptops. 2609 Example: machtype=lemote-yeeloong-2f-7inch 2610 2611 max_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory greater 2612 than or equal to this physical address is ignored. 2613 2614 maxcpus= [SMP] Maximum number of processors that an SMP kernel 2615 will bring up during bootup. maxcpus=n : n >= 0 limits 2616 the kernel to bring up 'n' processors. Surely after 2617 bootup you can bring up the other plugged cpu by executing 2618 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus 2619 only takes effect during system bootup. 2620 While n=0 is a special case, it is equivalent to "nosmp", 2621 which also disables the IO APIC. 2622 2623 max_loop= [LOOP] The number of loop block devices that get 2624 (loop.max_loop) unconditionally pre-created at init time. The default 2625 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead 2626 of statically allocating a predefined number, loop 2627 devices can be requested on-demand with the 2628 /dev/loop-control interface. 2629 2630 mce [X86-32] Machine Check Exception 2631 2632 mce=option [X86-64] See Documentation/x86/x86_64/boot-options.rst 2633 2634 md= [HW] RAID subsystems devices and level 2635 See Documentation/admin-guide/md.rst. 2636 2637 mdacon= [MDA] 2638 Format: <first>,<last> 2639 Specifies range of consoles to be captured by the MDA. 2640 2641 mds= [X86,INTEL] 2642 Control mitigation for the Micro-architectural Data 2643 Sampling (MDS) vulnerability. 2644 2645 Certain CPUs are vulnerable to an exploit against CPU 2646 internal buffers which can forward information to a 2647 disclosure gadget under certain conditions. 2648 2649 In vulnerable processors, the speculatively 2650 forwarded data can be used in a cache side channel 2651 attack, to access data to which the attacker does 2652 not have direct access. 2653 2654 This parameter controls the MDS mitigation. The 2655 options are: 2656 2657 full - Enable MDS mitigation on vulnerable CPUs 2658 full,nosmt - Enable MDS mitigation and disable 2659 SMT on vulnerable CPUs 2660 off - Unconditionally disable MDS mitigation 2661 2662 On TAA-affected machines, mds=off can be prevented by 2663 an active TAA mitigation as both vulnerabilities are 2664 mitigated with the same mechanism so in order to disable 2665 this mitigation, you need to specify tsx_async_abort=off 2666 too. 2667 2668 Not specifying this option is equivalent to 2669 mds=full. 2670 2671 For details see: Documentation/admin-guide/hw-vuln/mds.rst 2672 2673 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory 2674 Amount of memory to be used in cases as follows: 2675 2676 1 for test; 2677 2 when the kernel is not able to see the whole system memory; 2678 3 memory that lies after 'mem=' boundary is excluded from 2679 the hypervisor, then assigned to KVM guests. 2680 2681 [X86] Work as limiting max address. Use together 2682 with memmap= to avoid physical address space collisions. 2683 Without memmap= PCI devices could be placed at addresses 2684 belonging to unused RAM. 2685 2686 Note that this only takes effects during boot time since 2687 in above case 3, memory may need be hot added after boot 2688 if system memory of hypervisor is not sufficient. 2689 2690 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel 2691 memory. 2692 2693 memchunk=nn[KMG] 2694 [KNL,SH] Allow user to override the default size for 2695 per-device physically contiguous DMA buffers. 2696 2697 memhp_default_state=online/offline 2698 [KNL] Set the initial state for the memory hotplug 2699 onlining policy. If not specified, the default value is 2700 set according to the 2701 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config 2702 option. 2703 See Documentation/admin-guide/mm/memory-hotplug.rst. 2704 2705 memmap=exactmap [KNL,X86] Enable setting of an exact 2706 E820 memory map, as specified by the user. 2707 Such memmap=exactmap lines can be constructed based on 2708 BIOS output or other requirements. See the memmap=nn@ss 2709 option description. 2710 2711 memmap=nn[KMG]@ss[KMG] 2712 [KNL, X86, MIPS, XTENSA] Force usage of a specific region of memory. 2713 Region of memory to be used is from ss to ss+nn. 2714 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG], 2715 which limits max address to nn[KMG]. 2716 Multiple different regions can be specified, 2717 comma delimited. 2718 Example: 2719 memmap=100M@2G,100M#3G,1G!1024G 2720 2721 memmap=nn[KMG]#ss[KMG] 2722 [KNL,ACPI] Mark specific memory as ACPI data. 2723 Region of memory to be marked is from ss to ss+nn. 2724 2725 memmap=nn[KMG]$ss[KMG] 2726 [KNL,ACPI] Mark specific memory as reserved. 2727 Region of memory to be reserved is from ss to ss+nn. 2728 Example: Exclude memory from 0x18690000-0x1869ffff 2729 memmap=64K$0x18690000 2730 or 2731 memmap=0x10000$0x18690000 2732 Some bootloaders may need an escape character before '$', 2733 like Grub2, otherwise '$' and the following number 2734 will be eaten. 2735 2736 memmap=nn[KMG]!ss[KMG] 2737 [KNL,X86] Mark specific memory as protected. 2738 Region of memory to be used, from ss to ss+nn. 2739 The memory region may be marked as e820 type 12 (0xc) 2740 and is NVDIMM or ADR memory. 2741 2742 memmap=<size>%<offset>-<oldtype>+<newtype> 2743 [KNL,ACPI] Convert memory within the specified region 2744 from <oldtype> to <newtype>. If "-<oldtype>" is left 2745 out, the whole region will be marked as <newtype>, 2746 even if previously unavailable. If "+<newtype>" is left 2747 out, matching memory will be removed. Types are 2748 specified as e820 types, e.g., 1 = RAM, 2 = reserved, 2749 3 = ACPI, 12 = PRAM. 2750 2751 memory_corruption_check=0/1 [X86] 2752 Some BIOSes seem to corrupt the first 64k of 2753 memory when doing things like suspend/resume. 2754 Setting this option will scan the memory 2755 looking for corruption. Enabling this will 2756 both detect corruption and prevent the kernel 2757 from using the memory being corrupted. 2758 However, its intended as a diagnostic tool; if 2759 repeatable BIOS-originated corruption always 2760 affects the same memory, you can use memmap= 2761 to prevent the kernel from using that memory. 2762 2763 memory_corruption_check_size=size [X86] 2764 By default it checks for corruption in the low 2765 64k, making this memory unavailable for normal 2766 use. Use this parameter to scan for 2767 corruption in more or less memory. 2768 2769 memory_corruption_check_period=seconds [X86] 2770 By default it checks for corruption every 60 2771 seconds. Use this parameter to check at some 2772 other rate. 0 disables periodic checking. 2773 2774 memtest= [KNL,X86,ARM,PPC] Enable memtest 2775 Format: <integer> 2776 default : 0 <disable> 2777 Specifies the number of memtest passes to be 2778 performed. Each pass selects another test 2779 pattern from a given set of patterns. Memtest 2780 fills the memory with this pattern, validates 2781 memory contents and reserves bad memory 2782 regions that are detected. 2783 2784 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control 2785 Valid arguments: on, off 2786 Default (depends on kernel configuration option): 2787 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y) 2788 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n) 2789 mem_encrypt=on: Activate SME 2790 mem_encrypt=off: Do not activate SME 2791 2792 Refer to Documentation/virt/kvm/amd-memory-encryption.rst 2793 for details on when memory encryption can be activated. 2794 2795 mem_sleep_default= [SUSPEND] Default system suspend mode: 2796 s2idle - Suspend-To-Idle 2797 shallow - Power-On Suspend or equivalent (if supported) 2798 deep - Suspend-To-RAM or equivalent (if supported) 2799 See Documentation/admin-guide/pm/sleep-states.rst. 2800 2801 meye.*= [HW] Set MotionEye Camera parameters 2802 See Documentation/admin-guide/media/meye.rst. 2803 2804 mfgpt_irq= [IA-32] Specify the IRQ to use for the 2805 Multi-Function General Purpose Timers on AMD Geode 2806 platforms. 2807 2808 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when 2809 the BIOS has incorrectly applied a workaround. TinyBIOS 2810 version 0.98 is known to be affected, 0.99 fixes the 2811 problem by letting the user disable the workaround. 2812 2813 mga= [HW,DRM] 2814 2815 min_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory below this 2816 physical address is ignored. 2817 2818 mini2440= [ARM,HW,KNL] 2819 Format:[0..2][b][c][t] 2820 Default: "0tb" 2821 MINI2440 configuration specification: 2822 0 - The attached screen is the 3.5" TFT 2823 1 - The attached screen is the 7" TFT 2824 2 - The VGA Shield is attached (1024x768) 2825 Leaving out the screen size parameter will not load 2826 the TFT driver, and the framebuffer will be left 2827 unconfigured. 2828 b - Enable backlight. The TFT backlight pin will be 2829 linked to the kernel VESA blanking code and a GPIO 2830 LED. This parameter is not necessary when using the 2831 VGA shield. 2832 c - Enable the s3c camera interface. 2833 t - Reserved for enabling touchscreen support. The 2834 touchscreen support is not enabled in the mainstream 2835 kernel as of 2.6.30, a preliminary port can be found 2836 in the "bleeding edge" mini2440 support kernel at 2837 https://repo.or.cz/w/linux-2.6/mini2440.git 2838 2839 mitigations= 2840 [X86,PPC,S390,ARM64] Control optional mitigations for 2841 CPU vulnerabilities. This is a set of curated, 2842 arch-independent options, each of which is an 2843 aggregation of existing arch-specific options. 2844 2845 off 2846 Disable all optional CPU mitigations. This 2847 improves system performance, but it may also 2848 expose users to several CPU vulnerabilities. 2849 Equivalent to: nopti [X86,PPC] 2850 kpti=0 [ARM64] 2851 nospectre_v1 [X86,PPC] 2852 nobp=0 [S390] 2853 nospectre_v2 [X86,PPC,S390,ARM64] 2854 spectre_v2_user=off [X86] 2855 spec_store_bypass_disable=off [X86,PPC] 2856 ssbd=force-off [ARM64] 2857 l1tf=off [X86] 2858 mds=off [X86] 2859 tsx_async_abort=off [X86] 2860 kvm.nx_huge_pages=off [X86] 2861 no_entry_flush [PPC] 2862 no_uaccess_flush [PPC] 2863 2864 Exceptions: 2865 This does not have any effect on 2866 kvm.nx_huge_pages when 2867 kvm.nx_huge_pages=force. 2868 2869 auto (default) 2870 Mitigate all CPU vulnerabilities, but leave SMT 2871 enabled, even if it's vulnerable. This is for 2872 users who don't want to be surprised by SMT 2873 getting disabled across kernel upgrades, or who 2874 have other ways of avoiding SMT-based attacks. 2875 Equivalent to: (default behavior) 2876 2877 auto,nosmt 2878 Mitigate all CPU vulnerabilities, disabling SMT 2879 if needed. This is for users who always want to 2880 be fully mitigated, even if it means losing SMT. 2881 Equivalent to: l1tf=flush,nosmt [X86] 2882 mds=full,nosmt [X86] 2883 tsx_async_abort=full,nosmt [X86] 2884 2885 mminit_loglevel= 2886 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this 2887 parameter allows control of the logging verbosity for 2888 the additional memory initialisation checks. A value 2889 of 0 disables mminit logging and a level of 4 will 2890 log everything. Information is printed at KERN_DEBUG 2891 so loglevel=8 may also need to be specified. 2892 2893 module.sig_enforce 2894 [KNL] When CONFIG_MODULE_SIG is set, this means that 2895 modules without (valid) signatures will fail to load. 2896 Note that if CONFIG_MODULE_SIG_FORCE is set, that 2897 is always true, so this option does nothing. 2898 2899 module_blacklist= [KNL] Do not load a comma-separated list of 2900 modules. Useful for debugging problem modules. 2901 2902 mousedev.tap_time= 2903 [MOUSE] Maximum time between finger touching and 2904 leaving touchpad surface for touch to be considered 2905 a tap and be reported as a left button click (for 2906 touchpads working in absolute mode only). 2907 Format: <msecs> 2908 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices 2909 reporting absolute coordinates, such as tablets 2910 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices 2911 reporting absolute coordinates, such as tablets 2912 2913 movablecore= [KNL,X86,IA-64,PPC] 2914 Format: nn[KMGTPE] | nn% 2915 This parameter is the complement to kernelcore=, it 2916 specifies the amount of memory used for migratable 2917 allocations. If both kernelcore and movablecore is 2918 specified, then kernelcore will be at *least* the 2919 specified value but may be more. If movablecore on its 2920 own is specified, the administrator must be careful 2921 that the amount of memory usable for all allocations 2922 is not too small. 2923 2924 movable_node [KNL] Boot-time switch to make hotplugable memory 2925 NUMA nodes to be movable. This means that the memory 2926 of such nodes will be usable only for movable 2927 allocations which rules out almost all kernel 2928 allocations. Use with caution! 2929 2930 MTD_Partition= [MTD] 2931 Format: <name>,<region-number>,<size>,<offset> 2932 2933 MTD_Region= [MTD] Format: 2934 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>] 2935 2936 mtdparts= [MTD] 2937 See drivers/mtd/parsers/cmdlinepart.c 2938 2939 multitce=off [PPC] This parameter disables the use of the pSeries 2940 firmware feature for updating multiple TCE entries 2941 at a time. 2942 2943 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration 2944 2945 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock] 2946 2947 boundary - index of last SLC block on Flex-OneNAND. 2948 The remaining blocks are configured as MLC blocks. 2949 lock - Configure if Flex-OneNAND boundary should be locked. 2950 Once locked, the boundary cannot be changed. 2951 1 indicates lock status, 0 indicates unlock status. 2952 2953 mtdset= [ARM] 2954 ARM/S3C2412 JIVE boot control 2955 2956 See arch/arm/mach-s3c/mach-jive.c 2957 2958 mtouchusb.raw_coordinates= 2959 [HW] Make the MicroTouch USB driver use raw coordinates 2960 ('y', default) or cooked coordinates ('n') 2961 2962 mtrr_chunk_size=nn[KMG] [X86] 2963 used for mtrr cleanup. It is largest continuous chunk 2964 that could hold holes aka. UC entries. 2965 2966 mtrr_gran_size=nn[KMG] [X86] 2967 Used for mtrr cleanup. It is granularity of mtrr block. 2968 Default is 1. 2969 Large value could prevent small alignment from 2970 using up MTRRs. 2971 2972 mtrr_spare_reg_nr=n [X86] 2973 Format: <integer> 2974 Range: 0,7 : spare reg number 2975 Default : 1 2976 Used for mtrr cleanup. It is spare mtrr entries number. 2977 Set to 2 or more if your graphical card needs more. 2978 2979 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card 2980 2981 netdev= [NET] Network devices parameters 2982 Format: <irq>,<io>,<mem_start>,<mem_end>,<name> 2983 Note that mem_start is often overloaded to mean 2984 something different and driver-specific. 2985 This usage is only documented in each driver source 2986 file if at all. 2987 2988 nf_conntrack.acct= 2989 [NETFILTER] Enable connection tracking flow accounting 2990 0 to disable accounting 2991 1 to enable accounting 2992 Default value is 0. 2993 2994 nfsaddrs= [NFS] Deprecated. Use ip= instead. 2995 See Documentation/admin-guide/nfs/nfsroot.rst. 2996 2997 nfsroot= [NFS] nfs root filesystem for disk-less boxes. 2998 See Documentation/admin-guide/nfs/nfsroot.rst. 2999 3000 nfsrootdebug [NFS] enable nfsroot debugging messages. 3001 See Documentation/admin-guide/nfs/nfsroot.rst. 3002 3003 nfs.callback_nr_threads= 3004 [NFSv4] set the total number of threads that the 3005 NFS client will assign to service NFSv4 callback 3006 requests. 3007 3008 nfs.callback_tcpport= 3009 [NFS] set the TCP port on which the NFSv4 callback 3010 channel should listen. 3011 3012 nfs.cache_getent= 3013 [NFS] sets the pathname to the program which is used 3014 to update the NFS client cache entries. 3015 3016 nfs.cache_getent_timeout= 3017 [NFS] sets the timeout after which an attempt to 3018 update a cache entry is deemed to have failed. 3019 3020 nfs.idmap_cache_timeout= 3021 [NFS] set the maximum lifetime for idmapper cache 3022 entries. 3023 3024 nfs.enable_ino64= 3025 [NFS] enable 64-bit inode numbers. 3026 If zero, the NFS client will fake up a 32-bit inode 3027 number for the readdir() and stat() syscalls instead 3028 of returning the full 64-bit number. 3029 The default is to return 64-bit inode numbers. 3030 3031 nfs.max_session_cb_slots= 3032 [NFSv4.1] Sets the maximum number of session 3033 slots the client will assign to the callback 3034 channel. This determines the maximum number of 3035 callbacks the client will process in parallel for 3036 a particular server. 3037 3038 nfs.max_session_slots= 3039 [NFSv4.1] Sets the maximum number of session slots 3040 the client will attempt to negotiate with the server. 3041 This limits the number of simultaneous RPC requests 3042 that the client can send to the NFSv4.1 server. 3043 Note that there is little point in setting this 3044 value higher than the max_tcp_slot_table_limit. 3045 3046 nfs.nfs4_disable_idmapping= 3047 [NFSv4] When set to the default of '1', this option 3048 ensures that both the RPC level authentication 3049 scheme and the NFS level operations agree to use 3050 numeric uids/gids if the mount is using the 3051 'sec=sys' security flavour. In effect it is 3052 disabling idmapping, which can make migration from 3053 legacy NFSv2/v3 systems to NFSv4 easier. 3054 Servers that do not support this mode of operation 3055 will be autodetected by the client, and it will fall 3056 back to using the idmapper. 3057 To turn off this behaviour, set the value to '0'. 3058 nfs.nfs4_unique_id= 3059 [NFS4] Specify an additional fixed unique ident- 3060 ification string that NFSv4 clients can insert into 3061 their nfs_client_id4 string. This is typically a 3062 UUID that is generated at system install time. 3063 3064 nfs.send_implementation_id = 3065 [NFSv4.1] Send client implementation identification 3066 information in exchange_id requests. 3067 If zero, no implementation identification information 3068 will be sent. 3069 The default is to send the implementation identification 3070 information. 3071 3072 nfs.recover_lost_locks = 3073 [NFSv4] Attempt to recover locks that were lost due 3074 to a lease timeout on the server. Please note that 3075 doing this risks data corruption, since there are 3076 no guarantees that the file will remain unchanged 3077 after the locks are lost. 3078 If you want to enable the kernel legacy behaviour of 3079 attempting to recover these locks, then set this 3080 parameter to '1'. 3081 The default parameter value of '0' causes the kernel 3082 not to attempt recovery of lost locks. 3083 3084 nfs4.layoutstats_timer = 3085 [NFSv4.2] Change the rate at which the kernel sends 3086 layoutstats to the pNFS metadata server. 3087 3088 Setting this to value to 0 causes the kernel to use 3089 whatever value is the default set by the layout 3090 driver. A non-zero value sets the minimum interval 3091 in seconds between layoutstats transmissions. 3092 3093 nfsd.nfs4_disable_idmapping= 3094 [NFSv4] When set to the default of '1', the NFSv4 3095 server will return only numeric uids and gids to 3096 clients using auth_sys, and will accept numeric uids 3097 and gids from such clients. This is intended to ease 3098 migration from NFSv2/v3. 3099 3100 nmi_backtrace.backtrace_idle [KNL] 3101 Dump stacks even of idle CPUs in response to an 3102 NMI stack-backtrace request. 3103 3104 nmi_debug= [KNL,SH] Specify one or more actions to take 3105 when a NMI is triggered. 3106 Format: [state][,regs][,debounce][,die] 3107 3108 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels 3109 Format: [panic,][nopanic,][num] 3110 Valid num: 0 or 1 3111 0 - turn hardlockup detector in nmi_watchdog off 3112 1 - turn hardlockup detector in nmi_watchdog on 3113 When panic is specified, panic when an NMI watchdog 3114 timeout occurs (or 'nopanic' to not panic on an NMI 3115 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set) 3116 To disable both hard and soft lockup detectors, 3117 please see 'nowatchdog'. 3118 This is useful when you use a panic=... timeout and 3119 need the box quickly up again. 3120 3121 These settings can be accessed at runtime via 3122 the nmi_watchdog and hardlockup_panic sysctls. 3123 3124 netpoll.carrier_timeout= 3125 [NET] Specifies amount of time (in seconds) that 3126 netpoll should wait for a carrier. By default netpoll 3127 waits 4 seconds. 3128 3129 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths 3130 emulation library even if a 387 maths coprocessor 3131 is present. 3132 3133 no5lvl [X86-64] Disable 5-level paging mode. Forces 3134 kernel to use 4-level paging instead. 3135 3136 nofsgsbase [X86] Disables FSGSBASE instructions. 3137 3138 no_console_suspend 3139 [HW] Never suspend the console 3140 Disable suspending of consoles during suspend and 3141 hibernate operations. Once disabled, debugging 3142 messages can reach various consoles while the rest 3143 of the system is being put to sleep (ie, while 3144 debugging driver suspend/resume hooks). This may 3145 not work reliably with all consoles, but is known 3146 to work with serial and VGA consoles. 3147 To facilitate more flexible debugging, we also add 3148 console_suspend, a printk module parameter to control 3149 it. Users could use console_suspend (usually 3150 /sys/module/printk/parameters/console_suspend) to 3151 turn on/off it dynamically. 3152 3153 novmcoredd [KNL,KDUMP] 3154 Disable device dump. Device dump allows drivers to 3155 append dump data to vmcore so you can collect driver 3156 specified debug info. Drivers can append the data 3157 without any limit and this data is stored in memory, 3158 so this may cause significant memory stress. Disabling 3159 device dump can help save memory but the driver debug 3160 data will be no longer available. This parameter 3161 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP 3162 is set. 3163 3164 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien 3165 caches in the slab allocator. Saves per-node memory, 3166 but will impact performance. 3167 3168 noalign [KNL,ARM] 3169 3170 noaltinstr [S390] Disables alternative instructions patching 3171 (CPU alternatives feature). 3172 3173 noapic [SMP,APIC] Tells the kernel to not make use of any 3174 IOAPICs that may be present in the system. 3175 3176 noautogroup Disable scheduler automatic task group creation. 3177 3178 nobats [PPC] Do not use BATs for mapping kernel lowmem 3179 on "Classic" PPC cores. 3180 3181 nocache [ARM] 3182 3183 noclflush [BUGS=X86] Don't use the CLFLUSH instruction 3184 3185 nodelayacct [KNL] Disable per-task delay accounting 3186 3187 nodsp [SH] Disable hardware DSP at boot time. 3188 3189 noefi Disable EFI runtime services support. 3190 3191 no_entry_flush [PPC] Don't flush the L1-D cache when entering the kernel. 3192 3193 noexec [IA-64] 3194 3195 noexec [X86] 3196 On X86-32 available only on PAE configured kernels. 3197 noexec=on: enable non-executable mappings (default) 3198 noexec=off: disable non-executable mappings 3199 3200 nosmap [X86,PPC] 3201 Disable SMAP (Supervisor Mode Access Prevention) 3202 even if it is supported by processor. 3203 3204 nosmep [X86,PPC] 3205 Disable SMEP (Supervisor Mode Execution Prevention) 3206 even if it is supported by processor. 3207 3208 noexec32 [X86-64] 3209 This affects only 32-bit executables. 3210 noexec32=on: enable non-executable mappings (default) 3211 read doesn't imply executable mappings 3212 noexec32=off: disable non-executable mappings 3213 read implies executable mappings 3214 3215 nofpu [MIPS,SH] Disable hardware FPU at boot time. 3216 3217 nofxsr [BUGS=X86-32] Disables x86 floating point extended 3218 register save and restore. The kernel will only save 3219 legacy floating-point registers on task switch. 3220 3221 nohugeiomap [KNL,X86,PPC,ARM64] Disable kernel huge I/O mappings. 3222 3223 nosmt [KNL,S390] Disable symmetric multithreading (SMT). 3224 Equivalent to smt=1. 3225 3226 [KNL,X86] Disable symmetric multithreading (SMT). 3227 nosmt=force: Force disable SMT, cannot be undone 3228 via the sysfs control file. 3229 3230 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1 3231 (bounds check bypass). With this option data leaks are 3232 possible in the system. 3233 3234 nospectre_v2 [X86,PPC_FSL_BOOK3E,ARM64] Disable all mitigations for 3235 the Spectre variant 2 (indirect branch prediction) 3236 vulnerability. System may allow data leaks with this 3237 option. 3238 3239 nospec_store_bypass_disable 3240 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability 3241 3242 no_uaccess_flush 3243 [PPC] Don't flush the L1-D cache after accessing user data. 3244 3245 noxsave [BUGS=X86] Disables x86 extended register state save 3246 and restore using xsave. The kernel will fallback to 3247 enabling legacy floating-point and sse state. 3248 3249 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended 3250 register states. The kernel will fall back to use 3251 xsave to save the states. By using this parameter, 3252 performance of saving the states is degraded because 3253 xsave doesn't support modified optimization while 3254 xsaveopt supports it on xsaveopt enabled systems. 3255 3256 noxsaves [X86] Disables xsaves and xrstors used in saving and 3257 restoring x86 extended register state in compacted 3258 form of xsave area. The kernel will fall back to use 3259 xsaveopt and xrstor to save and restore the states 3260 in standard form of xsave area. By using this 3261 parameter, xsave area per process might occupy more 3262 memory on xsaves enabled systems. 3263 3264 nohlt [BUGS=ARM,SH] Tells the kernel that the sleep(SH) or 3265 wfi(ARM) instruction doesn't work correctly and not to 3266 use it. This is also useful when using JTAG debugger. 3267 3268 no_file_caps Tells the kernel not to honor file capabilities. The 3269 only way then for a file to be executed with privilege 3270 is to be setuid root or executed by root. 3271 3272 nohalt [IA-64] Tells the kernel not to use the power saving 3273 function PAL_HALT_LIGHT when idle. This increases 3274 power-consumption. On the positive side, it reduces 3275 interrupt wake-up latency, which may improve performance 3276 in certain environments such as networked servers or 3277 real-time systems. 3278 3279 nohibernate [HIBERNATION] Disable hibernation and resume. 3280 3281 nohz= [KNL] Boottime enable/disable dynamic ticks 3282 Valid arguments: on, off 3283 Default: on 3284 3285 nohz_full= [KNL,BOOT,SMP,ISOL] 3286 The argument is a cpu list, as described above. 3287 In kernels built with CONFIG_NO_HZ_FULL=y, set 3288 the specified list of CPUs whose tick will be stopped 3289 whenever possible. The boot CPU will be forced outside 3290 the range to maintain the timekeeping. Any CPUs 3291 in this list will have their RCU callbacks offloaded, 3292 just as if they had also been called out in the 3293 rcu_nocbs= boot parameter. 3294 3295 noiotrap [SH] Disables trapped I/O port accesses. 3296 3297 noirqdebug [X86-32] Disables the code which attempts to detect and 3298 disable unhandled interrupt sources. 3299 3300 no_timer_check [X86,APIC] Disables the code which tests for 3301 broken timer IRQ sources. 3302 3303 noisapnp [ISAPNP] Disables ISA PnP code. 3304 3305 noinitrd [RAM] Tells the kernel not to load any configured 3306 initial RAM disk. 3307 3308 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt 3309 remapping. 3310 [Deprecated - use intremap=off] 3311 3312 nointroute [IA-64] 3313 3314 noinvpcid [X86] Disable the INVPCID cpu feature. 3315 3316 nojitter [IA-64] Disables jitter checking for ITC timers. 3317 3318 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver 3319 3320 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page 3321 fault handling. 3322 3323 no-vmw-sched-clock 3324 [X86,PV_OPS] Disable paravirtualized VMware scheduler 3325 clock and use the default one. 3326 3327 no-steal-acc [X86,PV_OPS,ARM64] Disable paravirtualized steal time 3328 accounting. steal time is computed, but won't 3329 influence scheduler behaviour 3330 3331 nolapic [X86-32,APIC] Do not enable or use the local APIC. 3332 3333 nolapic_timer [X86-32,APIC] Do not use the local APIC timer. 3334 3335 noltlbs [PPC] Do not use large page/tlb entries for kernel 3336 lowmem mapping on PPC40x and PPC8xx 3337 3338 nomca [IA-64] Disable machine check abort handling 3339 3340 nomce [X86-32] Disable Machine Check Exception 3341 3342 nomfgpt [X86-32] Disable Multi-Function General Purpose 3343 Timer usage (for AMD Geode machines). 3344 3345 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to 3346 shutdown the other cpus. Instead use the REBOOT_VECTOR 3347 irq. 3348 3349 nomodule Disable module load 3350 3351 nopat [X86] Disable PAT (page attribute table extension of 3352 pagetables) support. 3353 3354 nopcid [X86-64] Disable the PCID cpu feature. 3355 3356 norandmaps Don't use address space randomization. Equivalent to 3357 echo 0 > /proc/sys/kernel/randomize_va_space 3358 3359 noreplace-smp [X86-32,SMP] Don't replace SMP instructions 3360 with UP alternatives 3361 3362 nordrand [X86] Disable kernel use of the RDRAND and 3363 RDSEED instructions even if they are supported 3364 by the processor. RDRAND and RDSEED are still 3365 available to user space applications. 3366 3367 noresume [SWSUSP] Disables resume and restores original swap 3368 space. 3369 3370 no-scroll [VGA] Disables scrollback. 3371 This is required for the Braillex ib80-piezo Braille 3372 reader made by F.H. Papenmeier (Germany). 3373 3374 nosbagart [IA-64] 3375 3376 nosep [BUGS=X86-32] Disables x86 SYSENTER/SYSEXIT support. 3377 3378 nosgx [X86-64,SGX] Disables Intel SGX kernel support. 3379 3380 nosmp [SMP] Tells an SMP kernel to act as a UP kernel, 3381 and disable the IO APIC. legacy for "maxcpus=0". 3382 3383 nosoftlockup [KNL] Disable the soft-lockup detector. 3384 3385 nosync [HW,M68K] Disables sync negotiation for all devices. 3386 3387 nowatchdog [KNL] Disable both lockup detectors, i.e. 3388 soft-lockup and NMI watchdog (hard-lockup). 3389 3390 nowb [ARM] 3391 3392 nox2apic [X86-64,APIC] Do not enable x2APIC mode. 3393 3394 cpu0_hotplug [X86] Turn on CPU0 hotplug feature when 3395 CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off. 3396 Some features depend on CPU0. Known dependencies are: 3397 1. Resume from suspend/hibernate depends on CPU0. 3398 Suspend/hibernate will fail if CPU0 is offline and you 3399 need to online CPU0 before suspend/hibernate. 3400 2. PIC interrupts also depend on CPU0. CPU0 can't be 3401 removed if a PIC interrupt is detected. 3402 It's said poweroff/reboot may depend on CPU0 on some 3403 machines although I haven't seen such issues so far 3404 after CPU0 is offline on a few tested machines. 3405 If the dependencies are under your control, you can 3406 turn on cpu0_hotplug. 3407 3408 nps_mtm_hs_ctr= [KNL,ARC] 3409 This parameter sets the maximum duration, in 3410 cycles, each HW thread of the CTOP can run 3411 without interruptions, before HW switches it. 3412 The actual maximum duration is 16 times this 3413 parameter's value. 3414 Format: integer between 1 and 255 3415 Default: 255 3416 3417 nptcg= [IA-64] Override max number of concurrent global TLB 3418 purges which is reported from either PAL_VM_SUMMARY or 3419 SAL PALO. 3420 3421 nr_cpus= [SMP] Maximum number of processors that an SMP kernel 3422 could support. nr_cpus=n : n >= 1 limits the kernel to 3423 support 'n' processors. It could be larger than the 3424 number of already plugged CPU during bootup, later in 3425 runtime you can physically add extra cpu until it reaches 3426 n. So during boot up some boot time memory for per-cpu 3427 variables need be pre-allocated for later physical cpu 3428 hot plugging. 3429 3430 nr_uarts= [SERIAL] maximum number of UARTs to be registered. 3431 3432 numa_balancing= [KNL,X86] Enable or disable automatic NUMA balancing. 3433 Allowed values are enable and disable 3434 3435 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA. 3436 'node', 'default' can be specified 3437 This can be set from sysctl after boot. 3438 See Documentation/admin-guide/sysctl/vm.rst for details. 3439 3440 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver. 3441 See Documentation/core-api/debugging-via-ohci1394.rst for more 3442 info. 3443 3444 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands 3445 Rather than timing out after 20 ms if an EC 3446 command is not properly ACKed, override the length 3447 of the timeout. We have interrupts disabled while 3448 waiting for the ACK, so if this is set too high 3449 interrupts *may* be lost! 3450 3451 omap_mux= [OMAP] Override bootloader pin multiplexing. 3452 Format: <mux_mode0.mode_name=value>... 3453 For example, to override I2C bus2: 3454 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100 3455 3456 oops=panic Always panic on oopses. Default is to just kill the 3457 process, but there is a small probability of 3458 deadlocking the machine. 3459 This will also cause panics on machine check exceptions. 3460 Useful together with panic=30 to trigger a reboot. 3461 3462 page_alloc.shuffle= 3463 [KNL] Boolean flag to control whether the page allocator 3464 should randomize its free lists. The randomization may 3465 be automatically enabled if the kernel detects it is 3466 running on a platform with a direct-mapped memory-side 3467 cache, and this parameter can be used to 3468 override/disable that behavior. The state of the flag 3469 can be read from sysfs at: 3470 /sys/module/page_alloc/parameters/shuffle. 3471 3472 page_owner= [KNL] Boot-time page_owner enabling option. 3473 Storage of the information about who allocated 3474 each page is disabled in default. With this switch, 3475 we can turn it on. 3476 on: enable the feature 3477 3478 page_poison= [KNL] Boot-time parameter changing the state of 3479 poisoning on the buddy allocator, available with 3480 CONFIG_PAGE_POISONING=y. 3481 off: turn off poisoning (default) 3482 on: turn on poisoning 3483 3484 panic= [KNL] Kernel behaviour on panic: delay <timeout> 3485 timeout > 0: seconds before rebooting 3486 timeout = 0: wait forever 3487 timeout < 0: reboot immediately 3488 Format: <timeout> 3489 3490 panic_print= Bitmask for printing system info when panic happens. 3491 User can chose combination of the following bits: 3492 bit 0: print all tasks info 3493 bit 1: print system memory info 3494 bit 2: print timer info 3495 bit 3: print locks info if CONFIG_LOCKDEP is on 3496 bit 4: print ftrace buffer 3497 bit 5: print all printk messages in buffer 3498 3499 panic_on_taint= Bitmask for conditionally calling panic() in add_taint() 3500 Format: <hex>[,nousertaint] 3501 Hexadecimal bitmask representing the set of TAINT flags 3502 that will cause the kernel to panic when add_taint() is 3503 called with any of the flags in this set. 3504 The optional switch "nousertaint" can be utilized to 3505 prevent userspace forced crashes by writing to sysctl 3506 /proc/sys/kernel/tainted any flagset matching with the 3507 bitmask set on panic_on_taint. 3508 See Documentation/admin-guide/tainted-kernels.rst for 3509 extra details on the taint flags that users can pick 3510 to compose the bitmask to assign to panic_on_taint. 3511 3512 panic_on_warn panic() instead of WARN(). Useful to cause kdump 3513 on a WARN(). 3514 3515 crash_kexec_post_notifiers 3516 Run kdump after running panic-notifiers and dumping 3517 kmsg. This only for the users who doubt kdump always 3518 succeeds in any situation. 3519 Note that this also increases risks of kdump failure, 3520 because some panic notifiers can make the crashed 3521 kernel more unstable. 3522 3523 parkbd.port= [HW] Parallel port number the keyboard adapter is 3524 connected to, default is 0. 3525 Format: <parport#> 3526 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation, 3527 0 for XT, 1 for AT (default is AT). 3528 Format: <mode> 3529 3530 parport= [HW,PPT] Specify parallel ports. 0 disables. 3531 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] } 3532 Use 'auto' to force the driver to use any 3533 IRQ/DMA settings detected (the default is to 3534 ignore detected IRQ/DMA settings because of 3535 possible conflicts). You can specify the base 3536 address, IRQ, and DMA settings; IRQ and DMA 3537 should be numbers, or 'auto' (for using detected 3538 settings on that particular port), or 'nofifo' 3539 (to avoid using a FIFO even if it is detected). 3540 Parallel ports are assigned in the order they 3541 are specified on the command line, starting 3542 with parport0. 3543 3544 parport_init_mode= [HW,PPT] 3545 Configure VIA parallel port to operate in 3546 a specific mode. This is necessary on Pegasos 3547 computer where firmware has no options for setting 3548 up parallel port mode and sets it to spp. 3549 Currently this function knows 686a and 8231 chips. 3550 Format: [spp|ps2|epp|ecp|ecpepp] 3551 3552 pause_on_oops= 3553 Halt all CPUs after the first oops has been printed for 3554 the specified number of seconds. This is to be used if 3555 your oopses keep scrolling off the screen. 3556 3557 pcbit= [HW,ISDN] 3558 3559 pcd. [PARIDE] 3560 See header of drivers/block/paride/pcd.c. 3561 See also Documentation/admin-guide/blockdev/paride.rst. 3562 3563 pci=option[,option...] [PCI] various PCI subsystem options. 3564 3565 Some options herein operate on a specific device 3566 or a set of devices (<pci_dev>). These are 3567 specified in one of the following formats: 3568 3569 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]* 3570 pci:<vendor>:<device>[:<subvendor>:<subdevice>] 3571 3572 Note: the first format specifies a PCI 3573 bus/device/function address which may change 3574 if new hardware is inserted, if motherboard 3575 firmware changes, or due to changes caused 3576 by other kernel parameters. If the 3577 domain is left unspecified, it is 3578 taken to be zero. Optionally, a path 3579 to a device through multiple device/function 3580 addresses can be specified after the base 3581 address (this is more robust against 3582 renumbering issues). The second format 3583 selects devices using IDs from the 3584 configuration space which may match multiple 3585 devices in the system. 3586 3587 earlydump dump PCI config space before the kernel 3588 changes anything 3589 off [X86] don't probe for the PCI bus 3590 bios [X86-32] force use of PCI BIOS, don't access 3591 the hardware directly. Use this if your machine 3592 has a non-standard PCI host bridge. 3593 nobios [X86-32] disallow use of PCI BIOS, only direct 3594 hardware access methods are allowed. Use this 3595 if you experience crashes upon bootup and you 3596 suspect they are caused by the BIOS. 3597 conf1 [X86] Force use of PCI Configuration Access 3598 Mechanism 1 (config address in IO port 0xCF8, 3599 data in IO port 0xCFC, both 32-bit). 3600 conf2 [X86] Force use of PCI Configuration Access 3601 Mechanism 2 (IO port 0xCF8 is an 8-bit port for 3602 the function, IO port 0xCFA, also 8-bit, sets 3603 bus number. The config space is then accessed 3604 through ports 0xC000-0xCFFF). 3605 See http://wiki.osdev.org/PCI for more info 3606 on the configuration access mechanisms. 3607 noaer [PCIE] If the PCIEAER kernel config parameter is 3608 enabled, this kernel boot option can be used to 3609 disable the use of PCIE advanced error reporting. 3610 nodomains [PCI] Disable support for multiple PCI 3611 root domains (aka PCI segments, in ACPI-speak). 3612 nommconf [X86] Disable use of MMCONFIG for PCI 3613 Configuration 3614 check_enable_amd_mmconf [X86] check for and enable 3615 properly configured MMIO access to PCI 3616 config space on AMD family 10h CPU 3617 nomsi [MSI] If the PCI_MSI kernel config parameter is 3618 enabled, this kernel boot option can be used to 3619 disable the use of MSI interrupts system-wide. 3620 noioapicquirk [APIC] Disable all boot interrupt quirks. 3621 Safety option to keep boot IRQs enabled. This 3622 should never be necessary. 3623 ioapicreroute [APIC] Enable rerouting of boot IRQs to the 3624 primary IO-APIC for bridges that cannot disable 3625 boot IRQs. This fixes a source of spurious IRQs 3626 when the system masks IRQs. 3627 noioapicreroute [APIC] Disable workaround that uses the 3628 boot IRQ equivalent of an IRQ that connects to 3629 a chipset where boot IRQs cannot be disabled. 3630 The opposite of ioapicreroute. 3631 biosirq [X86-32] Use PCI BIOS calls to get the interrupt 3632 routing table. These calls are known to be buggy 3633 on several machines and they hang the machine 3634 when used, but on other computers it's the only 3635 way to get the interrupt routing table. Try 3636 this option if the kernel is unable to allocate 3637 IRQs or discover secondary PCI buses on your 3638 motherboard. 3639 rom [X86] Assign address space to expansion ROMs. 3640 Use with caution as certain devices share 3641 address decoders between ROMs and other 3642 resources. 3643 norom [X86] Do not assign address space to 3644 expansion ROMs that do not already have 3645 BIOS assigned address ranges. 3646 nobar [X86] Do not assign address space to the 3647 BARs that weren't assigned by the BIOS. 3648 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be 3649 assigned automatically to PCI devices. You can 3650 make the kernel exclude IRQs of your ISA cards 3651 this way. 3652 pirqaddr=0xAAAAA [X86] Specify the physical address 3653 of the PIRQ table (normally generated 3654 by the BIOS) if it is outside the 3655 F0000h-100000h range. 3656 lastbus=N [X86] Scan all buses thru bus #N. Can be 3657 useful if the kernel is unable to find your 3658 secondary buses and you want to tell it 3659 explicitly which ones they are. 3660 assign-busses [X86] Always assign all PCI bus 3661 numbers ourselves, overriding 3662 whatever the firmware may have done. 3663 usepirqmask [X86] Honor the possible IRQ mask stored 3664 in the BIOS $PIR table. This is needed on 3665 some systems with broken BIOSes, notably 3666 some HP Pavilion N5400 and Omnibook XE3 3667 notebooks. This will have no effect if ACPI 3668 IRQ routing is enabled. 3669 noacpi [X86] Do not use ACPI for IRQ routing 3670 or for PCI scanning. 3671 use_crs [X86] Use PCI host bridge window information 3672 from ACPI. On BIOSes from 2008 or later, this 3673 is enabled by default. If you need to use this, 3674 please report a bug. 3675 nocrs [X86] Ignore PCI host bridge windows from ACPI. 3676 If you need to use this, please report a bug. 3677 routeirq Do IRQ routing for all PCI devices. 3678 This is normally done in pci_enable_device(), 3679 so this option is a temporary workaround 3680 for broken drivers that don't call it. 3681 skip_isa_align [X86] do not align io start addr, so can 3682 handle more pci cards 3683 noearly [X86] Don't do any early type 1 scanning. 3684 This might help on some broken boards which 3685 machine check when some devices' config space 3686 is read. But various workarounds are disabled 3687 and some IOMMU drivers will not work. 3688 bfsort Sort PCI devices into breadth-first order. 3689 This sorting is done to get a device 3690 order compatible with older (<= 2.4) kernels. 3691 nobfsort Don't sort PCI devices into breadth-first order. 3692 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size) 3693 tuning and use the BIOS-configured MPS defaults. 3694 pcie_bus_safe Set every device's MPS to the largest value 3695 supported by all devices below the root complex. 3696 pcie_bus_perf Set device MPS to the largest allowable MPS 3697 based on its parent bus. Also set MRRS (Max 3698 Read Request Size) to the largest supported 3699 value (no larger than the MPS that the device 3700 or bus can support) for best performance. 3701 pcie_bus_peer2peer Set every device's MPS to 128B, which 3702 every device is guaranteed to support. This 3703 configuration allows peer-to-peer DMA between 3704 any pair of devices, possibly at the cost of 3705 reduced performance. This also guarantees 3706 that hot-added devices will work. 3707 cbiosize=nn[KMG] The fixed amount of bus space which is 3708 reserved for the CardBus bridge's IO window. 3709 The default value is 256 bytes. 3710 cbmemsize=nn[KMG] The fixed amount of bus space which is 3711 reserved for the CardBus bridge's memory 3712 window. The default value is 64 megabytes. 3713 resource_alignment= 3714 Format: 3715 [<order of align>@]<pci_dev>[; ...] 3716 Specifies alignment and device to reassign 3717 aligned memory resources. How to 3718 specify the device is described above. 3719 If <order of align> is not specified, 3720 PAGE_SIZE is used as alignment. 3721 A PCI-PCI bridge can be specified if resource 3722 windows need to be expanded. 3723 To specify the alignment for several 3724 instances of a device, the PCI vendor, 3725 device, subvendor, and subdevice may be 3726 specified, e.g., 12@pci:8086:9c22:103c:198f 3727 for 4096-byte alignment. 3728 ecrc= Enable/disable PCIe ECRC (transaction layer 3729 end-to-end CRC checking). 3730 bios: Use BIOS/firmware settings. This is the 3731 the default. 3732 off: Turn ECRC off 3733 on: Turn ECRC on. 3734 hpiosize=nn[KMG] The fixed amount of bus space which is 3735 reserved for hotplug bridge's IO window. 3736 Default size is 256 bytes. 3737 hpmmiosize=nn[KMG] The fixed amount of bus space which is 3738 reserved for hotplug bridge's MMIO window. 3739 Default size is 2 megabytes. 3740 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is 3741 reserved for hotplug bridge's MMIO_PREF window. 3742 Default size is 2 megabytes. 3743 hpmemsize=nn[KMG] The fixed amount of bus space which is 3744 reserved for hotplug bridge's MMIO and 3745 MMIO_PREF window. 3746 Default size is 2 megabytes. 3747 hpbussize=nn The minimum amount of additional bus numbers 3748 reserved for buses below a hotplug bridge. 3749 Default is 1. 3750 realloc= Enable/disable reallocating PCI bridge resources 3751 if allocations done by BIOS are too small to 3752 accommodate resources required by all child 3753 devices. 3754 off: Turn realloc off 3755 on: Turn realloc on 3756 realloc same as realloc=on 3757 noari do not use PCIe ARI. 3758 noats [PCIE, Intel-IOMMU, AMD-IOMMU] 3759 do not use PCIe ATS (and IOMMU device IOTLB). 3760 pcie_scan_all Scan all possible PCIe devices. Otherwise we 3761 only look for one device below a PCIe downstream 3762 port. 3763 big_root_window Try to add a big 64bit memory window to the PCIe 3764 root complex on AMD CPUs. Some GFX hardware 3765 can resize a BAR to allow access to all VRAM. 3766 Adding the window is slightly risky (it may 3767 conflict with unreported devices), so this 3768 taints the kernel. 3769 disable_acs_redir=<pci_dev>[; ...] 3770 Specify one or more PCI devices (in the format 3771 specified above) separated by semicolons. 3772 Each device specified will have the PCI ACS 3773 redirect capabilities forced off which will 3774 allow P2P traffic between devices through 3775 bridges without forcing it upstream. Note: 3776 this removes isolation between devices and 3777 may put more devices in an IOMMU group. 3778 force_floating [S390] Force usage of floating interrupts. 3779 nomio [S390] Do not use MIO instructions. 3780 norid [S390] ignore the RID field and force use of 3781 one PCI domain per PCI function 3782 3783 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power 3784 Management. 3785 off Disable ASPM. 3786 force Enable ASPM even on devices that claim not to support it. 3787 WARNING: Forcing ASPM on may cause system lockups. 3788 3789 pcie_ports= [PCIE] PCIe port services handling: 3790 native Use native PCIe services (PME, AER, DPC, PCIe hotplug) 3791 even if the platform doesn't give the OS permission to 3792 use them. This may cause conflicts if the platform 3793 also tries to use these services. 3794 dpc-native Use native PCIe service for DPC only. May 3795 cause conflicts if firmware uses AER or DPC. 3796 compat Disable native PCIe services (PME, AER, DPC, PCIe 3797 hotplug). 3798 3799 pcie_port_pm= [PCIE] PCIe port power management handling: 3800 off Disable power management of all PCIe ports 3801 force Forcibly enable power management of all PCIe ports 3802 3803 pcie_pme= [PCIE,PM] Native PCIe PME signaling options: 3804 nomsi Do not use MSI for native PCIe PME signaling (this makes 3805 all PCIe root ports use INTx for all services). 3806 3807 pcmv= [HW,PCMCIA] BadgePAD 4 3808 3809 pd_ignore_unused 3810 [PM] 3811 Keep all power-domains already enabled by bootloader on, 3812 even if no driver has claimed them. This is useful 3813 for debug and development, but should not be 3814 needed on a platform with proper driver support. 3815 3816 pd. [PARIDE] 3817 See Documentation/admin-guide/blockdev/paride.rst. 3818 3819 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at 3820 boot time. 3821 Format: { 0 | 1 } 3822 See arch/parisc/kernel/pdc_chassis.c 3823 3824 percpu_alloc= Select which percpu first chunk allocator to use. 3825 Currently supported values are "embed" and "page". 3826 Archs may support subset or none of the selections. 3827 See comments in mm/percpu.c for details on each 3828 allocator. This parameter is primarily for debugging 3829 and performance comparison. 3830 3831 pf. [PARIDE] 3832 See Documentation/admin-guide/blockdev/paride.rst. 3833 3834 pg. [PARIDE] 3835 See Documentation/admin-guide/blockdev/paride.rst. 3836 3837 pirq= [SMP,APIC] Manual mp-table setup 3838 See Documentation/x86/i386/IO-APIC.rst. 3839 3840 plip= [PPT,NET] Parallel port network link 3841 Format: { parport<nr> | timid | 0 } 3842 See also Documentation/admin-guide/parport.rst. 3843 3844 pmtmr= [X86] Manual setup of pmtmr I/O Port. 3845 Override pmtimer IOPort with a hex value. 3846 e.g. pmtmr=0x508 3847 3848 pm_debug_messages [SUSPEND,KNL] 3849 Enable suspend/resume debug messages during boot up. 3850 3851 pnp.debug=1 [PNP] 3852 Enable PNP debug messages (depends on the 3853 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time 3854 via /sys/module/pnp/parameters/debug. We always show 3855 current resource usage; turning this on also shows 3856 possible settings and some assignment information. 3857 3858 pnpacpi= [ACPI] 3859 { off } 3860 3861 pnpbios= [ISAPNP] 3862 { on | off | curr | res | no-curr | no-res } 3863 3864 pnp_reserve_irq= 3865 [ISAPNP] Exclude IRQs for the autoconfiguration 3866 3867 pnp_reserve_dma= 3868 [ISAPNP] Exclude DMAs for the autoconfiguration 3869 3870 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration 3871 Ranges are in pairs (I/O port base and size). 3872 3873 pnp_reserve_mem= 3874 [ISAPNP] Exclude memory regions for the 3875 autoconfiguration. 3876 Ranges are in pairs (memory base and size). 3877 3878 ports= [IP_VS_FTP] IPVS ftp helper module 3879 Default is 21. 3880 Up to 8 (IP_VS_APP_MAX_PORTS) ports 3881 may be specified. 3882 Format: <port>,<port>.... 3883 3884 powersave=off [PPC] This option disables power saving features. 3885 It specifically disables cpuidle and sets the 3886 platform machine description specific power_save 3887 function to NULL. On Idle the CPU just reduces 3888 execution priority. 3889 3890 ppc_strict_facility_enable 3891 [PPC] This option catches any kernel floating point, 3892 Altivec, VSX and SPE outside of regions specifically 3893 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()). 3894 There is some performance impact when enabling this. 3895 3896 ppc_tm= [PPC] 3897 Format: {"off"} 3898 Disable Hardware Transactional Memory 3899 3900 preempt= [KNL] 3901 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC 3902 none - Limited to cond_resched() calls 3903 voluntary - Limited to cond_resched() and might_sleep() calls 3904 full - Any section that isn't explicitly preempt disabled 3905 can be preempted anytime. 3906 3907 print-fatal-signals= 3908 [KNL] debug: print fatal signals 3909 3910 If enabled, warn about various signal handling 3911 related application anomalies: too many signals, 3912 too many POSIX.1 timers, fatal signals causing a 3913 coredump - etc. 3914 3915 If you hit the warning due to signal overflow, 3916 you might want to try "ulimit -i unlimited". 3917 3918 default: off. 3919 3920 printk.always_kmsg_dump= 3921 Trigger kmsg_dump for cases other than kernel oops or 3922 panics 3923 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 3924 default: disabled 3925 3926 printk.devkmsg={on,off,ratelimit} 3927 Control writing to /dev/kmsg. 3928 on - unlimited logging to /dev/kmsg from userspace 3929 off - logging to /dev/kmsg disabled 3930 ratelimit - ratelimit the logging 3931 Default: ratelimit 3932 3933 printk.time= Show timing data prefixed to each printk message line 3934 Format: <bool> (1/Y/y=enable, 0/N/n=disable) 3935 3936 processor.max_cstate= [HW,ACPI] 3937 Limit processor to maximum C-state 3938 max_cstate=9 overrides any DMI blacklist limit. 3939 3940 processor.nocst [HW,ACPI] 3941 Ignore the _CST method to determine C-states, 3942 instead using the legacy FADT method 3943 3944 profile= [KNL] Enable kernel profiling via /proc/profile 3945 Format: [<profiletype>,]<number> 3946 Param: <profiletype>: "schedule", "sleep", or "kvm" 3947 [defaults to kernel profiling] 3948 Param: "schedule" - profile schedule points. 3949 Param: "sleep" - profile D-state sleeping (millisecs). 3950 Requires CONFIG_SCHEDSTATS 3951 Param: "kvm" - profile VM exits. 3952 Param: <number> - step/bucket size as a power of 2 for 3953 statistical time based profiling. 3954 3955 prompt_ramdisk= [RAM] [Deprecated] 3956 3957 prot_virt= [S390] enable hosting protected virtual machines 3958 isolated from the hypervisor (if hardware supports 3959 that). 3960 Format: <bool> 3961 3962 psi= [KNL] Enable or disable pressure stall information 3963 tracking. 3964 Format: <bool> 3965 3966 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to 3967 probe for; one of (bare|imps|exps|lifebook|any). 3968 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports 3969 per second. 3970 psmouse.resetafter= [HW,MOUSE] 3971 Try to reset the device after so many bad packets 3972 (0 = never). 3973 psmouse.resolution= 3974 [HW,MOUSE] Set desired mouse resolution, in dpi. 3975 psmouse.smartscroll= 3976 [HW,MOUSE] Controls Logitech smartscroll autorepeat. 3977 0 = disabled, 1 = enabled (default). 3978 3979 pstore.backend= Specify the name of the pstore backend to use 3980 3981 pt. [PARIDE] 3982 See Documentation/admin-guide/blockdev/paride.rst. 3983 3984 pti= [X86-64] Control Page Table Isolation of user and 3985 kernel address spaces. Disabling this feature 3986 removes hardening, but improves performance of 3987 system calls and interrupts. 3988 3989 on - unconditionally enable 3990 off - unconditionally disable 3991 auto - kernel detects whether your CPU model is 3992 vulnerable to issues that PTI mitigates 3993 3994 Not specifying this option is equivalent to pti=auto. 3995 3996 nopti [X86-64] 3997 Equivalent to pti=off 3998 3999 pty.legacy_count= 4000 [KNL] Number of legacy pty's. Overwrites compiled-in 4001 default number. 4002 4003 quiet [KNL] Disable most log messages 4004 4005 r128= [HW,DRM] 4006 4007 raid= [HW,RAID] 4008 See Documentation/admin-guide/md.rst. 4009 4010 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes 4011 See Documentation/admin-guide/blockdev/ramdisk.rst. 4012 4013 ramdisk_start= [RAM] RAM disk image start address 4014 4015 random.trust_cpu={on,off} 4016 [KNL] Enable or disable trusting the use of the 4017 CPU's random number generator (if available) to 4018 fully seed the kernel's CRNG. Default is controlled 4019 by CONFIG_RANDOM_TRUST_CPU. 4020 4021 ras=option[,option,...] [KNL] RAS-specific options 4022 4023 cec_disable [X86] 4024 Disable the Correctable Errors Collector, 4025 see CONFIG_RAS_CEC help text. 4026 4027 rcu_nocbs= [KNL] 4028 The argument is a cpu list, as described above, 4029 except that the string "all" can be used to 4030 specify every CPU on the system. 4031 4032 In kernels built with CONFIG_RCU_NOCB_CPU=y, set 4033 the specified list of CPUs to be no-callback CPUs. 4034 Invocation of these CPUs' RCU callbacks will be 4035 offloaded to "rcuox/N" kthreads created for that 4036 purpose, where "x" is "p" for RCU-preempt, and 4037 "s" for RCU-sched, and "N" is the CPU number. 4038 This reduces OS jitter on the offloaded CPUs, 4039 which can be useful for HPC and real-time 4040 workloads. It can also improve energy efficiency 4041 for asymmetric multiprocessors. 4042 4043 rcu_nocb_poll [KNL] 4044 Rather than requiring that offloaded CPUs 4045 (specified by rcu_nocbs= above) explicitly 4046 awaken the corresponding "rcuoN" kthreads, 4047 make these kthreads poll for callbacks. 4048 This improves the real-time response for the 4049 offloaded CPUs by relieving them of the need to 4050 wake up the corresponding kthread, but degrades 4051 energy efficiency by requiring that the kthreads 4052 periodically wake up to do the polling. 4053 4054 rcutree.blimit= [KNL] 4055 Set maximum number of finished RCU callbacks to 4056 process in one batch. 4057 4058 rcutree.dump_tree= [KNL] 4059 Dump the structure of the rcu_node combining tree 4060 out at early boot. This is used for diagnostic 4061 purposes, to verify correct tree setup. 4062 4063 rcutree.gp_cleanup_delay= [KNL] 4064 Set the number of jiffies to delay each step of 4065 RCU grace-period cleanup. 4066 4067 rcutree.gp_init_delay= [KNL] 4068 Set the number of jiffies to delay each step of 4069 RCU grace-period initialization. 4070 4071 rcutree.gp_preinit_delay= [KNL] 4072 Set the number of jiffies to delay each step of 4073 RCU grace-period pre-initialization, that is, 4074 the propagation of recent CPU-hotplug changes up 4075 the rcu_node combining tree. 4076 4077 rcutree.use_softirq= [KNL] 4078 If set to zero, move all RCU_SOFTIRQ processing to 4079 per-CPU rcuc kthreads. Defaults to a non-zero 4080 value, meaning that RCU_SOFTIRQ is used by default. 4081 Specify rcutree.use_softirq=0 to use rcuc kthreads. 4082 4083 But note that CONFIG_PREEMPT_RT=y kernels disable 4084 this kernel boot parameter, forcibly setting it 4085 to zero. 4086 4087 rcutree.rcu_fanout_exact= [KNL] 4088 Disable autobalancing of the rcu_node combining 4089 tree. This is used by rcutorture, and might 4090 possibly be useful for architectures having high 4091 cache-to-cache transfer latencies. 4092 4093 rcutree.rcu_fanout_leaf= [KNL] 4094 Change the number of CPUs assigned to each 4095 leaf rcu_node structure. Useful for very 4096 large systems, which will choose the value 64, 4097 and for NUMA systems with large remote-access 4098 latencies, which will choose a value aligned 4099 with the appropriate hardware boundaries. 4100 4101 rcutree.rcu_min_cached_objs= [KNL] 4102 Minimum number of objects which are cached and 4103 maintained per one CPU. Object size is equal 4104 to PAGE_SIZE. The cache allows to reduce the 4105 pressure to page allocator, also it makes the 4106 whole algorithm to behave better in low memory 4107 condition. 4108 4109 rcutree.jiffies_till_first_fqs= [KNL] 4110 Set delay from grace-period initialization to 4111 first attempt to force quiescent states. 4112 Units are jiffies, minimum value is zero, 4113 and maximum value is HZ. 4114 4115 rcutree.jiffies_till_next_fqs= [KNL] 4116 Set delay between subsequent attempts to force 4117 quiescent states. Units are jiffies, minimum 4118 value is one, and maximum value is HZ. 4119 4120 rcutree.jiffies_till_sched_qs= [KNL] 4121 Set required age in jiffies for a 4122 given grace period before RCU starts 4123 soliciting quiescent-state help from 4124 rcu_note_context_switch() and cond_resched(). 4125 If not specified, the kernel will calculate 4126 a value based on the most recent settings 4127 of rcutree.jiffies_till_first_fqs 4128 and rcutree.jiffies_till_next_fqs. 4129 This calculated value may be viewed in 4130 rcutree.jiffies_to_sched_qs. Any attempt to set 4131 rcutree.jiffies_to_sched_qs will be cheerfully 4132 overwritten. 4133 4134 rcutree.kthread_prio= [KNL,BOOT] 4135 Set the SCHED_FIFO priority of the RCU per-CPU 4136 kthreads (rcuc/N). This value is also used for 4137 the priority of the RCU boost threads (rcub/N) 4138 and for the RCU grace-period kthreads (rcu_bh, 4139 rcu_preempt, and rcu_sched). If RCU_BOOST is 4140 set, valid values are 1-99 and the default is 1 4141 (the least-favored priority). Otherwise, when 4142 RCU_BOOST is not set, valid values are 0-99 and 4143 the default is zero (non-realtime operation). 4144 4145 rcutree.rcu_nocb_gp_stride= [KNL] 4146 Set the number of NOCB callback kthreads in 4147 each group, which defaults to the square root 4148 of the number of CPUs. Larger numbers reduce 4149 the wakeup overhead on the global grace-period 4150 kthread, but increases that same overhead on 4151 each group's NOCB grace-period kthread. 4152 4153 rcutree.qhimark= [KNL] 4154 Set threshold of queued RCU callbacks beyond which 4155 batch limiting is disabled. 4156 4157 rcutree.qlowmark= [KNL] 4158 Set threshold of queued RCU callbacks below which 4159 batch limiting is re-enabled. 4160 4161 rcutree.qovld= [KNL] 4162 Set threshold of queued RCU callbacks beyond which 4163 RCU's force-quiescent-state scan will aggressively 4164 enlist help from cond_resched() and sched IPIs to 4165 help CPUs more quickly reach quiescent states. 4166 Set to less than zero to make this be set based 4167 on rcutree.qhimark at boot time and to zero to 4168 disable more aggressive help enlistment. 4169 4170 rcutree.rcu_idle_gp_delay= [KNL] 4171 Set wakeup interval for idle CPUs that have 4172 RCU callbacks (RCU_FAST_NO_HZ=y). 4173 4174 rcutree.rcu_kick_kthreads= [KNL] 4175 Cause the grace-period kthread to get an extra 4176 wake_up() if it sleeps three times longer than 4177 it should at force-quiescent-state time. 4178 This wake_up() will be accompanied by a 4179 WARN_ONCE() splat and an ftrace_dump(). 4180 4181 rcutree.rcu_unlock_delay= [KNL] 4182 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels, 4183 this specifies an rcu_read_unlock()-time delay 4184 in microseconds. This defaults to zero. 4185 Larger delays increase the probability of 4186 catching RCU pointer leaks, that is, buggy use 4187 of RCU-protected pointers after the relevant 4188 rcu_read_unlock() has completed. 4189 4190 rcutree.sysrq_rcu= [KNL] 4191 Commandeer a sysrq key to dump out Tree RCU's 4192 rcu_node tree with an eye towards determining 4193 why a new grace period has not yet started. 4194 4195 rcuscale.gp_async= [KNL] 4196 Measure performance of asynchronous 4197 grace-period primitives such as call_rcu(). 4198 4199 rcuscale.gp_async_max= [KNL] 4200 Specify the maximum number of outstanding 4201 callbacks per writer thread. When a writer 4202 thread exceeds this limit, it invokes the 4203 corresponding flavor of rcu_barrier() to allow 4204 previously posted callbacks to drain. 4205 4206 rcuscale.gp_exp= [KNL] 4207 Measure performance of expedited synchronous 4208 grace-period primitives. 4209 4210 rcuscale.holdoff= [KNL] 4211 Set test-start holdoff period. The purpose of 4212 this parameter is to delay the start of the 4213 test until boot completes in order to avoid 4214 interference. 4215 4216 rcuscale.kfree_rcu_test= [KNL] 4217 Set to measure performance of kfree_rcu() flooding. 4218 4219 rcuscale.kfree_nthreads= [KNL] 4220 The number of threads running loops of kfree_rcu(). 4221 4222 rcuscale.kfree_alloc_num= [KNL] 4223 Number of allocations and frees done in an iteration. 4224 4225 rcuscale.kfree_loops= [KNL] 4226 Number of loops doing rcuscale.kfree_alloc_num number 4227 of allocations and frees. 4228 4229 rcuscale.nreaders= [KNL] 4230 Set number of RCU readers. The value -1 selects 4231 N, where N is the number of CPUs. A value 4232 "n" less than -1 selects N-n+1, where N is again 4233 the number of CPUs. For example, -2 selects N 4234 (the number of CPUs), -3 selects N+1, and so on. 4235 A value of "n" less than or equal to -N selects 4236 a single reader. 4237 4238 rcuscale.nwriters= [KNL] 4239 Set number of RCU writers. The values operate 4240 the same as for rcuscale.nreaders. 4241 N, where N is the number of CPUs 4242 4243 rcuscale.perf_type= [KNL] 4244 Specify the RCU implementation to test. 4245 4246 rcuscale.shutdown= [KNL] 4247 Shut the system down after performance tests 4248 complete. This is useful for hands-off automated 4249 testing. 4250 4251 rcuscale.verbose= [KNL] 4252 Enable additional printk() statements. 4253 4254 rcuscale.writer_holdoff= [KNL] 4255 Write-side holdoff between grace periods, 4256 in microseconds. The default of zero says 4257 no holdoff. 4258 4259 rcutorture.fqs_duration= [KNL] 4260 Set duration of force_quiescent_state bursts 4261 in microseconds. 4262 4263 rcutorture.fqs_holdoff= [KNL] 4264 Set holdoff time within force_quiescent_state bursts 4265 in microseconds. 4266 4267 rcutorture.fqs_stutter= [KNL] 4268 Set wait time between force_quiescent_state bursts 4269 in seconds. 4270 4271 rcutorture.fwd_progress= [KNL] 4272 Enable RCU grace-period forward-progress testing 4273 for the types of RCU supporting this notion. 4274 4275 rcutorture.fwd_progress_div= [KNL] 4276 Specify the fraction of a CPU-stall-warning 4277 period to do tight-loop forward-progress testing. 4278 4279 rcutorture.fwd_progress_holdoff= [KNL] 4280 Number of seconds to wait between successive 4281 forward-progress tests. 4282 4283 rcutorture.fwd_progress_need_resched= [KNL] 4284 Enclose cond_resched() calls within checks for 4285 need_resched() during tight-loop forward-progress 4286 testing. 4287 4288 rcutorture.gp_cond= [KNL] 4289 Use conditional/asynchronous update-side 4290 primitives, if available. 4291 4292 rcutorture.gp_exp= [KNL] 4293 Use expedited update-side primitives, if available. 4294 4295 rcutorture.gp_normal= [KNL] 4296 Use normal (non-expedited) asynchronous 4297 update-side primitives, if available. 4298 4299 rcutorture.gp_sync= [KNL] 4300 Use normal (non-expedited) synchronous 4301 update-side primitives, if available. If all 4302 of rcutorture.gp_cond=, rcutorture.gp_exp=, 4303 rcutorture.gp_normal=, and rcutorture.gp_sync= 4304 are zero, rcutorture acts as if is interpreted 4305 they are all non-zero. 4306 4307 rcutorture.irqreader= [KNL] 4308 Run RCU readers from irq handlers, or, more 4309 accurately, from a timer handler. Not all RCU 4310 flavors take kindly to this sort of thing. 4311 4312 rcutorture.leakpointer= [KNL] 4313 Leak an RCU-protected pointer out of the reader. 4314 This can of course result in splats, and is 4315 intended to test the ability of things like 4316 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect 4317 such leaks. 4318 4319 rcutorture.n_barrier_cbs= [KNL] 4320 Set callbacks/threads for rcu_barrier() testing. 4321 4322 rcutorture.nfakewriters= [KNL] 4323 Set number of concurrent RCU writers. These just 4324 stress RCU, they don't participate in the actual 4325 test, hence the "fake". 4326 4327 rcutorture.nocbs_nthreads= [KNL] 4328 Set number of RCU callback-offload togglers. 4329 Zero (the default) disables toggling. 4330 4331 rcutorture.nocbs_toggle= [KNL] 4332 Set the delay in milliseconds between successive 4333 callback-offload toggling attempts. 4334 4335 rcutorture.nreaders= [KNL] 4336 Set number of RCU readers. The value -1 selects 4337 N-1, where N is the number of CPUs. A value 4338 "n" less than -1 selects N-n-2, where N is again 4339 the number of CPUs. For example, -2 selects N 4340 (the number of CPUs), -3 selects N+1, and so on. 4341 4342 rcutorture.object_debug= [KNL] 4343 Enable debug-object double-call_rcu() testing. 4344 4345 rcutorture.onoff_holdoff= [KNL] 4346 Set time (s) after boot for CPU-hotplug testing. 4347 4348 rcutorture.onoff_interval= [KNL] 4349 Set time (jiffies) between CPU-hotplug operations, 4350 or zero to disable CPU-hotplug testing. 4351 4352 rcutorture.read_exit= [KNL] 4353 Set the number of read-then-exit kthreads used 4354 to test the interaction of RCU updaters and 4355 task-exit processing. 4356 4357 rcutorture.read_exit_burst= [KNL] 4358 The number of times in a given read-then-exit 4359 episode that a set of read-then-exit kthreads 4360 is spawned. 4361 4362 rcutorture.read_exit_delay= [KNL] 4363 The delay, in seconds, between successive 4364 read-then-exit testing episodes. 4365 4366 rcutorture.shuffle_interval= [KNL] 4367 Set task-shuffle interval (s). Shuffling tasks 4368 allows some CPUs to go into dyntick-idle mode 4369 during the rcutorture test. 4370 4371 rcutorture.shutdown_secs= [KNL] 4372 Set time (s) after boot system shutdown. This 4373 is useful for hands-off automated testing. 4374 4375 rcutorture.stall_cpu= [KNL] 4376 Duration of CPU stall (s) to test RCU CPU stall 4377 warnings, zero to disable. 4378 4379 rcutorture.stall_cpu_block= [KNL] 4380 Sleep while stalling if set. This will result 4381 in warnings from preemptible RCU in addition 4382 to any other stall-related activity. 4383 4384 rcutorture.stall_cpu_holdoff= [KNL] 4385 Time to wait (s) after boot before inducing stall. 4386 4387 rcutorture.stall_cpu_irqsoff= [KNL] 4388 Disable interrupts while stalling if set. 4389 4390 rcutorture.stall_gp_kthread= [KNL] 4391 Duration (s) of forced sleep within RCU 4392 grace-period kthread to test RCU CPU stall 4393 warnings, zero to disable. If both stall_cpu 4394 and stall_gp_kthread are specified, the 4395 kthread is starved first, then the CPU. 4396 4397 rcutorture.stat_interval= [KNL] 4398 Time (s) between statistics printk()s. 4399 4400 rcutorture.stutter= [KNL] 4401 Time (s) to stutter testing, for example, specifying 4402 five seconds causes the test to run for five seconds, 4403 wait for five seconds, and so on. This tests RCU's 4404 ability to transition abruptly to and from idle. 4405 4406 rcutorture.test_boost= [KNL] 4407 Test RCU priority boosting? 0=no, 1=maybe, 2=yes. 4408 "Maybe" means test if the RCU implementation 4409 under test support RCU priority boosting. 4410 4411 rcutorture.test_boost_duration= [KNL] 4412 Duration (s) of each individual boost test. 4413 4414 rcutorture.test_boost_interval= [KNL] 4415 Interval (s) between each boost test. 4416 4417 rcutorture.test_no_idle_hz= [KNL] 4418 Test RCU's dyntick-idle handling. See also the 4419 rcutorture.shuffle_interval parameter. 4420 4421 rcutorture.torture_type= [KNL] 4422 Specify the RCU implementation to test. 4423 4424 rcutorture.verbose= [KNL] 4425 Enable additional printk() statements. 4426 4427 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL] 4428 Dump ftrace buffer after reporting RCU CPU 4429 stall warning. 4430 4431 rcupdate.rcu_cpu_stall_suppress= [KNL] 4432 Suppress RCU CPU stall warning messages. 4433 4434 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL] 4435 Suppress RCU CPU stall warning messages and 4436 rcutorture writer stall warnings that occur 4437 during early boot, that is, during the time 4438 before the init task is spawned. 4439 4440 rcupdate.rcu_cpu_stall_timeout= [KNL] 4441 Set timeout for RCU CPU stall warning messages. 4442 4443 rcupdate.rcu_expedited= [KNL] 4444 Use expedited grace-period primitives, for 4445 example, synchronize_rcu_expedited() instead 4446 of synchronize_rcu(). This reduces latency, 4447 but can increase CPU utilization, degrade 4448 real-time latency, and degrade energy efficiency. 4449 No effect on CONFIG_TINY_RCU kernels. 4450 4451 rcupdate.rcu_normal= [KNL] 4452 Use only normal grace-period primitives, 4453 for example, synchronize_rcu() instead of 4454 synchronize_rcu_expedited(). This improves 4455 real-time latency, CPU utilization, and 4456 energy efficiency, but can expose users to 4457 increased grace-period latency. This parameter 4458 overrides rcupdate.rcu_expedited. No effect on 4459 CONFIG_TINY_RCU kernels. 4460 4461 rcupdate.rcu_normal_after_boot= [KNL] 4462 Once boot has completed (that is, after 4463 rcu_end_inkernel_boot() has been invoked), use 4464 only normal grace-period primitives. No effect 4465 on CONFIG_TINY_RCU kernels. 4466 4467 But note that CONFIG_PREEMPT_RT=y kernels enables 4468 this kernel boot parameter, forcibly setting 4469 it to the value one, that is, converting any 4470 post-boot attempt at an expedited RCU grace 4471 period to instead use normal non-expedited 4472 grace-period processing. 4473 4474 rcupdate.rcu_task_ipi_delay= [KNL] 4475 Set time in jiffies during which RCU tasks will 4476 avoid sending IPIs, starting with the beginning 4477 of a given grace period. Setting a large 4478 number avoids disturbing real-time workloads, 4479 but lengthens grace periods. 4480 4481 rcupdate.rcu_task_stall_timeout= [KNL] 4482 Set timeout in jiffies for RCU task stall warning 4483 messages. Disable with a value less than or equal 4484 to zero. 4485 4486 rcupdate.rcu_self_test= [KNL] 4487 Run the RCU early boot self tests 4488 4489 rdinit= [KNL] 4490 Format: <full_path> 4491 Run specified binary instead of /init from the ramdisk, 4492 used for early userspace startup. See initrd. 4493 4494 rdrand= [X86] 4495 force - Override the decision by the kernel to hide the 4496 advertisement of RDRAND support (this affects 4497 certain AMD processors because of buggy BIOS 4498 support, specifically around the suspend/resume 4499 path). 4500 4501 rdt= [HW,X86,RDT] 4502 Turn on/off individual RDT features. List is: 4503 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp, 4504 mba. 4505 E.g. to turn on cmt and turn off mba use: 4506 rdt=cmt,!mba 4507 4508 reboot= [KNL] 4509 Format (x86 or x86_64): 4510 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] \ 4511 [[,]s[mp]#### \ 4512 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \ 4513 [[,]f[orce] 4514 Where reboot_mode is one of warm (soft) or cold (hard) or gpio 4515 (prefix with 'panic_' to set mode for panic 4516 reboot only), 4517 reboot_type is one of bios, acpi, kbd, triple, efi, or pci, 4518 reboot_force is either force or not specified, 4519 reboot_cpu is s[mp]#### with #### being the processor 4520 to be used for rebooting. 4521 4522 refscale.holdoff= [KNL] 4523 Set test-start holdoff period. The purpose of 4524 this parameter is to delay the start of the 4525 test until boot completes in order to avoid 4526 interference. 4527 4528 refscale.loops= [KNL] 4529 Set the number of loops over the synchronization 4530 primitive under test. Increasing this number 4531 reduces noise due to loop start/end overhead, 4532 but the default has already reduced the per-pass 4533 noise to a handful of picoseconds on ca. 2020 4534 x86 laptops. 4535 4536 refscale.nreaders= [KNL] 4537 Set number of readers. The default value of -1 4538 selects N, where N is roughly 75% of the number 4539 of CPUs. A value of zero is an interesting choice. 4540 4541 refscale.nruns= [KNL] 4542 Set number of runs, each of which is dumped onto 4543 the console log. 4544 4545 refscale.readdelay= [KNL] 4546 Set the read-side critical-section duration, 4547 measured in microseconds. 4548 4549 refscale.scale_type= [KNL] 4550 Specify the read-protection implementation to test. 4551 4552 refscale.shutdown= [KNL] 4553 Shut down the system at the end of the performance 4554 test. This defaults to 1 (shut it down) when 4555 refscale is built into the kernel and to 0 (leave 4556 it running) when refscale is built as a module. 4557 4558 refscale.verbose= [KNL] 4559 Enable additional printk() statements. 4560 4561 refscale.verbose_batched= [KNL] 4562 Batch the additional printk() statements. If zero 4563 (the default) or negative, print everything. Otherwise, 4564 print every Nth verbose statement, where N is the value 4565 specified. 4566 4567 relax_domain_level= 4568 [KNL, SMP] Set scheduler's default relax_domain_level. 4569 See Documentation/admin-guide/cgroup-v1/cpusets.rst. 4570 4571 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory 4572 Format: <base1>,<size1>[,<base2>,<size2>,...] 4573 Reserve I/O ports or memory so the kernel won't use 4574 them. If <base> is less than 0x10000, the region 4575 is assumed to be I/O ports; otherwise it is memory. 4576 4577 reservetop= [X86-32] 4578 Format: nn[KMG] 4579 Reserves a hole at the top of the kernel virtual 4580 address space. 4581 4582 reservelow= [X86] 4583 Format: nn[K] 4584 Set the amount of memory to reserve for BIOS at 4585 the bottom of the address space. 4586 4587 reset_devices [KNL] Force drivers to reset the underlying device 4588 during initialization. 4589 4590 resume= [SWSUSP] 4591 Specify the partition device for software suspend 4592 Format: 4593 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>} 4594 4595 resume_offset= [SWSUSP] 4596 Specify the offset from the beginning of the partition 4597 given by "resume=" at which the swap header is located, 4598 in <PAGE_SIZE> units (needed only for swap files). 4599 See Documentation/power/swsusp-and-swap-files.rst 4600 4601 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to 4602 read the resume files 4603 4604 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up. 4605 Useful for devices that are detected asynchronously 4606 (e.g. USB and MMC devices). 4607 4608 hibernate= [HIBERNATION] 4609 noresume Don't check if there's a hibernation image 4610 present during boot. 4611 nocompress Don't compress/decompress hibernation images. 4612 no Disable hibernation and resume. 4613 protect_image Turn on image protection during restoration 4614 (that will set all pages holding image data 4615 during restoration read-only). 4616 4617 retain_initrd [RAM] Keep initrd memory after extraction 4618 4619 rfkill.default_state= 4620 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm, 4621 etc. communication is blocked by default. 4622 1 Unblocked. 4623 4624 rfkill.master_switch_mode= 4625 0 The "airplane mode" button does nothing. 4626 1 The "airplane mode" button toggles between everything 4627 blocked and the previous configuration. 4628 2 The "airplane mode" button toggles between everything 4629 blocked and everything unblocked. 4630 4631 rhash_entries= [KNL,NET] 4632 Set number of hash buckets for route cache 4633 4634 ring3mwait=disable 4635 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported 4636 CPUs. 4637 4638 ro [KNL] Mount root device read-only on boot 4639 4640 rodata= [KNL] 4641 on Mark read-only kernel memory as read-only (default). 4642 off Leave read-only kernel memory writable for debugging. 4643 4644 rockchip.usb_uart 4645 Enable the uart passthrough on the designated usb port 4646 on Rockchip SoCs. When active, the signals of the 4647 debug-uart get routed to the D+ and D- pins of the usb 4648 port and the regular usb controller gets disabled. 4649 4650 root= [KNL] Root filesystem 4651 See name_to_dev_t comment in init/do_mounts.c. 4652 4653 rootdelay= [KNL] Delay (in seconds) to pause before attempting to 4654 mount the root filesystem 4655 4656 rootflags= [KNL] Set root filesystem mount option string 4657 4658 rootfstype= [KNL] Set root filesystem type 4659 4660 rootwait [KNL] Wait (indefinitely) for root device to show up. 4661 Useful for devices that are detected asynchronously 4662 (e.g. USB and MMC devices). 4663 4664 rproc_mem=nn[KMG][@address] 4665 [KNL,ARM,CMA] Remoteproc physical memory block. 4666 Memory area to be used by remote processor image, 4667 managed by CMA. 4668 4669 rw [KNL] Mount root device read-write on boot 4670 4671 S [KNL] Run init in single mode 4672 4673 s390_iommu= [HW,S390] 4674 Set s390 IOTLB flushing mode 4675 strict 4676 With strict flushing every unmap operation will result in 4677 an IOTLB flush. Default is lazy flushing before reuse, 4678 which is faster. 4679 4680 sa1100ir [NET] 4681 See drivers/net/irda/sa1100_ir.c. 4682 4683 sbni= [NET] Granch SBNI12 leased line adapter 4684 4685 sched_debug [KNL] Enables verbose scheduler debug messages. 4686 4687 schedstats= [KNL,X86] Enable or disable scheduled statistics. 4688 Allowed values are enable and disable. This feature 4689 incurs a small amount of overhead in the scheduler 4690 but is useful for debugging and performance tuning. 4691 4692 sched_thermal_decay_shift= 4693 [KNL, SMP] Set a decay shift for scheduler thermal 4694 pressure signal. Thermal pressure signal follows the 4695 default decay period of other scheduler pelt 4696 signals(usually 32 ms but configurable). Setting 4697 sched_thermal_decay_shift will left shift the decay 4698 period for the thermal pressure signal by the shift 4699 value. 4700 i.e. with the default pelt decay period of 32 ms 4701 sched_thermal_decay_shift thermal pressure decay pr 4702 1 64 ms 4703 2 128 ms 4704 and so on. 4705 Format: integer between 0 and 10 4706 Default is 0. 4707 4708 scftorture.holdoff= [KNL] 4709 Number of seconds to hold off before starting 4710 test. Defaults to zero for module insertion and 4711 to 10 seconds for built-in smp_call_function() 4712 tests. 4713 4714 scftorture.longwait= [KNL] 4715 Request ridiculously long waits randomly selected 4716 up to the chosen limit in seconds. Zero (the 4717 default) disables this feature. Please note 4718 that requesting even small non-zero numbers of 4719 seconds can result in RCU CPU stall warnings, 4720 softlockup complaints, and so on. 4721 4722 scftorture.nthreads= [KNL] 4723 Number of kthreads to spawn to invoke the 4724 smp_call_function() family of functions. 4725 The default of -1 specifies a number of kthreads 4726 equal to the number of CPUs. 4727 4728 scftorture.onoff_holdoff= [KNL] 4729 Number seconds to wait after the start of the 4730 test before initiating CPU-hotplug operations. 4731 4732 scftorture.onoff_interval= [KNL] 4733 Number seconds to wait between successive 4734 CPU-hotplug operations. Specifying zero (which 4735 is the default) disables CPU-hotplug operations. 4736 4737 scftorture.shutdown_secs= [KNL] 4738 The number of seconds following the start of the 4739 test after which to shut down the system. The 4740 default of zero avoids shutting down the system. 4741 Non-zero values are useful for automated tests. 4742 4743 scftorture.stat_interval= [KNL] 4744 The number of seconds between outputting the 4745 current test statistics to the console. A value 4746 of zero disables statistics output. 4747 4748 scftorture.stutter_cpus= [KNL] 4749 The number of jiffies to wait between each change 4750 to the set of CPUs under test. 4751 4752 scftorture.use_cpus_read_lock= [KNL] 4753 Use use_cpus_read_lock() instead of the default 4754 preempt_disable() to disable CPU hotplug 4755 while invoking one of the smp_call_function*() 4756 functions. 4757 4758 scftorture.verbose= [KNL] 4759 Enable additional printk() statements. 4760 4761 scftorture.weight_single= [KNL] 4762 The probability weighting to use for the 4763 smp_call_function_single() function with a zero 4764 "wait" parameter. A value of -1 selects the 4765 default if all other weights are -1. However, 4766 if at least one weight has some other value, a 4767 value of -1 will instead select a weight of zero. 4768 4769 scftorture.weight_single_wait= [KNL] 4770 The probability weighting to use for the 4771 smp_call_function_single() function with a 4772 non-zero "wait" parameter. See weight_single. 4773 4774 scftorture.weight_many= [KNL] 4775 The probability weighting to use for the 4776 smp_call_function_many() function with a zero 4777 "wait" parameter. See weight_single. 4778 Note well that setting a high probability for 4779 this weighting can place serious IPI load 4780 on the system. 4781 4782 scftorture.weight_many_wait= [KNL] 4783 The probability weighting to use for the 4784 smp_call_function_many() function with a 4785 non-zero "wait" parameter. See weight_single 4786 and weight_many. 4787 4788 scftorture.weight_all= [KNL] 4789 The probability weighting to use for the 4790 smp_call_function_all() function with a zero 4791 "wait" parameter. See weight_single and 4792 weight_many. 4793 4794 scftorture.weight_all_wait= [KNL] 4795 The probability weighting to use for the 4796 smp_call_function_all() function with a 4797 non-zero "wait" parameter. See weight_single 4798 and weight_many. 4799 4800 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate 4801 xtime_lock contention on larger systems, and/or RCU lock 4802 contention on all systems with CONFIG_MAXSMP set. 4803 Format: { "0" | "1" } 4804 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1" 4805 1 -- enable. 4806 Note: increases power consumption, thus should only be 4807 enabled if running jitter sensitive (HPC/RT) workloads. 4808 4809 security= [SECURITY] Choose a legacy "major" security module to 4810 enable at boot. This has been deprecated by the 4811 "lsm=" parameter. 4812 4813 selinux= [SELINUX] Disable or enable SELinux at boot time. 4814 Format: { "0" | "1" } 4815 See security/selinux/Kconfig help text. 4816 0 -- disable. 4817 1 -- enable. 4818 Default value is 1. 4819 4820 apparmor= [APPARMOR] Disable or enable AppArmor at boot time 4821 Format: { "0" | "1" } 4822 See security/apparmor/Kconfig help text 4823 0 -- disable. 4824 1 -- enable. 4825 Default value is set via kernel config option. 4826 4827 serialnumber [BUGS=X86-32] 4828 4829 shapers= [NET] 4830 Maximal number of shapers. 4831 4832 simeth= [IA-64] 4833 simscsi= 4834 4835 slram= [HW,MTD] 4836 4837 slab_nomerge [MM] 4838 Disable merging of slabs with similar size. May be 4839 necessary if there is some reason to distinguish 4840 allocs to different slabs, especially in hardened 4841 environments where the risk of heap overflows and 4842 layout control by attackers can usually be 4843 frustrated by disabling merging. This will reduce 4844 most of the exposure of a heap attack to a single 4845 cache (risks via metadata attacks are mostly 4846 unchanged). Debug options disable merging on their 4847 own. 4848 For more information see Documentation/vm/slub.rst. 4849 4850 slab_max_order= [MM, SLAB] 4851 Determines the maximum allowed order for slabs. 4852 A high setting may cause OOMs due to memory 4853 fragmentation. Defaults to 1 for systems with 4854 more than 32MB of RAM, 0 otherwise. 4855 4856 slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB] 4857 Enabling slub_debug allows one to determine the 4858 culprit if slab objects become corrupted. Enabling 4859 slub_debug can create guard zones around objects and 4860 may poison objects when not in use. Also tracks the 4861 last alloc / free. For more information see 4862 Documentation/vm/slub.rst. 4863 4864 slub_memcg_sysfs= [MM, SLUB] 4865 Determines whether to enable sysfs directories for 4866 memory cgroup sub-caches. 1 to enable, 0 to disable. 4867 The default is determined by CONFIG_SLUB_MEMCG_SYSFS_ON. 4868 Enabling this can lead to a very high number of debug 4869 directories and files being created under 4870 /sys/kernel/slub. 4871 4872 slub_max_order= [MM, SLUB] 4873 Determines the maximum allowed order for slabs. 4874 A high setting may cause OOMs due to memory 4875 fragmentation. For more information see 4876 Documentation/vm/slub.rst. 4877 4878 slub_min_objects= [MM, SLUB] 4879 The minimum number of objects per slab. SLUB will 4880 increase the slab order up to slub_max_order to 4881 generate a sufficiently large slab able to contain 4882 the number of objects indicated. The higher the number 4883 of objects the smaller the overhead of tracking slabs 4884 and the less frequently locks need to be acquired. 4885 For more information see Documentation/vm/slub.rst. 4886 4887 slub_min_order= [MM, SLUB] 4888 Determines the minimum page order for slabs. Must be 4889 lower than slub_max_order. 4890 For more information see Documentation/vm/slub.rst. 4891 4892 slub_nomerge [MM, SLUB] 4893 Same with slab_nomerge. This is supported for legacy. 4894 See slab_nomerge for more information. 4895 4896 smart2= [HW] 4897 Format: <io1>[,<io2>[,...,<io8>]] 4898 4899 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices 4900 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port 4901 smsc-ircc2.ircc_sir= [HW] SIR base I/O port 4902 smsc-ircc2.ircc_fir= [HW] FIR base I/O port 4903 smsc-ircc2.ircc_irq= [HW] IRQ line 4904 smsc-ircc2.ircc_dma= [HW] DMA channel 4905 smsc-ircc2.ircc_transceiver= [HW] Transceiver type: 4906 0: Toshiba Satellite 1800 (GP data pin select) 4907 1: Fast pin select (default) 4908 2: ATC IRMode 4909 4910 smt [KNL,S390] Set the maximum number of threads (logical 4911 CPUs) to use per physical CPU on systems capable of 4912 symmetric multithreading (SMT). Will be capped to the 4913 actual hardware limit. 4914 Format: <integer> 4915 Default: -1 (no limit) 4916 4917 softlockup_panic= 4918 [KNL] Should the soft-lockup detector generate panics. 4919 Format: 0 | 1 4920 4921 A value of 1 instructs the soft-lockup detector 4922 to panic the machine when a soft-lockup occurs. It is 4923 also controlled by the kernel.softlockup_panic sysctl 4924 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the 4925 respective build-time switch to that functionality. 4926 4927 softlockup_all_cpu_backtrace= 4928 [KNL] Should the soft-lockup detector generate 4929 backtraces on all cpus. 4930 Format: 0 | 1 4931 4932 sonypi.*= [HW] Sony Programmable I/O Control Device driver 4933 See Documentation/admin-guide/laptops/sonypi.rst 4934 4935 spectre_v2= [X86] Control mitigation of Spectre variant 2 4936 (indirect branch speculation) vulnerability. 4937 The default operation protects the kernel from 4938 user space attacks. 4939 4940 on - unconditionally enable, implies 4941 spectre_v2_user=on 4942 off - unconditionally disable, implies 4943 spectre_v2_user=off 4944 auto - kernel detects whether your CPU model is 4945 vulnerable 4946 4947 Selecting 'on' will, and 'auto' may, choose a 4948 mitigation method at run time according to the 4949 CPU, the available microcode, the setting of the 4950 CONFIG_RETPOLINE configuration option, and the 4951 compiler with which the kernel was built. 4952 4953 Selecting 'on' will also enable the mitigation 4954 against user space to user space task attacks. 4955 4956 Selecting 'off' will disable both the kernel and 4957 the user space protections. 4958 4959 Specific mitigations can also be selected manually: 4960 4961 retpoline - replace indirect branches 4962 retpoline,generic - google's original retpoline 4963 retpoline,amd - AMD-specific minimal thunk 4964 4965 Not specifying this option is equivalent to 4966 spectre_v2=auto. 4967 4968 spectre_v2_user= 4969 [X86] Control mitigation of Spectre variant 2 4970 (indirect branch speculation) vulnerability between 4971 user space tasks 4972 4973 on - Unconditionally enable mitigations. Is 4974 enforced by spectre_v2=on 4975 4976 off - Unconditionally disable mitigations. Is 4977 enforced by spectre_v2=off 4978 4979 prctl - Indirect branch speculation is enabled, 4980 but mitigation can be enabled via prctl 4981 per thread. The mitigation control state 4982 is inherited on fork. 4983 4984 prctl,ibpb 4985 - Like "prctl" above, but only STIBP is 4986 controlled per thread. IBPB is issued 4987 always when switching between different user 4988 space processes. 4989 4990 seccomp 4991 - Same as "prctl" above, but all seccomp 4992 threads will enable the mitigation unless 4993 they explicitly opt out. 4994 4995 seccomp,ibpb 4996 - Like "seccomp" above, but only STIBP is 4997 controlled per thread. IBPB is issued 4998 always when switching between different 4999 user space processes. 5000 5001 auto - Kernel selects the mitigation depending on 5002 the available CPU features and vulnerability. 5003 5004 Default mitigation: 5005 If CONFIG_SECCOMP=y then "seccomp", otherwise "prctl" 5006 5007 Not specifying this option is equivalent to 5008 spectre_v2_user=auto. 5009 5010 spec_store_bypass_disable= 5011 [HW] Control Speculative Store Bypass (SSB) Disable mitigation 5012 (Speculative Store Bypass vulnerability) 5013 5014 Certain CPUs are vulnerable to an exploit against a 5015 a common industry wide performance optimization known 5016 as "Speculative Store Bypass" in which recent stores 5017 to the same memory location may not be observed by 5018 later loads during speculative execution. The idea 5019 is that such stores are unlikely and that they can 5020 be detected prior to instruction retirement at the 5021 end of a particular speculation execution window. 5022 5023 In vulnerable processors, the speculatively forwarded 5024 store can be used in a cache side channel attack, for 5025 example to read memory to which the attacker does not 5026 directly have access (e.g. inside sandboxed code). 5027 5028 This parameter controls whether the Speculative Store 5029 Bypass optimization is used. 5030 5031 On x86 the options are: 5032 5033 on - Unconditionally disable Speculative Store Bypass 5034 off - Unconditionally enable Speculative Store Bypass 5035 auto - Kernel detects whether the CPU model contains an 5036 implementation of Speculative Store Bypass and 5037 picks the most appropriate mitigation. If the 5038 CPU is not vulnerable, "off" is selected. If the 5039 CPU is vulnerable the default mitigation is 5040 architecture and Kconfig dependent. See below. 5041 prctl - Control Speculative Store Bypass per thread 5042 via prctl. Speculative Store Bypass is enabled 5043 for a process by default. The state of the control 5044 is inherited on fork. 5045 seccomp - Same as "prctl" above, but all seccomp threads 5046 will disable SSB unless they explicitly opt out. 5047 5048 Default mitigations: 5049 X86: If CONFIG_SECCOMP=y "seccomp", otherwise "prctl" 5050 5051 On powerpc the options are: 5052 5053 on,auto - On Power8 and Power9 insert a store-forwarding 5054 barrier on kernel entry and exit. On Power7 5055 perform a software flush on kernel entry and 5056 exit. 5057 off - No action. 5058 5059 Not specifying this option is equivalent to 5060 spec_store_bypass_disable=auto. 5061 5062 spia_io_base= [HW,MTD] 5063 spia_fio_base= 5064 spia_pedr= 5065 spia_peddr= 5066 5067 split_lock_detect= 5068 [X86] Enable split lock detection 5069 5070 When enabled (and if hardware support is present), atomic 5071 instructions that access data across cache line 5072 boundaries will result in an alignment check exception. 5073 5074 off - not enabled 5075 5076 warn - the kernel will emit rate limited warnings 5077 about applications triggering the #AC 5078 exception. This mode is the default on CPUs 5079 that supports split lock detection. 5080 5081 fatal - the kernel will send SIGBUS to applications 5082 that trigger the #AC exception. 5083 5084 If an #AC exception is hit in the kernel or in 5085 firmware (i.e. not while executing in user mode) 5086 the kernel will oops in either "warn" or "fatal" 5087 mode. 5088 5089 srbds= [X86,INTEL] 5090 Control the Special Register Buffer Data Sampling 5091 (SRBDS) mitigation. 5092 5093 Certain CPUs are vulnerable to an MDS-like 5094 exploit which can leak bits from the random 5095 number generator. 5096 5097 By default, this issue is mitigated by 5098 microcode. However, the microcode fix can cause 5099 the RDRAND and RDSEED instructions to become 5100 much slower. Among other effects, this will 5101 result in reduced throughput from /dev/urandom. 5102 5103 The microcode mitigation can be disabled with 5104 the following option: 5105 5106 off: Disable mitigation and remove 5107 performance impact to RDRAND and RDSEED 5108 5109 srcutree.counter_wrap_check [KNL] 5110 Specifies how frequently to check for 5111 grace-period sequence counter wrap for the 5112 srcu_data structure's ->srcu_gp_seq_needed field. 5113 The greater the number of bits set in this kernel 5114 parameter, the less frequently counter wrap will 5115 be checked for. Note that the bottom two bits 5116 are ignored. 5117 5118 srcutree.exp_holdoff [KNL] 5119 Specifies how many nanoseconds must elapse 5120 since the end of the last SRCU grace period for 5121 a given srcu_struct until the next normal SRCU 5122 grace period will be considered for automatic 5123 expediting. Set to zero to disable automatic 5124 expediting. 5125 5126 ssbd= [ARM64,HW] 5127 Speculative Store Bypass Disable control 5128 5129 On CPUs that are vulnerable to the Speculative 5130 Store Bypass vulnerability and offer a 5131 firmware based mitigation, this parameter 5132 indicates how the mitigation should be used: 5133 5134 force-on: Unconditionally enable mitigation for 5135 for both kernel and userspace 5136 force-off: Unconditionally disable mitigation for 5137 for both kernel and userspace 5138 kernel: Always enable mitigation in the 5139 kernel, and offer a prctl interface 5140 to allow userspace to register its 5141 interest in being mitigated too. 5142 5143 stack_guard_gap= [MM] 5144 override the default stack gap protection. The value 5145 is in page units and it defines how many pages prior 5146 to (for stacks growing down) resp. after (for stacks 5147 growing up) the main stack are reserved for no other 5148 mapping. Default value is 256 pages. 5149 5150 stacktrace [FTRACE] 5151 Enabled the stack tracer on boot up. 5152 5153 stacktrace_filter=[function-list] 5154 [FTRACE] Limit the functions that the stack tracer 5155 will trace at boot up. function-list is a comma-separated 5156 list of functions. This list can be changed at run 5157 time by the stack_trace_filter file in the debugfs 5158 tracing directory. Note, this enables stack tracing 5159 and the stacktrace above is not needed. 5160 5161 sti= [PARISC,HW] 5162 Format: <num> 5163 Set the STI (builtin display/keyboard on the HP-PARISC 5164 machines) console (graphic card) which should be used 5165 as the initial boot-console. 5166 See also comment in drivers/video/console/sticore.c. 5167 5168 sti_font= [HW] 5169 See comment in drivers/video/console/sticore.c. 5170 5171 stifb= [HW] 5172 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]] 5173 5174 sunrpc.min_resvport= 5175 sunrpc.max_resvport= 5176 [NFS,SUNRPC] 5177 SunRPC servers often require that client requests 5178 originate from a privileged port (i.e. a port in the 5179 range 0 < portnr < 1024). 5180 An administrator who wishes to reserve some of these 5181 ports for other uses may adjust the range that the 5182 kernel's sunrpc client considers to be privileged 5183 using these two parameters to set the minimum and 5184 maximum port values. 5185 5186 sunrpc.svc_rpc_per_connection_limit= 5187 [NFS,SUNRPC] 5188 Limit the number of requests that the server will 5189 process in parallel from a single connection. 5190 The default value is 0 (no limit). 5191 5192 sunrpc.pool_mode= 5193 [NFS] 5194 Control how the NFS server code allocates CPUs to 5195 service thread pools. Depending on how many NICs 5196 you have and where their interrupts are bound, this 5197 option will affect which CPUs will do NFS serving. 5198 Note: this parameter cannot be changed while the 5199 NFS server is running. 5200 5201 auto the server chooses an appropriate mode 5202 automatically using heuristics 5203 global a single global pool contains all CPUs 5204 percpu one pool for each CPU 5205 pernode one pool for each NUMA node (equivalent 5206 to global on non-NUMA machines) 5207 5208 sunrpc.tcp_slot_table_entries= 5209 sunrpc.udp_slot_table_entries= 5210 [NFS,SUNRPC] 5211 Sets the upper limit on the number of simultaneous 5212 RPC calls that can be sent from the client to a 5213 server. Increasing these values may allow you to 5214 improve throughput, but will also increase the 5215 amount of memory reserved for use by the client. 5216 5217 suspend.pm_test_delay= 5218 [SUSPEND] 5219 Sets the number of seconds to remain in a suspend test 5220 mode before resuming the system (see 5221 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG 5222 is set. Default value is 5. 5223 5224 svm= [PPC] 5225 Format: { on | off | y | n | 1 | 0 } 5226 This parameter controls use of the Protected 5227 Execution Facility on pSeries. 5228 5229 swapaccount=[0|1] 5230 [KNL] Enable accounting of swap in memory resource 5231 controller if no parameter or 1 is given or disable 5232 it if 0 is given (See Documentation/admin-guide/cgroup-v1/memory.rst) 5233 5234 swiotlb= [ARM,IA-64,PPC,MIPS,X86] 5235 Format: { <int> | force | noforce } 5236 <int> -- Number of I/O TLB slabs 5237 force -- force using of bounce buffers even if they 5238 wouldn't be automatically used by the kernel 5239 noforce -- Never use bounce buffers (for debugging) 5240 5241 switches= [HW,M68k] 5242 5243 sysctl.*= [KNL] 5244 Set a sysctl parameter, right before loading the init 5245 process, as if the value was written to the respective 5246 /proc/sys/... file. Both '.' and '/' are recognized as 5247 separators. Unrecognized parameters and invalid values 5248 are reported in the kernel log. Sysctls registered 5249 later by a loaded module cannot be set this way. 5250 Example: sysctl.vm.swappiness=40 5251 5252 sysfs.deprecated=0|1 [KNL] 5253 Enable/disable old style sysfs layout for old udev 5254 on older distributions. When this option is enabled 5255 very new udev will not work anymore. When this option 5256 is disabled (or CONFIG_SYSFS_DEPRECATED not compiled) 5257 in older udev will not work anymore. 5258 Default depends on CONFIG_SYSFS_DEPRECATED_V2 set in 5259 the kernel configuration. 5260 5261 sysrq_always_enabled 5262 [KNL] 5263 Ignore sysrq setting - this boot parameter will 5264 neutralize any effect of /proc/sys/kernel/sysrq. 5265 Useful for debugging. 5266 5267 tcpmhash_entries= [KNL,NET] 5268 Set the number of tcp_metrics_hash slots. 5269 Default value is 8192 or 16384 depending on total 5270 ram pages. This is used to specify the TCP metrics 5271 cache size. See Documentation/networking/ip-sysctl.rst 5272 "tcp_no_metrics_save" section for more details. 5273 5274 tdfx= [HW,DRM] 5275 5276 test_suspend= [SUSPEND][,N] 5277 Specify "mem" (for Suspend-to-RAM) or "standby" (for 5278 standby suspend) or "freeze" (for suspend type freeze) 5279 as the system sleep state during system startup with 5280 the optional capability to repeat N number of times. 5281 The system is woken from this state using a 5282 wakeup-capable RTC alarm. 5283 5284 thash_entries= [KNL,NET] 5285 Set number of hash buckets for TCP connection 5286 5287 thermal.act= [HW,ACPI] 5288 -1: disable all active trip points in all thermal zones 5289 <degrees C>: override all lowest active trip points 5290 5291 thermal.crt= [HW,ACPI] 5292 -1: disable all critical trip points in all thermal zones 5293 <degrees C>: override all critical trip points 5294 5295 thermal.nocrt= [HW,ACPI] 5296 Set to disable actions on ACPI thermal zone 5297 critical and hot trip points. 5298 5299 thermal.off= [HW,ACPI] 5300 1: disable ACPI thermal control 5301 5302 thermal.psv= [HW,ACPI] 5303 -1: disable all passive trip points 5304 <degrees C>: override all passive trip points to this 5305 value 5306 5307 thermal.tzp= [HW,ACPI] 5308 Specify global default ACPI thermal zone polling rate 5309 <deci-seconds>: poll all this frequency 5310 0: no polling (default) 5311 5312 threadirqs [KNL] 5313 Force threading of all interrupt handlers except those 5314 marked explicitly IRQF_NO_THREAD. 5315 5316 topology= [S390] 5317 Format: {off | on} 5318 Specify if the kernel should make use of the cpu 5319 topology information if the hardware supports this. 5320 The scheduler will make use of this information and 5321 e.g. base its process migration decisions on it. 5322 Default is on. 5323 5324 topology_updates= [KNL, PPC, NUMA] 5325 Format: {off} 5326 Specify if the kernel should ignore (off) 5327 topology updates sent by the hypervisor to this 5328 LPAR. 5329 5330 torture.disable_onoff_at_boot= [KNL] 5331 Prevent the CPU-hotplug component of torturing 5332 until after init has spawned. 5333 5334 torture.ftrace_dump_at_shutdown= [KNL] 5335 Dump the ftrace buffer at torture-test shutdown, 5336 even if there were no errors. This can be a 5337 very costly operation when many torture tests 5338 are running concurrently, especially on systems 5339 with rotating-rust storage. 5340 5341 torture.verbose_sleep_frequency= [KNL] 5342 Specifies how many verbose printk()s should be 5343 emitted between each sleep. The default of zero 5344 disables verbose-printk() sleeping. 5345 5346 torture.verbose_sleep_duration= [KNL] 5347 Duration of each verbose-printk() sleep in jiffies. 5348 5349 tp720= [HW,PS2] 5350 5351 tpm_suspend_pcr=[HW,TPM] 5352 Format: integer pcr id 5353 Specify that at suspend time, the tpm driver 5354 should extend the specified pcr with zeros, 5355 as a workaround for some chips which fail to 5356 flush the last written pcr on TPM_SaveState. 5357 This will guarantee that all the other pcrs 5358 are saved. 5359 5360 trace_buf_size=nn[KMG] 5361 [FTRACE] will set tracing buffer size on each cpu. 5362 5363 trace_event=[event-list] 5364 [FTRACE] Set and start specified trace events in order 5365 to facilitate early boot debugging. The event-list is a 5366 comma-separated list of trace events to enable. See 5367 also Documentation/trace/events.rst 5368 5369 trace_options=[option-list] 5370 [FTRACE] Enable or disable tracer options at boot. 5371 The option-list is a comma delimited list of options 5372 that can be enabled or disabled just as if you were 5373 to echo the option name into 5374 5375 /sys/kernel/debug/tracing/trace_options 5376 5377 For example, to enable stacktrace option (to dump the 5378 stack trace of each event), add to the command line: 5379 5380 trace_options=stacktrace 5381 5382 See also Documentation/trace/ftrace.rst "trace options" 5383 section. 5384 5385 tp_printk[FTRACE] 5386 Have the tracepoints sent to printk as well as the 5387 tracing ring buffer. This is useful for early boot up 5388 where the system hangs or reboots and does not give the 5389 option for reading the tracing buffer or performing a 5390 ftrace_dump_on_oops. 5391 5392 To turn off having tracepoints sent to printk, 5393 echo 0 > /proc/sys/kernel/tracepoint_printk 5394 Note, echoing 1 into this file without the 5395 tracepoint_printk kernel cmdline option has no effect. 5396 5397 ** CAUTION ** 5398 5399 Having tracepoints sent to printk() and activating high 5400 frequency tracepoints such as irq or sched, can cause 5401 the system to live lock. 5402 5403 traceoff_on_warning 5404 [FTRACE] enable this option to disable tracing when a 5405 warning is hit. This turns off "tracing_on". Tracing can 5406 be enabled again by echoing '1' into the "tracing_on" 5407 file located in /sys/kernel/debug/tracing/ 5408 5409 This option is useful, as it disables the trace before 5410 the WARNING dump is called, which prevents the trace to 5411 be filled with content caused by the warning output. 5412 5413 This option can also be set at run time via the sysctl 5414 option: kernel/traceoff_on_warning 5415 5416 transparent_hugepage= 5417 [KNL] 5418 Format: [always|madvise|never] 5419 Can be used to control the default behavior of the system 5420 with respect to transparent hugepages. 5421 See Documentation/admin-guide/mm/transhuge.rst 5422 for more details. 5423 5424 tsc= Disable clocksource stability checks for TSC. 5425 Format: <string> 5426 [x86] reliable: mark tsc clocksource as reliable, this 5427 disables clocksource verification at runtime, as well 5428 as the stability checks done at bootup. Used to enable 5429 high-resolution timer mode on older hardware, and in 5430 virtualized environment. 5431 [x86] noirqtime: Do not use TSC to do irq accounting. 5432 Used to run time disable IRQ_TIME_ACCOUNTING on any 5433 platforms where RDTSC is slow and this accounting 5434 can add overhead. 5435 [x86] unstable: mark the TSC clocksource as unstable, this 5436 marks the TSC unconditionally unstable at bootup and 5437 avoids any further wobbles once the TSC watchdog notices. 5438 [x86] nowatchdog: disable clocksource watchdog. Used 5439 in situations with strict latency requirements (where 5440 interruptions from clocksource watchdog are not 5441 acceptable). 5442 5443 tsc_early_khz= [X86] Skip early TSC calibration and use the given 5444 value instead. Useful when the early TSC frequency discovery 5445 procedure is not reliable, such as on overclocked systems 5446 with CPUID.16h support and partial CPUID.15h support. 5447 Format: <unsigned int> 5448 5449 tsx= [X86] Control Transactional Synchronization 5450 Extensions (TSX) feature in Intel processors that 5451 support TSX control. 5452 5453 This parameter controls the TSX feature. The options are: 5454 5455 on - Enable TSX on the system. Although there are 5456 mitigations for all known security vulnerabilities, 5457 TSX has been known to be an accelerator for 5458 several previous speculation-related CVEs, and 5459 so there may be unknown security risks associated 5460 with leaving it enabled. 5461 5462 off - Disable TSX on the system. (Note that this 5463 option takes effect only on newer CPUs which are 5464 not vulnerable to MDS, i.e., have 5465 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get 5466 the new IA32_TSX_CTRL MSR through a microcode 5467 update. This new MSR allows for the reliable 5468 deactivation of the TSX functionality.) 5469 5470 auto - Disable TSX if X86_BUG_TAA is present, 5471 otherwise enable TSX on the system. 5472 5473 Not specifying this option is equivalent to tsx=off. 5474 5475 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst 5476 for more details. 5477 5478 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async 5479 Abort (TAA) vulnerability. 5480 5481 Similar to Micro-architectural Data Sampling (MDS) 5482 certain CPUs that support Transactional 5483 Synchronization Extensions (TSX) are vulnerable to an 5484 exploit against CPU internal buffers which can forward 5485 information to a disclosure gadget under certain 5486 conditions. 5487 5488 In vulnerable processors, the speculatively forwarded 5489 data can be used in a cache side channel attack, to 5490 access data to which the attacker does not have direct 5491 access. 5492 5493 This parameter controls the TAA mitigation. The 5494 options are: 5495 5496 full - Enable TAA mitigation on vulnerable CPUs 5497 if TSX is enabled. 5498 5499 full,nosmt - Enable TAA mitigation and disable SMT on 5500 vulnerable CPUs. If TSX is disabled, SMT 5501 is not disabled because CPU is not 5502 vulnerable to cross-thread TAA attacks. 5503 off - Unconditionally disable TAA mitigation 5504 5505 On MDS-affected machines, tsx_async_abort=off can be 5506 prevented by an active MDS mitigation as both vulnerabilities 5507 are mitigated with the same mechanism so in order to disable 5508 this mitigation, you need to specify mds=off too. 5509 5510 Not specifying this option is equivalent to 5511 tsx_async_abort=full. On CPUs which are MDS affected 5512 and deploy MDS mitigation, TAA mitigation is not 5513 required and doesn't provide any additional 5514 mitigation. 5515 5516 For details see: 5517 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst 5518 5519 turbografx.map[2|3]= [HW,JOY] 5520 TurboGraFX parallel port interface 5521 Format: 5522 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7> 5523 See also Documentation/input/devices/joystick-parport.rst 5524 5525 udbg-immortal [PPC] When debugging early kernel crashes that 5526 happen after console_init() and before a proper 5527 console driver takes over, this boot options might 5528 help "seeing" what's going on. 5529 5530 uhash_entries= [KNL,NET] 5531 Set number of hash buckets for UDP/UDP-Lite connections 5532 5533 uhci-hcd.ignore_oc= 5534 [USB] Ignore overcurrent events (default N). 5535 Some badly-designed motherboards generate lots of 5536 bogus events, for ports that aren't wired to 5537 anything. Set this parameter to avoid log spamming. 5538 Note that genuine overcurrent events won't be 5539 reported either. 5540 5541 unknown_nmi_panic 5542 [X86] Cause panic on unknown NMI. 5543 5544 usbcore.authorized_default= 5545 [USB] Default USB device authorization: 5546 (default -1 = authorized except for wireless USB, 5547 0 = not authorized, 1 = authorized, 2 = authorized 5548 if device connected to internal port) 5549 5550 usbcore.autosuspend= 5551 [USB] The autosuspend time delay (in seconds) used 5552 for newly-detected USB devices (default 2). This 5553 is the time required before an idle device will be 5554 autosuspended. Devices for which the delay is set 5555 to a negative value won't be autosuspended at all. 5556 5557 usbcore.usbfs_snoop= 5558 [USB] Set to log all usbfs traffic (default 0 = off). 5559 5560 usbcore.usbfs_snoop_max= 5561 [USB] Maximum number of bytes to snoop in each URB 5562 (default = 65536). 5563 5564 usbcore.blinkenlights= 5565 [USB] Set to cycle leds on hubs (default 0 = off). 5566 5567 usbcore.old_scheme_first= 5568 [USB] Start with the old device initialization 5569 scheme (default 0 = off). 5570 5571 usbcore.usbfs_memory_mb= 5572 [USB] Memory limit (in MB) for buffers allocated by 5573 usbfs (default = 16, 0 = max = 2047). 5574 5575 usbcore.use_both_schemes= 5576 [USB] Try the other device initialization scheme 5577 if the first one fails (default 1 = enabled). 5578 5579 usbcore.initial_descriptor_timeout= 5580 [USB] Specifies timeout for the initial 64-byte 5581 USB_REQ_GET_DESCRIPTOR request in milliseconds 5582 (default 5000 = 5.0 seconds). 5583 5584 usbcore.nousb [USB] Disable the USB subsystem 5585 5586 usbcore.quirks= 5587 [USB] A list of quirk entries to augment the built-in 5588 usb core quirk list. List entries are separated by 5589 commas. Each entry has the form 5590 VendorID:ProductID:Flags. The IDs are 4-digit hex 5591 numbers and Flags is a set of letters. Each letter 5592 will change the built-in quirk; setting it if it is 5593 clear and clearing it if it is set. The letters have 5594 the following meanings: 5595 a = USB_QUIRK_STRING_FETCH_255 (string 5596 descriptors must not be fetched using 5597 a 255-byte read); 5598 b = USB_QUIRK_RESET_RESUME (device can't resume 5599 correctly so reset it instead); 5600 c = USB_QUIRK_NO_SET_INTF (device can't handle 5601 Set-Interface requests); 5602 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't 5603 handle its Configuration or Interface 5604 strings); 5605 e = USB_QUIRK_RESET (device can't be reset 5606 (e.g morph devices), don't use reset); 5607 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has 5608 more interface descriptions than the 5609 bNumInterfaces count, and can't handle 5610 talking to these interfaces); 5611 g = USB_QUIRK_DELAY_INIT (device needs a pause 5612 during initialization, after we read 5613 the device descriptor); 5614 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For 5615 high speed and super speed interrupt 5616 endpoints, the USB 2.0 and USB 3.0 spec 5617 require the interval in microframes (1 5618 microframe = 125 microseconds) to be 5619 calculated as interval = 2 ^ 5620 (bInterval-1). 5621 Devices with this quirk report their 5622 bInterval as the result of this 5623 calculation instead of the exponent 5624 variable used in the calculation); 5625 i = USB_QUIRK_DEVICE_QUALIFIER (device can't 5626 handle device_qualifier descriptor 5627 requests); 5628 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device 5629 generates spurious wakeup, ignore 5630 remote wakeup capability); 5631 k = USB_QUIRK_NO_LPM (device can't handle Link 5632 Power Management); 5633 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL 5634 (Device reports its bInterval as linear 5635 frames instead of the USB 2.0 5636 calculation); 5637 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs 5638 to be disconnected before suspend to 5639 prevent spurious wakeup); 5640 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a 5641 pause after every control message); 5642 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra 5643 delay after resetting its port); 5644 Example: quirks=0781:5580:bk,0a5c:5834:gij 5645 5646 usbhid.mousepoll= 5647 [USBHID] The interval which mice are to be polled at. 5648 5649 usbhid.jspoll= 5650 [USBHID] The interval which joysticks are to be polled at. 5651 5652 usbhid.kbpoll= 5653 [USBHID] The interval which keyboards are to be polled at. 5654 5655 usb-storage.delay_use= 5656 [UMS] The delay in seconds before a new device is 5657 scanned for Logical Units (default 1). 5658 5659 usb-storage.quirks= 5660 [UMS] A list of quirks entries to supplement or 5661 override the built-in unusual_devs list. List 5662 entries are separated by commas. Each entry has 5663 the form VID:PID:Flags where VID and PID are Vendor 5664 and Product ID values (4-digit hex numbers) and 5665 Flags is a set of characters, each corresponding 5666 to a common usb-storage quirk flag as follows: 5667 a = SANE_SENSE (collect more than 18 bytes 5668 of sense data, not on uas); 5669 b = BAD_SENSE (don't collect more than 18 5670 bytes of sense data, not on uas); 5671 c = FIX_CAPACITY (decrease the reported 5672 device capacity by one sector); 5673 d = NO_READ_DISC_INFO (don't use 5674 READ_DISC_INFO command, not on uas); 5675 e = NO_READ_CAPACITY_16 (don't use 5676 READ_CAPACITY_16 command); 5677 f = NO_REPORT_OPCODES (don't use report opcodes 5678 command, uas only); 5679 g = MAX_SECTORS_240 (don't transfer more than 5680 240 sectors at a time, uas only); 5681 h = CAPACITY_HEURISTICS (decrease the 5682 reported device capacity by one 5683 sector if the number is odd); 5684 i = IGNORE_DEVICE (don't bind to this 5685 device); 5686 j = NO_REPORT_LUNS (don't use report luns 5687 command, uas only); 5688 k = NO_SAME (do not use WRITE_SAME, uas only) 5689 l = NOT_LOCKABLE (don't try to lock and 5690 unlock ejectable media, not on uas); 5691 m = MAX_SECTORS_64 (don't transfer more 5692 than 64 sectors = 32 KB at a time, 5693 not on uas); 5694 n = INITIAL_READ10 (force a retry of the 5695 initial READ(10) command, not on uas); 5696 o = CAPACITY_OK (accept the capacity 5697 reported by the device, not on uas); 5698 p = WRITE_CACHE (the device cache is ON 5699 by default, not on uas); 5700 r = IGNORE_RESIDUE (the device reports 5701 bogus residue values, not on uas); 5702 s = SINGLE_LUN (the device has only one 5703 Logical Unit); 5704 t = NO_ATA_1X (don't allow ATA(12) and ATA(16) 5705 commands, uas only); 5706 u = IGNORE_UAS (don't bind to the uas driver); 5707 w = NO_WP_DETECT (don't test whether the 5708 medium is write-protected). 5709 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE 5710 even if the device claims no cache, 5711 not on uas) 5712 Example: quirks=0419:aaf5:rl,0421:0433:rc 5713 5714 user_debug= [KNL,ARM] 5715 Format: <int> 5716 See arch/arm/Kconfig.debug help text. 5717 1 - undefined instruction events 5718 2 - system calls 5719 4 - invalid data aborts 5720 8 - SIGSEGV faults 5721 16 - SIGBUS faults 5722 Example: user_debug=31 5723 5724 userpte= 5725 [X86] Flags controlling user PTE allocations. 5726 5727 nohigh = do not allocate PTE pages in 5728 HIGHMEM regardless of setting 5729 of CONFIG_HIGHPTE. 5730 5731 vdso= [X86,SH] 5732 On X86_32, this is an alias for vdso32=. Otherwise: 5733 5734 vdso=1: enable VDSO (the default) 5735 vdso=0: disable VDSO mapping 5736 5737 vdso32= [X86] Control the 32-bit vDSO 5738 vdso32=1: enable 32-bit VDSO 5739 vdso32=0 or vdso32=2: disable 32-bit VDSO 5740 5741 See the help text for CONFIG_COMPAT_VDSO for more 5742 details. If CONFIG_COMPAT_VDSO is set, the default is 5743 vdso32=0; otherwise, the default is vdso32=1. 5744 5745 For compatibility with older kernels, vdso32=2 is an 5746 alias for vdso32=0. 5747 5748 Try vdso32=0 if you encounter an error that says: 5749 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed! 5750 5751 vector= [IA-64,SMP] 5752 vector=percpu: enable percpu vector domain 5753 5754 video= [FB] Frame buffer configuration 5755 See Documentation/fb/modedb.rst. 5756 5757 video.brightness_switch_enabled= [0,1] 5758 If set to 1, on receiving an ACPI notify event 5759 generated by hotkey, video driver will adjust brightness 5760 level and then send out the event to user space through 5761 the allocated input device; If set to 0, video driver 5762 will only send out the event without touching backlight 5763 brightness level. 5764 default: 1 5765 5766 virtio_mmio.device= 5767 [VMMIO] Memory mapped virtio (platform) device. 5768 5769 <size>@<baseaddr>:<irq>[:<id>] 5770 where: 5771 <size> := size (can use standard suffixes 5772 like K, M and G) 5773 <baseaddr> := physical base address 5774 <irq> := interrupt number (as passed to 5775 request_irq()) 5776 <id> := (optional) platform device id 5777 example: 5778 virtio_mmio.device=1K@0x100b0000:48:7 5779 5780 Can be used multiple times for multiple devices. 5781 5782 vga= [BOOT,X86-32] Select a particular video mode 5783 See Documentation/x86/boot.rst and 5784 Documentation/admin-guide/svga.rst. 5785 Use vga=ask for menu. 5786 This is actually a boot loader parameter; the value is 5787 passed to the kernel using a special protocol. 5788 5789 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y. 5790 May slow down system boot speed, especially when 5791 enabled on systems with a large amount of memory. 5792 All options are enabled by default, and this 5793 interface is meant to allow for selectively 5794 enabling or disabling specific virtual memory 5795 debugging features. 5796 5797 Available options are: 5798 P Enable page structure init time poisoning 5799 - Disable all of the above options 5800 5801 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact 5802 size of <nn>. This can be used to increase the 5803 minimum size (128MB on x86). It can also be used to 5804 decrease the size and leave more room for directly 5805 mapped kernel RAM. 5806 5807 vmcp_cma=nn[MG] [KNL,S390] 5808 Sets the memory size reserved for contiguous memory 5809 allocations for the vmcp device driver. 5810 5811 vmhalt= [KNL,S390] Perform z/VM CP command after system halt. 5812 Format: <command> 5813 5814 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic. 5815 Format: <command> 5816 5817 vmpoff= [KNL,S390] Perform z/VM CP command after power off. 5818 Format: <command> 5819 5820 vsyscall= [X86-64] 5821 Controls the behavior of vsyscalls (i.e. calls to 5822 fixed addresses of 0xffffffffff600x00 from legacy 5823 code). Most statically-linked binaries and older 5824 versions of glibc use these calls. Because these 5825 functions are at fixed addresses, they make nice 5826 targets for exploits that can control RIP. 5827 5828 emulate [default] Vsyscalls turn into traps and are 5829 emulated reasonably safely. The vsyscall 5830 page is readable. 5831 5832 xonly Vsyscalls turn into traps and are 5833 emulated reasonably safely. The vsyscall 5834 page is not readable. 5835 5836 none Vsyscalls don't work at all. This makes 5837 them quite hard to use for exploits but 5838 might break your system. 5839 5840 vt.color= [VT] Default text color. 5841 Format: 0xYX, X = foreground, Y = background. 5842 Default: 0x07 = light gray on black. 5843 5844 vt.cur_default= [VT] Default cursor shape. 5845 Format: 0xCCBBAA, where AA, BB, and CC are the same as 5846 the parameters of the <Esc>[?A;B;Cc escape sequence; 5847 see VGA-softcursor.txt. Default: 2 = underline. 5848 5849 vt.default_blu= [VT] 5850 Format: <blue0>,<blue1>,<blue2>,...,<blue15> 5851 Change the default blue palette of the console. 5852 This is a 16-member array composed of values 5853 ranging from 0-255. 5854 5855 vt.default_grn= [VT] 5856 Format: <green0>,<green1>,<green2>,...,<green15> 5857 Change the default green palette of the console. 5858 This is a 16-member array composed of values 5859 ranging from 0-255. 5860 5861 vt.default_red= [VT] 5862 Format: <red0>,<red1>,<red2>,...,<red15> 5863 Change the default red palette of the console. 5864 This is a 16-member array composed of values 5865 ranging from 0-255. 5866 5867 vt.default_utf8= 5868 [VT] 5869 Format=<0|1> 5870 Set system-wide default UTF-8 mode for all tty's. 5871 Default is 1, i.e. UTF-8 mode is enabled for all 5872 newly opened terminals. 5873 5874 vt.global_cursor_default= 5875 [VT] 5876 Format=<-1|0|1> 5877 Set system-wide default for whether a cursor 5878 is shown on new VTs. Default is -1, 5879 i.e. cursors will be created by default unless 5880 overridden by individual drivers. 0 will hide 5881 cursors, 1 will display them. 5882 5883 vt.italic= [VT] Default color for italic text; 0-15. 5884 Default: 2 = green. 5885 5886 vt.underline= [VT] Default color for underlined text; 0-15. 5887 Default: 3 = cyan. 5888 5889 watchdog timers [HW,WDT] For information on watchdog timers, 5890 see Documentation/watchdog/watchdog-parameters.rst 5891 or other driver-specific files in the 5892 Documentation/watchdog/ directory. 5893 5894 watchdog_thresh= 5895 [KNL] 5896 Set the hard lockup detector stall duration 5897 threshold in seconds. The soft lockup detector 5898 threshold is set to twice the value. A value of 0 5899 disables both lockup detectors. Default is 10 5900 seconds. 5901 5902 workqueue.watchdog_thresh= 5903 If CONFIG_WQ_WATCHDOG is configured, workqueue can 5904 warn stall conditions and dump internal state to 5905 help debugging. 0 disables workqueue stall 5906 detection; otherwise, it's the stall threshold 5907 duration in seconds. The default value is 30 and 5908 it can be updated at runtime by writing to the 5909 corresponding sysfs file. 5910 5911 workqueue.disable_numa 5912 By default, all work items queued to unbound 5913 workqueues are affine to the NUMA nodes they're 5914 issued on, which results in better behavior in 5915 general. If NUMA affinity needs to be disabled for 5916 whatever reason, this option can be used. Note 5917 that this also can be controlled per-workqueue for 5918 workqueues visible under /sys/bus/workqueue/. 5919 5920 workqueue.power_efficient 5921 Per-cpu workqueues are generally preferred because 5922 they show better performance thanks to cache 5923 locality; unfortunately, per-cpu workqueues tend to 5924 be more power hungry than unbound workqueues. 5925 5926 Enabling this makes the per-cpu workqueues which 5927 were observed to contribute significantly to power 5928 consumption unbound, leading to measurably lower 5929 power usage at the cost of small performance 5930 overhead. 5931 5932 The default value of this parameter is determined by 5933 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT. 5934 5935 workqueue.debug_force_rr_cpu 5936 Workqueue used to implicitly guarantee that work 5937 items queued without explicit CPU specified are put 5938 on the local CPU. This guarantee is no longer true 5939 and while local CPU is still preferred work items 5940 may be put on foreign CPUs. This debug option 5941 forces round-robin CPU selection to flush out 5942 usages which depend on the now broken guarantee. 5943 When enabled, memory and cache locality will be 5944 impacted. 5945 5946 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of 5947 default x2apic cluster mode on platforms 5948 supporting x2apic. 5949 5950 x86_intel_mid_timer= [X86-32,APBT] 5951 Choose timer option for x86 Intel MID platform. 5952 Two valid options are apbt timer only and lapic timer 5953 plus one apbt timer for broadcast timer. 5954 x86_intel_mid_timer=apbt_only | lapic_and_apbt 5955 5956 xen_512gb_limit [KNL,X86-64,XEN] 5957 Restricts the kernel running paravirtualized under Xen 5958 to use only up to 512 GB of RAM. The reason to do so is 5959 crash analysis tools and Xen tools for doing domain 5960 save/restore/migration must be enabled to handle larger 5961 domains. 5962 5963 xen_emul_unplug= [HW,X86,XEN] 5964 Unplug Xen emulated devices 5965 Format: [unplug0,][unplug1] 5966 ide-disks -- unplug primary master IDE devices 5967 aux-ide-disks -- unplug non-primary-master IDE devices 5968 nics -- unplug network devices 5969 all -- unplug all emulated devices (NICs and IDE disks) 5970 unnecessary -- unplugging emulated devices is 5971 unnecessary even if the host did not respond to 5972 the unplug protocol 5973 never -- do not unplug even if version check succeeds 5974 5975 xen_legacy_crash [X86,XEN] 5976 Crash from Xen panic notifier, without executing late 5977 panic() code such as dumping handler. 5978 5979 xen_nopvspin [X86,XEN] 5980 Disables the qspinlock slowpath using Xen PV optimizations. 5981 This parameter is obsoleted by "nopvspin" parameter, which 5982 has equivalent effect for XEN platform. 5983 5984 xen_nopv [X86] 5985 Disables the PV optimizations forcing the HVM guest to 5986 run as generic HVM guest with no PV drivers. 5987 This option is obsoleted by the "nopv" option, which 5988 has equivalent effect for XEN platform. 5989 5990 xen_no_vector_callback 5991 [KNL,X86,XEN] Disable the vector callback for Xen 5992 event channel interrupts. 5993 5994 xen_scrub_pages= [XEN] 5995 Boolean option to control scrubbing pages before giving them back 5996 to Xen, for use by other domains. Can be also changed at runtime 5997 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages. 5998 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT. 5999 6000 xen_timer_slop= [X86-64,XEN] 6001 Set the timer slop (in nanoseconds) for the virtual Xen 6002 timers (default is 100000). This adjusts the minimum 6003 delta of virtualized Xen timers, where lower values 6004 improve timer resolution at the expense of processing 6005 more timer interrupts. 6006 6007 xen.event_eoi_delay= [XEN] 6008 How long to delay EOI handling in case of event 6009 storms (jiffies). Default is 10. 6010 6011 xen.event_loop_timeout= [XEN] 6012 After which time (jiffies) the event handling loop 6013 should start to delay EOI handling. Default is 2. 6014 6015 xen.fifo_events= [XEN] 6016 Boolean parameter to disable using fifo event handling 6017 even if available. Normally fifo event handling is 6018 preferred over the 2-level event handling, as it is 6019 fairer and the number of possible event channels is 6020 much higher. Default is on (use fifo events). 6021 6022 nopv= [X86,XEN,KVM,HYPER_V,VMWARE] 6023 Disables the PV optimizations forcing the guest to run 6024 as generic guest with no PV drivers. Currently support 6025 XEN HVM, KVM, HYPER_V and VMWARE guest. 6026 6027 nopvspin [X86,XEN,KVM] 6028 Disables the qspinlock slow path using PV optimizations 6029 which allow the hypervisor to 'idle' the guest on lock 6030 contention. 6031 6032 xirc2ps_cs= [NET,PCMCIA] 6033 Format: 6034 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]] 6035 6036 xive= [PPC] 6037 By default on POWER9 and above, the kernel will 6038 natively use the XIVE interrupt controller. This option 6039 allows the fallback firmware mode to be used: 6040 6041 off Fallback to firmware control of XIVE interrupt 6042 controller on both pseries and powernv 6043 platforms. Only useful on POWER9 and above. 6044 6045 xhci-hcd.quirks [USB,KNL] 6046 A hex value specifying bitmask with supplemental xhci 6047 host controller quirks. Meaning of each bit can be 6048 consulted in header drivers/usb/host/xhci.h. 6049 6050 xmon [PPC] 6051 Format: { early | on | rw | ro | off } 6052 Controls if xmon debugger is enabled. Default is off. 6053 Passing only "xmon" is equivalent to "xmon=early". 6054 early Call xmon as early as possible on boot; xmon 6055 debugger is called from setup_arch(). 6056 on xmon debugger hooks will be installed so xmon 6057 is only called on a kernel crash. Default mode, 6058 i.e. either "ro" or "rw" mode, is controlled 6059 with CONFIG_XMON_DEFAULT_RO_MODE. 6060 rw xmon debugger hooks will be installed so xmon 6061 is called only on a kernel crash, mode is write, 6062 meaning SPR registers, memory and, other data 6063 can be written using xmon commands. 6064 ro same as "rw" option above but SPR registers, 6065 memory, and other data can't be written using 6066 xmon commands. 6067 off xmon is disabled. 6068