1 acpi= [HW,ACPI,X86,ARM64]
2 Advanced Configuration and Power Interface
3 Format: { force | on | off | strict | noirq | rsdt |
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"
16 See also Documentation/power/runtime_pm.txt, pci=noacpi
18 acpi_apic_instance= [ACPI, IOAPIC]
20 2: use 2nd APIC table, if available
21 1,0: use 1st APIC table
24 acpi_backlight= [HW,ACPI]
27 If set to vendor, prefer vendor specific driver
28 (e.g. thinkpad_acpi, sony_acpi, etc.) instead
29 of the ACPI video.ko driver.
31 acpi_force_32bit_fadt_addr
32 force FADT to use 32 bit addresses rather than the
33 64 bit X_* addresses. Some firmware have broken 64
34 bit addresses for force ACPI ignore these and use
35 the older legacy 32 bit addresses.
37 acpica_no_return_repair [HW, ACPI]
38 Disable AML predefined validation mechanism
39 This mechanism can repair the evaluation result to make
40 the return objects more ACPI specification compliant.
41 This option is useful for developers to identify the
42 root cause of an AML interpreter issue when the issue
43 has something to do with the repair mechanism.
45 acpi.debug_layer= [HW,ACPI,ACPI_DEBUG]
46 acpi.debug_level= [HW,ACPI,ACPI_DEBUG]
48 CONFIG_ACPI_DEBUG must be enabled to produce any ACPI
49 debug output. Bits in debug_layer correspond to a
50 _COMPONENT in an ACPI source file, e.g.,
51 #define _COMPONENT ACPI_PCI_COMPONENT
52 Bits in debug_level correspond to a level in
53 ACPI_DEBUG_PRINT statements, e.g.,
54 ACPI_DEBUG_PRINT((ACPI_DB_INFO, ...
55 The debug_level mask defaults to "info". See
56 Documentation/acpi/debug.txt for more information about
57 debug layers and levels.
59 Enable processor driver info messages:
60 acpi.debug_layer=0x20000000
61 Enable PCI/PCI interrupt routing info messages:
62 acpi.debug_layer=0x400000
63 Enable AML "Debug" output, i.e., stores to the Debug
64 object while interpreting AML:
65 acpi.debug_layer=0xffffffff acpi.debug_level=0x2
66 Enable all messages related to ACPI hardware:
67 acpi.debug_layer=0x2 acpi.debug_level=0xffffffff
69 Some values produce so much output that the system is
70 unusable. The "log_buf_len" parameter may be useful
71 if you need to capture more output.
73 acpi_enforce_resources= [ACPI]
75 Check for resource conflicts between native drivers
76 and ACPI OperationRegions (SystemIO and SystemMemory
77 only). IO ports and memory declared in ACPI might be
78 used by the ACPI subsystem in arbitrary AML code and
79 can interfere with legacy drivers.
80 strict (default): access to resources claimed by ACPI
81 is denied; legacy drivers trying to access reserved
82 resources will fail to bind to device using them.
83 lax: access to resources claimed by ACPI is allowed;
84 legacy drivers trying to access reserved resources
85 will bind successfully but a warning message is logged.
86 no: ACPI OperationRegions are not marked as reserved,
87 no further checks are performed.
89 acpi_force_table_verification [HW,ACPI]
90 Enable table checksum verification during early stage.
91 By default, this is disabled due to x86 early mapping
94 acpi_irq_balance [HW,ACPI]
95 ACPI will balance active IRQs
98 acpi_irq_nobalance [HW,ACPI]
99 ACPI will not move active IRQs (default)
102 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA
103 Format: <irq>,<irq>...
105 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for
107 Format: <irq>,<irq>...
109 acpi_mask_gpe= [HW,ACPI]
110 Due to the existence of _Lxx/_Exx, some GPEs triggered
111 by unsupported hardware/firmware features can result in
112 GPE floodings that cannot be automatically disabled by
114 This facility can be used to prevent such uncontrolled
117 Support masking of GPEs numbered from 0x00 to 0x7f.
119 acpi_no_auto_serialize [HW,ACPI]
120 Disable auto-serialization of AML methods
121 AML control methods that contain the opcodes to create
122 named objects will be marked as "Serialized" by the
123 auto-serialization feature.
124 This feature is enabled by default.
125 This option allows to turn off the feature.
127 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump
130 acpi_no_static_ssdt [HW,ACPI]
131 Disable installation of static SSDTs at early boot time
132 By default, SSDTs contained in the RSDT/XSDT will be
133 installed automatically and they will appear under
134 /sys/firmware/acpi/tables.
135 This option turns off this feature.
136 Note that specifying this option does not affect
137 dynamic table installation which will install SSDT
138 tables to /sys/firmware/acpi/tables/dynamic.
140 acpi_rsdp= [ACPI,EFI,KEXEC]
141 Pass the RSDP address to the kernel, mostly used
142 on machines running EFI runtime service to boot the
143 second kernel for kdump.
145 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS
146 Format: To spoof as Windows 98: ="Microsoft Windows"
148 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead
149 of 2 which is mandated by ACPI 6) as the supported ACPI
150 specification revision (when using this switch, it may
151 be necessary to carry out a cold reboot _twice_ in a
152 row to make it take effect on the platform firmware).
154 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings
155 acpi_osi="string1" # add string1
156 acpi_osi="!string2" # remove string2
157 acpi_osi=!* # remove all strings
158 acpi_osi=! # disable all built-in OS vendor
160 acpi_osi=!! # enable all built-in OS vendor
162 acpi_osi= # disable all strings
164 'acpi_osi=!' can be used in combination with single or
165 multiple 'acpi_osi="string1"' to support specific OS
166 vendor string(s). Note that such command can only
167 affect the default state of the OS vendor strings, thus
168 it cannot affect the default state of the feature group
169 strings and the current state of the OS vendor strings,
170 specifying it multiple times through kernel command line
171 is meaningless. This command is useful when one do not
172 care about the state of the feature group strings which
173 should be controlled by the OSPM.
175 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent
176 to 'acpi_osi="Windows 2000" acpi_osi=!', they all
177 can make '_OSI("Windows 2000")' TRUE.
179 'acpi_osi=' cannot be used in combination with other
180 'acpi_osi=' command lines, the _OSI method will not
181 exist in the ACPI namespace. NOTE that such command can
182 only affect the _OSI support state, thus specifying it
183 multiple times through kernel command line is also
186 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)'
189 'acpi_osi=!*' can be used in combination with single or
190 multiple 'acpi_osi="string1"' to support specific
191 string(s). Note that such command can affect the
192 current state of both the OS vendor strings and the
193 feature group strings, thus specifying it multiple times
194 through kernel command line is meaningful. But it may
195 still not able to affect the final state of a string if
196 there are quirks related to this string. This command
197 is useful when one want to control the state of the
198 feature group strings to debug BIOS issues related to
201 1. 'acpi_osi="Module Device" acpi_osi=!*' can make
202 '_OSI("Module Device")' FALSE.
203 2. 'acpi_osi=!* acpi_osi="Module Device"' can make
204 '_OSI("Module Device")' TRUE.
205 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is
207 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"'
209 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!',
210 they all will make '_OSI("Windows 2000")' TRUE.
213 Override the pmtimer bug detection: force the kernel
214 to assume that this machine's pmtimer latches its value
215 and always returns good values.
217 acpi_sci= [HW,ACPI] ACPI System Control Interrupt trigger mode
218 Format: { level | edge | high | low }
220 acpi_skip_timer_override [HW,ACPI]
221 Recognize and ignore IRQ0/pin2 Interrupt Override.
222 For broken nForce2 BIOS resulting in XT-PIC timer.
224 acpi_sleep= [HW,ACPI] Sleep options
225 Format: { s3_bios, s3_mode, s3_beep, s4_nohwsig,
226 old_ordering, nonvs, sci_force_enable }
227 See Documentation/power/video.txt for information on
229 s3_beep is for debugging; it makes the PC's speaker beep
230 as soon as the kernel's real-mode entry point is called.
231 s4_nohwsig prevents ACPI hardware signature from being
232 used during resume from hibernation.
233 old_ordering causes the ACPI 1.0 ordering of the _PTS
234 control method, with respect to putting devices into
235 low power states, to be enforced (the ACPI 2.0 ordering
236 of _PTS is used by default).
237 nonvs prevents the kernel from saving/restoring the
238 ACPI NVS memory during suspend/hibernation and resume.
239 sci_force_enable causes the kernel to set SCI_EN directly
240 on resume from S1/S3 (which is against the ACPI spec,
241 but some broken systems don't work without it).
243 acpi_use_timer_override [HW,ACPI]
244 Use timer override. For some broken Nvidia NF5 boards
245 that require a timer override, but don't have HPET
247 add_efi_memmap [EFI; X86] Include EFI memory map in
248 kernel's map of available physical RAM.
251 { off | try_unsupported }
252 off: disable AGP support
253 try_unsupported: try to drive unsupported chipsets
254 (may crash computer or cause data corruption)
257 See Documentation/sound/alsa/alsa-parameters.txt
260 Allow the default userspace alignment fault handler
261 behaviour to be specified. Bit 0 enables warnings,
262 bit 1 enables fixups, and bit 2 sends a segfault.
264 align_va_addr= [X86-64]
265 Align virtual addresses by clearing slice [14:12] when
266 allocating a VMA at process creation time. This option
267 gives you up to 3% performance improvement on AMD F15h
268 machines (where it is enabled by default) for a
269 CPU-intensive style benchmark, and it can vary highly in
270 a microbenchmark depending on workload and compiler.
272 32: only for 32-bit processes
273 64: only for 64-bit processes
274 on: enable for both 32- and 64-bit processes
275 off: disable for both 32- and 64-bit processes
277 alloc_snapshot [FTRACE]
278 Allocate the ftrace snapshot buffer on boot up when the
279 main buffer is allocated. This is handy if debugging
280 and you need to use tracing_snapshot() on boot up, and
281 do not want to use tracing_snapshot_alloc() as it needs
282 to be done where GFP_KERNEL allocations are allowed.
284 amd_iommu= [HW,X86-64]
285 Pass parameters to the AMD IOMMU driver in the system.
287 fullflush - enable flushing of IO/TLB entries when
288 they are unmapped. Otherwise they are
289 flushed before they will be reused, which
291 off - do not initialize any AMD IOMMU found in
293 force_isolation - Force device isolation for all
294 devices. The IOMMU driver is not
295 allowed anymore to lift isolation
296 requirements as needed. This option
297 does not override iommu=pt
299 amd_iommu_dump= [HW,X86-64]
300 Enable AMD IOMMU driver option to dump the ACPI table
301 for AMD IOMMU. With this option enabled, AMD IOMMU
302 driver will print ACPI tables for AMD IOMMU during
303 IOMMU initialization.
305 amd_iommu_intr= [HW,X86-64]
306 Specifies one of the following AMD IOMMU interrupt
308 legacy - Use legacy interrupt remapping mode.
309 vapic - Use virtual APIC mode, which allows IOMMU
310 to inject interrupts directly into guest.
311 This mode requires kvm-amd.avic=1.
312 (Default when IOMMU HW support is present.)
314 amijoy.map= [HW,JOY] Amiga joystick support
315 Map of devices attached to JOY0DAT and JOY1DAT
317 See also Documentation/input/joystick.txt
319 analog.map= [HW,JOY] Analog joystick and gamepad support
320 Specifies type or capabilities of an analog joystick
321 connected to one of 16 gameports
322 Format: <type1>,<type2>,..<type16>
325 Power management functions (SPARCstation-4/5 + deriv.)
327 Disable APC CPU standby support. SPARCstation-Fox does
328 not play well with APC CPU idle - disable it if you have
329 APC and your system crashes randomly.
331 apic= [APIC,X86-32] Advanced Programmable Interrupt Controller
332 Change the output verbosity whilst booting
333 Format: { quiet (default) | verbose | debug }
334 Change the amount of debugging information output
335 when initialising the APIC and IO-APIC components.
337 apic_extnmi= [APIC,X86] External NMI delivery setting
338 Format: { bsp (default) | all | none }
339 bsp: External NMI is delivered only to CPU 0
340 all: External NMIs are broadcast to all CPUs as a
342 none: External NMI is masked for all CPUs. This is
343 useful so that a dump capture kernel won't be
347 See Documentation/networking/ipv6.txt.
349 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
350 Limit apic dumping. The parameter defines the maximal
351 number of local apics being dumped. Also it is possible
352 to set it to "all" by meaning -- no limit here.
353 Format: { 1 (default) | 2 | ... | all }.
354 The parameter valid if only apic=debug or
355 apic=verbose is specified.
356 Example: apic=debug show_lapic=all
358 apm= [APM] Advanced Power Management
359 See header of arch/x86/kernel/apm_32.c.
361 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
362 Format: <io>,<irq>,<nodeID>
366 atarimouse= [HW,MOUSE] Atari Mouse
368 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
369 EzKey and similar keyboards
371 atkbd.reset= [HW] Reset keyboard during initialization
373 atkbd.set= [HW] Select keyboard code set
374 Format: <int> (2 = AT (default), 3 = PS/2)
376 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
379 atkbd.softraw= [HW] Choose between synthetic and real raw mode
380 Format: <bool> (0 = real, 1 = synthetic (default))
382 atkbd.softrepeat= [HW]
383 Use software keyboard repeat
385 audit= [KNL] Enable the audit sub-system
386 Format: { "0" | "1" } (0 = disabled, 1 = enabled)
387 0 - kernel audit is disabled and can not be enabled
388 until the next reboot
389 unset - kernel audit is initialized but disabled and
390 will be fully enabled by the userspace auditd.
391 1 - kernel audit is initialized and partially enabled,
392 storing at most audit_backlog_limit messages in
393 RAM until it is fully enabled by the userspace
397 audit_backlog_limit= [KNL] Set the audit queue size limit.
398 Format: <int> (must be >=0)
401 bau= [X86_UV] Enable the BAU on SGI UV. The default
402 behavior is to disable the BAU (i.e. bau=0).
403 Format: { "0" | "1" }
406 unset - Disable the BAU.
408 baycom_epp= [HW,AX25]
411 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
413 See header of drivers/net/hamradio/baycom_par.c.
415 baycom_ser_fdx= [HW,AX25]
416 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
417 Format: <io>,<irq>,<mode>[,<baud>]
418 See header of drivers/net/hamradio/baycom_ser_fdx.c.
420 baycom_ser_hdx= [HW,AX25]
421 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
422 Format: <io>,<irq>,<mode>
423 See header of drivers/net/hamradio/baycom_ser_hdx.c.
425 blkdevparts= Manual partition parsing of block device(s) for
426 embedded devices based on command line input.
427 See Documentation/block/cmdline-partition.txt
429 boot_delay= Milliseconds to delay each printk during boot.
430 Values larger than 10 seconds (10000) are changed to
434 bootmem_debug [KNL] Enable bootmem allocator debug messages.
437 Disable BERT OS support on buggy BIOSes.
439 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
440 bttv.radio= Most important insmod options are available as
442 bttv.pll= See Documentation/video4linux/bttv/Insmod-options
445 bulk_remove=off [PPC] This parameter disables the use of the pSeries
446 firmware feature for flushing multiple hpte entries
449 c101= [NET] Moxa C101 synchronous serial card
451 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
452 Sometimes CPU hardware bugs make them report the cache
453 size incorrectly. The kernel will attempt work arounds
454 to fix known problems, but for some CPUs it is not
455 possible to determine what the correct size should be.
456 This option provides an override for these situations.
458 ca_keys= [KEYS] This parameter identifies a specific key(s) on
459 the system trusted keyring to be used for certificate
461 format: { id:<keyid> | builtin }
463 cca= [MIPS] Override the kernel pages' cache coherency
464 algorithm. Accepted values range from 0 to 7
465 inclusive. See arch/mips/include/asm/pgtable-bits.h
466 for platform specific values (SB1, Loongson3 and
469 ccw_timeout_log [S390]
470 See Documentation/s390/CommonIO for details.
472 cgroup_disable= [KNL] Disable a particular controller
473 Format: {name of the controller(s) to disable}
474 The effects of cgroup_disable=foo are:
475 - foo isn't auto-mounted if you mount all cgroups in
477 - foo isn't visible as an individually mountable
479 {Currently only "memory" controller deal with this and
480 cut the overhead, others just disable the usage. So
481 only cgroup_disable=memory is actually worthy}
483 cgroup_no_v1= [KNL] Disable one, multiple, all cgroup controllers in v1
484 Format: { controller[,controller...] | "all" }
485 Like cgroup_disable, but only applies to cgroup v1;
486 the blacklisted controllers remain available in cgroup2.
488 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
490 nosocket -- Disable socket memory accounting.
491 nokmem -- Disable kernel memory accounting.
493 checkreqprot [SELINUX] Set initial checkreqprot flag value.
494 Format: { "0" | "1" }
495 See security/selinux/Kconfig help text.
496 0 -- check protection applied by kernel (includes
497 any implied execute protection).
498 1 -- check protection requested by application.
499 Default value is set via a kernel config option.
500 Value can be changed at runtime via
501 /selinux/checkreqprot.
504 See Documentation/s390/CommonIO for details.
507 Prevents the clock framework from automatically gating
508 clocks that have not been explicitly enabled by a Linux
509 device driver but are enabled in hardware at reset or
510 by the bootloader/firmware. Note that this does not
511 force such clocks to be always-on nor does it reserve
512 those clocks in any way. This parameter is useful for
513 debug and development, but should not be needed on a
514 platform with proper driver support. For more
515 information, see Documentation/clk.txt.
517 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
519 Forces specified clocksource (if available) to be used
520 when calculating gettimeofday(). If specified
521 clocksource is not available, it defaults to PIT.
522 Format: { pit | tsc | cyclone | pmtmr }
524 clocksource= Override the default clocksource
526 Override the default clocksource and use the clocksource
527 with the name specified.
528 Some clocksource names to choose from, depending on
530 [all] jiffies (this is the base, fallback clocksource)
532 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
533 pxa_timer,timer3,32k_counter,timer0_1
534 [X86-32] pit,hpet,tsc;
535 scx200_hrt on Geode; cyclone on IBM x440
543 clocksource.arm_arch_timer.evtstrm=
546 Enable/disable the eventstream feature of the ARM
547 architected timer so that code using WFE-based polling
548 loops can be debugged more effectively on production
551 clearcpuid=BITNUM [X86]
552 Disable CPUID feature X for the kernel. See
553 arch/x86/include/asm/cpufeatures.h for the valid bit
554 numbers. Note the Linux specific bits are not necessarily
555 stable over kernel options, but the vendor specific
557 Also note that user programs calling CPUID directly
558 or using the feature without checking anything
559 will still see it. This just prevents it from
560 being used by the kernel or shown in /proc/cpuinfo.
561 Also note the kernel might malfunction if you disable
564 cma=nn[MG]@[start[MG][-end[MG]]]
566 Sets the size of kernel global memory area for
567 contiguous memory allocations and optionally the
568 placement constraint by the physical address range of
569 memory allocations. A value of 0 disables CMA
570 altogether. For more information, see
571 include/linux/dma-contiguous.h
573 cmo_free_hint= [PPC] Format: { yes | no }
574 Specify whether pages are marked as being inactive
575 when they are freed. This is used in CMO environments
576 to determine OS memory pressure for page stealing by
580 coherent_pool=nn[KMG] [ARM,KNL]
581 Sets the size of memory pool for coherent, atomic dma
582 allocations, by default set to 256K.
584 code_bytes [X86] How many bytes of object code to print
589 com20020= [HW,NET] ARCnet - COM20020 chipset
591 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
593 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
597 ARCnet - COM90xx chipset (memory-mapped buffers)
598 Format: <io>[,<irq>[,<memstart>]]
600 condev= [HW,S390] console device
603 console= [KNL] Output console device and options.
605 tty<n> Use the virtual console device <n>.
609 Use the specified serial port. The options are of
610 the form "bbbbpnf", where "bbbb" is the baud rate,
611 "p" is parity ("n", "o", or "e"), "n" is number of
612 bits, and "f" is flow control ("r" for RTS or
613 omit it). Default is "9600n8".
615 See Documentation/admin-guide/serial-console.rst for more
617 Documentation/networking/netconsole.txt for an
620 uart[8250],io,<addr>[,options]
621 uart[8250],mmio,<addr>[,options]
622 uart[8250],mmio16,<addr>[,options]
623 uart[8250],mmio32,<addr>[,options]
624 uart[8250],0x<addr>[,options]
625 Start an early, polled-mode console on the 8250/16550
626 UART at the specified I/O port or MMIO address,
627 switching to the matching ttyS device later.
628 MMIO inter-register address stride is either 8-bit
629 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
630 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
631 to be equivalent to 'mmio'. 'options' are specified in
632 the same format described for ttyS above; if unspecified,
633 the h/w is not re-initialized.
635 hvc<n> Use the hypervisor console device <n>. This is for
636 both Xen and PowerPC hypervisors.
638 If the device connected to the port is not a TTY but a braille
639 device, prepend "brl," before the device type, for instance
641 For now, only VisioBraille is supported.
643 consoleblank= [KNL] The console blank (screen saver) timeout in
644 seconds. Defaults to 10*60 = 10mins. A value of 0
645 disables the blank timer.
648 [KNL] Change the default value for
649 /proc/<pid>/coredump_filter.
650 See also Documentation/filesystems/proc.txt.
652 coresight_cpu_debug.enable
655 Enable/disable the CPU sampling based debugging.
656 0: default value, disable debugging
657 1: enable debugging at boot time
659 cpuidle.off=1 [CPU_IDLE]
660 disable the cpuidle sub-system
662 cpufreq.off=1 [CPU_FREQ]
663 disable the cpufreq sub-system
666 [X86] Delay for N microsec between assert and de-assert
667 of APIC INIT to start processors. This delay occurs
668 on every CPU online, such as boot, and resume from suspend.
671 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
673 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
675 crashkernel=size[KMG][@offset[KMG]]
676 [KNL] Using kexec, Linux can switch to a 'crash kernel'
677 upon panic. This parameter reserves the physical
678 memory region [offset, offset + size] for that kernel
679 image. If '@offset' is omitted, then a suitable offset
680 is selected automatically. Check
681 Documentation/kdump/kdump.txt for further details.
683 crashkernel=range1:size1[,range2:size2,...][@offset]
684 [KNL] Same as above, but depends on the memory
685 in the running system. The syntax of range is
686 start-[end] where start and end are both
687 a memory unit (amount[KMG]). See also
688 Documentation/kdump/kdump.txt for an example.
690 crashkernel=size[KMG],high
691 [KNL, x86_64] range could be above 4G. Allow kernel
692 to allocate physical memory region from top, so could
693 be above 4G if system have more than 4G ram installed.
694 Otherwise memory region will be allocated below 4G, if
696 It will be ignored if crashkernel=X is specified.
697 crashkernel=size[KMG],low
698 [KNL, x86_64] range under 4G. When crashkernel=X,high
699 is passed, kernel could allocate physical memory region
700 above 4G, that cause second kernel crash on system
701 that require some amount of low memory, e.g. swiotlb
702 requires at least 64M+32K low memory, also enough extra
703 low memory is needed to make sure DMA buffers for 32-bit
704 devices won't run out. Kernel would try to allocate at
705 at least 256M below 4G automatically.
706 This one let user to specify own low range under 4G
707 for second kernel instead.
708 0: to disable low allocation.
709 It will be ignored when crashkernel=X,high is not used
710 or memory reserved is below 4G.
713 [KNL] Disable crypto self-tests
718 cs89x0_media= [HW,NET]
719 Format: { rj45 | aui | bnc }
722 See header of drivers/s390/block/dasd_devmap.c.
724 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
725 (one device per port)
726 Format: <port#>,<type>
727 See also Documentation/input/joystick-parport.txt
729 ddebug_query= [KNL,DYNAMIC_DEBUG] Enable debug messages at early boot
731 Documentation/admin-guide/dynamic-debug-howto.rst for
732 details. Deprecated, see dyndbg.
734 debug [KNL] Enable kernel debugging (events log level).
737 [KNL] verbose self-tests
739 Print debugging info while doing the locking API
741 We default to 0 (no extra messages), setting it to
742 1 will print _a lot_ more information - normally
743 only useful to kernel developers.
745 debug_objects [KNL] Enable object debugging
748 [KNL] Disable object debugging
750 debug_guardpage_minorder=
751 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
752 parameter allows control of the order of pages that will
753 be intentionally kept free (and hence protected) by the
754 buddy allocator. Bigger value increase the probability
755 of catching random memory corruption, but reduce the
756 amount of memory for normal system use. The maximum
757 possible value is MAX_ORDER/2. Setting this parameter
758 to 1 or 2 should be enough to identify most random
759 memory corruption problems caused by bugs in kernel or
760 driver code when a CPU writes to (or reads from) a
761 random memory location. Note that there exists a class
762 of memory corruptions problems caused by buggy H/W or
763 F/W or by drivers badly programing DMA (basically when
764 memory is written at bus level and the CPU MMU is
765 bypassed) which are not detectable by
766 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
767 tracking down these problems.
770 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
771 parameter enables the feature at boot time. In
772 default, it is disabled. We can avoid allocating huge
773 chunk of memory for debug pagealloc if we don't enable
774 it at boot time and the system will work mostly same
775 with the kernel built without CONFIG_DEBUG_PAGEALLOC.
776 on: enable the feature
778 debugpat [X86] Enable PAT debugging
780 decnet.addr= [HW,NET]
781 Format: <area>[,<node>]
782 See also Documentation/networking/decnet.txt.
785 [same as hugepagesz=] The size of the default
786 HugeTLB page size. This is the size represented by
787 the legacy /proc/ hugepages APIs, used for SHM, and
788 default size when mounting hugetlbfs filesystems.
789 Defaults to the default architecture's huge page size
793 Set number of hash buckets for dentry cache.
795 disable_1tb_segments [PPC]
796 Disables the use of 1TB hash page table segments. This
797 causes the kernel to fall back to 256MB segments which
798 can be useful when debugging issues that require an SLB
802 See Documentation/networking/ipv6.txt.
805 Disable RADIX MMU mode on POWER9
807 disable_cpu_apicid= [X86,APIC,SMP]
809 The number of initial APIC ID for the
810 corresponding CPU to be disabled at boot,
811 mostly used for the kdump 2nd kernel to
812 disable BSP to wake up multiple CPUs without
813 causing system reset or hang due to sending
816 disable_ddw [PPC/PSERIES]
817 Disable Dynamic DMA Window support. Use this if
818 to workaround buggy firmware.
821 See Documentation/networking/ipv6.txt.
823 disable_mtrr_cleanup [X86]
824 The kernel tries to adjust MTRR layout from continuous
825 to discrete, to make X server driver able to add WB
826 entry later. This parameter disables that.
828 disable_mtrr_trim [X86, Intel and AMD only]
829 By default the kernel will trim any uncacheable
830 memory out of your available memory pool based on
831 MTRR settings. This parameter disables that behavior,
832 possibly causing your machine to run very slowly.
834 disable_timer_pin_1 [X86]
835 Disable PIN 1 of APIC timer
836 Can be useful to work around chipset bugs.
838 dis_ucode_ldr [X86] Disable the microcode loader.
840 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
841 this option disables the debugging code at boot.
843 dma_debug_entries=<number>
844 This option allows to tune the number of preallocated
845 entries for DMA-API debugging code. One entry is
846 required per DMA-API allocation. Use this if the
847 DMA-API debugging code disables itself because the
848 architectural default is too low.
850 dma_debug_driver=<driver_name>
851 With this option the DMA-API debugging driver
852 filter feature can be enabled at boot time. Just
853 pass the driver to filter for as the parameter.
854 The filter can be disabled or changed to another
855 driver later using sysfs.
857 drm_kms_helper.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
858 Broken monitors, graphic adapters, KVMs and EDIDless
859 panels may send no or incorrect EDID data sets.
860 This parameter allows to specify an EDID data sets
861 in the /lib/firmware directory that are used instead.
862 Generic built-in EDID data sets are used, if one of
863 edid/1024x768.bin, edid/1280x1024.bin,
864 edid/1680x1050.bin, or edid/1920x1080.bin is given
865 and no file with the same name exists. Details and
866 instructions how to build your own EDID data are
867 available in Documentation/EDID/HOWTO.txt. An EDID
868 data set will only be used for a particular connector,
869 if its name and a colon are prepended to the EDID
870 name. Each connector may use a unique EDID data
871 set by separating the files with a comma. An EDID
872 data set with no connector name will be used for
873 any connectors not explicitly specified.
878 Format: {"off" | "known"}
879 Control how the dt_cpu_ftrs device-tree binding is
880 used for CPU feature discovery and setup (if it
882 off: Do not use it, fall back to legacy cpu table.
883 known: Do not pass through unknown features to guests
884 or userspace, only those that the kernel is aware of.
886 dump_apple_properties [X86]
887 Dump name and content of EFI device properties on
888 x86 Macs. Useful for driver authors to determine
889 what data is available or for reverse-engineering.
891 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
892 module.dyndbg[="val"]
893 Enable debug messages at boot time. See
894 Documentation/admin-guide/dynamic-debug-howto.rst
897 nompx [X86] Disables Intel Memory Protection Extensions.
898 See Documentation/x86/intel_mpx.txt for more
899 information about the feature.
901 nopku [X86] Disable Memory Protection Keys CPU feature found
904 module.async_probe [KNL]
905 Enable asynchronous probe on this module.
907 early_ioremap_debug [KNL]
908 Enable debug messages in early_ioremap support. This
909 is useful for tracking down temporary early mappings
910 which are not unmapped.
912 earlycon= [KNL] Output early console device and options.
914 When used with no options, the early console is
915 determined by the stdout-path property in device
918 cdns,<addr>[,options]
919 Start an early, polled-mode console on a Cadence
920 (xuartps) serial port at the specified address. Only
921 supported option is baud rate. If baud rate is not
922 specified, the serial port must already be setup and
925 uart[8250],io,<addr>[,options]
926 uart[8250],mmio,<addr>[,options]
927 uart[8250],mmio32,<addr>[,options]
928 uart[8250],mmio32be,<addr>[,options]
929 uart[8250],0x<addr>[,options]
930 Start an early, polled-mode console on the 8250/16550
931 UART at the specified I/O port or MMIO address.
932 MMIO inter-register address stride is either 8-bit
933 (mmio) or 32-bit (mmio32 or mmio32be).
934 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
935 to be equivalent to 'mmio'. 'options' are specified
936 in the same format described for "console=ttyS<n>"; if
937 unspecified, the h/w is not initialized.
941 Start an early, polled-mode console on a pl011 serial
942 port at the specified address. The pl011 serial port
943 must already be setup and configured. Options are not
944 yet supported. If 'mmio32' is specified, then only
945 the driver will use only 32-bit accessors to read/write
946 the device registers.
949 Start an early, polled-mode console on a meson serial
950 port at the specified address. The serial port must
951 already be setup and configured. Options are not yet
955 Start an early, polled-mode console on an msm serial
956 port at the specified address. The serial port
957 must already be setup and configured. Options are not
961 Start an early, polled-mode console on an msm serial
962 dm port at the specified address. The serial port
963 must already be setup and configured. Options are not
967 Start an early, polled-mode console on a serial port
968 of an Actions Semi SoC, such as S500 or S900, at the
969 specified address. The serial port must already be
970 setup and configured. Options are not yet supported.
972 smh Use ARM semihosting calls for early console.
980 Use early console provided by serial driver available
981 on Samsung SoCs, requires selecting proper type and
982 a correct base address of the selected UART port. The
983 serial port must already be setup and configured.
984 Options are not yet supported.
987 Start an early, polled-mode console on a lantiq serial
988 (lqasc) port at the specified address. The serial port
989 must already be setup and configured. Options are not
994 Use early console provided by Freescale LP UART driver
995 found on Freescale Vybrid and QorIQ LS1021A processors.
996 A valid base address must be provided, and the serial
997 port must already be setup and configured.
1000 Start an early, polled-mode console on the
1001 Armada 3700 serial port at the specified
1002 address. The serial port must already be setup
1003 and configured. Options are not yet supported.
1005 earlyprintk= [X86,SH,BLACKFIN,ARM,M68k,S390]
1010 earlyprintk=serial[,ttySn[,baudrate]]
1011 earlyprintk=serial[,0x...[,baudrate]]
1012 earlyprintk=ttySn[,baudrate]
1013 earlyprintk=dbgp[debugController#]
1014 earlyprintk=pciserial,bus:device.function[,baudrate]
1015 earlyprintk=xdbc[xhciController#]
1017 earlyprintk is useful when the kernel crashes before
1018 the normal console is initialized. It is not enabled by
1019 default because it has some cosmetic problems.
1021 Append ",keep" to not disable it when the real console
1024 Only one of vga, efi, serial, or usb debug port can
1027 Currently only ttyS0 and ttyS1 may be specified by
1028 name. Other I/O ports may be explicitly specified
1029 on some architectures (x86 and arm at least) by
1030 replacing ttySn with an I/O port address, like this:
1031 earlyprintk=serial,0x1008,115200
1032 You can find the port for a given device in
1033 /proc/tty/driver/serial:
1034 2: uart:ST16650V2 port:00001008 irq:18 ...
1036 Interaction with the standard serial driver is not
1039 The VGA and EFI output is eventually overwritten by
1042 The xen output can only be used by Xen PV guests.
1044 The sclp output can only be used on s390.
1046 edac_report= [HW,EDAC] Control how to report EDAC event
1047 Format: {"on" | "off" | "force"}
1048 on: enable EDAC to report H/W event. May be overridden
1049 by other higher priority error reporting module.
1050 off: disable H/W event reporting through EDAC.
1051 force: enforce the use of EDAC to report H/W event.
1054 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1057 This is designed to be used in conjunction with
1058 the boot argument: earlyprintk=vga
1061 Format: {"off" | "on" | "skip[mbr]"}
1064 Format: { "old_map", "nochunk", "noruntime", "debug" }
1065 old_map [X86-64]: switch to the old ioremap-based EFI
1066 runtime services mapping. 32-bit still uses this one by
1068 nochunk: disable reading files in "chunks" in the EFI
1069 boot stub, as chunking can cause problems with some
1070 firmware implementations.
1071 noruntime : disable EFI runtime services support
1072 debug: enable misc debug output
1074 efi_no_storage_paranoia [EFI; X86]
1075 Using this parameter you can use more than 50% of
1076 your efi variable storage. Use this parameter only if
1077 you are really sure that your UEFI does sane gc and
1078 fulfills the spec otherwise your board may brick.
1080 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1081 Add arbitrary attribute to specific memory range by
1082 updating original EFI memory map.
1083 Region of memory which aa attribute is added to is
1085 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1086 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1087 attribute is added to range 0x100000000-0x180000000 and
1088 0x10a0000000-0x1120000000.
1090 Using this parameter you can do debugging of EFI memmap
1091 related feature. For example, you can do debugging of
1092 Address Range Mirroring feature even if your box
1095 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1096 that is to be dynamically loaded by Linux. If there are
1097 multiple variables with the same name but with different
1098 vendor GUIDs, all of them will be loaded. See
1099 Documentation/acpi/ssdt-overlays.txt for details.
1102 eisa_irq_edge= [PARISC,HW]
1103 See header of drivers/parisc/eisa.c.
1106 See comment before function elanfreq_setup() in
1107 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1110 Format: {"cfq" | "deadline" | "noop"}
1111 See Documentation/block/cfq-iosched.txt and
1112 Documentation/block/deadline-iosched.txt for details.
1114 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1115 Specifies physical address of start of kernel core
1116 image elf header and optionally the size. Generally
1117 kexec loader will pass this option to capture kernel.
1118 See Documentation/kdump/kdump.txt for details.
1120 enable_mtrr_cleanup [X86]
1121 The kernel tries to adjust MTRR layout from continuous
1122 to discrete, to make X server driver able to add WB
1123 entry later. This parameter enables that.
1125 enable_timer_pin_1 [X86]
1126 Enable PIN 1 of APIC timer
1127 Can be useful to work around chipset bugs
1128 (in particular on some ATI chipsets).
1129 The kernel tries to set a reasonable default.
1131 enforcing [SELINUX] Set initial enforcing status.
1133 See security/selinux/Kconfig help text.
1134 0 -- permissive (log only, no denials).
1135 1 -- enforcing (deny and log).
1137 Value can be changed at runtime via /selinux/enforce.
1140 Disable Error Record Serialization Table (ERST)
1143 ether= [HW,NET] Ethernet cards parameters
1144 This option is obsoleted by the "netdev=" option, which
1145 has equivalent usage. See its documentation for details.
1149 Permit 'security.evm' to be updated regardless of
1150 current integrity status.
1154 fail_make_request=[KNL]
1155 General fault injection mechanism.
1156 Format: <interval>,<probability>,<space>,<times>
1157 See also Documentation/fault-injection/.
1160 See Documentation/blockdev/floppy.txt.
1162 force_pal_cache_flush
1163 [IA-64] Avoid check_sal_cache_flush which may hang on
1164 buggy SAL_CACHE_FLUSH implementations. Using this
1165 parameter will force ia64_sal_cache_flush to call
1166 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1169 Forcefully enable Physical Address Extension (PAE).
1170 Many Pentium M systems disable PAE but may have a
1171 functionally usable PAE implementation.
1172 Warning: use of this parameter will taint the kernel
1173 and may cause unknown problems.
1176 [FTRACE] will set and start the specified tracer
1177 as early as possible in order to facilitate early
1180 ftrace_dump_on_oops[=orig_cpu]
1181 [FTRACE] will dump the trace buffers on oops.
1182 If no parameter is passed, ftrace will dump
1183 buffers of all CPUs, but if you pass orig_cpu, it will
1184 dump only the buffer of the CPU that triggered the
1187 ftrace_filter=[function-list]
1188 [FTRACE] Limit the functions traced by the function
1189 tracer at boot up. function-list is a comma separated
1190 list of functions. This list can be changed at run
1191 time by the set_ftrace_filter file in the debugfs
1194 ftrace_notrace=[function-list]
1195 [FTRACE] Do not trace the functions specified in
1196 function-list. This list can be changed at run time
1197 by the set_ftrace_notrace file in the debugfs
1200 ftrace_graph_filter=[function-list]
1201 [FTRACE] Limit the top level callers functions traced
1202 by the function graph tracer at boot up.
1203 function-list is a comma separated list of functions
1204 that can be changed at run time by the
1205 set_graph_function file in the debugfs tracing directory.
1207 ftrace_graph_notrace=[function-list]
1208 [FTRACE] Do not trace from the functions specified in
1209 function-list. This list is a comma separated list of
1210 functions that can be changed at run time by the
1211 set_graph_notrace file in the debugfs tracing directory.
1213 ftrace_graph_max_depth=<uint>
1214 [FTRACE] Used with the function graph tracer. This is
1215 the max depth it will trace into a function. This value
1216 can be changed at run time by the max_graph_depth file
1217 in the tracefs tracing directory. default: 0 (no limit)
1220 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1221 support via parallel port (up to 5 devices per port)
1222 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1223 See also Documentation/input/joystick-parport.txt
1227 gart_fix_e820= [X86_64] disable the fix e820 for K8 GART
1231 gcov_persist= [GCOV] When non-zero (default), profiling data for
1232 kernel modules is saved and remains accessible via
1233 debugfs, even when the module is unloaded/reloaded.
1234 When zero, profiling data is discarded and associated
1235 debugfs files are removed at module unload time.
1237 goldfish [X86] Enable the goldfish android emulator platform.
1238 Don't use this when you are not running on the
1241 gpt [EFI] Forces disk with valid GPT signature but
1242 invalid Protective MBR to be treated as GPT. If the
1243 primary GPT is corrupted, it enables the backup/alternate
1244 GPT to be used instead.
1246 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1247 the "Enable 0" bit of the configuration register.
1250 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1251 the "Enable 0" bit of the configuration register.
1254 grcan.select= [HW] Select which physical interface to use.
1257 grcan.txsize= [HW] Sets the size of the tx buffer.
1258 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1260 grcan.rxsize= [HW] Sets the size of the rx buffer.
1261 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1264 gpio-mockup.gpio_mockup_ranges
1265 [HW] Sets the ranges of gpiochip of for this device.
1266 Format: <start1>,<end1>,<start2>,<end2>...
1268 hardlockup_all_cpu_backtrace=
1269 [KNL] Should the hard-lockup detector generate
1270 backtraces on all cpus.
1273 hashdist= [KNL,NUMA] Large hashes allocated during boot
1274 are distributed across NUMA nodes. Defaults on
1275 for 64-bit NUMA, off otherwise.
1276 Format: 0 | 1 (for off | on)
1278 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1280 hd= [EIDE] (E)IDE hard drive subsystem geometry
1281 Format: <cyl>,<head>,<sect>
1284 Disable Hardware Error Source Table (HEST) support;
1285 corresponding firmware-first mode error processing
1286 logic will be disabled.
1288 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1289 size of <nn>. This works even on boxes that have no
1290 highmem otherwise. This also works to reduce highmem
1291 size on bigger boxes.
1293 highres= [KNL] Enable/disable high resolution timer mode.
1294 Valid parameters: "on", "off"
1298 See Documentation/isdn/README.HiSax.
1302 hpet= [X86-32,HPET] option to control HPET usage
1303 Format: { enable (default) | disable | force |
1305 disable: disable HPET and use PIT instead
1306 force: allow force enabled of undocumented chips (ICH4,
1308 verbose: show contents of HPET registers during setup
1310 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1311 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1313 hugepages= [HW,X86-32,IA-64] HugeTLB pages to allocate at boot.
1314 hugepagesz= [HW,IA-64,PPC,X86-64] The size of the HugeTLB pages.
1315 On x86-64 and powerpc, this option can be specified
1316 multiple times interleaved with hugepages= to reserve
1317 huge pages of different sizes. Valid pages sizes on
1318 x86-64 are 2M (when the CPU supports "pse") and 1G
1319 (when the CPU supports the "pdpe1gb" cpuinfo flag).
1321 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1322 terminal devices. Valid values: 0..8
1323 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1324 If specified, z/VM IUCV HVC accepts connections
1325 from listed z/VM user IDs only.
1327 hwthread_map= [METAG] Comma-separated list of Linux cpu id to
1328 hardware thread id mappings.
1329 Format: <cpu>:<hwthread>
1332 Do not unregister boot console at start. This is only
1333 useful for debugging when something happens in the window
1334 between unregistering the boot console and initializing
1337 i2c_bus= [HW] Override the default board specific I2C bus speed
1338 or register an additional I2C bus that is not
1339 registered from board initialization code.
1343 i8042.debug [HW] Toggle i8042 debug mode
1344 i8042.unmask_kbd_data
1345 [HW] Enable printing of interrupt data from the KBD port
1346 (disabled by default, and as a pre-condition
1347 requires that i8042.debug=1 be enabled)
1348 i8042.direct [HW] Put keyboard port into non-translated mode
1349 i8042.dumbkbd [HW] Pretend that controller can only read data from
1350 keyboard and cannot control its state
1351 (Don't attempt to blink the leds)
1352 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1353 i8042.nokbd [HW] Don't check/create keyboard port
1354 i8042.noloop [HW] Disable the AUX Loopback command while probing
1356 i8042.nomux [HW] Don't check presence of an active multiplexing
1358 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1360 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1361 i8042.reset [HW] Reset the controller during init, cleanup and
1362 suspend-to-ram transitions, only during s2r
1363 transitions, or never reset
1364 Format: { 1 | Y | y | 0 | N | n }
1365 1, Y, y: always reset controller
1366 0, N, n: don't ever reset controller
1367 Default: only on s2r transitions on x86; most other
1368 architectures force reset to be always executed
1369 i8042.unlock [HW] Unlock (ignore) the keylock
1370 i8042.kbdreset [HW] Reset device connected to KBD port
1374 i8k.ignore_dmi [HW] Continue probing hardware even if DMI data
1375 indicates that the driver is running on unsupported
1377 i8k.force [HW] Activate i8k driver even if SMM BIOS signature
1378 does not match list of supported models.
1380 [HW] Report power status in /proc/i8k
1381 (disabled by default)
1382 i8k.restricted [HW] Allow controlling fans only if SYS_ADMIN
1385 i915.invert_brightness=
1386 [DRM] Invert the sense of the variable that is used to
1387 set the brightness of the panel backlight. Normally a
1388 brightness value of 0 indicates backlight switched off,
1389 and the maximum of the brightness value sets the backlight
1390 to maximum brightness. If this parameter is set to 0
1391 (default) and the machine requires it, or this parameter
1392 is set to 1, a brightness value of 0 sets the backlight
1393 to maximum brightness, and the maximum of the brightness
1394 value switches the backlight off.
1395 -1 -- never invert brightness
1396 0 -- machine default
1397 1 -- force brightness inversion
1400 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1402 ide-core.nodma= [HW] (E)IDE subsystem
1403 Format: =0.0 to prevent dma on hda, =0.1 hdb =1.0 hdc
1404 .vlb_clock .pci_clock .noflush .nohpa .noprobe .nowerr
1405 .cdrom .chs .ignore_cable are additional options
1406 See Documentation/ide/ide.txt.
1408 ide-generic.probe-mask= [HW] (E)IDE subsystem
1410 Probe mask for legacy ISA IDE ports. Depending on
1411 platform up to 6 ports are supported, enabled by
1412 setting corresponding bits in the mask to 1. The
1413 default value is 0x0, which has a special meaning.
1414 On systems that have PCI, it triggers scanning the
1415 PCI bus for the first and the second port, which
1416 are then probed. On systems without PCI the value
1417 of 0x0 enables probing the two first ports as if it
1420 ide-pci-generic.all-generic-ide [HW] (E)IDE subsystem
1421 Claim all unknown PCI IDE storage controllers.
1424 Format: idle=poll, idle=halt, idle=nomwait
1425 Poll forces a polling idle loop that can slightly
1426 improve the performance of waking up a idle CPU, but
1427 will use a lot of power and make the system run hot.
1429 idle=halt: Halt is forced to be used for CPU idle.
1430 In such case C2/C3 won't be used again.
1431 idle=nomwait: Disable mwait for CPU C-states
1433 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1434 Format: { strict | legacy | 2008 | relaxed }
1437 Choose which programs will be accepted for execution
1438 based on the IEEE 754 NaN encoding(s) supported by
1439 the FPU and the NaN encoding requested with the value
1440 of an ELF file header flag individually set by each
1441 binary. Hardware implementations are permitted to
1442 support either or both of the legacy and the 2008 NaN
1445 Available settings are as follows:
1446 strict accept binaries that request a NaN encoding
1447 supported by the FPU
1448 legacy only accept legacy-NaN binaries, if supported
1450 2008 only accept 2008-NaN binaries, if supported
1452 relaxed accept any binaries regardless of whether
1453 supported by the FPU
1455 The FPU emulator is always able to support both NaN
1456 encodings, so if no FPU hardware is present or it has
1457 been disabled with 'nofpu', then the settings of
1458 'legacy' and '2008' strap the emulator accordingly,
1459 'relaxed' straps the emulator for both legacy-NaN and
1460 2008-NaN, whereas 'strict' enables legacy-NaN only on
1461 legacy processors and both NaN encodings on MIPS32 or
1464 The setting for ABS.fmt/NEG.fmt instruction execution
1465 mode generally follows that for the NaN encoding,
1466 except where unsupported by hardware.
1468 ignore_loglevel [KNL]
1469 Ignore loglevel setting - this will print /all/
1470 kernel messages to the console. Useful for debugging.
1471 We also add it as printk module parameter, so users
1472 could change it dynamically, usually by
1473 /sys/module/printk/parameters/ignore_loglevel.
1476 Ignore RLIMIT_DATA setting for data mappings,
1477 print warning at first misuse. Can be changed via
1478 /sys/module/kernel/parameters/ignore_rlimit_data.
1480 ihash_entries= [KNL]
1481 Set number of hash buckets for inode cache.
1483 ima_appraise= [IMA] appraise integrity measurements
1484 Format: { "off" | "enforce" | "fix" | "log" }
1487 ima_appraise_tcb [IMA]
1488 The builtin appraise policy appraises all files
1491 ima_canonical_fmt [IMA]
1492 Use the canonical format for the binary runtime
1493 measurements, instead of host native format.
1496 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1500 The list of supported hash algorithms is defined
1501 in crypto/hash_info.h.
1504 The builtin policies to load during IMA setup.
1505 Format: "tcb | appraise_tcb | secure_boot"
1507 The "tcb" policy measures all programs exec'd, files
1508 mmap'd for exec, and all files opened with the read
1509 mode bit set by either the effective uid (euid=0) or
1512 The "appraise_tcb" policy appraises the integrity of
1513 all files owned by root. (This is the equivalent
1514 of ima_appraise_tcb.)
1516 The "secure_boot" policy appraises the integrity
1517 of files (eg. kexec kernel image, kernel modules,
1518 firmware, policy, etc) based on file signatures.
1520 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
1521 Load a policy which meets the needs of the Trusted
1522 Computing Base. This means IMA will measure all
1523 programs exec'd, files mmap'd for exec, and all files
1524 opened for read by uid=0.
1527 Select one of defined IMA measurements template formats.
1528 Formats: { "ima" | "ima-ng" | "ima-sig" }
1532 [IMA] Define a custom template format.
1533 Format: { "field1|...|fieldN" }
1535 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
1536 Format: <min_file_size>
1537 Set the minimal file size for using asynchronous hash.
1538 If left unspecified, ahash usage is disabled.
1540 ahash performance varies for different data sizes on
1541 different crypto accelerators. This option can be used
1542 to achieve the best performance for a particular HW.
1544 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
1546 Set hashing buffer size. Default: 4k.
1548 ahash performance varies for different chunk sizes on
1549 different crypto accelerators. This option can be used
1550 to achieve best performance for particular HW.
1554 Run specified binary instead of /sbin/init as init
1557 initcall_debug [KNL] Trace initcalls as they are executed. Useful
1558 for working out where the kernel is dying during
1561 initcall_blacklist= [KNL] Do not execute a comma-separated list of
1562 initcall functions. Useful for debugging built-in
1563 modules and initcalls.
1565 initrd= [BOOT] Specify the location of the initial ramdisk
1567 init_pkru= [x86] Specify the default memory protection keys rights
1568 register contents for all processes. 0x55555554 by
1569 default (disallow access to all but pkey 0). Can
1570 override in debugfs after boot.
1572 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
1575 int_pln_enable [x86] Enable power limit notification interrupt
1577 integrity_audit=[IMA]
1578 Format: { "0" | "1" }
1579 0 -- basic integrity auditing messages. (Default)
1580 1 -- additional integrity auditing messages.
1582 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
1584 Enable intel iommu driver.
1586 Disable intel iommu driver.
1587 igfx_off [Default Off]
1588 By default, gfx is mapped as normal device. If a gfx
1589 device has a dedicated DMAR unit, the DMAR unit is
1590 bypassed by not enabling DMAR with this option. In
1591 this case, gfx device will use physical address for
1594 With this option iommu will not optimize to look
1595 for io virtual address below 32-bit forcing dual
1596 address cycle on pci bus for cards supporting greater
1597 than 32-bit addressing. The default is to look
1598 for translation below 32-bit and if not available
1599 then look in the higher range.
1600 strict [Default Off]
1601 With this option on every unmap_single operation will
1602 result in a hardware IOTLB flush operation as opposed
1603 to batching them for performance.
1604 sp_off [Default Off]
1605 By default, super page will be supported if Intel IOMMU
1606 has the capability. With this option, super page will
1608 ecs_off [Default Off]
1609 By default, extended context tables will be supported if
1610 the hardware advertises that it has support both for the
1611 extended tables themselves, and also PASID support. With
1612 this option set, extended tables will not be used even
1613 on hardware which claims to support them.
1614 tboot_noforce [Default Off]
1615 Do not force the Intel IOMMU enabled under tboot.
1616 By default, tboot will force Intel IOMMU on, which
1617 could harm performance of some high-throughput
1618 devices like 40GBit network cards, even if identity
1620 Note that using this option lowers the security
1621 provided by tboot because it makes the system
1622 vulnerable to DMA attacks.
1624 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
1625 0 disables intel_idle and fall back on acpi_idle.
1626 1 to 9 specify maximum depth of C-state.
1630 Do not enable intel_pstate as the default
1631 scaling driver for the supported processors
1633 Use intel_pstate as a scaling driver, but configure it
1634 to work with generic cpufreq governors (instead of
1635 enabling its internal governor). This mode cannot be
1636 used along with the hardware-managed P-states (HWP)
1639 Enable intel_pstate on systems that prohibit it by default
1640 in favor of acpi-cpufreq. Forcing the intel_pstate driver
1641 instead of acpi-cpufreq may disable platform features, such
1642 as thermal controls and power capping, that rely on ACPI
1643 P-States information being indicated to OSPM and therefore
1644 should be used with caution. This option does not work with
1645 processors that aren't supported by the intel_pstate driver
1646 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
1648 Do not enable hardware P state control (HWP)
1651 Only load intel_pstate on systems which support
1652 hardware P state control (HWP) if available.
1654 Enforce ACPI _PPC performance limits. If the Fixed ACPI
1655 Description Table, specifies preferred power management
1656 profile as "Enterprise Server" or "Performance Server",
1657 then this feature is turned on by default.
1659 Allow per-logical-CPU P-State performance control limits using
1660 cpufreq sysfs interface
1662 intremap= [X86-64, Intel-IOMMU]
1663 on enable Interrupt Remapping (default)
1664 off disable Interrupt Remapping
1665 nosid disable Source ID checking
1667 BIOS x2APIC opt-out request will be ignored
1668 nopost disable Interrupt Posting
1670 iomem= Disable strict checking of access to MMIO memory
1671 strict regions from userspace.
1686 nobypass [PPC/POWERNV]
1687 Disable IOMMU bypass, using IOMMU for PCI devices.
1690 [ARM64] Configure DMA to bypass the IOMMU by default.
1691 Format: { "0" | "1" }
1692 0 - Use IOMMU translation for DMA.
1693 1 - Bypass the IOMMU for DMA.
1694 unset - Use IOMMU translation for DMA.
1696 io7= [HW] IO7 for Marvel based alpha systems
1697 See comment before marvel_specify_io7 in
1698 arch/alpha/kernel/core_marvel.c.
1700 io_delay= [X86] I/O delay method
1702 Standard port 0x80 based delay
1704 Alternate port 0xed based delay (needed on some systems)
1706 Simple two microseconds delay
1711 See Documentation/filesystems/nfs/nfsroot.txt.
1713 irqaffinity= [SMP] Set the default irq affinity mask
1714 The argument is a cpu list, as described above.
1717 When an interrupt is not handled search all handlers
1718 for it. Intended to get systems with badly broken
1722 When an interrupt is not handled search all handlers
1723 for it. Also check all handlers each timer
1724 interrupt. Intended to get systems with badly broken
1728 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
1730 isolcpus= [KNL,SMP] Isolate CPUs from the general scheduler.
1731 The argument is a cpu list, as described above.
1733 This option can be used to specify one or more CPUs
1734 to isolate from the general SMP balancing and scheduling
1735 algorithms. You can move a process onto or off an
1736 "isolated" CPU via the CPU affinity syscalls or cpuset.
1737 <cpu number> begins at 0 and the maximum value is
1738 "number of CPUs in system - 1".
1740 This option is the preferred way to isolate CPUs. The
1741 alternative -- manually setting the CPU mask of all
1742 tasks in the system -- can cause problems and
1743 suboptimal load balancer performance.
1747 ivrs_ioapic [HW,X86_64]
1748 Provide an override to the IOAPIC-ID<->DEVICE-ID
1749 mapping provided in the IVRS ACPI table. For
1750 example, to map IOAPIC-ID decimal 10 to
1751 PCI device 00:14.0 write the parameter as:
1752 ivrs_ioapic[10]=00:14.0
1754 ivrs_hpet [HW,X86_64]
1755 Provide an override to the HPET-ID<->DEVICE-ID
1756 mapping provided in the IVRS ACPI table. For
1757 example, to map HPET-ID decimal 0 to
1758 PCI device 00:14.0 write the parameter as:
1759 ivrs_hpet[0]=00:14.0
1761 ivrs_acpihid [HW,X86_64]
1762 Provide an override to the ACPI-HID:UID<->DEVICE-ID
1763 mapping provided in the IVRS ACPI table. For
1764 example, to map UART-HID:UID AMD0020:0 to
1765 PCI device 00:14.5 write the parameter as:
1766 ivrs_acpihid[00:14.5]=AMD0020:0
1768 js= [HW,JOY] Analog joystick
1769 See Documentation/input/joystick.txt.
1772 When CONFIG_RANDOMIZE_BASE is set, this disables
1773 kernel and module base offset ASLR (Address Space
1774 Layout Randomization).
1777 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
1778 report on every invalid memory access. Without this
1779 parameter KASAN will print report only for the first
1784 kernelcore= [KNL,X86,IA-64,PPC]
1785 Format: nn[KMGTPE] | "mirror"
1787 specifies the amount of memory usable by the kernel
1788 for non-movable allocations. The requested amount is
1789 spread evenly throughout all nodes in the system. The
1790 remaining memory in each node is used for Movable
1791 pages. In the event, a node is too small to have both
1792 kernelcore and Movable pages, kernelcore pages will
1793 take priority and other nodes will have a larger number
1794 of Movable pages. The Movable zone is used for the
1795 allocation of pages that may be reclaimed or moved
1796 by the page migration subsystem. This means that
1797 HugeTLB pages may not be allocated from this zone.
1798 Note that allocations like PTEs-from-HighMem still
1799 use the HighMem zone if it exists, and the Normal
1800 zone if it does not.
1802 Instead of specifying the amount of memory (nn[KMGTPE]),
1803 you can specify "mirror" option. In case "mirror"
1804 option is specified, mirrored (reliable) memory is used
1805 for non-movable allocations and remaining memory is used
1806 for Movable pages. nn[KMGTPE] and "mirror" are exclusive,
1807 so you can NOT specify nn[KMGTPE] and "mirror" at the same
1810 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
1811 Format: <Controller#>[,poll interval]
1812 The controller # is the number of the ehci usb debug
1813 port as it is probed via PCI. The poll interval is
1814 optional and is the number seconds in between
1815 each poll cycle to the debug port in case you need
1816 the functionality for interrupting the kernel with
1817 gdb or control-c on the dbgp connection. When
1818 not using this parameter you use sysrq-g to break into
1819 the kernel debugger.
1821 kgdboc= [KGDB,HW] kgdb over consoles.
1822 Requires a tty driver that supports console polling,
1823 or a supported polling keyboard driver (non-usb).
1824 Serial only format: <serial_device>[,baud]
1825 keyboard only format: kbd
1826 keyboard and serial format: kbd,<serial_device>[,baud]
1827 Optional Kernel mode setting:
1828 kms, kbd format: kms,kbd
1829 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
1831 kgdbwait [KGDB] Stop kernel execution and enter the
1832 kernel debugger at the earliest opportunity.
1834 kmac= [MIPS] korina ethernet MAC address.
1835 Configure the RouterBoard 532 series on-chip
1836 Ethernet adapter MAC address.
1838 kmemleak= [KNL] Boot-time kmemleak enable/disable
1839 Valid arguments: on, off
1841 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
1844 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
1845 Default is 0 (don't ignore, but inject #GP)
1847 kvm.mmu_audit= [KVM] This is a R/W parameter which allows audit
1851 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM.
1852 Default is 1 (enabled)
1854 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU)
1856 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode.
1858 kvm-arm.vgic_v3_group0_trap=
1859 [KVM,ARM] Trap guest accesses to GICv3 group-0
1862 kvm-arm.vgic_v3_group1_trap=
1863 [KVM,ARM] Trap guest accesses to GICv3 group-1
1866 kvm-arm.vgic_v3_common_trap=
1867 [KVM,ARM] Trap guest accesses to GICv3 common
1870 kvm-intel.ept= [KVM,Intel] Disable extended page tables
1871 (virtualized MMU) support on capable Intel chips.
1872 Default is 1 (enabled)
1874 kvm-intel.emulate_invalid_guest_state=
1875 [KVM,Intel] Enable emulation of invalid guest states
1876 Default is 0 (disabled)
1878 kvm-intel.flexpriority=
1879 [KVM,Intel] Disable FlexPriority feature (TPR shadow).
1880 Default is 1 (enabled)
1883 [KVM,Intel] Enable VMX nesting (nVMX).
1884 Default is 0 (disabled)
1886 kvm-intel.unrestricted_guest=
1887 [KVM,Intel] Disable unrestricted guest feature
1888 (virtualized real and unpaged mode) on capable
1889 Intel chips. Default is 1 (enabled)
1891 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
1894 Valid arguments: never, cond, always
1896 always: L1D cache flush on every VMENTER.
1897 cond: Flush L1D on VMENTER only when the code between
1898 VMEXIT and VMENTER can leak host memory.
1899 never: Disables the mitigation
1901 Default is cond (do L1 cache flush in specific instances)
1903 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification
1904 feature (tagged TLBs) on capable Intel chips.
1905 Default is 1 (enabled)
1907 l1tf= [X86] Control mitigation of the L1TF vulnerability on
1910 The kernel PTE inversion protection is unconditionally
1911 enabled and cannot be disabled.
1914 Provides all available mitigations for the
1915 L1TF vulnerability. Disables SMT and
1916 enables all mitigations in the
1917 hypervisors, i.e. unconditional L1D flush.
1919 SMT control and L1D flush control via the
1920 sysfs interface is still possible after
1921 boot. Hypervisors will issue a warning
1922 when the first VM is started in a
1923 potentially insecure configuration,
1924 i.e. SMT enabled or L1D flush disabled.
1927 Same as 'full', but disables SMT and L1D
1928 flush runtime control. Implies the
1929 'nosmt=force' command line option.
1930 (i.e. sysfs control of SMT is disabled.)
1933 Leaves SMT enabled and enables the default
1934 hypervisor mitigation, i.e. conditional
1937 SMT control and L1D flush control via the
1938 sysfs interface is still possible after
1939 boot. Hypervisors will issue a warning
1940 when the first VM is started in a
1941 potentially insecure configuration,
1942 i.e. SMT enabled or L1D flush disabled.
1946 Disables SMT and enables the default
1947 hypervisor mitigation.
1949 SMT control and L1D flush control via the
1950 sysfs interface is still possible after
1951 boot. Hypervisors will issue a warning
1952 when the first VM is started in a
1953 potentially insecure configuration,
1954 i.e. SMT enabled or L1D flush disabled.
1957 Same as 'flush', but hypervisors will not
1958 warn when a VM is started in a potentially
1959 insecure configuration.
1962 Disables hypervisor mitigations and doesn't
1967 For details see: Documentation/admin-guide/l1tf.rst
1973 lapic [X86-32,APIC] Enable the local APIC even if BIOS
1976 lapic= [x86,APIC] "notscdeadline" Do not use TSC deadline
1977 value for LAPIC timer one-shot implementation. Default
1978 back to the programmable timer unit in the LAPIC.
1980 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
1983 libata.dma= [LIBATA] DMA control
1984 libata.dma=0 Disable all PATA and SATA DMA
1985 libata.dma=1 PATA and SATA Disk DMA only
1986 libata.dma=2 ATAPI (CDROM) DMA only
1987 libata.dma=4 Compact Flash DMA only
1988 Combinations also work, so libata.dma=3 enables DMA
1989 for disks and CDROMs, but not CFs.
1991 libata.ignore_hpa= [LIBATA] Ignore HPA limit
1992 libata.ignore_hpa=0 keep BIOS limits (default)
1993 libata.ignore_hpa=1 ignore limits, using full disk
1995 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
1999 libata.force= [LIBATA] Force configurations. The format is comma
2000 separated list of "[ID:]VAL" where ID is
2001 PORT[.DEVICE]. PORT and DEVICE are decimal numbers
2002 matching port, link or device. Basically, it matches
2003 the ATA ID string printed on console by libata. If
2004 the whole ID part is omitted, the last PORT and DEVICE
2005 values are used. If ID hasn't been specified yet, the
2006 configuration applies to all ports, links and devices.
2008 If only DEVICE is omitted, the parameter applies to
2009 the port and all links and devices behind it. DEVICE
2010 number of 0 either selects the first device or the
2011 first fan-out link behind PMP device. It does not
2012 select the host link. DEVICE number of 15 selects the
2013 host link and device attached to it.
2015 The VAL specifies the configuration to force. As long
2016 as there's no ambiguity shortcut notation is allowed.
2017 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2018 The following configurations can be forced.
2020 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2021 Any ID with matching PORT is used.
2023 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2025 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2026 udma[/][16,25,33,44,66,100,133] notation is also
2029 * [no]ncq: Turn on or off NCQ.
2031 * [no]ncqtrim: Turn off queued DSM TRIM.
2033 * nohrst, nosrst, norst: suppress hard, soft
2036 * rstonce: only attempt one reset during
2037 hot-unplug link recovery
2039 * dump_id: dump IDENTIFY data.
2041 * atapi_dmadir: Enable ATAPI DMADIR bridge support
2043 * disable: Disable this device.
2045 If there are multiple matching configurations changing
2046 the same attribute, the last one is used.
2048 memblock=debug [KNL] Enable memblock debug messages.
2050 load_ramdisk= [RAM] List of ramdisks to load from floppy
2051 See Documentation/blockdev/ramdisk.txt.
2053 lockd.nlm_grace_period=P [NFS] Assign grace period.
2056 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2059 lockd.nlm_timeout=T [NFS] Assign timeout value.
2062 lockd.nlm_udpport=M [NFS] Assign UDP port.
2065 locktorture.nreaders_stress= [KNL]
2066 Set the number of locking read-acquisition kthreads.
2067 Defaults to being automatically set based on the
2068 number of online CPUs.
2070 locktorture.nwriters_stress= [KNL]
2071 Set the number of locking write-acquisition kthreads.
2073 locktorture.onoff_holdoff= [KNL]
2074 Set time (s) after boot for CPU-hotplug testing.
2076 locktorture.onoff_interval= [KNL]
2077 Set time (s) between CPU-hotplug operations, or
2078 zero to disable CPU-hotplug testing.
2080 locktorture.shuffle_interval= [KNL]
2081 Set task-shuffle interval (jiffies). Shuffling
2082 tasks allows some CPUs to go into dyntick-idle
2083 mode during the locktorture test.
2085 locktorture.shutdown_secs= [KNL]
2086 Set time (s) after boot system shutdown. This
2087 is useful for hands-off automated testing.
2089 locktorture.stat_interval= [KNL]
2090 Time (s) between statistics printk()s.
2092 locktorture.stutter= [KNL]
2093 Time (s) to stutter testing, for example,
2094 specifying five seconds causes the test to run for
2095 five seconds, wait for five seconds, and so on.
2096 This tests the locking primitive's ability to
2097 transition abruptly to and from idle.
2099 locktorture.torture_runnable= [BOOT]
2100 Start locktorture running at boot time.
2102 locktorture.torture_type= [KNL]
2103 Specify the locking implementation to test.
2105 locktorture.verbose= [KNL]
2106 Enable additional printk() statements.
2108 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
2111 loglevel= All Kernel Messages with a loglevel smaller than the
2112 console loglevel will be printed to the console. It can
2113 also be changed with klogd or other programs. The
2114 loglevels are defined as follows:
2116 0 (KERN_EMERG) system is unusable
2117 1 (KERN_ALERT) action must be taken immediately
2118 2 (KERN_CRIT) critical conditions
2119 3 (KERN_ERR) error conditions
2120 4 (KERN_WARNING) warning conditions
2121 5 (KERN_NOTICE) normal but significant condition
2122 6 (KERN_INFO) informational
2123 7 (KERN_DEBUG) debug-level messages
2125 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
2126 in bytes. n must be a power of two and greater
2127 than the minimal size. The minimal size is defined
2128 by LOG_BUF_SHIFT kernel config parameter. There is
2129 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2130 that allows to increase the default size depending on
2131 the number of CPUs. See init/Kconfig for more details.
2133 logo.nologo [FB] Disables display of the built-in Linux logo.
2134 This may be used to provide more screen space for
2135 kernel log messages and is useful when debugging
2136 kernel boot problems.
2138 lp=0 [LP] Specify parallel ports to use, e.g,
2139 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2140 lp=reset first parallel port). 'lp=0' disables the
2141 lp=auto printer driver. 'lp=reset' (which can be
2142 specified in addition to the ports) causes
2143 attached printers to be reset. Using
2144 lp=port1,port2,... specifies the parallel ports
2145 to associate lp devices with, starting with
2146 lp0. A port specification may be 'none' to skip
2147 that lp device, or a parport name such as
2148 'parport0'. Specifying 'lp=auto' instead of a
2149 port specification list means that device IDs
2150 from each port should be examined, to see if
2151 an IEEE 1284-compliant printer is attached; if
2152 so, the driver will manage that printer.
2153 See also header of drivers/char/lp.c.
2156 Sets loops_per_jiffy to given constant, thus avoiding
2157 time-consuming boot-time autodetection (up to 250 ms per
2158 CPU). 0 enables autodetection (default). To determine
2159 the correct value for your kernel, boot with normal
2160 autodetection and see what value is printed. Note that
2161 on SMP systems the preset will be applied to all CPUs,
2162 which is likely to cause problems if your CPUs need
2163 significantly divergent settings. An incorrect value
2164 will cause delays in the kernel to be wrong, leading to
2165 unpredictable I/O errors and other breakage. Although
2166 unlikely, in the extreme case this might damage your
2170 Format: <io>,<irq>,<dma>
2172 machvec= [IA-64] Force the use of a particular machine-vector
2173 (machvec) in a generic kernel.
2174 Example: machvec=hpzx1_swiotlb
2176 machtype= [Loongson] Share the same kernel image file between different
2178 Example: machtype=lemote-yeeloong-2f-7inch
2180 max_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory greater
2181 than or equal to this physical address is ignored.
2183 maxcpus= [SMP] Maximum number of processors that an SMP kernel
2184 will bring up during bootup. maxcpus=n : n >= 0 limits
2185 the kernel to bring up 'n' processors. Surely after
2186 bootup you can bring up the other plugged cpu by executing
2187 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
2188 only takes effect during system bootup.
2189 While n=0 is a special case, it is equivalent to "nosmp",
2190 which also disables the IO APIC.
2192 max_loop= [LOOP] The number of loop block devices that get
2193 (loop.max_loop) unconditionally pre-created at init time. The default
2194 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
2195 of statically allocating a predefined number, loop
2196 devices can be requested on-demand with the
2197 /dev/loop-control interface.
2199 mce [X86-32] Machine Check Exception
2201 mce=option [X86-64] See Documentation/x86/x86_64/boot-options.txt
2203 md= [HW] RAID subsystems devices and level
2204 See Documentation/admin-guide/md.rst.
2207 Format: <first>,<last>
2208 Specifies range of consoles to be captured by the MDA.
2210 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
2211 Amount of memory to be used when the kernel is not able
2212 to see the whole system memory or for test.
2213 [X86] Work as limiting max address. Use together
2214 with memmap= to avoid physical address space collisions.
2215 Without memmap= PCI devices could be placed at addresses
2216 belonging to unused RAM.
2218 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
2222 [KNL,SH] Allow user to override the default size for
2223 per-device physically contiguous DMA buffers.
2225 memhp_default_state=online/offline
2226 [KNL] Set the initial state for the memory hotplug
2227 onlining policy. If not specified, the default value is
2228 set according to the
2229 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
2231 See Documentation/memory-hotplug.txt.
2233 memmap=exactmap [KNL,X86] Enable setting of an exact
2234 E820 memory map, as specified by the user.
2235 Such memmap=exactmap lines can be constructed based on
2236 BIOS output or other requirements. See the memmap=nn@ss
2239 memmap=nn[KMG]@ss[KMG]
2240 [KNL] Force usage of a specific region of memory.
2241 Region of memory to be used is from ss to ss+nn.
2242 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
2243 which limits max address to nn[KMG].
2244 Multiple different regions can be specified,
2247 memmap=100M@2G,100M#3G,1G!1024G
2249 memmap=nn[KMG]#ss[KMG]
2250 [KNL,ACPI] Mark specific memory as ACPI data.
2251 Region of memory to be marked is from ss to ss+nn.
2253 memmap=nn[KMG]$ss[KMG]
2254 [KNL,ACPI] Mark specific memory as reserved.
2255 Region of memory to be reserved is from ss to ss+nn.
2256 Example: Exclude memory from 0x18690000-0x1869ffff
2257 memmap=64K$0x18690000
2259 memmap=0x10000$0x18690000
2260 Some bootloaders may need an escape character before '$',
2261 like Grub2, otherwise '$' and the following number
2264 memmap=nn[KMG]!ss[KMG]
2265 [KNL,X86] Mark specific memory as protected.
2266 Region of memory to be used, from ss to ss+nn.
2267 The memory region may be marked as e820 type 12 (0xc)
2268 and is NVDIMM or ADR memory.
2270 memory_corruption_check=0/1 [X86]
2271 Some BIOSes seem to corrupt the first 64k of
2272 memory when doing things like suspend/resume.
2273 Setting this option will scan the memory
2274 looking for corruption. Enabling this will
2275 both detect corruption and prevent the kernel
2276 from using the memory being corrupted.
2277 However, its intended as a diagnostic tool; if
2278 repeatable BIOS-originated corruption always
2279 affects the same memory, you can use memmap=
2280 to prevent the kernel from using that memory.
2282 memory_corruption_check_size=size [X86]
2283 By default it checks for corruption in the low
2284 64k, making this memory unavailable for normal
2285 use. Use this parameter to scan for
2286 corruption in more or less memory.
2288 memory_corruption_check_period=seconds [X86]
2289 By default it checks for corruption every 60
2290 seconds. Use this parameter to check at some
2291 other rate. 0 disables periodic checking.
2293 memtest= [KNL,X86,ARM] Enable memtest
2295 default : 0 <disable>
2296 Specifies the number of memtest passes to be
2297 performed. Each pass selects another test
2298 pattern from a given set of patterns. Memtest
2299 fills the memory with this pattern, validates
2300 memory contents and reserves bad memory
2301 regions that are detected.
2303 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
2304 Valid arguments: on, off
2305 Default (depends on kernel configuration option):
2306 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
2307 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
2308 mem_encrypt=on: Activate SME
2309 mem_encrypt=off: Do not activate SME
2311 Refer to Documentation/x86/amd-memory-encryption.txt
2312 for details on when memory encryption can be activated.
2314 mem_sleep_default= [SUSPEND] Default system suspend mode:
2315 s2idle - Suspend-To-Idle
2316 shallow - Power-On Suspend or equivalent (if supported)
2317 deep - Suspend-To-RAM or equivalent (if supported)
2318 See Documentation/power/states.txt.
2320 meye.*= [HW] Set MotionEye Camera parameters
2321 See Documentation/video4linux/meye.txt.
2323 mfgpt_irq= [IA-32] Specify the IRQ to use for the
2324 Multi-Function General Purpose Timers on AMD Geode
2327 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
2328 the BIOS has incorrectly applied a workaround. TinyBIOS
2329 version 0.98 is known to be affected, 0.99 fixes the
2330 problem by letting the user disable the workaround.
2334 min_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory below this
2335 physical address is ignored.
2337 mini2440= [ARM,HW,KNL]
2338 Format:[0..2][b][c][t]
2340 MINI2440 configuration specification:
2341 0 - The attached screen is the 3.5" TFT
2342 1 - The attached screen is the 7" TFT
2343 2 - The VGA Shield is attached (1024x768)
2344 Leaving out the screen size parameter will not load
2345 the TFT driver, and the framebuffer will be left
2347 b - Enable backlight. The TFT backlight pin will be
2348 linked to the kernel VESA blanking code and a GPIO
2349 LED. This parameter is not necessary when using the
2351 c - Enable the s3c camera interface.
2352 t - Reserved for enabling touchscreen support. The
2353 touchscreen support is not enabled in the mainstream
2354 kernel as of 2.6.30, a preliminary port can be found
2355 in the "bleeding edge" mini2440 support kernel at
2356 http://repo.or.cz/w/linux-2.6/mini2440.git
2359 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
2360 parameter allows control of the logging verbosity for
2361 the additional memory initialisation checks. A value
2362 of 0 disables mminit logging and a level of 4 will
2363 log everything. Information is printed at KERN_DEBUG
2364 so loglevel=8 may also need to be specified.
2367 [KNL] When CONFIG_MODULE_SIG is set, this means that
2368 modules without (valid) signatures will fail to load.
2369 Note that if CONFIG_MODULE_SIG_FORCE is set, that
2370 is always true, so this option does nothing.
2372 module_blacklist= [KNL] Do not load a comma-separated list of
2373 modules. Useful for debugging problem modules.
2376 [MOUSE] Maximum time between finger touching and
2377 leaving touchpad surface for touch to be considered
2378 a tap and be reported as a left button click (for
2379 touchpads working in absolute mode only).
2381 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
2382 reporting absolute coordinates, such as tablets
2383 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
2384 reporting absolute coordinates, such as tablets
2386 movablecore=nn[KMG] [KNL,X86,IA-64,PPC] This parameter
2387 is similar to kernelcore except it specifies the
2388 amount of memory used for migratable allocations.
2389 If both kernelcore and movablecore is specified,
2390 then kernelcore will be at *least* the specified
2391 value but may be more. If movablecore on its own
2392 is specified, the administrator must be careful
2393 that the amount of memory usable for all allocations
2396 movable_node [KNL] Boot-time switch to make hotplugable memory
2397 NUMA nodes to be movable. This means that the memory
2398 of such nodes will be usable only for movable
2399 allocations which rules out almost all kernel
2400 allocations. Use with caution!
2402 MTD_Partition= [MTD]
2403 Format: <name>,<region-number>,<size>,<offset>
2405 MTD_Region= [MTD] Format:
2406 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
2409 See drivers/mtd/cmdlinepart.c.
2411 multitce=off [PPC] This parameter disables the use of the pSeries
2412 firmware feature for updating multiple TCE entries
2415 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
2417 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
2419 boundary - index of last SLC block on Flex-OneNAND.
2420 The remaining blocks are configured as MLC blocks.
2421 lock - Configure if Flex-OneNAND boundary should be locked.
2422 Once locked, the boundary cannot be changed.
2423 1 indicates lock status, 0 indicates unlock status.
2426 ARM/S3C2412 JIVE boot control
2428 See arch/arm/mach-s3c2412/mach-jive.c
2430 mtouchusb.raw_coordinates=
2431 [HW] Make the MicroTouch USB driver use raw coordinates
2432 ('y', default) or cooked coordinates ('n')
2434 mtrr_chunk_size=nn[KMG] [X86]
2435 used for mtrr cleanup. It is largest continuous chunk
2436 that could hold holes aka. UC entries.
2438 mtrr_gran_size=nn[KMG] [X86]
2439 Used for mtrr cleanup. It is granularity of mtrr block.
2441 Large value could prevent small alignment from
2444 mtrr_spare_reg_nr=n [X86]
2446 Range: 0,7 : spare reg number
2448 Used for mtrr cleanup. It is spare mtrr entries number.
2449 Set to 2 or more if your graphical card needs more.
2451 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
2453 netdev= [NET] Network devices parameters
2454 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
2455 Note that mem_start is often overloaded to mean
2456 something different and driver-specific.
2457 This usage is only documented in each driver source
2461 [NETFILTER] Enable connection tracking flow accounting
2462 0 to disable accounting
2463 1 to enable accounting
2466 nfsaddrs= [NFS] Deprecated. Use ip= instead.
2467 See Documentation/filesystems/nfs/nfsroot.txt.
2469 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
2470 See Documentation/filesystems/nfs/nfsroot.txt.
2472 nfsrootdebug [NFS] enable nfsroot debugging messages.
2473 See Documentation/filesystems/nfs/nfsroot.txt.
2475 nfs.callback_nr_threads=
2476 [NFSv4] set the total number of threads that the
2477 NFS client will assign to service NFSv4 callback
2480 nfs.callback_tcpport=
2481 [NFS] set the TCP port on which the NFSv4 callback
2482 channel should listen.
2485 [NFS] sets the pathname to the program which is used
2486 to update the NFS client cache entries.
2488 nfs.cache_getent_timeout=
2489 [NFS] sets the timeout after which an attempt to
2490 update a cache entry is deemed to have failed.
2492 nfs.idmap_cache_timeout=
2493 [NFS] set the maximum lifetime for idmapper cache
2497 [NFS] enable 64-bit inode numbers.
2498 If zero, the NFS client will fake up a 32-bit inode
2499 number for the readdir() and stat() syscalls instead
2500 of returning the full 64-bit number.
2501 The default is to return 64-bit inode numbers.
2503 nfs.max_session_cb_slots=
2504 [NFSv4.1] Sets the maximum number of session
2505 slots the client will assign to the callback
2506 channel. This determines the maximum number of
2507 callbacks the client will process in parallel for
2508 a particular server.
2510 nfs.max_session_slots=
2511 [NFSv4.1] Sets the maximum number of session slots
2512 the client will attempt to negotiate with the server.
2513 This limits the number of simultaneous RPC requests
2514 that the client can send to the NFSv4.1 server.
2515 Note that there is little point in setting this
2516 value higher than the max_tcp_slot_table_limit.
2518 nfs.nfs4_disable_idmapping=
2519 [NFSv4] When set to the default of '1', this option
2520 ensures that both the RPC level authentication
2521 scheme and the NFS level operations agree to use
2522 numeric uids/gids if the mount is using the
2523 'sec=sys' security flavour. In effect it is
2524 disabling idmapping, which can make migration from
2525 legacy NFSv2/v3 systems to NFSv4 easier.
2526 Servers that do not support this mode of operation
2527 will be autodetected by the client, and it will fall
2528 back to using the idmapper.
2529 To turn off this behaviour, set the value to '0'.
2531 [NFS4] Specify an additional fixed unique ident-
2532 ification string that NFSv4 clients can insert into
2533 their nfs_client_id4 string. This is typically a
2534 UUID that is generated at system install time.
2536 nfs.send_implementation_id =
2537 [NFSv4.1] Send client implementation identification
2538 information in exchange_id requests.
2539 If zero, no implementation identification information
2541 The default is to send the implementation identification
2544 nfs.recover_lost_locks =
2545 [NFSv4] Attempt to recover locks that were lost due
2546 to a lease timeout on the server. Please note that
2547 doing this risks data corruption, since there are
2548 no guarantees that the file will remain unchanged
2549 after the locks are lost.
2550 If you want to enable the kernel legacy behaviour of
2551 attempting to recover these locks, then set this
2553 The default parameter value of '0' causes the kernel
2554 not to attempt recovery of lost locks.
2556 nfs4.layoutstats_timer =
2557 [NFSv4.2] Change the rate at which the kernel sends
2558 layoutstats to the pNFS metadata server.
2560 Setting this to value to 0 causes the kernel to use
2561 whatever value is the default set by the layout
2562 driver. A non-zero value sets the minimum interval
2563 in seconds between layoutstats transmissions.
2565 nfsd.nfs4_disable_idmapping=
2566 [NFSv4] When set to the default of '1', the NFSv4
2567 server will return only numeric uids and gids to
2568 clients using auth_sys, and will accept numeric uids
2569 and gids from such clients. This is intended to ease
2570 migration from NFSv2/v3.
2572 nmi_debug= [KNL,SH] Specify one or more actions to take
2573 when a NMI is triggered.
2574 Format: [state][,regs][,debounce][,die]
2576 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
2577 Format: [panic,][nopanic,][num]
2579 0 - turn hardlockup detector in nmi_watchdog off
2580 1 - turn hardlockup detector in nmi_watchdog on
2581 When panic is specified, panic when an NMI watchdog
2582 timeout occurs (or 'nopanic' to override the opposite
2583 default). To disable both hard and soft lockup detectors,
2584 please see 'nowatchdog'.
2585 This is useful when you use a panic=... timeout and
2586 need the box quickly up again.
2588 netpoll.carrier_timeout=
2589 [NET] Specifies amount of time (in seconds) that
2590 netpoll should wait for a carrier. By default netpoll
2593 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
2594 emulation library even if a 387 maths coprocessor
2598 [HW] Never suspend the console
2599 Disable suspending of consoles during suspend and
2600 hibernate operations. Once disabled, debugging
2601 messages can reach various consoles while the rest
2602 of the system is being put to sleep (ie, while
2603 debugging driver suspend/resume hooks). This may
2604 not work reliably with all consoles, but is known
2605 to work with serial and VGA consoles.
2606 To facilitate more flexible debugging, we also add
2607 console_suspend, a printk module parameter to control
2608 it. Users could use console_suspend (usually
2609 /sys/module/printk/parameters/console_suspend) to
2610 turn on/off it dynamically.
2612 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
2613 caches in the slab allocator. Saves per-node memory,
2614 but will impact performance.
2618 noaltinstr [S390] Disables alternative instructions patching
2619 (CPU alternatives feature).
2621 noapic [SMP,APIC] Tells the kernel to not make use of any
2622 IOAPICs that may be present in the system.
2624 noautogroup Disable scheduler automatic task group creation.
2626 nobats [PPC] Do not use BATs for mapping kernel lowmem
2627 on "Classic" PPC cores.
2631 noclflush [BUGS=X86] Don't use the CLFLUSH instruction
2633 nodelayacct [KNL] Disable per-task delay accounting
2635 nodsp [SH] Disable hardware DSP at boot time.
2637 noefi Disable EFI runtime services support.
2642 On X86-32 available only on PAE configured kernels.
2643 noexec=on: enable non-executable mappings (default)
2644 noexec=off: disable non-executable mappings
2647 Disable SMAP (Supervisor Mode Access Prevention)
2648 even if it is supported by processor.
2651 Disable SMEP (Supervisor Mode Execution Prevention)
2652 even if it is supported by processor.
2655 This affects only 32-bit executables.
2656 noexec32=on: enable non-executable mappings (default)
2657 read doesn't imply executable mappings
2658 noexec32=off: disable non-executable mappings
2659 read implies executable mappings
2661 nofpu [MIPS,SH] Disable hardware FPU at boot time.
2663 nofxsr [BUGS=X86-32] Disables x86 floating point extended
2664 register save and restore. The kernel will only save
2665 legacy floating-point registers on task switch.
2667 nohugeiomap [KNL,x86] Disable kernel huge I/O mappings.
2669 nosmt [KNL,S390] Disable symmetric multithreading (SMT).
2670 Equivalent to smt=1.
2672 [KNL,x86] Disable symmetric multithreading (SMT).
2673 nosmt=force: Force disable SMT, cannot be undone
2674 via the sysfs control file.
2676 nospectre_v2 [X86] Disable all mitigations for the Spectre variant 2
2677 (indirect branch prediction) vulnerability. System may
2678 allow data leaks with this option, which is equivalent
2681 nospec_store_bypass_disable
2682 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
2684 noxsave [BUGS=X86] Disables x86 extended register state save
2685 and restore using xsave. The kernel will fallback to
2686 enabling legacy floating-point and sse state.
2688 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
2689 register states. The kernel will fall back to use
2690 xsave to save the states. By using this parameter,
2691 performance of saving the states is degraded because
2692 xsave doesn't support modified optimization while
2693 xsaveopt supports it on xsaveopt enabled systems.
2695 noxsaves [X86] Disables xsaves and xrstors used in saving and
2696 restoring x86 extended register state in compacted
2697 form of xsave area. The kernel will fall back to use
2698 xsaveopt and xrstor to save and restore the states
2699 in standard form of xsave area. By using this
2700 parameter, xsave area per process might occupy more
2701 memory on xsaves enabled systems.
2703 nohlt [BUGS=ARM,SH] Tells the kernel that the sleep(SH) or
2704 wfi(ARM) instruction doesn't work correctly and not to
2705 use it. This is also useful when using JTAG debugger.
2707 no_file_caps Tells the kernel not to honor file capabilities. The
2708 only way then for a file to be executed with privilege
2709 is to be setuid root or executed by root.
2711 nohalt [IA-64] Tells the kernel not to use the power saving
2712 function PAL_HALT_LIGHT when idle. This increases
2713 power-consumption. On the positive side, it reduces
2714 interrupt wake-up latency, which may improve performance
2715 in certain environments such as networked servers or
2718 nohibernate [HIBERNATION] Disable hibernation and resume.
2720 nohz= [KNL] Boottime enable/disable dynamic ticks
2721 Valid arguments: on, off
2724 nohz_full= [KNL,BOOT]
2725 The argument is a cpu list, as described above.
2726 In kernels built with CONFIG_NO_HZ_FULL=y, set
2727 the specified list of CPUs whose tick will be stopped
2728 whenever possible. The boot CPU will be forced outside
2729 the range to maintain the timekeeping. Any CPUs
2730 in this list will have their RCU callbacks offloaded,
2731 just as if they had also been called out in the
2732 rcu_nocbs= boot parameter.
2734 noiotrap [SH] Disables trapped I/O port accesses.
2736 noirqdebug [X86-32] Disables the code which attempts to detect and
2737 disable unhandled interrupt sources.
2739 no_timer_check [X86,APIC] Disables the code which tests for
2740 broken timer IRQ sources.
2742 noisapnp [ISAPNP] Disables ISA PnP code.
2744 noinitrd [RAM] Tells the kernel not to load any configured
2747 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
2749 [Deprecated - use intremap=off]
2753 noinvpcid [X86] Disable the INVPCID cpu feature.
2755 nojitter [IA-64] Disables jitter checking for ITC timers.
2757 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
2759 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
2763 [X86,PV_OPS] Disable paravirtualized VMware scheduler
2764 clock and use the default one.
2766 no-steal-acc [X86,KVM] Disable paravirtualized steal time accounting.
2767 steal time is computed, but won't influence scheduler
2770 nolapic [X86-32,APIC] Do not enable or use the local APIC.
2772 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
2774 noltlbs [PPC] Do not use large page/tlb entries for kernel
2775 lowmem mapping on PPC40x and PPC8xx
2777 nomca [IA-64] Disable machine check abort handling
2779 nomce [X86-32] Disable Machine Check Exception
2781 nomfgpt [X86-32] Disable Multi-Function General Purpose
2782 Timer usage (for AMD Geode machines).
2784 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
2785 shutdown the other cpus. Instead use the REBOOT_VECTOR
2788 nomodule Disable module load
2790 nopat [X86] Disable PAT (page attribute table extension of
2791 pagetables) support.
2793 nopcid [X86-64] Disable the PCID cpu feature.
2795 norandmaps Don't use address space randomization. Equivalent to
2796 echo 0 > /proc/sys/kernel/randomize_va_space
2798 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
2799 with UP alternatives
2801 nordrand [X86] Disable kernel use of the RDRAND and
2802 RDSEED instructions even if they are supported
2803 by the processor. RDRAND and RDSEED are still
2804 available to user space applications.
2806 noresume [SWSUSP] Disables resume and restores original swap
2809 no-scroll [VGA] Disables scrollback.
2810 This is required for the Braillex ib80-piezo Braille
2811 reader made by F.H. Papenmeier (Germany).
2815 nosep [BUGS=X86-32] Disables x86 SYSENTER/SYSEXIT support.
2817 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
2818 and disable the IO APIC. legacy for "maxcpus=0".
2820 nosoftlockup [KNL] Disable the soft-lockup detector.
2822 nosync [HW,M68K] Disables sync negotiation for all devices.
2824 notsc [BUGS=X86-32] Disable Time Stamp Counter
2826 nowatchdog [KNL] Disable both lockup detectors, i.e.
2827 soft-lockup and NMI watchdog (hard-lockup).
2831 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
2833 cpu0_hotplug [X86] Turn on CPU0 hotplug feature when
2834 CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off.
2835 Some features depend on CPU0. Known dependencies are:
2836 1. Resume from suspend/hibernate depends on CPU0.
2837 Suspend/hibernate will fail if CPU0 is offline and you
2838 need to online CPU0 before suspend/hibernate.
2839 2. PIC interrupts also depend on CPU0. CPU0 can't be
2840 removed if a PIC interrupt is detected.
2841 It's said poweroff/reboot may depend on CPU0 on some
2842 machines although I haven't seen such issues so far
2843 after CPU0 is offline on a few tested machines.
2844 If the dependencies are under your control, you can
2845 turn on cpu0_hotplug.
2847 nps_mtm_hs_ctr= [KNL,ARC]
2848 This parameter sets the maximum duration, in
2849 cycles, each HW thread of the CTOP can run
2850 without interruptions, before HW switches it.
2851 The actual maximum duration is 16 times this
2853 Format: integer between 1 and 255
2856 nptcg= [IA-64] Override max number of concurrent global TLB
2857 purges which is reported from either PAL_VM_SUMMARY or
2860 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
2861 could support. nr_cpus=n : n >= 1 limits the kernel to
2862 support 'n' processors. It could be larger than the
2863 number of already plugged CPU during bootup, later in
2864 runtime you can physically add extra cpu until it reaches
2865 n. So during boot up some boot time memory for per-cpu
2866 variables need be pre-allocated for later physical cpu
2869 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
2871 numa_balancing= [KNL,X86] Enable or disable automatic NUMA balancing.
2872 Allowed values are enable and disable
2874 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
2875 'node', 'default' can be specified
2876 This can be set from sysctl after boot.
2877 See Documentation/sysctl/vm.txt for details.
2879 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
2880 See Documentation/debugging-via-ohci1394.txt for more
2883 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
2884 Rather than timing out after 20 ms if an EC
2885 command is not properly ACKed, override the length
2886 of the timeout. We have interrupts disabled while
2887 waiting for the ACK, so if this is set too high
2888 interrupts *may* be lost!
2890 omap_mux= [OMAP] Override bootloader pin multiplexing.
2891 Format: <mux_mode0.mode_name=value>...
2892 For example, to override I2C bus2:
2893 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
2895 oprofile.timer= [HW]
2896 Use timer interrupt instead of performance counters
2898 oprofile.cpu_type= Force an oprofile cpu type
2899 This might be useful if you have an older oprofile
2900 userland or if you want common events.
2901 Format: { arch_perfmon }
2902 arch_perfmon: [X86] Force use of architectural
2903 perfmon on Intel CPUs instead of the
2904 CPU specific event set.
2905 timer: [X86] Force use of architectural NMI
2906 timer mode (see also oprofile.timer
2907 for generic hr timer mode)
2909 oops=panic Always panic on oopses. Default is to just kill the
2910 process, but there is a small probability of
2911 deadlocking the machine.
2912 This will also cause panics on machine check exceptions.
2913 Useful together with panic=30 to trigger a reboot.
2916 See Documentation/sound/oss/oss-parameters.txt
2918 page_owner= [KNL] Boot-time page_owner enabling option.
2919 Storage of the information about who allocated
2920 each page is disabled in default. With this switch,
2922 on: enable the feature
2924 page_poison= [KNL] Boot-time parameter changing the state of
2925 poisoning on the buddy allocator.
2926 off: turn off poisoning
2927 on: turn on poisoning
2929 panic= [KNL] Kernel behaviour on panic: delay <timeout>
2930 timeout > 0: seconds before rebooting
2931 timeout = 0: wait forever
2932 timeout < 0: reboot immediately
2935 panic_on_warn panic() instead of WARN(). Useful to cause kdump
2938 crash_kexec_post_notifiers
2939 Run kdump after running panic-notifiers and dumping
2940 kmsg. This only for the users who doubt kdump always
2941 succeeds in any situation.
2942 Note that this also increases risks of kdump failure,
2943 because some panic notifiers can make the crashed
2944 kernel more unstable.
2946 parkbd.port= [HW] Parallel port number the keyboard adapter is
2947 connected to, default is 0.
2949 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
2950 0 for XT, 1 for AT (default is AT).
2953 parport= [HW,PPT] Specify parallel ports. 0 disables.
2954 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
2955 Use 'auto' to force the driver to use any
2956 IRQ/DMA settings detected (the default is to
2957 ignore detected IRQ/DMA settings because of
2958 possible conflicts). You can specify the base
2959 address, IRQ, and DMA settings; IRQ and DMA
2960 should be numbers, or 'auto' (for using detected
2961 settings on that particular port), or 'nofifo'
2962 (to avoid using a FIFO even if it is detected).
2963 Parallel ports are assigned in the order they
2964 are specified on the command line, starting
2967 parport_init_mode= [HW,PPT]
2968 Configure VIA parallel port to operate in
2969 a specific mode. This is necessary on Pegasos
2970 computer where firmware has no options for setting
2971 up parallel port mode and sets it to spp.
2972 Currently this function knows 686a and 8231 chips.
2973 Format: [spp|ps2|epp|ecp|ecpepp]
2976 Halt all CPUs after the first oops has been printed for
2977 the specified number of seconds. This is to be used if
2978 your oopses keep scrolling off the screen.
2983 See header of drivers/block/paride/pcd.c.
2984 See also Documentation/blockdev/paride.txt.
2986 pci=option[,option...] [PCI] various PCI subsystem options:
2987 earlydump [X86] dump PCI config space before the kernel
2989 off [X86] don't probe for the PCI bus
2990 bios [X86-32] force use of PCI BIOS, don't access
2991 the hardware directly. Use this if your machine
2992 has a non-standard PCI host bridge.
2993 nobios [X86-32] disallow use of PCI BIOS, only direct
2994 hardware access methods are allowed. Use this
2995 if you experience crashes upon bootup and you
2996 suspect they are caused by the BIOS.
2997 conf1 [X86] Force use of PCI Configuration Access
2998 Mechanism 1 (config address in IO port 0xCF8,
2999 data in IO port 0xCFC, both 32-bit).
3000 conf2 [X86] Force use of PCI Configuration Access
3001 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
3002 the function, IO port 0xCFA, also 8-bit, sets
3003 bus number. The config space is then accessed
3004 through ports 0xC000-0xCFFF).
3005 See http://wiki.osdev.org/PCI for more info
3006 on the configuration access mechanisms.
3007 noaer [PCIE] If the PCIEAER kernel config parameter is
3008 enabled, this kernel boot option can be used to
3009 disable the use of PCIE advanced error reporting.
3010 nodomains [PCI] Disable support for multiple PCI
3011 root domains (aka PCI segments, in ACPI-speak).
3012 nommconf [X86] Disable use of MMCONFIG for PCI
3014 check_enable_amd_mmconf [X86] check for and enable
3015 properly configured MMIO access to PCI
3016 config space on AMD family 10h CPU
3017 nomsi [MSI] If the PCI_MSI kernel config parameter is
3018 enabled, this kernel boot option can be used to
3019 disable the use of MSI interrupts system-wide.
3020 noioapicquirk [APIC] Disable all boot interrupt quirks.
3021 Safety option to keep boot IRQs enabled. This
3022 should never be necessary.
3023 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
3024 primary IO-APIC for bridges that cannot disable
3025 boot IRQs. This fixes a source of spurious IRQs
3026 when the system masks IRQs.
3027 noioapicreroute [APIC] Disable workaround that uses the
3028 boot IRQ equivalent of an IRQ that connects to
3029 a chipset where boot IRQs cannot be disabled.
3030 The opposite of ioapicreroute.
3031 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
3032 routing table. These calls are known to be buggy
3033 on several machines and they hang the machine
3034 when used, but on other computers it's the only
3035 way to get the interrupt routing table. Try
3036 this option if the kernel is unable to allocate
3037 IRQs or discover secondary PCI buses on your
3039 rom [X86] Assign address space to expansion ROMs.
3040 Use with caution as certain devices share
3041 address decoders between ROMs and other
3043 norom [X86] Do not assign address space to
3044 expansion ROMs that do not already have
3045 BIOS assigned address ranges.
3046 nobar [X86] Do not assign address space to the
3047 BARs that weren't assigned by the BIOS.
3048 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
3049 assigned automatically to PCI devices. You can
3050 make the kernel exclude IRQs of your ISA cards
3052 pirqaddr=0xAAAAA [X86] Specify the physical address
3053 of the PIRQ table (normally generated
3054 by the BIOS) if it is outside the
3055 F0000h-100000h range.
3056 lastbus=N [X86] Scan all buses thru bus #N. Can be
3057 useful if the kernel is unable to find your
3058 secondary buses and you want to tell it
3059 explicitly which ones they are.
3060 assign-busses [X86] Always assign all PCI bus
3061 numbers ourselves, overriding
3062 whatever the firmware may have done.
3063 usepirqmask [X86] Honor the possible IRQ mask stored
3064 in the BIOS $PIR table. This is needed on
3065 some systems with broken BIOSes, notably
3066 some HP Pavilion N5400 and Omnibook XE3
3067 notebooks. This will have no effect if ACPI
3068 IRQ routing is enabled.
3069 noacpi [X86] Do not use ACPI for IRQ routing
3070 or for PCI scanning.
3071 use_crs [X86] Use PCI host bridge window information
3072 from ACPI. On BIOSes from 2008 or later, this
3073 is enabled by default. If you need to use this,
3074 please report a bug.
3075 nocrs [X86] Ignore PCI host bridge windows from ACPI.
3076 If you need to use this, please report a bug.
3077 routeirq Do IRQ routing for all PCI devices.
3078 This is normally done in pci_enable_device(),
3079 so this option is a temporary workaround
3080 for broken drivers that don't call it.
3081 skip_isa_align [X86] do not align io start addr, so can
3082 handle more pci cards
3083 noearly [X86] Don't do any early type 1 scanning.
3084 This might help on some broken boards which
3085 machine check when some devices' config space
3086 is read. But various workarounds are disabled
3087 and some IOMMU drivers will not work.
3088 bfsort Sort PCI devices into breadth-first order.
3089 This sorting is done to get a device
3090 order compatible with older (<= 2.4) kernels.
3091 nobfsort Don't sort PCI devices into breadth-first order.
3092 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
3093 tuning and use the BIOS-configured MPS defaults.
3094 pcie_bus_safe Set every device's MPS to the largest value
3095 supported by all devices below the root complex.
3096 pcie_bus_perf Set device MPS to the largest allowable MPS
3097 based on its parent bus. Also set MRRS (Max
3098 Read Request Size) to the largest supported
3099 value (no larger than the MPS that the device
3100 or bus can support) for best performance.
3101 pcie_bus_peer2peer Set every device's MPS to 128B, which
3102 every device is guaranteed to support. This
3103 configuration allows peer-to-peer DMA between
3104 any pair of devices, possibly at the cost of
3105 reduced performance. This also guarantees
3106 that hot-added devices will work.
3107 cbiosize=nn[KMG] The fixed amount of bus space which is
3108 reserved for the CardBus bridge's IO window.
3109 The default value is 256 bytes.
3110 cbmemsize=nn[KMG] The fixed amount of bus space which is
3111 reserved for the CardBus bridge's memory
3112 window. The default value is 64 megabytes.
3115 [<order of align>@][<domain>:]<bus>:<slot>.<func>[; ...]
3116 [<order of align>@]pci:<vendor>:<device>\
3117 [:<subvendor>:<subdevice>][; ...]
3118 Specifies alignment and device to reassign
3119 aligned memory resources.
3120 If <order of align> is not specified,
3121 PAGE_SIZE is used as alignment.
3122 PCI-PCI bridge can be specified, if resource
3123 windows need to be expanded.
3124 To specify the alignment for several
3125 instances of a device, the PCI vendor,
3126 device, subvendor, and subdevice may be
3127 specified, e.g., 4096@pci:8086:9c22:103c:198f
3128 ecrc= Enable/disable PCIe ECRC (transaction layer
3129 end-to-end CRC checking).
3130 bios: Use BIOS/firmware settings. This is the
3134 hpiosize=nn[KMG] The fixed amount of bus space which is
3135 reserved for hotplug bridge's IO window.
3136 Default size is 256 bytes.
3137 hpmemsize=nn[KMG] The fixed amount of bus space which is
3138 reserved for hotplug bridge's memory window.
3139 Default size is 2 megabytes.
3140 hpbussize=nn The minimum amount of additional bus numbers
3141 reserved for buses below a hotplug bridge.
3143 realloc= Enable/disable reallocating PCI bridge resources
3144 if allocations done by BIOS are too small to
3145 accommodate resources required by all child
3147 off: Turn realloc off
3149 realloc same as realloc=on
3150 noari do not use PCIe ARI.
3151 pcie_scan_all Scan all possible PCIe devices. Otherwise we
3152 only look for one device below a PCIe downstream
3155 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
3158 force Enable ASPM even on devices that claim not to support it.
3159 WARNING: Forcing ASPM on may cause system lockups.
3161 pcie_hp= [PCIE] PCI Express Hotplug driver options:
3162 nomsi Do not use MSI for PCI Express Native Hotplug (this
3163 makes all PCIe ports use INTx for hotplug services).
3165 pcie_ports= [PCIE] PCIe ports handling:
3166 auto Ask the BIOS whether or not to use native PCIe services
3167 associated with PCIe ports (PME, hot-plug, AER). Use
3168 them only if that is allowed by the BIOS.
3169 native Use native PCIe services associated with PCIe ports
3171 compat Treat PCIe ports as PCI-to-PCI bridges, disable the PCIe
3174 pcie_port_pm= [PCIE] PCIe port power management handling:
3175 off Disable power management of all PCIe ports
3176 force Forcibly enable power management of all PCIe ports
3178 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
3179 nomsi Do not use MSI for native PCIe PME signaling (this makes
3180 all PCIe root ports use INTx for all services).
3182 pcmv= [HW,PCMCIA] BadgePAD 4
3186 Keep all power-domains already enabled by bootloader on,
3187 even if no driver has claimed them. This is useful
3188 for debug and development, but should not be
3189 needed on a platform with proper driver support.
3192 See Documentation/blockdev/paride.txt.
3194 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
3197 See arch/parisc/kernel/pdc_chassis.c
3199 percpu_alloc= Select which percpu first chunk allocator to use.
3200 Currently supported values are "embed" and "page".
3201 Archs may support subset or none of the selections.
3202 See comments in mm/percpu.c for details on each
3203 allocator. This parameter is primarily for debugging
3204 and performance comparison.
3207 See Documentation/blockdev/paride.txt.
3210 See Documentation/blockdev/paride.txt.
3212 pirq= [SMP,APIC] Manual mp-table setup
3213 See Documentation/x86/i386/IO-APIC.txt.
3215 plip= [PPT,NET] Parallel port network link
3216 Format: { parport<nr> | timid | 0 }
3217 See also Documentation/parport.txt.
3219 pmtmr= [X86] Manual setup of pmtmr I/O Port.
3220 Override pmtimer IOPort with a hex value.
3224 Enable PNP debug messages (depends on the
3225 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
3226 via /sys/module/pnp/parameters/debug. We always show
3227 current resource usage; turning this on also shows
3228 possible settings and some assignment information.
3234 { on | off | curr | res | no-curr | no-res }
3237 [ISAPNP] Exclude IRQs for the autoconfiguration
3240 [ISAPNP] Exclude DMAs for the autoconfiguration
3242 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
3243 Ranges are in pairs (I/O port base and size).
3246 [ISAPNP] Exclude memory regions for the
3248 Ranges are in pairs (memory base and size).
3250 ports= [IP_VS_FTP] IPVS ftp helper module
3252 Up to 8 (IP_VS_APP_MAX_PORTS) ports
3254 Format: <port>,<port>....
3256 powersave=off [PPC] This option disables power saving features.
3257 It specifically disables cpuidle and sets the
3258 platform machine description specific power_save
3259 function to NULL. On Idle the CPU just reduces
3262 ppc_strict_facility_enable
3263 [PPC] This option catches any kernel floating point,
3264 Altivec, VSX and SPE outside of regions specifically
3265 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
3266 There is some performance impact when enabling this.
3268 print-fatal-signals=
3269 [KNL] debug: print fatal signals
3271 If enabled, warn about various signal handling
3272 related application anomalies: too many signals,
3273 too many POSIX.1 timers, fatal signals causing a
3276 If you hit the warning due to signal overflow,
3277 you might want to try "ulimit -i unlimited".
3281 printk.always_kmsg_dump=
3282 Trigger kmsg_dump for cases other than kernel oops or
3284 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
3287 printk.devkmsg={on,off,ratelimit}
3288 Control writing to /dev/kmsg.
3289 on - unlimited logging to /dev/kmsg from userspace
3290 off - logging to /dev/kmsg disabled
3291 ratelimit - ratelimit the logging
3294 printk.time= Show timing data prefixed to each printk message line
3295 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
3297 processor.max_cstate= [HW,ACPI]
3298 Limit processor to maximum C-state
3299 max_cstate=9 overrides any DMI blacklist limit.
3301 processor.nocst [HW,ACPI]
3302 Ignore the _CST method to determine C-states,
3303 instead using the legacy FADT method
3305 profile= [KNL] Enable kernel profiling via /proc/profile
3306 Format: [schedule,]<number>
3307 Param: "schedule" - profile schedule points.
3308 Param: <number> - step/bucket size as a power of 2 for
3309 statistical time based profiling.
3310 Param: "sleep" - profile D-state sleeping (millisecs).
3311 Requires CONFIG_SCHEDSTATS
3312 Param: "kvm" - profile VM exits.
3314 prompt_ramdisk= [RAM] List of RAM disks to prompt for floppy disk
3316 See Documentation/blockdev/ramdisk.txt.
3318 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
3319 probe for; one of (bare|imps|exps|lifebook|any).
3320 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
3322 psmouse.resetafter= [HW,MOUSE]
3323 Try to reset the device after so many bad packets
3326 [HW,MOUSE] Set desired mouse resolution, in dpi.
3327 psmouse.smartscroll=
3328 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
3329 0 = disabled, 1 = enabled (default).
3331 pstore.backend= Specify the name of the pstore backend to use
3334 See Documentation/blockdev/paride.txt.
3336 pti= [X86_64] Control Page Table Isolation of user and
3337 kernel address spaces. Disabling this feature
3338 removes hardening, but improves performance of
3339 system calls and interrupts.
3341 on - unconditionally enable
3342 off - unconditionally disable
3343 auto - kernel detects whether your CPU model is
3344 vulnerable to issues that PTI mitigates
3346 Not specifying this option is equivalent to pti=auto.
3349 Equivalent to pti=off
3352 [KNL] Number of legacy pty's. Overwrites compiled-in
3355 quiet [KNL] Disable most log messages
3360 See Documentation/admin-guide/md.rst.
3362 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
3363 See Documentation/blockdev/ramdisk.txt.
3365 ras=option[,option,...] [KNL] RAS-specific options
3368 Disable the Correctable Errors Collector,
3369 see CONFIG_RAS_CEC help text.
3372 The argument is a cpu list, as described above.
3374 In kernels built with CONFIG_RCU_NOCB_CPU=y, set
3375 the specified list of CPUs to be no-callback CPUs.
3376 Invocation of these CPUs' RCU callbacks will
3377 be offloaded to "rcuox/N" kthreads created for
3378 that purpose, where "x" is "b" for RCU-bh, "p"
3379 for RCU-preempt, and "s" for RCU-sched, and "N"
3380 is the CPU number. This reduces OS jitter on the
3381 offloaded CPUs, which can be useful for HPC and
3382 real-time workloads. It can also improve energy
3383 efficiency for asymmetric multiprocessors.
3386 Rather than requiring that offloaded CPUs
3387 (specified by rcu_nocbs= above) explicitly
3388 awaken the corresponding "rcuoN" kthreads,
3389 make these kthreads poll for callbacks.
3390 This improves the real-time response for the
3391 offloaded CPUs by relieving them of the need to
3392 wake up the corresponding kthread, but degrades
3393 energy efficiency by requiring that the kthreads
3394 periodically wake up to do the polling.
3396 rcutree.blimit= [KNL]
3397 Set maximum number of finished RCU callbacks to
3398 process in one batch.
3400 rcutree.dump_tree= [KNL]
3401 Dump the structure of the rcu_node combining tree
3402 out at early boot. This is used for diagnostic
3403 purposes, to verify correct tree setup.
3405 rcutree.gp_cleanup_delay= [KNL]
3406 Set the number of jiffies to delay each step of
3407 RCU grace-period cleanup.
3409 rcutree.gp_init_delay= [KNL]
3410 Set the number of jiffies to delay each step of
3411 RCU grace-period initialization.
3413 rcutree.gp_preinit_delay= [KNL]
3414 Set the number of jiffies to delay each step of
3415 RCU grace-period pre-initialization, that is,
3416 the propagation of recent CPU-hotplug changes up
3417 the rcu_node combining tree.
3419 rcutree.rcu_fanout_exact= [KNL]
3420 Disable autobalancing of the rcu_node combining
3421 tree. This is used by rcutorture, and might
3422 possibly be useful for architectures having high
3423 cache-to-cache transfer latencies.
3425 rcutree.rcu_fanout_leaf= [KNL]
3426 Change the number of CPUs assigned to each
3427 leaf rcu_node structure. Useful for very
3428 large systems, which will choose the value 64,
3429 and for NUMA systems with large remote-access
3430 latencies, which will choose a value aligned
3431 with the appropriate hardware boundaries.
3433 rcutree.jiffies_till_sched_qs= [KNL]
3434 Set required age in jiffies for a
3435 given grace period before RCU starts
3436 soliciting quiescent-state help from
3437 rcu_note_context_switch().
3439 rcutree.jiffies_till_first_fqs= [KNL]
3440 Set delay from grace-period initialization to
3441 first attempt to force quiescent states.
3442 Units are jiffies, minimum value is zero,
3443 and maximum value is HZ.
3445 rcutree.jiffies_till_next_fqs= [KNL]
3446 Set delay between subsequent attempts to force
3447 quiescent states. Units are jiffies, minimum
3448 value is one, and maximum value is HZ.
3450 rcutree.kthread_prio= [KNL,BOOT]
3451 Set the SCHED_FIFO priority of the RCU per-CPU
3452 kthreads (rcuc/N). This value is also used for
3453 the priority of the RCU boost threads (rcub/N)
3454 and for the RCU grace-period kthreads (rcu_bh,
3455 rcu_preempt, and rcu_sched). If RCU_BOOST is
3456 set, valid values are 1-99 and the default is 1
3457 (the least-favored priority). Otherwise, when
3458 RCU_BOOST is not set, valid values are 0-99 and
3459 the default is zero (non-realtime operation).
3461 rcutree.rcu_nocb_leader_stride= [KNL]
3462 Set the number of NOCB kthread groups, which
3463 defaults to the square root of the number of
3464 CPUs. Larger numbers reduces the wakeup overhead
3465 on the per-CPU grace-period kthreads, but increases
3466 that same overhead on each group's leader.
3468 rcutree.qhimark= [KNL]
3469 Set threshold of queued RCU callbacks beyond which
3470 batch limiting is disabled.
3472 rcutree.qlowmark= [KNL]
3473 Set threshold of queued RCU callbacks below which
3474 batch limiting is re-enabled.
3476 rcutree.rcu_idle_gp_delay= [KNL]
3477 Set wakeup interval for idle CPUs that have
3478 RCU callbacks (RCU_FAST_NO_HZ=y).
3480 rcutree.rcu_idle_lazy_gp_delay= [KNL]
3481 Set wakeup interval for idle CPUs that have
3482 only "lazy" RCU callbacks (RCU_FAST_NO_HZ=y).
3483 Lazy RCU callbacks are those which RCU can
3484 prove do nothing more than free memory.
3486 rcutree.rcu_kick_kthreads= [KNL]
3487 Cause the grace-period kthread to get an extra
3488 wake_up() if it sleeps three times longer than
3489 it should at force-quiescent-state time.
3490 This wake_up() will be accompanied by a
3491 WARN_ONCE() splat and an ftrace_dump().
3493 rcuperf.gp_async= [KNL]
3494 Measure performance of asynchronous
3495 grace-period primitives such as call_rcu().
3497 rcuperf.gp_async_max= [KNL]
3498 Specify the maximum number of outstanding
3499 callbacks per writer thread. When a writer
3500 thread exceeds this limit, it invokes the
3501 corresponding flavor of rcu_barrier() to allow
3502 previously posted callbacks to drain.
3504 rcuperf.gp_exp= [KNL]
3505 Measure performance of expedited synchronous
3506 grace-period primitives.
3508 rcuperf.holdoff= [KNL]
3509 Set test-start holdoff period. The purpose of
3510 this parameter is to delay the start of the
3511 test until boot completes in order to avoid
3514 rcuperf.nreaders= [KNL]
3515 Set number of RCU readers. The value -1 selects
3516 N, where N is the number of CPUs. A value
3517 "n" less than -1 selects N-n+1, where N is again
3518 the number of CPUs. For example, -2 selects N
3519 (the number of CPUs), -3 selects N+1, and so on.
3520 A value of "n" less than or equal to -N selects
3523 rcuperf.nwriters= [KNL]
3524 Set number of RCU writers. The values operate
3525 the same as for rcuperf.nreaders.
3526 N, where N is the number of CPUs
3528 rcuperf.perf_runnable= [BOOT]
3529 Start rcuperf running at boot time.
3531 rcuperf.perf_type= [KNL]
3532 Specify the RCU implementation to test.
3534 rcuperf.shutdown= [KNL]
3535 Shut the system down after performance tests
3536 complete. This is useful for hands-off automated
3539 rcuperf.verbose= [KNL]
3540 Enable additional printk() statements.
3542 rcuperf.writer_holdoff= [KNL]
3543 Write-side holdoff between grace periods,
3544 in microseconds. The default of zero says
3547 rcutorture.cbflood_inter_holdoff= [KNL]
3548 Set holdoff time (jiffies) between successive
3549 callback-flood tests.
3551 rcutorture.cbflood_intra_holdoff= [KNL]
3552 Set holdoff time (jiffies) between successive
3553 bursts of callbacks within a given callback-flood
3556 rcutorture.cbflood_n_burst= [KNL]
3557 Set the number of bursts making up a given
3558 callback-flood test. Set this to zero to
3559 disable callback-flood testing.
3561 rcutorture.cbflood_n_per_burst= [KNL]
3562 Set the number of callbacks to be registered
3563 in a given burst of a callback-flood test.
3565 rcutorture.fqs_duration= [KNL]
3566 Set duration of force_quiescent_state bursts
3569 rcutorture.fqs_holdoff= [KNL]
3570 Set holdoff time within force_quiescent_state bursts
3573 rcutorture.fqs_stutter= [KNL]
3574 Set wait time between force_quiescent_state bursts
3577 rcutorture.gp_cond= [KNL]
3578 Use conditional/asynchronous update-side
3579 primitives, if available.
3581 rcutorture.gp_exp= [KNL]
3582 Use expedited update-side primitives, if available.
3584 rcutorture.gp_normal= [KNL]
3585 Use normal (non-expedited) asynchronous
3586 update-side primitives, if available.
3588 rcutorture.gp_sync= [KNL]
3589 Use normal (non-expedited) synchronous
3590 update-side primitives, if available. If all
3591 of rcutorture.gp_cond=, rcutorture.gp_exp=,
3592 rcutorture.gp_normal=, and rcutorture.gp_sync=
3593 are zero, rcutorture acts as if is interpreted
3594 they are all non-zero.
3596 rcutorture.n_barrier_cbs= [KNL]
3597 Set callbacks/threads for rcu_barrier() testing.
3599 rcutorture.nfakewriters= [KNL]
3600 Set number of concurrent RCU writers. These just
3601 stress RCU, they don't participate in the actual
3602 test, hence the "fake".
3604 rcutorture.nreaders= [KNL]
3605 Set number of RCU readers. The value -1 selects
3606 N-1, where N is the number of CPUs. A value
3607 "n" less than -1 selects N-n-2, where N is again
3608 the number of CPUs. For example, -2 selects N
3609 (the number of CPUs), -3 selects N+1, and so on.
3611 rcutorture.object_debug= [KNL]
3612 Enable debug-object double-call_rcu() testing.
3614 rcutorture.onoff_holdoff= [KNL]
3615 Set time (s) after boot for CPU-hotplug testing.
3617 rcutorture.onoff_interval= [KNL]
3618 Set time (s) between CPU-hotplug operations, or
3619 zero to disable CPU-hotplug testing.
3621 rcutorture.shuffle_interval= [KNL]
3622 Set task-shuffle interval (s). Shuffling tasks
3623 allows some CPUs to go into dyntick-idle mode
3624 during the rcutorture test.
3626 rcutorture.shutdown_secs= [KNL]
3627 Set time (s) after boot system shutdown. This
3628 is useful for hands-off automated testing.
3630 rcutorture.stall_cpu= [KNL]
3631 Duration of CPU stall (s) to test RCU CPU stall
3632 warnings, zero to disable.
3634 rcutorture.stall_cpu_holdoff= [KNL]
3635 Time to wait (s) after boot before inducing stall.
3637 rcutorture.stat_interval= [KNL]
3638 Time (s) between statistics printk()s.
3640 rcutorture.stutter= [KNL]
3641 Time (s) to stutter testing, for example, specifying
3642 five seconds causes the test to run for five seconds,
3643 wait for five seconds, and so on. This tests RCU's
3644 ability to transition abruptly to and from idle.
3646 rcutorture.test_boost= [KNL]
3647 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
3648 "Maybe" means test if the RCU implementation
3649 under test support RCU priority boosting.
3651 rcutorture.test_boost_duration= [KNL]
3652 Duration (s) of each individual boost test.
3654 rcutorture.test_boost_interval= [KNL]
3655 Interval (s) between each boost test.
3657 rcutorture.test_no_idle_hz= [KNL]
3658 Test RCU's dyntick-idle handling. See also the
3659 rcutorture.shuffle_interval parameter.
3661 rcutorture.torture_runnable= [BOOT]
3662 Start rcutorture running at boot time.
3664 rcutorture.torture_type= [KNL]
3665 Specify the RCU implementation to test.
3667 rcutorture.verbose= [KNL]
3668 Enable additional printk() statements.
3670 rcupdate.rcu_cpu_stall_suppress= [KNL]
3671 Suppress RCU CPU stall warning messages.
3673 rcupdate.rcu_cpu_stall_timeout= [KNL]
3674 Set timeout for RCU CPU stall warning messages.
3676 rcupdate.rcu_expedited= [KNL]
3677 Use expedited grace-period primitives, for
3678 example, synchronize_rcu_expedited() instead
3679 of synchronize_rcu(). This reduces latency,
3680 but can increase CPU utilization, degrade
3681 real-time latency, and degrade energy efficiency.
3682 No effect on CONFIG_TINY_RCU kernels.
3684 rcupdate.rcu_normal= [KNL]
3685 Use only normal grace-period primitives,
3686 for example, synchronize_rcu() instead of
3687 synchronize_rcu_expedited(). This improves
3688 real-time latency, CPU utilization, and
3689 energy efficiency, but can expose users to
3690 increased grace-period latency. This parameter
3691 overrides rcupdate.rcu_expedited. No effect on
3692 CONFIG_TINY_RCU kernels.
3694 rcupdate.rcu_normal_after_boot= [KNL]
3695 Once boot has completed (that is, after
3696 rcu_end_inkernel_boot() has been invoked), use
3697 only normal grace-period primitives. No effect
3698 on CONFIG_TINY_RCU kernels.
3700 rcupdate.rcu_task_stall_timeout= [KNL]
3701 Set timeout in jiffies for RCU task stall warning
3702 messages. Disable with a value less than or equal
3705 rcupdate.rcu_self_test= [KNL]
3706 Run the RCU early boot self tests
3708 rcupdate.rcu_self_test_bh= [KNL]
3709 Run the RCU bh early boot self tests
3711 rcupdate.rcu_self_test_sched= [KNL]
3712 Run the RCU sched early boot self tests
3716 Run specified binary instead of /init from the ramdisk,
3717 used for early userspace startup. See initrd.
3720 Turn on/off individual RDT features. List is:
3721 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, mba.
3722 E.g. to turn on cmt and turn off mba use:
3726 Format (x86 or x86_64):
3727 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] \
3729 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
3731 Where reboot_mode is one of warm (soft) or cold (hard) or gpio,
3732 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
3733 reboot_force is either force or not specified,
3734 reboot_cpu is s[mp]#### with #### being the processor
3735 to be used for rebooting.
3738 [KNL, SMP] Set scheduler's default relax_domain_level.
3739 See Documentation/cgroup-v1/cpusets.txt.
3741 reserve= [KNL,BUGS] Force the kernel to ignore some iomem area
3743 reservetop= [X86-32]
3745 Reserves a hole at the top of the kernel virtual
3750 Set the amount of memory to reserve for BIOS at
3751 the bottom of the address space.
3753 reset_devices [KNL] Force drivers to reset the underlying device
3754 during initialization.
3757 Specify the partition device for software suspend
3759 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
3761 resume_offset= [SWSUSP]
3762 Specify the offset from the beginning of the partition
3763 given by "resume=" at which the swap header is located,
3764 in <PAGE_SIZE> units (needed only for swap files).
3765 See Documentation/power/swsusp-and-swap-files.txt
3767 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
3768 read the resume files
3770 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
3771 Useful for devices that are detected asynchronously
3772 (e.g. USB and MMC devices).
3774 hibernate= [HIBERNATION]
3775 noresume Don't check if there's a hibernation image
3776 present during boot.
3777 nocompress Don't compress/decompress hibernation images.
3778 no Disable hibernation and resume.
3779 protect_image Turn on image protection during restoration
3780 (that will set all pages holding image data
3781 during restoration read-only).
3783 retain_initrd [RAM] Keep initrd memory after extraction
3785 rfkill.default_state=
3786 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
3787 etc. communication is blocked by default.
3790 rfkill.master_switch_mode=
3791 0 The "airplane mode" button does nothing.
3792 1 The "airplane mode" button toggles between everything
3793 blocked and the previous configuration.
3794 2 The "airplane mode" button toggles between everything
3795 blocked and everything unblocked.
3797 rhash_entries= [KNL,NET]
3798 Set number of hash buckets for route cache
3801 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
3804 ro [KNL] Mount root device read-only on boot
3807 on Mark read-only kernel memory as read-only (default).
3808 off Leave read-only kernel memory writable for debugging.
3811 Enable the uart passthrough on the designated usb port
3812 on Rockchip SoCs. When active, the signals of the
3813 debug-uart get routed to the D+ and D- pins of the usb
3814 port and the regular usb controller gets disabled.
3816 root= [KNL] Root filesystem
3817 See name_to_dev_t comment in init/do_mounts.c.
3819 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
3820 mount the root filesystem
3822 rootflags= [KNL] Set root filesystem mount option string
3824 rootfstype= [KNL] Set root filesystem type
3826 rootwait [KNL] Wait (indefinitely) for root device to show up.
3827 Useful for devices that are detected asynchronously
3828 (e.g. USB and MMC devices).
3830 rproc_mem=nn[KMG][@address]
3831 [KNL,ARM,CMA] Remoteproc physical memory block.
3832 Memory area to be used by remote processor image,
3835 rw [KNL] Mount root device read-write on boot
3837 S [KNL] Run init in single mode
3839 s390_iommu= [HW,S390]
3840 Set s390 IOTLB flushing mode
3842 With strict flushing every unmap operation will result in
3843 an IOTLB flush. Default is lazy flushing before reuse,
3847 See drivers/net/irda/sa1100_ir.c.
3849 sbni= [NET] Granch SBNI12 leased line adapter
3851 sched_debug [KNL] Enables verbose scheduler debug messages.
3853 schedstats= [KNL,X86] Enable or disable scheduled statistics.
3854 Allowed values are enable and disable. This feature
3855 incurs a small amount of overhead in the scheduler
3856 but is useful for debugging and performance tuning.
3858 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
3859 xtime_lock contention on larger systems, and/or RCU lock
3860 contention on all systems with CONFIG_MAXSMP set.
3861 Format: { "0" | "1" }
3862 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
3864 Note: increases power consumption, thus should only be
3865 enabled if running jitter sensitive (HPC/RT) workloads.
3867 security= [SECURITY] Choose a security module to enable at boot.
3868 If this boot parameter is not specified, only the first
3869 security module asking for security registration will be
3870 loaded. An invalid security module name will be treated
3871 as if no module has been chosen.
3873 selinux= [SELINUX] Disable or enable SELinux at boot time.
3874 Format: { "0" | "1" }
3875 See security/selinux/Kconfig help text.
3878 Default value is set via kernel config option.
3879 If enabled at boot time, /selinux/disable can be used
3880 later to disable prior to initial policy load.
3882 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
3883 Format: { "0" | "1" }
3884 See security/apparmor/Kconfig help text
3887 Default value is set via kernel config option.
3889 serialnumber [BUGS=X86-32]
3892 Maximal number of shapers.
3900 Disable merging of slabs with similar size. May be
3901 necessary if there is some reason to distinguish
3902 allocs to different slabs, especially in hardened
3903 environments where the risk of heap overflows and
3904 layout control by attackers can usually be
3905 frustrated by disabling merging. This will reduce
3906 most of the exposure of a heap attack to a single
3907 cache (risks via metadata attacks are mostly
3908 unchanged). Debug options disable merging on their
3910 For more information see Documentation/vm/slub.txt.
3912 slab_max_order= [MM, SLAB]
3913 Determines the maximum allowed order for slabs.
3914 A high setting may cause OOMs due to memory
3915 fragmentation. Defaults to 1 for systems with
3916 more than 32MB of RAM, 0 otherwise.
3918 slub_debug[=options[,slabs]] [MM, SLUB]
3919 Enabling slub_debug allows one to determine the
3920 culprit if slab objects become corrupted. Enabling
3921 slub_debug can create guard zones around objects and
3922 may poison objects when not in use. Also tracks the
3923 last alloc / free. For more information see
3924 Documentation/vm/slub.txt.
3926 slub_memcg_sysfs= [MM, SLUB]
3927 Determines whether to enable sysfs directories for
3928 memory cgroup sub-caches. 1 to enable, 0 to disable.
3929 The default is determined by CONFIG_SLUB_MEMCG_SYSFS_ON.
3930 Enabling this can lead to a very high number of debug
3931 directories and files being created under
3934 slub_max_order= [MM, SLUB]
3935 Determines the maximum allowed order for slabs.
3936 A high setting may cause OOMs due to memory
3937 fragmentation. For more information see
3938 Documentation/vm/slub.txt.
3940 slub_min_objects= [MM, SLUB]
3941 The minimum number of objects per slab. SLUB will
3942 increase the slab order up to slub_max_order to
3943 generate a sufficiently large slab able to contain
3944 the number of objects indicated. The higher the number
3945 of objects the smaller the overhead of tracking slabs
3946 and the less frequently locks need to be acquired.
3947 For more information see Documentation/vm/slub.txt.
3949 slub_min_order= [MM, SLUB]
3950 Determines the minimum page order for slabs. Must be
3951 lower than slub_max_order.
3952 For more information see Documentation/vm/slub.txt.
3954 slub_nomerge [MM, SLUB]
3955 Same with slab_nomerge. This is supported for legacy.
3956 See slab_nomerge for more information.
3959 Format: <io1>[,<io2>[,...,<io8>]]
3961 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
3962 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
3963 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
3964 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
3965 smsc-ircc2.ircc_irq= [HW] IRQ line
3966 smsc-ircc2.ircc_dma= [HW] DMA channel
3967 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
3968 0: Toshiba Satellite 1800 (GP data pin select)
3969 1: Fast pin select (default)
3972 smt [KNL,S390] Set the maximum number of threads (logical
3973 CPUs) to use per physical CPU on systems capable of
3974 symmetric multithreading (SMT). Will be capped to the
3975 actual hardware limit.
3977 Default: -1 (no limit)
3980 [KNL] Should the soft-lockup detector generate panics.
3983 softlockup_all_cpu_backtrace=
3984 [KNL] Should the soft-lockup detector generate
3985 backtraces on all cpus.
3988 sonypi.*= [HW] Sony Programmable I/O Control Device driver
3989 See Documentation/laptops/sonypi.txt
3991 spectre_v2= [X86] Control mitigation of Spectre variant 2
3992 (indirect branch speculation) vulnerability.
3994 on - unconditionally enable
3995 off - unconditionally disable
3996 auto - kernel detects whether your CPU model is
3999 Selecting 'on' will, and 'auto' may, choose a
4000 mitigation method at run time according to the
4001 CPU, the available microcode, the setting of the
4002 CONFIG_RETPOLINE configuration option, and the
4003 compiler with which the kernel was built.
4005 Specific mitigations can also be selected manually:
4007 retpoline - replace indirect branches
4008 retpoline,generic - google's original retpoline
4009 retpoline,amd - AMD-specific minimal thunk
4011 Not specifying this option is equivalent to
4014 spec_store_bypass_disable=
4015 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
4016 (Speculative Store Bypass vulnerability)
4018 Certain CPUs are vulnerable to an exploit against a
4019 a common industry wide performance optimization known
4020 as "Speculative Store Bypass" in which recent stores
4021 to the same memory location may not be observed by
4022 later loads during speculative execution. The idea
4023 is that such stores are unlikely and that they can
4024 be detected prior to instruction retirement at the
4025 end of a particular speculation execution window.
4027 In vulnerable processors, the speculatively forwarded
4028 store can be used in a cache side channel attack, for
4029 example to read memory to which the attacker does not
4030 directly have access (e.g. inside sandboxed code).
4032 This parameter controls whether the Speculative Store
4033 Bypass optimization is used.
4035 on - Unconditionally disable Speculative Store Bypass
4036 off - Unconditionally enable Speculative Store Bypass
4037 auto - Kernel detects whether the CPU model contains an
4038 implementation of Speculative Store Bypass and
4039 picks the most appropriate mitigation. If the
4040 CPU is not vulnerable, "off" is selected. If the
4041 CPU is vulnerable the default mitigation is
4042 architecture and Kconfig dependent. See below.
4043 prctl - Control Speculative Store Bypass per thread
4044 via prctl. Speculative Store Bypass is enabled
4045 for a process by default. The state of the control
4046 is inherited on fork.
4047 seccomp - Same as "prctl" above, but all seccomp threads
4048 will disable SSB unless they explicitly opt out.
4050 Not specifying this option is equivalent to
4051 spec_store_bypass_disable=auto.
4053 Default mitigations:
4054 X86: If CONFIG_SECCOMP=y "seccomp", otherwise "prctl"
4056 spia_io_base= [HW,MTD]
4061 srcutree.counter_wrap_check [KNL]
4062 Specifies how frequently to check for
4063 grace-period sequence counter wrap for the
4064 srcu_data structure's ->srcu_gp_seq_needed field.
4065 The greater the number of bits set in this kernel
4066 parameter, the less frequently counter wrap will
4067 be checked for. Note that the bottom two bits
4070 srcutree.exp_holdoff [KNL]
4071 Specifies how many nanoseconds must elapse
4072 since the end of the last SRCU grace period for
4073 a given srcu_struct until the next normal SRCU
4074 grace period will be considered for automatic
4075 expediting. Set to zero to disable automatic
4079 Speculative Store Bypass Disable control
4081 On CPUs that are vulnerable to the Speculative
4082 Store Bypass vulnerability and offer a
4083 firmware based mitigation, this parameter
4084 indicates how the mitigation should be used:
4086 force-on: Unconditionally enable mitigation for
4087 for both kernel and userspace
4088 force-off: Unconditionally disable mitigation for
4089 for both kernel and userspace
4090 kernel: Always enable mitigation in the
4091 kernel, and offer a prctl interface
4092 to allow userspace to register its
4093 interest in being mitigated too.
4095 stack_guard_gap= [MM]
4096 override the default stack gap protection. The value
4097 is in page units and it defines how many pages prior
4098 to (for stacks growing down) resp. after (for stacks
4099 growing up) the main stack are reserved for no other
4100 mapping. Default value is 256 pages.
4103 Enabled the stack tracer on boot up.
4105 stacktrace_filter=[function-list]
4106 [FTRACE] Limit the functions that the stack tracer
4107 will trace at boot up. function-list is a comma separated
4108 list of functions. This list can be changed at run
4109 time by the stack_trace_filter file in the debugfs
4110 tracing directory. Note, this enables stack tracing
4111 and the stacktrace above is not needed.
4115 Set the STI (builtin display/keyboard on the HP-PARISC
4116 machines) console (graphic card) which should be used
4117 as the initial boot-console.
4118 See also comment in drivers/video/console/sticore.c.
4121 See comment in drivers/video/console/sticore.c.
4124 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
4126 sunrpc.min_resvport=
4127 sunrpc.max_resvport=
4129 SunRPC servers often require that client requests
4130 originate from a privileged port (i.e. a port in the
4131 range 0 < portnr < 1024).
4132 An administrator who wishes to reserve some of these
4133 ports for other uses may adjust the range that the
4134 kernel's sunrpc client considers to be privileged
4135 using these two parameters to set the minimum and
4136 maximum port values.
4138 sunrpc.svc_rpc_per_connection_limit=
4140 Limit the number of requests that the server will
4141 process in parallel from a single connection.
4142 The default value is 0 (no limit).
4146 Control how the NFS server code allocates CPUs to
4147 service thread pools. Depending on how many NICs
4148 you have and where their interrupts are bound, this
4149 option will affect which CPUs will do NFS serving.
4150 Note: this parameter cannot be changed while the
4151 NFS server is running.
4153 auto the server chooses an appropriate mode
4154 automatically using heuristics
4155 global a single global pool contains all CPUs
4156 percpu one pool for each CPU
4157 pernode one pool for each NUMA node (equivalent
4158 to global on non-NUMA machines)
4160 sunrpc.tcp_slot_table_entries=
4161 sunrpc.udp_slot_table_entries=
4163 Sets the upper limit on the number of simultaneous
4164 RPC calls that can be sent from the client to a
4165 server. Increasing these values may allow you to
4166 improve throughput, but will also increase the
4167 amount of memory reserved for use by the client.
4169 suspend.pm_test_delay=
4171 Sets the number of seconds to remain in a suspend test
4172 mode before resuming the system (see
4173 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
4174 is set. Default value is 5.
4177 [KNL] Enable accounting of swap in memory resource
4178 controller if no parameter or 1 is given or disable
4179 it if 0 is given (See Documentation/cgroup-v1/memory.txt)
4181 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
4182 Format: { <int> | force | noforce }
4183 <int> -- Number of I/O TLB slabs
4184 force -- force using of bounce buffers even if they
4185 wouldn't be automatically used by the kernel
4186 noforce -- Never use bounce buffers (for debugging)
4190 sysfs.deprecated=0|1 [KNL]
4191 Enable/disable old style sysfs layout for old udev
4192 on older distributions. When this option is enabled
4193 very new udev will not work anymore. When this option
4194 is disabled (or CONFIG_SYSFS_DEPRECATED not compiled)
4195 in older udev will not work anymore.
4196 Default depends on CONFIG_SYSFS_DEPRECATED_V2 set in
4197 the kernel configuration.
4199 sysrq_always_enabled
4201 Ignore sysrq setting - this boot parameter will
4202 neutralize any effect of /proc/sys/kernel/sysrq.
4203 Useful for debugging.
4205 tcpmhash_entries= [KNL,NET]
4206 Set the number of tcp_metrics_hash slots.
4207 Default value is 8192 or 16384 depending on total
4208 ram pages. This is used to specify the TCP metrics
4209 cache size. See Documentation/networking/ip-sysctl.txt
4210 "tcp_no_metrics_save" section for more details.
4214 test_suspend= [SUSPEND][,N]
4215 Specify "mem" (for Suspend-to-RAM) or "standby" (for
4216 standby suspend) or "freeze" (for suspend type freeze)
4217 as the system sleep state during system startup with
4218 the optional capability to repeat N number of times.
4219 The system is woken from this state using a
4220 wakeup-capable RTC alarm.
4222 thash_entries= [KNL,NET]
4223 Set number of hash buckets for TCP connection
4225 thermal.act= [HW,ACPI]
4226 -1: disable all active trip points in all thermal zones
4227 <degrees C>: override all lowest active trip points
4229 thermal.crt= [HW,ACPI]
4230 -1: disable all critical trip points in all thermal zones
4231 <degrees C>: override all critical trip points
4233 thermal.nocrt= [HW,ACPI]
4234 Set to disable actions on ACPI thermal zone
4235 critical and hot trip points.
4237 thermal.off= [HW,ACPI]
4238 1: disable ACPI thermal control
4240 thermal.psv= [HW,ACPI]
4241 -1: disable all passive trip points
4242 <degrees C>: override all passive trip points to this
4245 thermal.tzp= [HW,ACPI]
4246 Specify global default ACPI thermal zone polling rate
4247 <deci-seconds>: poll all this frequency
4248 0: no polling (default)
4251 Force threading of all interrupt handlers except those
4252 marked explicitly IRQF_NO_THREAD.
4255 Enable the Transcendent memory driver if built-in.
4257 tmem.cleancache=0|1 [KNL, XEN]
4258 Default is on (1). Disable the usage of the cleancache
4259 API to send anonymous pages to the hypervisor.
4261 tmem.frontswap=0|1 [KNL, XEN]
4262 Default is on (1). Disable the usage of the frontswap
4263 API to send swap pages to the hypervisor. If disabled
4264 the selfballooning and selfshrinking are force disabled.
4266 tmem.selfballooning=0|1 [KNL, XEN]
4267 Default is on (1). Disable the driving of swap pages
4270 tmem.selfshrinking=0|1 [KNL, XEN]
4271 Default is on (1). Partial swapoff that immediately
4272 transfers pages from Xen hypervisor back to the
4273 kernel based on different criteria.
4277 Specify if the kernel should make use of the cpu
4278 topology information if the hardware supports this.
4279 The scheduler will make use of this information and
4280 e.g. base its process migration decisions on it.
4283 topology_updates= [KNL, PPC, NUMA]
4285 Specify if the kernel should ignore (off)
4286 topology updates sent by the hypervisor to this
4291 tpm_suspend_pcr=[HW,TPM]
4292 Format: integer pcr id
4293 Specify that at suspend time, the tpm driver
4294 should extend the specified pcr with zeros,
4295 as a workaround for some chips which fail to
4296 flush the last written pcr on TPM_SaveState.
4297 This will guarantee that all the other pcrs
4300 trace_buf_size=nn[KMG]
4301 [FTRACE] will set tracing buffer size on each cpu.
4303 trace_event=[event-list]
4304 [FTRACE] Set and start specified trace events in order
4305 to facilitate early boot debugging. The event-list is a
4306 comma separated list of trace events to enable. See
4307 also Documentation/trace/events.txt
4309 trace_options=[option-list]
4310 [FTRACE] Enable or disable tracer options at boot.
4311 The option-list is a comma delimited list of options
4312 that can be enabled or disabled just as if you were
4313 to echo the option name into
4315 /sys/kernel/debug/tracing/trace_options
4317 For example, to enable stacktrace option (to dump the
4318 stack trace of each event), add to the command line:
4320 trace_options=stacktrace
4322 See also Documentation/trace/ftrace.txt "trace options"
4326 Have the tracepoints sent to printk as well as the
4327 tracing ring buffer. This is useful for early boot up
4328 where the system hangs or reboots and does not give the
4329 option for reading the tracing buffer or performing a
4330 ftrace_dump_on_oops.
4332 To turn off having tracepoints sent to printk,
4333 echo 0 > /proc/sys/kernel/tracepoint_printk
4334 Note, echoing 1 into this file without the
4335 tracepoint_printk kernel cmdline option has no effect.
4339 Having tracepoints sent to printk() and activating high
4340 frequency tracepoints such as irq or sched, can cause
4341 the system to live lock.
4344 [FTRACE] enable this option to disable tracing when a
4345 warning is hit. This turns off "tracing_on". Tracing can
4346 be enabled again by echoing '1' into the "tracing_on"
4347 file located in /sys/kernel/debug/tracing/
4349 This option is useful, as it disables the trace before
4350 the WARNING dump is called, which prevents the trace to
4351 be filled with content caused by the warning output.
4353 This option can also be set at run time via the sysctl
4354 option: kernel/traceoff_on_warning
4356 transparent_hugepage=
4358 Format: [always|madvise|never]
4359 Can be used to control the default behavior of the system
4360 with respect to transparent hugepages.
4361 See Documentation/vm/transhuge.txt for more details.
4363 tsc= Disable clocksource stability checks for TSC.
4365 [x86] reliable: mark tsc clocksource as reliable, this
4366 disables clocksource verification at runtime, as well
4367 as the stability checks done at bootup. Used to enable
4368 high-resolution timer mode on older hardware, and in
4369 virtualized environment.
4370 [x86] noirqtime: Do not use TSC to do irq accounting.
4371 Used to run time disable IRQ_TIME_ACCOUNTING on any
4372 platforms where RDTSC is slow and this accounting
4375 turbografx.map[2|3]= [HW,JOY]
4376 TurboGraFX parallel port interface
4378 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
4379 See also Documentation/input/joystick-parport.txt
4381 udbg-immortal [PPC] When debugging early kernel crashes that
4382 happen after console_init() and before a proper
4383 console driver takes over, this boot options might
4384 help "seeing" what's going on.
4386 uhash_entries= [KNL,NET]
4387 Set number of hash buckets for UDP/UDP-Lite connections
4390 [USB] Ignore overcurrent events (default N).
4391 Some badly-designed motherboards generate lots of
4392 bogus events, for ports that aren't wired to
4393 anything. Set this parameter to avoid log spamming.
4394 Note that genuine overcurrent events won't be
4398 [X86] Cause panic on unknown NMI.
4400 usbcore.authorized_default=
4401 [USB] Default USB device authorization:
4402 (default -1 = authorized except for wireless USB,
4403 0 = not authorized, 1 = authorized)
4405 usbcore.autosuspend=
4406 [USB] The autosuspend time delay (in seconds) used
4407 for newly-detected USB devices (default 2). This
4408 is the time required before an idle device will be
4409 autosuspended. Devices for which the delay is set
4410 to a negative value won't be autosuspended at all.
4412 usbcore.usbfs_snoop=
4413 [USB] Set to log all usbfs traffic (default 0 = off).
4415 usbcore.usbfs_snoop_max=
4416 [USB] Maximum number of bytes to snoop in each URB
4419 usbcore.blinkenlights=
4420 [USB] Set to cycle leds on hubs (default 0 = off).
4422 usbcore.old_scheme_first=
4423 [USB] Start with the old device initialization
4424 scheme (default 0 = off).
4426 usbcore.usbfs_memory_mb=
4427 [USB] Memory limit (in MB) for buffers allocated by
4428 usbfs (default = 16, 0 = max = 2047).
4430 usbcore.use_both_schemes=
4431 [USB] Try the other device initialization scheme
4432 if the first one fails (default 1 = enabled).
4434 usbcore.initial_descriptor_timeout=
4435 [USB] Specifies timeout for the initial 64-byte
4436 USB_REQ_GET_DESCRIPTOR request in milliseconds
4437 (default 5000 = 5.0 seconds).
4439 usbcore.nousb [USB] Disable the USB subsystem
4442 [USBHID] The interval which mice are to be polled at.
4445 [USBHID] The interval which joysticks are to be polled at.
4447 usb-storage.delay_use=
4448 [UMS] The delay in seconds before a new device is
4449 scanned for Logical Units (default 1).
4452 [UMS] A list of quirks entries to supplement or
4453 override the built-in unusual_devs list. List
4454 entries are separated by commas. Each entry has
4455 the form VID:PID:Flags where VID and PID are Vendor
4456 and Product ID values (4-digit hex numbers) and
4457 Flags is a set of characters, each corresponding
4458 to a common usb-storage quirk flag as follows:
4459 a = SANE_SENSE (collect more than 18 bytes
4461 b = BAD_SENSE (don't collect more than 18
4462 bytes of sense data);
4463 c = FIX_CAPACITY (decrease the reported
4464 device capacity by one sector);
4465 d = NO_READ_DISC_INFO (don't use
4466 READ_DISC_INFO command);
4467 e = NO_READ_CAPACITY_16 (don't use
4468 READ_CAPACITY_16 command);
4469 f = NO_REPORT_OPCODES (don't use report opcodes
4471 g = MAX_SECTORS_240 (don't transfer more than
4472 240 sectors at a time, uas only);
4473 h = CAPACITY_HEURISTICS (decrease the
4474 reported device capacity by one
4475 sector if the number is odd);
4476 i = IGNORE_DEVICE (don't bind to this
4478 j = NO_REPORT_LUNS (don't use report luns
4480 l = NOT_LOCKABLE (don't try to lock and
4481 unlock ejectable media);
4482 m = MAX_SECTORS_64 (don't transfer more
4483 than 64 sectors = 32 KB at a time);
4484 n = INITIAL_READ10 (force a retry of the
4485 initial READ(10) command);
4486 o = CAPACITY_OK (accept the capacity
4487 reported by the device);
4488 p = WRITE_CACHE (the device cache is ON
4490 r = IGNORE_RESIDUE (the device reports
4491 bogus residue values);
4492 s = SINGLE_LUN (the device has only one
4494 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
4495 commands, uas only);
4496 u = IGNORE_UAS (don't bind to the uas driver);
4497 w = NO_WP_DETECT (don't test whether the
4498 medium is write-protected).
4499 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
4500 even if the device claims no cache)
4501 Example: quirks=0419:aaf5:rl,0421:0433:rc
4503 user_debug= [KNL,ARM]
4505 See arch/arm/Kconfig.debug help text.
4506 1 - undefined instruction events
4508 4 - invalid data aborts
4511 Example: user_debug=31
4514 [X86] Flags controlling user PTE allocations.
4516 nohigh = do not allocate PTE pages in
4517 HIGHMEM regardless of setting
4521 On X86_32, this is an alias for vdso32=. Otherwise:
4523 vdso=1: enable VDSO (the default)
4524 vdso=0: disable VDSO mapping
4526 vdso32= [X86] Control the 32-bit vDSO
4527 vdso32=1: enable 32-bit VDSO
4528 vdso32=0 or vdso32=2: disable 32-bit VDSO
4530 See the help text for CONFIG_COMPAT_VDSO for more
4531 details. If CONFIG_COMPAT_VDSO is set, the default is
4532 vdso32=0; otherwise, the default is vdso32=1.
4534 For compatibility with older kernels, vdso32=2 is an
4537 Try vdso32=0 if you encounter an error that says:
4538 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
4541 vector=percpu: enable percpu vector domain
4543 video= [FB] Frame buffer configuration
4544 See Documentation/fb/modedb.txt.
4546 video.brightness_switch_enabled= [0,1]
4547 If set to 1, on receiving an ACPI notify event
4548 generated by hotkey, video driver will adjust brightness
4549 level and then send out the event to user space through
4550 the allocated input device; If set to 0, video driver
4551 will only send out the event without touching backlight
4556 [VMMIO] Memory mapped virtio (platform) device.
4558 <size>@<baseaddr>:<irq>[:<id>]
4560 <size> := size (can use standard suffixes
4562 <baseaddr> := physical base address
4563 <irq> := interrupt number (as passed to
4565 <id> := (optional) platform device id
4567 virtio_mmio.device=1K@0x100b0000:48:7
4569 Can be used multiple times for multiple devices.
4571 vga= [BOOT,X86-32] Select a particular video mode
4572 See Documentation/x86/boot.txt and
4573 Documentation/svga.txt.
4574 Use vga=ask for menu.
4575 This is actually a boot loader parameter; the value is
4576 passed to the kernel using a special protocol.
4578 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
4579 size of <nn>. This can be used to increase the
4580 minimum size (128MB on x86). It can also be used to
4581 decrease the size and leave more room for directly
4584 vmcp_cma=nn[MG] [KNL,S390]
4585 Sets the memory size reserved for contiguous memory
4586 allocations for the vmcp device driver.
4588 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
4591 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
4594 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
4598 Controls the behavior of vsyscalls (i.e. calls to
4599 fixed addresses of 0xffffffffff600x00 from legacy
4600 code). Most statically-linked binaries and older
4601 versions of glibc use these calls. Because these
4602 functions are at fixed addresses, they make nice
4603 targets for exploits that can control RIP.
4605 emulate [default] Vsyscalls turn into traps and are
4606 emulated reasonably safely.
4608 native Vsyscalls are native syscall instructions.
4609 This is a little bit faster than trapping
4610 and makes a few dynamic recompilers work
4611 better than they would in emulation mode.
4612 It also makes exploits much easier to write.
4614 none Vsyscalls don't work at all. This makes
4615 them quite hard to use for exploits but
4616 might break your system.
4618 vt.color= [VT] Default text color.
4619 Format: 0xYX, X = foreground, Y = background.
4620 Default: 0x07 = light gray on black.
4622 vt.cur_default= [VT] Default cursor shape.
4623 Format: 0xCCBBAA, where AA, BB, and CC are the same as
4624 the parameters of the <Esc>[?A;B;Cc escape sequence;
4625 see VGA-softcursor.txt. Default: 2 = underline.
4627 vt.default_blu= [VT]
4628 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
4629 Change the default blue palette of the console.
4630 This is a 16-member array composed of values
4633 vt.default_grn= [VT]
4634 Format: <green0>,<green1>,<green2>,...,<green15>
4635 Change the default green palette of the console.
4636 This is a 16-member array composed of values
4639 vt.default_red= [VT]
4640 Format: <red0>,<red1>,<red2>,...,<red15>
4641 Change the default red palette of the console.
4642 This is a 16-member array composed of values
4648 Set system-wide default UTF-8 mode for all tty's.
4649 Default is 1, i.e. UTF-8 mode is enabled for all
4650 newly opened terminals.
4652 vt.global_cursor_default=
4655 Set system-wide default for whether a cursor
4656 is shown on new VTs. Default is -1,
4657 i.e. cursors will be created by default unless
4658 overridden by individual drivers. 0 will hide
4659 cursors, 1 will display them.
4661 vt.italic= [VT] Default color for italic text; 0-15.
4664 vt.underline= [VT] Default color for underlined text; 0-15.
4667 watchdog timers [HW,WDT] For information on watchdog timers,
4668 see Documentation/watchdog/watchdog-parameters.txt
4669 or other driver-specific files in the
4670 Documentation/watchdog/ directory.
4672 workqueue.watchdog_thresh=
4673 If CONFIG_WQ_WATCHDOG is configured, workqueue can
4674 warn stall conditions and dump internal state to
4675 help debugging. 0 disables workqueue stall
4676 detection; otherwise, it's the stall threshold
4677 duration in seconds. The default value is 30 and
4678 it can be updated at runtime by writing to the
4679 corresponding sysfs file.
4681 workqueue.disable_numa
4682 By default, all work items queued to unbound
4683 workqueues are affine to the NUMA nodes they're
4684 issued on, which results in better behavior in
4685 general. If NUMA affinity needs to be disabled for
4686 whatever reason, this option can be used. Note
4687 that this also can be controlled per-workqueue for
4688 workqueues visible under /sys/bus/workqueue/.
4690 workqueue.power_efficient
4691 Per-cpu workqueues are generally preferred because
4692 they show better performance thanks to cache
4693 locality; unfortunately, per-cpu workqueues tend to
4694 be more power hungry than unbound workqueues.
4696 Enabling this makes the per-cpu workqueues which
4697 were observed to contribute significantly to power
4698 consumption unbound, leading to measurably lower
4699 power usage at the cost of small performance
4702 The default value of this parameter is determined by
4703 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
4705 workqueue.debug_force_rr_cpu
4706 Workqueue used to implicitly guarantee that work
4707 items queued without explicit CPU specified are put
4708 on the local CPU. This guarantee is no longer true
4709 and while local CPU is still preferred work items
4710 may be put on foreign CPUs. This debug option
4711 forces round-robin CPU selection to flush out
4712 usages which depend on the now broken guarantee.
4713 When enabled, memory and cache locality will be
4716 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
4717 default x2apic cluster mode on platforms
4720 x86_intel_mid_timer= [X86-32,APBT]
4721 Choose timer option for x86 Intel MID platform.
4722 Two valid options are apbt timer only and lapic timer
4723 plus one apbt timer for broadcast timer.
4724 x86_intel_mid_timer=apbt_only | lapic_and_apbt
4726 xen_512gb_limit [KNL,X86-64,XEN]
4727 Restricts the kernel running paravirtualized under Xen
4728 to use only up to 512 GB of RAM. The reason to do so is
4729 crash analysis tools and Xen tools for doing domain
4730 save/restore/migration must be enabled to handle larger
4733 xen_emul_unplug= [HW,X86,XEN]
4734 Unplug Xen emulated devices
4735 Format: [unplug0,][unplug1]
4736 ide-disks -- unplug primary master IDE devices
4737 aux-ide-disks -- unplug non-primary-master IDE devices
4738 nics -- unplug network devices
4739 all -- unplug all emulated devices (NICs and IDE disks)
4740 unnecessary -- unplugging emulated devices is
4741 unnecessary even if the host did not respond to
4743 never -- do not unplug even if version check succeeds
4745 xen_nopvspin [X86,XEN]
4746 Disables the ticketlock slowpath using Xen PV
4750 Disables the PV optimizations forcing the HVM guest to
4751 run as generic HVM guest with no PV drivers.
4753 xirc2ps_cs= [NET,PCMCIA]
4755 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]