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_no_watchdog [HW,ACPI,WDT]
141 Ignore the ACPI-based watchdog interface (WDAT) and let
142 a native driver control the watchdog device instead.
144 acpi_rsdp= [ACPI,EFI,KEXEC]
145 Pass the RSDP address to the kernel, mostly used
146 on machines running EFI runtime service to boot the
147 second kernel for kdump.
149 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS
150 Format: To spoof as Windows 98: ="Microsoft Windows"
152 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead
153 of 2 which is mandated by ACPI 6) as the supported ACPI
154 specification revision (when using this switch, it may
155 be necessary to carry out a cold reboot _twice_ in a
156 row to make it take effect on the platform firmware).
158 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings
159 acpi_osi="string1" # add string1
160 acpi_osi="!string2" # remove string2
161 acpi_osi=!* # remove all strings
162 acpi_osi=! # disable all built-in OS vendor
164 acpi_osi=!! # enable all built-in OS vendor
166 acpi_osi= # disable all strings
168 'acpi_osi=!' can be used in combination with single or
169 multiple 'acpi_osi="string1"' to support specific OS
170 vendor string(s). Note that such command can only
171 affect the default state of the OS vendor strings, thus
172 it cannot affect the default state of the feature group
173 strings and the current state of the OS vendor strings,
174 specifying it multiple times through kernel command line
175 is meaningless. This command is useful when one do not
176 care about the state of the feature group strings which
177 should be controlled by the OSPM.
179 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent
180 to 'acpi_osi="Windows 2000" acpi_osi=!', they all
181 can make '_OSI("Windows 2000")' TRUE.
183 'acpi_osi=' cannot be used in combination with other
184 'acpi_osi=' command lines, the _OSI method will not
185 exist in the ACPI namespace. NOTE that such command can
186 only affect the _OSI support state, thus specifying it
187 multiple times through kernel command line is also
190 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)'
193 'acpi_osi=!*' can be used in combination with single or
194 multiple 'acpi_osi="string1"' to support specific
195 string(s). Note that such command can affect the
196 current state of both the OS vendor strings and the
197 feature group strings, thus specifying it multiple times
198 through kernel command line is meaningful. But it may
199 still not able to affect the final state of a string if
200 there are quirks related to this string. This command
201 is useful when one want to control the state of the
202 feature group strings to debug BIOS issues related to
205 1. 'acpi_osi="Module Device" acpi_osi=!*' can make
206 '_OSI("Module Device")' FALSE.
207 2. 'acpi_osi=!* acpi_osi="Module Device"' can make
208 '_OSI("Module Device")' TRUE.
209 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is
211 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"'
213 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!',
214 they all will make '_OSI("Windows 2000")' TRUE.
217 Override the pmtimer bug detection: force the kernel
218 to assume that this machine's pmtimer latches its value
219 and always returns good values.
221 acpi_sci= [HW,ACPI] ACPI System Control Interrupt trigger mode
222 Format: { level | edge | high | low }
224 acpi_skip_timer_override [HW,ACPI]
225 Recognize and ignore IRQ0/pin2 Interrupt Override.
226 For broken nForce2 BIOS resulting in XT-PIC timer.
228 acpi_sleep= [HW,ACPI] Sleep options
229 Format: { s3_bios, s3_mode, s3_beep, s4_nohwsig,
230 old_ordering, nonvs, sci_force_enable }
231 See Documentation/power/video.txt for information on
233 s3_beep is for debugging; it makes the PC's speaker beep
234 as soon as the kernel's real-mode entry point is called.
235 s4_nohwsig prevents ACPI hardware signature from being
236 used during resume from hibernation.
237 old_ordering causes the ACPI 1.0 ordering of the _PTS
238 control method, with respect to putting devices into
239 low power states, to be enforced (the ACPI 2.0 ordering
240 of _PTS is used by default).
241 nonvs prevents the kernel from saving/restoring the
242 ACPI NVS memory during suspend/hibernation and resume.
243 sci_force_enable causes the kernel to set SCI_EN directly
244 on resume from S1/S3 (which is against the ACPI spec,
245 but some broken systems don't work without it).
247 acpi_use_timer_override [HW,ACPI]
248 Use timer override. For some broken Nvidia NF5 boards
249 that require a timer override, but don't have HPET
251 add_efi_memmap [EFI; X86] Include EFI memory map in
252 kernel's map of available physical RAM.
255 { off | try_unsupported }
256 off: disable AGP support
257 try_unsupported: try to drive unsupported chipsets
258 (may crash computer or cause data corruption)
261 See Documentation/sound/alsa/alsa-parameters.txt
264 Allow the default userspace alignment fault handler
265 behaviour to be specified. Bit 0 enables warnings,
266 bit 1 enables fixups, and bit 2 sends a segfault.
268 align_va_addr= [X86-64]
269 Align virtual addresses by clearing slice [14:12] when
270 allocating a VMA at process creation time. This option
271 gives you up to 3% performance improvement on AMD F15h
272 machines (where it is enabled by default) for a
273 CPU-intensive style benchmark, and it can vary highly in
274 a microbenchmark depending on workload and compiler.
276 32: only for 32-bit processes
277 64: only for 64-bit processes
278 on: enable for both 32- and 64-bit processes
279 off: disable for both 32- and 64-bit processes
281 alloc_snapshot [FTRACE]
282 Allocate the ftrace snapshot buffer on boot up when the
283 main buffer is allocated. This is handy if debugging
284 and you need to use tracing_snapshot() on boot up, and
285 do not want to use tracing_snapshot_alloc() as it needs
286 to be done where GFP_KERNEL allocations are allowed.
288 amd_iommu= [HW,X86-64]
289 Pass parameters to the AMD IOMMU driver in the system.
291 fullflush - enable flushing of IO/TLB entries when
292 they are unmapped. Otherwise they are
293 flushed before they will be reused, which
295 off - do not initialize any AMD IOMMU found in
297 force_isolation - Force device isolation for all
298 devices. The IOMMU driver is not
299 allowed anymore to lift isolation
300 requirements as needed. This option
301 does not override iommu=pt
303 amd_iommu_dump= [HW,X86-64]
304 Enable AMD IOMMU driver option to dump the ACPI table
305 for AMD IOMMU. With this option enabled, AMD IOMMU
306 driver will print ACPI tables for AMD IOMMU during
307 IOMMU initialization.
309 amd_iommu_intr= [HW,X86-64]
310 Specifies one of the following AMD IOMMU interrupt
312 legacy - Use legacy interrupt remapping mode.
313 vapic - Use virtual APIC mode, which allows IOMMU
314 to inject interrupts directly into guest.
315 This mode requires kvm-amd.avic=1.
316 (Default when IOMMU HW support is present.)
318 amijoy.map= [HW,JOY] Amiga joystick support
319 Map of devices attached to JOY0DAT and JOY1DAT
321 See also Documentation/input/joystick.txt
323 analog.map= [HW,JOY] Analog joystick and gamepad support
324 Specifies type or capabilities of an analog joystick
325 connected to one of 16 gameports
326 Format: <type1>,<type2>,..<type16>
329 Power management functions (SPARCstation-4/5 + deriv.)
331 Disable APC CPU standby support. SPARCstation-Fox does
332 not play well with APC CPU idle - disable it if you have
333 APC and your system crashes randomly.
335 apic= [APIC,X86-32] Advanced Programmable Interrupt Controller
336 Change the output verbosity whilst booting
337 Format: { quiet (default) | verbose | debug }
338 Change the amount of debugging information output
339 when initialising the APIC and IO-APIC components.
341 apic_extnmi= [APIC,X86] External NMI delivery setting
342 Format: { bsp (default) | all | none }
343 bsp: External NMI is delivered only to CPU 0
344 all: External NMIs are broadcast to all CPUs as a
346 none: External NMI is masked for all CPUs. This is
347 useful so that a dump capture kernel won't be
351 See Documentation/networking/ipv6.txt.
353 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
354 Limit apic dumping. The parameter defines the maximal
355 number of local apics being dumped. Also it is possible
356 to set it to "all" by meaning -- no limit here.
357 Format: { 1 (default) | 2 | ... | all }.
358 The parameter valid if only apic=debug or
359 apic=verbose is specified.
360 Example: apic=debug show_lapic=all
362 apm= [APM] Advanced Power Management
363 See header of arch/x86/kernel/apm_32.c.
365 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
366 Format: <io>,<irq>,<nodeID>
370 atarimouse= [HW,MOUSE] Atari Mouse
372 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
373 EzKey and similar keyboards
375 atkbd.reset= [HW] Reset keyboard during initialization
377 atkbd.set= [HW] Select keyboard code set
378 Format: <int> (2 = AT (default), 3 = PS/2)
380 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
383 atkbd.softraw= [HW] Choose between synthetic and real raw mode
384 Format: <bool> (0 = real, 1 = synthetic (default))
386 atkbd.softrepeat= [HW]
387 Use software keyboard repeat
389 audit= [KNL] Enable the audit sub-system
390 Format: { "0" | "1" } (0 = disabled, 1 = enabled)
391 0 - kernel audit is disabled and can not be enabled
392 until the next reboot
393 unset - kernel audit is initialized but disabled and
394 will be fully enabled by the userspace auditd.
395 1 - kernel audit is initialized and partially enabled,
396 storing at most audit_backlog_limit messages in
397 RAM until it is fully enabled by the userspace
401 audit_backlog_limit= [KNL] Set the audit queue size limit.
402 Format: <int> (must be >=0)
405 bau= [X86_UV] Enable the BAU on SGI UV. The default
406 behavior is to disable the BAU (i.e. bau=0).
407 Format: { "0" | "1" }
410 unset - Disable the BAU.
412 baycom_epp= [HW,AX25]
415 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
417 See header of drivers/net/hamradio/baycom_par.c.
419 baycom_ser_fdx= [HW,AX25]
420 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
421 Format: <io>,<irq>,<mode>[,<baud>]
422 See header of drivers/net/hamradio/baycom_ser_fdx.c.
424 baycom_ser_hdx= [HW,AX25]
425 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
426 Format: <io>,<irq>,<mode>
427 See header of drivers/net/hamradio/baycom_ser_hdx.c.
429 blkdevparts= Manual partition parsing of block device(s) for
430 embedded devices based on command line input.
431 See Documentation/block/cmdline-partition.txt
433 boot_delay= Milliseconds to delay each printk during boot.
434 Values larger than 10 seconds (10000) are changed to
438 bootmem_debug [KNL] Enable bootmem allocator debug messages.
441 Disable BERT OS support on buggy BIOSes.
443 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
444 bttv.radio= Most important insmod options are available as
446 bttv.pll= See Documentation/video4linux/bttv/Insmod-options
449 bulk_remove=off [PPC] This parameter disables the use of the pSeries
450 firmware feature for flushing multiple hpte entries
453 c101= [NET] Moxa C101 synchronous serial card
455 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
456 Sometimes CPU hardware bugs make them report the cache
457 size incorrectly. The kernel will attempt work arounds
458 to fix known problems, but for some CPUs it is not
459 possible to determine what the correct size should be.
460 This option provides an override for these situations.
462 ca_keys= [KEYS] This parameter identifies a specific key(s) on
463 the system trusted keyring to be used for certificate
465 format: { id:<keyid> | builtin }
467 cca= [MIPS] Override the kernel pages' cache coherency
468 algorithm. Accepted values range from 0 to 7
469 inclusive. See arch/mips/include/asm/pgtable-bits.h
470 for platform specific values (SB1, Loongson3 and
473 ccw_timeout_log [S390]
474 See Documentation/s390/CommonIO for details.
476 cgroup_disable= [KNL] Disable a particular controller
477 Format: {name of the controller(s) to disable}
478 The effects of cgroup_disable=foo are:
479 - foo isn't auto-mounted if you mount all cgroups in
481 - foo isn't visible as an individually mountable
483 {Currently only "memory" controller deal with this and
484 cut the overhead, others just disable the usage. So
485 only cgroup_disable=memory is actually worthy}
487 cgroup_no_v1= [KNL] Disable one, multiple, all cgroup controllers in v1
488 Format: { controller[,controller...] | "all" }
489 Like cgroup_disable, but only applies to cgroup v1;
490 the blacklisted controllers remain available in cgroup2.
492 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
494 nosocket -- Disable socket memory accounting.
495 nokmem -- Disable kernel memory accounting.
497 checkreqprot [SELINUX] Set initial checkreqprot flag value.
498 Format: { "0" | "1" }
499 See security/selinux/Kconfig help text.
500 0 -- check protection applied by kernel (includes
501 any implied execute protection).
502 1 -- check protection requested by application.
503 Default value is set via a kernel config option.
504 Value can be changed at runtime via
505 /selinux/checkreqprot.
508 See Documentation/s390/CommonIO for details.
511 Prevents the clock framework from automatically gating
512 clocks that have not been explicitly enabled by a Linux
513 device driver but are enabled in hardware at reset or
514 by the bootloader/firmware. Note that this does not
515 force such clocks to be always-on nor does it reserve
516 those clocks in any way. This parameter is useful for
517 debug and development, but should not be needed on a
518 platform with proper driver support. For more
519 information, see Documentation/clk.txt.
521 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
523 Forces specified clocksource (if available) to be used
524 when calculating gettimeofday(). If specified
525 clocksource is not available, it defaults to PIT.
526 Format: { pit | tsc | cyclone | pmtmr }
528 clocksource= Override the default clocksource
530 Override the default clocksource and use the clocksource
531 with the name specified.
532 Some clocksource names to choose from, depending on
534 [all] jiffies (this is the base, fallback clocksource)
536 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
537 pxa_timer,timer3,32k_counter,timer0_1
538 [X86-32] pit,hpet,tsc;
539 scx200_hrt on Geode; cyclone on IBM x440
547 clocksource.arm_arch_timer.evtstrm=
550 Enable/disable the eventstream feature of the ARM
551 architected timer so that code using WFE-based polling
552 loops can be debugged more effectively on production
555 clearcpuid=BITNUM[,BITNUM...] [X86]
556 Disable CPUID feature X for the kernel. See
557 arch/x86/include/asm/cpufeatures.h for the valid bit
558 numbers. Note the Linux specific bits are not necessarily
559 stable over kernel options, but the vendor specific
561 Also note that user programs calling CPUID directly
562 or using the feature without checking anything
563 will still see it. This just prevents it from
564 being used by the kernel or shown in /proc/cpuinfo.
565 Also note the kernel might malfunction if you disable
568 cma=nn[MG]@[start[MG][-end[MG]]]
570 Sets the size of kernel global memory area for
571 contiguous memory allocations and optionally the
572 placement constraint by the physical address range of
573 memory allocations. A value of 0 disables CMA
574 altogether. For more information, see
575 include/linux/dma-contiguous.h
577 cmo_free_hint= [PPC] Format: { yes | no }
578 Specify whether pages are marked as being inactive
579 when they are freed. This is used in CMO environments
580 to determine OS memory pressure for page stealing by
584 coherent_pool=nn[KMG] [ARM,KNL]
585 Sets the size of memory pool for coherent, atomic dma
586 allocations, by default set to 256K.
588 code_bytes [X86] How many bytes of object code to print
593 com20020= [HW,NET] ARCnet - COM20020 chipset
595 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
597 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
601 ARCnet - COM90xx chipset (memory-mapped buffers)
602 Format: <io>[,<irq>[,<memstart>]]
604 condev= [HW,S390] console device
607 console= [KNL] Output console device and options.
609 tty<n> Use the virtual console device <n>.
613 Use the specified serial port. The options are of
614 the form "bbbbpnf", where "bbbb" is the baud rate,
615 "p" is parity ("n", "o", or "e"), "n" is number of
616 bits, and "f" is flow control ("r" for RTS or
617 omit it). Default is "9600n8".
619 See Documentation/admin-guide/serial-console.rst for more
621 Documentation/networking/netconsole.txt for an
624 uart[8250],io,<addr>[,options]
625 uart[8250],mmio,<addr>[,options]
626 uart[8250],mmio16,<addr>[,options]
627 uart[8250],mmio32,<addr>[,options]
628 uart[8250],0x<addr>[,options]
629 Start an early, polled-mode console on the 8250/16550
630 UART at the specified I/O port or MMIO address,
631 switching to the matching ttyS device later.
632 MMIO inter-register address stride is either 8-bit
633 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
634 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
635 to be equivalent to 'mmio'. 'options' are specified in
636 the same format described for ttyS above; if unspecified,
637 the h/w is not re-initialized.
639 hvc<n> Use the hypervisor console device <n>. This is for
640 both Xen and PowerPC hypervisors.
642 If the device connected to the port is not a TTY but a braille
643 device, prepend "brl," before the device type, for instance
645 For now, only VisioBraille is supported.
647 consoleblank= [KNL] The console blank (screen saver) timeout in
648 seconds. Defaults to 10*60 = 10mins. A value of 0
649 disables the blank timer.
652 [KNL] Change the default value for
653 /proc/<pid>/coredump_filter.
654 See also Documentation/filesystems/proc.txt.
656 coresight_cpu_debug.enable
659 Enable/disable the CPU sampling based debugging.
660 0: default value, disable debugging
661 1: enable debugging at boot time
663 cpuidle.off=1 [CPU_IDLE]
664 disable the cpuidle sub-system
666 cpufreq.off=1 [CPU_FREQ]
667 disable the cpufreq sub-system
670 [X86] Delay for N microsec between assert and de-assert
671 of APIC INIT to start processors. This delay occurs
672 on every CPU online, such as boot, and resume from suspend.
675 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
677 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
679 crashkernel=size[KMG][@offset[KMG]]
680 [KNL] Using kexec, Linux can switch to a 'crash kernel'
681 upon panic. This parameter reserves the physical
682 memory region [offset, offset + size] for that kernel
683 image. If '@offset' is omitted, then a suitable offset
684 is selected automatically. Check
685 Documentation/kdump/kdump.txt for further details.
687 crashkernel=range1:size1[,range2:size2,...][@offset]
688 [KNL] Same as above, but depends on the memory
689 in the running system. The syntax of range is
690 start-[end] where start and end are both
691 a memory unit (amount[KMG]). See also
692 Documentation/kdump/kdump.txt for an example.
694 crashkernel=size[KMG],high
695 [KNL, x86_64] range could be above 4G. Allow kernel
696 to allocate physical memory region from top, so could
697 be above 4G if system have more than 4G ram installed.
698 Otherwise memory region will be allocated below 4G, if
700 It will be ignored if crashkernel=X is specified.
701 crashkernel=size[KMG],low
702 [KNL, x86_64] range under 4G. When crashkernel=X,high
703 is passed, kernel could allocate physical memory region
704 above 4G, that cause second kernel crash on system
705 that require some amount of low memory, e.g. swiotlb
706 requires at least 64M+32K low memory, also enough extra
707 low memory is needed to make sure DMA buffers for 32-bit
708 devices won't run out. Kernel would try to allocate at
709 at least 256M below 4G automatically.
710 This one let user to specify own low range under 4G
711 for second kernel instead.
712 0: to disable low allocation.
713 It will be ignored when crashkernel=X,high is not used
714 or memory reserved is below 4G.
717 [KNL] Disable crypto self-tests
722 cs89x0_media= [HW,NET]
723 Format: { rj45 | aui | bnc }
726 See header of drivers/s390/block/dasd_devmap.c.
728 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
729 (one device per port)
730 Format: <port#>,<type>
731 See also Documentation/input/joystick-parport.txt
733 ddebug_query= [KNL,DYNAMIC_DEBUG] Enable debug messages at early boot
735 Documentation/admin-guide/dynamic-debug-howto.rst for
736 details. Deprecated, see dyndbg.
738 debug [KNL] Enable kernel debugging (events log level).
741 [KNL] verbose self-tests
743 Print debugging info while doing the locking API
745 We default to 0 (no extra messages), setting it to
746 1 will print _a lot_ more information - normally
747 only useful to kernel developers.
749 debug_objects [KNL] Enable object debugging
752 [KNL] Disable object debugging
754 debug_guardpage_minorder=
755 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
756 parameter allows control of the order of pages that will
757 be intentionally kept free (and hence protected) by the
758 buddy allocator. Bigger value increase the probability
759 of catching random memory corruption, but reduce the
760 amount of memory for normal system use. The maximum
761 possible value is MAX_ORDER/2. Setting this parameter
762 to 1 or 2 should be enough to identify most random
763 memory corruption problems caused by bugs in kernel or
764 driver code when a CPU writes to (or reads from) a
765 random memory location. Note that there exists a class
766 of memory corruptions problems caused by buggy H/W or
767 F/W or by drivers badly programing DMA (basically when
768 memory is written at bus level and the CPU MMU is
769 bypassed) which are not detectable by
770 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
771 tracking down these problems.
774 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
775 parameter enables the feature at boot time. In
776 default, it is disabled. We can avoid allocating huge
777 chunk of memory for debug pagealloc if we don't enable
778 it at boot time and the system will work mostly same
779 with the kernel built without CONFIG_DEBUG_PAGEALLOC.
780 on: enable the feature
782 debugpat [X86] Enable PAT debugging
784 decnet.addr= [HW,NET]
785 Format: <area>[,<node>]
786 See also Documentation/networking/decnet.txt.
789 [same as hugepagesz=] The size of the default
790 HugeTLB page size. This is the size represented by
791 the legacy /proc/ hugepages APIs, used for SHM, and
792 default size when mounting hugetlbfs filesystems.
793 Defaults to the default architecture's huge page size
797 Set number of hash buckets for dentry cache.
799 disable_1tb_segments [PPC]
800 Disables the use of 1TB hash page table segments. This
801 causes the kernel to fall back to 256MB segments which
802 can be useful when debugging issues that require an SLB
806 See Documentation/networking/ipv6.txt.
809 Disable RADIX MMU mode on POWER9
811 disable_cpu_apicid= [X86,APIC,SMP]
813 The number of initial APIC ID for the
814 corresponding CPU to be disabled at boot,
815 mostly used for the kdump 2nd kernel to
816 disable BSP to wake up multiple CPUs without
817 causing system reset or hang due to sending
820 disable_ddw [PPC/PSERIES]
821 Disable Dynamic DMA Window support. Use this if
822 to workaround buggy firmware.
825 See Documentation/networking/ipv6.txt.
827 disable_mtrr_cleanup [X86]
828 The kernel tries to adjust MTRR layout from continuous
829 to discrete, to make X server driver able to add WB
830 entry later. This parameter disables that.
832 disable_mtrr_trim [X86, Intel and AMD only]
833 By default the kernel will trim any uncacheable
834 memory out of your available memory pool based on
835 MTRR settings. This parameter disables that behavior,
836 possibly causing your machine to run very slowly.
838 disable_timer_pin_1 [X86]
839 Disable PIN 1 of APIC timer
840 Can be useful to work around chipset bugs.
842 dis_ucode_ldr [X86] Disable the microcode loader.
844 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
845 this option disables the debugging code at boot.
847 dma_debug_entries=<number>
848 This option allows to tune the number of preallocated
849 entries for DMA-API debugging code. One entry is
850 required per DMA-API allocation. Use this if the
851 DMA-API debugging code disables itself because the
852 architectural default is too low.
854 dma_debug_driver=<driver_name>
855 With this option the DMA-API debugging driver
856 filter feature can be enabled at boot time. Just
857 pass the driver to filter for as the parameter.
858 The filter can be disabled or changed to another
859 driver later using sysfs.
861 drm_kms_helper.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
862 Broken monitors, graphic adapters, KVMs and EDIDless
863 panels may send no or incorrect EDID data sets.
864 This parameter allows to specify an EDID data sets
865 in the /lib/firmware directory that are used instead.
866 Generic built-in EDID data sets are used, if one of
867 edid/1024x768.bin, edid/1280x1024.bin,
868 edid/1680x1050.bin, or edid/1920x1080.bin is given
869 and no file with the same name exists. Details and
870 instructions how to build your own EDID data are
871 available in Documentation/EDID/HOWTO.txt. An EDID
872 data set will only be used for a particular connector,
873 if its name and a colon are prepended to the EDID
874 name. Each connector may use a unique EDID data
875 set by separating the files with a comma. An EDID
876 data set with no connector name will be used for
877 any connectors not explicitly specified.
882 Format: {"off" | "known"}
883 Control how the dt_cpu_ftrs device-tree binding is
884 used for CPU feature discovery and setup (if it
886 off: Do not use it, fall back to legacy cpu table.
887 known: Do not pass through unknown features to guests
888 or userspace, only those that the kernel is aware of.
890 dump_apple_properties [X86]
891 Dump name and content of EFI device properties on
892 x86 Macs. Useful for driver authors to determine
893 what data is available or for reverse-engineering.
895 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
896 module.dyndbg[="val"]
897 Enable debug messages at boot time. See
898 Documentation/admin-guide/dynamic-debug-howto.rst
901 nompx [X86] Disables Intel Memory Protection Extensions.
902 See Documentation/x86/intel_mpx.txt for more
903 information about the feature.
905 nopku [X86] Disable Memory Protection Keys CPU feature found
908 module.async_probe [KNL]
909 Enable asynchronous probe on this module.
911 early_ioremap_debug [KNL]
912 Enable debug messages in early_ioremap support. This
913 is useful for tracking down temporary early mappings
914 which are not unmapped.
916 earlycon= [KNL] Output early console device and options.
918 When used with no options, the early console is
919 determined by the stdout-path property in device
922 cdns,<addr>[,options]
923 Start an early, polled-mode console on a Cadence
924 (xuartps) serial port at the specified address. Only
925 supported option is baud rate. If baud rate is not
926 specified, the serial port must already be setup and
929 uart[8250],io,<addr>[,options]
930 uart[8250],mmio,<addr>[,options]
931 uart[8250],mmio32,<addr>[,options]
932 uart[8250],mmio32be,<addr>[,options]
933 uart[8250],0x<addr>[,options]
934 Start an early, polled-mode console on the 8250/16550
935 UART at the specified I/O port or MMIO address.
936 MMIO inter-register address stride is either 8-bit
937 (mmio) or 32-bit (mmio32 or mmio32be).
938 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
939 to be equivalent to 'mmio'. 'options' are specified
940 in the same format described for "console=ttyS<n>"; if
941 unspecified, the h/w is not initialized.
945 Start an early, polled-mode console on a pl011 serial
946 port at the specified address. The pl011 serial port
947 must already be setup and configured. Options are not
948 yet supported. If 'mmio32' is specified, then only
949 the driver will use only 32-bit accessors to read/write
950 the device registers.
953 Start an early, polled-mode console on a meson serial
954 port at the specified address. The serial port must
955 already be setup and configured. Options are not yet
959 Start an early, polled-mode console on an msm serial
960 port at the specified address. The serial port
961 must already be setup and configured. Options are not
965 Start an early, polled-mode console on an msm serial
966 dm port at the specified address. The serial port
967 must already be setup and configured. Options are not
971 Start an early, polled-mode console on a serial port
972 of an Actions Semi SoC, such as S500 or S900, at the
973 specified address. The serial port must already be
974 setup and configured. Options are not yet supported.
976 smh Use ARM semihosting calls for early console.
984 Use early console provided by serial driver available
985 on Samsung SoCs, requires selecting proper type and
986 a correct base address of the selected UART port. The
987 serial port must already be setup and configured.
988 Options are not yet supported.
991 Start an early, polled-mode console on a lantiq serial
992 (lqasc) port at the specified address. The serial port
993 must already be setup and configured. Options are not
998 Use early console provided by Freescale LP UART driver
999 found on Freescale Vybrid and QorIQ LS1021A processors.
1000 A valid base address must be provided, and the serial
1001 port must already be setup and configured.
1004 Start an early, polled-mode console on the
1005 Armada 3700 serial port at the specified
1006 address. The serial port must already be setup
1007 and configured. Options are not yet supported.
1009 earlyprintk= [X86,SH,BLACKFIN,ARM,M68k,S390]
1014 earlyprintk=serial[,ttySn[,baudrate]]
1015 earlyprintk=serial[,0x...[,baudrate]]
1016 earlyprintk=ttySn[,baudrate]
1017 earlyprintk=dbgp[debugController#]
1018 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1019 earlyprintk=xdbc[xhciController#]
1021 earlyprintk is useful when the kernel crashes before
1022 the normal console is initialized. It is not enabled by
1023 default because it has some cosmetic problems.
1025 Append ",keep" to not disable it when the real console
1028 Only one of vga, efi, serial, or usb debug port can
1031 Currently only ttyS0 and ttyS1 may be specified by
1032 name. Other I/O ports may be explicitly specified
1033 on some architectures (x86 and arm at least) by
1034 replacing ttySn with an I/O port address, like this:
1035 earlyprintk=serial,0x1008,115200
1036 You can find the port for a given device in
1037 /proc/tty/driver/serial:
1038 2: uart:ST16650V2 port:00001008 irq:18 ...
1040 Interaction with the standard serial driver is not
1043 The VGA and EFI output is eventually overwritten by
1046 The xen output can only be used by Xen PV guests.
1048 The sclp output can only be used on s390.
1050 The optional "force" to "pciserial" enables use of a
1051 PCI device even when its classcode is not of the
1054 edac_report= [HW,EDAC] Control how to report EDAC event
1055 Format: {"on" | "off" | "force"}
1056 on: enable EDAC to report H/W event. May be overridden
1057 by other higher priority error reporting module.
1058 off: disable H/W event reporting through EDAC.
1059 force: enforce the use of EDAC to report H/W event.
1062 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1065 This is designed to be used in conjunction with
1066 the boot argument: earlyprintk=vga
1069 Format: {"off" | "on" | "skip[mbr]"}
1072 Format: { "old_map", "nochunk", "noruntime", "debug" }
1073 old_map [X86-64]: switch to the old ioremap-based EFI
1074 runtime services mapping. 32-bit still uses this one by
1076 nochunk: disable reading files in "chunks" in the EFI
1077 boot stub, as chunking can cause problems with some
1078 firmware implementations.
1079 noruntime : disable EFI runtime services support
1080 debug: enable misc debug output
1082 efi_no_storage_paranoia [EFI; X86]
1083 Using this parameter you can use more than 50% of
1084 your efi variable storage. Use this parameter only if
1085 you are really sure that your UEFI does sane gc and
1086 fulfills the spec otherwise your board may brick.
1088 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1089 Add arbitrary attribute to specific memory range by
1090 updating original EFI memory map.
1091 Region of memory which aa attribute is added to is
1093 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1094 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1095 attribute is added to range 0x100000000-0x180000000 and
1096 0x10a0000000-0x1120000000.
1098 Using this parameter you can do debugging of EFI memmap
1099 related feature. For example, you can do debugging of
1100 Address Range Mirroring feature even if your box
1103 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1104 that is to be dynamically loaded by Linux. If there are
1105 multiple variables with the same name but with different
1106 vendor GUIDs, all of them will be loaded. See
1107 Documentation/acpi/ssdt-overlays.txt for details.
1110 eisa_irq_edge= [PARISC,HW]
1111 See header of drivers/parisc/eisa.c.
1114 See comment before function elanfreq_setup() in
1115 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1118 Format: {"cfq" | "deadline" | "noop"}
1119 See Documentation/block/cfq-iosched.txt and
1120 Documentation/block/deadline-iosched.txt for details.
1122 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1123 Specifies physical address of start of kernel core
1124 image elf header and optionally the size. Generally
1125 kexec loader will pass this option to capture kernel.
1126 See Documentation/kdump/kdump.txt for details.
1128 enable_mtrr_cleanup [X86]
1129 The kernel tries to adjust MTRR layout from continuous
1130 to discrete, to make X server driver able to add WB
1131 entry later. This parameter enables that.
1133 enable_timer_pin_1 [X86]
1134 Enable PIN 1 of APIC timer
1135 Can be useful to work around chipset bugs
1136 (in particular on some ATI chipsets).
1137 The kernel tries to set a reasonable default.
1139 enforcing [SELINUX] Set initial enforcing status.
1141 See security/selinux/Kconfig help text.
1142 0 -- permissive (log only, no denials).
1143 1 -- enforcing (deny and log).
1145 Value can be changed at runtime via /selinux/enforce.
1148 Disable Error Record Serialization Table (ERST)
1151 ether= [HW,NET] Ethernet cards parameters
1152 This option is obsoleted by the "netdev=" option, which
1153 has equivalent usage. See its documentation for details.
1157 Permit 'security.evm' to be updated regardless of
1158 current integrity status.
1162 fail_make_request=[KNL]
1163 General fault injection mechanism.
1164 Format: <interval>,<probability>,<space>,<times>
1165 See also Documentation/fault-injection/.
1168 See Documentation/blockdev/floppy.txt.
1170 force_pal_cache_flush
1171 [IA-64] Avoid check_sal_cache_flush which may hang on
1172 buggy SAL_CACHE_FLUSH implementations. Using this
1173 parameter will force ia64_sal_cache_flush to call
1174 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1177 Forcefully enable Physical Address Extension (PAE).
1178 Many Pentium M systems disable PAE but may have a
1179 functionally usable PAE implementation.
1180 Warning: use of this parameter will taint the kernel
1181 and may cause unknown problems.
1184 [FTRACE] will set and start the specified tracer
1185 as early as possible in order to facilitate early
1188 ftrace_dump_on_oops[=orig_cpu]
1189 [FTRACE] will dump the trace buffers on oops.
1190 If no parameter is passed, ftrace will dump
1191 buffers of all CPUs, but if you pass orig_cpu, it will
1192 dump only the buffer of the CPU that triggered the
1195 ftrace_filter=[function-list]
1196 [FTRACE] Limit the functions traced by the function
1197 tracer at boot up. function-list is a comma separated
1198 list of functions. This list can be changed at run
1199 time by the set_ftrace_filter file in the debugfs
1202 ftrace_notrace=[function-list]
1203 [FTRACE] Do not trace the functions specified in
1204 function-list. This list can be changed at run time
1205 by the set_ftrace_notrace file in the debugfs
1208 ftrace_graph_filter=[function-list]
1209 [FTRACE] Limit the top level callers functions traced
1210 by the function graph tracer at boot up.
1211 function-list is a comma separated list of functions
1212 that can be changed at run time by the
1213 set_graph_function file in the debugfs tracing directory.
1215 ftrace_graph_notrace=[function-list]
1216 [FTRACE] Do not trace from the functions specified in
1217 function-list. This list is a comma separated list of
1218 functions that can be changed at run time by the
1219 set_graph_notrace file in the debugfs tracing directory.
1221 ftrace_graph_max_depth=<uint>
1222 [FTRACE] Used with the function graph tracer. This is
1223 the max depth it will trace into a function. This value
1224 can be changed at run time by the max_graph_depth file
1225 in the tracefs tracing directory. default: 0 (no limit)
1228 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1229 support via parallel port (up to 5 devices per port)
1230 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1231 See also Documentation/input/joystick-parport.txt
1235 gart_fix_e820= [X86_64] disable the fix e820 for K8 GART
1239 gcov_persist= [GCOV] When non-zero (default), profiling data for
1240 kernel modules is saved and remains accessible via
1241 debugfs, even when the module is unloaded/reloaded.
1242 When zero, profiling data is discarded and associated
1243 debugfs files are removed at module unload time.
1245 goldfish [X86] Enable the goldfish android emulator platform.
1246 Don't use this when you are not running on the
1249 gpt [EFI] Forces disk with valid GPT signature but
1250 invalid Protective MBR to be treated as GPT. If the
1251 primary GPT is corrupted, it enables the backup/alternate
1252 GPT to be used instead.
1254 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1255 the "Enable 0" bit of the configuration register.
1258 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1259 the "Enable 0" bit of the configuration register.
1262 grcan.select= [HW] Select which physical interface to use.
1265 grcan.txsize= [HW] Sets the size of the tx buffer.
1266 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1268 grcan.rxsize= [HW] Sets the size of the rx buffer.
1269 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1272 gpio-mockup.gpio_mockup_ranges
1273 [HW] Sets the ranges of gpiochip of for this device.
1274 Format: <start1>,<end1>,<start2>,<end2>...
1276 hardlockup_all_cpu_backtrace=
1277 [KNL] Should the hard-lockup detector generate
1278 backtraces on all cpus.
1281 hashdist= [KNL,NUMA] Large hashes allocated during boot
1282 are distributed across NUMA nodes. Defaults on
1283 for 64-bit NUMA, off otherwise.
1284 Format: 0 | 1 (for off | on)
1286 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1288 hd= [EIDE] (E)IDE hard drive subsystem geometry
1289 Format: <cyl>,<head>,<sect>
1292 Disable Hardware Error Source Table (HEST) support;
1293 corresponding firmware-first mode error processing
1294 logic will be disabled.
1296 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1297 size of <nn>. This works even on boxes that have no
1298 highmem otherwise. This also works to reduce highmem
1299 size on bigger boxes.
1301 highres= [KNL] Enable/disable high resolution timer mode.
1302 Valid parameters: "on", "off"
1306 See Documentation/isdn/README.HiSax.
1310 hpet= [X86-32,HPET] option to control HPET usage
1311 Format: { enable (default) | disable | force |
1313 disable: disable HPET and use PIT instead
1314 force: allow force enabled of undocumented chips (ICH4,
1316 verbose: show contents of HPET registers during setup
1318 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1319 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1321 hugepages= [HW,X86-32,IA-64] HugeTLB pages to allocate at boot.
1322 hugepagesz= [HW,IA-64,PPC,X86-64] The size of the HugeTLB pages.
1323 On x86-64 and powerpc, this option can be specified
1324 multiple times interleaved with hugepages= to reserve
1325 huge pages of different sizes. Valid pages sizes on
1326 x86-64 are 2M (when the CPU supports "pse") and 1G
1327 (when the CPU supports the "pdpe1gb" cpuinfo flag).
1329 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1330 terminal devices. Valid values: 0..8
1331 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1332 If specified, z/VM IUCV HVC accepts connections
1333 from listed z/VM user IDs only.
1335 hwthread_map= [METAG] Comma-separated list of Linux cpu id to
1336 hardware thread id mappings.
1337 Format: <cpu>:<hwthread>
1340 Do not unregister boot console at start. This is only
1341 useful for debugging when something happens in the window
1342 between unregistering the boot console and initializing
1345 i2c_bus= [HW] Override the default board specific I2C bus speed
1346 or register an additional I2C bus that is not
1347 registered from board initialization code.
1351 i8042.debug [HW] Toggle i8042 debug mode
1352 i8042.unmask_kbd_data
1353 [HW] Enable printing of interrupt data from the KBD port
1354 (disabled by default, and as a pre-condition
1355 requires that i8042.debug=1 be enabled)
1356 i8042.direct [HW] Put keyboard port into non-translated mode
1357 i8042.dumbkbd [HW] Pretend that controller can only read data from
1358 keyboard and cannot control its state
1359 (Don't attempt to blink the leds)
1360 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1361 i8042.nokbd [HW] Don't check/create keyboard port
1362 i8042.noloop [HW] Disable the AUX Loopback command while probing
1364 i8042.nomux [HW] Don't check presence of an active multiplexing
1366 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1368 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1369 i8042.reset [HW] Reset the controller during init, cleanup and
1370 suspend-to-ram transitions, only during s2r
1371 transitions, or never reset
1372 Format: { 1 | Y | y | 0 | N | n }
1373 1, Y, y: always reset controller
1374 0, N, n: don't ever reset controller
1375 Default: only on s2r transitions on x86; most other
1376 architectures force reset to be always executed
1377 i8042.unlock [HW] Unlock (ignore) the keylock
1378 i8042.kbdreset [HW] Reset device connected to KBD port
1382 i8k.ignore_dmi [HW] Continue probing hardware even if DMI data
1383 indicates that the driver is running on unsupported
1385 i8k.force [HW] Activate i8k driver even if SMM BIOS signature
1386 does not match list of supported models.
1388 [HW] Report power status in /proc/i8k
1389 (disabled by default)
1390 i8k.restricted [HW] Allow controlling fans only if SYS_ADMIN
1393 i915.invert_brightness=
1394 [DRM] Invert the sense of the variable that is used to
1395 set the brightness of the panel backlight. Normally a
1396 brightness value of 0 indicates backlight switched off,
1397 and the maximum of the brightness value sets the backlight
1398 to maximum brightness. If this parameter is set to 0
1399 (default) and the machine requires it, or this parameter
1400 is set to 1, a brightness value of 0 sets the backlight
1401 to maximum brightness, and the maximum of the brightness
1402 value switches the backlight off.
1403 -1 -- never invert brightness
1404 0 -- machine default
1405 1 -- force brightness inversion
1408 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1410 ide-core.nodma= [HW] (E)IDE subsystem
1411 Format: =0.0 to prevent dma on hda, =0.1 hdb =1.0 hdc
1412 .vlb_clock .pci_clock .noflush .nohpa .noprobe .nowerr
1413 .cdrom .chs .ignore_cable are additional options
1414 See Documentation/ide/ide.txt.
1416 ide-generic.probe-mask= [HW] (E)IDE subsystem
1418 Probe mask for legacy ISA IDE ports. Depending on
1419 platform up to 6 ports are supported, enabled by
1420 setting corresponding bits in the mask to 1. The
1421 default value is 0x0, which has a special meaning.
1422 On systems that have PCI, it triggers scanning the
1423 PCI bus for the first and the second port, which
1424 are then probed. On systems without PCI the value
1425 of 0x0 enables probing the two first ports as if it
1428 ide-pci-generic.all-generic-ide [HW] (E)IDE subsystem
1429 Claim all unknown PCI IDE storage controllers.
1432 Format: idle=poll, idle=halt, idle=nomwait
1433 Poll forces a polling idle loop that can slightly
1434 improve the performance of waking up a idle CPU, but
1435 will use a lot of power and make the system run hot.
1437 idle=halt: Halt is forced to be used for CPU idle.
1438 In such case C2/C3 won't be used again.
1439 idle=nomwait: Disable mwait for CPU C-states
1441 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1442 Format: { strict | legacy | 2008 | relaxed }
1445 Choose which programs will be accepted for execution
1446 based on the IEEE 754 NaN encoding(s) supported by
1447 the FPU and the NaN encoding requested with the value
1448 of an ELF file header flag individually set by each
1449 binary. Hardware implementations are permitted to
1450 support either or both of the legacy and the 2008 NaN
1453 Available settings are as follows:
1454 strict accept binaries that request a NaN encoding
1455 supported by the FPU
1456 legacy only accept legacy-NaN binaries, if supported
1458 2008 only accept 2008-NaN binaries, if supported
1460 relaxed accept any binaries regardless of whether
1461 supported by the FPU
1463 The FPU emulator is always able to support both NaN
1464 encodings, so if no FPU hardware is present or it has
1465 been disabled with 'nofpu', then the settings of
1466 'legacy' and '2008' strap the emulator accordingly,
1467 'relaxed' straps the emulator for both legacy-NaN and
1468 2008-NaN, whereas 'strict' enables legacy-NaN only on
1469 legacy processors and both NaN encodings on MIPS32 or
1472 The setting for ABS.fmt/NEG.fmt instruction execution
1473 mode generally follows that for the NaN encoding,
1474 except where unsupported by hardware.
1476 ignore_loglevel [KNL]
1477 Ignore loglevel setting - this will print /all/
1478 kernel messages to the console. Useful for debugging.
1479 We also add it as printk module parameter, so users
1480 could change it dynamically, usually by
1481 /sys/module/printk/parameters/ignore_loglevel.
1484 Ignore RLIMIT_DATA setting for data mappings,
1485 print warning at first misuse. Can be changed via
1486 /sys/module/kernel/parameters/ignore_rlimit_data.
1488 ihash_entries= [KNL]
1489 Set number of hash buckets for inode cache.
1491 ima_appraise= [IMA] appraise integrity measurements
1492 Format: { "off" | "enforce" | "fix" | "log" }
1495 ima_appraise_tcb [IMA]
1496 The builtin appraise policy appraises all files
1499 ima_canonical_fmt [IMA]
1500 Use the canonical format for the binary runtime
1501 measurements, instead of host native format.
1504 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1508 The list of supported hash algorithms is defined
1509 in crypto/hash_info.h.
1512 The builtin policies to load during IMA setup.
1513 Format: "tcb | appraise_tcb | secure_boot"
1515 The "tcb" policy measures all programs exec'd, files
1516 mmap'd for exec, and all files opened with the read
1517 mode bit set by either the effective uid (euid=0) or
1520 The "appraise_tcb" policy appraises the integrity of
1521 all files owned by root. (This is the equivalent
1522 of ima_appraise_tcb.)
1524 The "secure_boot" policy appraises the integrity
1525 of files (eg. kexec kernel image, kernel modules,
1526 firmware, policy, etc) based on file signatures.
1528 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
1529 Load a policy which meets the needs of the Trusted
1530 Computing Base. This means IMA will measure all
1531 programs exec'd, files mmap'd for exec, and all files
1532 opened for read by uid=0.
1535 Select one of defined IMA measurements template formats.
1536 Formats: { "ima" | "ima-ng" | "ima-sig" }
1540 [IMA] Define a custom template format.
1541 Format: { "field1|...|fieldN" }
1543 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
1544 Format: <min_file_size>
1545 Set the minimal file size for using asynchronous hash.
1546 If left unspecified, ahash usage is disabled.
1548 ahash performance varies for different data sizes on
1549 different crypto accelerators. This option can be used
1550 to achieve the best performance for a particular HW.
1552 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
1554 Set hashing buffer size. Default: 4k.
1556 ahash performance varies for different chunk sizes on
1557 different crypto accelerators. This option can be used
1558 to achieve best performance for particular HW.
1562 Run specified binary instead of /sbin/init as init
1565 initcall_debug [KNL] Trace initcalls as they are executed. Useful
1566 for working out where the kernel is dying during
1569 initcall_blacklist= [KNL] Do not execute a comma-separated list of
1570 initcall functions. Useful for debugging built-in
1571 modules and initcalls.
1573 initrd= [BOOT] Specify the location of the initial ramdisk
1575 init_pkru= [x86] Specify the default memory protection keys rights
1576 register contents for all processes. 0x55555554 by
1577 default (disallow access to all but pkey 0). Can
1578 override in debugfs after boot.
1580 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
1583 int_pln_enable [x86] Enable power limit notification interrupt
1585 integrity_audit=[IMA]
1586 Format: { "0" | "1" }
1587 0 -- basic integrity auditing messages. (Default)
1588 1 -- additional integrity auditing messages.
1590 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
1592 Enable intel iommu driver.
1594 Disable intel iommu driver.
1595 igfx_off [Default Off]
1596 By default, gfx is mapped as normal device. If a gfx
1597 device has a dedicated DMAR unit, the DMAR unit is
1598 bypassed by not enabling DMAR with this option. In
1599 this case, gfx device will use physical address for
1602 With this option iommu will not optimize to look
1603 for io virtual address below 32-bit forcing dual
1604 address cycle on pci bus for cards supporting greater
1605 than 32-bit addressing. The default is to look
1606 for translation below 32-bit and if not available
1607 then look in the higher range.
1608 strict [Default Off]
1609 With this option on every unmap_single operation will
1610 result in a hardware IOTLB flush operation as opposed
1611 to batching them for performance.
1612 sp_off [Default Off]
1613 By default, super page will be supported if Intel IOMMU
1614 has the capability. With this option, super page will
1616 ecs_off [Default Off]
1617 By default, extended context tables will be supported if
1618 the hardware advertises that it has support both for the
1619 extended tables themselves, and also PASID support. With
1620 this option set, extended tables will not be used even
1621 on hardware which claims to support them.
1622 tboot_noforce [Default Off]
1623 Do not force the Intel IOMMU enabled under tboot.
1624 By default, tboot will force Intel IOMMU on, which
1625 could harm performance of some high-throughput
1626 devices like 40GBit network cards, even if identity
1628 Note that using this option lowers the security
1629 provided by tboot because it makes the system
1630 vulnerable to DMA attacks.
1632 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
1633 0 disables intel_idle and fall back on acpi_idle.
1634 1 to 9 specify maximum depth of C-state.
1638 Do not enable intel_pstate as the default
1639 scaling driver for the supported processors
1641 Use intel_pstate as a scaling driver, but configure it
1642 to work with generic cpufreq governors (instead of
1643 enabling its internal governor). This mode cannot be
1644 used along with the hardware-managed P-states (HWP)
1647 Enable intel_pstate on systems that prohibit it by default
1648 in favor of acpi-cpufreq. Forcing the intel_pstate driver
1649 instead of acpi-cpufreq may disable platform features, such
1650 as thermal controls and power capping, that rely on ACPI
1651 P-States information being indicated to OSPM and therefore
1652 should be used with caution. This option does not work with
1653 processors that aren't supported by the intel_pstate driver
1654 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
1656 Do not enable hardware P state control (HWP)
1659 Only load intel_pstate on systems which support
1660 hardware P state control (HWP) if available.
1662 Enforce ACPI _PPC performance limits. If the Fixed ACPI
1663 Description Table, specifies preferred power management
1664 profile as "Enterprise Server" or "Performance Server",
1665 then this feature is turned on by default.
1667 Allow per-logical-CPU P-State performance control limits using
1668 cpufreq sysfs interface
1670 intremap= [X86-64, Intel-IOMMU]
1671 on enable Interrupt Remapping (default)
1672 off disable Interrupt Remapping
1673 nosid disable Source ID checking
1675 BIOS x2APIC opt-out request will be ignored
1676 nopost disable Interrupt Posting
1678 iomem= Disable strict checking of access to MMIO memory
1679 strict regions from userspace.
1694 nobypass [PPC/POWERNV]
1695 Disable IOMMU bypass, using IOMMU for PCI devices.
1698 [ARM64] Configure DMA to bypass the IOMMU by default.
1699 Format: { "0" | "1" }
1700 0 - Use IOMMU translation for DMA.
1701 1 - Bypass the IOMMU for DMA.
1702 unset - Use IOMMU translation for DMA.
1704 io7= [HW] IO7 for Marvel based alpha systems
1705 See comment before marvel_specify_io7 in
1706 arch/alpha/kernel/core_marvel.c.
1708 io_delay= [X86] I/O delay method
1710 Standard port 0x80 based delay
1712 Alternate port 0xed based delay (needed on some systems)
1714 Simple two microseconds delay
1719 See Documentation/filesystems/nfs/nfsroot.txt.
1721 irqaffinity= [SMP] Set the default irq affinity mask
1722 The argument is a cpu list, as described above.
1725 When an interrupt is not handled search all handlers
1726 for it. Intended to get systems with badly broken
1730 When an interrupt is not handled search all handlers
1731 for it. Also check all handlers each timer
1732 interrupt. Intended to get systems with badly broken
1736 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
1738 isolcpus= [KNL,SMP] Isolate CPUs from the general scheduler.
1739 The argument is a cpu list, as described above.
1741 This option can be used to specify one or more CPUs
1742 to isolate from the general SMP balancing and scheduling
1743 algorithms. You can move a process onto or off an
1744 "isolated" CPU via the CPU affinity syscalls or cpuset.
1745 <cpu number> begins at 0 and the maximum value is
1746 "number of CPUs in system - 1".
1748 This option is the preferred way to isolate CPUs. The
1749 alternative -- manually setting the CPU mask of all
1750 tasks in the system -- can cause problems and
1751 suboptimal load balancer performance.
1755 ivrs_ioapic [HW,X86_64]
1756 Provide an override to the IOAPIC-ID<->DEVICE-ID
1757 mapping provided in the IVRS ACPI table. For
1758 example, to map IOAPIC-ID decimal 10 to
1759 PCI device 00:14.0 write the parameter as:
1760 ivrs_ioapic[10]=00:14.0
1762 ivrs_hpet [HW,X86_64]
1763 Provide an override to the HPET-ID<->DEVICE-ID
1764 mapping provided in the IVRS ACPI table. For
1765 example, to map HPET-ID decimal 0 to
1766 PCI device 00:14.0 write the parameter as:
1767 ivrs_hpet[0]=00:14.0
1769 ivrs_acpihid [HW,X86_64]
1770 Provide an override to the ACPI-HID:UID<->DEVICE-ID
1771 mapping provided in the IVRS ACPI table. For
1772 example, to map UART-HID:UID AMD0020:0 to
1773 PCI device 00:14.5 write the parameter as:
1774 ivrs_acpihid[00:14.5]=AMD0020:0
1776 js= [HW,JOY] Analog joystick
1777 See Documentation/input/joystick.txt.
1780 When CONFIG_RANDOMIZE_BASE is set, this disables
1781 kernel and module base offset ASLR (Address Space
1782 Layout Randomization).
1785 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
1786 report on every invalid memory access. Without this
1787 parameter KASAN will print report only for the first
1792 kernelcore= [KNL,X86,IA-64,PPC]
1793 Format: nn[KMGTPE] | "mirror"
1795 specifies the amount of memory usable by the kernel
1796 for non-movable allocations. The requested amount is
1797 spread evenly throughout all nodes in the system. The
1798 remaining memory in each node is used for Movable
1799 pages. In the event, a node is too small to have both
1800 kernelcore and Movable pages, kernelcore pages will
1801 take priority and other nodes will have a larger number
1802 of Movable pages. The Movable zone is used for the
1803 allocation of pages that may be reclaimed or moved
1804 by the page migration subsystem. This means that
1805 HugeTLB pages may not be allocated from this zone.
1806 Note that allocations like PTEs-from-HighMem still
1807 use the HighMem zone if it exists, and the Normal
1808 zone if it does not.
1810 Instead of specifying the amount of memory (nn[KMGTPE]),
1811 you can specify "mirror" option. In case "mirror"
1812 option is specified, mirrored (reliable) memory is used
1813 for non-movable allocations and remaining memory is used
1814 for Movable pages. nn[KMGTPE] and "mirror" are exclusive,
1815 so you can NOT specify nn[KMGTPE] and "mirror" at the same
1818 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
1819 Format: <Controller#>[,poll interval]
1820 The controller # is the number of the ehci usb debug
1821 port as it is probed via PCI. The poll interval is
1822 optional and is the number seconds in between
1823 each poll cycle to the debug port in case you need
1824 the functionality for interrupting the kernel with
1825 gdb or control-c on the dbgp connection. When
1826 not using this parameter you use sysrq-g to break into
1827 the kernel debugger.
1829 kgdboc= [KGDB,HW] kgdb over consoles.
1830 Requires a tty driver that supports console polling,
1831 or a supported polling keyboard driver (non-usb).
1832 Serial only format: <serial_device>[,baud]
1833 keyboard only format: kbd
1834 keyboard and serial format: kbd,<serial_device>[,baud]
1835 Optional Kernel mode setting:
1836 kms, kbd format: kms,kbd
1837 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
1839 kgdbwait [KGDB] Stop kernel execution and enter the
1840 kernel debugger at the earliest opportunity.
1842 kmac= [MIPS] korina ethernet MAC address.
1843 Configure the RouterBoard 532 series on-chip
1844 Ethernet adapter MAC address.
1846 kmemleak= [KNL] Boot-time kmemleak enable/disable
1847 Valid arguments: on, off
1849 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
1852 kpti= [ARM64] Control page table isolation of user
1853 and kernel address spaces.
1854 Default: enabled on cores which need mitigation.
1858 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
1859 Default is 0 (don't ignore, but inject #GP)
1861 kvm.mmu_audit= [KVM] This is a R/W parameter which allows audit
1866 [KVM] Controls the software workaround for the
1867 X86_BUG_ITLB_MULTIHIT bug.
1868 force : Always deploy workaround.
1869 off : Never deploy workaround.
1870 auto : Deploy workaround based on the presence of
1871 X86_BUG_ITLB_MULTIHIT.
1875 If the software workaround is enabled for the host,
1876 guests do need not to enable it for nested guests.
1878 kvm.nx_huge_pages_recovery_ratio=
1879 [KVM] Controls how many 4KiB pages are periodically zapped
1880 back to huge pages. 0 disables the recovery, otherwise if
1881 the value is N KVM will zap 1/Nth of the 4KiB pages every
1882 minute. The default is 60.
1884 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM.
1885 Default is 1 (enabled)
1887 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU)
1889 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode.
1891 kvm-arm.vgic_v3_group0_trap=
1892 [KVM,ARM] Trap guest accesses to GICv3 group-0
1895 kvm-arm.vgic_v3_group1_trap=
1896 [KVM,ARM] Trap guest accesses to GICv3 group-1
1899 kvm-arm.vgic_v3_common_trap=
1900 [KVM,ARM] Trap guest accesses to GICv3 common
1903 kvm-intel.ept= [KVM,Intel] Disable extended page tables
1904 (virtualized MMU) support on capable Intel chips.
1905 Default is 1 (enabled)
1907 kvm-intel.emulate_invalid_guest_state=
1908 [KVM,Intel] Enable emulation of invalid guest states
1909 Default is 0 (disabled)
1911 kvm-intel.flexpriority=
1912 [KVM,Intel] Disable FlexPriority feature (TPR shadow).
1913 Default is 1 (enabled)
1916 [KVM,Intel] Enable VMX nesting (nVMX).
1917 Default is 0 (disabled)
1919 kvm-intel.unrestricted_guest=
1920 [KVM,Intel] Disable unrestricted guest feature
1921 (virtualized real and unpaged mode) on capable
1922 Intel chips. Default is 1 (enabled)
1924 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
1927 Valid arguments: never, cond, always
1929 always: L1D cache flush on every VMENTER.
1930 cond: Flush L1D on VMENTER only when the code between
1931 VMEXIT and VMENTER can leak host memory.
1932 never: Disables the mitigation
1934 Default is cond (do L1 cache flush in specific instances)
1936 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification
1937 feature (tagged TLBs) on capable Intel chips.
1938 Default is 1 (enabled)
1940 l1tf= [X86] Control mitigation of the L1TF vulnerability on
1943 The kernel PTE inversion protection is unconditionally
1944 enabled and cannot be disabled.
1947 Provides all available mitigations for the
1948 L1TF vulnerability. Disables SMT and
1949 enables all mitigations in the
1950 hypervisors, i.e. unconditional L1D flush.
1952 SMT control and L1D flush control via the
1953 sysfs interface is still possible after
1954 boot. Hypervisors will issue a warning
1955 when the first VM is started in a
1956 potentially insecure configuration,
1957 i.e. SMT enabled or L1D flush disabled.
1960 Same as 'full', but disables SMT and L1D
1961 flush runtime control. Implies the
1962 'nosmt=force' command line option.
1963 (i.e. sysfs control of SMT is disabled.)
1966 Leaves SMT enabled and enables the default
1967 hypervisor mitigation, i.e. conditional
1970 SMT control and L1D flush control via the
1971 sysfs interface is still possible after
1972 boot. Hypervisors will issue a warning
1973 when the first VM is started in a
1974 potentially insecure configuration,
1975 i.e. SMT enabled or L1D flush disabled.
1979 Disables SMT and enables the default
1980 hypervisor mitigation.
1982 SMT control and L1D flush control via the
1983 sysfs interface is still possible after
1984 boot. Hypervisors will issue a warning
1985 when the first VM is started in a
1986 potentially insecure configuration,
1987 i.e. SMT enabled or L1D flush disabled.
1990 Same as 'flush', but hypervisors will not
1991 warn when a VM is started in a potentially
1992 insecure configuration.
1995 Disables hypervisor mitigations and doesn't
1997 It also drops the swap size and available
1998 RAM limit restriction on both hypervisor and
2003 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2009 lapic [X86-32,APIC] Enable the local APIC even if BIOS
2012 lapic= [x86,APIC] "notscdeadline" Do not use TSC deadline
2013 value for LAPIC timer one-shot implementation. Default
2014 back to the programmable timer unit in the LAPIC.
2016 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
2019 libata.dma= [LIBATA] DMA control
2020 libata.dma=0 Disable all PATA and SATA DMA
2021 libata.dma=1 PATA and SATA Disk DMA only
2022 libata.dma=2 ATAPI (CDROM) DMA only
2023 libata.dma=4 Compact Flash DMA only
2024 Combinations also work, so libata.dma=3 enables DMA
2025 for disks and CDROMs, but not CFs.
2027 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2028 libata.ignore_hpa=0 keep BIOS limits (default)
2029 libata.ignore_hpa=1 ignore limits, using full disk
2031 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2035 libata.force= [LIBATA] Force configurations. The format is comma
2036 separated list of "[ID:]VAL" where ID is
2037 PORT[.DEVICE]. PORT and DEVICE are decimal numbers
2038 matching port, link or device. Basically, it matches
2039 the ATA ID string printed on console by libata. If
2040 the whole ID part is omitted, the last PORT and DEVICE
2041 values are used. If ID hasn't been specified yet, the
2042 configuration applies to all ports, links and devices.
2044 If only DEVICE is omitted, the parameter applies to
2045 the port and all links and devices behind it. DEVICE
2046 number of 0 either selects the first device or the
2047 first fan-out link behind PMP device. It does not
2048 select the host link. DEVICE number of 15 selects the
2049 host link and device attached to it.
2051 The VAL specifies the configuration to force. As long
2052 as there's no ambiguity shortcut notation is allowed.
2053 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2054 The following configurations can be forced.
2056 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2057 Any ID with matching PORT is used.
2059 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2061 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2062 udma[/][16,25,33,44,66,100,133] notation is also
2065 * [no]ncq: Turn on or off NCQ.
2067 * [no]ncqtrim: Turn off queued DSM TRIM.
2069 * nohrst, nosrst, norst: suppress hard, soft
2072 * rstonce: only attempt one reset during
2073 hot-unplug link recovery
2075 * dump_id: dump IDENTIFY data.
2077 * atapi_dmadir: Enable ATAPI DMADIR bridge support
2079 * disable: Disable this device.
2081 If there are multiple matching configurations changing
2082 the same attribute, the last one is used.
2084 memblock=debug [KNL] Enable memblock debug messages.
2086 load_ramdisk= [RAM] List of ramdisks to load from floppy
2087 See Documentation/blockdev/ramdisk.txt.
2089 lockd.nlm_grace_period=P [NFS] Assign grace period.
2092 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2095 lockd.nlm_timeout=T [NFS] Assign timeout value.
2098 lockd.nlm_udpport=M [NFS] Assign UDP port.
2101 locktorture.nreaders_stress= [KNL]
2102 Set the number of locking read-acquisition kthreads.
2103 Defaults to being automatically set based on the
2104 number of online CPUs.
2106 locktorture.nwriters_stress= [KNL]
2107 Set the number of locking write-acquisition kthreads.
2109 locktorture.onoff_holdoff= [KNL]
2110 Set time (s) after boot for CPU-hotplug testing.
2112 locktorture.onoff_interval= [KNL]
2113 Set time (s) between CPU-hotplug operations, or
2114 zero to disable CPU-hotplug testing.
2116 locktorture.shuffle_interval= [KNL]
2117 Set task-shuffle interval (jiffies). Shuffling
2118 tasks allows some CPUs to go into dyntick-idle
2119 mode during the locktorture test.
2121 locktorture.shutdown_secs= [KNL]
2122 Set time (s) after boot system shutdown. This
2123 is useful for hands-off automated testing.
2125 locktorture.stat_interval= [KNL]
2126 Time (s) between statistics printk()s.
2128 locktorture.stutter= [KNL]
2129 Time (s) to stutter testing, for example,
2130 specifying five seconds causes the test to run for
2131 five seconds, wait for five seconds, and so on.
2132 This tests the locking primitive's ability to
2133 transition abruptly to and from idle.
2135 locktorture.torture_runnable= [BOOT]
2136 Start locktorture running at boot time.
2138 locktorture.torture_type= [KNL]
2139 Specify the locking implementation to test.
2141 locktorture.verbose= [KNL]
2142 Enable additional printk() statements.
2144 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
2147 loglevel= All Kernel Messages with a loglevel smaller than the
2148 console loglevel will be printed to the console. It can
2149 also be changed with klogd or other programs. The
2150 loglevels are defined as follows:
2152 0 (KERN_EMERG) system is unusable
2153 1 (KERN_ALERT) action must be taken immediately
2154 2 (KERN_CRIT) critical conditions
2155 3 (KERN_ERR) error conditions
2156 4 (KERN_WARNING) warning conditions
2157 5 (KERN_NOTICE) normal but significant condition
2158 6 (KERN_INFO) informational
2159 7 (KERN_DEBUG) debug-level messages
2161 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
2162 in bytes. n must be a power of two and greater
2163 than the minimal size. The minimal size is defined
2164 by LOG_BUF_SHIFT kernel config parameter. There is
2165 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2166 that allows to increase the default size depending on
2167 the number of CPUs. See init/Kconfig for more details.
2169 logo.nologo [FB] Disables display of the built-in Linux logo.
2170 This may be used to provide more screen space for
2171 kernel log messages and is useful when debugging
2172 kernel boot problems.
2174 lp=0 [LP] Specify parallel ports to use, e.g,
2175 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2176 lp=reset first parallel port). 'lp=0' disables the
2177 lp=auto printer driver. 'lp=reset' (which can be
2178 specified in addition to the ports) causes
2179 attached printers to be reset. Using
2180 lp=port1,port2,... specifies the parallel ports
2181 to associate lp devices with, starting with
2182 lp0. A port specification may be 'none' to skip
2183 that lp device, or a parport name such as
2184 'parport0'. Specifying 'lp=auto' instead of a
2185 port specification list means that device IDs
2186 from each port should be examined, to see if
2187 an IEEE 1284-compliant printer is attached; if
2188 so, the driver will manage that printer.
2189 See also header of drivers/char/lp.c.
2192 Sets loops_per_jiffy to given constant, thus avoiding
2193 time-consuming boot-time autodetection (up to 250 ms per
2194 CPU). 0 enables autodetection (default). To determine
2195 the correct value for your kernel, boot with normal
2196 autodetection and see what value is printed. Note that
2197 on SMP systems the preset will be applied to all CPUs,
2198 which is likely to cause problems if your CPUs need
2199 significantly divergent settings. An incorrect value
2200 will cause delays in the kernel to be wrong, leading to
2201 unpredictable I/O errors and other breakage. Although
2202 unlikely, in the extreme case this might damage your
2206 Format: <io>,<irq>,<dma>
2208 machvec= [IA-64] Force the use of a particular machine-vector
2209 (machvec) in a generic kernel.
2210 Example: machvec=hpzx1_swiotlb
2212 machtype= [Loongson] Share the same kernel image file between different
2214 Example: machtype=lemote-yeeloong-2f-7inch
2216 max_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory greater
2217 than or equal to this physical address is ignored.
2219 maxcpus= [SMP] Maximum number of processors that an SMP kernel
2220 will bring up during bootup. maxcpus=n : n >= 0 limits
2221 the kernel to bring up 'n' processors. Surely after
2222 bootup you can bring up the other plugged cpu by executing
2223 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
2224 only takes effect during system bootup.
2225 While n=0 is a special case, it is equivalent to "nosmp",
2226 which also disables the IO APIC.
2228 max_loop= [LOOP] The number of loop block devices that get
2229 (loop.max_loop) unconditionally pre-created at init time. The default
2230 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
2231 of statically allocating a predefined number, loop
2232 devices can be requested on-demand with the
2233 /dev/loop-control interface.
2235 mce [X86-32] Machine Check Exception
2237 mce=option [X86-64] See Documentation/x86/x86_64/boot-options.txt
2239 md= [HW] RAID subsystems devices and level
2240 See Documentation/admin-guide/md.rst.
2243 Format: <first>,<last>
2244 Specifies range of consoles to be captured by the MDA.
2247 Control mitigation for the Micro-architectural Data
2248 Sampling (MDS) vulnerability.
2250 Certain CPUs are vulnerable to an exploit against CPU
2251 internal buffers which can forward information to a
2252 disclosure gadget under certain conditions.
2254 In vulnerable processors, the speculatively
2255 forwarded data can be used in a cache side channel
2256 attack, to access data to which the attacker does
2257 not have direct access.
2259 This parameter controls the MDS mitigation. The
2262 full - Enable MDS mitigation on vulnerable CPUs
2263 full,nosmt - Enable MDS mitigation and disable
2264 SMT on vulnerable CPUs
2265 off - Unconditionally disable MDS mitigation
2267 On TAA-affected machines, mds=off can be prevented by
2268 an active TAA mitigation as both vulnerabilities are
2269 mitigated with the same mechanism so in order to disable
2270 this mitigation, you need to specify tsx_async_abort=off
2273 Not specifying this option is equivalent to
2276 For details see: Documentation/admin-guide/hw-vuln/mds.rst
2278 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
2279 Amount of memory to be used when the kernel is not able
2280 to see the whole system memory or for test.
2281 [X86] Work as limiting max address. Use together
2282 with memmap= to avoid physical address space collisions.
2283 Without memmap= PCI devices could be placed at addresses
2284 belonging to unused RAM.
2286 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
2290 [KNL,SH] Allow user to override the default size for
2291 per-device physically contiguous DMA buffers.
2293 memhp_default_state=online/offline
2294 [KNL] Set the initial state for the memory hotplug
2295 onlining policy. If not specified, the default value is
2296 set according to the
2297 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
2299 See Documentation/memory-hotplug.txt.
2301 memmap=exactmap [KNL,X86] Enable setting of an exact
2302 E820 memory map, as specified by the user.
2303 Such memmap=exactmap lines can be constructed based on
2304 BIOS output or other requirements. See the memmap=nn@ss
2307 memmap=nn[KMG]@ss[KMG]
2308 [KNL] Force usage of a specific region of memory.
2309 Region of memory to be used is from ss to ss+nn.
2310 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
2311 which limits max address to nn[KMG].
2312 Multiple different regions can be specified,
2315 memmap=100M@2G,100M#3G,1G!1024G
2317 memmap=nn[KMG]#ss[KMG]
2318 [KNL,ACPI] Mark specific memory as ACPI data.
2319 Region of memory to be marked is from ss to ss+nn.
2321 memmap=nn[KMG]$ss[KMG]
2322 [KNL,ACPI] Mark specific memory as reserved.
2323 Region of memory to be reserved is from ss to ss+nn.
2324 Example: Exclude memory from 0x18690000-0x1869ffff
2325 memmap=64K$0x18690000
2327 memmap=0x10000$0x18690000
2328 Some bootloaders may need an escape character before '$',
2329 like Grub2, otherwise '$' and the following number
2332 memmap=nn[KMG]!ss[KMG]
2333 [KNL,X86] Mark specific memory as protected.
2334 Region of memory to be used, from ss to ss+nn.
2335 The memory region may be marked as e820 type 12 (0xc)
2336 and is NVDIMM or ADR memory.
2338 memory_corruption_check=0/1 [X86]
2339 Some BIOSes seem to corrupt the first 64k of
2340 memory when doing things like suspend/resume.
2341 Setting this option will scan the memory
2342 looking for corruption. Enabling this will
2343 both detect corruption and prevent the kernel
2344 from using the memory being corrupted.
2345 However, its intended as a diagnostic tool; if
2346 repeatable BIOS-originated corruption always
2347 affects the same memory, you can use memmap=
2348 to prevent the kernel from using that memory.
2350 memory_corruption_check_size=size [X86]
2351 By default it checks for corruption in the low
2352 64k, making this memory unavailable for normal
2353 use. Use this parameter to scan for
2354 corruption in more or less memory.
2356 memory_corruption_check_period=seconds [X86]
2357 By default it checks for corruption every 60
2358 seconds. Use this parameter to check at some
2359 other rate. 0 disables periodic checking.
2361 memtest= [KNL,X86,ARM] Enable memtest
2363 default : 0 <disable>
2364 Specifies the number of memtest passes to be
2365 performed. Each pass selects another test
2366 pattern from a given set of patterns. Memtest
2367 fills the memory with this pattern, validates
2368 memory contents and reserves bad memory
2369 regions that are detected.
2371 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
2372 Valid arguments: on, off
2373 Default (depends on kernel configuration option):
2374 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
2375 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
2376 mem_encrypt=on: Activate SME
2377 mem_encrypt=off: Do not activate SME
2379 Refer to Documentation/x86/amd-memory-encryption.txt
2380 for details on when memory encryption can be activated.
2382 mem_sleep_default= [SUSPEND] Default system suspend mode:
2383 s2idle - Suspend-To-Idle
2384 shallow - Power-On Suspend or equivalent (if supported)
2385 deep - Suspend-To-RAM or equivalent (if supported)
2386 See Documentation/power/states.txt.
2388 meye.*= [HW] Set MotionEye Camera parameters
2389 See Documentation/video4linux/meye.txt.
2391 mfgpt_irq= [IA-32] Specify the IRQ to use for the
2392 Multi-Function General Purpose Timers on AMD Geode
2395 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
2396 the BIOS has incorrectly applied a workaround. TinyBIOS
2397 version 0.98 is known to be affected, 0.99 fixes the
2398 problem by letting the user disable the workaround.
2402 min_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory below this
2403 physical address is ignored.
2405 mini2440= [ARM,HW,KNL]
2406 Format:[0..2][b][c][t]
2408 MINI2440 configuration specification:
2409 0 - The attached screen is the 3.5" TFT
2410 1 - The attached screen is the 7" TFT
2411 2 - The VGA Shield is attached (1024x768)
2412 Leaving out the screen size parameter will not load
2413 the TFT driver, and the framebuffer will be left
2415 b - Enable backlight. The TFT backlight pin will be
2416 linked to the kernel VESA blanking code and a GPIO
2417 LED. This parameter is not necessary when using the
2419 c - Enable the s3c camera interface.
2420 t - Reserved for enabling touchscreen support. The
2421 touchscreen support is not enabled in the mainstream
2422 kernel as of 2.6.30, a preliminary port can be found
2423 in the "bleeding edge" mini2440 support kernel at
2424 http://repo.or.cz/w/linux-2.6/mini2440.git
2427 [X86,PPC,S390,ARM64] Control optional mitigations for
2428 CPU vulnerabilities. This is a set of curated,
2429 arch-independent options, each of which is an
2430 aggregation of existing arch-specific options.
2433 Disable all optional CPU mitigations. This
2434 improves system performance, but it may also
2435 expose users to several CPU vulnerabilities.
2436 Equivalent to: nopti [X86,PPC]
2441 nospectre_v2 [X86,PPC,S390,ARM64]
2442 spectre_v2_user=off [X86]
2443 spec_store_bypass_disable=off [X86,PPC]
2444 ssbd=force-off [ARM64]
2447 tsx_async_abort=off [X86]
2448 kvm.nx_huge_pages=off [X86]
2449 no_entry_flush [PPC]
2450 no_uaccess_flush [PPC]
2453 This does not have any effect on
2454 kvm.nx_huge_pages when
2455 kvm.nx_huge_pages=force.
2458 Mitigate all CPU vulnerabilities, but leave SMT
2459 enabled, even if it's vulnerable. This is for
2460 users who don't want to be surprised by SMT
2461 getting disabled across kernel upgrades, or who
2462 have other ways of avoiding SMT-based attacks.
2463 Equivalent to: (default behavior)
2466 Mitigate all CPU vulnerabilities, disabling SMT
2467 if needed. This is for users who always want to
2468 be fully mitigated, even if it means losing SMT.
2469 Equivalent to: l1tf=flush,nosmt [X86]
2470 mds=full,nosmt [X86]
2471 tsx_async_abort=full,nosmt [X86]
2474 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
2475 parameter allows control of the logging verbosity for
2476 the additional memory initialisation checks. A value
2477 of 0 disables mminit logging and a level of 4 will
2478 log everything. Information is printed at KERN_DEBUG
2479 so loglevel=8 may also need to be specified.
2482 [KNL] When CONFIG_MODULE_SIG is set, this means that
2483 modules without (valid) signatures will fail to load.
2484 Note that if CONFIG_MODULE_SIG_FORCE is set, that
2485 is always true, so this option does nothing.
2487 module_blacklist= [KNL] Do not load a comma-separated list of
2488 modules. Useful for debugging problem modules.
2491 [MOUSE] Maximum time between finger touching and
2492 leaving touchpad surface for touch to be considered
2493 a tap and be reported as a left button click (for
2494 touchpads working in absolute mode only).
2496 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
2497 reporting absolute coordinates, such as tablets
2498 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
2499 reporting absolute coordinates, such as tablets
2501 movablecore=nn[KMG] [KNL,X86,IA-64,PPC] This parameter
2502 is similar to kernelcore except it specifies the
2503 amount of memory used for migratable allocations.
2504 If both kernelcore and movablecore is specified,
2505 then kernelcore will be at *least* the specified
2506 value but may be more. If movablecore on its own
2507 is specified, the administrator must be careful
2508 that the amount of memory usable for all allocations
2511 movable_node [KNL] Boot-time switch to make hotplugable memory
2512 NUMA nodes to be movable. This means that the memory
2513 of such nodes will be usable only for movable
2514 allocations which rules out almost all kernel
2515 allocations. Use with caution!
2517 MTD_Partition= [MTD]
2518 Format: <name>,<region-number>,<size>,<offset>
2520 MTD_Region= [MTD] Format:
2521 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
2524 See drivers/mtd/cmdlinepart.c.
2526 multitce=off [PPC] This parameter disables the use of the pSeries
2527 firmware feature for updating multiple TCE entries
2530 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
2532 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
2534 boundary - index of last SLC block on Flex-OneNAND.
2535 The remaining blocks are configured as MLC blocks.
2536 lock - Configure if Flex-OneNAND boundary should be locked.
2537 Once locked, the boundary cannot be changed.
2538 1 indicates lock status, 0 indicates unlock status.
2541 ARM/S3C2412 JIVE boot control
2543 See arch/arm/mach-s3c2412/mach-jive.c
2545 mtouchusb.raw_coordinates=
2546 [HW] Make the MicroTouch USB driver use raw coordinates
2547 ('y', default) or cooked coordinates ('n')
2549 mtrr_chunk_size=nn[KMG] [X86]
2550 used for mtrr cleanup. It is largest continuous chunk
2551 that could hold holes aka. UC entries.
2553 mtrr_gran_size=nn[KMG] [X86]
2554 Used for mtrr cleanup. It is granularity of mtrr block.
2556 Large value could prevent small alignment from
2559 mtrr_spare_reg_nr=n [X86]
2561 Range: 0,7 : spare reg number
2563 Used for mtrr cleanup. It is spare mtrr entries number.
2564 Set to 2 or more if your graphical card needs more.
2566 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
2568 netdev= [NET] Network devices parameters
2569 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
2570 Note that mem_start is often overloaded to mean
2571 something different and driver-specific.
2572 This usage is only documented in each driver source
2576 [NETFILTER] Enable connection tracking flow accounting
2577 0 to disable accounting
2578 1 to enable accounting
2581 nfsaddrs= [NFS] Deprecated. Use ip= instead.
2582 See Documentation/filesystems/nfs/nfsroot.txt.
2584 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
2585 See Documentation/filesystems/nfs/nfsroot.txt.
2587 nfsrootdebug [NFS] enable nfsroot debugging messages.
2588 See Documentation/filesystems/nfs/nfsroot.txt.
2590 nfs.callback_nr_threads=
2591 [NFSv4] set the total number of threads that the
2592 NFS client will assign to service NFSv4 callback
2595 nfs.callback_tcpport=
2596 [NFS] set the TCP port on which the NFSv4 callback
2597 channel should listen.
2600 [NFS] sets the pathname to the program which is used
2601 to update the NFS client cache entries.
2603 nfs.cache_getent_timeout=
2604 [NFS] sets the timeout after which an attempt to
2605 update a cache entry is deemed to have failed.
2607 nfs.idmap_cache_timeout=
2608 [NFS] set the maximum lifetime for idmapper cache
2612 [NFS] enable 64-bit inode numbers.
2613 If zero, the NFS client will fake up a 32-bit inode
2614 number for the readdir() and stat() syscalls instead
2615 of returning the full 64-bit number.
2616 The default is to return 64-bit inode numbers.
2618 nfs.max_session_cb_slots=
2619 [NFSv4.1] Sets the maximum number of session
2620 slots the client will assign to the callback
2621 channel. This determines the maximum number of
2622 callbacks the client will process in parallel for
2623 a particular server.
2625 nfs.max_session_slots=
2626 [NFSv4.1] Sets the maximum number of session slots
2627 the client will attempt to negotiate with the server.
2628 This limits the number of simultaneous RPC requests
2629 that the client can send to the NFSv4.1 server.
2630 Note that there is little point in setting this
2631 value higher than the max_tcp_slot_table_limit.
2633 nfs.nfs4_disable_idmapping=
2634 [NFSv4] When set to the default of '1', this option
2635 ensures that both the RPC level authentication
2636 scheme and the NFS level operations agree to use
2637 numeric uids/gids if the mount is using the
2638 'sec=sys' security flavour. In effect it is
2639 disabling idmapping, which can make migration from
2640 legacy NFSv2/v3 systems to NFSv4 easier.
2641 Servers that do not support this mode of operation
2642 will be autodetected by the client, and it will fall
2643 back to using the idmapper.
2644 To turn off this behaviour, set the value to '0'.
2646 [NFS4] Specify an additional fixed unique ident-
2647 ification string that NFSv4 clients can insert into
2648 their nfs_client_id4 string. This is typically a
2649 UUID that is generated at system install time.
2651 nfs.send_implementation_id =
2652 [NFSv4.1] Send client implementation identification
2653 information in exchange_id requests.
2654 If zero, no implementation identification information
2656 The default is to send the implementation identification
2659 nfs.recover_lost_locks =
2660 [NFSv4] Attempt to recover locks that were lost due
2661 to a lease timeout on the server. Please note that
2662 doing this risks data corruption, since there are
2663 no guarantees that the file will remain unchanged
2664 after the locks are lost.
2665 If you want to enable the kernel legacy behaviour of
2666 attempting to recover these locks, then set this
2668 The default parameter value of '0' causes the kernel
2669 not to attempt recovery of lost locks.
2671 nfs4.layoutstats_timer =
2672 [NFSv4.2] Change the rate at which the kernel sends
2673 layoutstats to the pNFS metadata server.
2675 Setting this to value to 0 causes the kernel to use
2676 whatever value is the default set by the layout
2677 driver. A non-zero value sets the minimum interval
2678 in seconds between layoutstats transmissions.
2680 nfsd.nfs4_disable_idmapping=
2681 [NFSv4] When set to the default of '1', the NFSv4
2682 server will return only numeric uids and gids to
2683 clients using auth_sys, and will accept numeric uids
2684 and gids from such clients. This is intended to ease
2685 migration from NFSv2/v3.
2687 nmi_debug= [KNL,SH] Specify one or more actions to take
2688 when a NMI is triggered.
2689 Format: [state][,regs][,debounce][,die]
2691 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
2692 Format: [panic,][nopanic,][num]
2694 0 - turn hardlockup detector in nmi_watchdog off
2695 1 - turn hardlockup detector in nmi_watchdog on
2696 When panic is specified, panic when an NMI watchdog
2697 timeout occurs (or 'nopanic' to override the opposite
2698 default). To disable both hard and soft lockup detectors,
2699 please see 'nowatchdog'.
2700 This is useful when you use a panic=... timeout and
2701 need the box quickly up again.
2703 netpoll.carrier_timeout=
2704 [NET] Specifies amount of time (in seconds) that
2705 netpoll should wait for a carrier. By default netpoll
2708 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
2709 emulation library even if a 387 maths coprocessor
2713 [HW] Never suspend the console
2714 Disable suspending of consoles during suspend and
2715 hibernate operations. Once disabled, debugging
2716 messages can reach various consoles while the rest
2717 of the system is being put to sleep (ie, while
2718 debugging driver suspend/resume hooks). This may
2719 not work reliably with all consoles, but is known
2720 to work with serial and VGA consoles.
2721 To facilitate more flexible debugging, we also add
2722 console_suspend, a printk module parameter to control
2723 it. Users could use console_suspend (usually
2724 /sys/module/printk/parameters/console_suspend) to
2725 turn on/off it dynamically.
2727 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
2728 caches in the slab allocator. Saves per-node memory,
2729 but will impact performance.
2733 noaltinstr [S390] Disables alternative instructions patching
2734 (CPU alternatives feature).
2736 noapic [SMP,APIC] Tells the kernel to not make use of any
2737 IOAPICs that may be present in the system.
2739 noautogroup Disable scheduler automatic task group creation.
2741 nobats [PPC] Do not use BATs for mapping kernel lowmem
2742 on "Classic" PPC cores.
2746 noclflush [BUGS=X86] Don't use the CLFLUSH instruction
2748 nodelayacct [KNL] Disable per-task delay accounting
2750 nodsp [SH] Disable hardware DSP at boot time.
2752 noefi Disable EFI runtime services support.
2754 no_entry_flush [PPC] Don't flush the L1-D cache when entering the kernel.
2759 On X86-32 available only on PAE configured kernels.
2760 noexec=on: enable non-executable mappings (default)
2761 noexec=off: disable non-executable mappings
2764 Disable SMAP (Supervisor Mode Access Prevention)
2765 even if it is supported by processor.
2768 Disable SMEP (Supervisor Mode Execution Prevention)
2769 even if it is supported by processor.
2772 This affects only 32-bit executables.
2773 noexec32=on: enable non-executable mappings (default)
2774 read doesn't imply executable mappings
2775 noexec32=off: disable non-executable mappings
2776 read implies executable mappings
2778 nofpu [MIPS,SH] Disable hardware FPU at boot time.
2780 nofxsr [BUGS=X86-32] Disables x86 floating point extended
2781 register save and restore. The kernel will only save
2782 legacy floating-point registers on task switch.
2784 nohugeiomap [KNL,x86] Disable kernel huge I/O mappings.
2786 nosmt [KNL,S390] Disable symmetric multithreading (SMT).
2787 Equivalent to smt=1.
2789 [KNL,x86] Disable symmetric multithreading (SMT).
2790 nosmt=force: Force disable SMT, cannot be undone
2791 via the sysfs control file.
2793 nospectre_v1 [X66, PPC] Disable mitigations for Spectre Variant 1
2794 (bounds check bypass). With this option data leaks
2795 are possible in the system.
2797 nospectre_v2 [X86,PPC_FSL_BOOK3E,ARM64] Disable all mitigations for
2798 the Spectre variant 2 (indirect branch prediction)
2799 vulnerability. System may allow data leaks with this
2802 nospec_store_bypass_disable
2803 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
2806 [PPC] Don't flush the L1-D cache after accessing user data.
2808 noxsave [BUGS=X86] Disables x86 extended register state save
2809 and restore using xsave. The kernel will fallback to
2810 enabling legacy floating-point and sse state.
2812 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
2813 register states. The kernel will fall back to use
2814 xsave to save the states. By using this parameter,
2815 performance of saving the states is degraded because
2816 xsave doesn't support modified optimization while
2817 xsaveopt supports it on xsaveopt enabled systems.
2819 noxsaves [X86] Disables xsaves and xrstors used in saving and
2820 restoring x86 extended register state in compacted
2821 form of xsave area. The kernel will fall back to use
2822 xsaveopt and xrstor to save and restore the states
2823 in standard form of xsave area. By using this
2824 parameter, xsave area per process might occupy more
2825 memory on xsaves enabled systems.
2827 nohlt [BUGS=ARM,SH] Tells the kernel that the sleep(SH) or
2828 wfi(ARM) instruction doesn't work correctly and not to
2829 use it. This is also useful when using JTAG debugger.
2831 no_file_caps Tells the kernel not to honor file capabilities. The
2832 only way then for a file to be executed with privilege
2833 is to be setuid root or executed by root.
2835 nohalt [IA-64] Tells the kernel not to use the power saving
2836 function PAL_HALT_LIGHT when idle. This increases
2837 power-consumption. On the positive side, it reduces
2838 interrupt wake-up latency, which may improve performance
2839 in certain environments such as networked servers or
2842 nohibernate [HIBERNATION] Disable hibernation and resume.
2844 nohz= [KNL] Boottime enable/disable dynamic ticks
2845 Valid arguments: on, off
2848 nohz_full= [KNL,BOOT]
2849 The argument is a cpu list, as described above.
2850 In kernels built with CONFIG_NO_HZ_FULL=y, set
2851 the specified list of CPUs whose tick will be stopped
2852 whenever possible. The boot CPU will be forced outside
2853 the range to maintain the timekeeping. Any CPUs
2854 in this list will have their RCU callbacks offloaded,
2855 just as if they had also been called out in the
2856 rcu_nocbs= boot parameter.
2858 noiotrap [SH] Disables trapped I/O port accesses.
2860 noirqdebug [X86-32] Disables the code which attempts to detect and
2861 disable unhandled interrupt sources.
2863 no_timer_check [X86,APIC] Disables the code which tests for
2864 broken timer IRQ sources.
2866 noisapnp [ISAPNP] Disables ISA PnP code.
2868 noinitrd [RAM] Tells the kernel not to load any configured
2871 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
2873 [Deprecated - use intremap=off]
2877 noinvpcid [X86] Disable the INVPCID cpu feature.
2879 nojitter [IA-64] Disables jitter checking for ITC timers.
2881 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
2883 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
2887 [X86,PV_OPS] Disable paravirtualized VMware scheduler
2888 clock and use the default one.
2890 no-steal-acc [X86,KVM] Disable paravirtualized steal time accounting.
2891 steal time is computed, but won't influence scheduler
2894 nolapic [X86-32,APIC] Do not enable or use the local APIC.
2896 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
2898 noltlbs [PPC] Do not use large page/tlb entries for kernel
2899 lowmem mapping on PPC40x and PPC8xx
2901 nomca [IA-64] Disable machine check abort handling
2903 nomce [X86-32] Disable Machine Check Exception
2905 nomfgpt [X86-32] Disable Multi-Function General Purpose
2906 Timer usage (for AMD Geode machines).
2908 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
2909 shutdown the other cpus. Instead use the REBOOT_VECTOR
2912 nomodule Disable module load
2914 nopat [X86] Disable PAT (page attribute table extension of
2915 pagetables) support.
2917 nopcid [X86-64] Disable the PCID cpu feature.
2919 norandmaps Don't use address space randomization. Equivalent to
2920 echo 0 > /proc/sys/kernel/randomize_va_space
2922 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
2923 with UP alternatives
2925 nordrand [X86] Disable kernel use of the RDRAND and
2926 RDSEED instructions even if they are supported
2927 by the processor. RDRAND and RDSEED are still
2928 available to user space applications.
2930 noresume [SWSUSP] Disables resume and restores original swap
2933 no-scroll [VGA] Disables scrollback.
2934 This is required for the Braillex ib80-piezo Braille
2935 reader made by F.H. Papenmeier (Germany).
2939 nosep [BUGS=X86-32] Disables x86 SYSENTER/SYSEXIT support.
2941 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
2942 and disable the IO APIC. legacy for "maxcpus=0".
2944 nosoftlockup [KNL] Disable the soft-lockup detector.
2946 nosync [HW,M68K] Disables sync negotiation for all devices.
2948 notsc [BUGS=X86-32] Disable Time Stamp Counter
2950 nowatchdog [KNL] Disable both lockup detectors, i.e.
2951 soft-lockup and NMI watchdog (hard-lockup).
2955 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
2957 cpu0_hotplug [X86] Turn on CPU0 hotplug feature when
2958 CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off.
2959 Some features depend on CPU0. Known dependencies are:
2960 1. Resume from suspend/hibernate depends on CPU0.
2961 Suspend/hibernate will fail if CPU0 is offline and you
2962 need to online CPU0 before suspend/hibernate.
2963 2. PIC interrupts also depend on CPU0. CPU0 can't be
2964 removed if a PIC interrupt is detected.
2965 It's said poweroff/reboot may depend on CPU0 on some
2966 machines although I haven't seen such issues so far
2967 after CPU0 is offline on a few tested machines.
2968 If the dependencies are under your control, you can
2969 turn on cpu0_hotplug.
2971 nps_mtm_hs_ctr= [KNL,ARC]
2972 This parameter sets the maximum duration, in
2973 cycles, each HW thread of the CTOP can run
2974 without interruptions, before HW switches it.
2975 The actual maximum duration is 16 times this
2977 Format: integer between 1 and 255
2980 nptcg= [IA-64] Override max number of concurrent global TLB
2981 purges which is reported from either PAL_VM_SUMMARY or
2984 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
2985 could support. nr_cpus=n : n >= 1 limits the kernel to
2986 support 'n' processors. It could be larger than the
2987 number of already plugged CPU during bootup, later in
2988 runtime you can physically add extra cpu until it reaches
2989 n. So during boot up some boot time memory for per-cpu
2990 variables need be pre-allocated for later physical cpu
2993 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
2995 numa_balancing= [KNL,X86] Enable or disable automatic NUMA balancing.
2996 Allowed values are enable and disable
2998 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
2999 'node', 'default' can be specified
3000 This can be set from sysctl after boot.
3001 See Documentation/sysctl/vm.txt for details.
3003 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
3004 See Documentation/debugging-via-ohci1394.txt for more
3007 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
3008 Rather than timing out after 20 ms if an EC
3009 command is not properly ACKed, override the length
3010 of the timeout. We have interrupts disabled while
3011 waiting for the ACK, so if this is set too high
3012 interrupts *may* be lost!
3014 omap_mux= [OMAP] Override bootloader pin multiplexing.
3015 Format: <mux_mode0.mode_name=value>...
3016 For example, to override I2C bus2:
3017 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
3019 oprofile.timer= [HW]
3020 Use timer interrupt instead of performance counters
3022 oprofile.cpu_type= Force an oprofile cpu type
3023 This might be useful if you have an older oprofile
3024 userland or if you want common events.
3025 Format: { arch_perfmon }
3026 arch_perfmon: [X86] Force use of architectural
3027 perfmon on Intel CPUs instead of the
3028 CPU specific event set.
3029 timer: [X86] Force use of architectural NMI
3030 timer mode (see also oprofile.timer
3031 for generic hr timer mode)
3033 oops=panic Always panic on oopses. Default is to just kill the
3034 process, but there is a small probability of
3035 deadlocking the machine.
3036 This will also cause panics on machine check exceptions.
3037 Useful together with panic=30 to trigger a reboot.
3040 See Documentation/sound/oss/oss-parameters.txt
3042 page_owner= [KNL] Boot-time page_owner enabling option.
3043 Storage of the information about who allocated
3044 each page is disabled in default. With this switch,
3046 on: enable the feature
3048 page_poison= [KNL] Boot-time parameter changing the state of
3049 poisoning on the buddy allocator.
3050 off: turn off poisoning
3051 on: turn on poisoning
3053 panic= [KNL] Kernel behaviour on panic: delay <timeout>
3054 timeout > 0: seconds before rebooting
3055 timeout = 0: wait forever
3056 timeout < 0: reboot immediately
3059 panic_on_warn panic() instead of WARN(). Useful to cause kdump
3062 crash_kexec_post_notifiers
3063 Run kdump after running panic-notifiers and dumping
3064 kmsg. This only for the users who doubt kdump always
3065 succeeds in any situation.
3066 Note that this also increases risks of kdump failure,
3067 because some panic notifiers can make the crashed
3068 kernel more unstable.
3070 parkbd.port= [HW] Parallel port number the keyboard adapter is
3071 connected to, default is 0.
3073 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
3074 0 for XT, 1 for AT (default is AT).
3077 parport= [HW,PPT] Specify parallel ports. 0 disables.
3078 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
3079 Use 'auto' to force the driver to use any
3080 IRQ/DMA settings detected (the default is to
3081 ignore detected IRQ/DMA settings because of
3082 possible conflicts). You can specify the base
3083 address, IRQ, and DMA settings; IRQ and DMA
3084 should be numbers, or 'auto' (for using detected
3085 settings on that particular port), or 'nofifo'
3086 (to avoid using a FIFO even if it is detected).
3087 Parallel ports are assigned in the order they
3088 are specified on the command line, starting
3091 parport_init_mode= [HW,PPT]
3092 Configure VIA parallel port to operate in
3093 a specific mode. This is necessary on Pegasos
3094 computer where firmware has no options for setting
3095 up parallel port mode and sets it to spp.
3096 Currently this function knows 686a and 8231 chips.
3097 Format: [spp|ps2|epp|ecp|ecpepp]
3100 Halt all CPUs after the first oops has been printed for
3101 the specified number of seconds. This is to be used if
3102 your oopses keep scrolling off the screen.
3107 See header of drivers/block/paride/pcd.c.
3108 See also Documentation/blockdev/paride.txt.
3110 pci=option[,option...] [PCI] various PCI subsystem options:
3111 earlydump [X86] dump PCI config space before the kernel
3113 off [X86] don't probe for the PCI bus
3114 bios [X86-32] force use of PCI BIOS, don't access
3115 the hardware directly. Use this if your machine
3116 has a non-standard PCI host bridge.
3117 nobios [X86-32] disallow use of PCI BIOS, only direct
3118 hardware access methods are allowed. Use this
3119 if you experience crashes upon bootup and you
3120 suspect they are caused by the BIOS.
3121 conf1 [X86] Force use of PCI Configuration Access
3122 Mechanism 1 (config address in IO port 0xCF8,
3123 data in IO port 0xCFC, both 32-bit).
3124 conf2 [X86] Force use of PCI Configuration Access
3125 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
3126 the function, IO port 0xCFA, also 8-bit, sets
3127 bus number. The config space is then accessed
3128 through ports 0xC000-0xCFFF).
3129 See http://wiki.osdev.org/PCI for more info
3130 on the configuration access mechanisms.
3131 noaer [PCIE] If the PCIEAER kernel config parameter is
3132 enabled, this kernel boot option can be used to
3133 disable the use of PCIE advanced error reporting.
3134 nodomains [PCI] Disable support for multiple PCI
3135 root domains (aka PCI segments, in ACPI-speak).
3136 nommconf [X86] Disable use of MMCONFIG for PCI
3138 check_enable_amd_mmconf [X86] check for and enable
3139 properly configured MMIO access to PCI
3140 config space on AMD family 10h CPU
3141 nomsi [MSI] If the PCI_MSI kernel config parameter is
3142 enabled, this kernel boot option can be used to
3143 disable the use of MSI interrupts system-wide.
3144 noioapicquirk [APIC] Disable all boot interrupt quirks.
3145 Safety option to keep boot IRQs enabled. This
3146 should never be necessary.
3147 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
3148 primary IO-APIC for bridges that cannot disable
3149 boot IRQs. This fixes a source of spurious IRQs
3150 when the system masks IRQs.
3151 noioapicreroute [APIC] Disable workaround that uses the
3152 boot IRQ equivalent of an IRQ that connects to
3153 a chipset where boot IRQs cannot be disabled.
3154 The opposite of ioapicreroute.
3155 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
3156 routing table. These calls are known to be buggy
3157 on several machines and they hang the machine
3158 when used, but on other computers it's the only
3159 way to get the interrupt routing table. Try
3160 this option if the kernel is unable to allocate
3161 IRQs or discover secondary PCI buses on your
3163 rom [X86] Assign address space to expansion ROMs.
3164 Use with caution as certain devices share
3165 address decoders between ROMs and other
3167 norom [X86] Do not assign address space to
3168 expansion ROMs that do not already have
3169 BIOS assigned address ranges.
3170 nobar [X86] Do not assign address space to the
3171 BARs that weren't assigned by the BIOS.
3172 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
3173 assigned automatically to PCI devices. You can
3174 make the kernel exclude IRQs of your ISA cards
3176 pirqaddr=0xAAAAA [X86] Specify the physical address
3177 of the PIRQ table (normally generated
3178 by the BIOS) if it is outside the
3179 F0000h-100000h range.
3180 lastbus=N [X86] Scan all buses thru bus #N. Can be
3181 useful if the kernel is unable to find your
3182 secondary buses and you want to tell it
3183 explicitly which ones they are.
3184 assign-busses [X86] Always assign all PCI bus
3185 numbers ourselves, overriding
3186 whatever the firmware may have done.
3187 usepirqmask [X86] Honor the possible IRQ mask stored
3188 in the BIOS $PIR table. This is needed on
3189 some systems with broken BIOSes, notably
3190 some HP Pavilion N5400 and Omnibook XE3
3191 notebooks. This will have no effect if ACPI
3192 IRQ routing is enabled.
3193 noacpi [X86] Do not use ACPI for IRQ routing
3194 or for PCI scanning.
3195 use_crs [X86] Use PCI host bridge window information
3196 from ACPI. On BIOSes from 2008 or later, this
3197 is enabled by default. If you need to use this,
3198 please report a bug.
3199 nocrs [X86] Ignore PCI host bridge windows from ACPI.
3200 If you need to use this, please report a bug.
3201 routeirq Do IRQ routing for all PCI devices.
3202 This is normally done in pci_enable_device(),
3203 so this option is a temporary workaround
3204 for broken drivers that don't call it.
3205 skip_isa_align [X86] do not align io start addr, so can
3206 handle more pci cards
3207 noearly [X86] Don't do any early type 1 scanning.
3208 This might help on some broken boards which
3209 machine check when some devices' config space
3210 is read. But various workarounds are disabled
3211 and some IOMMU drivers will not work.
3212 bfsort Sort PCI devices into breadth-first order.
3213 This sorting is done to get a device
3214 order compatible with older (<= 2.4) kernels.
3215 nobfsort Don't sort PCI devices into breadth-first order.
3216 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
3217 tuning and use the BIOS-configured MPS defaults.
3218 pcie_bus_safe Set every device's MPS to the largest value
3219 supported by all devices below the root complex.
3220 pcie_bus_perf Set device MPS to the largest allowable MPS
3221 based on its parent bus. Also set MRRS (Max
3222 Read Request Size) to the largest supported
3223 value (no larger than the MPS that the device
3224 or bus can support) for best performance.
3225 pcie_bus_peer2peer Set every device's MPS to 128B, which
3226 every device is guaranteed to support. This
3227 configuration allows peer-to-peer DMA between
3228 any pair of devices, possibly at the cost of
3229 reduced performance. This also guarantees
3230 that hot-added devices will work.
3231 cbiosize=nn[KMG] The fixed amount of bus space which is
3232 reserved for the CardBus bridge's IO window.
3233 The default value is 256 bytes.
3234 cbmemsize=nn[KMG] The fixed amount of bus space which is
3235 reserved for the CardBus bridge's memory
3236 window. The default value is 64 megabytes.
3239 [<order of align>@][<domain>:]<bus>:<slot>.<func>[; ...]
3240 [<order of align>@]pci:<vendor>:<device>\
3241 [:<subvendor>:<subdevice>][; ...]
3242 Specifies alignment and device to reassign
3243 aligned memory resources.
3244 If <order of align> is not specified,
3245 PAGE_SIZE is used as alignment.
3246 PCI-PCI bridge can be specified, if resource
3247 windows need to be expanded.
3248 To specify the alignment for several
3249 instances of a device, the PCI vendor,
3250 device, subvendor, and subdevice may be
3251 specified, e.g., 4096@pci:8086:9c22:103c:198f
3252 ecrc= Enable/disable PCIe ECRC (transaction layer
3253 end-to-end CRC checking).
3254 bios: Use BIOS/firmware settings. This is the
3258 hpiosize=nn[KMG] The fixed amount of bus space which is
3259 reserved for hotplug bridge's IO window.
3260 Default size is 256 bytes.
3261 hpmemsize=nn[KMG] The fixed amount of bus space which is
3262 reserved for hotplug bridge's memory window.
3263 Default size is 2 megabytes.
3264 hpbussize=nn The minimum amount of additional bus numbers
3265 reserved for buses below a hotplug bridge.
3267 realloc= Enable/disable reallocating PCI bridge resources
3268 if allocations done by BIOS are too small to
3269 accommodate resources required by all child
3271 off: Turn realloc off
3273 realloc same as realloc=on
3274 noari do not use PCIe ARI.
3275 pcie_scan_all Scan all possible PCIe devices. Otherwise we
3276 only look for one device below a PCIe downstream
3279 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
3282 force Enable ASPM even on devices that claim not to support it.
3283 WARNING: Forcing ASPM on may cause system lockups.
3285 pcie_hp= [PCIE] PCI Express Hotplug driver options:
3286 nomsi Do not use MSI for PCI Express Native Hotplug (this
3287 makes all PCIe ports use INTx for hotplug services).
3289 pcie_ports= [PCIE] PCIe ports handling:
3290 auto Ask the BIOS whether or not to use native PCIe services
3291 associated with PCIe ports (PME, hot-plug, AER). Use
3292 them only if that is allowed by the BIOS.
3293 native Use native PCIe services associated with PCIe ports
3295 compat Treat PCIe ports as PCI-to-PCI bridges, disable the PCIe
3298 pcie_port_pm= [PCIE] PCIe port power management handling:
3299 off Disable power management of all PCIe ports
3300 force Forcibly enable power management of all PCIe ports
3302 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
3303 nomsi Do not use MSI for native PCIe PME signaling (this makes
3304 all PCIe root ports use INTx for all services).
3306 pcmv= [HW,PCMCIA] BadgePAD 4
3310 Keep all power-domains already enabled by bootloader on,
3311 even if no driver has claimed them. This is useful
3312 for debug and development, but should not be
3313 needed on a platform with proper driver support.
3316 See Documentation/blockdev/paride.txt.
3318 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
3321 See arch/parisc/kernel/pdc_chassis.c
3323 percpu_alloc= Select which percpu first chunk allocator to use.
3324 Currently supported values are "embed" and "page".
3325 Archs may support subset or none of the selections.
3326 See comments in mm/percpu.c for details on each
3327 allocator. This parameter is primarily for debugging
3328 and performance comparison.
3331 See Documentation/blockdev/paride.txt.
3334 See Documentation/blockdev/paride.txt.
3336 pirq= [SMP,APIC] Manual mp-table setup
3337 See Documentation/x86/i386/IO-APIC.txt.
3339 plip= [PPT,NET] Parallel port network link
3340 Format: { parport<nr> | timid | 0 }
3341 See also Documentation/parport.txt.
3343 pmtmr= [X86] Manual setup of pmtmr I/O Port.
3344 Override pmtimer IOPort with a hex value.
3348 Enable PNP debug messages (depends on the
3349 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
3350 via /sys/module/pnp/parameters/debug. We always show
3351 current resource usage; turning this on also shows
3352 possible settings and some assignment information.
3358 { on | off | curr | res | no-curr | no-res }
3361 [ISAPNP] Exclude IRQs for the autoconfiguration
3364 [ISAPNP] Exclude DMAs for the autoconfiguration
3366 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
3367 Ranges are in pairs (I/O port base and size).
3370 [ISAPNP] Exclude memory regions for the
3372 Ranges are in pairs (memory base and size).
3374 ports= [IP_VS_FTP] IPVS ftp helper module
3376 Up to 8 (IP_VS_APP_MAX_PORTS) ports
3378 Format: <port>,<port>....
3380 powersave=off [PPC] This option disables power saving features.
3381 It specifically disables cpuidle and sets the
3382 platform machine description specific power_save
3383 function to NULL. On Idle the CPU just reduces
3386 ppc_strict_facility_enable
3387 [PPC] This option catches any kernel floating point,
3388 Altivec, VSX and SPE outside of regions specifically
3389 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
3390 There is some performance impact when enabling this.
3392 print-fatal-signals=
3393 [KNL] debug: print fatal signals
3395 If enabled, warn about various signal handling
3396 related application anomalies: too many signals,
3397 too many POSIX.1 timers, fatal signals causing a
3400 If you hit the warning due to signal overflow,
3401 you might want to try "ulimit -i unlimited".
3405 printk.always_kmsg_dump=
3406 Trigger kmsg_dump for cases other than kernel oops or
3408 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
3411 printk.devkmsg={on,off,ratelimit}
3412 Control writing to /dev/kmsg.
3413 on - unlimited logging to /dev/kmsg from userspace
3414 off - logging to /dev/kmsg disabled
3415 ratelimit - ratelimit the logging
3418 printk.time= Show timing data prefixed to each printk message line
3419 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
3421 processor.max_cstate= [HW,ACPI]
3422 Limit processor to maximum C-state
3423 max_cstate=9 overrides any DMI blacklist limit.
3425 processor.nocst [HW,ACPI]
3426 Ignore the _CST method to determine C-states,
3427 instead using the legacy FADT method
3429 profile= [KNL] Enable kernel profiling via /proc/profile
3430 Format: [schedule,]<number>
3431 Param: "schedule" - profile schedule points.
3432 Param: <number> - step/bucket size as a power of 2 for
3433 statistical time based profiling.
3434 Param: "sleep" - profile D-state sleeping (millisecs).
3435 Requires CONFIG_SCHEDSTATS
3436 Param: "kvm" - profile VM exits.
3438 prompt_ramdisk= [RAM] List of RAM disks to prompt for floppy disk
3440 See Documentation/blockdev/ramdisk.txt.
3442 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
3443 probe for; one of (bare|imps|exps|lifebook|any).
3444 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
3446 psmouse.resetafter= [HW,MOUSE]
3447 Try to reset the device after so many bad packets
3450 [HW,MOUSE] Set desired mouse resolution, in dpi.
3451 psmouse.smartscroll=
3452 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
3453 0 = disabled, 1 = enabled (default).
3455 pstore.backend= Specify the name of the pstore backend to use
3458 See Documentation/blockdev/paride.txt.
3460 pti= [X86_64] Control Page Table Isolation of user and
3461 kernel address spaces. Disabling this feature
3462 removes hardening, but improves performance of
3463 system calls and interrupts.
3465 on - unconditionally enable
3466 off - unconditionally disable
3467 auto - kernel detects whether your CPU model is
3468 vulnerable to issues that PTI mitigates
3470 Not specifying this option is equivalent to pti=auto.
3473 Equivalent to pti=off
3476 [KNL] Number of legacy pty's. Overwrites compiled-in
3479 quiet [KNL] Disable most log messages
3484 See Documentation/admin-guide/md.rst.
3486 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
3487 See Documentation/blockdev/ramdisk.txt.
3489 ras=option[,option,...] [KNL] RAS-specific options
3492 Disable the Correctable Errors Collector,
3493 see CONFIG_RAS_CEC help text.
3496 The argument is a cpu list, as described above.
3498 In kernels built with CONFIG_RCU_NOCB_CPU=y, set
3499 the specified list of CPUs to be no-callback CPUs.
3500 Invocation of these CPUs' RCU callbacks will
3501 be offloaded to "rcuox/N" kthreads created for
3502 that purpose, where "x" is "b" for RCU-bh, "p"
3503 for RCU-preempt, and "s" for RCU-sched, and "N"
3504 is the CPU number. This reduces OS jitter on the
3505 offloaded CPUs, which can be useful for HPC and
3506 real-time workloads. It can also improve energy
3507 efficiency for asymmetric multiprocessors.
3510 Rather than requiring that offloaded CPUs
3511 (specified by rcu_nocbs= above) explicitly
3512 awaken the corresponding "rcuoN" kthreads,
3513 make these kthreads poll for callbacks.
3514 This improves the real-time response for the
3515 offloaded CPUs by relieving them of the need to
3516 wake up the corresponding kthread, but degrades
3517 energy efficiency by requiring that the kthreads
3518 periodically wake up to do the polling.
3520 rcutree.blimit= [KNL]
3521 Set maximum number of finished RCU callbacks to
3522 process in one batch.
3524 rcutree.dump_tree= [KNL]
3525 Dump the structure of the rcu_node combining tree
3526 out at early boot. This is used for diagnostic
3527 purposes, to verify correct tree setup.
3529 rcutree.gp_cleanup_delay= [KNL]
3530 Set the number of jiffies to delay each step of
3531 RCU grace-period cleanup.
3533 rcutree.gp_init_delay= [KNL]
3534 Set the number of jiffies to delay each step of
3535 RCU grace-period initialization.
3537 rcutree.gp_preinit_delay= [KNL]
3538 Set the number of jiffies to delay each step of
3539 RCU grace-period pre-initialization, that is,
3540 the propagation of recent CPU-hotplug changes up
3541 the rcu_node combining tree.
3543 rcutree.rcu_fanout_exact= [KNL]
3544 Disable autobalancing of the rcu_node combining
3545 tree. This is used by rcutorture, and might
3546 possibly be useful for architectures having high
3547 cache-to-cache transfer latencies.
3549 rcutree.rcu_fanout_leaf= [KNL]
3550 Change the number of CPUs assigned to each
3551 leaf rcu_node structure. Useful for very
3552 large systems, which will choose the value 64,
3553 and for NUMA systems with large remote-access
3554 latencies, which will choose a value aligned
3555 with the appropriate hardware boundaries.
3557 rcutree.jiffies_till_sched_qs= [KNL]
3558 Set required age in jiffies for a
3559 given grace period before RCU starts
3560 soliciting quiescent-state help from
3561 rcu_note_context_switch().
3563 rcutree.jiffies_till_first_fqs= [KNL]
3564 Set delay from grace-period initialization to
3565 first attempt to force quiescent states.
3566 Units are jiffies, minimum value is zero,
3567 and maximum value is HZ.
3569 rcutree.jiffies_till_next_fqs= [KNL]
3570 Set delay between subsequent attempts to force
3571 quiescent states. Units are jiffies, minimum
3572 value is one, and maximum value is HZ.
3574 rcutree.kthread_prio= [KNL,BOOT]
3575 Set the SCHED_FIFO priority of the RCU per-CPU
3576 kthreads (rcuc/N). This value is also used for
3577 the priority of the RCU boost threads (rcub/N)
3578 and for the RCU grace-period kthreads (rcu_bh,
3579 rcu_preempt, and rcu_sched). If RCU_BOOST is
3580 set, valid values are 1-99 and the default is 1
3581 (the least-favored priority). Otherwise, when
3582 RCU_BOOST is not set, valid values are 0-99 and
3583 the default is zero (non-realtime operation).
3585 rcutree.rcu_nocb_leader_stride= [KNL]
3586 Set the number of NOCB kthread groups, which
3587 defaults to the square root of the number of
3588 CPUs. Larger numbers reduces the wakeup overhead
3589 on the per-CPU grace-period kthreads, but increases
3590 that same overhead on each group's leader.
3592 rcutree.qhimark= [KNL]
3593 Set threshold of queued RCU callbacks beyond which
3594 batch limiting is disabled.
3596 rcutree.qlowmark= [KNL]
3597 Set threshold of queued RCU callbacks below which
3598 batch limiting is re-enabled.
3600 rcutree.rcu_idle_gp_delay= [KNL]
3601 Set wakeup interval for idle CPUs that have
3602 RCU callbacks (RCU_FAST_NO_HZ=y).
3604 rcutree.rcu_idle_lazy_gp_delay= [KNL]
3605 Set wakeup interval for idle CPUs that have
3606 only "lazy" RCU callbacks (RCU_FAST_NO_HZ=y).
3607 Lazy RCU callbacks are those which RCU can
3608 prove do nothing more than free memory.
3610 rcutree.rcu_kick_kthreads= [KNL]
3611 Cause the grace-period kthread to get an extra
3612 wake_up() if it sleeps three times longer than
3613 it should at force-quiescent-state time.
3614 This wake_up() will be accompanied by a
3615 WARN_ONCE() splat and an ftrace_dump().
3617 rcuperf.gp_async= [KNL]
3618 Measure performance of asynchronous
3619 grace-period primitives such as call_rcu().
3621 rcuperf.gp_async_max= [KNL]
3622 Specify the maximum number of outstanding
3623 callbacks per writer thread. When a writer
3624 thread exceeds this limit, it invokes the
3625 corresponding flavor of rcu_barrier() to allow
3626 previously posted callbacks to drain.
3628 rcuperf.gp_exp= [KNL]
3629 Measure performance of expedited synchronous
3630 grace-period primitives.
3632 rcuperf.holdoff= [KNL]
3633 Set test-start holdoff period. The purpose of
3634 this parameter is to delay the start of the
3635 test until boot completes in order to avoid
3638 rcuperf.nreaders= [KNL]
3639 Set number of RCU readers. The value -1 selects
3640 N, where N is the number of CPUs. A value
3641 "n" less than -1 selects N-n+1, where N is again
3642 the number of CPUs. For example, -2 selects N
3643 (the number of CPUs), -3 selects N+1, and so on.
3644 A value of "n" less than or equal to -N selects
3647 rcuperf.nwriters= [KNL]
3648 Set number of RCU writers. The values operate
3649 the same as for rcuperf.nreaders.
3650 N, where N is the number of CPUs
3652 rcuperf.perf_runnable= [BOOT]
3653 Start rcuperf running at boot time.
3655 rcuperf.perf_type= [KNL]
3656 Specify the RCU implementation to test.
3658 rcuperf.shutdown= [KNL]
3659 Shut the system down after performance tests
3660 complete. This is useful for hands-off automated
3663 rcuperf.verbose= [KNL]
3664 Enable additional printk() statements.
3666 rcuperf.writer_holdoff= [KNL]
3667 Write-side holdoff between grace periods,
3668 in microseconds. The default of zero says
3671 rcutorture.cbflood_inter_holdoff= [KNL]
3672 Set holdoff time (jiffies) between successive
3673 callback-flood tests.
3675 rcutorture.cbflood_intra_holdoff= [KNL]
3676 Set holdoff time (jiffies) between successive
3677 bursts of callbacks within a given callback-flood
3680 rcutorture.cbflood_n_burst= [KNL]
3681 Set the number of bursts making up a given
3682 callback-flood test. Set this to zero to
3683 disable callback-flood testing.
3685 rcutorture.cbflood_n_per_burst= [KNL]
3686 Set the number of callbacks to be registered
3687 in a given burst of a callback-flood test.
3689 rcutorture.fqs_duration= [KNL]
3690 Set duration of force_quiescent_state bursts
3693 rcutorture.fqs_holdoff= [KNL]
3694 Set holdoff time within force_quiescent_state bursts
3697 rcutorture.fqs_stutter= [KNL]
3698 Set wait time between force_quiescent_state bursts
3701 rcutorture.gp_cond= [KNL]
3702 Use conditional/asynchronous update-side
3703 primitives, if available.
3705 rcutorture.gp_exp= [KNL]
3706 Use expedited update-side primitives, if available.
3708 rcutorture.gp_normal= [KNL]
3709 Use normal (non-expedited) asynchronous
3710 update-side primitives, if available.
3712 rcutorture.gp_sync= [KNL]
3713 Use normal (non-expedited) synchronous
3714 update-side primitives, if available. If all
3715 of rcutorture.gp_cond=, rcutorture.gp_exp=,
3716 rcutorture.gp_normal=, and rcutorture.gp_sync=
3717 are zero, rcutorture acts as if is interpreted
3718 they are all non-zero.
3720 rcutorture.n_barrier_cbs= [KNL]
3721 Set callbacks/threads for rcu_barrier() testing.
3723 rcutorture.nfakewriters= [KNL]
3724 Set number of concurrent RCU writers. These just
3725 stress RCU, they don't participate in the actual
3726 test, hence the "fake".
3728 rcutorture.nreaders= [KNL]
3729 Set number of RCU readers. The value -1 selects
3730 N-1, where N is the number of CPUs. A value
3731 "n" less than -1 selects N-n-2, where N is again
3732 the number of CPUs. For example, -2 selects N
3733 (the number of CPUs), -3 selects N+1, and so on.
3735 rcutorture.object_debug= [KNL]
3736 Enable debug-object double-call_rcu() testing.
3738 rcutorture.onoff_holdoff= [KNL]
3739 Set time (s) after boot for CPU-hotplug testing.
3741 rcutorture.onoff_interval= [KNL]
3742 Set time (s) between CPU-hotplug operations, or
3743 zero to disable CPU-hotplug testing.
3745 rcutorture.shuffle_interval= [KNL]
3746 Set task-shuffle interval (s). Shuffling tasks
3747 allows some CPUs to go into dyntick-idle mode
3748 during the rcutorture test.
3750 rcutorture.shutdown_secs= [KNL]
3751 Set time (s) after boot system shutdown. This
3752 is useful for hands-off automated testing.
3754 rcutorture.stall_cpu= [KNL]
3755 Duration of CPU stall (s) to test RCU CPU stall
3756 warnings, zero to disable.
3758 rcutorture.stall_cpu_holdoff= [KNL]
3759 Time to wait (s) after boot before inducing stall.
3761 rcutorture.stat_interval= [KNL]
3762 Time (s) between statistics printk()s.
3764 rcutorture.stutter= [KNL]
3765 Time (s) to stutter testing, for example, specifying
3766 five seconds causes the test to run for five seconds,
3767 wait for five seconds, and so on. This tests RCU's
3768 ability to transition abruptly to and from idle.
3770 rcutorture.test_boost= [KNL]
3771 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
3772 "Maybe" means test if the RCU implementation
3773 under test support RCU priority boosting.
3775 rcutorture.test_boost_duration= [KNL]
3776 Duration (s) of each individual boost test.
3778 rcutorture.test_boost_interval= [KNL]
3779 Interval (s) between each boost test.
3781 rcutorture.test_no_idle_hz= [KNL]
3782 Test RCU's dyntick-idle handling. See also the
3783 rcutorture.shuffle_interval parameter.
3785 rcutorture.torture_runnable= [BOOT]
3786 Start rcutorture running at boot time.
3788 rcutorture.torture_type= [KNL]
3789 Specify the RCU implementation to test.
3791 rcutorture.verbose= [KNL]
3792 Enable additional printk() statements.
3794 rcupdate.rcu_cpu_stall_suppress= [KNL]
3795 Suppress RCU CPU stall warning messages.
3797 rcupdate.rcu_cpu_stall_timeout= [KNL]
3798 Set timeout for RCU CPU stall warning messages.
3800 rcupdate.rcu_expedited= [KNL]
3801 Use expedited grace-period primitives, for
3802 example, synchronize_rcu_expedited() instead
3803 of synchronize_rcu(). This reduces latency,
3804 but can increase CPU utilization, degrade
3805 real-time latency, and degrade energy efficiency.
3806 No effect on CONFIG_TINY_RCU kernels.
3808 rcupdate.rcu_normal= [KNL]
3809 Use only normal grace-period primitives,
3810 for example, synchronize_rcu() instead of
3811 synchronize_rcu_expedited(). This improves
3812 real-time latency, CPU utilization, and
3813 energy efficiency, but can expose users to
3814 increased grace-period latency. This parameter
3815 overrides rcupdate.rcu_expedited. No effect on
3816 CONFIG_TINY_RCU kernels.
3818 rcupdate.rcu_normal_after_boot= [KNL]
3819 Once boot has completed (that is, after
3820 rcu_end_inkernel_boot() has been invoked), use
3821 only normal grace-period primitives. No effect
3822 on CONFIG_TINY_RCU kernels.
3824 rcupdate.rcu_task_stall_timeout= [KNL]
3825 Set timeout in jiffies for RCU task stall warning
3826 messages. Disable with a value less than or equal
3829 rcupdate.rcu_self_test= [KNL]
3830 Run the RCU early boot self tests
3832 rcupdate.rcu_self_test_bh= [KNL]
3833 Run the RCU bh early boot self tests
3835 rcupdate.rcu_self_test_sched= [KNL]
3836 Run the RCU sched early boot self tests
3840 Run specified binary instead of /init from the ramdisk,
3841 used for early userspace startup. See initrd.
3844 force - Override the decision by the kernel to hide the
3845 advertisement of RDRAND support (this affects
3846 certain AMD processors because of buggy BIOS
3847 support, specifically around the suspend/resume
3851 Turn on/off individual RDT features. List is:
3852 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, mba.
3853 E.g. to turn on cmt and turn off mba use:
3857 Format (x86 or x86_64):
3858 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] \
3860 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
3862 Where reboot_mode is one of warm (soft) or cold (hard) or gpio,
3863 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
3864 reboot_force is either force or not specified,
3865 reboot_cpu is s[mp]#### with #### being the processor
3866 to be used for rebooting.
3869 [KNL, SMP] Set scheduler's default relax_domain_level.
3870 See Documentation/cgroup-v1/cpusets.txt.
3872 reserve= [KNL,BUGS] Force the kernel to ignore some iomem area
3874 reservetop= [X86-32]
3876 Reserves a hole at the top of the kernel virtual
3881 Set the amount of memory to reserve for BIOS at
3882 the bottom of the address space.
3884 reset_devices [KNL] Force drivers to reset the underlying device
3885 during initialization.
3888 Specify the partition device for software suspend
3890 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
3892 resume_offset= [SWSUSP]
3893 Specify the offset from the beginning of the partition
3894 given by "resume=" at which the swap header is located,
3895 in <PAGE_SIZE> units (needed only for swap files).
3896 See Documentation/power/swsusp-and-swap-files.txt
3898 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
3899 read the resume files
3901 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
3902 Useful for devices that are detected asynchronously
3903 (e.g. USB and MMC devices).
3905 hibernate= [HIBERNATION]
3906 noresume Don't check if there's a hibernation image
3907 present during boot.
3908 nocompress Don't compress/decompress hibernation images.
3909 no Disable hibernation and resume.
3910 protect_image Turn on image protection during restoration
3911 (that will set all pages holding image data
3912 during restoration read-only).
3914 retain_initrd [RAM] Keep initrd memory after extraction
3916 rfkill.default_state=
3917 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
3918 etc. communication is blocked by default.
3921 rfkill.master_switch_mode=
3922 0 The "airplane mode" button does nothing.
3923 1 The "airplane mode" button toggles between everything
3924 blocked and the previous configuration.
3925 2 The "airplane mode" button toggles between everything
3926 blocked and everything unblocked.
3928 rhash_entries= [KNL,NET]
3929 Set number of hash buckets for route cache
3932 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
3935 ro [KNL] Mount root device read-only on boot
3938 on Mark read-only kernel memory as read-only (default).
3939 off Leave read-only kernel memory writable for debugging.
3942 Enable the uart passthrough on the designated usb port
3943 on Rockchip SoCs. When active, the signals of the
3944 debug-uart get routed to the D+ and D- pins of the usb
3945 port and the regular usb controller gets disabled.
3947 root= [KNL] Root filesystem
3948 See name_to_dev_t comment in init/do_mounts.c.
3950 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
3951 mount the root filesystem
3953 rootflags= [KNL] Set root filesystem mount option string
3955 rootfstype= [KNL] Set root filesystem type
3957 rootwait [KNL] Wait (indefinitely) for root device to show up.
3958 Useful for devices that are detected asynchronously
3959 (e.g. USB and MMC devices).
3961 rproc_mem=nn[KMG][@address]
3962 [KNL,ARM,CMA] Remoteproc physical memory block.
3963 Memory area to be used by remote processor image,
3966 rw [KNL] Mount root device read-write on boot
3968 S [KNL] Run init in single mode
3970 s390_iommu= [HW,S390]
3971 Set s390 IOTLB flushing mode
3973 With strict flushing every unmap operation will result in
3974 an IOTLB flush. Default is lazy flushing before reuse,
3978 See drivers/net/irda/sa1100_ir.c.
3980 sbni= [NET] Granch SBNI12 leased line adapter
3982 sched_debug [KNL] Enables verbose scheduler debug messages.
3984 schedstats= [KNL,X86] Enable or disable scheduled statistics.
3985 Allowed values are enable and disable. This feature
3986 incurs a small amount of overhead in the scheduler
3987 but is useful for debugging and performance tuning.
3989 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
3990 xtime_lock contention on larger systems, and/or RCU lock
3991 contention on all systems with CONFIG_MAXSMP set.
3992 Format: { "0" | "1" }
3993 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
3995 Note: increases power consumption, thus should only be
3996 enabled if running jitter sensitive (HPC/RT) workloads.
3998 security= [SECURITY] Choose a security module to enable at boot.
3999 If this boot parameter is not specified, only the first
4000 security module asking for security registration will be
4001 loaded. An invalid security module name will be treated
4002 as if no module has been chosen.
4004 selinux= [SELINUX] Disable or enable SELinux at boot time.
4005 Format: { "0" | "1" }
4006 See security/selinux/Kconfig help text.
4009 Default value is set via kernel config option.
4010 If enabled at boot time, /selinux/disable can be used
4011 later to disable prior to initial policy load.
4013 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
4014 Format: { "0" | "1" }
4015 See security/apparmor/Kconfig help text
4018 Default value is set via kernel config option.
4020 serialnumber [BUGS=X86-32]
4023 Maximal number of shapers.
4031 Disable merging of slabs with similar size. May be
4032 necessary if there is some reason to distinguish
4033 allocs to different slabs, especially in hardened
4034 environments where the risk of heap overflows and
4035 layout control by attackers can usually be
4036 frustrated by disabling merging. This will reduce
4037 most of the exposure of a heap attack to a single
4038 cache (risks via metadata attacks are mostly
4039 unchanged). Debug options disable merging on their
4041 For more information see Documentation/vm/slub.txt.
4043 slab_max_order= [MM, SLAB]
4044 Determines the maximum allowed order for slabs.
4045 A high setting may cause OOMs due to memory
4046 fragmentation. Defaults to 1 for systems with
4047 more than 32MB of RAM, 0 otherwise.
4049 slub_debug[=options[,slabs]] [MM, SLUB]
4050 Enabling slub_debug allows one to determine the
4051 culprit if slab objects become corrupted. Enabling
4052 slub_debug can create guard zones around objects and
4053 may poison objects when not in use. Also tracks the
4054 last alloc / free. For more information see
4055 Documentation/vm/slub.txt.
4057 slub_memcg_sysfs= [MM, SLUB]
4058 Determines whether to enable sysfs directories for
4059 memory cgroup sub-caches. 1 to enable, 0 to disable.
4060 The default is determined by CONFIG_SLUB_MEMCG_SYSFS_ON.
4061 Enabling this can lead to a very high number of debug
4062 directories and files being created under
4065 slub_max_order= [MM, SLUB]
4066 Determines the maximum allowed order for slabs.
4067 A high setting may cause OOMs due to memory
4068 fragmentation. For more information see
4069 Documentation/vm/slub.txt.
4071 slub_min_objects= [MM, SLUB]
4072 The minimum number of objects per slab. SLUB will
4073 increase the slab order up to slub_max_order to
4074 generate a sufficiently large slab able to contain
4075 the number of objects indicated. The higher the number
4076 of objects the smaller the overhead of tracking slabs
4077 and the less frequently locks need to be acquired.
4078 For more information see Documentation/vm/slub.txt.
4080 slub_min_order= [MM, SLUB]
4081 Determines the minimum page order for slabs. Must be
4082 lower than slub_max_order.
4083 For more information see Documentation/vm/slub.txt.
4085 slub_nomerge [MM, SLUB]
4086 Same with slab_nomerge. This is supported for legacy.
4087 See slab_nomerge for more information.
4090 Format: <io1>[,<io2>[,...,<io8>]]
4092 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
4093 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
4094 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
4095 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
4096 smsc-ircc2.ircc_irq= [HW] IRQ line
4097 smsc-ircc2.ircc_dma= [HW] DMA channel
4098 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
4099 0: Toshiba Satellite 1800 (GP data pin select)
4100 1: Fast pin select (default)
4103 smt [KNL,S390] Set the maximum number of threads (logical
4104 CPUs) to use per physical CPU on systems capable of
4105 symmetric multithreading (SMT). Will be capped to the
4106 actual hardware limit.
4108 Default: -1 (no limit)
4111 [KNL] Should the soft-lockup detector generate panics.
4114 softlockup_all_cpu_backtrace=
4115 [KNL] Should the soft-lockup detector generate
4116 backtraces on all cpus.
4119 sonypi.*= [HW] Sony Programmable I/O Control Device driver
4120 See Documentation/laptops/sonypi.txt
4122 spectre_v2= [X86] Control mitigation of Spectre variant 2
4123 (indirect branch speculation) vulnerability.
4124 The default operation protects the kernel from
4127 on - unconditionally enable, implies
4129 off - unconditionally disable, implies
4131 auto - kernel detects whether your CPU model is
4134 Selecting 'on' will, and 'auto' may, choose a
4135 mitigation method at run time according to the
4136 CPU, the available microcode, the setting of the
4137 CONFIG_RETPOLINE configuration option, and the
4138 compiler with which the kernel was built.
4140 Selecting 'on' will also enable the mitigation
4141 against user space to user space task attacks.
4143 Selecting 'off' will disable both the kernel and
4144 the user space protections.
4146 Specific mitigations can also be selected manually:
4148 retpoline - replace indirect branches
4149 retpoline,generic - google's original retpoline
4150 retpoline,amd - AMD-specific minimal thunk
4152 Not specifying this option is equivalent to
4156 [X86] Control mitigation of Spectre variant 2
4157 (indirect branch speculation) vulnerability between
4160 on - Unconditionally enable mitigations. Is
4161 enforced by spectre_v2=on
4163 off - Unconditionally disable mitigations. Is
4164 enforced by spectre_v2=off
4166 prctl - Indirect branch speculation is enabled,
4167 but mitigation can be enabled via prctl
4168 per thread. The mitigation control state
4169 is inherited on fork.
4172 - Like "prctl" above, but only STIBP is
4173 controlled per thread. IBPB is issued
4174 always when switching between different user
4178 - Same as "prctl" above, but all seccomp
4179 threads will enable the mitigation unless
4180 they explicitly opt out.
4183 - Like "seccomp" above, but only STIBP is
4184 controlled per thread. IBPB is issued
4185 always when switching between different
4186 user space processes.
4188 auto - Kernel selects the mitigation depending on
4189 the available CPU features and vulnerability.
4192 If CONFIG_SECCOMP=y then "seccomp", otherwise "prctl"
4194 Not specifying this option is equivalent to
4195 spectre_v2_user=auto.
4197 spec_store_bypass_disable=
4198 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
4199 (Speculative Store Bypass vulnerability)
4201 Certain CPUs are vulnerable to an exploit against a
4202 a common industry wide performance optimization known
4203 as "Speculative Store Bypass" in which recent stores
4204 to the same memory location may not be observed by
4205 later loads during speculative execution. The idea
4206 is that such stores are unlikely and that they can
4207 be detected prior to instruction retirement at the
4208 end of a particular speculation execution window.
4210 In vulnerable processors, the speculatively forwarded
4211 store can be used in a cache side channel attack, for
4212 example to read memory to which the attacker does not
4213 directly have access (e.g. inside sandboxed code).
4215 This parameter controls whether the Speculative Store
4216 Bypass optimization is used.
4218 on - Unconditionally disable Speculative Store Bypass
4219 off - Unconditionally enable Speculative Store Bypass
4220 auto - Kernel detects whether the CPU model contains an
4221 implementation of Speculative Store Bypass and
4222 picks the most appropriate mitigation. If the
4223 CPU is not vulnerable, "off" is selected. If the
4224 CPU is vulnerable the default mitigation is
4225 architecture and Kconfig dependent. See below.
4226 prctl - Control Speculative Store Bypass per thread
4227 via prctl. Speculative Store Bypass is enabled
4228 for a process by default. The state of the control
4229 is inherited on fork.
4230 seccomp - Same as "prctl" above, but all seccomp threads
4231 will disable SSB unless they explicitly opt out.
4233 Not specifying this option is equivalent to
4234 spec_store_bypass_disable=auto.
4236 Default mitigations:
4237 X86: If CONFIG_SECCOMP=y "seccomp", otherwise "prctl"
4239 spia_io_base= [HW,MTD]
4245 Control the Special Register Buffer Data Sampling
4248 Certain CPUs are vulnerable to an MDS-like
4249 exploit which can leak bits from the random
4252 By default, this issue is mitigated by
4253 microcode. However, the microcode fix can cause
4254 the RDRAND and RDSEED instructions to become
4255 much slower. Among other effects, this will
4256 result in reduced throughput from /dev/urandom.
4258 The microcode mitigation can be disabled with
4259 the following option:
4261 off: Disable mitigation and remove
4262 performance impact to RDRAND and RDSEED
4264 srcutree.counter_wrap_check [KNL]
4265 Specifies how frequently to check for
4266 grace-period sequence counter wrap for the
4267 srcu_data structure's ->srcu_gp_seq_needed field.
4268 The greater the number of bits set in this kernel
4269 parameter, the less frequently counter wrap will
4270 be checked for. Note that the bottom two bits
4273 srcutree.exp_holdoff [KNL]
4274 Specifies how many nanoseconds must elapse
4275 since the end of the last SRCU grace period for
4276 a given srcu_struct until the next normal SRCU
4277 grace period will be considered for automatic
4278 expediting. Set to zero to disable automatic
4282 Speculative Store Bypass Disable control
4284 On CPUs that are vulnerable to the Speculative
4285 Store Bypass vulnerability and offer a
4286 firmware based mitigation, this parameter
4287 indicates how the mitigation should be used:
4289 force-on: Unconditionally enable mitigation for
4290 for both kernel and userspace
4291 force-off: Unconditionally disable mitigation for
4292 for both kernel and userspace
4293 kernel: Always enable mitigation in the
4294 kernel, and offer a prctl interface
4295 to allow userspace to register its
4296 interest in being mitigated too.
4298 stack_guard_gap= [MM]
4299 override the default stack gap protection. The value
4300 is in page units and it defines how many pages prior
4301 to (for stacks growing down) resp. after (for stacks
4302 growing up) the main stack are reserved for no other
4303 mapping. Default value is 256 pages.
4306 Enabled the stack tracer on boot up.
4308 stacktrace_filter=[function-list]
4309 [FTRACE] Limit the functions that the stack tracer
4310 will trace at boot up. function-list is a comma separated
4311 list of functions. This list can be changed at run
4312 time by the stack_trace_filter file in the debugfs
4313 tracing directory. Note, this enables stack tracing
4314 and the stacktrace above is not needed.
4318 Set the STI (builtin display/keyboard on the HP-PARISC
4319 machines) console (graphic card) which should be used
4320 as the initial boot-console.
4321 See also comment in drivers/video/console/sticore.c.
4324 See comment in drivers/video/console/sticore.c.
4327 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
4329 sunrpc.min_resvport=
4330 sunrpc.max_resvport=
4332 SunRPC servers often require that client requests
4333 originate from a privileged port (i.e. a port in the
4334 range 0 < portnr < 1024).
4335 An administrator who wishes to reserve some of these
4336 ports for other uses may adjust the range that the
4337 kernel's sunrpc client considers to be privileged
4338 using these two parameters to set the minimum and
4339 maximum port values.
4341 sunrpc.svc_rpc_per_connection_limit=
4343 Limit the number of requests that the server will
4344 process in parallel from a single connection.
4345 The default value is 0 (no limit).
4349 Control how the NFS server code allocates CPUs to
4350 service thread pools. Depending on how many NICs
4351 you have and where their interrupts are bound, this
4352 option will affect which CPUs will do NFS serving.
4353 Note: this parameter cannot be changed while the
4354 NFS server is running.
4356 auto the server chooses an appropriate mode
4357 automatically using heuristics
4358 global a single global pool contains all CPUs
4359 percpu one pool for each CPU
4360 pernode one pool for each NUMA node (equivalent
4361 to global on non-NUMA machines)
4363 sunrpc.tcp_slot_table_entries=
4364 sunrpc.udp_slot_table_entries=
4366 Sets the upper limit on the number of simultaneous
4367 RPC calls that can be sent from the client to a
4368 server. Increasing these values may allow you to
4369 improve throughput, but will also increase the
4370 amount of memory reserved for use by the client.
4372 suspend.pm_test_delay=
4374 Sets the number of seconds to remain in a suspend test
4375 mode before resuming the system (see
4376 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
4377 is set. Default value is 5.
4380 [KNL] Enable accounting of swap in memory resource
4381 controller if no parameter or 1 is given or disable
4382 it if 0 is given (See Documentation/cgroup-v1/memory.txt)
4384 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
4385 Format: { <int> | force | noforce }
4386 <int> -- Number of I/O TLB slabs
4387 force -- force using of bounce buffers even if they
4388 wouldn't be automatically used by the kernel
4389 noforce -- Never use bounce buffers (for debugging)
4393 sysfs.deprecated=0|1 [KNL]
4394 Enable/disable old style sysfs layout for old udev
4395 on older distributions. When this option is enabled
4396 very new udev will not work anymore. When this option
4397 is disabled (or CONFIG_SYSFS_DEPRECATED not compiled)
4398 in older udev will not work anymore.
4399 Default depends on CONFIG_SYSFS_DEPRECATED_V2 set in
4400 the kernel configuration.
4402 sysrq_always_enabled
4404 Ignore sysrq setting - this boot parameter will
4405 neutralize any effect of /proc/sys/kernel/sysrq.
4406 Useful for debugging.
4408 tcpmhash_entries= [KNL,NET]
4409 Set the number of tcp_metrics_hash slots.
4410 Default value is 8192 or 16384 depending on total
4411 ram pages. This is used to specify the TCP metrics
4412 cache size. See Documentation/networking/ip-sysctl.txt
4413 "tcp_no_metrics_save" section for more details.
4417 test_suspend= [SUSPEND][,N]
4418 Specify "mem" (for Suspend-to-RAM) or "standby" (for
4419 standby suspend) or "freeze" (for suspend type freeze)
4420 as the system sleep state during system startup with
4421 the optional capability to repeat N number of times.
4422 The system is woken from this state using a
4423 wakeup-capable RTC alarm.
4425 thash_entries= [KNL,NET]
4426 Set number of hash buckets for TCP connection
4428 thermal.act= [HW,ACPI]
4429 -1: disable all active trip points in all thermal zones
4430 <degrees C>: override all lowest active trip points
4432 thermal.crt= [HW,ACPI]
4433 -1: disable all critical trip points in all thermal zones
4434 <degrees C>: override all critical trip points
4436 thermal.nocrt= [HW,ACPI]
4437 Set to disable actions on ACPI thermal zone
4438 critical and hot trip points.
4440 thermal.off= [HW,ACPI]
4441 1: disable ACPI thermal control
4443 thermal.psv= [HW,ACPI]
4444 -1: disable all passive trip points
4445 <degrees C>: override all passive trip points to this
4448 thermal.tzp= [HW,ACPI]
4449 Specify global default ACPI thermal zone polling rate
4450 <deci-seconds>: poll all this frequency
4451 0: no polling (default)
4454 Force threading of all interrupt handlers except those
4455 marked explicitly IRQF_NO_THREAD.
4458 Enable the Transcendent memory driver if built-in.
4460 tmem.cleancache=0|1 [KNL, XEN]
4461 Default is on (1). Disable the usage of the cleancache
4462 API to send anonymous pages to the hypervisor.
4464 tmem.frontswap=0|1 [KNL, XEN]
4465 Default is on (1). Disable the usage of the frontswap
4466 API to send swap pages to the hypervisor. If disabled
4467 the selfballooning and selfshrinking are force disabled.
4469 tmem.selfballooning=0|1 [KNL, XEN]
4470 Default is on (1). Disable the driving of swap pages
4473 tmem.selfshrinking=0|1 [KNL, XEN]
4474 Default is on (1). Partial swapoff that immediately
4475 transfers pages from Xen hypervisor back to the
4476 kernel based on different criteria.
4480 Specify if the kernel should make use of the cpu
4481 topology information if the hardware supports this.
4482 The scheduler will make use of this information and
4483 e.g. base its process migration decisions on it.
4486 topology_updates= [KNL, PPC, NUMA]
4488 Specify if the kernel should ignore (off)
4489 topology updates sent by the hypervisor to this
4494 tpm_suspend_pcr=[HW,TPM]
4495 Format: integer pcr id
4496 Specify that at suspend time, the tpm driver
4497 should extend the specified pcr with zeros,
4498 as a workaround for some chips which fail to
4499 flush the last written pcr on TPM_SaveState.
4500 This will guarantee that all the other pcrs
4503 trace_buf_size=nn[KMG]
4504 [FTRACE] will set tracing buffer size on each cpu.
4506 trace_event=[event-list]
4507 [FTRACE] Set and start specified trace events in order
4508 to facilitate early boot debugging. The event-list is a
4509 comma separated list of trace events to enable. See
4510 also Documentation/trace/events.txt
4512 trace_options=[option-list]
4513 [FTRACE] Enable or disable tracer options at boot.
4514 The option-list is a comma delimited list of options
4515 that can be enabled or disabled just as if you were
4516 to echo the option name into
4518 /sys/kernel/debug/tracing/trace_options
4520 For example, to enable stacktrace option (to dump the
4521 stack trace of each event), add to the command line:
4523 trace_options=stacktrace
4525 See also Documentation/trace/ftrace.txt "trace options"
4529 Have the tracepoints sent to printk as well as the
4530 tracing ring buffer. This is useful for early boot up
4531 where the system hangs or reboots and does not give the
4532 option for reading the tracing buffer or performing a
4533 ftrace_dump_on_oops.
4535 To turn off having tracepoints sent to printk,
4536 echo 0 > /proc/sys/kernel/tracepoint_printk
4537 Note, echoing 1 into this file without the
4538 tracepoint_printk kernel cmdline option has no effect.
4542 Having tracepoints sent to printk() and activating high
4543 frequency tracepoints such as irq or sched, can cause
4544 the system to live lock.
4547 [FTRACE] enable this option to disable tracing when a
4548 warning is hit. This turns off "tracing_on". Tracing can
4549 be enabled again by echoing '1' into the "tracing_on"
4550 file located in /sys/kernel/debug/tracing/
4552 This option is useful, as it disables the trace before
4553 the WARNING dump is called, which prevents the trace to
4554 be filled with content caused by the warning output.
4556 This option can also be set at run time via the sysctl
4557 option: kernel/traceoff_on_warning
4559 transparent_hugepage=
4561 Format: [always|madvise|never]
4562 Can be used to control the default behavior of the system
4563 with respect to transparent hugepages.
4564 See Documentation/vm/transhuge.txt for more details.
4566 tsc= Disable clocksource stability checks for TSC.
4568 [x86] reliable: mark tsc clocksource as reliable, this
4569 disables clocksource verification at runtime, as well
4570 as the stability checks done at bootup. Used to enable
4571 high-resolution timer mode on older hardware, and in
4572 virtualized environment.
4573 [x86] noirqtime: Do not use TSC to do irq accounting.
4574 Used to run time disable IRQ_TIME_ACCOUNTING on any
4575 platforms where RDTSC is slow and this accounting
4578 tsx= [X86] Control Transactional Synchronization
4579 Extensions (TSX) feature in Intel processors that
4580 support TSX control.
4582 This parameter controls the TSX feature. The options are:
4584 on - Enable TSX on the system. Although there are
4585 mitigations for all known security vulnerabilities,
4586 TSX has been known to be an accelerator for
4587 several previous speculation-related CVEs, and
4588 so there may be unknown security risks associated
4589 with leaving it enabled.
4591 off - Disable TSX on the system. (Note that this
4592 option takes effect only on newer CPUs which are
4593 not vulnerable to MDS, i.e., have
4594 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
4595 the new IA32_TSX_CTRL MSR through a microcode
4596 update. This new MSR allows for the reliable
4597 deactivation of the TSX functionality.)
4599 auto - Disable TSX if X86_BUG_TAA is present,
4600 otherwise enable TSX on the system.
4602 Not specifying this option is equivalent to tsx=off.
4604 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
4607 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
4608 Abort (TAA) vulnerability.
4610 Similar to Micro-architectural Data Sampling (MDS)
4611 certain CPUs that support Transactional
4612 Synchronization Extensions (TSX) are vulnerable to an
4613 exploit against CPU internal buffers which can forward
4614 information to a disclosure gadget under certain
4617 In vulnerable processors, the speculatively forwarded
4618 data can be used in a cache side channel attack, to
4619 access data to which the attacker does not have direct
4622 This parameter controls the TAA mitigation. The
4625 full - Enable TAA mitigation on vulnerable CPUs
4628 full,nosmt - Enable TAA mitigation and disable SMT on
4629 vulnerable CPUs. If TSX is disabled, SMT
4630 is not disabled because CPU is not
4631 vulnerable to cross-thread TAA attacks.
4632 off - Unconditionally disable TAA mitigation
4634 On MDS-affected machines, tsx_async_abort=off can be
4635 prevented by an active MDS mitigation as both vulnerabilities
4636 are mitigated with the same mechanism so in order to disable
4637 this mitigation, you need to specify mds=off too.
4639 Not specifying this option is equivalent to
4640 tsx_async_abort=full. On CPUs which are MDS affected
4641 and deploy MDS mitigation, TAA mitigation is not
4642 required and doesn't provide any additional
4646 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
4648 turbografx.map[2|3]= [HW,JOY]
4649 TurboGraFX parallel port interface
4651 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
4652 See also Documentation/input/joystick-parport.txt
4654 udbg-immortal [PPC] When debugging early kernel crashes that
4655 happen after console_init() and before a proper
4656 console driver takes over, this boot options might
4657 help "seeing" what's going on.
4659 uhash_entries= [KNL,NET]
4660 Set number of hash buckets for UDP/UDP-Lite connections
4663 [USB] Ignore overcurrent events (default N).
4664 Some badly-designed motherboards generate lots of
4665 bogus events, for ports that aren't wired to
4666 anything. Set this parameter to avoid log spamming.
4667 Note that genuine overcurrent events won't be
4671 [X86] Cause panic on unknown NMI.
4673 usbcore.authorized_default=
4674 [USB] Default USB device authorization:
4675 (default -1 = authorized except for wireless USB,
4676 0 = not authorized, 1 = authorized)
4678 usbcore.autosuspend=
4679 [USB] The autosuspend time delay (in seconds) used
4680 for newly-detected USB devices (default 2). This
4681 is the time required before an idle device will be
4682 autosuspended. Devices for which the delay is set
4683 to a negative value won't be autosuspended at all.
4685 usbcore.usbfs_snoop=
4686 [USB] Set to log all usbfs traffic (default 0 = off).
4688 usbcore.usbfs_snoop_max=
4689 [USB] Maximum number of bytes to snoop in each URB
4692 usbcore.blinkenlights=
4693 [USB] Set to cycle leds on hubs (default 0 = off).
4695 usbcore.old_scheme_first=
4696 [USB] Start with the old device initialization
4697 scheme (default 0 = off).
4699 usbcore.usbfs_memory_mb=
4700 [USB] Memory limit (in MB) for buffers allocated by
4701 usbfs (default = 16, 0 = max = 2047).
4703 usbcore.use_both_schemes=
4704 [USB] Try the other device initialization scheme
4705 if the first one fails (default 1 = enabled).
4707 usbcore.initial_descriptor_timeout=
4708 [USB] Specifies timeout for the initial 64-byte
4709 USB_REQ_GET_DESCRIPTOR request in milliseconds
4710 (default 5000 = 5.0 seconds).
4712 usbcore.nousb [USB] Disable the USB subsystem
4715 [USBHID] The interval which mice are to be polled at.
4718 [USBHID] The interval which joysticks are to be polled at.
4720 usb-storage.delay_use=
4721 [UMS] The delay in seconds before a new device is
4722 scanned for Logical Units (default 1).
4725 [UMS] A list of quirks entries to supplement or
4726 override the built-in unusual_devs list. List
4727 entries are separated by commas. Each entry has
4728 the form VID:PID:Flags where VID and PID are Vendor
4729 and Product ID values (4-digit hex numbers) and
4730 Flags is a set of characters, each corresponding
4731 to a common usb-storage quirk flag as follows:
4732 a = SANE_SENSE (collect more than 18 bytes
4733 of sense data, not on uas);
4734 b = BAD_SENSE (don't collect more than 18
4735 bytes of sense data, not on uas);
4736 c = FIX_CAPACITY (decrease the reported
4737 device capacity by one sector);
4738 d = NO_READ_DISC_INFO (don't use
4739 READ_DISC_INFO command, not on uas);
4740 e = NO_READ_CAPACITY_16 (don't use
4741 READ_CAPACITY_16 command);
4742 f = NO_REPORT_OPCODES (don't use report opcodes
4744 g = MAX_SECTORS_240 (don't transfer more than
4745 240 sectors at a time, uas only);
4746 h = CAPACITY_HEURISTICS (decrease the
4747 reported device capacity by one
4748 sector if the number is odd);
4749 i = IGNORE_DEVICE (don't bind to this
4751 j = NO_REPORT_LUNS (don't use report luns
4753 k = NO_SAME (do not use WRITE_SAME, uas only)
4754 l = NOT_LOCKABLE (don't try to lock and
4755 unlock ejectable media, not on uas);
4756 m = MAX_SECTORS_64 (don't transfer more
4757 than 64 sectors = 32 KB at a time,
4759 n = INITIAL_READ10 (force a retry of the
4760 initial READ(10) command, not on uas);
4761 o = CAPACITY_OK (accept the capacity
4762 reported by the device, not on uas);
4763 p = WRITE_CACHE (the device cache is ON
4764 by default, not on uas);
4765 r = IGNORE_RESIDUE (the device reports
4766 bogus residue values, not on uas);
4767 s = SINGLE_LUN (the device has only one
4769 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
4770 commands, uas only);
4771 u = IGNORE_UAS (don't bind to the uas driver);
4772 w = NO_WP_DETECT (don't test whether the
4773 medium is write-protected).
4774 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
4775 even if the device claims no cache,
4777 Example: quirks=0419:aaf5:rl,0421:0433:rc
4779 user_debug= [KNL,ARM]
4781 See arch/arm/Kconfig.debug help text.
4782 1 - undefined instruction events
4784 4 - invalid data aborts
4787 Example: user_debug=31
4790 [X86] Flags controlling user PTE allocations.
4792 nohigh = do not allocate PTE pages in
4793 HIGHMEM regardless of setting
4797 On X86_32, this is an alias for vdso32=. Otherwise:
4799 vdso=1: enable VDSO (the default)
4800 vdso=0: disable VDSO mapping
4802 vdso32= [X86] Control the 32-bit vDSO
4803 vdso32=1: enable 32-bit VDSO
4804 vdso32=0 or vdso32=2: disable 32-bit VDSO
4806 See the help text for CONFIG_COMPAT_VDSO for more
4807 details. If CONFIG_COMPAT_VDSO is set, the default is
4808 vdso32=0; otherwise, the default is vdso32=1.
4810 For compatibility with older kernels, vdso32=2 is an
4813 Try vdso32=0 if you encounter an error that says:
4814 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
4817 vector=percpu: enable percpu vector domain
4819 video= [FB] Frame buffer configuration
4820 See Documentation/fb/modedb.txt.
4822 video.brightness_switch_enabled= [0,1]
4823 If set to 1, on receiving an ACPI notify event
4824 generated by hotkey, video driver will adjust brightness
4825 level and then send out the event to user space through
4826 the allocated input device; If set to 0, video driver
4827 will only send out the event without touching backlight
4832 [VMMIO] Memory mapped virtio (platform) device.
4834 <size>@<baseaddr>:<irq>[:<id>]
4836 <size> := size (can use standard suffixes
4838 <baseaddr> := physical base address
4839 <irq> := interrupt number (as passed to
4841 <id> := (optional) platform device id
4843 virtio_mmio.device=1K@0x100b0000:48:7
4845 Can be used multiple times for multiple devices.
4847 vga= [BOOT,X86-32] Select a particular video mode
4848 See Documentation/x86/boot.txt and
4849 Documentation/svga.txt.
4850 Use vga=ask for menu.
4851 This is actually a boot loader parameter; the value is
4852 passed to the kernel using a special protocol.
4854 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
4855 size of <nn>. This can be used to increase the
4856 minimum size (128MB on x86). It can also be used to
4857 decrease the size and leave more room for directly
4860 vmcp_cma=nn[MG] [KNL,S390]
4861 Sets the memory size reserved for contiguous memory
4862 allocations for the vmcp device driver.
4864 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
4867 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
4870 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
4874 Controls the behavior of vsyscalls (i.e. calls to
4875 fixed addresses of 0xffffffffff600x00 from legacy
4876 code). Most statically-linked binaries and older
4877 versions of glibc use these calls. Because these
4878 functions are at fixed addresses, they make nice
4879 targets for exploits that can control RIP.
4881 emulate [default] Vsyscalls turn into traps and are
4882 emulated reasonably safely.
4884 native Vsyscalls are native syscall instructions.
4885 This is a little bit faster than trapping
4886 and makes a few dynamic recompilers work
4887 better than they would in emulation mode.
4888 It also makes exploits much easier to write.
4890 none Vsyscalls don't work at all. This makes
4891 them quite hard to use for exploits but
4892 might break your system.
4894 vt.color= [VT] Default text color.
4895 Format: 0xYX, X = foreground, Y = background.
4896 Default: 0x07 = light gray on black.
4898 vt.cur_default= [VT] Default cursor shape.
4899 Format: 0xCCBBAA, where AA, BB, and CC are the same as
4900 the parameters of the <Esc>[?A;B;Cc escape sequence;
4901 see VGA-softcursor.txt. Default: 2 = underline.
4903 vt.default_blu= [VT]
4904 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
4905 Change the default blue palette of the console.
4906 This is a 16-member array composed of values
4909 vt.default_grn= [VT]
4910 Format: <green0>,<green1>,<green2>,...,<green15>
4911 Change the default green palette of the console.
4912 This is a 16-member array composed of values
4915 vt.default_red= [VT]
4916 Format: <red0>,<red1>,<red2>,...,<red15>
4917 Change the default red palette of the console.
4918 This is a 16-member array composed of values
4924 Set system-wide default UTF-8 mode for all tty's.
4925 Default is 1, i.e. UTF-8 mode is enabled for all
4926 newly opened terminals.
4928 vt.global_cursor_default=
4931 Set system-wide default for whether a cursor
4932 is shown on new VTs. Default is -1,
4933 i.e. cursors will be created by default unless
4934 overridden by individual drivers. 0 will hide
4935 cursors, 1 will display them.
4937 vt.italic= [VT] Default color for italic text; 0-15.
4940 vt.underline= [VT] Default color for underlined text; 0-15.
4943 watchdog timers [HW,WDT] For information on watchdog timers,
4944 see Documentation/watchdog/watchdog-parameters.txt
4945 or other driver-specific files in the
4946 Documentation/watchdog/ directory.
4948 workqueue.watchdog_thresh=
4949 If CONFIG_WQ_WATCHDOG is configured, workqueue can
4950 warn stall conditions and dump internal state to
4951 help debugging. 0 disables workqueue stall
4952 detection; otherwise, it's the stall threshold
4953 duration in seconds. The default value is 30 and
4954 it can be updated at runtime by writing to the
4955 corresponding sysfs file.
4957 workqueue.disable_numa
4958 By default, all work items queued to unbound
4959 workqueues are affine to the NUMA nodes they're
4960 issued on, which results in better behavior in
4961 general. If NUMA affinity needs to be disabled for
4962 whatever reason, this option can be used. Note
4963 that this also can be controlled per-workqueue for
4964 workqueues visible under /sys/bus/workqueue/.
4966 workqueue.power_efficient
4967 Per-cpu workqueues are generally preferred because
4968 they show better performance thanks to cache
4969 locality; unfortunately, per-cpu workqueues tend to
4970 be more power hungry than unbound workqueues.
4972 Enabling this makes the per-cpu workqueues which
4973 were observed to contribute significantly to power
4974 consumption unbound, leading to measurably lower
4975 power usage at the cost of small performance
4978 The default value of this parameter is determined by
4979 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
4981 workqueue.debug_force_rr_cpu
4982 Workqueue used to implicitly guarantee that work
4983 items queued without explicit CPU specified are put
4984 on the local CPU. This guarantee is no longer true
4985 and while local CPU is still preferred work items
4986 may be put on foreign CPUs. This debug option
4987 forces round-robin CPU selection to flush out
4988 usages which depend on the now broken guarantee.
4989 When enabled, memory and cache locality will be
4992 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
4993 default x2apic cluster mode on platforms
4996 x86_intel_mid_timer= [X86-32,APBT]
4997 Choose timer option for x86 Intel MID platform.
4998 Two valid options are apbt timer only and lapic timer
4999 plus one apbt timer for broadcast timer.
5000 x86_intel_mid_timer=apbt_only | lapic_and_apbt
5002 xen_512gb_limit [KNL,X86-64,XEN]
5003 Restricts the kernel running paravirtualized under Xen
5004 to use only up to 512 GB of RAM. The reason to do so is
5005 crash analysis tools and Xen tools for doing domain
5006 save/restore/migration must be enabled to handle larger
5009 xen_emul_unplug= [HW,X86,XEN]
5010 Unplug Xen emulated devices
5011 Format: [unplug0,][unplug1]
5012 ide-disks -- unplug primary master IDE devices
5013 aux-ide-disks -- unplug non-primary-master IDE devices
5014 nics -- unplug network devices
5015 all -- unplug all emulated devices (NICs and IDE disks)
5016 unnecessary -- unplugging emulated devices is
5017 unnecessary even if the host did not respond to
5019 never -- do not unplug even if version check succeeds
5021 xen_legacy_crash [X86,XEN]
5022 Crash from Xen panic notifier, without executing late
5023 panic() code such as dumping handler.
5025 xen_nopvspin [X86,XEN]
5026 Disables the ticketlock slowpath using Xen PV
5030 Disables the PV optimizations forcing the HVM guest to
5031 run as generic HVM guest with no PV drivers.
5033 xen.balloon_boot_timeout= [XEN]
5034 The time (in seconds) to wait before giving up to boot
5035 in case initial ballooning fails to free enough memory.
5036 Applies only when running as HVM or PVH guest and
5037 started with less memory configured than allowed at
5038 max. Default is 180.
5040 xen.event_eoi_delay= [XEN]
5041 How long to delay EOI handling in case of event
5042 storms (jiffies). Default is 10.
5044 xen.event_loop_timeout= [XEN]
5045 After which time (jiffies) the event handling loop
5046 should start to delay EOI handling. Default is 2.
5048 xirc2ps_cs= [NET,PCMCIA]
5050 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]