4 The following is a consolidated list of the kernel parameters as
5 implemented by the __setup(), core_param() and module_param() macros
6 and sorted into English Dictionary order (defined as ignoring all
7 punctuation and sorting digits before letters in a case insensitive
8 manner), and with descriptions where known.
10 The kernel parses parameters from the kernel command line up to "--";
11 if it doesn't recognize a parameter and it doesn't contain a '.', the
12 parameter gets passed to init: parameters with '=' go into init's
13 environment, others are passed as command line arguments to init.
14 Everything after "--" is passed as an argument to init.
16 Module parameters can be specified in two ways: via the kernel command
17 line with a module name prefix, or via modprobe, e.g.:
19 (kernel command line) usbcore.blinkenlights=1
20 (modprobe command line) modprobe usbcore blinkenlights=1
22 Parameters for modules which are built into the kernel need to be
23 specified on the kernel command line. modprobe looks through the
24 kernel command line (/proc/cmdline) and collects module parameters
25 when it loads a module, so the kernel command line can be used for
28 Hyphens (dashes) and underscores are equivalent in parameter names, so
29 log_buf_len=1M print-fatal-signals=1
30 can also be entered as
31 log-buf-len=1M print_fatal_signals=1
33 Double-quotes can be used to protect spaces in values, e.g.:
34 param="spaces in here"
36 This document may not be entirely up to date and comprehensive. The command
37 "modinfo -p ${modulename}" shows a current list of all parameters of a loadable
38 module. Loadable modules, after being loaded into the running kernel, also
39 reveal their parameters in /sys/module/${modulename}/parameters/. Some of these
40 parameters may be changed at runtime by the command
41 "echo -n ${value} > /sys/module/${modulename}/parameters/${parm}".
43 The parameters listed below are only valid if certain kernel build options were
44 enabled and if respective hardware is present. The text in square brackets at
45 the beginning of each description states the restrictions within which a
46 parameter is applicable:
48 ACPI ACPI support is enabled.
49 AGP AGP (Accelerated Graphics Port) is enabled.
50 ALSA ALSA sound support is enabled.
51 APIC APIC support is enabled.
52 APM Advanced Power Management support is enabled.
53 ARM ARM architecture is enabled.
54 AVR32 AVR32 architecture is enabled.
55 AX25 Appropriate AX.25 support is enabled.
56 BLACKFIN Blackfin architecture is enabled.
57 CLK Common clock infrastructure is enabled.
58 CMA Contiguous Memory Area support is enabled.
59 DM Device mapper support is enabled.
60 DRM Direct Rendering Management support is enabled.
61 DYNAMIC_DEBUG Build in debug messages and enable them at runtime
62 EDD BIOS Enhanced Disk Drive Services (EDD) is enabled
63 EFI EFI Partitioning (GPT) is enabled
64 EIDE EIDE/ATAPI support is enabled.
65 EVM Extended Verification Module
66 FB The frame buffer device is enabled.
67 FTRACE Function tracing enabled.
68 GCOV GCOV profiling is enabled.
69 HW Appropriate hardware is enabled.
70 IA-64 IA-64 architecture is enabled.
71 IMA Integrity measurement architecture is enabled.
72 IOSCHED More than one I/O scheduler is enabled.
73 IP_PNP IP DHCP, BOOTP, or RARP is enabled.
74 IPV6 IPv6 support is enabled.
75 ISAPNP ISA PnP code is enabled.
76 ISDN Appropriate ISDN support is enabled.
77 JOY Appropriate joystick support is enabled.
78 KGDB Kernel debugger support is enabled.
79 KVM Kernel Virtual Machine support is enabled.
80 LIBATA Libata driver is enabled
81 LP Printer support is enabled.
82 LOOP Loopback device support is enabled.
83 M68k M68k architecture is enabled.
84 These options have more detailed description inside of
85 Documentation/m68k/kernel-options.txt.
86 MDA MDA console support is enabled.
87 MIPS MIPS architecture is enabled.
88 MOUSE Appropriate mouse support is enabled.
89 MSI Message Signaled Interrupts (PCI).
90 MTD MTD (Memory Technology Device) support is enabled.
91 NET Appropriate network support is enabled.
92 NUMA NUMA support is enabled.
93 NFS Appropriate NFS support is enabled.
94 OSS OSS sound support is enabled.
95 PV_OPS A paravirtualized kernel is enabled.
96 PARIDE The ParIDE (parallel port IDE) subsystem is enabled.
97 PARISC The PA-RISC architecture is enabled.
98 PCI PCI bus support is enabled.
99 PCIE PCI Express support is enabled.
100 PCMCIA The PCMCIA subsystem is enabled.
101 PNP Plug & Play support is enabled.
102 PPC PowerPC architecture is enabled.
103 PPT Parallel port support is enabled.
104 PS2 Appropriate PS/2 support is enabled.
105 RAM RAM disk support is enabled.
106 S390 S390 architecture is enabled.
107 SCSI Appropriate SCSI support is enabled.
108 A lot of drivers have their options described inside
109 the Documentation/scsi/ sub-directory.
110 SECURITY Different security models are enabled.
111 SELINUX SELinux support is enabled.
112 APPARMOR AppArmor support is enabled.
113 SERIAL Serial support is enabled.
114 SH SuperH architecture is enabled.
115 SMP The kernel is an SMP kernel.
116 SPARC Sparc architecture is enabled.
117 SWSUSP Software suspend (hibernation) is enabled.
118 SUSPEND System suspend states are enabled.
119 TPM TPM drivers are enabled.
120 TS Appropriate touchscreen support is enabled.
121 UMS USB Mass Storage support is enabled.
122 USB USB support is enabled.
123 USBHID USB Human Interface Device support is enabled.
124 V4L Video For Linux support is enabled.
125 VMMIO Driver for memory mapped virtio devices is enabled.
126 VGA The VGA console has been enabled.
127 VT Virtual terminal support is enabled.
128 WDT Watchdog support is enabled.
129 XT IBM PC/XT MFM hard disk support is enabled.
130 X86-32 X86-32, aka i386 architecture is enabled.
131 X86-64 X86-64 architecture is enabled.
132 More X86-64 boot options can be found in
133 Documentation/x86/x86_64/boot-options.txt .
134 X86 Either 32-bit or 64-bit x86 (same as X86-32+X86-64)
135 XEN Xen support is enabled
137 In addition, the following text indicates that the option:
139 BUGS= Relates to possible processor bugs on the said processor.
140 KNL Is a kernel start-up parameter.
141 BOOT Is a boot loader parameter.
143 Parameters denoted with BOOT are actually interpreted by the boot
144 loader, and have no meaning to the kernel directly.
145 Do not modify the syntax of boot loader parameters without extreme
146 need or coordination with <Documentation/x86/boot.txt>.
148 There are also arch-specific kernel-parameters not documented here.
149 See for example <Documentation/x86/x86_64/boot-options.txt>.
151 Note that ALL kernel parameters listed below are CASE SENSITIVE, and that
152 a trailing = on the name of any parameter states that that parameter will
153 be entered as an environment variable, whereas its absence indicates that
154 it will appear as a kernel argument readable via /proc/cmdline by programs
155 running once the system is up.
157 The number of kernel parameters is not limited, but the length of the
158 complete command line (parameters including spaces etc.) is limited to
159 a fixed number of characters. This limit depends on the architecture
160 and is between 256 and 4096 characters. It is defined in the file
161 ./include/asm/setup.h as COMMAND_LINE_SIZE.
163 Finally, the [KMG] suffix is commonly described after a number of kernel
164 parameter values. These 'K', 'M', and 'G' letters represent the _binary_
165 multipliers 'Kilo', 'Mega', and 'Giga', equalling 2^10, 2^20, and 2^30
166 bytes respectively. Such letter suffixes can also be entirely omitted.
169 acpi= [HW,ACPI,X86,ARM64]
170 Advanced Configuration and Power Interface
171 Format: { force | off | strict | noirq | rsdt |
173 force -- enable ACPI if default was off
174 off -- disable ACPI if default was on
175 noirq -- do not use ACPI for IRQ routing
176 strict -- Be less tolerant of platforms that are not
177 strictly ACPI specification compliant.
178 rsdt -- prefer RSDT over (default) XSDT
179 copy_dsdt -- copy DSDT to memory
180 For ARM64, ONLY "acpi=off" or "acpi=force" are available
182 See also Documentation/power/runtime_pm.txt, pci=noacpi
184 acpi_apic_instance= [ACPI, IOAPIC]
186 2: use 2nd APIC table, if available
187 1,0: use 1st APIC table
190 acpi_backlight= [HW,ACPI]
191 acpi_backlight=vendor
193 If set to vendor, prefer vendor specific driver
194 (e.g. thinkpad_acpi, sony_acpi, etc.) instead
195 of the ACPI video.ko driver.
197 acpica_no_return_repair [HW, ACPI]
198 Disable AML predefined validation mechanism
199 This mechanism can repair the evaluation result to make
200 the return objects more ACPI specification compliant.
201 This option is useful for developers to identify the
202 root cause of an AML interpreter issue when the issue
203 has something to do with the repair mechanism.
205 acpi.debug_layer= [HW,ACPI,ACPI_DEBUG]
206 acpi.debug_level= [HW,ACPI,ACPI_DEBUG]
208 CONFIG_ACPI_DEBUG must be enabled to produce any ACPI
209 debug output. Bits in debug_layer correspond to a
210 _COMPONENT in an ACPI source file, e.g.,
211 #define _COMPONENT ACPI_PCI_COMPONENT
212 Bits in debug_level correspond to a level in
213 ACPI_DEBUG_PRINT statements, e.g.,
214 ACPI_DEBUG_PRINT((ACPI_DB_INFO, ...
215 The debug_level mask defaults to "info". See
216 Documentation/acpi/debug.txt for more information about
217 debug layers and levels.
219 Enable processor driver info messages:
220 acpi.debug_layer=0x20000000
221 Enable PCI/PCI interrupt routing info messages:
222 acpi.debug_layer=0x400000
223 Enable AML "Debug" output, i.e., stores to the Debug
224 object while interpreting AML:
225 acpi.debug_layer=0xffffffff acpi.debug_level=0x2
226 Enable all messages related to ACPI hardware:
227 acpi.debug_layer=0x2 acpi.debug_level=0xffffffff
229 Some values produce so much output that the system is
230 unusable. The "log_buf_len" parameter may be useful
231 if you need to capture more output.
233 acpi_enforce_resources= [ACPI]
234 { strict | lax | no }
235 Check for resource conflicts between native drivers
236 and ACPI OperationRegions (SystemIO and SystemMemory
237 only). IO ports and memory declared in ACPI might be
238 used by the ACPI subsystem in arbitrary AML code and
239 can interfere with legacy drivers.
240 strict (default): access to resources claimed by ACPI
241 is denied; legacy drivers trying to access reserved
242 resources will fail to bind to device using them.
243 lax: access to resources claimed by ACPI is allowed;
244 legacy drivers trying to access reserved resources
245 will bind successfully but a warning message is logged.
246 no: ACPI OperationRegions are not marked as reserved,
247 no further checks are performed.
249 acpi_force_table_verification [HW,ACPI]
250 Enable table checksum verification during early stage.
251 By default, this is disabled due to x86 early mapping
254 acpi_irq_balance [HW,ACPI]
255 ACPI will balance active IRQs
258 acpi_irq_nobalance [HW,ACPI]
259 ACPI will not move active IRQs (default)
262 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA
263 Format: <irq>,<irq>...
265 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for
267 Format: <irq>,<irq>...
269 acpi_no_auto_serialize [HW,ACPI]
270 Disable auto-serialization of AML methods
271 AML control methods that contain the opcodes to create
272 named objects will be marked as "Serialized" by the
273 auto-serialization feature.
274 This feature is enabled by default.
275 This option allows to turn off the feature.
277 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump
280 acpi_no_static_ssdt [HW,ACPI]
281 Disable installation of static SSDTs at early boot time
282 By default, SSDTs contained in the RSDT/XSDT will be
283 installed automatically and they will appear under
284 /sys/firmware/acpi/tables.
285 This option turns off this feature.
286 Note that specifying this option does not affect
287 dynamic table installation which will install SSDT
288 tables to /sys/firmware/acpi/tables/dynamic.
290 acpi_rsdp= [ACPI,EFI,KEXEC]
291 Pass the RSDP address to the kernel, mostly used
292 on machines running EFI runtime service to boot the
293 second kernel for kdump.
295 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS
296 Format: To spoof as Windows 98: ="Microsoft Windows"
298 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead
299 of 2 which is mandated by ACPI 6) as the supported ACPI
300 specification revision (when using this switch, it may
301 be necessary to carry out a cold reboot _twice_ in a
302 row to make it take effect on the platform firmware).
304 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings
305 acpi_osi="string1" # add string1
306 acpi_osi="!string2" # remove string2
307 acpi_osi=!* # remove all strings
308 acpi_osi=! # disable all built-in OS vendor
310 acpi_osi= # disable all strings
312 'acpi_osi=!' can be used in combination with single or
313 multiple 'acpi_osi="string1"' to support specific OS
314 vendor string(s). Note that such command can only
315 affect the default state of the OS vendor strings, thus
316 it cannot affect the default state of the feature group
317 strings and the current state of the OS vendor strings,
318 specifying it multiple times through kernel command line
319 is meaningless. This command is useful when one do not
320 care about the state of the feature group strings which
321 should be controlled by the OSPM.
323 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent
324 to 'acpi_osi="Windows 2000" acpi_osi=!', they all
325 can make '_OSI("Windows 2000")' TRUE.
327 'acpi_osi=' cannot be used in combination with other
328 'acpi_osi=' command lines, the _OSI method will not
329 exist in the ACPI namespace. NOTE that such command can
330 only affect the _OSI support state, thus specifying it
331 multiple times through kernel command line is also
334 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)'
337 'acpi_osi=!*' can be used in combination with single or
338 multiple 'acpi_osi="string1"' to support specific
339 string(s). Note that such command can affect the
340 current state of both the OS vendor strings and the
341 feature group strings, thus specifying it multiple times
342 through kernel command line is meaningful. But it may
343 still not able to affect the final state of a string if
344 there are quirks related to this string. This command
345 is useful when one want to control the state of the
346 feature group strings to debug BIOS issues related to
349 1. 'acpi_osi="Module Device" acpi_osi=!*' can make
350 '_OSI("Module Device")' FALSE.
351 2. 'acpi_osi=!* acpi_osi="Module Device"' can make
352 '_OSI("Module Device")' TRUE.
353 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is
355 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"'
357 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!',
358 they all will make '_OSI("Windows 2000")' TRUE.
361 Override the pmtimer bug detection: force the kernel
362 to assume that this machine's pmtimer latches its value
363 and always returns good values.
365 acpi_sci= [HW,ACPI] ACPI System Control Interrupt trigger mode
366 Format: { level | edge | high | low }
368 acpi_skip_timer_override [HW,ACPI]
369 Recognize and ignore IRQ0/pin2 Interrupt Override.
370 For broken nForce2 BIOS resulting in XT-PIC timer.
372 acpi_sleep= [HW,ACPI] Sleep options
373 Format: { s3_bios, s3_mode, s3_beep, s4_nohwsig,
374 old_ordering, nonvs, sci_force_enable }
375 See Documentation/power/video.txt for information on
377 s3_beep is for debugging; it makes the PC's speaker beep
378 as soon as the kernel's real-mode entry point is called.
379 s4_nohwsig prevents ACPI hardware signature from being
380 used during resume from hibernation.
381 old_ordering causes the ACPI 1.0 ordering of the _PTS
382 control method, with respect to putting devices into
383 low power states, to be enforced (the ACPI 2.0 ordering
384 of _PTS is used by default).
385 nonvs prevents the kernel from saving/restoring the
386 ACPI NVS memory during suspend/hibernation and resume.
387 sci_force_enable causes the kernel to set SCI_EN directly
388 on resume from S1/S3 (which is against the ACPI spec,
389 but some broken systems don't work without it).
391 acpi_use_timer_override [HW,ACPI]
392 Use timer override. For some broken Nvidia NF5 boards
393 that require a timer override, but don't have HPET
395 add_efi_memmap [EFI; X86] Include EFI memory map in
396 kernel's map of available physical RAM.
399 { off | try_unsupported }
400 off: disable AGP support
401 try_unsupported: try to drive unsupported chipsets
402 (may crash computer or cause data corruption)
405 See Documentation/sound/alsa/alsa-parameters.txt
408 Allow the default userspace alignment fault handler
409 behaviour to be specified. Bit 0 enables warnings,
410 bit 1 enables fixups, and bit 2 sends a segfault.
412 align_va_addr= [X86-64]
413 Align virtual addresses by clearing slice [14:12] when
414 allocating a VMA at process creation time. This option
415 gives you up to 3% performance improvement on AMD F15h
416 machines (where it is enabled by default) for a
417 CPU-intensive style benchmark, and it can vary highly in
418 a microbenchmark depending on workload and compiler.
420 32: only for 32-bit processes
421 64: only for 64-bit processes
422 on: enable for both 32- and 64-bit processes
423 off: disable for both 32- and 64-bit processes
425 alloc_snapshot [FTRACE]
426 Allocate the ftrace snapshot buffer on boot up when the
427 main buffer is allocated. This is handy if debugging
428 and you need to use tracing_snapshot() on boot up, and
429 do not want to use tracing_snapshot_alloc() as it needs
430 to be done where GFP_KERNEL allocations are allowed.
432 amd_iommu= [HW,X86-64]
433 Pass parameters to the AMD IOMMU driver in the system.
435 fullflush - enable flushing of IO/TLB entries when
436 they are unmapped. Otherwise they are
437 flushed before they will be reused, which
439 off - do not initialize any AMD IOMMU found in
441 force_isolation - Force device isolation for all
442 devices. The IOMMU driver is not
443 allowed anymore to lift isolation
444 requirements as needed. This option
445 does not override iommu=pt
447 amd_iommu_dump= [HW,X86-64]
448 Enable AMD IOMMU driver option to dump the ACPI table
449 for AMD IOMMU. With this option enabled, AMD IOMMU
450 driver will print ACPI tables for AMD IOMMU during
451 IOMMU initialization.
453 amijoy.map= [HW,JOY] Amiga joystick support
454 Map of devices attached to JOY0DAT and JOY1DAT
456 See also Documentation/input/joystick.txt
458 analog.map= [HW,JOY] Analog joystick and gamepad support
459 Specifies type or capabilities of an analog joystick
460 connected to one of 16 gameports
461 Format: <type1>,<type2>,..<type16>
464 Power management functions (SPARCstation-4/5 + deriv.)
466 Disable APC CPU standby support. SPARCstation-Fox does
467 not play well with APC CPU idle - disable it if you have
468 APC and your system crashes randomly.
470 apic= [APIC,X86-32] Advanced Programmable Interrupt Controller
471 Change the output verbosity whilst booting
472 Format: { quiet (default) | verbose | debug }
473 Change the amount of debugging information output
474 when initialising the APIC and IO-APIC components.
477 See Documentation/networking/ipv6.txt.
479 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
480 Limit apic dumping. The parameter defines the maximal
481 number of local apics being dumped. Also it is possible
482 to set it to "all" by meaning -- no limit here.
483 Format: { 1 (default) | 2 | ... | all }.
484 The parameter valid if only apic=debug or
485 apic=verbose is specified.
486 Example: apic=debug show_lapic=all
488 apm= [APM] Advanced Power Management
489 See header of arch/x86/kernel/apm_32.c.
491 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
492 Format: <io>,<irq>,<nodeID>
496 atarimouse= [HW,MOUSE] Atari Mouse
498 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
499 EzKey and similar keyboards
501 atkbd.reset= [HW] Reset keyboard during initialization
503 atkbd.set= [HW] Select keyboard code set
504 Format: <int> (2 = AT (default), 3 = PS/2)
506 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
509 atkbd.softraw= [HW] Choose between synthetic and real raw mode
510 Format: <bool> (0 = real, 1 = synthetic (default))
512 atkbd.softrepeat= [HW]
513 Use software keyboard repeat
515 audit= [KNL] Enable the audit sub-system
516 Format: { "0" | "1" } (0 = disabled, 1 = enabled)
517 0 - kernel audit is disabled and can not be enabled
518 until the next reboot
519 unset - kernel audit is initialized but disabled and
520 will be fully enabled by the userspace auditd.
521 1 - kernel audit is initialized and partially enabled,
522 storing at most audit_backlog_limit messages in
523 RAM until it is fully enabled by the userspace
527 audit_backlog_limit= [KNL] Set the audit queue size limit.
528 Format: <int> (must be >=0)
531 baycom_epp= [HW,AX25]
534 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
536 See header of drivers/net/hamradio/baycom_par.c.
538 baycom_ser_fdx= [HW,AX25]
539 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
540 Format: <io>,<irq>,<mode>[,<baud>]
541 See header of drivers/net/hamradio/baycom_ser_fdx.c.
543 baycom_ser_hdx= [HW,AX25]
544 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
545 Format: <io>,<irq>,<mode>
546 See header of drivers/net/hamradio/baycom_ser_hdx.c.
548 blkdevparts= Manual partition parsing of block device(s) for
549 embedded devices based on command line input.
550 See Documentation/block/cmdline-partition.txt
552 boot_delay= Milliseconds to delay each printk during boot.
553 Values larger than 10 seconds (10000) are changed to
557 bootmem_debug [KNL] Enable bootmem allocator debug messages.
559 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
560 bttv.radio= Most important insmod options are available as
562 bttv.pll= See Documentation/video4linux/bttv/Insmod-options
565 bulk_remove=off [PPC] This parameter disables the use of the pSeries
566 firmware feature for flushing multiple hpte entries
569 c101= [NET] Moxa C101 synchronous serial card
571 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
572 Sometimes CPU hardware bugs make them report the cache
573 size incorrectly. The kernel will attempt work arounds
574 to fix known problems, but for some CPUs it is not
575 possible to determine what the correct size should be.
576 This option provides an override for these situations.
578 ca_keys= [KEYS] This parameter identifies a specific key(s) on
579 the system trusted keyring to be used for certificate
581 format: { id:<keyid> | builtin }
583 cca= [MIPS] Override the kernel pages' cache coherency
584 algorithm. Accepted values range from 0 to 7
585 inclusive. See arch/mips/include/asm/pgtable-bits.h
586 for platform specific values (SB1, Loongson3 and
589 ccw_timeout_log [S390]
590 See Documentation/s390/CommonIO for details.
592 cgroup_disable= [KNL] Disable a particular controller
593 Format: {name of the controller(s) to disable}
594 The effects of cgroup_disable=foo are:
595 - foo isn't auto-mounted if you mount all cgroups in
597 - foo isn't visible as an individually mountable
599 {Currently only "memory" controller deal with this and
600 cut the overhead, others just disable the usage. So
601 only cgroup_disable=memory is actually worthy}
603 checkreqprot [SELINUX] Set initial checkreqprot flag value.
604 Format: { "0" | "1" }
605 See security/selinux/Kconfig help text.
606 0 -- check protection applied by kernel (includes
607 any implied execute protection).
608 1 -- check protection requested by application.
609 Default value is set via a kernel config option.
610 Value can be changed at runtime via
611 /selinux/checkreqprot.
614 See Documentation/s390/CommonIO for details.
617 Prevents the clock framework from automatically gating
618 clocks that have not been explicitly enabled by a Linux
619 device driver but are enabled in hardware at reset or
620 by the bootloader/firmware. Note that this does not
621 force such clocks to be always-on nor does it reserve
622 those clocks in any way. This parameter is useful for
623 debug and development, but should not be needed on a
624 platform with proper driver support. For more
625 information, see Documentation/clk.txt.
627 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
629 Forces specified clocksource (if available) to be used
630 when calculating gettimeofday(). If specified
631 clocksource is not available, it defaults to PIT.
632 Format: { pit | tsc | cyclone | pmtmr }
634 clocksource= Override the default clocksource
636 Override the default clocksource and use the clocksource
637 with the name specified.
638 Some clocksource names to choose from, depending on
640 [all] jiffies (this is the base, fallback clocksource)
642 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
643 pxa_timer,timer3,32k_counter,timer0_1
645 [X86-32] pit,hpet,tsc;
646 scx200_hrt on Geode; cyclone on IBM x440
654 clearcpuid=BITNUM [X86]
655 Disable CPUID feature X for the kernel. See
656 arch/x86/include/asm/cpufeatures.h for the valid bit
657 numbers. Note the Linux specific bits are not necessarily
658 stable over kernel options, but the vendor specific
660 Also note that user programs calling CPUID directly
661 or using the feature without checking anything
662 will still see it. This just prevents it from
663 being used by the kernel or shown in /proc/cpuinfo.
664 Also note the kernel might malfunction if you disable
667 cma=nn[MG]@[start[MG][-end[MG]]]
669 Sets the size of kernel global memory area for
670 contiguous memory allocations and optionally the
671 placement constraint by the physical address range of
672 memory allocations. A value of 0 disables CMA
673 altogether. For more information, see
674 include/linux/dma-contiguous.h
676 cmo_free_hint= [PPC] Format: { yes | no }
677 Specify whether pages are marked as being inactive
678 when they are freed. This is used in CMO environments
679 to determine OS memory pressure for page stealing by
683 coherent_pool=nn[KMG] [ARM,KNL]
684 Sets the size of memory pool for coherent, atomic dma
685 allocations, by default set to 256K.
687 code_bytes [X86] How many bytes of object code to print
692 com20020= [HW,NET] ARCnet - COM20020 chipset
694 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
696 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
700 ARCnet - COM90xx chipset (memory-mapped buffers)
701 Format: <io>[,<irq>[,<memstart>]]
703 condev= [HW,S390] console device
706 console= [KNL] Output console device and options.
708 tty<n> Use the virtual console device <n>.
712 Use the specified serial port. The options are of
713 the form "bbbbpnf", where "bbbb" is the baud rate,
714 "p" is parity ("n", "o", or "e"), "n" is number of
715 bits, and "f" is flow control ("r" for RTS or
716 omit it). Default is "9600n8".
718 See Documentation/serial-console.txt for more
720 Documentation/networking/netconsole.txt for an
723 uart[8250],io,<addr>[,options]
724 uart[8250],mmio,<addr>[,options]
725 uart[8250],mmio32,<addr>[,options]
726 uart[8250],0x<addr>[,options]
727 Start an early, polled-mode console on the 8250/16550
728 UART at the specified I/O port or MMIO address,
729 switching to the matching ttyS device later.
730 MMIO inter-register address stride is either 8-bit
731 (mmio) or 32-bit (mmio32).
732 If none of [io|mmio|mmio32], <addr> is assumed to be
733 equivalent to 'mmio'. 'options' are specified in the
734 same format described for ttyS above; if unspecified,
735 the h/w is not re-initialized.
737 hvc<n> Use the hypervisor console device <n>. This is for
738 both Xen and PowerPC hypervisors.
740 If the device connected to the port is not a TTY but a braille
741 device, prepend "brl," before the device type, for instance
743 For now, only VisioBraille is supported.
745 consoleblank= [KNL] The console blank (screen saver) timeout in
746 seconds. Defaults to 10*60 = 10mins. A value of 0
747 disables the blank timer.
750 [KNL] Change the default value for
751 /proc/<pid>/coredump_filter.
752 See also Documentation/filesystems/proc.txt.
754 cpuidle.off=1 [CPU_IDLE]
755 disable the cpuidle sub-system
758 [X86] Delay for N microsec between assert and de-assert
759 of APIC INIT to start processors. This delay occurs
760 on every CPU online, such as boot, and resume from suspend.
763 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
765 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
767 crashkernel=size[KMG][@offset[KMG]]
768 [KNL] Using kexec, Linux can switch to a 'crash kernel'
769 upon panic. This parameter reserves the physical
770 memory region [offset, offset + size] for that kernel
771 image. If '@offset' is omitted, then a suitable offset
772 is selected automatically. Check
773 Documentation/kdump/kdump.txt for further details.
775 crashkernel=range1:size1[,range2:size2,...][@offset]
776 [KNL] Same as above, but depends on the memory
777 in the running system. The syntax of range is
778 start-[end] where start and end are both
779 a memory unit (amount[KMG]). See also
780 Documentation/kdump/kdump.txt for an example.
782 crashkernel=size[KMG],high
783 [KNL, x86_64] range could be above 4G. Allow kernel
784 to allocate physical memory region from top, so could
785 be above 4G if system have more than 4G ram installed.
786 Otherwise memory region will be allocated below 4G, if
788 It will be ignored if crashkernel=X is specified.
789 crashkernel=size[KMG],low
790 [KNL, x86_64] range under 4G. When crashkernel=X,high
791 is passed, kernel could allocate physical memory region
792 above 4G, that cause second kernel crash on system
793 that require some amount of low memory, e.g. swiotlb
794 requires at least 64M+32K low memory, also enough extra
795 low memory is needed to make sure DMA buffers for 32-bit
796 devices won't run out. Kernel would try to allocate at
797 at least 256M below 4G automatically.
798 This one let user to specify own low range under 4G
799 for second kernel instead.
800 0: to disable low allocation.
801 It will be ignored when crashkernel=X,high is not used
802 or memory reserved is below 4G.
807 cs89x0_media= [HW,NET]
808 Format: { rj45 | aui | bnc }
811 See header of drivers/s390/block/dasd_devmap.c.
813 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
814 (one device per port)
815 Format: <port#>,<type>
816 See also Documentation/input/joystick-parport.txt
818 ddebug_query= [KNL,DYNAMIC_DEBUG] Enable debug messages at early boot
819 time. See Documentation/dynamic-debug-howto.txt for
820 details. Deprecated, see dyndbg.
822 debug [KNL] Enable kernel debugging (events log level).
825 [KNL] verbose self-tests
827 Print debugging info while doing the locking API
829 We default to 0 (no extra messages), setting it to
830 1 will print _a lot_ more information - normally
831 only useful to kernel developers.
833 debug_objects [KNL] Enable object debugging
836 [KNL] Disable object debugging
838 debug_guardpage_minorder=
839 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
840 parameter allows control of the order of pages that will
841 be intentionally kept free (and hence protected) by the
842 buddy allocator. Bigger value increase the probability
843 of catching random memory corruption, but reduce the
844 amount of memory for normal system use. The maximum
845 possible value is MAX_ORDER/2. Setting this parameter
846 to 1 or 2 should be enough to identify most random
847 memory corruption problems caused by bugs in kernel or
848 driver code when a CPU writes to (or reads from) a
849 random memory location. Note that there exists a class
850 of memory corruptions problems caused by buggy H/W or
851 F/W or by drivers badly programing DMA (basically when
852 memory is written at bus level and the CPU MMU is
853 bypassed) which are not detectable by
854 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
855 tracking down these problems.
858 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
859 parameter enables the feature at boot time. In
860 default, it is disabled. We can avoid allocating huge
861 chunk of memory for debug pagealloc if we don't enable
862 it at boot time and the system will work mostly same
863 with the kernel built without CONFIG_DEBUG_PAGEALLOC.
864 on: enable the feature
866 debugpat [X86] Enable PAT debugging
868 decnet.addr= [HW,NET]
869 Format: <area>[,<node>]
870 See also Documentation/networking/decnet.txt.
873 [same as hugepagesz=] The size of the default
874 HugeTLB page size. This is the size represented by
875 the legacy /proc/ hugepages APIs, used for SHM, and
876 default size when mounting hugetlbfs filesystems.
877 Defaults to the default architecture's huge page size
881 Set number of hash buckets for dentry cache.
884 See Documentation/networking/ipv6.txt.
886 disable_cpu_apicid= [X86,APIC,SMP]
888 The number of initial APIC ID for the
889 corresponding CPU to be disabled at boot,
890 mostly used for the kdump 2nd kernel to
891 disable BSP to wake up multiple CPUs without
892 causing system reset or hang due to sending
895 disable_ddw [PPC/PSERIES]
896 Disable Dynamic DMA Window support. Use this if
897 to workaround buggy firmware.
900 See Documentation/networking/ipv6.txt.
902 disable_mtrr_cleanup [X86]
903 The kernel tries to adjust MTRR layout from continuous
904 to discrete, to make X server driver able to add WB
905 entry later. This parameter disables that.
907 disable_mtrr_trim [X86, Intel and AMD only]
908 By default the kernel will trim any uncacheable
909 memory out of your available memory pool based on
910 MTRR settings. This parameter disables that behavior,
911 possibly causing your machine to run very slowly.
913 disable_timer_pin_1 [X86]
914 Disable PIN 1 of APIC timer
915 Can be useful to work around chipset bugs.
917 dis_ucode_ldr [X86] Disable the microcode loader.
919 dm= [DM] Allows early creation of a device-mapper device.
920 See Documentation/device-mapper/boot.txt.
922 dmasound= [HW,OSS] Sound subsystem buff
924 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
925 this option disables the debugging code at boot.
927 dma_debug_entries=<number>
928 This option allows to tune the number of preallocated
929 entries for DMA-API debugging code. One entry is
930 required per DMA-API allocation. Use this if the
931 DMA-API debugging code disables itself because the
932 architectural default is too low.
934 dma_debug_driver=<driver_name>
935 With this option the DMA-API debugging driver
936 filter feature can be enabled at boot time. Just
937 pass the driver to filter for as the parameter.
938 The filter can be disabled or changed to another
939 driver later using sysfs.
941 drm_kms_helper.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
942 Broken monitors, graphic adapters, KVMs and EDIDless
943 panels may send no or incorrect EDID data sets.
944 This parameter allows to specify an EDID data sets
945 in the /lib/firmware directory that are used instead.
946 Generic built-in EDID data sets are used, if one of
947 edid/1024x768.bin, edid/1280x1024.bin,
948 edid/1680x1050.bin, or edid/1920x1080.bin is given
949 and no file with the same name exists. Details and
950 instructions how to build your own EDID data are
951 available in Documentation/EDID/HOWTO.txt. An EDID
952 data set will only be used for a particular connector,
953 if its name and a colon are prepended to the EDID
954 name. Each connector may use a unique EDID data
955 set by separating the files with a comma. An EDID
956 data set with no connector name will be used for
957 any connectors not explicitly specified.
961 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
962 module.dyndbg[="val"]
963 Enable debug messages at boot time. See
964 Documentation/dynamic-debug-howto.txt for details.
966 nompx [X86] Disables Intel Memory Protection Extensions.
967 See Documentation/x86/intel_mpx.txt for more
968 information about the feature.
971 module.async_probe [KNL]
972 Enable asynchronous probe on this module.
974 early_ioremap_debug [KNL]
975 Enable debug messages in early_ioremap support. This
976 is useful for tracking down temporary early mappings
977 which are not unmapped.
979 earlycon= [KNL] Output early console device and options.
981 When used with no options, the early console is
982 determined by the stdout-path property in device
986 Start an early, polled-mode console on a cadence serial
987 port at the specified address. The cadence serial port
988 must already be setup and configured. Options are not
991 uart[8250],io,<addr>[,options]
992 uart[8250],mmio,<addr>[,options]
993 uart[8250],mmio32,<addr>[,options]
994 uart[8250],mmio32be,<addr>[,options]
995 uart[8250],0x<addr>[,options]
996 Start an early, polled-mode console on the 8250/16550
997 UART at the specified I/O port or MMIO address.
998 MMIO inter-register address stride is either 8-bit
999 (mmio) or 32-bit (mmio32 or mmio32be).
1000 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1001 to be equivalent to 'mmio'. 'options' are specified
1002 in the same format described for "console=ttyS<n>"; if
1003 unspecified, the h/w is not initialized.
1006 Start an early, polled-mode console on a pl011 serial
1007 port at the specified address. The pl011 serial port
1008 must already be setup and configured. Options are not
1012 Start an early, polled-mode console on an msm serial
1013 port at the specified address. The serial port
1014 must already be setup and configured. Options are not
1017 msm_serial_dm,<addr>
1018 Start an early, polled-mode console on an msm serial
1019 dm port at the specified address. The serial port
1020 must already be setup and configured. Options are not
1023 smh Use ARM semihosting calls for early console.
1031 Use early console provided by serial driver available
1032 on Samsung SoCs, requires selecting proper type and
1033 a correct base address of the selected UART port. The
1034 serial port must already be setup and configured.
1035 Options are not yet supported.
1039 Use early console provided by Freescale LP UART driver
1040 found on Freescale Vybrid and QorIQ LS1021A processors.
1041 A valid base address must be provided, and the serial
1042 port must already be setup and configured.
1044 earlyprintk= [X86,SH,BLACKFIN,ARM,M68k]
1048 earlyprintk=serial[,ttySn[,baudrate]]
1049 earlyprintk=serial[,0x...[,baudrate]]
1050 earlyprintk=ttySn[,baudrate]
1051 earlyprintk=dbgp[debugController#]
1052 earlyprintk=pciserial,bus:device.function[,baudrate]
1054 earlyprintk is useful when the kernel crashes before
1055 the normal console is initialized. It is not enabled by
1056 default because it has some cosmetic problems.
1058 Append ",keep" to not disable it when the real console
1061 Only one of vga, efi, serial, or usb debug port can
1064 Currently only ttyS0 and ttyS1 may be specified by
1065 name. Other I/O ports may be explicitly specified
1066 on some architectures (x86 and arm at least) by
1067 replacing ttySn with an I/O port address, like this:
1068 earlyprintk=serial,0x1008,115200
1069 You can find the port for a given device in
1070 /proc/tty/driver/serial:
1071 2: uart:ST16650V2 port:00001008 irq:18 ...
1073 Interaction with the standard serial driver is not
1076 The VGA and EFI output is eventually overwritten by
1079 The xen output can only be used by Xen PV guests.
1081 edac_report= [HW,EDAC] Control how to report EDAC event
1082 Format: {"on" | "off" | "force"}
1083 on: enable EDAC to report H/W event. May be overridden
1084 by other higher priority error reporting module.
1085 off: disable H/W event reporting through EDAC.
1086 force: enforce the use of EDAC to report H/W event.
1089 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1092 This is designed to be used in conjunction with
1093 the boot argument: earlyprintk=vga
1096 Format: {"off" | "on" | "skip[mbr]"}
1099 Format: { "old_map", "nochunk", "noruntime", "debug" }
1100 old_map [X86-64]: switch to the old ioremap-based EFI
1101 runtime services mapping. 32-bit still uses this one by
1103 nochunk: disable reading files in "chunks" in the EFI
1104 boot stub, as chunking can cause problems with some
1105 firmware implementations.
1106 noruntime : disable EFI runtime services support
1107 debug: enable misc debug output
1109 efi_no_storage_paranoia [EFI; X86]
1110 Using this parameter you can use more than 50% of
1111 your efi variable storage. Use this parameter only if
1112 you are really sure that your UEFI does sane gc and
1113 fulfills the spec otherwise your board may brick.
1115 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1116 Add arbitrary attribute to specific memory range by
1117 updating original EFI memory map.
1118 Region of memory which aa attribute is added to is
1120 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1121 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1122 attribute is added to range 0x100000000-0x180000000 and
1123 0x10a0000000-0x1120000000.
1125 Using this parameter you can do debugging of EFI memmap
1126 related feature. For example, you can do debugging of
1127 Address Range Mirroring feature even if your box
1130 eisa_irq_edge= [PARISC,HW]
1131 See header of drivers/parisc/eisa.c.
1134 See comment before function elanfreq_setup() in
1135 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1138 Format: {"cfq" | "deadline" | "noop"}
1139 See Documentation/block/cfq-iosched.txt and
1140 Documentation/block/deadline-iosched.txt for details.
1142 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1143 Specifies physical address of start of kernel core
1144 image elf header and optionally the size. Generally
1145 kexec loader will pass this option to capture kernel.
1146 See Documentation/kdump/kdump.txt for details.
1148 enable_mtrr_cleanup [X86]
1149 The kernel tries to adjust MTRR layout from continuous
1150 to discrete, to make X server driver able to add WB
1151 entry later. This parameter enables that.
1153 enable_timer_pin_1 [X86]
1154 Enable PIN 1 of APIC timer
1155 Can be useful to work around chipset bugs
1156 (in particular on some ATI chipsets).
1157 The kernel tries to set a reasonable default.
1159 enforcing [SELINUX] Set initial enforcing status.
1161 See security/selinux/Kconfig help text.
1162 0 -- permissive (log only, no denials).
1163 1 -- enforcing (deny and log).
1165 Value can be changed at runtime via /selinux/enforce.
1168 Disable Error Record Serialization Table (ERST)
1171 ether= [HW,NET] Ethernet cards parameters
1172 This option is obsoleted by the "netdev=" option, which
1173 has equivalent usage. See its documentation for details.
1177 Permit 'security.evm' to be updated regardless of
1178 current integrity status.
1182 fail_make_request=[KNL]
1183 General fault injection mechanism.
1184 Format: <interval>,<probability>,<space>,<times>
1185 See also Documentation/fault-injection/.
1188 See Documentation/blockdev/floppy.txt.
1190 force_pal_cache_flush
1191 [IA-64] Avoid check_sal_cache_flush which may hang on
1192 buggy SAL_CACHE_FLUSH implementations. Using this
1193 parameter will force ia64_sal_cache_flush to call
1194 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1197 Forcefully enable Physical Address Extension (PAE).
1198 Many Pentium M systems disable PAE but may have a
1199 functionally usable PAE implementation.
1200 Warning: use of this parameter will taint the kernel
1201 and may cause unknown problems.
1204 [FTRACE] will set and start the specified tracer
1205 as early as possible in order to facilitate early
1208 ftrace_dump_on_oops[=orig_cpu]
1209 [FTRACE] will dump the trace buffers on oops.
1210 If no parameter is passed, ftrace will dump
1211 buffers of all CPUs, but if you pass orig_cpu, it will
1212 dump only the buffer of the CPU that triggered the
1215 ftrace_filter=[function-list]
1216 [FTRACE] Limit the functions traced by the function
1217 tracer at boot up. function-list is a comma separated
1218 list of functions. This list can be changed at run
1219 time by the set_ftrace_filter file in the debugfs
1222 ftrace_notrace=[function-list]
1223 [FTRACE] Do not trace the functions specified in
1224 function-list. This list can be changed at run time
1225 by the set_ftrace_notrace file in the debugfs
1228 ftrace_graph_filter=[function-list]
1229 [FTRACE] Limit the top level callers functions traced
1230 by the function graph tracer at boot up.
1231 function-list is a comma separated list of functions
1232 that can be changed at run time by the
1233 set_graph_function file in the debugfs tracing directory.
1235 ftrace_graph_notrace=[function-list]
1236 [FTRACE] Do not trace from the functions specified in
1237 function-list. This list is a comma separated list of
1238 functions that can be changed at run time by the
1239 set_graph_notrace file in the debugfs tracing directory.
1242 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1243 support via parallel port (up to 5 devices per port)
1244 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1245 See also Documentation/input/joystick-parport.txt
1249 gart_fix_e820= [X86_64] disable the fix e820 for K8 GART
1253 gcov_persist= [GCOV] When non-zero (default), profiling data for
1254 kernel modules is saved and remains accessible via
1255 debugfs, even when the module is unloaded/reloaded.
1256 When zero, profiling data is discarded and associated
1257 debugfs files are removed at module unload time.
1259 goldfish [X86] Enable the goldfish android emulator platform.
1260 Don't use this when you are not running on the
1263 gpt [EFI] Forces disk with valid GPT signature but
1264 invalid Protective MBR to be treated as GPT. If the
1265 primary GPT is corrupted, it enables the backup/alternate
1266 GPT to be used instead.
1268 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1269 the "Enable 0" bit of the configuration register.
1272 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1273 the "Enable 0" bit of the configuration register.
1276 grcan.select= [HW] Select which physical interface to use.
1279 grcan.txsize= [HW] Sets the size of the tx buffer.
1280 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1282 grcan.rxsize= [HW] Sets the size of the rx buffer.
1283 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1286 hardlockup_all_cpu_backtrace=
1287 [KNL] Should the hard-lockup detector generate
1288 backtraces on all cpus.
1291 hashdist= [KNL,NUMA] Large hashes allocated during boot
1292 are distributed across NUMA nodes. Defaults on
1293 for 64-bit NUMA, off otherwise.
1294 Format: 0 | 1 (for off | on)
1296 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1298 hd= [EIDE] (E)IDE hard drive subsystem geometry
1299 Format: <cyl>,<head>,<sect>
1302 Disable Hardware Error Source Table (HEST) support;
1303 corresponding firmware-first mode error processing
1304 logic will be disabled.
1306 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1307 size of <nn>. This works even on boxes that have no
1308 highmem otherwise. This also works to reduce highmem
1309 size on bigger boxes.
1311 highres= [KNL] Enable/disable high resolution timer mode.
1312 Valid parameters: "on", "off"
1316 See Documentation/isdn/README.HiSax.
1320 hpet= [X86-32,HPET] option to control HPET usage
1321 Format: { enable (default) | disable | force |
1323 disable: disable HPET and use PIT instead
1324 force: allow force enabled of undocumented chips (ICH4,
1326 verbose: show contents of HPET registers during setup
1328 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1329 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1331 hugepages= [HW,X86-32,IA-64] HugeTLB pages to allocate at boot.
1332 hugepagesz= [HW,IA-64,PPC,X86-64] The size of the HugeTLB pages.
1333 On x86-64 and powerpc, this option can be specified
1334 multiple times interleaved with hugepages= to reserve
1335 huge pages of different sizes. Valid pages sizes on
1336 x86-64 are 2M (when the CPU supports "pse") and 1G
1337 (when the CPU supports the "pdpe1gb" cpuinfo flag).
1339 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1340 terminal devices. Valid values: 0..8
1341 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1342 If specified, z/VM IUCV HVC accepts connections
1343 from listed z/VM user IDs only.
1345 hwthread_map= [METAG] Comma-separated list of Linux cpu id to
1346 hardware thread id mappings.
1347 Format: <cpu>:<hwthread>
1350 Do not unregister boot console at start. This is only
1351 useful for debugging when something happens in the window
1352 between unregistering the boot console and initializing
1355 i2c_bus= [HW] Override the default board specific I2C bus speed
1356 or register an additional I2C bus that is not
1357 registered from board initialization code.
1361 i8042.debug [HW] Toggle i8042 debug mode
1362 i8042.unmask_kbd_data
1363 [HW] Enable printing of interrupt data from the KBD port
1364 (disabled by default, and as a pre-condition
1365 requires that i8042.debug=1 be enabled)
1366 i8042.direct [HW] Put keyboard port into non-translated mode
1367 i8042.dumbkbd [HW] Pretend that controller can only read data from
1368 keyboard and cannot control its state
1369 (Don't attempt to blink the leds)
1370 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1371 i8042.nokbd [HW] Don't check/create keyboard port
1372 i8042.noloop [HW] Disable the AUX Loopback command while probing
1374 i8042.nomux [HW] Don't check presence of an active multiplexing
1376 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1378 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1379 i8042.reset [HW] Reset the controller during init, cleanup and
1380 suspend-to-ram transitions, only during s2r
1381 transitions, or never reset
1382 Format: { 1 | Y | y | 0 | N | n }
1383 1, Y, y: always reset controller
1384 0, N, n: don't ever reset controller
1385 Default: only on s2r transitions on x86; most other
1386 architectures force reset to be always executed
1387 i8042.unlock [HW] Unlock (ignore) the keylock
1388 i8042.kbdreset [HW] Reset device connected to KBD port
1392 i8k.ignore_dmi [HW] Continue probing hardware even if DMI data
1393 indicates that the driver is running on unsupported
1395 i8k.force [HW] Activate i8k driver even if SMM BIOS signature
1396 does not match list of supported models.
1398 [HW] Report power status in /proc/i8k
1399 (disabled by default)
1400 i8k.restricted [HW] Allow controlling fans only if SYS_ADMIN
1403 i915.invert_brightness=
1404 [DRM] Invert the sense of the variable that is used to
1405 set the brightness of the panel backlight. Normally a
1406 brightness value of 0 indicates backlight switched off,
1407 and the maximum of the brightness value sets the backlight
1408 to maximum brightness. If this parameter is set to 0
1409 (default) and the machine requires it, or this parameter
1410 is set to 1, a brightness value of 0 sets the backlight
1411 to maximum brightness, and the maximum of the brightness
1412 value switches the backlight off.
1413 -1 -- never invert brightness
1414 0 -- machine default
1415 1 -- force brightness inversion
1418 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1420 ide-core.nodma= [HW] (E)IDE subsystem
1421 Format: =0.0 to prevent dma on hda, =0.1 hdb =1.0 hdc
1422 .vlb_clock .pci_clock .noflush .nohpa .noprobe .nowerr
1423 .cdrom .chs .ignore_cable are additional options
1424 See Documentation/ide/ide.txt.
1426 ide-generic.probe-mask= [HW] (E)IDE subsystem
1428 Probe mask for legacy ISA IDE ports. Depending on
1429 platform up to 6 ports are supported, enabled by
1430 setting corresponding bits in the mask to 1. The
1431 default value is 0x0, which has a special meaning.
1432 On systems that have PCI, it triggers scanning the
1433 PCI bus for the first and the second port, which
1434 are then probed. On systems without PCI the value
1435 of 0x0 enables probing the two first ports as if it
1438 ide-pci-generic.all-generic-ide [HW] (E)IDE subsystem
1439 Claim all unknown PCI IDE storage controllers.
1442 Format: idle=poll, idle=halt, idle=nomwait
1443 Poll forces a polling idle loop that can slightly
1444 improve the performance of waking up a idle CPU, but
1445 will use a lot of power and make the system run hot.
1447 idle=halt: Halt is forced to be used for CPU idle.
1448 In such case C2/C3 won't be used again.
1449 idle=nomwait: Disable mwait for CPU C-states
1451 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1452 Format: { strict | legacy | 2008 | relaxed }
1455 Choose which programs will be accepted for execution
1456 based on the IEEE 754 NaN encoding(s) supported by
1457 the FPU and the NaN encoding requested with the value
1458 of an ELF file header flag individually set by each
1459 binary. Hardware implementations are permitted to
1460 support either or both of the legacy and the 2008 NaN
1463 Available settings are as follows:
1464 strict accept binaries that request a NaN encoding
1465 supported by the FPU
1466 legacy only accept legacy-NaN binaries, if supported
1468 2008 only accept 2008-NaN binaries, if supported
1470 relaxed accept any binaries regardless of whether
1471 supported by the FPU
1473 The FPU emulator is always able to support both NaN
1474 encodings, so if no FPU hardware is present or it has
1475 been disabled with 'nofpu', then the settings of
1476 'legacy' and '2008' strap the emulator accordingly,
1477 'relaxed' straps the emulator for both legacy-NaN and
1478 2008-NaN, whereas 'strict' enables legacy-NaN only on
1479 legacy processors and both NaN encodings on MIPS32 or
1482 The setting for ABS.fmt/NEG.fmt instruction execution
1483 mode generally follows that for the NaN encoding,
1484 except where unsupported by hardware.
1486 ignore_loglevel [KNL]
1487 Ignore loglevel setting - this will print /all/
1488 kernel messages to the console. Useful for debugging.
1489 We also add it as printk module parameter, so users
1490 could change it dynamically, usually by
1491 /sys/module/printk/parameters/ignore_loglevel.
1493 ihash_entries= [KNL]
1494 Set number of hash buckets for inode cache.
1496 ima_appraise= [IMA] appraise integrity measurements
1497 Format: { "off" | "enforce" | "fix" | "log" }
1500 ima_appraise_tcb [IMA]
1501 The builtin appraise policy appraises all files
1505 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1509 The list of supported hash algorithms is defined
1510 in crypto/hash_info.h.
1513 The builtin measurement policy to load during IMA
1514 setup. Specyfing "tcb" as the value, measures all
1515 programs exec'd, files mmap'd for exec, and all files
1516 opened with the read mode bit set by either the
1517 effective uid (euid=0) or uid=0.
1520 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
1521 Load a policy which meets the needs of the Trusted
1522 Computing Base. This means IMA will measure all
1523 programs exec'd, files mmap'd for exec, and all files
1524 opened for read by uid=0.
1527 Select one of defined IMA measurements template formats.
1528 Formats: { "ima" | "ima-ng" | "ima-sig" }
1532 [IMA] Define a custom template format.
1533 Format: { "field1|...|fieldN" }
1535 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
1536 Format: <min_file_size>
1537 Set the minimal file size for using asynchronous hash.
1538 If left unspecified, ahash usage is disabled.
1540 ahash performance varies for different data sizes on
1541 different crypto accelerators. This option can be used
1542 to achieve the best performance for a particular HW.
1544 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
1546 Set hashing buffer size. Default: 4k.
1548 ahash performance varies for different chunk sizes on
1549 different crypto accelerators. This option can be used
1550 to achieve best performance for particular HW.
1554 Run specified binary instead of /sbin/init as init
1557 initcall_debug [KNL] Trace initcalls as they are executed. Useful
1558 for working out where the kernel is dying during
1561 initcall_blacklist= [KNL] Do not execute a comma-separated list of
1562 initcall functions. Useful for debugging built-in
1563 modules and initcalls.
1565 initrd= [BOOT] Specify the location of the initial ramdisk
1567 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
1570 int_pln_enable [x86] Enable power limit notification interrupt
1572 integrity_audit=[IMA]
1573 Format: { "0" | "1" }
1574 0 -- basic integrity auditing messages. (Default)
1575 1 -- additional integrity auditing messages.
1577 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
1579 Enable intel iommu driver.
1581 Disable intel iommu driver.
1582 igfx_off [Default Off]
1583 By default, gfx is mapped as normal device. If a gfx
1584 device has a dedicated DMAR unit, the DMAR unit is
1585 bypassed by not enabling DMAR with this option. In
1586 this case, gfx device will use physical address for
1589 With this option iommu will not optimize to look
1590 for io virtual address below 32-bit forcing dual
1591 address cycle on pci bus for cards supporting greater
1592 than 32-bit addressing. The default is to look
1593 for translation below 32-bit and if not available
1594 then look in the higher range.
1595 strict [Default Off]
1596 With this option on every unmap_single operation will
1597 result in a hardware IOTLB flush operation as opposed
1598 to batching them for performance.
1599 sp_off [Default Off]
1600 By default, super page will be supported if Intel IOMMU
1601 has the capability. With this option, super page will
1603 ecs_off [Default Off]
1604 By default, extended context tables will be supported if
1605 the hardware advertises that it has support both for the
1606 extended tables themselves, and also PASID support. With
1607 this option set, extended tables will not be used even
1608 on hardware which claims to support them.
1610 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
1611 0 disables intel_idle and fall back on acpi_idle.
1612 1 to 6 specify maximum depth of C-state.
1616 Do not enable intel_pstate as the default
1617 scaling driver for the supported processors
1619 Enable intel_pstate on systems that prohibit it by default
1620 in favor of acpi-cpufreq. Forcing the intel_pstate driver
1621 instead of acpi-cpufreq may disable platform features, such
1622 as thermal controls and power capping, that rely on ACPI
1623 P-States information being indicated to OSPM and therefore
1624 should be used with caution. This option does not work with
1625 processors that aren't supported by the intel_pstate driver
1626 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
1628 Do not enable hardware P state control (HWP)
1631 Only load intel_pstate on systems which support
1632 hardware P state control (HWP) if available.
1634 intremap= [X86-64, Intel-IOMMU]
1635 on enable Interrupt Remapping (default)
1636 off disable Interrupt Remapping
1637 nosid disable Source ID checking
1639 BIOS x2APIC opt-out request will be ignored
1640 nopost disable Interrupt Posting
1642 iomem= Disable strict checking of access to MMIO memory
1643 strict regions from userspace.
1658 nobypass [PPC/POWERNV]
1659 Disable IOMMU bypass, using IOMMU for PCI devices.
1662 io7= [HW] IO7 for Marvel based alpha systems
1663 See comment before marvel_specify_io7 in
1664 arch/alpha/kernel/core_marvel.c.
1666 io_delay= [X86] I/O delay method
1668 Standard port 0x80 based delay
1670 Alternate port 0xed based delay (needed on some systems)
1672 Simple two microseconds delay
1677 See Documentation/filesystems/nfs/nfsroot.txt.
1680 When an interrupt is not handled search all handlers
1681 for it. Intended to get systems with badly broken
1685 When an interrupt is not handled search all handlers
1686 for it. Also check all handlers each timer
1687 interrupt. Intended to get systems with badly broken
1691 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
1693 isolcpus= [KNL,SMP] Isolate CPUs from the general scheduler.
1695 <cpu number>,...,<cpu number>
1697 <cpu number>-<cpu number>
1698 (must be a positive range in ascending order)
1700 <cpu number>,...,<cpu number>-<cpu number>
1702 This option can be used to specify one or more CPUs
1703 to isolate from the general SMP balancing and scheduling
1704 algorithms. You can move a process onto or off an
1705 "isolated" CPU via the CPU affinity syscalls or cpuset.
1706 <cpu number> begins at 0 and the maximum value is
1707 "number of CPUs in system - 1".
1709 This option is the preferred way to isolate CPUs. The
1710 alternative -- manually setting the CPU mask of all
1711 tasks in the system -- can cause problems and
1712 suboptimal load balancer performance.
1716 ivrs_ioapic [HW,X86_64]
1717 Provide an override to the IOAPIC-ID<->DEVICE-ID
1718 mapping provided in the IVRS ACPI table. For
1719 example, to map IOAPIC-ID decimal 10 to
1720 PCI device 00:14.0 write the parameter as:
1721 ivrs_ioapic[10]=00:14.0
1723 ivrs_hpet [HW,X86_64]
1724 Provide an override to the HPET-ID<->DEVICE-ID
1725 mapping provided in the IVRS ACPI table. For
1726 example, to map HPET-ID decimal 0 to
1727 PCI device 00:14.0 write the parameter as:
1728 ivrs_hpet[0]=00:14.0
1730 js= [HW,JOY] Analog joystick
1731 See Documentation/input/joystick.txt.
1734 Enable/disable kernel and module base offset ASLR
1735 (Address Space Layout Randomization) if built into
1736 the kernel. When CONFIG_HIBERNATION is selected,
1737 kASLR is disabled by default. When kASLR is enabled,
1738 hibernation will be disabled.
1742 kernelcore=nn[KMG] [KNL,X86,IA-64,PPC] This parameter
1743 specifies the amount of memory usable by the kernel
1744 for non-movable allocations. The requested amount is
1745 spread evenly throughout all nodes in the system. The
1746 remaining memory in each node is used for Movable
1747 pages. In the event, a node is too small to have both
1748 kernelcore and Movable pages, kernelcore pages will
1749 take priority and other nodes will have a larger number
1750 of Movable pages. The Movable zone is used for the
1751 allocation of pages that may be reclaimed or moved
1752 by the page migration subsystem. This means that
1753 HugeTLB pages may not be allocated from this zone.
1754 Note that allocations like PTEs-from-HighMem still
1755 use the HighMem zone if it exists, and the Normal
1756 zone if it does not.
1758 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
1759 Format: <Controller#>[,poll interval]
1760 The controller # is the number of the ehci usb debug
1761 port as it is probed via PCI. The poll interval is
1762 optional and is the number seconds in between
1763 each poll cycle to the debug port in case you need
1764 the functionality for interrupting the kernel with
1765 gdb or control-c on the dbgp connection. When
1766 not using this parameter you use sysrq-g to break into
1767 the kernel debugger.
1769 kgdboc= [KGDB,HW] kgdb over consoles.
1770 Requires a tty driver that supports console polling,
1771 or a supported polling keyboard driver (non-usb).
1772 Serial only format: <serial_device>[,baud]
1773 keyboard only format: kbd
1774 keyboard and serial format: kbd,<serial_device>[,baud]
1775 Optional Kernel mode setting:
1776 kms, kbd format: kms,kbd
1777 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
1779 kgdbwait [KGDB] Stop kernel execution and enter the
1780 kernel debugger at the earliest opportunity.
1782 kmac= [MIPS] korina ethernet MAC address.
1783 Configure the RouterBoard 532 series on-chip
1784 Ethernet adapter MAC address.
1786 kmemleak= [KNL] Boot-time kmemleak enable/disable
1787 Valid arguments: on, off
1789 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
1792 kmemcheck= [X86] Boot-time kmemcheck enable/disable/one-shot mode
1793 Valid arguments: 0, 1, 2
1794 kmemcheck=0 (disabled)
1795 kmemcheck=1 (enabled)
1796 kmemcheck=2 (one-shot mode)
1797 Default: 2 (one-shot mode)
1799 kstack=N [X86] Print N words from the kernel stack
1802 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
1803 Default is 0 (don't ignore, but inject #GP)
1805 kvm.mmu_audit= [KVM] This is a R/W parameter which allows audit
1809 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM.
1810 Default is 1 (enabled)
1812 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU)
1814 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode.
1816 kvm-intel.ept= [KVM,Intel] Disable extended page tables
1817 (virtualized MMU) support on capable Intel chips.
1818 Default is 1 (enabled)
1820 kvm-intel.emulate_invalid_guest_state=
1821 [KVM,Intel] Enable emulation of invalid guest states
1822 Default is 0 (disabled)
1824 kvm-intel.flexpriority=
1825 [KVM,Intel] Disable FlexPriority feature (TPR shadow).
1826 Default is 1 (enabled)
1829 [KVM,Intel] Enable VMX nesting (nVMX).
1830 Default is 0 (disabled)
1832 kvm-intel.unrestricted_guest=
1833 [KVM,Intel] Disable unrestricted guest feature
1834 (virtualized real and unpaged mode) on capable
1835 Intel chips. Default is 1 (enabled)
1837 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification
1838 feature (tagged TLBs) on capable Intel chips.
1839 Default is 1 (enabled)
1845 lapic [X86-32,APIC] Enable the local APIC even if BIOS
1848 lapic= [x86,APIC] "notscdeadline" Do not use TSC deadline
1849 value for LAPIC timer one-shot implementation. Default
1850 back to the programmable timer unit in the LAPIC.
1852 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
1855 libata.dma= [LIBATA] DMA control
1856 libata.dma=0 Disable all PATA and SATA DMA
1857 libata.dma=1 PATA and SATA Disk DMA only
1858 libata.dma=2 ATAPI (CDROM) DMA only
1859 libata.dma=4 Compact Flash DMA only
1860 Combinations also work, so libata.dma=3 enables DMA
1861 for disks and CDROMs, but not CFs.
1863 libata.ignore_hpa= [LIBATA] Ignore HPA limit
1864 libata.ignore_hpa=0 keep BIOS limits (default)
1865 libata.ignore_hpa=1 ignore limits, using full disk
1867 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
1871 libata.force= [LIBATA] Force configurations. The format is comma
1872 separated list of "[ID:]VAL" where ID is
1873 PORT[.DEVICE]. PORT and DEVICE are decimal numbers
1874 matching port, link or device. Basically, it matches
1875 the ATA ID string printed on console by libata. If
1876 the whole ID part is omitted, the last PORT and DEVICE
1877 values are used. If ID hasn't been specified yet, the
1878 configuration applies to all ports, links and devices.
1880 If only DEVICE is omitted, the parameter applies to
1881 the port and all links and devices behind it. DEVICE
1882 number of 0 either selects the first device or the
1883 first fan-out link behind PMP device. It does not
1884 select the host link. DEVICE number of 15 selects the
1885 host link and device attached to it.
1887 The VAL specifies the configuration to force. As long
1888 as there's no ambiguity shortcut notation is allowed.
1889 For example, both 1.5 and 1.5G would work for 1.5Gbps.
1890 The following configurations can be forced.
1892 * Cable type: 40c, 80c, short40c, unk, ign or sata.
1893 Any ID with matching PORT is used.
1895 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
1897 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
1898 udma[/][16,25,33,44,66,100,133] notation is also
1901 * [no]ncq: Turn on or off NCQ.
1903 * [no]ncqtrim: Turn off queued DSM TRIM.
1905 * nohrst, nosrst, norst: suppress hard, soft
1908 * rstonce: only attempt one reset during
1909 hot-unplug link recovery
1911 * dump_id: dump IDENTIFY data.
1913 * atapi_dmadir: Enable ATAPI DMADIR bridge support
1915 * disable: Disable this device.
1917 If there are multiple matching configurations changing
1918 the same attribute, the last one is used.
1920 memblock=debug [KNL] Enable memblock debug messages.
1922 load_ramdisk= [RAM] List of ramdisks to load from floppy
1923 See Documentation/blockdev/ramdisk.txt.
1925 lockd.nlm_grace_period=P [NFS] Assign grace period.
1928 lockd.nlm_tcpport=N [NFS] Assign TCP port.
1931 lockd.nlm_timeout=T [NFS] Assign timeout value.
1934 lockd.nlm_udpport=M [NFS] Assign UDP port.
1937 locktorture.nreaders_stress= [KNL]
1938 Set the number of locking read-acquisition kthreads.
1939 Defaults to being automatically set based on the
1940 number of online CPUs.
1942 locktorture.nwriters_stress= [KNL]
1943 Set the number of locking write-acquisition kthreads.
1945 locktorture.onoff_holdoff= [KNL]
1946 Set time (s) after boot for CPU-hotplug testing.
1948 locktorture.onoff_interval= [KNL]
1949 Set time (s) between CPU-hotplug operations, or
1950 zero to disable CPU-hotplug testing.
1952 locktorture.shuffle_interval= [KNL]
1953 Set task-shuffle interval (jiffies). Shuffling
1954 tasks allows some CPUs to go into dyntick-idle
1955 mode during the locktorture test.
1957 locktorture.shutdown_secs= [KNL]
1958 Set time (s) after boot system shutdown. This
1959 is useful for hands-off automated testing.
1961 locktorture.stat_interval= [KNL]
1962 Time (s) between statistics printk()s.
1964 locktorture.stutter= [KNL]
1965 Time (s) to stutter testing, for example,
1966 specifying five seconds causes the test to run for
1967 five seconds, wait for five seconds, and so on.
1968 This tests the locking primitive's ability to
1969 transition abruptly to and from idle.
1971 locktorture.torture_runnable= [BOOT]
1972 Start locktorture running at boot time.
1974 locktorture.torture_type= [KNL]
1975 Specify the locking implementation to test.
1977 locktorture.verbose= [KNL]
1978 Enable additional printk() statements.
1980 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
1983 loglevel= All Kernel Messages with a loglevel smaller than the
1984 console loglevel will be printed to the console. It can
1985 also be changed with klogd or other programs. The
1986 loglevels are defined as follows:
1988 0 (KERN_EMERG) system is unusable
1989 1 (KERN_ALERT) action must be taken immediately
1990 2 (KERN_CRIT) critical conditions
1991 3 (KERN_ERR) error conditions
1992 4 (KERN_WARNING) warning conditions
1993 5 (KERN_NOTICE) normal but significant condition
1994 6 (KERN_INFO) informational
1995 7 (KERN_DEBUG) debug-level messages
1997 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
1998 in bytes. n must be a power of two and greater
1999 than the minimal size. The minimal size is defined
2000 by LOG_BUF_SHIFT kernel config parameter. There is
2001 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2002 that allows to increase the default size depending on
2003 the number of CPUs. See init/Kconfig for more details.
2005 logo.nologo [FB] Disables display of the built-in Linux logo.
2006 This may be used to provide more screen space for
2007 kernel log messages and is useful when debugging
2008 kernel boot problems.
2010 lp=0 [LP] Specify parallel ports to use, e.g,
2011 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2012 lp=reset first parallel port). 'lp=0' disables the
2013 lp=auto printer driver. 'lp=reset' (which can be
2014 specified in addition to the ports) causes
2015 attached printers to be reset. Using
2016 lp=port1,port2,... specifies the parallel ports
2017 to associate lp devices with, starting with
2018 lp0. A port specification may be 'none' to skip
2019 that lp device, or a parport name such as
2020 'parport0'. Specifying 'lp=auto' instead of a
2021 port specification list means that device IDs
2022 from each port should be examined, to see if
2023 an IEEE 1284-compliant printer is attached; if
2024 so, the driver will manage that printer.
2025 See also header of drivers/char/lp.c.
2028 Sets loops_per_jiffy to given constant, thus avoiding
2029 time-consuming boot-time autodetection (up to 250 ms per
2030 CPU). 0 enables autodetection (default). To determine
2031 the correct value for your kernel, boot with normal
2032 autodetection and see what value is printed. Note that
2033 on SMP systems the preset will be applied to all CPUs,
2034 which is likely to cause problems if your CPUs need
2035 significantly divergent settings. An incorrect value
2036 will cause delays in the kernel to be wrong, leading to
2037 unpredictable I/O errors and other breakage. Although
2038 unlikely, in the extreme case this might damage your
2042 Format: <io>,<irq>,<dma>
2044 machvec= [IA-64] Force the use of a particular machine-vector
2045 (machvec) in a generic kernel.
2046 Example: machvec=hpzx1_swiotlb
2048 machtype= [Loongson] Share the same kernel image file between different
2050 Example: machtype=lemote-yeeloong-2f-7inch
2052 max_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory greater
2053 than or equal to this physical address is ignored.
2055 maxcpus= [SMP] Maximum number of processors that an SMP kernel
2056 should make use of. maxcpus=n : n >= 0 limits the
2057 kernel to using 'n' processors. n=0 is a special case,
2058 it is equivalent to "nosmp", which also disables
2061 max_loop= [LOOP] The number of loop block devices that get
2062 (loop.max_loop) unconditionally pre-created at init time. The default
2063 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
2064 of statically allocating a predefined number, loop
2065 devices can be requested on-demand with the
2066 /dev/loop-control interface.
2068 mce [X86-32] Machine Check Exception
2070 mce=option [X86-64] See Documentation/x86/x86_64/boot-options.txt
2072 md= [HW] RAID subsystems devices and level
2073 See Documentation/md.txt.
2076 Format: <first>,<last>
2077 Specifies range of consoles to be captured by the MDA.
2080 Control mitigation for the Micro-architectural Data
2081 Sampling (MDS) vulnerability.
2083 Certain CPUs are vulnerable to an exploit against CPU
2084 internal buffers which can forward information to a
2085 disclosure gadget under certain conditions.
2087 In vulnerable processors, the speculatively
2088 forwarded data can be used in a cache side channel
2089 attack, to access data to which the attacker does
2090 not have direct access.
2092 This parameter controls the MDS mitigation. The
2095 full - Enable MDS mitigation on vulnerable CPUs
2096 off - Unconditionally disable MDS mitigation
2098 Not specifying this option is equivalent to
2101 For details see: Documentation/hw-vuln/mds.rst
2103 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
2104 Amount of memory to be used when the kernel is not able
2105 to see the whole system memory or for test.
2106 [X86] Work as limiting max address. Use together
2107 with memmap= to avoid physical address space collisions.
2108 Without memmap= PCI devices could be placed at addresses
2109 belonging to unused RAM.
2111 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
2115 [KNL,SH] Allow user to override the default size for
2116 per-device physically contiguous DMA buffers.
2118 memmap=exactmap [KNL,X86] Enable setting of an exact
2119 E820 memory map, as specified by the user.
2120 Such memmap=exactmap lines can be constructed based on
2121 BIOS output or other requirements. See the memmap=nn@ss
2124 memmap=nn[KMG]@ss[KMG]
2125 [KNL] Force usage of a specific region of memory.
2126 Region of memory to be used is from ss to ss+nn.
2128 memmap=nn[KMG]#ss[KMG]
2129 [KNL,ACPI] Mark specific memory as ACPI data.
2130 Region of memory to be marked is from ss to ss+nn.
2132 memmap=nn[KMG]$ss[KMG]
2133 [KNL,ACPI] Mark specific memory as reserved.
2134 Region of memory to be reserved is from ss to ss+nn.
2135 Example: Exclude memory from 0x18690000-0x1869ffff
2136 memmap=64K$0x18690000
2138 memmap=0x10000$0x18690000
2140 memmap=nn[KMG]!ss[KMG]
2141 [KNL,X86] Mark specific memory as protected.
2142 Region of memory to be used, from ss to ss+nn.
2143 The memory region may be marked as e820 type 12 (0xc)
2144 and is NVDIMM or ADR memory.
2146 memory_corruption_check=0/1 [X86]
2147 Some BIOSes seem to corrupt the first 64k of
2148 memory when doing things like suspend/resume.
2149 Setting this option will scan the memory
2150 looking for corruption. Enabling this will
2151 both detect corruption and prevent the kernel
2152 from using the memory being corrupted.
2153 However, its intended as a diagnostic tool; if
2154 repeatable BIOS-originated corruption always
2155 affects the same memory, you can use memmap=
2156 to prevent the kernel from using that memory.
2158 memory_corruption_check_size=size [X86]
2159 By default it checks for corruption in the low
2160 64k, making this memory unavailable for normal
2161 use. Use this parameter to scan for
2162 corruption in more or less memory.
2164 memory_corruption_check_period=seconds [X86]
2165 By default it checks for corruption every 60
2166 seconds. Use this parameter to check at some
2167 other rate. 0 disables periodic checking.
2169 memtest= [KNL,X86,ARM] Enable memtest
2171 default : 0 <disable>
2172 Specifies the number of memtest passes to be
2173 performed. Each pass selects another test
2174 pattern from a given set of patterns. Memtest
2175 fills the memory with this pattern, validates
2176 memory contents and reserves bad memory
2177 regions that are detected.
2179 meye.*= [HW] Set MotionEye Camera parameters
2180 See Documentation/video4linux/meye.txt.
2182 mfgpt_irq= [IA-32] Specify the IRQ to use for the
2183 Multi-Function General Purpose Timers on AMD Geode
2186 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
2187 the BIOS has incorrectly applied a workaround. TinyBIOS
2188 version 0.98 is known to be affected, 0.99 fixes the
2189 problem by letting the user disable the workaround.
2193 min_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory below this
2194 physical address is ignored.
2196 mini2440= [ARM,HW,KNL]
2197 Format:[0..2][b][c][t]
2199 MINI2440 configuration specification:
2200 0 - The attached screen is the 3.5" TFT
2201 1 - The attached screen is the 7" TFT
2202 2 - The VGA Shield is attached (1024x768)
2203 Leaving out the screen size parameter will not load
2204 the TFT driver, and the framebuffer will be left
2206 b - Enable backlight. The TFT backlight pin will be
2207 linked to the kernel VESA blanking code and a GPIO
2208 LED. This parameter is not necessary when using the
2210 c - Enable the s3c camera interface.
2211 t - Reserved for enabling touchscreen support. The
2212 touchscreen support is not enabled in the mainstream
2213 kernel as of 2.6.30, a preliminary port can be found
2214 in the "bleeding edge" mini2440 support kernel at
2215 http://repo.or.cz/w/linux-2.6/mini2440.git
2218 [X86] Control optional mitigations for CPU
2219 vulnerabilities. This is a set of curated,
2220 arch-independent options, each of which is an
2221 aggregation of existing arch-specific options.
2224 Disable all optional CPU mitigations. This
2225 improves system performance, but it may also
2226 expose users to several CPU vulnerabilities.
2227 Equivalent to: nopti [X86]
2230 spectre_v2_user=off [X86]
2231 spec_store_bypass_disable=off [X86]
2233 tsx_async_abort=off [X86]
2236 Mitigate all CPU vulnerabilities, but leave SMT
2237 enabled, even if it's vulnerable. This is for
2238 users who don't want to be surprised by SMT
2239 getting disabled across kernel upgrades, or who
2240 have other ways of avoiding SMT-based attacks.
2241 Equivalent to: (default behavior)
2244 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
2245 parameter allows control of the logging verbosity for
2246 the additional memory initialisation checks. A value
2247 of 0 disables mminit logging and a level of 4 will
2248 log everything. Information is printed at KERN_DEBUG
2249 so loglevel=8 may also need to be specified.
2252 [KNL] When CONFIG_MODULE_SIG is set, this means that
2253 modules without (valid) signatures will fail to load.
2254 Note that if CONFIG_MODULE_SIG_FORCE is set, that
2255 is always true, so this option does nothing.
2258 [MOUSE] Maximum time between finger touching and
2259 leaving touchpad surface for touch to be considered
2260 a tap and be reported as a left button click (for
2261 touchpads working in absolute mode only).
2263 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
2264 reporting absolute coordinates, such as tablets
2265 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
2266 reporting absolute coordinates, such as tablets
2268 movablecore=nn[KMG] [KNL,X86,IA-64,PPC] This parameter
2269 is similar to kernelcore except it specifies the
2270 amount of memory used for migratable allocations.
2271 If both kernelcore and movablecore is specified,
2272 then kernelcore will be at *least* the specified
2273 value but may be more. If movablecore on its own
2274 is specified, the administrator must be careful
2275 that the amount of memory usable for all allocations
2278 movable_node [KNL,X86] Boot-time switch to enable the effects
2279 of CONFIG_MOVABLE_NODE=y. See mm/Kconfig for details.
2281 MTD_Partition= [MTD]
2282 Format: <name>,<region-number>,<size>,<offset>
2284 MTD_Region= [MTD] Format:
2285 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
2288 See drivers/mtd/cmdlinepart.c.
2290 multitce=off [PPC] This parameter disables the use of the pSeries
2291 firmware feature for updating multiple TCE entries
2294 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
2296 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
2298 boundary - index of last SLC block on Flex-OneNAND.
2299 The remaining blocks are configured as MLC blocks.
2300 lock - Configure if Flex-OneNAND boundary should be locked.
2301 Once locked, the boundary cannot be changed.
2302 1 indicates lock status, 0 indicates unlock status.
2305 ARM/S3C2412 JIVE boot control
2307 See arch/arm/mach-s3c2412/mach-jive.c
2309 mtouchusb.raw_coordinates=
2310 [HW] Make the MicroTouch USB driver use raw coordinates
2311 ('y', default) or cooked coordinates ('n')
2313 mtrr_chunk_size=nn[KMG] [X86]
2314 used for mtrr cleanup. It is largest continuous chunk
2315 that could hold holes aka. UC entries.
2317 mtrr_gran_size=nn[KMG] [X86]
2318 Used for mtrr cleanup. It is granularity of mtrr block.
2320 Large value could prevent small alignment from
2323 mtrr_spare_reg_nr=n [X86]
2325 Range: 0,7 : spare reg number
2327 Used for mtrr cleanup. It is spare mtrr entries number.
2328 Set to 2 or more if your graphical card needs more.
2330 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
2332 netdev= [NET] Network devices parameters
2333 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
2334 Note that mem_start is often overloaded to mean
2335 something different and driver-specific.
2336 This usage is only documented in each driver source
2340 [NETFILTER] Enable connection tracking flow accounting
2341 0 to disable accounting
2342 1 to enable accounting
2345 nfsaddrs= [NFS] Deprecated. Use ip= instead.
2346 See Documentation/filesystems/nfs/nfsroot.txt.
2348 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
2349 See Documentation/filesystems/nfs/nfsroot.txt.
2351 nfsrootdebug [NFS] enable nfsroot debugging messages.
2352 See Documentation/filesystems/nfs/nfsroot.txt.
2354 nfs.callback_tcpport=
2355 [NFS] set the TCP port on which the NFSv4 callback
2356 channel should listen.
2359 [NFS] sets the pathname to the program which is used
2360 to update the NFS client cache entries.
2362 nfs.cache_getent_timeout=
2363 [NFS] sets the timeout after which an attempt to
2364 update a cache entry is deemed to have failed.
2366 nfs.idmap_cache_timeout=
2367 [NFS] set the maximum lifetime for idmapper cache
2371 [NFS] enable 64-bit inode numbers.
2372 If zero, the NFS client will fake up a 32-bit inode
2373 number for the readdir() and stat() syscalls instead
2374 of returning the full 64-bit number.
2375 The default is to return 64-bit inode numbers.
2377 nfs.max_session_slots=
2378 [NFSv4.1] Sets the maximum number of session slots
2379 the client will attempt to negotiate with the server.
2380 This limits the number of simultaneous RPC requests
2381 that the client can send to the NFSv4.1 server.
2382 Note that there is little point in setting this
2383 value higher than the max_tcp_slot_table_limit.
2385 nfs.nfs4_disable_idmapping=
2386 [NFSv4] When set to the default of '1', this option
2387 ensures that both the RPC level authentication
2388 scheme and the NFS level operations agree to use
2389 numeric uids/gids if the mount is using the
2390 'sec=sys' security flavour. In effect it is
2391 disabling idmapping, which can make migration from
2392 legacy NFSv2/v3 systems to NFSv4 easier.
2393 Servers that do not support this mode of operation
2394 will be autodetected by the client, and it will fall
2395 back to using the idmapper.
2396 To turn off this behaviour, set the value to '0'.
2398 [NFS4] Specify an additional fixed unique ident-
2399 ification string that NFSv4 clients can insert into
2400 their nfs_client_id4 string. This is typically a
2401 UUID that is generated at system install time.
2403 nfs.send_implementation_id =
2404 [NFSv4.1] Send client implementation identification
2405 information in exchange_id requests.
2406 If zero, no implementation identification information
2408 The default is to send the implementation identification
2411 nfs.recover_lost_locks =
2412 [NFSv4] Attempt to recover locks that were lost due
2413 to a lease timeout on the server. Please note that
2414 doing this risks data corruption, since there are
2415 no guarantees that the file will remain unchanged
2416 after the locks are lost.
2417 If you want to enable the kernel legacy behaviour of
2418 attempting to recover these locks, then set this
2420 The default parameter value of '0' causes the kernel
2421 not to attempt recovery of lost locks.
2423 nfs4.layoutstats_timer =
2424 [NFSv4.2] Change the rate at which the kernel sends
2425 layoutstats to the pNFS metadata server.
2427 Setting this to value to 0 causes the kernel to use
2428 whatever value is the default set by the layout
2429 driver. A non-zero value sets the minimum interval
2430 in seconds between layoutstats transmissions.
2432 nfsd.nfs4_disable_idmapping=
2433 [NFSv4] When set to the default of '1', the NFSv4
2434 server will return only numeric uids and gids to
2435 clients using auth_sys, and will accept numeric uids
2436 and gids from such clients. This is intended to ease
2437 migration from NFSv2/v3.
2439 objlayoutdriver.osd_login_prog=
2440 [NFS] [OBJLAYOUT] sets the pathname to the program which
2441 is used to automatically discover and login into new
2442 osd-targets. Please see:
2443 Documentation/filesystems/pnfs.txt for more explanations
2445 nmi_debug= [KNL,AVR32,SH] Specify one or more actions to take
2446 when a NMI is triggered.
2447 Format: [state][,regs][,debounce][,die]
2449 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
2450 Format: [panic,][nopanic,][num]
2452 0 - turn hardlockup detector in nmi_watchdog off
2453 1 - turn hardlockup detector in nmi_watchdog on
2454 When panic is specified, panic when an NMI watchdog
2455 timeout occurs (or 'nopanic' to override the opposite
2456 default). To disable both hard and soft lockup detectors,
2457 please see 'nowatchdog'.
2458 This is useful when you use a panic=... timeout and
2459 need the box quickly up again.
2461 netpoll.carrier_timeout=
2462 [NET] Specifies amount of time (in seconds) that
2463 netpoll should wait for a carrier. By default netpoll
2466 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
2467 emulation library even if a 387 maths coprocessor
2471 [HW] Never suspend the console
2472 Disable suspending of consoles during suspend and
2473 hibernate operations. Once disabled, debugging
2474 messages can reach various consoles while the rest
2475 of the system is being put to sleep (ie, while
2476 debugging driver suspend/resume hooks). This may
2477 not work reliably with all consoles, but is known
2478 to work with serial and VGA consoles.
2479 To facilitate more flexible debugging, we also add
2480 console_suspend, a printk module parameter to control
2481 it. Users could use console_suspend (usually
2482 /sys/module/printk/parameters/console_suspend) to
2483 turn on/off it dynamically.
2485 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
2486 caches in the slab allocator. Saves per-node memory,
2487 but will impact performance.
2491 noaltinstr [S390] Disables alternative instructions patching
2492 (CPU alternatives feature).
2494 noapic [SMP,APIC] Tells the kernel to not make use of any
2495 IOAPICs that may be present in the system.
2497 noautogroup Disable scheduler automatic task group creation.
2499 nobats [PPC] Do not use BATs for mapping kernel lowmem
2500 on "Classic" PPC cores.
2504 noclflush [BUGS=X86] Don't use the CLFLUSH instruction
2506 nodelayacct [KNL] Disable per-task delay accounting
2508 nodisconnect [HW,SCSI,M68K] Disables SCSI disconnects.
2510 nodsp [SH] Disable hardware DSP at boot time.
2512 noefi Disable EFI runtime services support.
2517 On X86-32 available only on PAE configured kernels.
2518 noexec=on: enable non-executable mappings (default)
2519 noexec=off: disable non-executable mappings
2522 Force indicating stack and heap as non-executable or
2523 executable regardless of PT_GNU_STACK entry or CPU XI
2524 (execute inhibit) support. Valid valuess are: on, off.
2525 noexec=on: force indicating non-executable
2527 noexec=off: force indicating executable
2529 If this parameter is omitted, stack and heap will be
2530 indicated non-executable or executable as they are
2531 actually set up, which depends on PT_GNU_STACK entry
2532 and possibly other factors (for instance, CPU XI
2534 NOTE: Using noexec=on on a system without CPU XI
2535 support is not recommended since there is no actual
2536 HW support that provide non-executable stack/heap.
2537 Use only for debugging purposes and not in a
2538 production environment.
2541 Disable SMAP (Supervisor Mode Access Prevention)
2542 even if it is supported by processor.
2545 Disable SMEP (Supervisor Mode Execution Prevention)
2546 even if it is supported by processor.
2549 This affects only 32-bit executables.
2550 noexec32=on: enable non-executable mappings (default)
2551 read doesn't imply executable mappings
2552 noexec32=off: disable non-executable mappings
2553 read implies executable mappings
2555 nofpu [MIPS,SH] Disable hardware FPU at boot time.
2557 nofxsr [BUGS=X86-32] Disables x86 floating point extended
2558 register save and restore. The kernel will only save
2559 legacy floating-point registers on task switch.
2561 nohugeiomap [KNL,x86] Disable kernel huge I/O mappings.
2563 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1
2564 (bounds check bypass). With this option data leaks are
2565 possible in the system.
2567 nospectre_v2 [X86,PPC_FSL_BOOK3E] Disable all mitigations for the Spectre variant 2
2568 (indirect branch prediction) vulnerability. System may
2569 allow data leaks with this option, which is equivalent
2572 nospec_store_bypass_disable
2573 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
2575 noxsave [BUGS=X86] Disables x86 extended register state save
2576 and restore using xsave. The kernel will fallback to
2577 enabling legacy floating-point and sse state.
2579 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
2580 register states. The kernel will fall back to use
2581 xsave to save the states. By using this parameter,
2582 performance of saving the states is degraded because
2583 xsave doesn't support modified optimization while
2584 xsaveopt supports it on xsaveopt enabled systems.
2586 noxsaves [X86] Disables xsaves and xrstors used in saving and
2587 restoring x86 extended register state in compacted
2588 form of xsave area. The kernel will fall back to use
2589 xsaveopt and xrstor to save and restore the states
2590 in standard form of xsave area. By using this
2591 parameter, xsave area per process might occupy more
2592 memory on xsaves enabled systems.
2594 nohlt [BUGS=ARM,SH] Tells the kernel that the sleep(SH) or
2595 wfi(ARM) instruction doesn't work correctly and not to
2596 use it. This is also useful when using JTAG debugger.
2598 no_file_caps Tells the kernel not to honor file capabilities. The
2599 only way then for a file to be executed with privilege
2600 is to be setuid root or executed by root.
2602 nohalt [IA-64] Tells the kernel not to use the power saving
2603 function PAL_HALT_LIGHT when idle. This increases
2604 power-consumption. On the positive side, it reduces
2605 interrupt wake-up latency, which may improve performance
2606 in certain environments such as networked servers or
2609 nohibernate [HIBERNATION] Disable hibernation and resume.
2611 nohz= [KNL] Boottime enable/disable dynamic ticks
2612 Valid arguments: on, off
2615 nohz_full= [KNL,BOOT]
2616 In kernels built with CONFIG_NO_HZ_FULL=y, set
2617 the specified list of CPUs whose tick will be stopped
2618 whenever possible. The boot CPU will be forced outside
2619 the range to maintain the timekeeping.
2620 The CPUs in this range must also be included in the
2623 noiotrap [SH] Disables trapped I/O port accesses.
2625 noirqdebug [X86-32] Disables the code which attempts to detect and
2626 disable unhandled interrupt sources.
2628 no_timer_check [X86,APIC] Disables the code which tests for
2629 broken timer IRQ sources.
2631 noisapnp [ISAPNP] Disables ISA PnP code.
2633 noinitrd [RAM] Tells the kernel not to load any configured
2636 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
2638 [Deprecated - use intremap=off]
2642 noinvpcid [X86] Disable the INVPCID cpu feature.
2644 nojitter [IA-64] Disables jitter checking for ITC timers.
2646 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
2648 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
2651 no-steal-acc [X86,KVM] Disable paravirtualized steal time accounting.
2652 steal time is computed, but won't influence scheduler
2655 nolapic [X86-32,APIC] Do not enable or use the local APIC.
2657 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
2659 noltlbs [PPC] Do not use large page/tlb entries for kernel
2660 lowmem mapping on PPC40x.
2662 nomca [IA-64] Disable machine check abort handling
2664 nomce [X86-32] Disable Machine Check Exception
2666 nomfgpt [X86-32] Disable Multi-Function General Purpose
2667 Timer usage (for AMD Geode machines).
2669 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
2670 shutdown the other cpus. Instead use the REBOOT_VECTOR
2673 nomodule Disable module load
2675 nopat [X86] Disable PAT (page attribute table extension of
2676 pagetables) support.
2678 nopcid [X86-64] Disable the PCID cpu feature.
2680 norandmaps Don't use address space randomization. Equivalent to
2681 echo 0 > /proc/sys/kernel/randomize_va_space
2683 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
2684 with UP alternatives
2686 nordrand [X86] Disable kernel use of the RDRAND and
2687 RDSEED instructions even if they are supported
2688 by the processor. RDRAND and RDSEED are still
2689 available to user space applications.
2691 noresume [SWSUSP] Disables resume and restores original swap
2694 no-scroll [VGA] Disables scrollback.
2695 This is required for the Braillex ib80-piezo Braille
2696 reader made by F.H. Papenmeier (Germany).
2700 nosep [BUGS=X86-32] Disables x86 SYSENTER/SYSEXIT support.
2702 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
2703 and disable the IO APIC. legacy for "maxcpus=0".
2705 nosoftlockup [KNL] Disable the soft-lockup detector.
2707 nosync [HW,M68K] Disables sync negotiation for all devices.
2709 notsc [BUGS=X86-32] Disable Time Stamp Counter
2711 nousb [USB] Disable the USB subsystem
2713 nowatchdog [KNL] Disable both lockup detectors, i.e.
2714 soft-lockup and NMI watchdog (hard-lockup).
2718 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
2720 cpu0_hotplug [X86] Turn on CPU0 hotplug feature when
2721 CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off.
2722 Some features depend on CPU0. Known dependencies are:
2723 1. Resume from suspend/hibernate depends on CPU0.
2724 Suspend/hibernate will fail if CPU0 is offline and you
2725 need to online CPU0 before suspend/hibernate.
2726 2. PIC interrupts also depend on CPU0. CPU0 can't be
2727 removed if a PIC interrupt is detected.
2728 It's said poweroff/reboot may depend on CPU0 on some
2729 machines although I haven't seen such issues so far
2730 after CPU0 is offline on a few tested machines.
2731 If the dependencies are under your control, you can
2732 turn on cpu0_hotplug.
2734 nptcg= [IA-64] Override max number of concurrent global TLB
2735 purges which is reported from either PAL_VM_SUMMARY or
2738 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
2739 could support. nr_cpus=n : n >= 1 limits the kernel to
2740 supporting 'n' processors. Later in runtime you can not
2741 use hotplug cpu feature to put more cpu back to online.
2742 just like you compile the kernel NR_CPUS=n
2744 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
2746 numa_balancing= [KNL,X86] Enable or disable automatic NUMA balancing.
2747 Allowed values are enable and disable
2749 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
2750 one of ['zone', 'node', 'default'] can be specified
2751 This can be set from sysctl after boot.
2752 See Documentation/sysctl/vm.txt for details.
2754 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
2755 See Documentation/debugging-via-ohci1394.txt for more
2758 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
2759 Rather than timing out after 20 ms if an EC
2760 command is not properly ACKed, override the length
2761 of the timeout. We have interrupts disabled while
2762 waiting for the ACK, so if this is set too high
2763 interrupts *may* be lost!
2765 omap_mux= [OMAP] Override bootloader pin multiplexing.
2766 Format: <mux_mode0.mode_name=value>...
2767 For example, to override I2C bus2:
2768 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
2770 oprofile.timer= [HW]
2771 Use timer interrupt instead of performance counters
2773 oprofile.cpu_type= Force an oprofile cpu type
2774 This might be useful if you have an older oprofile
2775 userland or if you want common events.
2776 Format: { arch_perfmon }
2777 arch_perfmon: [X86] Force use of architectural
2778 perfmon on Intel CPUs instead of the
2779 CPU specific event set.
2780 timer: [X86] Force use of architectural NMI
2781 timer mode (see also oprofile.timer
2782 for generic hr timer mode)
2783 [s390] Force legacy basic mode sampling
2784 (report cpu_type "timer")
2786 oops=panic Always panic on oopses. Default is to just kill the
2787 process, but there is a small probability of
2788 deadlocking the machine.
2789 This will also cause panics on machine check exceptions.
2790 Useful together with panic=30 to trigger a reboot.
2793 See Documentation/sound/oss/oss-parameters.txt
2795 page_owner= [KNL] Boot-time page_owner enabling option.
2796 Storage of the information about who allocated
2797 each page is disabled in default. With this switch,
2799 on: enable the feature
2801 panic= [KNL] Kernel behaviour on panic: delay <timeout>
2802 timeout > 0: seconds before rebooting
2803 timeout = 0: wait forever
2804 timeout < 0: reboot immediately
2807 panic_on_warn panic() instead of WARN(). Useful to cause kdump
2810 crash_kexec_post_notifiers
2811 Run kdump after running panic-notifiers and dumping
2812 kmsg. This only for the users who doubt kdump always
2813 succeeds in any situation.
2814 Note that this also increases risks of kdump failure,
2815 because some panic notifiers can make the crashed
2816 kernel more unstable.
2818 parkbd.port= [HW] Parallel port number the keyboard adapter is
2819 connected to, default is 0.
2821 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
2822 0 for XT, 1 for AT (default is AT).
2825 parport= [HW,PPT] Specify parallel ports. 0 disables.
2826 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
2827 Use 'auto' to force the driver to use any
2828 IRQ/DMA settings detected (the default is to
2829 ignore detected IRQ/DMA settings because of
2830 possible conflicts). You can specify the base
2831 address, IRQ, and DMA settings; IRQ and DMA
2832 should be numbers, or 'auto' (for using detected
2833 settings on that particular port), or 'nofifo'
2834 (to avoid using a FIFO even if it is detected).
2835 Parallel ports are assigned in the order they
2836 are specified on the command line, starting
2839 parport_init_mode= [HW,PPT]
2840 Configure VIA parallel port to operate in
2841 a specific mode. This is necessary on Pegasos
2842 computer where firmware has no options for setting
2843 up parallel port mode and sets it to spp.
2844 Currently this function knows 686a and 8231 chips.
2845 Format: [spp|ps2|epp|ecp|ecpepp]
2848 Halt all CPUs after the first oops has been printed for
2849 the specified number of seconds. This is to be used if
2850 your oopses keep scrolling off the screen.
2855 See header of drivers/block/paride/pcd.c.
2856 See also Documentation/blockdev/paride.txt.
2858 pci=option[,option...] [PCI] various PCI subsystem options:
2859 earlydump [X86] dump PCI config space before the kernel
2861 off [X86] don't probe for the PCI bus
2862 bios [X86-32] force use of PCI BIOS, don't access
2863 the hardware directly. Use this if your machine
2864 has a non-standard PCI host bridge.
2865 nobios [X86-32] disallow use of PCI BIOS, only direct
2866 hardware access methods are allowed. Use this
2867 if you experience crashes upon bootup and you
2868 suspect they are caused by the BIOS.
2869 conf1 [X86] Force use of PCI Configuration
2871 conf2 [X86] Force use of PCI Configuration
2873 noaer [PCIE] If the PCIEAER kernel config parameter is
2874 enabled, this kernel boot option can be used to
2875 disable the use of PCIE advanced error reporting.
2876 nodomains [PCI] Disable support for multiple PCI
2877 root domains (aka PCI segments, in ACPI-speak).
2878 nommconf [X86] Disable use of MMCONFIG for PCI
2880 check_enable_amd_mmconf [X86] check for and enable
2881 properly configured MMIO access to PCI
2882 config space on AMD family 10h CPU
2883 nomsi [MSI] If the PCI_MSI kernel config parameter is
2884 enabled, this kernel boot option can be used to
2885 disable the use of MSI interrupts system-wide.
2886 noioapicquirk [APIC] Disable all boot interrupt quirks.
2887 Safety option to keep boot IRQs enabled. This
2888 should never be necessary.
2889 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
2890 primary IO-APIC for bridges that cannot disable
2891 boot IRQs. This fixes a source of spurious IRQs
2892 when the system masks IRQs.
2893 noioapicreroute [APIC] Disable workaround that uses the
2894 boot IRQ equivalent of an IRQ that connects to
2895 a chipset where boot IRQs cannot be disabled.
2896 The opposite of ioapicreroute.
2897 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
2898 routing table. These calls are known to be buggy
2899 on several machines and they hang the machine
2900 when used, but on other computers it's the only
2901 way to get the interrupt routing table. Try
2902 this option if the kernel is unable to allocate
2903 IRQs or discover secondary PCI buses on your
2905 rom [X86] Assign address space to expansion ROMs.
2906 Use with caution as certain devices share
2907 address decoders between ROMs and other
2909 norom [X86] Do not assign address space to
2910 expansion ROMs that do not already have
2911 BIOS assigned address ranges.
2912 nobar [X86] Do not assign address space to the
2913 BARs that weren't assigned by the BIOS.
2914 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
2915 assigned automatically to PCI devices. You can
2916 make the kernel exclude IRQs of your ISA cards
2918 pirqaddr=0xAAAAA [X86] Specify the physical address
2919 of the PIRQ table (normally generated
2920 by the BIOS) if it is outside the
2921 F0000h-100000h range.
2922 lastbus=N [X86] Scan all buses thru bus #N. Can be
2923 useful if the kernel is unable to find your
2924 secondary buses and you want to tell it
2925 explicitly which ones they are.
2926 assign-busses [X86] Always assign all PCI bus
2927 numbers ourselves, overriding
2928 whatever the firmware may have done.
2929 usepirqmask [X86] Honor the possible IRQ mask stored
2930 in the BIOS $PIR table. This is needed on
2931 some systems with broken BIOSes, notably
2932 some HP Pavilion N5400 and Omnibook XE3
2933 notebooks. This will have no effect if ACPI
2934 IRQ routing is enabled.
2935 noacpi [X86] Do not use ACPI for IRQ routing
2936 or for PCI scanning.
2937 use_crs [X86] Use PCI host bridge window information
2938 from ACPI. On BIOSes from 2008 or later, this
2939 is enabled by default. If you need to use this,
2940 please report a bug.
2941 nocrs [X86] Ignore PCI host bridge windows from ACPI.
2942 If you need to use this, please report a bug.
2943 routeirq Do IRQ routing for all PCI devices.
2944 This is normally done in pci_enable_device(),
2945 so this option is a temporary workaround
2946 for broken drivers that don't call it.
2947 skip_isa_align [X86] do not align io start addr, so can
2948 handle more pci cards
2949 firmware [ARM] Do not re-enumerate the bus but instead
2950 just use the configuration from the
2951 bootloader. This is currently used on
2952 IXP2000 systems where the bus has to be
2953 configured a certain way for adjunct CPUs.
2954 noearly [X86] Don't do any early type 1 scanning.
2955 This might help on some broken boards which
2956 machine check when some devices' config space
2957 is read. But various workarounds are disabled
2958 and some IOMMU drivers will not work.
2959 bfsort Sort PCI devices into breadth-first order.
2960 This sorting is done to get a device
2961 order compatible with older (<= 2.4) kernels.
2962 nobfsort Don't sort PCI devices into breadth-first order.
2963 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
2964 tuning and use the BIOS-configured MPS defaults.
2965 pcie_bus_safe Set every device's MPS to the largest value
2966 supported by all devices below the root complex.
2967 pcie_bus_perf Set device MPS to the largest allowable MPS
2968 based on its parent bus. Also set MRRS (Max
2969 Read Request Size) to the largest supported
2970 value (no larger than the MPS that the device
2971 or bus can support) for best performance.
2972 pcie_bus_peer2peer Set every device's MPS to 128B, which
2973 every device is guaranteed to support. This
2974 configuration allows peer-to-peer DMA between
2975 any pair of devices, possibly at the cost of
2976 reduced performance. This also guarantees
2977 that hot-added devices will work.
2978 cbiosize=nn[KMG] The fixed amount of bus space which is
2979 reserved for the CardBus bridge's IO window.
2980 The default value is 256 bytes.
2981 cbmemsize=nn[KMG] The fixed amount of bus space which is
2982 reserved for the CardBus bridge's memory
2983 window. The default value is 64 megabytes.
2986 [<order of align>@][<domain>:]<bus>:<slot>.<func>[; ...]
2987 Specifies alignment and device to reassign
2988 aligned memory resources.
2989 If <order of align> is not specified,
2990 PAGE_SIZE is used as alignment.
2991 PCI-PCI bridge can be specified, if resource
2992 windows need to be expanded.
2993 ecrc= Enable/disable PCIe ECRC (transaction layer
2994 end-to-end CRC checking).
2995 bios: Use BIOS/firmware settings. This is the
2999 hpiosize=nn[KMG] The fixed amount of bus space which is
3000 reserved for hotplug bridge's IO window.
3001 Default size is 256 bytes.
3002 hpmemsize=nn[KMG] The fixed amount of bus space which is
3003 reserved for hotplug bridge's memory window.
3004 Default size is 2 megabytes.
3005 realloc= Enable/disable reallocating PCI bridge resources
3006 if allocations done by BIOS are too small to
3007 accommodate resources required by all child
3009 off: Turn realloc off
3011 realloc same as realloc=on
3012 noari do not use PCIe ARI.
3013 pcie_scan_all Scan all possible PCIe devices. Otherwise we
3014 only look for one device below a PCIe downstream
3017 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
3020 force Enable ASPM even on devices that claim not to support it.
3021 WARNING: Forcing ASPM on may cause system lockups.
3023 pcie_hp= [PCIE] PCI Express Hotplug driver options:
3024 nomsi Do not use MSI for PCI Express Native Hotplug (this
3025 makes all PCIe ports use INTx for hotplug services).
3027 pcie_ports= [PCIE] PCIe ports handling:
3028 auto Ask the BIOS whether or not to use native PCIe services
3029 associated with PCIe ports (PME, hot-plug, AER). Use
3030 them only if that is allowed by the BIOS.
3031 native Use native PCIe services associated with PCIe ports
3033 compat Treat PCIe ports as PCI-to-PCI bridges, disable the PCIe
3036 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
3037 nomsi Do not use MSI for native PCIe PME signaling (this makes
3038 all PCIe root ports use INTx for all services).
3040 pcmv= [HW,PCMCIA] BadgePAD 4
3044 Keep all power-domains already enabled by bootloader on,
3045 even if no driver has claimed them. This is useful
3046 for debug and development, but should not be
3047 needed on a platform with proper driver support.
3050 See Documentation/blockdev/paride.txt.
3052 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
3055 See arch/parisc/kernel/pdc_chassis.c
3057 percpu_alloc= Select which percpu first chunk allocator to use.
3058 Currently supported values are "embed" and "page".
3059 Archs may support subset or none of the selections.
3060 See comments in mm/percpu.c for details on each
3061 allocator. This parameter is primarily for debugging
3062 and performance comparison.
3065 See Documentation/blockdev/paride.txt.
3068 See Documentation/blockdev/paride.txt.
3070 pirq= [SMP,APIC] Manual mp-table setup
3071 See Documentation/x86/i386/IO-APIC.txt.
3073 plip= [PPT,NET] Parallel port network link
3074 Format: { parport<nr> | timid | 0 }
3075 See also Documentation/parport.txt.
3077 pmtmr= [X86] Manual setup of pmtmr I/O Port.
3078 Override pmtimer IOPort with a hex value.
3082 Enable PNP debug messages (depends on the
3083 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
3084 via /sys/module/pnp/parameters/debug. We always show
3085 current resource usage; turning this on also shows
3086 possible settings and some assignment information.
3092 { on | off | curr | res | no-curr | no-res }
3095 [ISAPNP] Exclude IRQs for the autoconfiguration
3098 [ISAPNP] Exclude DMAs for the autoconfiguration
3100 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
3101 Ranges are in pairs (I/O port base and size).
3104 [ISAPNP] Exclude memory regions for the
3106 Ranges are in pairs (memory base and size).
3108 ports= [IP_VS_FTP] IPVS ftp helper module
3110 Up to 8 (IP_VS_APP_MAX_PORTS) ports
3112 Format: <port>,<port>....
3114 print-fatal-signals=
3115 [KNL] debug: print fatal signals
3117 If enabled, warn about various signal handling
3118 related application anomalies: too many signals,
3119 too many POSIX.1 timers, fatal signals causing a
3122 If you hit the warning due to signal overflow,
3123 you might want to try "ulimit -i unlimited".
3127 printk.always_kmsg_dump=
3128 Trigger kmsg_dump for cases other than kernel oops or
3130 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
3133 printk.time= Show timing data prefixed to each printk message line
3134 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
3136 processor.max_cstate= [HW,ACPI]
3137 Limit processor to maximum C-state
3138 max_cstate=9 overrides any DMI blacklist limit.
3140 processor.nocst [HW,ACPI]
3141 Ignore the _CST method to determine C-states,
3142 instead using the legacy FADT method
3144 profile= [KNL] Enable kernel profiling via /proc/profile
3145 Format: [schedule,]<number>
3146 Param: "schedule" - profile schedule points.
3147 Param: <number> - step/bucket size as a power of 2 for
3148 statistical time based profiling.
3149 Param: "sleep" - profile D-state sleeping (millisecs).
3150 Requires CONFIG_SCHEDSTATS
3151 Param: "kvm" - profile VM exits.
3153 prompt_ramdisk= [RAM] List of RAM disks to prompt for floppy disk
3155 See Documentation/blockdev/ramdisk.txt.
3157 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
3158 probe for; one of (bare|imps|exps|lifebook|any).
3159 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
3161 psmouse.resetafter= [HW,MOUSE]
3162 Try to reset the device after so many bad packets
3165 [HW,MOUSE] Set desired mouse resolution, in dpi.
3166 psmouse.smartscroll=
3167 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
3168 0 = disabled, 1 = enabled (default).
3170 pstore.backend= Specify the name of the pstore backend to use
3173 See Documentation/blockdev/paride.txt.
3175 pti= [X86_64] Control Page Table Isolation of user and
3176 kernel address spaces. Disabling this feature
3177 removes hardening, but improves performance of
3178 system calls and interrupts.
3180 on - unconditionally enable
3181 off - unconditionally disable
3182 auto - kernel detects whether your CPU model is
3183 vulnerable to issues that PTI mitigates
3185 Not specifying this option is equivalent to pti=auto.
3188 Equivalent to pti=off
3191 [KNL] Number of legacy pty's. Overwrites compiled-in
3194 quiet [KNL] Disable most log messages
3199 See Documentation/md.txt.
3201 ramdisk_blocksize= [RAM]
3202 See Documentation/blockdev/ramdisk.txt.
3204 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
3205 See Documentation/blockdev/ramdisk.txt.
3208 In kernels built with CONFIG_RCU_NOCB_CPU=y, set
3209 the specified list of CPUs to be no-callback CPUs.
3210 Invocation of these CPUs' RCU callbacks will
3211 be offloaded to "rcuox/N" kthreads created for
3212 that purpose, where "x" is "b" for RCU-bh, "p"
3213 for RCU-preempt, and "s" for RCU-sched, and "N"
3214 is the CPU number. This reduces OS jitter on the
3215 offloaded CPUs, which can be useful for HPC and
3216 real-time workloads. It can also improve energy
3217 efficiency for asymmetric multiprocessors.
3220 Rather than requiring that offloaded CPUs
3221 (specified by rcu_nocbs= above) explicitly
3222 awaken the corresponding "rcuoN" kthreads,
3223 make these kthreads poll for callbacks.
3224 This improves the real-time response for the
3225 offloaded CPUs by relieving them of the need to
3226 wake up the corresponding kthread, but degrades
3227 energy efficiency by requiring that the kthreads
3228 periodically wake up to do the polling.
3230 rcutree.blimit= [KNL]
3231 Set maximum number of finished RCU callbacks to
3232 process in one batch.
3234 rcutree.dump_tree= [KNL]
3235 Dump the structure of the rcu_node combining tree
3236 out at early boot. This is used for diagnostic
3237 purposes, to verify correct tree setup.
3239 rcutree.gp_cleanup_delay= [KNL]
3240 Set the number of jiffies to delay each step of
3241 RCU grace-period cleanup. This only has effect
3242 when CONFIG_RCU_TORTURE_TEST_SLOW_CLEANUP is set.
3244 rcutree.gp_init_delay= [KNL]
3245 Set the number of jiffies to delay each step of
3246 RCU grace-period initialization. This only has
3247 effect when CONFIG_RCU_TORTURE_TEST_SLOW_INIT
3250 rcutree.gp_preinit_delay= [KNL]
3251 Set the number of jiffies to delay each step of
3252 RCU grace-period pre-initialization, that is,
3253 the propagation of recent CPU-hotplug changes up
3254 the rcu_node combining tree. This only has effect
3255 when CONFIG_RCU_TORTURE_TEST_SLOW_PREINIT is set.
3257 rcutree.rcu_fanout_exact= [KNL]
3258 Disable autobalancing of the rcu_node combining
3259 tree. This is used by rcutorture, and might
3260 possibly be useful for architectures having high
3261 cache-to-cache transfer latencies.
3263 rcutree.rcu_fanout_leaf= [KNL]
3264 Change the number of CPUs assigned to each
3265 leaf rcu_node structure. Useful for very
3266 large systems, which will choose the value 64,
3267 and for NUMA systems with large remote-access
3268 latencies, which will choose a value aligned
3269 with the appropriate hardware boundaries.
3271 rcutree.jiffies_till_sched_qs= [KNL]
3272 Set required age in jiffies for a
3273 given grace period before RCU starts
3274 soliciting quiescent-state help from
3275 rcu_note_context_switch().
3277 rcutree.jiffies_till_first_fqs= [KNL]
3278 Set delay from grace-period initialization to
3279 first attempt to force quiescent states.
3280 Units are jiffies, minimum value is zero,
3281 and maximum value is HZ.
3283 rcutree.jiffies_till_next_fqs= [KNL]
3284 Set delay between subsequent attempts to force
3285 quiescent states. Units are jiffies, minimum
3286 value is one, and maximum value is HZ.
3288 rcutree.kthread_prio= [KNL,BOOT]
3289 Set the SCHED_FIFO priority of the RCU per-CPU
3290 kthreads (rcuc/N). This value is also used for
3291 the priority of the RCU boost threads (rcub/N)
3292 and for the RCU grace-period kthreads (rcu_bh,
3293 rcu_preempt, and rcu_sched). If RCU_BOOST is
3294 set, valid values are 1-99 and the default is 1
3295 (the least-favored priority). Otherwise, when
3296 RCU_BOOST is not set, valid values are 0-99 and
3297 the default is zero (non-realtime operation).
3299 rcutree.rcu_nocb_leader_stride= [KNL]
3300 Set the number of NOCB kthread groups, which
3301 defaults to the square root of the number of
3302 CPUs. Larger numbers reduces the wakeup overhead
3303 on the per-CPU grace-period kthreads, but increases
3304 that same overhead on each group's leader.
3306 rcutree.qhimark= [KNL]
3307 Set threshold of queued RCU callbacks beyond which
3308 batch limiting is disabled.
3310 rcutree.qlowmark= [KNL]
3311 Set threshold of queued RCU callbacks below which
3312 batch limiting is re-enabled.
3314 rcutree.rcu_idle_gp_delay= [KNL]
3315 Set wakeup interval for idle CPUs that have
3316 RCU callbacks (RCU_FAST_NO_HZ=y).
3318 rcutree.rcu_idle_lazy_gp_delay= [KNL]
3319 Set wakeup interval for idle CPUs that have
3320 only "lazy" RCU callbacks (RCU_FAST_NO_HZ=y).
3321 Lazy RCU callbacks are those which RCU can
3322 prove do nothing more than free memory.
3324 rcutorture.cbflood_inter_holdoff= [KNL]
3325 Set holdoff time (jiffies) between successive
3326 callback-flood tests.
3328 rcutorture.cbflood_intra_holdoff= [KNL]
3329 Set holdoff time (jiffies) between successive
3330 bursts of callbacks within a given callback-flood
3333 rcutorture.cbflood_n_burst= [KNL]
3334 Set the number of bursts making up a given
3335 callback-flood test. Set this to zero to
3336 disable callback-flood testing.
3338 rcutorture.cbflood_n_per_burst= [KNL]
3339 Set the number of callbacks to be registered
3340 in a given burst of a callback-flood test.
3342 rcutorture.fqs_duration= [KNL]
3343 Set duration of force_quiescent_state bursts
3346 rcutorture.fqs_holdoff= [KNL]
3347 Set holdoff time within force_quiescent_state bursts
3350 rcutorture.fqs_stutter= [KNL]
3351 Set wait time between force_quiescent_state bursts
3354 rcutorture.gp_cond= [KNL]
3355 Use conditional/asynchronous update-side
3356 primitives, if available.
3358 rcutorture.gp_exp= [KNL]
3359 Use expedited update-side primitives, if available.
3361 rcutorture.gp_normal= [KNL]
3362 Use normal (non-expedited) asynchronous
3363 update-side primitives, if available.
3365 rcutorture.gp_sync= [KNL]
3366 Use normal (non-expedited) synchronous
3367 update-side primitives, if available. If all
3368 of rcutorture.gp_cond=, rcutorture.gp_exp=,
3369 rcutorture.gp_normal=, and rcutorture.gp_sync=
3370 are zero, rcutorture acts as if is interpreted
3371 they are all non-zero.
3373 rcutorture.n_barrier_cbs= [KNL]
3374 Set callbacks/threads for rcu_barrier() testing.
3376 rcutorture.nfakewriters= [KNL]
3377 Set number of concurrent RCU writers. These just
3378 stress RCU, they don't participate in the actual
3379 test, hence the "fake".
3381 rcutorture.nreaders= [KNL]
3382 Set number of RCU readers. The value -1 selects
3383 N-1, where N is the number of CPUs. A value
3384 "n" less than -1 selects N-n-2, where N is again
3385 the number of CPUs. For example, -2 selects N
3386 (the number of CPUs), -3 selects N+1, and so on.
3388 rcutorture.object_debug= [KNL]
3389 Enable debug-object double-call_rcu() testing.
3391 rcutorture.onoff_holdoff= [KNL]
3392 Set time (s) after boot for CPU-hotplug testing.
3394 rcutorture.onoff_interval= [KNL]
3395 Set time (s) between CPU-hotplug operations, or
3396 zero to disable CPU-hotplug testing.
3398 rcutorture.shuffle_interval= [KNL]
3399 Set task-shuffle interval (s). Shuffling tasks
3400 allows some CPUs to go into dyntick-idle mode
3401 during the rcutorture test.
3403 rcutorture.shutdown_secs= [KNL]
3404 Set time (s) after boot system shutdown. This
3405 is useful for hands-off automated testing.
3407 rcutorture.stall_cpu= [KNL]
3408 Duration of CPU stall (s) to test RCU CPU stall
3409 warnings, zero to disable.
3411 rcutorture.stall_cpu_holdoff= [KNL]
3412 Time to wait (s) after boot before inducing stall.
3414 rcutorture.stat_interval= [KNL]
3415 Time (s) between statistics printk()s.
3417 rcutorture.stutter= [KNL]
3418 Time (s) to stutter testing, for example, specifying
3419 five seconds causes the test to run for five seconds,
3420 wait for five seconds, and so on. This tests RCU's
3421 ability to transition abruptly to and from idle.
3423 rcutorture.test_boost= [KNL]
3424 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
3425 "Maybe" means test if the RCU implementation
3426 under test support RCU priority boosting.
3428 rcutorture.test_boost_duration= [KNL]
3429 Duration (s) of each individual boost test.
3431 rcutorture.test_boost_interval= [KNL]
3432 Interval (s) between each boost test.
3434 rcutorture.test_no_idle_hz= [KNL]
3435 Test RCU's dyntick-idle handling. See also the
3436 rcutorture.shuffle_interval parameter.
3438 rcutorture.torture_runnable= [BOOT]
3439 Start rcutorture running at boot time.
3441 rcutorture.torture_type= [KNL]
3442 Specify the RCU implementation to test.
3444 rcutorture.verbose= [KNL]
3445 Enable additional printk() statements.
3447 rcupdate.rcu_expedited= [KNL]
3448 Use expedited grace-period primitives, for
3449 example, synchronize_rcu_expedited() instead
3450 of synchronize_rcu(). This reduces latency,
3451 but can increase CPU utilization, degrade
3452 real-time latency, and degrade energy efficiency.
3454 rcupdate.rcu_cpu_stall_suppress= [KNL]
3455 Suppress RCU CPU stall warning messages.
3457 rcupdate.rcu_cpu_stall_timeout= [KNL]
3458 Set timeout for RCU CPU stall warning messages.
3460 rcupdate.rcu_task_stall_timeout= [KNL]
3461 Set timeout in jiffies for RCU task stall warning
3462 messages. Disable with a value less than or equal
3465 rcupdate.rcu_self_test= [KNL]
3466 Run the RCU early boot self tests
3468 rcupdate.rcu_self_test_bh= [KNL]
3469 Run the RCU bh early boot self tests
3471 rcupdate.rcu_self_test_sched= [KNL]
3472 Run the RCU sched early boot self tests
3476 Run specified binary instead of /init from the ramdisk,
3477 used for early userspace startup. See initrd.
3480 force - Override the decision by the kernel to hide the
3481 advertisement of RDRAND support (this affects
3482 certain AMD processors because of buggy BIOS
3483 support, specifically around the suspend/resume
3487 Format (x86 or x86_64):
3488 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] \
3490 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
3492 Where reboot_mode is one of warm (soft) or cold (hard) or gpio,
3493 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
3494 reboot_force is either force or not specified,
3495 reboot_cpu is s[mp]#### with #### being the processor
3496 to be used for rebooting.
3499 [KNL, SMP] Set scheduler's default relax_domain_level.
3500 See Documentation/cgroups/cpusets.txt.
3502 relative_sleep_states=
3503 [SUSPEND] Use sleep state labeling where the deepest
3504 state available other than hibernation is always "mem".
3505 Format: { "0" | "1" }
3506 0 -- Traditional sleep state labels.
3507 1 -- Relative sleep state labels.
3509 reserve= [KNL,BUGS] Force the kernel to ignore some iomem area
3511 reservetop= [X86-32]
3513 Reserves a hole at the top of the kernel virtual
3518 Set the amount of memory to reserve for BIOS at
3519 the bottom of the address space.
3521 reset_devices [KNL] Force drivers to reset the underlying device
3522 during initialization.
3525 Specify the partition device for software suspend
3527 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
3529 resume_offset= [SWSUSP]
3530 Specify the offset from the beginning of the partition
3531 given by "resume=" at which the swap header is located,
3532 in <PAGE_SIZE> units (needed only for swap files).
3533 See Documentation/power/swsusp-and-swap-files.txt
3535 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
3536 read the resume files
3538 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
3539 Useful for devices that are detected asynchronously
3540 (e.g. USB and MMC devices).
3542 hibernate= [HIBERNATION]
3543 noresume Don't check if there's a hibernation image
3544 present during boot.
3545 nocompress Don't compress/decompress hibernation images.
3546 no Disable hibernation and resume.
3548 retain_initrd [RAM] Keep initrd memory after extraction
3550 rfkill.default_state=
3551 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
3552 etc. communication is blocked by default.
3555 rfkill.master_switch_mode=
3556 0 The "airplane mode" button does nothing.
3557 1 The "airplane mode" button toggles between everything
3558 blocked and the previous configuration.
3559 2 The "airplane mode" button toggles between everything
3560 blocked and everything unblocked.
3562 rhash_entries= [KNL,NET]
3563 Set number of hash buckets for route cache
3565 ro [KNL] Mount root device read-only on boot
3568 on Mark read-only kernel memory as read-only (default).
3569 off Leave read-only kernel memory writable for debugging.
3571 root= [KNL] Root filesystem
3572 See name_to_dev_t comment in init/do_mounts.c.
3574 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
3575 mount the root filesystem
3577 rootflags= [KNL] Set root filesystem mount option string
3579 rootfstype= [KNL] Set root filesystem type
3581 rootwait [KNL] Wait (indefinitely) for root device to show up.
3582 Useful for devices that are detected asynchronously
3583 (e.g. USB and MMC devices).
3585 rproc_mem=nn[KMG][@address]
3586 [KNL,ARM,CMA] Remoteproc physical memory block.
3587 Memory area to be used by remote processor image,
3590 rw [KNL] Mount root device read-write on boot
3592 S [KNL] Run init in single mode
3594 s390_iommu= [HW,S390]
3595 Set s390 IOTLB flushing mode
3597 With strict flushing every unmap operation will result in
3598 an IOTLB flush. Default is lazy flushing before reuse,
3602 See drivers/net/irda/sa1100_ir.c.
3604 sbni= [NET] Granch SBNI12 leased line adapter
3606 sched_debug [KNL] Enables verbose scheduler debug messages.
3608 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
3609 xtime_lock contention on larger systems, and/or RCU lock
3610 contention on all systems with CONFIG_MAXSMP set.
3611 Format: { "0" | "1" }
3612 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
3614 Note: increases power consumption, thus should only be
3615 enabled if running jitter sensitive (HPC/RT) workloads.
3617 security= [SECURITY] Choose a security module to enable at boot.
3618 If this boot parameter is not specified, only the first
3619 security module asking for security registration will be
3620 loaded. An invalid security module name will be treated
3621 as if no module has been chosen.
3623 selinux= [SELINUX] Disable or enable SELinux at boot time.
3624 Format: { "0" | "1" }
3625 See security/selinux/Kconfig help text.
3628 Default value is set via kernel config option.
3629 If enabled at boot time, /selinux/disable can be used
3630 later to disable prior to initial policy load.
3632 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
3633 Format: { "0" | "1" }
3634 See security/apparmor/Kconfig help text
3637 Default value is set via kernel config option.
3639 serialnumber [BUGS=X86-32]
3642 Maximal number of shapers.
3644 show_msr= [x86] show boot-time MSR settings
3645 Format: { <integer> }
3646 Show boot-time (BIOS-initialized) MSR settings.
3647 The parameter means the number of CPUs to show,
3648 for example 1 means boot CPU only.
3656 Disable merging of slabs with similar size. May be
3657 necessary if there is some reason to distinguish
3658 allocs to different slabs. Debug options disable
3659 merging on their own.
3660 For more information see Documentation/vm/slub.txt.
3662 slab_max_order= [MM, SLAB]
3663 Determines the maximum allowed order for slabs.
3664 A high setting may cause OOMs due to memory
3665 fragmentation. Defaults to 1 for systems with
3666 more than 32MB of RAM, 0 otherwise.
3668 slub_debug[=options[,slabs]] [MM, SLUB]
3669 Enabling slub_debug allows one to determine the
3670 culprit if slab objects become corrupted. Enabling
3671 slub_debug can create guard zones around objects and
3672 may poison objects when not in use. Also tracks the
3673 last alloc / free. For more information see
3674 Documentation/vm/slub.txt.
3676 slub_max_order= [MM, SLUB]
3677 Determines the maximum allowed order for slabs.
3678 A high setting may cause OOMs due to memory
3679 fragmentation. For more information see
3680 Documentation/vm/slub.txt.
3682 slub_min_objects= [MM, SLUB]
3683 The minimum number of objects per slab. SLUB will
3684 increase the slab order up to slub_max_order to
3685 generate a sufficiently large slab able to contain
3686 the number of objects indicated. The higher the number
3687 of objects the smaller the overhead of tracking slabs
3688 and the less frequently locks need to be acquired.
3689 For more information see Documentation/vm/slub.txt.
3691 slub_min_order= [MM, SLUB]
3692 Determines the minimum page order for slabs. Must be
3693 lower than slub_max_order.
3694 For more information see Documentation/vm/slub.txt.
3696 slub_nomerge [MM, SLUB]
3697 Same with slab_nomerge. This is supported for legacy.
3698 See slab_nomerge for more information.
3701 Format: <io1>[,<io2>[,...,<io8>]]
3703 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
3704 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
3705 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
3706 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
3707 smsc-ircc2.ircc_irq= [HW] IRQ line
3708 smsc-ircc2.ircc_dma= [HW] DMA channel
3709 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
3710 0: Toshiba Satellite 1800 (GP data pin select)
3711 1: Fast pin select (default)
3715 [KNL] Should the soft-lockup detector generate panics.
3718 softlockup_all_cpu_backtrace=
3719 [KNL] Should the soft-lockup detector generate
3720 backtraces on all cpus.
3723 sonypi.*= [HW] Sony Programmable I/O Control Device driver
3724 See Documentation/laptops/sonypi.txt
3726 spectre_v2= [X86] Control mitigation of Spectre variant 2
3727 (indirect branch speculation) vulnerability.
3728 The default operation protects the kernel from
3731 on - unconditionally enable, implies
3733 off - unconditionally disable, implies
3735 auto - kernel detects whether your CPU model is
3738 Selecting 'on' will, and 'auto' may, choose a
3739 mitigation method at run time according to the
3740 CPU, the available microcode, the setting of the
3741 CONFIG_RETPOLINE configuration option, and the
3742 compiler with which the kernel was built.
3744 Selecting 'on' will also enable the mitigation
3745 against user space to user space task attacks.
3747 Selecting 'off' will disable both the kernel and
3748 the user space protections.
3750 Specific mitigations can also be selected manually:
3752 retpoline - replace indirect branches
3753 retpoline,generic - google's original retpoline
3754 retpoline,amd - AMD-specific minimal thunk
3756 Not specifying this option is equivalent to
3760 [X86] Control mitigation of Spectre variant 2
3761 (indirect branch speculation) vulnerability between
3764 on - Unconditionally enable mitigations. Is
3765 enforced by spectre_v2=on
3767 off - Unconditionally disable mitigations. Is
3768 enforced by spectre_v2=off
3770 prctl - Indirect branch speculation is enabled,
3771 but mitigation can be enabled via prctl
3772 per thread. The mitigation control state
3773 is inherited on fork.
3776 - Like "prctl" above, but only STIBP is
3777 controlled per thread. IBPB is issued
3778 always when switching between different user
3782 - Same as "prctl" above, but all seccomp
3783 threads will enable the mitigation unless
3784 they explicitly opt out.
3787 - Like "seccomp" above, but only STIBP is
3788 controlled per thread. IBPB is issued
3789 always when switching between different
3790 user space processes.
3792 auto - Kernel selects the mitigation depending on
3793 the available CPU features and vulnerability.
3796 If CONFIG_SECCOMP=y then "seccomp", otherwise "prctl"
3798 Not specifying this option is equivalent to
3799 spectre_v2_user=auto.
3801 spec_store_bypass_disable=
3802 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
3803 (Speculative Store Bypass vulnerability)
3805 Certain CPUs are vulnerable to an exploit against a
3806 a common industry wide performance optimization known
3807 as "Speculative Store Bypass" in which recent stores
3808 to the same memory location may not be observed by
3809 later loads during speculative execution. The idea
3810 is that such stores are unlikely and that they can
3811 be detected prior to instruction retirement at the
3812 end of a particular speculation execution window.
3814 In vulnerable processors, the speculatively forwarded
3815 store can be used in a cache side channel attack, for
3816 example to read memory to which the attacker does not
3817 directly have access (e.g. inside sandboxed code).
3819 This parameter controls whether the Speculative Store
3820 Bypass optimization is used.
3822 on - Unconditionally disable Speculative Store Bypass
3823 off - Unconditionally enable Speculative Store Bypass
3824 auto - Kernel detects whether the CPU model contains an
3825 implementation of Speculative Store Bypass and
3826 picks the most appropriate mitigation. If the
3827 CPU is not vulnerable, "off" is selected. If the
3828 CPU is vulnerable the default mitigation is
3829 architecture and Kconfig dependent. See below.
3830 prctl - Control Speculative Store Bypass per thread
3831 via prctl. Speculative Store Bypass is enabled
3832 for a process by default. The state of the control
3833 is inherited on fork.
3834 seccomp - Same as "prctl" above, but all seccomp threads
3835 will disable SSB unless they explicitly opt out.
3837 Not specifying this option is equivalent to
3838 spec_store_bypass_disable=auto.
3840 Default mitigations:
3841 X86: If CONFIG_SECCOMP=y "seccomp", otherwise "prctl"
3843 spia_io_base= [HW,MTD]
3848 stack_guard_gap= [MM]
3849 override the default stack gap protection. The value
3850 is in page units and it defines how many pages prior
3851 to (for stacks growing down) resp. after (for stacks
3852 growing up) the main stack are reserved for no other
3853 mapping. Default value is 256 pages.
3856 Enabled the stack tracer on boot up.
3858 stacktrace_filter=[function-list]
3859 [FTRACE] Limit the functions that the stack tracer
3860 will trace at boot up. function-list is a comma separated
3861 list of functions. This list can be changed at run
3862 time by the stack_trace_filter file in the debugfs
3863 tracing directory. Note, this enables stack tracing
3864 and the stacktrace above is not needed.
3868 Set the STI (builtin display/keyboard on the HP-PARISC
3869 machines) console (graphic card) which should be used
3870 as the initial boot-console.
3871 See also comment in drivers/video/console/sticore.c.
3874 See comment in drivers/video/console/sticore.c.
3877 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
3879 sunrpc.min_resvport=
3880 sunrpc.max_resvport=
3882 SunRPC servers often require that client requests
3883 originate from a privileged port (i.e. a port in the
3884 range 0 < portnr < 1024).
3885 An administrator who wishes to reserve some of these
3886 ports for other uses may adjust the range that the
3887 kernel's sunrpc client considers to be privileged
3888 using these two parameters to set the minimum and
3889 maximum port values.
3893 Control how the NFS server code allocates CPUs to
3894 service thread pools. Depending on how many NICs
3895 you have and where their interrupts are bound, this
3896 option will affect which CPUs will do NFS serving.
3897 Note: this parameter cannot be changed while the
3898 NFS server is running.
3900 auto the server chooses an appropriate mode
3901 automatically using heuristics
3902 global a single global pool contains all CPUs
3903 percpu one pool for each CPU
3904 pernode one pool for each NUMA node (equivalent
3905 to global on non-NUMA machines)
3907 sunrpc.tcp_slot_table_entries=
3908 sunrpc.udp_slot_table_entries=
3910 Sets the upper limit on the number of simultaneous
3911 RPC calls that can be sent from the client to a
3912 server. Increasing these values may allow you to
3913 improve throughput, but will also increase the
3914 amount of memory reserved for use by the client.
3916 suspend.pm_test_delay=
3918 Sets the number of seconds to remain in a suspend test
3919 mode before resuming the system (see
3920 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
3921 is set. Default value is 5.
3924 [KNL] Enable accounting of swap in memory resource
3925 controller if no parameter or 1 is given or disable
3926 it if 0 is given (See Documentation/cgroups/memory.txt)
3928 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
3929 Format: { <int> | force }
3930 <int> -- Number of I/O TLB slabs
3931 force -- force using of bounce buffers even if they
3932 wouldn't be automatically used by the kernel
3936 sysfs.deprecated=0|1 [KNL]
3937 Enable/disable old style sysfs layout for old udev
3938 on older distributions. When this option is enabled
3939 very new udev will not work anymore. When this option
3940 is disabled (or CONFIG_SYSFS_DEPRECATED not compiled)
3941 in older udev will not work anymore.
3942 Default depends on CONFIG_SYSFS_DEPRECATED_V2 set in
3943 the kernel configuration.
3945 sysrq_always_enabled
3947 Ignore sysrq setting - this boot parameter will
3948 neutralize any effect of /proc/sys/kernel/sysrq.
3949 Useful for debugging.
3951 tcpmhash_entries= [KNL,NET]
3952 Set the number of tcp_metrics_hash slots.
3953 Default value is 8192 or 16384 depending on total
3954 ram pages. This is used to specify the TCP metrics
3955 cache size. See Documentation/networking/ip-sysctl.txt
3956 "tcp_no_metrics_save" section for more details.
3960 test_suspend= [SUSPEND][,N]
3961 Specify "mem" (for Suspend-to-RAM) or "standby" (for
3962 standby suspend) or "freeze" (for suspend type freeze)
3963 as the system sleep state during system startup with
3964 the optional capability to repeat N number of times.
3965 The system is woken from this state using a
3966 wakeup-capable RTC alarm.
3968 thash_entries= [KNL,NET]
3969 Set number of hash buckets for TCP connection
3971 thermal.act= [HW,ACPI]
3972 -1: disable all active trip points in all thermal zones
3973 <degrees C>: override all lowest active trip points
3975 thermal.crt= [HW,ACPI]
3976 -1: disable all critical trip points in all thermal zones
3977 <degrees C>: override all critical trip points
3979 thermal.nocrt= [HW,ACPI]
3980 Set to disable actions on ACPI thermal zone
3981 critical and hot trip points.
3983 thermal.off= [HW,ACPI]
3984 1: disable ACPI thermal control
3986 thermal.psv= [HW,ACPI]
3987 -1: disable all passive trip points
3988 <degrees C>: override all passive trip points to this
3991 thermal.tzp= [HW,ACPI]
3992 Specify global default ACPI thermal zone polling rate
3993 <deci-seconds>: poll all this frequency
3994 0: no polling (default)
3997 Force threading of all interrupt handlers except those
3998 marked explicitly IRQF_NO_THREAD.
4001 Enable the Transcendent memory driver if built-in.
4003 tmem.cleancache=0|1 [KNL, XEN]
4004 Default is on (1). Disable the usage of the cleancache
4005 API to send anonymous pages to the hypervisor.
4007 tmem.frontswap=0|1 [KNL, XEN]
4008 Default is on (1). Disable the usage of the frontswap
4009 API to send swap pages to the hypervisor. If disabled
4010 the selfballooning and selfshrinking are force disabled.
4012 tmem.selfballooning=0|1 [KNL, XEN]
4013 Default is on (1). Disable the driving of swap pages
4016 tmem.selfshrinking=0|1 [KNL, XEN]
4017 Default is on (1). Partial swapoff that immediately
4018 transfers pages from Xen hypervisor back to the
4019 kernel based on different criteria.
4023 Specify if the kernel should make use of the cpu
4024 topology information if the hardware supports this.
4025 The scheduler will make use of this information and
4026 e.g. base its process migration decisions on it.
4029 topology_updates= [KNL, PPC, NUMA]
4031 Specify if the kernel should ignore (off)
4032 topology updates sent by the hypervisor to this
4037 tpm_suspend_pcr=[HW,TPM]
4038 Format: integer pcr id
4039 Specify that at suspend time, the tpm driver
4040 should extend the specified pcr with zeros,
4041 as a workaround for some chips which fail to
4042 flush the last written pcr on TPM_SaveState.
4043 This will guarantee that all the other pcrs
4046 trace_buf_size=nn[KMG]
4047 [FTRACE] will set tracing buffer size on each cpu.
4049 trace_event=[event-list]
4050 [FTRACE] Set and start specified trace events in order
4051 to facilitate early boot debugging.
4052 See also Documentation/trace/events.txt
4054 trace_options=[option-list]
4055 [FTRACE] Enable or disable tracer options at boot.
4056 The option-list is a comma delimited list of options
4057 that can be enabled or disabled just as if you were
4058 to echo the option name into
4060 /sys/kernel/debug/tracing/trace_options
4062 For example, to enable stacktrace option (to dump the
4063 stack trace of each event), add to the command line:
4065 trace_options=stacktrace
4067 See also Documentation/trace/ftrace.txt "trace options"
4071 Have the tracepoints sent to printk as well as the
4072 tracing ring buffer. This is useful for early boot up
4073 where the system hangs or reboots and does not give the
4074 option for reading the tracing buffer or performing a
4075 ftrace_dump_on_oops.
4077 To turn off having tracepoints sent to printk,
4078 echo 0 > /proc/sys/kernel/tracepoint_printk
4079 Note, echoing 1 into this file without the
4080 tracepoint_printk kernel cmdline option has no effect.
4084 Having tracepoints sent to printk() and activating high
4085 frequency tracepoints such as irq or sched, can cause
4086 the system to live lock.
4089 [FTRACE] enable this option to disable tracing when a
4090 warning is hit. This turns off "tracing_on". Tracing can
4091 be enabled again by echoing '1' into the "tracing_on"
4092 file located in /sys/kernel/debug/tracing/
4094 This option is useful, as it disables the trace before
4095 the WARNING dump is called, which prevents the trace to
4096 be filled with content caused by the warning output.
4098 This option can also be set at run time via the sysctl
4099 option: kernel/traceoff_on_warning
4101 transparent_hugepage=
4103 Format: [always|madvise|never]
4104 Can be used to control the default behavior of the system
4105 with respect to transparent hugepages.
4106 See Documentation/vm/transhuge.txt for more details.
4108 tsc= Disable clocksource stability checks for TSC.
4110 [x86] reliable: mark tsc clocksource as reliable, this
4111 disables clocksource verification at runtime, as well
4112 as the stability checks done at bootup. Used to enable
4113 high-resolution timer mode on older hardware, and in
4114 virtualized environment.
4115 [x86] noirqtime: Do not use TSC to do irq accounting.
4116 Used to run time disable IRQ_TIME_ACCOUNTING on any
4117 platforms where RDTSC is slow and this accounting
4120 tsx= [X86] Control Transactional Synchronization
4121 Extensions (TSX) feature in Intel processors that
4122 support TSX control.
4124 This parameter controls the TSX feature. The options are:
4126 on - Enable TSX on the system. Although there are
4127 mitigations for all known security vulnerabilities,
4128 TSX has been known to be an accelerator for
4129 several previous speculation-related CVEs, and
4130 so there may be unknown security risks associated
4131 with leaving it enabled.
4133 off - Disable TSX on the system. (Note that this
4134 option takes effect only on newer CPUs which are
4135 not vulnerable to MDS, i.e., have
4136 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
4137 the new IA32_TSX_CTRL MSR through a microcode
4138 update. This new MSR allows for the reliable
4139 deactivation of the TSX functionality.)
4141 auto - Disable TSX if X86_BUG_TAA is present,
4142 otherwise enable TSX on the system.
4144 Not specifying this option is equivalent to tsx=off.
4146 See Documentation/hw-vuln/tsx_async_abort.rst
4149 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
4150 Abort (TAA) vulnerability.
4152 Similar to Micro-architectural Data Sampling (MDS)
4153 certain CPUs that support Transactional
4154 Synchronization Extensions (TSX) are vulnerable to an
4155 exploit against CPU internal buffers which can forward
4156 information to a disclosure gadget under certain
4159 In vulnerable processors, the speculatively forwarded
4160 data can be used in a cache side channel attack, to
4161 access data to which the attacker does not have direct
4164 This parameter controls the TAA mitigation. The
4167 full - Enable TAA mitigation on vulnerable CPUs
4170 off - Unconditionally disable TAA mitigation
4172 Not specifying this option is equivalent to
4173 tsx_async_abort=full. On CPUs which are MDS affected
4174 and deploy MDS mitigation, TAA mitigation is not
4175 required and doesn't provide any additional
4179 Documentation/hw-vuln/tsx_async_abort.rst
4181 turbografx.map[2|3]= [HW,JOY]
4182 TurboGraFX parallel port interface
4184 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
4185 See also Documentation/input/joystick-parport.txt
4187 udbg-immortal [PPC] When debugging early kernel crashes that
4188 happen after console_init() and before a proper
4189 console driver takes over, this boot options might
4190 help "seeing" what's going on.
4192 uhash_entries= [KNL,NET]
4193 Set number of hash buckets for UDP/UDP-Lite connections
4196 [USB] Ignore overcurrent events (default N).
4197 Some badly-designed motherboards generate lots of
4198 bogus events, for ports that aren't wired to
4199 anything. Set this parameter to avoid log spamming.
4200 Note that genuine overcurrent events won't be
4204 [X86] Cause panic on unknown NMI.
4206 usbcore.authorized_default=
4207 [USB] Default USB device authorization:
4208 (default -1 = authorized except for wireless USB,
4209 0 = not authorized, 1 = authorized)
4211 usbcore.autosuspend=
4212 [USB] The autosuspend time delay (in seconds) used
4213 for newly-detected USB devices (default 2). This
4214 is the time required before an idle device will be
4215 autosuspended. Devices for which the delay is set
4216 to a negative value won't be autosuspended at all.
4218 usbcore.usbfs_snoop=
4219 [USB] Set to log all usbfs traffic (default 0 = off).
4221 usbcore.blinkenlights=
4222 [USB] Set to cycle leds on hubs (default 0 = off).
4224 usbcore.old_scheme_first=
4225 [USB] Start with the old device initialization
4226 scheme (default 0 = off).
4228 usbcore.usbfs_memory_mb=
4229 [USB] Memory limit (in MB) for buffers allocated by
4230 usbfs (default = 16, 0 = max = 2047).
4232 usbcore.use_both_schemes=
4233 [USB] Try the other device initialization scheme
4234 if the first one fails (default 1 = enabled).
4236 usbcore.initial_descriptor_timeout=
4237 [USB] Specifies timeout for the initial 64-byte
4238 USB_REQ_GET_DESCRIPTOR request in milliseconds
4239 (default 5000 = 5.0 seconds).
4242 [USBHID] The interval which mice are to be polled at.
4244 usb-storage.delay_use=
4245 [UMS] The delay in seconds before a new device is
4246 scanned for Logical Units (default 1).
4249 [UMS] A list of quirks entries to supplement or
4250 override the built-in unusual_devs list. List
4251 entries are separated by commas. Each entry has
4252 the form VID:PID:Flags where VID and PID are Vendor
4253 and Product ID values (4-digit hex numbers) and
4254 Flags is a set of characters, each corresponding
4255 to a common usb-storage quirk flag as follows:
4256 a = SANE_SENSE (collect more than 18 bytes
4258 b = BAD_SENSE (don't collect more than 18
4259 bytes of sense data);
4260 c = FIX_CAPACITY (decrease the reported
4261 device capacity by one sector);
4262 d = NO_READ_DISC_INFO (don't use
4263 READ_DISC_INFO command);
4264 e = NO_READ_CAPACITY_16 (don't use
4265 READ_CAPACITY_16 command);
4266 f = NO_REPORT_OPCODES (don't use report opcodes
4268 g = MAX_SECTORS_240 (don't transfer more than
4269 240 sectors at a time, uas only);
4270 h = CAPACITY_HEURISTICS (decrease the
4271 reported device capacity by one
4272 sector if the number is odd);
4273 i = IGNORE_DEVICE (don't bind to this
4275 j = NO_REPORT_LUNS (don't use report luns
4277 l = NOT_LOCKABLE (don't try to lock and
4278 unlock ejectable media);
4279 m = MAX_SECTORS_64 (don't transfer more
4280 than 64 sectors = 32 KB at a time);
4281 n = INITIAL_READ10 (force a retry of the
4282 initial READ(10) command);
4283 o = CAPACITY_OK (accept the capacity
4284 reported by the device);
4285 p = WRITE_CACHE (the device cache is ON
4287 r = IGNORE_RESIDUE (the device reports
4288 bogus residue values);
4289 s = SINGLE_LUN (the device has only one
4291 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
4292 commands, uas only);
4293 u = IGNORE_UAS (don't bind to the uas driver);
4294 w = NO_WP_DETECT (don't test whether the
4295 medium is write-protected).
4296 Example: quirks=0419:aaf5:rl,0421:0433:rc
4298 user_debug= [KNL,ARM]
4300 See arch/arm/Kconfig.debug help text.
4301 1 - undefined instruction events
4303 4 - invalid data aborts
4306 Example: user_debug=31
4309 [X86] Flags controlling user PTE allocations.
4311 nohigh = do not allocate PTE pages in
4312 HIGHMEM regardless of setting
4316 On X86_32, this is an alias for vdso32=. Otherwise:
4318 vdso=1: enable VDSO (the default)
4319 vdso=0: disable VDSO mapping
4321 vdso32= [X86] Control the 32-bit vDSO
4322 vdso32=1: enable 32-bit VDSO
4323 vdso32=0 or vdso32=2: disable 32-bit VDSO
4325 See the help text for CONFIG_COMPAT_VDSO for more
4326 details. If CONFIG_COMPAT_VDSO is set, the default is
4327 vdso32=0; otherwise, the default is vdso32=1.
4329 For compatibility with older kernels, vdso32=2 is an
4332 Try vdso32=0 if you encounter an error that says:
4333 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
4336 vector=percpu: enable percpu vector domain
4338 video= [FB] Frame buffer configuration
4339 See Documentation/fb/modedb.txt.
4341 video.brightness_switch_enabled= [0,1]
4342 If set to 1, on receiving an ACPI notify event
4343 generated by hotkey, video driver will adjust brightness
4344 level and then send out the event to user space through
4345 the allocated input device; If set to 0, video driver
4346 will only send out the event without touching backlight
4351 [VMMIO] Memory mapped virtio (platform) device.
4353 <size>@<baseaddr>:<irq>[:<id>]
4355 <size> := size (can use standard suffixes
4357 <baseaddr> := physical base address
4358 <irq> := interrupt number (as passed to
4360 <id> := (optional) platform device id
4362 virtio_mmio.device=1K@0x100b0000:48:7
4364 Can be used multiple times for multiple devices.
4366 vga= [BOOT,X86-32] Select a particular video mode
4367 See Documentation/x86/boot.txt and
4368 Documentation/svga.txt.
4369 Use vga=ask for menu.
4370 This is actually a boot loader parameter; the value is
4371 passed to the kernel using a special protocol.
4373 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
4374 size of <nn>. This can be used to increase the
4375 minimum size (128MB on x86). It can also be used to
4376 decrease the size and leave more room for directly
4379 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
4382 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
4385 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
4389 Controls the behavior of vsyscalls (i.e. calls to
4390 fixed addresses of 0xffffffffff600x00 from legacy
4391 code). Most statically-linked binaries and older
4392 versions of glibc use these calls. Because these
4393 functions are at fixed addresses, they make nice
4394 targets for exploits that can control RIP.
4396 emulate [default] Vsyscalls turn into traps and are
4397 emulated reasonably safely.
4399 native Vsyscalls are native syscall instructions.
4400 This is a little bit faster than trapping
4401 and makes a few dynamic recompilers work
4402 better than they would in emulation mode.
4403 It also makes exploits much easier to write.
4405 none Vsyscalls don't work at all. This makes
4406 them quite hard to use for exploits but
4407 might break your system.
4409 vt.color= [VT] Default text color.
4410 Format: 0xYX, X = foreground, Y = background.
4411 Default: 0x07 = light gray on black.
4413 vt.cur_default= [VT] Default cursor shape.
4414 Format: 0xCCBBAA, where AA, BB, and CC are the same as
4415 the parameters of the <Esc>[?A;B;Cc escape sequence;
4416 see VGA-softcursor.txt. Default: 2 = underline.
4418 vt.default_blu= [VT]
4419 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
4420 Change the default blue palette of the console.
4421 This is a 16-member array composed of values
4424 vt.default_grn= [VT]
4425 Format: <green0>,<green1>,<green2>,...,<green15>
4426 Change the default green palette of the console.
4427 This is a 16-member array composed of values
4430 vt.default_red= [VT]
4431 Format: <red0>,<red1>,<red2>,...,<red15>
4432 Change the default red palette of the console.
4433 This is a 16-member array composed of values
4439 Set system-wide default UTF-8 mode for all tty's.
4440 Default is 1, i.e. UTF-8 mode is enabled for all
4441 newly opened terminals.
4443 vt.global_cursor_default=
4446 Set system-wide default for whether a cursor
4447 is shown on new VTs. Default is -1,
4448 i.e. cursors will be created by default unless
4449 overridden by individual drivers. 0 will hide
4450 cursors, 1 will display them.
4452 vt.italic= [VT] Default color for italic text; 0-15.
4455 vt.underline= [VT] Default color for underlined text; 0-15.
4458 watchdog timers [HW,WDT] For information on watchdog timers,
4459 see Documentation/watchdog/watchdog-parameters.txt
4460 or other driver-specific files in the
4461 Documentation/watchdog/ directory.
4463 workqueue.disable_numa
4464 By default, all work items queued to unbound
4465 workqueues are affine to the NUMA nodes they're
4466 issued on, which results in better behavior in
4467 general. If NUMA affinity needs to be disabled for
4468 whatever reason, this option can be used. Note
4469 that this also can be controlled per-workqueue for
4470 workqueues visible under /sys/bus/workqueue/.
4472 workqueue.power_efficient
4473 Per-cpu workqueues are generally preferred because
4474 they show better performance thanks to cache
4475 locality; unfortunately, per-cpu workqueues tend to
4476 be more power hungry than unbound workqueues.
4478 Enabling this makes the per-cpu workqueues which
4479 were observed to contribute significantly to power
4480 consumption unbound, leading to measurably lower
4481 power usage at the cost of small performance
4484 The default value of this parameter is determined by
4485 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
4487 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
4488 default x2apic cluster mode on platforms
4491 x86_intel_mid_timer= [X86-32,APBT]
4492 Choose timer option for x86 Intel MID platform.
4493 Two valid options are apbt timer only and lapic timer
4494 plus one apbt timer for broadcast timer.
4495 x86_intel_mid_timer=apbt_only | lapic_and_apbt
4497 xen_512gb_limit [KNL,X86-64,XEN]
4498 Restricts the kernel running paravirtualized under Xen
4499 to use only up to 512 GB of RAM. The reason to do so is
4500 crash analysis tools and Xen tools for doing domain
4501 save/restore/migration must be enabled to handle larger
4504 xen_emul_unplug= [HW,X86,XEN]
4505 Unplug Xen emulated devices
4506 Format: [unplug0,][unplug1]
4507 ide-disks -- unplug primary master IDE devices
4508 aux-ide-disks -- unplug non-primary-master IDE devices
4509 nics -- unplug network devices
4510 all -- unplug all emulated devices (NICs and IDE disks)
4511 unnecessary -- unplugging emulated devices is
4512 unnecessary even if the host did not respond to
4514 never -- do not unplug even if version check succeeds
4516 xen_nopvspin [X86,XEN]
4517 Disables the ticketlock slowpath using Xen PV
4521 Disables the PV optimizations forcing the HVM guest to
4522 run as generic HVM guest with no PV drivers.
4524 xirc2ps_cs= [NET,PCMCIA]
4526 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
4528 ______________________________________________________________________
4532 Add more DRM drivers.