1 acpi= [HW,ACPI,X86,ARM64]
2 Advanced Configuration and Power Interface
3 Format: { force | on | off | strict | noirq | rsdt |
5 force -- enable ACPI if default was off
6 on -- enable ACPI but allow fallback to DT [arm64]
7 off -- disable ACPI if default was on
8 noirq -- do not use ACPI for IRQ routing
9 strict -- Be less tolerant of platforms that are not
10 strictly ACPI specification compliant.
11 rsdt -- prefer RSDT over (default) XSDT
12 copy_dsdt -- copy DSDT to memory
13 For ARM64, ONLY "acpi=off", "acpi=on" or "acpi=force"
16 See also Documentation/power/runtime_pm.rst, pci=noacpi
18 acpi_apic_instance= [ACPI, IOAPIC]
20 2: use 2nd APIC table, if available
21 1,0: use 1st APIC table
24 acpi_backlight= [HW,ACPI]
25 { vendor | video | native | none }
26 If set to vendor, prefer vendor-specific driver
27 (e.g. thinkpad_acpi, sony_acpi, etc.) instead
28 of the ACPI video.ko driver.
29 If set to video, use the ACPI video.ko driver.
30 If set to native, use the device's native backlight mode.
31 If set to none, disable the ACPI backlight interface.
33 acpi_force_32bit_fadt_addr
34 force FADT to use 32 bit addresses rather than the
35 64 bit X_* addresses. Some firmware have broken 64
36 bit addresses for force ACPI ignore these and use
37 the older legacy 32 bit addresses.
39 acpica_no_return_repair [HW, ACPI]
40 Disable AML predefined validation mechanism
41 This mechanism can repair the evaluation result to make
42 the return objects more ACPI specification compliant.
43 This option is useful for developers to identify the
44 root cause of an AML interpreter issue when the issue
45 has something to do with the repair mechanism.
47 acpi.debug_layer= [HW,ACPI,ACPI_DEBUG]
48 acpi.debug_level= [HW,ACPI,ACPI_DEBUG]
50 CONFIG_ACPI_DEBUG must be enabled to produce any ACPI
51 debug output. Bits in debug_layer correspond to a
52 _COMPONENT in an ACPI source file, e.g.,
53 #define _COMPONENT ACPI_EVENTS
54 Bits in debug_level correspond to a level in
55 ACPI_DEBUG_PRINT statements, e.g.,
56 ACPI_DEBUG_PRINT((ACPI_DB_INFO, ...
57 The debug_level mask defaults to "info". See
58 Documentation/firmware-guide/acpi/debug.rst for more information about
59 debug layers and levels.
61 Enable processor driver info messages:
62 acpi.debug_layer=0x20000000
63 Enable AML "Debug" output, i.e., stores to the Debug
64 object while interpreting AML:
65 acpi.debug_layer=0xffffffff acpi.debug_level=0x2
66 Enable all messages related to ACPI hardware:
67 acpi.debug_layer=0x2 acpi.debug_level=0xffffffff
69 Some values produce so much output that the system is
70 unusable. The "log_buf_len" parameter may be useful
71 if you need to capture more output.
73 acpi_enforce_resources= [ACPI]
75 Check for resource conflicts between native drivers
76 and ACPI OperationRegions (SystemIO and SystemMemory
77 only). IO ports and memory declared in ACPI might be
78 used by the ACPI subsystem in arbitrary AML code and
79 can interfere with legacy drivers.
80 strict (default): access to resources claimed by ACPI
81 is denied; legacy drivers trying to access reserved
82 resources will fail to bind to device using them.
83 lax: access to resources claimed by ACPI is allowed;
84 legacy drivers trying to access reserved resources
85 will bind successfully but a warning message is logged.
86 no: ACPI OperationRegions are not marked as reserved,
87 no further checks are performed.
89 acpi_force_table_verification [HW,ACPI]
90 Enable table checksum verification during early stage.
91 By default, this is disabled due to x86 early mapping
94 acpi_irq_balance [HW,ACPI]
95 ACPI will balance active IRQs
98 acpi_irq_nobalance [HW,ACPI]
99 ACPI will not move active IRQs (default)
102 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA
103 Format: <irq>,<irq>...
105 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for
107 Format: <irq>,<irq>...
109 acpi_mask_gpe= [HW,ACPI]
110 Due to the existence of _Lxx/_Exx, some GPEs triggered
111 by unsupported hardware/firmware features can result in
112 GPE floodings that cannot be automatically disabled by
114 This facility can be used to prevent such uncontrolled
116 Format: <byte> or <bitmap-list>
118 acpi_no_auto_serialize [HW,ACPI]
119 Disable auto-serialization of AML methods
120 AML control methods that contain the opcodes to create
121 named objects will be marked as "Serialized" by the
122 auto-serialization feature.
123 This feature is enabled by default.
124 This option allows to turn off the feature.
126 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump
129 acpi_no_static_ssdt [HW,ACPI]
130 Disable installation of static SSDTs at early boot time
131 By default, SSDTs contained in the RSDT/XSDT will be
132 installed automatically and they will appear under
133 /sys/firmware/acpi/tables.
134 This option turns off this feature.
135 Note that specifying this option does not affect
136 dynamic table installation which will install SSDT
137 tables to /sys/firmware/acpi/tables/dynamic.
139 acpi_no_watchdog [HW,ACPI,WDT]
140 Ignore the ACPI-based watchdog interface (WDAT) and let
141 a native driver control the watchdog device instead.
143 acpi_rsdp= [ACPI,EFI,KEXEC]
144 Pass the RSDP address to the kernel, mostly used
145 on machines running EFI runtime service to boot the
146 second kernel for kdump.
148 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS
149 Format: To spoof as Windows 98: ="Microsoft Windows"
151 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead
152 of 2 which is mandated by ACPI 6) as the supported ACPI
153 specification revision (when using this switch, it may
154 be necessary to carry out a cold reboot _twice_ in a
155 row to make it take effect on the platform firmware).
157 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings
158 acpi_osi="string1" # add string1
159 acpi_osi="!string2" # remove string2
160 acpi_osi=!* # remove all strings
161 acpi_osi=! # disable all built-in OS vendor
163 acpi_osi=!! # enable all built-in OS vendor
165 acpi_osi= # disable all strings
167 'acpi_osi=!' can be used in combination with single or
168 multiple 'acpi_osi="string1"' to support specific OS
169 vendor string(s). Note that such command can only
170 affect the default state of the OS vendor strings, thus
171 it cannot affect the default state of the feature group
172 strings and the current state of the OS vendor strings,
173 specifying it multiple times through kernel command line
174 is meaningless. This command is useful when one do not
175 care about the state of the feature group strings which
176 should be controlled by the OSPM.
178 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent
179 to 'acpi_osi="Windows 2000" acpi_osi=!', they all
180 can make '_OSI("Windows 2000")' TRUE.
182 'acpi_osi=' cannot be used in combination with other
183 'acpi_osi=' command lines, the _OSI method will not
184 exist in the ACPI namespace. NOTE that such command can
185 only affect the _OSI support state, thus specifying it
186 multiple times through kernel command line is also
189 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)'
192 'acpi_osi=!*' can be used in combination with single or
193 multiple 'acpi_osi="string1"' to support specific
194 string(s). Note that such command can affect the
195 current state of both the OS vendor strings and the
196 feature group strings, thus specifying it multiple times
197 through kernel command line is meaningful. But it may
198 still not able to affect the final state of a string if
199 there are quirks related to this string. This command
200 is useful when one want to control the state of the
201 feature group strings to debug BIOS issues related to
204 1. 'acpi_osi="Module Device" acpi_osi=!*' can make
205 '_OSI("Module Device")' FALSE.
206 2. 'acpi_osi=!* acpi_osi="Module Device"' can make
207 '_OSI("Module Device")' TRUE.
208 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is
210 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"'
212 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!',
213 they all will make '_OSI("Windows 2000")' TRUE.
216 Override the pmtimer bug detection: force the kernel
217 to assume that this machine's pmtimer latches its value
218 and always returns good values.
220 acpi_sci= [HW,ACPI] ACPI System Control Interrupt trigger mode
221 Format: { level | edge | high | low }
223 acpi_skip_timer_override [HW,ACPI]
224 Recognize and ignore IRQ0/pin2 Interrupt Override.
225 For broken nForce2 BIOS resulting in XT-PIC timer.
227 acpi_sleep= [HW,ACPI] Sleep options
228 Format: { s3_bios, s3_mode, s3_beep, s4_nohwsig,
229 old_ordering, nonvs, sci_force_enable, nobl }
230 See Documentation/power/video.rst for information on
232 s3_beep is for debugging; it makes the PC's speaker beep
233 as soon as the kernel's real-mode entry point is called.
234 s4_nohwsig prevents ACPI hardware signature from being
235 used during resume from hibernation.
236 old_ordering causes the ACPI 1.0 ordering of the _PTS
237 control method, with respect to putting devices into
238 low power states, to be enforced (the ACPI 2.0 ordering
239 of _PTS is used by default).
240 nonvs prevents the kernel from saving/restoring the
241 ACPI NVS memory during suspend/hibernation and resume.
242 sci_force_enable causes the kernel to set SCI_EN directly
243 on resume from S1/S3 (which is against the ACPI spec,
244 but some broken systems don't work without it).
245 nobl causes the internal blacklist of systems known to
246 behave incorrectly in some ways with respect to system
247 suspend and resume to be ignored (use wisely).
249 acpi_use_timer_override [HW,ACPI]
250 Use timer override. For some broken Nvidia NF5 boards
251 that require a timer override, but don't have HPET
253 add_efi_memmap [EFI; X86] Include EFI memory map in
254 kernel's map of available physical RAM.
257 { off | try_unsupported }
258 off: disable AGP support
259 try_unsupported: try to drive unsupported chipsets
260 (may crash computer or cause data corruption)
263 See Documentation/sound/alsa-configuration.rst
266 Allow the default userspace alignment fault handler
267 behaviour to be specified. Bit 0 enables warnings,
268 bit 1 enables fixups, and bit 2 sends a segfault.
270 align_va_addr= [X86-64]
271 Align virtual addresses by clearing slice [14:12] when
272 allocating a VMA at process creation time. This option
273 gives you up to 3% performance improvement on AMD F15h
274 machines (where it is enabled by default) for a
275 CPU-intensive style benchmark, and it can vary highly in
276 a microbenchmark depending on workload and compiler.
278 32: only for 32-bit processes
279 64: only for 64-bit processes
280 on: enable for both 32- and 64-bit processes
281 off: disable for both 32- and 64-bit processes
283 alloc_snapshot [FTRACE]
284 Allocate the ftrace snapshot buffer on boot up when the
285 main buffer is allocated. This is handy if debugging
286 and you need to use tracing_snapshot() on boot up, and
287 do not want to use tracing_snapshot_alloc() as it needs
288 to be done where GFP_KERNEL allocations are allowed.
290 allow_mismatched_32bit_el0 [ARM64]
291 Allow execve() of 32-bit applications and setting of the
292 PER_LINUX32 personality on systems where only a strict
293 subset of the CPUs support 32-bit EL0. When this
294 parameter is present, the set of CPUs supporting 32-bit
295 EL0 is indicated by /sys/devices/system/cpu/aarch32_el0
296 and hot-unplug operations may be restricted.
298 amd_iommu= [HW,X86-64]
299 Pass parameters to the AMD IOMMU driver in the system.
301 fullflush - enable flushing of IO/TLB entries when
302 they are unmapped. Otherwise they are
303 flushed before they will be reused, which
305 off - do not initialize any AMD IOMMU found in
307 force_isolation - Force device isolation for all
308 devices. The IOMMU driver is not
309 allowed anymore to lift isolation
310 requirements as needed. This option
311 does not override iommu=pt
312 force_enable - Force enable the IOMMU on platforms known
313 to be buggy with IOMMU enabled. Use this
316 amd_iommu_dump= [HW,X86-64]
317 Enable AMD IOMMU driver option to dump the ACPI table
318 for AMD IOMMU. With this option enabled, AMD IOMMU
319 driver will print ACPI tables for AMD IOMMU during
320 IOMMU initialization.
322 amd_iommu_intr= [HW,X86-64]
323 Specifies one of the following AMD IOMMU interrupt
325 legacy - Use legacy interrupt remapping mode.
326 vapic - Use virtual APIC mode, which allows IOMMU
327 to inject interrupts directly into guest.
328 This mode requires kvm-amd.avic=1.
329 (Default when IOMMU HW support is present.)
331 amijoy.map= [HW,JOY] Amiga joystick support
332 Map of devices attached to JOY0DAT and JOY1DAT
334 See also Documentation/input/joydev/joystick.rst
336 analog.map= [HW,JOY] Analog joystick and gamepad support
337 Specifies type or capabilities of an analog joystick
338 connected to one of 16 gameports
339 Format: <type1>,<type2>,..<type16>
342 Power management functions (SPARCstation-4/5 + deriv.)
344 Disable APC CPU standby support. SPARCstation-Fox does
345 not play well with APC CPU idle - disable it if you have
346 APC and your system crashes randomly.
348 apic= [APIC,X86] Advanced Programmable Interrupt Controller
349 Change the output verbosity while booting
350 Format: { quiet (default) | verbose | debug }
351 Change the amount of debugging information output
352 when initialising the APIC and IO-APIC components.
353 For X86-32, this can also be used to specify an APIC
355 Format: apic=driver_name
356 Examples: apic=bigsmp
358 apic_extnmi= [APIC,X86] External NMI delivery setting
359 Format: { bsp (default) | all | none }
360 bsp: External NMI is delivered only to CPU 0
361 all: External NMIs are broadcast to all CPUs as a
363 none: External NMI is masked for all CPUs. This is
364 useful so that a dump capture kernel won't be
368 See Documentation/networking/ipv6.rst.
370 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
371 Limit apic dumping. The parameter defines the maximal
372 number of local apics being dumped. Also it is possible
373 to set it to "all" by meaning -- no limit here.
374 Format: { 1 (default) | 2 | ... | all }.
375 The parameter valid if only apic=debug or
376 apic=verbose is specified.
377 Example: apic=debug show_lapic=all
379 apm= [APM] Advanced Power Management
380 See header of arch/x86/kernel/apm_32.c.
382 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
383 Format: <io>,<irq>,<nodeID>
385 arm64.nobti [ARM64] Unconditionally disable Branch Target
386 Identification support
388 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication
393 atarimouse= [HW,MOUSE] Atari Mouse
395 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
396 EzKey and similar keyboards
398 atkbd.reset= [HW] Reset keyboard during initialization
400 atkbd.set= [HW] Select keyboard code set
401 Format: <int> (2 = AT (default), 3 = PS/2)
403 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
406 atkbd.softraw= [HW] Choose between synthetic and real raw mode
407 Format: <bool> (0 = real, 1 = synthetic (default))
409 atkbd.softrepeat= [HW]
410 Use software keyboard repeat
412 audit= [KNL] Enable the audit sub-system
413 Format: { "0" | "1" | "off" | "on" }
414 0 | off - kernel audit is disabled and can not be
415 enabled until the next reboot
416 unset - kernel audit is initialized but disabled and
417 will be fully enabled by the userspace auditd.
418 1 | on - kernel audit is initialized and partially
419 enabled, storing at most audit_backlog_limit
420 messages in RAM until it is fully enabled by the
424 audit_backlog_limit= [KNL] Set the audit queue size limit.
425 Format: <int> (must be >=0)
428 bau= [X86_UV] Enable the BAU on SGI UV. The default
429 behavior is to disable the BAU (i.e. bau=0).
430 Format: { "0" | "1" }
433 unset - Disable the BAU.
435 baycom_epp= [HW,AX25]
438 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
440 See header of drivers/net/hamradio/baycom_par.c.
442 baycom_ser_fdx= [HW,AX25]
443 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
444 Format: <io>,<irq>,<mode>[,<baud>]
445 See header of drivers/net/hamradio/baycom_ser_fdx.c.
447 baycom_ser_hdx= [HW,AX25]
448 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
449 Format: <io>,<irq>,<mode>
450 See header of drivers/net/hamradio/baycom_ser_hdx.c.
452 blkdevparts= Manual partition parsing of block device(s) for
453 embedded devices based on command line input.
454 See Documentation/block/cmdline-partition.rst
456 boot_delay= Milliseconds to delay each printk during boot.
457 Values larger than 10 seconds (10000) are changed to
462 Extended command line options can be added to an initrd
463 and this will cause the kernel to look for it.
465 See Documentation/admin-guide/bootconfig.rst
468 Disable BERT OS support on buggy BIOSes.
470 bgrt_disable [ACPI][X86]
471 Disable BGRT to avoid flickering OEM logo.
473 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
474 bttv.radio= Most important insmod options are available as
476 bttv.pll= See Documentation/admin-guide/media/bttv.rst
479 bulk_remove=off [PPC] This parameter disables the use of the pSeries
480 firmware feature for flushing multiple hpte entries
483 c101= [NET] Moxa C101 synchronous serial card
485 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
486 Sometimes CPU hardware bugs make them report the cache
487 size incorrectly. The kernel will attempt work arounds
488 to fix known problems, but for some CPUs it is not
489 possible to determine what the correct size should be.
490 This option provides an override for these situations.
493 [NET] Specifies amount of time (in seconds) that
494 the kernel should wait for a network carrier. By default
495 it waits 120 seconds.
497 ca_keys= [KEYS] This parameter identifies a specific key(s) on
498 the system trusted keyring to be used for certificate
500 format: { id:<keyid> | builtin }
502 cca= [MIPS] Override the kernel pages' cache coherency
503 algorithm. Accepted values range from 0 to 7
504 inclusive. See arch/mips/include/asm/pgtable-bits.h
505 for platform specific values (SB1, Loongson3 and
508 ccw_timeout_log [S390]
509 See Documentation/s390/common_io.rst for details.
511 cgroup_disable= [KNL] Disable a particular controller or optional feature
512 Format: {name of the controller(s) or feature(s) to disable}
513 The effects of cgroup_disable=foo are:
514 - foo isn't auto-mounted if you mount all cgroups in
516 - foo isn't visible as an individually mountable
518 - if foo is an optional feature then the feature is
519 disabled and corresponding cgroup files are not
521 {Currently only "memory" controller deal with this and
522 cut the overhead, others just disable the usage. So
523 only cgroup_disable=memory is actually worthy}
524 Specifying "pressure" disables per-cgroup pressure
525 stall information accounting feature
527 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
528 Format: { { controller | "all" | "named" }
529 [,{ controller | "all" | "named" }...] }
530 Like cgroup_disable, but only applies to cgroup v1;
531 the blacklisted controllers remain available in cgroup2.
532 "all" blacklists all controllers and "named" disables
533 named mounts. Specifying both "all" and "named" disables
536 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
538 nosocket -- Disable socket memory accounting.
539 nokmem -- Disable kernel memory accounting.
541 checkreqprot [SELINUX] Set initial checkreqprot flag value.
542 Format: { "0" | "1" }
543 See security/selinux/Kconfig help text.
544 0 -- check protection applied by kernel (includes
545 any implied execute protection).
546 1 -- check protection requested by application.
547 Default value is set via a kernel config option.
548 Value can be changed at runtime via
549 /sys/fs/selinux/checkreqprot.
550 Setting checkreqprot to 1 is deprecated.
553 See Documentation/s390/common_io.rst for details.
556 Prevents the clock framework from automatically gating
557 clocks that have not been explicitly enabled by a Linux
558 device driver but are enabled in hardware at reset or
559 by the bootloader/firmware. Note that this does not
560 force such clocks to be always-on nor does it reserve
561 those clocks in any way. This parameter is useful for
562 debug and development, but should not be needed on a
563 platform with proper driver support. For more
564 information, see Documentation/driver-api/clk.rst.
566 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
568 Forces specified clocksource (if available) to be used
569 when calculating gettimeofday(). If specified
570 clocksource is not available, it defaults to PIT.
571 Format: { pit | tsc | cyclone | pmtmr }
573 clocksource= Override the default clocksource
575 Override the default clocksource and use the clocksource
576 with the name specified.
577 Some clocksource names to choose from, depending on
579 [all] jiffies (this is the base, fallback clocksource)
581 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
582 pxa_timer,timer3,32k_counter,timer0_1
583 [X86-32] pit,hpet,tsc;
584 scx200_hrt on Geode; cyclone on IBM x440
592 clocksource.arm_arch_timer.evtstrm=
595 Enable/disable the eventstream feature of the ARM
596 architected timer so that code using WFE-based polling
597 loops can be debugged more effectively on production
600 clocksource.max_cswd_read_retries= [KNL]
601 Number of clocksource_watchdog() retries due to
602 external delays before the clock will be marked
603 unstable. Defaults to three retries, that is,
604 four attempts to read the clock under test.
606 clocksource.verify_n_cpus= [KNL]
607 Limit the number of CPUs checked for clocksources
608 marked with CLOCK_SOURCE_VERIFY_PERCPU that
609 are marked unstable due to excessive skew.
610 A negative value says to check all CPUs, while
611 zero says not to check any. Values larger than
612 nr_cpu_ids are silently truncated to nr_cpu_ids.
613 The actual CPUs are chosen randomly, with
614 no replacement if the same CPU is chosen twice.
616 clocksource-wdtest.holdoff= [KNL]
617 Set the time in seconds that the clocksource
618 watchdog test waits before commencing its tests.
619 Defaults to zero when built as a module and to
620 10 seconds when built into the kernel.
622 clearcpuid=BITNUM[,BITNUM...] [X86]
623 Disable CPUID feature X for the kernel. See
624 arch/x86/include/asm/cpufeatures.h for the valid bit
625 numbers. Note the Linux specific bits are not necessarily
626 stable over kernel options, but the vendor specific
628 Also note that user programs calling CPUID directly
629 or using the feature without checking anything
630 will still see it. This just prevents it from
631 being used by the kernel or shown in /proc/cpuinfo.
632 Also note the kernel might malfunction if you disable
635 cma=nn[MG]@[start[MG][-end[MG]]]
637 Sets the size of kernel global memory area for
638 contiguous memory allocations and optionally the
639 placement constraint by the physical address range of
640 memory allocations. A value of 0 disables CMA
641 altogether. For more information, see
642 kernel/dma/contiguous.c
646 Sets the size of kernel per-numa memory area for
647 contiguous memory allocations. A value of 0 disables
648 per-numa CMA altogether. And If this option is not
649 specificed, the default value is 0.
650 With per-numa CMA enabled, DMA users on node nid will
651 first try to allocate buffer from the pernuma area
652 which is located in node nid, if the allocation fails,
653 they will fallback to the global default memory area.
655 cmo_free_hint= [PPC] Format: { yes | no }
656 Specify whether pages are marked as being inactive
657 when they are freed. This is used in CMO environments
658 to determine OS memory pressure for page stealing by
662 coherent_pool=nn[KMG] [ARM,KNL]
663 Sets the size of memory pool for coherent, atomic dma
664 allocations, by default set to 256K.
666 com20020= [HW,NET] ARCnet - COM20020 chipset
668 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
670 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
674 ARCnet - COM90xx chipset (memory-mapped buffers)
675 Format: <io>[,<irq>[,<memstart>]]
677 condev= [HW,S390] console device
680 console= [KNL] Output console device and options.
682 tty<n> Use the virtual console device <n>.
686 Use the specified serial port. The options are of
687 the form "bbbbpnf", where "bbbb" is the baud rate,
688 "p" is parity ("n", "o", or "e"), "n" is number of
689 bits, and "f" is flow control ("r" for RTS or
690 omit it). Default is "9600n8".
692 See Documentation/admin-guide/serial-console.rst for more
694 Documentation/networking/netconsole.rst for an
697 uart[8250],io,<addr>[,options]
698 uart[8250],mmio,<addr>[,options]
699 uart[8250],mmio16,<addr>[,options]
700 uart[8250],mmio32,<addr>[,options]
701 uart[8250],0x<addr>[,options]
702 Start an early, polled-mode console on the 8250/16550
703 UART at the specified I/O port or MMIO address,
704 switching to the matching ttyS device later.
705 MMIO inter-register address stride is either 8-bit
706 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
707 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
708 to be equivalent to 'mmio'. 'options' are specified in
709 the same format described for ttyS above; if unspecified,
710 the h/w is not re-initialized.
712 hvc<n> Use the hypervisor console device <n>. This is for
713 both Xen and PowerPC hypervisors.
715 If the device connected to the port is not a TTY but a braille
716 device, prepend "brl," before the device type, for instance
718 For now, only VisioBraille is supported.
721 [KNL] Change console messages format
723 By default we print messages on consoles in
724 "[time stamp] text\n" format (time stamp may not be
725 printed, depending on CONFIG_PRINTK_TIME or
726 `printk_time' param).
728 Switch to syslog format: "<%u>[time stamp] text\n"
729 IOW, each message will have a facility and loglevel
730 prefix. The format is similar to one used by syslog()
731 syscall, or to executing "dmesg -S --raw" or to reading
734 consoleblank= [KNL] The console blank (screen saver) timeout in
735 seconds. A value of 0 disables the blank timer.
739 [KNL] Change the default value for
740 /proc/<pid>/coredump_filter.
741 See also Documentation/filesystems/proc.rst.
743 coresight_cpu_debug.enable
746 Enable/disable the CPU sampling based debugging.
747 0: default value, disable debugging
748 1: enable debugging at boot time
750 cpuidle.off=1 [CPU_IDLE]
751 disable the cpuidle sub-system
754 [CPU_IDLE] Name of the cpuidle governor to use.
756 cpufreq.off=1 [CPU_FREQ]
757 disable the cpufreq sub-system
759 cpufreq.default_governor=
760 [CPU_FREQ] Name of the default cpufreq governor or
761 policy to use. This governor must be registered in the
762 kernel before the cpufreq driver probes.
765 [X86] Delay for N microsec between assert and de-assert
766 of APIC INIT to start processors. This delay occurs
767 on every CPU online, such as boot, and resume from suspend.
770 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
772 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
774 crashkernel=size[KMG][@offset[KMG]]
775 [KNL] Using kexec, Linux can switch to a 'crash kernel'
776 upon panic. This parameter reserves the physical
777 memory region [offset, offset + size] for that kernel
778 image. If '@offset' is omitted, then a suitable offset
779 is selected automatically.
780 [KNL, X86-64] Select a region under 4G first, and
781 fall back to reserve region above 4G when '@offset'
782 hasn't been specified.
783 See Documentation/admin-guide/kdump/kdump.rst for further details.
785 crashkernel=range1:size1[,range2:size2,...][@offset]
786 [KNL] Same as above, but depends on the memory
787 in the running system. The syntax of range is
788 start-[end] where start and end are both
789 a memory unit (amount[KMG]). See also
790 Documentation/admin-guide/kdump/kdump.rst for an example.
792 crashkernel=size[KMG],high
793 [KNL, X86-64] range could be above 4G. Allow kernel
794 to allocate physical memory region from top, so could
795 be above 4G if system have more than 4G ram installed.
796 Otherwise memory region will be allocated below 4G, if
798 It will be ignored if crashkernel=X is specified.
799 crashkernel=size[KMG],low
800 [KNL, X86-64] range under 4G. When crashkernel=X,high
801 is passed, kernel could allocate physical memory region
802 above 4G, that cause second kernel crash on system
803 that require some amount of low memory, e.g. swiotlb
804 requires at least 64M+32K low memory, also enough extra
805 low memory is needed to make sure DMA buffers for 32-bit
806 devices won't run out. Kernel would try to allocate at
807 at least 256M below 4G automatically.
808 This one let user to specify own low range under 4G
809 for second kernel instead.
810 0: to disable low allocation.
811 It will be ignored when crashkernel=X,high is not used
812 or memory reserved is below 4G.
815 [KNL] Disable crypto self-tests
820 cs89x0_media= [HW,NET]
821 Format: { rj45 | aui | bnc }
823 csdlock_debug= [KNL] Enable debug add-ons of cross-CPU function call
824 handling. When switched on, additional debug data is
825 printed to the console in case a hanging CPU is
826 detected, and that CPU is pinged again in order to try
827 to resolve the hang situation.
828 0: disable csdlock debugging (default)
829 1: enable basic csdlock debugging (minor impact)
830 ext: enable extended csdlock debugging (more impact,
834 See header of drivers/s390/block/dasd_devmap.c.
836 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
837 (one device per port)
838 Format: <port#>,<type>
839 See also Documentation/input/devices/joystick-parport.rst
841 ddebug_query= [KNL,DYNAMIC_DEBUG] Enable debug messages at early boot
843 Documentation/admin-guide/dynamic-debug-howto.rst for
844 details. Deprecated, see dyndbg.
846 debug [KNL] Enable kernel debugging (events log level).
849 [KNL] Enable printing [hashed] pointers early in the
850 boot sequence. If enabled, we use a weak hash instead
851 of siphash to hash pointers. Use this option if you are
852 seeing instances of '(___ptrval___)') and need to see a
853 value (hashed pointer) instead. Cryptographically
854 insecure, please do not use on production kernels.
857 [KNL] verbose locking self-tests
859 Print debugging info while doing the locking API
861 Bitmask for the various LOCKTYPE_ tests. Defaults to 0
862 (no extra messages), setting it to -1 (all bits set)
863 will print _a_lot_ more information - normally only
864 useful to lockdep developers.
866 debug_objects [KNL] Enable object debugging
869 [KNL] Disable object debugging
871 debug_guardpage_minorder=
872 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
873 parameter allows control of the order of pages that will
874 be intentionally kept free (and hence protected) by the
875 buddy allocator. Bigger value increase the probability
876 of catching random memory corruption, but reduce the
877 amount of memory for normal system use. The maximum
878 possible value is MAX_ORDER/2. Setting this parameter
879 to 1 or 2 should be enough to identify most random
880 memory corruption problems caused by bugs in kernel or
881 driver code when a CPU writes to (or reads from) a
882 random memory location. Note that there exists a class
883 of memory corruptions problems caused by buggy H/W or
884 F/W or by drivers badly programing DMA (basically when
885 memory is written at bus level and the CPU MMU is
886 bypassed) which are not detectable by
887 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
888 tracking down these problems.
891 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
892 enables the feature at boot time. By default, it is
893 disabled and the system will work mostly the same as a
894 kernel built without CONFIG_DEBUG_PAGEALLOC.
895 Note: to get most of debug_pagealloc error reports, it's
896 useful to also enable the page_owner functionality.
897 on: enable the feature
899 debugfs= [KNL] This parameter enables what is exposed to userspace
900 and debugfs internal clients.
901 Format: { on, no-mount, off }
902 on: All functions are enabled.
904 Filesystem is not registered but kernel clients can
905 access APIs and a crashkernel can be used to read
906 its content. There is nothing to mount.
907 off: Filesystem is not registered and clients
908 get a -EPERM as result when trying to register files
909 or directories within debugfs.
910 This is equivalent of the runtime functionality if
911 debugfs was not enabled in the kernel at all.
912 Default value is set in build-time with a kernel configuration.
914 debugpat [X86] Enable PAT debugging
916 decnet.addr= [HW,NET]
917 Format: <area>[,<node>]
918 See also Documentation/networking/decnet.rst.
921 [HW] The size of the default HugeTLB page. This is
922 the size represented by the legacy /proc/ hugepages
923 APIs. In addition, this is the default hugetlb size
924 used for shmget(), mmap() and mounting hugetlbfs
925 filesystems. If not specified, defaults to the
926 architecture's default huge page size. Huge page
927 sizes are architecture dependent. See also
928 Documentation/admin-guide/mm/hugetlbpage.rst.
931 deferred_probe_timeout=
932 [KNL] Debugging option to set a timeout in seconds for
933 deferred probe to give up waiting on dependencies to
934 probe. Only specific dependencies (subsystems or
935 drivers) that have opted in will be ignored. A timeout of 0
936 will timeout at the end of initcalls. This option will also
937 dump out devices still on the deferred probe list after
941 Format: { on | off | def_only | inf_only | always }
942 on: s390 zlib hardware support for compression on
943 level 1 and decompression (default)
944 off: No s390 zlib hardware support
945 def_only: s390 zlib hardware support for deflate
946 only (compression on level 1)
947 inf_only: s390 zlib hardware support for inflate
949 always: Same as 'on' but ignores the selected compression
950 level always using hardware support (used for debugging)
953 Set number of hash buckets for dentry cache.
955 disable_1tb_segments [PPC]
956 Disables the use of 1TB hash page table segments. This
957 causes the kernel to fall back to 256MB segments which
958 can be useful when debugging issues that require an SLB
962 Limits the number of kernel SLB entries, and flushes
963 them frequently to increase the rate of SLB faults
967 See Documentation/networking/ipv6.rst.
970 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
971 hardening is enabled for this boot. Hardened
972 usercopy checking is used to protect the kernel
973 from reading or writing beyond known memory
974 allocation boundaries as a proactive defense
975 against bounds-checking flaws in the kernel's
976 copy_to_user()/copy_from_user() interface.
977 on Perform hardened usercopy checks (default).
978 off Disable hardened usercopy checks.
981 Disable RADIX MMU mode on POWER9
983 radix_hcall_invalidate=on [PPC/PSERIES]
984 Disable RADIX GTSE feature and use hcall for TLB
988 Disable TLBIE instruction. Currently does not work
989 with KVM, with HASH MMU, or with coherent accelerators.
991 disable_cpu_apicid= [X86,APIC,SMP]
993 The number of initial APIC ID for the
994 corresponding CPU to be disabled at boot,
995 mostly used for the kdump 2nd kernel to
996 disable BSP to wake up multiple CPUs without
997 causing system reset or hang due to sending
1000 disable_ddw [PPC/PSERIES]
1001 Disable Dynamic DMA Window support. Use this
1002 to workaround buggy firmware.
1004 disable_ipv6= [IPV6]
1005 See Documentation/networking/ipv6.rst.
1007 disable_mtrr_cleanup [X86]
1008 The kernel tries to adjust MTRR layout from continuous
1009 to discrete, to make X server driver able to add WB
1010 entry later. This parameter disables that.
1012 disable_mtrr_trim [X86, Intel and AMD only]
1013 By default the kernel will trim any uncacheable
1014 memory out of your available memory pool based on
1015 MTRR settings. This parameter disables that behavior,
1016 possibly causing your machine to run very slowly.
1018 disable_timer_pin_1 [X86]
1019 Disable PIN 1 of APIC timer
1020 Can be useful to work around chipset bugs.
1022 dis_ucode_ldr [X86] Disable the microcode loader.
1024 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
1025 this option disables the debugging code at boot.
1027 dma_debug_entries=<number>
1028 This option allows to tune the number of preallocated
1029 entries for DMA-API debugging code. One entry is
1030 required per DMA-API allocation. Use this if the
1031 DMA-API debugging code disables itself because the
1032 architectural default is too low.
1034 dma_debug_driver=<driver_name>
1035 With this option the DMA-API debugging driver
1036 filter feature can be enabled at boot time. Just
1037 pass the driver to filter for as the parameter.
1038 The filter can be disabled or changed to another
1039 driver later using sysfs.
1041 driver_async_probe= [KNL]
1042 List of driver names to be probed asynchronously.
1043 Format: <driver_name1>,<driver_name2>...
1045 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
1046 Broken monitors, graphic adapters, KVMs and EDIDless
1047 panels may send no or incorrect EDID data sets.
1048 This parameter allows to specify an EDID data sets
1049 in the /lib/firmware directory that are used instead.
1050 Generic built-in EDID data sets are used, if one of
1051 edid/1024x768.bin, edid/1280x1024.bin,
1052 edid/1680x1050.bin, or edid/1920x1080.bin is given
1053 and no file with the same name exists. Details and
1054 instructions how to build your own EDID data are
1055 available in Documentation/admin-guide/edid.rst. An EDID
1056 data set will only be used for a particular connector,
1057 if its name and a colon are prepended to the EDID
1058 name. Each connector may use a unique EDID data
1059 set by separating the files with a comma. An EDID
1060 data set with no connector name will be used for
1061 any connectors not explicitly specified.
1066 Format: {"off" | "known"}
1067 Control how the dt_cpu_ftrs device-tree binding is
1068 used for CPU feature discovery and setup (if it
1070 off: Do not use it, fall back to legacy cpu table.
1071 known: Do not pass through unknown features to guests
1072 or userspace, only those that the kernel is aware of.
1074 dump_apple_properties [X86]
1075 Dump name and content of EFI device properties on
1076 x86 Macs. Useful for driver authors to determine
1077 what data is available or for reverse-engineering.
1079 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
1080 <module>.dyndbg[="val"]
1081 Enable debug messages at boot time. See
1082 Documentation/admin-guide/dynamic-debug-howto.rst
1085 nopku [X86] Disable Memory Protection Keys CPU feature found
1088 <module>.async_probe [KNL]
1089 Enable asynchronous probe on this module.
1091 early_ioremap_debug [KNL]
1092 Enable debug messages in early_ioremap support. This
1093 is useful for tracking down temporary early mappings
1094 which are not unmapped.
1096 earlycon= [KNL] Output early console device and options.
1098 When used with no options, the early console is
1099 determined by stdout-path property in device tree's
1100 chosen node or the ACPI SPCR table if supported by
1103 cdns,<addr>[,options]
1104 Start an early, polled-mode console on a Cadence
1105 (xuartps) serial port at the specified address. Only
1106 supported option is baud rate. If baud rate is not
1107 specified, the serial port must already be setup and
1110 uart[8250],io,<addr>[,options]
1111 uart[8250],mmio,<addr>[,options]
1112 uart[8250],mmio32,<addr>[,options]
1113 uart[8250],mmio32be,<addr>[,options]
1114 uart[8250],0x<addr>[,options]
1115 Start an early, polled-mode console on the 8250/16550
1116 UART at the specified I/O port or MMIO address.
1117 MMIO inter-register address stride is either 8-bit
1118 (mmio) or 32-bit (mmio32 or mmio32be).
1119 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1120 to be equivalent to 'mmio'. 'options' are specified
1121 in the same format described for "console=ttyS<n>"; if
1122 unspecified, the h/w is not initialized.
1126 Start an early, polled-mode console on a pl011 serial
1127 port at the specified address. The pl011 serial port
1128 must already be setup and configured. Options are not
1129 yet supported. If 'mmio32' is specified, then only
1130 the driver will use only 32-bit accessors to read/write
1131 the device registers.
1134 Start an early console on a litex serial port at the
1135 specified address. The serial port must already be
1136 setup and configured. Options are not yet supported.
1139 Start an early, polled-mode console on a meson serial
1140 port at the specified address. The serial port must
1141 already be setup and configured. Options are not yet
1145 Start an early, polled-mode console on an msm serial
1146 port at the specified address. The serial port
1147 must already be setup and configured. Options are not
1150 msm_serial_dm,<addr>
1151 Start an early, polled-mode console on an msm serial
1152 dm port at the specified address. The serial port
1153 must already be setup and configured. Options are not
1157 Start an early, polled-mode console on a serial port
1158 of an Actions Semi SoC, such as S500 or S900, at the
1159 specified address. The serial port must already be
1160 setup and configured. Options are not yet supported.
1163 Start an early, polled-mode console on a serial port
1164 of an RDA Micro SoC, such as RDA8810PL, at the
1165 specified address. The serial port must already be
1166 setup and configured. Options are not yet supported.
1169 Use RISC-V SBI (Supervisor Binary Interface) for early
1172 smh Use ARM semihosting calls for early console.
1180 Use early console provided by serial driver available
1181 on Samsung SoCs, requires selecting proper type and
1182 a correct base address of the selected UART port. The
1183 serial port must already be setup and configured.
1184 Options are not yet supported.
1187 Start an early, polled-mode console on a lantiq serial
1188 (lqasc) port at the specified address. The serial port
1189 must already be setup and configured. Options are not
1194 Use early console provided by Freescale LP UART driver
1195 found on Freescale Vybrid and QorIQ LS1021A processors.
1196 A valid base address must be provided, and the serial
1197 port must already be setup and configured.
1201 Start an early, polled-mode, output-only console on the
1202 Freescale i.MX UART at the specified address. The UART
1203 must already be setup and configured.
1206 Start an early, polled-mode console on the
1207 Armada 3700 serial port at the specified
1208 address. The serial port must already be setup
1209 and configured. Options are not yet supported.
1212 Start an early, polled-mode console on a Qualcomm
1213 Generic Interface (GENI) based serial port at the
1214 specified address. The serial port must already be
1215 setup and configured. Options are not yet supported.
1218 Start an early, unaccelerated console on the EFI
1219 memory mapped framebuffer (if available). On cache
1220 coherent non-x86 systems that use system memory for
1221 the framebuffer, pass the 'ram' option so that it is
1222 mapped with the correct attributes.
1225 Use early console provided by Freescale LINFlexD UART
1226 serial driver for NXP S32V234 SoCs. A valid base
1227 address must be provided, and the serial port must
1228 already be setup and configured.
1230 earlyprintk= [X86,SH,ARM,M68k,S390]
1234 earlyprintk=serial[,ttySn[,baudrate]]
1235 earlyprintk=serial[,0x...[,baudrate]]
1236 earlyprintk=ttySn[,baudrate]
1237 earlyprintk=dbgp[debugController#]
1238 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1239 earlyprintk=xdbc[xhciController#]
1241 earlyprintk is useful when the kernel crashes before
1242 the normal console is initialized. It is not enabled by
1243 default because it has some cosmetic problems.
1245 Append ",keep" to not disable it when the real console
1248 Only one of vga, efi, serial, or usb debug port can
1251 Currently only ttyS0 and ttyS1 may be specified by
1252 name. Other I/O ports may be explicitly specified
1253 on some architectures (x86 and arm at least) by
1254 replacing ttySn with an I/O port address, like this:
1255 earlyprintk=serial,0x1008,115200
1256 You can find the port for a given device in
1257 /proc/tty/driver/serial:
1258 2: uart:ST16650V2 port:00001008 irq:18 ...
1260 Interaction with the standard serial driver is not
1263 The VGA and EFI output is eventually overwritten by
1266 The xen output can only be used by Xen PV guests.
1268 The sclp output can only be used on s390.
1270 The optional "force" to "pciserial" enables use of a
1271 PCI device even when its classcode is not of the
1274 edac_report= [HW,EDAC] Control how to report EDAC event
1275 Format: {"on" | "off" | "force"}
1276 on: enable EDAC to report H/W event. May be overridden
1277 by other higher priority error reporting module.
1278 off: disable H/W event reporting through EDAC.
1279 force: enforce the use of EDAC to report H/W event.
1282 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1285 This is designed to be used in conjunction with
1286 the boot argument: earlyprintk=vga
1288 This parameter works in place of the kgdboc parameter
1289 but can only be used if the backing tty is available
1290 very early in the boot process. For early debugging
1291 via a serial port see kgdboc_earlycon instead.
1294 Format: {"off" | "on" | "skip[mbr]"}
1297 Format: { "debug", "disable_early_pci_dma",
1298 "nochunk", "noruntime", "nosoftreserve",
1299 "novamap", "no_disable_early_pci_dma" }
1300 debug: enable misc debug output.
1301 disable_early_pci_dma: disable the busmaster bit on all
1302 PCI bridges while in the EFI boot stub.
1303 nochunk: disable reading files in "chunks" in the EFI
1304 boot stub, as chunking can cause problems with some
1305 firmware implementations.
1306 noruntime : disable EFI runtime services support
1307 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1308 attribute may cause the kernel to reserve the
1309 memory range for a memory mapping driver to
1310 claim. Specify efi=nosoftreserve to disable this
1311 reservation and treat the memory by its base type
1312 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1313 novamap: do not call SetVirtualAddressMap().
1314 no_disable_early_pci_dma: Leave the busmaster bit set
1315 on all PCI bridges while in the EFI boot stub
1317 efi_no_storage_paranoia [EFI; X86]
1318 Using this parameter you can use more than 50% of
1319 your efi variable storage. Use this parameter only if
1320 you are really sure that your UEFI does sane gc and
1321 fulfills the spec otherwise your board may brick.
1323 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1324 Add arbitrary attribute to specific memory range by
1325 updating original EFI memory map.
1326 Region of memory which aa attribute is added to is
1329 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1330 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1331 attribute is added to range 0x100000000-0x180000000 and
1332 0x10a0000000-0x1120000000.
1334 If efi_fake_mem=8G@9G:0x40000 is specified, the
1335 EFI_MEMORY_SP(0x40000) attribute is added to
1336 range 0x240000000-0x43fffffff.
1338 Using this parameter you can do debugging of EFI memmap
1339 related features. For example, you can do debugging of
1340 Address Range Mirroring feature even if your box
1341 doesn't support it, or mark specific memory as
1344 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1345 that is to be dynamically loaded by Linux. If there are
1346 multiple variables with the same name but with different
1347 vendor GUIDs, all of them will be loaded. See
1348 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1351 eisa_irq_edge= [PARISC,HW]
1352 See header of drivers/parisc/eisa.c.
1355 See comment before function elanfreq_setup() in
1356 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1358 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1359 Specifies physical address of start of kernel core
1360 image elf header and optionally the size. Generally
1361 kexec loader will pass this option to capture kernel.
1362 See Documentation/admin-guide/kdump/kdump.rst for details.
1364 enable_mtrr_cleanup [X86]
1365 The kernel tries to adjust MTRR layout from continuous
1366 to discrete, to make X server driver able to add WB
1367 entry later. This parameter enables that.
1369 enable_timer_pin_1 [X86]
1370 Enable PIN 1 of APIC timer
1371 Can be useful to work around chipset bugs
1372 (in particular on some ATI chipsets).
1373 The kernel tries to set a reasonable default.
1375 enforcing [SELINUX] Set initial enforcing status.
1377 See security/selinux/Kconfig help text.
1378 0 -- permissive (log only, no denials).
1379 1 -- enforcing (deny and log).
1381 Value can be changed at runtime via
1382 /sys/fs/selinux/enforce.
1385 Disable Error Record Serialization Table (ERST)
1388 ether= [HW,NET] Ethernet cards parameters
1389 This option is obsoleted by the "netdev=" option, which
1390 has equivalent usage. See its documentation for details.
1394 Permit 'security.evm' to be updated regardless of
1395 current integrity status.
1400 fail_make_request=[KNL]
1401 General fault injection mechanism.
1402 Format: <interval>,<probability>,<space>,<times>
1403 See also Documentation/fault-injection/.
1406 Format: { initns | none }
1407 See Documentation/admin-guide/sysctl/net.rst for
1408 fb_tunnels_only_for_init_ns
1411 See Documentation/admin-guide/blockdev/floppy.rst.
1413 force_pal_cache_flush
1414 [IA-64] Avoid check_sal_cache_flush which may hang on
1415 buggy SAL_CACHE_FLUSH implementations. Using this
1416 parameter will force ia64_sal_cache_flush to call
1417 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1420 Forcefully enable Physical Address Extension (PAE).
1421 Many Pentium M systems disable PAE but may have a
1422 functionally usable PAE implementation.
1423 Warning: use of this parameter will taint the kernel
1424 and may cause unknown problems.
1427 [FTRACE] will set and start the specified tracer
1428 as early as possible in order to facilitate early
1431 ftrace_dump_on_oops[=orig_cpu]
1432 [FTRACE] will dump the trace buffers on oops.
1433 If no parameter is passed, ftrace will dump
1434 buffers of all CPUs, but if you pass orig_cpu, it will
1435 dump only the buffer of the CPU that triggered the
1438 ftrace_filter=[function-list]
1439 [FTRACE] Limit the functions traced by the function
1440 tracer at boot up. function-list is a comma-separated
1441 list of functions. This list can be changed at run
1442 time by the set_ftrace_filter file in the debugfs
1445 ftrace_notrace=[function-list]
1446 [FTRACE] Do not trace the functions specified in
1447 function-list. This list can be changed at run time
1448 by the set_ftrace_notrace file in the debugfs
1451 ftrace_graph_filter=[function-list]
1452 [FTRACE] Limit the top level callers functions traced
1453 by the function graph tracer at boot up.
1454 function-list is a comma-separated list of functions
1455 that can be changed at run time by the
1456 set_graph_function file in the debugfs tracing directory.
1458 ftrace_graph_notrace=[function-list]
1459 [FTRACE] Do not trace from the functions specified in
1460 function-list. This list is a comma-separated list of
1461 functions that can be changed at run time by the
1462 set_graph_notrace file in the debugfs tracing directory.
1464 ftrace_graph_max_depth=<uint>
1465 [FTRACE] Used with the function graph tracer. This is
1466 the max depth it will trace into a function. This value
1467 can be changed at run time by the max_graph_depth file
1468 in the tracefs tracing directory. default: 0 (no limit)
1470 fw_devlink= [KNL] Create device links between consumer and supplier
1471 devices by scanning the firmware to infer the
1472 consumer/supplier relationships. This feature is
1473 especially useful when drivers are loaded as modules as
1474 it ensures proper ordering of tasks like device probing
1475 (suppliers first, then consumers), supplier boot state
1476 clean up (only after all consumers have probed),
1477 suspend/resume & runtime PM (consumers first, then
1479 Format: { off | permissive | on | rpm }
1480 off -- Don't create device links from firmware info.
1481 permissive -- Create device links from firmware info
1482 but use it only for ordering boot state clean
1483 up (sync_state() calls).
1484 on -- Create device links from firmware info and use it
1485 to enforce probe and suspend/resume ordering.
1486 rpm -- Like "on", but also use to order runtime PM.
1488 fw_devlink.strict=<bool>
1489 [KNL] Treat all inferred dependencies as mandatory
1490 dependencies. This only applies for fw_devlink=on|rpm.
1494 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1495 support via parallel port (up to 5 devices per port)
1496 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1497 See also Documentation/input/devices/joystick-parport.rst
1501 gart_fix_e820= [X86-64] disable the fix e820 for K8 GART
1505 gcov_persist= [GCOV] When non-zero (default), profiling data for
1506 kernel modules is saved and remains accessible via
1507 debugfs, even when the module is unloaded/reloaded.
1508 When zero, profiling data is discarded and associated
1509 debugfs files are removed at module unload time.
1511 goldfish [X86] Enable the goldfish android emulator platform.
1512 Don't use this when you are not running on the
1515 gpio-mockup.gpio_mockup_ranges
1516 [HW] Sets the ranges of gpiochip of for this device.
1517 Format: <start1>,<end1>,<start2>,<end2>...
1518 gpio-mockup.gpio_mockup_named_lines
1519 [HW] Let the driver know GPIO lines should be named.
1521 gpt [EFI] Forces disk with valid GPT signature but
1522 invalid Protective MBR to be treated as GPT. If the
1523 primary GPT is corrupted, it enables the backup/alternate
1524 GPT to be used instead.
1526 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1527 the "Enable 0" bit of the configuration register.
1530 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1531 the "Enable 0" bit of the configuration register.
1534 grcan.select= [HW] Select which physical interface to use.
1537 grcan.txsize= [HW] Sets the size of the tx buffer.
1538 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1540 grcan.rxsize= [HW] Sets the size of the rx buffer.
1541 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1544 hardlockup_all_cpu_backtrace=
1545 [KNL] Should the hard-lockup detector generate
1546 backtraces on all cpus.
1549 hashdist= [KNL,NUMA] Large hashes allocated during boot
1550 are distributed across NUMA nodes. Defaults on
1551 for 64-bit NUMA, off otherwise.
1552 Format: 0 | 1 (for off | on)
1554 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1556 hd= [EIDE] (E)IDE hard drive subsystem geometry
1557 Format: <cyl>,<head>,<sect>
1560 Disable Hardware Error Source Table (HEST) support;
1561 corresponding firmware-first mode error processing
1562 logic will be disabled.
1564 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1565 size of <nn>. This works even on boxes that have no
1566 highmem otherwise. This also works to reduce highmem
1567 size on bigger boxes.
1569 highres= [KNL] Enable/disable high resolution timer mode.
1570 Valid parameters: "on", "off"
1575 hpet= [X86-32,HPET] option to control HPET usage
1576 Format: { enable (default) | disable | force |
1578 disable: disable HPET and use PIT instead
1579 force: allow force enabled of undocumented chips (ICH4,
1581 verbose: show contents of HPET registers during setup
1583 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1584 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1586 hugetlb_cma= [HW,CMA] The size of a CMA area used for allocation
1587 of gigantic hugepages.
1590 Reserve a CMA area of given size and allocate gigantic
1591 hugepages using the CMA allocator. If enabled, the
1592 boot-time allocation of gigantic hugepages is skipped.
1594 hugepages= [HW] Number of HugeTLB pages to allocate at boot.
1595 If this follows hugepagesz (below), it specifies
1596 the number of pages of hugepagesz to be allocated.
1597 If this is the first HugeTLB parameter on the command
1598 line, it specifies the number of pages to allocate for
1599 the default huge page size. See also
1600 Documentation/admin-guide/mm/hugetlbpage.rst.
1604 [HW] The size of the HugeTLB pages. This is used in
1605 conjunction with hugepages (above) to allocate huge
1606 pages of a specific size at boot. The pair
1607 hugepagesz=X hugepages=Y can be specified once for
1608 each supported huge page size. Huge page sizes are
1609 architecture dependent. See also
1610 Documentation/admin-guide/mm/hugetlbpage.rst.
1613 hugetlb_free_vmemmap=
1614 [KNL] Reguires CONFIG_HUGETLB_PAGE_FREE_VMEMMAP
1616 Allows heavy hugetlb users to free up some more
1617 memory (6 * PAGE_SIZE for each 2MB hugetlb page).
1618 Format: { on | off (default) }
1620 on: enable the feature
1621 off: disable the feature
1623 Built with CONFIG_HUGETLB_PAGE_FREE_VMEMMAP_DEFAULT_ON=y,
1626 This is not compatible with memory_hotplug.memmap_on_memory.
1627 If both parameters are enabled, hugetlb_free_vmemmap takes
1628 precedence over memory_hotplug.memmap_on_memory.
1631 [KNL] Should the hung task detector generate panics.
1634 A value of 1 instructs the kernel to panic when a
1635 hung task is detected. The default value is controlled
1636 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1637 option. The value selected by this boot parameter can
1638 be changed later by the kernel.hung_task_panic sysctl.
1640 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1641 terminal devices. Valid values: 0..8
1642 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1643 If specified, z/VM IUCV HVC accepts connections
1644 from listed z/VM user IDs only.
1646 hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations
1647 which allow the hypervisor to 'idle' the
1648 guest on lock contention.
1651 Do not unregister boot console at start. This is only
1652 useful for debugging when something happens in the window
1653 between unregistering the boot console and initializing
1656 i2c_bus= [HW] Override the default board specific I2C bus speed
1657 or register an additional I2C bus that is not
1658 registered from board initialization code.
1662 i8042.debug [HW] Toggle i8042 debug mode
1663 i8042.unmask_kbd_data
1664 [HW] Enable printing of interrupt data from the KBD port
1665 (disabled by default, and as a pre-condition
1666 requires that i8042.debug=1 be enabled)
1667 i8042.direct [HW] Put keyboard port into non-translated mode
1668 i8042.dumbkbd [HW] Pretend that controller can only read data from
1669 keyboard and cannot control its state
1670 (Don't attempt to blink the leds)
1671 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1672 i8042.nokbd [HW] Don't check/create keyboard port
1673 i8042.noloop [HW] Disable the AUX Loopback command while probing
1675 i8042.nomux [HW] Don't check presence of an active multiplexing
1677 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1679 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1680 i8042.reset [HW] Reset the controller during init, cleanup and
1681 suspend-to-ram transitions, only during s2r
1682 transitions, or never reset
1683 Format: { 1 | Y | y | 0 | N | n }
1684 1, Y, y: always reset controller
1685 0, N, n: don't ever reset controller
1686 Default: only on s2r transitions on x86; most other
1687 architectures force reset to be always executed
1688 i8042.unlock [HW] Unlock (ignore) the keylock
1689 i8042.kbdreset [HW] Reset device connected to KBD port
1693 i8k.ignore_dmi [HW] Continue probing hardware even if DMI data
1694 indicates that the driver is running on unsupported
1696 i8k.force [HW] Activate i8k driver even if SMM BIOS signature
1697 does not match list of supported models.
1699 [HW] Report power status in /proc/i8k
1700 (disabled by default)
1701 i8k.restricted [HW] Allow controlling fans only if SYS_ADMIN
1704 i915.invert_brightness=
1705 [DRM] Invert the sense of the variable that is used to
1706 set the brightness of the panel backlight. Normally a
1707 brightness value of 0 indicates backlight switched off,
1708 and the maximum of the brightness value sets the backlight
1709 to maximum brightness. If this parameter is set to 0
1710 (default) and the machine requires it, or this parameter
1711 is set to 1, a brightness value of 0 sets the backlight
1712 to maximum brightness, and the maximum of the brightness
1713 value switches the backlight off.
1714 -1 -- never invert brightness
1715 0 -- machine default
1716 1 -- force brightness inversion
1719 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1721 ide-core.nodma= [HW] (E)IDE subsystem
1722 Format: =0.0 to prevent dma on hda, =0.1 hdb =1.0 hdc
1723 .vlb_clock .pci_clock .noflush .nohpa .noprobe .nowerr
1724 .cdrom .chs .ignore_cable are additional options
1725 See Documentation/ide/ide.rst.
1727 ide-generic.probe-mask= [HW] (E)IDE subsystem
1729 Probe mask for legacy ISA IDE ports. Depending on
1730 platform up to 6 ports are supported, enabled by
1731 setting corresponding bits in the mask to 1. The
1732 default value is 0x0, which has a special meaning.
1733 On systems that have PCI, it triggers scanning the
1734 PCI bus for the first and the second port, which
1735 are then probed. On systems without PCI the value
1736 of 0x0 enables probing the two first ports as if it
1739 ide-pci-generic.all-generic-ide [HW] (E)IDE subsystem
1740 Claim all unknown PCI IDE storage controllers.
1743 Format: idle=poll, idle=halt, idle=nomwait
1744 Poll forces a polling idle loop that can slightly
1745 improve the performance of waking up a idle CPU, but
1746 will use a lot of power and make the system run hot.
1748 idle=halt: Halt is forced to be used for CPU idle.
1749 In such case C2/C3 won't be used again.
1750 idle=nomwait: Disable mwait for CPU C-states
1754 Allow force disabling of Shared Virtual Memory (SVA)
1755 support for the idxd driver. By default it is set to
1758 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1759 Format: { strict | legacy | 2008 | relaxed }
1762 Choose which programs will be accepted for execution
1763 based on the IEEE 754 NaN encoding(s) supported by
1764 the FPU and the NaN encoding requested with the value
1765 of an ELF file header flag individually set by each
1766 binary. Hardware implementations are permitted to
1767 support either or both of the legacy and the 2008 NaN
1770 Available settings are as follows:
1771 strict accept binaries that request a NaN encoding
1772 supported by the FPU
1773 legacy only accept legacy-NaN binaries, if supported
1775 2008 only accept 2008-NaN binaries, if supported
1777 relaxed accept any binaries regardless of whether
1778 supported by the FPU
1780 The FPU emulator is always able to support both NaN
1781 encodings, so if no FPU hardware is present or it has
1782 been disabled with 'nofpu', then the settings of
1783 'legacy' and '2008' strap the emulator accordingly,
1784 'relaxed' straps the emulator for both legacy-NaN and
1785 2008-NaN, whereas 'strict' enables legacy-NaN only on
1786 legacy processors and both NaN encodings on MIPS32 or
1789 The setting for ABS.fmt/NEG.fmt instruction execution
1790 mode generally follows that for the NaN encoding,
1791 except where unsupported by hardware.
1793 ignore_loglevel [KNL]
1794 Ignore loglevel setting - this will print /all/
1795 kernel messages to the console. Useful for debugging.
1796 We also add it as printk module parameter, so users
1797 could change it dynamically, usually by
1798 /sys/module/printk/parameters/ignore_loglevel.
1801 Ignore RLIMIT_DATA setting for data mappings,
1802 print warning at first misuse. Can be changed via
1803 /sys/module/kernel/parameters/ignore_rlimit_data.
1805 ihash_entries= [KNL]
1806 Set number of hash buckets for inode cache.
1808 ima_appraise= [IMA] appraise integrity measurements
1809 Format: { "off" | "enforce" | "fix" | "log" }
1812 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
1813 The builtin appraise policy appraises all files
1816 ima_canonical_fmt [IMA]
1817 Use the canonical format for the binary runtime
1818 measurements, instead of host native format.
1821 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1825 The list of supported hash algorithms is defined
1826 in crypto/hash_info.h.
1829 The builtin policies to load during IMA setup.
1830 Format: "tcb | appraise_tcb | secure_boot |
1831 fail_securely | critical_data"
1833 The "tcb" policy measures all programs exec'd, files
1834 mmap'd for exec, and all files opened with the read
1835 mode bit set by either the effective uid (euid=0) or
1838 The "appraise_tcb" policy appraises the integrity of
1839 all files owned by root.
1841 The "secure_boot" policy appraises the integrity
1842 of files (eg. kexec kernel image, kernel modules,
1843 firmware, policy, etc) based on file signatures.
1845 The "fail_securely" policy forces file signature
1846 verification failure also on privileged mounted
1847 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
1850 The "critical_data" policy measures kernel integrity
1853 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
1854 Load a policy which meets the needs of the Trusted
1855 Computing Base. This means IMA will measure all
1856 programs exec'd, files mmap'd for exec, and all files
1857 opened for read by uid=0.
1860 Select one of defined IMA measurements template formats.
1861 Formats: { "ima" | "ima-ng" | "ima-sig" }
1865 [IMA] Define a custom template format.
1866 Format: { "field1|...|fieldN" }
1868 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
1869 Format: <min_file_size>
1870 Set the minimal file size for using asynchronous hash.
1871 If left unspecified, ahash usage is disabled.
1873 ahash performance varies for different data sizes on
1874 different crypto accelerators. This option can be used
1875 to achieve the best performance for a particular HW.
1877 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
1879 Set hashing buffer size. Default: 4k.
1881 ahash performance varies for different chunk sizes on
1882 different crypto accelerators. This option can be used
1883 to achieve best performance for particular HW.
1887 Run specified binary instead of /sbin/init as init
1890 initcall_debug [KNL] Trace initcalls as they are executed. Useful
1891 for working out where the kernel is dying during
1894 initcall_blacklist= [KNL] Do not execute a comma-separated list of
1895 initcall functions. Useful for debugging built-in
1896 modules and initcalls.
1898 initramfs_async= [KNL]
1901 This parameter controls whether the initramfs
1902 image is unpacked asynchronously, concurrently
1903 with devices being probed and
1904 initialized. This should normally just work,
1905 but as a debugging aid, one can get the
1906 historical behaviour of the initramfs
1907 unpacking being completed before device_ and
1910 initrd= [BOOT] Specify the location of the initial ramdisk
1912 initrdmem= [KNL] Specify a physical address and size from which to
1913 load the initrd. If an initrd is compiled in or
1914 specified in the bootparams, it takes priority over this
1916 Format: ss[KMG],nn[KMG]
1919 init_on_alloc= [MM] Fill newly allocated pages and heap objects with
1922 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
1924 init_on_free= [MM] Fill freed pages and heap objects with zeroes.
1926 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
1928 init_pkru= [X86] Specify the default memory protection keys rights
1929 register contents for all processes. 0x55555554 by
1930 default (disallow access to all but pkey 0). Can
1931 override in debugfs after boot.
1933 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
1936 int_pln_enable [X86] Enable power limit notification interrupt
1938 integrity_audit=[IMA]
1939 Format: { "0" | "1" }
1940 0 -- basic integrity auditing messages. (Default)
1941 1 -- additional integrity auditing messages.
1943 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
1945 Enable intel iommu driver.
1947 Disable intel iommu driver.
1948 igfx_off [Default Off]
1949 By default, gfx is mapped as normal device. If a gfx
1950 device has a dedicated DMAR unit, the DMAR unit is
1951 bypassed by not enabling DMAR with this option. In
1952 this case, gfx device will use physical address for
1954 strict [Default Off]
1955 With this option on every unmap_single operation will
1956 result in a hardware IOTLB flush operation as opposed
1957 to batching them for performance.
1958 sp_off [Default Off]
1959 By default, super page will be supported if Intel IOMMU
1960 has the capability. With this option, super page will
1963 By default, scalable mode will be disabled even if the
1964 hardware advertises that it has support for the scalable
1965 mode translation. With this option set, scalable mode
1966 will be used on hardware which claims to support it.
1967 tboot_noforce [Default Off]
1968 Do not force the Intel IOMMU enabled under tboot.
1969 By default, tboot will force Intel IOMMU on, which
1970 could harm performance of some high-throughput
1971 devices like 40GBit network cards, even if identity
1973 Note that using this option lowers the security
1974 provided by tboot because it makes the system
1975 vulnerable to DMA attacks.
1977 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
1978 0 disables intel_idle and fall back on acpi_idle.
1979 1 to 9 specify maximum depth of C-state.
1983 Do not enable intel_pstate as the default
1984 scaling driver for the supported processors
1986 Use intel_pstate as a scaling driver, but configure it
1987 to work with generic cpufreq governors (instead of
1988 enabling its internal governor). This mode cannot be
1989 used along with the hardware-managed P-states (HWP)
1992 Enable intel_pstate on systems that prohibit it by default
1993 in favor of acpi-cpufreq. Forcing the intel_pstate driver
1994 instead of acpi-cpufreq may disable platform features, such
1995 as thermal controls and power capping, that rely on ACPI
1996 P-States information being indicated to OSPM and therefore
1997 should be used with caution. This option does not work with
1998 processors that aren't supported by the intel_pstate driver
1999 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
2001 Do not enable hardware P state control (HWP)
2004 Only load intel_pstate on systems which support
2005 hardware P state control (HWP) if available.
2007 Enforce ACPI _PPC performance limits. If the Fixed ACPI
2008 Description Table, specifies preferred power management
2009 profile as "Enterprise Server" or "Performance Server",
2010 then this feature is turned on by default.
2012 Allow per-logical-CPU P-State performance control limits using
2013 cpufreq sysfs interface
2015 intremap= [X86-64, Intel-IOMMU]
2016 on enable Interrupt Remapping (default)
2017 off disable Interrupt Remapping
2018 nosid disable Source ID checking
2020 BIOS x2APIC opt-out request will be ignored
2021 nopost disable Interrupt Posting
2023 iomem= Disable strict checking of access to MMIO memory
2024 strict regions from userspace.
2039 nobypass [PPC/POWERNV]
2040 Disable IOMMU bypass, using IOMMU for PCI devices.
2042 iommu.forcedac= [ARM64, X86] Control IOVA allocation for PCI devices.
2043 Format: { "0" | "1" }
2044 0 - Try to allocate a 32-bit DMA address first, before
2045 falling back to the full range if needed.
2046 1 - Allocate directly from the full usable range,
2047 forcing Dual Address Cycle for PCI cards supporting
2048 greater than 32-bit addressing.
2050 iommu.strict= [ARM64, X86] Configure TLB invalidation behaviour
2051 Format: { "0" | "1" }
2053 Request that DMA unmap operations use deferred
2054 invalidation of hardware TLBs, for increased
2055 throughput at the cost of reduced device isolation.
2056 Will fall back to strict mode if not supported by
2057 the relevant IOMMU driver.
2058 1 - Strict mode (default).
2059 DMA unmap operations invalidate IOMMU hardware TLBs
2061 Note: on x86, the default behaviour depends on the
2062 equivalent driver-specific parameters, but a strict
2063 mode explicitly specified by either method takes
2067 [ARM64, X86] Configure DMA to bypass the IOMMU by default.
2068 Format: { "0" | "1" }
2069 0 - Use IOMMU translation for DMA.
2070 1 - Bypass the IOMMU for DMA.
2071 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
2073 io7= [HW] IO7 for Marvel-based Alpha systems
2074 See comment before marvel_specify_io7 in
2075 arch/alpha/kernel/core_marvel.c.
2077 io_delay= [X86] I/O delay method
2079 Standard port 0x80 based delay
2081 Alternate port 0xed based delay (needed on some systems)
2083 Simple two microseconds delay
2088 See Documentation/admin-guide/nfs/nfsroot.rst.
2090 ipcmni_extend [KNL] Extend the maximum number of unique System V
2091 IPC identifiers from 32,768 to 16,777,216.
2093 irqaffinity= [SMP] Set the default irq affinity mask
2094 The argument is a cpu list, as described above.
2096 irqchip.gicv2_force_probe=
2099 Force the kernel to look for the second 4kB page
2100 of a GICv2 controller even if the memory range
2101 exposed by the device tree is too small.
2103 irqchip.gicv3_nolpi=
2105 Force the kernel to ignore the availability of
2106 LPIs (and by consequence ITSs). Intended for system
2107 that use the kernel as a bootloader, and thus want
2108 to let secondary kernels in charge of setting up
2111 irqchip.gicv3_pseudo_nmi= [ARM64]
2112 Enables support for pseudo-NMIs in the kernel. This
2113 requires the kernel to be built with
2114 CONFIG_ARM64_PSEUDO_NMI.
2117 When an interrupt is not handled search all handlers
2118 for it. Intended to get systems with badly broken
2122 When an interrupt is not handled search all handlers
2123 for it. Also check all handlers each timer
2124 interrupt. Intended to get systems with badly broken
2128 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
2130 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
2131 [Deprecated - use cpusets instead]
2132 Format: [flag-list,]<cpu-list>
2134 Specify one or more CPUs to isolate from disturbances
2135 specified in the flag list (default: domain):
2138 Disable the tick when a single task runs.
2140 A residual 1Hz tick is offloaded to workqueues, which you
2141 need to affine to housekeeping through the global
2142 workqueue's affinity configured via the
2143 /sys/devices/virtual/workqueue/cpumask sysfs file, or
2144 by using the 'domain' flag described below.
2146 NOTE: by default the global workqueue runs on all CPUs,
2147 so to protect individual CPUs the 'cpumask' file has to
2148 be configured manually after bootup.
2151 Isolate from the general SMP balancing and scheduling
2152 algorithms. Note that performing domain isolation this way
2153 is irreversible: it's not possible to bring back a CPU to
2154 the domains once isolated through isolcpus. It's strongly
2155 advised to use cpusets instead to disable scheduler load
2156 balancing through the "cpuset.sched_load_balance" file.
2157 It offers a much more flexible interface where CPUs can
2158 move in and out of an isolated set anytime.
2160 You can move a process onto or off an "isolated" CPU via
2161 the CPU affinity syscalls or cpuset.
2162 <cpu number> begins at 0 and the maximum value is
2163 "number of CPUs in system - 1".
2167 Isolate from being targeted by managed interrupts
2168 which have an interrupt mask containing isolated
2169 CPUs. The affinity of managed interrupts is
2170 handled by the kernel and cannot be changed via
2171 the /proc/irq/* interfaces.
2173 This isolation is best effort and only effective
2174 if the automatically assigned interrupt mask of a
2175 device queue contains isolated and housekeeping
2176 CPUs. If housekeeping CPUs are online then such
2177 interrupts are directed to the housekeeping CPU
2178 so that IO submitted on the housekeeping CPU
2179 cannot disturb the isolated CPU.
2181 If a queue's affinity mask contains only isolated
2182 CPUs then this parameter has no effect on the
2183 interrupt routing decision, though interrupts are
2184 only delivered when tasks running on those
2185 isolated CPUs submit IO. IO submitted on
2186 housekeeping CPUs has no influence on those
2189 The format of <cpu-list> is described above.
2193 ivrs_ioapic [HW,X86-64]
2194 Provide an override to the IOAPIC-ID<->DEVICE-ID
2195 mapping provided in the IVRS ACPI table. For
2196 example, to map IOAPIC-ID decimal 10 to
2197 PCI device 00:14.0 write the parameter as:
2198 ivrs_ioapic[10]=00:14.0
2200 ivrs_hpet [HW,X86-64]
2201 Provide an override to the HPET-ID<->DEVICE-ID
2202 mapping provided in the IVRS ACPI table. For
2203 example, to map HPET-ID decimal 0 to
2204 PCI device 00:14.0 write the parameter as:
2205 ivrs_hpet[0]=00:14.0
2207 ivrs_acpihid [HW,X86-64]
2208 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2209 mapping provided in the IVRS ACPI table. For
2210 example, to map UART-HID:UID AMD0020:0 to
2211 PCI device 00:14.5 write the parameter as:
2212 ivrs_acpihid[00:14.5]=AMD0020:0
2214 js= [HW,JOY] Analog joystick
2215 See Documentation/input/joydev/joystick.rst.
2218 When CONFIG_RANDOMIZE_BASE is set, this disables
2219 kernel and module base offset ASLR (Address Space
2220 Layout Randomization).
2223 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2224 report on every invalid memory access. Without this
2225 parameter KASAN will print report only for the first
2230 kernelcore= [KNL,X86,IA-64,PPC]
2231 Format: nn[KMGTPE] | nn% | "mirror"
2232 This parameter specifies the amount of memory usable by
2233 the kernel for non-movable allocations. The requested
2234 amount is spread evenly throughout all nodes in the
2235 system as ZONE_NORMAL. The remaining memory is used for
2236 movable memory in its own zone, ZONE_MOVABLE. In the
2237 event, a node is too small to have both ZONE_NORMAL and
2238 ZONE_MOVABLE, kernelcore memory will take priority and
2239 other nodes will have a larger ZONE_MOVABLE.
2241 ZONE_MOVABLE is used for the allocation of pages that
2242 may be reclaimed or moved by the page migration
2243 subsystem. Note that allocations like PTEs-from-HighMem
2244 still use the HighMem zone if it exists, and the Normal
2245 zone if it does not.
2247 It is possible to specify the exact amount of memory in
2248 the form of "nn[KMGTPE]", a percentage of total system
2249 memory in the form of "nn%", or "mirror". If "mirror"
2250 option is specified, mirrored (reliable) memory is used
2251 for non-movable allocations and remaining memory is used
2252 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2253 are exclusive, so you cannot specify multiple forms.
2255 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
2256 Format: <Controller#>[,poll interval]
2257 The controller # is the number of the ehci usb debug
2258 port as it is probed via PCI. The poll interval is
2259 optional and is the number seconds in between
2260 each poll cycle to the debug port in case you need
2261 the functionality for interrupting the kernel with
2262 gdb or control-c on the dbgp connection. When
2263 not using this parameter you use sysrq-g to break into
2264 the kernel debugger.
2266 kgdboc= [KGDB,HW] kgdb over consoles.
2267 Requires a tty driver that supports console polling,
2268 or a supported polling keyboard driver (non-usb).
2269 Serial only format: <serial_device>[,baud]
2270 keyboard only format: kbd
2271 keyboard and serial format: kbd,<serial_device>[,baud]
2272 Optional Kernel mode setting:
2273 kms, kbd format: kms,kbd
2274 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2276 kgdboc_earlycon= [KGDB,HW]
2277 If the boot console provides the ability to read
2278 characters and can work in polling mode, you can use
2279 this parameter to tell kgdb to use it as a backend
2280 until the normal console is registered. Intended to
2281 be used together with the kgdboc parameter which
2282 specifies the normal console to transition to.
2284 The name of the early console should be specified
2285 as the value of this parameter. Note that the name of
2286 the early console might be different than the tty
2287 name passed to kgdboc. It's OK to leave the value
2288 blank and the first boot console that implements
2289 read() will be picked.
2291 kgdbwait [KGDB] Stop kernel execution and enter the
2292 kernel debugger at the earliest opportunity.
2294 kmac= [MIPS] Korina ethernet MAC address.
2295 Configure the RouterBoard 532 series on-chip
2296 Ethernet adapter MAC address.
2298 kmemleak= [KNL] Boot-time kmemleak enable/disable
2299 Valid arguments: on, off
2301 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2304 kprobe_event=[probe-list]
2305 [FTRACE] Add kprobe events and enable at boot time.
2306 The probe-list is a semicolon delimited list of probe
2307 definitions. Each definition is same as kprobe_events
2308 interface, but the parameters are comma delimited.
2309 For example, to add a kprobe event on vfs_read with
2310 arg1 and arg2, add to the command line;
2312 kprobe_event=p,vfs_read,$arg1,$arg2
2314 See also Documentation/trace/kprobetrace.rst "Kernel
2315 Boot Parameter" section.
2317 kpti= [ARM64] Control page table isolation of user
2318 and kernel address spaces.
2319 Default: enabled on cores which need mitigation.
2323 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2324 Default is 0 (don't ignore, but inject #GP)
2326 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2327 Default is false (don't support).
2329 kvm.mmu_audit= [KVM] This is a R/W parameter which allows audit
2334 [KVM] Controls the software workaround for the
2335 X86_BUG_ITLB_MULTIHIT bug.
2336 force : Always deploy workaround.
2337 off : Never deploy workaround.
2338 auto : Deploy workaround based on the presence of
2339 X86_BUG_ITLB_MULTIHIT.
2343 If the software workaround is enabled for the host,
2344 guests do need not to enable it for nested guests.
2346 kvm.nx_huge_pages_recovery_ratio=
2347 [KVM] Controls how many 4KiB pages are periodically zapped
2348 back to huge pages. 0 disables the recovery, otherwise if
2349 the value is N KVM will zap 1/Nth of the 4KiB pages every
2350 minute. The default is 60.
2352 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM.
2353 Default is 1 (enabled)
2355 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU)
2357 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode.
2360 [KVM,ARM] Select one of KVM/arm64's modes of operation.
2362 nvhe: Standard nVHE-based mode, without support for
2365 protected: nVHE-based mode with support for guests whose
2366 state is kept private from the host.
2367 Not valid if the kernel is running in EL2.
2369 Defaults to VHE/nVHE based on hardware support.
2371 kvm-arm.vgic_v3_group0_trap=
2372 [KVM,ARM] Trap guest accesses to GICv3 group-0
2375 kvm-arm.vgic_v3_group1_trap=
2376 [KVM,ARM] Trap guest accesses to GICv3 group-1
2379 kvm-arm.vgic_v3_common_trap=
2380 [KVM,ARM] Trap guest accesses to GICv3 common
2383 kvm-arm.vgic_v4_enable=
2384 [KVM,ARM] Allow use of GICv4 for direct injection of
2387 kvm_cma_resv_ratio=n [PPC]
2388 Reserves given percentage from system memory area for
2389 contiguous memory allocation for KVM hash pagetable
2391 By default it reserves 5% of total system memory.
2395 kvm-intel.ept= [KVM,Intel] Disable extended page tables
2396 (virtualized MMU) support on capable Intel chips.
2397 Default is 1 (enabled)
2399 kvm-intel.emulate_invalid_guest_state=
2400 [KVM,Intel] Enable emulation of invalid guest states
2401 Default is 0 (disabled)
2403 kvm-intel.flexpriority=
2404 [KVM,Intel] Disable FlexPriority feature (TPR shadow).
2405 Default is 1 (enabled)
2408 [KVM,Intel] Enable VMX nesting (nVMX).
2409 Default is 0 (disabled)
2411 kvm-intel.unrestricted_guest=
2412 [KVM,Intel] Disable unrestricted guest feature
2413 (virtualized real and unpaged mode) on capable
2414 Intel chips. Default is 1 (enabled)
2416 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2419 Valid arguments: never, cond, always
2421 always: L1D cache flush on every VMENTER.
2422 cond: Flush L1D on VMENTER only when the code between
2423 VMEXIT and VMENTER can leak host memory.
2424 never: Disables the mitigation
2426 Default is cond (do L1 cache flush in specific instances)
2428 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification
2429 feature (tagged TLBs) on capable Intel chips.
2430 Default is 1 (enabled)
2432 l1tf= [X86] Control mitigation of the L1TF vulnerability on
2435 The kernel PTE inversion protection is unconditionally
2436 enabled and cannot be disabled.
2439 Provides all available mitigations for the
2440 L1TF vulnerability. Disables SMT and
2441 enables all mitigations in the
2442 hypervisors, i.e. unconditional L1D flush.
2444 SMT control and L1D flush control via the
2445 sysfs interface is still possible after
2446 boot. Hypervisors will issue a warning
2447 when the first VM is started in a
2448 potentially insecure configuration,
2449 i.e. SMT enabled or L1D flush disabled.
2452 Same as 'full', but disables SMT and L1D
2453 flush runtime control. Implies the
2454 'nosmt=force' command line option.
2455 (i.e. sysfs control of SMT is disabled.)
2458 Leaves SMT enabled and enables the default
2459 hypervisor mitigation, i.e. conditional
2462 SMT control and L1D flush control via the
2463 sysfs interface is still possible after
2464 boot. Hypervisors will issue a warning
2465 when the first VM is started in a
2466 potentially insecure configuration,
2467 i.e. SMT enabled or L1D flush disabled.
2471 Disables SMT and enables the default
2472 hypervisor mitigation.
2474 SMT control and L1D flush control via the
2475 sysfs interface is still possible after
2476 boot. Hypervisors will issue a warning
2477 when the first VM is started in a
2478 potentially insecure configuration,
2479 i.e. SMT enabled or L1D flush disabled.
2482 Same as 'flush', but hypervisors will not
2483 warn when a VM is started in a potentially
2484 insecure configuration.
2487 Disables hypervisor mitigations and doesn't
2489 It also drops the swap size and available
2490 RAM limit restriction on both hypervisor and
2495 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2501 lapic [X86-32,APIC] Enable the local APIC even if BIOS
2504 lapic= [X86,APIC] Do not use TSC deadline
2505 value for LAPIC timer one-shot implementation. Default
2506 back to the programmable timer unit in the LAPIC.
2507 Format: notscdeadline
2509 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
2512 libata.dma= [LIBATA] DMA control
2513 libata.dma=0 Disable all PATA and SATA DMA
2514 libata.dma=1 PATA and SATA Disk DMA only
2515 libata.dma=2 ATAPI (CDROM) DMA only
2516 libata.dma=4 Compact Flash DMA only
2517 Combinations also work, so libata.dma=3 enables DMA
2518 for disks and CDROMs, but not CFs.
2520 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2521 libata.ignore_hpa=0 keep BIOS limits (default)
2522 libata.ignore_hpa=1 ignore limits, using full disk
2524 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2528 libata.force= [LIBATA] Force configurations. The format is comma-
2529 separated list of "[ID:]VAL" where ID is
2530 PORT[.DEVICE]. PORT and DEVICE are decimal numbers
2531 matching port, link or device. Basically, it matches
2532 the ATA ID string printed on console by libata. If
2533 the whole ID part is omitted, the last PORT and DEVICE
2534 values are used. If ID hasn't been specified yet, the
2535 configuration applies to all ports, links and devices.
2537 If only DEVICE is omitted, the parameter applies to
2538 the port and all links and devices behind it. DEVICE
2539 number of 0 either selects the first device or the
2540 first fan-out link behind PMP device. It does not
2541 select the host link. DEVICE number of 15 selects the
2542 host link and device attached to it.
2544 The VAL specifies the configuration to force. As long
2545 as there's no ambiguity shortcut notation is allowed.
2546 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2547 The following configurations can be forced.
2549 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2550 Any ID with matching PORT is used.
2552 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2554 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2555 udma[/][16,25,33,44,66,100,133] notation is also
2558 * [no]ncq: Turn on or off NCQ.
2560 * [no]ncqtrim: Turn off queued DSM TRIM.
2562 * nohrst, nosrst, norst: suppress hard, soft
2565 * rstonce: only attempt one reset during
2566 hot-unplug link recovery
2568 * dump_id: dump IDENTIFY data.
2570 * atapi_dmadir: Enable ATAPI DMADIR bridge support
2572 * disable: Disable this device.
2574 If there are multiple matching configurations changing
2575 the same attribute, the last one is used.
2577 memblock=debug [KNL] Enable memblock debug messages.
2579 load_ramdisk= [RAM] [Deprecated]
2581 lockd.nlm_grace_period=P [NFS] Assign grace period.
2584 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2587 lockd.nlm_timeout=T [NFS] Assign timeout value.
2590 lockd.nlm_udpport=M [NFS] Assign UDP port.
2593 lockdown= [SECURITY]
2594 { integrity | confidentiality }
2595 Enable the kernel lockdown feature. If set to
2596 integrity, kernel features that allow userland to
2597 modify the running kernel are disabled. If set to
2598 confidentiality, kernel features that allow userland
2599 to extract confidential information from the kernel
2602 locktorture.nreaders_stress= [KNL]
2603 Set the number of locking read-acquisition kthreads.
2604 Defaults to being automatically set based on the
2605 number of online CPUs.
2607 locktorture.nwriters_stress= [KNL]
2608 Set the number of locking write-acquisition kthreads.
2610 locktorture.onoff_holdoff= [KNL]
2611 Set time (s) after boot for CPU-hotplug testing.
2613 locktorture.onoff_interval= [KNL]
2614 Set time (s) between CPU-hotplug operations, or
2615 zero to disable CPU-hotplug testing.
2617 locktorture.shuffle_interval= [KNL]
2618 Set task-shuffle interval (jiffies). Shuffling
2619 tasks allows some CPUs to go into dyntick-idle
2620 mode during the locktorture test.
2622 locktorture.shutdown_secs= [KNL]
2623 Set time (s) after boot system shutdown. This
2624 is useful for hands-off automated testing.
2626 locktorture.stat_interval= [KNL]
2627 Time (s) between statistics printk()s.
2629 locktorture.stutter= [KNL]
2630 Time (s) to stutter testing, for example,
2631 specifying five seconds causes the test to run for
2632 five seconds, wait for five seconds, and so on.
2633 This tests the locking primitive's ability to
2634 transition abruptly to and from idle.
2636 locktorture.torture_type= [KNL]
2637 Specify the locking implementation to test.
2639 locktorture.verbose= [KNL]
2640 Enable additional printk() statements.
2642 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
2645 loglevel= All Kernel Messages with a loglevel smaller than the
2646 console loglevel will be printed to the console. It can
2647 also be changed with klogd or other programs. The
2648 loglevels are defined as follows:
2650 0 (KERN_EMERG) system is unusable
2651 1 (KERN_ALERT) action must be taken immediately
2652 2 (KERN_CRIT) critical conditions
2653 3 (KERN_ERR) error conditions
2654 4 (KERN_WARNING) warning conditions
2655 5 (KERN_NOTICE) normal but significant condition
2656 6 (KERN_INFO) informational
2657 7 (KERN_DEBUG) debug-level messages
2659 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
2660 in bytes. n must be a power of two and greater
2661 than the minimal size. The minimal size is defined
2662 by LOG_BUF_SHIFT kernel config parameter. There is
2663 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2664 that allows to increase the default size depending on
2665 the number of CPUs. See init/Kconfig for more details.
2667 logo.nologo [FB] Disables display of the built-in Linux logo.
2668 This may be used to provide more screen space for
2669 kernel log messages and is useful when debugging
2670 kernel boot problems.
2672 lp=0 [LP] Specify parallel ports to use, e.g,
2673 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2674 lp=reset first parallel port). 'lp=0' disables the
2675 lp=auto printer driver. 'lp=reset' (which can be
2676 specified in addition to the ports) causes
2677 attached printers to be reset. Using
2678 lp=port1,port2,... specifies the parallel ports
2679 to associate lp devices with, starting with
2680 lp0. A port specification may be 'none' to skip
2681 that lp device, or a parport name such as
2682 'parport0'. Specifying 'lp=auto' instead of a
2683 port specification list means that device IDs
2684 from each port should be examined, to see if
2685 an IEEE 1284-compliant printer is attached; if
2686 so, the driver will manage that printer.
2687 See also header of drivers/char/lp.c.
2690 Sets loops_per_jiffy to given constant, thus avoiding
2691 time-consuming boot-time autodetection (up to 250 ms per
2692 CPU). 0 enables autodetection (default). To determine
2693 the correct value for your kernel, boot with normal
2694 autodetection and see what value is printed. Note that
2695 on SMP systems the preset will be applied to all CPUs,
2696 which is likely to cause problems if your CPUs need
2697 significantly divergent settings. An incorrect value
2698 will cause delays in the kernel to be wrong, leading to
2699 unpredictable I/O errors and other breakage. Although
2700 unlikely, in the extreme case this might damage your
2704 Format: <io>,<irq>,<dma>
2706 lsm.debug [SECURITY] Enable LSM initialization debugging output.
2709 [SECURITY] Choose order of LSM initialization. This
2710 overrides CONFIG_LSM, and the "security=" parameter.
2712 machvec= [IA-64] Force the use of a particular machine-vector
2713 (machvec) in a generic kernel.
2714 Example: machvec=hpzx1
2716 machtype= [Loongson] Share the same kernel image file between
2717 different yeeloong laptops.
2718 Example: machtype=lemote-yeeloong-2f-7inch
2720 max_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory greater
2721 than or equal to this physical address is ignored.
2723 maxcpus= [SMP] Maximum number of processors that an SMP kernel
2724 will bring up during bootup. maxcpus=n : n >= 0 limits
2725 the kernel to bring up 'n' processors. Surely after
2726 bootup you can bring up the other plugged cpu by executing
2727 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
2728 only takes effect during system bootup.
2729 While n=0 is a special case, it is equivalent to "nosmp",
2730 which also disables the IO APIC.
2732 max_loop= [LOOP] The number of loop block devices that get
2733 (loop.max_loop) unconditionally pre-created at init time. The default
2734 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
2735 of statically allocating a predefined number, loop
2736 devices can be requested on-demand with the
2737 /dev/loop-control interface.
2739 mce [X86-32] Machine Check Exception
2741 mce=option [X86-64] See Documentation/x86/x86_64/boot-options.rst
2743 md= [HW] RAID subsystems devices and level
2744 See Documentation/admin-guide/md.rst.
2747 Format: <first>,<last>
2748 Specifies range of consoles to be captured by the MDA.
2751 Control mitigation for the Micro-architectural Data
2752 Sampling (MDS) vulnerability.
2754 Certain CPUs are vulnerable to an exploit against CPU
2755 internal buffers which can forward information to a
2756 disclosure gadget under certain conditions.
2758 In vulnerable processors, the speculatively
2759 forwarded data can be used in a cache side channel
2760 attack, to access data to which the attacker does
2761 not have direct access.
2763 This parameter controls the MDS mitigation. The
2766 full - Enable MDS mitigation on vulnerable CPUs
2767 full,nosmt - Enable MDS mitigation and disable
2768 SMT on vulnerable CPUs
2769 off - Unconditionally disable MDS mitigation
2771 On TAA-affected machines, mds=off can be prevented by
2772 an active TAA mitigation as both vulnerabilities are
2773 mitigated with the same mechanism so in order to disable
2774 this mitigation, you need to specify tsx_async_abort=off
2777 Not specifying this option is equivalent to
2780 For details see: Documentation/admin-guide/hw-vuln/mds.rst
2782 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
2783 Amount of memory to be used in cases as follows:
2786 2 when the kernel is not able to see the whole system memory;
2787 3 memory that lies after 'mem=' boundary is excluded from
2788 the hypervisor, then assigned to KVM guests.
2790 [X86] Work as limiting max address. Use together
2791 with memmap= to avoid physical address space collisions.
2792 Without memmap= PCI devices could be placed at addresses
2793 belonging to unused RAM.
2795 Note that this only takes effects during boot time since
2796 in above case 3, memory may need be hot added after boot
2797 if system memory of hypervisor is not sufficient.
2799 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
2803 [KNL,SH] Allow user to override the default size for
2804 per-device physically contiguous DMA buffers.
2806 memhp_default_state=online/offline
2807 [KNL] Set the initial state for the memory hotplug
2808 onlining policy. If not specified, the default value is
2809 set according to the
2810 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
2812 See Documentation/admin-guide/mm/memory-hotplug.rst.
2814 memmap=exactmap [KNL,X86] Enable setting of an exact
2815 E820 memory map, as specified by the user.
2816 Such memmap=exactmap lines can be constructed based on
2817 BIOS output or other requirements. See the memmap=nn@ss
2820 memmap=nn[KMG]@ss[KMG]
2821 [KNL, X86, MIPS, XTENSA] Force usage of a specific region of memory.
2822 Region of memory to be used is from ss to ss+nn.
2823 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
2824 which limits max address to nn[KMG].
2825 Multiple different regions can be specified,
2828 memmap=100M@2G,100M#3G,1G!1024G
2830 memmap=nn[KMG]#ss[KMG]
2831 [KNL,ACPI] Mark specific memory as ACPI data.
2832 Region of memory to be marked is from ss to ss+nn.
2834 memmap=nn[KMG]$ss[KMG]
2835 [KNL,ACPI] Mark specific memory as reserved.
2836 Region of memory to be reserved is from ss to ss+nn.
2837 Example: Exclude memory from 0x18690000-0x1869ffff
2838 memmap=64K$0x18690000
2840 memmap=0x10000$0x18690000
2841 Some bootloaders may need an escape character before '$',
2842 like Grub2, otherwise '$' and the following number
2845 memmap=nn[KMG]!ss[KMG]
2846 [KNL,X86] Mark specific memory as protected.
2847 Region of memory to be used, from ss to ss+nn.
2848 The memory region may be marked as e820 type 12 (0xc)
2849 and is NVDIMM or ADR memory.
2851 memmap=<size>%<offset>-<oldtype>+<newtype>
2852 [KNL,ACPI] Convert memory within the specified region
2853 from <oldtype> to <newtype>. If "-<oldtype>" is left
2854 out, the whole region will be marked as <newtype>,
2855 even if previously unavailable. If "+<newtype>" is left
2856 out, matching memory will be removed. Types are
2857 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
2858 3 = ACPI, 12 = PRAM.
2860 memory_corruption_check=0/1 [X86]
2861 Some BIOSes seem to corrupt the first 64k of
2862 memory when doing things like suspend/resume.
2863 Setting this option will scan the memory
2864 looking for corruption. Enabling this will
2865 both detect corruption and prevent the kernel
2866 from using the memory being corrupted.
2867 However, its intended as a diagnostic tool; if
2868 repeatable BIOS-originated corruption always
2869 affects the same memory, you can use memmap=
2870 to prevent the kernel from using that memory.
2872 memory_corruption_check_size=size [X86]
2873 By default it checks for corruption in the low
2874 64k, making this memory unavailable for normal
2875 use. Use this parameter to scan for
2876 corruption in more or less memory.
2878 memory_corruption_check_period=seconds [X86]
2879 By default it checks for corruption every 60
2880 seconds. Use this parameter to check at some
2881 other rate. 0 disables periodic checking.
2883 memory_hotplug.memmap_on_memory
2884 [KNL,X86,ARM] Boolean flag to enable this feature.
2885 Format: {on | off (default)}
2886 When enabled, runtime hotplugged memory will
2887 allocate its internal metadata (struct pages)
2888 from the hotadded memory which will allow to
2889 hotadd a lot of memory without requiring
2890 additional memory to do so.
2891 This feature is disabled by default because it
2892 has some implication on large (e.g. GB)
2893 allocations in some configurations (e.g. small
2895 The state of the flag can be read in
2896 /sys/module/memory_hotplug/parameters/memmap_on_memory.
2897 Note that even when enabled, there are a few cases where
2898 the feature is not effective.
2900 This is not compatible with hugetlb_free_vmemmap. If
2901 both parameters are enabled, hugetlb_free_vmemmap takes
2902 precedence over memory_hotplug.memmap_on_memory.
2904 memtest= [KNL,X86,ARM,PPC,RISCV] Enable memtest
2906 default : 0 <disable>
2907 Specifies the number of memtest passes to be
2908 performed. Each pass selects another test
2909 pattern from a given set of patterns. Memtest
2910 fills the memory with this pattern, validates
2911 memory contents and reserves bad memory
2912 regions that are detected.
2914 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
2915 Valid arguments: on, off
2916 Default (depends on kernel configuration option):
2917 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
2918 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
2919 mem_encrypt=on: Activate SME
2920 mem_encrypt=off: Do not activate SME
2922 Refer to Documentation/virt/kvm/amd-memory-encryption.rst
2923 for details on when memory encryption can be activated.
2925 mem_sleep_default= [SUSPEND] Default system suspend mode:
2926 s2idle - Suspend-To-Idle
2927 shallow - Power-On Suspend or equivalent (if supported)
2928 deep - Suspend-To-RAM or equivalent (if supported)
2929 See Documentation/admin-guide/pm/sleep-states.rst.
2931 meye.*= [HW] Set MotionEye Camera parameters
2932 See Documentation/admin-guide/media/meye.rst.
2934 mfgpt_irq= [IA-32] Specify the IRQ to use for the
2935 Multi-Function General Purpose Timers on AMD Geode
2938 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
2939 the BIOS has incorrectly applied a workaround. TinyBIOS
2940 version 0.98 is known to be affected, 0.99 fixes the
2941 problem by letting the user disable the workaround.
2945 min_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory below this
2946 physical address is ignored.
2948 mini2440= [ARM,HW,KNL]
2949 Format:[0..2][b][c][t]
2951 MINI2440 configuration specification:
2952 0 - The attached screen is the 3.5" TFT
2953 1 - The attached screen is the 7" TFT
2954 2 - The VGA Shield is attached (1024x768)
2955 Leaving out the screen size parameter will not load
2956 the TFT driver, and the framebuffer will be left
2958 b - Enable backlight. The TFT backlight pin will be
2959 linked to the kernel VESA blanking code and a GPIO
2960 LED. This parameter is not necessary when using the
2962 c - Enable the s3c camera interface.
2963 t - Reserved for enabling touchscreen support. The
2964 touchscreen support is not enabled in the mainstream
2965 kernel as of 2.6.30, a preliminary port can be found
2966 in the "bleeding edge" mini2440 support kernel at
2967 https://repo.or.cz/w/linux-2.6/mini2440.git
2970 [X86,PPC,S390,ARM64] Control optional mitigations for
2971 CPU vulnerabilities. This is a set of curated,
2972 arch-independent options, each of which is an
2973 aggregation of existing arch-specific options.
2976 Disable all optional CPU mitigations. This
2977 improves system performance, but it may also
2978 expose users to several CPU vulnerabilities.
2979 Equivalent to: nopti [X86,PPC]
2981 nospectre_v1 [X86,PPC]
2983 nospectre_v2 [X86,PPC,S390,ARM64]
2984 spectre_v2_user=off [X86]
2985 spec_store_bypass_disable=off [X86,PPC]
2986 ssbd=force-off [ARM64]
2989 tsx_async_abort=off [X86]
2990 kvm.nx_huge_pages=off [X86]
2991 no_entry_flush [PPC]
2992 no_uaccess_flush [PPC]
2995 This does not have any effect on
2996 kvm.nx_huge_pages when
2997 kvm.nx_huge_pages=force.
3000 Mitigate all CPU vulnerabilities, but leave SMT
3001 enabled, even if it's vulnerable. This is for
3002 users who don't want to be surprised by SMT
3003 getting disabled across kernel upgrades, or who
3004 have other ways of avoiding SMT-based attacks.
3005 Equivalent to: (default behavior)
3008 Mitigate all CPU vulnerabilities, disabling SMT
3009 if needed. This is for users who always want to
3010 be fully mitigated, even if it means losing SMT.
3011 Equivalent to: l1tf=flush,nosmt [X86]
3012 mds=full,nosmt [X86]
3013 tsx_async_abort=full,nosmt [X86]
3016 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
3017 parameter allows control of the logging verbosity for
3018 the additional memory initialisation checks. A value
3019 of 0 disables mminit logging and a level of 4 will
3020 log everything. Information is printed at KERN_DEBUG
3021 so loglevel=8 may also need to be specified.
3024 [KNL] When CONFIG_MODULE_SIG is set, this means that
3025 modules without (valid) signatures will fail to load.
3026 Note that if CONFIG_MODULE_SIG_FORCE is set, that
3027 is always true, so this option does nothing.
3029 module_blacklist= [KNL] Do not load a comma-separated list of
3030 modules. Useful for debugging problem modules.
3033 [MOUSE] Maximum time between finger touching and
3034 leaving touchpad surface for touch to be considered
3035 a tap and be reported as a left button click (for
3036 touchpads working in absolute mode only).
3038 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
3039 reporting absolute coordinates, such as tablets
3040 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
3041 reporting absolute coordinates, such as tablets
3043 movablecore= [KNL,X86,IA-64,PPC]
3044 Format: nn[KMGTPE] | nn%
3045 This parameter is the complement to kernelcore=, it
3046 specifies the amount of memory used for migratable
3047 allocations. If both kernelcore and movablecore is
3048 specified, then kernelcore will be at *least* the
3049 specified value but may be more. If movablecore on its
3050 own is specified, the administrator must be careful
3051 that the amount of memory usable for all allocations
3054 movable_node [KNL] Boot-time switch to make hotplugable memory
3055 NUMA nodes to be movable. This means that the memory
3056 of such nodes will be usable only for movable
3057 allocations which rules out almost all kernel
3058 allocations. Use with caution!
3060 MTD_Partition= [MTD]
3061 Format: <name>,<region-number>,<size>,<offset>
3063 MTD_Region= [MTD] Format:
3064 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
3067 See drivers/mtd/parsers/cmdlinepart.c
3069 multitce=off [PPC] This parameter disables the use of the pSeries
3070 firmware feature for updating multiple TCE entries
3073 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
3075 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
3077 boundary - index of last SLC block on Flex-OneNAND.
3078 The remaining blocks are configured as MLC blocks.
3079 lock - Configure if Flex-OneNAND boundary should be locked.
3080 Once locked, the boundary cannot be changed.
3081 1 indicates lock status, 0 indicates unlock status.
3084 ARM/S3C2412 JIVE boot control
3086 See arch/arm/mach-s3c/mach-jive.c
3088 mtouchusb.raw_coordinates=
3089 [HW] Make the MicroTouch USB driver use raw coordinates
3090 ('y', default) or cooked coordinates ('n')
3092 mtrr_chunk_size=nn[KMG] [X86]
3093 used for mtrr cleanup. It is largest continuous chunk
3094 that could hold holes aka. UC entries.
3096 mtrr_gran_size=nn[KMG] [X86]
3097 Used for mtrr cleanup. It is granularity of mtrr block.
3099 Large value could prevent small alignment from
3102 mtrr_spare_reg_nr=n [X86]
3104 Range: 0,7 : spare reg number
3106 Used for mtrr cleanup. It is spare mtrr entries number.
3107 Set to 2 or more if your graphical card needs more.
3109 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
3111 netdev= [NET] Network devices parameters
3112 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
3113 Note that mem_start is often overloaded to mean
3114 something different and driver-specific.
3115 This usage is only documented in each driver source
3119 [NETFILTER] Enable connection tracking flow accounting
3120 0 to disable accounting
3121 1 to enable accounting
3124 nfsaddrs= [NFS] Deprecated. Use ip= instead.
3125 See Documentation/admin-guide/nfs/nfsroot.rst.
3127 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
3128 See Documentation/admin-guide/nfs/nfsroot.rst.
3130 nfsrootdebug [NFS] enable nfsroot debugging messages.
3131 See Documentation/admin-guide/nfs/nfsroot.rst.
3133 nfs.callback_nr_threads=
3134 [NFSv4] set the total number of threads that the
3135 NFS client will assign to service NFSv4 callback
3138 nfs.callback_tcpport=
3139 [NFS] set the TCP port on which the NFSv4 callback
3140 channel should listen.
3143 [NFS] sets the pathname to the program which is used
3144 to update the NFS client cache entries.
3146 nfs.cache_getent_timeout=
3147 [NFS] sets the timeout after which an attempt to
3148 update a cache entry is deemed to have failed.
3150 nfs.idmap_cache_timeout=
3151 [NFS] set the maximum lifetime for idmapper cache
3155 [NFS] enable 64-bit inode numbers.
3156 If zero, the NFS client will fake up a 32-bit inode
3157 number for the readdir() and stat() syscalls instead
3158 of returning the full 64-bit number.
3159 The default is to return 64-bit inode numbers.
3161 nfs.max_session_cb_slots=
3162 [NFSv4.1] Sets the maximum number of session
3163 slots the client will assign to the callback
3164 channel. This determines the maximum number of
3165 callbacks the client will process in parallel for
3166 a particular server.
3168 nfs.max_session_slots=
3169 [NFSv4.1] Sets the maximum number of session slots
3170 the client will attempt to negotiate with the server.
3171 This limits the number of simultaneous RPC requests
3172 that the client can send to the NFSv4.1 server.
3173 Note that there is little point in setting this
3174 value higher than the max_tcp_slot_table_limit.
3176 nfs.nfs4_disable_idmapping=
3177 [NFSv4] When set to the default of '1', this option
3178 ensures that both the RPC level authentication
3179 scheme and the NFS level operations agree to use
3180 numeric uids/gids if the mount is using the
3181 'sec=sys' security flavour. In effect it is
3182 disabling idmapping, which can make migration from
3183 legacy NFSv2/v3 systems to NFSv4 easier.
3184 Servers that do not support this mode of operation
3185 will be autodetected by the client, and it will fall
3186 back to using the idmapper.
3187 To turn off this behaviour, set the value to '0'.
3189 [NFS4] Specify an additional fixed unique ident-
3190 ification string that NFSv4 clients can insert into
3191 their nfs_client_id4 string. This is typically a
3192 UUID that is generated at system install time.
3194 nfs.send_implementation_id =
3195 [NFSv4.1] Send client implementation identification
3196 information in exchange_id requests.
3197 If zero, no implementation identification information
3199 The default is to send the implementation identification
3202 nfs.recover_lost_locks =
3203 [NFSv4] Attempt to recover locks that were lost due
3204 to a lease timeout on the server. Please note that
3205 doing this risks data corruption, since there are
3206 no guarantees that the file will remain unchanged
3207 after the locks are lost.
3208 If you want to enable the kernel legacy behaviour of
3209 attempting to recover these locks, then set this
3211 The default parameter value of '0' causes the kernel
3212 not to attempt recovery of lost locks.
3214 nfs4.layoutstats_timer =
3215 [NFSv4.2] Change the rate at which the kernel sends
3216 layoutstats to the pNFS metadata server.
3218 Setting this to value to 0 causes the kernel to use
3219 whatever value is the default set by the layout
3220 driver. A non-zero value sets the minimum interval
3221 in seconds between layoutstats transmissions.
3223 nfsd.nfs4_disable_idmapping=
3224 [NFSv4] When set to the default of '1', the NFSv4
3225 server will return only numeric uids and gids to
3226 clients using auth_sys, and will accept numeric uids
3227 and gids from such clients. This is intended to ease
3228 migration from NFSv2/v3.
3230 nmi_backtrace.backtrace_idle [KNL]
3231 Dump stacks even of idle CPUs in response to an
3232 NMI stack-backtrace request.
3234 nmi_debug= [KNL,SH] Specify one or more actions to take
3235 when a NMI is triggered.
3236 Format: [state][,regs][,debounce][,die]
3238 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
3239 Format: [panic,][nopanic,][num]
3241 0 - turn hardlockup detector in nmi_watchdog off
3242 1 - turn hardlockup detector in nmi_watchdog on
3243 When panic is specified, panic when an NMI watchdog
3244 timeout occurs (or 'nopanic' to not panic on an NMI
3245 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
3246 To disable both hard and soft lockup detectors,
3247 please see 'nowatchdog'.
3248 This is useful when you use a panic=... timeout and
3249 need the box quickly up again.
3251 These settings can be accessed at runtime via
3252 the nmi_watchdog and hardlockup_panic sysctls.
3254 netpoll.carrier_timeout=
3255 [NET] Specifies amount of time (in seconds) that
3256 netpoll should wait for a carrier. By default netpoll
3259 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3260 emulation library even if a 387 maths coprocessor
3263 no5lvl [X86-64] Disable 5-level paging mode. Forces
3264 kernel to use 4-level paging instead.
3266 nofsgsbase [X86] Disables FSGSBASE instructions.
3269 [HW] Never suspend the console
3270 Disable suspending of consoles during suspend and
3271 hibernate operations. Once disabled, debugging
3272 messages can reach various consoles while the rest
3273 of the system is being put to sleep (ie, while
3274 debugging driver suspend/resume hooks). This may
3275 not work reliably with all consoles, but is known
3276 to work with serial and VGA consoles.
3277 To facilitate more flexible debugging, we also add
3278 console_suspend, a printk module parameter to control
3279 it. Users could use console_suspend (usually
3280 /sys/module/printk/parameters/console_suspend) to
3281 turn on/off it dynamically.
3283 novmcoredd [KNL,KDUMP]
3284 Disable device dump. Device dump allows drivers to
3285 append dump data to vmcore so you can collect driver
3286 specified debug info. Drivers can append the data
3287 without any limit and this data is stored in memory,
3288 so this may cause significant memory stress. Disabling
3289 device dump can help save memory but the driver debug
3290 data will be no longer available. This parameter
3291 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
3294 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
3295 caches in the slab allocator. Saves per-node memory,
3296 but will impact performance.
3300 noaltinstr [S390] Disables alternative instructions patching
3301 (CPU alternatives feature).
3303 noapic [SMP,APIC] Tells the kernel to not make use of any
3304 IOAPICs that may be present in the system.
3306 noautogroup Disable scheduler automatic task group creation.
3308 nobats [PPC] Do not use BATs for mapping kernel lowmem
3309 on "Classic" PPC cores.
3313 noclflush [BUGS=X86] Don't use the CLFLUSH instruction
3315 delayacct [KNL] Enable per-task delay accounting
3317 nodsp [SH] Disable hardware DSP at boot time.
3319 noefi Disable EFI runtime services support.
3321 no_entry_flush [PPC] Don't flush the L1-D cache when entering the kernel.
3326 On X86-32 available only on PAE configured kernels.
3327 noexec=on: enable non-executable mappings (default)
3328 noexec=off: disable non-executable mappings
3331 Disable SMAP (Supervisor Mode Access Prevention)
3332 even if it is supported by processor.
3335 Disable SMEP (Supervisor Mode Execution Prevention)
3336 even if it is supported by processor.
3339 This affects only 32-bit executables.
3340 noexec32=on: enable non-executable mappings (default)
3341 read doesn't imply executable mappings
3342 noexec32=off: disable non-executable mappings
3343 read implies executable mappings
3345 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3347 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3348 register save and restore. The kernel will only save
3349 legacy floating-point registers on task switch.
3351 nohugeiomap [KNL,X86,PPC,ARM64] Disable kernel huge I/O mappings.
3353 nohugevmalloc [PPC] Disable kernel huge vmalloc mappings.
3355 nosmt [KNL,S390] Disable symmetric multithreading (SMT).
3356 Equivalent to smt=1.
3358 [KNL,X86] Disable symmetric multithreading (SMT).
3359 nosmt=force: Force disable SMT, cannot be undone
3360 via the sysfs control file.
3362 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1
3363 (bounds check bypass). With this option data leaks are
3364 possible in the system.
3366 nospectre_v2 [X86,PPC_FSL_BOOK3E,ARM64] Disable all mitigations for
3367 the Spectre variant 2 (indirect branch prediction)
3368 vulnerability. System may allow data leaks with this
3371 nospec_store_bypass_disable
3372 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
3375 [PPC] Don't flush the L1-D cache after accessing user data.
3377 noxsave [BUGS=X86] Disables x86 extended register state save
3378 and restore using xsave. The kernel will fallback to
3379 enabling legacy floating-point and sse state.
3381 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
3382 register states. The kernel will fall back to use
3383 xsave to save the states. By using this parameter,
3384 performance of saving the states is degraded because
3385 xsave doesn't support modified optimization while
3386 xsaveopt supports it on xsaveopt enabled systems.
3388 noxsaves [X86] Disables xsaves and xrstors used in saving and
3389 restoring x86 extended register state in compacted
3390 form of xsave area. The kernel will fall back to use
3391 xsaveopt and xrstor to save and restore the states
3392 in standard form of xsave area. By using this
3393 parameter, xsave area per process might occupy more
3394 memory on xsaves enabled systems.
3396 nohlt [ARM,ARM64,MICROBLAZE,SH] Forces the kernel to busy wait
3397 in do_idle() and not use the arch_cpu_idle()
3398 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
3399 to be effective. This is useful on platforms where the
3400 sleep(SH) or wfi(ARM,ARM64) instructions do not work
3401 correctly or when doing power measurements to evalute
3402 the impact of the sleep instructions. This is also
3403 useful when using JTAG debugger.
3405 no_file_caps Tells the kernel not to honor file capabilities. The
3406 only way then for a file to be executed with privilege
3407 is to be setuid root or executed by root.
3409 nohalt [IA-64] Tells the kernel not to use the power saving
3410 function PAL_HALT_LIGHT when idle. This increases
3411 power-consumption. On the positive side, it reduces
3412 interrupt wake-up latency, which may improve performance
3413 in certain environments such as networked servers or
3417 Force pointers printed to the console or buffers to be
3418 unhashed. By default, when a pointer is printed via %p
3419 format string, that pointer is "hashed", i.e. obscured
3420 by hashing the pointer value. This is a security feature
3421 that hides actual kernel addresses from unprivileged
3422 users, but it also makes debugging the kernel more
3423 difficult since unequal pointers can no longer be
3424 compared. However, if this command-line option is
3425 specified, then all normal pointers will have their true
3426 value printed. Pointers printed via %pK may still be
3427 hashed. This option should only be specified when
3428 debugging the kernel. Please do not use on production
3431 nohibernate [HIBERNATION] Disable hibernation and resume.
3433 nohz= [KNL] Boottime enable/disable dynamic ticks
3434 Valid arguments: on, off
3437 nohz_full= [KNL,BOOT,SMP,ISOL]
3438 The argument is a cpu list, as described above.
3439 In kernels built with CONFIG_NO_HZ_FULL=y, set
3440 the specified list of CPUs whose tick will be stopped
3441 whenever possible. The boot CPU will be forced outside
3442 the range to maintain the timekeeping. Any CPUs
3443 in this list will have their RCU callbacks offloaded,
3444 just as if they had also been called out in the
3445 rcu_nocbs= boot parameter.
3447 noiotrap [SH] Disables trapped I/O port accesses.
3449 noirqdebug [X86-32] Disables the code which attempts to detect and
3450 disable unhandled interrupt sources.
3452 no_timer_check [X86,APIC] Disables the code which tests for
3453 broken timer IRQ sources.
3455 noisapnp [ISAPNP] Disables ISA PnP code.
3457 noinitrd [RAM] Tells the kernel not to load any configured
3460 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
3462 [Deprecated - use intremap=off]
3466 noinvpcid [X86] Disable the INVPCID cpu feature.
3468 nojitter [IA-64] Disables jitter checking for ITC timers.
3470 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
3472 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
3476 [X86,PV_OPS] Disable paravirtualized VMware scheduler
3477 clock and use the default one.
3479 no-steal-acc [X86,PV_OPS,ARM64] Disable paravirtualized steal time
3480 accounting. steal time is computed, but won't
3481 influence scheduler behaviour
3483 nolapic [X86-32,APIC] Do not enable or use the local APIC.
3485 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
3487 noltlbs [PPC] Do not use large page/tlb entries for kernel
3488 lowmem mapping on PPC40x and PPC8xx
3490 nomca [IA-64] Disable machine check abort handling
3492 nomce [X86-32] Disable Machine Check Exception
3494 nomfgpt [X86-32] Disable Multi-Function General Purpose
3495 Timer usage (for AMD Geode machines).
3497 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3498 shutdown the other cpus. Instead use the REBOOT_VECTOR
3501 nomodule Disable module load
3503 nopat [X86] Disable PAT (page attribute table extension of
3504 pagetables) support.
3506 nopcid [X86-64] Disable the PCID cpu feature.
3508 norandmaps Don't use address space randomization. Equivalent to
3509 echo 0 > /proc/sys/kernel/randomize_va_space
3511 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
3512 with UP alternatives
3514 nordrand [X86] Disable kernel use of the RDRAND and
3515 RDSEED instructions even if they are supported
3516 by the processor. RDRAND and RDSEED are still
3517 available to user space applications.
3519 noresume [SWSUSP] Disables resume and restores original swap
3522 no-scroll [VGA] Disables scrollback.
3523 This is required for the Braillex ib80-piezo Braille
3524 reader made by F.H. Papenmeier (Germany).
3528 nosep [BUGS=X86-32] Disables x86 SYSENTER/SYSEXIT support.
3530 nosgx [X86-64,SGX] Disables Intel SGX kernel support.
3532 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
3533 and disable the IO APIC. legacy for "maxcpus=0".
3535 nosoftlockup [KNL] Disable the soft-lockup detector.
3537 nosync [HW,M68K] Disables sync negotiation for all devices.
3539 nowatchdog [KNL] Disable both lockup detectors, i.e.
3540 soft-lockup and NMI watchdog (hard-lockup).
3544 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
3546 cpu0_hotplug [X86] Turn on CPU0 hotplug feature when
3547 CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off.
3548 Some features depend on CPU0. Known dependencies are:
3549 1. Resume from suspend/hibernate depends on CPU0.
3550 Suspend/hibernate will fail if CPU0 is offline and you
3551 need to online CPU0 before suspend/hibernate.
3552 2. PIC interrupts also depend on CPU0. CPU0 can't be
3553 removed if a PIC interrupt is detected.
3554 It's said poweroff/reboot may depend on CPU0 on some
3555 machines although I haven't seen such issues so far
3556 after CPU0 is offline on a few tested machines.
3557 If the dependencies are under your control, you can
3558 turn on cpu0_hotplug.
3560 nps_mtm_hs_ctr= [KNL,ARC]
3561 This parameter sets the maximum duration, in
3562 cycles, each HW thread of the CTOP can run
3563 without interruptions, before HW switches it.
3564 The actual maximum duration is 16 times this
3566 Format: integer between 1 and 255
3569 nptcg= [IA-64] Override max number of concurrent global TLB
3570 purges which is reported from either PAL_VM_SUMMARY or
3573 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
3574 could support. nr_cpus=n : n >= 1 limits the kernel to
3575 support 'n' processors. It could be larger than the
3576 number of already plugged CPU during bootup, later in
3577 runtime you can physically add extra cpu until it reaches
3578 n. So during boot up some boot time memory for per-cpu
3579 variables need be pre-allocated for later physical cpu
3582 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
3584 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86] Disable NUMA, Only
3585 set up a single NUMA node spanning all memory.
3587 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic
3589 Allowed values are enable and disable
3591 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
3592 'node', 'default' can be specified
3593 This can be set from sysctl after boot.
3594 See Documentation/admin-guide/sysctl/vm.rst for details.
3596 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
3597 See Documentation/core-api/debugging-via-ohci1394.rst for more
3600 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
3601 Rather than timing out after 20 ms if an EC
3602 command is not properly ACKed, override the length
3603 of the timeout. We have interrupts disabled while
3604 waiting for the ACK, so if this is set too high
3605 interrupts *may* be lost!
3607 omap_mux= [OMAP] Override bootloader pin multiplexing.
3608 Format: <mux_mode0.mode_name=value>...
3609 For example, to override I2C bus2:
3610 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
3612 oops=panic Always panic on oopses. Default is to just kill the
3613 process, but there is a small probability of
3614 deadlocking the machine.
3615 This will also cause panics on machine check exceptions.
3616 Useful together with panic=30 to trigger a reboot.
3619 [KNL] Boolean flag to control whether the page allocator
3620 should randomize its free lists. The randomization may
3621 be automatically enabled if the kernel detects it is
3622 running on a platform with a direct-mapped memory-side
3623 cache, and this parameter can be used to
3624 override/disable that behavior. The state of the flag
3625 can be read from sysfs at:
3626 /sys/module/page_alloc/parameters/shuffle.
3628 page_owner= [KNL] Boot-time page_owner enabling option.
3629 Storage of the information about who allocated
3630 each page is disabled in default. With this switch,
3632 on: enable the feature
3634 page_poison= [KNL] Boot-time parameter changing the state of
3635 poisoning on the buddy allocator, available with
3636 CONFIG_PAGE_POISONING=y.
3637 off: turn off poisoning (default)
3638 on: turn on poisoning
3640 page_reporting.page_reporting_order=
3641 [KNL] Minimal page reporting order
3643 Adjust the minimal page reporting order. The page
3644 reporting is disabled when it exceeds (MAX_ORDER-1).
3646 panic= [KNL] Kernel behaviour on panic: delay <timeout>
3647 timeout > 0: seconds before rebooting
3648 timeout = 0: wait forever
3649 timeout < 0: reboot immediately
3652 panic_print= Bitmask for printing system info when panic happens.
3653 User can chose combination of the following bits:
3654 bit 0: print all tasks info
3655 bit 1: print system memory info
3656 bit 2: print timer info
3657 bit 3: print locks info if CONFIG_LOCKDEP is on
3658 bit 4: print ftrace buffer
3659 bit 5: print all printk messages in buffer
3661 panic_on_taint= Bitmask for conditionally calling panic() in add_taint()
3662 Format: <hex>[,nousertaint]
3663 Hexadecimal bitmask representing the set of TAINT flags
3664 that will cause the kernel to panic when add_taint() is
3665 called with any of the flags in this set.
3666 The optional switch "nousertaint" can be utilized to
3667 prevent userspace forced crashes by writing to sysctl
3668 /proc/sys/kernel/tainted any flagset matching with the
3669 bitmask set on panic_on_taint.
3670 See Documentation/admin-guide/tainted-kernels.rst for
3671 extra details on the taint flags that users can pick
3672 to compose the bitmask to assign to panic_on_taint.
3674 panic_on_warn panic() instead of WARN(). Useful to cause kdump
3677 crash_kexec_post_notifiers
3678 Run kdump after running panic-notifiers and dumping
3679 kmsg. This only for the users who doubt kdump always
3680 succeeds in any situation.
3681 Note that this also increases risks of kdump failure,
3682 because some panic notifiers can make the crashed
3683 kernel more unstable.
3685 parkbd.port= [HW] Parallel port number the keyboard adapter is
3686 connected to, default is 0.
3688 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
3689 0 for XT, 1 for AT (default is AT).
3692 parport= [HW,PPT] Specify parallel ports. 0 disables.
3693 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
3694 Use 'auto' to force the driver to use any
3695 IRQ/DMA settings detected (the default is to
3696 ignore detected IRQ/DMA settings because of
3697 possible conflicts). You can specify the base
3698 address, IRQ, and DMA settings; IRQ and DMA
3699 should be numbers, or 'auto' (for using detected
3700 settings on that particular port), or 'nofifo'
3701 (to avoid using a FIFO even if it is detected).
3702 Parallel ports are assigned in the order they
3703 are specified on the command line, starting
3706 parport_init_mode= [HW,PPT]
3707 Configure VIA parallel port to operate in
3708 a specific mode. This is necessary on Pegasos
3709 computer where firmware has no options for setting
3710 up parallel port mode and sets it to spp.
3711 Currently this function knows 686a and 8231 chips.
3712 Format: [spp|ps2|epp|ecp|ecpepp]
3714 pata_legacy.all= [HW,LIBATA]
3716 Set to non-zero to probe primary and secondary ISA
3717 port ranges on PCI systems where no PCI PATA device
3718 has been found at either range. Disabled by default.
3720 pata_legacy.autospeed= [HW,LIBATA]
3722 Set to non-zero if a chip is present that snoops speed
3723 changes. Disabled by default.
3725 pata_legacy.ht6560a= [HW,LIBATA]
3727 Set to 1, 2, or 3 for HT 6560A on the primary channel,
3728 the secondary channel, or both channels respectively.
3729 Disabled by default.
3731 pata_legacy.ht6560b= [HW,LIBATA]
3733 Set to 1, 2, or 3 for HT 6560B on the primary channel,
3734 the secondary channel, or both channels respectively.
3735 Disabled by default.
3737 pata_legacy.iordy_mask= [HW,LIBATA]
3739 IORDY enable mask. Set individual bits to allow IORDY
3740 for the respective channel. Bit 0 is for the first
3741 legacy channel handled by this driver, bit 1 is for
3742 the second channel, and so on. The sequence will often
3743 correspond to the primary legacy channel, the secondary
3744 legacy channel, and so on, but the handling of a PCI
3745 bus and the use of other driver options may interfere
3746 with the sequence. By default IORDY is allowed across
3749 pata_legacy.opti82c46x= [HW,LIBATA]
3751 Set to 1, 2, or 3 for Opti 82c611A on the primary
3752 channel, the secondary channel, or both channels
3753 respectively. Disabled by default.
3755 pata_legacy.opti82c611a= [HW,LIBATA]
3757 Set to 1, 2, or 3 for Opti 82c465MV on the primary
3758 channel, the secondary channel, or both channels
3759 respectively. Disabled by default.
3761 pata_legacy.pio_mask= [HW,LIBATA]
3763 PIO mode mask for autospeed devices. Set individual
3764 bits to allow the use of the respective PIO modes.
3765 Bit 0 is for mode 0, bit 1 is for mode 1, and so on.
3766 All modes allowed by default.
3768 pata_legacy.probe_all= [HW,LIBATA]
3770 Set to non-zero to probe tertiary and further ISA
3771 port ranges on PCI systems. Disabled by default.
3773 pata_legacy.probe_mask= [HW,LIBATA]
3775 Probe mask for legacy ISA PATA ports. Depending on
3776 platform configuration and the use of other driver
3777 options up to 6 legacy ports are supported: 0x1f0,
3778 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing
3779 of individual ports can be disabled by setting the
3780 corresponding bits in the mask to 1. Bit 0 is for
3781 the first port in the list above (0x1f0), and so on.
3782 By default all supported ports are probed.
3784 pata_legacy.qdi= [HW,LIBATA]
3786 Set to non-zero to probe QDI controllers. By default
3787 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise.
3789 pata_legacy.winbond= [HW,LIBATA]
3791 Set to non-zero to probe Winbond controllers. Use
3792 the standard I/O port (0x130) if 1, otherwise the
3793 value given is the I/O port to use (typically 0x1b0).
3794 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE,
3797 pata_platform.pio_mask= [HW,LIBATA]
3799 Supported PIO mode mask. Set individual bits to allow
3800 the use of the respective PIO modes. Bit 0 is for
3801 mode 0, bit 1 is for mode 1, and so on. Mode 0 only
3805 Halt all CPUs after the first oops has been printed for
3806 the specified number of seconds. This is to be used if
3807 your oopses keep scrolling off the screen.
3812 See header of drivers/block/paride/pcd.c.
3813 See also Documentation/admin-guide/blockdev/paride.rst.
3815 pci=option[,option...] [PCI] various PCI subsystem options.
3817 Some options herein operate on a specific device
3818 or a set of devices (<pci_dev>). These are
3819 specified in one of the following formats:
3821 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
3822 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
3824 Note: the first format specifies a PCI
3825 bus/device/function address which may change
3826 if new hardware is inserted, if motherboard
3827 firmware changes, or due to changes caused
3828 by other kernel parameters. If the
3829 domain is left unspecified, it is
3830 taken to be zero. Optionally, a path
3831 to a device through multiple device/function
3832 addresses can be specified after the base
3833 address (this is more robust against
3834 renumbering issues). The second format
3835 selects devices using IDs from the
3836 configuration space which may match multiple
3837 devices in the system.
3839 earlydump dump PCI config space before the kernel
3841 off [X86] don't probe for the PCI bus
3842 bios [X86-32] force use of PCI BIOS, don't access
3843 the hardware directly. Use this if your machine
3844 has a non-standard PCI host bridge.
3845 nobios [X86-32] disallow use of PCI BIOS, only direct
3846 hardware access methods are allowed. Use this
3847 if you experience crashes upon bootup and you
3848 suspect they are caused by the BIOS.
3849 conf1 [X86] Force use of PCI Configuration Access
3850 Mechanism 1 (config address in IO port 0xCF8,
3851 data in IO port 0xCFC, both 32-bit).
3852 conf2 [X86] Force use of PCI Configuration Access
3853 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
3854 the function, IO port 0xCFA, also 8-bit, sets
3855 bus number. The config space is then accessed
3856 through ports 0xC000-0xCFFF).
3857 See http://wiki.osdev.org/PCI for more info
3858 on the configuration access mechanisms.
3859 noaer [PCIE] If the PCIEAER kernel config parameter is
3860 enabled, this kernel boot option can be used to
3861 disable the use of PCIE advanced error reporting.
3862 nodomains [PCI] Disable support for multiple PCI
3863 root domains (aka PCI segments, in ACPI-speak).
3864 nommconf [X86] Disable use of MMCONFIG for PCI
3866 check_enable_amd_mmconf [X86] check for and enable
3867 properly configured MMIO access to PCI
3868 config space on AMD family 10h CPU
3869 nomsi [MSI] If the PCI_MSI kernel config parameter is
3870 enabled, this kernel boot option can be used to
3871 disable the use of MSI interrupts system-wide.
3872 noioapicquirk [APIC] Disable all boot interrupt quirks.
3873 Safety option to keep boot IRQs enabled. This
3874 should never be necessary.
3875 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
3876 primary IO-APIC for bridges that cannot disable
3877 boot IRQs. This fixes a source of spurious IRQs
3878 when the system masks IRQs.
3879 noioapicreroute [APIC] Disable workaround that uses the
3880 boot IRQ equivalent of an IRQ that connects to
3881 a chipset where boot IRQs cannot be disabled.
3882 The opposite of ioapicreroute.
3883 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
3884 routing table. These calls are known to be buggy
3885 on several machines and they hang the machine
3886 when used, but on other computers it's the only
3887 way to get the interrupt routing table. Try
3888 this option if the kernel is unable to allocate
3889 IRQs or discover secondary PCI buses on your
3891 rom [X86] Assign address space to expansion ROMs.
3892 Use with caution as certain devices share
3893 address decoders between ROMs and other
3895 norom [X86] Do not assign address space to
3896 expansion ROMs that do not already have
3897 BIOS assigned address ranges.
3898 nobar [X86] Do not assign address space to the
3899 BARs that weren't assigned by the BIOS.
3900 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
3901 assigned automatically to PCI devices. You can
3902 make the kernel exclude IRQs of your ISA cards
3904 pirqaddr=0xAAAAA [X86] Specify the physical address
3905 of the PIRQ table (normally generated
3906 by the BIOS) if it is outside the
3907 F0000h-100000h range.
3908 lastbus=N [X86] Scan all buses thru bus #N. Can be
3909 useful if the kernel is unable to find your
3910 secondary buses and you want to tell it
3911 explicitly which ones they are.
3912 assign-busses [X86] Always assign all PCI bus
3913 numbers ourselves, overriding
3914 whatever the firmware may have done.
3915 usepirqmask [X86] Honor the possible IRQ mask stored
3916 in the BIOS $PIR table. This is needed on
3917 some systems with broken BIOSes, notably
3918 some HP Pavilion N5400 and Omnibook XE3
3919 notebooks. This will have no effect if ACPI
3920 IRQ routing is enabled.
3921 noacpi [X86] Do not use ACPI for IRQ routing
3922 or for PCI scanning.
3923 use_crs [X86] Use PCI host bridge window information
3924 from ACPI. On BIOSes from 2008 or later, this
3925 is enabled by default. If you need to use this,
3926 please report a bug.
3927 nocrs [X86] Ignore PCI host bridge windows from ACPI.
3928 If you need to use this, please report a bug.
3929 routeirq Do IRQ routing for all PCI devices.
3930 This is normally done in pci_enable_device(),
3931 so this option is a temporary workaround
3932 for broken drivers that don't call it.
3933 skip_isa_align [X86] do not align io start addr, so can
3934 handle more pci cards
3935 noearly [X86] Don't do any early type 1 scanning.
3936 This might help on some broken boards which
3937 machine check when some devices' config space
3938 is read. But various workarounds are disabled
3939 and some IOMMU drivers will not work.
3940 bfsort Sort PCI devices into breadth-first order.
3941 This sorting is done to get a device
3942 order compatible with older (<= 2.4) kernels.
3943 nobfsort Don't sort PCI devices into breadth-first order.
3944 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
3945 tuning and use the BIOS-configured MPS defaults.
3946 pcie_bus_safe Set every device's MPS to the largest value
3947 supported by all devices below the root complex.
3948 pcie_bus_perf Set device MPS to the largest allowable MPS
3949 based on its parent bus. Also set MRRS (Max
3950 Read Request Size) to the largest supported
3951 value (no larger than the MPS that the device
3952 or bus can support) for best performance.
3953 pcie_bus_peer2peer Set every device's MPS to 128B, which
3954 every device is guaranteed to support. This
3955 configuration allows peer-to-peer DMA between
3956 any pair of devices, possibly at the cost of
3957 reduced performance. This also guarantees
3958 that hot-added devices will work.
3959 cbiosize=nn[KMG] The fixed amount of bus space which is
3960 reserved for the CardBus bridge's IO window.
3961 The default value is 256 bytes.
3962 cbmemsize=nn[KMG] The fixed amount of bus space which is
3963 reserved for the CardBus bridge's memory
3964 window. The default value is 64 megabytes.
3967 [<order of align>@]<pci_dev>[; ...]
3968 Specifies alignment and device to reassign
3969 aligned memory resources. How to
3970 specify the device is described above.
3971 If <order of align> is not specified,
3972 PAGE_SIZE is used as alignment.
3973 A PCI-PCI bridge can be specified if resource
3974 windows need to be expanded.
3975 To specify the alignment for several
3976 instances of a device, the PCI vendor,
3977 device, subvendor, and subdevice may be
3978 specified, e.g., 12@pci:8086:9c22:103c:198f
3979 for 4096-byte alignment.
3980 ecrc= Enable/disable PCIe ECRC (transaction layer
3981 end-to-end CRC checking).
3982 bios: Use BIOS/firmware settings. This is the
3986 hpiosize=nn[KMG] The fixed amount of bus space which is
3987 reserved for hotplug bridge's IO window.
3988 Default size is 256 bytes.
3989 hpmmiosize=nn[KMG] The fixed amount of bus space which is
3990 reserved for hotplug bridge's MMIO window.
3991 Default size is 2 megabytes.
3992 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
3993 reserved for hotplug bridge's MMIO_PREF window.
3994 Default size is 2 megabytes.
3995 hpmemsize=nn[KMG] The fixed amount of bus space which is
3996 reserved for hotplug bridge's MMIO and
3998 Default size is 2 megabytes.
3999 hpbussize=nn The minimum amount of additional bus numbers
4000 reserved for buses below a hotplug bridge.
4002 realloc= Enable/disable reallocating PCI bridge resources
4003 if allocations done by BIOS are too small to
4004 accommodate resources required by all child
4006 off: Turn realloc off
4008 realloc same as realloc=on
4009 noari do not use PCIe ARI.
4010 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
4011 do not use PCIe ATS (and IOMMU device IOTLB).
4012 pcie_scan_all Scan all possible PCIe devices. Otherwise we
4013 only look for one device below a PCIe downstream
4015 big_root_window Try to add a big 64bit memory window to the PCIe
4016 root complex on AMD CPUs. Some GFX hardware
4017 can resize a BAR to allow access to all VRAM.
4018 Adding the window is slightly risky (it may
4019 conflict with unreported devices), so this
4021 disable_acs_redir=<pci_dev>[; ...]
4022 Specify one or more PCI devices (in the format
4023 specified above) separated by semicolons.
4024 Each device specified will have the PCI ACS
4025 redirect capabilities forced off which will
4026 allow P2P traffic between devices through
4027 bridges without forcing it upstream. Note:
4028 this removes isolation between devices and
4029 may put more devices in an IOMMU group.
4030 force_floating [S390] Force usage of floating interrupts.
4031 nomio [S390] Do not use MIO instructions.
4032 norid [S390] ignore the RID field and force use of
4033 one PCI domain per PCI function
4035 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
4038 force Enable ASPM even on devices that claim not to support it.
4039 WARNING: Forcing ASPM on may cause system lockups.
4041 pcie_ports= [PCIE] PCIe port services handling:
4042 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
4043 even if the platform doesn't give the OS permission to
4044 use them. This may cause conflicts if the platform
4045 also tries to use these services.
4046 dpc-native Use native PCIe service for DPC only. May
4047 cause conflicts if firmware uses AER or DPC.
4048 compat Disable native PCIe services (PME, AER, DPC, PCIe
4051 pcie_port_pm= [PCIE] PCIe port power management handling:
4052 off Disable power management of all PCIe ports
4053 force Forcibly enable power management of all PCIe ports
4055 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
4056 nomsi Do not use MSI for native PCIe PME signaling (this makes
4057 all PCIe root ports use INTx for all services).
4059 pcmv= [HW,PCMCIA] BadgePAD 4
4063 Keep all power-domains already enabled by bootloader on,
4064 even if no driver has claimed them. This is useful
4065 for debug and development, but should not be
4066 needed on a platform with proper driver support.
4069 See Documentation/admin-guide/blockdev/paride.rst.
4071 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
4074 See arch/parisc/kernel/pdc_chassis.c
4076 percpu_alloc= Select which percpu first chunk allocator to use.
4077 Currently supported values are "embed" and "page".
4078 Archs may support subset or none of the selections.
4079 See comments in mm/percpu.c for details on each
4080 allocator. This parameter is primarily for debugging
4081 and performance comparison.
4084 See Documentation/admin-guide/blockdev/paride.rst.
4087 See Documentation/admin-guide/blockdev/paride.rst.
4089 pirq= [SMP,APIC] Manual mp-table setup
4090 See Documentation/x86/i386/IO-APIC.rst.
4092 plip= [PPT,NET] Parallel port network link
4093 Format: { parport<nr> | timid | 0 }
4094 See also Documentation/admin-guide/parport.rst.
4096 pmtmr= [X86] Manual setup of pmtmr I/O Port.
4097 Override pmtimer IOPort with a hex value.
4100 pm_debug_messages [SUSPEND,KNL]
4101 Enable suspend/resume debug messages during boot up.
4104 Enable PNP debug messages (depends on the
4105 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
4106 via /sys/module/pnp/parameters/debug. We always show
4107 current resource usage; turning this on also shows
4108 possible settings and some assignment information.
4114 { on | off | curr | res | no-curr | no-res }
4117 [ISAPNP] Exclude IRQs for the autoconfiguration
4120 [ISAPNP] Exclude DMAs for the autoconfiguration
4122 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
4123 Ranges are in pairs (I/O port base and size).
4126 [ISAPNP] Exclude memory regions for the
4128 Ranges are in pairs (memory base and size).
4130 ports= [IP_VS_FTP] IPVS ftp helper module
4132 Up to 8 (IP_VS_APP_MAX_PORTS) ports
4134 Format: <port>,<port>....
4136 powersave=off [PPC] This option disables power saving features.
4137 It specifically disables cpuidle and sets the
4138 platform machine description specific power_save
4139 function to NULL. On Idle the CPU just reduces
4142 ppc_strict_facility_enable
4143 [PPC] This option catches any kernel floating point,
4144 Altivec, VSX and SPE outside of regions specifically
4145 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
4146 There is some performance impact when enabling this.
4150 Disable Hardware Transactional Memory
4153 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC
4154 none - Limited to cond_resched() calls
4155 voluntary - Limited to cond_resched() and might_sleep() calls
4156 full - Any section that isn't explicitly preempt disabled
4157 can be preempted anytime.
4159 print-fatal-signals=
4160 [KNL] debug: print fatal signals
4162 If enabled, warn about various signal handling
4163 related application anomalies: too many signals,
4164 too many POSIX.1 timers, fatal signals causing a
4167 If you hit the warning due to signal overflow,
4168 you might want to try "ulimit -i unlimited".
4172 printk.always_kmsg_dump=
4173 Trigger kmsg_dump for cases other than kernel oops or
4175 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4178 printk.devkmsg={on,off,ratelimit}
4179 Control writing to /dev/kmsg.
4180 on - unlimited logging to /dev/kmsg from userspace
4181 off - logging to /dev/kmsg disabled
4182 ratelimit - ratelimit the logging
4185 printk.time= Show timing data prefixed to each printk message line
4186 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4188 processor.max_cstate= [HW,ACPI]
4189 Limit processor to maximum C-state
4190 max_cstate=9 overrides any DMI blacklist limit.
4192 processor.nocst [HW,ACPI]
4193 Ignore the _CST method to determine C-states,
4194 instead using the legacy FADT method
4196 profile= [KNL] Enable kernel profiling via /proc/profile
4197 Format: [<profiletype>,]<number>
4198 Param: <profiletype>: "schedule", "sleep", or "kvm"
4199 [defaults to kernel profiling]
4200 Param: "schedule" - profile schedule points.
4201 Param: "sleep" - profile D-state sleeping (millisecs).
4202 Requires CONFIG_SCHEDSTATS
4203 Param: "kvm" - profile VM exits.
4204 Param: <number> - step/bucket size as a power of 2 for
4205 statistical time based profiling.
4207 prompt_ramdisk= [RAM] [Deprecated]
4209 prot_virt= [S390] enable hosting protected virtual machines
4210 isolated from the hypervisor (if hardware supports
4214 psi= [KNL] Enable or disable pressure stall information
4218 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
4219 probe for; one of (bare|imps|exps|lifebook|any).
4220 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
4222 psmouse.resetafter= [HW,MOUSE]
4223 Try to reset the device after so many bad packets
4226 [HW,MOUSE] Set desired mouse resolution, in dpi.
4227 psmouse.smartscroll=
4228 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
4229 0 = disabled, 1 = enabled (default).
4231 pstore.backend= Specify the name of the pstore backend to use
4234 See Documentation/admin-guide/blockdev/paride.rst.
4236 pti= [X86-64] Control Page Table Isolation of user and
4237 kernel address spaces. Disabling this feature
4238 removes hardening, but improves performance of
4239 system calls and interrupts.
4241 on - unconditionally enable
4242 off - unconditionally disable
4243 auto - kernel detects whether your CPU model is
4244 vulnerable to issues that PTI mitigates
4246 Not specifying this option is equivalent to pti=auto.
4249 Equivalent to pti=off
4252 [KNL] Number of legacy pty's. Overwrites compiled-in
4255 quiet [KNL] Disable most log messages
4260 See Documentation/admin-guide/md.rst.
4262 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
4263 See Documentation/admin-guide/blockdev/ramdisk.rst.
4265 ramdisk_start= [RAM] RAM disk image start address
4267 random.trust_cpu={on,off}
4268 [KNL] Enable or disable trusting the use of the
4269 CPU's random number generator (if available) to
4270 fully seed the kernel's CRNG. Default is controlled
4271 by CONFIG_RANDOM_TRUST_CPU.
4273 randomize_kstack_offset=
4274 [KNL] Enable or disable kernel stack offset
4275 randomization, which provides roughly 5 bits of
4276 entropy, frustrating memory corruption attacks
4277 that depend on stack address determinism or
4278 cross-syscall address exposures. This is only
4279 available on architectures that have defined
4280 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET.
4281 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4282 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT.
4284 ras=option[,option,...] [KNL] RAS-specific options
4287 Disable the Correctable Errors Collector,
4288 see CONFIG_RAS_CEC help text.
4291 The argument is a cpu list, as described above.
4293 In kernels built with CONFIG_RCU_NOCB_CPU=y, set
4294 the specified list of CPUs to be no-callback CPUs.
4295 Invocation of these CPUs' RCU callbacks will be
4296 offloaded to "rcuox/N" kthreads created for that
4297 purpose, where "x" is "p" for RCU-preempt, and
4298 "s" for RCU-sched, and "N" is the CPU number.
4299 This reduces OS jitter on the offloaded CPUs,
4300 which can be useful for HPC and real-time
4301 workloads. It can also improve energy efficiency
4302 for asymmetric multiprocessors.
4305 Rather than requiring that offloaded CPUs
4306 (specified by rcu_nocbs= above) explicitly
4307 awaken the corresponding "rcuoN" kthreads,
4308 make these kthreads poll for callbacks.
4309 This improves the real-time response for the
4310 offloaded CPUs by relieving them of the need to
4311 wake up the corresponding kthread, but degrades
4312 energy efficiency by requiring that the kthreads
4313 periodically wake up to do the polling.
4315 rcutree.blimit= [KNL]
4316 Set maximum number of finished RCU callbacks to
4317 process in one batch.
4319 rcutree.dump_tree= [KNL]
4320 Dump the structure of the rcu_node combining tree
4321 out at early boot. This is used for diagnostic
4322 purposes, to verify correct tree setup.
4324 rcutree.gp_cleanup_delay= [KNL]
4325 Set the number of jiffies to delay each step of
4326 RCU grace-period cleanup.
4328 rcutree.gp_init_delay= [KNL]
4329 Set the number of jiffies to delay each step of
4330 RCU grace-period initialization.
4332 rcutree.gp_preinit_delay= [KNL]
4333 Set the number of jiffies to delay each step of
4334 RCU grace-period pre-initialization, that is,
4335 the propagation of recent CPU-hotplug changes up
4336 the rcu_node combining tree.
4338 rcutree.use_softirq= [KNL]
4339 If set to zero, move all RCU_SOFTIRQ processing to
4340 per-CPU rcuc kthreads. Defaults to a non-zero
4341 value, meaning that RCU_SOFTIRQ is used by default.
4342 Specify rcutree.use_softirq=0 to use rcuc kthreads.
4344 But note that CONFIG_PREEMPT_RT=y kernels disable
4345 this kernel boot parameter, forcibly setting it
4348 rcutree.rcu_fanout_exact= [KNL]
4349 Disable autobalancing of the rcu_node combining
4350 tree. This is used by rcutorture, and might
4351 possibly be useful for architectures having high
4352 cache-to-cache transfer latencies.
4354 rcutree.rcu_fanout_leaf= [KNL]
4355 Change the number of CPUs assigned to each
4356 leaf rcu_node structure. Useful for very
4357 large systems, which will choose the value 64,
4358 and for NUMA systems with large remote-access
4359 latencies, which will choose a value aligned
4360 with the appropriate hardware boundaries.
4362 rcutree.rcu_min_cached_objs= [KNL]
4363 Minimum number of objects which are cached and
4364 maintained per one CPU. Object size is equal
4365 to PAGE_SIZE. The cache allows to reduce the
4366 pressure to page allocator, also it makes the
4367 whole algorithm to behave better in low memory
4370 rcutree.rcu_delay_page_cache_fill_msec= [KNL]
4371 Set the page-cache refill delay (in milliseconds)
4372 in response to low-memory conditions. The range
4373 of permitted values is in the range 0:100000.
4375 rcutree.jiffies_till_first_fqs= [KNL]
4376 Set delay from grace-period initialization to
4377 first attempt to force quiescent states.
4378 Units are jiffies, minimum value is zero,
4379 and maximum value is HZ.
4381 rcutree.jiffies_till_next_fqs= [KNL]
4382 Set delay between subsequent attempts to force
4383 quiescent states. Units are jiffies, minimum
4384 value is one, and maximum value is HZ.
4386 rcutree.jiffies_till_sched_qs= [KNL]
4387 Set required age in jiffies for a
4388 given grace period before RCU starts
4389 soliciting quiescent-state help from
4390 rcu_note_context_switch() and cond_resched().
4391 If not specified, the kernel will calculate
4392 a value based on the most recent settings
4393 of rcutree.jiffies_till_first_fqs
4394 and rcutree.jiffies_till_next_fqs.
4395 This calculated value may be viewed in
4396 rcutree.jiffies_to_sched_qs. Any attempt to set
4397 rcutree.jiffies_to_sched_qs will be cheerfully
4400 rcutree.kthread_prio= [KNL,BOOT]
4401 Set the SCHED_FIFO priority of the RCU per-CPU
4402 kthreads (rcuc/N). This value is also used for
4403 the priority of the RCU boost threads (rcub/N)
4404 and for the RCU grace-period kthreads (rcu_bh,
4405 rcu_preempt, and rcu_sched). If RCU_BOOST is
4406 set, valid values are 1-99 and the default is 1
4407 (the least-favored priority). Otherwise, when
4408 RCU_BOOST is not set, valid values are 0-99 and
4409 the default is zero (non-realtime operation).
4411 rcutree.rcu_nocb_gp_stride= [KNL]
4412 Set the number of NOCB callback kthreads in
4413 each group, which defaults to the square root
4414 of the number of CPUs. Larger numbers reduce
4415 the wakeup overhead on the global grace-period
4416 kthread, but increases that same overhead on
4417 each group's NOCB grace-period kthread.
4419 rcutree.qhimark= [KNL]
4420 Set threshold of queued RCU callbacks beyond which
4421 batch limiting is disabled.
4423 rcutree.qlowmark= [KNL]
4424 Set threshold of queued RCU callbacks below which
4425 batch limiting is re-enabled.
4427 rcutree.qovld= [KNL]
4428 Set threshold of queued RCU callbacks beyond which
4429 RCU's force-quiescent-state scan will aggressively
4430 enlist help from cond_resched() and sched IPIs to
4431 help CPUs more quickly reach quiescent states.
4432 Set to less than zero to make this be set based
4433 on rcutree.qhimark at boot time and to zero to
4434 disable more aggressive help enlistment.
4436 rcutree.rcu_idle_gp_delay= [KNL]
4437 Set wakeup interval for idle CPUs that have
4438 RCU callbacks (RCU_FAST_NO_HZ=y).
4440 rcutree.rcu_kick_kthreads= [KNL]
4441 Cause the grace-period kthread to get an extra
4442 wake_up() if it sleeps three times longer than
4443 it should at force-quiescent-state time.
4444 This wake_up() will be accompanied by a
4445 WARN_ONCE() splat and an ftrace_dump().
4447 rcutree.rcu_unlock_delay= [KNL]
4448 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels,
4449 this specifies an rcu_read_unlock()-time delay
4450 in microseconds. This defaults to zero.
4451 Larger delays increase the probability of
4452 catching RCU pointer leaks, that is, buggy use
4453 of RCU-protected pointers after the relevant
4454 rcu_read_unlock() has completed.
4456 rcutree.sysrq_rcu= [KNL]
4457 Commandeer a sysrq key to dump out Tree RCU's
4458 rcu_node tree with an eye towards determining
4459 why a new grace period has not yet started.
4461 rcuscale.gp_async= [KNL]
4462 Measure performance of asynchronous
4463 grace-period primitives such as call_rcu().
4465 rcuscale.gp_async_max= [KNL]
4466 Specify the maximum number of outstanding
4467 callbacks per writer thread. When a writer
4468 thread exceeds this limit, it invokes the
4469 corresponding flavor of rcu_barrier() to allow
4470 previously posted callbacks to drain.
4472 rcuscale.gp_exp= [KNL]
4473 Measure performance of expedited synchronous
4474 grace-period primitives.
4476 rcuscale.holdoff= [KNL]
4477 Set test-start holdoff period. The purpose of
4478 this parameter is to delay the start of the
4479 test until boot completes in order to avoid
4482 rcuscale.kfree_rcu_test= [KNL]
4483 Set to measure performance of kfree_rcu() flooding.
4485 rcuscale.kfree_rcu_test_double= [KNL]
4486 Test the double-argument variant of kfree_rcu().
4487 If this parameter has the same value as
4488 rcuscale.kfree_rcu_test_single, both the single-
4489 and double-argument variants are tested.
4491 rcuscale.kfree_rcu_test_single= [KNL]
4492 Test the single-argument variant of kfree_rcu().
4493 If this parameter has the same value as
4494 rcuscale.kfree_rcu_test_double, both the single-
4495 and double-argument variants are tested.
4497 rcuscale.kfree_nthreads= [KNL]
4498 The number of threads running loops of kfree_rcu().
4500 rcuscale.kfree_alloc_num= [KNL]
4501 Number of allocations and frees done in an iteration.
4503 rcuscale.kfree_loops= [KNL]
4504 Number of loops doing rcuscale.kfree_alloc_num number
4505 of allocations and frees.
4507 rcuscale.nreaders= [KNL]
4508 Set number of RCU readers. The value -1 selects
4509 N, where N is the number of CPUs. A value
4510 "n" less than -1 selects N-n+1, where N is again
4511 the number of CPUs. For example, -2 selects N
4512 (the number of CPUs), -3 selects N+1, and so on.
4513 A value of "n" less than or equal to -N selects
4516 rcuscale.nwriters= [KNL]
4517 Set number of RCU writers. The values operate
4518 the same as for rcuscale.nreaders.
4519 N, where N is the number of CPUs
4521 rcuscale.perf_type= [KNL]
4522 Specify the RCU implementation to test.
4524 rcuscale.shutdown= [KNL]
4525 Shut the system down after performance tests
4526 complete. This is useful for hands-off automated
4529 rcuscale.verbose= [KNL]
4530 Enable additional printk() statements.
4532 rcuscale.writer_holdoff= [KNL]
4533 Write-side holdoff between grace periods,
4534 in microseconds. The default of zero says
4537 rcutorture.fqs_duration= [KNL]
4538 Set duration of force_quiescent_state bursts
4541 rcutorture.fqs_holdoff= [KNL]
4542 Set holdoff time within force_quiescent_state bursts
4545 rcutorture.fqs_stutter= [KNL]
4546 Set wait time between force_quiescent_state bursts
4549 rcutorture.fwd_progress= [KNL]
4550 Enable RCU grace-period forward-progress testing
4551 for the types of RCU supporting this notion.
4553 rcutorture.fwd_progress_div= [KNL]
4554 Specify the fraction of a CPU-stall-warning
4555 period to do tight-loop forward-progress testing.
4557 rcutorture.fwd_progress_holdoff= [KNL]
4558 Number of seconds to wait between successive
4559 forward-progress tests.
4561 rcutorture.fwd_progress_need_resched= [KNL]
4562 Enclose cond_resched() calls within checks for
4563 need_resched() during tight-loop forward-progress
4566 rcutorture.gp_cond= [KNL]
4567 Use conditional/asynchronous update-side
4568 primitives, if available.
4570 rcutorture.gp_exp= [KNL]
4571 Use expedited update-side primitives, if available.
4573 rcutorture.gp_normal= [KNL]
4574 Use normal (non-expedited) asynchronous
4575 update-side primitives, if available.
4577 rcutorture.gp_sync= [KNL]
4578 Use normal (non-expedited) synchronous
4579 update-side primitives, if available. If all
4580 of rcutorture.gp_cond=, rcutorture.gp_exp=,
4581 rcutorture.gp_normal=, and rcutorture.gp_sync=
4582 are zero, rcutorture acts as if is interpreted
4583 they are all non-zero.
4585 rcutorture.irqreader= [KNL]
4586 Run RCU readers from irq handlers, or, more
4587 accurately, from a timer handler. Not all RCU
4588 flavors take kindly to this sort of thing.
4590 rcutorture.leakpointer= [KNL]
4591 Leak an RCU-protected pointer out of the reader.
4592 This can of course result in splats, and is
4593 intended to test the ability of things like
4594 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect
4597 rcutorture.n_barrier_cbs= [KNL]
4598 Set callbacks/threads for rcu_barrier() testing.
4600 rcutorture.nfakewriters= [KNL]
4601 Set number of concurrent RCU writers. These just
4602 stress RCU, they don't participate in the actual
4603 test, hence the "fake".
4605 rcutorture.nocbs_nthreads= [KNL]
4606 Set number of RCU callback-offload togglers.
4607 Zero (the default) disables toggling.
4609 rcutorture.nocbs_toggle= [KNL]
4610 Set the delay in milliseconds between successive
4611 callback-offload toggling attempts.
4613 rcutorture.nreaders= [KNL]
4614 Set number of RCU readers. The value -1 selects
4615 N-1, where N is the number of CPUs. A value
4616 "n" less than -1 selects N-n-2, where N is again
4617 the number of CPUs. For example, -2 selects N
4618 (the number of CPUs), -3 selects N+1, and so on.
4620 rcutorture.object_debug= [KNL]
4621 Enable debug-object double-call_rcu() testing.
4623 rcutorture.onoff_holdoff= [KNL]
4624 Set time (s) after boot for CPU-hotplug testing.
4626 rcutorture.onoff_interval= [KNL]
4627 Set time (jiffies) between CPU-hotplug operations,
4628 or zero to disable CPU-hotplug testing.
4630 rcutorture.read_exit= [KNL]
4631 Set the number of read-then-exit kthreads used
4632 to test the interaction of RCU updaters and
4633 task-exit processing.
4635 rcutorture.read_exit_burst= [KNL]
4636 The number of times in a given read-then-exit
4637 episode that a set of read-then-exit kthreads
4640 rcutorture.read_exit_delay= [KNL]
4641 The delay, in seconds, between successive
4642 read-then-exit testing episodes.
4644 rcutorture.shuffle_interval= [KNL]
4645 Set task-shuffle interval (s). Shuffling tasks
4646 allows some CPUs to go into dyntick-idle mode
4647 during the rcutorture test.
4649 rcutorture.shutdown_secs= [KNL]
4650 Set time (s) after boot system shutdown. This
4651 is useful for hands-off automated testing.
4653 rcutorture.stall_cpu= [KNL]
4654 Duration of CPU stall (s) to test RCU CPU stall
4655 warnings, zero to disable.
4657 rcutorture.stall_cpu_block= [KNL]
4658 Sleep while stalling if set. This will result
4659 in warnings from preemptible RCU in addition
4660 to any other stall-related activity.
4662 rcutorture.stall_cpu_holdoff= [KNL]
4663 Time to wait (s) after boot before inducing stall.
4665 rcutorture.stall_cpu_irqsoff= [KNL]
4666 Disable interrupts while stalling if set.
4668 rcutorture.stall_gp_kthread= [KNL]
4669 Duration (s) of forced sleep within RCU
4670 grace-period kthread to test RCU CPU stall
4671 warnings, zero to disable. If both stall_cpu
4672 and stall_gp_kthread are specified, the
4673 kthread is starved first, then the CPU.
4675 rcutorture.stat_interval= [KNL]
4676 Time (s) between statistics printk()s.
4678 rcutorture.stutter= [KNL]
4679 Time (s) to stutter testing, for example, specifying
4680 five seconds causes the test to run for five seconds,
4681 wait for five seconds, and so on. This tests RCU's
4682 ability to transition abruptly to and from idle.
4684 rcutorture.test_boost= [KNL]
4685 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
4686 "Maybe" means test if the RCU implementation
4687 under test support RCU priority boosting.
4689 rcutorture.test_boost_duration= [KNL]
4690 Duration (s) of each individual boost test.
4692 rcutorture.test_boost_interval= [KNL]
4693 Interval (s) between each boost test.
4695 rcutorture.test_no_idle_hz= [KNL]
4696 Test RCU's dyntick-idle handling. See also the
4697 rcutorture.shuffle_interval parameter.
4699 rcutorture.torture_type= [KNL]
4700 Specify the RCU implementation to test.
4702 rcutorture.verbose= [KNL]
4703 Enable additional printk() statements.
4705 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
4706 Dump ftrace buffer after reporting RCU CPU
4709 rcupdate.rcu_cpu_stall_suppress= [KNL]
4710 Suppress RCU CPU stall warning messages.
4712 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL]
4713 Suppress RCU CPU stall warning messages and
4714 rcutorture writer stall warnings that occur
4715 during early boot, that is, during the time
4716 before the init task is spawned.
4718 rcupdate.rcu_cpu_stall_timeout= [KNL]
4719 Set timeout for RCU CPU stall warning messages.
4721 rcupdate.rcu_expedited= [KNL]
4722 Use expedited grace-period primitives, for
4723 example, synchronize_rcu_expedited() instead
4724 of synchronize_rcu(). This reduces latency,
4725 but can increase CPU utilization, degrade
4726 real-time latency, and degrade energy efficiency.
4727 No effect on CONFIG_TINY_RCU kernels.
4729 rcupdate.rcu_normal= [KNL]
4730 Use only normal grace-period primitives,
4731 for example, synchronize_rcu() instead of
4732 synchronize_rcu_expedited(). This improves
4733 real-time latency, CPU utilization, and
4734 energy efficiency, but can expose users to
4735 increased grace-period latency. This parameter
4736 overrides rcupdate.rcu_expedited. No effect on
4737 CONFIG_TINY_RCU kernels.
4739 rcupdate.rcu_normal_after_boot= [KNL]
4740 Once boot has completed (that is, after
4741 rcu_end_inkernel_boot() has been invoked), use
4742 only normal grace-period primitives. No effect
4743 on CONFIG_TINY_RCU kernels.
4745 But note that CONFIG_PREEMPT_RT=y kernels enables
4746 this kernel boot parameter, forcibly setting
4747 it to the value one, that is, converting any
4748 post-boot attempt at an expedited RCU grace
4749 period to instead use normal non-expedited
4750 grace-period processing.
4752 rcupdate.rcu_task_ipi_delay= [KNL]
4753 Set time in jiffies during which RCU tasks will
4754 avoid sending IPIs, starting with the beginning
4755 of a given grace period. Setting a large
4756 number avoids disturbing real-time workloads,
4757 but lengthens grace periods.
4759 rcupdate.rcu_task_stall_timeout= [KNL]
4760 Set timeout in jiffies for RCU task stall warning
4761 messages. Disable with a value less than or equal
4764 rcupdate.rcu_self_test= [KNL]
4765 Run the RCU early boot self tests
4769 Run specified binary instead of /init from the ramdisk,
4770 used for early userspace startup. See initrd.
4773 force - Override the decision by the kernel to hide the
4774 advertisement of RDRAND support (this affects
4775 certain AMD processors because of buggy BIOS
4776 support, specifically around the suspend/resume
4780 Turn on/off individual RDT features. List is:
4781 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
4783 E.g. to turn on cmt and turn off mba use:
4787 Format (x86 or x86_64):
4788 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] \
4790 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
4792 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
4793 (prefix with 'panic_' to set mode for panic
4795 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
4796 reboot_force is either force or not specified,
4797 reboot_cpu is s[mp]#### with #### being the processor
4798 to be used for rebooting.
4800 refscale.holdoff= [KNL]
4801 Set test-start holdoff period. The purpose of
4802 this parameter is to delay the start of the
4803 test until boot completes in order to avoid
4806 refscale.loops= [KNL]
4807 Set the number of loops over the synchronization
4808 primitive under test. Increasing this number
4809 reduces noise due to loop start/end overhead,
4810 but the default has already reduced the per-pass
4811 noise to a handful of picoseconds on ca. 2020
4814 refscale.nreaders= [KNL]
4815 Set number of readers. The default value of -1
4816 selects N, where N is roughly 75% of the number
4817 of CPUs. A value of zero is an interesting choice.
4819 refscale.nruns= [KNL]
4820 Set number of runs, each of which is dumped onto
4823 refscale.readdelay= [KNL]
4824 Set the read-side critical-section duration,
4825 measured in microseconds.
4827 refscale.scale_type= [KNL]
4828 Specify the read-protection implementation to test.
4830 refscale.shutdown= [KNL]
4831 Shut down the system at the end of the performance
4832 test. This defaults to 1 (shut it down) when
4833 refscale is built into the kernel and to 0 (leave
4834 it running) when refscale is built as a module.
4836 refscale.verbose= [KNL]
4837 Enable additional printk() statements.
4839 refscale.verbose_batched= [KNL]
4840 Batch the additional printk() statements. If zero
4841 (the default) or negative, print everything. Otherwise,
4842 print every Nth verbose statement, where N is the value
4846 [KNL, SMP] Set scheduler's default relax_domain_level.
4847 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
4849 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
4850 Format: <base1>,<size1>[,<base2>,<size2>,...]
4851 Reserve I/O ports or memory so the kernel won't use
4852 them. If <base> is less than 0x10000, the region
4853 is assumed to be I/O ports; otherwise it is memory.
4855 reservetop= [X86-32]
4857 Reserves a hole at the top of the kernel virtual
4860 reset_devices [KNL] Force drivers to reset the underlying device
4861 during initialization.
4864 Specify the partition device for software suspend
4866 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
4868 resume_offset= [SWSUSP]
4869 Specify the offset from the beginning of the partition
4870 given by "resume=" at which the swap header is located,
4871 in <PAGE_SIZE> units (needed only for swap files).
4872 See Documentation/power/swsusp-and-swap-files.rst
4874 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
4875 read the resume files
4877 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
4878 Useful for devices that are detected asynchronously
4879 (e.g. USB and MMC devices).
4881 hibernate= [HIBERNATION]
4882 noresume Don't check if there's a hibernation image
4883 present during boot.
4884 nocompress Don't compress/decompress hibernation images.
4885 no Disable hibernation and resume.
4886 protect_image Turn on image protection during restoration
4887 (that will set all pages holding image data
4888 during restoration read-only).
4890 retain_initrd [RAM] Keep initrd memory after extraction
4892 rfkill.default_state=
4893 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
4894 etc. communication is blocked by default.
4897 rfkill.master_switch_mode=
4898 0 The "airplane mode" button does nothing.
4899 1 The "airplane mode" button toggles between everything
4900 blocked and the previous configuration.
4901 2 The "airplane mode" button toggles between everything
4902 blocked and everything unblocked.
4904 rhash_entries= [KNL,NET]
4905 Set number of hash buckets for route cache
4908 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
4911 ro [KNL] Mount root device read-only on boot
4914 on Mark read-only kernel memory as read-only (default).
4915 off Leave read-only kernel memory writable for debugging.
4918 Enable the uart passthrough on the designated usb port
4919 on Rockchip SoCs. When active, the signals of the
4920 debug-uart get routed to the D+ and D- pins of the usb
4921 port and the regular usb controller gets disabled.
4923 root= [KNL] Root filesystem
4924 See name_to_dev_t comment in init/do_mounts.c.
4926 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
4927 mount the root filesystem
4929 rootflags= [KNL] Set root filesystem mount option string
4931 rootfstype= [KNL] Set root filesystem type
4933 rootwait [KNL] Wait (indefinitely) for root device to show up.
4934 Useful for devices that are detected asynchronously
4935 (e.g. USB and MMC devices).
4937 rproc_mem=nn[KMG][@address]
4938 [KNL,ARM,CMA] Remoteproc physical memory block.
4939 Memory area to be used by remote processor image,
4942 rw [KNL] Mount root device read-write on boot
4944 S [KNL] Run init in single mode
4946 s390_iommu= [HW,S390]
4947 Set s390 IOTLB flushing mode
4949 With strict flushing every unmap operation will result in
4950 an IOTLB flush. Default is lazy flushing before reuse,
4954 See drivers/net/irda/sa1100_ir.c.
4956 sbni= [NET] Granch SBNI12 leased line adapter
4958 sched_verbose [KNL] Enables verbose scheduler debug messages.
4960 schedstats= [KNL,X86] Enable or disable scheduled statistics.
4961 Allowed values are enable and disable. This feature
4962 incurs a small amount of overhead in the scheduler
4963 but is useful for debugging and performance tuning.
4965 sched_thermal_decay_shift=
4966 [KNL, SMP] Set a decay shift for scheduler thermal
4967 pressure signal. Thermal pressure signal follows the
4968 default decay period of other scheduler pelt
4969 signals(usually 32 ms but configurable). Setting
4970 sched_thermal_decay_shift will left shift the decay
4971 period for the thermal pressure signal by the shift
4973 i.e. with the default pelt decay period of 32 ms
4974 sched_thermal_decay_shift thermal pressure decay pr
4978 Format: integer between 0 and 10
4981 scftorture.holdoff= [KNL]
4982 Number of seconds to hold off before starting
4983 test. Defaults to zero for module insertion and
4984 to 10 seconds for built-in smp_call_function()
4987 scftorture.longwait= [KNL]
4988 Request ridiculously long waits randomly selected
4989 up to the chosen limit in seconds. Zero (the
4990 default) disables this feature. Please note
4991 that requesting even small non-zero numbers of
4992 seconds can result in RCU CPU stall warnings,
4993 softlockup complaints, and so on.
4995 scftorture.nthreads= [KNL]
4996 Number of kthreads to spawn to invoke the
4997 smp_call_function() family of functions.
4998 The default of -1 specifies a number of kthreads
4999 equal to the number of CPUs.
5001 scftorture.onoff_holdoff= [KNL]
5002 Number seconds to wait after the start of the
5003 test before initiating CPU-hotplug operations.
5005 scftorture.onoff_interval= [KNL]
5006 Number seconds to wait between successive
5007 CPU-hotplug operations. Specifying zero (which
5008 is the default) disables CPU-hotplug operations.
5010 scftorture.shutdown_secs= [KNL]
5011 The number of seconds following the start of the
5012 test after which to shut down the system. The
5013 default of zero avoids shutting down the system.
5014 Non-zero values are useful for automated tests.
5016 scftorture.stat_interval= [KNL]
5017 The number of seconds between outputting the
5018 current test statistics to the console. A value
5019 of zero disables statistics output.
5021 scftorture.stutter_cpus= [KNL]
5022 The number of jiffies to wait between each change
5023 to the set of CPUs under test.
5025 scftorture.use_cpus_read_lock= [KNL]
5026 Use use_cpus_read_lock() instead of the default
5027 preempt_disable() to disable CPU hotplug
5028 while invoking one of the smp_call_function*()
5031 scftorture.verbose= [KNL]
5032 Enable additional printk() statements.
5034 scftorture.weight_single= [KNL]
5035 The probability weighting to use for the
5036 smp_call_function_single() function with a zero
5037 "wait" parameter. A value of -1 selects the
5038 default if all other weights are -1. However,
5039 if at least one weight has some other value, a
5040 value of -1 will instead select a weight of zero.
5042 scftorture.weight_single_wait= [KNL]
5043 The probability weighting to use for the
5044 smp_call_function_single() function with a
5045 non-zero "wait" parameter. See weight_single.
5047 scftorture.weight_many= [KNL]
5048 The probability weighting to use for the
5049 smp_call_function_many() function with a zero
5050 "wait" parameter. See weight_single.
5051 Note well that setting a high probability for
5052 this weighting can place serious IPI load
5055 scftorture.weight_many_wait= [KNL]
5056 The probability weighting to use for the
5057 smp_call_function_many() function with a
5058 non-zero "wait" parameter. See weight_single
5061 scftorture.weight_all= [KNL]
5062 The probability weighting to use for the
5063 smp_call_function_all() function with a zero
5064 "wait" parameter. See weight_single and
5067 scftorture.weight_all_wait= [KNL]
5068 The probability weighting to use for the
5069 smp_call_function_all() function with a
5070 non-zero "wait" parameter. See weight_single
5073 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
5074 xtime_lock contention on larger systems, and/or RCU lock
5075 contention on all systems with CONFIG_MAXSMP set.
5076 Format: { "0" | "1" }
5077 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
5079 Note: increases power consumption, thus should only be
5080 enabled if running jitter sensitive (HPC/RT) workloads.
5082 security= [SECURITY] Choose a legacy "major" security module to
5083 enable at boot. This has been deprecated by the
5086 selinux= [SELINUX] Disable or enable SELinux at boot time.
5087 Format: { "0" | "1" }
5088 See security/selinux/Kconfig help text.
5093 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
5094 Format: { "0" | "1" }
5095 See security/apparmor/Kconfig help text
5098 Default value is set via kernel config option.
5100 serialnumber [BUGS=X86-32]
5103 Maximal number of shapers.
5111 Enable merging of slabs with similar size when the
5112 kernel is built without CONFIG_SLAB_MERGE_DEFAULT.
5115 Disable merging of slabs with similar size. May be
5116 necessary if there is some reason to distinguish
5117 allocs to different slabs, especially in hardened
5118 environments where the risk of heap overflows and
5119 layout control by attackers can usually be
5120 frustrated by disabling merging. This will reduce
5121 most of the exposure of a heap attack to a single
5122 cache (risks via metadata attacks are mostly
5123 unchanged). Debug options disable merging on their
5125 For more information see Documentation/vm/slub.rst.
5127 slab_max_order= [MM, SLAB]
5128 Determines the maximum allowed order for slabs.
5129 A high setting may cause OOMs due to memory
5130 fragmentation. Defaults to 1 for systems with
5131 more than 32MB of RAM, 0 otherwise.
5133 slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB]
5134 Enabling slub_debug allows one to determine the
5135 culprit if slab objects become corrupted. Enabling
5136 slub_debug can create guard zones around objects and
5137 may poison objects when not in use. Also tracks the
5138 last alloc / free. For more information see
5139 Documentation/vm/slub.rst.
5141 slub_max_order= [MM, SLUB]
5142 Determines the maximum allowed order for slabs.
5143 A high setting may cause OOMs due to memory
5144 fragmentation. For more information see
5145 Documentation/vm/slub.rst.
5147 slub_min_objects= [MM, SLUB]
5148 The minimum number of objects per slab. SLUB will
5149 increase the slab order up to slub_max_order to
5150 generate a sufficiently large slab able to contain
5151 the number of objects indicated. The higher the number
5152 of objects the smaller the overhead of tracking slabs
5153 and the less frequently locks need to be acquired.
5154 For more information see Documentation/vm/slub.rst.
5156 slub_min_order= [MM, SLUB]
5157 Determines the minimum page order for slabs. Must be
5158 lower than slub_max_order.
5159 For more information see Documentation/vm/slub.rst.
5161 slub_merge [MM, SLUB]
5162 Same with slab_merge.
5164 slub_nomerge [MM, SLUB]
5165 Same with slab_nomerge. This is supported for legacy.
5166 See slab_nomerge for more information.
5169 Format: <io1>[,<io2>[,...,<io8>]]
5171 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
5172 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
5173 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
5174 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
5175 smsc-ircc2.ircc_irq= [HW] IRQ line
5176 smsc-ircc2.ircc_dma= [HW] DMA channel
5177 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
5178 0: Toshiba Satellite 1800 (GP data pin select)
5179 1: Fast pin select (default)
5182 smt [KNL,S390] Set the maximum number of threads (logical
5183 CPUs) to use per physical CPU on systems capable of
5184 symmetric multithreading (SMT). Will be capped to the
5185 actual hardware limit.
5187 Default: -1 (no limit)
5190 [KNL] Should the soft-lockup detector generate panics.
5193 A value of 1 instructs the soft-lockup detector
5194 to panic the machine when a soft-lockup occurs. It is
5195 also controlled by the kernel.softlockup_panic sysctl
5196 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
5197 respective build-time switch to that functionality.
5199 softlockup_all_cpu_backtrace=
5200 [KNL] Should the soft-lockup detector generate
5201 backtraces on all cpus.
5204 sonypi.*= [HW] Sony Programmable I/O Control Device driver
5205 See Documentation/admin-guide/laptops/sonypi.rst
5207 spectre_v2= [X86] Control mitigation of Spectre variant 2
5208 (indirect branch speculation) vulnerability.
5209 The default operation protects the kernel from
5212 on - unconditionally enable, implies
5214 off - unconditionally disable, implies
5216 auto - kernel detects whether your CPU model is
5219 Selecting 'on' will, and 'auto' may, choose a
5220 mitigation method at run time according to the
5221 CPU, the available microcode, the setting of the
5222 CONFIG_RETPOLINE configuration option, and the
5223 compiler with which the kernel was built.
5225 Selecting 'on' will also enable the mitigation
5226 against user space to user space task attacks.
5228 Selecting 'off' will disable both the kernel and
5229 the user space protections.
5231 Specific mitigations can also be selected manually:
5233 retpoline - replace indirect branches
5234 retpoline,generic - google's original retpoline
5235 retpoline,amd - AMD-specific minimal thunk
5237 Not specifying this option is equivalent to
5241 [X86] Control mitigation of Spectre variant 2
5242 (indirect branch speculation) vulnerability between
5245 on - Unconditionally enable mitigations. Is
5246 enforced by spectre_v2=on
5248 off - Unconditionally disable mitigations. Is
5249 enforced by spectre_v2=off
5251 prctl - Indirect branch speculation is enabled,
5252 but mitigation can be enabled via prctl
5253 per thread. The mitigation control state
5254 is inherited on fork.
5257 - Like "prctl" above, but only STIBP is
5258 controlled per thread. IBPB is issued
5259 always when switching between different user
5263 - Same as "prctl" above, but all seccomp
5264 threads will enable the mitigation unless
5265 they explicitly opt out.
5268 - Like "seccomp" above, but only STIBP is
5269 controlled per thread. IBPB is issued
5270 always when switching between different
5271 user space processes.
5273 auto - Kernel selects the mitigation depending on
5274 the available CPU features and vulnerability.
5277 If CONFIG_SECCOMP=y then "seccomp", otherwise "prctl"
5279 Not specifying this option is equivalent to
5280 spectre_v2_user=auto.
5282 spec_store_bypass_disable=
5283 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
5284 (Speculative Store Bypass vulnerability)
5286 Certain CPUs are vulnerable to an exploit against a
5287 a common industry wide performance optimization known
5288 as "Speculative Store Bypass" in which recent stores
5289 to the same memory location may not be observed by
5290 later loads during speculative execution. The idea
5291 is that such stores are unlikely and that they can
5292 be detected prior to instruction retirement at the
5293 end of a particular speculation execution window.
5295 In vulnerable processors, the speculatively forwarded
5296 store can be used in a cache side channel attack, for
5297 example to read memory to which the attacker does not
5298 directly have access (e.g. inside sandboxed code).
5300 This parameter controls whether the Speculative Store
5301 Bypass optimization is used.
5303 On x86 the options are:
5305 on - Unconditionally disable Speculative Store Bypass
5306 off - Unconditionally enable Speculative Store Bypass
5307 auto - Kernel detects whether the CPU model contains an
5308 implementation of Speculative Store Bypass and
5309 picks the most appropriate mitigation. If the
5310 CPU is not vulnerable, "off" is selected. If the
5311 CPU is vulnerable the default mitigation is
5312 architecture and Kconfig dependent. See below.
5313 prctl - Control Speculative Store Bypass per thread
5314 via prctl. Speculative Store Bypass is enabled
5315 for a process by default. The state of the control
5316 is inherited on fork.
5317 seccomp - Same as "prctl" above, but all seccomp threads
5318 will disable SSB unless they explicitly opt out.
5320 Default mitigations:
5321 X86: If CONFIG_SECCOMP=y "seccomp", otherwise "prctl"
5323 On powerpc the options are:
5325 on,auto - On Power8 and Power9 insert a store-forwarding
5326 barrier on kernel entry and exit. On Power7
5327 perform a software flush on kernel entry and
5331 Not specifying this option is equivalent to
5332 spec_store_bypass_disable=auto.
5334 spia_io_base= [HW,MTD]
5340 [X86] Enable split lock detection or bus lock detection
5342 When enabled (and if hardware support is present), atomic
5343 instructions that access data across cache line
5344 boundaries will result in an alignment check exception
5345 for split lock detection or a debug exception for
5350 warn - the kernel will emit rate-limited warnings
5351 about applications triggering the #AC
5352 exception or the #DB exception. This mode is
5353 the default on CPUs that support split lock
5354 detection or bus lock detection. Default
5355 behavior is by #AC if both features are
5356 enabled in hardware.
5358 fatal - the kernel will send SIGBUS to applications
5359 that trigger the #AC exception or the #DB
5360 exception. Default behavior is by #AC if
5361 both features are enabled in hardware.
5364 Set system wide rate limit to N bus locks
5365 per second for bus lock detection.
5368 N/A for split lock detection.
5371 If an #AC exception is hit in the kernel or in
5372 firmware (i.e. not while executing in user mode)
5373 the kernel will oops in either "warn" or "fatal"
5376 #DB exception for bus lock is triggered only when
5380 Control the Special Register Buffer Data Sampling
5383 Certain CPUs are vulnerable to an MDS-like
5384 exploit which can leak bits from the random
5387 By default, this issue is mitigated by
5388 microcode. However, the microcode fix can cause
5389 the RDRAND and RDSEED instructions to become
5390 much slower. Among other effects, this will
5391 result in reduced throughput from /dev/urandom.
5393 The microcode mitigation can be disabled with
5394 the following option:
5396 off: Disable mitigation and remove
5397 performance impact to RDRAND and RDSEED
5399 srcutree.counter_wrap_check [KNL]
5400 Specifies how frequently to check for
5401 grace-period sequence counter wrap for the
5402 srcu_data structure's ->srcu_gp_seq_needed field.
5403 The greater the number of bits set in this kernel
5404 parameter, the less frequently counter wrap will
5405 be checked for. Note that the bottom two bits
5408 srcutree.exp_holdoff [KNL]
5409 Specifies how many nanoseconds must elapse
5410 since the end of the last SRCU grace period for
5411 a given srcu_struct until the next normal SRCU
5412 grace period will be considered for automatic
5413 expediting. Set to zero to disable automatic
5417 Speculative Store Bypass Disable control
5419 On CPUs that are vulnerable to the Speculative
5420 Store Bypass vulnerability and offer a
5421 firmware based mitigation, this parameter
5422 indicates how the mitigation should be used:
5424 force-on: Unconditionally enable mitigation for
5425 for both kernel and userspace
5426 force-off: Unconditionally disable mitigation for
5427 for both kernel and userspace
5428 kernel: Always enable mitigation in the
5429 kernel, and offer a prctl interface
5430 to allow userspace to register its
5431 interest in being mitigated too.
5433 stack_guard_gap= [MM]
5434 override the default stack gap protection. The value
5435 is in page units and it defines how many pages prior
5436 to (for stacks growing down) resp. after (for stacks
5437 growing up) the main stack are reserved for no other
5438 mapping. Default value is 256 pages.
5440 stack_depot_disable= [KNL]
5441 Setting this to true through kernel command line will
5442 disable the stack depot thereby saving the static memory
5443 consumed by the stack hash table. By default this is set
5447 Enabled the stack tracer on boot up.
5449 stacktrace_filter=[function-list]
5450 [FTRACE] Limit the functions that the stack tracer
5451 will trace at boot up. function-list is a comma-separated
5452 list of functions. This list can be changed at run
5453 time by the stack_trace_filter file in the debugfs
5454 tracing directory. Note, this enables stack tracing
5455 and the stacktrace above is not needed.
5459 Set the STI (builtin display/keyboard on the HP-PARISC
5460 machines) console (graphic card) which should be used
5461 as the initial boot-console.
5462 See also comment in drivers/video/console/sticore.c.
5465 See comment in drivers/video/console/sticore.c.
5468 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
5470 sunrpc.min_resvport=
5471 sunrpc.max_resvport=
5473 SunRPC servers often require that client requests
5474 originate from a privileged port (i.e. a port in the
5475 range 0 < portnr < 1024).
5476 An administrator who wishes to reserve some of these
5477 ports for other uses may adjust the range that the
5478 kernel's sunrpc client considers to be privileged
5479 using these two parameters to set the minimum and
5480 maximum port values.
5482 sunrpc.svc_rpc_per_connection_limit=
5484 Limit the number of requests that the server will
5485 process in parallel from a single connection.
5486 The default value is 0 (no limit).
5490 Control how the NFS server code allocates CPUs to
5491 service thread pools. Depending on how many NICs
5492 you have and where their interrupts are bound, this
5493 option will affect which CPUs will do NFS serving.
5494 Note: this parameter cannot be changed while the
5495 NFS server is running.
5497 auto the server chooses an appropriate mode
5498 automatically using heuristics
5499 global a single global pool contains all CPUs
5500 percpu one pool for each CPU
5501 pernode one pool for each NUMA node (equivalent
5502 to global on non-NUMA machines)
5504 sunrpc.tcp_slot_table_entries=
5505 sunrpc.udp_slot_table_entries=
5507 Sets the upper limit on the number of simultaneous
5508 RPC calls that can be sent from the client to a
5509 server. Increasing these values may allow you to
5510 improve throughput, but will also increase the
5511 amount of memory reserved for use by the client.
5513 suspend.pm_test_delay=
5515 Sets the number of seconds to remain in a suspend test
5516 mode before resuming the system (see
5517 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
5518 is set. Default value is 5.
5521 Format: { on | off | y | n | 1 | 0 }
5522 This parameter controls use of the Protected
5523 Execution Facility on pSeries.
5526 [KNL] Enable accounting of swap in memory resource
5527 controller if no parameter or 1 is given or disable
5528 it if 0 is given (See Documentation/admin-guide/cgroup-v1/memory.rst)
5530 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
5531 Format: { <int> | force | noforce }
5532 <int> -- Number of I/O TLB slabs
5533 force -- force using of bounce buffers even if they
5534 wouldn't be automatically used by the kernel
5535 noforce -- Never use bounce buffers (for debugging)
5540 Set a sysctl parameter, right before loading the init
5541 process, as if the value was written to the respective
5542 /proc/sys/... file. Both '.' and '/' are recognized as
5543 separators. Unrecognized parameters and invalid values
5544 are reported in the kernel log. Sysctls registered
5545 later by a loaded module cannot be set this way.
5546 Example: sysctl.vm.swappiness=40
5548 sysfs.deprecated=0|1 [KNL]
5549 Enable/disable old style sysfs layout for old udev
5550 on older distributions. When this option is enabled
5551 very new udev will not work anymore. When this option
5552 is disabled (or CONFIG_SYSFS_DEPRECATED not compiled)
5553 in older udev will not work anymore.
5554 Default depends on CONFIG_SYSFS_DEPRECATED_V2 set in
5555 the kernel configuration.
5557 sysrq_always_enabled
5559 Ignore sysrq setting - this boot parameter will
5560 neutralize any effect of /proc/sys/kernel/sysrq.
5561 Useful for debugging.
5563 tcpmhash_entries= [KNL,NET]
5564 Set the number of tcp_metrics_hash slots.
5565 Default value is 8192 or 16384 depending on total
5566 ram pages. This is used to specify the TCP metrics
5567 cache size. See Documentation/networking/ip-sysctl.rst
5568 "tcp_no_metrics_save" section for more details.
5572 test_suspend= [SUSPEND][,N]
5573 Specify "mem" (for Suspend-to-RAM) or "standby" (for
5574 standby suspend) or "freeze" (for suspend type freeze)
5575 as the system sleep state during system startup with
5576 the optional capability to repeat N number of times.
5577 The system is woken from this state using a
5578 wakeup-capable RTC alarm.
5580 thash_entries= [KNL,NET]
5581 Set number of hash buckets for TCP connection
5583 thermal.act= [HW,ACPI]
5584 -1: disable all active trip points in all thermal zones
5585 <degrees C>: override all lowest active trip points
5587 thermal.crt= [HW,ACPI]
5588 -1: disable all critical trip points in all thermal zones
5589 <degrees C>: override all critical trip points
5591 thermal.nocrt= [HW,ACPI]
5592 Set to disable actions on ACPI thermal zone
5593 critical and hot trip points.
5595 thermal.off= [HW,ACPI]
5596 1: disable ACPI thermal control
5598 thermal.psv= [HW,ACPI]
5599 -1: disable all passive trip points
5600 <degrees C>: override all passive trip points to this
5603 thermal.tzp= [HW,ACPI]
5604 Specify global default ACPI thermal zone polling rate
5605 <deci-seconds>: poll all this frequency
5606 0: no polling (default)
5609 Force threading of all interrupt handlers except those
5610 marked explicitly IRQF_NO_THREAD.
5614 Specify if the kernel should make use of the cpu
5615 topology information if the hardware supports this.
5616 The scheduler will make use of this information and
5617 e.g. base its process migration decisions on it.
5620 topology_updates= [KNL, PPC, NUMA]
5622 Specify if the kernel should ignore (off)
5623 topology updates sent by the hypervisor to this
5626 torture.disable_onoff_at_boot= [KNL]
5627 Prevent the CPU-hotplug component of torturing
5628 until after init has spawned.
5630 torture.ftrace_dump_at_shutdown= [KNL]
5631 Dump the ftrace buffer at torture-test shutdown,
5632 even if there were no errors. This can be a
5633 very costly operation when many torture tests
5634 are running concurrently, especially on systems
5635 with rotating-rust storage.
5637 torture.verbose_sleep_frequency= [KNL]
5638 Specifies how many verbose printk()s should be
5639 emitted between each sleep. The default of zero
5640 disables verbose-printk() sleeping.
5642 torture.verbose_sleep_duration= [KNL]
5643 Duration of each verbose-printk() sleep in jiffies.
5647 tpm_suspend_pcr=[HW,TPM]
5648 Format: integer pcr id
5649 Specify that at suspend time, the tpm driver
5650 should extend the specified pcr with zeros,
5651 as a workaround for some chips which fail to
5652 flush the last written pcr on TPM_SaveState.
5653 This will guarantee that all the other pcrs
5656 trace_buf_size=nn[KMG]
5657 [FTRACE] will set tracing buffer size on each cpu.
5659 trace_event=[event-list]
5660 [FTRACE] Set and start specified trace events in order
5661 to facilitate early boot debugging. The event-list is a
5662 comma-separated list of trace events to enable. See
5663 also Documentation/trace/events.rst
5665 trace_options=[option-list]
5666 [FTRACE] Enable or disable tracer options at boot.
5667 The option-list is a comma delimited list of options
5668 that can be enabled or disabled just as if you were
5669 to echo the option name into
5671 /sys/kernel/debug/tracing/trace_options
5673 For example, to enable stacktrace option (to dump the
5674 stack trace of each event), add to the command line:
5676 trace_options=stacktrace
5678 See also Documentation/trace/ftrace.rst "trace options"
5682 Have the tracepoints sent to printk as well as the
5683 tracing ring buffer. This is useful for early boot up
5684 where the system hangs or reboots and does not give the
5685 option for reading the tracing buffer or performing a
5686 ftrace_dump_on_oops.
5688 To turn off having tracepoints sent to printk,
5689 echo 0 > /proc/sys/kernel/tracepoint_printk
5690 Note, echoing 1 into this file without the
5691 tracepoint_printk kernel cmdline option has no effect.
5693 The tp_printk_stop_on_boot (see below) can also be used
5694 to stop the printing of events to console at
5699 Having tracepoints sent to printk() and activating high
5700 frequency tracepoints such as irq or sched, can cause
5701 the system to live lock.
5703 tp_printk_stop_on_boot[FTRACE]
5704 When tp_printk (above) is set, it can cause a lot of noise
5705 on the console. It may be useful to only include the
5706 printing of events during boot up, as user space may
5707 make the system inoperable.
5709 This command line option will stop the printing of events
5710 to console at the late_initcall_sync() time frame.
5713 [FTRACE] enable this option to disable tracing when a
5714 warning is hit. This turns off "tracing_on". Tracing can
5715 be enabled again by echoing '1' into the "tracing_on"
5716 file located in /sys/kernel/debug/tracing/
5718 This option is useful, as it disables the trace before
5719 the WARNING dump is called, which prevents the trace to
5720 be filled with content caused by the warning output.
5722 This option can also be set at run time via the sysctl
5723 option: kernel/traceoff_on_warning
5725 transparent_hugepage=
5727 Format: [always|madvise|never]
5728 Can be used to control the default behavior of the system
5729 with respect to transparent hugepages.
5730 See Documentation/admin-guide/mm/transhuge.rst
5733 trusted.source= [KEYS]
5735 This parameter identifies the trust source as a backend
5736 for trusted keys implementation. Supported trust
5740 If not specified then it defaults to iterating through
5741 the trust source list starting with TPM and assigns the
5742 first trust source as a backend which is initialized
5743 successfully during iteration.
5745 tsc= Disable clocksource stability checks for TSC.
5747 [x86] reliable: mark tsc clocksource as reliable, this
5748 disables clocksource verification at runtime, as well
5749 as the stability checks done at bootup. Used to enable
5750 high-resolution timer mode on older hardware, and in
5751 virtualized environment.
5752 [x86] noirqtime: Do not use TSC to do irq accounting.
5753 Used to run time disable IRQ_TIME_ACCOUNTING on any
5754 platforms where RDTSC is slow and this accounting
5756 [x86] unstable: mark the TSC clocksource as unstable, this
5757 marks the TSC unconditionally unstable at bootup and
5758 avoids any further wobbles once the TSC watchdog notices.
5759 [x86] nowatchdog: disable clocksource watchdog. Used
5760 in situations with strict latency requirements (where
5761 interruptions from clocksource watchdog are not
5764 tsc_early_khz= [X86] Skip early TSC calibration and use the given
5765 value instead. Useful when the early TSC frequency discovery
5766 procedure is not reliable, such as on overclocked systems
5767 with CPUID.16h support and partial CPUID.15h support.
5768 Format: <unsigned int>
5770 tsx= [X86] Control Transactional Synchronization
5771 Extensions (TSX) feature in Intel processors that
5772 support TSX control.
5774 This parameter controls the TSX feature. The options are:
5776 on - Enable TSX on the system. Although there are
5777 mitigations for all known security vulnerabilities,
5778 TSX has been known to be an accelerator for
5779 several previous speculation-related CVEs, and
5780 so there may be unknown security risks associated
5781 with leaving it enabled.
5783 off - Disable TSX on the system. (Note that this
5784 option takes effect only on newer CPUs which are
5785 not vulnerable to MDS, i.e., have
5786 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
5787 the new IA32_TSX_CTRL MSR through a microcode
5788 update. This new MSR allows for the reliable
5789 deactivation of the TSX functionality.)
5791 auto - Disable TSX if X86_BUG_TAA is present,
5792 otherwise enable TSX on the system.
5794 Not specifying this option is equivalent to tsx=off.
5796 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5799 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
5800 Abort (TAA) vulnerability.
5802 Similar to Micro-architectural Data Sampling (MDS)
5803 certain CPUs that support Transactional
5804 Synchronization Extensions (TSX) are vulnerable to an
5805 exploit against CPU internal buffers which can forward
5806 information to a disclosure gadget under certain
5809 In vulnerable processors, the speculatively forwarded
5810 data can be used in a cache side channel attack, to
5811 access data to which the attacker does not have direct
5814 This parameter controls the TAA mitigation. The
5817 full - Enable TAA mitigation on vulnerable CPUs
5820 full,nosmt - Enable TAA mitigation and disable SMT on
5821 vulnerable CPUs. If TSX is disabled, SMT
5822 is not disabled because CPU is not
5823 vulnerable to cross-thread TAA attacks.
5824 off - Unconditionally disable TAA mitigation
5826 On MDS-affected machines, tsx_async_abort=off can be
5827 prevented by an active MDS mitigation as both vulnerabilities
5828 are mitigated with the same mechanism so in order to disable
5829 this mitigation, you need to specify mds=off too.
5831 Not specifying this option is equivalent to
5832 tsx_async_abort=full. On CPUs which are MDS affected
5833 and deploy MDS mitigation, TAA mitigation is not
5834 required and doesn't provide any additional
5838 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5840 turbografx.map[2|3]= [HW,JOY]
5841 TurboGraFX parallel port interface
5843 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
5844 See also Documentation/input/devices/joystick-parport.rst
5846 udbg-immortal [PPC] When debugging early kernel crashes that
5847 happen after console_init() and before a proper
5848 console driver takes over, this boot options might
5849 help "seeing" what's going on.
5851 uhash_entries= [KNL,NET]
5852 Set number of hash buckets for UDP/UDP-Lite connections
5855 [USB] Ignore overcurrent events (default N).
5856 Some badly-designed motherboards generate lots of
5857 bogus events, for ports that aren't wired to
5858 anything. Set this parameter to avoid log spamming.
5859 Note that genuine overcurrent events won't be
5863 [X86] Cause panic on unknown NMI.
5865 usbcore.authorized_default=
5866 [USB] Default USB device authorization:
5867 (default -1 = authorized except for wireless USB,
5868 0 = not authorized, 1 = authorized, 2 = authorized
5869 if device connected to internal port)
5871 usbcore.autosuspend=
5872 [USB] The autosuspend time delay (in seconds) used
5873 for newly-detected USB devices (default 2). This
5874 is the time required before an idle device will be
5875 autosuspended. Devices for which the delay is set
5876 to a negative value won't be autosuspended at all.
5878 usbcore.usbfs_snoop=
5879 [USB] Set to log all usbfs traffic (default 0 = off).
5881 usbcore.usbfs_snoop_max=
5882 [USB] Maximum number of bytes to snoop in each URB
5885 usbcore.blinkenlights=
5886 [USB] Set to cycle leds on hubs (default 0 = off).
5888 usbcore.old_scheme_first=
5889 [USB] Start with the old device initialization
5890 scheme (default 0 = off).
5892 usbcore.usbfs_memory_mb=
5893 [USB] Memory limit (in MB) for buffers allocated by
5894 usbfs (default = 16, 0 = max = 2047).
5896 usbcore.use_both_schemes=
5897 [USB] Try the other device initialization scheme
5898 if the first one fails (default 1 = enabled).
5900 usbcore.initial_descriptor_timeout=
5901 [USB] Specifies timeout for the initial 64-byte
5902 USB_REQ_GET_DESCRIPTOR request in milliseconds
5903 (default 5000 = 5.0 seconds).
5905 usbcore.nousb [USB] Disable the USB subsystem
5908 [USB] A list of quirk entries to augment the built-in
5909 usb core quirk list. List entries are separated by
5910 commas. Each entry has the form
5911 VendorID:ProductID:Flags. The IDs are 4-digit hex
5912 numbers and Flags is a set of letters. Each letter
5913 will change the built-in quirk; setting it if it is
5914 clear and clearing it if it is set. The letters have
5915 the following meanings:
5916 a = USB_QUIRK_STRING_FETCH_255 (string
5917 descriptors must not be fetched using
5919 b = USB_QUIRK_RESET_RESUME (device can't resume
5920 correctly so reset it instead);
5921 c = USB_QUIRK_NO_SET_INTF (device can't handle
5922 Set-Interface requests);
5923 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
5924 handle its Configuration or Interface
5926 e = USB_QUIRK_RESET (device can't be reset
5927 (e.g morph devices), don't use reset);
5928 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
5929 more interface descriptions than the
5930 bNumInterfaces count, and can't handle
5931 talking to these interfaces);
5932 g = USB_QUIRK_DELAY_INIT (device needs a pause
5933 during initialization, after we read
5934 the device descriptor);
5935 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
5936 high speed and super speed interrupt
5937 endpoints, the USB 2.0 and USB 3.0 spec
5938 require the interval in microframes (1
5939 microframe = 125 microseconds) to be
5940 calculated as interval = 2 ^
5942 Devices with this quirk report their
5943 bInterval as the result of this
5944 calculation instead of the exponent
5945 variable used in the calculation);
5946 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
5947 handle device_qualifier descriptor
5949 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
5950 generates spurious wakeup, ignore
5951 remote wakeup capability);
5952 k = USB_QUIRK_NO_LPM (device can't handle Link
5954 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
5955 (Device reports its bInterval as linear
5956 frames instead of the USB 2.0
5958 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
5959 to be disconnected before suspend to
5960 prevent spurious wakeup);
5961 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
5962 pause after every control message);
5963 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
5964 delay after resetting its port);
5965 Example: quirks=0781:5580:bk,0a5c:5834:gij
5968 [USBHID] The interval which mice are to be polled at.
5971 [USBHID] The interval which joysticks are to be polled at.
5974 [USBHID] The interval which keyboards are to be polled at.
5976 usb-storage.delay_use=
5977 [UMS] The delay in seconds before a new device is
5978 scanned for Logical Units (default 1).
5981 [UMS] A list of quirks entries to supplement or
5982 override the built-in unusual_devs list. List
5983 entries are separated by commas. Each entry has
5984 the form VID:PID:Flags where VID and PID are Vendor
5985 and Product ID values (4-digit hex numbers) and
5986 Flags is a set of characters, each corresponding
5987 to a common usb-storage quirk flag as follows:
5988 a = SANE_SENSE (collect more than 18 bytes
5989 of sense data, not on uas);
5990 b = BAD_SENSE (don't collect more than 18
5991 bytes of sense data, not on uas);
5992 c = FIX_CAPACITY (decrease the reported
5993 device capacity by one sector);
5994 d = NO_READ_DISC_INFO (don't use
5995 READ_DISC_INFO command, not on uas);
5996 e = NO_READ_CAPACITY_16 (don't use
5997 READ_CAPACITY_16 command);
5998 f = NO_REPORT_OPCODES (don't use report opcodes
6000 g = MAX_SECTORS_240 (don't transfer more than
6001 240 sectors at a time, uas only);
6002 h = CAPACITY_HEURISTICS (decrease the
6003 reported device capacity by one
6004 sector if the number is odd);
6005 i = IGNORE_DEVICE (don't bind to this
6007 j = NO_REPORT_LUNS (don't use report luns
6009 k = NO_SAME (do not use WRITE_SAME, uas only)
6010 l = NOT_LOCKABLE (don't try to lock and
6011 unlock ejectable media, not on uas);
6012 m = MAX_SECTORS_64 (don't transfer more
6013 than 64 sectors = 32 KB at a time,
6015 n = INITIAL_READ10 (force a retry of the
6016 initial READ(10) command, not on uas);
6017 o = CAPACITY_OK (accept the capacity
6018 reported by the device, not on uas);
6019 p = WRITE_CACHE (the device cache is ON
6020 by default, not on uas);
6021 r = IGNORE_RESIDUE (the device reports
6022 bogus residue values, not on uas);
6023 s = SINGLE_LUN (the device has only one
6025 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
6026 commands, uas only);
6027 u = IGNORE_UAS (don't bind to the uas driver);
6028 w = NO_WP_DETECT (don't test whether the
6029 medium is write-protected).
6030 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
6031 even if the device claims no cache,
6033 Example: quirks=0419:aaf5:rl,0421:0433:rc
6035 user_debug= [KNL,ARM]
6037 See arch/arm/Kconfig.debug help text.
6038 1 - undefined instruction events
6040 4 - invalid data aborts
6043 Example: user_debug=31
6046 [X86] Flags controlling user PTE allocations.
6048 nohigh = do not allocate PTE pages in
6049 HIGHMEM regardless of setting
6053 On X86_32, this is an alias for vdso32=. Otherwise:
6055 vdso=1: enable VDSO (the default)
6056 vdso=0: disable VDSO mapping
6058 vdso32= [X86] Control the 32-bit vDSO
6059 vdso32=1: enable 32-bit VDSO
6060 vdso32=0 or vdso32=2: disable 32-bit VDSO
6062 See the help text for CONFIG_COMPAT_VDSO for more
6063 details. If CONFIG_COMPAT_VDSO is set, the default is
6064 vdso32=0; otherwise, the default is vdso32=1.
6066 For compatibility with older kernels, vdso32=2 is an
6069 Try vdso32=0 if you encounter an error that says:
6070 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
6073 vector=percpu: enable percpu vector domain
6075 video= [FB] Frame buffer configuration
6076 See Documentation/fb/modedb.rst.
6078 video.brightness_switch_enabled= [0,1]
6079 If set to 1, on receiving an ACPI notify event
6080 generated by hotkey, video driver will adjust brightness
6081 level and then send out the event to user space through
6082 the allocated input device; If set to 0, video driver
6083 will only send out the event without touching backlight
6088 [VMMIO] Memory mapped virtio (platform) device.
6090 <size>@<baseaddr>:<irq>[:<id>]
6092 <size> := size (can use standard suffixes
6094 <baseaddr> := physical base address
6095 <irq> := interrupt number (as passed to
6097 <id> := (optional) platform device id
6099 virtio_mmio.device=1K@0x100b0000:48:7
6101 Can be used multiple times for multiple devices.
6103 vga= [BOOT,X86-32] Select a particular video mode
6104 See Documentation/x86/boot.rst and
6105 Documentation/admin-guide/svga.rst.
6106 Use vga=ask for menu.
6107 This is actually a boot loader parameter; the value is
6108 passed to the kernel using a special protocol.
6110 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
6111 May slow down system boot speed, especially when
6112 enabled on systems with a large amount of memory.
6113 All options are enabled by default, and this
6114 interface is meant to allow for selectively
6115 enabling or disabling specific virtual memory
6118 Available options are:
6119 P Enable page structure init time poisoning
6120 - Disable all of the above options
6122 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
6123 size of <nn>. This can be used to increase the
6124 minimum size (128MB on x86). It can also be used to
6125 decrease the size and leave more room for directly
6128 vmcp_cma=nn[MG] [KNL,S390]
6129 Sets the memory size reserved for contiguous memory
6130 allocations for the vmcp device driver.
6132 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
6135 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
6138 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
6142 Controls the behavior of vsyscalls (i.e. calls to
6143 fixed addresses of 0xffffffffff600x00 from legacy
6144 code). Most statically-linked binaries and older
6145 versions of glibc use these calls. Because these
6146 functions are at fixed addresses, they make nice
6147 targets for exploits that can control RIP.
6149 emulate [default] Vsyscalls turn into traps and are
6150 emulated reasonably safely. The vsyscall
6153 xonly Vsyscalls turn into traps and are
6154 emulated reasonably safely. The vsyscall
6155 page is not readable.
6157 none Vsyscalls don't work at all. This makes
6158 them quite hard to use for exploits but
6159 might break your system.
6161 vt.color= [VT] Default text color.
6162 Format: 0xYX, X = foreground, Y = background.
6163 Default: 0x07 = light gray on black.
6165 vt.cur_default= [VT] Default cursor shape.
6166 Format: 0xCCBBAA, where AA, BB, and CC are the same as
6167 the parameters of the <Esc>[?A;B;Cc escape sequence;
6168 see VGA-softcursor.txt. Default: 2 = underline.
6170 vt.default_blu= [VT]
6171 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
6172 Change the default blue palette of the console.
6173 This is a 16-member array composed of values
6176 vt.default_grn= [VT]
6177 Format: <green0>,<green1>,<green2>,...,<green15>
6178 Change the default green palette of the console.
6179 This is a 16-member array composed of values
6182 vt.default_red= [VT]
6183 Format: <red0>,<red1>,<red2>,...,<red15>
6184 Change the default red palette of the console.
6185 This is a 16-member array composed of values
6191 Set system-wide default UTF-8 mode for all tty's.
6192 Default is 1, i.e. UTF-8 mode is enabled for all
6193 newly opened terminals.
6195 vt.global_cursor_default=
6198 Set system-wide default for whether a cursor
6199 is shown on new VTs. Default is -1,
6200 i.e. cursors will be created by default unless
6201 overridden by individual drivers. 0 will hide
6202 cursors, 1 will display them.
6204 vt.italic= [VT] Default color for italic text; 0-15.
6207 vt.underline= [VT] Default color for underlined text; 0-15.
6210 watchdog timers [HW,WDT] For information on watchdog timers,
6211 see Documentation/watchdog/watchdog-parameters.rst
6212 or other driver-specific files in the
6213 Documentation/watchdog/ directory.
6217 Set the hard lockup detector stall duration
6218 threshold in seconds. The soft lockup detector
6219 threshold is set to twice the value. A value of 0
6220 disables both lockup detectors. Default is 10
6223 workqueue.watchdog_thresh=
6224 If CONFIG_WQ_WATCHDOG is configured, workqueue can
6225 warn stall conditions and dump internal state to
6226 help debugging. 0 disables workqueue stall
6227 detection; otherwise, it's the stall threshold
6228 duration in seconds. The default value is 30 and
6229 it can be updated at runtime by writing to the
6230 corresponding sysfs file.
6232 workqueue.disable_numa
6233 By default, all work items queued to unbound
6234 workqueues are affine to the NUMA nodes they're
6235 issued on, which results in better behavior in
6236 general. If NUMA affinity needs to be disabled for
6237 whatever reason, this option can be used. Note
6238 that this also can be controlled per-workqueue for
6239 workqueues visible under /sys/bus/workqueue/.
6241 workqueue.power_efficient
6242 Per-cpu workqueues are generally preferred because
6243 they show better performance thanks to cache
6244 locality; unfortunately, per-cpu workqueues tend to
6245 be more power hungry than unbound workqueues.
6247 Enabling this makes the per-cpu workqueues which
6248 were observed to contribute significantly to power
6249 consumption unbound, leading to measurably lower
6250 power usage at the cost of small performance
6253 The default value of this parameter is determined by
6254 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
6256 workqueue.debug_force_rr_cpu
6257 Workqueue used to implicitly guarantee that work
6258 items queued without explicit CPU specified are put
6259 on the local CPU. This guarantee is no longer true
6260 and while local CPU is still preferred work items
6261 may be put on foreign CPUs. This debug option
6262 forces round-robin CPU selection to flush out
6263 usages which depend on the now broken guarantee.
6264 When enabled, memory and cache locality will be
6267 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
6268 default x2apic cluster mode on platforms
6271 xen_512gb_limit [KNL,X86-64,XEN]
6272 Restricts the kernel running paravirtualized under Xen
6273 to use only up to 512 GB of RAM. The reason to do so is
6274 crash analysis tools and Xen tools for doing domain
6275 save/restore/migration must be enabled to handle larger
6278 xen_emul_unplug= [HW,X86,XEN]
6279 Unplug Xen emulated devices
6280 Format: [unplug0,][unplug1]
6281 ide-disks -- unplug primary master IDE devices
6282 aux-ide-disks -- unplug non-primary-master IDE devices
6283 nics -- unplug network devices
6284 all -- unplug all emulated devices (NICs and IDE disks)
6285 unnecessary -- unplugging emulated devices is
6286 unnecessary even if the host did not respond to
6288 never -- do not unplug even if version check succeeds
6290 xen_legacy_crash [X86,XEN]
6291 Crash from Xen panic notifier, without executing late
6292 panic() code such as dumping handler.
6294 xen_nopvspin [X86,XEN]
6295 Disables the qspinlock slowpath using Xen PV optimizations.
6296 This parameter is obsoleted by "nopvspin" parameter, which
6297 has equivalent effect for XEN platform.
6300 Disables the PV optimizations forcing the HVM guest to
6301 run as generic HVM guest with no PV drivers.
6302 This option is obsoleted by the "nopv" option, which
6303 has equivalent effect for XEN platform.
6305 xen_no_vector_callback
6306 [KNL,X86,XEN] Disable the vector callback for Xen
6307 event channel interrupts.
6309 xen_scrub_pages= [XEN]
6310 Boolean option to control scrubbing pages before giving them back
6311 to Xen, for use by other domains. Can be also changed at runtime
6312 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
6313 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
6315 xen_timer_slop= [X86-64,XEN]
6316 Set the timer slop (in nanoseconds) for the virtual Xen
6317 timers (default is 100000). This adjusts the minimum
6318 delta of virtualized Xen timers, where lower values
6319 improve timer resolution at the expense of processing
6320 more timer interrupts.
6322 xen.event_eoi_delay= [XEN]
6323 How long to delay EOI handling in case of event
6324 storms (jiffies). Default is 10.
6326 xen.event_loop_timeout= [XEN]
6327 After which time (jiffies) the event handling loop
6328 should start to delay EOI handling. Default is 2.
6330 xen.fifo_events= [XEN]
6331 Boolean parameter to disable using fifo event handling
6332 even if available. Normally fifo event handling is
6333 preferred over the 2-level event handling, as it is
6334 fairer and the number of possible event channels is
6335 much higher. Default is on (use fifo events).
6337 nopv= [X86,XEN,KVM,HYPER_V,VMWARE]
6338 Disables the PV optimizations forcing the guest to run
6339 as generic guest with no PV drivers. Currently support
6340 XEN HVM, KVM, HYPER_V and VMWARE guest.
6342 nopvspin [X86,XEN,KVM]
6343 Disables the qspinlock slow path using PV optimizations
6344 which allow the hypervisor to 'idle' the guest on lock
6347 xirc2ps_cs= [NET,PCMCIA]
6349 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
6352 By default on POWER9 and above, the kernel will
6353 natively use the XIVE interrupt controller. This option
6354 allows the fallback firmware mode to be used:
6356 off Fallback to firmware control of XIVE interrupt
6357 controller on both pseries and powernv
6358 platforms. Only useful on POWER9 and above.
6360 xhci-hcd.quirks [USB,KNL]
6361 A hex value specifying bitmask with supplemental xhci
6362 host controller quirks. Meaning of each bit can be
6363 consulted in header drivers/usb/host/xhci.h.
6366 Format: { early | on | rw | ro | off }
6367 Controls if xmon debugger is enabled. Default is off.
6368 Passing only "xmon" is equivalent to "xmon=early".
6369 early Call xmon as early as possible on boot; xmon
6370 debugger is called from setup_arch().
6371 on xmon debugger hooks will be installed so xmon
6372 is only called on a kernel crash. Default mode,
6373 i.e. either "ro" or "rw" mode, is controlled
6374 with CONFIG_XMON_DEFAULT_RO_MODE.
6375 rw xmon debugger hooks will be installed so xmon
6376 is called only on a kernel crash, mode is write,
6377 meaning SPR registers, memory and, other data
6378 can be written using xmon commands.
6379 ro same as "rw" option above but SPR registers,
6380 memory, and other data can't be written using
6382 off xmon is disabled.