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 amd_iommu= [HW,X86-64]
291 Pass parameters to the AMD IOMMU driver in the system.
293 fullflush - enable flushing of IO/TLB entries when
294 they are unmapped. Otherwise they are
295 flushed before they will be reused, which
297 off - do not initialize any AMD IOMMU found in
299 force_isolation - Force device isolation for all
300 devices. The IOMMU driver is not
301 allowed anymore to lift isolation
302 requirements as needed. This option
303 does not override iommu=pt
304 force_enable - Force enable the IOMMU on platforms known
305 to be buggy with IOMMU enabled. Use this
308 amd_iommu_dump= [HW,X86-64]
309 Enable AMD IOMMU driver option to dump the ACPI table
310 for AMD IOMMU. With this option enabled, AMD IOMMU
311 driver will print ACPI tables for AMD IOMMU during
312 IOMMU initialization.
314 amd_iommu_intr= [HW,X86-64]
315 Specifies one of the following AMD IOMMU interrupt
317 legacy - Use legacy interrupt remapping mode.
318 vapic - Use virtual APIC mode, which allows IOMMU
319 to inject interrupts directly into guest.
320 This mode requires kvm-amd.avic=1.
321 (Default when IOMMU HW support is present.)
323 amijoy.map= [HW,JOY] Amiga joystick support
324 Map of devices attached to JOY0DAT and JOY1DAT
326 See also Documentation/input/joydev/joystick.rst
328 analog.map= [HW,JOY] Analog joystick and gamepad support
329 Specifies type or capabilities of an analog joystick
330 connected to one of 16 gameports
331 Format: <type1>,<type2>,..<type16>
334 Power management functions (SPARCstation-4/5 + deriv.)
336 Disable APC CPU standby support. SPARCstation-Fox does
337 not play well with APC CPU idle - disable it if you have
338 APC and your system crashes randomly.
340 apic= [APIC,X86] Advanced Programmable Interrupt Controller
341 Change the output verbosity while booting
342 Format: { quiet (default) | verbose | debug }
343 Change the amount of debugging information output
344 when initialising the APIC and IO-APIC components.
345 For X86-32, this can also be used to specify an APIC
347 Format: apic=driver_name
348 Examples: apic=bigsmp
350 apic_extnmi= [APIC,X86] External NMI delivery setting
351 Format: { bsp (default) | all | none }
352 bsp: External NMI is delivered only to CPU 0
353 all: External NMIs are broadcast to all CPUs as a
355 none: External NMI is masked for all CPUs. This is
356 useful so that a dump capture kernel won't be
360 See Documentation/networking/ipv6.rst.
362 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
363 Limit apic dumping. The parameter defines the maximal
364 number of local apics being dumped. Also it is possible
365 to set it to "all" by meaning -- no limit here.
366 Format: { 1 (default) | 2 | ... | all }.
367 The parameter valid if only apic=debug or
368 apic=verbose is specified.
369 Example: apic=debug show_lapic=all
371 apm= [APM] Advanced Power Management
372 See header of arch/x86/kernel/apm_32.c.
374 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
375 Format: <io>,<irq>,<nodeID>
377 arm64.nobti [ARM64] Unconditionally disable Branch Target
378 Identification support
380 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication
383 arm64.nomte [ARM64] Unconditionally disable Memory Tagging Extension
388 atarimouse= [HW,MOUSE] Atari Mouse
390 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
391 EzKey and similar keyboards
393 atkbd.reset= [HW] Reset keyboard during initialization
395 atkbd.set= [HW] Select keyboard code set
396 Format: <int> (2 = AT (default), 3 = PS/2)
398 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
401 atkbd.softraw= [HW] Choose between synthetic and real raw mode
402 Format: <bool> (0 = real, 1 = synthetic (default))
404 atkbd.softrepeat= [HW]
405 Use software keyboard repeat
407 audit= [KNL] Enable the audit sub-system
408 Format: { "0" | "1" | "off" | "on" }
409 0 | off - kernel audit is disabled and can not be
410 enabled until the next reboot
411 unset - kernel audit is initialized but disabled and
412 will be fully enabled by the userspace auditd.
413 1 | on - kernel audit is initialized and partially
414 enabled, storing at most audit_backlog_limit
415 messages in RAM until it is fully enabled by the
419 audit_backlog_limit= [KNL] Set the audit queue size limit.
420 Format: <int> (must be >=0)
423 bau= [X86_UV] Enable the BAU on SGI UV. The default
424 behavior is to disable the BAU (i.e. bau=0).
425 Format: { "0" | "1" }
428 unset - Disable the BAU.
430 baycom_epp= [HW,AX25]
433 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
435 See header of drivers/net/hamradio/baycom_par.c.
437 baycom_ser_fdx= [HW,AX25]
438 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
439 Format: <io>,<irq>,<mode>[,<baud>]
440 See header of drivers/net/hamradio/baycom_ser_fdx.c.
442 baycom_ser_hdx= [HW,AX25]
443 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
444 Format: <io>,<irq>,<mode>
445 See header of drivers/net/hamradio/baycom_ser_hdx.c.
447 blkdevparts= Manual partition parsing of block device(s) for
448 embedded devices based on command line input.
449 See Documentation/block/cmdline-partition.rst
451 boot_delay= Milliseconds to delay each printk during boot.
452 Values larger than 10 seconds (10000) are changed to
457 Extended command line options can be added to an initrd
458 and this will cause the kernel to look for it.
460 See Documentation/admin-guide/bootconfig.rst
463 Disable BERT OS support on buggy BIOSes.
465 bgrt_disable [ACPI][X86]
466 Disable BGRT to avoid flickering OEM logo.
468 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
469 bttv.radio= Most important insmod options are available as
471 bttv.pll= See Documentation/admin-guide/media/bttv.rst
474 bulk_remove=off [PPC] This parameter disables the use of the pSeries
475 firmware feature for flushing multiple hpte entries
478 c101= [NET] Moxa C101 synchronous serial card
480 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
481 Sometimes CPU hardware bugs make them report the cache
482 size incorrectly. The kernel will attempt work arounds
483 to fix known problems, but for some CPUs it is not
484 possible to determine what the correct size should be.
485 This option provides an override for these situations.
488 [NET] Specifies amount of time (in seconds) that
489 the kernel should wait for a network carrier. By default
490 it waits 120 seconds.
492 ca_keys= [KEYS] This parameter identifies a specific key(s) on
493 the system trusted keyring to be used for certificate
495 format: { id:<keyid> | builtin }
497 cca= [MIPS] Override the kernel pages' cache coherency
498 algorithm. Accepted values range from 0 to 7
499 inclusive. See arch/mips/include/asm/pgtable-bits.h
500 for platform specific values (SB1, Loongson3 and
503 ccw_timeout_log [S390]
504 See Documentation/s390/common_io.rst for details.
506 cgroup_disable= [KNL] Disable a particular controller or optional feature
507 Format: {name of the controller(s) or feature(s) to disable}
508 The effects of cgroup_disable=foo are:
509 - foo isn't auto-mounted if you mount all cgroups in
511 - foo isn't visible as an individually mountable
513 - if foo is an optional feature then the feature is
514 disabled and corresponding cgroup files are not
516 {Currently only "memory" controller deal with this and
517 cut the overhead, others just disable the usage. So
518 only cgroup_disable=memory is actually worthy}
519 Specifying "pressure" disables per-cgroup pressure
520 stall information accounting feature
522 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
523 Format: { { controller | "all" | "named" }
524 [,{ controller | "all" | "named" }...] }
525 Like cgroup_disable, but only applies to cgroup v1;
526 the blacklisted controllers remain available in cgroup2.
527 "all" blacklists all controllers and "named" disables
528 named mounts. Specifying both "all" and "named" disables
531 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
533 nosocket -- Disable socket memory accounting.
534 nokmem -- Disable kernel memory accounting.
536 checkreqprot [SELINUX] Set initial checkreqprot flag value.
537 Format: { "0" | "1" }
538 See security/selinux/Kconfig help text.
539 0 -- check protection applied by kernel (includes
540 any implied execute protection).
541 1 -- check protection requested by application.
542 Default value is set via a kernel config option.
543 Value can be changed at runtime via
544 /sys/fs/selinux/checkreqprot.
545 Setting checkreqprot to 1 is deprecated.
548 See Documentation/s390/common_io.rst for details.
551 Prevents the clock framework from automatically gating
552 clocks that have not been explicitly enabled by a Linux
553 device driver but are enabled in hardware at reset or
554 by the bootloader/firmware. Note that this does not
555 force such clocks to be always-on nor does it reserve
556 those clocks in any way. This parameter is useful for
557 debug and development, but should not be needed on a
558 platform with proper driver support. For more
559 information, see Documentation/driver-api/clk.rst.
561 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
563 Forces specified clocksource (if available) to be used
564 when calculating gettimeofday(). If specified
565 clocksource is not available, it defaults to PIT.
566 Format: { pit | tsc | cyclone | pmtmr }
568 clocksource= Override the default clocksource
570 Override the default clocksource and use the clocksource
571 with the name specified.
572 Some clocksource names to choose from, depending on
574 [all] jiffies (this is the base, fallback clocksource)
576 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
577 pxa_timer,timer3,32k_counter,timer0_1
578 [X86-32] pit,hpet,tsc;
579 scx200_hrt on Geode; cyclone on IBM x440
587 clocksource.arm_arch_timer.evtstrm=
590 Enable/disable the eventstream feature of the ARM
591 architected timer so that code using WFE-based polling
592 loops can be debugged more effectively on production
595 clocksource.max_cswd_read_retries= [KNL]
596 Number of clocksource_watchdog() retries due to
597 external delays before the clock will be marked
598 unstable. Defaults to three retries, that is,
599 four attempts to read the clock under test.
601 clocksource.verify_n_cpus= [KNL]
602 Limit the number of CPUs checked for clocksources
603 marked with CLOCK_SOURCE_VERIFY_PERCPU that
604 are marked unstable due to excessive skew.
605 A negative value says to check all CPUs, while
606 zero says not to check any. Values larger than
607 nr_cpu_ids are silently truncated to nr_cpu_ids.
608 The actual CPUs are chosen randomly, with
609 no replacement if the same CPU is chosen twice.
611 clocksource-wdtest.holdoff= [KNL]
612 Set the time in seconds that the clocksource
613 watchdog test waits before commencing its tests.
614 Defaults to zero when built as a module and to
615 10 seconds when built into the kernel.
617 clearcpuid=BITNUM[,BITNUM...] [X86]
618 Disable CPUID feature X for the kernel. See
619 arch/x86/include/asm/cpufeatures.h for the valid bit
620 numbers. Note the Linux specific bits are not necessarily
621 stable over kernel options, but the vendor specific
623 Also note that user programs calling CPUID directly
624 or using the feature without checking anything
625 will still see it. This just prevents it from
626 being used by the kernel or shown in /proc/cpuinfo.
627 Also note the kernel might malfunction if you disable
630 cma=nn[MG]@[start[MG][-end[MG]]]
632 Sets the size of kernel global memory area for
633 contiguous memory allocations and optionally the
634 placement constraint by the physical address range of
635 memory allocations. A value of 0 disables CMA
636 altogether. For more information, see
637 kernel/dma/contiguous.c
641 Sets the size of kernel per-numa memory area for
642 contiguous memory allocations. A value of 0 disables
643 per-numa CMA altogether. And If this option is not
644 specificed, the default value is 0.
645 With per-numa CMA enabled, DMA users on node nid will
646 first try to allocate buffer from the pernuma area
647 which is located in node nid, if the allocation fails,
648 they will fallback to the global default memory area.
650 cmo_free_hint= [PPC] Format: { yes | no }
651 Specify whether pages are marked as being inactive
652 when they are freed. This is used in CMO environments
653 to determine OS memory pressure for page stealing by
657 coherent_pool=nn[KMG] [ARM,KNL]
658 Sets the size of memory pool for coherent, atomic dma
659 allocations, by default set to 256K.
661 com20020= [HW,NET] ARCnet - COM20020 chipset
663 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
665 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
669 ARCnet - COM90xx chipset (memory-mapped buffers)
670 Format: <io>[,<irq>[,<memstart>]]
672 condev= [HW,S390] console device
675 console= [KNL] Output console device and options.
677 tty<n> Use the virtual console device <n>.
681 Use the specified serial port. The options are of
682 the form "bbbbpnf", where "bbbb" is the baud rate,
683 "p" is parity ("n", "o", or "e"), "n" is number of
684 bits, and "f" is flow control ("r" for RTS or
685 omit it). Default is "9600n8".
687 See Documentation/admin-guide/serial-console.rst for more
689 Documentation/networking/netconsole.rst for an
692 uart[8250],io,<addr>[,options]
693 uart[8250],mmio,<addr>[,options]
694 uart[8250],mmio16,<addr>[,options]
695 uart[8250],mmio32,<addr>[,options]
696 uart[8250],0x<addr>[,options]
697 Start an early, polled-mode console on the 8250/16550
698 UART at the specified I/O port or MMIO address,
699 switching to the matching ttyS device later.
700 MMIO inter-register address stride is either 8-bit
701 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
702 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
703 to be equivalent to 'mmio'. 'options' are specified in
704 the same format described for ttyS above; if unspecified,
705 the h/w is not re-initialized.
707 hvc<n> Use the hypervisor console device <n>. This is for
708 both Xen and PowerPC hypervisors.
710 If the device connected to the port is not a TTY but a braille
711 device, prepend "brl," before the device type, for instance
713 For now, only VisioBraille is supported.
716 [KNL] Change console messages format
718 By default we print messages on consoles in
719 "[time stamp] text\n" format (time stamp may not be
720 printed, depending on CONFIG_PRINTK_TIME or
721 `printk_time' param).
723 Switch to syslog format: "<%u>[time stamp] text\n"
724 IOW, each message will have a facility and loglevel
725 prefix. The format is similar to one used by syslog()
726 syscall, or to executing "dmesg -S --raw" or to reading
729 consoleblank= [KNL] The console blank (screen saver) timeout in
730 seconds. A value of 0 disables the blank timer.
734 [KNL] Change the default value for
735 /proc/<pid>/coredump_filter.
736 See also Documentation/filesystems/proc.rst.
738 coresight_cpu_debug.enable
741 Enable/disable the CPU sampling based debugging.
742 0: default value, disable debugging
743 1: enable debugging at boot time
745 cpuidle.off=1 [CPU_IDLE]
746 disable the cpuidle sub-system
749 [CPU_IDLE] Name of the cpuidle governor to use.
751 cpufreq.off=1 [CPU_FREQ]
752 disable the cpufreq sub-system
754 cpufreq.default_governor=
755 [CPU_FREQ] Name of the default cpufreq governor or
756 policy to use. This governor must be registered in the
757 kernel before the cpufreq driver probes.
760 [X86] Delay for N microsec between assert and de-assert
761 of APIC INIT to start processors. This delay occurs
762 on every CPU online, such as boot, and resume from suspend.
765 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
767 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
769 crashkernel=size[KMG][@offset[KMG]]
770 [KNL] Using kexec, Linux can switch to a 'crash kernel'
771 upon panic. This parameter reserves the physical
772 memory region [offset, offset + size] for that kernel
773 image. If '@offset' is omitted, then a suitable offset
774 is selected automatically.
775 [KNL, X86-64] Select a region under 4G first, and
776 fall back to reserve region above 4G when '@offset'
777 hasn't been specified.
778 See Documentation/admin-guide/kdump/kdump.rst for further details.
780 crashkernel=range1:size1[,range2:size2,...][@offset]
781 [KNL] Same as above, but depends on the memory
782 in the running system. The syntax of range is
783 start-[end] where start and end are both
784 a memory unit (amount[KMG]). See also
785 Documentation/admin-guide/kdump/kdump.rst for an example.
787 crashkernel=size[KMG],high
788 [KNL, X86-64] range could be above 4G. Allow kernel
789 to allocate physical memory region from top, so could
790 be above 4G if system have more than 4G ram installed.
791 Otherwise memory region will be allocated below 4G, if
793 It will be ignored if crashkernel=X is specified.
794 crashkernel=size[KMG],low
795 [KNL, X86-64] range under 4G. When crashkernel=X,high
796 is passed, kernel could allocate physical memory region
797 above 4G, that cause second kernel crash on system
798 that require some amount of low memory, e.g. swiotlb
799 requires at least 64M+32K low memory, also enough extra
800 low memory is needed to make sure DMA buffers for 32-bit
801 devices won't run out. Kernel would try to allocate at
802 at least 256M below 4G automatically.
803 This one let user to specify own low range under 4G
804 for second kernel instead.
805 0: to disable low allocation.
806 It will be ignored when crashkernel=X,high is not used
807 or memory reserved is below 4G.
810 [KNL] Disable crypto self-tests
815 cs89x0_media= [HW,NET]
816 Format: { rj45 | aui | bnc }
818 csdlock_debug= [KNL] Enable debug add-ons of cross-CPU function call
819 handling. When switched on, additional debug data is
820 printed to the console in case a hanging CPU is
821 detected, and that CPU is pinged again in order to try
822 to resolve the hang situation.
823 0: disable csdlock debugging (default)
824 1: enable basic csdlock debugging (minor impact)
825 ext: enable extended csdlock debugging (more impact,
829 See header of drivers/s390/block/dasd_devmap.c.
831 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
832 (one device per port)
833 Format: <port#>,<type>
834 See also Documentation/input/devices/joystick-parport.rst
836 ddebug_query= [KNL,DYNAMIC_DEBUG] Enable debug messages at early boot
838 Documentation/admin-guide/dynamic-debug-howto.rst for
839 details. Deprecated, see dyndbg.
841 debug [KNL] Enable kernel debugging (events log level).
844 [KNL] Enable printing [hashed] pointers early in the
845 boot sequence. If enabled, we use a weak hash instead
846 of siphash to hash pointers. Use this option if you are
847 seeing instances of '(___ptrval___)') and need to see a
848 value (hashed pointer) instead. Cryptographically
849 insecure, please do not use on production kernels.
852 [KNL] verbose locking self-tests
854 Print debugging info while doing the locking API
856 Bitmask for the various LOCKTYPE_ tests. Defaults to 0
857 (no extra messages), setting it to -1 (all bits set)
858 will print _a_lot_ more information - normally only
859 useful to lockdep developers.
861 debug_objects [KNL] Enable object debugging
864 [KNL] Disable object debugging
866 debug_guardpage_minorder=
867 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
868 parameter allows control of the order of pages that will
869 be intentionally kept free (and hence protected) by the
870 buddy allocator. Bigger value increase the probability
871 of catching random memory corruption, but reduce the
872 amount of memory for normal system use. The maximum
873 possible value is MAX_ORDER/2. Setting this parameter
874 to 1 or 2 should be enough to identify most random
875 memory corruption problems caused by bugs in kernel or
876 driver code when a CPU writes to (or reads from) a
877 random memory location. Note that there exists a class
878 of memory corruptions problems caused by buggy H/W or
879 F/W or by drivers badly programing DMA (basically when
880 memory is written at bus level and the CPU MMU is
881 bypassed) which are not detectable by
882 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
883 tracking down these problems.
886 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
887 enables the feature at boot time. By default, it is
888 disabled and the system will work mostly the same as a
889 kernel built without CONFIG_DEBUG_PAGEALLOC.
890 Note: to get most of debug_pagealloc error reports, it's
891 useful to also enable the page_owner functionality.
892 on: enable the feature
894 debugfs= [KNL] This parameter enables what is exposed to userspace
895 and debugfs internal clients.
896 Format: { on, no-mount, off }
897 on: All functions are enabled.
899 Filesystem is not registered but kernel clients can
900 access APIs and a crashkernel can be used to read
901 its content. There is nothing to mount.
902 off: Filesystem is not registered and clients
903 get a -EPERM as result when trying to register files
904 or directories within debugfs.
905 This is equivalent of the runtime functionality if
906 debugfs was not enabled in the kernel at all.
907 Default value is set in build-time with a kernel configuration.
909 debugpat [X86] Enable PAT debugging
911 decnet.addr= [HW,NET]
912 Format: <area>[,<node>]
913 See also Documentation/networking/decnet.rst.
916 [HW] The size of the default HugeTLB page. This is
917 the size represented by the legacy /proc/ hugepages
918 APIs. In addition, this is the default hugetlb size
919 used for shmget(), mmap() and mounting hugetlbfs
920 filesystems. If not specified, defaults to the
921 architecture's default huge page size. Huge page
922 sizes are architecture dependent. See also
923 Documentation/admin-guide/mm/hugetlbpage.rst.
926 deferred_probe_timeout=
927 [KNL] Debugging option to set a timeout in seconds for
928 deferred probe to give up waiting on dependencies to
929 probe. Only specific dependencies (subsystems or
930 drivers) that have opted in will be ignored. A timeout of 0
931 will timeout at the end of initcalls. This option will also
932 dump out devices still on the deferred probe list after
936 Format: { on | off | def_only | inf_only | always }
937 on: s390 zlib hardware support for compression on
938 level 1 and decompression (default)
939 off: No s390 zlib hardware support
940 def_only: s390 zlib hardware support for deflate
941 only (compression on level 1)
942 inf_only: s390 zlib hardware support for inflate
944 always: Same as 'on' but ignores the selected compression
945 level always using hardware support (used for debugging)
948 Set number of hash buckets for dentry cache.
950 disable_1tb_segments [PPC]
951 Disables the use of 1TB hash page table segments. This
952 causes the kernel to fall back to 256MB segments which
953 can be useful when debugging issues that require an SLB
957 Limits the number of kernel SLB entries, and flushes
958 them frequently to increase the rate of SLB faults
962 See Documentation/networking/ipv6.rst.
965 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
966 hardening is enabled for this boot. Hardened
967 usercopy checking is used to protect the kernel
968 from reading or writing beyond known memory
969 allocation boundaries as a proactive defense
970 against bounds-checking flaws in the kernel's
971 copy_to_user()/copy_from_user() interface.
972 on Perform hardened usercopy checks (default).
973 off Disable hardened usercopy checks.
976 Disable RADIX MMU mode on POWER9
978 radix_hcall_invalidate=on [PPC/PSERIES]
979 Disable RADIX GTSE feature and use hcall for TLB
983 Disable TLBIE instruction. Currently does not work
984 with KVM, with HASH MMU, or with coherent accelerators.
986 disable_cpu_apicid= [X86,APIC,SMP]
988 The number of initial APIC ID for the
989 corresponding CPU to be disabled at boot,
990 mostly used for the kdump 2nd kernel to
991 disable BSP to wake up multiple CPUs without
992 causing system reset or hang due to sending
995 disable_ddw [PPC/PSERIES]
996 Disable Dynamic DMA Window support. Use this
997 to workaround buggy firmware.
1000 See Documentation/networking/ipv6.rst.
1002 disable_mtrr_cleanup [X86]
1003 The kernel tries to adjust MTRR layout from continuous
1004 to discrete, to make X server driver able to add WB
1005 entry later. This parameter disables that.
1007 disable_mtrr_trim [X86, Intel and AMD only]
1008 By default the kernel will trim any uncacheable
1009 memory out of your available memory pool based on
1010 MTRR settings. This parameter disables that behavior,
1011 possibly causing your machine to run very slowly.
1013 disable_timer_pin_1 [X86]
1014 Disable PIN 1 of APIC timer
1015 Can be useful to work around chipset bugs.
1017 dis_ucode_ldr [X86] Disable the microcode loader.
1019 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
1020 this option disables the debugging code at boot.
1022 dma_debug_entries=<number>
1023 This option allows to tune the number of preallocated
1024 entries for DMA-API debugging code. One entry is
1025 required per DMA-API allocation. Use this if the
1026 DMA-API debugging code disables itself because the
1027 architectural default is too low.
1029 dma_debug_driver=<driver_name>
1030 With this option the DMA-API debugging driver
1031 filter feature can be enabled at boot time. Just
1032 pass the driver to filter for as the parameter.
1033 The filter can be disabled or changed to another
1034 driver later using sysfs.
1036 driver_async_probe= [KNL]
1037 List of driver names to be probed asynchronously.
1038 Format: <driver_name1>,<driver_name2>...
1040 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
1041 Broken monitors, graphic adapters, KVMs and EDIDless
1042 panels may send no or incorrect EDID data sets.
1043 This parameter allows to specify an EDID data sets
1044 in the /lib/firmware directory that are used instead.
1045 Generic built-in EDID data sets are used, if one of
1046 edid/1024x768.bin, edid/1280x1024.bin,
1047 edid/1680x1050.bin, or edid/1920x1080.bin is given
1048 and no file with the same name exists. Details and
1049 instructions how to build your own EDID data are
1050 available in Documentation/admin-guide/edid.rst. An EDID
1051 data set will only be used for a particular connector,
1052 if its name and a colon are prepended to the EDID
1053 name. Each connector may use a unique EDID data
1054 set by separating the files with a comma. An EDID
1055 data set with no connector name will be used for
1056 any connectors not explicitly specified.
1061 Format: {"off" | "known"}
1062 Control how the dt_cpu_ftrs device-tree binding is
1063 used for CPU feature discovery and setup (if it
1065 off: Do not use it, fall back to legacy cpu table.
1066 known: Do not pass through unknown features to guests
1067 or userspace, only those that the kernel is aware of.
1069 dump_apple_properties [X86]
1070 Dump name and content of EFI device properties on
1071 x86 Macs. Useful for driver authors to determine
1072 what data is available or for reverse-engineering.
1074 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
1075 <module>.dyndbg[="val"]
1076 Enable debug messages at boot time. See
1077 Documentation/admin-guide/dynamic-debug-howto.rst
1080 nopku [X86] Disable Memory Protection Keys CPU feature found
1083 <module>.async_probe [KNL]
1084 Enable asynchronous probe on this module.
1086 early_ioremap_debug [KNL]
1087 Enable debug messages in early_ioremap support. This
1088 is useful for tracking down temporary early mappings
1089 which are not unmapped.
1091 earlycon= [KNL] Output early console device and options.
1093 When used with no options, the early console is
1094 determined by stdout-path property in device tree's
1095 chosen node or the ACPI SPCR table if supported by
1098 cdns,<addr>[,options]
1099 Start an early, polled-mode console on a Cadence
1100 (xuartps) serial port at the specified address. Only
1101 supported option is baud rate. If baud rate is not
1102 specified, the serial port must already be setup and
1105 uart[8250],io,<addr>[,options]
1106 uart[8250],mmio,<addr>[,options]
1107 uart[8250],mmio32,<addr>[,options]
1108 uart[8250],mmio32be,<addr>[,options]
1109 uart[8250],0x<addr>[,options]
1110 Start an early, polled-mode console on the 8250/16550
1111 UART at the specified I/O port or MMIO address.
1112 MMIO inter-register address stride is either 8-bit
1113 (mmio) or 32-bit (mmio32 or mmio32be).
1114 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1115 to be equivalent to 'mmio'. 'options' are specified
1116 in the same format described for "console=ttyS<n>"; if
1117 unspecified, the h/w is not initialized.
1121 Start an early, polled-mode console on a pl011 serial
1122 port at the specified address. The pl011 serial port
1123 must already be setup and configured. Options are not
1124 yet supported. If 'mmio32' is specified, then only
1125 the driver will use only 32-bit accessors to read/write
1126 the device registers.
1129 Start an early console on a litex serial port at the
1130 specified address. The serial port must already be
1131 setup and configured. Options are not yet supported.
1134 Start an early, polled-mode console on a meson serial
1135 port at the specified address. The serial port must
1136 already be setup and configured. Options are not yet
1140 Start an early, polled-mode console on an msm serial
1141 port at the specified address. The serial port
1142 must already be setup and configured. Options are not
1145 msm_serial_dm,<addr>
1146 Start an early, polled-mode console on an msm serial
1147 dm port at the specified address. The serial port
1148 must already be setup and configured. Options are not
1152 Start an early, polled-mode console on a serial port
1153 of an Actions Semi SoC, such as S500 or S900, at the
1154 specified address. The serial port must already be
1155 setup and configured. Options are not yet supported.
1158 Start an early, polled-mode console on a serial port
1159 of an RDA Micro SoC, such as RDA8810PL, at the
1160 specified address. The serial port must already be
1161 setup and configured. Options are not yet supported.
1164 Use RISC-V SBI (Supervisor Binary Interface) for early
1167 smh Use ARM semihosting calls for early console.
1175 Use early console provided by serial driver available
1176 on Samsung SoCs, requires selecting proper type and
1177 a correct base address of the selected UART port. The
1178 serial port must already be setup and configured.
1179 Options are not yet supported.
1182 Start an early, polled-mode console on a lantiq serial
1183 (lqasc) port at the specified address. The serial port
1184 must already be setup and configured. Options are not
1189 Use early console provided by Freescale LP UART driver
1190 found on Freescale Vybrid and QorIQ LS1021A processors.
1191 A valid base address must be provided, and the serial
1192 port must already be setup and configured.
1196 Start an early, polled-mode, output-only console on the
1197 Freescale i.MX UART at the specified address. The UART
1198 must already be setup and configured.
1201 Start an early, polled-mode console on the
1202 Armada 3700 serial port at the specified
1203 address. The serial port must already be setup
1204 and configured. Options are not yet supported.
1207 Start an early, polled-mode console on a Qualcomm
1208 Generic Interface (GENI) based serial port at the
1209 specified address. The serial port must already be
1210 setup and configured. Options are not yet supported.
1213 Start an early, unaccelerated console on the EFI
1214 memory mapped framebuffer (if available). On cache
1215 coherent non-x86 systems that use system memory for
1216 the framebuffer, pass the 'ram' option so that it is
1217 mapped with the correct attributes.
1220 Use early console provided by Freescale LINFlexD UART
1221 serial driver for NXP S32V234 SoCs. A valid base
1222 address must be provided, and the serial port must
1223 already be setup and configured.
1225 earlyprintk= [X86,SH,ARM,M68k,S390]
1229 earlyprintk=serial[,ttySn[,baudrate]]
1230 earlyprintk=serial[,0x...[,baudrate]]
1231 earlyprintk=ttySn[,baudrate]
1232 earlyprintk=dbgp[debugController#]
1233 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1234 earlyprintk=xdbc[xhciController#]
1236 earlyprintk is useful when the kernel crashes before
1237 the normal console is initialized. It is not enabled by
1238 default because it has some cosmetic problems.
1240 Append ",keep" to not disable it when the real console
1243 Only one of vga, efi, serial, or usb debug port can
1246 Currently only ttyS0 and ttyS1 may be specified by
1247 name. Other I/O ports may be explicitly specified
1248 on some architectures (x86 and arm at least) by
1249 replacing ttySn with an I/O port address, like this:
1250 earlyprintk=serial,0x1008,115200
1251 You can find the port for a given device in
1252 /proc/tty/driver/serial:
1253 2: uart:ST16650V2 port:00001008 irq:18 ...
1255 Interaction with the standard serial driver is not
1258 The VGA and EFI output is eventually overwritten by
1261 The xen output can only be used by Xen PV guests.
1263 The sclp output can only be used on s390.
1265 The optional "force" to "pciserial" enables use of a
1266 PCI device even when its classcode is not of the
1269 edac_report= [HW,EDAC] Control how to report EDAC event
1270 Format: {"on" | "off" | "force"}
1271 on: enable EDAC to report H/W event. May be overridden
1272 by other higher priority error reporting module.
1273 off: disable H/W event reporting through EDAC.
1274 force: enforce the use of EDAC to report H/W event.
1277 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1280 This is designed to be used in conjunction with
1281 the boot argument: earlyprintk=vga
1283 This parameter works in place of the kgdboc parameter
1284 but can only be used if the backing tty is available
1285 very early in the boot process. For early debugging
1286 via a serial port see kgdboc_earlycon instead.
1289 Format: {"off" | "on" | "skip[mbr]"}
1292 Format: { "debug", "disable_early_pci_dma",
1293 "nochunk", "noruntime", "nosoftreserve",
1294 "novamap", "no_disable_early_pci_dma" }
1295 debug: enable misc debug output.
1296 disable_early_pci_dma: disable the busmaster bit on all
1297 PCI bridges while in the EFI boot stub.
1298 nochunk: disable reading files in "chunks" in the EFI
1299 boot stub, as chunking can cause problems with some
1300 firmware implementations.
1301 noruntime : disable EFI runtime services support
1302 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1303 attribute may cause the kernel to reserve the
1304 memory range for a memory mapping driver to
1305 claim. Specify efi=nosoftreserve to disable this
1306 reservation and treat the memory by its base type
1307 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1308 novamap: do not call SetVirtualAddressMap().
1309 no_disable_early_pci_dma: Leave the busmaster bit set
1310 on all PCI bridges while in the EFI boot stub
1312 efi_no_storage_paranoia [EFI; X86]
1313 Using this parameter you can use more than 50% of
1314 your efi variable storage. Use this parameter only if
1315 you are really sure that your UEFI does sane gc and
1316 fulfills the spec otherwise your board may brick.
1318 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1319 Add arbitrary attribute to specific memory range by
1320 updating original EFI memory map.
1321 Region of memory which aa attribute is added to is
1324 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1325 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1326 attribute is added to range 0x100000000-0x180000000 and
1327 0x10a0000000-0x1120000000.
1329 If efi_fake_mem=8G@9G:0x40000 is specified, the
1330 EFI_MEMORY_SP(0x40000) attribute is added to
1331 range 0x240000000-0x43fffffff.
1333 Using this parameter you can do debugging of EFI memmap
1334 related features. For example, you can do debugging of
1335 Address Range Mirroring feature even if your box
1336 doesn't support it, or mark specific memory as
1339 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1340 that is to be dynamically loaded by Linux. If there are
1341 multiple variables with the same name but with different
1342 vendor GUIDs, all of them will be loaded. See
1343 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1346 eisa_irq_edge= [PARISC,HW]
1347 See header of drivers/parisc/eisa.c.
1350 See comment before function elanfreq_setup() in
1351 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1353 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1354 Specifies physical address of start of kernel core
1355 image elf header and optionally the size. Generally
1356 kexec loader will pass this option to capture kernel.
1357 See Documentation/admin-guide/kdump/kdump.rst for details.
1359 enable_mtrr_cleanup [X86]
1360 The kernel tries to adjust MTRR layout from continuous
1361 to discrete, to make X server driver able to add WB
1362 entry later. This parameter enables that.
1364 enable_timer_pin_1 [X86]
1365 Enable PIN 1 of APIC timer
1366 Can be useful to work around chipset bugs
1367 (in particular on some ATI chipsets).
1368 The kernel tries to set a reasonable default.
1370 enforcing [SELINUX] Set initial enforcing status.
1372 See security/selinux/Kconfig help text.
1373 0 -- permissive (log only, no denials).
1374 1 -- enforcing (deny and log).
1376 Value can be changed at runtime via
1377 /sys/fs/selinux/enforce.
1380 Disable Error Record Serialization Table (ERST)
1383 ether= [HW,NET] Ethernet cards parameters
1384 This option is obsoleted by the "netdev=" option, which
1385 has equivalent usage. See its documentation for details.
1389 Permit 'security.evm' to be updated regardless of
1390 current integrity status.
1395 fail_make_request=[KNL]
1396 General fault injection mechanism.
1397 Format: <interval>,<probability>,<space>,<times>
1398 See also Documentation/fault-injection/.
1401 Format: { initns | none }
1402 See Documentation/admin-guide/sysctl/net.rst for
1403 fb_tunnels_only_for_init_ns
1406 See Documentation/admin-guide/blockdev/floppy.rst.
1408 force_pal_cache_flush
1409 [IA-64] Avoid check_sal_cache_flush which may hang on
1410 buggy SAL_CACHE_FLUSH implementations. Using this
1411 parameter will force ia64_sal_cache_flush to call
1412 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1415 Forcefully enable Physical Address Extension (PAE).
1416 Many Pentium M systems disable PAE but may have a
1417 functionally usable PAE implementation.
1418 Warning: use of this parameter will taint the kernel
1419 and may cause unknown problems.
1422 [FTRACE] will set and start the specified tracer
1423 as early as possible in order to facilitate early
1426 ftrace_dump_on_oops[=orig_cpu]
1427 [FTRACE] will dump the trace buffers on oops.
1428 If no parameter is passed, ftrace will dump
1429 buffers of all CPUs, but if you pass orig_cpu, it will
1430 dump only the buffer of the CPU that triggered the
1433 ftrace_filter=[function-list]
1434 [FTRACE] Limit the functions traced by the function
1435 tracer at boot up. function-list is a comma-separated
1436 list of functions. This list can be changed at run
1437 time by the set_ftrace_filter file in the debugfs
1440 ftrace_notrace=[function-list]
1441 [FTRACE] Do not trace the functions specified in
1442 function-list. This list can be changed at run time
1443 by the set_ftrace_notrace file in the debugfs
1446 ftrace_graph_filter=[function-list]
1447 [FTRACE] Limit the top level callers functions traced
1448 by the function graph tracer at boot up.
1449 function-list is a comma-separated list of functions
1450 that can be changed at run time by the
1451 set_graph_function file in the debugfs tracing directory.
1453 ftrace_graph_notrace=[function-list]
1454 [FTRACE] Do not trace from the functions specified in
1455 function-list. This list is a comma-separated list of
1456 functions that can be changed at run time by the
1457 set_graph_notrace file in the debugfs tracing directory.
1459 ftrace_graph_max_depth=<uint>
1460 [FTRACE] Used with the function graph tracer. This is
1461 the max depth it will trace into a function. This value
1462 can be changed at run time by the max_graph_depth file
1463 in the tracefs tracing directory. default: 0 (no limit)
1465 fw_devlink= [KNL] Create device links between consumer and supplier
1466 devices by scanning the firmware to infer the
1467 consumer/supplier relationships. This feature is
1468 especially useful when drivers are loaded as modules as
1469 it ensures proper ordering of tasks like device probing
1470 (suppliers first, then consumers), supplier boot state
1471 clean up (only after all consumers have probed),
1472 suspend/resume & runtime PM (consumers first, then
1474 Format: { off | permissive | on | rpm }
1475 off -- Don't create device links from firmware info.
1476 permissive -- Create device links from firmware info
1477 but use it only for ordering boot state clean
1478 up (sync_state() calls).
1479 on -- Create device links from firmware info and use it
1480 to enforce probe and suspend/resume ordering.
1481 rpm -- Like "on", but also use to order runtime PM.
1483 fw_devlink.strict=<bool>
1484 [KNL] Treat all inferred dependencies as mandatory
1485 dependencies. This only applies for fw_devlink=on|rpm.
1489 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1490 support via parallel port (up to 5 devices per port)
1491 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1492 See also Documentation/input/devices/joystick-parport.rst
1496 gart_fix_e820= [X86-64] disable the fix e820 for K8 GART
1500 gcov_persist= [GCOV] When non-zero (default), profiling data for
1501 kernel modules is saved and remains accessible via
1502 debugfs, even when the module is unloaded/reloaded.
1503 When zero, profiling data is discarded and associated
1504 debugfs files are removed at module unload time.
1506 goldfish [X86] Enable the goldfish android emulator platform.
1507 Don't use this when you are not running on the
1510 gpio-mockup.gpio_mockup_ranges
1511 [HW] Sets the ranges of gpiochip of for this device.
1512 Format: <start1>,<end1>,<start2>,<end2>...
1513 gpio-mockup.gpio_mockup_named_lines
1514 [HW] Let the driver know GPIO lines should be named.
1516 gpt [EFI] Forces disk with valid GPT signature but
1517 invalid Protective MBR to be treated as GPT. If the
1518 primary GPT is corrupted, it enables the backup/alternate
1519 GPT to be used instead.
1521 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1522 the "Enable 0" bit of the configuration register.
1525 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1526 the "Enable 0" bit of the configuration register.
1529 grcan.select= [HW] Select which physical interface to use.
1532 grcan.txsize= [HW] Sets the size of the tx buffer.
1533 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1535 grcan.rxsize= [HW] Sets the size of the rx buffer.
1536 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1539 hardlockup_all_cpu_backtrace=
1540 [KNL] Should the hard-lockup detector generate
1541 backtraces on all cpus.
1544 hashdist= [KNL,NUMA] Large hashes allocated during boot
1545 are distributed across NUMA nodes. Defaults on
1546 for 64-bit NUMA, off otherwise.
1547 Format: 0 | 1 (for off | on)
1549 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1551 hd= [EIDE] (E)IDE hard drive subsystem geometry
1552 Format: <cyl>,<head>,<sect>
1555 Disable Hardware Error Source Table (HEST) support;
1556 corresponding firmware-first mode error processing
1557 logic will be disabled.
1559 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1560 size of <nn>. This works even on boxes that have no
1561 highmem otherwise. This also works to reduce highmem
1562 size on bigger boxes.
1564 highres= [KNL] Enable/disable high resolution timer mode.
1565 Valid parameters: "on", "off"
1570 hpet= [X86-32,HPET] option to control HPET usage
1571 Format: { enable (default) | disable | force |
1573 disable: disable HPET and use PIT instead
1574 force: allow force enabled of undocumented chips (ICH4,
1576 verbose: show contents of HPET registers during setup
1578 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1579 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1581 hugetlb_cma= [HW,CMA] The size of a CMA area used for allocation
1582 of gigantic hugepages.
1585 Reserve a CMA area of given size and allocate gigantic
1586 hugepages using the CMA allocator. If enabled, the
1587 boot-time allocation of gigantic hugepages is skipped.
1589 hugepages= [HW] Number of HugeTLB pages to allocate at boot.
1590 If this follows hugepagesz (below), it specifies
1591 the number of pages of hugepagesz to be allocated.
1592 If this is the first HugeTLB parameter on the command
1593 line, it specifies the number of pages to allocate for
1594 the default huge page size. See also
1595 Documentation/admin-guide/mm/hugetlbpage.rst.
1599 [HW] The size of the HugeTLB pages. This is used in
1600 conjunction with hugepages (above) to allocate huge
1601 pages of a specific size at boot. The pair
1602 hugepagesz=X hugepages=Y can be specified once for
1603 each supported huge page size. Huge page sizes are
1604 architecture dependent. See also
1605 Documentation/admin-guide/mm/hugetlbpage.rst.
1608 hugetlb_free_vmemmap=
1609 [KNL] Reguires CONFIG_HUGETLB_PAGE_FREE_VMEMMAP
1611 Allows heavy hugetlb users to free up some more
1612 memory (6 * PAGE_SIZE for each 2MB hugetlb page).
1613 Format: { on | off (default) }
1615 on: enable the feature
1616 off: disable the feature
1618 Built with CONFIG_HUGETLB_PAGE_FREE_VMEMMAP_DEFAULT_ON=y,
1621 This is not compatible with memory_hotplug.memmap_on_memory.
1622 If both parameters are enabled, hugetlb_free_vmemmap takes
1623 precedence over memory_hotplug.memmap_on_memory.
1626 [KNL] Should the hung task detector generate panics.
1629 A value of 1 instructs the kernel to panic when a
1630 hung task is detected. The default value is controlled
1631 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1632 option. The value selected by this boot parameter can
1633 be changed later by the kernel.hung_task_panic sysctl.
1635 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1636 terminal devices. Valid values: 0..8
1637 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1638 If specified, z/VM IUCV HVC accepts connections
1639 from listed z/VM user IDs only.
1641 hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations
1642 which allow the hypervisor to 'idle' the
1643 guest on lock contention.
1646 Do not unregister boot console at start. This is only
1647 useful for debugging when something happens in the window
1648 between unregistering the boot console and initializing
1651 i2c_bus= [HW] Override the default board specific I2C bus speed
1652 or register an additional I2C bus that is not
1653 registered from board initialization code.
1657 i8042.debug [HW] Toggle i8042 debug mode
1658 i8042.unmask_kbd_data
1659 [HW] Enable printing of interrupt data from the KBD port
1660 (disabled by default, and as a pre-condition
1661 requires that i8042.debug=1 be enabled)
1662 i8042.direct [HW] Put keyboard port into non-translated mode
1663 i8042.dumbkbd [HW] Pretend that controller can only read data from
1664 keyboard and cannot control its state
1665 (Don't attempt to blink the leds)
1666 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1667 i8042.nokbd [HW] Don't check/create keyboard port
1668 i8042.noloop [HW] Disable the AUX Loopback command while probing
1670 i8042.nomux [HW] Don't check presence of an active multiplexing
1672 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1674 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1675 i8042.reset [HW] Reset the controller during init, cleanup and
1676 suspend-to-ram transitions, only during s2r
1677 transitions, or never reset
1678 Format: { 1 | Y | y | 0 | N | n }
1679 1, Y, y: always reset controller
1680 0, N, n: don't ever reset controller
1681 Default: only on s2r transitions on x86; most other
1682 architectures force reset to be always executed
1683 i8042.unlock [HW] Unlock (ignore) the keylock
1684 i8042.kbdreset [HW] Reset device connected to KBD port
1688 i8k.ignore_dmi [HW] Continue probing hardware even if DMI data
1689 indicates that the driver is running on unsupported
1691 i8k.force [HW] Activate i8k driver even if SMM BIOS signature
1692 does not match list of supported models.
1694 [HW] Report power status in /proc/i8k
1695 (disabled by default)
1696 i8k.restricted [HW] Allow controlling fans only if SYS_ADMIN
1699 i915.invert_brightness=
1700 [DRM] Invert the sense of the variable that is used to
1701 set the brightness of the panel backlight. Normally a
1702 brightness value of 0 indicates backlight switched off,
1703 and the maximum of the brightness value sets the backlight
1704 to maximum brightness. If this parameter is set to 0
1705 (default) and the machine requires it, or this parameter
1706 is set to 1, a brightness value of 0 sets the backlight
1707 to maximum brightness, and the maximum of the brightness
1708 value switches the backlight off.
1709 -1 -- never invert brightness
1710 0 -- machine default
1711 1 -- force brightness inversion
1714 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1716 ide-core.nodma= [HW] (E)IDE subsystem
1717 Format: =0.0 to prevent dma on hda, =0.1 hdb =1.0 hdc
1718 .vlb_clock .pci_clock .noflush .nohpa .noprobe .nowerr
1719 .cdrom .chs .ignore_cable are additional options
1720 See Documentation/ide/ide.rst.
1722 ide-generic.probe-mask= [HW] (E)IDE subsystem
1724 Probe mask for legacy ISA IDE ports. Depending on
1725 platform up to 6 ports are supported, enabled by
1726 setting corresponding bits in the mask to 1. The
1727 default value is 0x0, which has a special meaning.
1728 On systems that have PCI, it triggers scanning the
1729 PCI bus for the first and the second port, which
1730 are then probed. On systems without PCI the value
1731 of 0x0 enables probing the two first ports as if it
1734 ide-pci-generic.all-generic-ide [HW] (E)IDE subsystem
1735 Claim all unknown PCI IDE storage controllers.
1738 Format: idle=poll, idle=halt, idle=nomwait
1739 Poll forces a polling idle loop that can slightly
1740 improve the performance of waking up a idle CPU, but
1741 will use a lot of power and make the system run hot.
1743 idle=halt: Halt is forced to be used for CPU idle.
1744 In such case C2/C3 won't be used again.
1745 idle=nomwait: Disable mwait for CPU C-states
1749 Allow force disabling of Shared Virtual Memory (SVA)
1750 support for the idxd driver. By default it is set to
1753 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1754 Format: { strict | legacy | 2008 | relaxed }
1757 Choose which programs will be accepted for execution
1758 based on the IEEE 754 NaN encoding(s) supported by
1759 the FPU and the NaN encoding requested with the value
1760 of an ELF file header flag individually set by each
1761 binary. Hardware implementations are permitted to
1762 support either or both of the legacy and the 2008 NaN
1765 Available settings are as follows:
1766 strict accept binaries that request a NaN encoding
1767 supported by the FPU
1768 legacy only accept legacy-NaN binaries, if supported
1770 2008 only accept 2008-NaN binaries, if supported
1772 relaxed accept any binaries regardless of whether
1773 supported by the FPU
1775 The FPU emulator is always able to support both NaN
1776 encodings, so if no FPU hardware is present or it has
1777 been disabled with 'nofpu', then the settings of
1778 'legacy' and '2008' strap the emulator accordingly,
1779 'relaxed' straps the emulator for both legacy-NaN and
1780 2008-NaN, whereas 'strict' enables legacy-NaN only on
1781 legacy processors and both NaN encodings on MIPS32 or
1784 The setting for ABS.fmt/NEG.fmt instruction execution
1785 mode generally follows that for the NaN encoding,
1786 except where unsupported by hardware.
1788 ignore_loglevel [KNL]
1789 Ignore loglevel setting - this will print /all/
1790 kernel messages to the console. Useful for debugging.
1791 We also add it as printk module parameter, so users
1792 could change it dynamically, usually by
1793 /sys/module/printk/parameters/ignore_loglevel.
1796 Ignore RLIMIT_DATA setting for data mappings,
1797 print warning at first misuse. Can be changed via
1798 /sys/module/kernel/parameters/ignore_rlimit_data.
1800 ihash_entries= [KNL]
1801 Set number of hash buckets for inode cache.
1803 ima_appraise= [IMA] appraise integrity measurements
1804 Format: { "off" | "enforce" | "fix" | "log" }
1807 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
1808 The builtin appraise policy appraises all files
1811 ima_canonical_fmt [IMA]
1812 Use the canonical format for the binary runtime
1813 measurements, instead of host native format.
1816 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1820 The list of supported hash algorithms is defined
1821 in crypto/hash_info.h.
1824 The builtin policies to load during IMA setup.
1825 Format: "tcb | appraise_tcb | secure_boot |
1826 fail_securely | critical_data"
1828 The "tcb" policy measures all programs exec'd, files
1829 mmap'd for exec, and all files opened with the read
1830 mode bit set by either the effective uid (euid=0) or
1833 The "appraise_tcb" policy appraises the integrity of
1834 all files owned by root.
1836 The "secure_boot" policy appraises the integrity
1837 of files (eg. kexec kernel image, kernel modules,
1838 firmware, policy, etc) based on file signatures.
1840 The "fail_securely" policy forces file signature
1841 verification failure also on privileged mounted
1842 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
1845 The "critical_data" policy measures kernel integrity
1848 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
1849 Load a policy which meets the needs of the Trusted
1850 Computing Base. This means IMA will measure all
1851 programs exec'd, files mmap'd for exec, and all files
1852 opened for read by uid=0.
1855 Select one of defined IMA measurements template formats.
1856 Formats: { "ima" | "ima-ng" | "ima-sig" }
1860 [IMA] Define a custom template format.
1861 Format: { "field1|...|fieldN" }
1863 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
1864 Format: <min_file_size>
1865 Set the minimal file size for using asynchronous hash.
1866 If left unspecified, ahash usage is disabled.
1868 ahash performance varies for different data sizes on
1869 different crypto accelerators. This option can be used
1870 to achieve the best performance for a particular HW.
1872 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
1874 Set hashing buffer size. Default: 4k.
1876 ahash performance varies for different chunk sizes on
1877 different crypto accelerators. This option can be used
1878 to achieve best performance for particular HW.
1882 Run specified binary instead of /sbin/init as init
1885 initcall_debug [KNL] Trace initcalls as they are executed. Useful
1886 for working out where the kernel is dying during
1889 initcall_blacklist= [KNL] Do not execute a comma-separated list of
1890 initcall functions. Useful for debugging built-in
1891 modules and initcalls.
1893 initramfs_async= [KNL]
1896 This parameter controls whether the initramfs
1897 image is unpacked asynchronously, concurrently
1898 with devices being probed and
1899 initialized. This should normally just work,
1900 but as a debugging aid, one can get the
1901 historical behaviour of the initramfs
1902 unpacking being completed before device_ and
1905 initrd= [BOOT] Specify the location of the initial ramdisk
1907 initrdmem= [KNL] Specify a physical address and size from which to
1908 load the initrd. If an initrd is compiled in or
1909 specified in the bootparams, it takes priority over this
1911 Format: ss[KMG],nn[KMG]
1914 init_on_alloc= [MM] Fill newly allocated pages and heap objects with
1917 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
1919 init_on_free= [MM] Fill freed pages and heap objects with zeroes.
1921 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
1923 init_pkru= [X86] Specify the default memory protection keys rights
1924 register contents for all processes. 0x55555554 by
1925 default (disallow access to all but pkey 0). Can
1926 override in debugfs after boot.
1928 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
1931 int_pln_enable [X86] Enable power limit notification interrupt
1933 integrity_audit=[IMA]
1934 Format: { "0" | "1" }
1935 0 -- basic integrity auditing messages. (Default)
1936 1 -- additional integrity auditing messages.
1938 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
1940 Enable intel iommu driver.
1942 Disable intel iommu driver.
1943 igfx_off [Default Off]
1944 By default, gfx is mapped as normal device. If a gfx
1945 device has a dedicated DMAR unit, the DMAR unit is
1946 bypassed by not enabling DMAR with this option. In
1947 this case, gfx device will use physical address for
1949 strict [Default Off]
1950 With this option on every unmap_single operation will
1951 result in a hardware IOTLB flush operation as opposed
1952 to batching them for performance.
1953 sp_off [Default Off]
1954 By default, super page will be supported if Intel IOMMU
1955 has the capability. With this option, super page will
1958 By default, scalable mode will be disabled even if the
1959 hardware advertises that it has support for the scalable
1960 mode translation. With this option set, scalable mode
1961 will be used on hardware which claims to support it.
1962 tboot_noforce [Default Off]
1963 Do not force the Intel IOMMU enabled under tboot.
1964 By default, tboot will force Intel IOMMU on, which
1965 could harm performance of some high-throughput
1966 devices like 40GBit network cards, even if identity
1968 Note that using this option lowers the security
1969 provided by tboot because it makes the system
1970 vulnerable to DMA attacks.
1972 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
1973 0 disables intel_idle and fall back on acpi_idle.
1974 1 to 9 specify maximum depth of C-state.
1978 Do not enable intel_pstate as the default
1979 scaling driver for the supported processors
1981 Use intel_pstate as a scaling driver, but configure it
1982 to work with generic cpufreq governors (instead of
1983 enabling its internal governor). This mode cannot be
1984 used along with the hardware-managed P-states (HWP)
1987 Enable intel_pstate on systems that prohibit it by default
1988 in favor of acpi-cpufreq. Forcing the intel_pstate driver
1989 instead of acpi-cpufreq may disable platform features, such
1990 as thermal controls and power capping, that rely on ACPI
1991 P-States information being indicated to OSPM and therefore
1992 should be used with caution. This option does not work with
1993 processors that aren't supported by the intel_pstate driver
1994 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
1996 Do not enable hardware P state control (HWP)
1999 Only load intel_pstate on systems which support
2000 hardware P state control (HWP) if available.
2002 Enforce ACPI _PPC performance limits. If the Fixed ACPI
2003 Description Table, specifies preferred power management
2004 profile as "Enterprise Server" or "Performance Server",
2005 then this feature is turned on by default.
2007 Allow per-logical-CPU P-State performance control limits using
2008 cpufreq sysfs interface
2010 intremap= [X86-64, Intel-IOMMU]
2011 on enable Interrupt Remapping (default)
2012 off disable Interrupt Remapping
2013 nosid disable Source ID checking
2015 BIOS x2APIC opt-out request will be ignored
2016 nopost disable Interrupt Posting
2018 iomem= Disable strict checking of access to MMIO memory
2019 strict regions from userspace.
2034 nobypass [PPC/POWERNV]
2035 Disable IOMMU bypass, using IOMMU for PCI devices.
2037 iommu.forcedac= [ARM64, X86] Control IOVA allocation for PCI devices.
2038 Format: { "0" | "1" }
2039 0 - Try to allocate a 32-bit DMA address first, before
2040 falling back to the full range if needed.
2041 1 - Allocate directly from the full usable range,
2042 forcing Dual Address Cycle for PCI cards supporting
2043 greater than 32-bit addressing.
2045 iommu.strict= [ARM64, X86] Configure TLB invalidation behaviour
2046 Format: { "0" | "1" }
2048 Request that DMA unmap operations use deferred
2049 invalidation of hardware TLBs, for increased
2050 throughput at the cost of reduced device isolation.
2051 Will fall back to strict mode if not supported by
2052 the relevant IOMMU driver.
2053 1 - Strict mode (default).
2054 DMA unmap operations invalidate IOMMU hardware TLBs
2056 Note: on x86, the default behaviour depends on the
2057 equivalent driver-specific parameters, but a strict
2058 mode explicitly specified by either method takes
2062 [ARM64, X86] Configure DMA to bypass the IOMMU by default.
2063 Format: { "0" | "1" }
2064 0 - Use IOMMU translation for DMA.
2065 1 - Bypass the IOMMU for DMA.
2066 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
2068 io7= [HW] IO7 for Marvel-based Alpha systems
2069 See comment before marvel_specify_io7 in
2070 arch/alpha/kernel/core_marvel.c.
2072 io_delay= [X86] I/O delay method
2074 Standard port 0x80 based delay
2076 Alternate port 0xed based delay (needed on some systems)
2078 Simple two microseconds delay
2083 See Documentation/admin-guide/nfs/nfsroot.rst.
2085 ipcmni_extend [KNL] Extend the maximum number of unique System V
2086 IPC identifiers from 32,768 to 16,777,216.
2088 irqaffinity= [SMP] Set the default irq affinity mask
2089 The argument is a cpu list, as described above.
2091 irqchip.gicv2_force_probe=
2094 Force the kernel to look for the second 4kB page
2095 of a GICv2 controller even if the memory range
2096 exposed by the device tree is too small.
2098 irqchip.gicv3_nolpi=
2100 Force the kernel to ignore the availability of
2101 LPIs (and by consequence ITSs). Intended for system
2102 that use the kernel as a bootloader, and thus want
2103 to let secondary kernels in charge of setting up
2106 irqchip.gicv3_pseudo_nmi= [ARM64]
2107 Enables support for pseudo-NMIs in the kernel. This
2108 requires the kernel to be built with
2109 CONFIG_ARM64_PSEUDO_NMI.
2112 When an interrupt is not handled search all handlers
2113 for it. Intended to get systems with badly broken
2117 When an interrupt is not handled search all handlers
2118 for it. Also check all handlers each timer
2119 interrupt. Intended to get systems with badly broken
2123 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
2125 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
2126 [Deprecated - use cpusets instead]
2127 Format: [flag-list,]<cpu-list>
2129 Specify one or more CPUs to isolate from disturbances
2130 specified in the flag list (default: domain):
2133 Disable the tick when a single task runs.
2135 A residual 1Hz tick is offloaded to workqueues, which you
2136 need to affine to housekeeping through the global
2137 workqueue's affinity configured via the
2138 /sys/devices/virtual/workqueue/cpumask sysfs file, or
2139 by using the 'domain' flag described below.
2141 NOTE: by default the global workqueue runs on all CPUs,
2142 so to protect individual CPUs the 'cpumask' file has to
2143 be configured manually after bootup.
2146 Isolate from the general SMP balancing and scheduling
2147 algorithms. Note that performing domain isolation this way
2148 is irreversible: it's not possible to bring back a CPU to
2149 the domains once isolated through isolcpus. It's strongly
2150 advised to use cpusets instead to disable scheduler load
2151 balancing through the "cpuset.sched_load_balance" file.
2152 It offers a much more flexible interface where CPUs can
2153 move in and out of an isolated set anytime.
2155 You can move a process onto or off an "isolated" CPU via
2156 the CPU affinity syscalls or cpuset.
2157 <cpu number> begins at 0 and the maximum value is
2158 "number of CPUs in system - 1".
2162 Isolate from being targeted by managed interrupts
2163 which have an interrupt mask containing isolated
2164 CPUs. The affinity of managed interrupts is
2165 handled by the kernel and cannot be changed via
2166 the /proc/irq/* interfaces.
2168 This isolation is best effort and only effective
2169 if the automatically assigned interrupt mask of a
2170 device queue contains isolated and housekeeping
2171 CPUs. If housekeeping CPUs are online then such
2172 interrupts are directed to the housekeeping CPU
2173 so that IO submitted on the housekeeping CPU
2174 cannot disturb the isolated CPU.
2176 If a queue's affinity mask contains only isolated
2177 CPUs then this parameter has no effect on the
2178 interrupt routing decision, though interrupts are
2179 only delivered when tasks running on those
2180 isolated CPUs submit IO. IO submitted on
2181 housekeeping CPUs has no influence on those
2184 The format of <cpu-list> is described above.
2188 ivrs_ioapic [HW,X86-64]
2189 Provide an override to the IOAPIC-ID<->DEVICE-ID
2190 mapping provided in the IVRS ACPI table. For
2191 example, to map IOAPIC-ID decimal 10 to
2192 PCI device 00:14.0 write the parameter as:
2193 ivrs_ioapic[10]=00:14.0
2195 ivrs_hpet [HW,X86-64]
2196 Provide an override to the HPET-ID<->DEVICE-ID
2197 mapping provided in the IVRS ACPI table. For
2198 example, to map HPET-ID decimal 0 to
2199 PCI device 00:14.0 write the parameter as:
2200 ivrs_hpet[0]=00:14.0
2202 ivrs_acpihid [HW,X86-64]
2203 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2204 mapping provided in the IVRS ACPI table. For
2205 example, to map UART-HID:UID AMD0020:0 to
2206 PCI device 00:14.5 write the parameter as:
2207 ivrs_acpihid[00:14.5]=AMD0020:0
2209 js= [HW,JOY] Analog joystick
2210 See Documentation/input/joydev/joystick.rst.
2213 When CONFIG_RANDOMIZE_BASE is set, this disables
2214 kernel and module base offset ASLR (Address Space
2215 Layout Randomization).
2218 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2219 report on every invalid memory access. Without this
2220 parameter KASAN will print report only for the first
2225 kernelcore= [KNL,X86,IA-64,PPC]
2226 Format: nn[KMGTPE] | nn% | "mirror"
2227 This parameter specifies the amount of memory usable by
2228 the kernel for non-movable allocations. The requested
2229 amount is spread evenly throughout all nodes in the
2230 system as ZONE_NORMAL. The remaining memory is used for
2231 movable memory in its own zone, ZONE_MOVABLE. In the
2232 event, a node is too small to have both ZONE_NORMAL and
2233 ZONE_MOVABLE, kernelcore memory will take priority and
2234 other nodes will have a larger ZONE_MOVABLE.
2236 ZONE_MOVABLE is used for the allocation of pages that
2237 may be reclaimed or moved by the page migration
2238 subsystem. Note that allocations like PTEs-from-HighMem
2239 still use the HighMem zone if it exists, and the Normal
2240 zone if it does not.
2242 It is possible to specify the exact amount of memory in
2243 the form of "nn[KMGTPE]", a percentage of total system
2244 memory in the form of "nn%", or "mirror". If "mirror"
2245 option is specified, mirrored (reliable) memory is used
2246 for non-movable allocations and remaining memory is used
2247 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2248 are exclusive, so you cannot specify multiple forms.
2250 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
2251 Format: <Controller#>[,poll interval]
2252 The controller # is the number of the ehci usb debug
2253 port as it is probed via PCI. The poll interval is
2254 optional and is the number seconds in between
2255 each poll cycle to the debug port in case you need
2256 the functionality for interrupting the kernel with
2257 gdb or control-c on the dbgp connection. When
2258 not using this parameter you use sysrq-g to break into
2259 the kernel debugger.
2261 kgdboc= [KGDB,HW] kgdb over consoles.
2262 Requires a tty driver that supports console polling,
2263 or a supported polling keyboard driver (non-usb).
2264 Serial only format: <serial_device>[,baud]
2265 keyboard only format: kbd
2266 keyboard and serial format: kbd,<serial_device>[,baud]
2267 Optional Kernel mode setting:
2268 kms, kbd format: kms,kbd
2269 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2271 kgdboc_earlycon= [KGDB,HW]
2272 If the boot console provides the ability to read
2273 characters and can work in polling mode, you can use
2274 this parameter to tell kgdb to use it as a backend
2275 until the normal console is registered. Intended to
2276 be used together with the kgdboc parameter which
2277 specifies the normal console to transition to.
2279 The name of the early console should be specified
2280 as the value of this parameter. Note that the name of
2281 the early console might be different than the tty
2282 name passed to kgdboc. It's OK to leave the value
2283 blank and the first boot console that implements
2284 read() will be picked.
2286 kgdbwait [KGDB] Stop kernel execution and enter the
2287 kernel debugger at the earliest opportunity.
2289 kmac= [MIPS] Korina ethernet MAC address.
2290 Configure the RouterBoard 532 series on-chip
2291 Ethernet adapter MAC address.
2293 kmemleak= [KNL] Boot-time kmemleak enable/disable
2294 Valid arguments: on, off
2296 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2299 kprobe_event=[probe-list]
2300 [FTRACE] Add kprobe events and enable at boot time.
2301 The probe-list is a semicolon delimited list of probe
2302 definitions. Each definition is same as kprobe_events
2303 interface, but the parameters are comma delimited.
2304 For example, to add a kprobe event on vfs_read with
2305 arg1 and arg2, add to the command line;
2307 kprobe_event=p,vfs_read,$arg1,$arg2
2309 See also Documentation/trace/kprobetrace.rst "Kernel
2310 Boot Parameter" section.
2312 kpti= [ARM64] Control page table isolation of user
2313 and kernel address spaces.
2314 Default: enabled on cores which need mitigation.
2318 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2319 Default is 0 (don't ignore, but inject #GP)
2321 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2322 Default is false (don't support).
2324 kvm.mmu_audit= [KVM] This is a R/W parameter which allows audit
2329 [KVM] Controls the software workaround for the
2330 X86_BUG_ITLB_MULTIHIT bug.
2331 force : Always deploy workaround.
2332 off : Never deploy workaround.
2333 auto : Deploy workaround based on the presence of
2334 X86_BUG_ITLB_MULTIHIT.
2338 If the software workaround is enabled for the host,
2339 guests do need not to enable it for nested guests.
2341 kvm.nx_huge_pages_recovery_ratio=
2342 [KVM] Controls how many 4KiB pages are periodically zapped
2343 back to huge pages. 0 disables the recovery, otherwise if
2344 the value is N KVM will zap 1/Nth of the 4KiB pages every
2345 minute. The default is 60.
2347 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM.
2348 Default is 1 (enabled)
2350 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU)
2352 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode.
2355 [KVM,ARM] Select one of KVM/arm64's modes of operation.
2357 nvhe: Standard nVHE-based mode, without support for
2360 protected: nVHE-based mode with support for guests whose
2361 state is kept private from the host.
2362 Not valid if the kernel is running in EL2.
2364 Defaults to VHE/nVHE based on hardware support.
2366 kvm-arm.vgic_v3_group0_trap=
2367 [KVM,ARM] Trap guest accesses to GICv3 group-0
2370 kvm-arm.vgic_v3_group1_trap=
2371 [KVM,ARM] Trap guest accesses to GICv3 group-1
2374 kvm-arm.vgic_v3_common_trap=
2375 [KVM,ARM] Trap guest accesses to GICv3 common
2378 kvm-arm.vgic_v4_enable=
2379 [KVM,ARM] Allow use of GICv4 for direct injection of
2382 kvm_cma_resv_ratio=n [PPC]
2383 Reserves given percentage from system memory area for
2384 contiguous memory allocation for KVM hash pagetable
2386 By default it reserves 5% of total system memory.
2390 kvm-intel.ept= [KVM,Intel] Disable extended page tables
2391 (virtualized MMU) support on capable Intel chips.
2392 Default is 1 (enabled)
2394 kvm-intel.emulate_invalid_guest_state=
2395 [KVM,Intel] Enable emulation of invalid guest states
2396 Default is 0 (disabled)
2398 kvm-intel.flexpriority=
2399 [KVM,Intel] Disable FlexPriority feature (TPR shadow).
2400 Default is 1 (enabled)
2403 [KVM,Intel] Enable VMX nesting (nVMX).
2404 Default is 0 (disabled)
2406 kvm-intel.unrestricted_guest=
2407 [KVM,Intel] Disable unrestricted guest feature
2408 (virtualized real and unpaged mode) on capable
2409 Intel chips. Default is 1 (enabled)
2411 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2414 Valid arguments: never, cond, always
2416 always: L1D cache flush on every VMENTER.
2417 cond: Flush L1D on VMENTER only when the code between
2418 VMEXIT and VMENTER can leak host memory.
2419 never: Disables the mitigation
2421 Default is cond (do L1 cache flush in specific instances)
2423 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification
2424 feature (tagged TLBs) on capable Intel chips.
2425 Default is 1 (enabled)
2427 l1tf= [X86] Control mitigation of the L1TF vulnerability on
2430 The kernel PTE inversion protection is unconditionally
2431 enabled and cannot be disabled.
2434 Provides all available mitigations for the
2435 L1TF vulnerability. Disables SMT and
2436 enables all mitigations in the
2437 hypervisors, i.e. unconditional L1D flush.
2439 SMT control and L1D flush control via the
2440 sysfs interface is still possible after
2441 boot. Hypervisors will issue a warning
2442 when the first VM is started in a
2443 potentially insecure configuration,
2444 i.e. SMT enabled or L1D flush disabled.
2447 Same as 'full', but disables SMT and L1D
2448 flush runtime control. Implies the
2449 'nosmt=force' command line option.
2450 (i.e. sysfs control of SMT is disabled.)
2453 Leaves SMT enabled and enables the default
2454 hypervisor mitigation, i.e. conditional
2457 SMT control and L1D flush control via the
2458 sysfs interface is still possible after
2459 boot. Hypervisors will issue a warning
2460 when the first VM is started in a
2461 potentially insecure configuration,
2462 i.e. SMT enabled or L1D flush disabled.
2466 Disables SMT and enables the default
2467 hypervisor mitigation.
2469 SMT control and L1D flush control via the
2470 sysfs interface is still possible after
2471 boot. Hypervisors will issue a warning
2472 when the first VM is started in a
2473 potentially insecure configuration,
2474 i.e. SMT enabled or L1D flush disabled.
2477 Same as 'flush', but hypervisors will not
2478 warn when a VM is started in a potentially
2479 insecure configuration.
2482 Disables hypervisor mitigations and doesn't
2484 It also drops the swap size and available
2485 RAM limit restriction on both hypervisor and
2490 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2496 lapic [X86-32,APIC] Enable the local APIC even if BIOS
2499 lapic= [X86,APIC] Do not use TSC deadline
2500 value for LAPIC timer one-shot implementation. Default
2501 back to the programmable timer unit in the LAPIC.
2502 Format: notscdeadline
2504 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
2507 libata.dma= [LIBATA] DMA control
2508 libata.dma=0 Disable all PATA and SATA DMA
2509 libata.dma=1 PATA and SATA Disk DMA only
2510 libata.dma=2 ATAPI (CDROM) DMA only
2511 libata.dma=4 Compact Flash DMA only
2512 Combinations also work, so libata.dma=3 enables DMA
2513 for disks and CDROMs, but not CFs.
2515 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2516 libata.ignore_hpa=0 keep BIOS limits (default)
2517 libata.ignore_hpa=1 ignore limits, using full disk
2519 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2523 libata.force= [LIBATA] Force configurations. The format is comma-
2524 separated list of "[ID:]VAL" where ID is
2525 PORT[.DEVICE]. PORT and DEVICE are decimal numbers
2526 matching port, link or device. Basically, it matches
2527 the ATA ID string printed on console by libata. If
2528 the whole ID part is omitted, the last PORT and DEVICE
2529 values are used. If ID hasn't been specified yet, the
2530 configuration applies to all ports, links and devices.
2532 If only DEVICE is omitted, the parameter applies to
2533 the port and all links and devices behind it. DEVICE
2534 number of 0 either selects the first device or the
2535 first fan-out link behind PMP device. It does not
2536 select the host link. DEVICE number of 15 selects the
2537 host link and device attached to it.
2539 The VAL specifies the configuration to force. As long
2540 as there's no ambiguity shortcut notation is allowed.
2541 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2542 The following configurations can be forced.
2544 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2545 Any ID with matching PORT is used.
2547 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2549 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2550 udma[/][16,25,33,44,66,100,133] notation is also
2553 * [no]ncq: Turn on or off NCQ.
2555 * [no]ncqtrim: Turn off queued DSM TRIM.
2557 * nohrst, nosrst, norst: suppress hard, soft
2560 * rstonce: only attempt one reset during
2561 hot-unplug link recovery
2563 * dump_id: dump IDENTIFY data.
2565 * atapi_dmadir: Enable ATAPI DMADIR bridge support
2567 * disable: Disable this device.
2569 If there are multiple matching configurations changing
2570 the same attribute, the last one is used.
2572 memblock=debug [KNL] Enable memblock debug messages.
2574 load_ramdisk= [RAM] [Deprecated]
2576 lockd.nlm_grace_period=P [NFS] Assign grace period.
2579 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2582 lockd.nlm_timeout=T [NFS] Assign timeout value.
2585 lockd.nlm_udpport=M [NFS] Assign UDP port.
2588 lockdown= [SECURITY]
2589 { integrity | confidentiality }
2590 Enable the kernel lockdown feature. If set to
2591 integrity, kernel features that allow userland to
2592 modify the running kernel are disabled. If set to
2593 confidentiality, kernel features that allow userland
2594 to extract confidential information from the kernel
2597 locktorture.nreaders_stress= [KNL]
2598 Set the number of locking read-acquisition kthreads.
2599 Defaults to being automatically set based on the
2600 number of online CPUs.
2602 locktorture.nwriters_stress= [KNL]
2603 Set the number of locking write-acquisition kthreads.
2605 locktorture.onoff_holdoff= [KNL]
2606 Set time (s) after boot for CPU-hotplug testing.
2608 locktorture.onoff_interval= [KNL]
2609 Set time (s) between CPU-hotplug operations, or
2610 zero to disable CPU-hotplug testing.
2612 locktorture.shuffle_interval= [KNL]
2613 Set task-shuffle interval (jiffies). Shuffling
2614 tasks allows some CPUs to go into dyntick-idle
2615 mode during the locktorture test.
2617 locktorture.shutdown_secs= [KNL]
2618 Set time (s) after boot system shutdown. This
2619 is useful for hands-off automated testing.
2621 locktorture.stat_interval= [KNL]
2622 Time (s) between statistics printk()s.
2624 locktorture.stutter= [KNL]
2625 Time (s) to stutter testing, for example,
2626 specifying five seconds causes the test to run for
2627 five seconds, wait for five seconds, and so on.
2628 This tests the locking primitive's ability to
2629 transition abruptly to and from idle.
2631 locktorture.torture_type= [KNL]
2632 Specify the locking implementation to test.
2634 locktorture.verbose= [KNL]
2635 Enable additional printk() statements.
2637 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
2640 loglevel= All Kernel Messages with a loglevel smaller than the
2641 console loglevel will be printed to the console. It can
2642 also be changed with klogd or other programs. The
2643 loglevels are defined as follows:
2645 0 (KERN_EMERG) system is unusable
2646 1 (KERN_ALERT) action must be taken immediately
2647 2 (KERN_CRIT) critical conditions
2648 3 (KERN_ERR) error conditions
2649 4 (KERN_WARNING) warning conditions
2650 5 (KERN_NOTICE) normal but significant condition
2651 6 (KERN_INFO) informational
2652 7 (KERN_DEBUG) debug-level messages
2654 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
2655 in bytes. n must be a power of two and greater
2656 than the minimal size. The minimal size is defined
2657 by LOG_BUF_SHIFT kernel config parameter. There is
2658 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2659 that allows to increase the default size depending on
2660 the number of CPUs. See init/Kconfig for more details.
2662 logo.nologo [FB] Disables display of the built-in Linux logo.
2663 This may be used to provide more screen space for
2664 kernel log messages and is useful when debugging
2665 kernel boot problems.
2667 lp=0 [LP] Specify parallel ports to use, e.g,
2668 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2669 lp=reset first parallel port). 'lp=0' disables the
2670 lp=auto printer driver. 'lp=reset' (which can be
2671 specified in addition to the ports) causes
2672 attached printers to be reset. Using
2673 lp=port1,port2,... specifies the parallel ports
2674 to associate lp devices with, starting with
2675 lp0. A port specification may be 'none' to skip
2676 that lp device, or a parport name such as
2677 'parport0'. Specifying 'lp=auto' instead of a
2678 port specification list means that device IDs
2679 from each port should be examined, to see if
2680 an IEEE 1284-compliant printer is attached; if
2681 so, the driver will manage that printer.
2682 See also header of drivers/char/lp.c.
2685 Sets loops_per_jiffy to given constant, thus avoiding
2686 time-consuming boot-time autodetection (up to 250 ms per
2687 CPU). 0 enables autodetection (default). To determine
2688 the correct value for your kernel, boot with normal
2689 autodetection and see what value is printed. Note that
2690 on SMP systems the preset will be applied to all CPUs,
2691 which is likely to cause problems if your CPUs need
2692 significantly divergent settings. An incorrect value
2693 will cause delays in the kernel to be wrong, leading to
2694 unpredictable I/O errors and other breakage. Although
2695 unlikely, in the extreme case this might damage your
2699 Format: <io>,<irq>,<dma>
2701 lsm.debug [SECURITY] Enable LSM initialization debugging output.
2704 [SECURITY] Choose order of LSM initialization. This
2705 overrides CONFIG_LSM, and the "security=" parameter.
2707 machvec= [IA-64] Force the use of a particular machine-vector
2708 (machvec) in a generic kernel.
2709 Example: machvec=hpzx1
2711 machtype= [Loongson] Share the same kernel image file between
2712 different yeeloong laptops.
2713 Example: machtype=lemote-yeeloong-2f-7inch
2715 max_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory greater
2716 than or equal to this physical address is ignored.
2718 maxcpus= [SMP] Maximum number of processors that an SMP kernel
2719 will bring up during bootup. maxcpus=n : n >= 0 limits
2720 the kernel to bring up 'n' processors. Surely after
2721 bootup you can bring up the other plugged cpu by executing
2722 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
2723 only takes effect during system bootup.
2724 While n=0 is a special case, it is equivalent to "nosmp",
2725 which also disables the IO APIC.
2727 max_loop= [LOOP] The number of loop block devices that get
2728 (loop.max_loop) unconditionally pre-created at init time. The default
2729 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
2730 of statically allocating a predefined number, loop
2731 devices can be requested on-demand with the
2732 /dev/loop-control interface.
2734 mce [X86-32] Machine Check Exception
2736 mce=option [X86-64] See Documentation/x86/x86_64/boot-options.rst
2738 md= [HW] RAID subsystems devices and level
2739 See Documentation/admin-guide/md.rst.
2742 Format: <first>,<last>
2743 Specifies range of consoles to be captured by the MDA.
2746 Control mitigation for the Micro-architectural Data
2747 Sampling (MDS) vulnerability.
2749 Certain CPUs are vulnerable to an exploit against CPU
2750 internal buffers which can forward information to a
2751 disclosure gadget under certain conditions.
2753 In vulnerable processors, the speculatively
2754 forwarded data can be used in a cache side channel
2755 attack, to access data to which the attacker does
2756 not have direct access.
2758 This parameter controls the MDS mitigation. The
2761 full - Enable MDS mitigation on vulnerable CPUs
2762 full,nosmt - Enable MDS mitigation and disable
2763 SMT on vulnerable CPUs
2764 off - Unconditionally disable MDS mitigation
2766 On TAA-affected machines, mds=off can be prevented by
2767 an active TAA mitigation as both vulnerabilities are
2768 mitigated with the same mechanism so in order to disable
2769 this mitigation, you need to specify tsx_async_abort=off
2772 Not specifying this option is equivalent to
2775 For details see: Documentation/admin-guide/hw-vuln/mds.rst
2777 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
2778 Amount of memory to be used in cases as follows:
2781 2 when the kernel is not able to see the whole system memory;
2782 3 memory that lies after 'mem=' boundary is excluded from
2783 the hypervisor, then assigned to KVM guests.
2785 [X86] Work as limiting max address. Use together
2786 with memmap= to avoid physical address space collisions.
2787 Without memmap= PCI devices could be placed at addresses
2788 belonging to unused RAM.
2790 Note that this only takes effects during boot time since
2791 in above case 3, memory may need be hot added after boot
2792 if system memory of hypervisor is not sufficient.
2794 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
2798 [KNL,SH] Allow user to override the default size for
2799 per-device physically contiguous DMA buffers.
2801 memhp_default_state=online/offline
2802 [KNL] Set the initial state for the memory hotplug
2803 onlining policy. If not specified, the default value is
2804 set according to the
2805 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
2807 See Documentation/admin-guide/mm/memory-hotplug.rst.
2809 memmap=exactmap [KNL,X86] Enable setting of an exact
2810 E820 memory map, as specified by the user.
2811 Such memmap=exactmap lines can be constructed based on
2812 BIOS output or other requirements. See the memmap=nn@ss
2815 memmap=nn[KMG]@ss[KMG]
2816 [KNL, X86, MIPS, XTENSA] Force usage of a specific region of memory.
2817 Region of memory to be used is from ss to ss+nn.
2818 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
2819 which limits max address to nn[KMG].
2820 Multiple different regions can be specified,
2823 memmap=100M@2G,100M#3G,1G!1024G
2825 memmap=nn[KMG]#ss[KMG]
2826 [KNL,ACPI] Mark specific memory as ACPI data.
2827 Region of memory to be marked is from ss to ss+nn.
2829 memmap=nn[KMG]$ss[KMG]
2830 [KNL,ACPI] Mark specific memory as reserved.
2831 Region of memory to be reserved is from ss to ss+nn.
2832 Example: Exclude memory from 0x18690000-0x1869ffff
2833 memmap=64K$0x18690000
2835 memmap=0x10000$0x18690000
2836 Some bootloaders may need an escape character before '$',
2837 like Grub2, otherwise '$' and the following number
2840 memmap=nn[KMG]!ss[KMG]
2841 [KNL,X86] Mark specific memory as protected.
2842 Region of memory to be used, from ss to ss+nn.
2843 The memory region may be marked as e820 type 12 (0xc)
2844 and is NVDIMM or ADR memory.
2846 memmap=<size>%<offset>-<oldtype>+<newtype>
2847 [KNL,ACPI] Convert memory within the specified region
2848 from <oldtype> to <newtype>. If "-<oldtype>" is left
2849 out, the whole region will be marked as <newtype>,
2850 even if previously unavailable. If "+<newtype>" is left
2851 out, matching memory will be removed. Types are
2852 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
2853 3 = ACPI, 12 = PRAM.
2855 memory_corruption_check=0/1 [X86]
2856 Some BIOSes seem to corrupt the first 64k of
2857 memory when doing things like suspend/resume.
2858 Setting this option will scan the memory
2859 looking for corruption. Enabling this will
2860 both detect corruption and prevent the kernel
2861 from using the memory being corrupted.
2862 However, its intended as a diagnostic tool; if
2863 repeatable BIOS-originated corruption always
2864 affects the same memory, you can use memmap=
2865 to prevent the kernel from using that memory.
2867 memory_corruption_check_size=size [X86]
2868 By default it checks for corruption in the low
2869 64k, making this memory unavailable for normal
2870 use. Use this parameter to scan for
2871 corruption in more or less memory.
2873 memory_corruption_check_period=seconds [X86]
2874 By default it checks for corruption every 60
2875 seconds. Use this parameter to check at some
2876 other rate. 0 disables periodic checking.
2878 memory_hotplug.memmap_on_memory
2879 [KNL,X86,ARM] Boolean flag to enable this feature.
2880 Format: {on | off (default)}
2881 When enabled, runtime hotplugged memory will
2882 allocate its internal metadata (struct pages)
2883 from the hotadded memory which will allow to
2884 hotadd a lot of memory without requiring
2885 additional memory to do so.
2886 This feature is disabled by default because it
2887 has some implication on large (e.g. GB)
2888 allocations in some configurations (e.g. small
2890 The state of the flag can be read in
2891 /sys/module/memory_hotplug/parameters/memmap_on_memory.
2892 Note that even when enabled, there are a few cases where
2893 the feature is not effective.
2895 This is not compatible with hugetlb_free_vmemmap. If
2896 both parameters are enabled, hugetlb_free_vmemmap takes
2897 precedence over memory_hotplug.memmap_on_memory.
2899 memtest= [KNL,X86,ARM,PPC,RISCV] Enable memtest
2901 default : 0 <disable>
2902 Specifies the number of memtest passes to be
2903 performed. Each pass selects another test
2904 pattern from a given set of patterns. Memtest
2905 fills the memory with this pattern, validates
2906 memory contents and reserves bad memory
2907 regions that are detected.
2909 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
2910 Valid arguments: on, off
2911 Default (depends on kernel configuration option):
2912 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
2913 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
2914 mem_encrypt=on: Activate SME
2915 mem_encrypt=off: Do not activate SME
2917 Refer to Documentation/virt/kvm/amd-memory-encryption.rst
2918 for details on when memory encryption can be activated.
2920 mem_sleep_default= [SUSPEND] Default system suspend mode:
2921 s2idle - Suspend-To-Idle
2922 shallow - Power-On Suspend or equivalent (if supported)
2923 deep - Suspend-To-RAM or equivalent (if supported)
2924 See Documentation/admin-guide/pm/sleep-states.rst.
2926 meye.*= [HW] Set MotionEye Camera parameters
2927 See Documentation/admin-guide/media/meye.rst.
2929 mfgpt_irq= [IA-32] Specify the IRQ to use for the
2930 Multi-Function General Purpose Timers on AMD Geode
2933 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
2934 the BIOS has incorrectly applied a workaround. TinyBIOS
2935 version 0.98 is known to be affected, 0.99 fixes the
2936 problem by letting the user disable the workaround.
2940 min_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory below this
2941 physical address is ignored.
2943 mini2440= [ARM,HW,KNL]
2944 Format:[0..2][b][c][t]
2946 MINI2440 configuration specification:
2947 0 - The attached screen is the 3.5" TFT
2948 1 - The attached screen is the 7" TFT
2949 2 - The VGA Shield is attached (1024x768)
2950 Leaving out the screen size parameter will not load
2951 the TFT driver, and the framebuffer will be left
2953 b - Enable backlight. The TFT backlight pin will be
2954 linked to the kernel VESA blanking code and a GPIO
2955 LED. This parameter is not necessary when using the
2957 c - Enable the s3c camera interface.
2958 t - Reserved for enabling touchscreen support. The
2959 touchscreen support is not enabled in the mainstream
2960 kernel as of 2.6.30, a preliminary port can be found
2961 in the "bleeding edge" mini2440 support kernel at
2962 https://repo.or.cz/w/linux-2.6/mini2440.git
2965 [X86,PPC,S390,ARM64] Control optional mitigations for
2966 CPU vulnerabilities. This is a set of curated,
2967 arch-independent options, each of which is an
2968 aggregation of existing arch-specific options.
2971 Disable all optional CPU mitigations. This
2972 improves system performance, but it may also
2973 expose users to several CPU vulnerabilities.
2974 Equivalent to: nopti [X86,PPC]
2976 nospectre_v1 [X86,PPC]
2978 nospectre_v2 [X86,PPC,S390,ARM64]
2979 spectre_v2_user=off [X86]
2980 spec_store_bypass_disable=off [X86,PPC]
2981 ssbd=force-off [ARM64]
2984 tsx_async_abort=off [X86]
2985 kvm.nx_huge_pages=off [X86]
2986 no_entry_flush [PPC]
2987 no_uaccess_flush [PPC]
2990 This does not have any effect on
2991 kvm.nx_huge_pages when
2992 kvm.nx_huge_pages=force.
2995 Mitigate all CPU vulnerabilities, but leave SMT
2996 enabled, even if it's vulnerable. This is for
2997 users who don't want to be surprised by SMT
2998 getting disabled across kernel upgrades, or who
2999 have other ways of avoiding SMT-based attacks.
3000 Equivalent to: (default behavior)
3003 Mitigate all CPU vulnerabilities, disabling SMT
3004 if needed. This is for users who always want to
3005 be fully mitigated, even if it means losing SMT.
3006 Equivalent to: l1tf=flush,nosmt [X86]
3007 mds=full,nosmt [X86]
3008 tsx_async_abort=full,nosmt [X86]
3011 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
3012 parameter allows control of the logging verbosity for
3013 the additional memory initialisation checks. A value
3014 of 0 disables mminit logging and a level of 4 will
3015 log everything. Information is printed at KERN_DEBUG
3016 so loglevel=8 may also need to be specified.
3019 [KNL] When CONFIG_MODULE_SIG is set, this means that
3020 modules without (valid) signatures will fail to load.
3021 Note that if CONFIG_MODULE_SIG_FORCE is set, that
3022 is always true, so this option does nothing.
3024 module_blacklist= [KNL] Do not load a comma-separated list of
3025 modules. Useful for debugging problem modules.
3028 [MOUSE] Maximum time between finger touching and
3029 leaving touchpad surface for touch to be considered
3030 a tap and be reported as a left button click (for
3031 touchpads working in absolute mode only).
3033 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
3034 reporting absolute coordinates, such as tablets
3035 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
3036 reporting absolute coordinates, such as tablets
3038 movablecore= [KNL,X86,IA-64,PPC]
3039 Format: nn[KMGTPE] | nn%
3040 This parameter is the complement to kernelcore=, it
3041 specifies the amount of memory used for migratable
3042 allocations. If both kernelcore and movablecore is
3043 specified, then kernelcore will be at *least* the
3044 specified value but may be more. If movablecore on its
3045 own is specified, the administrator must be careful
3046 that the amount of memory usable for all allocations
3049 movable_node [KNL] Boot-time switch to make hotplugable memory
3050 NUMA nodes to be movable. This means that the memory
3051 of such nodes will be usable only for movable
3052 allocations which rules out almost all kernel
3053 allocations. Use with caution!
3055 MTD_Partition= [MTD]
3056 Format: <name>,<region-number>,<size>,<offset>
3058 MTD_Region= [MTD] Format:
3059 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
3062 See drivers/mtd/parsers/cmdlinepart.c
3064 multitce=off [PPC] This parameter disables the use of the pSeries
3065 firmware feature for updating multiple TCE entries
3068 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
3070 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
3072 boundary - index of last SLC block on Flex-OneNAND.
3073 The remaining blocks are configured as MLC blocks.
3074 lock - Configure if Flex-OneNAND boundary should be locked.
3075 Once locked, the boundary cannot be changed.
3076 1 indicates lock status, 0 indicates unlock status.
3079 ARM/S3C2412 JIVE boot control
3081 See arch/arm/mach-s3c/mach-jive.c
3083 mtouchusb.raw_coordinates=
3084 [HW] Make the MicroTouch USB driver use raw coordinates
3085 ('y', default) or cooked coordinates ('n')
3087 mtrr_chunk_size=nn[KMG] [X86]
3088 used for mtrr cleanup. It is largest continuous chunk
3089 that could hold holes aka. UC entries.
3091 mtrr_gran_size=nn[KMG] [X86]
3092 Used for mtrr cleanup. It is granularity of mtrr block.
3094 Large value could prevent small alignment from
3097 mtrr_spare_reg_nr=n [X86]
3099 Range: 0,7 : spare reg number
3101 Used for mtrr cleanup. It is spare mtrr entries number.
3102 Set to 2 or more if your graphical card needs more.
3104 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
3106 netdev= [NET] Network devices parameters
3107 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
3108 Note that mem_start is often overloaded to mean
3109 something different and driver-specific.
3110 This usage is only documented in each driver source
3114 [NETFILTER] Enable connection tracking flow accounting
3115 0 to disable accounting
3116 1 to enable accounting
3119 nfsaddrs= [NFS] Deprecated. Use ip= instead.
3120 See Documentation/admin-guide/nfs/nfsroot.rst.
3122 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
3123 See Documentation/admin-guide/nfs/nfsroot.rst.
3125 nfsrootdebug [NFS] enable nfsroot debugging messages.
3126 See Documentation/admin-guide/nfs/nfsroot.rst.
3128 nfs.callback_nr_threads=
3129 [NFSv4] set the total number of threads that the
3130 NFS client will assign to service NFSv4 callback
3133 nfs.callback_tcpport=
3134 [NFS] set the TCP port on which the NFSv4 callback
3135 channel should listen.
3138 [NFS] sets the pathname to the program which is used
3139 to update the NFS client cache entries.
3141 nfs.cache_getent_timeout=
3142 [NFS] sets the timeout after which an attempt to
3143 update a cache entry is deemed to have failed.
3145 nfs.idmap_cache_timeout=
3146 [NFS] set the maximum lifetime for idmapper cache
3150 [NFS] enable 64-bit inode numbers.
3151 If zero, the NFS client will fake up a 32-bit inode
3152 number for the readdir() and stat() syscalls instead
3153 of returning the full 64-bit number.
3154 The default is to return 64-bit inode numbers.
3156 nfs.max_session_cb_slots=
3157 [NFSv4.1] Sets the maximum number of session
3158 slots the client will assign to the callback
3159 channel. This determines the maximum number of
3160 callbacks the client will process in parallel for
3161 a particular server.
3163 nfs.max_session_slots=
3164 [NFSv4.1] Sets the maximum number of session slots
3165 the client will attempt to negotiate with the server.
3166 This limits the number of simultaneous RPC requests
3167 that the client can send to the NFSv4.1 server.
3168 Note that there is little point in setting this
3169 value higher than the max_tcp_slot_table_limit.
3171 nfs.nfs4_disable_idmapping=
3172 [NFSv4] When set to the default of '1', this option
3173 ensures that both the RPC level authentication
3174 scheme and the NFS level operations agree to use
3175 numeric uids/gids if the mount is using the
3176 'sec=sys' security flavour. In effect it is
3177 disabling idmapping, which can make migration from
3178 legacy NFSv2/v3 systems to NFSv4 easier.
3179 Servers that do not support this mode of operation
3180 will be autodetected by the client, and it will fall
3181 back to using the idmapper.
3182 To turn off this behaviour, set the value to '0'.
3184 [NFS4] Specify an additional fixed unique ident-
3185 ification string that NFSv4 clients can insert into
3186 their nfs_client_id4 string. This is typically a
3187 UUID that is generated at system install time.
3189 nfs.send_implementation_id =
3190 [NFSv4.1] Send client implementation identification
3191 information in exchange_id requests.
3192 If zero, no implementation identification information
3194 The default is to send the implementation identification
3197 nfs.recover_lost_locks =
3198 [NFSv4] Attempt to recover locks that were lost due
3199 to a lease timeout on the server. Please note that
3200 doing this risks data corruption, since there are
3201 no guarantees that the file will remain unchanged
3202 after the locks are lost.
3203 If you want to enable the kernel legacy behaviour of
3204 attempting to recover these locks, then set this
3206 The default parameter value of '0' causes the kernel
3207 not to attempt recovery of lost locks.
3209 nfs4.layoutstats_timer =
3210 [NFSv4.2] Change the rate at which the kernel sends
3211 layoutstats to the pNFS metadata server.
3213 Setting this to value to 0 causes the kernel to use
3214 whatever value is the default set by the layout
3215 driver. A non-zero value sets the minimum interval
3216 in seconds between layoutstats transmissions.
3218 nfsd.nfs4_disable_idmapping=
3219 [NFSv4] When set to the default of '1', the NFSv4
3220 server will return only numeric uids and gids to
3221 clients using auth_sys, and will accept numeric uids
3222 and gids from such clients. This is intended to ease
3223 migration from NFSv2/v3.
3225 nmi_backtrace.backtrace_idle [KNL]
3226 Dump stacks even of idle CPUs in response to an
3227 NMI stack-backtrace request.
3229 nmi_debug= [KNL,SH] Specify one or more actions to take
3230 when a NMI is triggered.
3231 Format: [state][,regs][,debounce][,die]
3233 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
3234 Format: [panic,][nopanic,][num]
3236 0 - turn hardlockup detector in nmi_watchdog off
3237 1 - turn hardlockup detector in nmi_watchdog on
3238 When panic is specified, panic when an NMI watchdog
3239 timeout occurs (or 'nopanic' to not panic on an NMI
3240 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
3241 To disable both hard and soft lockup detectors,
3242 please see 'nowatchdog'.
3243 This is useful when you use a panic=... timeout and
3244 need the box quickly up again.
3246 These settings can be accessed at runtime via
3247 the nmi_watchdog and hardlockup_panic sysctls.
3249 netpoll.carrier_timeout=
3250 [NET] Specifies amount of time (in seconds) that
3251 netpoll should wait for a carrier. By default netpoll
3254 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3255 emulation library even if a 387 maths coprocessor
3258 no5lvl [X86-64] Disable 5-level paging mode. Forces
3259 kernel to use 4-level paging instead.
3261 nofsgsbase [X86] Disables FSGSBASE instructions.
3264 [HW] Never suspend the console
3265 Disable suspending of consoles during suspend and
3266 hibernate operations. Once disabled, debugging
3267 messages can reach various consoles while the rest
3268 of the system is being put to sleep (ie, while
3269 debugging driver suspend/resume hooks). This may
3270 not work reliably with all consoles, but is known
3271 to work with serial and VGA consoles.
3272 To facilitate more flexible debugging, we also add
3273 console_suspend, a printk module parameter to control
3274 it. Users could use console_suspend (usually
3275 /sys/module/printk/parameters/console_suspend) to
3276 turn on/off it dynamically.
3278 novmcoredd [KNL,KDUMP]
3279 Disable device dump. Device dump allows drivers to
3280 append dump data to vmcore so you can collect driver
3281 specified debug info. Drivers can append the data
3282 without any limit and this data is stored in memory,
3283 so this may cause significant memory stress. Disabling
3284 device dump can help save memory but the driver debug
3285 data will be no longer available. This parameter
3286 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
3289 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
3290 caches in the slab allocator. Saves per-node memory,
3291 but will impact performance.
3295 noaltinstr [S390] Disables alternative instructions patching
3296 (CPU alternatives feature).
3298 noapic [SMP,APIC] Tells the kernel to not make use of any
3299 IOAPICs that may be present in the system.
3301 noautogroup Disable scheduler automatic task group creation.
3303 nobats [PPC] Do not use BATs for mapping kernel lowmem
3304 on "Classic" PPC cores.
3308 noclflush [BUGS=X86] Don't use the CLFLUSH instruction
3310 delayacct [KNL] Enable per-task delay accounting
3312 nodsp [SH] Disable hardware DSP at boot time.
3314 noefi Disable EFI runtime services support.
3316 no_entry_flush [PPC] Don't flush the L1-D cache when entering the kernel.
3321 On X86-32 available only on PAE configured kernels.
3322 noexec=on: enable non-executable mappings (default)
3323 noexec=off: disable non-executable mappings
3326 Disable SMAP (Supervisor Mode Access Prevention)
3327 even if it is supported by processor.
3330 Disable SMEP (Supervisor Mode Execution Prevention)
3331 even if it is supported by processor.
3334 This affects only 32-bit executables.
3335 noexec32=on: enable non-executable mappings (default)
3336 read doesn't imply executable mappings
3337 noexec32=off: disable non-executable mappings
3338 read implies executable mappings
3340 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3342 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3343 register save and restore. The kernel will only save
3344 legacy floating-point registers on task switch.
3346 nohugeiomap [KNL,X86,PPC,ARM64] Disable kernel huge I/O mappings.
3348 nohugevmalloc [PPC] Disable kernel huge vmalloc mappings.
3350 nosmt [KNL,S390] Disable symmetric multithreading (SMT).
3351 Equivalent to smt=1.
3353 [KNL,X86] Disable symmetric multithreading (SMT).
3354 nosmt=force: Force disable SMT, cannot be undone
3355 via the sysfs control file.
3357 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1
3358 (bounds check bypass). With this option data leaks are
3359 possible in the system.
3361 nospectre_v2 [X86,PPC_FSL_BOOK3E,ARM64] Disable all mitigations for
3362 the Spectre variant 2 (indirect branch prediction)
3363 vulnerability. System may allow data leaks with this
3366 nospec_store_bypass_disable
3367 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
3370 [PPC] Don't flush the L1-D cache after accessing user data.
3372 noxsave [BUGS=X86] Disables x86 extended register state save
3373 and restore using xsave. The kernel will fallback to
3374 enabling legacy floating-point and sse state.
3376 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
3377 register states. The kernel will fall back to use
3378 xsave to save the states. By using this parameter,
3379 performance of saving the states is degraded because
3380 xsave doesn't support modified optimization while
3381 xsaveopt supports it on xsaveopt enabled systems.
3383 noxsaves [X86] Disables xsaves and xrstors used in saving and
3384 restoring x86 extended register state in compacted
3385 form of xsave area. The kernel will fall back to use
3386 xsaveopt and xrstor to save and restore the states
3387 in standard form of xsave area. By using this
3388 parameter, xsave area per process might occupy more
3389 memory on xsaves enabled systems.
3391 nohlt [ARM,ARM64,MICROBLAZE,SH] Forces the kernel to busy wait
3392 in do_idle() and not use the arch_cpu_idle()
3393 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
3394 to be effective. This is useful on platforms where the
3395 sleep(SH) or wfi(ARM,ARM64) instructions do not work
3396 correctly or when doing power measurements to evalute
3397 the impact of the sleep instructions. This is also
3398 useful when using JTAG debugger.
3400 no_file_caps Tells the kernel not to honor file capabilities. The
3401 only way then for a file to be executed with privilege
3402 is to be setuid root or executed by root.
3404 nohalt [IA-64] Tells the kernel not to use the power saving
3405 function PAL_HALT_LIGHT when idle. This increases
3406 power-consumption. On the positive side, it reduces
3407 interrupt wake-up latency, which may improve performance
3408 in certain environments such as networked servers or
3412 Force pointers printed to the console or buffers to be
3413 unhashed. By default, when a pointer is printed via %p
3414 format string, that pointer is "hashed", i.e. obscured
3415 by hashing the pointer value. This is a security feature
3416 that hides actual kernel addresses from unprivileged
3417 users, but it also makes debugging the kernel more
3418 difficult since unequal pointers can no longer be
3419 compared. However, if this command-line option is
3420 specified, then all normal pointers will have their true
3421 value printed. Pointers printed via %pK may still be
3422 hashed. This option should only be specified when
3423 debugging the kernel. Please do not use on production
3426 nohibernate [HIBERNATION] Disable hibernation and resume.
3428 nohz= [KNL] Boottime enable/disable dynamic ticks
3429 Valid arguments: on, off
3432 nohz_full= [KNL,BOOT,SMP,ISOL]
3433 The argument is a cpu list, as described above.
3434 In kernels built with CONFIG_NO_HZ_FULL=y, set
3435 the specified list of CPUs whose tick will be stopped
3436 whenever possible. The boot CPU will be forced outside
3437 the range to maintain the timekeeping. Any CPUs
3438 in this list will have their RCU callbacks offloaded,
3439 just as if they had also been called out in the
3440 rcu_nocbs= boot parameter.
3442 noiotrap [SH] Disables trapped I/O port accesses.
3444 noirqdebug [X86-32] Disables the code which attempts to detect and
3445 disable unhandled interrupt sources.
3447 no_timer_check [X86,APIC] Disables the code which tests for
3448 broken timer IRQ sources.
3450 noisapnp [ISAPNP] Disables ISA PnP code.
3452 noinitrd [RAM] Tells the kernel not to load any configured
3455 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
3457 [Deprecated - use intremap=off]
3461 noinvpcid [X86] Disable the INVPCID cpu feature.
3463 nojitter [IA-64] Disables jitter checking for ITC timers.
3465 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
3467 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
3471 [X86,PV_OPS] Disable paravirtualized VMware scheduler
3472 clock and use the default one.
3474 no-steal-acc [X86,PV_OPS,ARM64] Disable paravirtualized steal time
3475 accounting. steal time is computed, but won't
3476 influence scheduler behaviour
3478 nolapic [X86-32,APIC] Do not enable or use the local APIC.
3480 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
3482 noltlbs [PPC] Do not use large page/tlb entries for kernel
3483 lowmem mapping on PPC40x and PPC8xx
3485 nomca [IA-64] Disable machine check abort handling
3487 nomce [X86-32] Disable Machine Check Exception
3489 nomfgpt [X86-32] Disable Multi-Function General Purpose
3490 Timer usage (for AMD Geode machines).
3492 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3493 shutdown the other cpus. Instead use the REBOOT_VECTOR
3496 nomodule Disable module load
3498 nopat [X86] Disable PAT (page attribute table extension of
3499 pagetables) support.
3501 nopcid [X86-64] Disable the PCID cpu feature.
3503 norandmaps Don't use address space randomization. Equivalent to
3504 echo 0 > /proc/sys/kernel/randomize_va_space
3506 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
3507 with UP alternatives
3509 nordrand [X86] Disable kernel use of the RDRAND and
3510 RDSEED instructions even if they are supported
3511 by the processor. RDRAND and RDSEED are still
3512 available to user space applications.
3514 noresume [SWSUSP] Disables resume and restores original swap
3517 no-scroll [VGA] Disables scrollback.
3518 This is required for the Braillex ib80-piezo Braille
3519 reader made by F.H. Papenmeier (Germany).
3523 nosep [BUGS=X86-32] Disables x86 SYSENTER/SYSEXIT support.
3525 nosgx [X86-64,SGX] Disables Intel SGX kernel support.
3527 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
3528 and disable the IO APIC. legacy for "maxcpus=0".
3530 nosoftlockup [KNL] Disable the soft-lockup detector.
3532 nosync [HW,M68K] Disables sync negotiation for all devices.
3534 nowatchdog [KNL] Disable both lockup detectors, i.e.
3535 soft-lockup and NMI watchdog (hard-lockup).
3539 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
3541 cpu0_hotplug [X86] Turn on CPU0 hotplug feature when
3542 CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off.
3543 Some features depend on CPU0. Known dependencies are:
3544 1. Resume from suspend/hibernate depends on CPU0.
3545 Suspend/hibernate will fail if CPU0 is offline and you
3546 need to online CPU0 before suspend/hibernate.
3547 2. PIC interrupts also depend on CPU0. CPU0 can't be
3548 removed if a PIC interrupt is detected.
3549 It's said poweroff/reboot may depend on CPU0 on some
3550 machines although I haven't seen such issues so far
3551 after CPU0 is offline on a few tested machines.
3552 If the dependencies are under your control, you can
3553 turn on cpu0_hotplug.
3555 nps_mtm_hs_ctr= [KNL,ARC]
3556 This parameter sets the maximum duration, in
3557 cycles, each HW thread of the CTOP can run
3558 without interruptions, before HW switches it.
3559 The actual maximum duration is 16 times this
3561 Format: integer between 1 and 255
3564 nptcg= [IA-64] Override max number of concurrent global TLB
3565 purges which is reported from either PAL_VM_SUMMARY or
3568 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
3569 could support. nr_cpus=n : n >= 1 limits the kernel to
3570 support 'n' processors. It could be larger than the
3571 number of already plugged CPU during bootup, later in
3572 runtime you can physically add extra cpu until it reaches
3573 n. So during boot up some boot time memory for per-cpu
3574 variables need be pre-allocated for later physical cpu
3577 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
3579 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86] Disable NUMA, Only
3580 set up a single NUMA node spanning all memory.
3582 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic
3584 Allowed values are enable and disable
3586 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
3587 'node', 'default' can be specified
3588 This can be set from sysctl after boot.
3589 See Documentation/admin-guide/sysctl/vm.rst for details.
3591 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
3592 See Documentation/core-api/debugging-via-ohci1394.rst for more
3595 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
3596 Rather than timing out after 20 ms if an EC
3597 command is not properly ACKed, override the length
3598 of the timeout. We have interrupts disabled while
3599 waiting for the ACK, so if this is set too high
3600 interrupts *may* be lost!
3602 omap_mux= [OMAP] Override bootloader pin multiplexing.
3603 Format: <mux_mode0.mode_name=value>...
3604 For example, to override I2C bus2:
3605 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
3607 oops=panic Always panic on oopses. Default is to just kill the
3608 process, but there is a small probability of
3609 deadlocking the machine.
3610 This will also cause panics on machine check exceptions.
3611 Useful together with panic=30 to trigger a reboot.
3614 [KNL] Boolean flag to control whether the page allocator
3615 should randomize its free lists. The randomization may
3616 be automatically enabled if the kernel detects it is
3617 running on a platform with a direct-mapped memory-side
3618 cache, and this parameter can be used to
3619 override/disable that behavior. The state of the flag
3620 can be read from sysfs at:
3621 /sys/module/page_alloc/parameters/shuffle.
3623 page_owner= [KNL] Boot-time page_owner enabling option.
3624 Storage of the information about who allocated
3625 each page is disabled in default. With this switch,
3627 on: enable the feature
3629 page_poison= [KNL] Boot-time parameter changing the state of
3630 poisoning on the buddy allocator, available with
3631 CONFIG_PAGE_POISONING=y.
3632 off: turn off poisoning (default)
3633 on: turn on poisoning
3635 page_reporting.page_reporting_order=
3636 [KNL] Minimal page reporting order
3638 Adjust the minimal page reporting order. The page
3639 reporting is disabled when it exceeds (MAX_ORDER-1).
3641 panic= [KNL] Kernel behaviour on panic: delay <timeout>
3642 timeout > 0: seconds before rebooting
3643 timeout = 0: wait forever
3644 timeout < 0: reboot immediately
3647 panic_print= Bitmask for printing system info when panic happens.
3648 User can chose combination of the following bits:
3649 bit 0: print all tasks info
3650 bit 1: print system memory info
3651 bit 2: print timer info
3652 bit 3: print locks info if CONFIG_LOCKDEP is on
3653 bit 4: print ftrace buffer
3654 bit 5: print all printk messages in buffer
3656 panic_on_taint= Bitmask for conditionally calling panic() in add_taint()
3657 Format: <hex>[,nousertaint]
3658 Hexadecimal bitmask representing the set of TAINT flags
3659 that will cause the kernel to panic when add_taint() is
3660 called with any of the flags in this set.
3661 The optional switch "nousertaint" can be utilized to
3662 prevent userspace forced crashes by writing to sysctl
3663 /proc/sys/kernel/tainted any flagset matching with the
3664 bitmask set on panic_on_taint.
3665 See Documentation/admin-guide/tainted-kernels.rst for
3666 extra details on the taint flags that users can pick
3667 to compose the bitmask to assign to panic_on_taint.
3669 panic_on_warn panic() instead of WARN(). Useful to cause kdump
3672 crash_kexec_post_notifiers
3673 Run kdump after running panic-notifiers and dumping
3674 kmsg. This only for the users who doubt kdump always
3675 succeeds in any situation.
3676 Note that this also increases risks of kdump failure,
3677 because some panic notifiers can make the crashed
3678 kernel more unstable.
3680 parkbd.port= [HW] Parallel port number the keyboard adapter is
3681 connected to, default is 0.
3683 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
3684 0 for XT, 1 for AT (default is AT).
3687 parport= [HW,PPT] Specify parallel ports. 0 disables.
3688 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
3689 Use 'auto' to force the driver to use any
3690 IRQ/DMA settings detected (the default is to
3691 ignore detected IRQ/DMA settings because of
3692 possible conflicts). You can specify the base
3693 address, IRQ, and DMA settings; IRQ and DMA
3694 should be numbers, or 'auto' (for using detected
3695 settings on that particular port), or 'nofifo'
3696 (to avoid using a FIFO even if it is detected).
3697 Parallel ports are assigned in the order they
3698 are specified on the command line, starting
3701 parport_init_mode= [HW,PPT]
3702 Configure VIA parallel port to operate in
3703 a specific mode. This is necessary on Pegasos
3704 computer where firmware has no options for setting
3705 up parallel port mode and sets it to spp.
3706 Currently this function knows 686a and 8231 chips.
3707 Format: [spp|ps2|epp|ecp|ecpepp]
3709 pata_legacy.all= [HW,LIBATA]
3711 Set to non-zero to probe primary and secondary ISA
3712 port ranges on PCI systems where no PCI PATA device
3713 has been found at either range. Disabled by default.
3715 pata_legacy.autospeed= [HW,LIBATA]
3717 Set to non-zero if a chip is present that snoops speed
3718 changes. Disabled by default.
3720 pata_legacy.ht6560a= [HW,LIBATA]
3722 Set to 1, 2, or 3 for HT 6560A on the primary channel,
3723 the secondary channel, or both channels respectively.
3724 Disabled by default.
3726 pata_legacy.ht6560b= [HW,LIBATA]
3728 Set to 1, 2, or 3 for HT 6560B on the primary channel,
3729 the secondary channel, or both channels respectively.
3730 Disabled by default.
3732 pata_legacy.iordy_mask= [HW,LIBATA]
3734 IORDY enable mask. Set individual bits to allow IORDY
3735 for the respective channel. Bit 0 is for the first
3736 legacy channel handled by this driver, bit 1 is for
3737 the second channel, and so on. The sequence will often
3738 correspond to the primary legacy channel, the secondary
3739 legacy channel, and so on, but the handling of a PCI
3740 bus and the use of other driver options may interfere
3741 with the sequence. By default IORDY is allowed across
3744 pata_legacy.opti82c46x= [HW,LIBATA]
3746 Set to 1, 2, or 3 for Opti 82c611A on the primary
3747 channel, the secondary channel, or both channels
3748 respectively. Disabled by default.
3750 pata_legacy.opti82c611a= [HW,LIBATA]
3752 Set to 1, 2, or 3 for Opti 82c465MV on the primary
3753 channel, the secondary channel, or both channels
3754 respectively. Disabled by default.
3756 pata_legacy.pio_mask= [HW,LIBATA]
3758 PIO mode mask for autospeed devices. Set individual
3759 bits to allow the use of the respective PIO modes.
3760 Bit 0 is for mode 0, bit 1 is for mode 1, and so on.
3761 All modes allowed by default.
3763 pata_legacy.probe_all= [HW,LIBATA]
3765 Set to non-zero to probe tertiary and further ISA
3766 port ranges on PCI systems. Disabled by default.
3768 pata_legacy.probe_mask= [HW,LIBATA]
3770 Probe mask for legacy ISA PATA ports. Depending on
3771 platform configuration and the use of other driver
3772 options up to 6 legacy ports are supported: 0x1f0,
3773 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing
3774 of individual ports can be disabled by setting the
3775 corresponding bits in the mask to 1. Bit 0 is for
3776 the first port in the list above (0x1f0), and so on.
3777 By default all supported ports are probed.
3779 pata_legacy.qdi= [HW,LIBATA]
3781 Set to non-zero to probe QDI controllers. By default
3782 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise.
3784 pata_legacy.winbond= [HW,LIBATA]
3786 Set to non-zero to probe Winbond controllers. Use
3787 the standard I/O port (0x130) if 1, otherwise the
3788 value given is the I/O port to use (typically 0x1b0).
3789 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE,
3792 pata_platform.pio_mask= [HW,LIBATA]
3794 Supported PIO mode mask. Set individual bits to allow
3795 the use of the respective PIO modes. Bit 0 is for
3796 mode 0, bit 1 is for mode 1, and so on. Mode 0 only
3800 Halt all CPUs after the first oops has been printed for
3801 the specified number of seconds. This is to be used if
3802 your oopses keep scrolling off the screen.
3807 See header of drivers/block/paride/pcd.c.
3808 See also Documentation/admin-guide/blockdev/paride.rst.
3810 pci=option[,option...] [PCI] various PCI subsystem options.
3812 Some options herein operate on a specific device
3813 or a set of devices (<pci_dev>). These are
3814 specified in one of the following formats:
3816 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
3817 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
3819 Note: the first format specifies a PCI
3820 bus/device/function address which may change
3821 if new hardware is inserted, if motherboard
3822 firmware changes, or due to changes caused
3823 by other kernel parameters. If the
3824 domain is left unspecified, it is
3825 taken to be zero. Optionally, a path
3826 to a device through multiple device/function
3827 addresses can be specified after the base
3828 address (this is more robust against
3829 renumbering issues). The second format
3830 selects devices using IDs from the
3831 configuration space which may match multiple
3832 devices in the system.
3834 earlydump dump PCI config space before the kernel
3836 off [X86] don't probe for the PCI bus
3837 bios [X86-32] force use of PCI BIOS, don't access
3838 the hardware directly. Use this if your machine
3839 has a non-standard PCI host bridge.
3840 nobios [X86-32] disallow use of PCI BIOS, only direct
3841 hardware access methods are allowed. Use this
3842 if you experience crashes upon bootup and you
3843 suspect they are caused by the BIOS.
3844 conf1 [X86] Force use of PCI Configuration Access
3845 Mechanism 1 (config address in IO port 0xCF8,
3846 data in IO port 0xCFC, both 32-bit).
3847 conf2 [X86] Force use of PCI Configuration Access
3848 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
3849 the function, IO port 0xCFA, also 8-bit, sets
3850 bus number. The config space is then accessed
3851 through ports 0xC000-0xCFFF).
3852 See http://wiki.osdev.org/PCI for more info
3853 on the configuration access mechanisms.
3854 noaer [PCIE] If the PCIEAER kernel config parameter is
3855 enabled, this kernel boot option can be used to
3856 disable the use of PCIE advanced error reporting.
3857 nodomains [PCI] Disable support for multiple PCI
3858 root domains (aka PCI segments, in ACPI-speak).
3859 nommconf [X86] Disable use of MMCONFIG for PCI
3861 check_enable_amd_mmconf [X86] check for and enable
3862 properly configured MMIO access to PCI
3863 config space on AMD family 10h CPU
3864 nomsi [MSI] If the PCI_MSI kernel config parameter is
3865 enabled, this kernel boot option can be used to
3866 disable the use of MSI interrupts system-wide.
3867 noioapicquirk [APIC] Disable all boot interrupt quirks.
3868 Safety option to keep boot IRQs enabled. This
3869 should never be necessary.
3870 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
3871 primary IO-APIC for bridges that cannot disable
3872 boot IRQs. This fixes a source of spurious IRQs
3873 when the system masks IRQs.
3874 noioapicreroute [APIC] Disable workaround that uses the
3875 boot IRQ equivalent of an IRQ that connects to
3876 a chipset where boot IRQs cannot be disabled.
3877 The opposite of ioapicreroute.
3878 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
3879 routing table. These calls are known to be buggy
3880 on several machines and they hang the machine
3881 when used, but on other computers it's the only
3882 way to get the interrupt routing table. Try
3883 this option if the kernel is unable to allocate
3884 IRQs or discover secondary PCI buses on your
3886 rom [X86] Assign address space to expansion ROMs.
3887 Use with caution as certain devices share
3888 address decoders between ROMs and other
3890 norom [X86] Do not assign address space to
3891 expansion ROMs that do not already have
3892 BIOS assigned address ranges.
3893 nobar [X86] Do not assign address space to the
3894 BARs that weren't assigned by the BIOS.
3895 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
3896 assigned automatically to PCI devices. You can
3897 make the kernel exclude IRQs of your ISA cards
3899 pirqaddr=0xAAAAA [X86] Specify the physical address
3900 of the PIRQ table (normally generated
3901 by the BIOS) if it is outside the
3902 F0000h-100000h range.
3903 lastbus=N [X86] Scan all buses thru bus #N. Can be
3904 useful if the kernel is unable to find your
3905 secondary buses and you want to tell it
3906 explicitly which ones they are.
3907 assign-busses [X86] Always assign all PCI bus
3908 numbers ourselves, overriding
3909 whatever the firmware may have done.
3910 usepirqmask [X86] Honor the possible IRQ mask stored
3911 in the BIOS $PIR table. This is needed on
3912 some systems with broken BIOSes, notably
3913 some HP Pavilion N5400 and Omnibook XE3
3914 notebooks. This will have no effect if ACPI
3915 IRQ routing is enabled.
3916 noacpi [X86] Do not use ACPI for IRQ routing
3917 or for PCI scanning.
3918 use_crs [X86] Use PCI host bridge window information
3919 from ACPI. On BIOSes from 2008 or later, this
3920 is enabled by default. If you need to use this,
3921 please report a bug.
3922 nocrs [X86] Ignore PCI host bridge windows from ACPI.
3923 If you need to use this, please report a bug.
3924 routeirq Do IRQ routing for all PCI devices.
3925 This is normally done in pci_enable_device(),
3926 so this option is a temporary workaround
3927 for broken drivers that don't call it.
3928 skip_isa_align [X86] do not align io start addr, so can
3929 handle more pci cards
3930 noearly [X86] Don't do any early type 1 scanning.
3931 This might help on some broken boards which
3932 machine check when some devices' config space
3933 is read. But various workarounds are disabled
3934 and some IOMMU drivers will not work.
3935 bfsort Sort PCI devices into breadth-first order.
3936 This sorting is done to get a device
3937 order compatible with older (<= 2.4) kernels.
3938 nobfsort Don't sort PCI devices into breadth-first order.
3939 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
3940 tuning and use the BIOS-configured MPS defaults.
3941 pcie_bus_safe Set every device's MPS to the largest value
3942 supported by all devices below the root complex.
3943 pcie_bus_perf Set device MPS to the largest allowable MPS
3944 based on its parent bus. Also set MRRS (Max
3945 Read Request Size) to the largest supported
3946 value (no larger than the MPS that the device
3947 or bus can support) for best performance.
3948 pcie_bus_peer2peer Set every device's MPS to 128B, which
3949 every device is guaranteed to support. This
3950 configuration allows peer-to-peer DMA between
3951 any pair of devices, possibly at the cost of
3952 reduced performance. This also guarantees
3953 that hot-added devices will work.
3954 cbiosize=nn[KMG] The fixed amount of bus space which is
3955 reserved for the CardBus bridge's IO window.
3956 The default value is 256 bytes.
3957 cbmemsize=nn[KMG] The fixed amount of bus space which is
3958 reserved for the CardBus bridge's memory
3959 window. The default value is 64 megabytes.
3962 [<order of align>@]<pci_dev>[; ...]
3963 Specifies alignment and device to reassign
3964 aligned memory resources. How to
3965 specify the device is described above.
3966 If <order of align> is not specified,
3967 PAGE_SIZE is used as alignment.
3968 A PCI-PCI bridge can be specified if resource
3969 windows need to be expanded.
3970 To specify the alignment for several
3971 instances of a device, the PCI vendor,
3972 device, subvendor, and subdevice may be
3973 specified, e.g., 12@pci:8086:9c22:103c:198f
3974 for 4096-byte alignment.
3975 ecrc= Enable/disable PCIe ECRC (transaction layer
3976 end-to-end CRC checking).
3977 bios: Use BIOS/firmware settings. This is the
3981 hpiosize=nn[KMG] The fixed amount of bus space which is
3982 reserved for hotplug bridge's IO window.
3983 Default size is 256 bytes.
3984 hpmmiosize=nn[KMG] The fixed amount of bus space which is
3985 reserved for hotplug bridge's MMIO window.
3986 Default size is 2 megabytes.
3987 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
3988 reserved for hotplug bridge's MMIO_PREF window.
3989 Default size is 2 megabytes.
3990 hpmemsize=nn[KMG] The fixed amount of bus space which is
3991 reserved for hotplug bridge's MMIO and
3993 Default size is 2 megabytes.
3994 hpbussize=nn The minimum amount of additional bus numbers
3995 reserved for buses below a hotplug bridge.
3997 realloc= Enable/disable reallocating PCI bridge resources
3998 if allocations done by BIOS are too small to
3999 accommodate resources required by all child
4001 off: Turn realloc off
4003 realloc same as realloc=on
4004 noari do not use PCIe ARI.
4005 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
4006 do not use PCIe ATS (and IOMMU device IOTLB).
4007 pcie_scan_all Scan all possible PCIe devices. Otherwise we
4008 only look for one device below a PCIe downstream
4010 big_root_window Try to add a big 64bit memory window to the PCIe
4011 root complex on AMD CPUs. Some GFX hardware
4012 can resize a BAR to allow access to all VRAM.
4013 Adding the window is slightly risky (it may
4014 conflict with unreported devices), so this
4016 disable_acs_redir=<pci_dev>[; ...]
4017 Specify one or more PCI devices (in the format
4018 specified above) separated by semicolons.
4019 Each device specified will have the PCI ACS
4020 redirect capabilities forced off which will
4021 allow P2P traffic between devices through
4022 bridges without forcing it upstream. Note:
4023 this removes isolation between devices and
4024 may put more devices in an IOMMU group.
4025 force_floating [S390] Force usage of floating interrupts.
4026 nomio [S390] Do not use MIO instructions.
4027 norid [S390] ignore the RID field and force use of
4028 one PCI domain per PCI function
4030 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
4033 force Enable ASPM even on devices that claim not to support it.
4034 WARNING: Forcing ASPM on may cause system lockups.
4036 pcie_ports= [PCIE] PCIe port services handling:
4037 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
4038 even if the platform doesn't give the OS permission to
4039 use them. This may cause conflicts if the platform
4040 also tries to use these services.
4041 dpc-native Use native PCIe service for DPC only. May
4042 cause conflicts if firmware uses AER or DPC.
4043 compat Disable native PCIe services (PME, AER, DPC, PCIe
4046 pcie_port_pm= [PCIE] PCIe port power management handling:
4047 off Disable power management of all PCIe ports
4048 force Forcibly enable power management of all PCIe ports
4050 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
4051 nomsi Do not use MSI for native PCIe PME signaling (this makes
4052 all PCIe root ports use INTx for all services).
4054 pcmv= [HW,PCMCIA] BadgePAD 4
4058 Keep all power-domains already enabled by bootloader on,
4059 even if no driver has claimed them. This is useful
4060 for debug and development, but should not be
4061 needed on a platform with proper driver support.
4064 See Documentation/admin-guide/blockdev/paride.rst.
4066 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
4069 See arch/parisc/kernel/pdc_chassis.c
4071 percpu_alloc= Select which percpu first chunk allocator to use.
4072 Currently supported values are "embed" and "page".
4073 Archs may support subset or none of the selections.
4074 See comments in mm/percpu.c for details on each
4075 allocator. This parameter is primarily for debugging
4076 and performance comparison.
4079 See Documentation/admin-guide/blockdev/paride.rst.
4082 See Documentation/admin-guide/blockdev/paride.rst.
4084 pirq= [SMP,APIC] Manual mp-table setup
4085 See Documentation/x86/i386/IO-APIC.rst.
4087 plip= [PPT,NET] Parallel port network link
4088 Format: { parport<nr> | timid | 0 }
4089 See also Documentation/admin-guide/parport.rst.
4091 pmtmr= [X86] Manual setup of pmtmr I/O Port.
4092 Override pmtimer IOPort with a hex value.
4095 pm_debug_messages [SUSPEND,KNL]
4096 Enable suspend/resume debug messages during boot up.
4099 Enable PNP debug messages (depends on the
4100 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
4101 via /sys/module/pnp/parameters/debug. We always show
4102 current resource usage; turning this on also shows
4103 possible settings and some assignment information.
4109 { on | off | curr | res | no-curr | no-res }
4112 [ISAPNP] Exclude IRQs for the autoconfiguration
4115 [ISAPNP] Exclude DMAs for the autoconfiguration
4117 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
4118 Ranges are in pairs (I/O port base and size).
4121 [ISAPNP] Exclude memory regions for the
4123 Ranges are in pairs (memory base and size).
4125 ports= [IP_VS_FTP] IPVS ftp helper module
4127 Up to 8 (IP_VS_APP_MAX_PORTS) ports
4129 Format: <port>,<port>....
4131 powersave=off [PPC] This option disables power saving features.
4132 It specifically disables cpuidle and sets the
4133 platform machine description specific power_save
4134 function to NULL. On Idle the CPU just reduces
4137 ppc_strict_facility_enable
4138 [PPC] This option catches any kernel floating point,
4139 Altivec, VSX and SPE outside of regions specifically
4140 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
4141 There is some performance impact when enabling this.
4145 Disable Hardware Transactional Memory
4148 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC
4149 none - Limited to cond_resched() calls
4150 voluntary - Limited to cond_resched() and might_sleep() calls
4151 full - Any section that isn't explicitly preempt disabled
4152 can be preempted anytime.
4154 print-fatal-signals=
4155 [KNL] debug: print fatal signals
4157 If enabled, warn about various signal handling
4158 related application anomalies: too many signals,
4159 too many POSIX.1 timers, fatal signals causing a
4162 If you hit the warning due to signal overflow,
4163 you might want to try "ulimit -i unlimited".
4167 printk.always_kmsg_dump=
4168 Trigger kmsg_dump for cases other than kernel oops or
4170 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4173 printk.devkmsg={on,off,ratelimit}
4174 Control writing to /dev/kmsg.
4175 on - unlimited logging to /dev/kmsg from userspace
4176 off - logging to /dev/kmsg disabled
4177 ratelimit - ratelimit the logging
4180 printk.time= Show timing data prefixed to each printk message line
4181 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4183 processor.max_cstate= [HW,ACPI]
4184 Limit processor to maximum C-state
4185 max_cstate=9 overrides any DMI blacklist limit.
4187 processor.nocst [HW,ACPI]
4188 Ignore the _CST method to determine C-states,
4189 instead using the legacy FADT method
4191 profile= [KNL] Enable kernel profiling via /proc/profile
4192 Format: [<profiletype>,]<number>
4193 Param: <profiletype>: "schedule", "sleep", or "kvm"
4194 [defaults to kernel profiling]
4195 Param: "schedule" - profile schedule points.
4196 Param: "sleep" - profile D-state sleeping (millisecs).
4197 Requires CONFIG_SCHEDSTATS
4198 Param: "kvm" - profile VM exits.
4199 Param: <number> - step/bucket size as a power of 2 for
4200 statistical time based profiling.
4202 prompt_ramdisk= [RAM] [Deprecated]
4204 prot_virt= [S390] enable hosting protected virtual machines
4205 isolated from the hypervisor (if hardware supports
4209 psi= [KNL] Enable or disable pressure stall information
4213 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
4214 probe for; one of (bare|imps|exps|lifebook|any).
4215 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
4217 psmouse.resetafter= [HW,MOUSE]
4218 Try to reset the device after so many bad packets
4221 [HW,MOUSE] Set desired mouse resolution, in dpi.
4222 psmouse.smartscroll=
4223 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
4224 0 = disabled, 1 = enabled (default).
4226 pstore.backend= Specify the name of the pstore backend to use
4229 See Documentation/admin-guide/blockdev/paride.rst.
4231 pti= [X86-64] Control Page Table Isolation of user and
4232 kernel address spaces. Disabling this feature
4233 removes hardening, but improves performance of
4234 system calls and interrupts.
4236 on - unconditionally enable
4237 off - unconditionally disable
4238 auto - kernel detects whether your CPU model is
4239 vulnerable to issues that PTI mitigates
4241 Not specifying this option is equivalent to pti=auto.
4244 Equivalent to pti=off
4247 [KNL] Number of legacy pty's. Overwrites compiled-in
4250 quiet [KNL] Disable most log messages
4255 See Documentation/admin-guide/md.rst.
4257 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
4258 See Documentation/admin-guide/blockdev/ramdisk.rst.
4260 ramdisk_start= [RAM] RAM disk image start address
4262 random.trust_cpu={on,off}
4263 [KNL] Enable or disable trusting the use of the
4264 CPU's random number generator (if available) to
4265 fully seed the kernel's CRNG. Default is controlled
4266 by CONFIG_RANDOM_TRUST_CPU.
4268 randomize_kstack_offset=
4269 [KNL] Enable or disable kernel stack offset
4270 randomization, which provides roughly 5 bits of
4271 entropy, frustrating memory corruption attacks
4272 that depend on stack address determinism or
4273 cross-syscall address exposures. This is only
4274 available on architectures that have defined
4275 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET.
4276 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4277 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT.
4279 ras=option[,option,...] [KNL] RAS-specific options
4282 Disable the Correctable Errors Collector,
4283 see CONFIG_RAS_CEC help text.
4286 The argument is a cpu list, as described above.
4288 In kernels built with CONFIG_RCU_NOCB_CPU=y, set
4289 the specified list of CPUs to be no-callback CPUs.
4290 Invocation of these CPUs' RCU callbacks will be
4291 offloaded to "rcuox/N" kthreads created for that
4292 purpose, where "x" is "p" for RCU-preempt, and
4293 "s" for RCU-sched, and "N" is the CPU number.
4294 This reduces OS jitter on the offloaded CPUs,
4295 which can be useful for HPC and real-time
4296 workloads. It can also improve energy efficiency
4297 for asymmetric multiprocessors.
4300 Rather than requiring that offloaded CPUs
4301 (specified by rcu_nocbs= above) explicitly
4302 awaken the corresponding "rcuoN" kthreads,
4303 make these kthreads poll for callbacks.
4304 This improves the real-time response for the
4305 offloaded CPUs by relieving them of the need to
4306 wake up the corresponding kthread, but degrades
4307 energy efficiency by requiring that the kthreads
4308 periodically wake up to do the polling.
4310 rcutree.blimit= [KNL]
4311 Set maximum number of finished RCU callbacks to
4312 process in one batch.
4314 rcutree.dump_tree= [KNL]
4315 Dump the structure of the rcu_node combining tree
4316 out at early boot. This is used for diagnostic
4317 purposes, to verify correct tree setup.
4319 rcutree.gp_cleanup_delay= [KNL]
4320 Set the number of jiffies to delay each step of
4321 RCU grace-period cleanup.
4323 rcutree.gp_init_delay= [KNL]
4324 Set the number of jiffies to delay each step of
4325 RCU grace-period initialization.
4327 rcutree.gp_preinit_delay= [KNL]
4328 Set the number of jiffies to delay each step of
4329 RCU grace-period pre-initialization, that is,
4330 the propagation of recent CPU-hotplug changes up
4331 the rcu_node combining tree.
4333 rcutree.use_softirq= [KNL]
4334 If set to zero, move all RCU_SOFTIRQ processing to
4335 per-CPU rcuc kthreads. Defaults to a non-zero
4336 value, meaning that RCU_SOFTIRQ is used by default.
4337 Specify rcutree.use_softirq=0 to use rcuc kthreads.
4339 But note that CONFIG_PREEMPT_RT=y kernels disable
4340 this kernel boot parameter, forcibly setting it
4343 rcutree.rcu_fanout_exact= [KNL]
4344 Disable autobalancing of the rcu_node combining
4345 tree. This is used by rcutorture, and might
4346 possibly be useful for architectures having high
4347 cache-to-cache transfer latencies.
4349 rcutree.rcu_fanout_leaf= [KNL]
4350 Change the number of CPUs assigned to each
4351 leaf rcu_node structure. Useful for very
4352 large systems, which will choose the value 64,
4353 and for NUMA systems with large remote-access
4354 latencies, which will choose a value aligned
4355 with the appropriate hardware boundaries.
4357 rcutree.rcu_min_cached_objs= [KNL]
4358 Minimum number of objects which are cached and
4359 maintained per one CPU. Object size is equal
4360 to PAGE_SIZE. The cache allows to reduce the
4361 pressure to page allocator, also it makes the
4362 whole algorithm to behave better in low memory
4365 rcutree.rcu_delay_page_cache_fill_msec= [KNL]
4366 Set the page-cache refill delay (in milliseconds)
4367 in response to low-memory conditions. The range
4368 of permitted values is in the range 0:100000.
4370 rcutree.jiffies_till_first_fqs= [KNL]
4371 Set delay from grace-period initialization to
4372 first attempt to force quiescent states.
4373 Units are jiffies, minimum value is zero,
4374 and maximum value is HZ.
4376 rcutree.jiffies_till_next_fqs= [KNL]
4377 Set delay between subsequent attempts to force
4378 quiescent states. Units are jiffies, minimum
4379 value is one, and maximum value is HZ.
4381 rcutree.jiffies_till_sched_qs= [KNL]
4382 Set required age in jiffies for a
4383 given grace period before RCU starts
4384 soliciting quiescent-state help from
4385 rcu_note_context_switch() and cond_resched().
4386 If not specified, the kernel will calculate
4387 a value based on the most recent settings
4388 of rcutree.jiffies_till_first_fqs
4389 and rcutree.jiffies_till_next_fqs.
4390 This calculated value may be viewed in
4391 rcutree.jiffies_to_sched_qs. Any attempt to set
4392 rcutree.jiffies_to_sched_qs will be cheerfully
4395 rcutree.kthread_prio= [KNL,BOOT]
4396 Set the SCHED_FIFO priority of the RCU per-CPU
4397 kthreads (rcuc/N). This value is also used for
4398 the priority of the RCU boost threads (rcub/N)
4399 and for the RCU grace-period kthreads (rcu_bh,
4400 rcu_preempt, and rcu_sched). If RCU_BOOST is
4401 set, valid values are 1-99 and the default is 1
4402 (the least-favored priority). Otherwise, when
4403 RCU_BOOST is not set, valid values are 0-99 and
4404 the default is zero (non-realtime operation).
4406 rcutree.rcu_nocb_gp_stride= [KNL]
4407 Set the number of NOCB callback kthreads in
4408 each group, which defaults to the square root
4409 of the number of CPUs. Larger numbers reduce
4410 the wakeup overhead on the global grace-period
4411 kthread, but increases that same overhead on
4412 each group's NOCB grace-period kthread.
4414 rcutree.qhimark= [KNL]
4415 Set threshold of queued RCU callbacks beyond which
4416 batch limiting is disabled.
4418 rcutree.qlowmark= [KNL]
4419 Set threshold of queued RCU callbacks below which
4420 batch limiting is re-enabled.
4422 rcutree.qovld= [KNL]
4423 Set threshold of queued RCU callbacks beyond which
4424 RCU's force-quiescent-state scan will aggressively
4425 enlist help from cond_resched() and sched IPIs to
4426 help CPUs more quickly reach quiescent states.
4427 Set to less than zero to make this be set based
4428 on rcutree.qhimark at boot time and to zero to
4429 disable more aggressive help enlistment.
4431 rcutree.rcu_idle_gp_delay= [KNL]
4432 Set wakeup interval for idle CPUs that have
4433 RCU callbacks (RCU_FAST_NO_HZ=y).
4435 rcutree.rcu_kick_kthreads= [KNL]
4436 Cause the grace-period kthread to get an extra
4437 wake_up() if it sleeps three times longer than
4438 it should at force-quiescent-state time.
4439 This wake_up() will be accompanied by a
4440 WARN_ONCE() splat and an ftrace_dump().
4442 rcutree.rcu_unlock_delay= [KNL]
4443 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels,
4444 this specifies an rcu_read_unlock()-time delay
4445 in microseconds. This defaults to zero.
4446 Larger delays increase the probability of
4447 catching RCU pointer leaks, that is, buggy use
4448 of RCU-protected pointers after the relevant
4449 rcu_read_unlock() has completed.
4451 rcutree.sysrq_rcu= [KNL]
4452 Commandeer a sysrq key to dump out Tree RCU's
4453 rcu_node tree with an eye towards determining
4454 why a new grace period has not yet started.
4456 rcuscale.gp_async= [KNL]
4457 Measure performance of asynchronous
4458 grace-period primitives such as call_rcu().
4460 rcuscale.gp_async_max= [KNL]
4461 Specify the maximum number of outstanding
4462 callbacks per writer thread. When a writer
4463 thread exceeds this limit, it invokes the
4464 corresponding flavor of rcu_barrier() to allow
4465 previously posted callbacks to drain.
4467 rcuscale.gp_exp= [KNL]
4468 Measure performance of expedited synchronous
4469 grace-period primitives.
4471 rcuscale.holdoff= [KNL]
4472 Set test-start holdoff period. The purpose of
4473 this parameter is to delay the start of the
4474 test until boot completes in order to avoid
4477 rcuscale.kfree_rcu_test= [KNL]
4478 Set to measure performance of kfree_rcu() flooding.
4480 rcuscale.kfree_rcu_test_double= [KNL]
4481 Test the double-argument variant of kfree_rcu().
4482 If this parameter has the same value as
4483 rcuscale.kfree_rcu_test_single, both the single-
4484 and double-argument variants are tested.
4486 rcuscale.kfree_rcu_test_single= [KNL]
4487 Test the single-argument variant of kfree_rcu().
4488 If this parameter has the same value as
4489 rcuscale.kfree_rcu_test_double, both the single-
4490 and double-argument variants are tested.
4492 rcuscale.kfree_nthreads= [KNL]
4493 The number of threads running loops of kfree_rcu().
4495 rcuscale.kfree_alloc_num= [KNL]
4496 Number of allocations and frees done in an iteration.
4498 rcuscale.kfree_loops= [KNL]
4499 Number of loops doing rcuscale.kfree_alloc_num number
4500 of allocations and frees.
4502 rcuscale.nreaders= [KNL]
4503 Set number of RCU readers. The value -1 selects
4504 N, where N is the number of CPUs. A value
4505 "n" less than -1 selects N-n+1, where N is again
4506 the number of CPUs. For example, -2 selects N
4507 (the number of CPUs), -3 selects N+1, and so on.
4508 A value of "n" less than or equal to -N selects
4511 rcuscale.nwriters= [KNL]
4512 Set number of RCU writers. The values operate
4513 the same as for rcuscale.nreaders.
4514 N, where N is the number of CPUs
4516 rcuscale.perf_type= [KNL]
4517 Specify the RCU implementation to test.
4519 rcuscale.shutdown= [KNL]
4520 Shut the system down after performance tests
4521 complete. This is useful for hands-off automated
4524 rcuscale.verbose= [KNL]
4525 Enable additional printk() statements.
4527 rcuscale.writer_holdoff= [KNL]
4528 Write-side holdoff between grace periods,
4529 in microseconds. The default of zero says
4532 rcutorture.fqs_duration= [KNL]
4533 Set duration of force_quiescent_state bursts
4536 rcutorture.fqs_holdoff= [KNL]
4537 Set holdoff time within force_quiescent_state bursts
4540 rcutorture.fqs_stutter= [KNL]
4541 Set wait time between force_quiescent_state bursts
4544 rcutorture.fwd_progress= [KNL]
4545 Enable RCU grace-period forward-progress testing
4546 for the types of RCU supporting this notion.
4548 rcutorture.fwd_progress_div= [KNL]
4549 Specify the fraction of a CPU-stall-warning
4550 period to do tight-loop forward-progress testing.
4552 rcutorture.fwd_progress_holdoff= [KNL]
4553 Number of seconds to wait between successive
4554 forward-progress tests.
4556 rcutorture.fwd_progress_need_resched= [KNL]
4557 Enclose cond_resched() calls within checks for
4558 need_resched() during tight-loop forward-progress
4561 rcutorture.gp_cond= [KNL]
4562 Use conditional/asynchronous update-side
4563 primitives, if available.
4565 rcutorture.gp_exp= [KNL]
4566 Use expedited update-side primitives, if available.
4568 rcutorture.gp_normal= [KNL]
4569 Use normal (non-expedited) asynchronous
4570 update-side primitives, if available.
4572 rcutorture.gp_sync= [KNL]
4573 Use normal (non-expedited) synchronous
4574 update-side primitives, if available. If all
4575 of rcutorture.gp_cond=, rcutorture.gp_exp=,
4576 rcutorture.gp_normal=, and rcutorture.gp_sync=
4577 are zero, rcutorture acts as if is interpreted
4578 they are all non-zero.
4580 rcutorture.irqreader= [KNL]
4581 Run RCU readers from irq handlers, or, more
4582 accurately, from a timer handler. Not all RCU
4583 flavors take kindly to this sort of thing.
4585 rcutorture.leakpointer= [KNL]
4586 Leak an RCU-protected pointer out of the reader.
4587 This can of course result in splats, and is
4588 intended to test the ability of things like
4589 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect
4592 rcutorture.n_barrier_cbs= [KNL]
4593 Set callbacks/threads for rcu_barrier() testing.
4595 rcutorture.nfakewriters= [KNL]
4596 Set number of concurrent RCU writers. These just
4597 stress RCU, they don't participate in the actual
4598 test, hence the "fake".
4600 rcutorture.nocbs_nthreads= [KNL]
4601 Set number of RCU callback-offload togglers.
4602 Zero (the default) disables toggling.
4604 rcutorture.nocbs_toggle= [KNL]
4605 Set the delay in milliseconds between successive
4606 callback-offload toggling attempts.
4608 rcutorture.nreaders= [KNL]
4609 Set number of RCU readers. The value -1 selects
4610 N-1, where N is the number of CPUs. A value
4611 "n" less than -1 selects N-n-2, where N is again
4612 the number of CPUs. For example, -2 selects N
4613 (the number of CPUs), -3 selects N+1, and so on.
4615 rcutorture.object_debug= [KNL]
4616 Enable debug-object double-call_rcu() testing.
4618 rcutorture.onoff_holdoff= [KNL]
4619 Set time (s) after boot for CPU-hotplug testing.
4621 rcutorture.onoff_interval= [KNL]
4622 Set time (jiffies) between CPU-hotplug operations,
4623 or zero to disable CPU-hotplug testing.
4625 rcutorture.read_exit= [KNL]
4626 Set the number of read-then-exit kthreads used
4627 to test the interaction of RCU updaters and
4628 task-exit processing.
4630 rcutorture.read_exit_burst= [KNL]
4631 The number of times in a given read-then-exit
4632 episode that a set of read-then-exit kthreads
4635 rcutorture.read_exit_delay= [KNL]
4636 The delay, in seconds, between successive
4637 read-then-exit testing episodes.
4639 rcutorture.shuffle_interval= [KNL]
4640 Set task-shuffle interval (s). Shuffling tasks
4641 allows some CPUs to go into dyntick-idle mode
4642 during the rcutorture test.
4644 rcutorture.shutdown_secs= [KNL]
4645 Set time (s) after boot system shutdown. This
4646 is useful for hands-off automated testing.
4648 rcutorture.stall_cpu= [KNL]
4649 Duration of CPU stall (s) to test RCU CPU stall
4650 warnings, zero to disable.
4652 rcutorture.stall_cpu_block= [KNL]
4653 Sleep while stalling if set. This will result
4654 in warnings from preemptible RCU in addition
4655 to any other stall-related activity.
4657 rcutorture.stall_cpu_holdoff= [KNL]
4658 Time to wait (s) after boot before inducing stall.
4660 rcutorture.stall_cpu_irqsoff= [KNL]
4661 Disable interrupts while stalling if set.
4663 rcutorture.stall_gp_kthread= [KNL]
4664 Duration (s) of forced sleep within RCU
4665 grace-period kthread to test RCU CPU stall
4666 warnings, zero to disable. If both stall_cpu
4667 and stall_gp_kthread are specified, the
4668 kthread is starved first, then the CPU.
4670 rcutorture.stat_interval= [KNL]
4671 Time (s) between statistics printk()s.
4673 rcutorture.stutter= [KNL]
4674 Time (s) to stutter testing, for example, specifying
4675 five seconds causes the test to run for five seconds,
4676 wait for five seconds, and so on. This tests RCU's
4677 ability to transition abruptly to and from idle.
4679 rcutorture.test_boost= [KNL]
4680 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
4681 "Maybe" means test if the RCU implementation
4682 under test support RCU priority boosting.
4684 rcutorture.test_boost_duration= [KNL]
4685 Duration (s) of each individual boost test.
4687 rcutorture.test_boost_interval= [KNL]
4688 Interval (s) between each boost test.
4690 rcutorture.test_no_idle_hz= [KNL]
4691 Test RCU's dyntick-idle handling. See also the
4692 rcutorture.shuffle_interval parameter.
4694 rcutorture.torture_type= [KNL]
4695 Specify the RCU implementation to test.
4697 rcutorture.verbose= [KNL]
4698 Enable additional printk() statements.
4700 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
4701 Dump ftrace buffer after reporting RCU CPU
4704 rcupdate.rcu_cpu_stall_suppress= [KNL]
4705 Suppress RCU CPU stall warning messages.
4707 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL]
4708 Suppress RCU CPU stall warning messages and
4709 rcutorture writer stall warnings that occur
4710 during early boot, that is, during the time
4711 before the init task is spawned.
4713 rcupdate.rcu_cpu_stall_timeout= [KNL]
4714 Set timeout for RCU CPU stall warning messages.
4716 rcupdate.rcu_expedited= [KNL]
4717 Use expedited grace-period primitives, for
4718 example, synchronize_rcu_expedited() instead
4719 of synchronize_rcu(). This reduces latency,
4720 but can increase CPU utilization, degrade
4721 real-time latency, and degrade energy efficiency.
4722 No effect on CONFIG_TINY_RCU kernels.
4724 rcupdate.rcu_normal= [KNL]
4725 Use only normal grace-period primitives,
4726 for example, synchronize_rcu() instead of
4727 synchronize_rcu_expedited(). This improves
4728 real-time latency, CPU utilization, and
4729 energy efficiency, but can expose users to
4730 increased grace-period latency. This parameter
4731 overrides rcupdate.rcu_expedited. No effect on
4732 CONFIG_TINY_RCU kernels.
4734 rcupdate.rcu_normal_after_boot= [KNL]
4735 Once boot has completed (that is, after
4736 rcu_end_inkernel_boot() has been invoked), use
4737 only normal grace-period primitives. No effect
4738 on CONFIG_TINY_RCU kernels.
4740 But note that CONFIG_PREEMPT_RT=y kernels enables
4741 this kernel boot parameter, forcibly setting
4742 it to the value one, that is, converting any
4743 post-boot attempt at an expedited RCU grace
4744 period to instead use normal non-expedited
4745 grace-period processing.
4747 rcupdate.rcu_task_ipi_delay= [KNL]
4748 Set time in jiffies during which RCU tasks will
4749 avoid sending IPIs, starting with the beginning
4750 of a given grace period. Setting a large
4751 number avoids disturbing real-time workloads,
4752 but lengthens grace periods.
4754 rcupdate.rcu_task_stall_timeout= [KNL]
4755 Set timeout in jiffies for RCU task stall warning
4756 messages. Disable with a value less than or equal
4759 rcupdate.rcu_self_test= [KNL]
4760 Run the RCU early boot self tests
4764 Run specified binary instead of /init from the ramdisk,
4765 used for early userspace startup. See initrd.
4768 force - Override the decision by the kernel to hide the
4769 advertisement of RDRAND support (this affects
4770 certain AMD processors because of buggy BIOS
4771 support, specifically around the suspend/resume
4775 Turn on/off individual RDT features. List is:
4776 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
4778 E.g. to turn on cmt and turn off mba use:
4782 Format (x86 or x86_64):
4783 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] \
4785 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
4787 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
4788 (prefix with 'panic_' to set mode for panic
4790 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
4791 reboot_force is either force or not specified,
4792 reboot_cpu is s[mp]#### with #### being the processor
4793 to be used for rebooting.
4795 refscale.holdoff= [KNL]
4796 Set test-start holdoff period. The purpose of
4797 this parameter is to delay the start of the
4798 test until boot completes in order to avoid
4801 refscale.loops= [KNL]
4802 Set the number of loops over the synchronization
4803 primitive under test. Increasing this number
4804 reduces noise due to loop start/end overhead,
4805 but the default has already reduced the per-pass
4806 noise to a handful of picoseconds on ca. 2020
4809 refscale.nreaders= [KNL]
4810 Set number of readers. The default value of -1
4811 selects N, where N is roughly 75% of the number
4812 of CPUs. A value of zero is an interesting choice.
4814 refscale.nruns= [KNL]
4815 Set number of runs, each of which is dumped onto
4818 refscale.readdelay= [KNL]
4819 Set the read-side critical-section duration,
4820 measured in microseconds.
4822 refscale.scale_type= [KNL]
4823 Specify the read-protection implementation to test.
4825 refscale.shutdown= [KNL]
4826 Shut down the system at the end of the performance
4827 test. This defaults to 1 (shut it down) when
4828 refscale is built into the kernel and to 0 (leave
4829 it running) when refscale is built as a module.
4831 refscale.verbose= [KNL]
4832 Enable additional printk() statements.
4834 refscale.verbose_batched= [KNL]
4835 Batch the additional printk() statements. If zero
4836 (the default) or negative, print everything. Otherwise,
4837 print every Nth verbose statement, where N is the value
4841 [KNL, SMP] Set scheduler's default relax_domain_level.
4842 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
4844 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
4845 Format: <base1>,<size1>[,<base2>,<size2>,...]
4846 Reserve I/O ports or memory so the kernel won't use
4847 them. If <base> is less than 0x10000, the region
4848 is assumed to be I/O ports; otherwise it is memory.
4850 reservetop= [X86-32]
4852 Reserves a hole at the top of the kernel virtual
4855 reset_devices [KNL] Force drivers to reset the underlying device
4856 during initialization.
4859 Specify the partition device for software suspend
4861 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
4863 resume_offset= [SWSUSP]
4864 Specify the offset from the beginning of the partition
4865 given by "resume=" at which the swap header is located,
4866 in <PAGE_SIZE> units (needed only for swap files).
4867 See Documentation/power/swsusp-and-swap-files.rst
4869 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
4870 read the resume files
4872 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
4873 Useful for devices that are detected asynchronously
4874 (e.g. USB and MMC devices).
4876 hibernate= [HIBERNATION]
4877 noresume Don't check if there's a hibernation image
4878 present during boot.
4879 nocompress Don't compress/decompress hibernation images.
4880 no Disable hibernation and resume.
4881 protect_image Turn on image protection during restoration
4882 (that will set all pages holding image data
4883 during restoration read-only).
4885 retain_initrd [RAM] Keep initrd memory after extraction
4887 rfkill.default_state=
4888 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
4889 etc. communication is blocked by default.
4892 rfkill.master_switch_mode=
4893 0 The "airplane mode" button does nothing.
4894 1 The "airplane mode" button toggles between everything
4895 blocked and the previous configuration.
4896 2 The "airplane mode" button toggles between everything
4897 blocked and everything unblocked.
4899 rhash_entries= [KNL,NET]
4900 Set number of hash buckets for route cache
4903 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
4906 ro [KNL] Mount root device read-only on boot
4909 on Mark read-only kernel memory as read-only (default).
4910 off Leave read-only kernel memory writable for debugging.
4913 Enable the uart passthrough on the designated usb port
4914 on Rockchip SoCs. When active, the signals of the
4915 debug-uart get routed to the D+ and D- pins of the usb
4916 port and the regular usb controller gets disabled.
4918 root= [KNL] Root filesystem
4919 See name_to_dev_t comment in init/do_mounts.c.
4921 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
4922 mount the root filesystem
4924 rootflags= [KNL] Set root filesystem mount option string
4926 rootfstype= [KNL] Set root filesystem type
4928 rootwait [KNL] Wait (indefinitely) for root device to show up.
4929 Useful for devices that are detected asynchronously
4930 (e.g. USB and MMC devices).
4932 rproc_mem=nn[KMG][@address]
4933 [KNL,ARM,CMA] Remoteproc physical memory block.
4934 Memory area to be used by remote processor image,
4937 rw [KNL] Mount root device read-write on boot
4939 S [KNL] Run init in single mode
4941 s390_iommu= [HW,S390]
4942 Set s390 IOTLB flushing mode
4944 With strict flushing every unmap operation will result in
4945 an IOTLB flush. Default is lazy flushing before reuse,
4949 See drivers/net/irda/sa1100_ir.c.
4951 sbni= [NET] Granch SBNI12 leased line adapter
4953 sched_verbose [KNL] Enables verbose scheduler debug messages.
4955 schedstats= [KNL,X86] Enable or disable scheduled statistics.
4956 Allowed values are enable and disable. This feature
4957 incurs a small amount of overhead in the scheduler
4958 but is useful for debugging and performance tuning.
4960 sched_thermal_decay_shift=
4961 [KNL, SMP] Set a decay shift for scheduler thermal
4962 pressure signal. Thermal pressure signal follows the
4963 default decay period of other scheduler pelt
4964 signals(usually 32 ms but configurable). Setting
4965 sched_thermal_decay_shift will left shift the decay
4966 period for the thermal pressure signal by the shift
4968 i.e. with the default pelt decay period of 32 ms
4969 sched_thermal_decay_shift thermal pressure decay pr
4973 Format: integer between 0 and 10
4976 scftorture.holdoff= [KNL]
4977 Number of seconds to hold off before starting
4978 test. Defaults to zero for module insertion and
4979 to 10 seconds for built-in smp_call_function()
4982 scftorture.longwait= [KNL]
4983 Request ridiculously long waits randomly selected
4984 up to the chosen limit in seconds. Zero (the
4985 default) disables this feature. Please note
4986 that requesting even small non-zero numbers of
4987 seconds can result in RCU CPU stall warnings,
4988 softlockup complaints, and so on.
4990 scftorture.nthreads= [KNL]
4991 Number of kthreads to spawn to invoke the
4992 smp_call_function() family of functions.
4993 The default of -1 specifies a number of kthreads
4994 equal to the number of CPUs.
4996 scftorture.onoff_holdoff= [KNL]
4997 Number seconds to wait after the start of the
4998 test before initiating CPU-hotplug operations.
5000 scftorture.onoff_interval= [KNL]
5001 Number seconds to wait between successive
5002 CPU-hotplug operations. Specifying zero (which
5003 is the default) disables CPU-hotplug operations.
5005 scftorture.shutdown_secs= [KNL]
5006 The number of seconds following the start of the
5007 test after which to shut down the system. The
5008 default of zero avoids shutting down the system.
5009 Non-zero values are useful for automated tests.
5011 scftorture.stat_interval= [KNL]
5012 The number of seconds between outputting the
5013 current test statistics to the console. A value
5014 of zero disables statistics output.
5016 scftorture.stutter_cpus= [KNL]
5017 The number of jiffies to wait between each change
5018 to the set of CPUs under test.
5020 scftorture.use_cpus_read_lock= [KNL]
5021 Use use_cpus_read_lock() instead of the default
5022 preempt_disable() to disable CPU hotplug
5023 while invoking one of the smp_call_function*()
5026 scftorture.verbose= [KNL]
5027 Enable additional printk() statements.
5029 scftorture.weight_single= [KNL]
5030 The probability weighting to use for the
5031 smp_call_function_single() function with a zero
5032 "wait" parameter. A value of -1 selects the
5033 default if all other weights are -1. However,
5034 if at least one weight has some other value, a
5035 value of -1 will instead select a weight of zero.
5037 scftorture.weight_single_wait= [KNL]
5038 The probability weighting to use for the
5039 smp_call_function_single() function with a
5040 non-zero "wait" parameter. See weight_single.
5042 scftorture.weight_many= [KNL]
5043 The probability weighting to use for the
5044 smp_call_function_many() function with a zero
5045 "wait" parameter. See weight_single.
5046 Note well that setting a high probability for
5047 this weighting can place serious IPI load
5050 scftorture.weight_many_wait= [KNL]
5051 The probability weighting to use for the
5052 smp_call_function_many() function with a
5053 non-zero "wait" parameter. See weight_single
5056 scftorture.weight_all= [KNL]
5057 The probability weighting to use for the
5058 smp_call_function_all() function with a zero
5059 "wait" parameter. See weight_single and
5062 scftorture.weight_all_wait= [KNL]
5063 The probability weighting to use for the
5064 smp_call_function_all() function with a
5065 non-zero "wait" parameter. See weight_single
5068 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
5069 xtime_lock contention on larger systems, and/or RCU lock
5070 contention on all systems with CONFIG_MAXSMP set.
5071 Format: { "0" | "1" }
5072 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
5074 Note: increases power consumption, thus should only be
5075 enabled if running jitter sensitive (HPC/RT) workloads.
5077 security= [SECURITY] Choose a legacy "major" security module to
5078 enable at boot. This has been deprecated by the
5081 selinux= [SELINUX] Disable or enable SELinux at boot time.
5082 Format: { "0" | "1" }
5083 See security/selinux/Kconfig help text.
5088 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
5089 Format: { "0" | "1" }
5090 See security/apparmor/Kconfig help text
5093 Default value is set via kernel config option.
5095 serialnumber [BUGS=X86-32]
5098 Maximal number of shapers.
5106 Enable merging of slabs with similar size when the
5107 kernel is built without CONFIG_SLAB_MERGE_DEFAULT.
5110 Disable merging of slabs with similar size. May be
5111 necessary if there is some reason to distinguish
5112 allocs to different slabs, especially in hardened
5113 environments where the risk of heap overflows and
5114 layout control by attackers can usually be
5115 frustrated by disabling merging. This will reduce
5116 most of the exposure of a heap attack to a single
5117 cache (risks via metadata attacks are mostly
5118 unchanged). Debug options disable merging on their
5120 For more information see Documentation/vm/slub.rst.
5122 slab_max_order= [MM, SLAB]
5123 Determines the maximum allowed order for slabs.
5124 A high setting may cause OOMs due to memory
5125 fragmentation. Defaults to 1 for systems with
5126 more than 32MB of RAM, 0 otherwise.
5128 slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB]
5129 Enabling slub_debug allows one to determine the
5130 culprit if slab objects become corrupted. Enabling
5131 slub_debug can create guard zones around objects and
5132 may poison objects when not in use. Also tracks the
5133 last alloc / free. For more information see
5134 Documentation/vm/slub.rst.
5136 slub_max_order= [MM, SLUB]
5137 Determines the maximum allowed order for slabs.
5138 A high setting may cause OOMs due to memory
5139 fragmentation. For more information see
5140 Documentation/vm/slub.rst.
5142 slub_min_objects= [MM, SLUB]
5143 The minimum number of objects per slab. SLUB will
5144 increase the slab order up to slub_max_order to
5145 generate a sufficiently large slab able to contain
5146 the number of objects indicated. The higher the number
5147 of objects the smaller the overhead of tracking slabs
5148 and the less frequently locks need to be acquired.
5149 For more information see Documentation/vm/slub.rst.
5151 slub_min_order= [MM, SLUB]
5152 Determines the minimum page order for slabs. Must be
5153 lower than slub_max_order.
5154 For more information see Documentation/vm/slub.rst.
5156 slub_merge [MM, SLUB]
5157 Same with slab_merge.
5159 slub_nomerge [MM, SLUB]
5160 Same with slab_nomerge. This is supported for legacy.
5161 See slab_nomerge for more information.
5164 Format: <io1>[,<io2>[,...,<io8>]]
5166 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
5167 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
5168 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
5169 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
5170 smsc-ircc2.ircc_irq= [HW] IRQ line
5171 smsc-ircc2.ircc_dma= [HW] DMA channel
5172 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
5173 0: Toshiba Satellite 1800 (GP data pin select)
5174 1: Fast pin select (default)
5177 smt [KNL,S390] Set the maximum number of threads (logical
5178 CPUs) to use per physical CPU on systems capable of
5179 symmetric multithreading (SMT). Will be capped to the
5180 actual hardware limit.
5182 Default: -1 (no limit)
5185 [KNL] Should the soft-lockup detector generate panics.
5188 A value of 1 instructs the soft-lockup detector
5189 to panic the machine when a soft-lockup occurs. It is
5190 also controlled by the kernel.softlockup_panic sysctl
5191 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
5192 respective build-time switch to that functionality.
5194 softlockup_all_cpu_backtrace=
5195 [KNL] Should the soft-lockup detector generate
5196 backtraces on all cpus.
5199 sonypi.*= [HW] Sony Programmable I/O Control Device driver
5200 See Documentation/admin-guide/laptops/sonypi.rst
5202 spectre_v2= [X86] Control mitigation of Spectre variant 2
5203 (indirect branch speculation) vulnerability.
5204 The default operation protects the kernel from
5207 on - unconditionally enable, implies
5209 off - unconditionally disable, implies
5211 auto - kernel detects whether your CPU model is
5214 Selecting 'on' will, and 'auto' may, choose a
5215 mitigation method at run time according to the
5216 CPU, the available microcode, the setting of the
5217 CONFIG_RETPOLINE configuration option, and the
5218 compiler with which the kernel was built.
5220 Selecting 'on' will also enable the mitigation
5221 against user space to user space task attacks.
5223 Selecting 'off' will disable both the kernel and
5224 the user space protections.
5226 Specific mitigations can also be selected manually:
5228 retpoline - replace indirect branches
5229 retpoline,generic - google's original retpoline
5230 retpoline,amd - AMD-specific minimal thunk
5232 Not specifying this option is equivalent to
5236 [X86] Control mitigation of Spectre variant 2
5237 (indirect branch speculation) vulnerability between
5240 on - Unconditionally enable mitigations. Is
5241 enforced by spectre_v2=on
5243 off - Unconditionally disable mitigations. Is
5244 enforced by spectre_v2=off
5246 prctl - Indirect branch speculation is enabled,
5247 but mitigation can be enabled via prctl
5248 per thread. The mitigation control state
5249 is inherited on fork.
5252 - Like "prctl" above, but only STIBP is
5253 controlled per thread. IBPB is issued
5254 always when switching between different user
5258 - Same as "prctl" above, but all seccomp
5259 threads will enable the mitigation unless
5260 they explicitly opt out.
5263 - Like "seccomp" above, but only STIBP is
5264 controlled per thread. IBPB is issued
5265 always when switching between different
5266 user space processes.
5268 auto - Kernel selects the mitigation depending on
5269 the available CPU features and vulnerability.
5272 If CONFIG_SECCOMP=y then "seccomp", otherwise "prctl"
5274 Not specifying this option is equivalent to
5275 spectre_v2_user=auto.
5277 spec_store_bypass_disable=
5278 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
5279 (Speculative Store Bypass vulnerability)
5281 Certain CPUs are vulnerable to an exploit against a
5282 a common industry wide performance optimization known
5283 as "Speculative Store Bypass" in which recent stores
5284 to the same memory location may not be observed by
5285 later loads during speculative execution. The idea
5286 is that such stores are unlikely and that they can
5287 be detected prior to instruction retirement at the
5288 end of a particular speculation execution window.
5290 In vulnerable processors, the speculatively forwarded
5291 store can be used in a cache side channel attack, for
5292 example to read memory to which the attacker does not
5293 directly have access (e.g. inside sandboxed code).
5295 This parameter controls whether the Speculative Store
5296 Bypass optimization is used.
5298 On x86 the options are:
5300 on - Unconditionally disable Speculative Store Bypass
5301 off - Unconditionally enable Speculative Store Bypass
5302 auto - Kernel detects whether the CPU model contains an
5303 implementation of Speculative Store Bypass and
5304 picks the most appropriate mitigation. If the
5305 CPU is not vulnerable, "off" is selected. If the
5306 CPU is vulnerable the default mitigation is
5307 architecture and Kconfig dependent. See below.
5308 prctl - Control Speculative Store Bypass per thread
5309 via prctl. Speculative Store Bypass is enabled
5310 for a process by default. The state of the control
5311 is inherited on fork.
5312 seccomp - Same as "prctl" above, but all seccomp threads
5313 will disable SSB unless they explicitly opt out.
5315 Default mitigations:
5316 X86: If CONFIG_SECCOMP=y "seccomp", otherwise "prctl"
5318 On powerpc the options are:
5320 on,auto - On Power8 and Power9 insert a store-forwarding
5321 barrier on kernel entry and exit. On Power7
5322 perform a software flush on kernel entry and
5326 Not specifying this option is equivalent to
5327 spec_store_bypass_disable=auto.
5329 spia_io_base= [HW,MTD]
5335 [X86] Enable split lock detection or bus lock detection
5337 When enabled (and if hardware support is present), atomic
5338 instructions that access data across cache line
5339 boundaries will result in an alignment check exception
5340 for split lock detection or a debug exception for
5345 warn - the kernel will emit rate-limited warnings
5346 about applications triggering the #AC
5347 exception or the #DB exception. This mode is
5348 the default on CPUs that support split lock
5349 detection or bus lock detection. Default
5350 behavior is by #AC if both features are
5351 enabled in hardware.
5353 fatal - the kernel will send SIGBUS to applications
5354 that trigger the #AC exception or the #DB
5355 exception. Default behavior is by #AC if
5356 both features are enabled in hardware.
5359 Set system wide rate limit to N bus locks
5360 per second for bus lock detection.
5363 N/A for split lock detection.
5366 If an #AC exception is hit in the kernel or in
5367 firmware (i.e. not while executing in user mode)
5368 the kernel will oops in either "warn" or "fatal"
5371 #DB exception for bus lock is triggered only when
5375 Control the Special Register Buffer Data Sampling
5378 Certain CPUs are vulnerable to an MDS-like
5379 exploit which can leak bits from the random
5382 By default, this issue is mitigated by
5383 microcode. However, the microcode fix can cause
5384 the RDRAND and RDSEED instructions to become
5385 much slower. Among other effects, this will
5386 result in reduced throughput from /dev/urandom.
5388 The microcode mitigation can be disabled with
5389 the following option:
5391 off: Disable mitigation and remove
5392 performance impact to RDRAND and RDSEED
5394 srcutree.counter_wrap_check [KNL]
5395 Specifies how frequently to check for
5396 grace-period sequence counter wrap for the
5397 srcu_data structure's ->srcu_gp_seq_needed field.
5398 The greater the number of bits set in this kernel
5399 parameter, the less frequently counter wrap will
5400 be checked for. Note that the bottom two bits
5403 srcutree.exp_holdoff [KNL]
5404 Specifies how many nanoseconds must elapse
5405 since the end of the last SRCU grace period for
5406 a given srcu_struct until the next normal SRCU
5407 grace period will be considered for automatic
5408 expediting. Set to zero to disable automatic
5412 Speculative Store Bypass Disable control
5414 On CPUs that are vulnerable to the Speculative
5415 Store Bypass vulnerability and offer a
5416 firmware based mitigation, this parameter
5417 indicates how the mitigation should be used:
5419 force-on: Unconditionally enable mitigation for
5420 for both kernel and userspace
5421 force-off: Unconditionally disable mitigation for
5422 for both kernel and userspace
5423 kernel: Always enable mitigation in the
5424 kernel, and offer a prctl interface
5425 to allow userspace to register its
5426 interest in being mitigated too.
5428 stack_guard_gap= [MM]
5429 override the default stack gap protection. The value
5430 is in page units and it defines how many pages prior
5431 to (for stacks growing down) resp. after (for stacks
5432 growing up) the main stack are reserved for no other
5433 mapping. Default value is 256 pages.
5435 stack_depot_disable= [KNL]
5436 Setting this to true through kernel command line will
5437 disable the stack depot thereby saving the static memory
5438 consumed by the stack hash table. By default this is set
5442 Enabled the stack tracer on boot up.
5444 stacktrace_filter=[function-list]
5445 [FTRACE] Limit the functions that the stack tracer
5446 will trace at boot up. function-list is a comma-separated
5447 list of functions. This list can be changed at run
5448 time by the stack_trace_filter file in the debugfs
5449 tracing directory. Note, this enables stack tracing
5450 and the stacktrace above is not needed.
5454 Set the STI (builtin display/keyboard on the HP-PARISC
5455 machines) console (graphic card) which should be used
5456 as the initial boot-console.
5457 See also comment in drivers/video/console/sticore.c.
5460 See comment in drivers/video/console/sticore.c.
5463 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
5465 sunrpc.min_resvport=
5466 sunrpc.max_resvport=
5468 SunRPC servers often require that client requests
5469 originate from a privileged port (i.e. a port in the
5470 range 0 < portnr < 1024).
5471 An administrator who wishes to reserve some of these
5472 ports for other uses may adjust the range that the
5473 kernel's sunrpc client considers to be privileged
5474 using these two parameters to set the minimum and
5475 maximum port values.
5477 sunrpc.svc_rpc_per_connection_limit=
5479 Limit the number of requests that the server will
5480 process in parallel from a single connection.
5481 The default value is 0 (no limit).
5485 Control how the NFS server code allocates CPUs to
5486 service thread pools. Depending on how many NICs
5487 you have and where their interrupts are bound, this
5488 option will affect which CPUs will do NFS serving.
5489 Note: this parameter cannot be changed while the
5490 NFS server is running.
5492 auto the server chooses an appropriate mode
5493 automatically using heuristics
5494 global a single global pool contains all CPUs
5495 percpu one pool for each CPU
5496 pernode one pool for each NUMA node (equivalent
5497 to global on non-NUMA machines)
5499 sunrpc.tcp_slot_table_entries=
5500 sunrpc.udp_slot_table_entries=
5502 Sets the upper limit on the number of simultaneous
5503 RPC calls that can be sent from the client to a
5504 server. Increasing these values may allow you to
5505 improve throughput, but will also increase the
5506 amount of memory reserved for use by the client.
5508 suspend.pm_test_delay=
5510 Sets the number of seconds to remain in a suspend test
5511 mode before resuming the system (see
5512 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
5513 is set. Default value is 5.
5516 Format: { on | off | y | n | 1 | 0 }
5517 This parameter controls use of the Protected
5518 Execution Facility on pSeries.
5521 [KNL] Enable accounting of swap in memory resource
5522 controller if no parameter or 1 is given or disable
5523 it if 0 is given (See Documentation/admin-guide/cgroup-v1/memory.rst)
5525 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
5526 Format: { <int> | force | noforce }
5527 <int> -- Number of I/O TLB slabs
5528 force -- force using of bounce buffers even if they
5529 wouldn't be automatically used by the kernel
5530 noforce -- Never use bounce buffers (for debugging)
5535 Set a sysctl parameter, right before loading the init
5536 process, as if the value was written to the respective
5537 /proc/sys/... file. Both '.' and '/' are recognized as
5538 separators. Unrecognized parameters and invalid values
5539 are reported in the kernel log. Sysctls registered
5540 later by a loaded module cannot be set this way.
5541 Example: sysctl.vm.swappiness=40
5543 sysfs.deprecated=0|1 [KNL]
5544 Enable/disable old style sysfs layout for old udev
5545 on older distributions. When this option is enabled
5546 very new udev will not work anymore. When this option
5547 is disabled (or CONFIG_SYSFS_DEPRECATED not compiled)
5548 in older udev will not work anymore.
5549 Default depends on CONFIG_SYSFS_DEPRECATED_V2 set in
5550 the kernel configuration.
5552 sysrq_always_enabled
5554 Ignore sysrq setting - this boot parameter will
5555 neutralize any effect of /proc/sys/kernel/sysrq.
5556 Useful for debugging.
5558 tcpmhash_entries= [KNL,NET]
5559 Set the number of tcp_metrics_hash slots.
5560 Default value is 8192 or 16384 depending on total
5561 ram pages. This is used to specify the TCP metrics
5562 cache size. See Documentation/networking/ip-sysctl.rst
5563 "tcp_no_metrics_save" section for more details.
5567 test_suspend= [SUSPEND][,N]
5568 Specify "mem" (for Suspend-to-RAM) or "standby" (for
5569 standby suspend) or "freeze" (for suspend type freeze)
5570 as the system sleep state during system startup with
5571 the optional capability to repeat N number of times.
5572 The system is woken from this state using a
5573 wakeup-capable RTC alarm.
5575 thash_entries= [KNL,NET]
5576 Set number of hash buckets for TCP connection
5578 thermal.act= [HW,ACPI]
5579 -1: disable all active trip points in all thermal zones
5580 <degrees C>: override all lowest active trip points
5582 thermal.crt= [HW,ACPI]
5583 -1: disable all critical trip points in all thermal zones
5584 <degrees C>: override all critical trip points
5586 thermal.nocrt= [HW,ACPI]
5587 Set to disable actions on ACPI thermal zone
5588 critical and hot trip points.
5590 thermal.off= [HW,ACPI]
5591 1: disable ACPI thermal control
5593 thermal.psv= [HW,ACPI]
5594 -1: disable all passive trip points
5595 <degrees C>: override all passive trip points to this
5598 thermal.tzp= [HW,ACPI]
5599 Specify global default ACPI thermal zone polling rate
5600 <deci-seconds>: poll all this frequency
5601 0: no polling (default)
5604 Force threading of all interrupt handlers except those
5605 marked explicitly IRQF_NO_THREAD.
5609 Specify if the kernel should make use of the cpu
5610 topology information if the hardware supports this.
5611 The scheduler will make use of this information and
5612 e.g. base its process migration decisions on it.
5615 topology_updates= [KNL, PPC, NUMA]
5617 Specify if the kernel should ignore (off)
5618 topology updates sent by the hypervisor to this
5621 torture.disable_onoff_at_boot= [KNL]
5622 Prevent the CPU-hotplug component of torturing
5623 until after init has spawned.
5625 torture.ftrace_dump_at_shutdown= [KNL]
5626 Dump the ftrace buffer at torture-test shutdown,
5627 even if there were no errors. This can be a
5628 very costly operation when many torture tests
5629 are running concurrently, especially on systems
5630 with rotating-rust storage.
5632 torture.verbose_sleep_frequency= [KNL]
5633 Specifies how many verbose printk()s should be
5634 emitted between each sleep. The default of zero
5635 disables verbose-printk() sleeping.
5637 torture.verbose_sleep_duration= [KNL]
5638 Duration of each verbose-printk() sleep in jiffies.
5642 tpm_suspend_pcr=[HW,TPM]
5643 Format: integer pcr id
5644 Specify that at suspend time, the tpm driver
5645 should extend the specified pcr with zeros,
5646 as a workaround for some chips which fail to
5647 flush the last written pcr on TPM_SaveState.
5648 This will guarantee that all the other pcrs
5651 trace_buf_size=nn[KMG]
5652 [FTRACE] will set tracing buffer size on each cpu.
5654 trace_event=[event-list]
5655 [FTRACE] Set and start specified trace events in order
5656 to facilitate early boot debugging. The event-list is a
5657 comma-separated list of trace events to enable. See
5658 also Documentation/trace/events.rst
5660 trace_options=[option-list]
5661 [FTRACE] Enable or disable tracer options at boot.
5662 The option-list is a comma delimited list of options
5663 that can be enabled or disabled just as if you were
5664 to echo the option name into
5666 /sys/kernel/debug/tracing/trace_options
5668 For example, to enable stacktrace option (to dump the
5669 stack trace of each event), add to the command line:
5671 trace_options=stacktrace
5673 See also Documentation/trace/ftrace.rst "trace options"
5677 Have the tracepoints sent to printk as well as the
5678 tracing ring buffer. This is useful for early boot up
5679 where the system hangs or reboots and does not give the
5680 option for reading the tracing buffer or performing a
5681 ftrace_dump_on_oops.
5683 To turn off having tracepoints sent to printk,
5684 echo 0 > /proc/sys/kernel/tracepoint_printk
5685 Note, echoing 1 into this file without the
5686 tracepoint_printk kernel cmdline option has no effect.
5688 The tp_printk_stop_on_boot (see below) can also be used
5689 to stop the printing of events to console at
5694 Having tracepoints sent to printk() and activating high
5695 frequency tracepoints such as irq or sched, can cause
5696 the system to live lock.
5698 tp_printk_stop_on_boot[FTRACE]
5699 When tp_printk (above) is set, it can cause a lot of noise
5700 on the console. It may be useful to only include the
5701 printing of events during boot up, as user space may
5702 make the system inoperable.
5704 This command line option will stop the printing of events
5705 to console at the late_initcall_sync() time frame.
5708 [FTRACE] enable this option to disable tracing when a
5709 warning is hit. This turns off "tracing_on". Tracing can
5710 be enabled again by echoing '1' into the "tracing_on"
5711 file located in /sys/kernel/debug/tracing/
5713 This option is useful, as it disables the trace before
5714 the WARNING dump is called, which prevents the trace to
5715 be filled with content caused by the warning output.
5717 This option can also be set at run time via the sysctl
5718 option: kernel/traceoff_on_warning
5720 transparent_hugepage=
5722 Format: [always|madvise|never]
5723 Can be used to control the default behavior of the system
5724 with respect to transparent hugepages.
5725 See Documentation/admin-guide/mm/transhuge.rst
5728 trusted.source= [KEYS]
5730 This parameter identifies the trust source as a backend
5731 for trusted keys implementation. Supported trust
5735 If not specified then it defaults to iterating through
5736 the trust source list starting with TPM and assigns the
5737 first trust source as a backend which is initialized
5738 successfully during iteration.
5740 tsc= Disable clocksource stability checks for TSC.
5742 [x86] reliable: mark tsc clocksource as reliable, this
5743 disables clocksource verification at runtime, as well
5744 as the stability checks done at bootup. Used to enable
5745 high-resolution timer mode on older hardware, and in
5746 virtualized environment.
5747 [x86] noirqtime: Do not use TSC to do irq accounting.
5748 Used to run time disable IRQ_TIME_ACCOUNTING on any
5749 platforms where RDTSC is slow and this accounting
5751 [x86] unstable: mark the TSC clocksource as unstable, this
5752 marks the TSC unconditionally unstable at bootup and
5753 avoids any further wobbles once the TSC watchdog notices.
5754 [x86] nowatchdog: disable clocksource watchdog. Used
5755 in situations with strict latency requirements (where
5756 interruptions from clocksource watchdog are not
5759 tsc_early_khz= [X86] Skip early TSC calibration and use the given
5760 value instead. Useful when the early TSC frequency discovery
5761 procedure is not reliable, such as on overclocked systems
5762 with CPUID.16h support and partial CPUID.15h support.
5763 Format: <unsigned int>
5765 tsx= [X86] Control Transactional Synchronization
5766 Extensions (TSX) feature in Intel processors that
5767 support TSX control.
5769 This parameter controls the TSX feature. The options are:
5771 on - Enable TSX on the system. Although there are
5772 mitigations for all known security vulnerabilities,
5773 TSX has been known to be an accelerator for
5774 several previous speculation-related CVEs, and
5775 so there may be unknown security risks associated
5776 with leaving it enabled.
5778 off - Disable TSX on the system. (Note that this
5779 option takes effect only on newer CPUs which are
5780 not vulnerable to MDS, i.e., have
5781 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
5782 the new IA32_TSX_CTRL MSR through a microcode
5783 update. This new MSR allows for the reliable
5784 deactivation of the TSX functionality.)
5786 auto - Disable TSX if X86_BUG_TAA is present,
5787 otherwise enable TSX on the system.
5789 Not specifying this option is equivalent to tsx=off.
5791 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5794 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
5795 Abort (TAA) vulnerability.
5797 Similar to Micro-architectural Data Sampling (MDS)
5798 certain CPUs that support Transactional
5799 Synchronization Extensions (TSX) are vulnerable to an
5800 exploit against CPU internal buffers which can forward
5801 information to a disclosure gadget under certain
5804 In vulnerable processors, the speculatively forwarded
5805 data can be used in a cache side channel attack, to
5806 access data to which the attacker does not have direct
5809 This parameter controls the TAA mitigation. The
5812 full - Enable TAA mitigation on vulnerable CPUs
5815 full,nosmt - Enable TAA mitigation and disable SMT on
5816 vulnerable CPUs. If TSX is disabled, SMT
5817 is not disabled because CPU is not
5818 vulnerable to cross-thread TAA attacks.
5819 off - Unconditionally disable TAA mitigation
5821 On MDS-affected machines, tsx_async_abort=off can be
5822 prevented by an active MDS mitigation as both vulnerabilities
5823 are mitigated with the same mechanism so in order to disable
5824 this mitigation, you need to specify mds=off too.
5826 Not specifying this option is equivalent to
5827 tsx_async_abort=full. On CPUs which are MDS affected
5828 and deploy MDS mitigation, TAA mitigation is not
5829 required and doesn't provide any additional
5833 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5835 turbografx.map[2|3]= [HW,JOY]
5836 TurboGraFX parallel port interface
5838 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
5839 See also Documentation/input/devices/joystick-parport.rst
5841 udbg-immortal [PPC] When debugging early kernel crashes that
5842 happen after console_init() and before a proper
5843 console driver takes over, this boot options might
5844 help "seeing" what's going on.
5846 uhash_entries= [KNL,NET]
5847 Set number of hash buckets for UDP/UDP-Lite connections
5850 [USB] Ignore overcurrent events (default N).
5851 Some badly-designed motherboards generate lots of
5852 bogus events, for ports that aren't wired to
5853 anything. Set this parameter to avoid log spamming.
5854 Note that genuine overcurrent events won't be
5858 [X86] Cause panic on unknown NMI.
5860 usbcore.authorized_default=
5861 [USB] Default USB device authorization:
5862 (default -1 = authorized except for wireless USB,
5863 0 = not authorized, 1 = authorized, 2 = authorized
5864 if device connected to internal port)
5866 usbcore.autosuspend=
5867 [USB] The autosuspend time delay (in seconds) used
5868 for newly-detected USB devices (default 2). This
5869 is the time required before an idle device will be
5870 autosuspended. Devices for which the delay is set
5871 to a negative value won't be autosuspended at all.
5873 usbcore.usbfs_snoop=
5874 [USB] Set to log all usbfs traffic (default 0 = off).
5876 usbcore.usbfs_snoop_max=
5877 [USB] Maximum number of bytes to snoop in each URB
5880 usbcore.blinkenlights=
5881 [USB] Set to cycle leds on hubs (default 0 = off).
5883 usbcore.old_scheme_first=
5884 [USB] Start with the old device initialization
5885 scheme (default 0 = off).
5887 usbcore.usbfs_memory_mb=
5888 [USB] Memory limit (in MB) for buffers allocated by
5889 usbfs (default = 16, 0 = max = 2047).
5891 usbcore.use_both_schemes=
5892 [USB] Try the other device initialization scheme
5893 if the first one fails (default 1 = enabled).
5895 usbcore.initial_descriptor_timeout=
5896 [USB] Specifies timeout for the initial 64-byte
5897 USB_REQ_GET_DESCRIPTOR request in milliseconds
5898 (default 5000 = 5.0 seconds).
5900 usbcore.nousb [USB] Disable the USB subsystem
5903 [USB] A list of quirk entries to augment the built-in
5904 usb core quirk list. List entries are separated by
5905 commas. Each entry has the form
5906 VendorID:ProductID:Flags. The IDs are 4-digit hex
5907 numbers and Flags is a set of letters. Each letter
5908 will change the built-in quirk; setting it if it is
5909 clear and clearing it if it is set. The letters have
5910 the following meanings:
5911 a = USB_QUIRK_STRING_FETCH_255 (string
5912 descriptors must not be fetched using
5914 b = USB_QUIRK_RESET_RESUME (device can't resume
5915 correctly so reset it instead);
5916 c = USB_QUIRK_NO_SET_INTF (device can't handle
5917 Set-Interface requests);
5918 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
5919 handle its Configuration or Interface
5921 e = USB_QUIRK_RESET (device can't be reset
5922 (e.g morph devices), don't use reset);
5923 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
5924 more interface descriptions than the
5925 bNumInterfaces count, and can't handle
5926 talking to these interfaces);
5927 g = USB_QUIRK_DELAY_INIT (device needs a pause
5928 during initialization, after we read
5929 the device descriptor);
5930 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
5931 high speed and super speed interrupt
5932 endpoints, the USB 2.0 and USB 3.0 spec
5933 require the interval in microframes (1
5934 microframe = 125 microseconds) to be
5935 calculated as interval = 2 ^
5937 Devices with this quirk report their
5938 bInterval as the result of this
5939 calculation instead of the exponent
5940 variable used in the calculation);
5941 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
5942 handle device_qualifier descriptor
5944 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
5945 generates spurious wakeup, ignore
5946 remote wakeup capability);
5947 k = USB_QUIRK_NO_LPM (device can't handle Link
5949 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
5950 (Device reports its bInterval as linear
5951 frames instead of the USB 2.0
5953 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
5954 to be disconnected before suspend to
5955 prevent spurious wakeup);
5956 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
5957 pause after every control message);
5958 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
5959 delay after resetting its port);
5960 Example: quirks=0781:5580:bk,0a5c:5834:gij
5963 [USBHID] The interval which mice are to be polled at.
5966 [USBHID] The interval which joysticks are to be polled at.
5969 [USBHID] The interval which keyboards are to be polled at.
5971 usb-storage.delay_use=
5972 [UMS] The delay in seconds before a new device is
5973 scanned for Logical Units (default 1).
5976 [UMS] A list of quirks entries to supplement or
5977 override the built-in unusual_devs list. List
5978 entries are separated by commas. Each entry has
5979 the form VID:PID:Flags where VID and PID are Vendor
5980 and Product ID values (4-digit hex numbers) and
5981 Flags is a set of characters, each corresponding
5982 to a common usb-storage quirk flag as follows:
5983 a = SANE_SENSE (collect more than 18 bytes
5984 of sense data, not on uas);
5985 b = BAD_SENSE (don't collect more than 18
5986 bytes of sense data, not on uas);
5987 c = FIX_CAPACITY (decrease the reported
5988 device capacity by one sector);
5989 d = NO_READ_DISC_INFO (don't use
5990 READ_DISC_INFO command, not on uas);
5991 e = NO_READ_CAPACITY_16 (don't use
5992 READ_CAPACITY_16 command);
5993 f = NO_REPORT_OPCODES (don't use report opcodes
5995 g = MAX_SECTORS_240 (don't transfer more than
5996 240 sectors at a time, uas only);
5997 h = CAPACITY_HEURISTICS (decrease the
5998 reported device capacity by one
5999 sector if the number is odd);
6000 i = IGNORE_DEVICE (don't bind to this
6002 j = NO_REPORT_LUNS (don't use report luns
6004 k = NO_SAME (do not use WRITE_SAME, uas only)
6005 l = NOT_LOCKABLE (don't try to lock and
6006 unlock ejectable media, not on uas);
6007 m = MAX_SECTORS_64 (don't transfer more
6008 than 64 sectors = 32 KB at a time,
6010 n = INITIAL_READ10 (force a retry of the
6011 initial READ(10) command, not on uas);
6012 o = CAPACITY_OK (accept the capacity
6013 reported by the device, not on uas);
6014 p = WRITE_CACHE (the device cache is ON
6015 by default, not on uas);
6016 r = IGNORE_RESIDUE (the device reports
6017 bogus residue values, not on uas);
6018 s = SINGLE_LUN (the device has only one
6020 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
6021 commands, uas only);
6022 u = IGNORE_UAS (don't bind to the uas driver);
6023 w = NO_WP_DETECT (don't test whether the
6024 medium is write-protected).
6025 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
6026 even if the device claims no cache,
6028 Example: quirks=0419:aaf5:rl,0421:0433:rc
6030 user_debug= [KNL,ARM]
6032 See arch/arm/Kconfig.debug help text.
6033 1 - undefined instruction events
6035 4 - invalid data aborts
6038 Example: user_debug=31
6041 [X86] Flags controlling user PTE allocations.
6043 nohigh = do not allocate PTE pages in
6044 HIGHMEM regardless of setting
6048 On X86_32, this is an alias for vdso32=. Otherwise:
6050 vdso=1: enable VDSO (the default)
6051 vdso=0: disable VDSO mapping
6053 vdso32= [X86] Control the 32-bit vDSO
6054 vdso32=1: enable 32-bit VDSO
6055 vdso32=0 or vdso32=2: disable 32-bit VDSO
6057 See the help text for CONFIG_COMPAT_VDSO for more
6058 details. If CONFIG_COMPAT_VDSO is set, the default is
6059 vdso32=0; otherwise, the default is vdso32=1.
6061 For compatibility with older kernels, vdso32=2 is an
6064 Try vdso32=0 if you encounter an error that says:
6065 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
6068 vector=percpu: enable percpu vector domain
6070 video= [FB] Frame buffer configuration
6071 See Documentation/fb/modedb.rst.
6073 video.brightness_switch_enabled= [0,1]
6074 If set to 1, on receiving an ACPI notify event
6075 generated by hotkey, video driver will adjust brightness
6076 level and then send out the event to user space through
6077 the allocated input device; If set to 0, video driver
6078 will only send out the event without touching backlight
6083 [VMMIO] Memory mapped virtio (platform) device.
6085 <size>@<baseaddr>:<irq>[:<id>]
6087 <size> := size (can use standard suffixes
6089 <baseaddr> := physical base address
6090 <irq> := interrupt number (as passed to
6092 <id> := (optional) platform device id
6094 virtio_mmio.device=1K@0x100b0000:48:7
6096 Can be used multiple times for multiple devices.
6098 vga= [BOOT,X86-32] Select a particular video mode
6099 See Documentation/x86/boot.rst and
6100 Documentation/admin-guide/svga.rst.
6101 Use vga=ask for menu.
6102 This is actually a boot loader parameter; the value is
6103 passed to the kernel using a special protocol.
6105 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
6106 May slow down system boot speed, especially when
6107 enabled on systems with a large amount of memory.
6108 All options are enabled by default, and this
6109 interface is meant to allow for selectively
6110 enabling or disabling specific virtual memory
6113 Available options are:
6114 P Enable page structure init time poisoning
6115 - Disable all of the above options
6117 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
6118 size of <nn>. This can be used to increase the
6119 minimum size (128MB on x86). It can also be used to
6120 decrease the size and leave more room for directly
6123 vmcp_cma=nn[MG] [KNL,S390]
6124 Sets the memory size reserved for contiguous memory
6125 allocations for the vmcp device driver.
6127 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
6130 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
6133 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
6137 Controls the behavior of vsyscalls (i.e. calls to
6138 fixed addresses of 0xffffffffff600x00 from legacy
6139 code). Most statically-linked binaries and older
6140 versions of glibc use these calls. Because these
6141 functions are at fixed addresses, they make nice
6142 targets for exploits that can control RIP.
6144 emulate [default] Vsyscalls turn into traps and are
6145 emulated reasonably safely. The vsyscall
6148 xonly Vsyscalls turn into traps and are
6149 emulated reasonably safely. The vsyscall
6150 page is not readable.
6152 none Vsyscalls don't work at all. This makes
6153 them quite hard to use for exploits but
6154 might break your system.
6156 vt.color= [VT] Default text color.
6157 Format: 0xYX, X = foreground, Y = background.
6158 Default: 0x07 = light gray on black.
6160 vt.cur_default= [VT] Default cursor shape.
6161 Format: 0xCCBBAA, where AA, BB, and CC are the same as
6162 the parameters of the <Esc>[?A;B;Cc escape sequence;
6163 see VGA-softcursor.txt. Default: 2 = underline.
6165 vt.default_blu= [VT]
6166 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
6167 Change the default blue palette of the console.
6168 This is a 16-member array composed of values
6171 vt.default_grn= [VT]
6172 Format: <green0>,<green1>,<green2>,...,<green15>
6173 Change the default green palette of the console.
6174 This is a 16-member array composed of values
6177 vt.default_red= [VT]
6178 Format: <red0>,<red1>,<red2>,...,<red15>
6179 Change the default red palette of the console.
6180 This is a 16-member array composed of values
6186 Set system-wide default UTF-8 mode for all tty's.
6187 Default is 1, i.e. UTF-8 mode is enabled for all
6188 newly opened terminals.
6190 vt.global_cursor_default=
6193 Set system-wide default for whether a cursor
6194 is shown on new VTs. Default is -1,
6195 i.e. cursors will be created by default unless
6196 overridden by individual drivers. 0 will hide
6197 cursors, 1 will display them.
6199 vt.italic= [VT] Default color for italic text; 0-15.
6202 vt.underline= [VT] Default color for underlined text; 0-15.
6205 watchdog timers [HW,WDT] For information on watchdog timers,
6206 see Documentation/watchdog/watchdog-parameters.rst
6207 or other driver-specific files in the
6208 Documentation/watchdog/ directory.
6212 Set the hard lockup detector stall duration
6213 threshold in seconds. The soft lockup detector
6214 threshold is set to twice the value. A value of 0
6215 disables both lockup detectors. Default is 10
6218 workqueue.watchdog_thresh=
6219 If CONFIG_WQ_WATCHDOG is configured, workqueue can
6220 warn stall conditions and dump internal state to
6221 help debugging. 0 disables workqueue stall
6222 detection; otherwise, it's the stall threshold
6223 duration in seconds. The default value is 30 and
6224 it can be updated at runtime by writing to the
6225 corresponding sysfs file.
6227 workqueue.disable_numa
6228 By default, all work items queued to unbound
6229 workqueues are affine to the NUMA nodes they're
6230 issued on, which results in better behavior in
6231 general. If NUMA affinity needs to be disabled for
6232 whatever reason, this option can be used. Note
6233 that this also can be controlled per-workqueue for
6234 workqueues visible under /sys/bus/workqueue/.
6236 workqueue.power_efficient
6237 Per-cpu workqueues are generally preferred because
6238 they show better performance thanks to cache
6239 locality; unfortunately, per-cpu workqueues tend to
6240 be more power hungry than unbound workqueues.
6242 Enabling this makes the per-cpu workqueues which
6243 were observed to contribute significantly to power
6244 consumption unbound, leading to measurably lower
6245 power usage at the cost of small performance
6248 The default value of this parameter is determined by
6249 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
6251 workqueue.debug_force_rr_cpu
6252 Workqueue used to implicitly guarantee that work
6253 items queued without explicit CPU specified are put
6254 on the local CPU. This guarantee is no longer true
6255 and while local CPU is still preferred work items
6256 may be put on foreign CPUs. This debug option
6257 forces round-robin CPU selection to flush out
6258 usages which depend on the now broken guarantee.
6259 When enabled, memory and cache locality will be
6262 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
6263 default x2apic cluster mode on platforms
6266 xen_512gb_limit [KNL,X86-64,XEN]
6267 Restricts the kernel running paravirtualized under Xen
6268 to use only up to 512 GB of RAM. The reason to do so is
6269 crash analysis tools and Xen tools for doing domain
6270 save/restore/migration must be enabled to handle larger
6273 xen_emul_unplug= [HW,X86,XEN]
6274 Unplug Xen emulated devices
6275 Format: [unplug0,][unplug1]
6276 ide-disks -- unplug primary master IDE devices
6277 aux-ide-disks -- unplug non-primary-master IDE devices
6278 nics -- unplug network devices
6279 all -- unplug all emulated devices (NICs and IDE disks)
6280 unnecessary -- unplugging emulated devices is
6281 unnecessary even if the host did not respond to
6283 never -- do not unplug even if version check succeeds
6285 xen_legacy_crash [X86,XEN]
6286 Crash from Xen panic notifier, without executing late
6287 panic() code such as dumping handler.
6289 xen_nopvspin [X86,XEN]
6290 Disables the qspinlock slowpath using Xen PV optimizations.
6291 This parameter is obsoleted by "nopvspin" parameter, which
6292 has equivalent effect for XEN platform.
6295 Disables the PV optimizations forcing the HVM guest to
6296 run as generic HVM guest with no PV drivers.
6297 This option is obsoleted by the "nopv" option, which
6298 has equivalent effect for XEN platform.
6300 xen_no_vector_callback
6301 [KNL,X86,XEN] Disable the vector callback for Xen
6302 event channel interrupts.
6304 xen_scrub_pages= [XEN]
6305 Boolean option to control scrubbing pages before giving them back
6306 to Xen, for use by other domains. Can be also changed at runtime
6307 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
6308 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
6310 xen_timer_slop= [X86-64,XEN]
6311 Set the timer slop (in nanoseconds) for the virtual Xen
6312 timers (default is 100000). This adjusts the minimum
6313 delta of virtualized Xen timers, where lower values
6314 improve timer resolution at the expense of processing
6315 more timer interrupts.
6317 xen.event_eoi_delay= [XEN]
6318 How long to delay EOI handling in case of event
6319 storms (jiffies). Default is 10.
6321 xen.event_loop_timeout= [XEN]
6322 After which time (jiffies) the event handling loop
6323 should start to delay EOI handling. Default is 2.
6325 xen.fifo_events= [XEN]
6326 Boolean parameter to disable using fifo event handling
6327 even if available. Normally fifo event handling is
6328 preferred over the 2-level event handling, as it is
6329 fairer and the number of possible event channels is
6330 much higher. Default is on (use fifo events).
6332 nopv= [X86,XEN,KVM,HYPER_V,VMWARE]
6333 Disables the PV optimizations forcing the guest to run
6334 as generic guest with no PV drivers. Currently support
6335 XEN HVM, KVM, HYPER_V and VMWARE guest.
6337 nopvspin [X86,XEN,KVM]
6338 Disables the qspinlock slow path using PV optimizations
6339 which allow the hypervisor to 'idle' the guest on lock
6342 xirc2ps_cs= [NET,PCMCIA]
6344 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
6347 By default on POWER9 and above, the kernel will
6348 natively use the XIVE interrupt controller. This option
6349 allows the fallback firmware mode to be used:
6351 off Fallback to firmware control of XIVE interrupt
6352 controller on both pseries and powernv
6353 platforms. Only useful on POWER9 and above.
6355 xhci-hcd.quirks [USB,KNL]
6356 A hex value specifying bitmask with supplemental xhci
6357 host controller quirks. Meaning of each bit can be
6358 consulted in header drivers/usb/host/xhci.h.
6361 Format: { early | on | rw | ro | off }
6362 Controls if xmon debugger is enabled. Default is off.
6363 Passing only "xmon" is equivalent to "xmon=early".
6364 early Call xmon as early as possible on boot; xmon
6365 debugger is called from setup_arch().
6366 on xmon debugger hooks will be installed so xmon
6367 is only called on a kernel crash. Default mode,
6368 i.e. either "ro" or "rw" mode, is controlled
6369 with CONFIG_XMON_DEFAULT_RO_MODE.
6370 rw xmon debugger hooks will be installed so xmon
6371 is called only on a kernel crash, mode is write,
6372 meaning SPR registers, memory and, other data
6373 can be written using xmon commands.
6374 ro same as "rw" option above but SPR registers,
6375 memory, and other data can't be written using
6377 off xmon is disabled.