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
305 amd_iommu_dump= [HW,X86-64]
306 Enable AMD IOMMU driver option to dump the ACPI table
307 for AMD IOMMU. With this option enabled, AMD IOMMU
308 driver will print ACPI tables for AMD IOMMU during
309 IOMMU initialization.
311 amd_iommu_intr= [HW,X86-64]
312 Specifies one of the following AMD IOMMU interrupt
314 legacy - Use legacy interrupt remapping mode.
315 vapic - Use virtual APIC mode, which allows IOMMU
316 to inject interrupts directly into guest.
317 This mode requires kvm-amd.avic=1.
318 (Default when IOMMU HW support is present.)
320 amijoy.map= [HW,JOY] Amiga joystick support
321 Map of devices attached to JOY0DAT and JOY1DAT
323 See also Documentation/input/joydev/joystick.rst
325 analog.map= [HW,JOY] Analog joystick and gamepad support
326 Specifies type or capabilities of an analog joystick
327 connected to one of 16 gameports
328 Format: <type1>,<type2>,..<type16>
331 Power management functions (SPARCstation-4/5 + deriv.)
333 Disable APC CPU standby support. SPARCstation-Fox does
334 not play well with APC CPU idle - disable it if you have
335 APC and your system crashes randomly.
337 apic= [APIC,X86] Advanced Programmable Interrupt Controller
338 Change the output verbosity while booting
339 Format: { quiet (default) | verbose | debug }
340 Change the amount of debugging information output
341 when initialising the APIC and IO-APIC components.
342 For X86-32, this can also be used to specify an APIC
344 Format: apic=driver_name
345 Examples: apic=bigsmp
347 apic_extnmi= [APIC,X86] External NMI delivery setting
348 Format: { bsp (default) | all | none }
349 bsp: External NMI is delivered only to CPU 0
350 all: External NMIs are broadcast to all CPUs as a
352 none: External NMI is masked for all CPUs. This is
353 useful so that a dump capture kernel won't be
357 See Documentation/networking/ipv6.rst.
359 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
360 Limit apic dumping. The parameter defines the maximal
361 number of local apics being dumped. Also it is possible
362 to set it to "all" by meaning -- no limit here.
363 Format: { 1 (default) | 2 | ... | all }.
364 The parameter valid if only apic=debug or
365 apic=verbose is specified.
366 Example: apic=debug show_lapic=all
368 apm= [APM] Advanced Power Management
369 See header of arch/x86/kernel/apm_32.c.
371 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
372 Format: <io>,<irq>,<nodeID>
374 arm64.nobti [ARM64] Unconditionally disable Branch Target
375 Identification support
377 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication
382 atarimouse= [HW,MOUSE] Atari Mouse
384 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
385 EzKey and similar keyboards
387 atkbd.reset= [HW] Reset keyboard during initialization
389 atkbd.set= [HW] Select keyboard code set
390 Format: <int> (2 = AT (default), 3 = PS/2)
392 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
395 atkbd.softraw= [HW] Choose between synthetic and real raw mode
396 Format: <bool> (0 = real, 1 = synthetic (default))
398 atkbd.softrepeat= [HW]
399 Use software keyboard repeat
401 audit= [KNL] Enable the audit sub-system
402 Format: { "0" | "1" | "off" | "on" }
403 0 | off - kernel audit is disabled and can not be
404 enabled until the next reboot
405 unset - kernel audit is initialized but disabled and
406 will be fully enabled by the userspace auditd.
407 1 | on - kernel audit is initialized and partially
408 enabled, storing at most audit_backlog_limit
409 messages in RAM until it is fully enabled by the
413 audit_backlog_limit= [KNL] Set the audit queue size limit.
414 Format: <int> (must be >=0)
417 bau= [X86_UV] Enable the BAU on SGI UV. The default
418 behavior is to disable the BAU (i.e. bau=0).
419 Format: { "0" | "1" }
422 unset - Disable the BAU.
424 baycom_epp= [HW,AX25]
427 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
429 See header of drivers/net/hamradio/baycom_par.c.
431 baycom_ser_fdx= [HW,AX25]
432 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
433 Format: <io>,<irq>,<mode>[,<baud>]
434 See header of drivers/net/hamradio/baycom_ser_fdx.c.
436 baycom_ser_hdx= [HW,AX25]
437 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
438 Format: <io>,<irq>,<mode>
439 See header of drivers/net/hamradio/baycom_ser_hdx.c.
441 blkdevparts= Manual partition parsing of block device(s) for
442 embedded devices based on command line input.
443 See Documentation/block/cmdline-partition.rst
445 boot_delay= Milliseconds to delay each printk during boot.
446 Values larger than 10 seconds (10000) are changed to
451 Extended command line options can be added to an initrd
452 and this will cause the kernel to look for it.
454 See Documentation/admin-guide/bootconfig.rst
457 Disable BERT OS support on buggy BIOSes.
459 bgrt_disable [ACPI][X86]
460 Disable BGRT to avoid flickering OEM logo.
462 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
463 bttv.radio= Most important insmod options are available as
465 bttv.pll= See Documentation/admin-guide/media/bttv.rst
468 bulk_remove=off [PPC] This parameter disables the use of the pSeries
469 firmware feature for flushing multiple hpte entries
472 c101= [NET] Moxa C101 synchronous serial card
474 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
475 Sometimes CPU hardware bugs make them report the cache
476 size incorrectly. The kernel will attempt work arounds
477 to fix known problems, but for some CPUs it is not
478 possible to determine what the correct size should be.
479 This option provides an override for these situations.
482 [NET] Specifies amount of time (in seconds) that
483 the kernel should wait for a network carrier. By default
484 it waits 120 seconds.
486 ca_keys= [KEYS] This parameter identifies a specific key(s) on
487 the system trusted keyring to be used for certificate
489 format: { id:<keyid> | builtin }
491 cca= [MIPS] Override the kernel pages' cache coherency
492 algorithm. Accepted values range from 0 to 7
493 inclusive. See arch/mips/include/asm/pgtable-bits.h
494 for platform specific values (SB1, Loongson3 and
497 ccw_timeout_log [S390]
498 See Documentation/s390/common_io.rst for details.
500 cgroup_disable= [KNL] Disable a particular controller
501 Format: {name of the controller(s) to disable}
502 The effects of cgroup_disable=foo are:
503 - foo isn't auto-mounted if you mount all cgroups in
505 - foo isn't visible as an individually mountable
507 {Currently only "memory" controller deal with this and
508 cut the overhead, others just disable the usage. So
509 only cgroup_disable=memory is actually worthy}
511 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
512 Format: { { controller | "all" | "named" }
513 [,{ controller | "all" | "named" }...] }
514 Like cgroup_disable, but only applies to cgroup v1;
515 the blacklisted controllers remain available in cgroup2.
516 "all" blacklists all controllers and "named" disables
517 named mounts. Specifying both "all" and "named" disables
520 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
522 nosocket -- Disable socket memory accounting.
523 nokmem -- Disable kernel memory accounting.
525 checkreqprot [SELINUX] Set initial checkreqprot flag value.
526 Format: { "0" | "1" }
527 See security/selinux/Kconfig help text.
528 0 -- check protection applied by kernel (includes
529 any implied execute protection).
530 1 -- check protection requested by application.
531 Default value is set via a kernel config option.
532 Value can be changed at runtime via
533 /sys/fs/selinux/checkreqprot.
534 Setting checkreqprot to 1 is deprecated.
537 See Documentation/s390/common_io.rst for details.
540 Prevents the clock framework from automatically gating
541 clocks that have not been explicitly enabled by a Linux
542 device driver but are enabled in hardware at reset or
543 by the bootloader/firmware. Note that this does not
544 force such clocks to be always-on nor does it reserve
545 those clocks in any way. This parameter is useful for
546 debug and development, but should not be needed on a
547 platform with proper driver support. For more
548 information, see Documentation/driver-api/clk.rst.
550 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
552 Forces specified clocksource (if available) to be used
553 when calculating gettimeofday(). If specified
554 clocksource is not available, it defaults to PIT.
555 Format: { pit | tsc | cyclone | pmtmr }
557 clocksource= Override the default clocksource
559 Override the default clocksource and use the clocksource
560 with the name specified.
561 Some clocksource names to choose from, depending on
563 [all] jiffies (this is the base, fallback clocksource)
565 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
566 pxa_timer,timer3,32k_counter,timer0_1
567 [X86-32] pit,hpet,tsc;
568 scx200_hrt on Geode; cyclone on IBM x440
576 clocksource.arm_arch_timer.evtstrm=
579 Enable/disable the eventstream feature of the ARM
580 architected timer so that code using WFE-based polling
581 loops can be debugged more effectively on production
584 clocksource.max_cswd_read_retries= [KNL]
585 Number of clocksource_watchdog() retries due to
586 external delays before the clock will be marked
587 unstable. Defaults to three retries, that is,
588 four attempts to read the clock under test.
590 clocksource.verify_n_cpus= [KNL]
591 Limit the number of CPUs checked for clocksources
592 marked with CLOCK_SOURCE_VERIFY_PERCPU that
593 are marked unstable due to excessive skew.
594 A negative value says to check all CPUs, while
595 zero says not to check any. Values larger than
596 nr_cpu_ids are silently truncated to nr_cpu_ids.
597 The actual CPUs are chosen randomly, with
598 no replacement if the same CPU is chosen twice.
600 clocksource-wdtest.holdoff= [KNL]
601 Set the time in seconds that the clocksource
602 watchdog test waits before commencing its tests.
603 Defaults to zero when built as a module and to
604 10 seconds when built into the kernel.
606 clearcpuid=BITNUM[,BITNUM...] [X86]
607 Disable CPUID feature X for the kernel. See
608 arch/x86/include/asm/cpufeatures.h for the valid bit
609 numbers. Note the Linux specific bits are not necessarily
610 stable over kernel options, but the vendor specific
612 Also note that user programs calling CPUID directly
613 or using the feature without checking anything
614 will still see it. This just prevents it from
615 being used by the kernel or shown in /proc/cpuinfo.
616 Also note the kernel might malfunction if you disable
619 cma=nn[MG]@[start[MG][-end[MG]]]
621 Sets the size of kernel global memory area for
622 contiguous memory allocations and optionally the
623 placement constraint by the physical address range of
624 memory allocations. A value of 0 disables CMA
625 altogether. For more information, see
626 kernel/dma/contiguous.c
630 Sets the size of kernel per-numa memory area for
631 contiguous memory allocations. A value of 0 disables
632 per-numa CMA altogether. And If this option is not
633 specificed, the default value is 0.
634 With per-numa CMA enabled, DMA users on node nid will
635 first try to allocate buffer from the pernuma area
636 which is located in node nid, if the allocation fails,
637 they will fallback to the global default memory area.
639 cmo_free_hint= [PPC] Format: { yes | no }
640 Specify whether pages are marked as being inactive
641 when they are freed. This is used in CMO environments
642 to determine OS memory pressure for page stealing by
646 coherent_pool=nn[KMG] [ARM,KNL]
647 Sets the size of memory pool for coherent, atomic dma
648 allocations, by default set to 256K.
650 com20020= [HW,NET] ARCnet - COM20020 chipset
652 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
654 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
658 ARCnet - COM90xx chipset (memory-mapped buffers)
659 Format: <io>[,<irq>[,<memstart>]]
661 condev= [HW,S390] console device
664 console= [KNL] Output console device and options.
666 tty<n> Use the virtual console device <n>.
670 Use the specified serial port. The options are of
671 the form "bbbbpnf", where "bbbb" is the baud rate,
672 "p" is parity ("n", "o", or "e"), "n" is number of
673 bits, and "f" is flow control ("r" for RTS or
674 omit it). Default is "9600n8".
676 See Documentation/admin-guide/serial-console.rst for more
678 Documentation/networking/netconsole.rst for an
681 uart[8250],io,<addr>[,options]
682 uart[8250],mmio,<addr>[,options]
683 uart[8250],mmio16,<addr>[,options]
684 uart[8250],mmio32,<addr>[,options]
685 uart[8250],0x<addr>[,options]
686 Start an early, polled-mode console on the 8250/16550
687 UART at the specified I/O port or MMIO address,
688 switching to the matching ttyS device later.
689 MMIO inter-register address stride is either 8-bit
690 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
691 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
692 to be equivalent to 'mmio'. 'options' are specified in
693 the same format described for ttyS above; if unspecified,
694 the h/w is not re-initialized.
696 hvc<n> Use the hypervisor console device <n>. This is for
697 both Xen and PowerPC hypervisors.
699 If the device connected to the port is not a TTY but a braille
700 device, prepend "brl," before the device type, for instance
702 For now, only VisioBraille is supported.
705 [KNL] Change console messages format
707 By default we print messages on consoles in
708 "[time stamp] text\n" format (time stamp may not be
709 printed, depending on CONFIG_PRINTK_TIME or
710 `printk_time' param).
712 Switch to syslog format: "<%u>[time stamp] text\n"
713 IOW, each message will have a facility and loglevel
714 prefix. The format is similar to one used by syslog()
715 syscall, or to executing "dmesg -S --raw" or to reading
718 consoleblank= [KNL] The console blank (screen saver) timeout in
719 seconds. A value of 0 disables the blank timer.
723 [KNL] Change the default value for
724 /proc/<pid>/coredump_filter.
725 See also Documentation/filesystems/proc.rst.
727 coresight_cpu_debug.enable
730 Enable/disable the CPU sampling based debugging.
731 0: default value, disable debugging
732 1: enable debugging at boot time
734 cpuidle.off=1 [CPU_IDLE]
735 disable the cpuidle sub-system
738 [CPU_IDLE] Name of the cpuidle governor to use.
740 cpufreq.off=1 [CPU_FREQ]
741 disable the cpufreq sub-system
743 cpufreq.default_governor=
744 [CPU_FREQ] Name of the default cpufreq governor or
745 policy to use. This governor must be registered in the
746 kernel before the cpufreq driver probes.
749 [X86] Delay for N microsec between assert and de-assert
750 of APIC INIT to start processors. This delay occurs
751 on every CPU online, such as boot, and resume from suspend.
754 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
756 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
758 crashkernel=size[KMG][@offset[KMG]]
759 [KNL] Using kexec, Linux can switch to a 'crash kernel'
760 upon panic. This parameter reserves the physical
761 memory region [offset, offset + size] for that kernel
762 image. If '@offset' is omitted, then a suitable offset
763 is selected automatically.
764 [KNL, X86-64] Select a region under 4G first, and
765 fall back to reserve region above 4G when '@offset'
766 hasn't been specified.
767 See Documentation/admin-guide/kdump/kdump.rst for further details.
769 crashkernel=range1:size1[,range2:size2,...][@offset]
770 [KNL] Same as above, but depends on the memory
771 in the running system. The syntax of range is
772 start-[end] where start and end are both
773 a memory unit (amount[KMG]). See also
774 Documentation/admin-guide/kdump/kdump.rst for an example.
776 crashkernel=size[KMG],high
777 [KNL, X86-64] range could be above 4G. Allow kernel
778 to allocate physical memory region from top, so could
779 be above 4G if system have more than 4G ram installed.
780 Otherwise memory region will be allocated below 4G, if
782 It will be ignored if crashkernel=X is specified.
783 crashkernel=size[KMG],low
784 [KNL, X86-64] range under 4G. When crashkernel=X,high
785 is passed, kernel could allocate physical memory region
786 above 4G, that cause second kernel crash on system
787 that require some amount of low memory, e.g. swiotlb
788 requires at least 64M+32K low memory, also enough extra
789 low memory is needed to make sure DMA buffers for 32-bit
790 devices won't run out. Kernel would try to allocate at
791 at least 256M below 4G automatically.
792 This one let user to specify own low range under 4G
793 for second kernel instead.
794 0: to disable low allocation.
795 It will be ignored when crashkernel=X,high is not used
796 or memory reserved is below 4G.
799 [KNL] Disable crypto self-tests
804 cs89x0_media= [HW,NET]
805 Format: { rj45 | aui | bnc }
807 csdlock_debug= [KNL] Enable debug add-ons of cross-CPU function call
808 handling. When switched on, additional debug data is
809 printed to the console in case a hanging CPU is
810 detected, and that CPU is pinged again in order to try
811 to resolve the hang situation.
812 0: disable csdlock debugging (default)
813 1: enable basic csdlock debugging (minor impact)
814 ext: enable extended csdlock debugging (more impact,
818 See header of drivers/s390/block/dasd_devmap.c.
820 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
821 (one device per port)
822 Format: <port#>,<type>
823 See also Documentation/input/devices/joystick-parport.rst
825 ddebug_query= [KNL,DYNAMIC_DEBUG] Enable debug messages at early boot
827 Documentation/admin-guide/dynamic-debug-howto.rst for
828 details. Deprecated, see dyndbg.
830 debug [KNL] Enable kernel debugging (events log level).
833 [KNL] Enable printing [hashed] pointers early in the
834 boot sequence. If enabled, we use a weak hash instead
835 of siphash to hash pointers. Use this option if you are
836 seeing instances of '(___ptrval___)') and need to see a
837 value (hashed pointer) instead. Cryptographically
838 insecure, please do not use on production kernels.
841 [KNL] verbose locking self-tests
843 Print debugging info while doing the locking API
845 Bitmask for the various LOCKTYPE_ tests. Defaults to 0
846 (no extra messages), setting it to -1 (all bits set)
847 will print _a_lot_ more information - normally only
848 useful to lockdep developers.
850 debug_objects [KNL] Enable object debugging
853 [KNL] Disable object debugging
855 debug_guardpage_minorder=
856 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
857 parameter allows control of the order of pages that will
858 be intentionally kept free (and hence protected) by the
859 buddy allocator. Bigger value increase the probability
860 of catching random memory corruption, but reduce the
861 amount of memory for normal system use. The maximum
862 possible value is MAX_ORDER/2. Setting this parameter
863 to 1 or 2 should be enough to identify most random
864 memory corruption problems caused by bugs in kernel or
865 driver code when a CPU writes to (or reads from) a
866 random memory location. Note that there exists a class
867 of memory corruptions problems caused by buggy H/W or
868 F/W or by drivers badly programing DMA (basically when
869 memory is written at bus level and the CPU MMU is
870 bypassed) which are not detectable by
871 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
872 tracking down these problems.
875 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
876 enables the feature at boot time. By default, it is
877 disabled and the system will work mostly the same as a
878 kernel built without CONFIG_DEBUG_PAGEALLOC.
879 Note: to get most of debug_pagealloc error reports, it's
880 useful to also enable the page_owner functionality.
881 on: enable the feature
883 debugfs= [KNL] This parameter enables what is exposed to userspace
884 and debugfs internal clients.
885 Format: { on, no-mount, off }
886 on: All functions are enabled.
888 Filesystem is not registered but kernel clients can
889 access APIs and a crashkernel can be used to read
890 its content. There is nothing to mount.
891 off: Filesystem is not registered and clients
892 get a -EPERM as result when trying to register files
893 or directories within debugfs.
894 This is equivalent of the runtime functionality if
895 debugfs was not enabled in the kernel at all.
896 Default value is set in build-time with a kernel configuration.
898 debugpat [X86] Enable PAT debugging
900 decnet.addr= [HW,NET]
901 Format: <area>[,<node>]
902 See also Documentation/networking/decnet.rst.
905 [HW] The size of the default HugeTLB page. This is
906 the size represented by the legacy /proc/ hugepages
907 APIs. In addition, this is the default hugetlb size
908 used for shmget(), mmap() and mounting hugetlbfs
909 filesystems. If not specified, defaults to the
910 architecture's default huge page size. Huge page
911 sizes are architecture dependent. See also
912 Documentation/admin-guide/mm/hugetlbpage.rst.
915 deferred_probe_timeout=
916 [KNL] Debugging option to set a timeout in seconds for
917 deferred probe to give up waiting on dependencies to
918 probe. Only specific dependencies (subsystems or
919 drivers) that have opted in will be ignored. A timeout of 0
920 will timeout at the end of initcalls. This option will also
921 dump out devices still on the deferred probe list after
925 Format: { on | off | def_only | inf_only | always }
926 on: s390 zlib hardware support for compression on
927 level 1 and decompression (default)
928 off: No s390 zlib hardware support
929 def_only: s390 zlib hardware support for deflate
930 only (compression on level 1)
931 inf_only: s390 zlib hardware support for inflate
933 always: Same as 'on' but ignores the selected compression
934 level always using hardware support (used for debugging)
937 Set number of hash buckets for dentry cache.
939 disable_1tb_segments [PPC]
940 Disables the use of 1TB hash page table segments. This
941 causes the kernel to fall back to 256MB segments which
942 can be useful when debugging issues that require an SLB
946 Limits the number of kernel SLB entries, and flushes
947 them frequently to increase the rate of SLB faults
951 See Documentation/networking/ipv6.rst.
954 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
955 hardening is enabled for this boot. Hardened
956 usercopy checking is used to protect the kernel
957 from reading or writing beyond known memory
958 allocation boundaries as a proactive defense
959 against bounds-checking flaws in the kernel's
960 copy_to_user()/copy_from_user() interface.
961 on Perform hardened usercopy checks (default).
962 off Disable hardened usercopy checks.
965 Disable RADIX MMU mode on POWER9
967 radix_hcall_invalidate=on [PPC/PSERIES]
968 Disable RADIX GTSE feature and use hcall for TLB
972 Disable TLBIE instruction. Currently does not work
973 with KVM, with HASH MMU, or with coherent accelerators.
975 disable_cpu_apicid= [X86,APIC,SMP]
977 The number of initial APIC ID for the
978 corresponding CPU to be disabled at boot,
979 mostly used for the kdump 2nd kernel to
980 disable BSP to wake up multiple CPUs without
981 causing system reset or hang due to sending
984 disable_ddw [PPC/PSERIES]
985 Disable Dynamic DMA Window support. Use this
986 to workaround buggy firmware.
989 See Documentation/networking/ipv6.rst.
991 disable_mtrr_cleanup [X86]
992 The kernel tries to adjust MTRR layout from continuous
993 to discrete, to make X server driver able to add WB
994 entry later. This parameter disables that.
996 disable_mtrr_trim [X86, Intel and AMD only]
997 By default the kernel will trim any uncacheable
998 memory out of your available memory pool based on
999 MTRR settings. This parameter disables that behavior,
1000 possibly causing your machine to run very slowly.
1002 disable_timer_pin_1 [X86]
1003 Disable PIN 1 of APIC timer
1004 Can be useful to work around chipset bugs.
1006 dis_ucode_ldr [X86] Disable the microcode loader.
1008 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
1009 this option disables the debugging code at boot.
1011 dma_debug_entries=<number>
1012 This option allows to tune the number of preallocated
1013 entries for DMA-API debugging code. One entry is
1014 required per DMA-API allocation. Use this if the
1015 DMA-API debugging code disables itself because the
1016 architectural default is too low.
1018 dma_debug_driver=<driver_name>
1019 With this option the DMA-API debugging driver
1020 filter feature can be enabled at boot time. Just
1021 pass the driver to filter for as the parameter.
1022 The filter can be disabled or changed to another
1023 driver later using sysfs.
1025 driver_async_probe= [KNL]
1026 List of driver names to be probed asynchronously.
1027 Format: <driver_name1>,<driver_name2>...
1029 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
1030 Broken monitors, graphic adapters, KVMs and EDIDless
1031 panels may send no or incorrect EDID data sets.
1032 This parameter allows to specify an EDID data sets
1033 in the /lib/firmware directory that are used instead.
1034 Generic built-in EDID data sets are used, if one of
1035 edid/1024x768.bin, edid/1280x1024.bin,
1036 edid/1680x1050.bin, or edid/1920x1080.bin is given
1037 and no file with the same name exists. Details and
1038 instructions how to build your own EDID data are
1039 available in Documentation/admin-guide/edid.rst. An EDID
1040 data set will only be used for a particular connector,
1041 if its name and a colon are prepended to the EDID
1042 name. Each connector may use a unique EDID data
1043 set by separating the files with a comma. An EDID
1044 data set with no connector name will be used for
1045 any connectors not explicitly specified.
1050 Format: {"off" | "known"}
1051 Control how the dt_cpu_ftrs device-tree binding is
1052 used for CPU feature discovery and setup (if it
1054 off: Do not use it, fall back to legacy cpu table.
1055 known: Do not pass through unknown features to guests
1056 or userspace, only those that the kernel is aware of.
1058 dump_apple_properties [X86]
1059 Dump name and content of EFI device properties on
1060 x86 Macs. Useful for driver authors to determine
1061 what data is available or for reverse-engineering.
1063 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
1064 <module>.dyndbg[="val"]
1065 Enable debug messages at boot time. See
1066 Documentation/admin-guide/dynamic-debug-howto.rst
1069 nopku [X86] Disable Memory Protection Keys CPU feature found
1072 <module>.async_probe [KNL]
1073 Enable asynchronous probe on this module.
1075 early_ioremap_debug [KNL]
1076 Enable debug messages in early_ioremap support. This
1077 is useful for tracking down temporary early mappings
1078 which are not unmapped.
1080 earlycon= [KNL] Output early console device and options.
1082 When used with no options, the early console is
1083 determined by stdout-path property in device tree's
1084 chosen node or the ACPI SPCR table if supported by
1087 cdns,<addr>[,options]
1088 Start an early, polled-mode console on a Cadence
1089 (xuartps) serial port at the specified address. Only
1090 supported option is baud rate. If baud rate is not
1091 specified, the serial port must already be setup and
1094 uart[8250],io,<addr>[,options]
1095 uart[8250],mmio,<addr>[,options]
1096 uart[8250],mmio32,<addr>[,options]
1097 uart[8250],mmio32be,<addr>[,options]
1098 uart[8250],0x<addr>[,options]
1099 Start an early, polled-mode console on the 8250/16550
1100 UART at the specified I/O port or MMIO address.
1101 MMIO inter-register address stride is either 8-bit
1102 (mmio) or 32-bit (mmio32 or mmio32be).
1103 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1104 to be equivalent to 'mmio'. 'options' are specified
1105 in the same format described for "console=ttyS<n>"; if
1106 unspecified, the h/w is not initialized.
1110 Start an early, polled-mode console on a pl011 serial
1111 port at the specified address. The pl011 serial port
1112 must already be setup and configured. Options are not
1113 yet supported. If 'mmio32' is specified, then only
1114 the driver will use only 32-bit accessors to read/write
1115 the device registers.
1118 Start an early, polled-mode console on a meson serial
1119 port at the specified address. The serial port must
1120 already be setup and configured. Options are not yet
1124 Start an early, polled-mode console on an msm serial
1125 port at the specified address. The serial port
1126 must already be setup and configured. Options are not
1129 msm_serial_dm,<addr>
1130 Start an early, polled-mode console on an msm serial
1131 dm port at the specified address. The serial port
1132 must already be setup and configured. Options are not
1136 Start an early, polled-mode console on a serial port
1137 of an Actions Semi SoC, such as S500 or S900, at the
1138 specified address. The serial port must already be
1139 setup and configured. Options are not yet supported.
1142 Start an early, polled-mode console on a serial port
1143 of an RDA Micro SoC, such as RDA8810PL, at the
1144 specified address. The serial port must already be
1145 setup and configured. Options are not yet supported.
1148 Use RISC-V SBI (Supervisor Binary Interface) for early
1151 smh Use ARM semihosting calls for early console.
1159 Use early console provided by serial driver available
1160 on Samsung SoCs, requires selecting proper type and
1161 a correct base address of the selected UART port. The
1162 serial port must already be setup and configured.
1163 Options are not yet supported.
1166 Start an early, polled-mode console on a lantiq serial
1167 (lqasc) port at the specified address. The serial port
1168 must already be setup and configured. Options are not
1173 Use early console provided by Freescale LP UART driver
1174 found on Freescale Vybrid and QorIQ LS1021A processors.
1175 A valid base address must be provided, and the serial
1176 port must already be setup and configured.
1180 Start an early, polled-mode, output-only console on the
1181 Freescale i.MX UART at the specified address. The UART
1182 must already be setup and configured.
1185 Start an early, polled-mode console on the
1186 Armada 3700 serial port at the specified
1187 address. The serial port must already be setup
1188 and configured. Options are not yet supported.
1191 Start an early, polled-mode console on a Qualcomm
1192 Generic Interface (GENI) based serial port at the
1193 specified address. The serial port must already be
1194 setup and configured. Options are not yet supported.
1197 Start an early, unaccelerated console on the EFI
1198 memory mapped framebuffer (if available). On cache
1199 coherent non-x86 systems that use system memory for
1200 the framebuffer, pass the 'ram' option so that it is
1201 mapped with the correct attributes.
1204 Use early console provided by Freescale LINFlexD UART
1205 serial driver for NXP S32V234 SoCs. A valid base
1206 address must be provided, and the serial port must
1207 already be setup and configured.
1209 earlyprintk= [X86,SH,ARM,M68k,S390]
1213 earlyprintk=serial[,ttySn[,baudrate]]
1214 earlyprintk=serial[,0x...[,baudrate]]
1215 earlyprintk=ttySn[,baudrate]
1216 earlyprintk=dbgp[debugController#]
1217 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1218 earlyprintk=xdbc[xhciController#]
1220 earlyprintk is useful when the kernel crashes before
1221 the normal console is initialized. It is not enabled by
1222 default because it has some cosmetic problems.
1224 Append ",keep" to not disable it when the real console
1227 Only one of vga, efi, serial, or usb debug port can
1230 Currently only ttyS0 and ttyS1 may be specified by
1231 name. Other I/O ports may be explicitly specified
1232 on some architectures (x86 and arm at least) by
1233 replacing ttySn with an I/O port address, like this:
1234 earlyprintk=serial,0x1008,115200
1235 You can find the port for a given device in
1236 /proc/tty/driver/serial:
1237 2: uart:ST16650V2 port:00001008 irq:18 ...
1239 Interaction with the standard serial driver is not
1242 The VGA and EFI output is eventually overwritten by
1245 The xen output can only be used by Xen PV guests.
1247 The sclp output can only be used on s390.
1249 The optional "force" to "pciserial" enables use of a
1250 PCI device even when its classcode is not of the
1253 edac_report= [HW,EDAC] Control how to report EDAC event
1254 Format: {"on" | "off" | "force"}
1255 on: enable EDAC to report H/W event. May be overridden
1256 by other higher priority error reporting module.
1257 off: disable H/W event reporting through EDAC.
1258 force: enforce the use of EDAC to report H/W event.
1261 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1264 This is designed to be used in conjunction with
1265 the boot argument: earlyprintk=vga
1267 This parameter works in place of the kgdboc parameter
1268 but can only be used if the backing tty is available
1269 very early in the boot process. For early debugging
1270 via a serial port see kgdboc_earlycon instead.
1273 Format: {"off" | "on" | "skip[mbr]"}
1276 Format: { "debug", "disable_early_pci_dma",
1277 "nochunk", "noruntime", "nosoftreserve",
1278 "novamap", "no_disable_early_pci_dma" }
1279 debug: enable misc debug output.
1280 disable_early_pci_dma: disable the busmaster bit on all
1281 PCI bridges while in the EFI boot stub.
1282 nochunk: disable reading files in "chunks" in the EFI
1283 boot stub, as chunking can cause problems with some
1284 firmware implementations.
1285 noruntime : disable EFI runtime services support
1286 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1287 attribute may cause the kernel to reserve the
1288 memory range for a memory mapping driver to
1289 claim. Specify efi=nosoftreserve to disable this
1290 reservation and treat the memory by its base type
1291 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1292 novamap: do not call SetVirtualAddressMap().
1293 no_disable_early_pci_dma: Leave the busmaster bit set
1294 on all PCI bridges while in the EFI boot stub
1296 efi_no_storage_paranoia [EFI; X86]
1297 Using this parameter you can use more than 50% of
1298 your efi variable storage. Use this parameter only if
1299 you are really sure that your UEFI does sane gc and
1300 fulfills the spec otherwise your board may brick.
1302 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1303 Add arbitrary attribute to specific memory range by
1304 updating original EFI memory map.
1305 Region of memory which aa attribute is added to is
1308 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1309 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1310 attribute is added to range 0x100000000-0x180000000 and
1311 0x10a0000000-0x1120000000.
1313 If efi_fake_mem=8G@9G:0x40000 is specified, the
1314 EFI_MEMORY_SP(0x40000) attribute is added to
1315 range 0x240000000-0x43fffffff.
1317 Using this parameter you can do debugging of EFI memmap
1318 related features. For example, you can do debugging of
1319 Address Range Mirroring feature even if your box
1320 doesn't support it, or mark specific memory as
1323 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1324 that is to be dynamically loaded by Linux. If there are
1325 multiple variables with the same name but with different
1326 vendor GUIDs, all of them will be loaded. See
1327 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1330 eisa_irq_edge= [PARISC,HW]
1331 See header of drivers/parisc/eisa.c.
1334 See comment before function elanfreq_setup() in
1335 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1337 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1338 Specifies physical address of start of kernel core
1339 image elf header and optionally the size. Generally
1340 kexec loader will pass this option to capture kernel.
1341 See Documentation/admin-guide/kdump/kdump.rst for details.
1343 enable_mtrr_cleanup [X86]
1344 The kernel tries to adjust MTRR layout from continuous
1345 to discrete, to make X server driver able to add WB
1346 entry later. This parameter enables that.
1348 enable_timer_pin_1 [X86]
1349 Enable PIN 1 of APIC timer
1350 Can be useful to work around chipset bugs
1351 (in particular on some ATI chipsets).
1352 The kernel tries to set a reasonable default.
1354 enforcing [SELINUX] Set initial enforcing status.
1356 See security/selinux/Kconfig help text.
1357 0 -- permissive (log only, no denials).
1358 1 -- enforcing (deny and log).
1360 Value can be changed at runtime via
1361 /sys/fs/selinux/enforce.
1364 Disable Error Record Serialization Table (ERST)
1367 ether= [HW,NET] Ethernet cards parameters
1368 This option is obsoleted by the "netdev=" option, which
1369 has equivalent usage. See its documentation for details.
1373 Permit 'security.evm' to be updated regardless of
1374 current integrity status.
1379 fail_make_request=[KNL]
1380 General fault injection mechanism.
1381 Format: <interval>,<probability>,<space>,<times>
1382 See also Documentation/fault-injection/.
1385 Format: { initns | none }
1386 See Documentation/admin-guide/sysctl/net.rst for
1387 fb_tunnels_only_for_init_ns
1390 See Documentation/admin-guide/blockdev/floppy.rst.
1392 force_pal_cache_flush
1393 [IA-64] Avoid check_sal_cache_flush which may hang on
1394 buggy SAL_CACHE_FLUSH implementations. Using this
1395 parameter will force ia64_sal_cache_flush to call
1396 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1399 Forcefully enable Physical Address Extension (PAE).
1400 Many Pentium M systems disable PAE but may have a
1401 functionally usable PAE implementation.
1402 Warning: use of this parameter will taint the kernel
1403 and may cause unknown problems.
1406 [FTRACE] will set and start the specified tracer
1407 as early as possible in order to facilitate early
1410 ftrace_dump_on_oops[=orig_cpu]
1411 [FTRACE] will dump the trace buffers on oops.
1412 If no parameter is passed, ftrace will dump
1413 buffers of all CPUs, but if you pass orig_cpu, it will
1414 dump only the buffer of the CPU that triggered the
1417 ftrace_filter=[function-list]
1418 [FTRACE] Limit the functions traced by the function
1419 tracer at boot up. function-list is a comma-separated
1420 list of functions. This list can be changed at run
1421 time by the set_ftrace_filter file in the debugfs
1424 ftrace_notrace=[function-list]
1425 [FTRACE] Do not trace the functions specified in
1426 function-list. This list can be changed at run time
1427 by the set_ftrace_notrace file in the debugfs
1430 ftrace_graph_filter=[function-list]
1431 [FTRACE] Limit the top level callers functions traced
1432 by the function graph tracer at boot up.
1433 function-list is a comma-separated list of functions
1434 that can be changed at run time by the
1435 set_graph_function file in the debugfs tracing directory.
1437 ftrace_graph_notrace=[function-list]
1438 [FTRACE] Do not trace from the functions specified in
1439 function-list. This list is a comma-separated list of
1440 functions that can be changed at run time by the
1441 set_graph_notrace file in the debugfs tracing directory.
1443 ftrace_graph_max_depth=<uint>
1444 [FTRACE] Used with the function graph tracer. This is
1445 the max depth it will trace into a function. This value
1446 can be changed at run time by the max_graph_depth file
1447 in the tracefs tracing directory. default: 0 (no limit)
1449 fw_devlink= [KNL] Create device links between consumer and supplier
1450 devices by scanning the firmware to infer the
1451 consumer/supplier relationships. This feature is
1452 especially useful when drivers are loaded as modules as
1453 it ensures proper ordering of tasks like device probing
1454 (suppliers first, then consumers), supplier boot state
1455 clean up (only after all consumers have probed),
1456 suspend/resume & runtime PM (consumers first, then
1458 Format: { off | permissive | on | rpm }
1459 off -- Don't create device links from firmware info.
1460 permissive -- Create device links from firmware info
1461 but use it only for ordering boot state clean
1462 up (sync_state() calls).
1463 on -- Create device links from firmware info and use it
1464 to enforce probe and suspend/resume ordering.
1465 rpm -- Like "on", but also use to order runtime PM.
1467 fw_devlink.strict=<bool>
1468 [KNL] Treat all inferred dependencies as mandatory
1469 dependencies. This only applies for fw_devlink=on|rpm.
1473 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1474 support via parallel port (up to 5 devices per port)
1475 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1476 See also Documentation/input/devices/joystick-parport.rst
1480 gart_fix_e820= [X86-64] disable the fix e820 for K8 GART
1484 gcov_persist= [GCOV] When non-zero (default), profiling data for
1485 kernel modules is saved and remains accessible via
1486 debugfs, even when the module is unloaded/reloaded.
1487 When zero, profiling data is discarded and associated
1488 debugfs files are removed at module unload time.
1490 goldfish [X86] Enable the goldfish android emulator platform.
1491 Don't use this when you are not running on the
1494 gpio-mockup.gpio_mockup_ranges
1495 [HW] Sets the ranges of gpiochip of for this device.
1496 Format: <start1>,<end1>,<start2>,<end2>...
1497 gpio-mockup.gpio_mockup_named_lines
1498 [HW] Let the driver know GPIO lines should be named.
1500 gpt [EFI] Forces disk with valid GPT signature but
1501 invalid Protective MBR to be treated as GPT. If the
1502 primary GPT is corrupted, it enables the backup/alternate
1503 GPT to be used instead.
1505 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1506 the "Enable 0" bit of the configuration register.
1509 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1510 the "Enable 0" bit of the configuration register.
1513 grcan.select= [HW] Select which physical interface to use.
1516 grcan.txsize= [HW] Sets the size of the tx buffer.
1517 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1519 grcan.rxsize= [HW] Sets the size of the rx buffer.
1520 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1523 hardlockup_all_cpu_backtrace=
1524 [KNL] Should the hard-lockup detector generate
1525 backtraces on all cpus.
1528 hashdist= [KNL,NUMA] Large hashes allocated during boot
1529 are distributed across NUMA nodes. Defaults on
1530 for 64-bit NUMA, off otherwise.
1531 Format: 0 | 1 (for off | on)
1533 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1535 hd= [EIDE] (E)IDE hard drive subsystem geometry
1536 Format: <cyl>,<head>,<sect>
1539 Disable Hardware Error Source Table (HEST) support;
1540 corresponding firmware-first mode error processing
1541 logic will be disabled.
1543 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1544 size of <nn>. This works even on boxes that have no
1545 highmem otherwise. This also works to reduce highmem
1546 size on bigger boxes.
1548 highres= [KNL] Enable/disable high resolution timer mode.
1549 Valid parameters: "on", "off"
1554 hpet= [X86-32,HPET] option to control HPET usage
1555 Format: { enable (default) | disable | force |
1557 disable: disable HPET and use PIT instead
1558 force: allow force enabled of undocumented chips (ICH4,
1560 verbose: show contents of HPET registers during setup
1562 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1563 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1565 hugetlb_cma= [HW,CMA] The size of a CMA area used for allocation
1566 of gigantic hugepages.
1569 Reserve a CMA area of given size and allocate gigantic
1570 hugepages using the CMA allocator. If enabled, the
1571 boot-time allocation of gigantic hugepages is skipped.
1573 hugepages= [HW] Number of HugeTLB pages to allocate at boot.
1574 If this follows hugepagesz (below), it specifies
1575 the number of pages of hugepagesz to be allocated.
1576 If this is the first HugeTLB parameter on the command
1577 line, it specifies the number of pages to allocate for
1578 the default huge page size. See also
1579 Documentation/admin-guide/mm/hugetlbpage.rst.
1583 [HW] The size of the HugeTLB pages. This is used in
1584 conjunction with hugepages (above) to allocate huge
1585 pages of a specific size at boot. The pair
1586 hugepagesz=X hugepages=Y can be specified once for
1587 each supported huge page size. Huge page sizes are
1588 architecture dependent. See also
1589 Documentation/admin-guide/mm/hugetlbpage.rst.
1593 [KNL] Should the hung task detector generate panics.
1596 A value of 1 instructs the kernel to panic when a
1597 hung task is detected. The default value is controlled
1598 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1599 option. The value selected by this boot parameter can
1600 be changed later by the kernel.hung_task_panic sysctl.
1602 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1603 terminal devices. Valid values: 0..8
1604 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1605 If specified, z/VM IUCV HVC accepts connections
1606 from listed z/VM user IDs only.
1608 hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations
1609 which allow the hypervisor to 'idle' the
1610 guest on lock contention.
1613 Do not unregister boot console at start. This is only
1614 useful for debugging when something happens in the window
1615 between unregistering the boot console and initializing
1618 i2c_bus= [HW] Override the default board specific I2C bus speed
1619 or register an additional I2C bus that is not
1620 registered from board initialization code.
1624 i8042.debug [HW] Toggle i8042 debug mode
1625 i8042.unmask_kbd_data
1626 [HW] Enable printing of interrupt data from the KBD port
1627 (disabled by default, and as a pre-condition
1628 requires that i8042.debug=1 be enabled)
1629 i8042.direct [HW] Put keyboard port into non-translated mode
1630 i8042.dumbkbd [HW] Pretend that controller can only read data from
1631 keyboard and cannot control its state
1632 (Don't attempt to blink the leds)
1633 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1634 i8042.nokbd [HW] Don't check/create keyboard port
1635 i8042.noloop [HW] Disable the AUX Loopback command while probing
1637 i8042.nomux [HW] Don't check presence of an active multiplexing
1639 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1641 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1642 i8042.reset [HW] Reset the controller during init, cleanup and
1643 suspend-to-ram transitions, only during s2r
1644 transitions, or never reset
1645 Format: { 1 | Y | y | 0 | N | n }
1646 1, Y, y: always reset controller
1647 0, N, n: don't ever reset controller
1648 Default: only on s2r transitions on x86; most other
1649 architectures force reset to be always executed
1650 i8042.unlock [HW] Unlock (ignore) the keylock
1651 i8042.kbdreset [HW] Reset device connected to KBD port
1655 i8k.ignore_dmi [HW] Continue probing hardware even if DMI data
1656 indicates that the driver is running on unsupported
1658 i8k.force [HW] Activate i8k driver even if SMM BIOS signature
1659 does not match list of supported models.
1661 [HW] Report power status in /proc/i8k
1662 (disabled by default)
1663 i8k.restricted [HW] Allow controlling fans only if SYS_ADMIN
1666 i915.invert_brightness=
1667 [DRM] Invert the sense of the variable that is used to
1668 set the brightness of the panel backlight. Normally a
1669 brightness value of 0 indicates backlight switched off,
1670 and the maximum of the brightness value sets the backlight
1671 to maximum brightness. If this parameter is set to 0
1672 (default) and the machine requires it, or this parameter
1673 is set to 1, a brightness value of 0 sets the backlight
1674 to maximum brightness, and the maximum of the brightness
1675 value switches the backlight off.
1676 -1 -- never invert brightness
1677 0 -- machine default
1678 1 -- force brightness inversion
1681 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1683 ide-core.nodma= [HW] (E)IDE subsystem
1684 Format: =0.0 to prevent dma on hda, =0.1 hdb =1.0 hdc
1685 .vlb_clock .pci_clock .noflush .nohpa .noprobe .nowerr
1686 .cdrom .chs .ignore_cable are additional options
1687 See Documentation/ide/ide.rst.
1689 ide-generic.probe-mask= [HW] (E)IDE subsystem
1691 Probe mask for legacy ISA IDE ports. Depending on
1692 platform up to 6 ports are supported, enabled by
1693 setting corresponding bits in the mask to 1. The
1694 default value is 0x0, which has a special meaning.
1695 On systems that have PCI, it triggers scanning the
1696 PCI bus for the first and the second port, which
1697 are then probed. On systems without PCI the value
1698 of 0x0 enables probing the two first ports as if it
1701 ide-pci-generic.all-generic-ide [HW] (E)IDE subsystem
1702 Claim all unknown PCI IDE storage controllers.
1705 Format: idle=poll, idle=halt, idle=nomwait
1706 Poll forces a polling idle loop that can slightly
1707 improve the performance of waking up a idle CPU, but
1708 will use a lot of power and make the system run hot.
1710 idle=halt: Halt is forced to be used for CPU idle.
1711 In such case C2/C3 won't be used again.
1712 idle=nomwait: Disable mwait for CPU C-states
1716 Allow force disabling of Shared Virtual Memory (SVA)
1717 support for the idxd driver. By default it is set to
1720 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1721 Format: { strict | legacy | 2008 | relaxed }
1724 Choose which programs will be accepted for execution
1725 based on the IEEE 754 NaN encoding(s) supported by
1726 the FPU and the NaN encoding requested with the value
1727 of an ELF file header flag individually set by each
1728 binary. Hardware implementations are permitted to
1729 support either or both of the legacy and the 2008 NaN
1732 Available settings are as follows:
1733 strict accept binaries that request a NaN encoding
1734 supported by the FPU
1735 legacy only accept legacy-NaN binaries, if supported
1737 2008 only accept 2008-NaN binaries, if supported
1739 relaxed accept any binaries regardless of whether
1740 supported by the FPU
1742 The FPU emulator is always able to support both NaN
1743 encodings, so if no FPU hardware is present or it has
1744 been disabled with 'nofpu', then the settings of
1745 'legacy' and '2008' strap the emulator accordingly,
1746 'relaxed' straps the emulator for both legacy-NaN and
1747 2008-NaN, whereas 'strict' enables legacy-NaN only on
1748 legacy processors and both NaN encodings on MIPS32 or
1751 The setting for ABS.fmt/NEG.fmt instruction execution
1752 mode generally follows that for the NaN encoding,
1753 except where unsupported by hardware.
1755 ignore_loglevel [KNL]
1756 Ignore loglevel setting - this will print /all/
1757 kernel messages to the console. Useful for debugging.
1758 We also add it as printk module parameter, so users
1759 could change it dynamically, usually by
1760 /sys/module/printk/parameters/ignore_loglevel.
1763 Ignore RLIMIT_DATA setting for data mappings,
1764 print warning at first misuse. Can be changed via
1765 /sys/module/kernel/parameters/ignore_rlimit_data.
1767 ihash_entries= [KNL]
1768 Set number of hash buckets for inode cache.
1770 ima_appraise= [IMA] appraise integrity measurements
1771 Format: { "off" | "enforce" | "fix" | "log" }
1774 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
1775 The builtin appraise policy appraises all files
1778 ima_canonical_fmt [IMA]
1779 Use the canonical format for the binary runtime
1780 measurements, instead of host native format.
1783 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1787 The list of supported hash algorithms is defined
1788 in crypto/hash_info.h.
1791 The builtin policies to load during IMA setup.
1792 Format: "tcb | appraise_tcb | secure_boot |
1793 fail_securely | critical_data"
1795 The "tcb" policy measures all programs exec'd, files
1796 mmap'd for exec, and all files opened with the read
1797 mode bit set by either the effective uid (euid=0) or
1800 The "appraise_tcb" policy appraises the integrity of
1801 all files owned by root.
1803 The "secure_boot" policy appraises the integrity
1804 of files (eg. kexec kernel image, kernel modules,
1805 firmware, policy, etc) based on file signatures.
1807 The "fail_securely" policy forces file signature
1808 verification failure also on privileged mounted
1809 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
1812 The "critical_data" policy measures kernel integrity
1815 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
1816 Load a policy which meets the needs of the Trusted
1817 Computing Base. This means IMA will measure all
1818 programs exec'd, files mmap'd for exec, and all files
1819 opened for read by uid=0.
1822 Select one of defined IMA measurements template formats.
1823 Formats: { "ima" | "ima-ng" | "ima-sig" }
1827 [IMA] Define a custom template format.
1828 Format: { "field1|...|fieldN" }
1830 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
1831 Format: <min_file_size>
1832 Set the minimal file size for using asynchronous hash.
1833 If left unspecified, ahash usage is disabled.
1835 ahash performance varies for different data sizes on
1836 different crypto accelerators. This option can be used
1837 to achieve the best performance for a particular HW.
1839 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
1841 Set hashing buffer size. Default: 4k.
1843 ahash performance varies for different chunk sizes on
1844 different crypto accelerators. This option can be used
1845 to achieve best performance for particular HW.
1849 Run specified binary instead of /sbin/init as init
1852 initcall_debug [KNL] Trace initcalls as they are executed. Useful
1853 for working out where the kernel is dying during
1856 initcall_blacklist= [KNL] Do not execute a comma-separated list of
1857 initcall functions. Useful for debugging built-in
1858 modules and initcalls.
1860 initramfs_async= [KNL]
1863 This parameter controls whether the initramfs
1864 image is unpacked asynchronously, concurrently
1865 with devices being probed and
1866 initialized. This should normally just work,
1867 but as a debugging aid, one can get the
1868 historical behaviour of the initramfs
1869 unpacking being completed before device_ and
1872 initrd= [BOOT] Specify the location of the initial ramdisk
1874 initrdmem= [KNL] Specify a physical address and size from which to
1875 load the initrd. If an initrd is compiled in or
1876 specified in the bootparams, it takes priority over this
1878 Format: ss[KMG],nn[KMG]
1881 init_on_alloc= [MM] Fill newly allocated pages and heap objects with
1884 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
1886 init_on_free= [MM] Fill freed pages and heap objects with zeroes.
1888 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
1890 init_pkru= [X86] Specify the default memory protection keys rights
1891 register contents for all processes. 0x55555554 by
1892 default (disallow access to all but pkey 0). Can
1893 override in debugfs after boot.
1895 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
1898 int_pln_enable [X86] Enable power limit notification interrupt
1900 integrity_audit=[IMA]
1901 Format: { "0" | "1" }
1902 0 -- basic integrity auditing messages. (Default)
1903 1 -- additional integrity auditing messages.
1905 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
1907 Enable intel iommu driver.
1909 Disable intel iommu driver.
1910 igfx_off [Default Off]
1911 By default, gfx is mapped as normal device. If a gfx
1912 device has a dedicated DMAR unit, the DMAR unit is
1913 bypassed by not enabling DMAR with this option. In
1914 this case, gfx device will use physical address for
1916 strict [Default Off]
1917 With this option on every unmap_single operation will
1918 result in a hardware IOTLB flush operation as opposed
1919 to batching them for performance.
1920 sp_off [Default Off]
1921 By default, super page will be supported if Intel IOMMU
1922 has the capability. With this option, super page will
1925 By default, scalable mode will be disabled even if the
1926 hardware advertises that it has support for the scalable
1927 mode translation. With this option set, scalable mode
1928 will be used on hardware which claims to support it.
1929 tboot_noforce [Default Off]
1930 Do not force the Intel IOMMU enabled under tboot.
1931 By default, tboot will force Intel IOMMU on, which
1932 could harm performance of some high-throughput
1933 devices like 40GBit network cards, even if identity
1935 Note that using this option lowers the security
1936 provided by tboot because it makes the system
1937 vulnerable to DMA attacks.
1939 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
1940 0 disables intel_idle and fall back on acpi_idle.
1941 1 to 9 specify maximum depth of C-state.
1945 Do not enable intel_pstate as the default
1946 scaling driver for the supported processors
1948 Use intel_pstate as a scaling driver, but configure it
1949 to work with generic cpufreq governors (instead of
1950 enabling its internal governor). This mode cannot be
1951 used along with the hardware-managed P-states (HWP)
1954 Enable intel_pstate on systems that prohibit it by default
1955 in favor of acpi-cpufreq. Forcing the intel_pstate driver
1956 instead of acpi-cpufreq may disable platform features, such
1957 as thermal controls and power capping, that rely on ACPI
1958 P-States information being indicated to OSPM and therefore
1959 should be used with caution. This option does not work with
1960 processors that aren't supported by the intel_pstate driver
1961 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
1963 Do not enable hardware P state control (HWP)
1966 Only load intel_pstate on systems which support
1967 hardware P state control (HWP) if available.
1969 Enforce ACPI _PPC performance limits. If the Fixed ACPI
1970 Description Table, specifies preferred power management
1971 profile as "Enterprise Server" or "Performance Server",
1972 then this feature is turned on by default.
1974 Allow per-logical-CPU P-State performance control limits using
1975 cpufreq sysfs interface
1977 intremap= [X86-64, Intel-IOMMU]
1978 on enable Interrupt Remapping (default)
1979 off disable Interrupt Remapping
1980 nosid disable Source ID checking
1982 BIOS x2APIC opt-out request will be ignored
1983 nopost disable Interrupt Posting
1985 iomem= Disable strict checking of access to MMIO memory
1986 strict regions from userspace.
2001 nobypass [PPC/POWERNV]
2002 Disable IOMMU bypass, using IOMMU for PCI devices.
2004 iommu.forcedac= [ARM64, X86] Control IOVA allocation for PCI devices.
2005 Format: { "0" | "1" }
2006 0 - Try to allocate a 32-bit DMA address first, before
2007 falling back to the full range if needed.
2008 1 - Allocate directly from the full usable range,
2009 forcing Dual Address Cycle for PCI cards supporting
2010 greater than 32-bit addressing.
2012 iommu.strict= [ARM64] Configure TLB invalidation behaviour
2013 Format: { "0" | "1" }
2015 Request that DMA unmap operations use deferred
2016 invalidation of hardware TLBs, for increased
2017 throughput at the cost of reduced device isolation.
2018 Will fall back to strict mode if not supported by
2019 the relevant IOMMU driver.
2020 1 - Strict mode (default).
2021 DMA unmap operations invalidate IOMMU hardware TLBs
2025 [ARM64, X86] Configure DMA to bypass the IOMMU by default.
2026 Format: { "0" | "1" }
2027 0 - Use IOMMU translation for DMA.
2028 1 - Bypass the IOMMU for DMA.
2029 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
2031 io7= [HW] IO7 for Marvel-based Alpha systems
2032 See comment before marvel_specify_io7 in
2033 arch/alpha/kernel/core_marvel.c.
2035 io_delay= [X86] I/O delay method
2037 Standard port 0x80 based delay
2039 Alternate port 0xed based delay (needed on some systems)
2041 Simple two microseconds delay
2046 See Documentation/admin-guide/nfs/nfsroot.rst.
2048 ipcmni_extend [KNL] Extend the maximum number of unique System V
2049 IPC identifiers from 32,768 to 16,777,216.
2051 irqaffinity= [SMP] Set the default irq affinity mask
2052 The argument is a cpu list, as described above.
2054 irqchip.gicv2_force_probe=
2057 Force the kernel to look for the second 4kB page
2058 of a GICv2 controller even if the memory range
2059 exposed by the device tree is too small.
2061 irqchip.gicv3_nolpi=
2063 Force the kernel to ignore the availability of
2064 LPIs (and by consequence ITSs). Intended for system
2065 that use the kernel as a bootloader, and thus want
2066 to let secondary kernels in charge of setting up
2069 irqchip.gicv3_pseudo_nmi= [ARM64]
2070 Enables support for pseudo-NMIs in the kernel. This
2071 requires the kernel to be built with
2072 CONFIG_ARM64_PSEUDO_NMI.
2075 When an interrupt is not handled search all handlers
2076 for it. Intended to get systems with badly broken
2080 When an interrupt is not handled search all handlers
2081 for it. Also check all handlers each timer
2082 interrupt. Intended to get systems with badly broken
2086 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
2088 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
2089 [Deprecated - use cpusets instead]
2090 Format: [flag-list,]<cpu-list>
2092 Specify one or more CPUs to isolate from disturbances
2093 specified in the flag list (default: domain):
2096 Disable the tick when a single task runs.
2098 A residual 1Hz tick is offloaded to workqueues, which you
2099 need to affine to housekeeping through the global
2100 workqueue's affinity configured via the
2101 /sys/devices/virtual/workqueue/cpumask sysfs file, or
2102 by using the 'domain' flag described below.
2104 NOTE: by default the global workqueue runs on all CPUs,
2105 so to protect individual CPUs the 'cpumask' file has to
2106 be configured manually after bootup.
2109 Isolate from the general SMP balancing and scheduling
2110 algorithms. Note that performing domain isolation this way
2111 is irreversible: it's not possible to bring back a CPU to
2112 the domains once isolated through isolcpus. It's strongly
2113 advised to use cpusets instead to disable scheduler load
2114 balancing through the "cpuset.sched_load_balance" file.
2115 It offers a much more flexible interface where CPUs can
2116 move in and out of an isolated set anytime.
2118 You can move a process onto or off an "isolated" CPU via
2119 the CPU affinity syscalls or cpuset.
2120 <cpu number> begins at 0 and the maximum value is
2121 "number of CPUs in system - 1".
2125 Isolate from being targeted by managed interrupts
2126 which have an interrupt mask containing isolated
2127 CPUs. The affinity of managed interrupts is
2128 handled by the kernel and cannot be changed via
2129 the /proc/irq/* interfaces.
2131 This isolation is best effort and only effective
2132 if the automatically assigned interrupt mask of a
2133 device queue contains isolated and housekeeping
2134 CPUs. If housekeeping CPUs are online then such
2135 interrupts are directed to the housekeeping CPU
2136 so that IO submitted on the housekeeping CPU
2137 cannot disturb the isolated CPU.
2139 If a queue's affinity mask contains only isolated
2140 CPUs then this parameter has no effect on the
2141 interrupt routing decision, though interrupts are
2142 only delivered when tasks running on those
2143 isolated CPUs submit IO. IO submitted on
2144 housekeeping CPUs has no influence on those
2147 The format of <cpu-list> is described above.
2151 ivrs_ioapic [HW,X86-64]
2152 Provide an override to the IOAPIC-ID<->DEVICE-ID
2153 mapping provided in the IVRS ACPI table. For
2154 example, to map IOAPIC-ID decimal 10 to
2155 PCI device 00:14.0 write the parameter as:
2156 ivrs_ioapic[10]=00:14.0
2158 ivrs_hpet [HW,X86-64]
2159 Provide an override to the HPET-ID<->DEVICE-ID
2160 mapping provided in the IVRS ACPI table. For
2161 example, to map HPET-ID decimal 0 to
2162 PCI device 00:14.0 write the parameter as:
2163 ivrs_hpet[0]=00:14.0
2165 ivrs_acpihid [HW,X86-64]
2166 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2167 mapping provided in the IVRS ACPI table. For
2168 example, to map UART-HID:UID AMD0020:0 to
2169 PCI device 00:14.5 write the parameter as:
2170 ivrs_acpihid[00:14.5]=AMD0020:0
2172 js= [HW,JOY] Analog joystick
2173 See Documentation/input/joydev/joystick.rst.
2176 When CONFIG_RANDOMIZE_BASE is set, this disables
2177 kernel and module base offset ASLR (Address Space
2178 Layout Randomization).
2181 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2182 report on every invalid memory access. Without this
2183 parameter KASAN will print report only for the first
2188 kernelcore= [KNL,X86,IA-64,PPC]
2189 Format: nn[KMGTPE] | nn% | "mirror"
2190 This parameter specifies the amount of memory usable by
2191 the kernel for non-movable allocations. The requested
2192 amount is spread evenly throughout all nodes in the
2193 system as ZONE_NORMAL. The remaining memory is used for
2194 movable memory in its own zone, ZONE_MOVABLE. In the
2195 event, a node is too small to have both ZONE_NORMAL and
2196 ZONE_MOVABLE, kernelcore memory will take priority and
2197 other nodes will have a larger ZONE_MOVABLE.
2199 ZONE_MOVABLE is used for the allocation of pages that
2200 may be reclaimed or moved by the page migration
2201 subsystem. Note that allocations like PTEs-from-HighMem
2202 still use the HighMem zone if it exists, and the Normal
2203 zone if it does not.
2205 It is possible to specify the exact amount of memory in
2206 the form of "nn[KMGTPE]", a percentage of total system
2207 memory in the form of "nn%", or "mirror". If "mirror"
2208 option is specified, mirrored (reliable) memory is used
2209 for non-movable allocations and remaining memory is used
2210 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2211 are exclusive, so you cannot specify multiple forms.
2213 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
2214 Format: <Controller#>[,poll interval]
2215 The controller # is the number of the ehci usb debug
2216 port as it is probed via PCI. The poll interval is
2217 optional and is the number seconds in between
2218 each poll cycle to the debug port in case you need
2219 the functionality for interrupting the kernel with
2220 gdb or control-c on the dbgp connection. When
2221 not using this parameter you use sysrq-g to break into
2222 the kernel debugger.
2224 kgdboc= [KGDB,HW] kgdb over consoles.
2225 Requires a tty driver that supports console polling,
2226 or a supported polling keyboard driver (non-usb).
2227 Serial only format: <serial_device>[,baud]
2228 keyboard only format: kbd
2229 keyboard and serial format: kbd,<serial_device>[,baud]
2230 Optional Kernel mode setting:
2231 kms, kbd format: kms,kbd
2232 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2234 kgdboc_earlycon= [KGDB,HW]
2235 If the boot console provides the ability to read
2236 characters and can work in polling mode, you can use
2237 this parameter to tell kgdb to use it as a backend
2238 until the normal console is registered. Intended to
2239 be used together with the kgdboc parameter which
2240 specifies the normal console to transition to.
2242 The name of the early console should be specified
2243 as the value of this parameter. Note that the name of
2244 the early console might be different than the tty
2245 name passed to kgdboc. It's OK to leave the value
2246 blank and the first boot console that implements
2247 read() will be picked.
2249 kgdbwait [KGDB] Stop kernel execution and enter the
2250 kernel debugger at the earliest opportunity.
2252 kmac= [MIPS] Korina ethernet MAC address.
2253 Configure the RouterBoard 532 series on-chip
2254 Ethernet adapter MAC address.
2256 kmemleak= [KNL] Boot-time kmemleak enable/disable
2257 Valid arguments: on, off
2259 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2262 kprobe_event=[probe-list]
2263 [FTRACE] Add kprobe events and enable at boot time.
2264 The probe-list is a semicolon delimited list of probe
2265 definitions. Each definition is same as kprobe_events
2266 interface, but the parameters are comma delimited.
2267 For example, to add a kprobe event on vfs_read with
2268 arg1 and arg2, add to the command line;
2270 kprobe_event=p,vfs_read,$arg1,$arg2
2272 See also Documentation/trace/kprobetrace.rst "Kernel
2273 Boot Parameter" section.
2275 kpti= [ARM64] Control page table isolation of user
2276 and kernel address spaces.
2277 Default: enabled on cores which need mitigation.
2281 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2282 Default is 0 (don't ignore, but inject #GP)
2284 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2285 Default is false (don't support).
2287 kvm.mmu_audit= [KVM] This is a R/W parameter which allows audit
2292 [KVM] Controls the software workaround for the
2293 X86_BUG_ITLB_MULTIHIT bug.
2294 force : Always deploy workaround.
2295 off : Never deploy workaround.
2296 auto : Deploy workaround based on the presence of
2297 X86_BUG_ITLB_MULTIHIT.
2301 If the software workaround is enabled for the host,
2302 guests do need not to enable it for nested guests.
2304 kvm.nx_huge_pages_recovery_ratio=
2305 [KVM] Controls how many 4KiB pages are periodically zapped
2306 back to huge pages. 0 disables the recovery, otherwise if
2307 the value is N KVM will zap 1/Nth of the 4KiB pages every
2308 minute. The default is 60.
2310 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM.
2311 Default is 1 (enabled)
2313 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU)
2315 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode.
2318 [KVM,ARM] Select one of KVM/arm64's modes of operation.
2320 nvhe: Standard nVHE-based mode, without support for
2323 protected: nVHE-based mode with support for guests whose
2324 state is kept private from the host.
2325 Not valid if the kernel is running in EL2.
2327 Defaults to VHE/nVHE based on hardware support.
2329 kvm-arm.vgic_v3_group0_trap=
2330 [KVM,ARM] Trap guest accesses to GICv3 group-0
2333 kvm-arm.vgic_v3_group1_trap=
2334 [KVM,ARM] Trap guest accesses to GICv3 group-1
2337 kvm-arm.vgic_v3_common_trap=
2338 [KVM,ARM] Trap guest accesses to GICv3 common
2341 kvm-arm.vgic_v4_enable=
2342 [KVM,ARM] Allow use of GICv4 for direct injection of
2345 kvm_cma_resv_ratio=n [PPC]
2346 Reserves given percentage from system memory area for
2347 contiguous memory allocation for KVM hash pagetable
2349 By default it reserves 5% of total system memory.
2353 kvm-intel.ept= [KVM,Intel] Disable extended page tables
2354 (virtualized MMU) support on capable Intel chips.
2355 Default is 1 (enabled)
2357 kvm-intel.emulate_invalid_guest_state=
2358 [KVM,Intel] Enable emulation of invalid guest states
2359 Default is 0 (disabled)
2361 kvm-intel.flexpriority=
2362 [KVM,Intel] Disable FlexPriority feature (TPR shadow).
2363 Default is 1 (enabled)
2366 [KVM,Intel] Enable VMX nesting (nVMX).
2367 Default is 0 (disabled)
2369 kvm-intel.unrestricted_guest=
2370 [KVM,Intel] Disable unrestricted guest feature
2371 (virtualized real and unpaged mode) on capable
2372 Intel chips. Default is 1 (enabled)
2374 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2377 Valid arguments: never, cond, always
2379 always: L1D cache flush on every VMENTER.
2380 cond: Flush L1D on VMENTER only when the code between
2381 VMEXIT and VMENTER can leak host memory.
2382 never: Disables the mitigation
2384 Default is cond (do L1 cache flush in specific instances)
2386 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification
2387 feature (tagged TLBs) on capable Intel chips.
2388 Default is 1 (enabled)
2390 l1tf= [X86] Control mitigation of the L1TF vulnerability on
2393 The kernel PTE inversion protection is unconditionally
2394 enabled and cannot be disabled.
2397 Provides all available mitigations for the
2398 L1TF vulnerability. Disables SMT and
2399 enables all mitigations in the
2400 hypervisors, i.e. unconditional L1D flush.
2402 SMT control and L1D flush control via the
2403 sysfs interface is still possible after
2404 boot. Hypervisors will issue a warning
2405 when the first VM is started in a
2406 potentially insecure configuration,
2407 i.e. SMT enabled or L1D flush disabled.
2410 Same as 'full', but disables SMT and L1D
2411 flush runtime control. Implies the
2412 'nosmt=force' command line option.
2413 (i.e. sysfs control of SMT is disabled.)
2416 Leaves SMT enabled and enables the default
2417 hypervisor mitigation, i.e. conditional
2420 SMT control and L1D flush control via the
2421 sysfs interface is still possible after
2422 boot. Hypervisors will issue a warning
2423 when the first VM is started in a
2424 potentially insecure configuration,
2425 i.e. SMT enabled or L1D flush disabled.
2429 Disables SMT and enables the default
2430 hypervisor mitigation.
2432 SMT control and L1D flush control via the
2433 sysfs interface is still possible after
2434 boot. Hypervisors will issue a warning
2435 when the first VM is started in a
2436 potentially insecure configuration,
2437 i.e. SMT enabled or L1D flush disabled.
2440 Same as 'flush', but hypervisors will not
2441 warn when a VM is started in a potentially
2442 insecure configuration.
2445 Disables hypervisor mitigations and doesn't
2447 It also drops the swap size and available
2448 RAM limit restriction on both hypervisor and
2453 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2459 lapic [X86-32,APIC] Enable the local APIC even if BIOS
2462 lapic= [X86,APIC] Do not use TSC deadline
2463 value for LAPIC timer one-shot implementation. Default
2464 back to the programmable timer unit in the LAPIC.
2465 Format: notscdeadline
2467 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
2470 libata.dma= [LIBATA] DMA control
2471 libata.dma=0 Disable all PATA and SATA DMA
2472 libata.dma=1 PATA and SATA Disk DMA only
2473 libata.dma=2 ATAPI (CDROM) DMA only
2474 libata.dma=4 Compact Flash DMA only
2475 Combinations also work, so libata.dma=3 enables DMA
2476 for disks and CDROMs, but not CFs.
2478 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2479 libata.ignore_hpa=0 keep BIOS limits (default)
2480 libata.ignore_hpa=1 ignore limits, using full disk
2482 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2486 libata.force= [LIBATA] Force configurations. The format is comma-
2487 separated list of "[ID:]VAL" where ID is
2488 PORT[.DEVICE]. PORT and DEVICE are decimal numbers
2489 matching port, link or device. Basically, it matches
2490 the ATA ID string printed on console by libata. If
2491 the whole ID part is omitted, the last PORT and DEVICE
2492 values are used. If ID hasn't been specified yet, the
2493 configuration applies to all ports, links and devices.
2495 If only DEVICE is omitted, the parameter applies to
2496 the port and all links and devices behind it. DEVICE
2497 number of 0 either selects the first device or the
2498 first fan-out link behind PMP device. It does not
2499 select the host link. DEVICE number of 15 selects the
2500 host link and device attached to it.
2502 The VAL specifies the configuration to force. As long
2503 as there's no ambiguity shortcut notation is allowed.
2504 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2505 The following configurations can be forced.
2507 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2508 Any ID with matching PORT is used.
2510 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2512 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2513 udma[/][16,25,33,44,66,100,133] notation is also
2516 * [no]ncq: Turn on or off NCQ.
2518 * [no]ncqtrim: Turn off queued DSM TRIM.
2520 * nohrst, nosrst, norst: suppress hard, soft
2523 * rstonce: only attempt one reset during
2524 hot-unplug link recovery
2526 * dump_id: dump IDENTIFY data.
2528 * atapi_dmadir: Enable ATAPI DMADIR bridge support
2530 * disable: Disable this device.
2532 If there are multiple matching configurations changing
2533 the same attribute, the last one is used.
2535 memblock=debug [KNL] Enable memblock debug messages.
2537 load_ramdisk= [RAM] [Deprecated]
2539 lockd.nlm_grace_period=P [NFS] Assign grace period.
2542 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2545 lockd.nlm_timeout=T [NFS] Assign timeout value.
2548 lockd.nlm_udpport=M [NFS] Assign UDP port.
2551 lockdown= [SECURITY]
2552 { integrity | confidentiality }
2553 Enable the kernel lockdown feature. If set to
2554 integrity, kernel features that allow userland to
2555 modify the running kernel are disabled. If set to
2556 confidentiality, kernel features that allow userland
2557 to extract confidential information from the kernel
2560 locktorture.nreaders_stress= [KNL]
2561 Set the number of locking read-acquisition kthreads.
2562 Defaults to being automatically set based on the
2563 number of online CPUs.
2565 locktorture.nwriters_stress= [KNL]
2566 Set the number of locking write-acquisition kthreads.
2568 locktorture.onoff_holdoff= [KNL]
2569 Set time (s) after boot for CPU-hotplug testing.
2571 locktorture.onoff_interval= [KNL]
2572 Set time (s) between CPU-hotplug operations, or
2573 zero to disable CPU-hotplug testing.
2575 locktorture.shuffle_interval= [KNL]
2576 Set task-shuffle interval (jiffies). Shuffling
2577 tasks allows some CPUs to go into dyntick-idle
2578 mode during the locktorture test.
2580 locktorture.shutdown_secs= [KNL]
2581 Set time (s) after boot system shutdown. This
2582 is useful for hands-off automated testing.
2584 locktorture.stat_interval= [KNL]
2585 Time (s) between statistics printk()s.
2587 locktorture.stutter= [KNL]
2588 Time (s) to stutter testing, for example,
2589 specifying five seconds causes the test to run for
2590 five seconds, wait for five seconds, and so on.
2591 This tests the locking primitive's ability to
2592 transition abruptly to and from idle.
2594 locktorture.torture_type= [KNL]
2595 Specify the locking implementation to test.
2597 locktorture.verbose= [KNL]
2598 Enable additional printk() statements.
2600 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
2603 loglevel= All Kernel Messages with a loglevel smaller than the
2604 console loglevel will be printed to the console. It can
2605 also be changed with klogd or other programs. The
2606 loglevels are defined as follows:
2608 0 (KERN_EMERG) system is unusable
2609 1 (KERN_ALERT) action must be taken immediately
2610 2 (KERN_CRIT) critical conditions
2611 3 (KERN_ERR) error conditions
2612 4 (KERN_WARNING) warning conditions
2613 5 (KERN_NOTICE) normal but significant condition
2614 6 (KERN_INFO) informational
2615 7 (KERN_DEBUG) debug-level messages
2617 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
2618 in bytes. n must be a power of two and greater
2619 than the minimal size. The minimal size is defined
2620 by LOG_BUF_SHIFT kernel config parameter. There is
2621 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2622 that allows to increase the default size depending on
2623 the number of CPUs. See init/Kconfig for more details.
2625 logo.nologo [FB] Disables display of the built-in Linux logo.
2626 This may be used to provide more screen space for
2627 kernel log messages and is useful when debugging
2628 kernel boot problems.
2630 lp=0 [LP] Specify parallel ports to use, e.g,
2631 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2632 lp=reset first parallel port). 'lp=0' disables the
2633 lp=auto printer driver. 'lp=reset' (which can be
2634 specified in addition to the ports) causes
2635 attached printers to be reset. Using
2636 lp=port1,port2,... specifies the parallel ports
2637 to associate lp devices with, starting with
2638 lp0. A port specification may be 'none' to skip
2639 that lp device, or a parport name such as
2640 'parport0'. Specifying 'lp=auto' instead of a
2641 port specification list means that device IDs
2642 from each port should be examined, to see if
2643 an IEEE 1284-compliant printer is attached; if
2644 so, the driver will manage that printer.
2645 See also header of drivers/char/lp.c.
2648 Sets loops_per_jiffy to given constant, thus avoiding
2649 time-consuming boot-time autodetection (up to 250 ms per
2650 CPU). 0 enables autodetection (default). To determine
2651 the correct value for your kernel, boot with normal
2652 autodetection and see what value is printed. Note that
2653 on SMP systems the preset will be applied to all CPUs,
2654 which is likely to cause problems if your CPUs need
2655 significantly divergent settings. An incorrect value
2656 will cause delays in the kernel to be wrong, leading to
2657 unpredictable I/O errors and other breakage. Although
2658 unlikely, in the extreme case this might damage your
2662 Format: <io>,<irq>,<dma>
2664 lsm.debug [SECURITY] Enable LSM initialization debugging output.
2667 [SECURITY] Choose order of LSM initialization. This
2668 overrides CONFIG_LSM, and the "security=" parameter.
2670 machvec= [IA-64] Force the use of a particular machine-vector
2671 (machvec) in a generic kernel.
2672 Example: machvec=hpzx1
2674 machtype= [Loongson] Share the same kernel image file between
2675 different yeeloong laptops.
2676 Example: machtype=lemote-yeeloong-2f-7inch
2678 max_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory greater
2679 than or equal to this physical address is ignored.
2681 maxcpus= [SMP] Maximum number of processors that an SMP kernel
2682 will bring up during bootup. maxcpus=n : n >= 0 limits
2683 the kernel to bring up 'n' processors. Surely after
2684 bootup you can bring up the other plugged cpu by executing
2685 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
2686 only takes effect during system bootup.
2687 While n=0 is a special case, it is equivalent to "nosmp",
2688 which also disables the IO APIC.
2690 max_loop= [LOOP] The number of loop block devices that get
2691 (loop.max_loop) unconditionally pre-created at init time. The default
2692 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
2693 of statically allocating a predefined number, loop
2694 devices can be requested on-demand with the
2695 /dev/loop-control interface.
2697 mce [X86-32] Machine Check Exception
2699 mce=option [X86-64] See Documentation/x86/x86_64/boot-options.rst
2701 md= [HW] RAID subsystems devices and level
2702 See Documentation/admin-guide/md.rst.
2705 Format: <first>,<last>
2706 Specifies range of consoles to be captured by the MDA.
2709 Control mitigation for the Micro-architectural Data
2710 Sampling (MDS) vulnerability.
2712 Certain CPUs are vulnerable to an exploit against CPU
2713 internal buffers which can forward information to a
2714 disclosure gadget under certain conditions.
2716 In vulnerable processors, the speculatively
2717 forwarded data can be used in a cache side channel
2718 attack, to access data to which the attacker does
2719 not have direct access.
2721 This parameter controls the MDS mitigation. The
2724 full - Enable MDS mitigation on vulnerable CPUs
2725 full,nosmt - Enable MDS mitigation and disable
2726 SMT on vulnerable CPUs
2727 off - Unconditionally disable MDS mitigation
2729 On TAA-affected machines, mds=off can be prevented by
2730 an active TAA mitigation as both vulnerabilities are
2731 mitigated with the same mechanism so in order to disable
2732 this mitigation, you need to specify tsx_async_abort=off
2735 Not specifying this option is equivalent to
2738 For details see: Documentation/admin-guide/hw-vuln/mds.rst
2740 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
2741 Amount of memory to be used in cases as follows:
2744 2 when the kernel is not able to see the whole system memory;
2745 3 memory that lies after 'mem=' boundary is excluded from
2746 the hypervisor, then assigned to KVM guests.
2748 [X86] Work as limiting max address. Use together
2749 with memmap= to avoid physical address space collisions.
2750 Without memmap= PCI devices could be placed at addresses
2751 belonging to unused RAM.
2753 Note that this only takes effects during boot time since
2754 in above case 3, memory may need be hot added after boot
2755 if system memory of hypervisor is not sufficient.
2757 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
2761 [KNL,SH] Allow user to override the default size for
2762 per-device physically contiguous DMA buffers.
2764 memhp_default_state=online/offline
2765 [KNL] Set the initial state for the memory hotplug
2766 onlining policy. If not specified, the default value is
2767 set according to the
2768 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
2770 See Documentation/admin-guide/mm/memory-hotplug.rst.
2772 memmap=exactmap [KNL,X86] Enable setting of an exact
2773 E820 memory map, as specified by the user.
2774 Such memmap=exactmap lines can be constructed based on
2775 BIOS output or other requirements. See the memmap=nn@ss
2778 memmap=nn[KMG]@ss[KMG]
2779 [KNL, X86, MIPS, XTENSA] Force usage of a specific region of memory.
2780 Region of memory to be used is from ss to ss+nn.
2781 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
2782 which limits max address to nn[KMG].
2783 Multiple different regions can be specified,
2786 memmap=100M@2G,100M#3G,1G!1024G
2788 memmap=nn[KMG]#ss[KMG]
2789 [KNL,ACPI] Mark specific memory as ACPI data.
2790 Region of memory to be marked is from ss to ss+nn.
2792 memmap=nn[KMG]$ss[KMG]
2793 [KNL,ACPI] Mark specific memory as reserved.
2794 Region of memory to be reserved is from ss to ss+nn.
2795 Example: Exclude memory from 0x18690000-0x1869ffff
2796 memmap=64K$0x18690000
2798 memmap=0x10000$0x18690000
2799 Some bootloaders may need an escape character before '$',
2800 like Grub2, otherwise '$' and the following number
2803 memmap=nn[KMG]!ss[KMG]
2804 [KNL,X86] Mark specific memory as protected.
2805 Region of memory to be used, from ss to ss+nn.
2806 The memory region may be marked as e820 type 12 (0xc)
2807 and is NVDIMM or ADR memory.
2809 memmap=<size>%<offset>-<oldtype>+<newtype>
2810 [KNL,ACPI] Convert memory within the specified region
2811 from <oldtype> to <newtype>. If "-<oldtype>" is left
2812 out, the whole region will be marked as <newtype>,
2813 even if previously unavailable. If "+<newtype>" is left
2814 out, matching memory will be removed. Types are
2815 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
2816 3 = ACPI, 12 = PRAM.
2818 memory_corruption_check=0/1 [X86]
2819 Some BIOSes seem to corrupt the first 64k of
2820 memory when doing things like suspend/resume.
2821 Setting this option will scan the memory
2822 looking for corruption. Enabling this will
2823 both detect corruption and prevent the kernel
2824 from using the memory being corrupted.
2825 However, its intended as a diagnostic tool; if
2826 repeatable BIOS-originated corruption always
2827 affects the same memory, you can use memmap=
2828 to prevent the kernel from using that memory.
2830 memory_corruption_check_size=size [X86]
2831 By default it checks for corruption in the low
2832 64k, making this memory unavailable for normal
2833 use. Use this parameter to scan for
2834 corruption in more or less memory.
2836 memory_corruption_check_period=seconds [X86]
2837 By default it checks for corruption every 60
2838 seconds. Use this parameter to check at some
2839 other rate. 0 disables periodic checking.
2841 memory_hotplug.memmap_on_memory
2842 [KNL,X86,ARM] Boolean flag to enable this feature.
2843 Format: {on | off (default)}
2844 When enabled, runtime hotplugged memory will
2845 allocate its internal metadata (struct pages)
2846 from the hotadded memory which will allow to
2847 hotadd a lot of memory without requiring
2848 additional memory to do so.
2849 This feature is disabled by default because it
2850 has some implication on large (e.g. GB)
2851 allocations in some configurations (e.g. small
2853 The state of the flag can be read in
2854 /sys/module/memory_hotplug/parameters/memmap_on_memory.
2855 Note that even when enabled, there are a few cases where
2856 the feature is not effective.
2858 memtest= [KNL,X86,ARM,PPC,RISCV] Enable memtest
2860 default : 0 <disable>
2861 Specifies the number of memtest passes to be
2862 performed. Each pass selects another test
2863 pattern from a given set of patterns. Memtest
2864 fills the memory with this pattern, validates
2865 memory contents and reserves bad memory
2866 regions that are detected.
2868 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
2869 Valid arguments: on, off
2870 Default (depends on kernel configuration option):
2871 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
2872 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
2873 mem_encrypt=on: Activate SME
2874 mem_encrypt=off: Do not activate SME
2876 Refer to Documentation/virt/kvm/amd-memory-encryption.rst
2877 for details on when memory encryption can be activated.
2879 mem_sleep_default= [SUSPEND] Default system suspend mode:
2880 s2idle - Suspend-To-Idle
2881 shallow - Power-On Suspend or equivalent (if supported)
2882 deep - Suspend-To-RAM or equivalent (if supported)
2883 See Documentation/admin-guide/pm/sleep-states.rst.
2885 meye.*= [HW] Set MotionEye Camera parameters
2886 See Documentation/admin-guide/media/meye.rst.
2888 mfgpt_irq= [IA-32] Specify the IRQ to use for the
2889 Multi-Function General Purpose Timers on AMD Geode
2892 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
2893 the BIOS has incorrectly applied a workaround. TinyBIOS
2894 version 0.98 is known to be affected, 0.99 fixes the
2895 problem by letting the user disable the workaround.
2899 min_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory below this
2900 physical address is ignored.
2902 mini2440= [ARM,HW,KNL]
2903 Format:[0..2][b][c][t]
2905 MINI2440 configuration specification:
2906 0 - The attached screen is the 3.5" TFT
2907 1 - The attached screen is the 7" TFT
2908 2 - The VGA Shield is attached (1024x768)
2909 Leaving out the screen size parameter will not load
2910 the TFT driver, and the framebuffer will be left
2912 b - Enable backlight. The TFT backlight pin will be
2913 linked to the kernel VESA blanking code and a GPIO
2914 LED. This parameter is not necessary when using the
2916 c - Enable the s3c camera interface.
2917 t - Reserved for enabling touchscreen support. The
2918 touchscreen support is not enabled in the mainstream
2919 kernel as of 2.6.30, a preliminary port can be found
2920 in the "bleeding edge" mini2440 support kernel at
2921 https://repo.or.cz/w/linux-2.6/mini2440.git
2924 [X86,PPC,S390,ARM64] Control optional mitigations for
2925 CPU vulnerabilities. This is a set of curated,
2926 arch-independent options, each of which is an
2927 aggregation of existing arch-specific options.
2930 Disable all optional CPU mitigations. This
2931 improves system performance, but it may also
2932 expose users to several CPU vulnerabilities.
2933 Equivalent to: nopti [X86,PPC]
2935 nospectre_v1 [X86,PPC]
2937 nospectre_v2 [X86,PPC,S390,ARM64]
2938 spectre_v2_user=off [X86]
2939 spec_store_bypass_disable=off [X86,PPC]
2940 ssbd=force-off [ARM64]
2943 tsx_async_abort=off [X86]
2944 kvm.nx_huge_pages=off [X86]
2945 no_entry_flush [PPC]
2946 no_uaccess_flush [PPC]
2949 This does not have any effect on
2950 kvm.nx_huge_pages when
2951 kvm.nx_huge_pages=force.
2954 Mitigate all CPU vulnerabilities, but leave SMT
2955 enabled, even if it's vulnerable. This is for
2956 users who don't want to be surprised by SMT
2957 getting disabled across kernel upgrades, or who
2958 have other ways of avoiding SMT-based attacks.
2959 Equivalent to: (default behavior)
2962 Mitigate all CPU vulnerabilities, disabling SMT
2963 if needed. This is for users who always want to
2964 be fully mitigated, even if it means losing SMT.
2965 Equivalent to: l1tf=flush,nosmt [X86]
2966 mds=full,nosmt [X86]
2967 tsx_async_abort=full,nosmt [X86]
2970 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
2971 parameter allows control of the logging verbosity for
2972 the additional memory initialisation checks. A value
2973 of 0 disables mminit logging and a level of 4 will
2974 log everything. Information is printed at KERN_DEBUG
2975 so loglevel=8 may also need to be specified.
2978 [KNL] When CONFIG_MODULE_SIG is set, this means that
2979 modules without (valid) signatures will fail to load.
2980 Note that if CONFIG_MODULE_SIG_FORCE is set, that
2981 is always true, so this option does nothing.
2983 module_blacklist= [KNL] Do not load a comma-separated list of
2984 modules. Useful for debugging problem modules.
2987 [MOUSE] Maximum time between finger touching and
2988 leaving touchpad surface for touch to be considered
2989 a tap and be reported as a left button click (for
2990 touchpads working in absolute mode only).
2992 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
2993 reporting absolute coordinates, such as tablets
2994 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
2995 reporting absolute coordinates, such as tablets
2997 movablecore= [KNL,X86,IA-64,PPC]
2998 Format: nn[KMGTPE] | nn%
2999 This parameter is the complement to kernelcore=, it
3000 specifies the amount of memory used for migratable
3001 allocations. If both kernelcore and movablecore is
3002 specified, then kernelcore will be at *least* the
3003 specified value but may be more. If movablecore on its
3004 own is specified, the administrator must be careful
3005 that the amount of memory usable for all allocations
3008 movable_node [KNL] Boot-time switch to make hotplugable memory
3009 NUMA nodes to be movable. This means that the memory
3010 of such nodes will be usable only for movable
3011 allocations which rules out almost all kernel
3012 allocations. Use with caution!
3014 MTD_Partition= [MTD]
3015 Format: <name>,<region-number>,<size>,<offset>
3017 MTD_Region= [MTD] Format:
3018 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
3021 See drivers/mtd/parsers/cmdlinepart.c
3023 multitce=off [PPC] This parameter disables the use of the pSeries
3024 firmware feature for updating multiple TCE entries
3027 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
3029 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
3031 boundary - index of last SLC block on Flex-OneNAND.
3032 The remaining blocks are configured as MLC blocks.
3033 lock - Configure if Flex-OneNAND boundary should be locked.
3034 Once locked, the boundary cannot be changed.
3035 1 indicates lock status, 0 indicates unlock status.
3038 ARM/S3C2412 JIVE boot control
3040 See arch/arm/mach-s3c/mach-jive.c
3042 mtouchusb.raw_coordinates=
3043 [HW] Make the MicroTouch USB driver use raw coordinates
3044 ('y', default) or cooked coordinates ('n')
3046 mtrr_chunk_size=nn[KMG] [X86]
3047 used for mtrr cleanup. It is largest continuous chunk
3048 that could hold holes aka. UC entries.
3050 mtrr_gran_size=nn[KMG] [X86]
3051 Used for mtrr cleanup. It is granularity of mtrr block.
3053 Large value could prevent small alignment from
3056 mtrr_spare_reg_nr=n [X86]
3058 Range: 0,7 : spare reg number
3060 Used for mtrr cleanup. It is spare mtrr entries number.
3061 Set to 2 or more if your graphical card needs more.
3063 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
3065 netdev= [NET] Network devices parameters
3066 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
3067 Note that mem_start is often overloaded to mean
3068 something different and driver-specific.
3069 This usage is only documented in each driver source
3073 [NETFILTER] Enable connection tracking flow accounting
3074 0 to disable accounting
3075 1 to enable accounting
3078 nfsaddrs= [NFS] Deprecated. Use ip= instead.
3079 See Documentation/admin-guide/nfs/nfsroot.rst.
3081 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
3082 See Documentation/admin-guide/nfs/nfsroot.rst.
3084 nfsrootdebug [NFS] enable nfsroot debugging messages.
3085 See Documentation/admin-guide/nfs/nfsroot.rst.
3087 nfs.callback_nr_threads=
3088 [NFSv4] set the total number of threads that the
3089 NFS client will assign to service NFSv4 callback
3092 nfs.callback_tcpport=
3093 [NFS] set the TCP port on which the NFSv4 callback
3094 channel should listen.
3097 [NFS] sets the pathname to the program which is used
3098 to update the NFS client cache entries.
3100 nfs.cache_getent_timeout=
3101 [NFS] sets the timeout after which an attempt to
3102 update a cache entry is deemed to have failed.
3104 nfs.idmap_cache_timeout=
3105 [NFS] set the maximum lifetime for idmapper cache
3109 [NFS] enable 64-bit inode numbers.
3110 If zero, the NFS client will fake up a 32-bit inode
3111 number for the readdir() and stat() syscalls instead
3112 of returning the full 64-bit number.
3113 The default is to return 64-bit inode numbers.
3115 nfs.max_session_cb_slots=
3116 [NFSv4.1] Sets the maximum number of session
3117 slots the client will assign to the callback
3118 channel. This determines the maximum number of
3119 callbacks the client will process in parallel for
3120 a particular server.
3122 nfs.max_session_slots=
3123 [NFSv4.1] Sets the maximum number of session slots
3124 the client will attempt to negotiate with the server.
3125 This limits the number of simultaneous RPC requests
3126 that the client can send to the NFSv4.1 server.
3127 Note that there is little point in setting this
3128 value higher than the max_tcp_slot_table_limit.
3130 nfs.nfs4_disable_idmapping=
3131 [NFSv4] When set to the default of '1', this option
3132 ensures that both the RPC level authentication
3133 scheme and the NFS level operations agree to use
3134 numeric uids/gids if the mount is using the
3135 'sec=sys' security flavour. In effect it is
3136 disabling idmapping, which can make migration from
3137 legacy NFSv2/v3 systems to NFSv4 easier.
3138 Servers that do not support this mode of operation
3139 will be autodetected by the client, and it will fall
3140 back to using the idmapper.
3141 To turn off this behaviour, set the value to '0'.
3143 [NFS4] Specify an additional fixed unique ident-
3144 ification string that NFSv4 clients can insert into
3145 their nfs_client_id4 string. This is typically a
3146 UUID that is generated at system install time.
3148 nfs.send_implementation_id =
3149 [NFSv4.1] Send client implementation identification
3150 information in exchange_id requests.
3151 If zero, no implementation identification information
3153 The default is to send the implementation identification
3156 nfs.recover_lost_locks =
3157 [NFSv4] Attempt to recover locks that were lost due
3158 to a lease timeout on the server. Please note that
3159 doing this risks data corruption, since there are
3160 no guarantees that the file will remain unchanged
3161 after the locks are lost.
3162 If you want to enable the kernel legacy behaviour of
3163 attempting to recover these locks, then set this
3165 The default parameter value of '0' causes the kernel
3166 not to attempt recovery of lost locks.
3168 nfs4.layoutstats_timer =
3169 [NFSv4.2] Change the rate at which the kernel sends
3170 layoutstats to the pNFS metadata server.
3172 Setting this to value to 0 causes the kernel to use
3173 whatever value is the default set by the layout
3174 driver. A non-zero value sets the minimum interval
3175 in seconds between layoutstats transmissions.
3177 nfsd.nfs4_disable_idmapping=
3178 [NFSv4] When set to the default of '1', the NFSv4
3179 server will return only numeric uids and gids to
3180 clients using auth_sys, and will accept numeric uids
3181 and gids from such clients. This is intended to ease
3182 migration from NFSv2/v3.
3184 nmi_backtrace.backtrace_idle [KNL]
3185 Dump stacks even of idle CPUs in response to an
3186 NMI stack-backtrace request.
3188 nmi_debug= [KNL,SH] Specify one or more actions to take
3189 when a NMI is triggered.
3190 Format: [state][,regs][,debounce][,die]
3192 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
3193 Format: [panic,][nopanic,][num]
3195 0 - turn hardlockup detector in nmi_watchdog off
3196 1 - turn hardlockup detector in nmi_watchdog on
3197 When panic is specified, panic when an NMI watchdog
3198 timeout occurs (or 'nopanic' to not panic on an NMI
3199 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
3200 To disable both hard and soft lockup detectors,
3201 please see 'nowatchdog'.
3202 This is useful when you use a panic=... timeout and
3203 need the box quickly up again.
3205 These settings can be accessed at runtime via
3206 the nmi_watchdog and hardlockup_panic sysctls.
3208 netpoll.carrier_timeout=
3209 [NET] Specifies amount of time (in seconds) that
3210 netpoll should wait for a carrier. By default netpoll
3213 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3214 emulation library even if a 387 maths coprocessor
3217 no5lvl [X86-64] Disable 5-level paging mode. Forces
3218 kernel to use 4-level paging instead.
3220 nofsgsbase [X86] Disables FSGSBASE instructions.
3223 [HW] Never suspend the console
3224 Disable suspending of consoles during suspend and
3225 hibernate operations. Once disabled, debugging
3226 messages can reach various consoles while the rest
3227 of the system is being put to sleep (ie, while
3228 debugging driver suspend/resume hooks). This may
3229 not work reliably with all consoles, but is known
3230 to work with serial and VGA consoles.
3231 To facilitate more flexible debugging, we also add
3232 console_suspend, a printk module parameter to control
3233 it. Users could use console_suspend (usually
3234 /sys/module/printk/parameters/console_suspend) to
3235 turn on/off it dynamically.
3237 novmcoredd [KNL,KDUMP]
3238 Disable device dump. Device dump allows drivers to
3239 append dump data to vmcore so you can collect driver
3240 specified debug info. Drivers can append the data
3241 without any limit and this data is stored in memory,
3242 so this may cause significant memory stress. Disabling
3243 device dump can help save memory but the driver debug
3244 data will be no longer available. This parameter
3245 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
3248 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
3249 caches in the slab allocator. Saves per-node memory,
3250 but will impact performance.
3254 noaltinstr [S390] Disables alternative instructions patching
3255 (CPU alternatives feature).
3257 noapic [SMP,APIC] Tells the kernel to not make use of any
3258 IOAPICs that may be present in the system.
3260 noautogroup Disable scheduler automatic task group creation.
3262 nobats [PPC] Do not use BATs for mapping kernel lowmem
3263 on "Classic" PPC cores.
3267 noclflush [BUGS=X86] Don't use the CLFLUSH instruction
3269 delayacct [KNL] Enable per-task delay accounting
3271 nodsp [SH] Disable hardware DSP at boot time.
3273 noefi Disable EFI runtime services support.
3275 no_entry_flush [PPC] Don't flush the L1-D cache when entering the kernel.
3280 On X86-32 available only on PAE configured kernels.
3281 noexec=on: enable non-executable mappings (default)
3282 noexec=off: disable non-executable mappings
3285 Disable SMAP (Supervisor Mode Access Prevention)
3286 even if it is supported by processor.
3289 Disable SMEP (Supervisor Mode Execution Prevention)
3290 even if it is supported by processor.
3293 This affects only 32-bit executables.
3294 noexec32=on: enable non-executable mappings (default)
3295 read doesn't imply executable mappings
3296 noexec32=off: disable non-executable mappings
3297 read implies executable mappings
3299 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3301 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3302 register save and restore. The kernel will only save
3303 legacy floating-point registers on task switch.
3305 nohugeiomap [KNL,X86,PPC,ARM64] Disable kernel huge I/O mappings.
3307 nohugevmalloc [PPC] Disable kernel huge vmalloc mappings.
3309 nosmt [KNL,S390] Disable symmetric multithreading (SMT).
3310 Equivalent to smt=1.
3312 [KNL,X86] Disable symmetric multithreading (SMT).
3313 nosmt=force: Force disable SMT, cannot be undone
3314 via the sysfs control file.
3316 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1
3317 (bounds check bypass). With this option data leaks are
3318 possible in the system.
3320 nospectre_v2 [X86,PPC_FSL_BOOK3E,ARM64] Disable all mitigations for
3321 the Spectre variant 2 (indirect branch prediction)
3322 vulnerability. System may allow data leaks with this
3325 nospec_store_bypass_disable
3326 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
3329 [PPC] Don't flush the L1-D cache after accessing user data.
3331 noxsave [BUGS=X86] Disables x86 extended register state save
3332 and restore using xsave. The kernel will fallback to
3333 enabling legacy floating-point and sse state.
3335 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
3336 register states. The kernel will fall back to use
3337 xsave to save the states. By using this parameter,
3338 performance of saving the states is degraded because
3339 xsave doesn't support modified optimization while
3340 xsaveopt supports it on xsaveopt enabled systems.
3342 noxsaves [X86] Disables xsaves and xrstors used in saving and
3343 restoring x86 extended register state in compacted
3344 form of xsave area. The kernel will fall back to use
3345 xsaveopt and xrstor to save and restore the states
3346 in standard form of xsave area. By using this
3347 parameter, xsave area per process might occupy more
3348 memory on xsaves enabled systems.
3350 nohlt [ARM,ARM64,MICROBLAZE,SH] Forces the kernel to busy wait
3351 in do_idle() and not use the arch_cpu_idle()
3352 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
3353 to be effective. This is useful on platforms where the
3354 sleep(SH) or wfi(ARM,ARM64) instructions do not work
3355 correctly or when doing power measurements to evalute
3356 the impact of the sleep instructions. This is also
3357 useful when using JTAG debugger.
3359 no_file_caps Tells the kernel not to honor file capabilities. The
3360 only way then for a file to be executed with privilege
3361 is to be setuid root or executed by root.
3363 nohalt [IA-64] Tells the kernel not to use the power saving
3364 function PAL_HALT_LIGHT when idle. This increases
3365 power-consumption. On the positive side, it reduces
3366 interrupt wake-up latency, which may improve performance
3367 in certain environments such as networked servers or
3371 Force pointers printed to the console or buffers to be
3372 unhashed. By default, when a pointer is printed via %p
3373 format string, that pointer is "hashed", i.e. obscured
3374 by hashing the pointer value. This is a security feature
3375 that hides actual kernel addresses from unprivileged
3376 users, but it also makes debugging the kernel more
3377 difficult since unequal pointers can no longer be
3378 compared. However, if this command-line option is
3379 specified, then all normal pointers will have their true
3380 value printed. Pointers printed via %pK may still be
3381 hashed. This option should only be specified when
3382 debugging the kernel. Please do not use on production
3385 nohibernate [HIBERNATION] Disable hibernation and resume.
3387 nohz= [KNL] Boottime enable/disable dynamic ticks
3388 Valid arguments: on, off
3391 nohz_full= [KNL,BOOT,SMP,ISOL]
3392 The argument is a cpu list, as described above.
3393 In kernels built with CONFIG_NO_HZ_FULL=y, set
3394 the specified list of CPUs whose tick will be stopped
3395 whenever possible. The boot CPU will be forced outside
3396 the range to maintain the timekeeping. Any CPUs
3397 in this list will have their RCU callbacks offloaded,
3398 just as if they had also been called out in the
3399 rcu_nocbs= boot parameter.
3401 noiotrap [SH] Disables trapped I/O port accesses.
3403 noirqdebug [X86-32] Disables the code which attempts to detect and
3404 disable unhandled interrupt sources.
3406 no_timer_check [X86,APIC] Disables the code which tests for
3407 broken timer IRQ sources.
3409 noisapnp [ISAPNP] Disables ISA PnP code.
3411 noinitrd [RAM] Tells the kernel not to load any configured
3414 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
3416 [Deprecated - use intremap=off]
3420 noinvpcid [X86] Disable the INVPCID cpu feature.
3422 nojitter [IA-64] Disables jitter checking for ITC timers.
3424 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
3426 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
3430 [X86,PV_OPS] Disable paravirtualized VMware scheduler
3431 clock and use the default one.
3433 no-steal-acc [X86,PV_OPS,ARM64] Disable paravirtualized steal time
3434 accounting. steal time is computed, but won't
3435 influence scheduler behaviour
3437 nolapic [X86-32,APIC] Do not enable or use the local APIC.
3439 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
3441 noltlbs [PPC] Do not use large page/tlb entries for kernel
3442 lowmem mapping on PPC40x and PPC8xx
3444 nomca [IA-64] Disable machine check abort handling
3446 nomce [X86-32] Disable Machine Check Exception
3448 nomfgpt [X86-32] Disable Multi-Function General Purpose
3449 Timer usage (for AMD Geode machines).
3451 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3452 shutdown the other cpus. Instead use the REBOOT_VECTOR
3455 nomodule Disable module load
3457 nopat [X86] Disable PAT (page attribute table extension of
3458 pagetables) support.
3460 nopcid [X86-64] Disable the PCID cpu feature.
3462 norandmaps Don't use address space randomization. Equivalent to
3463 echo 0 > /proc/sys/kernel/randomize_va_space
3465 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
3466 with UP alternatives
3468 nordrand [X86] Disable kernel use of the RDRAND and
3469 RDSEED instructions even if they are supported
3470 by the processor. RDRAND and RDSEED are still
3471 available to user space applications.
3473 noresume [SWSUSP] Disables resume and restores original swap
3476 no-scroll [VGA] Disables scrollback.
3477 This is required for the Braillex ib80-piezo Braille
3478 reader made by F.H. Papenmeier (Germany).
3482 nosep [BUGS=X86-32] Disables x86 SYSENTER/SYSEXIT support.
3484 nosgx [X86-64,SGX] Disables Intel SGX kernel support.
3486 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
3487 and disable the IO APIC. legacy for "maxcpus=0".
3489 nosoftlockup [KNL] Disable the soft-lockup detector.
3491 nosync [HW,M68K] Disables sync negotiation for all devices.
3493 nowatchdog [KNL] Disable both lockup detectors, i.e.
3494 soft-lockup and NMI watchdog (hard-lockup).
3498 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
3500 cpu0_hotplug [X86] Turn on CPU0 hotplug feature when
3501 CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off.
3502 Some features depend on CPU0. Known dependencies are:
3503 1. Resume from suspend/hibernate depends on CPU0.
3504 Suspend/hibernate will fail if CPU0 is offline and you
3505 need to online CPU0 before suspend/hibernate.
3506 2. PIC interrupts also depend on CPU0. CPU0 can't be
3507 removed if a PIC interrupt is detected.
3508 It's said poweroff/reboot may depend on CPU0 on some
3509 machines although I haven't seen such issues so far
3510 after CPU0 is offline on a few tested machines.
3511 If the dependencies are under your control, you can
3512 turn on cpu0_hotplug.
3514 nps_mtm_hs_ctr= [KNL,ARC]
3515 This parameter sets the maximum duration, in
3516 cycles, each HW thread of the CTOP can run
3517 without interruptions, before HW switches it.
3518 The actual maximum duration is 16 times this
3520 Format: integer between 1 and 255
3523 nptcg= [IA-64] Override max number of concurrent global TLB
3524 purges which is reported from either PAL_VM_SUMMARY or
3527 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
3528 could support. nr_cpus=n : n >= 1 limits the kernel to
3529 support 'n' processors. It could be larger than the
3530 number of already plugged CPU during bootup, later in
3531 runtime you can physically add extra cpu until it reaches
3532 n. So during boot up some boot time memory for per-cpu
3533 variables need be pre-allocated for later physical cpu
3536 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
3538 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86] Disable NUMA, Only
3539 set up a single NUMA node spanning all memory.
3541 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic
3543 Allowed values are enable and disable
3545 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
3546 'node', 'default' can be specified
3547 This can be set from sysctl after boot.
3548 See Documentation/admin-guide/sysctl/vm.rst for details.
3550 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
3551 See Documentation/core-api/debugging-via-ohci1394.rst for more
3554 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
3555 Rather than timing out after 20 ms if an EC
3556 command is not properly ACKed, override the length
3557 of the timeout. We have interrupts disabled while
3558 waiting for the ACK, so if this is set too high
3559 interrupts *may* be lost!
3561 omap_mux= [OMAP] Override bootloader pin multiplexing.
3562 Format: <mux_mode0.mode_name=value>...
3563 For example, to override I2C bus2:
3564 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
3566 oops=panic Always panic on oopses. Default is to just kill the
3567 process, but there is a small probability of
3568 deadlocking the machine.
3569 This will also cause panics on machine check exceptions.
3570 Useful together with panic=30 to trigger a reboot.
3573 [KNL] Boolean flag to control whether the page allocator
3574 should randomize its free lists. The randomization may
3575 be automatically enabled if the kernel detects it is
3576 running on a platform with a direct-mapped memory-side
3577 cache, and this parameter can be used to
3578 override/disable that behavior. The state of the flag
3579 can be read from sysfs at:
3580 /sys/module/page_alloc/parameters/shuffle.
3582 page_owner= [KNL] Boot-time page_owner enabling option.
3583 Storage of the information about who allocated
3584 each page is disabled in default. With this switch,
3586 on: enable the feature
3588 page_poison= [KNL] Boot-time parameter changing the state of
3589 poisoning on the buddy allocator, available with
3590 CONFIG_PAGE_POISONING=y.
3591 off: turn off poisoning (default)
3592 on: turn on poisoning
3594 panic= [KNL] Kernel behaviour on panic: delay <timeout>
3595 timeout > 0: seconds before rebooting
3596 timeout = 0: wait forever
3597 timeout < 0: reboot immediately
3600 panic_print= Bitmask for printing system info when panic happens.
3601 User can chose combination of the following bits:
3602 bit 0: print all tasks info
3603 bit 1: print system memory info
3604 bit 2: print timer info
3605 bit 3: print locks info if CONFIG_LOCKDEP is on
3606 bit 4: print ftrace buffer
3607 bit 5: print all printk messages in buffer
3609 panic_on_taint= Bitmask for conditionally calling panic() in add_taint()
3610 Format: <hex>[,nousertaint]
3611 Hexadecimal bitmask representing the set of TAINT flags
3612 that will cause the kernel to panic when add_taint() is
3613 called with any of the flags in this set.
3614 The optional switch "nousertaint" can be utilized to
3615 prevent userspace forced crashes by writing to sysctl
3616 /proc/sys/kernel/tainted any flagset matching with the
3617 bitmask set on panic_on_taint.
3618 See Documentation/admin-guide/tainted-kernels.rst for
3619 extra details on the taint flags that users can pick
3620 to compose the bitmask to assign to panic_on_taint.
3622 panic_on_warn panic() instead of WARN(). Useful to cause kdump
3625 crash_kexec_post_notifiers
3626 Run kdump after running panic-notifiers and dumping
3627 kmsg. This only for the users who doubt kdump always
3628 succeeds in any situation.
3629 Note that this also increases risks of kdump failure,
3630 because some panic notifiers can make the crashed
3631 kernel more unstable.
3633 parkbd.port= [HW] Parallel port number the keyboard adapter is
3634 connected to, default is 0.
3636 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
3637 0 for XT, 1 for AT (default is AT).
3640 parport= [HW,PPT] Specify parallel ports. 0 disables.
3641 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
3642 Use 'auto' to force the driver to use any
3643 IRQ/DMA settings detected (the default is to
3644 ignore detected IRQ/DMA settings because of
3645 possible conflicts). You can specify the base
3646 address, IRQ, and DMA settings; IRQ and DMA
3647 should be numbers, or 'auto' (for using detected
3648 settings on that particular port), or 'nofifo'
3649 (to avoid using a FIFO even if it is detected).
3650 Parallel ports are assigned in the order they
3651 are specified on the command line, starting
3654 parport_init_mode= [HW,PPT]
3655 Configure VIA parallel port to operate in
3656 a specific mode. This is necessary on Pegasos
3657 computer where firmware has no options for setting
3658 up parallel port mode and sets it to spp.
3659 Currently this function knows 686a and 8231 chips.
3660 Format: [spp|ps2|epp|ecp|ecpepp]
3662 pata_legacy.all= [HW,LIBATA]
3664 Set to non-zero to probe primary and secondary ISA
3665 port ranges on PCI systems where no PCI PATA device
3666 has been found at either range. Disabled by default.
3668 pata_legacy.autospeed= [HW,LIBATA]
3670 Set to non-zero if a chip is present that snoops speed
3671 changes. Disabled by default.
3673 pata_legacy.ht6560a= [HW,LIBATA]
3675 Set to 1, 2, or 3 for HT 6560A on the primary channel,
3676 the secondary channel, or both channels respectively.
3677 Disabled by default.
3679 pata_legacy.ht6560b= [HW,LIBATA]
3681 Set to 1, 2, or 3 for HT 6560B on the primary channel,
3682 the secondary channel, or both channels respectively.
3683 Disabled by default.
3685 pata_legacy.iordy_mask= [HW,LIBATA]
3687 IORDY enable mask. Set individual bits to allow IORDY
3688 for the respective channel. Bit 0 is for the first
3689 legacy channel handled by this driver, bit 1 is for
3690 the second channel, and so on. The sequence will often
3691 correspond to the primary legacy channel, the secondary
3692 legacy channel, and so on, but the handling of a PCI
3693 bus and the use of other driver options may interfere
3694 with the sequence. By default IORDY is allowed across
3697 pata_legacy.opti82c46x= [HW,LIBATA]
3699 Set to 1, 2, or 3 for Opti 82c611A on the primary
3700 channel, the secondary channel, or both channels
3701 respectively. Disabled by default.
3703 pata_legacy.opti82c611a= [HW,LIBATA]
3705 Set to 1, 2, or 3 for Opti 82c465MV on the primary
3706 channel, the secondary channel, or both channels
3707 respectively. Disabled by default.
3709 pata_legacy.pio_mask= [HW,LIBATA]
3711 PIO mode mask for autospeed devices. Set individual
3712 bits to allow the use of the respective PIO modes.
3713 Bit 0 is for mode 0, bit 1 is for mode 1, and so on.
3714 All modes allowed by default.
3716 pata_legacy.probe_all= [HW,LIBATA]
3718 Set to non-zero to probe tertiary and further ISA
3719 port ranges on PCI systems. Disabled by default.
3721 pata_legacy.probe_mask= [HW,LIBATA]
3723 Probe mask for legacy ISA PATA ports. Depending on
3724 platform configuration and the use of other driver
3725 options up to 6 legacy ports are supported: 0x1f0,
3726 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing
3727 of individual ports can be disabled by setting the
3728 corresponding bits in the mask to 1. Bit 0 is for
3729 the first port in the list above (0x1f0), and so on.
3730 By default all supported ports are probed.
3732 pata_legacy.qdi= [HW,LIBATA]
3734 Set to non-zero to probe QDI controllers. By default
3735 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise.
3737 pata_legacy.winbond= [HW,LIBATA]
3739 Set to non-zero to probe Winbond controllers. Use
3740 the standard I/O port (0x130) if 1, otherwise the
3741 value given is the I/O port to use (typically 0x1b0).
3742 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE,
3745 pata_platform.pio_mask= [HW,LIBATA]
3747 Supported PIO mode mask. Set individual bits to allow
3748 the use of the respective PIO modes. Bit 0 is for
3749 mode 0, bit 1 is for mode 1, and so on. Mode 0 only
3753 Halt all CPUs after the first oops has been printed for
3754 the specified number of seconds. This is to be used if
3755 your oopses keep scrolling off the screen.
3760 See header of drivers/block/paride/pcd.c.
3761 See also Documentation/admin-guide/blockdev/paride.rst.
3763 pci=option[,option...] [PCI] various PCI subsystem options.
3765 Some options herein operate on a specific device
3766 or a set of devices (<pci_dev>). These are
3767 specified in one of the following formats:
3769 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
3770 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
3772 Note: the first format specifies a PCI
3773 bus/device/function address which may change
3774 if new hardware is inserted, if motherboard
3775 firmware changes, or due to changes caused
3776 by other kernel parameters. If the
3777 domain is left unspecified, it is
3778 taken to be zero. Optionally, a path
3779 to a device through multiple device/function
3780 addresses can be specified after the base
3781 address (this is more robust against
3782 renumbering issues). The second format
3783 selects devices using IDs from the
3784 configuration space which may match multiple
3785 devices in the system.
3787 earlydump dump PCI config space before the kernel
3789 off [X86] don't probe for the PCI bus
3790 bios [X86-32] force use of PCI BIOS, don't access
3791 the hardware directly. Use this if your machine
3792 has a non-standard PCI host bridge.
3793 nobios [X86-32] disallow use of PCI BIOS, only direct
3794 hardware access methods are allowed. Use this
3795 if you experience crashes upon bootup and you
3796 suspect they are caused by the BIOS.
3797 conf1 [X86] Force use of PCI Configuration Access
3798 Mechanism 1 (config address in IO port 0xCF8,
3799 data in IO port 0xCFC, both 32-bit).
3800 conf2 [X86] Force use of PCI Configuration Access
3801 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
3802 the function, IO port 0xCFA, also 8-bit, sets
3803 bus number. The config space is then accessed
3804 through ports 0xC000-0xCFFF).
3805 See http://wiki.osdev.org/PCI for more info
3806 on the configuration access mechanisms.
3807 noaer [PCIE] If the PCIEAER kernel config parameter is
3808 enabled, this kernel boot option can be used to
3809 disable the use of PCIE advanced error reporting.
3810 nodomains [PCI] Disable support for multiple PCI
3811 root domains (aka PCI segments, in ACPI-speak).
3812 nommconf [X86] Disable use of MMCONFIG for PCI
3814 check_enable_amd_mmconf [X86] check for and enable
3815 properly configured MMIO access to PCI
3816 config space on AMD family 10h CPU
3817 nomsi [MSI] If the PCI_MSI kernel config parameter is
3818 enabled, this kernel boot option can be used to
3819 disable the use of MSI interrupts system-wide.
3820 noioapicquirk [APIC] Disable all boot interrupt quirks.
3821 Safety option to keep boot IRQs enabled. This
3822 should never be necessary.
3823 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
3824 primary IO-APIC for bridges that cannot disable
3825 boot IRQs. This fixes a source of spurious IRQs
3826 when the system masks IRQs.
3827 noioapicreroute [APIC] Disable workaround that uses the
3828 boot IRQ equivalent of an IRQ that connects to
3829 a chipset where boot IRQs cannot be disabled.
3830 The opposite of ioapicreroute.
3831 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
3832 routing table. These calls are known to be buggy
3833 on several machines and they hang the machine
3834 when used, but on other computers it's the only
3835 way to get the interrupt routing table. Try
3836 this option if the kernel is unable to allocate
3837 IRQs or discover secondary PCI buses on your
3839 rom [X86] Assign address space to expansion ROMs.
3840 Use with caution as certain devices share
3841 address decoders between ROMs and other
3843 norom [X86] Do not assign address space to
3844 expansion ROMs that do not already have
3845 BIOS assigned address ranges.
3846 nobar [X86] Do not assign address space to the
3847 BARs that weren't assigned by the BIOS.
3848 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
3849 assigned automatically to PCI devices. You can
3850 make the kernel exclude IRQs of your ISA cards
3852 pirqaddr=0xAAAAA [X86] Specify the physical address
3853 of the PIRQ table (normally generated
3854 by the BIOS) if it is outside the
3855 F0000h-100000h range.
3856 lastbus=N [X86] Scan all buses thru bus #N. Can be
3857 useful if the kernel is unable to find your
3858 secondary buses and you want to tell it
3859 explicitly which ones they are.
3860 assign-busses [X86] Always assign all PCI bus
3861 numbers ourselves, overriding
3862 whatever the firmware may have done.
3863 usepirqmask [X86] Honor the possible IRQ mask stored
3864 in the BIOS $PIR table. This is needed on
3865 some systems with broken BIOSes, notably
3866 some HP Pavilion N5400 and Omnibook XE3
3867 notebooks. This will have no effect if ACPI
3868 IRQ routing is enabled.
3869 noacpi [X86] Do not use ACPI for IRQ routing
3870 or for PCI scanning.
3871 use_crs [X86] Use PCI host bridge window information
3872 from ACPI. On BIOSes from 2008 or later, this
3873 is enabled by default. If you need to use this,
3874 please report a bug.
3875 nocrs [X86] Ignore PCI host bridge windows from ACPI.
3876 If you need to use this, please report a bug.
3877 routeirq Do IRQ routing for all PCI devices.
3878 This is normally done in pci_enable_device(),
3879 so this option is a temporary workaround
3880 for broken drivers that don't call it.
3881 skip_isa_align [X86] do not align io start addr, so can
3882 handle more pci cards
3883 noearly [X86] Don't do any early type 1 scanning.
3884 This might help on some broken boards which
3885 machine check when some devices' config space
3886 is read. But various workarounds are disabled
3887 and some IOMMU drivers will not work.
3888 bfsort Sort PCI devices into breadth-first order.
3889 This sorting is done to get a device
3890 order compatible with older (<= 2.4) kernels.
3891 nobfsort Don't sort PCI devices into breadth-first order.
3892 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
3893 tuning and use the BIOS-configured MPS defaults.
3894 pcie_bus_safe Set every device's MPS to the largest value
3895 supported by all devices below the root complex.
3896 pcie_bus_perf Set device MPS to the largest allowable MPS
3897 based on its parent bus. Also set MRRS (Max
3898 Read Request Size) to the largest supported
3899 value (no larger than the MPS that the device
3900 or bus can support) for best performance.
3901 pcie_bus_peer2peer Set every device's MPS to 128B, which
3902 every device is guaranteed to support. This
3903 configuration allows peer-to-peer DMA between
3904 any pair of devices, possibly at the cost of
3905 reduced performance. This also guarantees
3906 that hot-added devices will work.
3907 cbiosize=nn[KMG] The fixed amount of bus space which is
3908 reserved for the CardBus bridge's IO window.
3909 The default value is 256 bytes.
3910 cbmemsize=nn[KMG] The fixed amount of bus space which is
3911 reserved for the CardBus bridge's memory
3912 window. The default value is 64 megabytes.
3915 [<order of align>@]<pci_dev>[; ...]
3916 Specifies alignment and device to reassign
3917 aligned memory resources. How to
3918 specify the device is described above.
3919 If <order of align> is not specified,
3920 PAGE_SIZE is used as alignment.
3921 A PCI-PCI bridge can be specified if resource
3922 windows need to be expanded.
3923 To specify the alignment for several
3924 instances of a device, the PCI vendor,
3925 device, subvendor, and subdevice may be
3926 specified, e.g., 12@pci:8086:9c22:103c:198f
3927 for 4096-byte alignment.
3928 ecrc= Enable/disable PCIe ECRC (transaction layer
3929 end-to-end CRC checking).
3930 bios: Use BIOS/firmware settings. This is the
3934 hpiosize=nn[KMG] The fixed amount of bus space which is
3935 reserved for hotplug bridge's IO window.
3936 Default size is 256 bytes.
3937 hpmmiosize=nn[KMG] The fixed amount of bus space which is
3938 reserved for hotplug bridge's MMIO window.
3939 Default size is 2 megabytes.
3940 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
3941 reserved for hotplug bridge's MMIO_PREF window.
3942 Default size is 2 megabytes.
3943 hpmemsize=nn[KMG] The fixed amount of bus space which is
3944 reserved for hotplug bridge's MMIO and
3946 Default size is 2 megabytes.
3947 hpbussize=nn The minimum amount of additional bus numbers
3948 reserved for buses below a hotplug bridge.
3950 realloc= Enable/disable reallocating PCI bridge resources
3951 if allocations done by BIOS are too small to
3952 accommodate resources required by all child
3954 off: Turn realloc off
3956 realloc same as realloc=on
3957 noari do not use PCIe ARI.
3958 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
3959 do not use PCIe ATS (and IOMMU device IOTLB).
3960 pcie_scan_all Scan all possible PCIe devices. Otherwise we
3961 only look for one device below a PCIe downstream
3963 big_root_window Try to add a big 64bit memory window to the PCIe
3964 root complex on AMD CPUs. Some GFX hardware
3965 can resize a BAR to allow access to all VRAM.
3966 Adding the window is slightly risky (it may
3967 conflict with unreported devices), so this
3969 disable_acs_redir=<pci_dev>[; ...]
3970 Specify one or more PCI devices (in the format
3971 specified above) separated by semicolons.
3972 Each device specified will have the PCI ACS
3973 redirect capabilities forced off which will
3974 allow P2P traffic between devices through
3975 bridges without forcing it upstream. Note:
3976 this removes isolation between devices and
3977 may put more devices in an IOMMU group.
3978 force_floating [S390] Force usage of floating interrupts.
3979 nomio [S390] Do not use MIO instructions.
3980 norid [S390] ignore the RID field and force use of
3981 one PCI domain per PCI function
3983 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
3986 force Enable ASPM even on devices that claim not to support it.
3987 WARNING: Forcing ASPM on may cause system lockups.
3989 pcie_ports= [PCIE] PCIe port services handling:
3990 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
3991 even if the platform doesn't give the OS permission to
3992 use them. This may cause conflicts if the platform
3993 also tries to use these services.
3994 dpc-native Use native PCIe service for DPC only. May
3995 cause conflicts if firmware uses AER or DPC.
3996 compat Disable native PCIe services (PME, AER, DPC, PCIe
3999 pcie_port_pm= [PCIE] PCIe port power management handling:
4000 off Disable power management of all PCIe ports
4001 force Forcibly enable power management of all PCIe ports
4003 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
4004 nomsi Do not use MSI for native PCIe PME signaling (this makes
4005 all PCIe root ports use INTx for all services).
4007 pcmv= [HW,PCMCIA] BadgePAD 4
4011 Keep all power-domains already enabled by bootloader on,
4012 even if no driver has claimed them. This is useful
4013 for debug and development, but should not be
4014 needed on a platform with proper driver support.
4017 See Documentation/admin-guide/blockdev/paride.rst.
4019 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
4022 See arch/parisc/kernel/pdc_chassis.c
4024 percpu_alloc= Select which percpu first chunk allocator to use.
4025 Currently supported values are "embed" and "page".
4026 Archs may support subset or none of the selections.
4027 See comments in mm/percpu.c for details on each
4028 allocator. This parameter is primarily for debugging
4029 and performance comparison.
4032 See Documentation/admin-guide/blockdev/paride.rst.
4035 See Documentation/admin-guide/blockdev/paride.rst.
4037 pirq= [SMP,APIC] Manual mp-table setup
4038 See Documentation/x86/i386/IO-APIC.rst.
4040 plip= [PPT,NET] Parallel port network link
4041 Format: { parport<nr> | timid | 0 }
4042 See also Documentation/admin-guide/parport.rst.
4044 pmtmr= [X86] Manual setup of pmtmr I/O Port.
4045 Override pmtimer IOPort with a hex value.
4048 pm_debug_messages [SUSPEND,KNL]
4049 Enable suspend/resume debug messages during boot up.
4052 Enable PNP debug messages (depends on the
4053 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
4054 via /sys/module/pnp/parameters/debug. We always show
4055 current resource usage; turning this on also shows
4056 possible settings and some assignment information.
4062 { on | off | curr | res | no-curr | no-res }
4065 [ISAPNP] Exclude IRQs for the autoconfiguration
4068 [ISAPNP] Exclude DMAs for the autoconfiguration
4070 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
4071 Ranges are in pairs (I/O port base and size).
4074 [ISAPNP] Exclude memory regions for the
4076 Ranges are in pairs (memory base and size).
4078 ports= [IP_VS_FTP] IPVS ftp helper module
4080 Up to 8 (IP_VS_APP_MAX_PORTS) ports
4082 Format: <port>,<port>....
4084 powersave=off [PPC] This option disables power saving features.
4085 It specifically disables cpuidle and sets the
4086 platform machine description specific power_save
4087 function to NULL. On Idle the CPU just reduces
4090 ppc_strict_facility_enable
4091 [PPC] This option catches any kernel floating point,
4092 Altivec, VSX and SPE outside of regions specifically
4093 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
4094 There is some performance impact when enabling this.
4098 Disable Hardware Transactional Memory
4101 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC
4102 none - Limited to cond_resched() calls
4103 voluntary - Limited to cond_resched() and might_sleep() calls
4104 full - Any section that isn't explicitly preempt disabled
4105 can be preempted anytime.
4107 print-fatal-signals=
4108 [KNL] debug: print fatal signals
4110 If enabled, warn about various signal handling
4111 related application anomalies: too many signals,
4112 too many POSIX.1 timers, fatal signals causing a
4115 If you hit the warning due to signal overflow,
4116 you might want to try "ulimit -i unlimited".
4120 printk.always_kmsg_dump=
4121 Trigger kmsg_dump for cases other than kernel oops or
4123 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4126 printk.devkmsg={on,off,ratelimit}
4127 Control writing to /dev/kmsg.
4128 on - unlimited logging to /dev/kmsg from userspace
4129 off - logging to /dev/kmsg disabled
4130 ratelimit - ratelimit the logging
4133 printk.time= Show timing data prefixed to each printk message line
4134 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4136 processor.max_cstate= [HW,ACPI]
4137 Limit processor to maximum C-state
4138 max_cstate=9 overrides any DMI blacklist limit.
4140 processor.nocst [HW,ACPI]
4141 Ignore the _CST method to determine C-states,
4142 instead using the legacy FADT method
4144 profile= [KNL] Enable kernel profiling via /proc/profile
4145 Format: [<profiletype>,]<number>
4146 Param: <profiletype>: "schedule", "sleep", or "kvm"
4147 [defaults to kernel profiling]
4148 Param: "schedule" - profile schedule points.
4149 Param: "sleep" - profile D-state sleeping (millisecs).
4150 Requires CONFIG_SCHEDSTATS
4151 Param: "kvm" - profile VM exits.
4152 Param: <number> - step/bucket size as a power of 2 for
4153 statistical time based profiling.
4155 prompt_ramdisk= [RAM] [Deprecated]
4157 prot_virt= [S390] enable hosting protected virtual machines
4158 isolated from the hypervisor (if hardware supports
4162 psi= [KNL] Enable or disable pressure stall information
4166 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
4167 probe for; one of (bare|imps|exps|lifebook|any).
4168 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
4170 psmouse.resetafter= [HW,MOUSE]
4171 Try to reset the device after so many bad packets
4174 [HW,MOUSE] Set desired mouse resolution, in dpi.
4175 psmouse.smartscroll=
4176 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
4177 0 = disabled, 1 = enabled (default).
4179 pstore.backend= Specify the name of the pstore backend to use
4182 See Documentation/admin-guide/blockdev/paride.rst.
4184 pti= [X86-64] Control Page Table Isolation of user and
4185 kernel address spaces. Disabling this feature
4186 removes hardening, but improves performance of
4187 system calls and interrupts.
4189 on - unconditionally enable
4190 off - unconditionally disable
4191 auto - kernel detects whether your CPU model is
4192 vulnerable to issues that PTI mitigates
4194 Not specifying this option is equivalent to pti=auto.
4197 Equivalent to pti=off
4200 [KNL] Number of legacy pty's. Overwrites compiled-in
4203 quiet [KNL] Disable most log messages
4208 See Documentation/admin-guide/md.rst.
4210 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
4211 See Documentation/admin-guide/blockdev/ramdisk.rst.
4213 ramdisk_start= [RAM] RAM disk image start address
4215 random.trust_cpu={on,off}
4216 [KNL] Enable or disable trusting the use of the
4217 CPU's random number generator (if available) to
4218 fully seed the kernel's CRNG. Default is controlled
4219 by CONFIG_RANDOM_TRUST_CPU.
4221 randomize_kstack_offset=
4222 [KNL] Enable or disable kernel stack offset
4223 randomization, which provides roughly 5 bits of
4224 entropy, frustrating memory corruption attacks
4225 that depend on stack address determinism or
4226 cross-syscall address exposures. This is only
4227 available on architectures that have defined
4228 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET.
4229 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4230 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT.
4232 ras=option[,option,...] [KNL] RAS-specific options
4235 Disable the Correctable Errors Collector,
4236 see CONFIG_RAS_CEC help text.
4239 The argument is a cpu list, as described above.
4241 In kernels built with CONFIG_RCU_NOCB_CPU=y, set
4242 the specified list of CPUs to be no-callback CPUs.
4243 Invocation of these CPUs' RCU callbacks will be
4244 offloaded to "rcuox/N" kthreads created for that
4245 purpose, where "x" is "p" for RCU-preempt, and
4246 "s" for RCU-sched, and "N" is the CPU number.
4247 This reduces OS jitter on the offloaded CPUs,
4248 which can be useful for HPC and real-time
4249 workloads. It can also improve energy efficiency
4250 for asymmetric multiprocessors.
4253 Rather than requiring that offloaded CPUs
4254 (specified by rcu_nocbs= above) explicitly
4255 awaken the corresponding "rcuoN" kthreads,
4256 make these kthreads poll for callbacks.
4257 This improves the real-time response for the
4258 offloaded CPUs by relieving them of the need to
4259 wake up the corresponding kthread, but degrades
4260 energy efficiency by requiring that the kthreads
4261 periodically wake up to do the polling.
4263 rcutree.blimit= [KNL]
4264 Set maximum number of finished RCU callbacks to
4265 process in one batch.
4267 rcutree.dump_tree= [KNL]
4268 Dump the structure of the rcu_node combining tree
4269 out at early boot. This is used for diagnostic
4270 purposes, to verify correct tree setup.
4272 rcutree.gp_cleanup_delay= [KNL]
4273 Set the number of jiffies to delay each step of
4274 RCU grace-period cleanup.
4276 rcutree.gp_init_delay= [KNL]
4277 Set the number of jiffies to delay each step of
4278 RCU grace-period initialization.
4280 rcutree.gp_preinit_delay= [KNL]
4281 Set the number of jiffies to delay each step of
4282 RCU grace-period pre-initialization, that is,
4283 the propagation of recent CPU-hotplug changes up
4284 the rcu_node combining tree.
4286 rcutree.use_softirq= [KNL]
4287 If set to zero, move all RCU_SOFTIRQ processing to
4288 per-CPU rcuc kthreads. Defaults to a non-zero
4289 value, meaning that RCU_SOFTIRQ is used by default.
4290 Specify rcutree.use_softirq=0 to use rcuc kthreads.
4292 But note that CONFIG_PREEMPT_RT=y kernels disable
4293 this kernel boot parameter, forcibly setting it
4296 rcutree.rcu_fanout_exact= [KNL]
4297 Disable autobalancing of the rcu_node combining
4298 tree. This is used by rcutorture, and might
4299 possibly be useful for architectures having high
4300 cache-to-cache transfer latencies.
4302 rcutree.rcu_fanout_leaf= [KNL]
4303 Change the number of CPUs assigned to each
4304 leaf rcu_node structure. Useful for very
4305 large systems, which will choose the value 64,
4306 and for NUMA systems with large remote-access
4307 latencies, which will choose a value aligned
4308 with the appropriate hardware boundaries.
4310 rcutree.rcu_min_cached_objs= [KNL]
4311 Minimum number of objects which are cached and
4312 maintained per one CPU. Object size is equal
4313 to PAGE_SIZE. The cache allows to reduce the
4314 pressure to page allocator, also it makes the
4315 whole algorithm to behave better in low memory
4318 rcutree.jiffies_till_first_fqs= [KNL]
4319 Set delay from grace-period initialization to
4320 first attempt to force quiescent states.
4321 Units are jiffies, minimum value is zero,
4322 and maximum value is HZ.
4324 rcutree.jiffies_till_next_fqs= [KNL]
4325 Set delay between subsequent attempts to force
4326 quiescent states. Units are jiffies, minimum
4327 value is one, and maximum value is HZ.
4329 rcutree.jiffies_till_sched_qs= [KNL]
4330 Set required age in jiffies for a
4331 given grace period before RCU starts
4332 soliciting quiescent-state help from
4333 rcu_note_context_switch() and cond_resched().
4334 If not specified, the kernel will calculate
4335 a value based on the most recent settings
4336 of rcutree.jiffies_till_first_fqs
4337 and rcutree.jiffies_till_next_fqs.
4338 This calculated value may be viewed in
4339 rcutree.jiffies_to_sched_qs. Any attempt to set
4340 rcutree.jiffies_to_sched_qs will be cheerfully
4343 rcutree.kthread_prio= [KNL,BOOT]
4344 Set the SCHED_FIFO priority of the RCU per-CPU
4345 kthreads (rcuc/N). This value is also used for
4346 the priority of the RCU boost threads (rcub/N)
4347 and for the RCU grace-period kthreads (rcu_bh,
4348 rcu_preempt, and rcu_sched). If RCU_BOOST is
4349 set, valid values are 1-99 and the default is 1
4350 (the least-favored priority). Otherwise, when
4351 RCU_BOOST is not set, valid values are 0-99 and
4352 the default is zero (non-realtime operation).
4354 rcutree.rcu_nocb_gp_stride= [KNL]
4355 Set the number of NOCB callback kthreads in
4356 each group, which defaults to the square root
4357 of the number of CPUs. Larger numbers reduce
4358 the wakeup overhead on the global grace-period
4359 kthread, but increases that same overhead on
4360 each group's NOCB grace-period kthread.
4362 rcutree.qhimark= [KNL]
4363 Set threshold of queued RCU callbacks beyond which
4364 batch limiting is disabled.
4366 rcutree.qlowmark= [KNL]
4367 Set threshold of queued RCU callbacks below which
4368 batch limiting is re-enabled.
4370 rcutree.qovld= [KNL]
4371 Set threshold of queued RCU callbacks beyond which
4372 RCU's force-quiescent-state scan will aggressively
4373 enlist help from cond_resched() and sched IPIs to
4374 help CPUs more quickly reach quiescent states.
4375 Set to less than zero to make this be set based
4376 on rcutree.qhimark at boot time and to zero to
4377 disable more aggressive help enlistment.
4379 rcutree.rcu_idle_gp_delay= [KNL]
4380 Set wakeup interval for idle CPUs that have
4381 RCU callbacks (RCU_FAST_NO_HZ=y).
4383 rcutree.rcu_kick_kthreads= [KNL]
4384 Cause the grace-period kthread to get an extra
4385 wake_up() if it sleeps three times longer than
4386 it should at force-quiescent-state time.
4387 This wake_up() will be accompanied by a
4388 WARN_ONCE() splat and an ftrace_dump().
4390 rcutree.rcu_unlock_delay= [KNL]
4391 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels,
4392 this specifies an rcu_read_unlock()-time delay
4393 in microseconds. This defaults to zero.
4394 Larger delays increase the probability of
4395 catching RCU pointer leaks, that is, buggy use
4396 of RCU-protected pointers after the relevant
4397 rcu_read_unlock() has completed.
4399 rcutree.sysrq_rcu= [KNL]
4400 Commandeer a sysrq key to dump out Tree RCU's
4401 rcu_node tree with an eye towards determining
4402 why a new grace period has not yet started.
4404 rcuscale.gp_async= [KNL]
4405 Measure performance of asynchronous
4406 grace-period primitives such as call_rcu().
4408 rcuscale.gp_async_max= [KNL]
4409 Specify the maximum number of outstanding
4410 callbacks per writer thread. When a writer
4411 thread exceeds this limit, it invokes the
4412 corresponding flavor of rcu_barrier() to allow
4413 previously posted callbacks to drain.
4415 rcuscale.gp_exp= [KNL]
4416 Measure performance of expedited synchronous
4417 grace-period primitives.
4419 rcuscale.holdoff= [KNL]
4420 Set test-start holdoff period. The purpose of
4421 this parameter is to delay the start of the
4422 test until boot completes in order to avoid
4425 rcuscale.kfree_rcu_test= [KNL]
4426 Set to measure performance of kfree_rcu() flooding.
4428 rcuscale.kfree_rcu_test_double= [KNL]
4429 Test the double-argument variant of kfree_rcu().
4430 If this parameter has the same value as
4431 rcuscale.kfree_rcu_test_single, both the single-
4432 and double-argument variants are tested.
4434 rcuscale.kfree_rcu_test_single= [KNL]
4435 Test the single-argument variant of kfree_rcu().
4436 If this parameter has the same value as
4437 rcuscale.kfree_rcu_test_double, both the single-
4438 and double-argument variants are tested.
4440 rcuscale.kfree_nthreads= [KNL]
4441 The number of threads running loops of kfree_rcu().
4443 rcuscale.kfree_alloc_num= [KNL]
4444 Number of allocations and frees done in an iteration.
4446 rcuscale.kfree_loops= [KNL]
4447 Number of loops doing rcuscale.kfree_alloc_num number
4448 of allocations and frees.
4450 rcuscale.nreaders= [KNL]
4451 Set number of RCU readers. The value -1 selects
4452 N, where N is the number of CPUs. A value
4453 "n" less than -1 selects N-n+1, where N is again
4454 the number of CPUs. For example, -2 selects N
4455 (the number of CPUs), -3 selects N+1, and so on.
4456 A value of "n" less than or equal to -N selects
4459 rcuscale.nwriters= [KNL]
4460 Set number of RCU writers. The values operate
4461 the same as for rcuscale.nreaders.
4462 N, where N is the number of CPUs
4464 rcuscale.perf_type= [KNL]
4465 Specify the RCU implementation to test.
4467 rcuscale.shutdown= [KNL]
4468 Shut the system down after performance tests
4469 complete. This is useful for hands-off automated
4472 rcuscale.verbose= [KNL]
4473 Enable additional printk() statements.
4475 rcuscale.writer_holdoff= [KNL]
4476 Write-side holdoff between grace periods,
4477 in microseconds. The default of zero says
4480 rcutorture.fqs_duration= [KNL]
4481 Set duration of force_quiescent_state bursts
4484 rcutorture.fqs_holdoff= [KNL]
4485 Set holdoff time within force_quiescent_state bursts
4488 rcutorture.fqs_stutter= [KNL]
4489 Set wait time between force_quiescent_state bursts
4492 rcutorture.fwd_progress= [KNL]
4493 Enable RCU grace-period forward-progress testing
4494 for the types of RCU supporting this notion.
4496 rcutorture.fwd_progress_div= [KNL]
4497 Specify the fraction of a CPU-stall-warning
4498 period to do tight-loop forward-progress testing.
4500 rcutorture.fwd_progress_holdoff= [KNL]
4501 Number of seconds to wait between successive
4502 forward-progress tests.
4504 rcutorture.fwd_progress_need_resched= [KNL]
4505 Enclose cond_resched() calls within checks for
4506 need_resched() during tight-loop forward-progress
4509 rcutorture.gp_cond= [KNL]
4510 Use conditional/asynchronous update-side
4511 primitives, if available.
4513 rcutorture.gp_exp= [KNL]
4514 Use expedited update-side primitives, if available.
4516 rcutorture.gp_normal= [KNL]
4517 Use normal (non-expedited) asynchronous
4518 update-side primitives, if available.
4520 rcutorture.gp_sync= [KNL]
4521 Use normal (non-expedited) synchronous
4522 update-side primitives, if available. If all
4523 of rcutorture.gp_cond=, rcutorture.gp_exp=,
4524 rcutorture.gp_normal=, and rcutorture.gp_sync=
4525 are zero, rcutorture acts as if is interpreted
4526 they are all non-zero.
4528 rcutorture.irqreader= [KNL]
4529 Run RCU readers from irq handlers, or, more
4530 accurately, from a timer handler. Not all RCU
4531 flavors take kindly to this sort of thing.
4533 rcutorture.leakpointer= [KNL]
4534 Leak an RCU-protected pointer out of the reader.
4535 This can of course result in splats, and is
4536 intended to test the ability of things like
4537 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect
4540 rcutorture.n_barrier_cbs= [KNL]
4541 Set callbacks/threads for rcu_barrier() testing.
4543 rcutorture.nfakewriters= [KNL]
4544 Set number of concurrent RCU writers. These just
4545 stress RCU, they don't participate in the actual
4546 test, hence the "fake".
4548 rcutorture.nocbs_nthreads= [KNL]
4549 Set number of RCU callback-offload togglers.
4550 Zero (the default) disables toggling.
4552 rcutorture.nocbs_toggle= [KNL]
4553 Set the delay in milliseconds between successive
4554 callback-offload toggling attempts.
4556 rcutorture.nreaders= [KNL]
4557 Set number of RCU readers. The value -1 selects
4558 N-1, where N is the number of CPUs. A value
4559 "n" less than -1 selects N-n-2, where N is again
4560 the number of CPUs. For example, -2 selects N
4561 (the number of CPUs), -3 selects N+1, and so on.
4563 rcutorture.object_debug= [KNL]
4564 Enable debug-object double-call_rcu() testing.
4566 rcutorture.onoff_holdoff= [KNL]
4567 Set time (s) after boot for CPU-hotplug testing.
4569 rcutorture.onoff_interval= [KNL]
4570 Set time (jiffies) between CPU-hotplug operations,
4571 or zero to disable CPU-hotplug testing.
4573 rcutorture.read_exit= [KNL]
4574 Set the number of read-then-exit kthreads used
4575 to test the interaction of RCU updaters and
4576 task-exit processing.
4578 rcutorture.read_exit_burst= [KNL]
4579 The number of times in a given read-then-exit
4580 episode that a set of read-then-exit kthreads
4583 rcutorture.read_exit_delay= [KNL]
4584 The delay, in seconds, between successive
4585 read-then-exit testing episodes.
4587 rcutorture.shuffle_interval= [KNL]
4588 Set task-shuffle interval (s). Shuffling tasks
4589 allows some CPUs to go into dyntick-idle mode
4590 during the rcutorture test.
4592 rcutorture.shutdown_secs= [KNL]
4593 Set time (s) after boot system shutdown. This
4594 is useful for hands-off automated testing.
4596 rcutorture.stall_cpu= [KNL]
4597 Duration of CPU stall (s) to test RCU CPU stall
4598 warnings, zero to disable.
4600 rcutorture.stall_cpu_block= [KNL]
4601 Sleep while stalling if set. This will result
4602 in warnings from preemptible RCU in addition
4603 to any other stall-related activity.
4605 rcutorture.stall_cpu_holdoff= [KNL]
4606 Time to wait (s) after boot before inducing stall.
4608 rcutorture.stall_cpu_irqsoff= [KNL]
4609 Disable interrupts while stalling if set.
4611 rcutorture.stall_gp_kthread= [KNL]
4612 Duration (s) of forced sleep within RCU
4613 grace-period kthread to test RCU CPU stall
4614 warnings, zero to disable. If both stall_cpu
4615 and stall_gp_kthread are specified, the
4616 kthread is starved first, then the CPU.
4618 rcutorture.stat_interval= [KNL]
4619 Time (s) between statistics printk()s.
4621 rcutorture.stutter= [KNL]
4622 Time (s) to stutter testing, for example, specifying
4623 five seconds causes the test to run for five seconds,
4624 wait for five seconds, and so on. This tests RCU's
4625 ability to transition abruptly to and from idle.
4627 rcutorture.test_boost= [KNL]
4628 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
4629 "Maybe" means test if the RCU implementation
4630 under test support RCU priority boosting.
4632 rcutorture.test_boost_duration= [KNL]
4633 Duration (s) of each individual boost test.
4635 rcutorture.test_boost_interval= [KNL]
4636 Interval (s) between each boost test.
4638 rcutorture.test_no_idle_hz= [KNL]
4639 Test RCU's dyntick-idle handling. See also the
4640 rcutorture.shuffle_interval parameter.
4642 rcutorture.torture_type= [KNL]
4643 Specify the RCU implementation to test.
4645 rcutorture.verbose= [KNL]
4646 Enable additional printk() statements.
4648 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
4649 Dump ftrace buffer after reporting RCU CPU
4652 rcupdate.rcu_cpu_stall_suppress= [KNL]
4653 Suppress RCU CPU stall warning messages.
4655 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL]
4656 Suppress RCU CPU stall warning messages and
4657 rcutorture writer stall warnings that occur
4658 during early boot, that is, during the time
4659 before the init task is spawned.
4661 rcupdate.rcu_cpu_stall_timeout= [KNL]
4662 Set timeout for RCU CPU stall warning messages.
4664 rcupdate.rcu_expedited= [KNL]
4665 Use expedited grace-period primitives, for
4666 example, synchronize_rcu_expedited() instead
4667 of synchronize_rcu(). This reduces latency,
4668 but can increase CPU utilization, degrade
4669 real-time latency, and degrade energy efficiency.
4670 No effect on CONFIG_TINY_RCU kernels.
4672 rcupdate.rcu_normal= [KNL]
4673 Use only normal grace-period primitives,
4674 for example, synchronize_rcu() instead of
4675 synchronize_rcu_expedited(). This improves
4676 real-time latency, CPU utilization, and
4677 energy efficiency, but can expose users to
4678 increased grace-period latency. This parameter
4679 overrides rcupdate.rcu_expedited. No effect on
4680 CONFIG_TINY_RCU kernels.
4682 rcupdate.rcu_normal_after_boot= [KNL]
4683 Once boot has completed (that is, after
4684 rcu_end_inkernel_boot() has been invoked), use
4685 only normal grace-period primitives. No effect
4686 on CONFIG_TINY_RCU kernels.
4688 But note that CONFIG_PREEMPT_RT=y kernels enables
4689 this kernel boot parameter, forcibly setting
4690 it to the value one, that is, converting any
4691 post-boot attempt at an expedited RCU grace
4692 period to instead use normal non-expedited
4693 grace-period processing.
4695 rcupdate.rcu_task_ipi_delay= [KNL]
4696 Set time in jiffies during which RCU tasks will
4697 avoid sending IPIs, starting with the beginning
4698 of a given grace period. Setting a large
4699 number avoids disturbing real-time workloads,
4700 but lengthens grace periods.
4702 rcupdate.rcu_task_stall_timeout= [KNL]
4703 Set timeout in jiffies for RCU task stall warning
4704 messages. Disable with a value less than or equal
4707 rcupdate.rcu_self_test= [KNL]
4708 Run the RCU early boot self tests
4712 Run specified binary instead of /init from the ramdisk,
4713 used for early userspace startup. See initrd.
4716 force - Override the decision by the kernel to hide the
4717 advertisement of RDRAND support (this affects
4718 certain AMD processors because of buggy BIOS
4719 support, specifically around the suspend/resume
4723 Turn on/off individual RDT features. List is:
4724 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
4726 E.g. to turn on cmt and turn off mba use:
4730 Format (x86 or x86_64):
4731 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] \
4733 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
4735 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
4736 (prefix with 'panic_' to set mode for panic
4738 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
4739 reboot_force is either force or not specified,
4740 reboot_cpu is s[mp]#### with #### being the processor
4741 to be used for rebooting.
4743 refscale.holdoff= [KNL]
4744 Set test-start holdoff period. The purpose of
4745 this parameter is to delay the start of the
4746 test until boot completes in order to avoid
4749 refscale.loops= [KNL]
4750 Set the number of loops over the synchronization
4751 primitive under test. Increasing this number
4752 reduces noise due to loop start/end overhead,
4753 but the default has already reduced the per-pass
4754 noise to a handful of picoseconds on ca. 2020
4757 refscale.nreaders= [KNL]
4758 Set number of readers. The default value of -1
4759 selects N, where N is roughly 75% of the number
4760 of CPUs. A value of zero is an interesting choice.
4762 refscale.nruns= [KNL]
4763 Set number of runs, each of which is dumped onto
4766 refscale.readdelay= [KNL]
4767 Set the read-side critical-section duration,
4768 measured in microseconds.
4770 refscale.scale_type= [KNL]
4771 Specify the read-protection implementation to test.
4773 refscale.shutdown= [KNL]
4774 Shut down the system at the end of the performance
4775 test. This defaults to 1 (shut it down) when
4776 refscale is built into the kernel and to 0 (leave
4777 it running) when refscale is built as a module.
4779 refscale.verbose= [KNL]
4780 Enable additional printk() statements.
4782 refscale.verbose_batched= [KNL]
4783 Batch the additional printk() statements. If zero
4784 (the default) or negative, print everything. Otherwise,
4785 print every Nth verbose statement, where N is the value
4789 [KNL, SMP] Set scheduler's default relax_domain_level.
4790 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
4792 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
4793 Format: <base1>,<size1>[,<base2>,<size2>,...]
4794 Reserve I/O ports or memory so the kernel won't use
4795 them. If <base> is less than 0x10000, the region
4796 is assumed to be I/O ports; otherwise it is memory.
4798 reservetop= [X86-32]
4800 Reserves a hole at the top of the kernel virtual
4803 reset_devices [KNL] Force drivers to reset the underlying device
4804 during initialization.
4807 Specify the partition device for software suspend
4809 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
4811 resume_offset= [SWSUSP]
4812 Specify the offset from the beginning of the partition
4813 given by "resume=" at which the swap header is located,
4814 in <PAGE_SIZE> units (needed only for swap files).
4815 See Documentation/power/swsusp-and-swap-files.rst
4817 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
4818 read the resume files
4820 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
4821 Useful for devices that are detected asynchronously
4822 (e.g. USB and MMC devices).
4824 hibernate= [HIBERNATION]
4825 noresume Don't check if there's a hibernation image
4826 present during boot.
4827 nocompress Don't compress/decompress hibernation images.
4828 no Disable hibernation and resume.
4829 protect_image Turn on image protection during restoration
4830 (that will set all pages holding image data
4831 during restoration read-only).
4833 retain_initrd [RAM] Keep initrd memory after extraction
4835 rfkill.default_state=
4836 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
4837 etc. communication is blocked by default.
4840 rfkill.master_switch_mode=
4841 0 The "airplane mode" button does nothing.
4842 1 The "airplane mode" button toggles between everything
4843 blocked and the previous configuration.
4844 2 The "airplane mode" button toggles between everything
4845 blocked and everything unblocked.
4847 rhash_entries= [KNL,NET]
4848 Set number of hash buckets for route cache
4851 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
4854 ro [KNL] Mount root device read-only on boot
4857 on Mark read-only kernel memory as read-only (default).
4858 off Leave read-only kernel memory writable for debugging.
4861 Enable the uart passthrough on the designated usb port
4862 on Rockchip SoCs. When active, the signals of the
4863 debug-uart get routed to the D+ and D- pins of the usb
4864 port and the regular usb controller gets disabled.
4866 root= [KNL] Root filesystem
4867 See name_to_dev_t comment in init/do_mounts.c.
4869 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
4870 mount the root filesystem
4872 rootflags= [KNL] Set root filesystem mount option string
4874 rootfstype= [KNL] Set root filesystem type
4876 rootwait [KNL] Wait (indefinitely) for root device to show up.
4877 Useful for devices that are detected asynchronously
4878 (e.g. USB and MMC devices).
4880 rproc_mem=nn[KMG][@address]
4881 [KNL,ARM,CMA] Remoteproc physical memory block.
4882 Memory area to be used by remote processor image,
4885 rw [KNL] Mount root device read-write on boot
4887 S [KNL] Run init in single mode
4889 s390_iommu= [HW,S390]
4890 Set s390 IOTLB flushing mode
4892 With strict flushing every unmap operation will result in
4893 an IOTLB flush. Default is lazy flushing before reuse,
4897 See drivers/net/irda/sa1100_ir.c.
4899 sbni= [NET] Granch SBNI12 leased line adapter
4901 sched_verbose [KNL] Enables verbose scheduler debug messages.
4903 schedstats= [KNL,X86] Enable or disable scheduled statistics.
4904 Allowed values are enable and disable. This feature
4905 incurs a small amount of overhead in the scheduler
4906 but is useful for debugging and performance tuning.
4908 sched_thermal_decay_shift=
4909 [KNL, SMP] Set a decay shift for scheduler thermal
4910 pressure signal. Thermal pressure signal follows the
4911 default decay period of other scheduler pelt
4912 signals(usually 32 ms but configurable). Setting
4913 sched_thermal_decay_shift will left shift the decay
4914 period for the thermal pressure signal by the shift
4916 i.e. with the default pelt decay period of 32 ms
4917 sched_thermal_decay_shift thermal pressure decay pr
4921 Format: integer between 0 and 10
4924 scftorture.holdoff= [KNL]
4925 Number of seconds to hold off before starting
4926 test. Defaults to zero for module insertion and
4927 to 10 seconds for built-in smp_call_function()
4930 scftorture.longwait= [KNL]
4931 Request ridiculously long waits randomly selected
4932 up to the chosen limit in seconds. Zero (the
4933 default) disables this feature. Please note
4934 that requesting even small non-zero numbers of
4935 seconds can result in RCU CPU stall warnings,
4936 softlockup complaints, and so on.
4938 scftorture.nthreads= [KNL]
4939 Number of kthreads to spawn to invoke the
4940 smp_call_function() family of functions.
4941 The default of -1 specifies a number of kthreads
4942 equal to the number of CPUs.
4944 scftorture.onoff_holdoff= [KNL]
4945 Number seconds to wait after the start of the
4946 test before initiating CPU-hotplug operations.
4948 scftorture.onoff_interval= [KNL]
4949 Number seconds to wait between successive
4950 CPU-hotplug operations. Specifying zero (which
4951 is the default) disables CPU-hotplug operations.
4953 scftorture.shutdown_secs= [KNL]
4954 The number of seconds following the start of the
4955 test after which to shut down the system. The
4956 default of zero avoids shutting down the system.
4957 Non-zero values are useful for automated tests.
4959 scftorture.stat_interval= [KNL]
4960 The number of seconds between outputting the
4961 current test statistics to the console. A value
4962 of zero disables statistics output.
4964 scftorture.stutter_cpus= [KNL]
4965 The number of jiffies to wait between each change
4966 to the set of CPUs under test.
4968 scftorture.use_cpus_read_lock= [KNL]
4969 Use use_cpus_read_lock() instead of the default
4970 preempt_disable() to disable CPU hotplug
4971 while invoking one of the smp_call_function*()
4974 scftorture.verbose= [KNL]
4975 Enable additional printk() statements.
4977 scftorture.weight_single= [KNL]
4978 The probability weighting to use for the
4979 smp_call_function_single() function with a zero
4980 "wait" parameter. A value of -1 selects the
4981 default if all other weights are -1. However,
4982 if at least one weight has some other value, a
4983 value of -1 will instead select a weight of zero.
4985 scftorture.weight_single_wait= [KNL]
4986 The probability weighting to use for the
4987 smp_call_function_single() function with a
4988 non-zero "wait" parameter. See weight_single.
4990 scftorture.weight_many= [KNL]
4991 The probability weighting to use for the
4992 smp_call_function_many() function with a zero
4993 "wait" parameter. See weight_single.
4994 Note well that setting a high probability for
4995 this weighting can place serious IPI load
4998 scftorture.weight_many_wait= [KNL]
4999 The probability weighting to use for the
5000 smp_call_function_many() function with a
5001 non-zero "wait" parameter. See weight_single
5004 scftorture.weight_all= [KNL]
5005 The probability weighting to use for the
5006 smp_call_function_all() function with a zero
5007 "wait" parameter. See weight_single and
5010 scftorture.weight_all_wait= [KNL]
5011 The probability weighting to use for the
5012 smp_call_function_all() function with a
5013 non-zero "wait" parameter. See weight_single
5016 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
5017 xtime_lock contention on larger systems, and/or RCU lock
5018 contention on all systems with CONFIG_MAXSMP set.
5019 Format: { "0" | "1" }
5020 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
5022 Note: increases power consumption, thus should only be
5023 enabled if running jitter sensitive (HPC/RT) workloads.
5025 security= [SECURITY] Choose a legacy "major" security module to
5026 enable at boot. This has been deprecated by the
5029 selinux= [SELINUX] Disable or enable SELinux at boot time.
5030 Format: { "0" | "1" }
5031 See security/selinux/Kconfig help text.
5036 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
5037 Format: { "0" | "1" }
5038 See security/apparmor/Kconfig help text
5041 Default value is set via kernel config option.
5043 serialnumber [BUGS=X86-32]
5046 Maximal number of shapers.
5054 Enable merging of slabs with similar size when the
5055 kernel is built without CONFIG_SLAB_MERGE_DEFAULT.
5058 Disable merging of slabs with similar size. May be
5059 necessary if there is some reason to distinguish
5060 allocs to different slabs, especially in hardened
5061 environments where the risk of heap overflows and
5062 layout control by attackers can usually be
5063 frustrated by disabling merging. This will reduce
5064 most of the exposure of a heap attack to a single
5065 cache (risks via metadata attacks are mostly
5066 unchanged). Debug options disable merging on their
5068 For more information see Documentation/vm/slub.rst.
5070 slab_max_order= [MM, SLAB]
5071 Determines the maximum allowed order for slabs.
5072 A high setting may cause OOMs due to memory
5073 fragmentation. Defaults to 1 for systems with
5074 more than 32MB of RAM, 0 otherwise.
5076 slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB]
5077 Enabling slub_debug allows one to determine the
5078 culprit if slab objects become corrupted. Enabling
5079 slub_debug can create guard zones around objects and
5080 may poison objects when not in use. Also tracks the
5081 last alloc / free. For more information see
5082 Documentation/vm/slub.rst.
5084 slub_max_order= [MM, SLUB]
5085 Determines the maximum allowed order for slabs.
5086 A high setting may cause OOMs due to memory
5087 fragmentation. For more information see
5088 Documentation/vm/slub.rst.
5090 slub_min_objects= [MM, SLUB]
5091 The minimum number of objects per slab. SLUB will
5092 increase the slab order up to slub_max_order to
5093 generate a sufficiently large slab able to contain
5094 the number of objects indicated. The higher the number
5095 of objects the smaller the overhead of tracking slabs
5096 and the less frequently locks need to be acquired.
5097 For more information see Documentation/vm/slub.rst.
5099 slub_min_order= [MM, SLUB]
5100 Determines the minimum page order for slabs. Must be
5101 lower than slub_max_order.
5102 For more information see Documentation/vm/slub.rst.
5104 slub_merge [MM, SLUB]
5105 Same with slab_merge.
5107 slub_nomerge [MM, SLUB]
5108 Same with slab_nomerge. This is supported for legacy.
5109 See slab_nomerge for more information.
5112 Format: <io1>[,<io2>[,...,<io8>]]
5114 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
5115 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
5116 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
5117 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
5118 smsc-ircc2.ircc_irq= [HW] IRQ line
5119 smsc-ircc2.ircc_dma= [HW] DMA channel
5120 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
5121 0: Toshiba Satellite 1800 (GP data pin select)
5122 1: Fast pin select (default)
5125 smt [KNL,S390] Set the maximum number of threads (logical
5126 CPUs) to use per physical CPU on systems capable of
5127 symmetric multithreading (SMT). Will be capped to the
5128 actual hardware limit.
5130 Default: -1 (no limit)
5133 [KNL] Should the soft-lockup detector generate panics.
5136 A value of 1 instructs the soft-lockup detector
5137 to panic the machine when a soft-lockup occurs. It is
5138 also controlled by the kernel.softlockup_panic sysctl
5139 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
5140 respective build-time switch to that functionality.
5142 softlockup_all_cpu_backtrace=
5143 [KNL] Should the soft-lockup detector generate
5144 backtraces on all cpus.
5147 sonypi.*= [HW] Sony Programmable I/O Control Device driver
5148 See Documentation/admin-guide/laptops/sonypi.rst
5150 spectre_v2= [X86] Control mitigation of Spectre variant 2
5151 (indirect branch speculation) vulnerability.
5152 The default operation protects the kernel from
5155 on - unconditionally enable, implies
5157 off - unconditionally disable, implies
5159 auto - kernel detects whether your CPU model is
5162 Selecting 'on' will, and 'auto' may, choose a
5163 mitigation method at run time according to the
5164 CPU, the available microcode, the setting of the
5165 CONFIG_RETPOLINE configuration option, and the
5166 compiler with which the kernel was built.
5168 Selecting 'on' will also enable the mitigation
5169 against user space to user space task attacks.
5171 Selecting 'off' will disable both the kernel and
5172 the user space protections.
5174 Specific mitigations can also be selected manually:
5176 retpoline - replace indirect branches
5177 retpoline,generic - google's original retpoline
5178 retpoline,amd - AMD-specific minimal thunk
5180 Not specifying this option is equivalent to
5184 [X86] Control mitigation of Spectre variant 2
5185 (indirect branch speculation) vulnerability between
5188 on - Unconditionally enable mitigations. Is
5189 enforced by spectre_v2=on
5191 off - Unconditionally disable mitigations. Is
5192 enforced by spectre_v2=off
5194 prctl - Indirect branch speculation is enabled,
5195 but mitigation can be enabled via prctl
5196 per thread. The mitigation control state
5197 is inherited on fork.
5200 - Like "prctl" above, but only STIBP is
5201 controlled per thread. IBPB is issued
5202 always when switching between different user
5206 - Same as "prctl" above, but all seccomp
5207 threads will enable the mitigation unless
5208 they explicitly opt out.
5211 - Like "seccomp" above, but only STIBP is
5212 controlled per thread. IBPB is issued
5213 always when switching between different
5214 user space processes.
5216 auto - Kernel selects the mitigation depending on
5217 the available CPU features and vulnerability.
5220 If CONFIG_SECCOMP=y then "seccomp", otherwise "prctl"
5222 Not specifying this option is equivalent to
5223 spectre_v2_user=auto.
5225 spec_store_bypass_disable=
5226 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
5227 (Speculative Store Bypass vulnerability)
5229 Certain CPUs are vulnerable to an exploit against a
5230 a common industry wide performance optimization known
5231 as "Speculative Store Bypass" in which recent stores
5232 to the same memory location may not be observed by
5233 later loads during speculative execution. The idea
5234 is that such stores are unlikely and that they can
5235 be detected prior to instruction retirement at the
5236 end of a particular speculation execution window.
5238 In vulnerable processors, the speculatively forwarded
5239 store can be used in a cache side channel attack, for
5240 example to read memory to which the attacker does not
5241 directly have access (e.g. inside sandboxed code).
5243 This parameter controls whether the Speculative Store
5244 Bypass optimization is used.
5246 On x86 the options are:
5248 on - Unconditionally disable Speculative Store Bypass
5249 off - Unconditionally enable Speculative Store Bypass
5250 auto - Kernel detects whether the CPU model contains an
5251 implementation of Speculative Store Bypass and
5252 picks the most appropriate mitigation. If the
5253 CPU is not vulnerable, "off" is selected. If the
5254 CPU is vulnerable the default mitigation is
5255 architecture and Kconfig dependent. See below.
5256 prctl - Control Speculative Store Bypass per thread
5257 via prctl. Speculative Store Bypass is enabled
5258 for a process by default. The state of the control
5259 is inherited on fork.
5260 seccomp - Same as "prctl" above, but all seccomp threads
5261 will disable SSB unless they explicitly opt out.
5263 Default mitigations:
5264 X86: If CONFIG_SECCOMP=y "seccomp", otherwise "prctl"
5266 On powerpc the options are:
5268 on,auto - On Power8 and Power9 insert a store-forwarding
5269 barrier on kernel entry and exit. On Power7
5270 perform a software flush on kernel entry and
5274 Not specifying this option is equivalent to
5275 spec_store_bypass_disable=auto.
5277 spia_io_base= [HW,MTD]
5283 [X86] Enable split lock detection or bus lock detection
5285 When enabled (and if hardware support is present), atomic
5286 instructions that access data across cache line
5287 boundaries will result in an alignment check exception
5288 for split lock detection or a debug exception for
5293 warn - the kernel will emit rate-limited warnings
5294 about applications triggering the #AC
5295 exception or the #DB exception. This mode is
5296 the default on CPUs that support split lock
5297 detection or bus lock detection. Default
5298 behavior is by #AC if both features are
5299 enabled in hardware.
5301 fatal - the kernel will send SIGBUS to applications
5302 that trigger the #AC exception or the #DB
5303 exception. Default behavior is by #AC if
5304 both features are enabled in hardware.
5307 Set system wide rate limit to N bus locks
5308 per second for bus lock detection.
5311 N/A for split lock detection.
5314 If an #AC exception is hit in the kernel or in
5315 firmware (i.e. not while executing in user mode)
5316 the kernel will oops in either "warn" or "fatal"
5319 #DB exception for bus lock is triggered only when
5323 Control the Special Register Buffer Data Sampling
5326 Certain CPUs are vulnerable to an MDS-like
5327 exploit which can leak bits from the random
5330 By default, this issue is mitigated by
5331 microcode. However, the microcode fix can cause
5332 the RDRAND and RDSEED instructions to become
5333 much slower. Among other effects, this will
5334 result in reduced throughput from /dev/urandom.
5336 The microcode mitigation can be disabled with
5337 the following option:
5339 off: Disable mitigation and remove
5340 performance impact to RDRAND and RDSEED
5342 srcutree.counter_wrap_check [KNL]
5343 Specifies how frequently to check for
5344 grace-period sequence counter wrap for the
5345 srcu_data structure's ->srcu_gp_seq_needed field.
5346 The greater the number of bits set in this kernel
5347 parameter, the less frequently counter wrap will
5348 be checked for. Note that the bottom two bits
5351 srcutree.exp_holdoff [KNL]
5352 Specifies how many nanoseconds must elapse
5353 since the end of the last SRCU grace period for
5354 a given srcu_struct until the next normal SRCU
5355 grace period will be considered for automatic
5356 expediting. Set to zero to disable automatic
5360 Speculative Store Bypass Disable control
5362 On CPUs that are vulnerable to the Speculative
5363 Store Bypass vulnerability and offer a
5364 firmware based mitigation, this parameter
5365 indicates how the mitigation should be used:
5367 force-on: Unconditionally enable mitigation for
5368 for both kernel and userspace
5369 force-off: Unconditionally disable mitigation for
5370 for both kernel and userspace
5371 kernel: Always enable mitigation in the
5372 kernel, and offer a prctl interface
5373 to allow userspace to register its
5374 interest in being mitigated too.
5376 stack_guard_gap= [MM]
5377 override the default stack gap protection. The value
5378 is in page units and it defines how many pages prior
5379 to (for stacks growing down) resp. after (for stacks
5380 growing up) the main stack are reserved for no other
5381 mapping. Default value is 256 pages.
5383 stack_depot_disable= [KNL]
5384 Setting this to true through kernel command line will
5385 disable the stack depot thereby saving the static memory
5386 consumed by the stack hash table. By default this is set
5390 Enabled the stack tracer on boot up.
5392 stacktrace_filter=[function-list]
5393 [FTRACE] Limit the functions that the stack tracer
5394 will trace at boot up. function-list is a comma-separated
5395 list of functions. This list can be changed at run
5396 time by the stack_trace_filter file in the debugfs
5397 tracing directory. Note, this enables stack tracing
5398 and the stacktrace above is not needed.
5402 Set the STI (builtin display/keyboard on the HP-PARISC
5403 machines) console (graphic card) which should be used
5404 as the initial boot-console.
5405 See also comment in drivers/video/console/sticore.c.
5408 See comment in drivers/video/console/sticore.c.
5411 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
5413 sunrpc.min_resvport=
5414 sunrpc.max_resvport=
5416 SunRPC servers often require that client requests
5417 originate from a privileged port (i.e. a port in the
5418 range 0 < portnr < 1024).
5419 An administrator who wishes to reserve some of these
5420 ports for other uses may adjust the range that the
5421 kernel's sunrpc client considers to be privileged
5422 using these two parameters to set the minimum and
5423 maximum port values.
5425 sunrpc.svc_rpc_per_connection_limit=
5427 Limit the number of requests that the server will
5428 process in parallel from a single connection.
5429 The default value is 0 (no limit).
5433 Control how the NFS server code allocates CPUs to
5434 service thread pools. Depending on how many NICs
5435 you have and where their interrupts are bound, this
5436 option will affect which CPUs will do NFS serving.
5437 Note: this parameter cannot be changed while the
5438 NFS server is running.
5440 auto the server chooses an appropriate mode
5441 automatically using heuristics
5442 global a single global pool contains all CPUs
5443 percpu one pool for each CPU
5444 pernode one pool for each NUMA node (equivalent
5445 to global on non-NUMA machines)
5447 sunrpc.tcp_slot_table_entries=
5448 sunrpc.udp_slot_table_entries=
5450 Sets the upper limit on the number of simultaneous
5451 RPC calls that can be sent from the client to a
5452 server. Increasing these values may allow you to
5453 improve throughput, but will also increase the
5454 amount of memory reserved for use by the client.
5456 suspend.pm_test_delay=
5458 Sets the number of seconds to remain in a suspend test
5459 mode before resuming the system (see
5460 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
5461 is set. Default value is 5.
5464 Format: { on | off | y | n | 1 | 0 }
5465 This parameter controls use of the Protected
5466 Execution Facility on pSeries.
5469 [KNL] Enable accounting of swap in memory resource
5470 controller if no parameter or 1 is given or disable
5471 it if 0 is given (See Documentation/admin-guide/cgroup-v1/memory.rst)
5473 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
5474 Format: { <int> | force | noforce }
5475 <int> -- Number of I/O TLB slabs
5476 force -- force using of bounce buffers even if they
5477 wouldn't be automatically used by the kernel
5478 noforce -- Never use bounce buffers (for debugging)
5483 Set a sysctl parameter, right before loading the init
5484 process, as if the value was written to the respective
5485 /proc/sys/... file. Both '.' and '/' are recognized as
5486 separators. Unrecognized parameters and invalid values
5487 are reported in the kernel log. Sysctls registered
5488 later by a loaded module cannot be set this way.
5489 Example: sysctl.vm.swappiness=40
5491 sysfs.deprecated=0|1 [KNL]
5492 Enable/disable old style sysfs layout for old udev
5493 on older distributions. When this option is enabled
5494 very new udev will not work anymore. When this option
5495 is disabled (or CONFIG_SYSFS_DEPRECATED not compiled)
5496 in older udev will not work anymore.
5497 Default depends on CONFIG_SYSFS_DEPRECATED_V2 set in
5498 the kernel configuration.
5500 sysrq_always_enabled
5502 Ignore sysrq setting - this boot parameter will
5503 neutralize any effect of /proc/sys/kernel/sysrq.
5504 Useful for debugging.
5506 tcpmhash_entries= [KNL,NET]
5507 Set the number of tcp_metrics_hash slots.
5508 Default value is 8192 or 16384 depending on total
5509 ram pages. This is used to specify the TCP metrics
5510 cache size. See Documentation/networking/ip-sysctl.rst
5511 "tcp_no_metrics_save" section for more details.
5515 test_suspend= [SUSPEND][,N]
5516 Specify "mem" (for Suspend-to-RAM) or "standby" (for
5517 standby suspend) or "freeze" (for suspend type freeze)
5518 as the system sleep state during system startup with
5519 the optional capability to repeat N number of times.
5520 The system is woken from this state using a
5521 wakeup-capable RTC alarm.
5523 thash_entries= [KNL,NET]
5524 Set number of hash buckets for TCP connection
5526 thermal.act= [HW,ACPI]
5527 -1: disable all active trip points in all thermal zones
5528 <degrees C>: override all lowest active trip points
5530 thermal.crt= [HW,ACPI]
5531 -1: disable all critical trip points in all thermal zones
5532 <degrees C>: override all critical trip points
5534 thermal.nocrt= [HW,ACPI]
5535 Set to disable actions on ACPI thermal zone
5536 critical and hot trip points.
5538 thermal.off= [HW,ACPI]
5539 1: disable ACPI thermal control
5541 thermal.psv= [HW,ACPI]
5542 -1: disable all passive trip points
5543 <degrees C>: override all passive trip points to this
5546 thermal.tzp= [HW,ACPI]
5547 Specify global default ACPI thermal zone polling rate
5548 <deci-seconds>: poll all this frequency
5549 0: no polling (default)
5552 Force threading of all interrupt handlers except those
5553 marked explicitly IRQF_NO_THREAD.
5557 Specify if the kernel should make use of the cpu
5558 topology information if the hardware supports this.
5559 The scheduler will make use of this information and
5560 e.g. base its process migration decisions on it.
5563 topology_updates= [KNL, PPC, NUMA]
5565 Specify if the kernel should ignore (off)
5566 topology updates sent by the hypervisor to this
5569 torture.disable_onoff_at_boot= [KNL]
5570 Prevent the CPU-hotplug component of torturing
5571 until after init has spawned.
5573 torture.ftrace_dump_at_shutdown= [KNL]
5574 Dump the ftrace buffer at torture-test shutdown,
5575 even if there were no errors. This can be a
5576 very costly operation when many torture tests
5577 are running concurrently, especially on systems
5578 with rotating-rust storage.
5580 torture.verbose_sleep_frequency= [KNL]
5581 Specifies how many verbose printk()s should be
5582 emitted between each sleep. The default of zero
5583 disables verbose-printk() sleeping.
5585 torture.verbose_sleep_duration= [KNL]
5586 Duration of each verbose-printk() sleep in jiffies.
5590 tpm_suspend_pcr=[HW,TPM]
5591 Format: integer pcr id
5592 Specify that at suspend time, the tpm driver
5593 should extend the specified pcr with zeros,
5594 as a workaround for some chips which fail to
5595 flush the last written pcr on TPM_SaveState.
5596 This will guarantee that all the other pcrs
5599 trace_buf_size=nn[KMG]
5600 [FTRACE] will set tracing buffer size on each cpu.
5602 trace_event=[event-list]
5603 [FTRACE] Set and start specified trace events in order
5604 to facilitate early boot debugging. The event-list is a
5605 comma-separated list of trace events to enable. See
5606 also Documentation/trace/events.rst
5608 trace_options=[option-list]
5609 [FTRACE] Enable or disable tracer options at boot.
5610 The option-list is a comma delimited list of options
5611 that can be enabled or disabled just as if you were
5612 to echo the option name into
5614 /sys/kernel/debug/tracing/trace_options
5616 For example, to enable stacktrace option (to dump the
5617 stack trace of each event), add to the command line:
5619 trace_options=stacktrace
5621 See also Documentation/trace/ftrace.rst "trace options"
5625 Have the tracepoints sent to printk as well as the
5626 tracing ring buffer. This is useful for early boot up
5627 where the system hangs or reboots and does not give the
5628 option for reading the tracing buffer or performing a
5629 ftrace_dump_on_oops.
5631 To turn off having tracepoints sent to printk,
5632 echo 0 > /proc/sys/kernel/tracepoint_printk
5633 Note, echoing 1 into this file without the
5634 tracepoint_printk kernel cmdline option has no effect.
5638 Having tracepoints sent to printk() and activating high
5639 frequency tracepoints such as irq or sched, can cause
5640 the system to live lock.
5643 [FTRACE] enable this option to disable tracing when a
5644 warning is hit. This turns off "tracing_on". Tracing can
5645 be enabled again by echoing '1' into the "tracing_on"
5646 file located in /sys/kernel/debug/tracing/
5648 This option is useful, as it disables the trace before
5649 the WARNING dump is called, which prevents the trace to
5650 be filled with content caused by the warning output.
5652 This option can also be set at run time via the sysctl
5653 option: kernel/traceoff_on_warning
5655 transparent_hugepage=
5657 Format: [always|madvise|never]
5658 Can be used to control the default behavior of the system
5659 with respect to transparent hugepages.
5660 See Documentation/admin-guide/mm/transhuge.rst
5663 trusted.source= [KEYS]
5665 This parameter identifies the trust source as a backend
5666 for trusted keys implementation. Supported trust
5670 If not specified then it defaults to iterating through
5671 the trust source list starting with TPM and assigns the
5672 first trust source as a backend which is initialized
5673 successfully during iteration.
5675 tsc= Disable clocksource stability checks for TSC.
5677 [x86] reliable: mark tsc clocksource as reliable, this
5678 disables clocksource verification at runtime, as well
5679 as the stability checks done at bootup. Used to enable
5680 high-resolution timer mode on older hardware, and in
5681 virtualized environment.
5682 [x86] noirqtime: Do not use TSC to do irq accounting.
5683 Used to run time disable IRQ_TIME_ACCOUNTING on any
5684 platforms where RDTSC is slow and this accounting
5686 [x86] unstable: mark the TSC clocksource as unstable, this
5687 marks the TSC unconditionally unstable at bootup and
5688 avoids any further wobbles once the TSC watchdog notices.
5689 [x86] nowatchdog: disable clocksource watchdog. Used
5690 in situations with strict latency requirements (where
5691 interruptions from clocksource watchdog are not
5694 tsc_early_khz= [X86] Skip early TSC calibration and use the given
5695 value instead. Useful when the early TSC frequency discovery
5696 procedure is not reliable, such as on overclocked systems
5697 with CPUID.16h support and partial CPUID.15h support.
5698 Format: <unsigned int>
5700 tsx= [X86] Control Transactional Synchronization
5701 Extensions (TSX) feature in Intel processors that
5702 support TSX control.
5704 This parameter controls the TSX feature. The options are:
5706 on - Enable TSX on the system. Although there are
5707 mitigations for all known security vulnerabilities,
5708 TSX has been known to be an accelerator for
5709 several previous speculation-related CVEs, and
5710 so there may be unknown security risks associated
5711 with leaving it enabled.
5713 off - Disable TSX on the system. (Note that this
5714 option takes effect only on newer CPUs which are
5715 not vulnerable to MDS, i.e., have
5716 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
5717 the new IA32_TSX_CTRL MSR through a microcode
5718 update. This new MSR allows for the reliable
5719 deactivation of the TSX functionality.)
5721 auto - Disable TSX if X86_BUG_TAA is present,
5722 otherwise enable TSX on the system.
5724 Not specifying this option is equivalent to tsx=off.
5726 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5729 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
5730 Abort (TAA) vulnerability.
5732 Similar to Micro-architectural Data Sampling (MDS)
5733 certain CPUs that support Transactional
5734 Synchronization Extensions (TSX) are vulnerable to an
5735 exploit against CPU internal buffers which can forward
5736 information to a disclosure gadget under certain
5739 In vulnerable processors, the speculatively forwarded
5740 data can be used in a cache side channel attack, to
5741 access data to which the attacker does not have direct
5744 This parameter controls the TAA mitigation. The
5747 full - Enable TAA mitigation on vulnerable CPUs
5750 full,nosmt - Enable TAA mitigation and disable SMT on
5751 vulnerable CPUs. If TSX is disabled, SMT
5752 is not disabled because CPU is not
5753 vulnerable to cross-thread TAA attacks.
5754 off - Unconditionally disable TAA mitigation
5756 On MDS-affected machines, tsx_async_abort=off can be
5757 prevented by an active MDS mitigation as both vulnerabilities
5758 are mitigated with the same mechanism so in order to disable
5759 this mitigation, you need to specify mds=off too.
5761 Not specifying this option is equivalent to
5762 tsx_async_abort=full. On CPUs which are MDS affected
5763 and deploy MDS mitigation, TAA mitigation is not
5764 required and doesn't provide any additional
5768 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5770 turbografx.map[2|3]= [HW,JOY]
5771 TurboGraFX parallel port interface
5773 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
5774 See also Documentation/input/devices/joystick-parport.rst
5776 udbg-immortal [PPC] When debugging early kernel crashes that
5777 happen after console_init() and before a proper
5778 console driver takes over, this boot options might
5779 help "seeing" what's going on.
5781 uhash_entries= [KNL,NET]
5782 Set number of hash buckets for UDP/UDP-Lite connections
5785 [USB] Ignore overcurrent events (default N).
5786 Some badly-designed motherboards generate lots of
5787 bogus events, for ports that aren't wired to
5788 anything. Set this parameter to avoid log spamming.
5789 Note that genuine overcurrent events won't be
5793 [X86] Cause panic on unknown NMI.
5795 usbcore.authorized_default=
5796 [USB] Default USB device authorization:
5797 (default -1 = authorized except for wireless USB,
5798 0 = not authorized, 1 = authorized, 2 = authorized
5799 if device connected to internal port)
5801 usbcore.autosuspend=
5802 [USB] The autosuspend time delay (in seconds) used
5803 for newly-detected USB devices (default 2). This
5804 is the time required before an idle device will be
5805 autosuspended. Devices for which the delay is set
5806 to a negative value won't be autosuspended at all.
5808 usbcore.usbfs_snoop=
5809 [USB] Set to log all usbfs traffic (default 0 = off).
5811 usbcore.usbfs_snoop_max=
5812 [USB] Maximum number of bytes to snoop in each URB
5815 usbcore.blinkenlights=
5816 [USB] Set to cycle leds on hubs (default 0 = off).
5818 usbcore.old_scheme_first=
5819 [USB] Start with the old device initialization
5820 scheme (default 0 = off).
5822 usbcore.usbfs_memory_mb=
5823 [USB] Memory limit (in MB) for buffers allocated by
5824 usbfs (default = 16, 0 = max = 2047).
5826 usbcore.use_both_schemes=
5827 [USB] Try the other device initialization scheme
5828 if the first one fails (default 1 = enabled).
5830 usbcore.initial_descriptor_timeout=
5831 [USB] Specifies timeout for the initial 64-byte
5832 USB_REQ_GET_DESCRIPTOR request in milliseconds
5833 (default 5000 = 5.0 seconds).
5835 usbcore.nousb [USB] Disable the USB subsystem
5838 [USB] A list of quirk entries to augment the built-in
5839 usb core quirk list. List entries are separated by
5840 commas. Each entry has the form
5841 VendorID:ProductID:Flags. The IDs are 4-digit hex
5842 numbers and Flags is a set of letters. Each letter
5843 will change the built-in quirk; setting it if it is
5844 clear and clearing it if it is set. The letters have
5845 the following meanings:
5846 a = USB_QUIRK_STRING_FETCH_255 (string
5847 descriptors must not be fetched using
5849 b = USB_QUIRK_RESET_RESUME (device can't resume
5850 correctly so reset it instead);
5851 c = USB_QUIRK_NO_SET_INTF (device can't handle
5852 Set-Interface requests);
5853 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
5854 handle its Configuration or Interface
5856 e = USB_QUIRK_RESET (device can't be reset
5857 (e.g morph devices), don't use reset);
5858 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
5859 more interface descriptions than the
5860 bNumInterfaces count, and can't handle
5861 talking to these interfaces);
5862 g = USB_QUIRK_DELAY_INIT (device needs a pause
5863 during initialization, after we read
5864 the device descriptor);
5865 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
5866 high speed and super speed interrupt
5867 endpoints, the USB 2.0 and USB 3.0 spec
5868 require the interval in microframes (1
5869 microframe = 125 microseconds) to be
5870 calculated as interval = 2 ^
5872 Devices with this quirk report their
5873 bInterval as the result of this
5874 calculation instead of the exponent
5875 variable used in the calculation);
5876 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
5877 handle device_qualifier descriptor
5879 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
5880 generates spurious wakeup, ignore
5881 remote wakeup capability);
5882 k = USB_QUIRK_NO_LPM (device can't handle Link
5884 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
5885 (Device reports its bInterval as linear
5886 frames instead of the USB 2.0
5888 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
5889 to be disconnected before suspend to
5890 prevent spurious wakeup);
5891 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
5892 pause after every control message);
5893 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
5894 delay after resetting its port);
5895 Example: quirks=0781:5580:bk,0a5c:5834:gij
5898 [USBHID] The interval which mice are to be polled at.
5901 [USBHID] The interval which joysticks are to be polled at.
5904 [USBHID] The interval which keyboards are to be polled at.
5906 usb-storage.delay_use=
5907 [UMS] The delay in seconds before a new device is
5908 scanned for Logical Units (default 1).
5911 [UMS] A list of quirks entries to supplement or
5912 override the built-in unusual_devs list. List
5913 entries are separated by commas. Each entry has
5914 the form VID:PID:Flags where VID and PID are Vendor
5915 and Product ID values (4-digit hex numbers) and
5916 Flags is a set of characters, each corresponding
5917 to a common usb-storage quirk flag as follows:
5918 a = SANE_SENSE (collect more than 18 bytes
5919 of sense data, not on uas);
5920 b = BAD_SENSE (don't collect more than 18
5921 bytes of sense data, not on uas);
5922 c = FIX_CAPACITY (decrease the reported
5923 device capacity by one sector);
5924 d = NO_READ_DISC_INFO (don't use
5925 READ_DISC_INFO command, not on uas);
5926 e = NO_READ_CAPACITY_16 (don't use
5927 READ_CAPACITY_16 command);
5928 f = NO_REPORT_OPCODES (don't use report opcodes
5930 g = MAX_SECTORS_240 (don't transfer more than
5931 240 sectors at a time, uas only);
5932 h = CAPACITY_HEURISTICS (decrease the
5933 reported device capacity by one
5934 sector if the number is odd);
5935 i = IGNORE_DEVICE (don't bind to this
5937 j = NO_REPORT_LUNS (don't use report luns
5939 k = NO_SAME (do not use WRITE_SAME, uas only)
5940 l = NOT_LOCKABLE (don't try to lock and
5941 unlock ejectable media, not on uas);
5942 m = MAX_SECTORS_64 (don't transfer more
5943 than 64 sectors = 32 KB at a time,
5945 n = INITIAL_READ10 (force a retry of the
5946 initial READ(10) command, not on uas);
5947 o = CAPACITY_OK (accept the capacity
5948 reported by the device, not on uas);
5949 p = WRITE_CACHE (the device cache is ON
5950 by default, not on uas);
5951 r = IGNORE_RESIDUE (the device reports
5952 bogus residue values, not on uas);
5953 s = SINGLE_LUN (the device has only one
5955 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
5956 commands, uas only);
5957 u = IGNORE_UAS (don't bind to the uas driver);
5958 w = NO_WP_DETECT (don't test whether the
5959 medium is write-protected).
5960 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
5961 even if the device claims no cache,
5963 Example: quirks=0419:aaf5:rl,0421:0433:rc
5965 user_debug= [KNL,ARM]
5967 See arch/arm/Kconfig.debug help text.
5968 1 - undefined instruction events
5970 4 - invalid data aborts
5973 Example: user_debug=31
5976 [X86] Flags controlling user PTE allocations.
5978 nohigh = do not allocate PTE pages in
5979 HIGHMEM regardless of setting
5983 On X86_32, this is an alias for vdso32=. Otherwise:
5985 vdso=1: enable VDSO (the default)
5986 vdso=0: disable VDSO mapping
5988 vdso32= [X86] Control the 32-bit vDSO
5989 vdso32=1: enable 32-bit VDSO
5990 vdso32=0 or vdso32=2: disable 32-bit VDSO
5992 See the help text for CONFIG_COMPAT_VDSO for more
5993 details. If CONFIG_COMPAT_VDSO is set, the default is
5994 vdso32=0; otherwise, the default is vdso32=1.
5996 For compatibility with older kernels, vdso32=2 is an
5999 Try vdso32=0 if you encounter an error that says:
6000 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
6003 vector=percpu: enable percpu vector domain
6005 video= [FB] Frame buffer configuration
6006 See Documentation/fb/modedb.rst.
6008 video.brightness_switch_enabled= [0,1]
6009 If set to 1, on receiving an ACPI notify event
6010 generated by hotkey, video driver will adjust brightness
6011 level and then send out the event to user space through
6012 the allocated input device; If set to 0, video driver
6013 will only send out the event without touching backlight
6018 [VMMIO] Memory mapped virtio (platform) device.
6020 <size>@<baseaddr>:<irq>[:<id>]
6022 <size> := size (can use standard suffixes
6024 <baseaddr> := physical base address
6025 <irq> := interrupt number (as passed to
6027 <id> := (optional) platform device id
6029 virtio_mmio.device=1K@0x100b0000:48:7
6031 Can be used multiple times for multiple devices.
6033 vga= [BOOT,X86-32] Select a particular video mode
6034 See Documentation/x86/boot.rst and
6035 Documentation/admin-guide/svga.rst.
6036 Use vga=ask for menu.
6037 This is actually a boot loader parameter; the value is
6038 passed to the kernel using a special protocol.
6040 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
6041 May slow down system boot speed, especially when
6042 enabled on systems with a large amount of memory.
6043 All options are enabled by default, and this
6044 interface is meant to allow for selectively
6045 enabling or disabling specific virtual memory
6048 Available options are:
6049 P Enable page structure init time poisoning
6050 - Disable all of the above options
6052 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
6053 size of <nn>. This can be used to increase the
6054 minimum size (128MB on x86). It can also be used to
6055 decrease the size and leave more room for directly
6058 vmcp_cma=nn[MG] [KNL,S390]
6059 Sets the memory size reserved for contiguous memory
6060 allocations for the vmcp device driver.
6062 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
6065 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
6068 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
6072 Controls the behavior of vsyscalls (i.e. calls to
6073 fixed addresses of 0xffffffffff600x00 from legacy
6074 code). Most statically-linked binaries and older
6075 versions of glibc use these calls. Because these
6076 functions are at fixed addresses, they make nice
6077 targets for exploits that can control RIP.
6079 emulate [default] Vsyscalls turn into traps and are
6080 emulated reasonably safely. The vsyscall
6083 xonly Vsyscalls turn into traps and are
6084 emulated reasonably safely. The vsyscall
6085 page is not readable.
6087 none Vsyscalls don't work at all. This makes
6088 them quite hard to use for exploits but
6089 might break your system.
6091 vt.color= [VT] Default text color.
6092 Format: 0xYX, X = foreground, Y = background.
6093 Default: 0x07 = light gray on black.
6095 vt.cur_default= [VT] Default cursor shape.
6096 Format: 0xCCBBAA, where AA, BB, and CC are the same as
6097 the parameters of the <Esc>[?A;B;Cc escape sequence;
6098 see VGA-softcursor.txt. Default: 2 = underline.
6100 vt.default_blu= [VT]
6101 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
6102 Change the default blue palette of the console.
6103 This is a 16-member array composed of values
6106 vt.default_grn= [VT]
6107 Format: <green0>,<green1>,<green2>,...,<green15>
6108 Change the default green palette of the console.
6109 This is a 16-member array composed of values
6112 vt.default_red= [VT]
6113 Format: <red0>,<red1>,<red2>,...,<red15>
6114 Change the default red palette of the console.
6115 This is a 16-member array composed of values
6121 Set system-wide default UTF-8 mode for all tty's.
6122 Default is 1, i.e. UTF-8 mode is enabled for all
6123 newly opened terminals.
6125 vt.global_cursor_default=
6128 Set system-wide default for whether a cursor
6129 is shown on new VTs. Default is -1,
6130 i.e. cursors will be created by default unless
6131 overridden by individual drivers. 0 will hide
6132 cursors, 1 will display them.
6134 vt.italic= [VT] Default color for italic text; 0-15.
6137 vt.underline= [VT] Default color for underlined text; 0-15.
6140 watchdog timers [HW,WDT] For information on watchdog timers,
6141 see Documentation/watchdog/watchdog-parameters.rst
6142 or other driver-specific files in the
6143 Documentation/watchdog/ directory.
6147 Set the hard lockup detector stall duration
6148 threshold in seconds. The soft lockup detector
6149 threshold is set to twice the value. A value of 0
6150 disables both lockup detectors. Default is 10
6153 workqueue.watchdog_thresh=
6154 If CONFIG_WQ_WATCHDOG is configured, workqueue can
6155 warn stall conditions and dump internal state to
6156 help debugging. 0 disables workqueue stall
6157 detection; otherwise, it's the stall threshold
6158 duration in seconds. The default value is 30 and
6159 it can be updated at runtime by writing to the
6160 corresponding sysfs file.
6162 workqueue.disable_numa
6163 By default, all work items queued to unbound
6164 workqueues are affine to the NUMA nodes they're
6165 issued on, which results in better behavior in
6166 general. If NUMA affinity needs to be disabled for
6167 whatever reason, this option can be used. Note
6168 that this also can be controlled per-workqueue for
6169 workqueues visible under /sys/bus/workqueue/.
6171 workqueue.power_efficient
6172 Per-cpu workqueues are generally preferred because
6173 they show better performance thanks to cache
6174 locality; unfortunately, per-cpu workqueues tend to
6175 be more power hungry than unbound workqueues.
6177 Enabling this makes the per-cpu workqueues which
6178 were observed to contribute significantly to power
6179 consumption unbound, leading to measurably lower
6180 power usage at the cost of small performance
6183 The default value of this parameter is determined by
6184 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
6186 workqueue.debug_force_rr_cpu
6187 Workqueue used to implicitly guarantee that work
6188 items queued without explicit CPU specified are put
6189 on the local CPU. This guarantee is no longer true
6190 and while local CPU is still preferred work items
6191 may be put on foreign CPUs. This debug option
6192 forces round-robin CPU selection to flush out
6193 usages which depend on the now broken guarantee.
6194 When enabled, memory and cache locality will be
6197 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
6198 default x2apic cluster mode on platforms
6201 xen_512gb_limit [KNL,X86-64,XEN]
6202 Restricts the kernel running paravirtualized under Xen
6203 to use only up to 512 GB of RAM. The reason to do so is
6204 crash analysis tools and Xen tools for doing domain
6205 save/restore/migration must be enabled to handle larger
6208 xen_emul_unplug= [HW,X86,XEN]
6209 Unplug Xen emulated devices
6210 Format: [unplug0,][unplug1]
6211 ide-disks -- unplug primary master IDE devices
6212 aux-ide-disks -- unplug non-primary-master IDE devices
6213 nics -- unplug network devices
6214 all -- unplug all emulated devices (NICs and IDE disks)
6215 unnecessary -- unplugging emulated devices is
6216 unnecessary even if the host did not respond to
6218 never -- do not unplug even if version check succeeds
6220 xen_legacy_crash [X86,XEN]
6221 Crash from Xen panic notifier, without executing late
6222 panic() code such as dumping handler.
6224 xen_nopvspin [X86,XEN]
6225 Disables the qspinlock slowpath using Xen PV optimizations.
6226 This parameter is obsoleted by "nopvspin" parameter, which
6227 has equivalent effect for XEN platform.
6230 Disables the PV optimizations forcing the HVM guest to
6231 run as generic HVM guest with no PV drivers.
6232 This option is obsoleted by the "nopv" option, which
6233 has equivalent effect for XEN platform.
6235 xen_no_vector_callback
6236 [KNL,X86,XEN] Disable the vector callback for Xen
6237 event channel interrupts.
6239 xen_scrub_pages= [XEN]
6240 Boolean option to control scrubbing pages before giving them back
6241 to Xen, for use by other domains. Can be also changed at runtime
6242 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
6243 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
6245 xen_timer_slop= [X86-64,XEN]
6246 Set the timer slop (in nanoseconds) for the virtual Xen
6247 timers (default is 100000). This adjusts the minimum
6248 delta of virtualized Xen timers, where lower values
6249 improve timer resolution at the expense of processing
6250 more timer interrupts.
6252 xen.event_eoi_delay= [XEN]
6253 How long to delay EOI handling in case of event
6254 storms (jiffies). Default is 10.
6256 xen.event_loop_timeout= [XEN]
6257 After which time (jiffies) the event handling loop
6258 should start to delay EOI handling. Default is 2.
6260 xen.fifo_events= [XEN]
6261 Boolean parameter to disable using fifo event handling
6262 even if available. Normally fifo event handling is
6263 preferred over the 2-level event handling, as it is
6264 fairer and the number of possible event channels is
6265 much higher. Default is on (use fifo events).
6267 nopv= [X86,XEN,KVM,HYPER_V,VMWARE]
6268 Disables the PV optimizations forcing the guest to run
6269 as generic guest with no PV drivers. Currently support
6270 XEN HVM, KVM, HYPER_V and VMWARE guest.
6272 nopvspin [X86,XEN,KVM]
6273 Disables the qspinlock slow path using PV optimizations
6274 which allow the hypervisor to 'idle' the guest on lock
6277 xirc2ps_cs= [NET,PCMCIA]
6279 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
6282 By default on POWER9 and above, the kernel will
6283 natively use the XIVE interrupt controller. This option
6284 allows the fallback firmware mode to be used:
6286 off Fallback to firmware control of XIVE interrupt
6287 controller on both pseries and powernv
6288 platforms. Only useful on POWER9 and above.
6290 xhci-hcd.quirks [USB,KNL]
6291 A hex value specifying bitmask with supplemental xhci
6292 host controller quirks. Meaning of each bit can be
6293 consulted in header drivers/usb/host/xhci.h.
6296 Format: { early | on | rw | ro | off }
6297 Controls if xmon debugger is enabled. Default is off.
6298 Passing only "xmon" is equivalent to "xmon=early".
6299 early Call xmon as early as possible on boot; xmon
6300 debugger is called from setup_arch().
6301 on xmon debugger hooks will be installed so xmon
6302 is only called on a kernel crash. Default mode,
6303 i.e. either "ro" or "rw" mode, is controlled
6304 with CONFIG_XMON_DEFAULT_RO_MODE.
6305 rw xmon debugger hooks will be installed so xmon
6306 is called only on a kernel crash, mode is write,
6307 meaning SPR registers, memory and, other data
6308 can be written using xmon commands.
6309 ro same as "rw" option above but SPR registers,
6310 memory, and other data can't be written using
6312 off xmon is disabled.