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 or optional feature
501 Format: {name of the controller(s) or feature(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 - if foo is an optional feature then the feature is
508 disabled and corresponding cgroup files are not
510 {Currently only "memory" controller deal with this and
511 cut the overhead, others just disable the usage. So
512 only cgroup_disable=memory is actually worthy}
513 Specifying "pressure" disables per-cgroup pressure
514 stall information accounting feature
516 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
517 Format: { { controller | "all" | "named" }
518 [,{ controller | "all" | "named" }...] }
519 Like cgroup_disable, but only applies to cgroup v1;
520 the blacklisted controllers remain available in cgroup2.
521 "all" blacklists all controllers and "named" disables
522 named mounts. Specifying both "all" and "named" disables
525 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
527 nosocket -- Disable socket memory accounting.
528 nokmem -- Disable kernel memory accounting.
530 checkreqprot [SELINUX] Set initial checkreqprot flag value.
531 Format: { "0" | "1" }
532 See security/selinux/Kconfig help text.
533 0 -- check protection applied by kernel (includes
534 any implied execute protection).
535 1 -- check protection requested by application.
536 Default value is set via a kernel config option.
537 Value can be changed at runtime via
538 /sys/fs/selinux/checkreqprot.
539 Setting checkreqprot to 1 is deprecated.
542 See Documentation/s390/common_io.rst for details.
545 Prevents the clock framework from automatically gating
546 clocks that have not been explicitly enabled by a Linux
547 device driver but are enabled in hardware at reset or
548 by the bootloader/firmware. Note that this does not
549 force such clocks to be always-on nor does it reserve
550 those clocks in any way. This parameter is useful for
551 debug and development, but should not be needed on a
552 platform with proper driver support. For more
553 information, see Documentation/driver-api/clk.rst.
555 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
557 Forces specified clocksource (if available) to be used
558 when calculating gettimeofday(). If specified
559 clocksource is not available, it defaults to PIT.
560 Format: { pit | tsc | cyclone | pmtmr }
562 clocksource= Override the default clocksource
564 Override the default clocksource and use the clocksource
565 with the name specified.
566 Some clocksource names to choose from, depending on
568 [all] jiffies (this is the base, fallback clocksource)
570 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
571 pxa_timer,timer3,32k_counter,timer0_1
572 [X86-32] pit,hpet,tsc;
573 scx200_hrt on Geode; cyclone on IBM x440
581 clocksource.arm_arch_timer.evtstrm=
584 Enable/disable the eventstream feature of the ARM
585 architected timer so that code using WFE-based polling
586 loops can be debugged more effectively on production
589 clocksource.max_cswd_read_retries= [KNL]
590 Number of clocksource_watchdog() retries due to
591 external delays before the clock will be marked
592 unstable. Defaults to three retries, that is,
593 four attempts to read the clock under test.
595 clocksource.verify_n_cpus= [KNL]
596 Limit the number of CPUs checked for clocksources
597 marked with CLOCK_SOURCE_VERIFY_PERCPU that
598 are marked unstable due to excessive skew.
599 A negative value says to check all CPUs, while
600 zero says not to check any. Values larger than
601 nr_cpu_ids are silently truncated to nr_cpu_ids.
602 The actual CPUs are chosen randomly, with
603 no replacement if the same CPU is chosen twice.
605 clocksource-wdtest.holdoff= [KNL]
606 Set the time in seconds that the clocksource
607 watchdog test waits before commencing its tests.
608 Defaults to zero when built as a module and to
609 10 seconds when built into the kernel.
611 clearcpuid=BITNUM[,BITNUM...] [X86]
612 Disable CPUID feature X for the kernel. See
613 arch/x86/include/asm/cpufeatures.h for the valid bit
614 numbers. Note the Linux specific bits are not necessarily
615 stable over kernel options, but the vendor specific
617 Also note that user programs calling CPUID directly
618 or using the feature without checking anything
619 will still see it. This just prevents it from
620 being used by the kernel or shown in /proc/cpuinfo.
621 Also note the kernel might malfunction if you disable
624 cma=nn[MG]@[start[MG][-end[MG]]]
626 Sets the size of kernel global memory area for
627 contiguous memory allocations and optionally the
628 placement constraint by the physical address range of
629 memory allocations. A value of 0 disables CMA
630 altogether. For more information, see
631 kernel/dma/contiguous.c
635 Sets the size of kernel per-numa memory area for
636 contiguous memory allocations. A value of 0 disables
637 per-numa CMA altogether. And If this option is not
638 specificed, the default value is 0.
639 With per-numa CMA enabled, DMA users on node nid will
640 first try to allocate buffer from the pernuma area
641 which is located in node nid, if the allocation fails,
642 they will fallback to the global default memory area.
644 cmo_free_hint= [PPC] Format: { yes | no }
645 Specify whether pages are marked as being inactive
646 when they are freed. This is used in CMO environments
647 to determine OS memory pressure for page stealing by
651 coherent_pool=nn[KMG] [ARM,KNL]
652 Sets the size of memory pool for coherent, atomic dma
653 allocations, by default set to 256K.
655 com20020= [HW,NET] ARCnet - COM20020 chipset
657 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
659 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
663 ARCnet - COM90xx chipset (memory-mapped buffers)
664 Format: <io>[,<irq>[,<memstart>]]
666 condev= [HW,S390] console device
669 console= [KNL] Output console device and options.
671 tty<n> Use the virtual console device <n>.
675 Use the specified serial port. The options are of
676 the form "bbbbpnf", where "bbbb" is the baud rate,
677 "p" is parity ("n", "o", or "e"), "n" is number of
678 bits, and "f" is flow control ("r" for RTS or
679 omit it). Default is "9600n8".
681 See Documentation/admin-guide/serial-console.rst for more
683 Documentation/networking/netconsole.rst for an
686 uart[8250],io,<addr>[,options]
687 uart[8250],mmio,<addr>[,options]
688 uart[8250],mmio16,<addr>[,options]
689 uart[8250],mmio32,<addr>[,options]
690 uart[8250],0x<addr>[,options]
691 Start an early, polled-mode console on the 8250/16550
692 UART at the specified I/O port or MMIO address,
693 switching to the matching ttyS device later.
694 MMIO inter-register address stride is either 8-bit
695 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
696 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
697 to be equivalent to 'mmio'. 'options' are specified in
698 the same format described for ttyS above; if unspecified,
699 the h/w is not re-initialized.
701 hvc<n> Use the hypervisor console device <n>. This is for
702 both Xen and PowerPC hypervisors.
704 If the device connected to the port is not a TTY but a braille
705 device, prepend "brl," before the device type, for instance
707 For now, only VisioBraille is supported.
710 [KNL] Change console messages format
712 By default we print messages on consoles in
713 "[time stamp] text\n" format (time stamp may not be
714 printed, depending on CONFIG_PRINTK_TIME or
715 `printk_time' param).
717 Switch to syslog format: "<%u>[time stamp] text\n"
718 IOW, each message will have a facility and loglevel
719 prefix. The format is similar to one used by syslog()
720 syscall, or to executing "dmesg -S --raw" or to reading
723 consoleblank= [KNL] The console blank (screen saver) timeout in
724 seconds. A value of 0 disables the blank timer.
728 [KNL] Change the default value for
729 /proc/<pid>/coredump_filter.
730 See also Documentation/filesystems/proc.rst.
732 coresight_cpu_debug.enable
735 Enable/disable the CPU sampling based debugging.
736 0: default value, disable debugging
737 1: enable debugging at boot time
739 cpuidle.off=1 [CPU_IDLE]
740 disable the cpuidle sub-system
743 [CPU_IDLE] Name of the cpuidle governor to use.
745 cpufreq.off=1 [CPU_FREQ]
746 disable the cpufreq sub-system
748 cpufreq.default_governor=
749 [CPU_FREQ] Name of the default cpufreq governor or
750 policy to use. This governor must be registered in the
751 kernel before the cpufreq driver probes.
754 [X86] Delay for N microsec between assert and de-assert
755 of APIC INIT to start processors. This delay occurs
756 on every CPU online, such as boot, and resume from suspend.
759 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
761 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
763 crashkernel=size[KMG][@offset[KMG]]
764 [KNL] Using kexec, Linux can switch to a 'crash kernel'
765 upon panic. This parameter reserves the physical
766 memory region [offset, offset + size] for that kernel
767 image. If '@offset' is omitted, then a suitable offset
768 is selected automatically.
769 [KNL, X86-64] Select a region under 4G first, and
770 fall back to reserve region above 4G when '@offset'
771 hasn't been specified.
772 See Documentation/admin-guide/kdump/kdump.rst for further details.
774 crashkernel=range1:size1[,range2:size2,...][@offset]
775 [KNL] Same as above, but depends on the memory
776 in the running system. The syntax of range is
777 start-[end] where start and end are both
778 a memory unit (amount[KMG]). See also
779 Documentation/admin-guide/kdump/kdump.rst for an example.
781 crashkernel=size[KMG],high
782 [KNL, X86-64] range could be above 4G. Allow kernel
783 to allocate physical memory region from top, so could
784 be above 4G if system have more than 4G ram installed.
785 Otherwise memory region will be allocated below 4G, if
787 It will be ignored if crashkernel=X is specified.
788 crashkernel=size[KMG],low
789 [KNL, X86-64] range under 4G. When crashkernel=X,high
790 is passed, kernel could allocate physical memory region
791 above 4G, that cause second kernel crash on system
792 that require some amount of low memory, e.g. swiotlb
793 requires at least 64M+32K low memory, also enough extra
794 low memory is needed to make sure DMA buffers for 32-bit
795 devices won't run out. Kernel would try to allocate at
796 at least 256M below 4G automatically.
797 This one let user to specify own low range under 4G
798 for second kernel instead.
799 0: to disable low allocation.
800 It will be ignored when crashkernel=X,high is not used
801 or memory reserved is below 4G.
804 [KNL] Disable crypto self-tests
809 cs89x0_media= [HW,NET]
810 Format: { rj45 | aui | bnc }
812 csdlock_debug= [KNL] Enable debug add-ons of cross-CPU function call
813 handling. When switched on, additional debug data is
814 printed to the console in case a hanging CPU is
815 detected, and that CPU is pinged again in order to try
816 to resolve the hang situation.
817 0: disable csdlock debugging (default)
818 1: enable basic csdlock debugging (minor impact)
819 ext: enable extended csdlock debugging (more impact,
823 See header of drivers/s390/block/dasd_devmap.c.
825 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
826 (one device per port)
827 Format: <port#>,<type>
828 See also Documentation/input/devices/joystick-parport.rst
830 ddebug_query= [KNL,DYNAMIC_DEBUG] Enable debug messages at early boot
832 Documentation/admin-guide/dynamic-debug-howto.rst for
833 details. Deprecated, see dyndbg.
835 debug [KNL] Enable kernel debugging (events log level).
838 [KNL] Enable printing [hashed] pointers early in the
839 boot sequence. If enabled, we use a weak hash instead
840 of siphash to hash pointers. Use this option if you are
841 seeing instances of '(___ptrval___)') and need to see a
842 value (hashed pointer) instead. Cryptographically
843 insecure, please do not use on production kernels.
846 [KNL] verbose locking self-tests
848 Print debugging info while doing the locking API
850 Bitmask for the various LOCKTYPE_ tests. Defaults to 0
851 (no extra messages), setting it to -1 (all bits set)
852 will print _a_lot_ more information - normally only
853 useful to lockdep developers.
855 debug_objects [KNL] Enable object debugging
858 [KNL] Disable object debugging
860 debug_guardpage_minorder=
861 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
862 parameter allows control of the order of pages that will
863 be intentionally kept free (and hence protected) by the
864 buddy allocator. Bigger value increase the probability
865 of catching random memory corruption, but reduce the
866 amount of memory for normal system use. The maximum
867 possible value is MAX_ORDER/2. Setting this parameter
868 to 1 or 2 should be enough to identify most random
869 memory corruption problems caused by bugs in kernel or
870 driver code when a CPU writes to (or reads from) a
871 random memory location. Note that there exists a class
872 of memory corruptions problems caused by buggy H/W or
873 F/W or by drivers badly programing DMA (basically when
874 memory is written at bus level and the CPU MMU is
875 bypassed) which are not detectable by
876 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
877 tracking down these problems.
880 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
881 enables the feature at boot time. By default, it is
882 disabled and the system will work mostly the same as a
883 kernel built without CONFIG_DEBUG_PAGEALLOC.
884 Note: to get most of debug_pagealloc error reports, it's
885 useful to also enable the page_owner functionality.
886 on: enable the feature
888 debugfs= [KNL] This parameter enables what is exposed to userspace
889 and debugfs internal clients.
890 Format: { on, no-mount, off }
891 on: All functions are enabled.
893 Filesystem is not registered but kernel clients can
894 access APIs and a crashkernel can be used to read
895 its content. There is nothing to mount.
896 off: Filesystem is not registered and clients
897 get a -EPERM as result when trying to register files
898 or directories within debugfs.
899 This is equivalent of the runtime functionality if
900 debugfs was not enabled in the kernel at all.
901 Default value is set in build-time with a kernel configuration.
903 debugpat [X86] Enable PAT debugging
905 decnet.addr= [HW,NET]
906 Format: <area>[,<node>]
907 See also Documentation/networking/decnet.rst.
910 [HW] The size of the default HugeTLB page. This is
911 the size represented by the legacy /proc/ hugepages
912 APIs. In addition, this is the default hugetlb size
913 used for shmget(), mmap() and mounting hugetlbfs
914 filesystems. If not specified, defaults to the
915 architecture's default huge page size. Huge page
916 sizes are architecture dependent. See also
917 Documentation/admin-guide/mm/hugetlbpage.rst.
920 deferred_probe_timeout=
921 [KNL] Debugging option to set a timeout in seconds for
922 deferred probe to give up waiting on dependencies to
923 probe. Only specific dependencies (subsystems or
924 drivers) that have opted in will be ignored. A timeout of 0
925 will timeout at the end of initcalls. This option will also
926 dump out devices still on the deferred probe list after
930 Format: { on | off | def_only | inf_only | always }
931 on: s390 zlib hardware support for compression on
932 level 1 and decompression (default)
933 off: No s390 zlib hardware support
934 def_only: s390 zlib hardware support for deflate
935 only (compression on level 1)
936 inf_only: s390 zlib hardware support for inflate
938 always: Same as 'on' but ignores the selected compression
939 level always using hardware support (used for debugging)
942 Set number of hash buckets for dentry cache.
944 disable_1tb_segments [PPC]
945 Disables the use of 1TB hash page table segments. This
946 causes the kernel to fall back to 256MB segments which
947 can be useful when debugging issues that require an SLB
951 Limits the number of kernel SLB entries, and flushes
952 them frequently to increase the rate of SLB faults
956 See Documentation/networking/ipv6.rst.
959 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
960 hardening is enabled for this boot. Hardened
961 usercopy checking is used to protect the kernel
962 from reading or writing beyond known memory
963 allocation boundaries as a proactive defense
964 against bounds-checking flaws in the kernel's
965 copy_to_user()/copy_from_user() interface.
966 on Perform hardened usercopy checks (default).
967 off Disable hardened usercopy checks.
970 Disable RADIX MMU mode on POWER9
972 radix_hcall_invalidate=on [PPC/PSERIES]
973 Disable RADIX GTSE feature and use hcall for TLB
977 Disable TLBIE instruction. Currently does not work
978 with KVM, with HASH MMU, or with coherent accelerators.
980 disable_cpu_apicid= [X86,APIC,SMP]
982 The number of initial APIC ID for the
983 corresponding CPU to be disabled at boot,
984 mostly used for the kdump 2nd kernel to
985 disable BSP to wake up multiple CPUs without
986 causing system reset or hang due to sending
989 disable_ddw [PPC/PSERIES]
990 Disable Dynamic DMA Window support. Use this
991 to workaround buggy firmware.
994 See Documentation/networking/ipv6.rst.
996 disable_mtrr_cleanup [X86]
997 The kernel tries to adjust MTRR layout from continuous
998 to discrete, to make X server driver able to add WB
999 entry later. This parameter disables that.
1001 disable_mtrr_trim [X86, Intel and AMD only]
1002 By default the kernel will trim any uncacheable
1003 memory out of your available memory pool based on
1004 MTRR settings. This parameter disables that behavior,
1005 possibly causing your machine to run very slowly.
1007 disable_timer_pin_1 [X86]
1008 Disable PIN 1 of APIC timer
1009 Can be useful to work around chipset bugs.
1011 dis_ucode_ldr [X86] Disable the microcode loader.
1013 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
1014 this option disables the debugging code at boot.
1016 dma_debug_entries=<number>
1017 This option allows to tune the number of preallocated
1018 entries for DMA-API debugging code. One entry is
1019 required per DMA-API allocation. Use this if the
1020 DMA-API debugging code disables itself because the
1021 architectural default is too low.
1023 dma_debug_driver=<driver_name>
1024 With this option the DMA-API debugging driver
1025 filter feature can be enabled at boot time. Just
1026 pass the driver to filter for as the parameter.
1027 The filter can be disabled or changed to another
1028 driver later using sysfs.
1030 driver_async_probe= [KNL]
1031 List of driver names to be probed asynchronously.
1032 Format: <driver_name1>,<driver_name2>...
1034 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
1035 Broken monitors, graphic adapters, KVMs and EDIDless
1036 panels may send no or incorrect EDID data sets.
1037 This parameter allows to specify an EDID data sets
1038 in the /lib/firmware directory that are used instead.
1039 Generic built-in EDID data sets are used, if one of
1040 edid/1024x768.bin, edid/1280x1024.bin,
1041 edid/1680x1050.bin, or edid/1920x1080.bin is given
1042 and no file with the same name exists. Details and
1043 instructions how to build your own EDID data are
1044 available in Documentation/admin-guide/edid.rst. An EDID
1045 data set will only be used for a particular connector,
1046 if its name and a colon are prepended to the EDID
1047 name. Each connector may use a unique EDID data
1048 set by separating the files with a comma. An EDID
1049 data set with no connector name will be used for
1050 any connectors not explicitly specified.
1055 Format: {"off" | "known"}
1056 Control how the dt_cpu_ftrs device-tree binding is
1057 used for CPU feature discovery and setup (if it
1059 off: Do not use it, fall back to legacy cpu table.
1060 known: Do not pass through unknown features to guests
1061 or userspace, only those that the kernel is aware of.
1063 dump_apple_properties [X86]
1064 Dump name and content of EFI device properties on
1065 x86 Macs. Useful for driver authors to determine
1066 what data is available or for reverse-engineering.
1068 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
1069 <module>.dyndbg[="val"]
1070 Enable debug messages at boot time. See
1071 Documentation/admin-guide/dynamic-debug-howto.rst
1074 nopku [X86] Disable Memory Protection Keys CPU feature found
1077 <module>.async_probe [KNL]
1078 Enable asynchronous probe on this module.
1080 early_ioremap_debug [KNL]
1081 Enable debug messages in early_ioremap support. This
1082 is useful for tracking down temporary early mappings
1083 which are not unmapped.
1085 earlycon= [KNL] Output early console device and options.
1087 When used with no options, the early console is
1088 determined by stdout-path property in device tree's
1089 chosen node or the ACPI SPCR table if supported by
1092 cdns,<addr>[,options]
1093 Start an early, polled-mode console on a Cadence
1094 (xuartps) serial port at the specified address. Only
1095 supported option is baud rate. If baud rate is not
1096 specified, the serial port must already be setup and
1099 uart[8250],io,<addr>[,options]
1100 uart[8250],mmio,<addr>[,options]
1101 uart[8250],mmio32,<addr>[,options]
1102 uart[8250],mmio32be,<addr>[,options]
1103 uart[8250],0x<addr>[,options]
1104 Start an early, polled-mode console on the 8250/16550
1105 UART at the specified I/O port or MMIO address.
1106 MMIO inter-register address stride is either 8-bit
1107 (mmio) or 32-bit (mmio32 or mmio32be).
1108 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1109 to be equivalent to 'mmio'. 'options' are specified
1110 in the same format described for "console=ttyS<n>"; if
1111 unspecified, the h/w is not initialized.
1115 Start an early, polled-mode console on a pl011 serial
1116 port at the specified address. The pl011 serial port
1117 must already be setup and configured. Options are not
1118 yet supported. If 'mmio32' is specified, then only
1119 the driver will use only 32-bit accessors to read/write
1120 the device registers.
1123 Start an early, polled-mode console on a meson serial
1124 port at the specified address. The serial port must
1125 already be setup and configured. Options are not yet
1129 Start an early, polled-mode console on an msm serial
1130 port at the specified address. The serial port
1131 must already be setup and configured. Options are not
1134 msm_serial_dm,<addr>
1135 Start an early, polled-mode console on an msm serial
1136 dm port at the specified address. The serial port
1137 must already be setup and configured. Options are not
1141 Start an early, polled-mode console on a serial port
1142 of an Actions Semi SoC, such as S500 or S900, at the
1143 specified address. The serial port must already be
1144 setup and configured. Options are not yet supported.
1147 Start an early, polled-mode console on a serial port
1148 of an RDA Micro SoC, such as RDA8810PL, at the
1149 specified address. The serial port must already be
1150 setup and configured. Options are not yet supported.
1153 Use RISC-V SBI (Supervisor Binary Interface) for early
1156 smh Use ARM semihosting calls for early console.
1164 Use early console provided by serial driver available
1165 on Samsung SoCs, requires selecting proper type and
1166 a correct base address of the selected UART port. The
1167 serial port must already be setup and configured.
1168 Options are not yet supported.
1171 Start an early, polled-mode console on a lantiq serial
1172 (lqasc) port at the specified address. The serial port
1173 must already be setup and configured. Options are not
1178 Use early console provided by Freescale LP UART driver
1179 found on Freescale Vybrid and QorIQ LS1021A processors.
1180 A valid base address must be provided, and the serial
1181 port must already be setup and configured.
1185 Start an early, polled-mode, output-only console on the
1186 Freescale i.MX UART at the specified address. The UART
1187 must already be setup and configured.
1190 Start an early, polled-mode console on the
1191 Armada 3700 serial port at the specified
1192 address. The serial port must already be setup
1193 and configured. Options are not yet supported.
1196 Start an early, polled-mode console on a Qualcomm
1197 Generic Interface (GENI) based serial port at the
1198 specified address. The serial port must already be
1199 setup and configured. Options are not yet supported.
1202 Start an early, unaccelerated console on the EFI
1203 memory mapped framebuffer (if available). On cache
1204 coherent non-x86 systems that use system memory for
1205 the framebuffer, pass the 'ram' option so that it is
1206 mapped with the correct attributes.
1209 Use early console provided by Freescale LINFlexD UART
1210 serial driver for NXP S32V234 SoCs. A valid base
1211 address must be provided, and the serial port must
1212 already be setup and configured.
1214 earlyprintk= [X86,SH,ARM,M68k,S390]
1218 earlyprintk=serial[,ttySn[,baudrate]]
1219 earlyprintk=serial[,0x...[,baudrate]]
1220 earlyprintk=ttySn[,baudrate]
1221 earlyprintk=dbgp[debugController#]
1222 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1223 earlyprintk=xdbc[xhciController#]
1225 earlyprintk is useful when the kernel crashes before
1226 the normal console is initialized. It is not enabled by
1227 default because it has some cosmetic problems.
1229 Append ",keep" to not disable it when the real console
1232 Only one of vga, efi, serial, or usb debug port can
1235 Currently only ttyS0 and ttyS1 may be specified by
1236 name. Other I/O ports may be explicitly specified
1237 on some architectures (x86 and arm at least) by
1238 replacing ttySn with an I/O port address, like this:
1239 earlyprintk=serial,0x1008,115200
1240 You can find the port for a given device in
1241 /proc/tty/driver/serial:
1242 2: uart:ST16650V2 port:00001008 irq:18 ...
1244 Interaction with the standard serial driver is not
1247 The VGA and EFI output is eventually overwritten by
1250 The xen output can only be used by Xen PV guests.
1252 The sclp output can only be used on s390.
1254 The optional "force" to "pciserial" enables use of a
1255 PCI device even when its classcode is not of the
1258 edac_report= [HW,EDAC] Control how to report EDAC event
1259 Format: {"on" | "off" | "force"}
1260 on: enable EDAC to report H/W event. May be overridden
1261 by other higher priority error reporting module.
1262 off: disable H/W event reporting through EDAC.
1263 force: enforce the use of EDAC to report H/W event.
1266 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1269 This is designed to be used in conjunction with
1270 the boot argument: earlyprintk=vga
1272 This parameter works in place of the kgdboc parameter
1273 but can only be used if the backing tty is available
1274 very early in the boot process. For early debugging
1275 via a serial port see kgdboc_earlycon instead.
1278 Format: {"off" | "on" | "skip[mbr]"}
1281 Format: { "debug", "disable_early_pci_dma",
1282 "nochunk", "noruntime", "nosoftreserve",
1283 "novamap", "no_disable_early_pci_dma" }
1284 debug: enable misc debug output.
1285 disable_early_pci_dma: disable the busmaster bit on all
1286 PCI bridges while in the EFI boot stub.
1287 nochunk: disable reading files in "chunks" in the EFI
1288 boot stub, as chunking can cause problems with some
1289 firmware implementations.
1290 noruntime : disable EFI runtime services support
1291 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1292 attribute may cause the kernel to reserve the
1293 memory range for a memory mapping driver to
1294 claim. Specify efi=nosoftreserve to disable this
1295 reservation and treat the memory by its base type
1296 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1297 novamap: do not call SetVirtualAddressMap().
1298 no_disable_early_pci_dma: Leave the busmaster bit set
1299 on all PCI bridges while in the EFI boot stub
1301 efi_no_storage_paranoia [EFI; X86]
1302 Using this parameter you can use more than 50% of
1303 your efi variable storage. Use this parameter only if
1304 you are really sure that your UEFI does sane gc and
1305 fulfills the spec otherwise your board may brick.
1307 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1308 Add arbitrary attribute to specific memory range by
1309 updating original EFI memory map.
1310 Region of memory which aa attribute is added to is
1313 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1314 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1315 attribute is added to range 0x100000000-0x180000000 and
1316 0x10a0000000-0x1120000000.
1318 If efi_fake_mem=8G@9G:0x40000 is specified, the
1319 EFI_MEMORY_SP(0x40000) attribute is added to
1320 range 0x240000000-0x43fffffff.
1322 Using this parameter you can do debugging of EFI memmap
1323 related features. For example, you can do debugging of
1324 Address Range Mirroring feature even if your box
1325 doesn't support it, or mark specific memory as
1328 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1329 that is to be dynamically loaded by Linux. If there are
1330 multiple variables with the same name but with different
1331 vendor GUIDs, all of them will be loaded. See
1332 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1335 eisa_irq_edge= [PARISC,HW]
1336 See header of drivers/parisc/eisa.c.
1339 See comment before function elanfreq_setup() in
1340 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1342 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1343 Specifies physical address of start of kernel core
1344 image elf header and optionally the size. Generally
1345 kexec loader will pass this option to capture kernel.
1346 See Documentation/admin-guide/kdump/kdump.rst for details.
1348 enable_mtrr_cleanup [X86]
1349 The kernel tries to adjust MTRR layout from continuous
1350 to discrete, to make X server driver able to add WB
1351 entry later. This parameter enables that.
1353 enable_timer_pin_1 [X86]
1354 Enable PIN 1 of APIC timer
1355 Can be useful to work around chipset bugs
1356 (in particular on some ATI chipsets).
1357 The kernel tries to set a reasonable default.
1359 enforcing [SELINUX] Set initial enforcing status.
1361 See security/selinux/Kconfig help text.
1362 0 -- permissive (log only, no denials).
1363 1 -- enforcing (deny and log).
1365 Value can be changed at runtime via
1366 /sys/fs/selinux/enforce.
1369 Disable Error Record Serialization Table (ERST)
1372 ether= [HW,NET] Ethernet cards parameters
1373 This option is obsoleted by the "netdev=" option, which
1374 has equivalent usage. See its documentation for details.
1378 Permit 'security.evm' to be updated regardless of
1379 current integrity status.
1384 fail_make_request=[KNL]
1385 General fault injection mechanism.
1386 Format: <interval>,<probability>,<space>,<times>
1387 See also Documentation/fault-injection/.
1390 Format: { initns | none }
1391 See Documentation/admin-guide/sysctl/net.rst for
1392 fb_tunnels_only_for_init_ns
1395 See Documentation/admin-guide/blockdev/floppy.rst.
1397 force_pal_cache_flush
1398 [IA-64] Avoid check_sal_cache_flush which may hang on
1399 buggy SAL_CACHE_FLUSH implementations. Using this
1400 parameter will force ia64_sal_cache_flush to call
1401 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1404 Forcefully enable Physical Address Extension (PAE).
1405 Many Pentium M systems disable PAE but may have a
1406 functionally usable PAE implementation.
1407 Warning: use of this parameter will taint the kernel
1408 and may cause unknown problems.
1411 [FTRACE] will set and start the specified tracer
1412 as early as possible in order to facilitate early
1415 ftrace_dump_on_oops[=orig_cpu]
1416 [FTRACE] will dump the trace buffers on oops.
1417 If no parameter is passed, ftrace will dump
1418 buffers of all CPUs, but if you pass orig_cpu, it will
1419 dump only the buffer of the CPU that triggered the
1422 ftrace_filter=[function-list]
1423 [FTRACE] Limit the functions traced by the function
1424 tracer at boot up. function-list is a comma-separated
1425 list of functions. This list can be changed at run
1426 time by the set_ftrace_filter file in the debugfs
1429 ftrace_notrace=[function-list]
1430 [FTRACE] Do not trace the functions specified in
1431 function-list. This list can be changed at run time
1432 by the set_ftrace_notrace file in the debugfs
1435 ftrace_graph_filter=[function-list]
1436 [FTRACE] Limit the top level callers functions traced
1437 by the function graph tracer at boot up.
1438 function-list is a comma-separated list of functions
1439 that can be changed at run time by the
1440 set_graph_function file in the debugfs tracing directory.
1442 ftrace_graph_notrace=[function-list]
1443 [FTRACE] Do not trace from the functions specified in
1444 function-list. This list is a comma-separated list of
1445 functions that can be changed at run time by the
1446 set_graph_notrace file in the debugfs tracing directory.
1448 ftrace_graph_max_depth=<uint>
1449 [FTRACE] Used with the function graph tracer. This is
1450 the max depth it will trace into a function. This value
1451 can be changed at run time by the max_graph_depth file
1452 in the tracefs tracing directory. default: 0 (no limit)
1454 fw_devlink= [KNL] Create device links between consumer and supplier
1455 devices by scanning the firmware to infer the
1456 consumer/supplier relationships. This feature is
1457 especially useful when drivers are loaded as modules as
1458 it ensures proper ordering of tasks like device probing
1459 (suppliers first, then consumers), supplier boot state
1460 clean up (only after all consumers have probed),
1461 suspend/resume & runtime PM (consumers first, then
1463 Format: { off | permissive | on | rpm }
1464 off -- Don't create device links from firmware info.
1465 permissive -- Create device links from firmware info
1466 but use it only for ordering boot state clean
1467 up (sync_state() calls).
1468 on -- Create device links from firmware info and use it
1469 to enforce probe and suspend/resume ordering.
1470 rpm -- Like "on", but also use to order runtime PM.
1472 fw_devlink.strict=<bool>
1473 [KNL] Treat all inferred dependencies as mandatory
1474 dependencies. This only applies for fw_devlink=on|rpm.
1478 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1479 support via parallel port (up to 5 devices per port)
1480 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1481 See also Documentation/input/devices/joystick-parport.rst
1485 gart_fix_e820= [X86-64] disable the fix e820 for K8 GART
1489 gcov_persist= [GCOV] When non-zero (default), profiling data for
1490 kernel modules is saved and remains accessible via
1491 debugfs, even when the module is unloaded/reloaded.
1492 When zero, profiling data is discarded and associated
1493 debugfs files are removed at module unload time.
1495 goldfish [X86] Enable the goldfish android emulator platform.
1496 Don't use this when you are not running on the
1499 gpio-mockup.gpio_mockup_ranges
1500 [HW] Sets the ranges of gpiochip of for this device.
1501 Format: <start1>,<end1>,<start2>,<end2>...
1502 gpio-mockup.gpio_mockup_named_lines
1503 [HW] Let the driver know GPIO lines should be named.
1505 gpt [EFI] Forces disk with valid GPT signature but
1506 invalid Protective MBR to be treated as GPT. If the
1507 primary GPT is corrupted, it enables the backup/alternate
1508 GPT to be used instead.
1510 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1511 the "Enable 0" bit of the configuration register.
1514 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1515 the "Enable 0" bit of the configuration register.
1518 grcan.select= [HW] Select which physical interface to use.
1521 grcan.txsize= [HW] Sets the size of the tx buffer.
1522 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1524 grcan.rxsize= [HW] Sets the size of the rx buffer.
1525 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1528 hardlockup_all_cpu_backtrace=
1529 [KNL] Should the hard-lockup detector generate
1530 backtraces on all cpus.
1533 hashdist= [KNL,NUMA] Large hashes allocated during boot
1534 are distributed across NUMA nodes. Defaults on
1535 for 64-bit NUMA, off otherwise.
1536 Format: 0 | 1 (for off | on)
1538 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1540 hd= [EIDE] (E)IDE hard drive subsystem geometry
1541 Format: <cyl>,<head>,<sect>
1544 Disable Hardware Error Source Table (HEST) support;
1545 corresponding firmware-first mode error processing
1546 logic will be disabled.
1548 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1549 size of <nn>. This works even on boxes that have no
1550 highmem otherwise. This also works to reduce highmem
1551 size on bigger boxes.
1553 highres= [KNL] Enable/disable high resolution timer mode.
1554 Valid parameters: "on", "off"
1559 hpet= [X86-32,HPET] option to control HPET usage
1560 Format: { enable (default) | disable | force |
1562 disable: disable HPET and use PIT instead
1563 force: allow force enabled of undocumented chips (ICH4,
1565 verbose: show contents of HPET registers during setup
1567 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1568 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1570 hugetlb_cma= [HW,CMA] The size of a CMA area used for allocation
1571 of gigantic hugepages.
1574 Reserve a CMA area of given size and allocate gigantic
1575 hugepages using the CMA allocator. If enabled, the
1576 boot-time allocation of gigantic hugepages is skipped.
1578 hugepages= [HW] Number of HugeTLB pages to allocate at boot.
1579 If this follows hugepagesz (below), it specifies
1580 the number of pages of hugepagesz to be allocated.
1581 If this is the first HugeTLB parameter on the command
1582 line, it specifies the number of pages to allocate for
1583 the default huge page size. See also
1584 Documentation/admin-guide/mm/hugetlbpage.rst.
1588 [HW] The size of the HugeTLB pages. This is used in
1589 conjunction with hugepages (above) to allocate huge
1590 pages of a specific size at boot. The pair
1591 hugepagesz=X hugepages=Y can be specified once for
1592 each supported huge page size. Huge page sizes are
1593 architecture dependent. See also
1594 Documentation/admin-guide/mm/hugetlbpage.rst.
1598 [KNL] Should the hung task detector generate panics.
1601 A value of 1 instructs the kernel to panic when a
1602 hung task is detected. The default value is controlled
1603 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1604 option. The value selected by this boot parameter can
1605 be changed later by the kernel.hung_task_panic sysctl.
1607 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1608 terminal devices. Valid values: 0..8
1609 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1610 If specified, z/VM IUCV HVC accepts connections
1611 from listed z/VM user IDs only.
1613 hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations
1614 which allow the hypervisor to 'idle' the
1615 guest on lock contention.
1618 Do not unregister boot console at start. This is only
1619 useful for debugging when something happens in the window
1620 between unregistering the boot console and initializing
1623 i2c_bus= [HW] Override the default board specific I2C bus speed
1624 or register an additional I2C bus that is not
1625 registered from board initialization code.
1629 i8042.debug [HW] Toggle i8042 debug mode
1630 i8042.unmask_kbd_data
1631 [HW] Enable printing of interrupt data from the KBD port
1632 (disabled by default, and as a pre-condition
1633 requires that i8042.debug=1 be enabled)
1634 i8042.direct [HW] Put keyboard port into non-translated mode
1635 i8042.dumbkbd [HW] Pretend that controller can only read data from
1636 keyboard and cannot control its state
1637 (Don't attempt to blink the leds)
1638 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1639 i8042.nokbd [HW] Don't check/create keyboard port
1640 i8042.noloop [HW] Disable the AUX Loopback command while probing
1642 i8042.nomux [HW] Don't check presence of an active multiplexing
1644 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1646 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1647 i8042.reset [HW] Reset the controller during init, cleanup and
1648 suspend-to-ram transitions, only during s2r
1649 transitions, or never reset
1650 Format: { 1 | Y | y | 0 | N | n }
1651 1, Y, y: always reset controller
1652 0, N, n: don't ever reset controller
1653 Default: only on s2r transitions on x86; most other
1654 architectures force reset to be always executed
1655 i8042.unlock [HW] Unlock (ignore) the keylock
1656 i8042.kbdreset [HW] Reset device connected to KBD port
1660 i8k.ignore_dmi [HW] Continue probing hardware even if DMI data
1661 indicates that the driver is running on unsupported
1663 i8k.force [HW] Activate i8k driver even if SMM BIOS signature
1664 does not match list of supported models.
1666 [HW] Report power status in /proc/i8k
1667 (disabled by default)
1668 i8k.restricted [HW] Allow controlling fans only if SYS_ADMIN
1671 i915.invert_brightness=
1672 [DRM] Invert the sense of the variable that is used to
1673 set the brightness of the panel backlight. Normally a
1674 brightness value of 0 indicates backlight switched off,
1675 and the maximum of the brightness value sets the backlight
1676 to maximum brightness. If this parameter is set to 0
1677 (default) and the machine requires it, or this parameter
1678 is set to 1, a brightness value of 0 sets the backlight
1679 to maximum brightness, and the maximum of the brightness
1680 value switches the backlight off.
1681 -1 -- never invert brightness
1682 0 -- machine default
1683 1 -- force brightness inversion
1686 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1688 ide-core.nodma= [HW] (E)IDE subsystem
1689 Format: =0.0 to prevent dma on hda, =0.1 hdb =1.0 hdc
1690 .vlb_clock .pci_clock .noflush .nohpa .noprobe .nowerr
1691 .cdrom .chs .ignore_cable are additional options
1692 See Documentation/ide/ide.rst.
1694 ide-generic.probe-mask= [HW] (E)IDE subsystem
1696 Probe mask for legacy ISA IDE ports. Depending on
1697 platform up to 6 ports are supported, enabled by
1698 setting corresponding bits in the mask to 1. The
1699 default value is 0x0, which has a special meaning.
1700 On systems that have PCI, it triggers scanning the
1701 PCI bus for the first and the second port, which
1702 are then probed. On systems without PCI the value
1703 of 0x0 enables probing the two first ports as if it
1706 ide-pci-generic.all-generic-ide [HW] (E)IDE subsystem
1707 Claim all unknown PCI IDE storage controllers.
1710 Format: idle=poll, idle=halt, idle=nomwait
1711 Poll forces a polling idle loop that can slightly
1712 improve the performance of waking up a idle CPU, but
1713 will use a lot of power and make the system run hot.
1715 idle=halt: Halt is forced to be used for CPU idle.
1716 In such case C2/C3 won't be used again.
1717 idle=nomwait: Disable mwait for CPU C-states
1721 Allow force disabling of Shared Virtual Memory (SVA)
1722 support for the idxd driver. By default it is set to
1725 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1726 Format: { strict | legacy | 2008 | relaxed }
1729 Choose which programs will be accepted for execution
1730 based on the IEEE 754 NaN encoding(s) supported by
1731 the FPU and the NaN encoding requested with the value
1732 of an ELF file header flag individually set by each
1733 binary. Hardware implementations are permitted to
1734 support either or both of the legacy and the 2008 NaN
1737 Available settings are as follows:
1738 strict accept binaries that request a NaN encoding
1739 supported by the FPU
1740 legacy only accept legacy-NaN binaries, if supported
1742 2008 only accept 2008-NaN binaries, if supported
1744 relaxed accept any binaries regardless of whether
1745 supported by the FPU
1747 The FPU emulator is always able to support both NaN
1748 encodings, so if no FPU hardware is present or it has
1749 been disabled with 'nofpu', then the settings of
1750 'legacy' and '2008' strap the emulator accordingly,
1751 'relaxed' straps the emulator for both legacy-NaN and
1752 2008-NaN, whereas 'strict' enables legacy-NaN only on
1753 legacy processors and both NaN encodings on MIPS32 or
1756 The setting for ABS.fmt/NEG.fmt instruction execution
1757 mode generally follows that for the NaN encoding,
1758 except where unsupported by hardware.
1760 ignore_loglevel [KNL]
1761 Ignore loglevel setting - this will print /all/
1762 kernel messages to the console. Useful for debugging.
1763 We also add it as printk module parameter, so users
1764 could change it dynamically, usually by
1765 /sys/module/printk/parameters/ignore_loglevel.
1768 Ignore RLIMIT_DATA setting for data mappings,
1769 print warning at first misuse. Can be changed via
1770 /sys/module/kernel/parameters/ignore_rlimit_data.
1772 ihash_entries= [KNL]
1773 Set number of hash buckets for inode cache.
1775 ima_appraise= [IMA] appraise integrity measurements
1776 Format: { "off" | "enforce" | "fix" | "log" }
1779 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
1780 The builtin appraise policy appraises all files
1783 ima_canonical_fmt [IMA]
1784 Use the canonical format for the binary runtime
1785 measurements, instead of host native format.
1788 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1792 The list of supported hash algorithms is defined
1793 in crypto/hash_info.h.
1796 The builtin policies to load during IMA setup.
1797 Format: "tcb | appraise_tcb | secure_boot |
1798 fail_securely | critical_data"
1800 The "tcb" policy measures all programs exec'd, files
1801 mmap'd for exec, and all files opened with the read
1802 mode bit set by either the effective uid (euid=0) or
1805 The "appraise_tcb" policy appraises the integrity of
1806 all files owned by root.
1808 The "secure_boot" policy appraises the integrity
1809 of files (eg. kexec kernel image, kernel modules,
1810 firmware, policy, etc) based on file signatures.
1812 The "fail_securely" policy forces file signature
1813 verification failure also on privileged mounted
1814 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
1817 The "critical_data" policy measures kernel integrity
1820 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
1821 Load a policy which meets the needs of the Trusted
1822 Computing Base. This means IMA will measure all
1823 programs exec'd, files mmap'd for exec, and all files
1824 opened for read by uid=0.
1827 Select one of defined IMA measurements template formats.
1828 Formats: { "ima" | "ima-ng" | "ima-sig" }
1832 [IMA] Define a custom template format.
1833 Format: { "field1|...|fieldN" }
1835 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
1836 Format: <min_file_size>
1837 Set the minimal file size for using asynchronous hash.
1838 If left unspecified, ahash usage is disabled.
1840 ahash performance varies for different data sizes on
1841 different crypto accelerators. This option can be used
1842 to achieve the best performance for a particular HW.
1844 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
1846 Set hashing buffer size. Default: 4k.
1848 ahash performance varies for different chunk sizes on
1849 different crypto accelerators. This option can be used
1850 to achieve best performance for particular HW.
1854 Run specified binary instead of /sbin/init as init
1857 initcall_debug [KNL] Trace initcalls as they are executed. Useful
1858 for working out where the kernel is dying during
1861 initcall_blacklist= [KNL] Do not execute a comma-separated list of
1862 initcall functions. Useful for debugging built-in
1863 modules and initcalls.
1865 initramfs_async= [KNL]
1868 This parameter controls whether the initramfs
1869 image is unpacked asynchronously, concurrently
1870 with devices being probed and
1871 initialized. This should normally just work,
1872 but as a debugging aid, one can get the
1873 historical behaviour of the initramfs
1874 unpacking being completed before device_ and
1877 initrd= [BOOT] Specify the location of the initial ramdisk
1879 initrdmem= [KNL] Specify a physical address and size from which to
1880 load the initrd. If an initrd is compiled in or
1881 specified in the bootparams, it takes priority over this
1883 Format: ss[KMG],nn[KMG]
1886 init_on_alloc= [MM] Fill newly allocated pages and heap objects with
1889 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
1891 init_on_free= [MM] Fill freed pages and heap objects with zeroes.
1893 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
1895 init_pkru= [X86] Specify the default memory protection keys rights
1896 register contents for all processes. 0x55555554 by
1897 default (disallow access to all but pkey 0). Can
1898 override in debugfs after boot.
1900 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
1903 int_pln_enable [X86] Enable power limit notification interrupt
1905 integrity_audit=[IMA]
1906 Format: { "0" | "1" }
1907 0 -- basic integrity auditing messages. (Default)
1908 1 -- additional integrity auditing messages.
1910 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
1912 Enable intel iommu driver.
1914 Disable intel iommu driver.
1915 igfx_off [Default Off]
1916 By default, gfx is mapped as normal device. If a gfx
1917 device has a dedicated DMAR unit, the DMAR unit is
1918 bypassed by not enabling DMAR with this option. In
1919 this case, gfx device will use physical address for
1921 strict [Default Off]
1922 With this option on every unmap_single operation will
1923 result in a hardware IOTLB flush operation as opposed
1924 to batching them for performance.
1925 sp_off [Default Off]
1926 By default, super page will be supported if Intel IOMMU
1927 has the capability. With this option, super page will
1930 By default, scalable mode will be disabled even if the
1931 hardware advertises that it has support for the scalable
1932 mode translation. With this option set, scalable mode
1933 will be used on hardware which claims to support it.
1934 tboot_noforce [Default Off]
1935 Do not force the Intel IOMMU enabled under tboot.
1936 By default, tboot will force Intel IOMMU on, which
1937 could harm performance of some high-throughput
1938 devices like 40GBit network cards, even if identity
1940 Note that using this option lowers the security
1941 provided by tboot because it makes the system
1942 vulnerable to DMA attacks.
1944 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
1945 0 disables intel_idle and fall back on acpi_idle.
1946 1 to 9 specify maximum depth of C-state.
1950 Do not enable intel_pstate as the default
1951 scaling driver for the supported processors
1953 Use intel_pstate as a scaling driver, but configure it
1954 to work with generic cpufreq governors (instead of
1955 enabling its internal governor). This mode cannot be
1956 used along with the hardware-managed P-states (HWP)
1959 Enable intel_pstate on systems that prohibit it by default
1960 in favor of acpi-cpufreq. Forcing the intel_pstate driver
1961 instead of acpi-cpufreq may disable platform features, such
1962 as thermal controls and power capping, that rely on ACPI
1963 P-States information being indicated to OSPM and therefore
1964 should be used with caution. This option does not work with
1965 processors that aren't supported by the intel_pstate driver
1966 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
1968 Do not enable hardware P state control (HWP)
1971 Only load intel_pstate on systems which support
1972 hardware P state control (HWP) if available.
1974 Enforce ACPI _PPC performance limits. If the Fixed ACPI
1975 Description Table, specifies preferred power management
1976 profile as "Enterprise Server" or "Performance Server",
1977 then this feature is turned on by default.
1979 Allow per-logical-CPU P-State performance control limits using
1980 cpufreq sysfs interface
1982 intremap= [X86-64, Intel-IOMMU]
1983 on enable Interrupt Remapping (default)
1984 off disable Interrupt Remapping
1985 nosid disable Source ID checking
1987 BIOS x2APIC opt-out request will be ignored
1988 nopost disable Interrupt Posting
1990 iomem= Disable strict checking of access to MMIO memory
1991 strict regions from userspace.
2006 nobypass [PPC/POWERNV]
2007 Disable IOMMU bypass, using IOMMU for PCI devices.
2009 iommu.forcedac= [ARM64, X86] Control IOVA allocation for PCI devices.
2010 Format: { "0" | "1" }
2011 0 - Try to allocate a 32-bit DMA address first, before
2012 falling back to the full range if needed.
2013 1 - Allocate directly from the full usable range,
2014 forcing Dual Address Cycle for PCI cards supporting
2015 greater than 32-bit addressing.
2017 iommu.strict= [ARM64] Configure TLB invalidation behaviour
2018 Format: { "0" | "1" }
2020 Request that DMA unmap operations use deferred
2021 invalidation of hardware TLBs, for increased
2022 throughput at the cost of reduced device isolation.
2023 Will fall back to strict mode if not supported by
2024 the relevant IOMMU driver.
2025 1 - Strict mode (default).
2026 DMA unmap operations invalidate IOMMU hardware TLBs
2030 [ARM64, X86] Configure DMA to bypass the IOMMU by default.
2031 Format: { "0" | "1" }
2032 0 - Use IOMMU translation for DMA.
2033 1 - Bypass the IOMMU for DMA.
2034 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
2036 io7= [HW] IO7 for Marvel-based Alpha systems
2037 See comment before marvel_specify_io7 in
2038 arch/alpha/kernel/core_marvel.c.
2040 io_delay= [X86] I/O delay method
2042 Standard port 0x80 based delay
2044 Alternate port 0xed based delay (needed on some systems)
2046 Simple two microseconds delay
2051 See Documentation/admin-guide/nfs/nfsroot.rst.
2053 ipcmni_extend [KNL] Extend the maximum number of unique System V
2054 IPC identifiers from 32,768 to 16,777,216.
2056 irqaffinity= [SMP] Set the default irq affinity mask
2057 The argument is a cpu list, as described above.
2059 irqchip.gicv2_force_probe=
2062 Force the kernel to look for the second 4kB page
2063 of a GICv2 controller even if the memory range
2064 exposed by the device tree is too small.
2066 irqchip.gicv3_nolpi=
2068 Force the kernel to ignore the availability of
2069 LPIs (and by consequence ITSs). Intended for system
2070 that use the kernel as a bootloader, and thus want
2071 to let secondary kernels in charge of setting up
2074 irqchip.gicv3_pseudo_nmi= [ARM64]
2075 Enables support for pseudo-NMIs in the kernel. This
2076 requires the kernel to be built with
2077 CONFIG_ARM64_PSEUDO_NMI.
2080 When an interrupt is not handled search all handlers
2081 for it. Intended to get systems with badly broken
2085 When an interrupt is not handled search all handlers
2086 for it. Also check all handlers each timer
2087 interrupt. Intended to get systems with badly broken
2091 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
2093 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
2094 [Deprecated - use cpusets instead]
2095 Format: [flag-list,]<cpu-list>
2097 Specify one or more CPUs to isolate from disturbances
2098 specified in the flag list (default: domain):
2101 Disable the tick when a single task runs.
2103 A residual 1Hz tick is offloaded to workqueues, which you
2104 need to affine to housekeeping through the global
2105 workqueue's affinity configured via the
2106 /sys/devices/virtual/workqueue/cpumask sysfs file, or
2107 by using the 'domain' flag described below.
2109 NOTE: by default the global workqueue runs on all CPUs,
2110 so to protect individual CPUs the 'cpumask' file has to
2111 be configured manually after bootup.
2114 Isolate from the general SMP balancing and scheduling
2115 algorithms. Note that performing domain isolation this way
2116 is irreversible: it's not possible to bring back a CPU to
2117 the domains once isolated through isolcpus. It's strongly
2118 advised to use cpusets instead to disable scheduler load
2119 balancing through the "cpuset.sched_load_balance" file.
2120 It offers a much more flexible interface where CPUs can
2121 move in and out of an isolated set anytime.
2123 You can move a process onto or off an "isolated" CPU via
2124 the CPU affinity syscalls or cpuset.
2125 <cpu number> begins at 0 and the maximum value is
2126 "number of CPUs in system - 1".
2130 Isolate from being targeted by managed interrupts
2131 which have an interrupt mask containing isolated
2132 CPUs. The affinity of managed interrupts is
2133 handled by the kernel and cannot be changed via
2134 the /proc/irq/* interfaces.
2136 This isolation is best effort and only effective
2137 if the automatically assigned interrupt mask of a
2138 device queue contains isolated and housekeeping
2139 CPUs. If housekeeping CPUs are online then such
2140 interrupts are directed to the housekeeping CPU
2141 so that IO submitted on the housekeeping CPU
2142 cannot disturb the isolated CPU.
2144 If a queue's affinity mask contains only isolated
2145 CPUs then this parameter has no effect on the
2146 interrupt routing decision, though interrupts are
2147 only delivered when tasks running on those
2148 isolated CPUs submit IO. IO submitted on
2149 housekeeping CPUs has no influence on those
2152 The format of <cpu-list> is described above.
2156 ivrs_ioapic [HW,X86-64]
2157 Provide an override to the IOAPIC-ID<->DEVICE-ID
2158 mapping provided in the IVRS ACPI table. For
2159 example, to map IOAPIC-ID decimal 10 to
2160 PCI device 00:14.0 write the parameter as:
2161 ivrs_ioapic[10]=00:14.0
2163 ivrs_hpet [HW,X86-64]
2164 Provide an override to the HPET-ID<->DEVICE-ID
2165 mapping provided in the IVRS ACPI table. For
2166 example, to map HPET-ID decimal 0 to
2167 PCI device 00:14.0 write the parameter as:
2168 ivrs_hpet[0]=00:14.0
2170 ivrs_acpihid [HW,X86-64]
2171 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2172 mapping provided in the IVRS ACPI table. For
2173 example, to map UART-HID:UID AMD0020:0 to
2174 PCI device 00:14.5 write the parameter as:
2175 ivrs_acpihid[00:14.5]=AMD0020:0
2177 js= [HW,JOY] Analog joystick
2178 See Documentation/input/joydev/joystick.rst.
2181 When CONFIG_RANDOMIZE_BASE is set, this disables
2182 kernel and module base offset ASLR (Address Space
2183 Layout Randomization).
2186 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2187 report on every invalid memory access. Without this
2188 parameter KASAN will print report only for the first
2193 kernelcore= [KNL,X86,IA-64,PPC]
2194 Format: nn[KMGTPE] | nn% | "mirror"
2195 This parameter specifies the amount of memory usable by
2196 the kernel for non-movable allocations. The requested
2197 amount is spread evenly throughout all nodes in the
2198 system as ZONE_NORMAL. The remaining memory is used for
2199 movable memory in its own zone, ZONE_MOVABLE. In the
2200 event, a node is too small to have both ZONE_NORMAL and
2201 ZONE_MOVABLE, kernelcore memory will take priority and
2202 other nodes will have a larger ZONE_MOVABLE.
2204 ZONE_MOVABLE is used for the allocation of pages that
2205 may be reclaimed or moved by the page migration
2206 subsystem. Note that allocations like PTEs-from-HighMem
2207 still use the HighMem zone if it exists, and the Normal
2208 zone if it does not.
2210 It is possible to specify the exact amount of memory in
2211 the form of "nn[KMGTPE]", a percentage of total system
2212 memory in the form of "nn%", or "mirror". If "mirror"
2213 option is specified, mirrored (reliable) memory is used
2214 for non-movable allocations and remaining memory is used
2215 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2216 are exclusive, so you cannot specify multiple forms.
2218 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
2219 Format: <Controller#>[,poll interval]
2220 The controller # is the number of the ehci usb debug
2221 port as it is probed via PCI. The poll interval is
2222 optional and is the number seconds in between
2223 each poll cycle to the debug port in case you need
2224 the functionality for interrupting the kernel with
2225 gdb or control-c on the dbgp connection. When
2226 not using this parameter you use sysrq-g to break into
2227 the kernel debugger.
2229 kgdboc= [KGDB,HW] kgdb over consoles.
2230 Requires a tty driver that supports console polling,
2231 or a supported polling keyboard driver (non-usb).
2232 Serial only format: <serial_device>[,baud]
2233 keyboard only format: kbd
2234 keyboard and serial format: kbd,<serial_device>[,baud]
2235 Optional Kernel mode setting:
2236 kms, kbd format: kms,kbd
2237 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2239 kgdboc_earlycon= [KGDB,HW]
2240 If the boot console provides the ability to read
2241 characters and can work in polling mode, you can use
2242 this parameter to tell kgdb to use it as a backend
2243 until the normal console is registered. Intended to
2244 be used together with the kgdboc parameter which
2245 specifies the normal console to transition to.
2247 The name of the early console should be specified
2248 as the value of this parameter. Note that the name of
2249 the early console might be different than the tty
2250 name passed to kgdboc. It's OK to leave the value
2251 blank and the first boot console that implements
2252 read() will be picked.
2254 kgdbwait [KGDB] Stop kernel execution and enter the
2255 kernel debugger at the earliest opportunity.
2257 kmac= [MIPS] Korina ethernet MAC address.
2258 Configure the RouterBoard 532 series on-chip
2259 Ethernet adapter MAC address.
2261 kmemleak= [KNL] Boot-time kmemleak enable/disable
2262 Valid arguments: on, off
2264 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2267 kprobe_event=[probe-list]
2268 [FTRACE] Add kprobe events and enable at boot time.
2269 The probe-list is a semicolon delimited list of probe
2270 definitions. Each definition is same as kprobe_events
2271 interface, but the parameters are comma delimited.
2272 For example, to add a kprobe event on vfs_read with
2273 arg1 and arg2, add to the command line;
2275 kprobe_event=p,vfs_read,$arg1,$arg2
2277 See also Documentation/trace/kprobetrace.rst "Kernel
2278 Boot Parameter" section.
2280 kpti= [ARM64] Control page table isolation of user
2281 and kernel address spaces.
2282 Default: enabled on cores which need mitigation.
2286 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2287 Default is 0 (don't ignore, but inject #GP)
2289 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2290 Default is false (don't support).
2292 kvm.mmu_audit= [KVM] This is a R/W parameter which allows audit
2297 [KVM] Controls the software workaround for the
2298 X86_BUG_ITLB_MULTIHIT bug.
2299 force : Always deploy workaround.
2300 off : Never deploy workaround.
2301 auto : Deploy workaround based on the presence of
2302 X86_BUG_ITLB_MULTIHIT.
2306 If the software workaround is enabled for the host,
2307 guests do need not to enable it for nested guests.
2309 kvm.nx_huge_pages_recovery_ratio=
2310 [KVM] Controls how many 4KiB pages are periodically zapped
2311 back to huge pages. 0 disables the recovery, otherwise if
2312 the value is N KVM will zap 1/Nth of the 4KiB pages every
2313 minute. The default is 60.
2315 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM.
2316 Default is 1 (enabled)
2318 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU)
2320 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode.
2323 [KVM,ARM] Select one of KVM/arm64's modes of operation.
2325 nvhe: Standard nVHE-based mode, without support for
2328 protected: nVHE-based mode with support for guests whose
2329 state is kept private from the host.
2330 Not valid if the kernel is running in EL2.
2332 Defaults to VHE/nVHE based on hardware support.
2334 kvm-arm.vgic_v3_group0_trap=
2335 [KVM,ARM] Trap guest accesses to GICv3 group-0
2338 kvm-arm.vgic_v3_group1_trap=
2339 [KVM,ARM] Trap guest accesses to GICv3 group-1
2342 kvm-arm.vgic_v3_common_trap=
2343 [KVM,ARM] Trap guest accesses to GICv3 common
2346 kvm-arm.vgic_v4_enable=
2347 [KVM,ARM] Allow use of GICv4 for direct injection of
2350 kvm_cma_resv_ratio=n [PPC]
2351 Reserves given percentage from system memory area for
2352 contiguous memory allocation for KVM hash pagetable
2354 By default it reserves 5% of total system memory.
2358 kvm-intel.ept= [KVM,Intel] Disable extended page tables
2359 (virtualized MMU) support on capable Intel chips.
2360 Default is 1 (enabled)
2362 kvm-intel.emulate_invalid_guest_state=
2363 [KVM,Intel] Enable emulation of invalid guest states
2364 Default is 0 (disabled)
2366 kvm-intel.flexpriority=
2367 [KVM,Intel] Disable FlexPriority feature (TPR shadow).
2368 Default is 1 (enabled)
2371 [KVM,Intel] Enable VMX nesting (nVMX).
2372 Default is 0 (disabled)
2374 kvm-intel.unrestricted_guest=
2375 [KVM,Intel] Disable unrestricted guest feature
2376 (virtualized real and unpaged mode) on capable
2377 Intel chips. Default is 1 (enabled)
2379 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2382 Valid arguments: never, cond, always
2384 always: L1D cache flush on every VMENTER.
2385 cond: Flush L1D on VMENTER only when the code between
2386 VMEXIT and VMENTER can leak host memory.
2387 never: Disables the mitigation
2389 Default is cond (do L1 cache flush in specific instances)
2391 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification
2392 feature (tagged TLBs) on capable Intel chips.
2393 Default is 1 (enabled)
2395 l1tf= [X86] Control mitigation of the L1TF vulnerability on
2398 The kernel PTE inversion protection is unconditionally
2399 enabled and cannot be disabled.
2402 Provides all available mitigations for the
2403 L1TF vulnerability. Disables SMT and
2404 enables all mitigations in the
2405 hypervisors, i.e. unconditional L1D flush.
2407 SMT control and L1D flush control via the
2408 sysfs interface is still possible after
2409 boot. Hypervisors will issue a warning
2410 when the first VM is started in a
2411 potentially insecure configuration,
2412 i.e. SMT enabled or L1D flush disabled.
2415 Same as 'full', but disables SMT and L1D
2416 flush runtime control. Implies the
2417 'nosmt=force' command line option.
2418 (i.e. sysfs control of SMT is disabled.)
2421 Leaves SMT enabled and enables the default
2422 hypervisor mitigation, i.e. conditional
2425 SMT control and L1D flush control via the
2426 sysfs interface is still possible after
2427 boot. Hypervisors will issue a warning
2428 when the first VM is started in a
2429 potentially insecure configuration,
2430 i.e. SMT enabled or L1D flush disabled.
2434 Disables SMT and enables the default
2435 hypervisor mitigation.
2437 SMT control and L1D flush control via the
2438 sysfs interface is still possible after
2439 boot. Hypervisors will issue a warning
2440 when the first VM is started in a
2441 potentially insecure configuration,
2442 i.e. SMT enabled or L1D flush disabled.
2445 Same as 'flush', but hypervisors will not
2446 warn when a VM is started in a potentially
2447 insecure configuration.
2450 Disables hypervisor mitigations and doesn't
2452 It also drops the swap size and available
2453 RAM limit restriction on both hypervisor and
2458 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2464 lapic [X86-32,APIC] Enable the local APIC even if BIOS
2467 lapic= [X86,APIC] Do not use TSC deadline
2468 value for LAPIC timer one-shot implementation. Default
2469 back to the programmable timer unit in the LAPIC.
2470 Format: notscdeadline
2472 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
2475 libata.dma= [LIBATA] DMA control
2476 libata.dma=0 Disable all PATA and SATA DMA
2477 libata.dma=1 PATA and SATA Disk DMA only
2478 libata.dma=2 ATAPI (CDROM) DMA only
2479 libata.dma=4 Compact Flash DMA only
2480 Combinations also work, so libata.dma=3 enables DMA
2481 for disks and CDROMs, but not CFs.
2483 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2484 libata.ignore_hpa=0 keep BIOS limits (default)
2485 libata.ignore_hpa=1 ignore limits, using full disk
2487 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2491 libata.force= [LIBATA] Force configurations. The format is comma-
2492 separated list of "[ID:]VAL" where ID is
2493 PORT[.DEVICE]. PORT and DEVICE are decimal numbers
2494 matching port, link or device. Basically, it matches
2495 the ATA ID string printed on console by libata. If
2496 the whole ID part is omitted, the last PORT and DEVICE
2497 values are used. If ID hasn't been specified yet, the
2498 configuration applies to all ports, links and devices.
2500 If only DEVICE is omitted, the parameter applies to
2501 the port and all links and devices behind it. DEVICE
2502 number of 0 either selects the first device or the
2503 first fan-out link behind PMP device. It does not
2504 select the host link. DEVICE number of 15 selects the
2505 host link and device attached to it.
2507 The VAL specifies the configuration to force. As long
2508 as there's no ambiguity shortcut notation is allowed.
2509 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2510 The following configurations can be forced.
2512 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2513 Any ID with matching PORT is used.
2515 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2517 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2518 udma[/][16,25,33,44,66,100,133] notation is also
2521 * [no]ncq: Turn on or off NCQ.
2523 * [no]ncqtrim: Turn off queued DSM TRIM.
2525 * nohrst, nosrst, norst: suppress hard, soft
2528 * rstonce: only attempt one reset during
2529 hot-unplug link recovery
2531 * dump_id: dump IDENTIFY data.
2533 * atapi_dmadir: Enable ATAPI DMADIR bridge support
2535 * disable: Disable this device.
2537 If there are multiple matching configurations changing
2538 the same attribute, the last one is used.
2540 memblock=debug [KNL] Enable memblock debug messages.
2542 load_ramdisk= [RAM] [Deprecated]
2544 lockd.nlm_grace_period=P [NFS] Assign grace period.
2547 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2550 lockd.nlm_timeout=T [NFS] Assign timeout value.
2553 lockd.nlm_udpport=M [NFS] Assign UDP port.
2556 lockdown= [SECURITY]
2557 { integrity | confidentiality }
2558 Enable the kernel lockdown feature. If set to
2559 integrity, kernel features that allow userland to
2560 modify the running kernel are disabled. If set to
2561 confidentiality, kernel features that allow userland
2562 to extract confidential information from the kernel
2565 locktorture.nreaders_stress= [KNL]
2566 Set the number of locking read-acquisition kthreads.
2567 Defaults to being automatically set based on the
2568 number of online CPUs.
2570 locktorture.nwriters_stress= [KNL]
2571 Set the number of locking write-acquisition kthreads.
2573 locktorture.onoff_holdoff= [KNL]
2574 Set time (s) after boot for CPU-hotplug testing.
2576 locktorture.onoff_interval= [KNL]
2577 Set time (s) between CPU-hotplug operations, or
2578 zero to disable CPU-hotplug testing.
2580 locktorture.shuffle_interval= [KNL]
2581 Set task-shuffle interval (jiffies). Shuffling
2582 tasks allows some CPUs to go into dyntick-idle
2583 mode during the locktorture test.
2585 locktorture.shutdown_secs= [KNL]
2586 Set time (s) after boot system shutdown. This
2587 is useful for hands-off automated testing.
2589 locktorture.stat_interval= [KNL]
2590 Time (s) between statistics printk()s.
2592 locktorture.stutter= [KNL]
2593 Time (s) to stutter testing, for example,
2594 specifying five seconds causes the test to run for
2595 five seconds, wait for five seconds, and so on.
2596 This tests the locking primitive's ability to
2597 transition abruptly to and from idle.
2599 locktorture.torture_type= [KNL]
2600 Specify the locking implementation to test.
2602 locktorture.verbose= [KNL]
2603 Enable additional printk() statements.
2605 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
2608 loglevel= All Kernel Messages with a loglevel smaller than the
2609 console loglevel will be printed to the console. It can
2610 also be changed with klogd or other programs. The
2611 loglevels are defined as follows:
2613 0 (KERN_EMERG) system is unusable
2614 1 (KERN_ALERT) action must be taken immediately
2615 2 (KERN_CRIT) critical conditions
2616 3 (KERN_ERR) error conditions
2617 4 (KERN_WARNING) warning conditions
2618 5 (KERN_NOTICE) normal but significant condition
2619 6 (KERN_INFO) informational
2620 7 (KERN_DEBUG) debug-level messages
2622 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
2623 in bytes. n must be a power of two and greater
2624 than the minimal size. The minimal size is defined
2625 by LOG_BUF_SHIFT kernel config parameter. There is
2626 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2627 that allows to increase the default size depending on
2628 the number of CPUs. See init/Kconfig for more details.
2630 logo.nologo [FB] Disables display of the built-in Linux logo.
2631 This may be used to provide more screen space for
2632 kernel log messages and is useful when debugging
2633 kernel boot problems.
2635 lp=0 [LP] Specify parallel ports to use, e.g,
2636 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2637 lp=reset first parallel port). 'lp=0' disables the
2638 lp=auto printer driver. 'lp=reset' (which can be
2639 specified in addition to the ports) causes
2640 attached printers to be reset. Using
2641 lp=port1,port2,... specifies the parallel ports
2642 to associate lp devices with, starting with
2643 lp0. A port specification may be 'none' to skip
2644 that lp device, or a parport name such as
2645 'parport0'. Specifying 'lp=auto' instead of a
2646 port specification list means that device IDs
2647 from each port should be examined, to see if
2648 an IEEE 1284-compliant printer is attached; if
2649 so, the driver will manage that printer.
2650 See also header of drivers/char/lp.c.
2653 Sets loops_per_jiffy to given constant, thus avoiding
2654 time-consuming boot-time autodetection (up to 250 ms per
2655 CPU). 0 enables autodetection (default). To determine
2656 the correct value for your kernel, boot with normal
2657 autodetection and see what value is printed. Note that
2658 on SMP systems the preset will be applied to all CPUs,
2659 which is likely to cause problems if your CPUs need
2660 significantly divergent settings. An incorrect value
2661 will cause delays in the kernel to be wrong, leading to
2662 unpredictable I/O errors and other breakage. Although
2663 unlikely, in the extreme case this might damage your
2667 Format: <io>,<irq>,<dma>
2669 lsm.debug [SECURITY] Enable LSM initialization debugging output.
2672 [SECURITY] Choose order of LSM initialization. This
2673 overrides CONFIG_LSM, and the "security=" parameter.
2675 machvec= [IA-64] Force the use of a particular machine-vector
2676 (machvec) in a generic kernel.
2677 Example: machvec=hpzx1
2679 machtype= [Loongson] Share the same kernel image file between
2680 different yeeloong laptops.
2681 Example: machtype=lemote-yeeloong-2f-7inch
2683 max_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory greater
2684 than or equal to this physical address is ignored.
2686 maxcpus= [SMP] Maximum number of processors that an SMP kernel
2687 will bring up during bootup. maxcpus=n : n >= 0 limits
2688 the kernel to bring up 'n' processors. Surely after
2689 bootup you can bring up the other plugged cpu by executing
2690 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
2691 only takes effect during system bootup.
2692 While n=0 is a special case, it is equivalent to "nosmp",
2693 which also disables the IO APIC.
2695 max_loop= [LOOP] The number of loop block devices that get
2696 (loop.max_loop) unconditionally pre-created at init time. The default
2697 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
2698 of statically allocating a predefined number, loop
2699 devices can be requested on-demand with the
2700 /dev/loop-control interface.
2702 mce [X86-32] Machine Check Exception
2704 mce=option [X86-64] See Documentation/x86/x86_64/boot-options.rst
2706 md= [HW] RAID subsystems devices and level
2707 See Documentation/admin-guide/md.rst.
2710 Format: <first>,<last>
2711 Specifies range of consoles to be captured by the MDA.
2714 Control mitigation for the Micro-architectural Data
2715 Sampling (MDS) vulnerability.
2717 Certain CPUs are vulnerable to an exploit against CPU
2718 internal buffers which can forward information to a
2719 disclosure gadget under certain conditions.
2721 In vulnerable processors, the speculatively
2722 forwarded data can be used in a cache side channel
2723 attack, to access data to which the attacker does
2724 not have direct access.
2726 This parameter controls the MDS mitigation. The
2729 full - Enable MDS mitigation on vulnerable CPUs
2730 full,nosmt - Enable MDS mitigation and disable
2731 SMT on vulnerable CPUs
2732 off - Unconditionally disable MDS mitigation
2734 On TAA-affected machines, mds=off can be prevented by
2735 an active TAA mitigation as both vulnerabilities are
2736 mitigated with the same mechanism so in order to disable
2737 this mitigation, you need to specify tsx_async_abort=off
2740 Not specifying this option is equivalent to
2743 For details see: Documentation/admin-guide/hw-vuln/mds.rst
2745 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
2746 Amount of memory to be used in cases as follows:
2749 2 when the kernel is not able to see the whole system memory;
2750 3 memory that lies after 'mem=' boundary is excluded from
2751 the hypervisor, then assigned to KVM guests.
2753 [X86] Work as limiting max address. Use together
2754 with memmap= to avoid physical address space collisions.
2755 Without memmap= PCI devices could be placed at addresses
2756 belonging to unused RAM.
2758 Note that this only takes effects during boot time since
2759 in above case 3, memory may need be hot added after boot
2760 if system memory of hypervisor is not sufficient.
2762 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
2766 [KNL,SH] Allow user to override the default size for
2767 per-device physically contiguous DMA buffers.
2769 memhp_default_state=online/offline
2770 [KNL] Set the initial state for the memory hotplug
2771 onlining policy. If not specified, the default value is
2772 set according to the
2773 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
2775 See Documentation/admin-guide/mm/memory-hotplug.rst.
2777 memmap=exactmap [KNL,X86] Enable setting of an exact
2778 E820 memory map, as specified by the user.
2779 Such memmap=exactmap lines can be constructed based on
2780 BIOS output or other requirements. See the memmap=nn@ss
2783 memmap=nn[KMG]@ss[KMG]
2784 [KNL, X86, MIPS, XTENSA] Force usage of a specific region of memory.
2785 Region of memory to be used is from ss to ss+nn.
2786 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
2787 which limits max address to nn[KMG].
2788 Multiple different regions can be specified,
2791 memmap=100M@2G,100M#3G,1G!1024G
2793 memmap=nn[KMG]#ss[KMG]
2794 [KNL,ACPI] Mark specific memory as ACPI data.
2795 Region of memory to be marked is from ss to ss+nn.
2797 memmap=nn[KMG]$ss[KMG]
2798 [KNL,ACPI] Mark specific memory as reserved.
2799 Region of memory to be reserved is from ss to ss+nn.
2800 Example: Exclude memory from 0x18690000-0x1869ffff
2801 memmap=64K$0x18690000
2803 memmap=0x10000$0x18690000
2804 Some bootloaders may need an escape character before '$',
2805 like Grub2, otherwise '$' and the following number
2808 memmap=nn[KMG]!ss[KMG]
2809 [KNL,X86] Mark specific memory as protected.
2810 Region of memory to be used, from ss to ss+nn.
2811 The memory region may be marked as e820 type 12 (0xc)
2812 and is NVDIMM or ADR memory.
2814 memmap=<size>%<offset>-<oldtype>+<newtype>
2815 [KNL,ACPI] Convert memory within the specified region
2816 from <oldtype> to <newtype>. If "-<oldtype>" is left
2817 out, the whole region will be marked as <newtype>,
2818 even if previously unavailable. If "+<newtype>" is left
2819 out, matching memory will be removed. Types are
2820 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
2821 3 = ACPI, 12 = PRAM.
2823 memory_corruption_check=0/1 [X86]
2824 Some BIOSes seem to corrupt the first 64k of
2825 memory when doing things like suspend/resume.
2826 Setting this option will scan the memory
2827 looking for corruption. Enabling this will
2828 both detect corruption and prevent the kernel
2829 from using the memory being corrupted.
2830 However, its intended as a diagnostic tool; if
2831 repeatable BIOS-originated corruption always
2832 affects the same memory, you can use memmap=
2833 to prevent the kernel from using that memory.
2835 memory_corruption_check_size=size [X86]
2836 By default it checks for corruption in the low
2837 64k, making this memory unavailable for normal
2838 use. Use this parameter to scan for
2839 corruption in more or less memory.
2841 memory_corruption_check_period=seconds [X86]
2842 By default it checks for corruption every 60
2843 seconds. Use this parameter to check at some
2844 other rate. 0 disables periodic checking.
2846 memory_hotplug.memmap_on_memory
2847 [KNL,X86,ARM] Boolean flag to enable this feature.
2848 Format: {on | off (default)}
2849 When enabled, runtime hotplugged memory will
2850 allocate its internal metadata (struct pages)
2851 from the hotadded memory which will allow to
2852 hotadd a lot of memory without requiring
2853 additional memory to do so.
2854 This feature is disabled by default because it
2855 has some implication on large (e.g. GB)
2856 allocations in some configurations (e.g. small
2858 The state of the flag can be read in
2859 /sys/module/memory_hotplug/parameters/memmap_on_memory.
2860 Note that even when enabled, there are a few cases where
2861 the feature is not effective.
2863 memtest= [KNL,X86,ARM,PPC,RISCV] Enable memtest
2865 default : 0 <disable>
2866 Specifies the number of memtest passes to be
2867 performed. Each pass selects another test
2868 pattern from a given set of patterns. Memtest
2869 fills the memory with this pattern, validates
2870 memory contents and reserves bad memory
2871 regions that are detected.
2873 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
2874 Valid arguments: on, off
2875 Default (depends on kernel configuration option):
2876 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
2877 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
2878 mem_encrypt=on: Activate SME
2879 mem_encrypt=off: Do not activate SME
2881 Refer to Documentation/virt/kvm/amd-memory-encryption.rst
2882 for details on when memory encryption can be activated.
2884 mem_sleep_default= [SUSPEND] Default system suspend mode:
2885 s2idle - Suspend-To-Idle
2886 shallow - Power-On Suspend or equivalent (if supported)
2887 deep - Suspend-To-RAM or equivalent (if supported)
2888 See Documentation/admin-guide/pm/sleep-states.rst.
2890 meye.*= [HW] Set MotionEye Camera parameters
2891 See Documentation/admin-guide/media/meye.rst.
2893 mfgpt_irq= [IA-32] Specify the IRQ to use for the
2894 Multi-Function General Purpose Timers on AMD Geode
2897 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
2898 the BIOS has incorrectly applied a workaround. TinyBIOS
2899 version 0.98 is known to be affected, 0.99 fixes the
2900 problem by letting the user disable the workaround.
2904 min_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory below this
2905 physical address is ignored.
2907 mini2440= [ARM,HW,KNL]
2908 Format:[0..2][b][c][t]
2910 MINI2440 configuration specification:
2911 0 - The attached screen is the 3.5" TFT
2912 1 - The attached screen is the 7" TFT
2913 2 - The VGA Shield is attached (1024x768)
2914 Leaving out the screen size parameter will not load
2915 the TFT driver, and the framebuffer will be left
2917 b - Enable backlight. The TFT backlight pin will be
2918 linked to the kernel VESA blanking code and a GPIO
2919 LED. This parameter is not necessary when using the
2921 c - Enable the s3c camera interface.
2922 t - Reserved for enabling touchscreen support. The
2923 touchscreen support is not enabled in the mainstream
2924 kernel as of 2.6.30, a preliminary port can be found
2925 in the "bleeding edge" mini2440 support kernel at
2926 https://repo.or.cz/w/linux-2.6/mini2440.git
2929 [X86,PPC,S390,ARM64] Control optional mitigations for
2930 CPU vulnerabilities. This is a set of curated,
2931 arch-independent options, each of which is an
2932 aggregation of existing arch-specific options.
2935 Disable all optional CPU mitigations. This
2936 improves system performance, but it may also
2937 expose users to several CPU vulnerabilities.
2938 Equivalent to: nopti [X86,PPC]
2940 nospectre_v1 [X86,PPC]
2942 nospectre_v2 [X86,PPC,S390,ARM64]
2943 spectre_v2_user=off [X86]
2944 spec_store_bypass_disable=off [X86,PPC]
2945 ssbd=force-off [ARM64]
2948 tsx_async_abort=off [X86]
2949 kvm.nx_huge_pages=off [X86]
2950 no_entry_flush [PPC]
2951 no_uaccess_flush [PPC]
2954 This does not have any effect on
2955 kvm.nx_huge_pages when
2956 kvm.nx_huge_pages=force.
2959 Mitigate all CPU vulnerabilities, but leave SMT
2960 enabled, even if it's vulnerable. This is for
2961 users who don't want to be surprised by SMT
2962 getting disabled across kernel upgrades, or who
2963 have other ways of avoiding SMT-based attacks.
2964 Equivalent to: (default behavior)
2967 Mitigate all CPU vulnerabilities, disabling SMT
2968 if needed. This is for users who always want to
2969 be fully mitigated, even if it means losing SMT.
2970 Equivalent to: l1tf=flush,nosmt [X86]
2971 mds=full,nosmt [X86]
2972 tsx_async_abort=full,nosmt [X86]
2975 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
2976 parameter allows control of the logging verbosity for
2977 the additional memory initialisation checks. A value
2978 of 0 disables mminit logging and a level of 4 will
2979 log everything. Information is printed at KERN_DEBUG
2980 so loglevel=8 may also need to be specified.
2983 [KNL] When CONFIG_MODULE_SIG is set, this means that
2984 modules without (valid) signatures will fail to load.
2985 Note that if CONFIG_MODULE_SIG_FORCE is set, that
2986 is always true, so this option does nothing.
2988 module_blacklist= [KNL] Do not load a comma-separated list of
2989 modules. Useful for debugging problem modules.
2992 [MOUSE] Maximum time between finger touching and
2993 leaving touchpad surface for touch to be considered
2994 a tap and be reported as a left button click (for
2995 touchpads working in absolute mode only).
2997 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
2998 reporting absolute coordinates, such as tablets
2999 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
3000 reporting absolute coordinates, such as tablets
3002 movablecore= [KNL,X86,IA-64,PPC]
3003 Format: nn[KMGTPE] | nn%
3004 This parameter is the complement to kernelcore=, it
3005 specifies the amount of memory used for migratable
3006 allocations. If both kernelcore and movablecore is
3007 specified, then kernelcore will be at *least* the
3008 specified value but may be more. If movablecore on its
3009 own is specified, the administrator must be careful
3010 that the amount of memory usable for all allocations
3013 movable_node [KNL] Boot-time switch to make hotplugable memory
3014 NUMA nodes to be movable. This means that the memory
3015 of such nodes will be usable only for movable
3016 allocations which rules out almost all kernel
3017 allocations. Use with caution!
3019 MTD_Partition= [MTD]
3020 Format: <name>,<region-number>,<size>,<offset>
3022 MTD_Region= [MTD] Format:
3023 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
3026 See drivers/mtd/parsers/cmdlinepart.c
3028 multitce=off [PPC] This parameter disables the use of the pSeries
3029 firmware feature for updating multiple TCE entries
3032 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
3034 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
3036 boundary - index of last SLC block on Flex-OneNAND.
3037 The remaining blocks are configured as MLC blocks.
3038 lock - Configure if Flex-OneNAND boundary should be locked.
3039 Once locked, the boundary cannot be changed.
3040 1 indicates lock status, 0 indicates unlock status.
3043 ARM/S3C2412 JIVE boot control
3045 See arch/arm/mach-s3c/mach-jive.c
3047 mtouchusb.raw_coordinates=
3048 [HW] Make the MicroTouch USB driver use raw coordinates
3049 ('y', default) or cooked coordinates ('n')
3051 mtrr_chunk_size=nn[KMG] [X86]
3052 used for mtrr cleanup. It is largest continuous chunk
3053 that could hold holes aka. UC entries.
3055 mtrr_gran_size=nn[KMG] [X86]
3056 Used for mtrr cleanup. It is granularity of mtrr block.
3058 Large value could prevent small alignment from
3061 mtrr_spare_reg_nr=n [X86]
3063 Range: 0,7 : spare reg number
3065 Used for mtrr cleanup. It is spare mtrr entries number.
3066 Set to 2 or more if your graphical card needs more.
3068 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
3070 netdev= [NET] Network devices parameters
3071 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
3072 Note that mem_start is often overloaded to mean
3073 something different and driver-specific.
3074 This usage is only documented in each driver source
3078 [NETFILTER] Enable connection tracking flow accounting
3079 0 to disable accounting
3080 1 to enable accounting
3083 nfsaddrs= [NFS] Deprecated. Use ip= instead.
3084 See Documentation/admin-guide/nfs/nfsroot.rst.
3086 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
3087 See Documentation/admin-guide/nfs/nfsroot.rst.
3089 nfsrootdebug [NFS] enable nfsroot debugging messages.
3090 See Documentation/admin-guide/nfs/nfsroot.rst.
3092 nfs.callback_nr_threads=
3093 [NFSv4] set the total number of threads that the
3094 NFS client will assign to service NFSv4 callback
3097 nfs.callback_tcpport=
3098 [NFS] set the TCP port on which the NFSv4 callback
3099 channel should listen.
3102 [NFS] sets the pathname to the program which is used
3103 to update the NFS client cache entries.
3105 nfs.cache_getent_timeout=
3106 [NFS] sets the timeout after which an attempt to
3107 update a cache entry is deemed to have failed.
3109 nfs.idmap_cache_timeout=
3110 [NFS] set the maximum lifetime for idmapper cache
3114 [NFS] enable 64-bit inode numbers.
3115 If zero, the NFS client will fake up a 32-bit inode
3116 number for the readdir() and stat() syscalls instead
3117 of returning the full 64-bit number.
3118 The default is to return 64-bit inode numbers.
3120 nfs.max_session_cb_slots=
3121 [NFSv4.1] Sets the maximum number of session
3122 slots the client will assign to the callback
3123 channel. This determines the maximum number of
3124 callbacks the client will process in parallel for
3125 a particular server.
3127 nfs.max_session_slots=
3128 [NFSv4.1] Sets the maximum number of session slots
3129 the client will attempt to negotiate with the server.
3130 This limits the number of simultaneous RPC requests
3131 that the client can send to the NFSv4.1 server.
3132 Note that there is little point in setting this
3133 value higher than the max_tcp_slot_table_limit.
3135 nfs.nfs4_disable_idmapping=
3136 [NFSv4] When set to the default of '1', this option
3137 ensures that both the RPC level authentication
3138 scheme and the NFS level operations agree to use
3139 numeric uids/gids if the mount is using the
3140 'sec=sys' security flavour. In effect it is
3141 disabling idmapping, which can make migration from
3142 legacy NFSv2/v3 systems to NFSv4 easier.
3143 Servers that do not support this mode of operation
3144 will be autodetected by the client, and it will fall
3145 back to using the idmapper.
3146 To turn off this behaviour, set the value to '0'.
3148 [NFS4] Specify an additional fixed unique ident-
3149 ification string that NFSv4 clients can insert into
3150 their nfs_client_id4 string. This is typically a
3151 UUID that is generated at system install time.
3153 nfs.send_implementation_id =
3154 [NFSv4.1] Send client implementation identification
3155 information in exchange_id requests.
3156 If zero, no implementation identification information
3158 The default is to send the implementation identification
3161 nfs.recover_lost_locks =
3162 [NFSv4] Attempt to recover locks that were lost due
3163 to a lease timeout on the server. Please note that
3164 doing this risks data corruption, since there are
3165 no guarantees that the file will remain unchanged
3166 after the locks are lost.
3167 If you want to enable the kernel legacy behaviour of
3168 attempting to recover these locks, then set this
3170 The default parameter value of '0' causes the kernel
3171 not to attempt recovery of lost locks.
3173 nfs4.layoutstats_timer =
3174 [NFSv4.2] Change the rate at which the kernel sends
3175 layoutstats to the pNFS metadata server.
3177 Setting this to value to 0 causes the kernel to use
3178 whatever value is the default set by the layout
3179 driver. A non-zero value sets the minimum interval
3180 in seconds between layoutstats transmissions.
3182 nfsd.nfs4_disable_idmapping=
3183 [NFSv4] When set to the default of '1', the NFSv4
3184 server will return only numeric uids and gids to
3185 clients using auth_sys, and will accept numeric uids
3186 and gids from such clients. This is intended to ease
3187 migration from NFSv2/v3.
3189 nmi_backtrace.backtrace_idle [KNL]
3190 Dump stacks even of idle CPUs in response to an
3191 NMI stack-backtrace request.
3193 nmi_debug= [KNL,SH] Specify one or more actions to take
3194 when a NMI is triggered.
3195 Format: [state][,regs][,debounce][,die]
3197 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
3198 Format: [panic,][nopanic,][num]
3200 0 - turn hardlockup detector in nmi_watchdog off
3201 1 - turn hardlockup detector in nmi_watchdog on
3202 When panic is specified, panic when an NMI watchdog
3203 timeout occurs (or 'nopanic' to not panic on an NMI
3204 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
3205 To disable both hard and soft lockup detectors,
3206 please see 'nowatchdog'.
3207 This is useful when you use a panic=... timeout and
3208 need the box quickly up again.
3210 These settings can be accessed at runtime via
3211 the nmi_watchdog and hardlockup_panic sysctls.
3213 netpoll.carrier_timeout=
3214 [NET] Specifies amount of time (in seconds) that
3215 netpoll should wait for a carrier. By default netpoll
3218 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3219 emulation library even if a 387 maths coprocessor
3222 no5lvl [X86-64] Disable 5-level paging mode. Forces
3223 kernel to use 4-level paging instead.
3225 nofsgsbase [X86] Disables FSGSBASE instructions.
3228 [HW] Never suspend the console
3229 Disable suspending of consoles during suspend and
3230 hibernate operations. Once disabled, debugging
3231 messages can reach various consoles while the rest
3232 of the system is being put to sleep (ie, while
3233 debugging driver suspend/resume hooks). This may
3234 not work reliably with all consoles, but is known
3235 to work with serial and VGA consoles.
3236 To facilitate more flexible debugging, we also add
3237 console_suspend, a printk module parameter to control
3238 it. Users could use console_suspend (usually
3239 /sys/module/printk/parameters/console_suspend) to
3240 turn on/off it dynamically.
3242 novmcoredd [KNL,KDUMP]
3243 Disable device dump. Device dump allows drivers to
3244 append dump data to vmcore so you can collect driver
3245 specified debug info. Drivers can append the data
3246 without any limit and this data is stored in memory,
3247 so this may cause significant memory stress. Disabling
3248 device dump can help save memory but the driver debug
3249 data will be no longer available. This parameter
3250 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
3253 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
3254 caches in the slab allocator. Saves per-node memory,
3255 but will impact performance.
3259 noaltinstr [S390] Disables alternative instructions patching
3260 (CPU alternatives feature).
3262 noapic [SMP,APIC] Tells the kernel to not make use of any
3263 IOAPICs that may be present in the system.
3265 noautogroup Disable scheduler automatic task group creation.
3267 nobats [PPC] Do not use BATs for mapping kernel lowmem
3268 on "Classic" PPC cores.
3272 noclflush [BUGS=X86] Don't use the CLFLUSH instruction
3274 delayacct [KNL] Enable per-task delay accounting
3276 nodsp [SH] Disable hardware DSP at boot time.
3278 noefi Disable EFI runtime services support.
3280 no_entry_flush [PPC] Don't flush the L1-D cache when entering the kernel.
3285 On X86-32 available only on PAE configured kernels.
3286 noexec=on: enable non-executable mappings (default)
3287 noexec=off: disable non-executable mappings
3290 Disable SMAP (Supervisor Mode Access Prevention)
3291 even if it is supported by processor.
3294 Disable SMEP (Supervisor Mode Execution Prevention)
3295 even if it is supported by processor.
3298 This affects only 32-bit executables.
3299 noexec32=on: enable non-executable mappings (default)
3300 read doesn't imply executable mappings
3301 noexec32=off: disable non-executable mappings
3302 read implies executable mappings
3304 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3306 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3307 register save and restore. The kernel will only save
3308 legacy floating-point registers on task switch.
3310 nohugeiomap [KNL,X86,PPC,ARM64] Disable kernel huge I/O mappings.
3312 nohugevmalloc [PPC] Disable kernel huge vmalloc mappings.
3314 nosmt [KNL,S390] Disable symmetric multithreading (SMT).
3315 Equivalent to smt=1.
3317 [KNL,X86] Disable symmetric multithreading (SMT).
3318 nosmt=force: Force disable SMT, cannot be undone
3319 via the sysfs control file.
3321 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1
3322 (bounds check bypass). With this option data leaks are
3323 possible in the system.
3325 nospectre_v2 [X86,PPC_FSL_BOOK3E,ARM64] Disable all mitigations for
3326 the Spectre variant 2 (indirect branch prediction)
3327 vulnerability. System may allow data leaks with this
3330 nospec_store_bypass_disable
3331 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
3334 [PPC] Don't flush the L1-D cache after accessing user data.
3336 noxsave [BUGS=X86] Disables x86 extended register state save
3337 and restore using xsave. The kernel will fallback to
3338 enabling legacy floating-point and sse state.
3340 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
3341 register states. The kernel will fall back to use
3342 xsave to save the states. By using this parameter,
3343 performance of saving the states is degraded because
3344 xsave doesn't support modified optimization while
3345 xsaveopt supports it on xsaveopt enabled systems.
3347 noxsaves [X86] Disables xsaves and xrstors used in saving and
3348 restoring x86 extended register state in compacted
3349 form of xsave area. The kernel will fall back to use
3350 xsaveopt and xrstor to save and restore the states
3351 in standard form of xsave area. By using this
3352 parameter, xsave area per process might occupy more
3353 memory on xsaves enabled systems.
3355 nohlt [ARM,ARM64,MICROBLAZE,SH] Forces the kernel to busy wait
3356 in do_idle() and not use the arch_cpu_idle()
3357 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
3358 to be effective. This is useful on platforms where the
3359 sleep(SH) or wfi(ARM,ARM64) instructions do not work
3360 correctly or when doing power measurements to evalute
3361 the impact of the sleep instructions. This is also
3362 useful when using JTAG debugger.
3364 no_file_caps Tells the kernel not to honor file capabilities. The
3365 only way then for a file to be executed with privilege
3366 is to be setuid root or executed by root.
3368 nohalt [IA-64] Tells the kernel not to use the power saving
3369 function PAL_HALT_LIGHT when idle. This increases
3370 power-consumption. On the positive side, it reduces
3371 interrupt wake-up latency, which may improve performance
3372 in certain environments such as networked servers or
3376 Force pointers printed to the console or buffers to be
3377 unhashed. By default, when a pointer is printed via %p
3378 format string, that pointer is "hashed", i.e. obscured
3379 by hashing the pointer value. This is a security feature
3380 that hides actual kernel addresses from unprivileged
3381 users, but it also makes debugging the kernel more
3382 difficult since unequal pointers can no longer be
3383 compared. However, if this command-line option is
3384 specified, then all normal pointers will have their true
3385 value printed. Pointers printed via %pK may still be
3386 hashed. This option should only be specified when
3387 debugging the kernel. Please do not use on production
3390 nohibernate [HIBERNATION] Disable hibernation and resume.
3392 nohz= [KNL] Boottime enable/disable dynamic ticks
3393 Valid arguments: on, off
3396 nohz_full= [KNL,BOOT,SMP,ISOL]
3397 The argument is a cpu list, as described above.
3398 In kernels built with CONFIG_NO_HZ_FULL=y, set
3399 the specified list of CPUs whose tick will be stopped
3400 whenever possible. The boot CPU will be forced outside
3401 the range to maintain the timekeeping. Any CPUs
3402 in this list will have their RCU callbacks offloaded,
3403 just as if they had also been called out in the
3404 rcu_nocbs= boot parameter.
3406 noiotrap [SH] Disables trapped I/O port accesses.
3408 noirqdebug [X86-32] Disables the code which attempts to detect and
3409 disable unhandled interrupt sources.
3411 no_timer_check [X86,APIC] Disables the code which tests for
3412 broken timer IRQ sources.
3414 noisapnp [ISAPNP] Disables ISA PnP code.
3416 noinitrd [RAM] Tells the kernel not to load any configured
3419 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
3421 [Deprecated - use intremap=off]
3425 noinvpcid [X86] Disable the INVPCID cpu feature.
3427 nojitter [IA-64] Disables jitter checking for ITC timers.
3429 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
3431 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
3435 [X86,PV_OPS] Disable paravirtualized VMware scheduler
3436 clock and use the default one.
3438 no-steal-acc [X86,PV_OPS,ARM64] Disable paravirtualized steal time
3439 accounting. steal time is computed, but won't
3440 influence scheduler behaviour
3442 nolapic [X86-32,APIC] Do not enable or use the local APIC.
3444 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
3446 noltlbs [PPC] Do not use large page/tlb entries for kernel
3447 lowmem mapping on PPC40x and PPC8xx
3449 nomca [IA-64] Disable machine check abort handling
3451 nomce [X86-32] Disable Machine Check Exception
3453 nomfgpt [X86-32] Disable Multi-Function General Purpose
3454 Timer usage (for AMD Geode machines).
3456 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3457 shutdown the other cpus. Instead use the REBOOT_VECTOR
3460 nomodule Disable module load
3462 nopat [X86] Disable PAT (page attribute table extension of
3463 pagetables) support.
3465 nopcid [X86-64] Disable the PCID cpu feature.
3467 norandmaps Don't use address space randomization. Equivalent to
3468 echo 0 > /proc/sys/kernel/randomize_va_space
3470 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
3471 with UP alternatives
3473 nordrand [X86] Disable kernel use of the RDRAND and
3474 RDSEED instructions even if they are supported
3475 by the processor. RDRAND and RDSEED are still
3476 available to user space applications.
3478 noresume [SWSUSP] Disables resume and restores original swap
3481 no-scroll [VGA] Disables scrollback.
3482 This is required for the Braillex ib80-piezo Braille
3483 reader made by F.H. Papenmeier (Germany).
3487 nosep [BUGS=X86-32] Disables x86 SYSENTER/SYSEXIT support.
3489 nosgx [X86-64,SGX] Disables Intel SGX kernel support.
3491 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
3492 and disable the IO APIC. legacy for "maxcpus=0".
3494 nosoftlockup [KNL] Disable the soft-lockup detector.
3496 nosync [HW,M68K] Disables sync negotiation for all devices.
3498 nowatchdog [KNL] Disable both lockup detectors, i.e.
3499 soft-lockup and NMI watchdog (hard-lockup).
3503 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
3505 cpu0_hotplug [X86] Turn on CPU0 hotplug feature when
3506 CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off.
3507 Some features depend on CPU0. Known dependencies are:
3508 1. Resume from suspend/hibernate depends on CPU0.
3509 Suspend/hibernate will fail if CPU0 is offline and you
3510 need to online CPU0 before suspend/hibernate.
3511 2. PIC interrupts also depend on CPU0. CPU0 can't be
3512 removed if a PIC interrupt is detected.
3513 It's said poweroff/reboot may depend on CPU0 on some
3514 machines although I haven't seen such issues so far
3515 after CPU0 is offline on a few tested machines.
3516 If the dependencies are under your control, you can
3517 turn on cpu0_hotplug.
3519 nps_mtm_hs_ctr= [KNL,ARC]
3520 This parameter sets the maximum duration, in
3521 cycles, each HW thread of the CTOP can run
3522 without interruptions, before HW switches it.
3523 The actual maximum duration is 16 times this
3525 Format: integer between 1 and 255
3528 nptcg= [IA-64] Override max number of concurrent global TLB
3529 purges which is reported from either PAL_VM_SUMMARY or
3532 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
3533 could support. nr_cpus=n : n >= 1 limits the kernel to
3534 support 'n' processors. It could be larger than the
3535 number of already plugged CPU during bootup, later in
3536 runtime you can physically add extra cpu until it reaches
3537 n. So during boot up some boot time memory for per-cpu
3538 variables need be pre-allocated for later physical cpu
3541 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
3543 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86] Disable NUMA, Only
3544 set up a single NUMA node spanning all memory.
3546 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic
3548 Allowed values are enable and disable
3550 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
3551 'node', 'default' can be specified
3552 This can be set from sysctl after boot.
3553 See Documentation/admin-guide/sysctl/vm.rst for details.
3555 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
3556 See Documentation/core-api/debugging-via-ohci1394.rst for more
3559 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
3560 Rather than timing out after 20 ms if an EC
3561 command is not properly ACKed, override the length
3562 of the timeout. We have interrupts disabled while
3563 waiting for the ACK, so if this is set too high
3564 interrupts *may* be lost!
3566 omap_mux= [OMAP] Override bootloader pin multiplexing.
3567 Format: <mux_mode0.mode_name=value>...
3568 For example, to override I2C bus2:
3569 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
3571 oops=panic Always panic on oopses. Default is to just kill the
3572 process, but there is a small probability of
3573 deadlocking the machine.
3574 This will also cause panics on machine check exceptions.
3575 Useful together with panic=30 to trigger a reboot.
3578 [KNL] Boolean flag to control whether the page allocator
3579 should randomize its free lists. The randomization may
3580 be automatically enabled if the kernel detects it is
3581 running on a platform with a direct-mapped memory-side
3582 cache, and this parameter can be used to
3583 override/disable that behavior. The state of the flag
3584 can be read from sysfs at:
3585 /sys/module/page_alloc/parameters/shuffle.
3587 page_owner= [KNL] Boot-time page_owner enabling option.
3588 Storage of the information about who allocated
3589 each page is disabled in default. With this switch,
3591 on: enable the feature
3593 page_poison= [KNL] Boot-time parameter changing the state of
3594 poisoning on the buddy allocator, available with
3595 CONFIG_PAGE_POISONING=y.
3596 off: turn off poisoning (default)
3597 on: turn on poisoning
3599 page_reporting.page_reporting_order=
3600 [KNL] Minimal page reporting order
3602 Adjust the minimal page reporting order. The page
3603 reporting is disabled when it exceeds (MAX_ORDER-1).
3605 panic= [KNL] Kernel behaviour on panic: delay <timeout>
3606 timeout > 0: seconds before rebooting
3607 timeout = 0: wait forever
3608 timeout < 0: reboot immediately
3611 panic_print= Bitmask for printing system info when panic happens.
3612 User can chose combination of the following bits:
3613 bit 0: print all tasks info
3614 bit 1: print system memory info
3615 bit 2: print timer info
3616 bit 3: print locks info if CONFIG_LOCKDEP is on
3617 bit 4: print ftrace buffer
3618 bit 5: print all printk messages in buffer
3620 panic_on_taint= Bitmask for conditionally calling panic() in add_taint()
3621 Format: <hex>[,nousertaint]
3622 Hexadecimal bitmask representing the set of TAINT flags
3623 that will cause the kernel to panic when add_taint() is
3624 called with any of the flags in this set.
3625 The optional switch "nousertaint" can be utilized to
3626 prevent userspace forced crashes by writing to sysctl
3627 /proc/sys/kernel/tainted any flagset matching with the
3628 bitmask set on panic_on_taint.
3629 See Documentation/admin-guide/tainted-kernels.rst for
3630 extra details on the taint flags that users can pick
3631 to compose the bitmask to assign to panic_on_taint.
3633 panic_on_warn panic() instead of WARN(). Useful to cause kdump
3636 crash_kexec_post_notifiers
3637 Run kdump after running panic-notifiers and dumping
3638 kmsg. This only for the users who doubt kdump always
3639 succeeds in any situation.
3640 Note that this also increases risks of kdump failure,
3641 because some panic notifiers can make the crashed
3642 kernel more unstable.
3644 parkbd.port= [HW] Parallel port number the keyboard adapter is
3645 connected to, default is 0.
3647 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
3648 0 for XT, 1 for AT (default is AT).
3651 parport= [HW,PPT] Specify parallel ports. 0 disables.
3652 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
3653 Use 'auto' to force the driver to use any
3654 IRQ/DMA settings detected (the default is to
3655 ignore detected IRQ/DMA settings because of
3656 possible conflicts). You can specify the base
3657 address, IRQ, and DMA settings; IRQ and DMA
3658 should be numbers, or 'auto' (for using detected
3659 settings on that particular port), or 'nofifo'
3660 (to avoid using a FIFO even if it is detected).
3661 Parallel ports are assigned in the order they
3662 are specified on the command line, starting
3665 parport_init_mode= [HW,PPT]
3666 Configure VIA parallel port to operate in
3667 a specific mode. This is necessary on Pegasos
3668 computer where firmware has no options for setting
3669 up parallel port mode and sets it to spp.
3670 Currently this function knows 686a and 8231 chips.
3671 Format: [spp|ps2|epp|ecp|ecpepp]
3673 pata_legacy.all= [HW,LIBATA]
3675 Set to non-zero to probe primary and secondary ISA
3676 port ranges on PCI systems where no PCI PATA device
3677 has been found at either range. Disabled by default.
3679 pata_legacy.autospeed= [HW,LIBATA]
3681 Set to non-zero if a chip is present that snoops speed
3682 changes. Disabled by default.
3684 pata_legacy.ht6560a= [HW,LIBATA]
3686 Set to 1, 2, or 3 for HT 6560A on the primary channel,
3687 the secondary channel, or both channels respectively.
3688 Disabled by default.
3690 pata_legacy.ht6560b= [HW,LIBATA]
3692 Set to 1, 2, or 3 for HT 6560B on the primary channel,
3693 the secondary channel, or both channels respectively.
3694 Disabled by default.
3696 pata_legacy.iordy_mask= [HW,LIBATA]
3698 IORDY enable mask. Set individual bits to allow IORDY
3699 for the respective channel. Bit 0 is for the first
3700 legacy channel handled by this driver, bit 1 is for
3701 the second channel, and so on. The sequence will often
3702 correspond to the primary legacy channel, the secondary
3703 legacy channel, and so on, but the handling of a PCI
3704 bus and the use of other driver options may interfere
3705 with the sequence. By default IORDY is allowed across
3708 pata_legacy.opti82c46x= [HW,LIBATA]
3710 Set to 1, 2, or 3 for Opti 82c611A on the primary
3711 channel, the secondary channel, or both channels
3712 respectively. Disabled by default.
3714 pata_legacy.opti82c611a= [HW,LIBATA]
3716 Set to 1, 2, or 3 for Opti 82c465MV on the primary
3717 channel, the secondary channel, or both channels
3718 respectively. Disabled by default.
3720 pata_legacy.pio_mask= [HW,LIBATA]
3722 PIO mode mask for autospeed devices. Set individual
3723 bits to allow the use of the respective PIO modes.
3724 Bit 0 is for mode 0, bit 1 is for mode 1, and so on.
3725 All modes allowed by default.
3727 pata_legacy.probe_all= [HW,LIBATA]
3729 Set to non-zero to probe tertiary and further ISA
3730 port ranges on PCI systems. Disabled by default.
3732 pata_legacy.probe_mask= [HW,LIBATA]
3734 Probe mask for legacy ISA PATA ports. Depending on
3735 platform configuration and the use of other driver
3736 options up to 6 legacy ports are supported: 0x1f0,
3737 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing
3738 of individual ports can be disabled by setting the
3739 corresponding bits in the mask to 1. Bit 0 is for
3740 the first port in the list above (0x1f0), and so on.
3741 By default all supported ports are probed.
3743 pata_legacy.qdi= [HW,LIBATA]
3745 Set to non-zero to probe QDI controllers. By default
3746 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise.
3748 pata_legacy.winbond= [HW,LIBATA]
3750 Set to non-zero to probe Winbond controllers. Use
3751 the standard I/O port (0x130) if 1, otherwise the
3752 value given is the I/O port to use (typically 0x1b0).
3753 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE,
3756 pata_platform.pio_mask= [HW,LIBATA]
3758 Supported PIO mode mask. Set individual bits to allow
3759 the use of the respective PIO modes. Bit 0 is for
3760 mode 0, bit 1 is for mode 1, and so on. Mode 0 only
3764 Halt all CPUs after the first oops has been printed for
3765 the specified number of seconds. This is to be used if
3766 your oopses keep scrolling off the screen.
3771 See header of drivers/block/paride/pcd.c.
3772 See also Documentation/admin-guide/blockdev/paride.rst.
3774 pci=option[,option...] [PCI] various PCI subsystem options.
3776 Some options herein operate on a specific device
3777 or a set of devices (<pci_dev>). These are
3778 specified in one of the following formats:
3780 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
3781 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
3783 Note: the first format specifies a PCI
3784 bus/device/function address which may change
3785 if new hardware is inserted, if motherboard
3786 firmware changes, or due to changes caused
3787 by other kernel parameters. If the
3788 domain is left unspecified, it is
3789 taken to be zero. Optionally, a path
3790 to a device through multiple device/function
3791 addresses can be specified after the base
3792 address (this is more robust against
3793 renumbering issues). The second format
3794 selects devices using IDs from the
3795 configuration space which may match multiple
3796 devices in the system.
3798 earlydump dump PCI config space before the kernel
3800 off [X86] don't probe for the PCI bus
3801 bios [X86-32] force use of PCI BIOS, don't access
3802 the hardware directly. Use this if your machine
3803 has a non-standard PCI host bridge.
3804 nobios [X86-32] disallow use of PCI BIOS, only direct
3805 hardware access methods are allowed. Use this
3806 if you experience crashes upon bootup and you
3807 suspect they are caused by the BIOS.
3808 conf1 [X86] Force use of PCI Configuration Access
3809 Mechanism 1 (config address in IO port 0xCF8,
3810 data in IO port 0xCFC, both 32-bit).
3811 conf2 [X86] Force use of PCI Configuration Access
3812 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
3813 the function, IO port 0xCFA, also 8-bit, sets
3814 bus number. The config space is then accessed
3815 through ports 0xC000-0xCFFF).
3816 See http://wiki.osdev.org/PCI for more info
3817 on the configuration access mechanisms.
3818 noaer [PCIE] If the PCIEAER kernel config parameter is
3819 enabled, this kernel boot option can be used to
3820 disable the use of PCIE advanced error reporting.
3821 nodomains [PCI] Disable support for multiple PCI
3822 root domains (aka PCI segments, in ACPI-speak).
3823 nommconf [X86] Disable use of MMCONFIG for PCI
3825 check_enable_amd_mmconf [X86] check for and enable
3826 properly configured MMIO access to PCI
3827 config space on AMD family 10h CPU
3828 nomsi [MSI] If the PCI_MSI kernel config parameter is
3829 enabled, this kernel boot option can be used to
3830 disable the use of MSI interrupts system-wide.
3831 noioapicquirk [APIC] Disable all boot interrupt quirks.
3832 Safety option to keep boot IRQs enabled. This
3833 should never be necessary.
3834 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
3835 primary IO-APIC for bridges that cannot disable
3836 boot IRQs. This fixes a source of spurious IRQs
3837 when the system masks IRQs.
3838 noioapicreroute [APIC] Disable workaround that uses the
3839 boot IRQ equivalent of an IRQ that connects to
3840 a chipset where boot IRQs cannot be disabled.
3841 The opposite of ioapicreroute.
3842 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
3843 routing table. These calls are known to be buggy
3844 on several machines and they hang the machine
3845 when used, but on other computers it's the only
3846 way to get the interrupt routing table. Try
3847 this option if the kernel is unable to allocate
3848 IRQs or discover secondary PCI buses on your
3850 rom [X86] Assign address space to expansion ROMs.
3851 Use with caution as certain devices share
3852 address decoders between ROMs and other
3854 norom [X86] Do not assign address space to
3855 expansion ROMs that do not already have
3856 BIOS assigned address ranges.
3857 nobar [X86] Do not assign address space to the
3858 BARs that weren't assigned by the BIOS.
3859 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
3860 assigned automatically to PCI devices. You can
3861 make the kernel exclude IRQs of your ISA cards
3863 pirqaddr=0xAAAAA [X86] Specify the physical address
3864 of the PIRQ table (normally generated
3865 by the BIOS) if it is outside the
3866 F0000h-100000h range.
3867 lastbus=N [X86] Scan all buses thru bus #N. Can be
3868 useful if the kernel is unable to find your
3869 secondary buses and you want to tell it
3870 explicitly which ones they are.
3871 assign-busses [X86] Always assign all PCI bus
3872 numbers ourselves, overriding
3873 whatever the firmware may have done.
3874 usepirqmask [X86] Honor the possible IRQ mask stored
3875 in the BIOS $PIR table. This is needed on
3876 some systems with broken BIOSes, notably
3877 some HP Pavilion N5400 and Omnibook XE3
3878 notebooks. This will have no effect if ACPI
3879 IRQ routing is enabled.
3880 noacpi [X86] Do not use ACPI for IRQ routing
3881 or for PCI scanning.
3882 use_crs [X86] Use PCI host bridge window information
3883 from ACPI. On BIOSes from 2008 or later, this
3884 is enabled by default. If you need to use this,
3885 please report a bug.
3886 nocrs [X86] Ignore PCI host bridge windows from ACPI.
3887 If you need to use this, please report a bug.
3888 routeirq Do IRQ routing for all PCI devices.
3889 This is normally done in pci_enable_device(),
3890 so this option is a temporary workaround
3891 for broken drivers that don't call it.
3892 skip_isa_align [X86] do not align io start addr, so can
3893 handle more pci cards
3894 noearly [X86] Don't do any early type 1 scanning.
3895 This might help on some broken boards which
3896 machine check when some devices' config space
3897 is read. But various workarounds are disabled
3898 and some IOMMU drivers will not work.
3899 bfsort Sort PCI devices into breadth-first order.
3900 This sorting is done to get a device
3901 order compatible with older (<= 2.4) kernels.
3902 nobfsort Don't sort PCI devices into breadth-first order.
3903 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
3904 tuning and use the BIOS-configured MPS defaults.
3905 pcie_bus_safe Set every device's MPS to the largest value
3906 supported by all devices below the root complex.
3907 pcie_bus_perf Set device MPS to the largest allowable MPS
3908 based on its parent bus. Also set MRRS (Max
3909 Read Request Size) to the largest supported
3910 value (no larger than the MPS that the device
3911 or bus can support) for best performance.
3912 pcie_bus_peer2peer Set every device's MPS to 128B, which
3913 every device is guaranteed to support. This
3914 configuration allows peer-to-peer DMA between
3915 any pair of devices, possibly at the cost of
3916 reduced performance. This also guarantees
3917 that hot-added devices will work.
3918 cbiosize=nn[KMG] The fixed amount of bus space which is
3919 reserved for the CardBus bridge's IO window.
3920 The default value is 256 bytes.
3921 cbmemsize=nn[KMG] The fixed amount of bus space which is
3922 reserved for the CardBus bridge's memory
3923 window. The default value is 64 megabytes.
3926 [<order of align>@]<pci_dev>[; ...]
3927 Specifies alignment and device to reassign
3928 aligned memory resources. How to
3929 specify the device is described above.
3930 If <order of align> is not specified,
3931 PAGE_SIZE is used as alignment.
3932 A PCI-PCI bridge can be specified if resource
3933 windows need to be expanded.
3934 To specify the alignment for several
3935 instances of a device, the PCI vendor,
3936 device, subvendor, and subdevice may be
3937 specified, e.g., 12@pci:8086:9c22:103c:198f
3938 for 4096-byte alignment.
3939 ecrc= Enable/disable PCIe ECRC (transaction layer
3940 end-to-end CRC checking).
3941 bios: Use BIOS/firmware settings. This is the
3945 hpiosize=nn[KMG] The fixed amount of bus space which is
3946 reserved for hotplug bridge's IO window.
3947 Default size is 256 bytes.
3948 hpmmiosize=nn[KMG] The fixed amount of bus space which is
3949 reserved for hotplug bridge's MMIO window.
3950 Default size is 2 megabytes.
3951 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
3952 reserved for hotplug bridge's MMIO_PREF window.
3953 Default size is 2 megabytes.
3954 hpmemsize=nn[KMG] The fixed amount of bus space which is
3955 reserved for hotplug bridge's MMIO and
3957 Default size is 2 megabytes.
3958 hpbussize=nn The minimum amount of additional bus numbers
3959 reserved for buses below a hotplug bridge.
3961 realloc= Enable/disable reallocating PCI bridge resources
3962 if allocations done by BIOS are too small to
3963 accommodate resources required by all child
3965 off: Turn realloc off
3967 realloc same as realloc=on
3968 noari do not use PCIe ARI.
3969 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
3970 do not use PCIe ATS (and IOMMU device IOTLB).
3971 pcie_scan_all Scan all possible PCIe devices. Otherwise we
3972 only look for one device below a PCIe downstream
3974 big_root_window Try to add a big 64bit memory window to the PCIe
3975 root complex on AMD CPUs. Some GFX hardware
3976 can resize a BAR to allow access to all VRAM.
3977 Adding the window is slightly risky (it may
3978 conflict with unreported devices), so this
3980 disable_acs_redir=<pci_dev>[; ...]
3981 Specify one or more PCI devices (in the format
3982 specified above) separated by semicolons.
3983 Each device specified will have the PCI ACS
3984 redirect capabilities forced off which will
3985 allow P2P traffic between devices through
3986 bridges without forcing it upstream. Note:
3987 this removes isolation between devices and
3988 may put more devices in an IOMMU group.
3989 force_floating [S390] Force usage of floating interrupts.
3990 nomio [S390] Do not use MIO instructions.
3991 norid [S390] ignore the RID field and force use of
3992 one PCI domain per PCI function
3994 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
3997 force Enable ASPM even on devices that claim not to support it.
3998 WARNING: Forcing ASPM on may cause system lockups.
4000 pcie_ports= [PCIE] PCIe port services handling:
4001 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
4002 even if the platform doesn't give the OS permission to
4003 use them. This may cause conflicts if the platform
4004 also tries to use these services.
4005 dpc-native Use native PCIe service for DPC only. May
4006 cause conflicts if firmware uses AER or DPC.
4007 compat Disable native PCIe services (PME, AER, DPC, PCIe
4010 pcie_port_pm= [PCIE] PCIe port power management handling:
4011 off Disable power management of all PCIe ports
4012 force Forcibly enable power management of all PCIe ports
4014 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
4015 nomsi Do not use MSI for native PCIe PME signaling (this makes
4016 all PCIe root ports use INTx for all services).
4018 pcmv= [HW,PCMCIA] BadgePAD 4
4022 Keep all power-domains already enabled by bootloader on,
4023 even if no driver has claimed them. This is useful
4024 for debug and development, but should not be
4025 needed on a platform with proper driver support.
4028 See Documentation/admin-guide/blockdev/paride.rst.
4030 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
4033 See arch/parisc/kernel/pdc_chassis.c
4035 percpu_alloc= Select which percpu first chunk allocator to use.
4036 Currently supported values are "embed" and "page".
4037 Archs may support subset or none of the selections.
4038 See comments in mm/percpu.c for details on each
4039 allocator. This parameter is primarily for debugging
4040 and performance comparison.
4043 See Documentation/admin-guide/blockdev/paride.rst.
4046 See Documentation/admin-guide/blockdev/paride.rst.
4048 pirq= [SMP,APIC] Manual mp-table setup
4049 See Documentation/x86/i386/IO-APIC.rst.
4051 plip= [PPT,NET] Parallel port network link
4052 Format: { parport<nr> | timid | 0 }
4053 See also Documentation/admin-guide/parport.rst.
4055 pmtmr= [X86] Manual setup of pmtmr I/O Port.
4056 Override pmtimer IOPort with a hex value.
4059 pm_debug_messages [SUSPEND,KNL]
4060 Enable suspend/resume debug messages during boot up.
4063 Enable PNP debug messages (depends on the
4064 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
4065 via /sys/module/pnp/parameters/debug. We always show
4066 current resource usage; turning this on also shows
4067 possible settings and some assignment information.
4073 { on | off | curr | res | no-curr | no-res }
4076 [ISAPNP] Exclude IRQs for the autoconfiguration
4079 [ISAPNP] Exclude DMAs for the autoconfiguration
4081 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
4082 Ranges are in pairs (I/O port base and size).
4085 [ISAPNP] Exclude memory regions for the
4087 Ranges are in pairs (memory base and size).
4089 ports= [IP_VS_FTP] IPVS ftp helper module
4091 Up to 8 (IP_VS_APP_MAX_PORTS) ports
4093 Format: <port>,<port>....
4095 powersave=off [PPC] This option disables power saving features.
4096 It specifically disables cpuidle and sets the
4097 platform machine description specific power_save
4098 function to NULL. On Idle the CPU just reduces
4101 ppc_strict_facility_enable
4102 [PPC] This option catches any kernel floating point,
4103 Altivec, VSX and SPE outside of regions specifically
4104 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
4105 There is some performance impact when enabling this.
4109 Disable Hardware Transactional Memory
4112 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC
4113 none - Limited to cond_resched() calls
4114 voluntary - Limited to cond_resched() and might_sleep() calls
4115 full - Any section that isn't explicitly preempt disabled
4116 can be preempted anytime.
4118 print-fatal-signals=
4119 [KNL] debug: print fatal signals
4121 If enabled, warn about various signal handling
4122 related application anomalies: too many signals,
4123 too many POSIX.1 timers, fatal signals causing a
4126 If you hit the warning due to signal overflow,
4127 you might want to try "ulimit -i unlimited".
4131 printk.always_kmsg_dump=
4132 Trigger kmsg_dump for cases other than kernel oops or
4134 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4137 printk.devkmsg={on,off,ratelimit}
4138 Control writing to /dev/kmsg.
4139 on - unlimited logging to /dev/kmsg from userspace
4140 off - logging to /dev/kmsg disabled
4141 ratelimit - ratelimit the logging
4144 printk.time= Show timing data prefixed to each printk message line
4145 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4147 processor.max_cstate= [HW,ACPI]
4148 Limit processor to maximum C-state
4149 max_cstate=9 overrides any DMI blacklist limit.
4151 processor.nocst [HW,ACPI]
4152 Ignore the _CST method to determine C-states,
4153 instead using the legacy FADT method
4155 profile= [KNL] Enable kernel profiling via /proc/profile
4156 Format: [<profiletype>,]<number>
4157 Param: <profiletype>: "schedule", "sleep", or "kvm"
4158 [defaults to kernel profiling]
4159 Param: "schedule" - profile schedule points.
4160 Param: "sleep" - profile D-state sleeping (millisecs).
4161 Requires CONFIG_SCHEDSTATS
4162 Param: "kvm" - profile VM exits.
4163 Param: <number> - step/bucket size as a power of 2 for
4164 statistical time based profiling.
4166 prompt_ramdisk= [RAM] [Deprecated]
4168 prot_virt= [S390] enable hosting protected virtual machines
4169 isolated from the hypervisor (if hardware supports
4173 psi= [KNL] Enable or disable pressure stall information
4177 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
4178 probe for; one of (bare|imps|exps|lifebook|any).
4179 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
4181 psmouse.resetafter= [HW,MOUSE]
4182 Try to reset the device after so many bad packets
4185 [HW,MOUSE] Set desired mouse resolution, in dpi.
4186 psmouse.smartscroll=
4187 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
4188 0 = disabled, 1 = enabled (default).
4190 pstore.backend= Specify the name of the pstore backend to use
4193 See Documentation/admin-guide/blockdev/paride.rst.
4195 pti= [X86-64] Control Page Table Isolation of user and
4196 kernel address spaces. Disabling this feature
4197 removes hardening, but improves performance of
4198 system calls and interrupts.
4200 on - unconditionally enable
4201 off - unconditionally disable
4202 auto - kernel detects whether your CPU model is
4203 vulnerable to issues that PTI mitigates
4205 Not specifying this option is equivalent to pti=auto.
4208 Equivalent to pti=off
4211 [KNL] Number of legacy pty's. Overwrites compiled-in
4214 quiet [KNL] Disable most log messages
4219 See Documentation/admin-guide/md.rst.
4221 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
4222 See Documentation/admin-guide/blockdev/ramdisk.rst.
4224 ramdisk_start= [RAM] RAM disk image start address
4226 random.trust_cpu={on,off}
4227 [KNL] Enable or disable trusting the use of the
4228 CPU's random number generator (if available) to
4229 fully seed the kernel's CRNG. Default is controlled
4230 by CONFIG_RANDOM_TRUST_CPU.
4232 randomize_kstack_offset=
4233 [KNL] Enable or disable kernel stack offset
4234 randomization, which provides roughly 5 bits of
4235 entropy, frustrating memory corruption attacks
4236 that depend on stack address determinism or
4237 cross-syscall address exposures. This is only
4238 available on architectures that have defined
4239 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET.
4240 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4241 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT.
4243 ras=option[,option,...] [KNL] RAS-specific options
4246 Disable the Correctable Errors Collector,
4247 see CONFIG_RAS_CEC help text.
4250 The argument is a cpu list, as described above.
4252 In kernels built with CONFIG_RCU_NOCB_CPU=y, set
4253 the specified list of CPUs to be no-callback CPUs.
4254 Invocation of these CPUs' RCU callbacks will be
4255 offloaded to "rcuox/N" kthreads created for that
4256 purpose, where "x" is "p" for RCU-preempt, and
4257 "s" for RCU-sched, and "N" is the CPU number.
4258 This reduces OS jitter on the offloaded CPUs,
4259 which can be useful for HPC and real-time
4260 workloads. It can also improve energy efficiency
4261 for asymmetric multiprocessors.
4264 Rather than requiring that offloaded CPUs
4265 (specified by rcu_nocbs= above) explicitly
4266 awaken the corresponding "rcuoN" kthreads,
4267 make these kthreads poll for callbacks.
4268 This improves the real-time response for the
4269 offloaded CPUs by relieving them of the need to
4270 wake up the corresponding kthread, but degrades
4271 energy efficiency by requiring that the kthreads
4272 periodically wake up to do the polling.
4274 rcutree.blimit= [KNL]
4275 Set maximum number of finished RCU callbacks to
4276 process in one batch.
4278 rcutree.dump_tree= [KNL]
4279 Dump the structure of the rcu_node combining tree
4280 out at early boot. This is used for diagnostic
4281 purposes, to verify correct tree setup.
4283 rcutree.gp_cleanup_delay= [KNL]
4284 Set the number of jiffies to delay each step of
4285 RCU grace-period cleanup.
4287 rcutree.gp_init_delay= [KNL]
4288 Set the number of jiffies to delay each step of
4289 RCU grace-period initialization.
4291 rcutree.gp_preinit_delay= [KNL]
4292 Set the number of jiffies to delay each step of
4293 RCU grace-period pre-initialization, that is,
4294 the propagation of recent CPU-hotplug changes up
4295 the rcu_node combining tree.
4297 rcutree.use_softirq= [KNL]
4298 If set to zero, move all RCU_SOFTIRQ processing to
4299 per-CPU rcuc kthreads. Defaults to a non-zero
4300 value, meaning that RCU_SOFTIRQ is used by default.
4301 Specify rcutree.use_softirq=0 to use rcuc kthreads.
4303 But note that CONFIG_PREEMPT_RT=y kernels disable
4304 this kernel boot parameter, forcibly setting it
4307 rcutree.rcu_fanout_exact= [KNL]
4308 Disable autobalancing of the rcu_node combining
4309 tree. This is used by rcutorture, and might
4310 possibly be useful for architectures having high
4311 cache-to-cache transfer latencies.
4313 rcutree.rcu_fanout_leaf= [KNL]
4314 Change the number of CPUs assigned to each
4315 leaf rcu_node structure. Useful for very
4316 large systems, which will choose the value 64,
4317 and for NUMA systems with large remote-access
4318 latencies, which will choose a value aligned
4319 with the appropriate hardware boundaries.
4321 rcutree.rcu_min_cached_objs= [KNL]
4322 Minimum number of objects which are cached and
4323 maintained per one CPU. Object size is equal
4324 to PAGE_SIZE. The cache allows to reduce the
4325 pressure to page allocator, also it makes the
4326 whole algorithm to behave better in low memory
4329 rcutree.jiffies_till_first_fqs= [KNL]
4330 Set delay from grace-period initialization to
4331 first attempt to force quiescent states.
4332 Units are jiffies, minimum value is zero,
4333 and maximum value is HZ.
4335 rcutree.jiffies_till_next_fqs= [KNL]
4336 Set delay between subsequent attempts to force
4337 quiescent states. Units are jiffies, minimum
4338 value is one, and maximum value is HZ.
4340 rcutree.jiffies_till_sched_qs= [KNL]
4341 Set required age in jiffies for a
4342 given grace period before RCU starts
4343 soliciting quiescent-state help from
4344 rcu_note_context_switch() and cond_resched().
4345 If not specified, the kernel will calculate
4346 a value based on the most recent settings
4347 of rcutree.jiffies_till_first_fqs
4348 and rcutree.jiffies_till_next_fqs.
4349 This calculated value may be viewed in
4350 rcutree.jiffies_to_sched_qs. Any attempt to set
4351 rcutree.jiffies_to_sched_qs will be cheerfully
4354 rcutree.kthread_prio= [KNL,BOOT]
4355 Set the SCHED_FIFO priority of the RCU per-CPU
4356 kthreads (rcuc/N). This value is also used for
4357 the priority of the RCU boost threads (rcub/N)
4358 and for the RCU grace-period kthreads (rcu_bh,
4359 rcu_preempt, and rcu_sched). If RCU_BOOST is
4360 set, valid values are 1-99 and the default is 1
4361 (the least-favored priority). Otherwise, when
4362 RCU_BOOST is not set, valid values are 0-99 and
4363 the default is zero (non-realtime operation).
4365 rcutree.rcu_nocb_gp_stride= [KNL]
4366 Set the number of NOCB callback kthreads in
4367 each group, which defaults to the square root
4368 of the number of CPUs. Larger numbers reduce
4369 the wakeup overhead on the global grace-period
4370 kthread, but increases that same overhead on
4371 each group's NOCB grace-period kthread.
4373 rcutree.qhimark= [KNL]
4374 Set threshold of queued RCU callbacks beyond which
4375 batch limiting is disabled.
4377 rcutree.qlowmark= [KNL]
4378 Set threshold of queued RCU callbacks below which
4379 batch limiting is re-enabled.
4381 rcutree.qovld= [KNL]
4382 Set threshold of queued RCU callbacks beyond which
4383 RCU's force-quiescent-state scan will aggressively
4384 enlist help from cond_resched() and sched IPIs to
4385 help CPUs more quickly reach quiescent states.
4386 Set to less than zero to make this be set based
4387 on rcutree.qhimark at boot time and to zero to
4388 disable more aggressive help enlistment.
4390 rcutree.rcu_idle_gp_delay= [KNL]
4391 Set wakeup interval for idle CPUs that have
4392 RCU callbacks (RCU_FAST_NO_HZ=y).
4394 rcutree.rcu_kick_kthreads= [KNL]
4395 Cause the grace-period kthread to get an extra
4396 wake_up() if it sleeps three times longer than
4397 it should at force-quiescent-state time.
4398 This wake_up() will be accompanied by a
4399 WARN_ONCE() splat and an ftrace_dump().
4401 rcutree.rcu_unlock_delay= [KNL]
4402 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels,
4403 this specifies an rcu_read_unlock()-time delay
4404 in microseconds. This defaults to zero.
4405 Larger delays increase the probability of
4406 catching RCU pointer leaks, that is, buggy use
4407 of RCU-protected pointers after the relevant
4408 rcu_read_unlock() has completed.
4410 rcutree.sysrq_rcu= [KNL]
4411 Commandeer a sysrq key to dump out Tree RCU's
4412 rcu_node tree with an eye towards determining
4413 why a new grace period has not yet started.
4415 rcuscale.gp_async= [KNL]
4416 Measure performance of asynchronous
4417 grace-period primitives such as call_rcu().
4419 rcuscale.gp_async_max= [KNL]
4420 Specify the maximum number of outstanding
4421 callbacks per writer thread. When a writer
4422 thread exceeds this limit, it invokes the
4423 corresponding flavor of rcu_barrier() to allow
4424 previously posted callbacks to drain.
4426 rcuscale.gp_exp= [KNL]
4427 Measure performance of expedited synchronous
4428 grace-period primitives.
4430 rcuscale.holdoff= [KNL]
4431 Set test-start holdoff period. The purpose of
4432 this parameter is to delay the start of the
4433 test until boot completes in order to avoid
4436 rcuscale.kfree_rcu_test= [KNL]
4437 Set to measure performance of kfree_rcu() flooding.
4439 rcuscale.kfree_rcu_test_double= [KNL]
4440 Test the double-argument variant of kfree_rcu().
4441 If this parameter has the same value as
4442 rcuscale.kfree_rcu_test_single, both the single-
4443 and double-argument variants are tested.
4445 rcuscale.kfree_rcu_test_single= [KNL]
4446 Test the single-argument variant of kfree_rcu().
4447 If this parameter has the same value as
4448 rcuscale.kfree_rcu_test_double, both the single-
4449 and double-argument variants are tested.
4451 rcuscale.kfree_nthreads= [KNL]
4452 The number of threads running loops of kfree_rcu().
4454 rcuscale.kfree_alloc_num= [KNL]
4455 Number of allocations and frees done in an iteration.
4457 rcuscale.kfree_loops= [KNL]
4458 Number of loops doing rcuscale.kfree_alloc_num number
4459 of allocations and frees.
4461 rcuscale.nreaders= [KNL]
4462 Set number of RCU readers. The value -1 selects
4463 N, where N is the number of CPUs. A value
4464 "n" less than -1 selects N-n+1, where N is again
4465 the number of CPUs. For example, -2 selects N
4466 (the number of CPUs), -3 selects N+1, and so on.
4467 A value of "n" less than or equal to -N selects
4470 rcuscale.nwriters= [KNL]
4471 Set number of RCU writers. The values operate
4472 the same as for rcuscale.nreaders.
4473 N, where N is the number of CPUs
4475 rcuscale.perf_type= [KNL]
4476 Specify the RCU implementation to test.
4478 rcuscale.shutdown= [KNL]
4479 Shut the system down after performance tests
4480 complete. This is useful for hands-off automated
4483 rcuscale.verbose= [KNL]
4484 Enable additional printk() statements.
4486 rcuscale.writer_holdoff= [KNL]
4487 Write-side holdoff between grace periods,
4488 in microseconds. The default of zero says
4491 rcutorture.fqs_duration= [KNL]
4492 Set duration of force_quiescent_state bursts
4495 rcutorture.fqs_holdoff= [KNL]
4496 Set holdoff time within force_quiescent_state bursts
4499 rcutorture.fqs_stutter= [KNL]
4500 Set wait time between force_quiescent_state bursts
4503 rcutorture.fwd_progress= [KNL]
4504 Enable RCU grace-period forward-progress testing
4505 for the types of RCU supporting this notion.
4507 rcutorture.fwd_progress_div= [KNL]
4508 Specify the fraction of a CPU-stall-warning
4509 period to do tight-loop forward-progress testing.
4511 rcutorture.fwd_progress_holdoff= [KNL]
4512 Number of seconds to wait between successive
4513 forward-progress tests.
4515 rcutorture.fwd_progress_need_resched= [KNL]
4516 Enclose cond_resched() calls within checks for
4517 need_resched() during tight-loop forward-progress
4520 rcutorture.gp_cond= [KNL]
4521 Use conditional/asynchronous update-side
4522 primitives, if available.
4524 rcutorture.gp_exp= [KNL]
4525 Use expedited update-side primitives, if available.
4527 rcutorture.gp_normal= [KNL]
4528 Use normal (non-expedited) asynchronous
4529 update-side primitives, if available.
4531 rcutorture.gp_sync= [KNL]
4532 Use normal (non-expedited) synchronous
4533 update-side primitives, if available. If all
4534 of rcutorture.gp_cond=, rcutorture.gp_exp=,
4535 rcutorture.gp_normal=, and rcutorture.gp_sync=
4536 are zero, rcutorture acts as if is interpreted
4537 they are all non-zero.
4539 rcutorture.irqreader= [KNL]
4540 Run RCU readers from irq handlers, or, more
4541 accurately, from a timer handler. Not all RCU
4542 flavors take kindly to this sort of thing.
4544 rcutorture.leakpointer= [KNL]
4545 Leak an RCU-protected pointer out of the reader.
4546 This can of course result in splats, and is
4547 intended to test the ability of things like
4548 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect
4551 rcutorture.n_barrier_cbs= [KNL]
4552 Set callbacks/threads for rcu_barrier() testing.
4554 rcutorture.nfakewriters= [KNL]
4555 Set number of concurrent RCU writers. These just
4556 stress RCU, they don't participate in the actual
4557 test, hence the "fake".
4559 rcutorture.nocbs_nthreads= [KNL]
4560 Set number of RCU callback-offload togglers.
4561 Zero (the default) disables toggling.
4563 rcutorture.nocbs_toggle= [KNL]
4564 Set the delay in milliseconds between successive
4565 callback-offload toggling attempts.
4567 rcutorture.nreaders= [KNL]
4568 Set number of RCU readers. The value -1 selects
4569 N-1, where N is the number of CPUs. A value
4570 "n" less than -1 selects N-n-2, where N is again
4571 the number of CPUs. For example, -2 selects N
4572 (the number of CPUs), -3 selects N+1, and so on.
4574 rcutorture.object_debug= [KNL]
4575 Enable debug-object double-call_rcu() testing.
4577 rcutorture.onoff_holdoff= [KNL]
4578 Set time (s) after boot for CPU-hotplug testing.
4580 rcutorture.onoff_interval= [KNL]
4581 Set time (jiffies) between CPU-hotplug operations,
4582 or zero to disable CPU-hotplug testing.
4584 rcutorture.read_exit= [KNL]
4585 Set the number of read-then-exit kthreads used
4586 to test the interaction of RCU updaters and
4587 task-exit processing.
4589 rcutorture.read_exit_burst= [KNL]
4590 The number of times in a given read-then-exit
4591 episode that a set of read-then-exit kthreads
4594 rcutorture.read_exit_delay= [KNL]
4595 The delay, in seconds, between successive
4596 read-then-exit testing episodes.
4598 rcutorture.shuffle_interval= [KNL]
4599 Set task-shuffle interval (s). Shuffling tasks
4600 allows some CPUs to go into dyntick-idle mode
4601 during the rcutorture test.
4603 rcutorture.shutdown_secs= [KNL]
4604 Set time (s) after boot system shutdown. This
4605 is useful for hands-off automated testing.
4607 rcutorture.stall_cpu= [KNL]
4608 Duration of CPU stall (s) to test RCU CPU stall
4609 warnings, zero to disable.
4611 rcutorture.stall_cpu_block= [KNL]
4612 Sleep while stalling if set. This will result
4613 in warnings from preemptible RCU in addition
4614 to any other stall-related activity.
4616 rcutorture.stall_cpu_holdoff= [KNL]
4617 Time to wait (s) after boot before inducing stall.
4619 rcutorture.stall_cpu_irqsoff= [KNL]
4620 Disable interrupts while stalling if set.
4622 rcutorture.stall_gp_kthread= [KNL]
4623 Duration (s) of forced sleep within RCU
4624 grace-period kthread to test RCU CPU stall
4625 warnings, zero to disable. If both stall_cpu
4626 and stall_gp_kthread are specified, the
4627 kthread is starved first, then the CPU.
4629 rcutorture.stat_interval= [KNL]
4630 Time (s) between statistics printk()s.
4632 rcutorture.stutter= [KNL]
4633 Time (s) to stutter testing, for example, specifying
4634 five seconds causes the test to run for five seconds,
4635 wait for five seconds, and so on. This tests RCU's
4636 ability to transition abruptly to and from idle.
4638 rcutorture.test_boost= [KNL]
4639 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
4640 "Maybe" means test if the RCU implementation
4641 under test support RCU priority boosting.
4643 rcutorture.test_boost_duration= [KNL]
4644 Duration (s) of each individual boost test.
4646 rcutorture.test_boost_interval= [KNL]
4647 Interval (s) between each boost test.
4649 rcutorture.test_no_idle_hz= [KNL]
4650 Test RCU's dyntick-idle handling. See also the
4651 rcutorture.shuffle_interval parameter.
4653 rcutorture.torture_type= [KNL]
4654 Specify the RCU implementation to test.
4656 rcutorture.verbose= [KNL]
4657 Enable additional printk() statements.
4659 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
4660 Dump ftrace buffer after reporting RCU CPU
4663 rcupdate.rcu_cpu_stall_suppress= [KNL]
4664 Suppress RCU CPU stall warning messages.
4666 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL]
4667 Suppress RCU CPU stall warning messages and
4668 rcutorture writer stall warnings that occur
4669 during early boot, that is, during the time
4670 before the init task is spawned.
4672 rcupdate.rcu_cpu_stall_timeout= [KNL]
4673 Set timeout for RCU CPU stall warning messages.
4675 rcupdate.rcu_expedited= [KNL]
4676 Use expedited grace-period primitives, for
4677 example, synchronize_rcu_expedited() instead
4678 of synchronize_rcu(). This reduces latency,
4679 but can increase CPU utilization, degrade
4680 real-time latency, and degrade energy efficiency.
4681 No effect on CONFIG_TINY_RCU kernels.
4683 rcupdate.rcu_normal= [KNL]
4684 Use only normal grace-period primitives,
4685 for example, synchronize_rcu() instead of
4686 synchronize_rcu_expedited(). This improves
4687 real-time latency, CPU utilization, and
4688 energy efficiency, but can expose users to
4689 increased grace-period latency. This parameter
4690 overrides rcupdate.rcu_expedited. No effect on
4691 CONFIG_TINY_RCU kernels.
4693 rcupdate.rcu_normal_after_boot= [KNL]
4694 Once boot has completed (that is, after
4695 rcu_end_inkernel_boot() has been invoked), use
4696 only normal grace-period primitives. No effect
4697 on CONFIG_TINY_RCU kernels.
4699 But note that CONFIG_PREEMPT_RT=y kernels enables
4700 this kernel boot parameter, forcibly setting
4701 it to the value one, that is, converting any
4702 post-boot attempt at an expedited RCU grace
4703 period to instead use normal non-expedited
4704 grace-period processing.
4706 rcupdate.rcu_task_ipi_delay= [KNL]
4707 Set time in jiffies during which RCU tasks will
4708 avoid sending IPIs, starting with the beginning
4709 of a given grace period. Setting a large
4710 number avoids disturbing real-time workloads,
4711 but lengthens grace periods.
4713 rcupdate.rcu_task_stall_timeout= [KNL]
4714 Set timeout in jiffies for RCU task stall warning
4715 messages. Disable with a value less than or equal
4718 rcupdate.rcu_self_test= [KNL]
4719 Run the RCU early boot self tests
4723 Run specified binary instead of /init from the ramdisk,
4724 used for early userspace startup. See initrd.
4727 force - Override the decision by the kernel to hide the
4728 advertisement of RDRAND support (this affects
4729 certain AMD processors because of buggy BIOS
4730 support, specifically around the suspend/resume
4734 Turn on/off individual RDT features. List is:
4735 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
4737 E.g. to turn on cmt and turn off mba use:
4741 Format (x86 or x86_64):
4742 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] \
4744 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
4746 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
4747 (prefix with 'panic_' to set mode for panic
4749 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
4750 reboot_force is either force or not specified,
4751 reboot_cpu is s[mp]#### with #### being the processor
4752 to be used for rebooting.
4754 refscale.holdoff= [KNL]
4755 Set test-start holdoff period. The purpose of
4756 this parameter is to delay the start of the
4757 test until boot completes in order to avoid
4760 refscale.loops= [KNL]
4761 Set the number of loops over the synchronization
4762 primitive under test. Increasing this number
4763 reduces noise due to loop start/end overhead,
4764 but the default has already reduced the per-pass
4765 noise to a handful of picoseconds on ca. 2020
4768 refscale.nreaders= [KNL]
4769 Set number of readers. The default value of -1
4770 selects N, where N is roughly 75% of the number
4771 of CPUs. A value of zero is an interesting choice.
4773 refscale.nruns= [KNL]
4774 Set number of runs, each of which is dumped onto
4777 refscale.readdelay= [KNL]
4778 Set the read-side critical-section duration,
4779 measured in microseconds.
4781 refscale.scale_type= [KNL]
4782 Specify the read-protection implementation to test.
4784 refscale.shutdown= [KNL]
4785 Shut down the system at the end of the performance
4786 test. This defaults to 1 (shut it down) when
4787 refscale is built into the kernel and to 0 (leave
4788 it running) when refscale is built as a module.
4790 refscale.verbose= [KNL]
4791 Enable additional printk() statements.
4793 refscale.verbose_batched= [KNL]
4794 Batch the additional printk() statements. If zero
4795 (the default) or negative, print everything. Otherwise,
4796 print every Nth verbose statement, where N is the value
4800 [KNL, SMP] Set scheduler's default relax_domain_level.
4801 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
4803 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
4804 Format: <base1>,<size1>[,<base2>,<size2>,...]
4805 Reserve I/O ports or memory so the kernel won't use
4806 them. If <base> is less than 0x10000, the region
4807 is assumed to be I/O ports; otherwise it is memory.
4809 reservetop= [X86-32]
4811 Reserves a hole at the top of the kernel virtual
4814 reset_devices [KNL] Force drivers to reset the underlying device
4815 during initialization.
4818 Specify the partition device for software suspend
4820 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
4822 resume_offset= [SWSUSP]
4823 Specify the offset from the beginning of the partition
4824 given by "resume=" at which the swap header is located,
4825 in <PAGE_SIZE> units (needed only for swap files).
4826 See Documentation/power/swsusp-and-swap-files.rst
4828 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
4829 read the resume files
4831 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
4832 Useful for devices that are detected asynchronously
4833 (e.g. USB and MMC devices).
4835 hibernate= [HIBERNATION]
4836 noresume Don't check if there's a hibernation image
4837 present during boot.
4838 nocompress Don't compress/decompress hibernation images.
4839 no Disable hibernation and resume.
4840 protect_image Turn on image protection during restoration
4841 (that will set all pages holding image data
4842 during restoration read-only).
4844 retain_initrd [RAM] Keep initrd memory after extraction
4846 rfkill.default_state=
4847 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
4848 etc. communication is blocked by default.
4851 rfkill.master_switch_mode=
4852 0 The "airplane mode" button does nothing.
4853 1 The "airplane mode" button toggles between everything
4854 blocked and the previous configuration.
4855 2 The "airplane mode" button toggles between everything
4856 blocked and everything unblocked.
4858 rhash_entries= [KNL,NET]
4859 Set number of hash buckets for route cache
4862 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
4865 ro [KNL] Mount root device read-only on boot
4868 on Mark read-only kernel memory as read-only (default).
4869 off Leave read-only kernel memory writable for debugging.
4872 Enable the uart passthrough on the designated usb port
4873 on Rockchip SoCs. When active, the signals of the
4874 debug-uart get routed to the D+ and D- pins of the usb
4875 port and the regular usb controller gets disabled.
4877 root= [KNL] Root filesystem
4878 See name_to_dev_t comment in init/do_mounts.c.
4880 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
4881 mount the root filesystem
4883 rootflags= [KNL] Set root filesystem mount option string
4885 rootfstype= [KNL] Set root filesystem type
4887 rootwait [KNL] Wait (indefinitely) for root device to show up.
4888 Useful for devices that are detected asynchronously
4889 (e.g. USB and MMC devices).
4891 rproc_mem=nn[KMG][@address]
4892 [KNL,ARM,CMA] Remoteproc physical memory block.
4893 Memory area to be used by remote processor image,
4896 rw [KNL] Mount root device read-write on boot
4898 S [KNL] Run init in single mode
4900 s390_iommu= [HW,S390]
4901 Set s390 IOTLB flushing mode
4903 With strict flushing every unmap operation will result in
4904 an IOTLB flush. Default is lazy flushing before reuse,
4908 See drivers/net/irda/sa1100_ir.c.
4910 sbni= [NET] Granch SBNI12 leased line adapter
4912 sched_verbose [KNL] Enables verbose scheduler debug messages.
4914 schedstats= [KNL,X86] Enable or disable scheduled statistics.
4915 Allowed values are enable and disable. This feature
4916 incurs a small amount of overhead in the scheduler
4917 but is useful for debugging and performance tuning.
4919 sched_thermal_decay_shift=
4920 [KNL, SMP] Set a decay shift for scheduler thermal
4921 pressure signal. Thermal pressure signal follows the
4922 default decay period of other scheduler pelt
4923 signals(usually 32 ms but configurable). Setting
4924 sched_thermal_decay_shift will left shift the decay
4925 period for the thermal pressure signal by the shift
4927 i.e. with the default pelt decay period of 32 ms
4928 sched_thermal_decay_shift thermal pressure decay pr
4932 Format: integer between 0 and 10
4935 scftorture.holdoff= [KNL]
4936 Number of seconds to hold off before starting
4937 test. Defaults to zero for module insertion and
4938 to 10 seconds for built-in smp_call_function()
4941 scftorture.longwait= [KNL]
4942 Request ridiculously long waits randomly selected
4943 up to the chosen limit in seconds. Zero (the
4944 default) disables this feature. Please note
4945 that requesting even small non-zero numbers of
4946 seconds can result in RCU CPU stall warnings,
4947 softlockup complaints, and so on.
4949 scftorture.nthreads= [KNL]
4950 Number of kthreads to spawn to invoke the
4951 smp_call_function() family of functions.
4952 The default of -1 specifies a number of kthreads
4953 equal to the number of CPUs.
4955 scftorture.onoff_holdoff= [KNL]
4956 Number seconds to wait after the start of the
4957 test before initiating CPU-hotplug operations.
4959 scftorture.onoff_interval= [KNL]
4960 Number seconds to wait between successive
4961 CPU-hotplug operations. Specifying zero (which
4962 is the default) disables CPU-hotplug operations.
4964 scftorture.shutdown_secs= [KNL]
4965 The number of seconds following the start of the
4966 test after which to shut down the system. The
4967 default of zero avoids shutting down the system.
4968 Non-zero values are useful for automated tests.
4970 scftorture.stat_interval= [KNL]
4971 The number of seconds between outputting the
4972 current test statistics to the console. A value
4973 of zero disables statistics output.
4975 scftorture.stutter_cpus= [KNL]
4976 The number of jiffies to wait between each change
4977 to the set of CPUs under test.
4979 scftorture.use_cpus_read_lock= [KNL]
4980 Use use_cpus_read_lock() instead of the default
4981 preempt_disable() to disable CPU hotplug
4982 while invoking one of the smp_call_function*()
4985 scftorture.verbose= [KNL]
4986 Enable additional printk() statements.
4988 scftorture.weight_single= [KNL]
4989 The probability weighting to use for the
4990 smp_call_function_single() function with a zero
4991 "wait" parameter. A value of -1 selects the
4992 default if all other weights are -1. However,
4993 if at least one weight has some other value, a
4994 value of -1 will instead select a weight of zero.
4996 scftorture.weight_single_wait= [KNL]
4997 The probability weighting to use for the
4998 smp_call_function_single() function with a
4999 non-zero "wait" parameter. See weight_single.
5001 scftorture.weight_many= [KNL]
5002 The probability weighting to use for the
5003 smp_call_function_many() function with a zero
5004 "wait" parameter. See weight_single.
5005 Note well that setting a high probability for
5006 this weighting can place serious IPI load
5009 scftorture.weight_many_wait= [KNL]
5010 The probability weighting to use for the
5011 smp_call_function_many() function with a
5012 non-zero "wait" parameter. See weight_single
5015 scftorture.weight_all= [KNL]
5016 The probability weighting to use for the
5017 smp_call_function_all() function with a zero
5018 "wait" parameter. See weight_single and
5021 scftorture.weight_all_wait= [KNL]
5022 The probability weighting to use for the
5023 smp_call_function_all() function with a
5024 non-zero "wait" parameter. See weight_single
5027 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
5028 xtime_lock contention on larger systems, and/or RCU lock
5029 contention on all systems with CONFIG_MAXSMP set.
5030 Format: { "0" | "1" }
5031 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
5033 Note: increases power consumption, thus should only be
5034 enabled if running jitter sensitive (HPC/RT) workloads.
5036 security= [SECURITY] Choose a legacy "major" security module to
5037 enable at boot. This has been deprecated by the
5040 selinux= [SELINUX] Disable or enable SELinux at boot time.
5041 Format: { "0" | "1" }
5042 See security/selinux/Kconfig help text.
5047 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
5048 Format: { "0" | "1" }
5049 See security/apparmor/Kconfig help text
5052 Default value is set via kernel config option.
5054 serialnumber [BUGS=X86-32]
5057 Maximal number of shapers.
5065 Enable merging of slabs with similar size when the
5066 kernel is built without CONFIG_SLAB_MERGE_DEFAULT.
5069 Disable merging of slabs with similar size. May be
5070 necessary if there is some reason to distinguish
5071 allocs to different slabs, especially in hardened
5072 environments where the risk of heap overflows and
5073 layout control by attackers can usually be
5074 frustrated by disabling merging. This will reduce
5075 most of the exposure of a heap attack to a single
5076 cache (risks via metadata attacks are mostly
5077 unchanged). Debug options disable merging on their
5079 For more information see Documentation/vm/slub.rst.
5081 slab_max_order= [MM, SLAB]
5082 Determines the maximum allowed order for slabs.
5083 A high setting may cause OOMs due to memory
5084 fragmentation. Defaults to 1 for systems with
5085 more than 32MB of RAM, 0 otherwise.
5087 slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB]
5088 Enabling slub_debug allows one to determine the
5089 culprit if slab objects become corrupted. Enabling
5090 slub_debug can create guard zones around objects and
5091 may poison objects when not in use. Also tracks the
5092 last alloc / free. For more information see
5093 Documentation/vm/slub.rst.
5095 slub_max_order= [MM, SLUB]
5096 Determines the maximum allowed order for slabs.
5097 A high setting may cause OOMs due to memory
5098 fragmentation. For more information see
5099 Documentation/vm/slub.rst.
5101 slub_min_objects= [MM, SLUB]
5102 The minimum number of objects per slab. SLUB will
5103 increase the slab order up to slub_max_order to
5104 generate a sufficiently large slab able to contain
5105 the number of objects indicated. The higher the number
5106 of objects the smaller the overhead of tracking slabs
5107 and the less frequently locks need to be acquired.
5108 For more information see Documentation/vm/slub.rst.
5110 slub_min_order= [MM, SLUB]
5111 Determines the minimum page order for slabs. Must be
5112 lower than slub_max_order.
5113 For more information see Documentation/vm/slub.rst.
5115 slub_merge [MM, SLUB]
5116 Same with slab_merge.
5118 slub_nomerge [MM, SLUB]
5119 Same with slab_nomerge. This is supported for legacy.
5120 See slab_nomerge for more information.
5123 Format: <io1>[,<io2>[,...,<io8>]]
5125 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
5126 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
5127 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
5128 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
5129 smsc-ircc2.ircc_irq= [HW] IRQ line
5130 smsc-ircc2.ircc_dma= [HW] DMA channel
5131 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
5132 0: Toshiba Satellite 1800 (GP data pin select)
5133 1: Fast pin select (default)
5136 smt [KNL,S390] Set the maximum number of threads (logical
5137 CPUs) to use per physical CPU on systems capable of
5138 symmetric multithreading (SMT). Will be capped to the
5139 actual hardware limit.
5141 Default: -1 (no limit)
5144 [KNL] Should the soft-lockup detector generate panics.
5147 A value of 1 instructs the soft-lockup detector
5148 to panic the machine when a soft-lockup occurs. It is
5149 also controlled by the kernel.softlockup_panic sysctl
5150 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
5151 respective build-time switch to that functionality.
5153 softlockup_all_cpu_backtrace=
5154 [KNL] Should the soft-lockup detector generate
5155 backtraces on all cpus.
5158 sonypi.*= [HW] Sony Programmable I/O Control Device driver
5159 See Documentation/admin-guide/laptops/sonypi.rst
5161 spectre_v2= [X86] Control mitigation of Spectre variant 2
5162 (indirect branch speculation) vulnerability.
5163 The default operation protects the kernel from
5166 on - unconditionally enable, implies
5168 off - unconditionally disable, implies
5170 auto - kernel detects whether your CPU model is
5173 Selecting 'on' will, and 'auto' may, choose a
5174 mitigation method at run time according to the
5175 CPU, the available microcode, the setting of the
5176 CONFIG_RETPOLINE configuration option, and the
5177 compiler with which the kernel was built.
5179 Selecting 'on' will also enable the mitigation
5180 against user space to user space task attacks.
5182 Selecting 'off' will disable both the kernel and
5183 the user space protections.
5185 Specific mitigations can also be selected manually:
5187 retpoline - replace indirect branches
5188 retpoline,generic - google's original retpoline
5189 retpoline,amd - AMD-specific minimal thunk
5191 Not specifying this option is equivalent to
5195 [X86] Control mitigation of Spectre variant 2
5196 (indirect branch speculation) vulnerability between
5199 on - Unconditionally enable mitigations. Is
5200 enforced by spectre_v2=on
5202 off - Unconditionally disable mitigations. Is
5203 enforced by spectre_v2=off
5205 prctl - Indirect branch speculation is enabled,
5206 but mitigation can be enabled via prctl
5207 per thread. The mitigation control state
5208 is inherited on fork.
5211 - Like "prctl" above, but only STIBP is
5212 controlled per thread. IBPB is issued
5213 always when switching between different user
5217 - Same as "prctl" above, but all seccomp
5218 threads will enable the mitigation unless
5219 they explicitly opt out.
5222 - Like "seccomp" above, but only STIBP is
5223 controlled per thread. IBPB is issued
5224 always when switching between different
5225 user space processes.
5227 auto - Kernel selects the mitigation depending on
5228 the available CPU features and vulnerability.
5231 If CONFIG_SECCOMP=y then "seccomp", otherwise "prctl"
5233 Not specifying this option is equivalent to
5234 spectre_v2_user=auto.
5236 spec_store_bypass_disable=
5237 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
5238 (Speculative Store Bypass vulnerability)
5240 Certain CPUs are vulnerable to an exploit against a
5241 a common industry wide performance optimization known
5242 as "Speculative Store Bypass" in which recent stores
5243 to the same memory location may not be observed by
5244 later loads during speculative execution. The idea
5245 is that such stores are unlikely and that they can
5246 be detected prior to instruction retirement at the
5247 end of a particular speculation execution window.
5249 In vulnerable processors, the speculatively forwarded
5250 store can be used in a cache side channel attack, for
5251 example to read memory to which the attacker does not
5252 directly have access (e.g. inside sandboxed code).
5254 This parameter controls whether the Speculative Store
5255 Bypass optimization is used.
5257 On x86 the options are:
5259 on - Unconditionally disable Speculative Store Bypass
5260 off - Unconditionally enable Speculative Store Bypass
5261 auto - Kernel detects whether the CPU model contains an
5262 implementation of Speculative Store Bypass and
5263 picks the most appropriate mitigation. If the
5264 CPU is not vulnerable, "off" is selected. If the
5265 CPU is vulnerable the default mitigation is
5266 architecture and Kconfig dependent. See below.
5267 prctl - Control Speculative Store Bypass per thread
5268 via prctl. Speculative Store Bypass is enabled
5269 for a process by default. The state of the control
5270 is inherited on fork.
5271 seccomp - Same as "prctl" above, but all seccomp threads
5272 will disable SSB unless they explicitly opt out.
5274 Default mitigations:
5275 X86: If CONFIG_SECCOMP=y "seccomp", otherwise "prctl"
5277 On powerpc the options are:
5279 on,auto - On Power8 and Power9 insert a store-forwarding
5280 barrier on kernel entry and exit. On Power7
5281 perform a software flush on kernel entry and
5285 Not specifying this option is equivalent to
5286 spec_store_bypass_disable=auto.
5288 spia_io_base= [HW,MTD]
5294 [X86] Enable split lock detection or bus lock detection
5296 When enabled (and if hardware support is present), atomic
5297 instructions that access data across cache line
5298 boundaries will result in an alignment check exception
5299 for split lock detection or a debug exception for
5304 warn - the kernel will emit rate-limited warnings
5305 about applications triggering the #AC
5306 exception or the #DB exception. This mode is
5307 the default on CPUs that support split lock
5308 detection or bus lock detection. Default
5309 behavior is by #AC if both features are
5310 enabled in hardware.
5312 fatal - the kernel will send SIGBUS to applications
5313 that trigger the #AC exception or the #DB
5314 exception. Default behavior is by #AC if
5315 both features are enabled in hardware.
5318 Set system wide rate limit to N bus locks
5319 per second for bus lock detection.
5322 N/A for split lock detection.
5325 If an #AC exception is hit in the kernel or in
5326 firmware (i.e. not while executing in user mode)
5327 the kernel will oops in either "warn" or "fatal"
5330 #DB exception for bus lock is triggered only when
5334 Control the Special Register Buffer Data Sampling
5337 Certain CPUs are vulnerable to an MDS-like
5338 exploit which can leak bits from the random
5341 By default, this issue is mitigated by
5342 microcode. However, the microcode fix can cause
5343 the RDRAND and RDSEED instructions to become
5344 much slower. Among other effects, this will
5345 result in reduced throughput from /dev/urandom.
5347 The microcode mitigation can be disabled with
5348 the following option:
5350 off: Disable mitigation and remove
5351 performance impact to RDRAND and RDSEED
5353 srcutree.counter_wrap_check [KNL]
5354 Specifies how frequently to check for
5355 grace-period sequence counter wrap for the
5356 srcu_data structure's ->srcu_gp_seq_needed field.
5357 The greater the number of bits set in this kernel
5358 parameter, the less frequently counter wrap will
5359 be checked for. Note that the bottom two bits
5362 srcutree.exp_holdoff [KNL]
5363 Specifies how many nanoseconds must elapse
5364 since the end of the last SRCU grace period for
5365 a given srcu_struct until the next normal SRCU
5366 grace period will be considered for automatic
5367 expediting. Set to zero to disable automatic
5371 Speculative Store Bypass Disable control
5373 On CPUs that are vulnerable to the Speculative
5374 Store Bypass vulnerability and offer a
5375 firmware based mitigation, this parameter
5376 indicates how the mitigation should be used:
5378 force-on: Unconditionally enable mitigation for
5379 for both kernel and userspace
5380 force-off: Unconditionally disable mitigation for
5381 for both kernel and userspace
5382 kernel: Always enable mitigation in the
5383 kernel, and offer a prctl interface
5384 to allow userspace to register its
5385 interest in being mitigated too.
5387 stack_guard_gap= [MM]
5388 override the default stack gap protection. The value
5389 is in page units and it defines how many pages prior
5390 to (for stacks growing down) resp. after (for stacks
5391 growing up) the main stack are reserved for no other
5392 mapping. Default value is 256 pages.
5394 stack_depot_disable= [KNL]
5395 Setting this to true through kernel command line will
5396 disable the stack depot thereby saving the static memory
5397 consumed by the stack hash table. By default this is set
5401 Enabled the stack tracer on boot up.
5403 stacktrace_filter=[function-list]
5404 [FTRACE] Limit the functions that the stack tracer
5405 will trace at boot up. function-list is a comma-separated
5406 list of functions. This list can be changed at run
5407 time by the stack_trace_filter file in the debugfs
5408 tracing directory. Note, this enables stack tracing
5409 and the stacktrace above is not needed.
5413 Set the STI (builtin display/keyboard on the HP-PARISC
5414 machines) console (graphic card) which should be used
5415 as the initial boot-console.
5416 See also comment in drivers/video/console/sticore.c.
5419 See comment in drivers/video/console/sticore.c.
5422 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
5424 sunrpc.min_resvport=
5425 sunrpc.max_resvport=
5427 SunRPC servers often require that client requests
5428 originate from a privileged port (i.e. a port in the
5429 range 0 < portnr < 1024).
5430 An administrator who wishes to reserve some of these
5431 ports for other uses may adjust the range that the
5432 kernel's sunrpc client considers to be privileged
5433 using these two parameters to set the minimum and
5434 maximum port values.
5436 sunrpc.svc_rpc_per_connection_limit=
5438 Limit the number of requests that the server will
5439 process in parallel from a single connection.
5440 The default value is 0 (no limit).
5444 Control how the NFS server code allocates CPUs to
5445 service thread pools. Depending on how many NICs
5446 you have and where their interrupts are bound, this
5447 option will affect which CPUs will do NFS serving.
5448 Note: this parameter cannot be changed while the
5449 NFS server is running.
5451 auto the server chooses an appropriate mode
5452 automatically using heuristics
5453 global a single global pool contains all CPUs
5454 percpu one pool for each CPU
5455 pernode one pool for each NUMA node (equivalent
5456 to global on non-NUMA machines)
5458 sunrpc.tcp_slot_table_entries=
5459 sunrpc.udp_slot_table_entries=
5461 Sets the upper limit on the number of simultaneous
5462 RPC calls that can be sent from the client to a
5463 server. Increasing these values may allow you to
5464 improve throughput, but will also increase the
5465 amount of memory reserved for use by the client.
5467 suspend.pm_test_delay=
5469 Sets the number of seconds to remain in a suspend test
5470 mode before resuming the system (see
5471 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
5472 is set. Default value is 5.
5475 Format: { on | off | y | n | 1 | 0 }
5476 This parameter controls use of the Protected
5477 Execution Facility on pSeries.
5480 [KNL] Enable accounting of swap in memory resource
5481 controller if no parameter or 1 is given or disable
5482 it if 0 is given (See Documentation/admin-guide/cgroup-v1/memory.rst)
5484 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
5485 Format: { <int> | force | noforce }
5486 <int> -- Number of I/O TLB slabs
5487 force -- force using of bounce buffers even if they
5488 wouldn't be automatically used by the kernel
5489 noforce -- Never use bounce buffers (for debugging)
5494 Set a sysctl parameter, right before loading the init
5495 process, as if the value was written to the respective
5496 /proc/sys/... file. Both '.' and '/' are recognized as
5497 separators. Unrecognized parameters and invalid values
5498 are reported in the kernel log. Sysctls registered
5499 later by a loaded module cannot be set this way.
5500 Example: sysctl.vm.swappiness=40
5502 sysfs.deprecated=0|1 [KNL]
5503 Enable/disable old style sysfs layout for old udev
5504 on older distributions. When this option is enabled
5505 very new udev will not work anymore. When this option
5506 is disabled (or CONFIG_SYSFS_DEPRECATED not compiled)
5507 in older udev will not work anymore.
5508 Default depends on CONFIG_SYSFS_DEPRECATED_V2 set in
5509 the kernel configuration.
5511 sysrq_always_enabled
5513 Ignore sysrq setting - this boot parameter will
5514 neutralize any effect of /proc/sys/kernel/sysrq.
5515 Useful for debugging.
5517 tcpmhash_entries= [KNL,NET]
5518 Set the number of tcp_metrics_hash slots.
5519 Default value is 8192 or 16384 depending on total
5520 ram pages. This is used to specify the TCP metrics
5521 cache size. See Documentation/networking/ip-sysctl.rst
5522 "tcp_no_metrics_save" section for more details.
5526 test_suspend= [SUSPEND][,N]
5527 Specify "mem" (for Suspend-to-RAM) or "standby" (for
5528 standby suspend) or "freeze" (for suspend type freeze)
5529 as the system sleep state during system startup with
5530 the optional capability to repeat N number of times.
5531 The system is woken from this state using a
5532 wakeup-capable RTC alarm.
5534 thash_entries= [KNL,NET]
5535 Set number of hash buckets for TCP connection
5537 thermal.act= [HW,ACPI]
5538 -1: disable all active trip points in all thermal zones
5539 <degrees C>: override all lowest active trip points
5541 thermal.crt= [HW,ACPI]
5542 -1: disable all critical trip points in all thermal zones
5543 <degrees C>: override all critical trip points
5545 thermal.nocrt= [HW,ACPI]
5546 Set to disable actions on ACPI thermal zone
5547 critical and hot trip points.
5549 thermal.off= [HW,ACPI]
5550 1: disable ACPI thermal control
5552 thermal.psv= [HW,ACPI]
5553 -1: disable all passive trip points
5554 <degrees C>: override all passive trip points to this
5557 thermal.tzp= [HW,ACPI]
5558 Specify global default ACPI thermal zone polling rate
5559 <deci-seconds>: poll all this frequency
5560 0: no polling (default)
5563 Force threading of all interrupt handlers except those
5564 marked explicitly IRQF_NO_THREAD.
5568 Specify if the kernel should make use of the cpu
5569 topology information if the hardware supports this.
5570 The scheduler will make use of this information and
5571 e.g. base its process migration decisions on it.
5574 topology_updates= [KNL, PPC, NUMA]
5576 Specify if the kernel should ignore (off)
5577 topology updates sent by the hypervisor to this
5580 torture.disable_onoff_at_boot= [KNL]
5581 Prevent the CPU-hotplug component of torturing
5582 until after init has spawned.
5584 torture.ftrace_dump_at_shutdown= [KNL]
5585 Dump the ftrace buffer at torture-test shutdown,
5586 even if there were no errors. This can be a
5587 very costly operation when many torture tests
5588 are running concurrently, especially on systems
5589 with rotating-rust storage.
5591 torture.verbose_sleep_frequency= [KNL]
5592 Specifies how many verbose printk()s should be
5593 emitted between each sleep. The default of zero
5594 disables verbose-printk() sleeping.
5596 torture.verbose_sleep_duration= [KNL]
5597 Duration of each verbose-printk() sleep in jiffies.
5601 tpm_suspend_pcr=[HW,TPM]
5602 Format: integer pcr id
5603 Specify that at suspend time, the tpm driver
5604 should extend the specified pcr with zeros,
5605 as a workaround for some chips which fail to
5606 flush the last written pcr on TPM_SaveState.
5607 This will guarantee that all the other pcrs
5610 trace_buf_size=nn[KMG]
5611 [FTRACE] will set tracing buffer size on each cpu.
5613 trace_event=[event-list]
5614 [FTRACE] Set and start specified trace events in order
5615 to facilitate early boot debugging. The event-list is a
5616 comma-separated list of trace events to enable. See
5617 also Documentation/trace/events.rst
5619 trace_options=[option-list]
5620 [FTRACE] Enable or disable tracer options at boot.
5621 The option-list is a comma delimited list of options
5622 that can be enabled or disabled just as if you were
5623 to echo the option name into
5625 /sys/kernel/debug/tracing/trace_options
5627 For example, to enable stacktrace option (to dump the
5628 stack trace of each event), add to the command line:
5630 trace_options=stacktrace
5632 See also Documentation/trace/ftrace.rst "trace options"
5636 Have the tracepoints sent to printk as well as the
5637 tracing ring buffer. This is useful for early boot up
5638 where the system hangs or reboots and does not give the
5639 option for reading the tracing buffer or performing a
5640 ftrace_dump_on_oops.
5642 To turn off having tracepoints sent to printk,
5643 echo 0 > /proc/sys/kernel/tracepoint_printk
5644 Note, echoing 1 into this file without the
5645 tracepoint_printk kernel cmdline option has no effect.
5649 Having tracepoints sent to printk() and activating high
5650 frequency tracepoints such as irq or sched, can cause
5651 the system to live lock.
5654 [FTRACE] enable this option to disable tracing when a
5655 warning is hit. This turns off "tracing_on". Tracing can
5656 be enabled again by echoing '1' into the "tracing_on"
5657 file located in /sys/kernel/debug/tracing/
5659 This option is useful, as it disables the trace before
5660 the WARNING dump is called, which prevents the trace to
5661 be filled with content caused by the warning output.
5663 This option can also be set at run time via the sysctl
5664 option: kernel/traceoff_on_warning
5666 transparent_hugepage=
5668 Format: [always|madvise|never]
5669 Can be used to control the default behavior of the system
5670 with respect to transparent hugepages.
5671 See Documentation/admin-guide/mm/transhuge.rst
5674 trusted.source= [KEYS]
5676 This parameter identifies the trust source as a backend
5677 for trusted keys implementation. Supported trust
5681 If not specified then it defaults to iterating through
5682 the trust source list starting with TPM and assigns the
5683 first trust source as a backend which is initialized
5684 successfully during iteration.
5686 tsc= Disable clocksource stability checks for TSC.
5688 [x86] reliable: mark tsc clocksource as reliable, this
5689 disables clocksource verification at runtime, as well
5690 as the stability checks done at bootup. Used to enable
5691 high-resolution timer mode on older hardware, and in
5692 virtualized environment.
5693 [x86] noirqtime: Do not use TSC to do irq accounting.
5694 Used to run time disable IRQ_TIME_ACCOUNTING on any
5695 platforms where RDTSC is slow and this accounting
5697 [x86] unstable: mark the TSC clocksource as unstable, this
5698 marks the TSC unconditionally unstable at bootup and
5699 avoids any further wobbles once the TSC watchdog notices.
5700 [x86] nowatchdog: disable clocksource watchdog. Used
5701 in situations with strict latency requirements (where
5702 interruptions from clocksource watchdog are not
5705 tsc_early_khz= [X86] Skip early TSC calibration and use the given
5706 value instead. Useful when the early TSC frequency discovery
5707 procedure is not reliable, such as on overclocked systems
5708 with CPUID.16h support and partial CPUID.15h support.
5709 Format: <unsigned int>
5711 tsx= [X86] Control Transactional Synchronization
5712 Extensions (TSX) feature in Intel processors that
5713 support TSX control.
5715 This parameter controls the TSX feature. The options are:
5717 on - Enable TSX on the system. Although there are
5718 mitigations for all known security vulnerabilities,
5719 TSX has been known to be an accelerator for
5720 several previous speculation-related CVEs, and
5721 so there may be unknown security risks associated
5722 with leaving it enabled.
5724 off - Disable TSX on the system. (Note that this
5725 option takes effect only on newer CPUs which are
5726 not vulnerable to MDS, i.e., have
5727 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
5728 the new IA32_TSX_CTRL MSR through a microcode
5729 update. This new MSR allows for the reliable
5730 deactivation of the TSX functionality.)
5732 auto - Disable TSX if X86_BUG_TAA is present,
5733 otherwise enable TSX on the system.
5735 Not specifying this option is equivalent to tsx=off.
5737 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5740 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
5741 Abort (TAA) vulnerability.
5743 Similar to Micro-architectural Data Sampling (MDS)
5744 certain CPUs that support Transactional
5745 Synchronization Extensions (TSX) are vulnerable to an
5746 exploit against CPU internal buffers which can forward
5747 information to a disclosure gadget under certain
5750 In vulnerable processors, the speculatively forwarded
5751 data can be used in a cache side channel attack, to
5752 access data to which the attacker does not have direct
5755 This parameter controls the TAA mitigation. The
5758 full - Enable TAA mitigation on vulnerable CPUs
5761 full,nosmt - Enable TAA mitigation and disable SMT on
5762 vulnerable CPUs. If TSX is disabled, SMT
5763 is not disabled because CPU is not
5764 vulnerable to cross-thread TAA attacks.
5765 off - Unconditionally disable TAA mitigation
5767 On MDS-affected machines, tsx_async_abort=off can be
5768 prevented by an active MDS mitigation as both vulnerabilities
5769 are mitigated with the same mechanism so in order to disable
5770 this mitigation, you need to specify mds=off too.
5772 Not specifying this option is equivalent to
5773 tsx_async_abort=full. On CPUs which are MDS affected
5774 and deploy MDS mitigation, TAA mitigation is not
5775 required and doesn't provide any additional
5779 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5781 turbografx.map[2|3]= [HW,JOY]
5782 TurboGraFX parallel port interface
5784 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
5785 See also Documentation/input/devices/joystick-parport.rst
5787 udbg-immortal [PPC] When debugging early kernel crashes that
5788 happen after console_init() and before a proper
5789 console driver takes over, this boot options might
5790 help "seeing" what's going on.
5792 uhash_entries= [KNL,NET]
5793 Set number of hash buckets for UDP/UDP-Lite connections
5796 [USB] Ignore overcurrent events (default N).
5797 Some badly-designed motherboards generate lots of
5798 bogus events, for ports that aren't wired to
5799 anything. Set this parameter to avoid log spamming.
5800 Note that genuine overcurrent events won't be
5804 [X86] Cause panic on unknown NMI.
5806 usbcore.authorized_default=
5807 [USB] Default USB device authorization:
5808 (default -1 = authorized except for wireless USB,
5809 0 = not authorized, 1 = authorized, 2 = authorized
5810 if device connected to internal port)
5812 usbcore.autosuspend=
5813 [USB] The autosuspend time delay (in seconds) used
5814 for newly-detected USB devices (default 2). This
5815 is the time required before an idle device will be
5816 autosuspended. Devices for which the delay is set
5817 to a negative value won't be autosuspended at all.
5819 usbcore.usbfs_snoop=
5820 [USB] Set to log all usbfs traffic (default 0 = off).
5822 usbcore.usbfs_snoop_max=
5823 [USB] Maximum number of bytes to snoop in each URB
5826 usbcore.blinkenlights=
5827 [USB] Set to cycle leds on hubs (default 0 = off).
5829 usbcore.old_scheme_first=
5830 [USB] Start with the old device initialization
5831 scheme (default 0 = off).
5833 usbcore.usbfs_memory_mb=
5834 [USB] Memory limit (in MB) for buffers allocated by
5835 usbfs (default = 16, 0 = max = 2047).
5837 usbcore.use_both_schemes=
5838 [USB] Try the other device initialization scheme
5839 if the first one fails (default 1 = enabled).
5841 usbcore.initial_descriptor_timeout=
5842 [USB] Specifies timeout for the initial 64-byte
5843 USB_REQ_GET_DESCRIPTOR request in milliseconds
5844 (default 5000 = 5.0 seconds).
5846 usbcore.nousb [USB] Disable the USB subsystem
5849 [USB] A list of quirk entries to augment the built-in
5850 usb core quirk list. List entries are separated by
5851 commas. Each entry has the form
5852 VendorID:ProductID:Flags. The IDs are 4-digit hex
5853 numbers and Flags is a set of letters. Each letter
5854 will change the built-in quirk; setting it if it is
5855 clear and clearing it if it is set. The letters have
5856 the following meanings:
5857 a = USB_QUIRK_STRING_FETCH_255 (string
5858 descriptors must not be fetched using
5860 b = USB_QUIRK_RESET_RESUME (device can't resume
5861 correctly so reset it instead);
5862 c = USB_QUIRK_NO_SET_INTF (device can't handle
5863 Set-Interface requests);
5864 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
5865 handle its Configuration or Interface
5867 e = USB_QUIRK_RESET (device can't be reset
5868 (e.g morph devices), don't use reset);
5869 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
5870 more interface descriptions than the
5871 bNumInterfaces count, and can't handle
5872 talking to these interfaces);
5873 g = USB_QUIRK_DELAY_INIT (device needs a pause
5874 during initialization, after we read
5875 the device descriptor);
5876 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
5877 high speed and super speed interrupt
5878 endpoints, the USB 2.0 and USB 3.0 spec
5879 require the interval in microframes (1
5880 microframe = 125 microseconds) to be
5881 calculated as interval = 2 ^
5883 Devices with this quirk report their
5884 bInterval as the result of this
5885 calculation instead of the exponent
5886 variable used in the calculation);
5887 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
5888 handle device_qualifier descriptor
5890 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
5891 generates spurious wakeup, ignore
5892 remote wakeup capability);
5893 k = USB_QUIRK_NO_LPM (device can't handle Link
5895 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
5896 (Device reports its bInterval as linear
5897 frames instead of the USB 2.0
5899 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
5900 to be disconnected before suspend to
5901 prevent spurious wakeup);
5902 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
5903 pause after every control message);
5904 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
5905 delay after resetting its port);
5906 Example: quirks=0781:5580:bk,0a5c:5834:gij
5909 [USBHID] The interval which mice are to be polled at.
5912 [USBHID] The interval which joysticks are to be polled at.
5915 [USBHID] The interval which keyboards are to be polled at.
5917 usb-storage.delay_use=
5918 [UMS] The delay in seconds before a new device is
5919 scanned for Logical Units (default 1).
5922 [UMS] A list of quirks entries to supplement or
5923 override the built-in unusual_devs list. List
5924 entries are separated by commas. Each entry has
5925 the form VID:PID:Flags where VID and PID are Vendor
5926 and Product ID values (4-digit hex numbers) and
5927 Flags is a set of characters, each corresponding
5928 to a common usb-storage quirk flag as follows:
5929 a = SANE_SENSE (collect more than 18 bytes
5930 of sense data, not on uas);
5931 b = BAD_SENSE (don't collect more than 18
5932 bytes of sense data, not on uas);
5933 c = FIX_CAPACITY (decrease the reported
5934 device capacity by one sector);
5935 d = NO_READ_DISC_INFO (don't use
5936 READ_DISC_INFO command, not on uas);
5937 e = NO_READ_CAPACITY_16 (don't use
5938 READ_CAPACITY_16 command);
5939 f = NO_REPORT_OPCODES (don't use report opcodes
5941 g = MAX_SECTORS_240 (don't transfer more than
5942 240 sectors at a time, uas only);
5943 h = CAPACITY_HEURISTICS (decrease the
5944 reported device capacity by one
5945 sector if the number is odd);
5946 i = IGNORE_DEVICE (don't bind to this
5948 j = NO_REPORT_LUNS (don't use report luns
5950 k = NO_SAME (do not use WRITE_SAME, uas only)
5951 l = NOT_LOCKABLE (don't try to lock and
5952 unlock ejectable media, not on uas);
5953 m = MAX_SECTORS_64 (don't transfer more
5954 than 64 sectors = 32 KB at a time,
5956 n = INITIAL_READ10 (force a retry of the
5957 initial READ(10) command, not on uas);
5958 o = CAPACITY_OK (accept the capacity
5959 reported by the device, not on uas);
5960 p = WRITE_CACHE (the device cache is ON
5961 by default, not on uas);
5962 r = IGNORE_RESIDUE (the device reports
5963 bogus residue values, not on uas);
5964 s = SINGLE_LUN (the device has only one
5966 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
5967 commands, uas only);
5968 u = IGNORE_UAS (don't bind to the uas driver);
5969 w = NO_WP_DETECT (don't test whether the
5970 medium is write-protected).
5971 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
5972 even if the device claims no cache,
5974 Example: quirks=0419:aaf5:rl,0421:0433:rc
5976 user_debug= [KNL,ARM]
5978 See arch/arm/Kconfig.debug help text.
5979 1 - undefined instruction events
5981 4 - invalid data aborts
5984 Example: user_debug=31
5987 [X86] Flags controlling user PTE allocations.
5989 nohigh = do not allocate PTE pages in
5990 HIGHMEM regardless of setting
5994 On X86_32, this is an alias for vdso32=. Otherwise:
5996 vdso=1: enable VDSO (the default)
5997 vdso=0: disable VDSO mapping
5999 vdso32= [X86] Control the 32-bit vDSO
6000 vdso32=1: enable 32-bit VDSO
6001 vdso32=0 or vdso32=2: disable 32-bit VDSO
6003 See the help text for CONFIG_COMPAT_VDSO for more
6004 details. If CONFIG_COMPAT_VDSO is set, the default is
6005 vdso32=0; otherwise, the default is vdso32=1.
6007 For compatibility with older kernels, vdso32=2 is an
6010 Try vdso32=0 if you encounter an error that says:
6011 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
6014 vector=percpu: enable percpu vector domain
6016 video= [FB] Frame buffer configuration
6017 See Documentation/fb/modedb.rst.
6019 video.brightness_switch_enabled= [0,1]
6020 If set to 1, on receiving an ACPI notify event
6021 generated by hotkey, video driver will adjust brightness
6022 level and then send out the event to user space through
6023 the allocated input device; If set to 0, video driver
6024 will only send out the event without touching backlight
6029 [VMMIO] Memory mapped virtio (platform) device.
6031 <size>@<baseaddr>:<irq>[:<id>]
6033 <size> := size (can use standard suffixes
6035 <baseaddr> := physical base address
6036 <irq> := interrupt number (as passed to
6038 <id> := (optional) platform device id
6040 virtio_mmio.device=1K@0x100b0000:48:7
6042 Can be used multiple times for multiple devices.
6044 vga= [BOOT,X86-32] Select a particular video mode
6045 See Documentation/x86/boot.rst and
6046 Documentation/admin-guide/svga.rst.
6047 Use vga=ask for menu.
6048 This is actually a boot loader parameter; the value is
6049 passed to the kernel using a special protocol.
6051 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
6052 May slow down system boot speed, especially when
6053 enabled on systems with a large amount of memory.
6054 All options are enabled by default, and this
6055 interface is meant to allow for selectively
6056 enabling or disabling specific virtual memory
6059 Available options are:
6060 P Enable page structure init time poisoning
6061 - Disable all of the above options
6063 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
6064 size of <nn>. This can be used to increase the
6065 minimum size (128MB on x86). It can also be used to
6066 decrease the size and leave more room for directly
6069 vmcp_cma=nn[MG] [KNL,S390]
6070 Sets the memory size reserved for contiguous memory
6071 allocations for the vmcp device driver.
6073 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
6076 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
6079 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
6083 Controls the behavior of vsyscalls (i.e. calls to
6084 fixed addresses of 0xffffffffff600x00 from legacy
6085 code). Most statically-linked binaries and older
6086 versions of glibc use these calls. Because these
6087 functions are at fixed addresses, they make nice
6088 targets for exploits that can control RIP.
6090 emulate [default] Vsyscalls turn into traps and are
6091 emulated reasonably safely. The vsyscall
6094 xonly Vsyscalls turn into traps and are
6095 emulated reasonably safely. The vsyscall
6096 page is not readable.
6098 none Vsyscalls don't work at all. This makes
6099 them quite hard to use for exploits but
6100 might break your system.
6102 vt.color= [VT] Default text color.
6103 Format: 0xYX, X = foreground, Y = background.
6104 Default: 0x07 = light gray on black.
6106 vt.cur_default= [VT] Default cursor shape.
6107 Format: 0xCCBBAA, where AA, BB, and CC are the same as
6108 the parameters of the <Esc>[?A;B;Cc escape sequence;
6109 see VGA-softcursor.txt. Default: 2 = underline.
6111 vt.default_blu= [VT]
6112 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
6113 Change the default blue palette of the console.
6114 This is a 16-member array composed of values
6117 vt.default_grn= [VT]
6118 Format: <green0>,<green1>,<green2>,...,<green15>
6119 Change the default green palette of the console.
6120 This is a 16-member array composed of values
6123 vt.default_red= [VT]
6124 Format: <red0>,<red1>,<red2>,...,<red15>
6125 Change the default red palette of the console.
6126 This is a 16-member array composed of values
6132 Set system-wide default UTF-8 mode for all tty's.
6133 Default is 1, i.e. UTF-8 mode is enabled for all
6134 newly opened terminals.
6136 vt.global_cursor_default=
6139 Set system-wide default for whether a cursor
6140 is shown on new VTs. Default is -1,
6141 i.e. cursors will be created by default unless
6142 overridden by individual drivers. 0 will hide
6143 cursors, 1 will display them.
6145 vt.italic= [VT] Default color for italic text; 0-15.
6148 vt.underline= [VT] Default color for underlined text; 0-15.
6151 watchdog timers [HW,WDT] For information on watchdog timers,
6152 see Documentation/watchdog/watchdog-parameters.rst
6153 or other driver-specific files in the
6154 Documentation/watchdog/ directory.
6158 Set the hard lockup detector stall duration
6159 threshold in seconds. The soft lockup detector
6160 threshold is set to twice the value. A value of 0
6161 disables both lockup detectors. Default is 10
6164 workqueue.watchdog_thresh=
6165 If CONFIG_WQ_WATCHDOG is configured, workqueue can
6166 warn stall conditions and dump internal state to
6167 help debugging. 0 disables workqueue stall
6168 detection; otherwise, it's the stall threshold
6169 duration in seconds. The default value is 30 and
6170 it can be updated at runtime by writing to the
6171 corresponding sysfs file.
6173 workqueue.disable_numa
6174 By default, all work items queued to unbound
6175 workqueues are affine to the NUMA nodes they're
6176 issued on, which results in better behavior in
6177 general. If NUMA affinity needs to be disabled for
6178 whatever reason, this option can be used. Note
6179 that this also can be controlled per-workqueue for
6180 workqueues visible under /sys/bus/workqueue/.
6182 workqueue.power_efficient
6183 Per-cpu workqueues are generally preferred because
6184 they show better performance thanks to cache
6185 locality; unfortunately, per-cpu workqueues tend to
6186 be more power hungry than unbound workqueues.
6188 Enabling this makes the per-cpu workqueues which
6189 were observed to contribute significantly to power
6190 consumption unbound, leading to measurably lower
6191 power usage at the cost of small performance
6194 The default value of this parameter is determined by
6195 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
6197 workqueue.debug_force_rr_cpu
6198 Workqueue used to implicitly guarantee that work
6199 items queued without explicit CPU specified are put
6200 on the local CPU. This guarantee is no longer true
6201 and while local CPU is still preferred work items
6202 may be put on foreign CPUs. This debug option
6203 forces round-robin CPU selection to flush out
6204 usages which depend on the now broken guarantee.
6205 When enabled, memory and cache locality will be
6208 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
6209 default x2apic cluster mode on platforms
6212 xen_512gb_limit [KNL,X86-64,XEN]
6213 Restricts the kernel running paravirtualized under Xen
6214 to use only up to 512 GB of RAM. The reason to do so is
6215 crash analysis tools and Xen tools for doing domain
6216 save/restore/migration must be enabled to handle larger
6219 xen_emul_unplug= [HW,X86,XEN]
6220 Unplug Xen emulated devices
6221 Format: [unplug0,][unplug1]
6222 ide-disks -- unplug primary master IDE devices
6223 aux-ide-disks -- unplug non-primary-master IDE devices
6224 nics -- unplug network devices
6225 all -- unplug all emulated devices (NICs and IDE disks)
6226 unnecessary -- unplugging emulated devices is
6227 unnecessary even if the host did not respond to
6229 never -- do not unplug even if version check succeeds
6231 xen_legacy_crash [X86,XEN]
6232 Crash from Xen panic notifier, without executing late
6233 panic() code such as dumping handler.
6235 xen_nopvspin [X86,XEN]
6236 Disables the qspinlock slowpath using Xen PV optimizations.
6237 This parameter is obsoleted by "nopvspin" parameter, which
6238 has equivalent effect for XEN platform.
6241 Disables the PV optimizations forcing the HVM guest to
6242 run as generic HVM guest with no PV drivers.
6243 This option is obsoleted by the "nopv" option, which
6244 has equivalent effect for XEN platform.
6246 xen_no_vector_callback
6247 [KNL,X86,XEN] Disable the vector callback for Xen
6248 event channel interrupts.
6250 xen_scrub_pages= [XEN]
6251 Boolean option to control scrubbing pages before giving them back
6252 to Xen, for use by other domains. Can be also changed at runtime
6253 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
6254 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
6256 xen_timer_slop= [X86-64,XEN]
6257 Set the timer slop (in nanoseconds) for the virtual Xen
6258 timers (default is 100000). This adjusts the minimum
6259 delta of virtualized Xen timers, where lower values
6260 improve timer resolution at the expense of processing
6261 more timer interrupts.
6263 xen.event_eoi_delay= [XEN]
6264 How long to delay EOI handling in case of event
6265 storms (jiffies). Default is 10.
6267 xen.event_loop_timeout= [XEN]
6268 After which time (jiffies) the event handling loop
6269 should start to delay EOI handling. Default is 2.
6271 xen.fifo_events= [XEN]
6272 Boolean parameter to disable using fifo event handling
6273 even if available. Normally fifo event handling is
6274 preferred over the 2-level event handling, as it is
6275 fairer and the number of possible event channels is
6276 much higher. Default is on (use fifo events).
6278 nopv= [X86,XEN,KVM,HYPER_V,VMWARE]
6279 Disables the PV optimizations forcing the guest to run
6280 as generic guest with no PV drivers. Currently support
6281 XEN HVM, KVM, HYPER_V and VMWARE guest.
6283 nopvspin [X86,XEN,KVM]
6284 Disables the qspinlock slow path using PV optimizations
6285 which allow the hypervisor to 'idle' the guest on lock
6288 xirc2ps_cs= [NET,PCMCIA]
6290 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
6293 By default on POWER9 and above, the kernel will
6294 natively use the XIVE interrupt controller. This option
6295 allows the fallback firmware mode to be used:
6297 off Fallback to firmware control of XIVE interrupt
6298 controller on both pseries and powernv
6299 platforms. Only useful on POWER9 and above.
6301 xhci-hcd.quirks [USB,KNL]
6302 A hex value specifying bitmask with supplemental xhci
6303 host controller quirks. Meaning of each bit can be
6304 consulted in header drivers/usb/host/xhci.h.
6307 Format: { early | on | rw | ro | off }
6308 Controls if xmon debugger is enabled. Default is off.
6309 Passing only "xmon" is equivalent to "xmon=early".
6310 early Call xmon as early as possible on boot; xmon
6311 debugger is called from setup_arch().
6312 on xmon debugger hooks will be installed so xmon
6313 is only called on a kernel crash. Default mode,
6314 i.e. either "ro" or "rw" mode, is controlled
6315 with CONFIG_XMON_DEFAULT_RO_MODE.
6316 rw xmon debugger hooks will be installed so xmon
6317 is called only on a kernel crash, mode is write,
6318 meaning SPR registers, memory and, other data
6319 can be written using xmon commands.
6320 ro same as "rw" option above but SPR registers,
6321 memory, and other data can't be written using
6323 off xmon is disabled.