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_hwsig,
229 s4_nohwsig, old_ordering, nonvs,
230 sci_force_enable, nobl }
231 See Documentation/power/video.rst for information on
233 s3_beep is for debugging; it makes the PC's speaker beep
234 as soon as the kernel's real-mode entry point is called.
235 s4_hwsig causes the kernel to check the ACPI hardware
236 signature during resume from hibernation, and gracefully
237 refuse to resume if it has changed. This complies with
238 the ACPI specification but not with reality, since
239 Windows does not do this and many laptops do change it
240 on docking. So the default behaviour is to allow resume
241 and simply warn when the signature changes, unless the
242 s4_hwsig option is enabled.
243 s4_nohwsig prevents ACPI hardware signature from being
244 used (or even warned about) during resume.
245 old_ordering causes the ACPI 1.0 ordering of the _PTS
246 control method, with respect to putting devices into
247 low power states, to be enforced (the ACPI 2.0 ordering
248 of _PTS is used by default).
249 nonvs prevents the kernel from saving/restoring the
250 ACPI NVS memory during suspend/hibernation and resume.
251 sci_force_enable causes the kernel to set SCI_EN directly
252 on resume from S1/S3 (which is against the ACPI spec,
253 but some broken systems don't work without it).
254 nobl causes the internal blacklist of systems known to
255 behave incorrectly in some ways with respect to system
256 suspend and resume to be ignored (use wisely).
258 acpi_use_timer_override [HW,ACPI]
259 Use timer override. For some broken Nvidia NF5 boards
260 that require a timer override, but don't have HPET
262 add_efi_memmap [EFI; X86] Include EFI memory map in
263 kernel's map of available physical RAM.
266 { off | try_unsupported }
267 off: disable AGP support
268 try_unsupported: try to drive unsupported chipsets
269 (may crash computer or cause data corruption)
272 See Documentation/sound/alsa-configuration.rst
275 Allow the default userspace alignment fault handler
276 behaviour to be specified. Bit 0 enables warnings,
277 bit 1 enables fixups, and bit 2 sends a segfault.
279 align_va_addr= [X86-64]
280 Align virtual addresses by clearing slice [14:12] when
281 allocating a VMA at process creation time. This option
282 gives you up to 3% performance improvement on AMD F15h
283 machines (where it is enabled by default) for a
284 CPU-intensive style benchmark, and it can vary highly in
285 a microbenchmark depending on workload and compiler.
287 32: only for 32-bit processes
288 64: only for 64-bit processes
289 on: enable for both 32- and 64-bit processes
290 off: disable for both 32- and 64-bit processes
292 alloc_snapshot [FTRACE]
293 Allocate the ftrace snapshot buffer on boot up when the
294 main buffer is allocated. This is handy if debugging
295 and you need to use tracing_snapshot() on boot up, and
296 do not want to use tracing_snapshot_alloc() as it needs
297 to be done where GFP_KERNEL allocations are allowed.
299 allow_mismatched_32bit_el0 [ARM64]
300 Allow execve() of 32-bit applications and setting of the
301 PER_LINUX32 personality on systems where only a strict
302 subset of the CPUs support 32-bit EL0. When this
303 parameter is present, the set of CPUs supporting 32-bit
304 EL0 is indicated by /sys/devices/system/cpu/aarch32_el0
305 and hot-unplug operations may be restricted.
307 See Documentation/arm64/asymmetric-32bit.rst for more
310 amd_iommu= [HW,X86-64]
311 Pass parameters to the AMD IOMMU driver in the system.
313 fullflush - Deprecated, equivalent to iommu.strict=1
314 off - do not initialize any AMD IOMMU found in
316 force_isolation - Force device isolation for all
317 devices. The IOMMU driver is not
318 allowed anymore to lift isolation
319 requirements as needed. This option
320 does not override iommu=pt
321 force_enable - Force enable the IOMMU on platforms known
322 to be buggy with IOMMU enabled. Use this
324 pgtbl_v1 - Use v1 page table for DMA-API (Default).
325 pgtbl_v2 - Use v2 page table for DMA-API.
327 amd_iommu_dump= [HW,X86-64]
328 Enable AMD IOMMU driver option to dump the ACPI table
329 for AMD IOMMU. With this option enabled, AMD IOMMU
330 driver will print ACPI tables for AMD IOMMU during
331 IOMMU initialization.
333 amd_iommu_intr= [HW,X86-64]
334 Specifies one of the following AMD IOMMU interrupt
336 legacy - Use legacy interrupt remapping mode.
337 vapic - Use virtual APIC mode, which allows IOMMU
338 to inject interrupts directly into guest.
339 This mode requires kvm-amd.avic=1.
340 (Default when IOMMU HW support is present.)
342 amijoy.map= [HW,JOY] Amiga joystick support
343 Map of devices attached to JOY0DAT and JOY1DAT
345 See also Documentation/input/joydev/joystick.rst
347 analog.map= [HW,JOY] Analog joystick and gamepad support
348 Specifies type or capabilities of an analog joystick
349 connected to one of 16 gameports
350 Format: <type1>,<type2>,..<type16>
353 Power management functions (SPARCstation-4/5 + deriv.)
355 Disable APC CPU standby support. SPARCstation-Fox does
356 not play well with APC CPU idle - disable it if you have
357 APC and your system crashes randomly.
359 apic= [APIC,X86] Advanced Programmable Interrupt Controller
360 Change the output verbosity while booting
361 Format: { quiet (default) | verbose | debug }
362 Change the amount of debugging information output
363 when initialising the APIC and IO-APIC components.
364 For X86-32, this can also be used to specify an APIC
366 Format: apic=driver_name
367 Examples: apic=bigsmp
369 apic_extnmi= [APIC,X86] External NMI delivery setting
370 Format: { bsp (default) | all | none }
371 bsp: External NMI is delivered only to CPU 0
372 all: External NMIs are broadcast to all CPUs as a
374 none: External NMI is masked for all CPUs. This is
375 useful so that a dump capture kernel won't be
379 See Documentation/networking/ipv6.rst.
381 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
382 Limit apic dumping. The parameter defines the maximal
383 number of local apics being dumped. Also it is possible
384 to set it to "all" by meaning -- no limit here.
385 Format: { 1 (default) | 2 | ... | all }.
386 The parameter valid if only apic=debug or
387 apic=verbose is specified.
388 Example: apic=debug show_lapic=all
390 apm= [APM] Advanced Power Management
391 See header of arch/x86/kernel/apm_32.c.
393 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
394 Format: <io>,<irq>,<nodeID>
396 arm64.nobti [ARM64] Unconditionally disable Branch Target
397 Identification support
399 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication
402 arm64.nomte [ARM64] Unconditionally disable Memory Tagging Extension
405 arm64.nosve [ARM64] Unconditionally disable Scalable Vector
408 arm64.nosme [ARM64] Unconditionally disable Scalable Matrix
413 atarimouse= [HW,MOUSE] Atari Mouse
415 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
416 EzKey and similar keyboards
418 atkbd.reset= [HW] Reset keyboard during initialization
420 atkbd.set= [HW] Select keyboard code set
421 Format: <int> (2 = AT (default), 3 = PS/2)
423 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
426 atkbd.softraw= [HW] Choose between synthetic and real raw mode
427 Format: <bool> (0 = real, 1 = synthetic (default))
429 atkbd.softrepeat= [HW]
430 Use software keyboard repeat
432 audit= [KNL] Enable the audit sub-system
433 Format: { "0" | "1" | "off" | "on" }
434 0 | off - kernel audit is disabled and can not be
435 enabled until the next reboot
436 unset - kernel audit is initialized but disabled and
437 will be fully enabled by the userspace auditd.
438 1 | on - kernel audit is initialized and partially
439 enabled, storing at most audit_backlog_limit
440 messages in RAM until it is fully enabled by the
444 audit_backlog_limit= [KNL] Set the audit queue size limit.
445 Format: <int> (must be >=0)
448 bau= [X86_UV] Enable the BAU on SGI UV. The default
449 behavior is to disable the BAU (i.e. bau=0).
450 Format: { "0" | "1" }
453 unset - Disable the BAU.
455 baycom_epp= [HW,AX25]
458 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
460 See header of drivers/net/hamradio/baycom_par.c.
462 baycom_ser_fdx= [HW,AX25]
463 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
464 Format: <io>,<irq>,<mode>[,<baud>]
465 See header of drivers/net/hamradio/baycom_ser_fdx.c.
467 baycom_ser_hdx= [HW,AX25]
468 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
469 Format: <io>,<irq>,<mode>
470 See header of drivers/net/hamradio/baycom_ser_hdx.c.
473 Disable BERT OS support on buggy BIOSes.
475 bgrt_disable [ACPI][X86]
476 Disable BGRT to avoid flickering OEM logo.
478 blkdevparts= Manual partition parsing of block device(s) for
479 embedded devices based on command line input.
480 See Documentation/block/cmdline-partition.rst
482 boot_delay= Milliseconds to delay each printk during boot.
483 Values larger than 10 seconds (10000) are changed to
488 Extended command line options can be added to an initrd
489 and this will cause the kernel to look for it.
491 See Documentation/admin-guide/bootconfig.rst
493 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
494 bttv.radio= Most important insmod options are available as
496 bttv.pll= See Documentation/admin-guide/media/bttv.rst
499 bulk_remove=off [PPC] This parameter disables the use of the pSeries
500 firmware feature for flushing multiple hpte entries
503 c101= [NET] Moxa C101 synchronous serial card
505 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
506 Sometimes CPU hardware bugs make them report the cache
507 size incorrectly. The kernel will attempt work arounds
508 to fix known problems, but for some CPUs it is not
509 possible to determine what the correct size should be.
510 This option provides an override for these situations.
513 [NET] Specifies amount of time (in seconds) that
514 the kernel should wait for a network carrier. By default
515 it waits 120 seconds.
517 ca_keys= [KEYS] This parameter identifies a specific key(s) on
518 the system trusted keyring to be used for certificate
520 format: { id:<keyid> | builtin }
522 cca= [MIPS] Override the kernel pages' cache coherency
523 algorithm. Accepted values range from 0 to 7
524 inclusive. See arch/mips/include/asm/pgtable-bits.h
525 for platform specific values (SB1, Loongson3 and
528 ccw_timeout_log [S390]
529 See Documentation/s390/common_io.rst for details.
531 cgroup_disable= [KNL] Disable a particular controller or optional feature
532 Format: {name of the controller(s) or feature(s) to disable}
533 The effects of cgroup_disable=foo are:
534 - foo isn't auto-mounted if you mount all cgroups in
536 - foo isn't visible as an individually mountable
538 - if foo is an optional feature then the feature is
539 disabled and corresponding cgroup files are not
541 {Currently only "memory" controller deal with this and
542 cut the overhead, others just disable the usage. So
543 only cgroup_disable=memory is actually worthy}
544 Specifying "pressure" disables per-cgroup pressure
545 stall information accounting feature
547 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
548 Format: { { controller | "all" | "named" }
549 [,{ controller | "all" | "named" }...] }
550 Like cgroup_disable, but only applies to cgroup v1;
551 the blacklisted controllers remain available in cgroup2.
552 "all" blacklists all controllers and "named" disables
553 named mounts. Specifying both "all" and "named" disables
556 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
558 nosocket -- Disable socket memory accounting.
559 nokmem -- Disable kernel memory accounting.
561 checkreqprot= [SELINUX] Set initial checkreqprot flag value.
562 Format: { "0" | "1" }
563 See security/selinux/Kconfig help text.
564 0 -- check protection applied by kernel (includes
565 any implied execute protection).
566 1 -- check protection requested by application.
567 Default value is set via a kernel config option.
568 Value can be changed at runtime via
569 /sys/fs/selinux/checkreqprot.
570 Setting checkreqprot to 1 is deprecated.
573 See Documentation/s390/common_io.rst for details.
575 clearcpuid=X[,X...] [X86]
576 Disable CPUID feature X for the kernel. See
577 arch/x86/include/asm/cpufeatures.h for the valid bit
578 numbers X. Note the Linux-specific bits are not necessarily
579 stable over kernel options, but the vendor-specific
581 X can also be a string as appearing in the flags: line
582 in /proc/cpuinfo which does not have the above
583 instability issue. However, not all features have names
585 Note that using this option will taint your kernel.
586 Also note that user programs calling CPUID directly
587 or using the feature without checking anything
588 will still see it. This just prevents it from
589 being used by the kernel or shown in /proc/cpuinfo.
590 Also note the kernel might malfunction if you disable
595 Prevents the clock framework from automatically gating
596 clocks that have not been explicitly enabled by a Linux
597 device driver but are enabled in hardware at reset or
598 by the bootloader/firmware. Note that this does not
599 force such clocks to be always-on nor does it reserve
600 those clocks in any way. This parameter is useful for
601 debug and development, but should not be needed on a
602 platform with proper driver support. For more
603 information, see Documentation/driver-api/clk.rst.
605 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
607 Forces specified clocksource (if available) to be used
608 when calculating gettimeofday(). If specified
609 clocksource is not available, it defaults to PIT.
610 Format: { pit | tsc | cyclone | pmtmr }
612 clocksource= Override the default clocksource
614 Override the default clocksource and use the clocksource
615 with the name specified.
616 Some clocksource names to choose from, depending on
618 [all] jiffies (this is the base, fallback clocksource)
620 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
621 pxa_timer,timer3,32k_counter,timer0_1
622 [X86-32] pit,hpet,tsc;
623 scx200_hrt on Geode; cyclone on IBM x440
631 clocksource.arm_arch_timer.evtstrm=
634 Enable/disable the eventstream feature of the ARM
635 architected timer so that code using WFE-based polling
636 loops can be debugged more effectively on production
639 clocksource.max_cswd_read_retries= [KNL]
640 Number of clocksource_watchdog() retries due to
641 external delays before the clock will be marked
642 unstable. Defaults to two retries, that is,
643 three attempts to read the clock under test.
645 clocksource.verify_n_cpus= [KNL]
646 Limit the number of CPUs checked for clocksources
647 marked with CLOCK_SOURCE_VERIFY_PERCPU that
648 are marked unstable due to excessive skew.
649 A negative value says to check all CPUs, while
650 zero says not to check any. Values larger than
651 nr_cpu_ids are silently truncated to nr_cpu_ids.
652 The actual CPUs are chosen randomly, with
653 no replacement if the same CPU is chosen twice.
655 clocksource-wdtest.holdoff= [KNL]
656 Set the time in seconds that the clocksource
657 watchdog test waits before commencing its tests.
658 Defaults to zero when built as a module and to
659 10 seconds when built into the kernel.
661 cma=nn[MG]@[start[MG][-end[MG]]]
663 Sets the size of kernel global memory area for
664 contiguous memory allocations and optionally the
665 placement constraint by the physical address range of
666 memory allocations. A value of 0 disables CMA
667 altogether. For more information, see
668 kernel/dma/contiguous.c
672 Sets the size of kernel per-numa memory area for
673 contiguous memory allocations. A value of 0 disables
674 per-numa CMA altogether. And If this option is not
675 specificed, the default value is 0.
676 With per-numa CMA enabled, DMA users on node nid will
677 first try to allocate buffer from the pernuma area
678 which is located in node nid, if the allocation fails,
679 they will fallback to the global default memory area.
681 cmo_free_hint= [PPC] Format: { yes | no }
682 Specify whether pages are marked as being inactive
683 when they are freed. This is used in CMO environments
684 to determine OS memory pressure for page stealing by
688 coherent_pool=nn[KMG] [ARM,KNL]
689 Sets the size of memory pool for coherent, atomic dma
690 allocations, by default set to 256K.
692 com20020= [HW,NET] ARCnet - COM20020 chipset
694 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
696 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
700 ARCnet - COM90xx chipset (memory-mapped buffers)
701 Format: <io>[,<irq>[,<memstart>]]
703 condev= [HW,S390] console device
706 con3215_drop= [S390] 3215 console drop mode.
708 When set to true, drop data on the 3215 console when
709 the console buffer is full. In this case the
710 operator using a 3270 terminal emulator (for example
711 x3270) does not have to enter the clear key for the
712 console output to advance and the kernel to continue.
713 This leads to a much faster boot time when a 3270
714 terminal emulator is active. If no 3270 terminal
715 emulator is used, this parameter has no effect.
717 console= [KNL] Output console device and options.
719 tty<n> Use the virtual console device <n>.
723 Use the specified serial port. The options are of
724 the form "bbbbpnf", where "bbbb" is the baud rate,
725 "p" is parity ("n", "o", or "e"), "n" is number of
726 bits, and "f" is flow control ("r" for RTS or
727 omit it). Default is "9600n8".
729 See Documentation/admin-guide/serial-console.rst for more
731 Documentation/networking/netconsole.rst for an
734 uart[8250],io,<addr>[,options]
735 uart[8250],mmio,<addr>[,options]
736 uart[8250],mmio16,<addr>[,options]
737 uart[8250],mmio32,<addr>[,options]
738 uart[8250],0x<addr>[,options]
739 Start an early, polled-mode console on the 8250/16550
740 UART at the specified I/O port or MMIO address,
741 switching to the matching ttyS device later.
742 MMIO inter-register address stride is either 8-bit
743 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
744 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
745 to be equivalent to 'mmio'. 'options' are specified in
746 the same format described for ttyS above; if unspecified,
747 the h/w is not re-initialized.
749 hvc<n> Use the hypervisor console device <n>. This is for
750 both Xen and PowerPC hypervisors.
753 Use to disable console output, i.e., to have kernel
754 console messages discarded.
755 This must be the only console= parameter used on the
758 If the device connected to the port is not a TTY but a braille
759 device, prepend "brl," before the device type, for instance
761 For now, only VisioBraille is supported.
764 [KNL] Change console messages format
766 By default we print messages on consoles in
767 "[time stamp] text\n" format (time stamp may not be
768 printed, depending on CONFIG_PRINTK_TIME or
769 `printk_time' param).
771 Switch to syslog format: "<%u>[time stamp] text\n"
772 IOW, each message will have a facility and loglevel
773 prefix. The format is similar to one used by syslog()
774 syscall, or to executing "dmesg -S --raw" or to reading
777 consoleblank= [KNL] The console blank (screen saver) timeout in
778 seconds. A value of 0 disables the blank timer.
782 [KNL] Change the default value for
783 /proc/<pid>/coredump_filter.
784 See also Documentation/filesystems/proc.rst.
786 coresight_cpu_debug.enable
789 Enable/disable the CPU sampling based debugging.
790 0: default value, disable debugging
791 1: enable debugging at boot time
793 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
795 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
797 cpu0_hotplug [X86] Turn on CPU0 hotplug feature when
798 CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off.
799 Some features depend on CPU0. Known dependencies are:
800 1. Resume from suspend/hibernate depends on CPU0.
801 Suspend/hibernate will fail if CPU0 is offline and you
802 need to online CPU0 before suspend/hibernate.
803 2. PIC interrupts also depend on CPU0. CPU0 can't be
804 removed if a PIC interrupt is detected.
805 It's said poweroff/reboot may depend on CPU0 on some
806 machines although I haven't seen such issues so far
807 after CPU0 is offline on a few tested machines.
808 If the dependencies are under your control, you can
809 turn on cpu0_hotplug.
811 cpuidle.off=1 [CPU_IDLE]
812 disable the cpuidle sub-system
815 [CPU_IDLE] Name of the cpuidle governor to use.
817 cpufreq.off=1 [CPU_FREQ]
818 disable the cpufreq sub-system
820 cpufreq.default_governor=
821 [CPU_FREQ] Name of the default cpufreq governor or
822 policy to use. This governor must be registered in the
823 kernel before the cpufreq driver probes.
826 [X86] Delay for N microsec between assert and de-assert
827 of APIC INIT to start processors. This delay occurs
828 on every CPU online, such as boot, and resume from suspend.
831 crash_kexec_post_notifiers
832 Run kdump after running panic-notifiers and dumping
833 kmsg. This only for the users who doubt kdump always
834 succeeds in any situation.
835 Note that this also increases risks of kdump failure,
836 because some panic notifiers can make the crashed
837 kernel more unstable.
839 crashkernel=size[KMG][@offset[KMG]]
840 [KNL] Using kexec, Linux can switch to a 'crash kernel'
841 upon panic. This parameter reserves the physical
842 memory region [offset, offset + size] for that kernel
843 image. If '@offset' is omitted, then a suitable offset
844 is selected automatically.
845 [KNL, X86-64, ARM64] Select a region under 4G first, and
846 fall back to reserve region above 4G when '@offset'
847 hasn't been specified.
848 See Documentation/admin-guide/kdump/kdump.rst for further details.
850 crashkernel=range1:size1[,range2:size2,...][@offset]
851 [KNL] Same as above, but depends on the memory
852 in the running system. The syntax of range is
853 start-[end] where start and end are both
854 a memory unit (amount[KMG]). See also
855 Documentation/admin-guide/kdump/kdump.rst for an example.
857 crashkernel=size[KMG],high
858 [KNL, X86-64, ARM64] range could be above 4G. Allow kernel
859 to allocate physical memory region from top, so could
860 be above 4G if system have more than 4G ram installed.
861 Otherwise memory region will be allocated below 4G, if
863 It will be ignored if crashkernel=X is specified.
864 crashkernel=size[KMG],low
865 [KNL, X86-64, ARM64] range under 4G. When crashkernel=X,high
866 is passed, kernel could allocate physical memory region
867 above 4G, that cause second kernel crash on system
868 that require some amount of low memory, e.g. swiotlb
869 requires at least 64M+32K low memory, also enough extra
870 low memory is needed to make sure DMA buffers for 32-bit
871 devices won't run out. Kernel would try to allocate
872 default size of memory below 4G automatically. The default
873 size is platform dependent.
874 --> x86: max(swiotlb_size_or_default() + 8MiB, 256MiB)
876 This one lets the user specify own low range under 4G
877 for second kernel instead.
878 0: to disable low allocation.
879 It will be ignored when crashkernel=X,high is not used
880 or memory reserved is below 4G.
883 [KNL] Disable crypto self-tests
888 cs89x0_media= [HW,NET]
889 Format: { rj45 | aui | bnc }
891 csdlock_debug= [KNL] Enable debug add-ons of cross-CPU function call
892 handling. When switched on, additional debug data is
893 printed to the console in case a hanging CPU is
894 detected, and that CPU is pinged again in order to try
895 to resolve the hang situation.
896 0: disable csdlock debugging (default)
897 1: enable basic csdlock debugging (minor impact)
898 ext: enable extended csdlock debugging (more impact,
902 See header of drivers/s390/block/dasd_devmap.c.
904 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
905 (one device per port)
906 Format: <port#>,<type>
907 See also Documentation/input/devices/joystick-parport.rst
909 debug [KNL] Enable kernel debugging (events log level).
912 [KNL] Enable printing [hashed] pointers early in the
913 boot sequence. If enabled, we use a weak hash instead
914 of siphash to hash pointers. Use this option if you are
915 seeing instances of '(___ptrval___)') and need to see a
916 value (hashed pointer) instead. Cryptographically
917 insecure, please do not use on production kernels.
920 [KNL] verbose locking self-tests
922 Print debugging info while doing the locking API
924 Bitmask for the various LOCKTYPE_ tests. Defaults to 0
925 (no extra messages), setting it to -1 (all bits set)
926 will print _a_lot_ more information - normally only
927 useful to lockdep developers.
929 debug_objects [KNL] Enable object debugging
932 [KNL] Disable object debugging
934 debug_guardpage_minorder=
935 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
936 parameter allows control of the order of pages that will
937 be intentionally kept free (and hence protected) by the
938 buddy allocator. Bigger value increase the probability
939 of catching random memory corruption, but reduce the
940 amount of memory for normal system use. The maximum
941 possible value is MAX_ORDER/2. Setting this parameter
942 to 1 or 2 should be enough to identify most random
943 memory corruption problems caused by bugs in kernel or
944 driver code when a CPU writes to (or reads from) a
945 random memory location. Note that there exists a class
946 of memory corruptions problems caused by buggy H/W or
947 F/W or by drivers badly programing DMA (basically when
948 memory is written at bus level and the CPU MMU is
949 bypassed) which are not detectable by
950 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
951 tracking down these problems.
954 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
955 enables the feature at boot time. By default, it is
956 disabled and the system will work mostly the same as a
957 kernel built without CONFIG_DEBUG_PAGEALLOC.
958 Note: to get most of debug_pagealloc error reports, it's
959 useful to also enable the page_owner functionality.
960 on: enable the feature
962 debugfs= [KNL] This parameter enables what is exposed to userspace
963 and debugfs internal clients.
964 Format: { on, no-mount, off }
965 on: All functions are enabled.
967 Filesystem is not registered but kernel clients can
968 access APIs and a crashkernel can be used to read
969 its content. There is nothing to mount.
970 off: Filesystem is not registered and clients
971 get a -EPERM as result when trying to register files
972 or directories within debugfs.
973 This is equivalent of the runtime functionality if
974 debugfs was not enabled in the kernel at all.
975 Default value is set in build-time with a kernel configuration.
977 debugpat [X86] Enable PAT debugging
980 [HW] The size of the default HugeTLB page. This is
981 the size represented by the legacy /proc/ hugepages
982 APIs. In addition, this is the default hugetlb size
983 used for shmget(), mmap() and mounting hugetlbfs
984 filesystems. If not specified, defaults to the
985 architecture's default huge page size. Huge page
986 sizes are architecture dependent. See also
987 Documentation/admin-guide/mm/hugetlbpage.rst.
990 deferred_probe_timeout=
991 [KNL] Debugging option to set a timeout in seconds for
992 deferred probe to give up waiting on dependencies to
993 probe. Only specific dependencies (subsystems or
994 drivers) that have opted in will be ignored. A timeout
995 of 0 will timeout at the end of initcalls. If the time
996 out hasn't expired, it'll be restarted by each
997 successful driver registration. This option will also
998 dump out devices still on the deferred probe list after
1001 delayacct [KNL] Enable per-task delay accounting
1003 dell_smm_hwmon.ignore_dmi=
1004 [HW] Continue probing hardware even if DMI data
1005 indicates that the driver is running on unsupported
1008 dell_smm_hwmon.force=
1009 [HW] Activate driver even if SMM BIOS signature does
1010 not match list of supported models and enable otherwise
1011 blacklisted features.
1013 dell_smm_hwmon.power_status=
1014 [HW] Report power status in /proc/i8k
1015 (disabled by default).
1017 dell_smm_hwmon.restricted=
1018 [HW] Allow controlling fans only if SYS_ADMIN
1021 dell_smm_hwmon.fan_mult=
1022 [HW] Factor to multiply fan speed with.
1024 dell_smm_hwmon.fan_max=
1025 [HW] Maximum configurable fan speed.
1028 Format: { on | off | def_only | inf_only | always }
1029 on: s390 zlib hardware support for compression on
1030 level 1 and decompression (default)
1031 off: No s390 zlib hardware support
1032 def_only: s390 zlib hardware support for deflate
1033 only (compression on level 1)
1034 inf_only: s390 zlib hardware support for inflate
1035 only (decompression)
1036 always: Same as 'on' but ignores the selected compression
1037 level always using hardware support (used for debugging)
1039 dhash_entries= [KNL]
1040 Set number of hash buckets for dentry cache.
1042 disable_1tb_segments [PPC]
1043 Disables the use of 1TB hash page table segments. This
1044 causes the kernel to fall back to 256MB segments which
1045 can be useful when debugging issues that require an SLB
1049 Limits the number of kernel SLB entries, and flushes
1050 them frequently to increase the rate of SLB faults
1051 on kernel addresses.
1054 Limits the number of kernel HPT entries in the hash
1055 page table to increase the rate of hash page table
1056 faults on kernel addresses.
1059 See Documentation/networking/ipv6.rst.
1062 Disable RADIX MMU mode on POWER9
1064 radix_hcall_invalidate=on [PPC/PSERIES]
1065 Disable RADIX GTSE feature and use hcall for TLB
1069 Disable TLBIE instruction. Currently does not work
1070 with KVM, with HASH MMU, or with coherent accelerators.
1072 disable_cpu_apicid= [X86,APIC,SMP]
1074 The number of initial APIC ID for the
1075 corresponding CPU to be disabled at boot,
1076 mostly used for the kdump 2nd kernel to
1077 disable BSP to wake up multiple CPUs without
1078 causing system reset or hang due to sending
1079 INIT from AP to BSP.
1081 disable_ddw [PPC/PSERIES]
1082 Disable Dynamic DMA Window support. Use this
1083 to workaround buggy firmware.
1085 disable_ipv6= [IPV6]
1086 See Documentation/networking/ipv6.rst.
1088 disable_mtrr_cleanup [X86]
1089 The kernel tries to adjust MTRR layout from continuous
1090 to discrete, to make X server driver able to add WB
1091 entry later. This parameter disables that.
1093 disable_mtrr_trim [X86, Intel and AMD only]
1094 By default the kernel will trim any uncacheable
1095 memory out of your available memory pool based on
1096 MTRR settings. This parameter disables that behavior,
1097 possibly causing your machine to run very slowly.
1099 disable_timer_pin_1 [X86]
1100 Disable PIN 1 of APIC timer
1101 Can be useful to work around chipset bugs.
1103 dis_ucode_ldr [X86] Disable the microcode loader.
1105 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
1106 this option disables the debugging code at boot.
1108 dma_debug_entries=<number>
1109 This option allows to tune the number of preallocated
1110 entries for DMA-API debugging code. One entry is
1111 required per DMA-API allocation. Use this if the
1112 DMA-API debugging code disables itself because the
1113 architectural default is too low.
1115 dma_debug_driver=<driver_name>
1116 With this option the DMA-API debugging driver
1117 filter feature can be enabled at boot time. Just
1118 pass the driver to filter for as the parameter.
1119 The filter can be disabled or changed to another
1120 driver later using sysfs.
1122 driver_async_probe= [KNL]
1123 List of driver names to be probed asynchronously. *
1124 matches with all driver names. If * is specified, the
1125 rest of the listed driver names are those that will NOT
1127 Format: <driver_name1>,<driver_name2>...
1129 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
1130 Broken monitors, graphic adapters, KVMs and EDIDless
1131 panels may send no or incorrect EDID data sets.
1132 This parameter allows to specify an EDID data sets
1133 in the /lib/firmware directory that are used instead.
1134 Generic built-in EDID data sets are used, if one of
1135 edid/1024x768.bin, edid/1280x1024.bin,
1136 edid/1680x1050.bin, or edid/1920x1080.bin is given
1137 and no file with the same name exists. Details and
1138 instructions how to build your own EDID data are
1139 available in Documentation/admin-guide/edid.rst. An EDID
1140 data set will only be used for a particular connector,
1141 if its name and a colon are prepended to the EDID
1142 name. Each connector may use a unique EDID data
1143 set by separating the files with a comma. An EDID
1144 data set with no connector name will be used for
1145 any connectors not explicitly specified.
1150 Format: {"off" | "known"}
1151 Control how the dt_cpu_ftrs device-tree binding is
1152 used for CPU feature discovery and setup (if it
1154 off: Do not use it, fall back to legacy cpu table.
1155 known: Do not pass through unknown features to guests
1156 or userspace, only those that the kernel is aware of.
1158 dump_apple_properties [X86]
1159 Dump name and content of EFI device properties on
1160 x86 Macs. Useful for driver authors to determine
1161 what data is available or for reverse-engineering.
1163 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
1164 <module>.dyndbg[="val"]
1165 Enable debug messages at boot time. See
1166 Documentation/admin-guide/dynamic-debug-howto.rst
1169 nopku [X86] Disable Memory Protection Keys CPU feature found
1172 <module>.async_probe[=<bool>] [KNL]
1173 If no <bool> value is specified or if the value
1174 specified is not a valid <bool>, enable asynchronous
1175 probe on this module. Otherwise, enable/disable
1176 asynchronous probe on this module as indicated by the
1177 <bool> value. See also: module.async_probe
1179 early_ioremap_debug [KNL]
1180 Enable debug messages in early_ioremap support. This
1181 is useful for tracking down temporary early mappings
1182 which are not unmapped.
1184 earlycon= [KNL] Output early console device and options.
1186 When used with no options, the early console is
1187 determined by stdout-path property in device tree's
1188 chosen node or the ACPI SPCR table if supported by
1191 cdns,<addr>[,options]
1192 Start an early, polled-mode console on a Cadence
1193 (xuartps) serial port at the specified address. Only
1194 supported option is baud rate. If baud rate is not
1195 specified, the serial port must already be setup and
1198 uart[8250],io,<addr>[,options]
1199 uart[8250],mmio,<addr>[,options]
1200 uart[8250],mmio32,<addr>[,options]
1201 uart[8250],mmio32be,<addr>[,options]
1202 uart[8250],0x<addr>[,options]
1203 Start an early, polled-mode console on the 8250/16550
1204 UART at the specified I/O port or MMIO address.
1205 MMIO inter-register address stride is either 8-bit
1206 (mmio) or 32-bit (mmio32 or mmio32be).
1207 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1208 to be equivalent to 'mmio'. 'options' are specified
1209 in the same format described for "console=ttyS<n>"; if
1210 unspecified, the h/w is not initialized.
1214 Start an early, polled-mode console on a pl011 serial
1215 port at the specified address. The pl011 serial port
1216 must already be setup and configured. Options are not
1217 yet supported. If 'mmio32' is specified, then only
1218 the driver will use only 32-bit accessors to read/write
1219 the device registers.
1222 Start an early console on a litex serial port at the
1223 specified address. The serial port must already be
1224 setup and configured. Options are not yet supported.
1227 Start an early, polled-mode console on a meson serial
1228 port at the specified address. The serial port must
1229 already be setup and configured. Options are not yet
1233 Start an early, polled-mode console on an msm serial
1234 port at the specified address. The serial port
1235 must already be setup and configured. Options are not
1238 msm_serial_dm,<addr>
1239 Start an early, polled-mode console on an msm serial
1240 dm port at the specified address. The serial port
1241 must already be setup and configured. Options are not
1245 Start an early, polled-mode console on a serial port
1246 of an Actions Semi SoC, such as S500 or S900, at the
1247 specified address. The serial port must already be
1248 setup and configured. Options are not yet supported.
1251 Start an early, polled-mode console on a serial port
1252 of an RDA Micro SoC, such as RDA8810PL, at the
1253 specified address. The serial port must already be
1254 setup and configured. Options are not yet supported.
1257 Use RISC-V SBI (Supervisor Binary Interface) for early
1260 smh Use ARM semihosting calls for early console.
1268 Use early console provided by serial driver available
1269 on Samsung SoCs, requires selecting proper type and
1270 a correct base address of the selected UART port. The
1271 serial port must already be setup and configured.
1272 Options are not yet supported.
1275 Start an early, polled-mode console on a lantiq serial
1276 (lqasc) port at the specified address. The serial port
1277 must already be setup and configured. Options are not
1282 Use early console provided by Freescale LP UART driver
1283 found on Freescale Vybrid and QorIQ LS1021A processors.
1284 A valid base address must be provided, and the serial
1285 port must already be setup and configured.
1289 Start an early, polled-mode, output-only console on the
1290 Freescale i.MX UART at the specified address. The UART
1291 must already be setup and configured.
1294 Start an early, polled-mode console on the
1295 Armada 3700 serial port at the specified
1296 address. The serial port must already be setup
1297 and configured. Options are not yet supported.
1300 Start an early, polled-mode console on a Qualcomm
1301 Generic Interface (GENI) based serial port at the
1302 specified address. The serial port must already be
1303 setup and configured. Options are not yet supported.
1306 Start an early, unaccelerated console on the EFI
1307 memory mapped framebuffer (if available). On cache
1308 coherent non-x86 systems that use system memory for
1309 the framebuffer, pass the 'ram' option so that it is
1310 mapped with the correct attributes.
1313 Use early console provided by Freescale LINFlexD UART
1314 serial driver for NXP S32V234 SoCs. A valid base
1315 address must be provided, and the serial port must
1316 already be setup and configured.
1318 earlyprintk= [X86,SH,ARM,M68k,S390]
1322 earlyprintk=serial[,ttySn[,baudrate]]
1323 earlyprintk=serial[,0x...[,baudrate]]
1324 earlyprintk=ttySn[,baudrate]
1325 earlyprintk=dbgp[debugController#]
1326 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1327 earlyprintk=xdbc[xhciController#]
1329 earlyprintk is useful when the kernel crashes before
1330 the normal console is initialized. It is not enabled by
1331 default because it has some cosmetic problems.
1333 Append ",keep" to not disable it when the real console
1336 Only one of vga, serial, or usb debug port can
1339 Currently only ttyS0 and ttyS1 may be specified by
1340 name. Other I/O ports may be explicitly specified
1341 on some architectures (x86 and arm at least) by
1342 replacing ttySn with an I/O port address, like this:
1343 earlyprintk=serial,0x1008,115200
1344 You can find the port for a given device in
1345 /proc/tty/driver/serial:
1346 2: uart:ST16650V2 port:00001008 irq:18 ...
1348 Interaction with the standard serial driver is not
1351 The VGA output is eventually overwritten by
1354 The xen option can only be used in Xen domains.
1356 The sclp output can only be used on s390.
1358 The optional "force" to "pciserial" enables use of a
1359 PCI device even when its classcode is not of the
1362 edac_report= [HW,EDAC] Control how to report EDAC event
1363 Format: {"on" | "off" | "force"}
1364 on: enable EDAC to report H/W event. May be overridden
1365 by other higher priority error reporting module.
1366 off: disable H/W event reporting through EDAC.
1367 force: enforce the use of EDAC to report H/W event.
1371 Format: {"off" | "on" | "skip[mbr]"}
1374 Format: { "debug", "disable_early_pci_dma",
1375 "nochunk", "noruntime", "nosoftreserve",
1376 "novamap", "no_disable_early_pci_dma" }
1377 debug: enable misc debug output.
1378 disable_early_pci_dma: disable the busmaster bit on all
1379 PCI bridges while in the EFI boot stub.
1380 nochunk: disable reading files in "chunks" in the EFI
1381 boot stub, as chunking can cause problems with some
1382 firmware implementations.
1383 noruntime : disable EFI runtime services support
1384 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1385 attribute may cause the kernel to reserve the
1386 memory range for a memory mapping driver to
1387 claim. Specify efi=nosoftreserve to disable this
1388 reservation and treat the memory by its base type
1389 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1390 novamap: do not call SetVirtualAddressMap().
1391 no_disable_early_pci_dma: Leave the busmaster bit set
1392 on all PCI bridges while in the EFI boot stub
1394 efi_no_storage_paranoia [EFI; X86]
1395 Using this parameter you can use more than 50% of
1396 your efi variable storage. Use this parameter only if
1397 you are really sure that your UEFI does sane gc and
1398 fulfills the spec otherwise your board may brick.
1400 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1401 Add arbitrary attribute to specific memory range by
1402 updating original EFI memory map.
1403 Region of memory which aa attribute is added to is
1406 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1407 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1408 attribute is added to range 0x100000000-0x180000000 and
1409 0x10a0000000-0x1120000000.
1411 If efi_fake_mem=8G@9G:0x40000 is specified, the
1412 EFI_MEMORY_SP(0x40000) attribute is added to
1413 range 0x240000000-0x43fffffff.
1415 Using this parameter you can do debugging of EFI memmap
1416 related features. For example, you can do debugging of
1417 Address Range Mirroring feature even if your box
1418 doesn't support it, or mark specific memory as
1421 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1422 that is to be dynamically loaded by Linux. If there are
1423 multiple variables with the same name but with different
1424 vendor GUIDs, all of them will be loaded. See
1425 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1428 eisa_irq_edge= [PARISC,HW]
1429 See header of drivers/parisc/eisa.c.
1431 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1434 This is designed to be used in conjunction with
1435 the boot argument: earlyprintk=vga
1437 This parameter works in place of the kgdboc parameter
1438 but can only be used if the backing tty is available
1439 very early in the boot process. For early debugging
1440 via a serial port see kgdboc_earlycon instead.
1443 See comment before function elanfreq_setup() in
1444 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1446 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1447 Specifies physical address of start of kernel core
1448 image elf header and optionally the size. Generally
1449 kexec loader will pass this option to capture kernel.
1450 See Documentation/admin-guide/kdump/kdump.rst for details.
1452 enable_mtrr_cleanup [X86]
1453 The kernel tries to adjust MTRR layout from continuous
1454 to discrete, to make X server driver able to add WB
1455 entry later. This parameter enables that.
1457 enable_timer_pin_1 [X86]
1458 Enable PIN 1 of APIC timer
1459 Can be useful to work around chipset bugs
1460 (in particular on some ATI chipsets).
1461 The kernel tries to set a reasonable default.
1463 enforcing= [SELINUX] Set initial enforcing status.
1465 See security/selinux/Kconfig help text.
1466 0 -- permissive (log only, no denials).
1467 1 -- enforcing (deny and log).
1469 Value can be changed at runtime via
1470 /sys/fs/selinux/enforce.
1473 Disable Error Record Serialization Table (ERST)
1476 ether= [HW,NET] Ethernet cards parameters
1477 This option is obsoleted by the "netdev=" option, which
1478 has equivalent usage. See its documentation for details.
1482 Permit 'security.evm' to be updated regardless of
1483 current integrity status.
1485 early_page_ext [KNL] Enforces page_ext initialization to earlier
1486 stages so cover more early boot allocations.
1487 Please note that as side effect some optimizations
1488 might be disabled to achieve that (e.g. parallelized
1489 memory initialization is disabled) so the boot process
1490 might take longer, especially on systems with a lot of
1491 memory. Available with CONFIG_PAGE_EXTENSION=y.
1496 fail_make_request=[KNL]
1497 General fault injection mechanism.
1498 Format: <interval>,<probability>,<space>,<times>
1499 See also Documentation/fault-injection/.
1502 Format: { initns | none }
1503 See Documentation/admin-guide/sysctl/net.rst for
1504 fb_tunnels_only_for_init_ns
1507 See Documentation/admin-guide/blockdev/floppy.rst.
1509 force_pal_cache_flush
1510 [IA-64] Avoid check_sal_cache_flush which may hang on
1511 buggy SAL_CACHE_FLUSH implementations. Using this
1512 parameter will force ia64_sal_cache_flush to call
1513 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1516 Forcefully enable Physical Address Extension (PAE).
1517 Many Pentium M systems disable PAE but may have a
1518 functionally usable PAE implementation.
1519 Warning: use of this parameter will taint the kernel
1520 and may cause unknown problems.
1523 [FTRACE] will set and start the specified tracer
1524 as early as possible in order to facilitate early
1527 ftrace_boot_snapshot
1528 [FTRACE] On boot up, a snapshot will be taken of the
1529 ftrace ring buffer that can be read at:
1530 /sys/kernel/tracing/snapshot.
1531 This is useful if you need tracing information from kernel
1532 boot up that is likely to be overridden by user space
1533 start up functionality.
1535 ftrace_dump_on_oops[=orig_cpu]
1536 [FTRACE] will dump the trace buffers on oops.
1537 If no parameter is passed, ftrace will dump
1538 buffers of all CPUs, but if you pass orig_cpu, it will
1539 dump only the buffer of the CPU that triggered the
1542 ftrace_filter=[function-list]
1543 [FTRACE] Limit the functions traced by the function
1544 tracer at boot up. function-list is a comma-separated
1545 list of functions. This list can be changed at run
1546 time by the set_ftrace_filter file in the debugfs
1549 ftrace_notrace=[function-list]
1550 [FTRACE] Do not trace the functions specified in
1551 function-list. This list can be changed at run time
1552 by the set_ftrace_notrace file in the debugfs
1555 ftrace_graph_filter=[function-list]
1556 [FTRACE] Limit the top level callers functions traced
1557 by the function graph tracer at boot up.
1558 function-list is a comma-separated list of functions
1559 that can be changed at run time by the
1560 set_graph_function file in the debugfs tracing directory.
1562 ftrace_graph_notrace=[function-list]
1563 [FTRACE] Do not trace from the functions specified in
1564 function-list. This list is a comma-separated list of
1565 functions that can be changed at run time by the
1566 set_graph_notrace file in the debugfs tracing directory.
1568 ftrace_graph_max_depth=<uint>
1569 [FTRACE] Used with the function graph tracer. This is
1570 the max depth it will trace into a function. This value
1571 can be changed at run time by the max_graph_depth file
1572 in the tracefs tracing directory. default: 0 (no limit)
1574 fw_devlink= [KNL] Create device links between consumer and supplier
1575 devices by scanning the firmware to infer the
1576 consumer/supplier relationships. This feature is
1577 especially useful when drivers are loaded as modules as
1578 it ensures proper ordering of tasks like device probing
1579 (suppliers first, then consumers), supplier boot state
1580 clean up (only after all consumers have probed),
1581 suspend/resume & runtime PM (consumers first, then
1583 Format: { off | permissive | on | rpm }
1584 off -- Don't create device links from firmware info.
1585 permissive -- Create device links from firmware info
1586 but use it only for ordering boot state clean
1587 up (sync_state() calls).
1588 on -- Create device links from firmware info and use it
1589 to enforce probe and suspend/resume ordering.
1590 rpm -- Like "on", but also use to order runtime PM.
1592 fw_devlink.strict=<bool>
1593 [KNL] Treat all inferred dependencies as mandatory
1594 dependencies. This only applies for fw_devlink=on|rpm.
1598 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1599 support via parallel port (up to 5 devices per port)
1600 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1601 See also Documentation/input/devices/joystick-parport.rst
1605 gart_fix_e820= [X86-64] disable the fix e820 for K8 GART
1609 gcov_persist= [GCOV] When non-zero (default), profiling data for
1610 kernel modules is saved and remains accessible via
1611 debugfs, even when the module is unloaded/reloaded.
1612 When zero, profiling data is discarded and associated
1613 debugfs files are removed at module unload time.
1615 goldfish [X86] Enable the goldfish android emulator platform.
1616 Don't use this when you are not running on the
1619 gpio-mockup.gpio_mockup_ranges
1620 [HW] Sets the ranges of gpiochip of for this device.
1621 Format: <start1>,<end1>,<start2>,<end2>...
1622 gpio-mockup.gpio_mockup_named_lines
1623 [HW] Let the driver know GPIO lines should be named.
1625 gpt [EFI] Forces disk with valid GPT signature but
1626 invalid Protective MBR to be treated as GPT. If the
1627 primary GPT is corrupted, it enables the backup/alternate
1628 GPT to be used instead.
1630 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1631 the "Enable 0" bit of the configuration register.
1634 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1635 the "Enable 0" bit of the configuration register.
1638 grcan.select= [HW] Select which physical interface to use.
1641 grcan.txsize= [HW] Sets the size of the tx buffer.
1642 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1644 grcan.rxsize= [HW] Sets the size of the rx buffer.
1645 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1649 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
1650 hardening is enabled for this boot. Hardened
1651 usercopy checking is used to protect the kernel
1652 from reading or writing beyond known memory
1653 allocation boundaries as a proactive defense
1654 against bounds-checking flaws in the kernel's
1655 copy_to_user()/copy_from_user() interface.
1656 on Perform hardened usercopy checks (default).
1657 off Disable hardened usercopy checks.
1659 hardlockup_all_cpu_backtrace=
1660 [KNL] Should the hard-lockup detector generate
1661 backtraces on all cpus.
1664 hashdist= [KNL,NUMA] Large hashes allocated during boot
1665 are distributed across NUMA nodes. Defaults on
1666 for 64-bit NUMA, off otherwise.
1667 Format: 0 | 1 (for off | on)
1669 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1671 hd= [EIDE] (E)IDE hard drive subsystem geometry
1672 Format: <cyl>,<head>,<sect>
1675 Disable Hardware Error Source Table (HEST) support;
1676 corresponding firmware-first mode error processing
1677 logic will be disabled.
1679 hibernate= [HIBERNATION]
1680 noresume Don't check if there's a hibernation image
1681 present during boot.
1682 nocompress Don't compress/decompress hibernation images.
1683 no Disable hibernation and resume.
1684 protect_image Turn on image protection during restoration
1685 (that will set all pages holding image data
1686 during restoration read-only).
1688 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1689 size of <nn>. This works even on boxes that have no
1690 highmem otherwise. This also works to reduce highmem
1691 size on bigger boxes.
1693 highres= [KNL] Enable/disable high resolution timer mode.
1694 Valid parameters: "on", "off"
1699 hostname= [KNL] Set the hostname (aka UTS nodename).
1701 This allows setting the system's hostname during early
1702 startup. This sets the name returned by gethostname.
1703 Using this parameter to set the hostname makes it
1704 possible to ensure the hostname is correctly set before
1705 any userspace processes run, avoiding the possibility
1706 that a process may call gethostname before the hostname
1707 has been explicitly set, resulting in the calling
1708 process getting an incorrect result. The string must
1709 not exceed the maximum allowed hostname length (usually
1710 64 characters) and will be truncated otherwise.
1712 hpet= [X86-32,HPET] option to control HPET usage
1713 Format: { enable (default) | disable | force |
1715 disable: disable HPET and use PIT instead
1716 force: allow force enabled of undocumented chips (ICH4,
1718 verbose: show contents of HPET registers during setup
1720 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1721 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1723 hugepages= [HW] Number of HugeTLB pages to allocate at boot.
1724 If this follows hugepagesz (below), it specifies
1725 the number of pages of hugepagesz to be allocated.
1726 If this is the first HugeTLB parameter on the command
1727 line, it specifies the number of pages to allocate for
1728 the default huge page size. If using node format, the
1729 number of pages to allocate per-node can be specified.
1730 See also Documentation/admin-guide/mm/hugetlbpage.rst.
1731 Format: <integer> or (node format)
1732 <node>:<integer>[,<node>:<integer>]
1735 [HW] The size of the HugeTLB pages. This is used in
1736 conjunction with hugepages (above) to allocate huge
1737 pages of a specific size at boot. The pair
1738 hugepagesz=X hugepages=Y can be specified once for
1739 each supported huge page size. Huge page sizes are
1740 architecture dependent. See also
1741 Documentation/admin-guide/mm/hugetlbpage.rst.
1744 hugetlb_cma= [HW,CMA] The size of a CMA area used for allocation
1745 of gigantic hugepages. Or using node format, the size
1746 of a CMA area per node can be specified.
1747 Format: nn[KMGTPE] or (node format)
1748 <node>:nn[KMGTPE][,<node>:nn[KMGTPE]]
1750 Reserve a CMA area of given size and allocate gigantic
1751 hugepages using the CMA allocator. If enabled, the
1752 boot-time allocation of gigantic hugepages is skipped.
1754 hugetlb_free_vmemmap=
1755 [KNL] Reguires CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP
1757 Control if HugeTLB Vmemmap Optimization (HVO) is enabled.
1758 Allows heavy hugetlb users to free up some more
1759 memory (7 * PAGE_SIZE for each 2MB hugetlb page).
1760 Format: { on | off (default) }
1765 Built with CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP_DEFAULT_ON=y,
1768 Note that the vmemmap pages may be allocated from the added
1769 memory block itself when memory_hotplug.memmap_on_memory is
1770 enabled, those vmemmap pages cannot be optimized even if this
1771 feature is enabled. Other vmemmap pages not allocated from
1772 the added memory block itself do not be affected.
1775 [KNL] Should the hung task detector generate panics.
1778 A value of 1 instructs the kernel to panic when a
1779 hung task is detected. The default value is controlled
1780 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1781 option. The value selected by this boot parameter can
1782 be changed later by the kernel.hung_task_panic sysctl.
1784 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1785 terminal devices. Valid values: 0..8
1786 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1787 If specified, z/VM IUCV HVC accepts connections
1788 from listed z/VM user IDs only.
1790 hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations
1791 which allow the hypervisor to 'idle' the
1792 guest on lock contention.
1795 Do not unregister boot console at start. This is only
1796 useful for debugging when something happens in the window
1797 between unregistering the boot console and initializing
1800 i2c_bus= [HW] Override the default board specific I2C bus speed
1801 or register an additional I2C bus that is not
1802 registered from board initialization code.
1806 i8042.debug [HW] Toggle i8042 debug mode
1807 i8042.unmask_kbd_data
1808 [HW] Enable printing of interrupt data from the KBD port
1809 (disabled by default, and as a pre-condition
1810 requires that i8042.debug=1 be enabled)
1811 i8042.direct [HW] Put keyboard port into non-translated mode
1812 i8042.dumbkbd [HW] Pretend that controller can only read data from
1813 keyboard and cannot control its state
1814 (Don't attempt to blink the leds)
1815 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1816 i8042.nokbd [HW] Don't check/create keyboard port
1817 i8042.noloop [HW] Disable the AUX Loopback command while probing
1819 i8042.nomux [HW] Don't check presence of an active multiplexing
1821 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1823 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1824 i8042.reset [HW] Reset the controller during init, cleanup and
1825 suspend-to-ram transitions, only during s2r
1826 transitions, or never reset
1827 Format: { 1 | Y | y | 0 | N | n }
1828 1, Y, y: always reset controller
1829 0, N, n: don't ever reset controller
1830 Default: only on s2r transitions on x86; most other
1831 architectures force reset to be always executed
1832 i8042.unlock [HW] Unlock (ignore) the keylock
1833 i8042.kbdreset [HW] Reset device connected to KBD port
1835 [HW] Allow deferred probing upon i8042 probe errors
1839 i915.invert_brightness=
1840 [DRM] Invert the sense of the variable that is used to
1841 set the brightness of the panel backlight. Normally a
1842 brightness value of 0 indicates backlight switched off,
1843 and the maximum of the brightness value sets the backlight
1844 to maximum brightness. If this parameter is set to 0
1845 (default) and the machine requires it, or this parameter
1846 is set to 1, a brightness value of 0 sets the backlight
1847 to maximum brightness, and the maximum of the brightness
1848 value switches the backlight off.
1849 -1 -- never invert brightness
1850 0 -- machine default
1851 1 -- force brightness inversion
1854 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1858 Format: idle=poll, idle=halt, idle=nomwait
1859 Poll forces a polling idle loop that can slightly
1860 improve the performance of waking up a idle CPU, but
1861 will use a lot of power and make the system run hot.
1863 idle=halt: Halt is forced to be used for CPU idle.
1864 In such case C2/C3 won't be used again.
1865 idle=nomwait: Disable mwait for CPU C-states
1869 Allow force disabling of Shared Virtual Memory (SVA)
1870 support for the idxd driver. By default it is set to
1873 idxd.tc_override= [HW]
1875 Allow override of default traffic class configuration
1876 for the device. By default it is set to false (0).
1878 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1879 Format: { strict | legacy | 2008 | relaxed }
1882 Choose which programs will be accepted for execution
1883 based on the IEEE 754 NaN encoding(s) supported by
1884 the FPU and the NaN encoding requested with the value
1885 of an ELF file header flag individually set by each
1886 binary. Hardware implementations are permitted to
1887 support either or both of the legacy and the 2008 NaN
1890 Available settings are as follows:
1891 strict accept binaries that request a NaN encoding
1892 supported by the FPU
1893 legacy only accept legacy-NaN binaries, if supported
1895 2008 only accept 2008-NaN binaries, if supported
1897 relaxed accept any binaries regardless of whether
1898 supported by the FPU
1900 The FPU emulator is always able to support both NaN
1901 encodings, so if no FPU hardware is present or it has
1902 been disabled with 'nofpu', then the settings of
1903 'legacy' and '2008' strap the emulator accordingly,
1904 'relaxed' straps the emulator for both legacy-NaN and
1905 2008-NaN, whereas 'strict' enables legacy-NaN only on
1906 legacy processors and both NaN encodings on MIPS32 or
1909 The setting for ABS.fmt/NEG.fmt instruction execution
1910 mode generally follows that for the NaN encoding,
1911 except where unsupported by hardware.
1913 ignore_loglevel [KNL]
1914 Ignore loglevel setting - this will print /all/
1915 kernel messages to the console. Useful for debugging.
1916 We also add it as printk module parameter, so users
1917 could change it dynamically, usually by
1918 /sys/module/printk/parameters/ignore_loglevel.
1921 Ignore RLIMIT_DATA setting for data mappings,
1922 print warning at first misuse. Can be changed via
1923 /sys/module/kernel/parameters/ignore_rlimit_data.
1925 ihash_entries= [KNL]
1926 Set number of hash buckets for inode cache.
1928 ima_appraise= [IMA] appraise integrity measurements
1929 Format: { "off" | "enforce" | "fix" | "log" }
1932 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
1933 The builtin appraise policy appraises all files
1936 ima_canonical_fmt [IMA]
1937 Use the canonical format for the binary runtime
1938 measurements, instead of host native format.
1941 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1945 The list of supported hash algorithms is defined
1946 in crypto/hash_info.h.
1949 The builtin policies to load during IMA setup.
1950 Format: "tcb | appraise_tcb | secure_boot |
1951 fail_securely | critical_data"
1953 The "tcb" policy measures all programs exec'd, files
1954 mmap'd for exec, and all files opened with the read
1955 mode bit set by either the effective uid (euid=0) or
1958 The "appraise_tcb" policy appraises the integrity of
1959 all files owned by root.
1961 The "secure_boot" policy appraises the integrity
1962 of files (eg. kexec kernel image, kernel modules,
1963 firmware, policy, etc) based on file signatures.
1965 The "fail_securely" policy forces file signature
1966 verification failure also on privileged mounted
1967 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
1970 The "critical_data" policy measures kernel integrity
1973 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
1974 Load a policy which meets the needs of the Trusted
1975 Computing Base. This means IMA will measure all
1976 programs exec'd, files mmap'd for exec, and all files
1977 opened for read by uid=0.
1980 Select one of defined IMA measurements template formats.
1981 Formats: { "ima" | "ima-ng" | "ima-ngv2" | "ima-sig" |
1986 [IMA] Define a custom template format.
1987 Format: { "field1|...|fieldN" }
1989 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
1990 Format: <min_file_size>
1991 Set the minimal file size for using asynchronous hash.
1992 If left unspecified, ahash usage is disabled.
1994 ahash performance varies for different data sizes on
1995 different crypto accelerators. This option can be used
1996 to achieve the best performance for a particular HW.
1998 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
2000 Set hashing buffer size. Default: 4k.
2002 ahash performance varies for different chunk sizes on
2003 different crypto accelerators. This option can be used
2004 to achieve best performance for particular HW.
2008 Run specified binary instead of /sbin/init as init
2011 initcall_debug [KNL] Trace initcalls as they are executed. Useful
2012 for working out where the kernel is dying during
2015 initcall_blacklist= [KNL] Do not execute a comma-separated list of
2016 initcall functions. Useful for debugging built-in
2017 modules and initcalls.
2019 initramfs_async= [KNL]
2022 This parameter controls whether the initramfs
2023 image is unpacked asynchronously, concurrently
2024 with devices being probed and
2025 initialized. This should normally just work,
2026 but as a debugging aid, one can get the
2027 historical behaviour of the initramfs
2028 unpacking being completed before device_ and
2031 initrd= [BOOT] Specify the location of the initial ramdisk
2033 initrdmem= [KNL] Specify a physical address and size from which to
2034 load the initrd. If an initrd is compiled in or
2035 specified in the bootparams, it takes priority over this
2037 Format: ss[KMG],nn[KMG]
2040 init_on_alloc= [MM] Fill newly allocated pages and heap objects with
2043 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
2045 init_on_free= [MM] Fill freed pages and heap objects with zeroes.
2047 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
2049 init_pkru= [X86] Specify the default memory protection keys rights
2050 register contents for all processes. 0x55555554 by
2051 default (disallow access to all but pkey 0). Can
2052 override in debugfs after boot.
2054 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
2057 int_pln_enable [X86] Enable power limit notification interrupt
2059 integrity_audit=[IMA]
2060 Format: { "0" | "1" }
2061 0 -- basic integrity auditing messages. (Default)
2062 1 -- additional integrity auditing messages.
2064 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
2066 Enable intel iommu driver.
2068 Disable intel iommu driver.
2069 igfx_off [Default Off]
2070 By default, gfx is mapped as normal device. If a gfx
2071 device has a dedicated DMAR unit, the DMAR unit is
2072 bypassed by not enabling DMAR with this option. In
2073 this case, gfx device will use physical address for
2075 strict [Default Off]
2076 Deprecated, equivalent to iommu.strict=1.
2077 sp_off [Default Off]
2078 By default, super page will be supported if Intel IOMMU
2079 has the capability. With this option, super page will
2082 Enable the Intel IOMMU scalable mode if the hardware
2083 advertises that it has support for the scalable mode
2086 Disallow use of the Intel IOMMU scalable mode.
2087 tboot_noforce [Default Off]
2088 Do not force the Intel IOMMU enabled under tboot.
2089 By default, tboot will force Intel IOMMU on, which
2090 could harm performance of some high-throughput
2091 devices like 40GBit network cards, even if identity
2093 Note that using this option lowers the security
2094 provided by tboot because it makes the system
2095 vulnerable to DMA attacks.
2097 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
2098 0 disables intel_idle and fall back on acpi_idle.
2099 1 to 9 specify maximum depth of C-state.
2103 Do not enable intel_pstate as the default
2104 scaling driver for the supported processors
2106 Use intel_pstate as a scaling driver, but configure it
2107 to work with generic cpufreq governors (instead of
2108 enabling its internal governor). This mode cannot be
2109 used along with the hardware-managed P-states (HWP)
2112 Enable intel_pstate on systems that prohibit it by default
2113 in favor of acpi-cpufreq. Forcing the intel_pstate driver
2114 instead of acpi-cpufreq may disable platform features, such
2115 as thermal controls and power capping, that rely on ACPI
2116 P-States information being indicated to OSPM and therefore
2117 should be used with caution. This option does not work with
2118 processors that aren't supported by the intel_pstate driver
2119 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
2121 Do not enable hardware P state control (HWP)
2124 Only load intel_pstate on systems which support
2125 hardware P state control (HWP) if available.
2127 Enforce ACPI _PPC performance limits. If the Fixed ACPI
2128 Description Table, specifies preferred power management
2129 profile as "Enterprise Server" or "Performance Server",
2130 then this feature is turned on by default.
2132 Allow per-logical-CPU P-State performance control limits using
2133 cpufreq sysfs interface
2135 intremap= [X86-64, Intel-IOMMU]
2136 on enable Interrupt Remapping (default)
2137 off disable Interrupt Remapping
2138 nosid disable Source ID checking
2140 BIOS x2APIC opt-out request will be ignored
2141 nopost disable Interrupt Posting
2143 iomem= Disable strict checking of access to MMIO memory
2144 strict regions from userspace.
2159 nobypass [PPC/POWERNV]
2160 Disable IOMMU bypass, using IOMMU for PCI devices.
2162 iommu.forcedac= [ARM64, X86] Control IOVA allocation for PCI devices.
2163 Format: { "0" | "1" }
2164 0 - Try to allocate a 32-bit DMA address first, before
2165 falling back to the full range if needed.
2166 1 - Allocate directly from the full usable range,
2167 forcing Dual Address Cycle for PCI cards supporting
2168 greater than 32-bit addressing.
2170 iommu.strict= [ARM64, X86] Configure TLB invalidation behaviour
2171 Format: { "0" | "1" }
2173 Request that DMA unmap operations use deferred
2174 invalidation of hardware TLBs, for increased
2175 throughput at the cost of reduced device isolation.
2176 Will fall back to strict mode if not supported by
2177 the relevant IOMMU driver.
2179 DMA unmap operations invalidate IOMMU hardware TLBs
2181 unset - Use value of CONFIG_IOMMU_DEFAULT_DMA_{LAZY,STRICT}.
2182 Note: on x86, strict mode specified via one of the
2183 legacy driver-specific options takes precedence.
2186 [ARM64, X86] Configure DMA to bypass the IOMMU by default.
2187 Format: { "0" | "1" }
2188 0 - Use IOMMU translation for DMA.
2189 1 - Bypass the IOMMU for DMA.
2190 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
2192 io7= [HW] IO7 for Marvel-based Alpha systems
2193 See comment before marvel_specify_io7 in
2194 arch/alpha/kernel/core_marvel.c.
2196 io_delay= [X86] I/O delay method
2198 Standard port 0x80 based delay
2200 Alternate port 0xed based delay (needed on some systems)
2202 Simple two microseconds delay
2207 See Documentation/admin-guide/nfs/nfsroot.rst.
2209 ipcmni_extend [KNL] Extend the maximum number of unique System V
2210 IPC identifiers from 32,768 to 16,777,216.
2212 irqaffinity= [SMP] Set the default irq affinity mask
2213 The argument is a cpu list, as described above.
2215 irqchip.gicv2_force_probe=
2218 Force the kernel to look for the second 4kB page
2219 of a GICv2 controller even if the memory range
2220 exposed by the device tree is too small.
2222 irqchip.gicv3_nolpi=
2224 Force the kernel to ignore the availability of
2225 LPIs (and by consequence ITSs). Intended for system
2226 that use the kernel as a bootloader, and thus want
2227 to let secondary kernels in charge of setting up
2230 irqchip.gicv3_pseudo_nmi= [ARM64]
2231 Enables support for pseudo-NMIs in the kernel. This
2232 requires the kernel to be built with
2233 CONFIG_ARM64_PSEUDO_NMI.
2236 When an interrupt is not handled search all handlers
2237 for it. Intended to get systems with badly broken
2241 When an interrupt is not handled search all handlers
2242 for it. Also check all handlers each timer
2243 interrupt. Intended to get systems with badly broken
2247 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
2249 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
2250 [Deprecated - use cpusets instead]
2251 Format: [flag-list,]<cpu-list>
2253 Specify one or more CPUs to isolate from disturbances
2254 specified in the flag list (default: domain):
2257 Disable the tick when a single task runs.
2259 A residual 1Hz tick is offloaded to workqueues, which you
2260 need to affine to housekeeping through the global
2261 workqueue's affinity configured via the
2262 /sys/devices/virtual/workqueue/cpumask sysfs file, or
2263 by using the 'domain' flag described below.
2265 NOTE: by default the global workqueue runs on all CPUs,
2266 so to protect individual CPUs the 'cpumask' file has to
2267 be configured manually after bootup.
2270 Isolate from the general SMP balancing and scheduling
2271 algorithms. Note that performing domain isolation this way
2272 is irreversible: it's not possible to bring back a CPU to
2273 the domains once isolated through isolcpus. It's strongly
2274 advised to use cpusets instead to disable scheduler load
2275 balancing through the "cpuset.sched_load_balance" file.
2276 It offers a much more flexible interface where CPUs can
2277 move in and out of an isolated set anytime.
2279 You can move a process onto or off an "isolated" CPU via
2280 the CPU affinity syscalls or cpuset.
2281 <cpu number> begins at 0 and the maximum value is
2282 "number of CPUs in system - 1".
2286 Isolate from being targeted by managed interrupts
2287 which have an interrupt mask containing isolated
2288 CPUs. The affinity of managed interrupts is
2289 handled by the kernel and cannot be changed via
2290 the /proc/irq/* interfaces.
2292 This isolation is best effort and only effective
2293 if the automatically assigned interrupt mask of a
2294 device queue contains isolated and housekeeping
2295 CPUs. If housekeeping CPUs are online then such
2296 interrupts are directed to the housekeeping CPU
2297 so that IO submitted on the housekeeping CPU
2298 cannot disturb the isolated CPU.
2300 If a queue's affinity mask contains only isolated
2301 CPUs then this parameter has no effect on the
2302 interrupt routing decision, though interrupts are
2303 only delivered when tasks running on those
2304 isolated CPUs submit IO. IO submitted on
2305 housekeeping CPUs has no influence on those
2308 The format of <cpu-list> is described above.
2312 ivrs_ioapic [HW,X86-64]
2313 Provide an override to the IOAPIC-ID<->DEVICE-ID
2314 mapping provided in the IVRS ACPI table.
2315 By default, PCI segment is 0, and can be omitted.
2317 * To map IOAPIC-ID decimal 10 to PCI device 00:14.0
2318 write the parameter as:
2319 ivrs_ioapic[10]=00:14.0
2320 * To map IOAPIC-ID decimal 10 to PCI segment 0x1 and
2321 PCI device 00:14.0 write the parameter as:
2322 ivrs_ioapic[10]=0001:00:14.0
2324 ivrs_hpet [HW,X86-64]
2325 Provide an override to the HPET-ID<->DEVICE-ID
2326 mapping provided in the IVRS ACPI table.
2327 By default, PCI segment is 0, and can be omitted.
2329 * To map HPET-ID decimal 0 to PCI device 00:14.0
2330 write the parameter as:
2331 ivrs_hpet[0]=00:14.0
2332 * To map HPET-ID decimal 10 to PCI segment 0x1 and
2333 PCI device 00:14.0 write the parameter as:
2334 ivrs_ioapic[10]=0001:00:14.0
2336 ivrs_acpihid [HW,X86-64]
2337 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2338 mapping provided in the IVRS ACPI table.
2340 For example, to map UART-HID:UID AMD0020:0 to
2341 PCI segment 0x1 and PCI device ID 00:14.5,
2342 write the parameter as:
2343 ivrs_acpihid[0001:00:14.5]=AMD0020:0
2345 By default, PCI segment is 0, and can be omitted.
2346 For example, PCI device 00:14.5 write the parameter as:
2347 ivrs_acpihid[00:14.5]=AMD0020:0
2349 js= [HW,JOY] Analog joystick
2350 See Documentation/input/joydev/joystick.rst.
2353 When CONFIG_RANDOMIZE_BASE is set, this disables
2354 kernel and module base offset ASLR (Address Space
2355 Layout Randomization).
2358 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2359 report on every invalid memory access. Without this
2360 parameter KASAN will print report only for the first
2365 kernelcore= [KNL,X86,IA-64,PPC]
2366 Format: nn[KMGTPE] | nn% | "mirror"
2367 This parameter specifies the amount of memory usable by
2368 the kernel for non-movable allocations. The requested
2369 amount is spread evenly throughout all nodes in the
2370 system as ZONE_NORMAL. The remaining memory is used for
2371 movable memory in its own zone, ZONE_MOVABLE. In the
2372 event, a node is too small to have both ZONE_NORMAL and
2373 ZONE_MOVABLE, kernelcore memory will take priority and
2374 other nodes will have a larger ZONE_MOVABLE.
2376 ZONE_MOVABLE is used for the allocation of pages that
2377 may be reclaimed or moved by the page migration
2378 subsystem. Note that allocations like PTEs-from-HighMem
2379 still use the HighMem zone if it exists, and the Normal
2380 zone if it does not.
2382 It is possible to specify the exact amount of memory in
2383 the form of "nn[KMGTPE]", a percentage of total system
2384 memory in the form of "nn%", or "mirror". If "mirror"
2385 option is specified, mirrored (reliable) memory is used
2386 for non-movable allocations and remaining memory is used
2387 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2388 are exclusive, so you cannot specify multiple forms.
2390 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
2391 Format: <Controller#>[,poll interval]
2392 The controller # is the number of the ehci usb debug
2393 port as it is probed via PCI. The poll interval is
2394 optional and is the number seconds in between
2395 each poll cycle to the debug port in case you need
2396 the functionality for interrupting the kernel with
2397 gdb or control-c on the dbgp connection. When
2398 not using this parameter you use sysrq-g to break into
2399 the kernel debugger.
2401 kgdboc= [KGDB,HW] kgdb over consoles.
2402 Requires a tty driver that supports console polling,
2403 or a supported polling keyboard driver (non-usb).
2404 Serial only format: <serial_device>[,baud]
2405 keyboard only format: kbd
2406 keyboard and serial format: kbd,<serial_device>[,baud]
2407 Optional Kernel mode setting:
2408 kms, kbd format: kms,kbd
2409 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2411 kgdboc_earlycon= [KGDB,HW]
2412 If the boot console provides the ability to read
2413 characters and can work in polling mode, you can use
2414 this parameter to tell kgdb to use it as a backend
2415 until the normal console is registered. Intended to
2416 be used together with the kgdboc parameter which
2417 specifies the normal console to transition to.
2419 The name of the early console should be specified
2420 as the value of this parameter. Note that the name of
2421 the early console might be different than the tty
2422 name passed to kgdboc. It's OK to leave the value
2423 blank and the first boot console that implements
2424 read() will be picked.
2426 kgdbwait [KGDB] Stop kernel execution and enter the
2427 kernel debugger at the earliest opportunity.
2429 kmac= [MIPS] Korina ethernet MAC address.
2430 Configure the RouterBoard 532 series on-chip
2431 Ethernet adapter MAC address.
2433 kmemleak= [KNL] Boot-time kmemleak enable/disable
2434 Valid arguments: on, off
2436 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2439 kprobe_event=[probe-list]
2440 [FTRACE] Add kprobe events and enable at boot time.
2441 The probe-list is a semicolon delimited list of probe
2442 definitions. Each definition is same as kprobe_events
2443 interface, but the parameters are comma delimited.
2444 For example, to add a kprobe event on vfs_read with
2445 arg1 and arg2, add to the command line;
2447 kprobe_event=p,vfs_read,$arg1,$arg2
2449 See also Documentation/trace/kprobetrace.rst "Kernel
2450 Boot Parameter" section.
2452 kpti= [ARM64] Control page table isolation of user
2453 and kernel address spaces.
2454 Default: enabled on cores which need mitigation.
2458 kunit.enable= [KUNIT] Enable executing KUnit tests. Requires
2459 CONFIG_KUNIT to be set to be fully enabled. The
2460 default value can be overridden via
2461 KUNIT_DEFAULT_ENABLED.
2462 Default is 1 (enabled)
2464 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2465 Default is 0 (don't ignore, but inject #GP)
2467 kvm.eager_page_split=
2468 [KVM,X86] Controls whether or not KVM will try to
2469 proactively split all huge pages during dirty logging.
2470 Eager page splitting reduces interruptions to vCPU
2471 execution by eliminating the write-protection faults
2472 and MMU lock contention that would otherwise be
2473 required to split huge pages lazily.
2475 VM workloads that rarely perform writes or that write
2476 only to a small region of VM memory may benefit from
2477 disabling eager page splitting to allow huge pages to
2478 still be used for reads.
2480 The behavior of eager page splitting depends on whether
2481 KVM_DIRTY_LOG_INITIALLY_SET is enabled or disabled. If
2482 disabled, all huge pages in a memslot will be eagerly
2483 split when dirty logging is enabled on that memslot. If
2484 enabled, eager page splitting will be performed during
2485 the KVM_CLEAR_DIRTY ioctl, and only for the pages being
2488 Eager page splitting is only supported when kvm.tdp_mmu=Y.
2492 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2493 Default is false (don't support).
2496 [KVM] Controls the software workaround for the
2497 X86_BUG_ITLB_MULTIHIT bug.
2498 force : Always deploy workaround.
2499 off : Never deploy workaround.
2500 auto : Deploy workaround based on the presence of
2501 X86_BUG_ITLB_MULTIHIT.
2505 If the software workaround is enabled for the host,
2506 guests do need not to enable it for nested guests.
2508 kvm.nx_huge_pages_recovery_ratio=
2509 [KVM] Controls how many 4KiB pages are periodically zapped
2510 back to huge pages. 0 disables the recovery, otherwise if
2511 the value is N KVM will zap 1/Nth of the 4KiB pages every
2512 period (see below). The default is 60.
2514 kvm.nx_huge_pages_recovery_period_ms=
2515 [KVM] Controls the time period at which KVM zaps 4KiB pages
2516 back to huge pages. If the value is a non-zero N, KVM will
2517 zap a portion (see ratio above) of the pages every N msecs.
2518 If the value is 0 (the default), KVM will pick a period based
2519 on the ratio, such that a page is zapped after 1 hour on average.
2521 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM.
2522 Default is 1 (enabled)
2524 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU)
2526 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode.
2529 [KVM,ARM] Select one of KVM/arm64's modes of operation.
2531 none: Forcefully disable KVM.
2533 nvhe: Standard nVHE-based mode, without support for
2536 protected: nVHE-based mode with support for guests whose
2537 state is kept private from the host.
2539 Defaults to VHE/nVHE based on hardware support. Setting
2540 mode to "protected" will disable kexec and hibernation
2543 kvm-arm.vgic_v3_group0_trap=
2544 [KVM,ARM] Trap guest accesses to GICv3 group-0
2547 kvm-arm.vgic_v3_group1_trap=
2548 [KVM,ARM] Trap guest accesses to GICv3 group-1
2551 kvm-arm.vgic_v3_common_trap=
2552 [KVM,ARM] Trap guest accesses to GICv3 common
2555 kvm-arm.vgic_v4_enable=
2556 [KVM,ARM] Allow use of GICv4 for direct injection of
2559 kvm_cma_resv_ratio=n [PPC]
2560 Reserves given percentage from system memory area for
2561 contiguous memory allocation for KVM hash pagetable
2563 By default it reserves 5% of total system memory.
2567 kvm-intel.ept= [KVM,Intel] Disable extended page tables
2568 (virtualized MMU) support on capable Intel chips.
2569 Default is 1 (enabled)
2571 kvm-intel.emulate_invalid_guest_state=
2572 [KVM,Intel] Disable emulation of invalid guest state.
2573 Ignored if kvm-intel.enable_unrestricted_guest=1, as
2574 guest state is never invalid for unrestricted guests.
2575 This param doesn't apply to nested guests (L2), as KVM
2576 never emulates invalid L2 guest state.
2577 Default is 1 (enabled)
2579 kvm-intel.flexpriority=
2580 [KVM,Intel] Disable FlexPriority feature (TPR shadow).
2581 Default is 1 (enabled)
2584 [KVM,Intel] Enable VMX nesting (nVMX).
2585 Default is 0 (disabled)
2587 kvm-intel.unrestricted_guest=
2588 [KVM,Intel] Disable unrestricted guest feature
2589 (virtualized real and unpaged mode) on capable
2590 Intel chips. Default is 1 (enabled)
2592 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2595 Valid arguments: never, cond, always
2597 always: L1D cache flush on every VMENTER.
2598 cond: Flush L1D on VMENTER only when the code between
2599 VMEXIT and VMENTER can leak host memory.
2600 never: Disables the mitigation
2602 Default is cond (do L1 cache flush in specific instances)
2604 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification
2605 feature (tagged TLBs) on capable Intel chips.
2606 Default is 1 (enabled)
2608 l1d_flush= [X86,INTEL]
2609 Control mitigation for L1D based snooping vulnerability.
2611 Certain CPUs are vulnerable to an exploit against CPU
2612 internal buffers which can forward information to a
2613 disclosure gadget under certain conditions.
2615 In vulnerable processors, the speculatively
2616 forwarded data can be used in a cache side channel
2617 attack, to access data to which the attacker does
2618 not have direct access.
2620 This parameter controls the mitigation. The
2623 on - enable the interface for the mitigation
2625 l1tf= [X86] Control mitigation of the L1TF vulnerability on
2628 The kernel PTE inversion protection is unconditionally
2629 enabled and cannot be disabled.
2632 Provides all available mitigations for the
2633 L1TF vulnerability. Disables SMT and
2634 enables all mitigations in the
2635 hypervisors, i.e. unconditional L1D flush.
2637 SMT control and L1D flush control via the
2638 sysfs interface is still possible after
2639 boot. Hypervisors will issue a warning
2640 when the first VM is started in a
2641 potentially insecure configuration,
2642 i.e. SMT enabled or L1D flush disabled.
2645 Same as 'full', but disables SMT and L1D
2646 flush runtime control. Implies the
2647 'nosmt=force' command line option.
2648 (i.e. sysfs control of SMT is disabled.)
2651 Leaves SMT enabled and enables the default
2652 hypervisor mitigation, i.e. conditional
2655 SMT control and L1D flush control via the
2656 sysfs interface is still possible after
2657 boot. Hypervisors will issue a warning
2658 when the first VM is started in a
2659 potentially insecure configuration,
2660 i.e. SMT enabled or L1D flush disabled.
2664 Disables SMT and enables the default
2665 hypervisor mitigation.
2667 SMT control and L1D flush control via the
2668 sysfs interface is still possible after
2669 boot. Hypervisors will issue a warning
2670 when the first VM is started in a
2671 potentially insecure configuration,
2672 i.e. SMT enabled or L1D flush disabled.
2675 Same as 'flush', but hypervisors will not
2676 warn when a VM is started in a potentially
2677 insecure configuration.
2680 Disables hypervisor mitigations and doesn't
2682 It also drops the swap size and available
2683 RAM limit restriction on both hypervisor and
2688 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2694 lapic [X86-32,APIC] Enable the local APIC even if BIOS
2697 lapic= [X86,APIC] Do not use TSC deadline
2698 value for LAPIC timer one-shot implementation. Default
2699 back to the programmable timer unit in the LAPIC.
2700 Format: notscdeadline
2702 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
2705 libata.dma= [LIBATA] DMA control
2706 libata.dma=0 Disable all PATA and SATA DMA
2707 libata.dma=1 PATA and SATA Disk DMA only
2708 libata.dma=2 ATAPI (CDROM) DMA only
2709 libata.dma=4 Compact Flash DMA only
2710 Combinations also work, so libata.dma=3 enables DMA
2711 for disks and CDROMs, but not CFs.
2713 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2714 libata.ignore_hpa=0 keep BIOS limits (default)
2715 libata.ignore_hpa=1 ignore limits, using full disk
2717 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2721 libata.force= [LIBATA] Force configurations. The format is a comma-
2722 separated list of "[ID:]VAL" where ID is PORT[.DEVICE].
2723 PORT and DEVICE are decimal numbers matching port, link
2724 or device. Basically, it matches the ATA ID string
2725 printed on console by libata. If the whole ID part is
2726 omitted, the last PORT and DEVICE values are used. If
2727 ID hasn't been specified yet, the configuration applies
2728 to all ports, links and devices.
2730 If only DEVICE is omitted, the parameter applies to
2731 the port and all links and devices behind it. DEVICE
2732 number of 0 either selects the first device or the
2733 first fan-out link behind PMP device. It does not
2734 select the host link. DEVICE number of 15 selects the
2735 host link and device attached to it.
2737 The VAL specifies the configuration to force. As long
2738 as there is no ambiguity, shortcut notation is allowed.
2739 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2740 The following configurations can be forced.
2742 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2743 Any ID with matching PORT is used.
2745 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2747 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2748 udma[/][16,25,33,44,66,100,133] notation is also
2751 * nohrst, nosrst, norst: suppress hard, soft and both
2754 * rstonce: only attempt one reset during hot-unplug
2757 * [no]dbdelay: Enable or disable the extra 200ms delay
2758 before debouncing a link PHY and device presence
2761 * [no]ncq: Turn on or off NCQ.
2763 * [no]ncqtrim: Enable or disable queued DSM TRIM.
2765 * [no]ncqati: Enable or disable NCQ trim on ATI chipset.
2767 * [no]trim: Enable or disable (unqueued) TRIM.
2769 * trim_zero: Indicate that TRIM command zeroes data.
2771 * max_trim_128m: Set 128M maximum trim size limit.
2773 * [no]dma: Turn on or off DMA transfers.
2775 * atapi_dmadir: Enable ATAPI DMADIR bridge support.
2777 * atapi_mod16_dma: Enable the use of ATAPI DMA for
2778 commands that are not a multiple of 16 bytes.
2780 * [no]dmalog: Enable or disable the use of the
2781 READ LOG DMA EXT command to access logs.
2783 * [no]iddevlog: Enable or disable access to the
2784 identify device data log.
2786 * [no]logdir: Enable or disable access to the general
2787 purpose log directory.
2789 * max_sec_128: Set transfer size limit to 128 sectors.
2791 * max_sec_1024: Set or clear transfer size limit to
2794 * max_sec_lba48: Set or clear transfer size limit to
2797 * [no]lpm: Enable or disable link power management.
2799 * [no]setxfer: Indicate if transfer speed mode setting
2802 * dump_id: Dump IDENTIFY data.
2804 * disable: Disable this device.
2806 If there are multiple matching configurations changing
2807 the same attribute, the last one is used.
2809 load_ramdisk= [RAM] [Deprecated]
2811 lockd.nlm_grace_period=P [NFS] Assign grace period.
2814 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2817 lockd.nlm_timeout=T [NFS] Assign timeout value.
2820 lockd.nlm_udpport=M [NFS] Assign UDP port.
2823 lockdown= [SECURITY]
2824 { integrity | confidentiality }
2825 Enable the kernel lockdown feature. If set to
2826 integrity, kernel features that allow userland to
2827 modify the running kernel are disabled. If set to
2828 confidentiality, kernel features that allow userland
2829 to extract confidential information from the kernel
2832 locktorture.nreaders_stress= [KNL]
2833 Set the number of locking read-acquisition kthreads.
2834 Defaults to being automatically set based on the
2835 number of online CPUs.
2837 locktorture.nwriters_stress= [KNL]
2838 Set the number of locking write-acquisition kthreads.
2840 locktorture.onoff_holdoff= [KNL]
2841 Set time (s) after boot for CPU-hotplug testing.
2843 locktorture.onoff_interval= [KNL]
2844 Set time (s) between CPU-hotplug operations, or
2845 zero to disable CPU-hotplug testing.
2847 locktorture.shuffle_interval= [KNL]
2848 Set task-shuffle interval (jiffies). Shuffling
2849 tasks allows some CPUs to go into dyntick-idle
2850 mode during the locktorture test.
2852 locktorture.shutdown_secs= [KNL]
2853 Set time (s) after boot system shutdown. This
2854 is useful for hands-off automated testing.
2856 locktorture.stat_interval= [KNL]
2857 Time (s) between statistics printk()s.
2859 locktorture.stutter= [KNL]
2860 Time (s) to stutter testing, for example,
2861 specifying five seconds causes the test to run for
2862 five seconds, wait for five seconds, and so on.
2863 This tests the locking primitive's ability to
2864 transition abruptly to and from idle.
2866 locktorture.torture_type= [KNL]
2867 Specify the locking implementation to test.
2869 locktorture.verbose= [KNL]
2870 Enable additional printk() statements.
2872 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
2875 loglevel= All Kernel Messages with a loglevel smaller than the
2876 console loglevel will be printed to the console. It can
2877 also be changed with klogd or other programs. The
2878 loglevels are defined as follows:
2880 0 (KERN_EMERG) system is unusable
2881 1 (KERN_ALERT) action must be taken immediately
2882 2 (KERN_CRIT) critical conditions
2883 3 (KERN_ERR) error conditions
2884 4 (KERN_WARNING) warning conditions
2885 5 (KERN_NOTICE) normal but significant condition
2886 6 (KERN_INFO) informational
2887 7 (KERN_DEBUG) debug-level messages
2889 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
2890 in bytes. n must be a power of two and greater
2891 than the minimal size. The minimal size is defined
2892 by LOG_BUF_SHIFT kernel config parameter. There is
2893 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2894 that allows to increase the default size depending on
2895 the number of CPUs. See init/Kconfig for more details.
2897 logo.nologo [FB] Disables display of the built-in Linux logo.
2898 This may be used to provide more screen space for
2899 kernel log messages and is useful when debugging
2900 kernel boot problems.
2902 lp=0 [LP] Specify parallel ports to use, e.g,
2903 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2904 lp=reset first parallel port). 'lp=0' disables the
2905 lp=auto printer driver. 'lp=reset' (which can be
2906 specified in addition to the ports) causes
2907 attached printers to be reset. Using
2908 lp=port1,port2,... specifies the parallel ports
2909 to associate lp devices with, starting with
2910 lp0. A port specification may be 'none' to skip
2911 that lp device, or a parport name such as
2912 'parport0'. Specifying 'lp=auto' instead of a
2913 port specification list means that device IDs
2914 from each port should be examined, to see if
2915 an IEEE 1284-compliant printer is attached; if
2916 so, the driver will manage that printer.
2917 See also header of drivers/char/lp.c.
2920 Sets loops_per_jiffy to given constant, thus avoiding
2921 time-consuming boot-time autodetection (up to 250 ms per
2922 CPU). 0 enables autodetection (default). To determine
2923 the correct value for your kernel, boot with normal
2924 autodetection and see what value is printed. Note that
2925 on SMP systems the preset will be applied to all CPUs,
2926 which is likely to cause problems if your CPUs need
2927 significantly divergent settings. An incorrect value
2928 will cause delays in the kernel to be wrong, leading to
2929 unpredictable I/O errors and other breakage. Although
2930 unlikely, in the extreme case this might damage your
2934 Format: <io>,<irq>,<dma>
2936 lsm.debug [SECURITY] Enable LSM initialization debugging output.
2939 [SECURITY] Choose order of LSM initialization. This
2940 overrides CONFIG_LSM, and the "security=" parameter.
2942 machvec= [IA-64] Force the use of a particular machine-vector
2943 (machvec) in a generic kernel.
2944 Example: machvec=hpzx1
2946 machtype= [Loongson] Share the same kernel image file between
2947 different yeeloong laptops.
2948 Example: machtype=lemote-yeeloong-2f-7inch
2950 max_addr=nn[KMG] [KNL,BOOT,IA-64] All physical memory greater
2951 than or equal to this physical address is ignored.
2953 maxcpus= [SMP] Maximum number of processors that an SMP kernel
2954 will bring up during bootup. maxcpus=n : n >= 0 limits
2955 the kernel to bring up 'n' processors. Surely after
2956 bootup you can bring up the other plugged cpu by executing
2957 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
2958 only takes effect during system bootup.
2959 While n=0 is a special case, it is equivalent to "nosmp",
2960 which also disables the IO APIC.
2962 max_loop= [LOOP] The number of loop block devices that get
2963 (loop.max_loop) unconditionally pre-created at init time. The default
2964 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
2965 of statically allocating a predefined number, loop
2966 devices can be requested on-demand with the
2967 /dev/loop-control interface.
2969 mce [X86-32] Machine Check Exception
2971 mce=option [X86-64] See Documentation/x86/x86_64/boot-options.rst
2973 md= [HW] RAID subsystems devices and level
2974 See Documentation/admin-guide/md.rst.
2977 Format: <first>,<last>
2978 Specifies range of consoles to be captured by the MDA.
2981 Control mitigation for the Micro-architectural Data
2982 Sampling (MDS) vulnerability.
2984 Certain CPUs are vulnerable to an exploit against CPU
2985 internal buffers which can forward information to a
2986 disclosure gadget under certain conditions.
2988 In vulnerable processors, the speculatively
2989 forwarded data can be used in a cache side channel
2990 attack, to access data to which the attacker does
2991 not have direct access.
2993 This parameter controls the MDS mitigation. The
2996 full - Enable MDS mitigation on vulnerable CPUs
2997 full,nosmt - Enable MDS mitigation and disable
2998 SMT on vulnerable CPUs
2999 off - Unconditionally disable MDS mitigation
3001 On TAA-affected machines, mds=off can be prevented by
3002 an active TAA mitigation as both vulnerabilities are
3003 mitigated with the same mechanism so in order to disable
3004 this mitigation, you need to specify tsx_async_abort=off
3007 Not specifying this option is equivalent to
3010 For details see: Documentation/admin-guide/hw-vuln/mds.rst
3012 mem=nn[KMG] [HEXAGON] Set the memory size.
3013 Must be specified, otherwise memory size will be 0.
3015 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
3016 Amount of memory to be used in cases as follows:
3019 2 when the kernel is not able to see the whole system memory;
3020 3 memory that lies after 'mem=' boundary is excluded from
3021 the hypervisor, then assigned to KVM guests.
3022 4 to limit the memory available for kdump kernel.
3024 [ARC,MICROBLAZE] - the limit applies only to low memory,
3025 high memory is not affected.
3027 [ARM64] - only limits memory covered by the linear
3028 mapping. The NOMAP regions are not affected.
3030 [X86] Work as limiting max address. Use together
3031 with memmap= to avoid physical address space collisions.
3032 Without memmap= PCI devices could be placed at addresses
3033 belonging to unused RAM.
3035 Note that this only takes effects during boot time since
3036 in above case 3, memory may need be hot added after boot
3037 if system memory of hypervisor is not sufficient.
3040 [ARM,MIPS] - override the memory layout reported by
3042 Define a memory region of size nn[KMG] starting at
3044 Multiple different regions can be specified with
3045 multiple mem= parameters on the command line.
3047 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
3050 memblock=debug [KNL] Enable memblock debug messages.
3053 [KNL,SH] Allow user to override the default size for
3054 per-device physically contiguous DMA buffers.
3056 memhp_default_state=online/offline
3057 [KNL] Set the initial state for the memory hotplug
3058 onlining policy. If not specified, the default value is
3059 set according to the
3060 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
3062 See Documentation/admin-guide/mm/memory-hotplug.rst.
3064 memmap=exactmap [KNL,X86] Enable setting of an exact
3065 E820 memory map, as specified by the user.
3066 Such memmap=exactmap lines can be constructed based on
3067 BIOS output or other requirements. See the memmap=nn@ss
3070 memmap=nn[KMG]@ss[KMG]
3071 [KNL, X86, MIPS, XTENSA] Force usage of a specific region of memory.
3072 Region of memory to be used is from ss to ss+nn.
3073 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
3074 which limits max address to nn[KMG].
3075 Multiple different regions can be specified,
3078 memmap=100M@2G,100M#3G,1G!1024G
3080 memmap=nn[KMG]#ss[KMG]
3081 [KNL,ACPI] Mark specific memory as ACPI data.
3082 Region of memory to be marked is from ss to ss+nn.
3084 memmap=nn[KMG]$ss[KMG]
3085 [KNL,ACPI] Mark specific memory as reserved.
3086 Region of memory to be reserved is from ss to ss+nn.
3087 Example: Exclude memory from 0x18690000-0x1869ffff
3088 memmap=64K$0x18690000
3090 memmap=0x10000$0x18690000
3091 Some bootloaders may need an escape character before '$',
3092 like Grub2, otherwise '$' and the following number
3095 memmap=nn[KMG]!ss[KMG]
3096 [KNL,X86] Mark specific memory as protected.
3097 Region of memory to be used, from ss to ss+nn.
3098 The memory region may be marked as e820 type 12 (0xc)
3099 and is NVDIMM or ADR memory.
3101 memmap=<size>%<offset>-<oldtype>+<newtype>
3102 [KNL,ACPI] Convert memory within the specified region
3103 from <oldtype> to <newtype>. If "-<oldtype>" is left
3104 out, the whole region will be marked as <newtype>,
3105 even if previously unavailable. If "+<newtype>" is left
3106 out, matching memory will be removed. Types are
3107 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
3108 3 = ACPI, 12 = PRAM.
3110 memory_corruption_check=0/1 [X86]
3111 Some BIOSes seem to corrupt the first 64k of
3112 memory when doing things like suspend/resume.
3113 Setting this option will scan the memory
3114 looking for corruption. Enabling this will
3115 both detect corruption and prevent the kernel
3116 from using the memory being corrupted.
3117 However, its intended as a diagnostic tool; if
3118 repeatable BIOS-originated corruption always
3119 affects the same memory, you can use memmap=
3120 to prevent the kernel from using that memory.
3122 memory_corruption_check_size=size [X86]
3123 By default it checks for corruption in the low
3124 64k, making this memory unavailable for normal
3125 use. Use this parameter to scan for
3126 corruption in more or less memory.
3128 memory_corruption_check_period=seconds [X86]
3129 By default it checks for corruption every 60
3130 seconds. Use this parameter to check at some
3131 other rate. 0 disables periodic checking.
3133 memory_hotplug.memmap_on_memory
3134 [KNL,X86,ARM] Boolean flag to enable this feature.
3135 Format: {on | off (default)}
3136 When enabled, runtime hotplugged memory will
3137 allocate its internal metadata (struct pages,
3138 those vmemmap pages cannot be optimized even
3139 if hugetlb_free_vmemmap is enabled) from the
3140 hotadded memory which will allow to hotadd a
3141 lot of memory without requiring additional
3143 This feature is disabled by default because it
3144 has some implication on large (e.g. GB)
3145 allocations in some configurations (e.g. small
3147 The state of the flag can be read in
3148 /sys/module/memory_hotplug/parameters/memmap_on_memory.
3149 Note that even when enabled, there are a few cases where
3150 the feature is not effective.
3152 memtest= [KNL,X86,ARM,M68K,PPC,RISCV] Enable memtest
3154 default : 0 <disable>
3155 Specifies the number of memtest passes to be
3156 performed. Each pass selects another test
3157 pattern from a given set of patterns. Memtest
3158 fills the memory with this pattern, validates
3159 memory contents and reserves bad memory
3160 regions that are detected.
3162 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
3163 Valid arguments: on, off
3164 Default (depends on kernel configuration option):
3165 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
3166 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
3167 mem_encrypt=on: Activate SME
3168 mem_encrypt=off: Do not activate SME
3170 Refer to Documentation/virt/kvm/x86/amd-memory-encryption.rst
3171 for details on when memory encryption can be activated.
3173 mem_sleep_default= [SUSPEND] Default system suspend mode:
3174 s2idle - Suspend-To-Idle
3175 shallow - Power-On Suspend or equivalent (if supported)
3176 deep - Suspend-To-RAM or equivalent (if supported)
3177 See Documentation/admin-guide/pm/sleep-states.rst.
3179 meye.*= [HW] Set MotionEye Camera parameters
3180 See Documentation/admin-guide/media/meye.rst.
3182 mfgpt_irq= [IA-32] Specify the IRQ to use for the
3183 Multi-Function General Purpose Timers on AMD Geode
3186 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
3187 the BIOS has incorrectly applied a workaround. TinyBIOS
3188 version 0.98 is known to be affected, 0.99 fixes the
3189 problem by letting the user disable the workaround.
3193 min_addr=nn[KMG] [KNL,BOOT,IA-64] All physical memory below this
3194 physical address is ignored.
3196 mini2440= [ARM,HW,KNL]
3197 Format:[0..2][b][c][t]
3199 MINI2440 configuration specification:
3200 0 - The attached screen is the 3.5" TFT
3201 1 - The attached screen is the 7" TFT
3202 2 - The VGA Shield is attached (1024x768)
3203 Leaving out the screen size parameter will not load
3204 the TFT driver, and the framebuffer will be left
3206 b - Enable backlight. The TFT backlight pin will be
3207 linked to the kernel VESA blanking code and a GPIO
3208 LED. This parameter is not necessary when using the
3210 c - Enable the s3c camera interface.
3211 t - Reserved for enabling touchscreen support. The
3212 touchscreen support is not enabled in the mainstream
3213 kernel as of 2.6.30, a preliminary port can be found
3214 in the "bleeding edge" mini2440 support kernel at
3215 https://repo.or.cz/w/linux-2.6/mini2440.git
3218 [X86,PPC,S390,ARM64] Control optional mitigations for
3219 CPU vulnerabilities. This is a set of curated,
3220 arch-independent options, each of which is an
3221 aggregation of existing arch-specific options.
3224 Disable all optional CPU mitigations. This
3225 improves system performance, but it may also
3226 expose users to several CPU vulnerabilities.
3227 Equivalent to: nopti [X86,PPC]
3228 if nokaslr then kpti=0 [ARM64]
3229 nospectre_v1 [X86,PPC]
3231 nospectre_v2 [X86,PPC,S390,ARM64]
3232 spectre_v2_user=off [X86]
3233 spec_store_bypass_disable=off [X86,PPC]
3234 ssbd=force-off [ARM64]
3235 nospectre_bhb [ARM64]
3238 tsx_async_abort=off [X86]
3239 kvm.nx_huge_pages=off [X86]
3240 srbds=off [X86,INTEL]
3241 no_entry_flush [PPC]
3242 no_uaccess_flush [PPC]
3243 mmio_stale_data=off [X86]
3247 This does not have any effect on
3248 kvm.nx_huge_pages when
3249 kvm.nx_huge_pages=force.
3252 Mitigate all CPU vulnerabilities, but leave SMT
3253 enabled, even if it's vulnerable. This is for
3254 users who don't want to be surprised by SMT
3255 getting disabled across kernel upgrades, or who
3256 have other ways of avoiding SMT-based attacks.
3257 Equivalent to: (default behavior)
3260 Mitigate all CPU vulnerabilities, disabling SMT
3261 if needed. This is for users who always want to
3262 be fully mitigated, even if it means losing SMT.
3263 Equivalent to: l1tf=flush,nosmt [X86]
3264 mds=full,nosmt [X86]
3265 tsx_async_abort=full,nosmt [X86]
3266 mmio_stale_data=full,nosmt [X86]
3267 retbleed=auto,nosmt [X86]
3270 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
3271 parameter allows control of the logging verbosity for
3272 the additional memory initialisation checks. A value
3273 of 0 disables mminit logging and a level of 4 will
3274 log everything. Information is printed at KERN_DEBUG
3275 so loglevel=8 may also need to be specified.
3278 [X86,INTEL] Control mitigation for the Processor
3279 MMIO Stale Data vulnerabilities.
3281 Processor MMIO Stale Data is a class of
3282 vulnerabilities that may expose data after an MMIO
3283 operation. Exposed data could originate or end in
3284 the same CPU buffers as affected by MDS and TAA.
3285 Therefore, similar to MDS and TAA, the mitigation
3286 is to clear the affected CPU buffers.
3288 This parameter controls the mitigation. The
3291 full - Enable mitigation on vulnerable CPUs
3293 full,nosmt - Enable mitigation and disable SMT on
3296 off - Unconditionally disable mitigation
3298 On MDS or TAA affected machines,
3299 mmio_stale_data=off can be prevented by an active
3300 MDS or TAA mitigation as these vulnerabilities are
3301 mitigated with the same mechanism so in order to
3302 disable this mitigation, you need to specify
3303 mds=off and tsx_async_abort=off too.
3305 Not specifying this option is equivalent to
3306 mmio_stale_data=full.
3309 Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst
3311 module.async_probe=<bool>
3312 [KNL] When set to true, modules will use async probing
3313 by default. To enable/disable async probing for a
3314 specific module, use the module specific control that
3315 is documented under <module>.async_probe. When both
3316 module.async_probe and <module>.async_probe are
3317 specified, <module>.async_probe takes precedence for
3318 the specific module.
3321 [KNL] When CONFIG_MODULE_SIG is set, this means that
3322 modules without (valid) signatures will fail to load.
3323 Note that if CONFIG_MODULE_SIG_FORCE is set, that
3324 is always true, so this option does nothing.
3326 module_blacklist= [KNL] Do not load a comma-separated list of
3327 modules. Useful for debugging problem modules.
3330 [MOUSE] Maximum time between finger touching and
3331 leaving touchpad surface for touch to be considered
3332 a tap and be reported as a left button click (for
3333 touchpads working in absolute mode only).
3335 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
3336 reporting absolute coordinates, such as tablets
3337 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
3338 reporting absolute coordinates, such as tablets
3340 movablecore= [KNL,X86,IA-64,PPC]
3341 Format: nn[KMGTPE] | nn%
3342 This parameter is the complement to kernelcore=, it
3343 specifies the amount of memory used for migratable
3344 allocations. If both kernelcore and movablecore is
3345 specified, then kernelcore will be at *least* the
3346 specified value but may be more. If movablecore on its
3347 own is specified, the administrator must be careful
3348 that the amount of memory usable for all allocations
3351 movable_node [KNL] Boot-time switch to make hotplugable memory
3352 NUMA nodes to be movable. This means that the memory
3353 of such nodes will be usable only for movable
3354 allocations which rules out almost all kernel
3355 allocations. Use with caution!
3357 MTD_Partition= [MTD]
3358 Format: <name>,<region-number>,<size>,<offset>
3360 MTD_Region= [MTD] Format:
3361 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
3364 See drivers/mtd/parsers/cmdlinepart.c
3367 ARM/S3C2412 JIVE boot control
3369 See arch/arm/mach-s3c/mach-jive.c
3371 mtouchusb.raw_coordinates=
3372 [HW] Make the MicroTouch USB driver use raw coordinates
3373 ('y', default) or cooked coordinates ('n')
3375 mtrr_chunk_size=nn[KMG] [X86]
3376 used for mtrr cleanup. It is largest continuous chunk
3377 that could hold holes aka. UC entries.
3379 mtrr_gran_size=nn[KMG] [X86]
3380 Used for mtrr cleanup. It is granularity of mtrr block.
3382 Large value could prevent small alignment from
3385 mtrr_spare_reg_nr=n [X86]
3387 Range: 0,7 : spare reg number
3389 Used for mtrr cleanup. It is spare mtrr entries number.
3390 Set to 2 or more if your graphical card needs more.
3392 multitce=off [PPC] This parameter disables the use of the pSeries
3393 firmware feature for updating multiple TCE entries
3396 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
3398 netdev= [NET] Network devices parameters
3399 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
3400 Note that mem_start is often overloaded to mean
3401 something different and driver-specific.
3402 This usage is only documented in each driver source
3405 netpoll.carrier_timeout=
3406 [NET] Specifies amount of time (in seconds) that
3407 netpoll should wait for a carrier. By default netpoll
3411 [NETFILTER] Enable connection tracking flow accounting
3412 0 to disable accounting
3413 1 to enable accounting
3416 nfsaddrs= [NFS] Deprecated. Use ip= instead.
3417 See Documentation/admin-guide/nfs/nfsroot.rst.
3419 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
3420 See Documentation/admin-guide/nfs/nfsroot.rst.
3422 nfsrootdebug [NFS] enable nfsroot debugging messages.
3423 See Documentation/admin-guide/nfs/nfsroot.rst.
3425 nfs.callback_nr_threads=
3426 [NFSv4] set the total number of threads that the
3427 NFS client will assign to service NFSv4 callback
3430 nfs.callback_tcpport=
3431 [NFS] set the TCP port on which the NFSv4 callback
3432 channel should listen.
3435 [NFS] sets the pathname to the program which is used
3436 to update the NFS client cache entries.
3438 nfs.cache_getent_timeout=
3439 [NFS] sets the timeout after which an attempt to
3440 update a cache entry is deemed to have failed.
3442 nfs.idmap_cache_timeout=
3443 [NFS] set the maximum lifetime for idmapper cache
3447 [NFS] enable 64-bit inode numbers.
3448 If zero, the NFS client will fake up a 32-bit inode
3449 number for the readdir() and stat() syscalls instead
3450 of returning the full 64-bit number.
3451 The default is to return 64-bit inode numbers.
3453 nfs.max_session_cb_slots=
3454 [NFSv4.1] Sets the maximum number of session
3455 slots the client will assign to the callback
3456 channel. This determines the maximum number of
3457 callbacks the client will process in parallel for
3458 a particular server.
3460 nfs.max_session_slots=
3461 [NFSv4.1] Sets the maximum number of session slots
3462 the client will attempt to negotiate with the server.
3463 This limits the number of simultaneous RPC requests
3464 that the client can send to the NFSv4.1 server.
3465 Note that there is little point in setting this
3466 value higher than the max_tcp_slot_table_limit.
3468 nfs.nfs4_disable_idmapping=
3469 [NFSv4] When set to the default of '1', this option
3470 ensures that both the RPC level authentication
3471 scheme and the NFS level operations agree to use
3472 numeric uids/gids if the mount is using the
3473 'sec=sys' security flavour. In effect it is
3474 disabling idmapping, which can make migration from
3475 legacy NFSv2/v3 systems to NFSv4 easier.
3476 Servers that do not support this mode of operation
3477 will be autodetected by the client, and it will fall
3478 back to using the idmapper.
3479 To turn off this behaviour, set the value to '0'.
3481 [NFS4] Specify an additional fixed unique ident-
3482 ification string that NFSv4 clients can insert into
3483 their nfs_client_id4 string. This is typically a
3484 UUID that is generated at system install time.
3486 nfs.send_implementation_id =
3487 [NFSv4.1] Send client implementation identification
3488 information in exchange_id requests.
3489 If zero, no implementation identification information
3491 The default is to send the implementation identification
3494 nfs.recover_lost_locks =
3495 [NFSv4] Attempt to recover locks that were lost due
3496 to a lease timeout on the server. Please note that
3497 doing this risks data corruption, since there are
3498 no guarantees that the file will remain unchanged
3499 after the locks are lost.
3500 If you want to enable the kernel legacy behaviour of
3501 attempting to recover these locks, then set this
3503 The default parameter value of '0' causes the kernel
3504 not to attempt recovery of lost locks.
3506 nfs4.layoutstats_timer =
3507 [NFSv4.2] Change the rate at which the kernel sends
3508 layoutstats to the pNFS metadata server.
3510 Setting this to value to 0 causes the kernel to use
3511 whatever value is the default set by the layout
3512 driver. A non-zero value sets the minimum interval
3513 in seconds between layoutstats transmissions.
3515 nfsd.inter_copy_offload_enable =
3516 [NFSv4.2] When set to 1, the server will support
3517 server-to-server copies for which this server is
3518 the destination of the copy.
3520 nfsd.nfsd4_ssc_umount_timeout =
3521 [NFSv4.2] When used as the destination of a
3522 server-to-server copy, knfsd temporarily mounts
3523 the source server. It caches the mount in case
3524 it will be needed again, and discards it if not
3525 used for the number of milliseconds specified by
3528 nfsd.nfs4_disable_idmapping=
3529 [NFSv4] When set to the default of '1', the NFSv4
3530 server will return only numeric uids and gids to
3531 clients using auth_sys, and will accept numeric uids
3532 and gids from such clients. This is intended to ease
3533 migration from NFSv2/v3.
3536 nmi_backtrace.backtrace_idle [KNL]
3537 Dump stacks even of idle CPUs in response to an
3538 NMI stack-backtrace request.
3540 nmi_debug= [KNL,SH] Specify one or more actions to take
3541 when a NMI is triggered.
3542 Format: [state][,regs][,debounce][,die]
3544 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
3545 Format: [panic,][nopanic,][num]
3547 0 - turn hardlockup detector in nmi_watchdog off
3548 1 - turn hardlockup detector in nmi_watchdog on
3549 When panic is specified, panic when an NMI watchdog
3550 timeout occurs (or 'nopanic' to not panic on an NMI
3551 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
3552 To disable both hard and soft lockup detectors,
3553 please see 'nowatchdog'.
3554 This is useful when you use a panic=... timeout and
3555 need the box quickly up again.
3557 These settings can be accessed at runtime via
3558 the nmi_watchdog and hardlockup_panic sysctls.
3560 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3561 emulation library even if a 387 maths coprocessor
3564 no5lvl [X86-64] Disable 5-level paging mode. Forces
3565 kernel to use 4-level paging instead.
3567 nofsgsbase [X86] Disables FSGSBASE instructions.
3570 [HW] Never suspend the console
3571 Disable suspending of consoles during suspend and
3572 hibernate operations. Once disabled, debugging
3573 messages can reach various consoles while the rest
3574 of the system is being put to sleep (ie, while
3575 debugging driver suspend/resume hooks). This may
3576 not work reliably with all consoles, but is known
3577 to work with serial and VGA consoles.
3578 To facilitate more flexible debugging, we also add
3579 console_suspend, a printk module parameter to control
3580 it. Users could use console_suspend (usually
3581 /sys/module/printk/parameters/console_suspend) to
3582 turn on/off it dynamically.
3584 novmcoredd [KNL,KDUMP]
3585 Disable device dump. Device dump allows drivers to
3586 append dump data to vmcore so you can collect driver
3587 specified debug info. Drivers can append the data
3588 without any limit and this data is stored in memory,
3589 so this may cause significant memory stress. Disabling
3590 device dump can help save memory but the driver debug
3591 data will be no longer available. This parameter
3592 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
3595 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
3596 caches in the slab allocator. Saves per-node memory,
3597 but will impact performance.
3601 noaltinstr [S390] Disables alternative instructions patching
3602 (CPU alternatives feature).
3604 noapic [SMP,APIC] Tells the kernel to not make use of any
3605 IOAPICs that may be present in the system.
3607 noautogroup Disable scheduler automatic task group creation.
3611 nodsp [SH] Disable hardware DSP at boot time.
3613 noefi Disable EFI runtime services support.
3615 no_entry_flush [PPC] Don't flush the L1-D cache when entering the kernel.
3620 Disable SMAP (Supervisor Mode Access Prevention)
3621 even if it is supported by processor.
3624 Disable SMEP (Supervisor Mode Execution Prevention)
3625 even if it is supported by processor.
3628 This affects only 32-bit executables.
3629 noexec32=on: enable non-executable mappings (default)
3630 read doesn't imply executable mappings
3631 noexec32=off: disable non-executable mappings
3632 read implies executable mappings
3634 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3636 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3637 register save and restore. The kernel will only save
3638 legacy floating-point registers on task switch.
3640 nohugeiomap [KNL,X86,PPC,ARM64] Disable kernel huge I/O mappings.
3642 nohugevmalloc [KNL,X86,PPC,ARM64] Disable kernel huge vmalloc mappings.
3644 nosmt [KNL,S390] Disable symmetric multithreading (SMT).
3645 Equivalent to smt=1.
3647 [KNL,X86] Disable symmetric multithreading (SMT).
3648 nosmt=force: Force disable SMT, cannot be undone
3649 via the sysfs control file.
3651 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1
3652 (bounds check bypass). With this option data leaks are
3653 possible in the system.
3655 nospectre_v2 [X86,PPC_E500,ARM64] Disable all mitigations for
3656 the Spectre variant 2 (indirect branch prediction)
3657 vulnerability. System may allow data leaks with this
3660 nospectre_bhb [ARM64] Disable all mitigations for Spectre-BHB (branch
3661 history injection) vulnerability. System may allow data leaks
3664 nospec_store_bypass_disable
3665 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
3668 [PPC] Don't flush the L1-D cache after accessing user data.
3670 noxsave [BUGS=X86] Disables x86 extended register state save
3671 and restore using xsave. The kernel will fallback to
3672 enabling legacy floating-point and sse state.
3674 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
3675 register states. The kernel will fall back to use
3676 xsave to save the states. By using this parameter,
3677 performance of saving the states is degraded because
3678 xsave doesn't support modified optimization while
3679 xsaveopt supports it on xsaveopt enabled systems.
3681 noxsaves [X86] Disables xsaves and xrstors used in saving and
3682 restoring x86 extended register state in compacted
3683 form of xsave area. The kernel will fall back to use
3684 xsaveopt and xrstor to save and restore the states
3685 in standard form of xsave area. By using this
3686 parameter, xsave area per process might occupy more
3687 memory on xsaves enabled systems.
3689 nohlt [ARM,ARM64,MICROBLAZE,SH] Forces the kernel to busy wait
3690 in do_idle() and not use the arch_cpu_idle()
3691 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
3692 to be effective. This is useful on platforms where the
3693 sleep(SH) or wfi(ARM,ARM64) instructions do not work
3694 correctly or when doing power measurements to evalute
3695 the impact of the sleep instructions. This is also
3696 useful when using JTAG debugger.
3698 no_file_caps Tells the kernel not to honor file capabilities. The
3699 only way then for a file to be executed with privilege
3700 is to be setuid root or executed by root.
3702 nohalt [IA-64] Tells the kernel not to use the power saving
3703 function PAL_HALT_LIGHT when idle. This increases
3704 power-consumption. On the positive side, it reduces
3705 interrupt wake-up latency, which may improve performance
3706 in certain environments such as networked servers or
3710 Force pointers printed to the console or buffers to be
3711 unhashed. By default, when a pointer is printed via %p
3712 format string, that pointer is "hashed", i.e. obscured
3713 by hashing the pointer value. This is a security feature
3714 that hides actual kernel addresses from unprivileged
3715 users, but it also makes debugging the kernel more
3716 difficult since unequal pointers can no longer be
3717 compared. However, if this command-line option is
3718 specified, then all normal pointers will have their true
3719 value printed. This option should only be specified when
3720 debugging the kernel. Please do not use on production
3723 nohibernate [HIBERNATION] Disable hibernation and resume.
3725 nohz= [KNL] Boottime enable/disable dynamic ticks
3726 Valid arguments: on, off
3729 nohz_full= [KNL,BOOT,SMP,ISOL]
3730 The argument is a cpu list, as described above.
3731 In kernels built with CONFIG_NO_HZ_FULL=y, set
3732 the specified list of CPUs whose tick will be stopped
3733 whenever possible. The boot CPU will be forced outside
3734 the range to maintain the timekeeping. Any CPUs
3735 in this list will have their RCU callbacks offloaded,
3736 just as if they had also been called out in the
3737 rcu_nocbs= boot parameter.
3739 Note that this argument takes precedence over
3740 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option.
3742 noiotrap [SH] Disables trapped I/O port accesses.
3744 noirqdebug [X86-32] Disables the code which attempts to detect and
3745 disable unhandled interrupt sources.
3747 no_timer_check [X86,APIC] Disables the code which tests for
3748 broken timer IRQ sources.
3750 noisapnp [ISAPNP] Disables ISA PnP code.
3752 noinitrd [RAM] Tells the kernel not to load any configured
3755 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
3757 [Deprecated - use intremap=off]
3761 noinvpcid [X86] Disable the INVPCID cpu feature.
3763 nojitter [IA-64] Disables jitter checking for ITC timers.
3765 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
3767 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
3771 [X86,PV_OPS] Disable paravirtualized VMware scheduler
3772 clock and use the default one.
3774 no-steal-acc [X86,PV_OPS,ARM64,PPC/PSERIES] Disable paravirtualized
3775 steal time accounting. steal time is computed, but
3776 won't influence scheduler behaviour
3778 nolapic [X86-32,APIC] Do not enable or use the local APIC.
3780 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
3782 nomca [IA-64] Disable machine check abort handling
3784 nomce [X86-32] Disable Machine Check Exception
3786 nomfgpt [X86-32] Disable Multi-Function General Purpose
3787 Timer usage (for AMD Geode machines).
3789 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3790 shutdown the other cpus. Instead use the REBOOT_VECTOR
3793 nomodeset Disable kernel modesetting. Most systems' firmware
3794 sets up a display mode and provides framebuffer memory
3795 for output. With nomodeset, DRM and fbdev drivers will
3796 not load if they could possibly displace the pre-
3797 initialized output. Only the system framebuffer will
3798 be available for use. The respective drivers will not
3799 perform display-mode changes or accelerated rendering.
3801 Useful as error fallback, or for testing and debugging.
3803 nomodule Disable module load
3805 nopat [X86] Disable PAT (page attribute table extension of
3806 pagetables) support.
3808 nopcid [X86-64] Disable the PCID cpu feature.
3810 norandmaps Don't use address space randomization. Equivalent to
3811 echo 0 > /proc/sys/kernel/randomize_va_space
3813 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
3814 with UP alternatives
3816 noresume [SWSUSP] Disables resume and restores original swap
3819 no-scroll [VGA] Disables scrollback.
3820 This is required for the Braillex ib80-piezo Braille
3821 reader made by F.H. Papenmeier (Germany).
3825 nosgx [X86-64,SGX] Disables Intel SGX kernel support.
3827 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
3828 and disable the IO APIC. legacy for "maxcpus=0".
3830 nosoftlockup [KNL] Disable the soft-lockup detector.
3832 nosync [HW,M68K] Disables sync negotiation for all devices.
3834 nowatchdog [KNL] Disable both lockup detectors, i.e.
3835 soft-lockup and NMI watchdog (hard-lockup).
3839 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
3841 NOTE: this parameter will be ignored on systems with the
3842 LEGACY_XAPIC_DISABLED bit set in the
3843 IA32_XAPIC_DISABLE_STATUS MSR.
3845 nps_mtm_hs_ctr= [KNL,ARC]
3846 This parameter sets the maximum duration, in
3847 cycles, each HW thread of the CTOP can run
3848 without interruptions, before HW switches it.
3849 The actual maximum duration is 16 times this
3851 Format: integer between 1 and 255
3854 nptcg= [IA-64] Override max number of concurrent global TLB
3855 purges which is reported from either PAL_VM_SUMMARY or
3858 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
3859 could support. nr_cpus=n : n >= 1 limits the kernel to
3860 support 'n' processors. It could be larger than the
3861 number of already plugged CPU during bootup, later in
3862 runtime you can physically add extra cpu until it reaches
3863 n. So during boot up some boot time memory for per-cpu
3864 variables need be pre-allocated for later physical cpu
3867 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
3869 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86] Disable NUMA, Only
3870 set up a single NUMA node spanning all memory.
3872 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic
3874 Allowed values are enable and disable
3876 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
3877 'node', 'default' can be specified
3878 This can be set from sysctl after boot.
3879 See Documentation/admin-guide/sysctl/vm.rst for details.
3881 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
3882 See Documentation/core-api/debugging-via-ohci1394.rst for more
3885 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
3886 Rather than timing out after 20 ms if an EC
3887 command is not properly ACKed, override the length
3888 of the timeout. We have interrupts disabled while
3889 waiting for the ACK, so if this is set too high
3890 interrupts *may* be lost!
3892 omap_mux= [OMAP] Override bootloader pin multiplexing.
3893 Format: <mux_mode0.mode_name=value>...
3894 For example, to override I2C bus2:
3895 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
3897 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
3899 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
3901 boundary - index of last SLC block on Flex-OneNAND.
3902 The remaining blocks are configured as MLC blocks.
3903 lock - Configure if Flex-OneNAND boundary should be locked.
3904 Once locked, the boundary cannot be changed.
3905 1 indicates lock status, 0 indicates unlock status.
3907 oops=panic Always panic on oopses. Default is to just kill the
3908 process, but there is a small probability of
3909 deadlocking the machine.
3910 This will also cause panics on machine check exceptions.
3911 Useful together with panic=30 to trigger a reboot.
3914 [KNL] Boolean flag to control whether the page allocator
3915 should randomize its free lists. The randomization may
3916 be automatically enabled if the kernel detects it is
3917 running on a platform with a direct-mapped memory-side
3918 cache, and this parameter can be used to
3919 override/disable that behavior. The state of the flag
3920 can be read from sysfs at:
3921 /sys/module/page_alloc/parameters/shuffle.
3923 page_owner= [KNL] Boot-time page_owner enabling option.
3924 Storage of the information about who allocated
3925 each page is disabled in default. With this switch,
3927 on: enable the feature
3929 page_poison= [KNL] Boot-time parameter changing the state of
3930 poisoning on the buddy allocator, available with
3931 CONFIG_PAGE_POISONING=y.
3932 off: turn off poisoning (default)
3933 on: turn on poisoning
3935 page_reporting.page_reporting_order=
3936 [KNL] Minimal page reporting order
3938 Adjust the minimal page reporting order. The page
3939 reporting is disabled when it exceeds (MAX_ORDER-1).
3941 panic= [KNL] Kernel behaviour on panic: delay <timeout>
3942 timeout > 0: seconds before rebooting
3943 timeout = 0: wait forever
3944 timeout < 0: reboot immediately
3947 panic_print= Bitmask for printing system info when panic happens.
3948 User can chose combination of the following bits:
3949 bit 0: print all tasks info
3950 bit 1: print system memory info
3951 bit 2: print timer info
3952 bit 3: print locks info if CONFIG_LOCKDEP is on
3953 bit 4: print ftrace buffer
3954 bit 5: print all printk messages in buffer
3955 bit 6: print all CPUs backtrace (if available in the arch)
3956 *Be aware* that this option may print a _lot_ of lines,
3957 so there are risks of losing older messages in the log.
3958 Use this option carefully, maybe worth to setup a
3959 bigger log buffer with "log_buf_len" along with this.
3961 panic_on_taint= Bitmask for conditionally calling panic() in add_taint()
3962 Format: <hex>[,nousertaint]
3963 Hexadecimal bitmask representing the set of TAINT flags
3964 that will cause the kernel to panic when add_taint() is
3965 called with any of the flags in this set.
3966 The optional switch "nousertaint" can be utilized to
3967 prevent userspace forced crashes by writing to sysctl
3968 /proc/sys/kernel/tainted any flagset matching with the
3969 bitmask set on panic_on_taint.
3970 See Documentation/admin-guide/tainted-kernels.rst for
3971 extra details on the taint flags that users can pick
3972 to compose the bitmask to assign to panic_on_taint.
3974 panic_on_warn panic() instead of WARN(). Useful to cause kdump
3977 parkbd.port= [HW] Parallel port number the keyboard adapter is
3978 connected to, default is 0.
3980 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
3981 0 for XT, 1 for AT (default is AT).
3984 parport= [HW,PPT] Specify parallel ports. 0 disables.
3985 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
3986 Use 'auto' to force the driver to use any
3987 IRQ/DMA settings detected (the default is to
3988 ignore detected IRQ/DMA settings because of
3989 possible conflicts). You can specify the base
3990 address, IRQ, and DMA settings; IRQ and DMA
3991 should be numbers, or 'auto' (for using detected
3992 settings on that particular port), or 'nofifo'
3993 (to avoid using a FIFO even if it is detected).
3994 Parallel ports are assigned in the order they
3995 are specified on the command line, starting
3998 parport_init_mode= [HW,PPT]
3999 Configure VIA parallel port to operate in
4000 a specific mode. This is necessary on Pegasos
4001 computer where firmware has no options for setting
4002 up parallel port mode and sets it to spp.
4003 Currently this function knows 686a and 8231 chips.
4004 Format: [spp|ps2|epp|ecp|ecpepp]
4006 pata_legacy.all= [HW,LIBATA]
4008 Set to non-zero to probe primary and secondary ISA
4009 port ranges on PCI systems where no PCI PATA device
4010 has been found at either range. Disabled by default.
4012 pata_legacy.autospeed= [HW,LIBATA]
4014 Set to non-zero if a chip is present that snoops speed
4015 changes. Disabled by default.
4017 pata_legacy.ht6560a= [HW,LIBATA]
4019 Set to 1, 2, or 3 for HT 6560A on the primary channel,
4020 the secondary channel, or both channels respectively.
4021 Disabled by default.
4023 pata_legacy.ht6560b= [HW,LIBATA]
4025 Set to 1, 2, or 3 for HT 6560B on the primary channel,
4026 the secondary channel, or both channels respectively.
4027 Disabled by default.
4029 pata_legacy.iordy_mask= [HW,LIBATA]
4031 IORDY enable mask. Set individual bits to allow IORDY
4032 for the respective channel. Bit 0 is for the first
4033 legacy channel handled by this driver, bit 1 is for
4034 the second channel, and so on. The sequence will often
4035 correspond to the primary legacy channel, the secondary
4036 legacy channel, and so on, but the handling of a PCI
4037 bus and the use of other driver options may interfere
4038 with the sequence. By default IORDY is allowed across
4041 pata_legacy.opti82c46x= [HW,LIBATA]
4043 Set to 1, 2, or 3 for Opti 82c611A on the primary
4044 channel, the secondary channel, or both channels
4045 respectively. Disabled by default.
4047 pata_legacy.opti82c611a= [HW,LIBATA]
4049 Set to 1, 2, or 3 for Opti 82c465MV on the primary
4050 channel, the secondary channel, or both channels
4051 respectively. Disabled by default.
4053 pata_legacy.pio_mask= [HW,LIBATA]
4055 PIO mode mask for autospeed devices. Set individual
4056 bits to allow the use of the respective PIO modes.
4057 Bit 0 is for mode 0, bit 1 is for mode 1, and so on.
4058 All modes allowed by default.
4060 pata_legacy.probe_all= [HW,LIBATA]
4062 Set to non-zero to probe tertiary and further ISA
4063 port ranges on PCI systems. Disabled by default.
4065 pata_legacy.probe_mask= [HW,LIBATA]
4067 Probe mask for legacy ISA PATA ports. Depending on
4068 platform configuration and the use of other driver
4069 options up to 6 legacy ports are supported: 0x1f0,
4070 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing
4071 of individual ports can be disabled by setting the
4072 corresponding bits in the mask to 1. Bit 0 is for
4073 the first port in the list above (0x1f0), and so on.
4074 By default all supported ports are probed.
4076 pata_legacy.qdi= [HW,LIBATA]
4078 Set to non-zero to probe QDI controllers. By default
4079 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise.
4081 pata_legacy.winbond= [HW,LIBATA]
4083 Set to non-zero to probe Winbond controllers. Use
4084 the standard I/O port (0x130) if 1, otherwise the
4085 value given is the I/O port to use (typically 0x1b0).
4086 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE,
4089 pata_platform.pio_mask= [HW,LIBATA]
4091 Supported PIO mode mask. Set individual bits to allow
4092 the use of the respective PIO modes. Bit 0 is for
4093 mode 0, bit 1 is for mode 1, and so on. Mode 0 only
4097 Halt all CPUs after the first oops has been printed for
4098 the specified number of seconds. This is to be used if
4099 your oopses keep scrolling off the screen.
4104 See header of drivers/block/paride/pcd.c.
4105 See also Documentation/admin-guide/blockdev/paride.rst.
4107 pci=option[,option...] [PCI] various PCI subsystem options.
4109 Some options herein operate on a specific device
4110 or a set of devices (<pci_dev>). These are
4111 specified in one of the following formats:
4113 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
4114 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
4116 Note: the first format specifies a PCI
4117 bus/device/function address which may change
4118 if new hardware is inserted, if motherboard
4119 firmware changes, or due to changes caused
4120 by other kernel parameters. If the
4121 domain is left unspecified, it is
4122 taken to be zero. Optionally, a path
4123 to a device through multiple device/function
4124 addresses can be specified after the base
4125 address (this is more robust against
4126 renumbering issues). The second format
4127 selects devices using IDs from the
4128 configuration space which may match multiple
4129 devices in the system.
4131 earlydump dump PCI config space before the kernel
4133 off [X86] don't probe for the PCI bus
4134 bios [X86-32] force use of PCI BIOS, don't access
4135 the hardware directly. Use this if your machine
4136 has a non-standard PCI host bridge.
4137 nobios [X86-32] disallow use of PCI BIOS, only direct
4138 hardware access methods are allowed. Use this
4139 if you experience crashes upon bootup and you
4140 suspect they are caused by the BIOS.
4141 conf1 [X86] Force use of PCI Configuration Access
4142 Mechanism 1 (config address in IO port 0xCF8,
4143 data in IO port 0xCFC, both 32-bit).
4144 conf2 [X86] Force use of PCI Configuration Access
4145 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
4146 the function, IO port 0xCFA, also 8-bit, sets
4147 bus number. The config space is then accessed
4148 through ports 0xC000-0xCFFF).
4149 See http://wiki.osdev.org/PCI for more info
4150 on the configuration access mechanisms.
4151 noaer [PCIE] If the PCIEAER kernel config parameter is
4152 enabled, this kernel boot option can be used to
4153 disable the use of PCIE advanced error reporting.
4154 nodomains [PCI] Disable support for multiple PCI
4155 root domains (aka PCI segments, in ACPI-speak).
4156 nommconf [X86] Disable use of MMCONFIG for PCI
4158 check_enable_amd_mmconf [X86] check for and enable
4159 properly configured MMIO access to PCI
4160 config space on AMD family 10h CPU
4161 nomsi [MSI] If the PCI_MSI kernel config parameter is
4162 enabled, this kernel boot option can be used to
4163 disable the use of MSI interrupts system-wide.
4164 noioapicquirk [APIC] Disable all boot interrupt quirks.
4165 Safety option to keep boot IRQs enabled. This
4166 should never be necessary.
4167 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
4168 primary IO-APIC for bridges that cannot disable
4169 boot IRQs. This fixes a source of spurious IRQs
4170 when the system masks IRQs.
4171 noioapicreroute [APIC] Disable workaround that uses the
4172 boot IRQ equivalent of an IRQ that connects to
4173 a chipset where boot IRQs cannot be disabled.
4174 The opposite of ioapicreroute.
4175 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
4176 routing table. These calls are known to be buggy
4177 on several machines and they hang the machine
4178 when used, but on other computers it's the only
4179 way to get the interrupt routing table. Try
4180 this option if the kernel is unable to allocate
4181 IRQs or discover secondary PCI buses on your
4183 rom [X86] Assign address space to expansion ROMs.
4184 Use with caution as certain devices share
4185 address decoders between ROMs and other
4187 norom [X86] Do not assign address space to
4188 expansion ROMs that do not already have
4189 BIOS assigned address ranges.
4190 nobar [X86] Do not assign address space to the
4191 BARs that weren't assigned by the BIOS.
4192 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
4193 assigned automatically to PCI devices. You can
4194 make the kernel exclude IRQs of your ISA cards
4196 pirqaddr=0xAAAAA [X86] Specify the physical address
4197 of the PIRQ table (normally generated
4198 by the BIOS) if it is outside the
4199 F0000h-100000h range.
4200 lastbus=N [X86] Scan all buses thru bus #N. Can be
4201 useful if the kernel is unable to find your
4202 secondary buses and you want to tell it
4203 explicitly which ones they are.
4204 assign-busses [X86] Always assign all PCI bus
4205 numbers ourselves, overriding
4206 whatever the firmware may have done.
4207 usepirqmask [X86] Honor the possible IRQ mask stored
4208 in the BIOS $PIR table. This is needed on
4209 some systems with broken BIOSes, notably
4210 some HP Pavilion N5400 and Omnibook XE3
4211 notebooks. This will have no effect if ACPI
4212 IRQ routing is enabled.
4213 noacpi [X86] Do not use ACPI for IRQ routing
4214 or for PCI scanning.
4215 use_crs [X86] Use PCI host bridge window information
4216 from ACPI. On BIOSes from 2008 or later, this
4217 is enabled by default. If you need to use this,
4218 please report a bug.
4219 nocrs [X86] Ignore PCI host bridge windows from ACPI.
4220 If you need to use this, please report a bug.
4221 use_e820 [X86] Use E820 reservations to exclude parts of
4222 PCI host bridge windows. This is a workaround
4223 for BIOS defects in host bridge _CRS methods.
4224 If you need to use this, please report a bug to
4225 <linux-pci@vger.kernel.org>.
4226 no_e820 [X86] Ignore E820 reservations for PCI host
4227 bridge windows. This is the default on modern
4228 hardware. If you need to use this, please report
4229 a bug to <linux-pci@vger.kernel.org>.
4230 routeirq Do IRQ routing for all PCI devices.
4231 This is normally done in pci_enable_device(),
4232 so this option is a temporary workaround
4233 for broken drivers that don't call it.
4234 skip_isa_align [X86] do not align io start addr, so can
4235 handle more pci cards
4236 noearly [X86] Don't do any early type 1 scanning.
4237 This might help on some broken boards which
4238 machine check when some devices' config space
4239 is read. But various workarounds are disabled
4240 and some IOMMU drivers will not work.
4241 bfsort Sort PCI devices into breadth-first order.
4242 This sorting is done to get a device
4243 order compatible with older (<= 2.4) kernels.
4244 nobfsort Don't sort PCI devices into breadth-first order.
4245 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
4246 tuning and use the BIOS-configured MPS defaults.
4247 pcie_bus_safe Set every device's MPS to the largest value
4248 supported by all devices below the root complex.
4249 pcie_bus_perf Set device MPS to the largest allowable MPS
4250 based on its parent bus. Also set MRRS (Max
4251 Read Request Size) to the largest supported
4252 value (no larger than the MPS that the device
4253 or bus can support) for best performance.
4254 pcie_bus_peer2peer Set every device's MPS to 128B, which
4255 every device is guaranteed to support. This
4256 configuration allows peer-to-peer DMA between
4257 any pair of devices, possibly at the cost of
4258 reduced performance. This also guarantees
4259 that hot-added devices will work.
4260 cbiosize=nn[KMG] The fixed amount of bus space which is
4261 reserved for the CardBus bridge's IO window.
4262 The default value is 256 bytes.
4263 cbmemsize=nn[KMG] The fixed amount of bus space which is
4264 reserved for the CardBus bridge's memory
4265 window. The default value is 64 megabytes.
4268 [<order of align>@]<pci_dev>[; ...]
4269 Specifies alignment and device to reassign
4270 aligned memory resources. How to
4271 specify the device is described above.
4272 If <order of align> is not specified,
4273 PAGE_SIZE is used as alignment.
4274 A PCI-PCI bridge can be specified if resource
4275 windows need to be expanded.
4276 To specify the alignment for several
4277 instances of a device, the PCI vendor,
4278 device, subvendor, and subdevice may be
4279 specified, e.g., 12@pci:8086:9c22:103c:198f
4280 for 4096-byte alignment.
4281 ecrc= Enable/disable PCIe ECRC (transaction layer
4282 end-to-end CRC checking).
4283 bios: Use BIOS/firmware settings. This is the
4287 hpiosize=nn[KMG] The fixed amount of bus space which is
4288 reserved for hotplug bridge's IO window.
4289 Default size is 256 bytes.
4290 hpmmiosize=nn[KMG] The fixed amount of bus space which is
4291 reserved for hotplug bridge's MMIO window.
4292 Default size is 2 megabytes.
4293 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
4294 reserved for hotplug bridge's MMIO_PREF window.
4295 Default size is 2 megabytes.
4296 hpmemsize=nn[KMG] The fixed amount of bus space which is
4297 reserved for hotplug bridge's MMIO and
4299 Default size is 2 megabytes.
4300 hpbussize=nn The minimum amount of additional bus numbers
4301 reserved for buses below a hotplug bridge.
4303 realloc= Enable/disable reallocating PCI bridge resources
4304 if allocations done by BIOS are too small to
4305 accommodate resources required by all child
4307 off: Turn realloc off
4309 realloc same as realloc=on
4310 noari do not use PCIe ARI.
4311 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
4312 do not use PCIe ATS (and IOMMU device IOTLB).
4313 pcie_scan_all Scan all possible PCIe devices. Otherwise we
4314 only look for one device below a PCIe downstream
4316 big_root_window Try to add a big 64bit memory window to the PCIe
4317 root complex on AMD CPUs. Some GFX hardware
4318 can resize a BAR to allow access to all VRAM.
4319 Adding the window is slightly risky (it may
4320 conflict with unreported devices), so this
4322 disable_acs_redir=<pci_dev>[; ...]
4323 Specify one or more PCI devices (in the format
4324 specified above) separated by semicolons.
4325 Each device specified will have the PCI ACS
4326 redirect capabilities forced off which will
4327 allow P2P traffic between devices through
4328 bridges without forcing it upstream. Note:
4329 this removes isolation between devices and
4330 may put more devices in an IOMMU group.
4331 force_floating [S390] Force usage of floating interrupts.
4332 nomio [S390] Do not use MIO instructions.
4333 norid [S390] ignore the RID field and force use of
4334 one PCI domain per PCI function
4336 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
4339 force Enable ASPM even on devices that claim not to support it.
4340 WARNING: Forcing ASPM on may cause system lockups.
4342 pcie_ports= [PCIE] PCIe port services handling:
4343 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
4344 even if the platform doesn't give the OS permission to
4345 use them. This may cause conflicts if the platform
4346 also tries to use these services.
4347 dpc-native Use native PCIe service for DPC only. May
4348 cause conflicts if firmware uses AER or DPC.
4349 compat Disable native PCIe services (PME, AER, DPC, PCIe
4352 pcie_port_pm= [PCIE] PCIe port power management handling:
4353 off Disable power management of all PCIe ports
4354 force Forcibly enable power management of all PCIe ports
4356 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
4357 nomsi Do not use MSI for native PCIe PME signaling (this makes
4358 all PCIe root ports use INTx for all services).
4360 pcmv= [HW,PCMCIA] BadgePAD 4
4364 Keep all power-domains already enabled by bootloader on,
4365 even if no driver has claimed them. This is useful
4366 for debug and development, but should not be
4367 needed on a platform with proper driver support.
4370 See Documentation/admin-guide/blockdev/paride.rst.
4372 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
4375 See arch/parisc/kernel/pdc_chassis.c
4377 percpu_alloc= Select which percpu first chunk allocator to use.
4378 Currently supported values are "embed" and "page".
4379 Archs may support subset or none of the selections.
4380 See comments in mm/percpu.c for details on each
4381 allocator. This parameter is primarily for debugging
4382 and performance comparison.
4385 See Documentation/admin-guide/blockdev/paride.rst.
4388 See Documentation/admin-guide/blockdev/paride.rst.
4390 pirq= [SMP,APIC] Manual mp-table setup
4391 See Documentation/x86/i386/IO-APIC.rst.
4393 plip= [PPT,NET] Parallel port network link
4394 Format: { parport<nr> | timid | 0 }
4395 See also Documentation/admin-guide/parport.rst.
4397 pmtmr= [X86] Manual setup of pmtmr I/O Port.
4398 Override pmtimer IOPort with a hex value.
4401 pmu_override= [PPC] Override the PMU.
4402 This option takes over the PMU facility, so it is no
4403 longer usable by perf. Setting this option starts the
4404 PMU counters by setting MMCR0 to 0 (the FC bit is
4405 cleared). If a number is given, then MMCR1 is set to
4406 that number, otherwise (e.g., 'pmu_override=on'), MMCR1
4409 pm_debug_messages [SUSPEND,KNL]
4410 Enable suspend/resume debug messages during boot up.
4413 Enable PNP debug messages (depends on the
4414 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
4415 via /sys/module/pnp/parameters/debug. We always show
4416 current resource usage; turning this on also shows
4417 possible settings and some assignment information.
4423 { on | off | curr | res | no-curr | no-res }
4426 [ISAPNP] Exclude IRQs for the autoconfiguration
4429 [ISAPNP] Exclude DMAs for the autoconfiguration
4431 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
4432 Ranges are in pairs (I/O port base and size).
4435 [ISAPNP] Exclude memory regions for the
4437 Ranges are in pairs (memory base and size).
4439 ports= [IP_VS_FTP] IPVS ftp helper module
4441 Up to 8 (IP_VS_APP_MAX_PORTS) ports
4443 Format: <port>,<port>....
4445 powersave=off [PPC] This option disables power saving features.
4446 It specifically disables cpuidle and sets the
4447 platform machine description specific power_save
4448 function to NULL. On Idle the CPU just reduces
4451 ppc_strict_facility_enable
4452 [PPC] This option catches any kernel floating point,
4453 Altivec, VSX and SPE outside of regions specifically
4454 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
4455 There is some performance impact when enabling this.
4459 Disable Hardware Transactional Memory
4462 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC
4463 none - Limited to cond_resched() calls
4464 voluntary - Limited to cond_resched() and might_sleep() calls
4465 full - Any section that isn't explicitly preempt disabled
4466 can be preempted anytime.
4468 print-fatal-signals=
4469 [KNL] debug: print fatal signals
4471 If enabled, warn about various signal handling
4472 related application anomalies: too many signals,
4473 too many POSIX.1 timers, fatal signals causing a
4476 If you hit the warning due to signal overflow,
4477 you might want to try "ulimit -i unlimited".
4481 printk.always_kmsg_dump=
4482 Trigger kmsg_dump for cases other than kernel oops or
4484 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4487 printk.console_no_auto_verbose=
4488 Disable console loglevel raise on oops, panic
4489 or lockdep-detected issues (only if lock debug is on).
4490 With an exception to setups with low baudrate on
4491 serial console, keeping this 0 is a good choice
4492 in order to provide more debug information.
4494 default: 0 (auto_verbose is enabled)
4496 printk.devkmsg={on,off,ratelimit}
4497 Control writing to /dev/kmsg.
4498 on - unlimited logging to /dev/kmsg from userspace
4499 off - logging to /dev/kmsg disabled
4500 ratelimit - ratelimit the logging
4503 printk.time= Show timing data prefixed to each printk message line
4504 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4506 processor.max_cstate= [HW,ACPI]
4507 Limit processor to maximum C-state
4508 max_cstate=9 overrides any DMI blacklist limit.
4510 processor.nocst [HW,ACPI]
4511 Ignore the _CST method to determine C-states,
4512 instead using the legacy FADT method
4514 profile= [KNL] Enable kernel profiling via /proc/profile
4515 Format: [<profiletype>,]<number>
4516 Param: <profiletype>: "schedule", "sleep", or "kvm"
4517 [defaults to kernel profiling]
4518 Param: "schedule" - profile schedule points.
4519 Param: "sleep" - profile D-state sleeping (millisecs).
4520 Requires CONFIG_SCHEDSTATS
4521 Param: "kvm" - profile VM exits.
4522 Param: <number> - step/bucket size as a power of 2 for
4523 statistical time based profiling.
4525 prompt_ramdisk= [RAM] [Deprecated]
4527 prot_virt= [S390] enable hosting protected virtual machines
4528 isolated from the hypervisor (if hardware supports
4532 psi= [KNL] Enable or disable pressure stall information
4536 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
4537 probe for; one of (bare|imps|exps|lifebook|any).
4538 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
4540 psmouse.resetafter= [HW,MOUSE]
4541 Try to reset the device after so many bad packets
4544 [HW,MOUSE] Set desired mouse resolution, in dpi.
4545 psmouse.smartscroll=
4546 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
4547 0 = disabled, 1 = enabled (default).
4549 pstore.backend= Specify the name of the pstore backend to use
4552 See Documentation/admin-guide/blockdev/paride.rst.
4554 pti= [X86-64] Control Page Table Isolation of user and
4555 kernel address spaces. Disabling this feature
4556 removes hardening, but improves performance of
4557 system calls and interrupts.
4559 on - unconditionally enable
4560 off - unconditionally disable
4561 auto - kernel detects whether your CPU model is
4562 vulnerable to issues that PTI mitigates
4564 Not specifying this option is equivalent to pti=auto.
4567 Equivalent to pti=off
4570 [KNL] Number of legacy pty's. Overwrites compiled-in
4573 quiet [KNL] Disable most log messages
4578 See Documentation/admin-guide/md.rst.
4580 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
4581 See Documentation/admin-guide/blockdev/ramdisk.rst.
4583 ramdisk_start= [RAM] RAM disk image start address
4585 random.trust_cpu=off
4586 [KNL] Disable trusting the use of the CPU's
4587 random number generator (if available) to
4588 initialize the kernel's RNG.
4590 random.trust_bootloader=off
4591 [KNL] Disable trusting the use of the a seed
4592 passed by the bootloader (if available) to
4593 initialize the kernel's RNG.
4595 randomize_kstack_offset=
4596 [KNL] Enable or disable kernel stack offset
4597 randomization, which provides roughly 5 bits of
4598 entropy, frustrating memory corruption attacks
4599 that depend on stack address determinism or
4600 cross-syscall address exposures. This is only
4601 available on architectures that have defined
4602 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET.
4603 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4604 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT.
4606 ras=option[,option,...] [KNL] RAS-specific options
4609 Disable the Correctable Errors Collector,
4610 see CONFIG_RAS_CEC help text.
4612 rcu_nocbs[=cpu-list]
4613 [KNL] The optional argument is a cpu list,
4616 In kernels built with CONFIG_RCU_NOCB_CPU=y,
4617 enable the no-callback CPU mode, which prevents
4618 such CPUs' callbacks from being invoked in
4619 softirq context. Invocation of such CPUs' RCU
4620 callbacks will instead be offloaded to "rcuox/N"
4621 kthreads created for that purpose, where "x" is
4622 "p" for RCU-preempt, "s" for RCU-sched, and "g"
4623 for the kthreads that mediate grace periods; and
4624 "N" is the CPU number. This reduces OS jitter on
4625 the offloaded CPUs, which can be useful for HPC
4626 and real-time workloads. It can also improve
4627 energy efficiency for asymmetric multiprocessors.
4629 If a cpulist is passed as an argument, the specified
4630 list of CPUs is set to no-callback mode from boot.
4632 Otherwise, if the '=' sign and the cpulist
4633 arguments are omitted, no CPU will be set to
4634 no-callback mode from boot but the mode may be
4635 toggled at runtime via cpusets.
4637 Note that this argument takes precedence over
4638 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option.
4641 Rather than requiring that offloaded CPUs
4642 (specified by rcu_nocbs= above) explicitly
4643 awaken the corresponding "rcuoN" kthreads,
4644 make these kthreads poll for callbacks.
4645 This improves the real-time response for the
4646 offloaded CPUs by relieving them of the need to
4647 wake up the corresponding kthread, but degrades
4648 energy efficiency by requiring that the kthreads
4649 periodically wake up to do the polling.
4651 rcutree.blimit= [KNL]
4652 Set maximum number of finished RCU callbacks to
4653 process in one batch.
4655 rcutree.dump_tree= [KNL]
4656 Dump the structure of the rcu_node combining tree
4657 out at early boot. This is used for diagnostic
4658 purposes, to verify correct tree setup.
4660 rcutree.gp_cleanup_delay= [KNL]
4661 Set the number of jiffies to delay each step of
4662 RCU grace-period cleanup.
4664 rcutree.gp_init_delay= [KNL]
4665 Set the number of jiffies to delay each step of
4666 RCU grace-period initialization.
4668 rcutree.gp_preinit_delay= [KNL]
4669 Set the number of jiffies to delay each step of
4670 RCU grace-period pre-initialization, that is,
4671 the propagation of recent CPU-hotplug changes up
4672 the rcu_node combining tree.
4674 rcutree.use_softirq= [KNL]
4675 If set to zero, move all RCU_SOFTIRQ processing to
4676 per-CPU rcuc kthreads. Defaults to a non-zero
4677 value, meaning that RCU_SOFTIRQ is used by default.
4678 Specify rcutree.use_softirq=0 to use rcuc kthreads.
4680 But note that CONFIG_PREEMPT_RT=y kernels disable
4681 this kernel boot parameter, forcibly setting it
4684 rcutree.rcu_fanout_exact= [KNL]
4685 Disable autobalancing of the rcu_node combining
4686 tree. This is used by rcutorture, and might
4687 possibly be useful for architectures having high
4688 cache-to-cache transfer latencies.
4690 rcutree.rcu_fanout_leaf= [KNL]
4691 Change the number of CPUs assigned to each
4692 leaf rcu_node structure. Useful for very
4693 large systems, which will choose the value 64,
4694 and for NUMA systems with large remote-access
4695 latencies, which will choose a value aligned
4696 with the appropriate hardware boundaries.
4698 rcutree.rcu_min_cached_objs= [KNL]
4699 Minimum number of objects which are cached and
4700 maintained per one CPU. Object size is equal
4701 to PAGE_SIZE. The cache allows to reduce the
4702 pressure to page allocator, also it makes the
4703 whole algorithm to behave better in low memory
4706 rcutree.rcu_delay_page_cache_fill_msec= [KNL]
4707 Set the page-cache refill delay (in milliseconds)
4708 in response to low-memory conditions. The range
4709 of permitted values is in the range 0:100000.
4711 rcutree.jiffies_till_first_fqs= [KNL]
4712 Set delay from grace-period initialization to
4713 first attempt to force quiescent states.
4714 Units are jiffies, minimum value is zero,
4715 and maximum value is HZ.
4717 rcutree.jiffies_till_next_fqs= [KNL]
4718 Set delay between subsequent attempts to force
4719 quiescent states. Units are jiffies, minimum
4720 value is one, and maximum value is HZ.
4722 rcutree.jiffies_till_sched_qs= [KNL]
4723 Set required age in jiffies for a
4724 given grace period before RCU starts
4725 soliciting quiescent-state help from
4726 rcu_note_context_switch() and cond_resched().
4727 If not specified, the kernel will calculate
4728 a value based on the most recent settings
4729 of rcutree.jiffies_till_first_fqs
4730 and rcutree.jiffies_till_next_fqs.
4731 This calculated value may be viewed in
4732 rcutree.jiffies_to_sched_qs. Any attempt to set
4733 rcutree.jiffies_to_sched_qs will be cheerfully
4736 rcutree.kthread_prio= [KNL,BOOT]
4737 Set the SCHED_FIFO priority of the RCU per-CPU
4738 kthreads (rcuc/N). This value is also used for
4739 the priority of the RCU boost threads (rcub/N)
4740 and for the RCU grace-period kthreads (rcu_bh,
4741 rcu_preempt, and rcu_sched). If RCU_BOOST is
4742 set, valid values are 1-99 and the default is 1
4743 (the least-favored priority). Otherwise, when
4744 RCU_BOOST is not set, valid values are 0-99 and
4745 the default is zero (non-realtime operation).
4746 When RCU_NOCB_CPU is set, also adjust the
4747 priority of NOCB callback kthreads.
4749 rcutree.rcu_divisor= [KNL]
4750 Set the shift-right count to use to compute
4751 the callback-invocation batch limit bl from
4752 the number of callbacks queued on this CPU.
4753 The result will be bounded below by the value of
4754 the rcutree.blimit kernel parameter. Every bl
4755 callbacks, the softirq handler will exit in
4756 order to allow the CPU to do other work.
4758 Please note that this callback-invocation batch
4759 limit applies only to non-offloaded callback
4760 invocation. Offloaded callbacks are instead
4761 invoked in the context of an rcuoc kthread, which
4762 scheduler will preempt as it does any other task.
4764 rcutree.nocb_nobypass_lim_per_jiffy= [KNL]
4765 On callback-offloaded (rcu_nocbs) CPUs,
4766 RCU reduces the lock contention that would
4767 otherwise be caused by callback floods through
4768 use of the ->nocb_bypass list. However, in the
4769 common non-flooded case, RCU queues directly to
4770 the main ->cblist in order to avoid the extra
4771 overhead of the ->nocb_bypass list and its lock.
4772 But if there are too many callbacks queued during
4773 a single jiffy, RCU pre-queues the callbacks into
4774 the ->nocb_bypass queue. The definition of "too
4775 many" is supplied by this kernel boot parameter.
4777 rcutree.rcu_nocb_gp_stride= [KNL]
4778 Set the number of NOCB callback kthreads in
4779 each group, which defaults to the square root
4780 of the number of CPUs. Larger numbers reduce
4781 the wakeup overhead on the global grace-period
4782 kthread, but increases that same overhead on
4783 each group's NOCB grace-period kthread.
4785 rcutree.qhimark= [KNL]
4786 Set threshold of queued RCU callbacks beyond which
4787 batch limiting is disabled.
4789 rcutree.qlowmark= [KNL]
4790 Set threshold of queued RCU callbacks below which
4791 batch limiting is re-enabled.
4793 rcutree.qovld= [KNL]
4794 Set threshold of queued RCU callbacks beyond which
4795 RCU's force-quiescent-state scan will aggressively
4796 enlist help from cond_resched() and sched IPIs to
4797 help CPUs more quickly reach quiescent states.
4798 Set to less than zero to make this be set based
4799 on rcutree.qhimark at boot time and to zero to
4800 disable more aggressive help enlistment.
4802 rcutree.rcu_kick_kthreads= [KNL]
4803 Cause the grace-period kthread to get an extra
4804 wake_up() if it sleeps three times longer than
4805 it should at force-quiescent-state time.
4806 This wake_up() will be accompanied by a
4807 WARN_ONCE() splat and an ftrace_dump().
4809 rcutree.rcu_unlock_delay= [KNL]
4810 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels,
4811 this specifies an rcu_read_unlock()-time delay
4812 in microseconds. This defaults to zero.
4813 Larger delays increase the probability of
4814 catching RCU pointer leaks, that is, buggy use
4815 of RCU-protected pointers after the relevant
4816 rcu_read_unlock() has completed.
4818 rcutree.sysrq_rcu= [KNL]
4819 Commandeer a sysrq key to dump out Tree RCU's
4820 rcu_node tree with an eye towards determining
4821 why a new grace period has not yet started.
4823 rcuscale.gp_async= [KNL]
4824 Measure performance of asynchronous
4825 grace-period primitives such as call_rcu().
4827 rcuscale.gp_async_max= [KNL]
4828 Specify the maximum number of outstanding
4829 callbacks per writer thread. When a writer
4830 thread exceeds this limit, it invokes the
4831 corresponding flavor of rcu_barrier() to allow
4832 previously posted callbacks to drain.
4834 rcuscale.gp_exp= [KNL]
4835 Measure performance of expedited synchronous
4836 grace-period primitives.
4838 rcuscale.holdoff= [KNL]
4839 Set test-start holdoff period. The purpose of
4840 this parameter is to delay the start of the
4841 test until boot completes in order to avoid
4844 rcuscale.kfree_rcu_test= [KNL]
4845 Set to measure performance of kfree_rcu() flooding.
4847 rcuscale.kfree_rcu_test_double= [KNL]
4848 Test the double-argument variant of kfree_rcu().
4849 If this parameter has the same value as
4850 rcuscale.kfree_rcu_test_single, both the single-
4851 and double-argument variants are tested.
4853 rcuscale.kfree_rcu_test_single= [KNL]
4854 Test the single-argument variant of kfree_rcu().
4855 If this parameter has the same value as
4856 rcuscale.kfree_rcu_test_double, both the single-
4857 and double-argument variants are tested.
4859 rcuscale.kfree_nthreads= [KNL]
4860 The number of threads running loops of kfree_rcu().
4862 rcuscale.kfree_alloc_num= [KNL]
4863 Number of allocations and frees done in an iteration.
4865 rcuscale.kfree_loops= [KNL]
4866 Number of loops doing rcuscale.kfree_alloc_num number
4867 of allocations and frees.
4869 rcuscale.nreaders= [KNL]
4870 Set number of RCU readers. The value -1 selects
4871 N, where N is the number of CPUs. A value
4872 "n" less than -1 selects N-n+1, where N is again
4873 the number of CPUs. For example, -2 selects N
4874 (the number of CPUs), -3 selects N+1, and so on.
4875 A value of "n" less than or equal to -N selects
4878 rcuscale.nwriters= [KNL]
4879 Set number of RCU writers. The values operate
4880 the same as for rcuscale.nreaders.
4881 N, where N is the number of CPUs
4883 rcuscale.perf_type= [KNL]
4884 Specify the RCU implementation to test.
4886 rcuscale.shutdown= [KNL]
4887 Shut the system down after performance tests
4888 complete. This is useful for hands-off automated
4891 rcuscale.verbose= [KNL]
4892 Enable additional printk() statements.
4894 rcuscale.writer_holdoff= [KNL]
4895 Write-side holdoff between grace periods,
4896 in microseconds. The default of zero says
4899 rcutorture.fqs_duration= [KNL]
4900 Set duration of force_quiescent_state bursts
4903 rcutorture.fqs_holdoff= [KNL]
4904 Set holdoff time within force_quiescent_state bursts
4907 rcutorture.fqs_stutter= [KNL]
4908 Set wait time between force_quiescent_state bursts
4911 rcutorture.fwd_progress= [KNL]
4912 Specifies the number of kthreads to be used
4913 for RCU grace-period forward-progress testing
4914 for the types of RCU supporting this notion.
4915 Defaults to 1 kthread, values less than zero or
4916 greater than the number of CPUs cause the number
4919 rcutorture.fwd_progress_div= [KNL]
4920 Specify the fraction of a CPU-stall-warning
4921 period to do tight-loop forward-progress testing.
4923 rcutorture.fwd_progress_holdoff= [KNL]
4924 Number of seconds to wait between successive
4925 forward-progress tests.
4927 rcutorture.fwd_progress_need_resched= [KNL]
4928 Enclose cond_resched() calls within checks for
4929 need_resched() during tight-loop forward-progress
4932 rcutorture.gp_cond= [KNL]
4933 Use conditional/asynchronous update-side
4934 primitives, if available.
4936 rcutorture.gp_exp= [KNL]
4937 Use expedited update-side primitives, if available.
4939 rcutorture.gp_normal= [KNL]
4940 Use normal (non-expedited) asynchronous
4941 update-side primitives, if available.
4943 rcutorture.gp_sync= [KNL]
4944 Use normal (non-expedited) synchronous
4945 update-side primitives, if available. If all
4946 of rcutorture.gp_cond=, rcutorture.gp_exp=,
4947 rcutorture.gp_normal=, and rcutorture.gp_sync=
4948 are zero, rcutorture acts as if is interpreted
4949 they are all non-zero.
4951 rcutorture.irqreader= [KNL]
4952 Run RCU readers from irq handlers, or, more
4953 accurately, from a timer handler. Not all RCU
4954 flavors take kindly to this sort of thing.
4956 rcutorture.leakpointer= [KNL]
4957 Leak an RCU-protected pointer out of the reader.
4958 This can of course result in splats, and is
4959 intended to test the ability of things like
4960 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect
4963 rcutorture.n_barrier_cbs= [KNL]
4964 Set callbacks/threads for rcu_barrier() testing.
4966 rcutorture.nfakewriters= [KNL]
4967 Set number of concurrent RCU writers. These just
4968 stress RCU, they don't participate in the actual
4969 test, hence the "fake".
4971 rcutorture.nocbs_nthreads= [KNL]
4972 Set number of RCU callback-offload togglers.
4973 Zero (the default) disables toggling.
4975 rcutorture.nocbs_toggle= [KNL]
4976 Set the delay in milliseconds between successive
4977 callback-offload toggling attempts.
4979 rcutorture.nreaders= [KNL]
4980 Set number of RCU readers. The value -1 selects
4981 N-1, where N is the number of CPUs. A value
4982 "n" less than -1 selects N-n-2, where N is again
4983 the number of CPUs. For example, -2 selects N
4984 (the number of CPUs), -3 selects N+1, and so on.
4986 rcutorture.object_debug= [KNL]
4987 Enable debug-object double-call_rcu() testing.
4989 rcutorture.onoff_holdoff= [KNL]
4990 Set time (s) after boot for CPU-hotplug testing.
4992 rcutorture.onoff_interval= [KNL]
4993 Set time (jiffies) between CPU-hotplug operations,
4994 or zero to disable CPU-hotplug testing.
4996 rcutorture.read_exit= [KNL]
4997 Set the number of read-then-exit kthreads used
4998 to test the interaction of RCU updaters and
4999 task-exit processing.
5001 rcutorture.read_exit_burst= [KNL]
5002 The number of times in a given read-then-exit
5003 episode that a set of read-then-exit kthreads
5006 rcutorture.read_exit_delay= [KNL]
5007 The delay, in seconds, between successive
5008 read-then-exit testing episodes.
5010 rcutorture.shuffle_interval= [KNL]
5011 Set task-shuffle interval (s). Shuffling tasks
5012 allows some CPUs to go into dyntick-idle mode
5013 during the rcutorture test.
5015 rcutorture.shutdown_secs= [KNL]
5016 Set time (s) after boot system shutdown. This
5017 is useful for hands-off automated testing.
5019 rcutorture.stall_cpu= [KNL]
5020 Duration of CPU stall (s) to test RCU CPU stall
5021 warnings, zero to disable.
5023 rcutorture.stall_cpu_block= [KNL]
5024 Sleep while stalling if set. This will result
5025 in warnings from preemptible RCU in addition
5026 to any other stall-related activity.
5028 rcutorture.stall_cpu_holdoff= [KNL]
5029 Time to wait (s) after boot before inducing stall.
5031 rcutorture.stall_cpu_irqsoff= [KNL]
5032 Disable interrupts while stalling if set.
5034 rcutorture.stall_gp_kthread= [KNL]
5035 Duration (s) of forced sleep within RCU
5036 grace-period kthread to test RCU CPU stall
5037 warnings, zero to disable. If both stall_cpu
5038 and stall_gp_kthread are specified, the
5039 kthread is starved first, then the CPU.
5041 rcutorture.stat_interval= [KNL]
5042 Time (s) between statistics printk()s.
5044 rcutorture.stutter= [KNL]
5045 Time (s) to stutter testing, for example, specifying
5046 five seconds causes the test to run for five seconds,
5047 wait for five seconds, and so on. This tests RCU's
5048 ability to transition abruptly to and from idle.
5050 rcutorture.test_boost= [KNL]
5051 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
5052 "Maybe" means test if the RCU implementation
5053 under test support RCU priority boosting.
5055 rcutorture.test_boost_duration= [KNL]
5056 Duration (s) of each individual boost test.
5058 rcutorture.test_boost_interval= [KNL]
5059 Interval (s) between each boost test.
5061 rcutorture.test_no_idle_hz= [KNL]
5062 Test RCU's dyntick-idle handling. See also the
5063 rcutorture.shuffle_interval parameter.
5065 rcutorture.torture_type= [KNL]
5066 Specify the RCU implementation to test.
5068 rcutorture.verbose= [KNL]
5069 Enable additional printk() statements.
5071 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
5072 Dump ftrace buffer after reporting RCU CPU
5075 rcupdate.rcu_cpu_stall_suppress= [KNL]
5076 Suppress RCU CPU stall warning messages.
5078 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL]
5079 Suppress RCU CPU stall warning messages and
5080 rcutorture writer stall warnings that occur
5081 during early boot, that is, during the time
5082 before the init task is spawned.
5084 rcupdate.rcu_cpu_stall_timeout= [KNL]
5085 Set timeout for RCU CPU stall warning messages.
5086 The value is in seconds and the maximum allowed
5087 value is 300 seconds.
5089 rcupdate.rcu_exp_cpu_stall_timeout= [KNL]
5090 Set timeout for expedited RCU CPU stall warning
5091 messages. The value is in milliseconds
5092 and the maximum allowed value is 21000
5093 milliseconds. Please note that this value is
5094 adjusted to an arch timer tick resolution.
5095 Setting this to zero causes the value from
5096 rcupdate.rcu_cpu_stall_timeout to be used (after
5097 conversion from seconds to milliseconds).
5099 rcupdate.rcu_expedited= [KNL]
5100 Use expedited grace-period primitives, for
5101 example, synchronize_rcu_expedited() instead
5102 of synchronize_rcu(). This reduces latency,
5103 but can increase CPU utilization, degrade
5104 real-time latency, and degrade energy efficiency.
5105 No effect on CONFIG_TINY_RCU kernels.
5107 rcupdate.rcu_normal= [KNL]
5108 Use only normal grace-period primitives,
5109 for example, synchronize_rcu() instead of
5110 synchronize_rcu_expedited(). This improves
5111 real-time latency, CPU utilization, and
5112 energy efficiency, but can expose users to
5113 increased grace-period latency. This parameter
5114 overrides rcupdate.rcu_expedited. No effect on
5115 CONFIG_TINY_RCU kernels.
5117 rcupdate.rcu_normal_after_boot= [KNL]
5118 Once boot has completed (that is, after
5119 rcu_end_inkernel_boot() has been invoked), use
5120 only normal grace-period primitives. No effect
5121 on CONFIG_TINY_RCU kernels.
5123 But note that CONFIG_PREEMPT_RT=y kernels enables
5124 this kernel boot parameter, forcibly setting
5125 it to the value one, that is, converting any
5126 post-boot attempt at an expedited RCU grace
5127 period to instead use normal non-expedited
5128 grace-period processing.
5130 rcupdate.rcu_task_collapse_lim= [KNL]
5131 Set the maximum number of callbacks present
5132 at the beginning of a grace period that allows
5133 the RCU Tasks flavors to collapse back to using
5134 a single callback queue. This switching only
5135 occurs when rcupdate.rcu_task_enqueue_lim is
5136 set to the default value of -1.
5138 rcupdate.rcu_task_contend_lim= [KNL]
5139 Set the minimum number of callback-queuing-time
5140 lock-contention events per jiffy required to
5141 cause the RCU Tasks flavors to switch to per-CPU
5142 callback queuing. This switching only occurs
5143 when rcupdate.rcu_task_enqueue_lim is set to
5144 the default value of -1.
5146 rcupdate.rcu_task_enqueue_lim= [KNL]
5147 Set the number of callback queues to use for the
5148 RCU Tasks family of RCU flavors. The default
5149 of -1 allows this to be automatically (and
5150 dynamically) adjusted. This parameter is intended
5153 rcupdate.rcu_task_ipi_delay= [KNL]
5154 Set time in jiffies during which RCU tasks will
5155 avoid sending IPIs, starting with the beginning
5156 of a given grace period. Setting a large
5157 number avoids disturbing real-time workloads,
5158 but lengthens grace periods.
5160 rcupdate.rcu_task_stall_info= [KNL]
5161 Set initial timeout in jiffies for RCU task stall
5162 informational messages, which give some indication
5163 of the problem for those not patient enough to
5164 wait for ten minutes. Informational messages are
5165 only printed prior to the stall-warning message
5166 for a given grace period. Disable with a value
5167 less than or equal to zero. Defaults to ten
5168 seconds. A change in value does not take effect
5169 until the beginning of the next grace period.
5171 rcupdate.rcu_task_stall_info_mult= [KNL]
5172 Multiplier for time interval between successive
5173 RCU task stall informational messages for a given
5174 RCU tasks grace period. This value is clamped
5175 to one through ten, inclusive. It defaults to
5176 the value three, so that the first informational
5177 message is printed 10 seconds into the grace
5178 period, the second at 40 seconds, the third at
5179 160 seconds, and then the stall warning at 600
5180 seconds would prevent a fourth at 640 seconds.
5182 rcupdate.rcu_task_stall_timeout= [KNL]
5183 Set timeout in jiffies for RCU task stall
5184 warning messages. Disable with a value less
5185 than or equal to zero. Defaults to ten minutes.
5186 A change in value does not take effect until
5187 the beginning of the next grace period.
5189 rcupdate.rcu_self_test= [KNL]
5190 Run the RCU early boot self tests
5194 Run specified binary instead of /init from the ramdisk,
5195 used for early userspace startup. See initrd.
5198 force - Override the decision by the kernel to hide the
5199 advertisement of RDRAND support (this affects
5200 certain AMD processors because of buggy BIOS
5201 support, specifically around the suspend/resume
5205 Turn on/off individual RDT features. List is:
5206 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
5208 E.g. to turn on cmt and turn off mba use:
5212 Format (x86 or x86_64):
5213 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] | d[efault] \
5215 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
5217 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
5218 (prefix with 'panic_' to set mode for panic
5220 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
5221 reboot_force is either force or not specified,
5222 reboot_cpu is s[mp]#### with #### being the processor
5223 to be used for rebooting.
5225 refscale.holdoff= [KNL]
5226 Set test-start holdoff period. The purpose of
5227 this parameter is to delay the start of the
5228 test until boot completes in order to avoid
5231 refscale.loops= [KNL]
5232 Set the number of loops over the synchronization
5233 primitive under test. Increasing this number
5234 reduces noise due to loop start/end overhead,
5235 but the default has already reduced the per-pass
5236 noise to a handful of picoseconds on ca. 2020
5239 refscale.nreaders= [KNL]
5240 Set number of readers. The default value of -1
5241 selects N, where N is roughly 75% of the number
5242 of CPUs. A value of zero is an interesting choice.
5244 refscale.nruns= [KNL]
5245 Set number of runs, each of which is dumped onto
5248 refscale.readdelay= [KNL]
5249 Set the read-side critical-section duration,
5250 measured in microseconds.
5252 refscale.scale_type= [KNL]
5253 Specify the read-protection implementation to test.
5255 refscale.shutdown= [KNL]
5256 Shut down the system at the end of the performance
5257 test. This defaults to 1 (shut it down) when
5258 refscale is built into the kernel and to 0 (leave
5259 it running) when refscale is built as a module.
5261 refscale.verbose= [KNL]
5262 Enable additional printk() statements.
5264 refscale.verbose_batched= [KNL]
5265 Batch the additional printk() statements. If zero
5266 (the default) or negative, print everything. Otherwise,
5267 print every Nth verbose statement, where N is the value
5271 [KNL, SMP] Set scheduler's default relax_domain_level.
5272 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
5274 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
5275 Format: <base1>,<size1>[,<base2>,<size2>,...]
5276 Reserve I/O ports or memory so the kernel won't use
5277 them. If <base> is less than 0x10000, the region
5278 is assumed to be I/O ports; otherwise it is memory.
5280 reservetop= [X86-32]
5282 Reserves a hole at the top of the kernel virtual
5285 reset_devices [KNL] Force drivers to reset the underlying device
5286 during initialization.
5289 Specify the partition device for software suspend
5291 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
5293 resume_offset= [SWSUSP]
5294 Specify the offset from the beginning of the partition
5295 given by "resume=" at which the swap header is located,
5296 in <PAGE_SIZE> units (needed only for swap files).
5297 See Documentation/power/swsusp-and-swap-files.rst
5299 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
5300 read the resume files
5302 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
5303 Useful for devices that are detected asynchronously
5304 (e.g. USB and MMC devices).
5306 retain_initrd [RAM] Keep initrd memory after extraction
5308 retbleed= [X86] Control mitigation of RETBleed (Arbitrary
5309 Speculative Code Execution with Return Instructions)
5312 AMD-based UNRET and IBPB mitigations alone do not stop
5313 sibling threads from influencing the predictions of other
5314 sibling threads. For that reason, STIBP is used on pro-
5315 cessors that support it, and mitigate SMT on processors
5319 auto - automatically select a migitation
5320 auto,nosmt - automatically select a mitigation,
5321 disabling SMT if necessary for
5322 the full mitigation (only on Zen1
5323 and older without STIBP).
5324 ibpb - On AMD, mitigate short speculation
5325 windows on basic block boundaries too.
5326 Safe, highest perf impact. It also
5327 enables STIBP if present. Not suitable
5329 ibpb,nosmt - Like "ibpb" above but will disable SMT
5330 when STIBP is not available. This is
5331 the alternative for systems which do not
5333 unret - Force enable untrained return thunks,
5334 only effective on AMD f15h-f17h based
5336 unret,nosmt - Like unret, but will disable SMT when STIBP
5337 is not available. This is the alternative for
5338 systems which do not have STIBP.
5340 Selecting 'auto' will choose a mitigation method at run
5341 time according to the CPU.
5343 Not specifying this option is equivalent to retbleed=auto.
5345 rfkill.default_state=
5346 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
5347 etc. communication is blocked by default.
5350 rfkill.master_switch_mode=
5351 0 The "airplane mode" button does nothing.
5352 1 The "airplane mode" button toggles between everything
5353 blocked and the previous configuration.
5354 2 The "airplane mode" button toggles between everything
5355 blocked and everything unblocked.
5357 rhash_entries= [KNL,NET]
5358 Set number of hash buckets for route cache
5361 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
5364 ro [KNL] Mount root device read-only on boot
5367 on Mark read-only kernel memory as read-only (default).
5368 off Leave read-only kernel memory writable for debugging.
5369 full Mark read-only kernel memory and aliases as read-only
5373 Enable the uart passthrough on the designated usb port
5374 on Rockchip SoCs. When active, the signals of the
5375 debug-uart get routed to the D+ and D- pins of the usb
5376 port and the regular usb controller gets disabled.
5378 root= [KNL] Root filesystem
5379 See name_to_dev_t comment in init/do_mounts.c.
5381 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
5382 mount the root filesystem
5384 rootflags= [KNL] Set root filesystem mount option string
5386 rootfstype= [KNL] Set root filesystem type
5388 rootwait [KNL] Wait (indefinitely) for root device to show up.
5389 Useful for devices that are detected asynchronously
5390 (e.g. USB and MMC devices).
5392 rproc_mem=nn[KMG][@address]
5393 [KNL,ARM,CMA] Remoteproc physical memory block.
5394 Memory area to be used by remote processor image,
5397 rw [KNL] Mount root device read-write on boot
5399 S [KNL] Run init in single mode
5401 s390_iommu= [HW,S390]
5402 Set s390 IOTLB flushing mode
5404 With strict flushing every unmap operation will result in
5405 an IOTLB flush. Default is lazy flushing before reuse,
5408 s390_iommu_aperture= [KNL,S390]
5409 Specifies the size of the per device DMA address space
5410 accessible through the DMA and IOMMU APIs as a decimal
5411 factor of the size of main memory.
5412 The default is 1 meaning that one can concurrently use
5413 as many DMA addresses as physical memory is installed,
5414 if supported by hardware, and thus map all of memory
5415 once. With a value of 2 one can map all of memory twice
5416 and so on. As a special case a factor of 0 imposes no
5417 restrictions other than those given by hardware at the
5418 cost of significant additional memory use for tables.
5421 See drivers/net/irda/sa1100_ir.c.
5423 sched_verbose [KNL] Enables verbose scheduler debug messages.
5425 schedstats= [KNL,X86] Enable or disable scheduled statistics.
5426 Allowed values are enable and disable. This feature
5427 incurs a small amount of overhead in the scheduler
5428 but is useful for debugging and performance tuning.
5430 sched_thermal_decay_shift=
5431 [KNL, SMP] Set a decay shift for scheduler thermal
5432 pressure signal. Thermal pressure signal follows the
5433 default decay period of other scheduler pelt
5434 signals(usually 32 ms but configurable). Setting
5435 sched_thermal_decay_shift will left shift the decay
5436 period for the thermal pressure signal by the shift
5438 i.e. with the default pelt decay period of 32 ms
5439 sched_thermal_decay_shift thermal pressure decay pr
5443 Format: integer between 0 and 10
5446 scftorture.holdoff= [KNL]
5447 Number of seconds to hold off before starting
5448 test. Defaults to zero for module insertion and
5449 to 10 seconds for built-in smp_call_function()
5452 scftorture.longwait= [KNL]
5453 Request ridiculously long waits randomly selected
5454 up to the chosen limit in seconds. Zero (the
5455 default) disables this feature. Please note
5456 that requesting even small non-zero numbers of
5457 seconds can result in RCU CPU stall warnings,
5458 softlockup complaints, and so on.
5460 scftorture.nthreads= [KNL]
5461 Number of kthreads to spawn to invoke the
5462 smp_call_function() family of functions.
5463 The default of -1 specifies a number of kthreads
5464 equal to the number of CPUs.
5466 scftorture.onoff_holdoff= [KNL]
5467 Number seconds to wait after the start of the
5468 test before initiating CPU-hotplug operations.
5470 scftorture.onoff_interval= [KNL]
5471 Number seconds to wait between successive
5472 CPU-hotplug operations. Specifying zero (which
5473 is the default) disables CPU-hotplug operations.
5475 scftorture.shutdown_secs= [KNL]
5476 The number of seconds following the start of the
5477 test after which to shut down the system. The
5478 default of zero avoids shutting down the system.
5479 Non-zero values are useful for automated tests.
5481 scftorture.stat_interval= [KNL]
5482 The number of seconds between outputting the
5483 current test statistics to the console. A value
5484 of zero disables statistics output.
5486 scftorture.stutter_cpus= [KNL]
5487 The number of jiffies to wait between each change
5488 to the set of CPUs under test.
5490 scftorture.use_cpus_read_lock= [KNL]
5491 Use use_cpus_read_lock() instead of the default
5492 preempt_disable() to disable CPU hotplug
5493 while invoking one of the smp_call_function*()
5496 scftorture.verbose= [KNL]
5497 Enable additional printk() statements.
5499 scftorture.weight_single= [KNL]
5500 The probability weighting to use for the
5501 smp_call_function_single() function with a zero
5502 "wait" parameter. A value of -1 selects the
5503 default if all other weights are -1. However,
5504 if at least one weight has some other value, a
5505 value of -1 will instead select a weight of zero.
5507 scftorture.weight_single_wait= [KNL]
5508 The probability weighting to use for the
5509 smp_call_function_single() function with a
5510 non-zero "wait" parameter. See weight_single.
5512 scftorture.weight_many= [KNL]
5513 The probability weighting to use for the
5514 smp_call_function_many() function with a zero
5515 "wait" parameter. See weight_single.
5516 Note well that setting a high probability for
5517 this weighting can place serious IPI load
5520 scftorture.weight_many_wait= [KNL]
5521 The probability weighting to use for the
5522 smp_call_function_many() function with a
5523 non-zero "wait" parameter. See weight_single
5526 scftorture.weight_all= [KNL]
5527 The probability weighting to use for the
5528 smp_call_function_all() function with a zero
5529 "wait" parameter. See weight_single and
5532 scftorture.weight_all_wait= [KNL]
5533 The probability weighting to use for the
5534 smp_call_function_all() function with a
5535 non-zero "wait" parameter. See weight_single
5538 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
5539 xtime_lock contention on larger systems, and/or RCU lock
5540 contention on all systems with CONFIG_MAXSMP set.
5541 Format: { "0" | "1" }
5542 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
5544 Note: increases power consumption, thus should only be
5545 enabled if running jitter sensitive (HPC/RT) workloads.
5547 security= [SECURITY] Choose a legacy "major" security module to
5548 enable at boot. This has been deprecated by the
5551 selinux= [SELINUX] Disable or enable SELinux at boot time.
5552 Format: { "0" | "1" }
5553 See security/selinux/Kconfig help text.
5558 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
5559 Format: { "0" | "1" }
5560 See security/apparmor/Kconfig help text
5563 Default value is set via kernel config option.
5565 serialnumber [BUGS=X86-32]
5567 sev=option[,option...] [X86-64] See Documentation/x86/x86_64/boot-options.rst
5570 Maximal number of shapers.
5578 Enable merging of slabs with similar size when the
5579 kernel is built without CONFIG_SLAB_MERGE_DEFAULT.
5582 Disable merging of slabs with similar size. May be
5583 necessary if there is some reason to distinguish
5584 allocs to different slabs, especially in hardened
5585 environments where the risk of heap overflows and
5586 layout control by attackers can usually be
5587 frustrated by disabling merging. This will reduce
5588 most of the exposure of a heap attack to a single
5589 cache (risks via metadata attacks are mostly
5590 unchanged). Debug options disable merging on their
5592 For more information see Documentation/mm/slub.rst.
5594 slab_max_order= [MM, SLAB]
5595 Determines the maximum allowed order for slabs.
5596 A high setting may cause OOMs due to memory
5597 fragmentation. Defaults to 1 for systems with
5598 more than 32MB of RAM, 0 otherwise.
5600 slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB]
5601 Enabling slub_debug allows one to determine the
5602 culprit if slab objects become corrupted. Enabling
5603 slub_debug can create guard zones around objects and
5604 may poison objects when not in use. Also tracks the
5605 last alloc / free. For more information see
5606 Documentation/mm/slub.rst.
5608 slub_max_order= [MM, SLUB]
5609 Determines the maximum allowed order for slabs.
5610 A high setting may cause OOMs due to memory
5611 fragmentation. For more information see
5612 Documentation/mm/slub.rst.
5614 slub_min_objects= [MM, SLUB]
5615 The minimum number of objects per slab. SLUB will
5616 increase the slab order up to slub_max_order to
5617 generate a sufficiently large slab able to contain
5618 the number of objects indicated. The higher the number
5619 of objects the smaller the overhead of tracking slabs
5620 and the less frequently locks need to be acquired.
5621 For more information see Documentation/mm/slub.rst.
5623 slub_min_order= [MM, SLUB]
5624 Determines the minimum page order for slabs. Must be
5625 lower than slub_max_order.
5626 For more information see Documentation/mm/slub.rst.
5628 slub_merge [MM, SLUB]
5629 Same with slab_merge.
5631 slub_nomerge [MM, SLUB]
5632 Same with slab_nomerge. This is supported for legacy.
5633 See slab_nomerge for more information.
5636 Format: <io1>[,<io2>[,...,<io8>]]
5638 smp.csd_lock_timeout= [KNL]
5639 Specify the period of time in milliseconds
5640 that smp_call_function() and friends will wait
5641 for a CPU to release the CSD lock. This is
5642 useful when diagnosing bugs involving CPUs
5643 disabling interrupts for extended periods
5644 of time. Defaults to 5,000 milliseconds, and
5645 setting a value of zero disables this feature.
5646 This feature may be more efficiently disabled
5647 using the csdlock_debug- kernel parameter.
5649 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
5650 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
5651 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
5652 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
5653 smsc-ircc2.ircc_irq= [HW] IRQ line
5654 smsc-ircc2.ircc_dma= [HW] DMA channel
5655 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
5656 0: Toshiba Satellite 1800 (GP data pin select)
5657 1: Fast pin select (default)
5660 smt= [KNL,S390] Set the maximum number of threads (logical
5661 CPUs) to use per physical CPU on systems capable of
5662 symmetric multithreading (SMT). Will be capped to the
5663 actual hardware limit.
5665 Default: -1 (no limit)
5668 [KNL] Should the soft-lockup detector generate panics.
5671 A value of 1 instructs the soft-lockup detector
5672 to panic the machine when a soft-lockup occurs. It is
5673 also controlled by the kernel.softlockup_panic sysctl
5674 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
5675 respective build-time switch to that functionality.
5677 softlockup_all_cpu_backtrace=
5678 [KNL] Should the soft-lockup detector generate
5679 backtraces on all cpus.
5682 sonypi.*= [HW] Sony Programmable I/O Control Device driver
5683 See Documentation/admin-guide/laptops/sonypi.rst
5685 spectre_v2= [X86] Control mitigation of Spectre variant 2
5686 (indirect branch speculation) vulnerability.
5687 The default operation protects the kernel from
5690 on - unconditionally enable, implies
5692 off - unconditionally disable, implies
5694 auto - kernel detects whether your CPU model is
5697 Selecting 'on' will, and 'auto' may, choose a
5698 mitigation method at run time according to the
5699 CPU, the available microcode, the setting of the
5700 CONFIG_RETPOLINE configuration option, and the
5701 compiler with which the kernel was built.
5703 Selecting 'on' will also enable the mitigation
5704 against user space to user space task attacks.
5706 Selecting 'off' will disable both the kernel and
5707 the user space protections.
5709 Specific mitigations can also be selected manually:
5711 retpoline - replace indirect branches
5712 retpoline,generic - Retpolines
5713 retpoline,lfence - LFENCE; indirect branch
5714 retpoline,amd - alias for retpoline,lfence
5715 eibrs - enhanced IBRS
5716 eibrs,retpoline - enhanced IBRS + Retpolines
5717 eibrs,lfence - enhanced IBRS + LFENCE
5718 ibrs - use IBRS to protect kernel
5720 Not specifying this option is equivalent to
5724 [X86] Control mitigation of Spectre variant 2
5725 (indirect branch speculation) vulnerability between
5728 on - Unconditionally enable mitigations. Is
5729 enforced by spectre_v2=on
5731 off - Unconditionally disable mitigations. Is
5732 enforced by spectre_v2=off
5734 prctl - Indirect branch speculation is enabled,
5735 but mitigation can be enabled via prctl
5736 per thread. The mitigation control state
5737 is inherited on fork.
5740 - Like "prctl" above, but only STIBP is
5741 controlled per thread. IBPB is issued
5742 always when switching between different user
5746 - Same as "prctl" above, but all seccomp
5747 threads will enable the mitigation unless
5748 they explicitly opt out.
5751 - Like "seccomp" above, but only STIBP is
5752 controlled per thread. IBPB is issued
5753 always when switching between different
5754 user space processes.
5756 auto - Kernel selects the mitigation depending on
5757 the available CPU features and vulnerability.
5759 Default mitigation: "prctl"
5761 Not specifying this option is equivalent to
5762 spectre_v2_user=auto.
5764 spec_store_bypass_disable=
5765 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
5766 (Speculative Store Bypass vulnerability)
5768 Certain CPUs are vulnerable to an exploit against a
5769 a common industry wide performance optimization known
5770 as "Speculative Store Bypass" in which recent stores
5771 to the same memory location may not be observed by
5772 later loads during speculative execution. The idea
5773 is that such stores are unlikely and that they can
5774 be detected prior to instruction retirement at the
5775 end of a particular speculation execution window.
5777 In vulnerable processors, the speculatively forwarded
5778 store can be used in a cache side channel attack, for
5779 example to read memory to which the attacker does not
5780 directly have access (e.g. inside sandboxed code).
5782 This parameter controls whether the Speculative Store
5783 Bypass optimization is used.
5785 On x86 the options are:
5787 on - Unconditionally disable Speculative Store Bypass
5788 off - Unconditionally enable Speculative Store Bypass
5789 auto - Kernel detects whether the CPU model contains an
5790 implementation of Speculative Store Bypass and
5791 picks the most appropriate mitigation. If the
5792 CPU is not vulnerable, "off" is selected. If the
5793 CPU is vulnerable the default mitigation is
5794 architecture and Kconfig dependent. See below.
5795 prctl - Control Speculative Store Bypass per thread
5796 via prctl. Speculative Store Bypass is enabled
5797 for a process by default. The state of the control
5798 is inherited on fork.
5799 seccomp - Same as "prctl" above, but all seccomp threads
5800 will disable SSB unless they explicitly opt out.
5802 Default mitigations:
5805 On powerpc the options are:
5807 on,auto - On Power8 and Power9 insert a store-forwarding
5808 barrier on kernel entry and exit. On Power7
5809 perform a software flush on kernel entry and
5813 Not specifying this option is equivalent to
5814 spec_store_bypass_disable=auto.
5816 spia_io_base= [HW,MTD]
5822 [X86] Enable split lock detection or bus lock detection
5824 When enabled (and if hardware support is present), atomic
5825 instructions that access data across cache line
5826 boundaries will result in an alignment check exception
5827 for split lock detection or a debug exception for
5832 warn - the kernel will emit rate-limited warnings
5833 about applications triggering the #AC
5834 exception or the #DB exception. This mode is
5835 the default on CPUs that support split lock
5836 detection or bus lock detection. Default
5837 behavior is by #AC if both features are
5838 enabled in hardware.
5840 fatal - the kernel will send SIGBUS to applications
5841 that trigger the #AC exception or the #DB
5842 exception. Default behavior is by #AC if
5843 both features are enabled in hardware.
5846 Set system wide rate limit to N bus locks
5847 per second for bus lock detection.
5850 N/A for split lock detection.
5853 If an #AC exception is hit in the kernel or in
5854 firmware (i.e. not while executing in user mode)
5855 the kernel will oops in either "warn" or "fatal"
5858 #DB exception for bus lock is triggered only when
5862 Control the Special Register Buffer Data Sampling
5865 Certain CPUs are vulnerable to an MDS-like
5866 exploit which can leak bits from the random
5869 By default, this issue is mitigated by
5870 microcode. However, the microcode fix can cause
5871 the RDRAND and RDSEED instructions to become
5872 much slower. Among other effects, this will
5873 result in reduced throughput from /dev/urandom.
5875 The microcode mitigation can be disabled with
5876 the following option:
5878 off: Disable mitigation and remove
5879 performance impact to RDRAND and RDSEED
5881 srcutree.big_cpu_lim [KNL]
5882 Specifies the number of CPUs constituting a
5883 large system, such that srcu_struct structures
5884 should immediately allocate an srcu_node array.
5885 This kernel-boot parameter defaults to 128,
5886 but takes effect only when the low-order four
5887 bits of srcutree.convert_to_big is equal to 3
5890 srcutree.convert_to_big [KNL]
5891 Specifies under what conditions an SRCU tree
5892 srcu_struct structure will be converted to big
5893 form, that is, with an rcu_node tree:
5896 1: At init_srcu_struct() time.
5897 2: When rcutorture decides to.
5898 3: Decide at boot time (default).
5899 0x1X: Above plus if high contention.
5901 Either way, the srcu_node tree will be sized based
5902 on the actual runtime number of CPUs (nr_cpu_ids)
5903 instead of the compile-time CONFIG_NR_CPUS.
5905 srcutree.counter_wrap_check [KNL]
5906 Specifies how frequently to check for
5907 grace-period sequence counter wrap for the
5908 srcu_data structure's ->srcu_gp_seq_needed field.
5909 The greater the number of bits set in this kernel
5910 parameter, the less frequently counter wrap will
5911 be checked for. Note that the bottom two bits
5914 srcutree.exp_holdoff [KNL]
5915 Specifies how many nanoseconds must elapse
5916 since the end of the last SRCU grace period for
5917 a given srcu_struct until the next normal SRCU
5918 grace period will be considered for automatic
5919 expediting. Set to zero to disable automatic
5922 srcutree.srcu_max_nodelay [KNL]
5923 Specifies the number of no-delay instances
5924 per jiffy for which the SRCU grace period
5925 worker thread will be rescheduled with zero
5926 delay. Beyond this limit, worker thread will
5927 be rescheduled with a sleep delay of one jiffy.
5929 srcutree.srcu_max_nodelay_phase [KNL]
5930 Specifies the per-grace-period phase, number of
5931 non-sleeping polls of readers. Beyond this limit,
5932 grace period worker thread will be rescheduled
5933 with a sleep delay of one jiffy, between each
5934 rescan of the readers, for a grace period phase.
5936 srcutree.srcu_retry_check_delay [KNL]
5937 Specifies number of microseconds of non-sleeping
5938 delay between each non-sleeping poll of readers.
5940 srcutree.small_contention_lim [KNL]
5941 Specifies the number of update-side contention
5942 events per jiffy will be tolerated before
5943 initiating a conversion of an srcu_struct
5944 structure to big form. Note that the value of
5945 srcutree.convert_to_big must have the 0x10 bit
5946 set for contention-based conversions to occur.
5949 Speculative Store Bypass Disable control
5951 On CPUs that are vulnerable to the Speculative
5952 Store Bypass vulnerability and offer a
5953 firmware based mitigation, this parameter
5954 indicates how the mitigation should be used:
5956 force-on: Unconditionally enable mitigation for
5957 for both kernel and userspace
5958 force-off: Unconditionally disable mitigation for
5959 for both kernel and userspace
5960 kernel: Always enable mitigation in the
5961 kernel, and offer a prctl interface
5962 to allow userspace to register its
5963 interest in being mitigated too.
5965 stack_guard_gap= [MM]
5966 override the default stack gap protection. The value
5967 is in page units and it defines how many pages prior
5968 to (for stacks growing down) resp. after (for stacks
5969 growing up) the main stack are reserved for no other
5970 mapping. Default value is 256 pages.
5972 stack_depot_disable= [KNL]
5973 Setting this to true through kernel command line will
5974 disable the stack depot thereby saving the static memory
5975 consumed by the stack hash table. By default this is set
5979 Enabled the stack tracer on boot up.
5981 stacktrace_filter=[function-list]
5982 [FTRACE] Limit the functions that the stack tracer
5983 will trace at boot up. function-list is a comma-separated
5984 list of functions. This list can be changed at run
5985 time by the stack_trace_filter file in the debugfs
5986 tracing directory. Note, this enables stack tracing
5987 and the stacktrace above is not needed.
5991 Set the STI (builtin display/keyboard on the HP-PARISC
5992 machines) console (graphic card) which should be used
5993 as the initial boot-console.
5994 See also comment in drivers/video/console/sticore.c.
5997 See comment in drivers/video/console/sticore.c.
6000 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
6005 Enable or disable strict sigaltstack size checks
6006 against the required signal frame size which
6007 depends on the supported FPU features. This can
6008 be used to filter out binaries which have
6009 not yet been made aware of AT_MINSIGSTKSZ.
6011 sunrpc.min_resvport=
6012 sunrpc.max_resvport=
6014 SunRPC servers often require that client requests
6015 originate from a privileged port (i.e. a port in the
6016 range 0 < portnr < 1024).
6017 An administrator who wishes to reserve some of these
6018 ports for other uses may adjust the range that the
6019 kernel's sunrpc client considers to be privileged
6020 using these two parameters to set the minimum and
6021 maximum port values.
6023 sunrpc.svc_rpc_per_connection_limit=
6025 Limit the number of requests that the server will
6026 process in parallel from a single connection.
6027 The default value is 0 (no limit).
6031 Control how the NFS server code allocates CPUs to
6032 service thread pools. Depending on how many NICs
6033 you have and where their interrupts are bound, this
6034 option will affect which CPUs will do NFS serving.
6035 Note: this parameter cannot be changed while the
6036 NFS server is running.
6038 auto the server chooses an appropriate mode
6039 automatically using heuristics
6040 global a single global pool contains all CPUs
6041 percpu one pool for each CPU
6042 pernode one pool for each NUMA node (equivalent
6043 to global on non-NUMA machines)
6045 sunrpc.tcp_slot_table_entries=
6046 sunrpc.udp_slot_table_entries=
6048 Sets the upper limit on the number of simultaneous
6049 RPC calls that can be sent from the client to a
6050 server. Increasing these values may allow you to
6051 improve throughput, but will also increase the
6052 amount of memory reserved for use by the client.
6054 suspend.pm_test_delay=
6056 Sets the number of seconds to remain in a suspend test
6057 mode before resuming the system (see
6058 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
6059 is set. Default value is 5.
6062 Format: { on | off | y | n | 1 | 0 }
6063 This parameter controls use of the Protected
6064 Execution Facility on pSeries.
6066 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
6067 Format: { <int> [,<int>] | force | noforce }
6068 <int> -- Number of I/O TLB slabs
6069 <int> -- Second integer after comma. Number of swiotlb
6070 areas with their own lock. Will be rounded up
6072 force -- force using of bounce buffers even if they
6073 wouldn't be automatically used by the kernel
6074 noforce -- Never use bounce buffers (for debugging)
6079 Set a sysctl parameter, right before loading the init
6080 process, as if the value was written to the respective
6081 /proc/sys/... file. Both '.' and '/' are recognized as
6082 separators. Unrecognized parameters and invalid values
6083 are reported in the kernel log. Sysctls registered
6084 later by a loaded module cannot be set this way.
6085 Example: sysctl.vm.swappiness=40
6087 sysfs.deprecated=0|1 [KNL]
6088 Enable/disable old style sysfs layout for old udev
6089 on older distributions. When this option is enabled
6090 very new udev will not work anymore. When this option
6091 is disabled (or CONFIG_SYSFS_DEPRECATED not compiled)
6092 in older udev will not work anymore.
6093 Default depends on CONFIG_SYSFS_DEPRECATED_V2 set in
6094 the kernel configuration.
6096 sysrq_always_enabled
6098 Ignore sysrq setting - this boot parameter will
6099 neutralize any effect of /proc/sys/kernel/sysrq.
6100 Useful for debugging.
6102 tcpmhash_entries= [KNL,NET]
6103 Set the number of tcp_metrics_hash slots.
6104 Default value is 8192 or 16384 depending on total
6105 ram pages. This is used to specify the TCP metrics
6106 cache size. See Documentation/networking/ip-sysctl.rst
6107 "tcp_no_metrics_save" section for more details.
6111 test_suspend= [SUSPEND]
6112 Format: { "mem" | "standby" | "freeze" }[,N]
6113 Specify "mem" (for Suspend-to-RAM) or "standby" (for
6114 standby suspend) or "freeze" (for suspend type freeze)
6115 as the system sleep state during system startup with
6116 the optional capability to repeat N number of times.
6117 The system is woken from this state using a
6118 wakeup-capable RTC alarm.
6120 thash_entries= [KNL,NET]
6121 Set number of hash buckets for TCP connection
6123 thermal.act= [HW,ACPI]
6124 -1: disable all active trip points in all thermal zones
6125 <degrees C>: override all lowest active trip points
6127 thermal.crt= [HW,ACPI]
6128 -1: disable all critical trip points in all thermal zones
6129 <degrees C>: override all critical trip points
6131 thermal.nocrt= [HW,ACPI]
6132 Set to disable actions on ACPI thermal zone
6133 critical and hot trip points.
6135 thermal.off= [HW,ACPI]
6136 1: disable ACPI thermal control
6138 thermal.psv= [HW,ACPI]
6139 -1: disable all passive trip points
6140 <degrees C>: override all passive trip points to this
6143 thermal.tzp= [HW,ACPI]
6144 Specify global default ACPI thermal zone polling rate
6145 <deci-seconds>: poll all this frequency
6146 0: no polling (default)
6149 Force threading of all interrupt handlers except those
6150 marked explicitly IRQF_NO_THREAD.
6154 Specify if the kernel should make use of the cpu
6155 topology information if the hardware supports this.
6156 The scheduler will make use of this information and
6157 e.g. base its process migration decisions on it.
6160 topology_updates= [KNL, PPC, NUMA]
6162 Specify if the kernel should ignore (off)
6163 topology updates sent by the hypervisor to this
6166 torture.disable_onoff_at_boot= [KNL]
6167 Prevent the CPU-hotplug component of torturing
6168 until after init has spawned.
6170 torture.ftrace_dump_at_shutdown= [KNL]
6171 Dump the ftrace buffer at torture-test shutdown,
6172 even if there were no errors. This can be a
6173 very costly operation when many torture tests
6174 are running concurrently, especially on systems
6175 with rotating-rust storage.
6177 torture.verbose_sleep_frequency= [KNL]
6178 Specifies how many verbose printk()s should be
6179 emitted between each sleep. The default of zero
6180 disables verbose-printk() sleeping.
6182 torture.verbose_sleep_duration= [KNL]
6183 Duration of each verbose-printk() sleep in jiffies.
6187 tpm_suspend_pcr=[HW,TPM]
6188 Format: integer pcr id
6189 Specify that at suspend time, the tpm driver
6190 should extend the specified pcr with zeros,
6191 as a workaround for some chips which fail to
6192 flush the last written pcr on TPM_SaveState.
6193 This will guarantee that all the other pcrs
6197 Have the tracepoints sent to printk as well as the
6198 tracing ring buffer. This is useful for early boot up
6199 where the system hangs or reboots and does not give the
6200 option for reading the tracing buffer or performing a
6201 ftrace_dump_on_oops.
6203 To turn off having tracepoints sent to printk,
6204 echo 0 > /proc/sys/kernel/tracepoint_printk
6205 Note, echoing 1 into this file without the
6206 tracepoint_printk kernel cmdline option has no effect.
6208 The tp_printk_stop_on_boot (see below) can also be used
6209 to stop the printing of events to console at
6214 Having tracepoints sent to printk() and activating high
6215 frequency tracepoints such as irq or sched, can cause
6216 the system to live lock.
6218 tp_printk_stop_on_boot [FTRACE]
6219 When tp_printk (above) is set, it can cause a lot of noise
6220 on the console. It may be useful to only include the
6221 printing of events during boot up, as user space may
6222 make the system inoperable.
6224 This command line option will stop the printing of events
6225 to console at the late_initcall_sync() time frame.
6227 trace_buf_size=nn[KMG]
6228 [FTRACE] will set tracing buffer size on each cpu.
6230 trace_clock= [FTRACE] Set the clock used for tracing events
6232 local - Use the per CPU time stamp counter
6233 (converted into nanoseconds). Fast, but
6234 depending on the architecture, may not be
6235 in sync between CPUs.
6236 global - Event time stamps are synchronize across
6237 CPUs. May be slower than the local clock,
6238 but better for some race conditions.
6239 counter - Simple counting of events (1, 2, ..)
6240 note, some counts may be skipped due to the
6241 infrastructure grabbing the clock more than
6243 uptime - Use jiffies as the time stamp.
6244 perf - Use the same clock that perf uses.
6245 mono - Use ktime_get_mono_fast_ns() for time stamps.
6246 mono_raw - Use ktime_get_raw_fast_ns() for time
6248 boot - Use ktime_get_boot_fast_ns() for time stamps.
6249 Architectures may add more clocks. See
6250 Documentation/trace/ftrace.rst for more details.
6252 trace_event=[event-list]
6253 [FTRACE] Set and start specified trace events in order
6254 to facilitate early boot debugging. The event-list is a
6255 comma-separated list of trace events to enable. See
6256 also Documentation/trace/events.rst
6258 trace_options=[option-list]
6259 [FTRACE] Enable or disable tracer options at boot.
6260 The option-list is a comma delimited list of options
6261 that can be enabled or disabled just as if you were
6262 to echo the option name into
6264 /sys/kernel/debug/tracing/trace_options
6266 For example, to enable stacktrace option (to dump the
6267 stack trace of each event), add to the command line:
6269 trace_options=stacktrace
6271 See also Documentation/trace/ftrace.rst "trace options"
6274 trace_trigger=[trigger-list]
6275 [FTRACE] Add a event trigger on specific events.
6276 Set a trigger on top of a specific event, with an optional
6279 The format is is "trace_trigger=<event>.<trigger>[ if <filter>],..."
6280 Where more than one trigger may be specified that are comma deliminated.
6284 trace_trigger="sched_switch.stacktrace if prev_state == 2"
6286 The above will enable the "stacktrace" trigger on the "sched_switch"
6287 event but only trigger it if the "prev_state" of the "sched_switch"
6288 event is "2" (TASK_UNINTERUPTIBLE).
6290 See also "Event triggers" in Documentation/trace/events.rst
6294 [FTRACE] enable this option to disable tracing when a
6295 warning is hit. This turns off "tracing_on". Tracing can
6296 be enabled again by echoing '1' into the "tracing_on"
6297 file located in /sys/kernel/debug/tracing/
6299 This option is useful, as it disables the trace before
6300 the WARNING dump is called, which prevents the trace to
6301 be filled with content caused by the warning output.
6303 This option can also be set at run time via the sysctl
6304 option: kernel/traceoff_on_warning
6306 transparent_hugepage=
6308 Format: [always|madvise|never]
6309 Can be used to control the default behavior of the system
6310 with respect to transparent hugepages.
6311 See Documentation/admin-guide/mm/transhuge.rst
6314 trusted.source= [KEYS]
6316 This parameter identifies the trust source as a backend
6317 for trusted keys implementation. Supported trust
6322 If not specified then it defaults to iterating through
6323 the trust source list starting with TPM and assigns the
6324 first trust source as a backend which is initialized
6325 successfully during iteration.
6329 The RNG used to generate key material for trusted keys.
6332 - the same value as trusted.source: "tpm" or "tee"
6334 If not specified, "default" is used. In this case,
6335 the RNG's choice is left to each individual trust source.
6337 tsc= Disable clocksource stability checks for TSC.
6339 [x86] reliable: mark tsc clocksource as reliable, this
6340 disables clocksource verification at runtime, as well
6341 as the stability checks done at bootup. Used to enable
6342 high-resolution timer mode on older hardware, and in
6343 virtualized environment.
6344 [x86] noirqtime: Do not use TSC to do irq accounting.
6345 Used to run time disable IRQ_TIME_ACCOUNTING on any
6346 platforms where RDTSC is slow and this accounting
6348 [x86] unstable: mark the TSC clocksource as unstable, this
6349 marks the TSC unconditionally unstable at bootup and
6350 avoids any further wobbles once the TSC watchdog notices.
6351 [x86] nowatchdog: disable clocksource watchdog. Used
6352 in situations with strict latency requirements (where
6353 interruptions from clocksource watchdog are not
6356 tsc_early_khz= [X86] Skip early TSC calibration and use the given
6357 value instead. Useful when the early TSC frequency discovery
6358 procedure is not reliable, such as on overclocked systems
6359 with CPUID.16h support and partial CPUID.15h support.
6360 Format: <unsigned int>
6362 tsx= [X86] Control Transactional Synchronization
6363 Extensions (TSX) feature in Intel processors that
6364 support TSX control.
6366 This parameter controls the TSX feature. The options are:
6368 on - Enable TSX on the system. Although there are
6369 mitigations for all known security vulnerabilities,
6370 TSX has been known to be an accelerator for
6371 several previous speculation-related CVEs, and
6372 so there may be unknown security risks associated
6373 with leaving it enabled.
6375 off - Disable TSX on the system. (Note that this
6376 option takes effect only on newer CPUs which are
6377 not vulnerable to MDS, i.e., have
6378 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
6379 the new IA32_TSX_CTRL MSR through a microcode
6380 update. This new MSR allows for the reliable
6381 deactivation of the TSX functionality.)
6383 auto - Disable TSX if X86_BUG_TAA is present,
6384 otherwise enable TSX on the system.
6386 Not specifying this option is equivalent to tsx=off.
6388 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6391 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
6392 Abort (TAA) vulnerability.
6394 Similar to Micro-architectural Data Sampling (MDS)
6395 certain CPUs that support Transactional
6396 Synchronization Extensions (TSX) are vulnerable to an
6397 exploit against CPU internal buffers which can forward
6398 information to a disclosure gadget under certain
6401 In vulnerable processors, the speculatively forwarded
6402 data can be used in a cache side channel attack, to
6403 access data to which the attacker does not have direct
6406 This parameter controls the TAA mitigation. The
6409 full - Enable TAA mitigation on vulnerable CPUs
6412 full,nosmt - Enable TAA mitigation and disable SMT on
6413 vulnerable CPUs. If TSX is disabled, SMT
6414 is not disabled because CPU is not
6415 vulnerable to cross-thread TAA attacks.
6416 off - Unconditionally disable TAA mitigation
6418 On MDS-affected machines, tsx_async_abort=off can be
6419 prevented by an active MDS mitigation as both vulnerabilities
6420 are mitigated with the same mechanism so in order to disable
6421 this mitigation, you need to specify mds=off too.
6423 Not specifying this option is equivalent to
6424 tsx_async_abort=full. On CPUs which are MDS affected
6425 and deploy MDS mitigation, TAA mitigation is not
6426 required and doesn't provide any additional
6430 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6432 turbografx.map[2|3]= [HW,JOY]
6433 TurboGraFX parallel port interface
6435 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
6436 See also Documentation/input/devices/joystick-parport.rst
6438 udbg-immortal [PPC] When debugging early kernel crashes that
6439 happen after console_init() and before a proper
6440 console driver takes over, this boot options might
6441 help "seeing" what's going on.
6443 uhash_entries= [KNL,NET]
6444 Set number of hash buckets for UDP/UDP-Lite connections
6447 [USB] Ignore overcurrent events (default N).
6448 Some badly-designed motherboards generate lots of
6449 bogus events, for ports that aren't wired to
6450 anything. Set this parameter to avoid log spamming.
6451 Note that genuine overcurrent events won't be
6455 [X86] Cause panic on unknown NMI.
6457 usbcore.authorized_default=
6458 [USB] Default USB device authorization:
6459 (default -1 = authorized except for wireless USB,
6460 0 = not authorized, 1 = authorized, 2 = authorized
6461 if device connected to internal port)
6463 usbcore.autosuspend=
6464 [USB] The autosuspend time delay (in seconds) used
6465 for newly-detected USB devices (default 2). This
6466 is the time required before an idle device will be
6467 autosuspended. Devices for which the delay is set
6468 to a negative value won't be autosuspended at all.
6470 usbcore.usbfs_snoop=
6471 [USB] Set to log all usbfs traffic (default 0 = off).
6473 usbcore.usbfs_snoop_max=
6474 [USB] Maximum number of bytes to snoop in each URB
6477 usbcore.blinkenlights=
6478 [USB] Set to cycle leds on hubs (default 0 = off).
6480 usbcore.old_scheme_first=
6481 [USB] Start with the old device initialization
6482 scheme (default 0 = off).
6484 usbcore.usbfs_memory_mb=
6485 [USB] Memory limit (in MB) for buffers allocated by
6486 usbfs (default = 16, 0 = max = 2047).
6488 usbcore.use_both_schemes=
6489 [USB] Try the other device initialization scheme
6490 if the first one fails (default 1 = enabled).
6492 usbcore.initial_descriptor_timeout=
6493 [USB] Specifies timeout for the initial 64-byte
6494 USB_REQ_GET_DESCRIPTOR request in milliseconds
6495 (default 5000 = 5.0 seconds).
6497 usbcore.nousb [USB] Disable the USB subsystem
6500 [USB] A list of quirk entries to augment the built-in
6501 usb core quirk list. List entries are separated by
6502 commas. Each entry has the form
6503 VendorID:ProductID:Flags. The IDs are 4-digit hex
6504 numbers and Flags is a set of letters. Each letter
6505 will change the built-in quirk; setting it if it is
6506 clear and clearing it if it is set. The letters have
6507 the following meanings:
6508 a = USB_QUIRK_STRING_FETCH_255 (string
6509 descriptors must not be fetched using
6511 b = USB_QUIRK_RESET_RESUME (device can't resume
6512 correctly so reset it instead);
6513 c = USB_QUIRK_NO_SET_INTF (device can't handle
6514 Set-Interface requests);
6515 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
6516 handle its Configuration or Interface
6518 e = USB_QUIRK_RESET (device can't be reset
6519 (e.g morph devices), don't use reset);
6520 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
6521 more interface descriptions than the
6522 bNumInterfaces count, and can't handle
6523 talking to these interfaces);
6524 g = USB_QUIRK_DELAY_INIT (device needs a pause
6525 during initialization, after we read
6526 the device descriptor);
6527 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
6528 high speed and super speed interrupt
6529 endpoints, the USB 2.0 and USB 3.0 spec
6530 require the interval in microframes (1
6531 microframe = 125 microseconds) to be
6532 calculated as interval = 2 ^
6534 Devices with this quirk report their
6535 bInterval as the result of this
6536 calculation instead of the exponent
6537 variable used in the calculation);
6538 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
6539 handle device_qualifier descriptor
6541 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
6542 generates spurious wakeup, ignore
6543 remote wakeup capability);
6544 k = USB_QUIRK_NO_LPM (device can't handle Link
6546 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
6547 (Device reports its bInterval as linear
6548 frames instead of the USB 2.0
6550 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
6551 to be disconnected before suspend to
6552 prevent spurious wakeup);
6553 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
6554 pause after every control message);
6555 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
6556 delay after resetting its port);
6557 Example: quirks=0781:5580:bk,0a5c:5834:gij
6560 [USBHID] The interval which mice are to be polled at.
6563 [USBHID] The interval which joysticks are to be polled at.
6566 [USBHID] The interval which keyboards are to be polled at.
6568 usb-storage.delay_use=
6569 [UMS] The delay in seconds before a new device is
6570 scanned for Logical Units (default 1).
6573 [UMS] A list of quirks entries to supplement or
6574 override the built-in unusual_devs list. List
6575 entries are separated by commas. Each entry has
6576 the form VID:PID:Flags where VID and PID are Vendor
6577 and Product ID values (4-digit hex numbers) and
6578 Flags is a set of characters, each corresponding
6579 to a common usb-storage quirk flag as follows:
6580 a = SANE_SENSE (collect more than 18 bytes
6581 of sense data, not on uas);
6582 b = BAD_SENSE (don't collect more than 18
6583 bytes of sense data, not on uas);
6584 c = FIX_CAPACITY (decrease the reported
6585 device capacity by one sector);
6586 d = NO_READ_DISC_INFO (don't use
6587 READ_DISC_INFO command, not on uas);
6588 e = NO_READ_CAPACITY_16 (don't use
6589 READ_CAPACITY_16 command);
6590 f = NO_REPORT_OPCODES (don't use report opcodes
6592 g = MAX_SECTORS_240 (don't transfer more than
6593 240 sectors at a time, uas only);
6594 h = CAPACITY_HEURISTICS (decrease the
6595 reported device capacity by one
6596 sector if the number is odd);
6597 i = IGNORE_DEVICE (don't bind to this
6599 j = NO_REPORT_LUNS (don't use report luns
6601 k = NO_SAME (do not use WRITE_SAME, uas only)
6602 l = NOT_LOCKABLE (don't try to lock and
6603 unlock ejectable media, not on uas);
6604 m = MAX_SECTORS_64 (don't transfer more
6605 than 64 sectors = 32 KB at a time,
6607 n = INITIAL_READ10 (force a retry of the
6608 initial READ(10) command, not on uas);
6609 o = CAPACITY_OK (accept the capacity
6610 reported by the device, not on uas);
6611 p = WRITE_CACHE (the device cache is ON
6612 by default, not on uas);
6613 r = IGNORE_RESIDUE (the device reports
6614 bogus residue values, not on uas);
6615 s = SINGLE_LUN (the device has only one
6617 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
6618 commands, uas only);
6619 u = IGNORE_UAS (don't bind to the uas driver);
6620 w = NO_WP_DETECT (don't test whether the
6621 medium is write-protected).
6622 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
6623 even if the device claims no cache,
6625 Example: quirks=0419:aaf5:rl,0421:0433:rc
6627 user_debug= [KNL,ARM]
6629 See arch/arm/Kconfig.debug help text.
6630 1 - undefined instruction events
6632 4 - invalid data aborts
6635 Example: user_debug=31
6638 [X86] Flags controlling user PTE allocations.
6640 nohigh = do not allocate PTE pages in
6641 HIGHMEM regardless of setting
6644 vdso= [X86,SH,SPARC]
6645 On X86_32, this is an alias for vdso32=. Otherwise:
6647 vdso=1: enable VDSO (the default)
6648 vdso=0: disable VDSO mapping
6650 vdso32= [X86] Control the 32-bit vDSO
6651 vdso32=1: enable 32-bit VDSO
6652 vdso32=0 or vdso32=2: disable 32-bit VDSO
6654 See the help text for CONFIG_COMPAT_VDSO for more
6655 details. If CONFIG_COMPAT_VDSO is set, the default is
6656 vdso32=0; otherwise, the default is vdso32=1.
6658 For compatibility with older kernels, vdso32=2 is an
6661 Try vdso32=0 if you encounter an error that says:
6662 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
6665 vector=percpu: enable percpu vector domain
6667 video= [FB] Frame buffer configuration
6668 See Documentation/fb/modedb.rst.
6670 video.brightness_switch_enabled= [ACPI]
6672 If set to 1, on receiving an ACPI notify event
6673 generated by hotkey, video driver will adjust brightness
6674 level and then send out the event to user space through
6675 the allocated input device. If set to 0, video driver
6676 will only send out the event without touching backlight
6681 [VMMIO] Memory mapped virtio (platform) device.
6683 <size>@<baseaddr>:<irq>[:<id>]
6685 <size> := size (can use standard suffixes
6687 <baseaddr> := physical base address
6688 <irq> := interrupt number (as passed to
6690 <id> := (optional) platform device id
6692 virtio_mmio.device=1K@0x100b0000:48:7
6694 Can be used multiple times for multiple devices.
6696 vga= [BOOT,X86-32] Select a particular video mode
6697 See Documentation/x86/boot.rst and
6698 Documentation/admin-guide/svga.rst.
6699 Use vga=ask for menu.
6700 This is actually a boot loader parameter; the value is
6701 passed to the kernel using a special protocol.
6703 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
6704 May slow down system boot speed, especially when
6705 enabled on systems with a large amount of memory.
6706 All options are enabled by default, and this
6707 interface is meant to allow for selectively
6708 enabling or disabling specific virtual memory
6711 Available options are:
6712 P Enable page structure init time poisoning
6713 - Disable all of the above options
6715 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
6716 size of <nn>. This can be used to increase the
6717 minimum size (128MB on x86). It can also be used to
6718 decrease the size and leave more room for directly
6721 vmcp_cma=nn[MG] [KNL,S390]
6722 Sets the memory size reserved for contiguous memory
6723 allocations for the vmcp device driver.
6725 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
6728 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
6731 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
6735 Controls the behavior of vsyscalls (i.e. calls to
6736 fixed addresses of 0xffffffffff600x00 from legacy
6737 code). Most statically-linked binaries and older
6738 versions of glibc use these calls. Because these
6739 functions are at fixed addresses, they make nice
6740 targets for exploits that can control RIP.
6742 emulate [default] Vsyscalls turn into traps and are
6743 emulated reasonably safely. The vsyscall
6746 xonly Vsyscalls turn into traps and are
6747 emulated reasonably safely. The vsyscall
6748 page is not readable.
6750 none Vsyscalls don't work at all. This makes
6751 them quite hard to use for exploits but
6752 might break your system.
6754 vt.color= [VT] Default text color.
6755 Format: 0xYX, X = foreground, Y = background.
6756 Default: 0x07 = light gray on black.
6758 vt.cur_default= [VT] Default cursor shape.
6759 Format: 0xCCBBAA, where AA, BB, and CC are the same as
6760 the parameters of the <Esc>[?A;B;Cc escape sequence;
6761 see VGA-softcursor.txt. Default: 2 = underline.
6763 vt.default_blu= [VT]
6764 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
6765 Change the default blue palette of the console.
6766 This is a 16-member array composed of values
6769 vt.default_grn= [VT]
6770 Format: <green0>,<green1>,<green2>,...,<green15>
6771 Change the default green palette of the console.
6772 This is a 16-member array composed of values
6775 vt.default_red= [VT]
6776 Format: <red0>,<red1>,<red2>,...,<red15>
6777 Change the default red palette of the console.
6778 This is a 16-member array composed of values
6784 Set system-wide default UTF-8 mode for all tty's.
6785 Default is 1, i.e. UTF-8 mode is enabled for all
6786 newly opened terminals.
6788 vt.global_cursor_default=
6791 Set system-wide default for whether a cursor
6792 is shown on new VTs. Default is -1,
6793 i.e. cursors will be created by default unless
6794 overridden by individual drivers. 0 will hide
6795 cursors, 1 will display them.
6797 vt.italic= [VT] Default color for italic text; 0-15.
6800 vt.underline= [VT] Default color for underlined text; 0-15.
6803 watchdog timers [HW,WDT] For information on watchdog timers,
6804 see Documentation/watchdog/watchdog-parameters.rst
6805 or other driver-specific files in the
6806 Documentation/watchdog/ directory.
6810 Set the hard lockup detector stall duration
6811 threshold in seconds. The soft lockup detector
6812 threshold is set to twice the value. A value of 0
6813 disables both lockup detectors. Default is 10
6816 workqueue.watchdog_thresh=
6817 If CONFIG_WQ_WATCHDOG is configured, workqueue can
6818 warn stall conditions and dump internal state to
6819 help debugging. 0 disables workqueue stall
6820 detection; otherwise, it's the stall threshold
6821 duration in seconds. The default value is 30 and
6822 it can be updated at runtime by writing to the
6823 corresponding sysfs file.
6825 workqueue.disable_numa
6826 By default, all work items queued to unbound
6827 workqueues are affine to the NUMA nodes they're
6828 issued on, which results in better behavior in
6829 general. If NUMA affinity needs to be disabled for
6830 whatever reason, this option can be used. Note
6831 that this also can be controlled per-workqueue for
6832 workqueues visible under /sys/bus/workqueue/.
6834 workqueue.power_efficient
6835 Per-cpu workqueues are generally preferred because
6836 they show better performance thanks to cache
6837 locality; unfortunately, per-cpu workqueues tend to
6838 be more power hungry than unbound workqueues.
6840 Enabling this makes the per-cpu workqueues which
6841 were observed to contribute significantly to power
6842 consumption unbound, leading to measurably lower
6843 power usage at the cost of small performance
6846 The default value of this parameter is determined by
6847 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
6849 workqueue.debug_force_rr_cpu
6850 Workqueue used to implicitly guarantee that work
6851 items queued without explicit CPU specified are put
6852 on the local CPU. This guarantee is no longer true
6853 and while local CPU is still preferred work items
6854 may be put on foreign CPUs. This debug option
6855 forces round-robin CPU selection to flush out
6856 usages which depend on the now broken guarantee.
6857 When enabled, memory and cache locality will be
6860 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
6861 default x2apic cluster mode on platforms
6864 xen_512gb_limit [KNL,X86-64,XEN]
6865 Restricts the kernel running paravirtualized under Xen
6866 to use only up to 512 GB of RAM. The reason to do so is
6867 crash analysis tools and Xen tools for doing domain
6868 save/restore/migration must be enabled to handle larger
6871 xen_emul_unplug= [HW,X86,XEN]
6872 Unplug Xen emulated devices
6873 Format: [unplug0,][unplug1]
6874 ide-disks -- unplug primary master IDE devices
6875 aux-ide-disks -- unplug non-primary-master IDE devices
6876 nics -- unplug network devices
6877 all -- unplug all emulated devices (NICs and IDE disks)
6878 unnecessary -- unplugging emulated devices is
6879 unnecessary even if the host did not respond to
6881 never -- do not unplug even if version check succeeds
6883 xen_legacy_crash [X86,XEN]
6884 Crash from Xen panic notifier, without executing late
6885 panic() code such as dumping handler.
6887 xen_msr_safe= [X86,XEN]
6889 Select whether to always use non-faulting (safe) MSR
6890 access functions when running as Xen PV guest. The
6891 default value is controlled by CONFIG_XEN_PV_MSR_SAFE.
6893 xen_nopvspin [X86,XEN]
6894 Disables the qspinlock slowpath using Xen PV optimizations.
6895 This parameter is obsoleted by "nopvspin" parameter, which
6896 has equivalent effect for XEN platform.
6899 Disables the PV optimizations forcing the HVM guest to
6900 run as generic HVM guest with no PV drivers.
6901 This option is obsoleted by the "nopv" option, which
6902 has equivalent effect for XEN platform.
6904 xen_no_vector_callback
6905 [KNL,X86,XEN] Disable the vector callback for Xen
6906 event channel interrupts.
6908 xen_scrub_pages= [XEN]
6909 Boolean option to control scrubbing pages before giving them back
6910 to Xen, for use by other domains. Can be also changed at runtime
6911 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
6912 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
6914 xen_timer_slop= [X86-64,XEN]
6915 Set the timer slop (in nanoseconds) for the virtual Xen
6916 timers (default is 100000). This adjusts the minimum
6917 delta of virtualized Xen timers, where lower values
6918 improve timer resolution at the expense of processing
6919 more timer interrupts.
6921 xen.balloon_boot_timeout= [XEN]
6922 The time (in seconds) to wait before giving up to boot
6923 in case initial ballooning fails to free enough memory.
6924 Applies only when running as HVM or PVH guest and
6925 started with less memory configured than allowed at
6926 max. Default is 180.
6928 xen.event_eoi_delay= [XEN]
6929 How long to delay EOI handling in case of event
6930 storms (jiffies). Default is 10.
6932 xen.event_loop_timeout= [XEN]
6933 After which time (jiffies) the event handling loop
6934 should start to delay EOI handling. Default is 2.
6936 xen.fifo_events= [XEN]
6937 Boolean parameter to disable using fifo event handling
6938 even if available. Normally fifo event handling is
6939 preferred over the 2-level event handling, as it is
6940 fairer and the number of possible event channels is
6941 much higher. Default is on (use fifo events).
6943 nopv= [X86,XEN,KVM,HYPER_V,VMWARE]
6944 Disables the PV optimizations forcing the guest to run
6945 as generic guest with no PV drivers. Currently support
6946 XEN HVM, KVM, HYPER_V and VMWARE guest.
6948 nopvspin [X86,XEN,KVM]
6949 Disables the qspinlock slow path using PV optimizations
6950 which allow the hypervisor to 'idle' the guest on lock
6953 xirc2ps_cs= [NET,PCMCIA]
6955 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
6958 By default on POWER9 and above, the kernel will
6959 natively use the XIVE interrupt controller. This option
6960 allows the fallback firmware mode to be used:
6962 off Fallback to firmware control of XIVE interrupt
6963 controller on both pseries and powernv
6964 platforms. Only useful on POWER9 and above.
6966 xive.store-eoi=off [PPC]
6967 By default on POWER10 and above, the kernel will use
6968 stores for EOI handling when the XIVE interrupt mode
6969 is active. This option allows the XIVE driver to use
6970 loads instead, as on POWER9.
6972 xhci-hcd.quirks [USB,KNL]
6973 A hex value specifying bitmask with supplemental xhci
6974 host controller quirks. Meaning of each bit can be
6975 consulted in header drivers/usb/host/xhci.h.
6978 Format: { early | on | rw | ro | off }
6979 Controls if xmon debugger is enabled. Default is off.
6980 Passing only "xmon" is equivalent to "xmon=early".
6981 early Call xmon as early as possible on boot; xmon
6982 debugger is called from setup_arch().
6983 on xmon debugger hooks will be installed so xmon
6984 is only called on a kernel crash. Default mode,
6985 i.e. either "ro" or "rw" mode, is controlled
6986 with CONFIG_XMON_DEFAULT_RO_MODE.
6987 rw xmon debugger hooks will be installed so xmon
6988 is called only on a kernel crash, mode is write,
6989 meaning SPR registers, memory and, other data
6990 can be written using xmon commands.
6991 ro same as "rw" option above but SPR registers,
6992 memory, and other data can't be written using
6994 off xmon is disabled.
6998 Do not enable amd_pstate as the default
6999 scaling driver for the supported processors
7001 Use amd_pstate as a scaling driver, driver requests a
7002 desired performance on this abstract scale and the power
7003 management firmware translates the requests into actual
7004 hardware states (core frequency, data fabric and memory