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.)
344 Do not enable amd_pstate as the default
345 scaling driver for the supported processors
347 Use amd_pstate with passive mode as a scaling driver.
348 In this mode autonomous selection is disabled.
349 Driver requests a desired performance level and platform
350 tries to match the same performance level if it is
351 satisfied by guaranteed performance level.
353 Use amd_pstate_epp driver instance as the scaling driver,
354 driver provides a hint to the hardware if software wants
355 to bias toward performance (0x0) or energy efficiency (0xff)
356 to the CPPC firmware. then CPPC power algorithm will
357 calculate the runtime workload and adjust the realtime cores
360 Activate guided autonomous mode. Driver requests minimum and
361 maximum performance level and the platform autonomously
362 selects a performance level in this range and appropriate
363 to the current workload.
365 amijoy.map= [HW,JOY] Amiga joystick support
366 Map of devices attached to JOY0DAT and JOY1DAT
368 See also Documentation/input/joydev/joystick.rst
370 analog.map= [HW,JOY] Analog joystick and gamepad support
371 Specifies type or capabilities of an analog joystick
372 connected to one of 16 gameports
373 Format: <type1>,<type2>,..<type16>
376 Power management functions (SPARCstation-4/5 + deriv.)
378 Disable APC CPU standby support. SPARCstation-Fox does
379 not play well with APC CPU idle - disable it if you have
380 APC and your system crashes randomly.
382 apic= [APIC,X86] Advanced Programmable Interrupt Controller
383 Change the output verbosity while booting
384 Format: { quiet (default) | verbose | debug }
385 Change the amount of debugging information output
386 when initialising the APIC and IO-APIC components.
387 For X86-32, this can also be used to specify an APIC
389 Format: apic=driver_name
390 Examples: apic=bigsmp
392 apic_extnmi= [APIC,X86] External NMI delivery setting
393 Format: { bsp (default) | all | none }
394 bsp: External NMI is delivered only to CPU 0
395 all: External NMIs are broadcast to all CPUs as a
397 none: External NMI is masked for all CPUs. This is
398 useful so that a dump capture kernel won't be
402 See Documentation/networking/ipv6.rst.
404 apm= [APM] Advanced Power Management
405 See header of arch/x86/kernel/apm_32.c.
407 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
408 Format: { "0" | "1" }
409 See security/apparmor/Kconfig help text
412 Default value is set via kernel config option.
414 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
415 Format: <io>,<irq>,<nodeID>
417 arm64.nobti [ARM64] Unconditionally disable Branch Target
418 Identification support
420 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication
423 arm64.nomte [ARM64] Unconditionally disable Memory Tagging Extension
426 arm64.nosve [ARM64] Unconditionally disable Scalable Vector
429 arm64.nosme [ARM64] Unconditionally disable Scalable Matrix
432 arm64.nomops [ARM64] Unconditionally disable Memory Copy and Memory
433 Set instructions support
437 atarimouse= [HW,MOUSE] Atari Mouse
439 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
440 EzKey and similar keyboards
442 atkbd.reset= [HW] Reset keyboard during initialization
444 atkbd.set= [HW] Select keyboard code set
445 Format: <int> (2 = AT (default), 3 = PS/2)
447 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
450 atkbd.softraw= [HW] Choose between synthetic and real raw mode
451 Format: <bool> (0 = real, 1 = synthetic (default))
453 atkbd.softrepeat= [HW]
454 Use software keyboard repeat
456 audit= [KNL] Enable the audit sub-system
457 Format: { "0" | "1" | "off" | "on" }
458 0 | off - kernel audit is disabled and can not be
459 enabled until the next reboot
460 unset - kernel audit is initialized but disabled and
461 will be fully enabled by the userspace auditd.
462 1 | on - kernel audit is initialized and partially
463 enabled, storing at most audit_backlog_limit
464 messages in RAM until it is fully enabled by the
468 audit_backlog_limit= [KNL] Set the audit queue size limit.
469 Format: <int> (must be >=0)
472 bau= [X86_UV] Enable the BAU on SGI UV. The default
473 behavior is to disable the BAU (i.e. bau=0).
474 Format: { "0" | "1" }
477 unset - Disable the BAU.
479 baycom_epp= [HW,AX25]
482 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
484 See header of drivers/net/hamradio/baycom_par.c.
486 baycom_ser_fdx= [HW,AX25]
487 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
488 Format: <io>,<irq>,<mode>[,<baud>]
489 See header of drivers/net/hamradio/baycom_ser_fdx.c.
491 baycom_ser_hdx= [HW,AX25]
492 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
493 Format: <io>,<irq>,<mode>
494 See header of drivers/net/hamradio/baycom_ser_hdx.c.
497 Disable BERT OS support on buggy BIOSes.
499 bgrt_disable [ACPI][X86]
500 Disable BGRT to avoid flickering OEM logo.
502 blkdevparts= Manual partition parsing of block device(s) for
503 embedded devices based on command line input.
504 See Documentation/block/cmdline-partition.rst
506 boot_delay= Milliseconds to delay each printk during boot.
507 Only works if CONFIG_BOOT_PRINTK_DELAY is enabled,
508 and you may also have to specify "lpj=". Boot_delay
509 values larger than 10 seconds (10000) are assumed
510 erroneous and ignored.
514 Extended command line options can be added to an initrd
515 and this will cause the kernel to look for it.
517 See Documentation/admin-guide/bootconfig.rst
519 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
520 bttv.radio= Most important insmod options are available as
522 bttv.pll= See Documentation/admin-guide/media/bttv.rst
525 bulk_remove=off [PPC] This parameter disables the use of the pSeries
526 firmware feature for flushing multiple hpte entries
529 c101= [NET] Moxa C101 synchronous serial card
531 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
532 Sometimes CPU hardware bugs make them report the cache
533 size incorrectly. The kernel will attempt work arounds
534 to fix known problems, but for some CPUs it is not
535 possible to determine what the correct size should be.
536 This option provides an override for these situations.
539 [NET] Specifies amount of time (in seconds) that
540 the kernel should wait for a network carrier. By default
541 it waits 120 seconds.
543 ca_keys= [KEYS] This parameter identifies a specific key(s) on
544 the system trusted keyring to be used for certificate
546 format: { id:<keyid> | builtin }
548 cca= [MIPS] Override the kernel pages' cache coherency
549 algorithm. Accepted values range from 0 to 7
550 inclusive. See arch/mips/include/asm/pgtable-bits.h
551 for platform specific values (SB1, Loongson3 and
554 ccw_timeout_log [S390]
555 See Documentation/s390/common_io.rst for details.
557 cgroup_disable= [KNL] Disable a particular controller or optional feature
558 Format: {name of the controller(s) or feature(s) to disable}
559 The effects of cgroup_disable=foo are:
560 - foo isn't auto-mounted if you mount all cgroups in
562 - foo isn't visible as an individually mountable
564 - if foo is an optional feature then the feature is
565 disabled and corresponding cgroup files are not
567 {Currently only "memory" controller deal with this and
568 cut the overhead, others just disable the usage. So
569 only cgroup_disable=memory is actually worthy}
570 Specifying "pressure" disables per-cgroup pressure
571 stall information accounting feature
573 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
574 Format: { { controller | "all" | "named" }
575 [,{ controller | "all" | "named" }...] }
576 Like cgroup_disable, but only applies to cgroup v1;
577 the blacklisted controllers remain available in cgroup2.
578 "all" blacklists all controllers and "named" disables
579 named mounts. Specifying both "all" and "named" disables
582 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
584 nosocket -- Disable socket memory accounting.
585 nokmem -- Disable kernel memory accounting.
586 nobpf -- Disable BPF memory accounting.
588 checkreqprot= [SELINUX] Set initial checkreqprot flag value.
589 Format: { "0" | "1" }
590 See security/selinux/Kconfig help text.
591 0 -- check protection applied by kernel (includes
592 any implied execute protection).
593 1 -- check protection requested by application.
594 Default value is set via a kernel config option.
595 Value can be changed at runtime via
596 /sys/fs/selinux/checkreqprot.
597 Setting checkreqprot to 1 is deprecated.
600 See Documentation/s390/common_io.rst for details.
602 clearcpuid=X[,X...] [X86]
603 Disable CPUID feature X for the kernel. See
604 arch/x86/include/asm/cpufeatures.h for the valid bit
605 numbers X. Note the Linux-specific bits are not necessarily
606 stable over kernel options, but the vendor-specific
608 X can also be a string as appearing in the flags: line
609 in /proc/cpuinfo which does not have the above
610 instability issue. However, not all features have names
612 Note that using this option will taint your kernel.
613 Also note that user programs calling CPUID directly
614 or using the feature without checking anything
615 will still see it. This just prevents it from
616 being used by the kernel or shown in /proc/cpuinfo.
617 Also note the kernel might malfunction if you disable
622 Prevents the clock framework from automatically gating
623 clocks that have not been explicitly enabled by a Linux
624 device driver but are enabled in hardware at reset or
625 by the bootloader/firmware. Note that this does not
626 force such clocks to be always-on nor does it reserve
627 those clocks in any way. This parameter is useful for
628 debug and development, but should not be needed on a
629 platform with proper driver support. For more
630 information, see Documentation/driver-api/clk.rst.
632 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
634 Forces specified clocksource (if available) to be used
635 when calculating gettimeofday(). If specified
636 clocksource is not available, it defaults to PIT.
637 Format: { pit | tsc | cyclone | pmtmr }
639 clocksource= Override the default clocksource
641 Override the default clocksource and use the clocksource
642 with the name specified.
643 Some clocksource names to choose from, depending on
645 [all] jiffies (this is the base, fallback clocksource)
647 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
648 pxa_timer,timer3,32k_counter,timer0_1
649 [X86-32] pit,hpet,tsc;
650 scx200_hrt on Geode; cyclone on IBM x440
658 clocksource.arm_arch_timer.evtstrm=
661 Enable/disable the eventstream feature of the ARM
662 architected timer so that code using WFE-based polling
663 loops can be debugged more effectively on production
666 clocksource.max_cswd_read_retries= [KNL]
667 Number of clocksource_watchdog() retries due to
668 external delays before the clock will be marked
669 unstable. Defaults to two retries, that is,
670 three attempts to read the clock under test.
672 clocksource.verify_n_cpus= [KNL]
673 Limit the number of CPUs checked for clocksources
674 marked with CLOCK_SOURCE_VERIFY_PERCPU that
675 are marked unstable due to excessive skew.
676 A negative value says to check all CPUs, while
677 zero says not to check any. Values larger than
678 nr_cpu_ids are silently truncated to nr_cpu_ids.
679 The actual CPUs are chosen randomly, with
680 no replacement if the same CPU is chosen twice.
682 clocksource-wdtest.holdoff= [KNL]
683 Set the time in seconds that the clocksource
684 watchdog test waits before commencing its tests.
685 Defaults to zero when built as a module and to
686 10 seconds when built into the kernel.
688 cma=nn[MG]@[start[MG][-end[MG]]]
690 Sets the size of kernel global memory area for
691 contiguous memory allocations and optionally the
692 placement constraint by the physical address range of
693 memory allocations. A value of 0 disables CMA
694 altogether. For more information, see
695 kernel/dma/contiguous.c
699 Sets the size of kernel per-numa memory area for
700 contiguous memory allocations. A value of 0 disables
701 per-numa CMA altogether. And If this option is not
702 specified, the default value is 0.
703 With per-numa CMA enabled, DMA users on node nid will
704 first try to allocate buffer from the pernuma area
705 which is located in node nid, if the allocation fails,
706 they will fallback to the global default memory area.
708 cmo_free_hint= [PPC] Format: { yes | no }
709 Specify whether pages are marked as being inactive
710 when they are freed. This is used in CMO environments
711 to determine OS memory pressure for page stealing by
715 coherent_pool=nn[KMG] [ARM,KNL]
716 Sets the size of memory pool for coherent, atomic dma
717 allocations, by default set to 256K.
719 com20020= [HW,NET] ARCnet - COM20020 chipset
721 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
723 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
727 ARCnet - COM90xx chipset (memory-mapped buffers)
728 Format: <io>[,<irq>[,<memstart>]]
730 condev= [HW,S390] console device
733 con3215_drop= [S390] 3215 console drop mode.
735 When set to true, drop data on the 3215 console when
736 the console buffer is full. In this case the
737 operator using a 3270 terminal emulator (for example
738 x3270) does not have to enter the clear key for the
739 console output to advance and the kernel to continue.
740 This leads to a much faster boot time when a 3270
741 terminal emulator is active. If no 3270 terminal
742 emulator is used, this parameter has no effect.
744 console= [KNL] Output console device and options.
746 tty<n> Use the virtual console device <n>.
750 Use the specified serial port. The options are of
751 the form "bbbbpnf", where "bbbb" is the baud rate,
752 "p" is parity ("n", "o", or "e"), "n" is number of
753 bits, and "f" is flow control ("r" for RTS or
754 omit it). Default is "9600n8".
756 See Documentation/admin-guide/serial-console.rst for more
758 Documentation/networking/netconsole.rst for an
761 uart[8250],io,<addr>[,options]
762 uart[8250],mmio,<addr>[,options]
763 uart[8250],mmio16,<addr>[,options]
764 uart[8250],mmio32,<addr>[,options]
765 uart[8250],0x<addr>[,options]
766 Start an early, polled-mode console on the 8250/16550
767 UART at the specified I/O port or MMIO address,
768 switching to the matching ttyS device later.
769 MMIO inter-register address stride is either 8-bit
770 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
771 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
772 to be equivalent to 'mmio'. 'options' are specified in
773 the same format described for ttyS above; if unspecified,
774 the h/w is not re-initialized.
776 hvc<n> Use the hypervisor console device <n>. This is for
777 both Xen and PowerPC hypervisors.
780 Use to disable console output, i.e., to have kernel
781 console messages discarded.
782 This must be the only console= parameter used on the
785 If the device connected to the port is not a TTY but a braille
786 device, prepend "brl," before the device type, for instance
788 For now, only VisioBraille is supported.
791 [KNL] Change console messages format
793 By default we print messages on consoles in
794 "[time stamp] text\n" format (time stamp may not be
795 printed, depending on CONFIG_PRINTK_TIME or
796 `printk_time' param).
798 Switch to syslog format: "<%u>[time stamp] text\n"
799 IOW, each message will have a facility and loglevel
800 prefix. The format is similar to one used by syslog()
801 syscall, or to executing "dmesg -S --raw" or to reading
804 consoleblank= [KNL] The console blank (screen saver) timeout in
805 seconds. A value of 0 disables the blank timer.
809 [KNL] Change the default value for
810 /proc/<pid>/coredump_filter.
811 See also Documentation/filesystems/proc.rst.
813 coresight_cpu_debug.enable
816 Enable/disable the CPU sampling based debugging.
817 0: default value, disable debugging
818 1: enable debugging at boot time
820 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
822 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
824 cpuidle.off=1 [CPU_IDLE]
825 disable the cpuidle sub-system
828 [CPU_IDLE] Name of the cpuidle governor to use.
830 cpufreq.off=1 [CPU_FREQ]
831 disable the cpufreq sub-system
833 cpufreq.default_governor=
834 [CPU_FREQ] Name of the default cpufreq governor or
835 policy to use. This governor must be registered in the
836 kernel before the cpufreq driver probes.
839 [X86] Delay for N microsec between assert and de-assert
840 of APIC INIT to start processors. This delay occurs
841 on every CPU online, such as boot, and resume from suspend.
845 [SMP] Enable/disable parallel bringup of secondary CPUs
847 Default is enabled if CONFIG_HOTPLUG_PARALLEL=y. Otherwise
848 the parameter has no effect.
850 crash_kexec_post_notifiers
851 Run kdump after running panic-notifiers and dumping
852 kmsg. This only for the users who doubt kdump always
853 succeeds in any situation.
854 Note that this also increases risks of kdump failure,
855 because some panic notifiers can make the crashed
856 kernel more unstable.
858 crashkernel=size[KMG][@offset[KMG]]
859 [KNL] Using kexec, Linux can switch to a 'crash kernel'
860 upon panic. This parameter reserves the physical
861 memory region [offset, offset + size] for that kernel
862 image. If '@offset' is omitted, then a suitable offset
863 is selected automatically.
864 [KNL, X86-64, ARM64] Select a region under 4G first, and
865 fall back to reserve region above 4G when '@offset'
866 hasn't been specified.
867 See Documentation/admin-guide/kdump/kdump.rst for further details.
869 crashkernel=range1:size1[,range2:size2,...][@offset]
870 [KNL] Same as above, but depends on the memory
871 in the running system. The syntax of range is
872 start-[end] where start and end are both
873 a memory unit (amount[KMG]). See also
874 Documentation/admin-guide/kdump/kdump.rst for an example.
876 crashkernel=size[KMG],high
877 [KNL, X86-64, ARM64] range could be above 4G. Allow kernel
878 to allocate physical memory region from top, so could
879 be above 4G if system have more than 4G ram installed.
880 Otherwise memory region will be allocated below 4G, if
882 It will be ignored if crashkernel=X is specified.
883 crashkernel=size[KMG],low
884 [KNL, X86-64, ARM64] range under 4G. When crashkernel=X,high
885 is passed, kernel could allocate physical memory region
886 above 4G, that cause second kernel crash on system
887 that require some amount of low memory, e.g. swiotlb
888 requires at least 64M+32K low memory, also enough extra
889 low memory is needed to make sure DMA buffers for 32-bit
890 devices won't run out. Kernel would try to allocate
891 default size of memory below 4G automatically. The default
892 size is platform dependent.
893 --> x86: max(swiotlb_size_or_default() + 8MiB, 256MiB)
895 This one lets the user specify own low range under 4G
896 for second kernel instead.
897 0: to disable low allocation.
898 It will be ignored when crashkernel=X,high is not used
899 or memory reserved is below 4G.
902 [KNL] Disable crypto self-tests
907 cs89x0_media= [HW,NET]
908 Format: { rj45 | aui | bnc }
910 csdlock_debug= [KNL] Enable or disable debug add-ons of cross-CPU
911 function call handling. When switched on,
912 additional debug data is printed to the console
913 in case a hanging CPU is detected, and that
914 CPU is pinged again in order to try to resolve
915 the hang situation. The default value of this
916 option depends on the CSD_LOCK_WAIT_DEBUG_DEFAULT
920 See header of drivers/s390/block/dasd_devmap.c.
922 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
923 (one device per port)
924 Format: <port#>,<type>
925 See also Documentation/input/devices/joystick-parport.rst
927 debug [KNL] Enable kernel debugging (events log level).
930 [KNL] Enable printing [hashed] pointers early in the
931 boot sequence. If enabled, we use a weak hash instead
932 of siphash to hash pointers. Use this option if you are
933 seeing instances of '(___ptrval___)') and need to see a
934 value (hashed pointer) instead. Cryptographically
935 insecure, please do not use on production kernels.
938 [KNL] verbose locking self-tests
940 Print debugging info while doing the locking API
942 Bitmask for the various LOCKTYPE_ tests. Defaults to 0
943 (no extra messages), setting it to -1 (all bits set)
944 will print _a_lot_ more information - normally only
945 useful to lockdep developers.
947 debug_objects [KNL] Enable object debugging
949 debug_guardpage_minorder=
950 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
951 parameter allows control of the order of pages that will
952 be intentionally kept free (and hence protected) by the
953 buddy allocator. Bigger value increase the probability
954 of catching random memory corruption, but reduce the
955 amount of memory for normal system use. The maximum
956 possible value is MAX_ORDER/2. Setting this parameter
957 to 1 or 2 should be enough to identify most random
958 memory corruption problems caused by bugs in kernel or
959 driver code when a CPU writes to (or reads from) a
960 random memory location. Note that there exists a class
961 of memory corruptions problems caused by buggy H/W or
962 F/W or by drivers badly programming DMA (basically when
963 memory is written at bus level and the CPU MMU is
964 bypassed) which are not detectable by
965 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
966 tracking down these problems.
969 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
970 enables the feature at boot time. By default, it is
971 disabled and the system will work mostly the same as a
972 kernel built without CONFIG_DEBUG_PAGEALLOC.
973 Note: to get most of debug_pagealloc error reports, it's
974 useful to also enable the page_owner functionality.
975 on: enable the feature
977 debugfs= [KNL] This parameter enables what is exposed to userspace
978 and debugfs internal clients.
979 Format: { on, no-mount, off }
980 on: All functions are enabled.
982 Filesystem is not registered but kernel clients can
983 access APIs and a crashkernel can be used to read
984 its content. There is nothing to mount.
985 off: Filesystem is not registered and clients
986 get a -EPERM as result when trying to register files
987 or directories within debugfs.
988 This is equivalent of the runtime functionality if
989 debugfs was not enabled in the kernel at all.
990 Default value is set in build-time with a kernel configuration.
992 debugpat [X86] Enable PAT debugging
995 [HW] The size of the default HugeTLB page. This is
996 the size represented by the legacy /proc/ hugepages
997 APIs. In addition, this is the default hugetlb size
998 used for shmget(), mmap() and mounting hugetlbfs
999 filesystems. If not specified, defaults to the
1000 architecture's default huge page size. Huge page
1001 sizes are architecture dependent. See also
1002 Documentation/admin-guide/mm/hugetlbpage.rst.
1005 deferred_probe_timeout=
1006 [KNL] Debugging option to set a timeout in seconds for
1007 deferred probe to give up waiting on dependencies to
1008 probe. Only specific dependencies (subsystems or
1009 drivers) that have opted in will be ignored. A timeout
1010 of 0 will timeout at the end of initcalls. If the time
1011 out hasn't expired, it'll be restarted by each
1012 successful driver registration. This option will also
1013 dump out devices still on the deferred probe list after
1016 delayacct [KNL] Enable per-task delay accounting
1018 dell_smm_hwmon.ignore_dmi=
1019 [HW] Continue probing hardware even if DMI data
1020 indicates that the driver is running on unsupported
1023 dell_smm_hwmon.force=
1024 [HW] Activate driver even if SMM BIOS signature does
1025 not match list of supported models and enable otherwise
1026 blacklisted features.
1028 dell_smm_hwmon.power_status=
1029 [HW] Report power status in /proc/i8k
1030 (disabled by default).
1032 dell_smm_hwmon.restricted=
1033 [HW] Allow controlling fans only if SYS_ADMIN
1036 dell_smm_hwmon.fan_mult=
1037 [HW] Factor to multiply fan speed with.
1039 dell_smm_hwmon.fan_max=
1040 [HW] Maximum configurable fan speed.
1043 Format: { on | off | def_only | inf_only | always }
1044 on: s390 zlib hardware support for compression on
1045 level 1 and decompression (default)
1046 off: No s390 zlib hardware support
1047 def_only: s390 zlib hardware support for deflate
1048 only (compression on level 1)
1049 inf_only: s390 zlib hardware support for inflate
1050 only (decompression)
1051 always: Same as 'on' but ignores the selected compression
1052 level always using hardware support (used for debugging)
1054 dhash_entries= [KNL]
1055 Set number of hash buckets for dentry cache.
1057 disable_1tb_segments [PPC]
1058 Disables the use of 1TB hash page table segments. This
1059 causes the kernel to fall back to 256MB segments which
1060 can be useful when debugging issues that require an SLB
1064 See Documentation/networking/ipv6.rst.
1067 Disable RADIX MMU mode on POWER9
1070 Disable TLBIE instruction. Currently does not work
1071 with KVM, with HASH MMU, or with coherent accelerators.
1073 disable_cpu_apicid= [X86,APIC,SMP]
1075 The number of initial APIC ID for the
1076 corresponding CPU to be disabled at boot,
1077 mostly used for the kdump 2nd kernel to
1078 disable BSP to wake up multiple CPUs without
1079 causing system reset or hang due to sending
1080 INIT from AP to BSP.
1082 disable_ddw [PPC/PSERIES]
1083 Disable Dynamic DMA Window support. Use this
1084 to workaround buggy firmware.
1086 disable_ipv6= [IPV6]
1087 See Documentation/networking/ipv6.rst.
1089 disable_mtrr_cleanup [X86]
1090 The kernel tries to adjust MTRR layout from continuous
1091 to discrete, to make X server driver able to add WB
1092 entry later. This parameter disables that.
1094 disable_mtrr_trim [X86, Intel and AMD only]
1095 By default the kernel will trim any uncacheable
1096 memory out of your available memory pool based on
1097 MTRR settings. This parameter disables that behavior,
1098 possibly causing your machine to run very slowly.
1100 disable_timer_pin_1 [X86]
1101 Disable PIN 1 of APIC timer
1102 Can be useful to work around chipset bugs.
1104 dis_ucode_ldr [X86] Disable the microcode loader.
1106 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
1107 this option disables the debugging code at boot.
1109 dma_debug_entries=<number>
1110 This option allows to tune the number of preallocated
1111 entries for DMA-API debugging code. One entry is
1112 required per DMA-API allocation. Use this if the
1113 DMA-API debugging code disables itself because the
1114 architectural default is too low.
1116 dma_debug_driver=<driver_name>
1117 With this option the DMA-API debugging driver
1118 filter feature can be enabled at boot time. Just
1119 pass the driver to filter for as the parameter.
1120 The filter can be disabled or changed to another
1121 driver later using sysfs.
1123 driver_async_probe= [KNL]
1124 List of driver names to be probed asynchronously. *
1125 matches with all driver names. If * is specified, the
1126 rest of the listed driver names are those that will NOT
1128 Format: <driver_name1>,<driver_name2>...
1130 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
1131 Broken monitors, graphic adapters, KVMs and EDIDless
1132 panels may send no or incorrect EDID data sets.
1133 This parameter allows to specify an EDID data sets
1134 in the /lib/firmware directory that are used instead.
1135 Generic built-in EDID data sets are used, if one of
1136 edid/1024x768.bin, edid/1280x1024.bin,
1137 edid/1680x1050.bin, or edid/1920x1080.bin is given
1138 and no file with the same name exists. Details and
1139 instructions how to build your own EDID data are
1140 available in Documentation/admin-guide/edid.rst. An EDID
1141 data set will only be used for a particular connector,
1142 if its name and a colon are prepended to the EDID
1143 name. Each connector may use a unique EDID data
1144 set by separating the files with a comma. An EDID
1145 data set with no connector name will be used for
1146 any connectors not explicitly specified.
1151 Format: {"off" | "known"}
1152 Control how the dt_cpu_ftrs device-tree binding is
1153 used for CPU feature discovery and setup (if it
1155 off: Do not use it, fall back to legacy cpu table.
1156 known: Do not pass through unknown features to guests
1157 or userspace, only those that the kernel is aware of.
1159 dump_apple_properties [X86]
1160 Dump name and content of EFI device properties on
1161 x86 Macs. Useful for driver authors to determine
1162 what data is available or for reverse-engineering.
1164 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
1165 <module>.dyndbg[="val"]
1166 Enable debug messages at boot time. See
1167 Documentation/admin-guide/dynamic-debug-howto.rst
1170 early_ioremap_debug [KNL]
1171 Enable debug messages in early_ioremap support. This
1172 is useful for tracking down temporary early mappings
1173 which are not unmapped.
1175 earlycon= [KNL] Output early console device and options.
1177 When used with no options, the early console is
1178 determined by stdout-path property in device tree's
1179 chosen node or the ACPI SPCR table if supported by
1182 cdns,<addr>[,options]
1183 Start an early, polled-mode console on a Cadence
1184 (xuartps) serial port at the specified address. Only
1185 supported option is baud rate. If baud rate is not
1186 specified, the serial port must already be setup and
1189 uart[8250],io,<addr>[,options[,uartclk]]
1190 uart[8250],mmio,<addr>[,options[,uartclk]]
1191 uart[8250],mmio32,<addr>[,options[,uartclk]]
1192 uart[8250],mmio32be,<addr>[,options[,uartclk]]
1193 uart[8250],0x<addr>[,options]
1194 Start an early, polled-mode console on the 8250/16550
1195 UART at the specified I/O port or MMIO address.
1196 MMIO inter-register address stride is either 8-bit
1197 (mmio) or 32-bit (mmio32 or mmio32be).
1198 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1199 to be equivalent to 'mmio'. 'options' are specified
1200 in the same format described for "console=ttyS<n>"; if
1201 unspecified, the h/w is not initialized. 'uartclk' is
1202 the uart clock frequency; if unspecified, it is set
1203 to 'BASE_BAUD' * 16.
1207 Start an early, polled-mode console on a pl011 serial
1208 port at the specified address. The pl011 serial port
1209 must already be setup and configured. Options are not
1210 yet supported. If 'mmio32' is specified, then only
1211 the driver will use only 32-bit accessors to read/write
1212 the device registers.
1215 Start an early console on a litex serial port at the
1216 specified address. The serial port must already be
1217 setup and configured. Options are not yet supported.
1220 Start an early, polled-mode console on a meson serial
1221 port at the specified address. The serial port must
1222 already be setup and configured. Options are not yet
1226 Start an early, polled-mode console on an msm serial
1227 port at the specified address. The serial port
1228 must already be setup and configured. Options are not
1231 msm_serial_dm,<addr>
1232 Start an early, polled-mode console on an msm serial
1233 dm port at the specified address. The serial port
1234 must already be setup and configured. Options are not
1238 Start an early, polled-mode console on a serial port
1239 of an Actions Semi SoC, such as S500 or S900, at the
1240 specified address. The serial port must already be
1241 setup and configured. Options are not yet supported.
1244 Start an early, polled-mode console on a serial port
1245 of an RDA Micro SoC, such as RDA8810PL, at the
1246 specified address. The serial port must already be
1247 setup and configured. Options are not yet supported.
1250 Use RISC-V SBI (Supervisor Binary Interface) for early
1253 smh Use ARM semihosting calls for early console.
1261 Use early console provided by serial driver available
1262 on Samsung SoCs, requires selecting proper type and
1263 a correct base address of the selected UART port. The
1264 serial port must already be setup and configured.
1265 Options are not yet supported.
1268 Start an early, polled-mode console on a lantiq serial
1269 (lqasc) port at the specified address. The serial port
1270 must already be setup and configured. Options are not
1275 Use early console provided by Freescale LP UART driver
1276 found on Freescale Vybrid and QorIQ LS1021A processors.
1277 A valid base address must be provided, and the serial
1278 port must already be setup and configured.
1282 Start an early, polled-mode, output-only console on the
1283 Freescale i.MX UART at the specified address. The UART
1284 must already be setup and configured.
1287 Start an early, polled-mode console on the
1288 Armada 3700 serial port at the specified
1289 address. The serial port must already be setup
1290 and configured. Options are not yet supported.
1293 Start an early, polled-mode console on a Qualcomm
1294 Generic Interface (GENI) based serial port at the
1295 specified address. The serial port must already be
1296 setup and configured. Options are not yet supported.
1299 Start an early, unaccelerated console on the EFI
1300 memory mapped framebuffer (if available). On cache
1301 coherent non-x86 systems that use system memory for
1302 the framebuffer, pass the 'ram' option so that it is
1303 mapped with the correct attributes.
1306 Use early console provided by Freescale LINFlexD UART
1307 serial driver for NXP S32V234 SoCs. A valid base
1308 address must be provided, and the serial port must
1309 already be setup and configured.
1311 earlyprintk= [X86,SH,ARM,M68k,S390]
1315 earlyprintk=serial[,ttySn[,baudrate]]
1316 earlyprintk=serial[,0x...[,baudrate]]
1317 earlyprintk=ttySn[,baudrate]
1318 earlyprintk=dbgp[debugController#]
1319 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1320 earlyprintk=xdbc[xhciController#]
1322 earlyprintk is useful when the kernel crashes before
1323 the normal console is initialized. It is not enabled by
1324 default because it has some cosmetic problems.
1326 Append ",keep" to not disable it when the real console
1329 Only one of vga, serial, or usb debug port can
1332 Currently only ttyS0 and ttyS1 may be specified by
1333 name. Other I/O ports may be explicitly specified
1334 on some architectures (x86 and arm at least) by
1335 replacing ttySn with an I/O port address, like this:
1336 earlyprintk=serial,0x1008,115200
1337 You can find the port for a given device in
1338 /proc/tty/driver/serial:
1339 2: uart:ST16650V2 port:00001008 irq:18 ...
1341 Interaction with the standard serial driver is not
1344 The VGA output is eventually overwritten by
1347 The xen option can only be used in Xen domains.
1349 The sclp output can only be used on s390.
1351 The optional "force" to "pciserial" enables use of a
1352 PCI device even when its classcode is not of the
1355 edac_report= [HW,EDAC] Control how to report EDAC event
1356 Format: {"on" | "off" | "force"}
1357 on: enable EDAC to report H/W event. May be overridden
1358 by other higher priority error reporting module.
1359 off: disable H/W event reporting through EDAC.
1360 force: enforce the use of EDAC to report H/W event.
1364 Format: {"off" | "on" | "skip[mbr]"}
1367 Format: { "debug", "disable_early_pci_dma",
1368 "nochunk", "noruntime", "nosoftreserve",
1369 "novamap", "no_disable_early_pci_dma" }
1370 debug: enable misc debug output.
1371 disable_early_pci_dma: disable the busmaster bit on all
1372 PCI bridges while in the EFI boot stub.
1373 nochunk: disable reading files in "chunks" in the EFI
1374 boot stub, as chunking can cause problems with some
1375 firmware implementations.
1376 noruntime : disable EFI runtime services support
1377 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1378 attribute may cause the kernel to reserve the
1379 memory range for a memory mapping driver to
1380 claim. Specify efi=nosoftreserve to disable this
1381 reservation and treat the memory by its base type
1382 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1383 novamap: do not call SetVirtualAddressMap().
1384 no_disable_early_pci_dma: Leave the busmaster bit set
1385 on all PCI bridges while in the EFI boot stub
1387 efi_no_storage_paranoia [EFI; X86]
1388 Using this parameter you can use more than 50% of
1389 your efi variable storage. Use this parameter only if
1390 you are really sure that your UEFI does sane gc and
1391 fulfills the spec otherwise your board may brick.
1393 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1394 Add arbitrary attribute to specific memory range by
1395 updating original EFI memory map.
1396 Region of memory which aa attribute is added to is
1399 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1400 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1401 attribute is added to range 0x100000000-0x180000000 and
1402 0x10a0000000-0x1120000000.
1404 If efi_fake_mem=8G@9G:0x40000 is specified, the
1405 EFI_MEMORY_SP(0x40000) attribute is added to
1406 range 0x240000000-0x43fffffff.
1408 Using this parameter you can do debugging of EFI memmap
1409 related features. For example, you can do debugging of
1410 Address Range Mirroring feature even if your box
1411 doesn't support it, or mark specific memory as
1414 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1415 that is to be dynamically loaded by Linux. If there are
1416 multiple variables with the same name but with different
1417 vendor GUIDs, all of them will be loaded. See
1418 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1421 eisa_irq_edge= [PARISC,HW]
1422 See header of drivers/parisc/eisa.c.
1424 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1427 This is designed to be used in conjunction with
1428 the boot argument: earlyprintk=vga
1430 This parameter works in place of the kgdboc parameter
1431 but can only be used if the backing tty is available
1432 very early in the boot process. For early debugging
1433 via a serial port see kgdboc_earlycon instead.
1436 See comment before function elanfreq_setup() in
1437 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1439 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1440 Specifies physical address of start of kernel core
1441 image elf header and optionally the size. Generally
1442 kexec loader will pass this option to capture kernel.
1443 See Documentation/admin-guide/kdump/kdump.rst for details.
1445 enable_mtrr_cleanup [X86]
1446 The kernel tries to adjust MTRR layout from continuous
1447 to discrete, to make X server driver able to add WB
1448 entry later. This parameter enables that.
1450 enable_timer_pin_1 [X86]
1451 Enable PIN 1 of APIC timer
1452 Can be useful to work around chipset bugs
1453 (in particular on some ATI chipsets).
1454 The kernel tries to set a reasonable default.
1456 enforcing= [SELINUX] Set initial enforcing status.
1458 See security/selinux/Kconfig help text.
1459 0 -- permissive (log only, no denials).
1460 1 -- enforcing (deny and log).
1462 Value can be changed at runtime via
1463 /sys/fs/selinux/enforce.
1466 Disable Error Record Serialization Table (ERST)
1469 ether= [HW,NET] Ethernet cards parameters
1470 This option is obsoleted by the "netdev=" option, which
1471 has equivalent usage. See its documentation for details.
1475 Permit 'security.evm' to be updated regardless of
1476 current integrity status.
1478 early_page_ext [KNL] Enforces page_ext initialization to earlier
1479 stages so cover more early boot allocations.
1480 Please note that as side effect some optimizations
1481 might be disabled to achieve that (e.g. parallelized
1482 memory initialization is disabled) so the boot process
1483 might take longer, especially on systems with a lot of
1484 memory. Available with CONFIG_PAGE_EXTENSION=y.
1489 fail_make_request=[KNL]
1490 General fault injection mechanism.
1491 Format: <interval>,<probability>,<space>,<times>
1492 See also Documentation/fault-injection/.
1495 Format: { initns | none }
1496 See Documentation/admin-guide/sysctl/net.rst for
1497 fb_tunnels_only_for_init_ns
1500 See Documentation/admin-guide/blockdev/floppy.rst.
1502 force_pal_cache_flush
1503 [IA-64] Avoid check_sal_cache_flush which may hang on
1504 buggy SAL_CACHE_FLUSH implementations. Using this
1505 parameter will force ia64_sal_cache_flush to call
1506 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1509 Forcefully enable Physical Address Extension (PAE).
1510 Many Pentium M systems disable PAE but may have a
1511 functionally usable PAE implementation.
1512 Warning: use of this parameter will taint the kernel
1513 and may cause unknown problems.
1516 [FTRACE] will set and start the specified tracer
1517 as early as possible in order to facilitate early
1520 ftrace_boot_snapshot
1521 [FTRACE] On boot up, a snapshot will be taken of the
1522 ftrace ring buffer that can be read at:
1523 /sys/kernel/tracing/snapshot.
1524 This is useful if you need tracing information from kernel
1525 boot up that is likely to be overridden by user space
1526 start up functionality.
1528 Optionally, the snapshot can also be defined for a tracing
1529 instance that was created by the trace_instance= command
1532 trace_instance=foo,sched_switch ftrace_boot_snapshot=foo
1534 The above will cause the "foo" tracing instance to trigger
1535 a snapshot at the end of boot up.
1537 ftrace_dump_on_oops[=orig_cpu]
1538 [FTRACE] will dump the trace buffers on oops.
1539 If no parameter is passed, ftrace will dump
1540 buffers of all CPUs, but if you pass orig_cpu, it will
1541 dump only the buffer of the CPU that triggered the
1544 ftrace_filter=[function-list]
1545 [FTRACE] Limit the functions traced by the function
1546 tracer at boot up. function-list is a comma-separated
1547 list of functions. This list can be changed at run
1548 time by the set_ftrace_filter file in the debugfs
1551 ftrace_notrace=[function-list]
1552 [FTRACE] Do not trace the functions specified in
1553 function-list. This list can be changed at run time
1554 by the set_ftrace_notrace file in the debugfs
1557 ftrace_graph_filter=[function-list]
1558 [FTRACE] Limit the top level callers functions traced
1559 by the function graph tracer at boot up.
1560 function-list is a comma-separated list of functions
1561 that can be changed at run time by the
1562 set_graph_function file in the debugfs tracing directory.
1564 ftrace_graph_notrace=[function-list]
1565 [FTRACE] Do not trace from the functions specified in
1566 function-list. This list is a comma-separated list of
1567 functions that can be changed at run time by the
1568 set_graph_notrace file in the debugfs tracing directory.
1570 ftrace_graph_max_depth=<uint>
1571 [FTRACE] Used with the function graph tracer. This is
1572 the max depth it will trace into a function. This value
1573 can be changed at run time by the max_graph_depth file
1574 in the tracefs tracing directory. default: 0 (no limit)
1576 fw_devlink= [KNL] Create device links between consumer and supplier
1577 devices by scanning the firmware to infer the
1578 consumer/supplier relationships. This feature is
1579 especially useful when drivers are loaded as modules as
1580 it ensures proper ordering of tasks like device probing
1581 (suppliers first, then consumers), supplier boot state
1582 clean up (only after all consumers have probed),
1583 suspend/resume & runtime PM (consumers first, then
1585 Format: { off | permissive | on | rpm }
1586 off -- Don't create device links from firmware info.
1587 permissive -- Create device links from firmware info
1588 but use it only for ordering boot state clean
1589 up (sync_state() calls).
1590 on -- Create device links from firmware info and use it
1591 to enforce probe and suspend/resume ordering.
1592 rpm -- Like "on", but also use to order runtime PM.
1594 fw_devlink.strict=<bool>
1595 [KNL] Treat all inferred dependencies as mandatory
1596 dependencies. This only applies for fw_devlink=on|rpm.
1599 fw_devlink.sync_state =
1600 [KNL] When all devices that could probe have finished
1601 probing, this parameter controls what to do with
1602 devices that haven't yet received their sync_state()
1604 Format: { strict | timeout }
1605 strict -- Default. Continue waiting on consumers to
1607 timeout -- Give up waiting on consumers and call
1608 sync_state() on any devices that haven't yet
1609 received their sync_state() calls after
1610 deferred_probe_timeout has expired or by
1611 late_initcall() if !CONFIG_MODULES.
1614 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1615 support via parallel port (up to 5 devices per port)
1616 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1617 See also Documentation/input/devices/joystick-parport.rst
1621 gart_fix_e820= [X86-64] disable the fix e820 for K8 GART
1625 gcov_persist= [GCOV] When non-zero (default), profiling data for
1626 kernel modules is saved and remains accessible via
1627 debugfs, even when the module is unloaded/reloaded.
1628 When zero, profiling data is discarded and associated
1629 debugfs files are removed at module unload time.
1631 goldfish [X86] Enable the goldfish android emulator platform.
1632 Don't use this when you are not running on the
1635 gpio-mockup.gpio_mockup_ranges
1636 [HW] Sets the ranges of gpiochip of for this device.
1637 Format: <start1>,<end1>,<start2>,<end2>...
1638 gpio-mockup.gpio_mockup_named_lines
1639 [HW] Let the driver know GPIO lines should be named.
1641 gpt [EFI] Forces disk with valid GPT signature but
1642 invalid Protective MBR to be treated as GPT. If the
1643 primary GPT is corrupted, it enables the backup/alternate
1644 GPT to be used instead.
1646 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1647 the "Enable 0" bit of the configuration register.
1650 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1651 the "Enable 0" bit of the configuration register.
1654 grcan.select= [HW] Select which physical interface to use.
1657 grcan.txsize= [HW] Sets the size of the tx buffer.
1658 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1660 grcan.rxsize= [HW] Sets the size of the rx buffer.
1661 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1665 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
1666 hardening is enabled for this boot. Hardened
1667 usercopy checking is used to protect the kernel
1668 from reading or writing beyond known memory
1669 allocation boundaries as a proactive defense
1670 against bounds-checking flaws in the kernel's
1671 copy_to_user()/copy_from_user() interface.
1672 on Perform hardened usercopy checks (default).
1673 off Disable hardened usercopy checks.
1675 hardlockup_all_cpu_backtrace=
1676 [KNL] Should the hard-lockup detector generate
1677 backtraces on all cpus.
1680 hashdist= [KNL,NUMA] Large hashes allocated during boot
1681 are distributed across NUMA nodes. Defaults on
1682 for 64-bit NUMA, off otherwise.
1683 Format: 0 | 1 (for off | on)
1685 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1687 hd= [EIDE] (E)IDE hard drive subsystem geometry
1688 Format: <cyl>,<head>,<sect>
1691 Disable Hardware Error Source Table (HEST) support;
1692 corresponding firmware-first mode error processing
1693 logic will be disabled.
1695 hibernate= [HIBERNATION]
1696 noresume Don't check if there's a hibernation image
1697 present during boot.
1698 nocompress Don't compress/decompress hibernation images.
1699 no Disable hibernation and resume.
1700 protect_image Turn on image protection during restoration
1701 (that will set all pages holding image data
1702 during restoration read-only).
1704 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1705 size of <nn>. This works even on boxes that have no
1706 highmem otherwise. This also works to reduce highmem
1707 size on bigger boxes.
1709 highres= [KNL] Enable/disable high resolution timer mode.
1710 Valid parameters: "on", "off"
1715 hostname= [KNL] Set the hostname (aka UTS nodename).
1717 This allows setting the system's hostname during early
1718 startup. This sets the name returned by gethostname.
1719 Using this parameter to set the hostname makes it
1720 possible to ensure the hostname is correctly set before
1721 any userspace processes run, avoiding the possibility
1722 that a process may call gethostname before the hostname
1723 has been explicitly set, resulting in the calling
1724 process getting an incorrect result. The string must
1725 not exceed the maximum allowed hostname length (usually
1726 64 characters) and will be truncated otherwise.
1728 hpet= [X86-32,HPET] option to control HPET usage
1729 Format: { enable (default) | disable | force |
1731 disable: disable HPET and use PIT instead
1732 force: allow force enabled of undocumented chips (ICH4,
1734 verbose: show contents of HPET registers during setup
1736 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1737 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1739 hugepages= [HW] Number of HugeTLB pages to allocate at boot.
1740 If this follows hugepagesz (below), it specifies
1741 the number of pages of hugepagesz to be allocated.
1742 If this is the first HugeTLB parameter on the command
1743 line, it specifies the number of pages to allocate for
1744 the default huge page size. If using node format, the
1745 number of pages to allocate per-node can be specified.
1746 See also Documentation/admin-guide/mm/hugetlbpage.rst.
1747 Format: <integer> or (node format)
1748 <node>:<integer>[,<node>:<integer>]
1751 [HW] The size of the HugeTLB pages. This is used in
1752 conjunction with hugepages (above) to allocate huge
1753 pages of a specific size at boot. The pair
1754 hugepagesz=X hugepages=Y can be specified once for
1755 each supported huge page size. Huge page sizes are
1756 architecture dependent. See also
1757 Documentation/admin-guide/mm/hugetlbpage.rst.
1760 hugetlb_cma= [HW,CMA] The size of a CMA area used for allocation
1761 of gigantic hugepages. Or using node format, the size
1762 of a CMA area per node can be specified.
1763 Format: nn[KMGTPE] or (node format)
1764 <node>:nn[KMGTPE][,<node>:nn[KMGTPE]]
1766 Reserve a CMA area of given size and allocate gigantic
1767 hugepages using the CMA allocator. If enabled, the
1768 boot-time allocation of gigantic hugepages is skipped.
1770 hugetlb_free_vmemmap=
1771 [KNL] Requires CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP
1773 Control if HugeTLB Vmemmap Optimization (HVO) is enabled.
1774 Allows heavy hugetlb users to free up some more
1775 memory (7 * PAGE_SIZE for each 2MB hugetlb page).
1776 Format: { on | off (default) }
1781 Built with CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP_DEFAULT_ON=y,
1784 Note that the vmemmap pages may be allocated from the added
1785 memory block itself when memory_hotplug.memmap_on_memory is
1786 enabled, those vmemmap pages cannot be optimized even if this
1787 feature is enabled. Other vmemmap pages not allocated from
1788 the added memory block itself do not be affected.
1791 [KNL] Should the hung task detector generate panics.
1794 A value of 1 instructs the kernel to panic when a
1795 hung task is detected. The default value is controlled
1796 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1797 option. The value selected by this boot parameter can
1798 be changed later by the kernel.hung_task_panic sysctl.
1800 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1801 terminal devices. Valid values: 0..8
1802 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1803 If specified, z/VM IUCV HVC accepts connections
1804 from listed z/VM user IDs only.
1806 hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations
1807 which allow the hypervisor to 'idle' the
1808 guest on lock contention.
1810 i2c_bus= [HW] Override the default board specific I2C bus speed
1811 or register an additional I2C bus that is not
1812 registered from board initialization code.
1816 i8042.debug [HW] Toggle i8042 debug mode
1817 i8042.unmask_kbd_data
1818 [HW] Enable printing of interrupt data from the KBD port
1819 (disabled by default, and as a pre-condition
1820 requires that i8042.debug=1 be enabled)
1821 i8042.direct [HW] Put keyboard port into non-translated mode
1822 i8042.dumbkbd [HW] Pretend that controller can only read data from
1823 keyboard and cannot control its state
1824 (Don't attempt to blink the leds)
1825 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1826 i8042.nokbd [HW] Don't check/create keyboard port
1827 i8042.noloop [HW] Disable the AUX Loopback command while probing
1829 i8042.nomux [HW] Don't check presence of an active multiplexing
1831 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1833 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1834 i8042.reset [HW] Reset the controller during init, cleanup and
1835 suspend-to-ram transitions, only during s2r
1836 transitions, or never reset
1837 Format: { 1 | Y | y | 0 | N | n }
1838 1, Y, y: always reset controller
1839 0, N, n: don't ever reset controller
1840 Default: only on s2r transitions on x86; most other
1841 architectures force reset to be always executed
1842 i8042.unlock [HW] Unlock (ignore) the keylock
1843 i8042.kbdreset [HW] Reset device connected to KBD port
1845 [HW] Allow deferred probing upon i8042 probe errors
1849 i915.invert_brightness=
1850 [DRM] Invert the sense of the variable that is used to
1851 set the brightness of the panel backlight. Normally a
1852 brightness value of 0 indicates backlight switched off,
1853 and the maximum of the brightness value sets the backlight
1854 to maximum brightness. If this parameter is set to 0
1855 (default) and the machine requires it, or this parameter
1856 is set to 1, a brightness value of 0 sets the backlight
1857 to maximum brightness, and the maximum of the brightness
1858 value switches the backlight off.
1859 -1 -- never invert brightness
1860 0 -- machine default
1861 1 -- force brightness inversion
1864 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1868 Format: idle=poll, idle=halt, idle=nomwait
1869 Poll forces a polling idle loop that can slightly
1870 improve the performance of waking up a idle CPU, but
1871 will use a lot of power and make the system run hot.
1873 idle=halt: Halt is forced to be used for CPU idle.
1874 In such case C2/C3 won't be used again.
1875 idle=nomwait: Disable mwait for CPU C-states
1879 Allow force disabling of Shared Virtual Memory (SVA)
1880 support for the idxd driver. By default it is set to
1883 idxd.tc_override= [HW]
1885 Allow override of default traffic class configuration
1886 for the device. By default it is set to false (0).
1888 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1889 Format: { strict | legacy | 2008 | relaxed }
1892 Choose which programs will be accepted for execution
1893 based on the IEEE 754 NaN encoding(s) supported by
1894 the FPU and the NaN encoding requested with the value
1895 of an ELF file header flag individually set by each
1896 binary. Hardware implementations are permitted to
1897 support either or both of the legacy and the 2008 NaN
1900 Available settings are as follows:
1901 strict accept binaries that request a NaN encoding
1902 supported by the FPU
1903 legacy only accept legacy-NaN binaries, if supported
1905 2008 only accept 2008-NaN binaries, if supported
1907 relaxed accept any binaries regardless of whether
1908 supported by the FPU
1910 The FPU emulator is always able to support both NaN
1911 encodings, so if no FPU hardware is present or it has
1912 been disabled with 'nofpu', then the settings of
1913 'legacy' and '2008' strap the emulator accordingly,
1914 'relaxed' straps the emulator for both legacy-NaN and
1915 2008-NaN, whereas 'strict' enables legacy-NaN only on
1916 legacy processors and both NaN encodings on MIPS32 or
1919 The setting for ABS.fmt/NEG.fmt instruction execution
1920 mode generally follows that for the NaN encoding,
1921 except where unsupported by hardware.
1923 ignore_loglevel [KNL]
1924 Ignore loglevel setting - this will print /all/
1925 kernel messages to the console. Useful for debugging.
1926 We also add it as printk module parameter, so users
1927 could change it dynamically, usually by
1928 /sys/module/printk/parameters/ignore_loglevel.
1931 Ignore RLIMIT_DATA setting for data mappings,
1932 print warning at first misuse. Can be changed via
1933 /sys/module/kernel/parameters/ignore_rlimit_data.
1935 ihash_entries= [KNL]
1936 Set number of hash buckets for inode cache.
1938 ima_appraise= [IMA] appraise integrity measurements
1939 Format: { "off" | "enforce" | "fix" | "log" }
1942 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
1943 The builtin appraise policy appraises all files
1946 ima_canonical_fmt [IMA]
1947 Use the canonical format for the binary runtime
1948 measurements, instead of host native format.
1951 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1955 The list of supported hash algorithms is defined
1956 in crypto/hash_info.h.
1959 The builtin policies to load during IMA setup.
1960 Format: "tcb | appraise_tcb | secure_boot |
1961 fail_securely | critical_data"
1963 The "tcb" policy measures all programs exec'd, files
1964 mmap'd for exec, and all files opened with the read
1965 mode bit set by either the effective uid (euid=0) or
1968 The "appraise_tcb" policy appraises the integrity of
1969 all files owned by root.
1971 The "secure_boot" policy appraises the integrity
1972 of files (eg. kexec kernel image, kernel modules,
1973 firmware, policy, etc) based on file signatures.
1975 The "fail_securely" policy forces file signature
1976 verification failure also on privileged mounted
1977 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
1980 The "critical_data" policy measures kernel integrity
1983 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
1984 Load a policy which meets the needs of the Trusted
1985 Computing Base. This means IMA will measure all
1986 programs exec'd, files mmap'd for exec, and all files
1987 opened for read by uid=0.
1990 Select one of defined IMA measurements template formats.
1991 Formats: { "ima" | "ima-ng" | "ima-ngv2" | "ima-sig" |
1996 [IMA] Define a custom template format.
1997 Format: { "field1|...|fieldN" }
1999 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
2000 Format: <min_file_size>
2001 Set the minimal file size for using asynchronous hash.
2002 If left unspecified, ahash usage is disabled.
2004 ahash performance varies for different data sizes on
2005 different crypto accelerators. This option can be used
2006 to achieve the best performance for a particular HW.
2008 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
2010 Set hashing buffer size. Default: 4k.
2012 ahash performance varies for different chunk sizes on
2013 different crypto accelerators. This option can be used
2014 to achieve best performance for particular HW.
2018 Run specified binary instead of /sbin/init as init
2021 initcall_debug [KNL] Trace initcalls as they are executed. Useful
2022 for working out where the kernel is dying during
2025 initcall_blacklist= [KNL] Do not execute a comma-separated list of
2026 initcall functions. Useful for debugging built-in
2027 modules and initcalls.
2029 initramfs_async= [KNL]
2032 This parameter controls whether the initramfs
2033 image is unpacked asynchronously, concurrently
2034 with devices being probed and
2035 initialized. This should normally just work,
2036 but as a debugging aid, one can get the
2037 historical behaviour of the initramfs
2038 unpacking being completed before device_ and
2041 initrd= [BOOT] Specify the location of the initial ramdisk
2043 initrdmem= [KNL] Specify a physical address and size from which to
2044 load the initrd. If an initrd is compiled in or
2045 specified in the bootparams, it takes priority over this
2047 Format: ss[KMG],nn[KMG]
2050 init_on_alloc= [MM] Fill newly allocated pages and heap objects with
2053 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
2055 init_on_free= [MM] Fill freed pages and heap objects with zeroes.
2057 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
2059 init_pkru= [X86] Specify the default memory protection keys rights
2060 register contents for all processes. 0x55555554 by
2061 default (disallow access to all but pkey 0). Can
2062 override in debugfs after boot.
2064 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
2067 int_pln_enable [X86] Enable power limit notification interrupt
2069 integrity_audit=[IMA]
2070 Format: { "0" | "1" }
2071 0 -- basic integrity auditing messages. (Default)
2072 1 -- additional integrity auditing messages.
2074 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
2076 Enable intel iommu driver.
2078 Disable intel iommu driver.
2079 igfx_off [Default Off]
2080 By default, gfx is mapped as normal device. If a gfx
2081 device has a dedicated DMAR unit, the DMAR unit is
2082 bypassed by not enabling DMAR with this option. In
2083 this case, gfx device will use physical address for
2085 strict [Default Off]
2086 Deprecated, equivalent to iommu.strict=1.
2087 sp_off [Default Off]
2088 By default, super page will be supported if Intel IOMMU
2089 has the capability. With this option, super page will
2092 Enable the Intel IOMMU scalable mode if the hardware
2093 advertises that it has support for the scalable mode
2096 Disallow use of the Intel IOMMU scalable mode.
2097 tboot_noforce [Default Off]
2098 Do not force the Intel IOMMU enabled under tboot.
2099 By default, tboot will force Intel IOMMU on, which
2100 could harm performance of some high-throughput
2101 devices like 40GBit network cards, even if identity
2103 Note that using this option lowers the security
2104 provided by tboot because it makes the system
2105 vulnerable to DMA attacks.
2107 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
2108 0 disables intel_idle and fall back on acpi_idle.
2109 1 to 9 specify maximum depth of C-state.
2113 Do not enable intel_pstate as the default
2114 scaling driver for the supported processors
2116 Use intel_pstate as a scaling driver, but configure it
2117 to work with generic cpufreq governors (instead of
2118 enabling its internal governor). This mode cannot be
2119 used along with the hardware-managed P-states (HWP)
2122 Enable intel_pstate on systems that prohibit it by default
2123 in favor of acpi-cpufreq. Forcing the intel_pstate driver
2124 instead of acpi-cpufreq may disable platform features, such
2125 as thermal controls and power capping, that rely on ACPI
2126 P-States information being indicated to OSPM and therefore
2127 should be used with caution. This option does not work with
2128 processors that aren't supported by the intel_pstate driver
2129 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
2131 Do not enable hardware P state control (HWP)
2134 Only load intel_pstate on systems which support
2135 hardware P state control (HWP) if available.
2137 Enforce ACPI _PPC performance limits. If the Fixed ACPI
2138 Description Table, specifies preferred power management
2139 profile as "Enterprise Server" or "Performance Server",
2140 then this feature is turned on by default.
2142 Allow per-logical-CPU P-State performance control limits using
2143 cpufreq sysfs interface
2145 intremap= [X86-64, Intel-IOMMU]
2146 on enable Interrupt Remapping (default)
2147 off disable Interrupt Remapping
2148 nosid disable Source ID checking
2150 BIOS x2APIC opt-out request will be ignored
2151 nopost disable Interrupt Posting
2153 iomem= Disable strict checking of access to MMIO memory
2154 strict regions from userspace.
2169 nobypass [PPC/POWERNV]
2170 Disable IOMMU bypass, using IOMMU for PCI devices.
2172 iommu.forcedac= [ARM64, X86] Control IOVA allocation for PCI devices.
2173 Format: { "0" | "1" }
2174 0 - Try to allocate a 32-bit DMA address first, before
2175 falling back to the full range if needed.
2176 1 - Allocate directly from the full usable range,
2177 forcing Dual Address Cycle for PCI cards supporting
2178 greater than 32-bit addressing.
2180 iommu.strict= [ARM64, X86] Configure TLB invalidation behaviour
2181 Format: { "0" | "1" }
2183 Request that DMA unmap operations use deferred
2184 invalidation of hardware TLBs, for increased
2185 throughput at the cost of reduced device isolation.
2186 Will fall back to strict mode if not supported by
2187 the relevant IOMMU driver.
2189 DMA unmap operations invalidate IOMMU hardware TLBs
2191 unset - Use value of CONFIG_IOMMU_DEFAULT_DMA_{LAZY,STRICT}.
2192 Note: on x86, strict mode specified via one of the
2193 legacy driver-specific options takes precedence.
2196 [ARM64, X86] Configure DMA to bypass the IOMMU by default.
2197 Format: { "0" | "1" }
2198 0 - Use IOMMU translation for DMA.
2199 1 - Bypass the IOMMU for DMA.
2200 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
2202 io7= [HW] IO7 for Marvel-based Alpha systems
2203 See comment before marvel_specify_io7 in
2204 arch/alpha/kernel/core_marvel.c.
2206 io_delay= [X86] I/O delay method
2208 Standard port 0x80 based delay
2210 Alternate port 0xed based delay (needed on some systems)
2212 Simple two microseconds delay
2217 See Documentation/admin-guide/nfs/nfsroot.rst.
2219 ipcmni_extend [KNL] Extend the maximum number of unique System V
2220 IPC identifiers from 32,768 to 16,777,216.
2222 irqaffinity= [SMP] Set the default irq affinity mask
2223 The argument is a cpu list, as described above.
2225 irqchip.gicv2_force_probe=
2228 Force the kernel to look for the second 4kB page
2229 of a GICv2 controller even if the memory range
2230 exposed by the device tree is too small.
2232 irqchip.gicv3_nolpi=
2234 Force the kernel to ignore the availability of
2235 LPIs (and by consequence ITSs). Intended for system
2236 that use the kernel as a bootloader, and thus want
2237 to let secondary kernels in charge of setting up
2240 irqchip.gicv3_pseudo_nmi= [ARM64]
2241 Enables support for pseudo-NMIs in the kernel. This
2242 requires the kernel to be built with
2243 CONFIG_ARM64_PSEUDO_NMI.
2246 When an interrupt is not handled search all handlers
2247 for it. Intended to get systems with badly broken
2251 When an interrupt is not handled search all handlers
2252 for it. Also check all handlers each timer
2253 interrupt. Intended to get systems with badly broken
2257 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
2259 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
2260 [Deprecated - use cpusets instead]
2261 Format: [flag-list,]<cpu-list>
2263 Specify one or more CPUs to isolate from disturbances
2264 specified in the flag list (default: domain):
2267 Disable the tick when a single task runs.
2269 A residual 1Hz tick is offloaded to workqueues, which you
2270 need to affine to housekeeping through the global
2271 workqueue's affinity configured via the
2272 /sys/devices/virtual/workqueue/cpumask sysfs file, or
2273 by using the 'domain' flag described below.
2275 NOTE: by default the global workqueue runs on all CPUs,
2276 so to protect individual CPUs the 'cpumask' file has to
2277 be configured manually after bootup.
2280 Isolate from the general SMP balancing and scheduling
2281 algorithms. Note that performing domain isolation this way
2282 is irreversible: it's not possible to bring back a CPU to
2283 the domains once isolated through isolcpus. It's strongly
2284 advised to use cpusets instead to disable scheduler load
2285 balancing through the "cpuset.sched_load_balance" file.
2286 It offers a much more flexible interface where CPUs can
2287 move in and out of an isolated set anytime.
2289 You can move a process onto or off an "isolated" CPU via
2290 the CPU affinity syscalls or cpuset.
2291 <cpu number> begins at 0 and the maximum value is
2292 "number of CPUs in system - 1".
2296 Isolate from being targeted by managed interrupts
2297 which have an interrupt mask containing isolated
2298 CPUs. The affinity of managed interrupts is
2299 handled by the kernel and cannot be changed via
2300 the /proc/irq/* interfaces.
2302 This isolation is best effort and only effective
2303 if the automatically assigned interrupt mask of a
2304 device queue contains isolated and housekeeping
2305 CPUs. If housekeeping CPUs are online then such
2306 interrupts are directed to the housekeeping CPU
2307 so that IO submitted on the housekeeping CPU
2308 cannot disturb the isolated CPU.
2310 If a queue's affinity mask contains only isolated
2311 CPUs then this parameter has no effect on the
2312 interrupt routing decision, though interrupts are
2313 only delivered when tasks running on those
2314 isolated CPUs submit IO. IO submitted on
2315 housekeeping CPUs has no influence on those
2318 The format of <cpu-list> is described above.
2322 ivrs_ioapic [HW,X86-64]
2323 Provide an override to the IOAPIC-ID<->DEVICE-ID
2324 mapping provided in the IVRS ACPI table.
2325 By default, PCI segment is 0, and can be omitted.
2327 For example, to map IOAPIC-ID decimal 10 to
2328 PCI segment 0x1 and PCI device 00:14.0,
2329 write the parameter as:
2330 ivrs_ioapic=10@0001:00:14.0
2333 * To map IOAPIC-ID decimal 10 to PCI device 00:14.0
2334 write the parameter as:
2335 ivrs_ioapic[10]=00:14.0
2336 * To map IOAPIC-ID decimal 10 to PCI segment 0x1 and
2337 PCI device 00:14.0 write the parameter as:
2338 ivrs_ioapic[10]=0001:00:14.0
2340 ivrs_hpet [HW,X86-64]
2341 Provide an override to the HPET-ID<->DEVICE-ID
2342 mapping provided in the IVRS ACPI table.
2343 By default, PCI segment is 0, and can be omitted.
2345 For example, to map HPET-ID decimal 10 to
2346 PCI segment 0x1 and PCI device 00:14.0,
2347 write the parameter as:
2348 ivrs_hpet=10@0001:00:14.0
2351 * To map HPET-ID decimal 0 to PCI device 00:14.0
2352 write the parameter as:
2353 ivrs_hpet[0]=00:14.0
2354 * To map HPET-ID decimal 10 to PCI segment 0x1 and
2355 PCI device 00:14.0 write the parameter as:
2356 ivrs_ioapic[10]=0001:00:14.0
2358 ivrs_acpihid [HW,X86-64]
2359 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2360 mapping provided in the IVRS ACPI table.
2361 By default, PCI segment is 0, and can be omitted.
2363 For example, to map UART-HID:UID AMD0020:0 to
2364 PCI segment 0x1 and PCI device ID 00:14.5,
2365 write the parameter as:
2366 ivrs_acpihid=AMD0020:0@0001:00:14.5
2369 * To map UART-HID:UID AMD0020:0 to PCI segment is 0,
2370 PCI device ID 00:14.5, write the parameter as:
2371 ivrs_acpihid[00:14.5]=AMD0020:0
2372 * To map UART-HID:UID AMD0020:0 to PCI segment 0x1 and
2373 PCI device ID 00:14.5, write the parameter as:
2374 ivrs_acpihid[0001:00:14.5]=AMD0020:0
2376 js= [HW,JOY] Analog joystick
2377 See Documentation/input/joydev/joystick.rst.
2380 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2381 report on every invalid memory access. Without this
2382 parameter KASAN will print report only for the first
2386 Do not unregister boot console at start. This is only
2387 useful for debugging when something happens in the window
2388 between unregistering the boot console and initializing
2393 kernelcore= [KNL,X86,IA-64,PPC]
2394 Format: nn[KMGTPE] | nn% | "mirror"
2395 This parameter specifies the amount of memory usable by
2396 the kernel for non-movable allocations. The requested
2397 amount is spread evenly throughout all nodes in the
2398 system as ZONE_NORMAL. The remaining memory is used for
2399 movable memory in its own zone, ZONE_MOVABLE. In the
2400 event, a node is too small to have both ZONE_NORMAL and
2401 ZONE_MOVABLE, kernelcore memory will take priority and
2402 other nodes will have a larger ZONE_MOVABLE.
2404 ZONE_MOVABLE is used for the allocation of pages that
2405 may be reclaimed or moved by the page migration
2406 subsystem. Note that allocations like PTEs-from-HighMem
2407 still use the HighMem zone if it exists, and the Normal
2408 zone if it does not.
2410 It is possible to specify the exact amount of memory in
2411 the form of "nn[KMGTPE]", a percentage of total system
2412 memory in the form of "nn%", or "mirror". If "mirror"
2413 option is specified, mirrored (reliable) memory is used
2414 for non-movable allocations and remaining memory is used
2415 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2416 are exclusive, so you cannot specify multiple forms.
2418 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
2419 Format: <Controller#>[,poll interval]
2420 The controller # is the number of the ehci usb debug
2421 port as it is probed via PCI. The poll interval is
2422 optional and is the number seconds in between
2423 each poll cycle to the debug port in case you need
2424 the functionality for interrupting the kernel with
2425 gdb or control-c on the dbgp connection. When
2426 not using this parameter you use sysrq-g to break into
2427 the kernel debugger.
2429 kgdboc= [KGDB,HW] kgdb over consoles.
2430 Requires a tty driver that supports console polling,
2431 or a supported polling keyboard driver (non-usb).
2432 Serial only format: <serial_device>[,baud]
2433 keyboard only format: kbd
2434 keyboard and serial format: kbd,<serial_device>[,baud]
2435 Optional Kernel mode setting:
2436 kms, kbd format: kms,kbd
2437 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2439 kgdboc_earlycon= [KGDB,HW]
2440 If the boot console provides the ability to read
2441 characters and can work in polling mode, you can use
2442 this parameter to tell kgdb to use it as a backend
2443 until the normal console is registered. Intended to
2444 be used together with the kgdboc parameter which
2445 specifies the normal console to transition to.
2447 The name of the early console should be specified
2448 as the value of this parameter. Note that the name of
2449 the early console might be different than the tty
2450 name passed to kgdboc. It's OK to leave the value
2451 blank and the first boot console that implements
2452 read() will be picked.
2454 kgdbwait [KGDB] Stop kernel execution and enter the
2455 kernel debugger at the earliest opportunity.
2457 kmac= [MIPS] Korina ethernet MAC address.
2458 Configure the RouterBoard 532 series on-chip
2459 Ethernet adapter MAC address.
2461 kmemleak= [KNL] Boot-time kmemleak enable/disable
2462 Valid arguments: on, off
2464 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2467 kprobe_event=[probe-list]
2468 [FTRACE] Add kprobe events and enable at boot time.
2469 The probe-list is a semicolon delimited list of probe
2470 definitions. Each definition is same as kprobe_events
2471 interface, but the parameters are comma delimited.
2472 For example, to add a kprobe event on vfs_read with
2473 arg1 and arg2, add to the command line;
2475 kprobe_event=p,vfs_read,$arg1,$arg2
2477 See also Documentation/trace/kprobetrace.rst "Kernel
2478 Boot Parameter" section.
2480 kpti= [ARM64] Control page table isolation of user
2481 and kernel address spaces.
2482 Default: enabled on cores which need mitigation.
2486 kunit.enable= [KUNIT] Enable executing KUnit tests. Requires
2487 CONFIG_KUNIT to be set to be fully enabled. The
2488 default value can be overridden via
2489 KUNIT_DEFAULT_ENABLED.
2490 Default is 1 (enabled)
2492 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2493 Default is 0 (don't ignore, but inject #GP)
2495 kvm.eager_page_split=
2496 [KVM,X86] Controls whether or not KVM will try to
2497 proactively split all huge pages during dirty logging.
2498 Eager page splitting reduces interruptions to vCPU
2499 execution by eliminating the write-protection faults
2500 and MMU lock contention that would otherwise be
2501 required to split huge pages lazily.
2503 VM workloads that rarely perform writes or that write
2504 only to a small region of VM memory may benefit from
2505 disabling eager page splitting to allow huge pages to
2506 still be used for reads.
2508 The behavior of eager page splitting depends on whether
2509 KVM_DIRTY_LOG_INITIALLY_SET is enabled or disabled. If
2510 disabled, all huge pages in a memslot will be eagerly
2511 split when dirty logging is enabled on that memslot. If
2512 enabled, eager page splitting will be performed during
2513 the KVM_CLEAR_DIRTY ioctl, and only for the pages being
2516 Eager page splitting is only supported when kvm.tdp_mmu=Y.
2520 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2521 Default is false (don't support).
2524 [KVM] Controls the software workaround for the
2525 X86_BUG_ITLB_MULTIHIT bug.
2526 force : Always deploy workaround.
2527 off : Never deploy workaround.
2528 auto : Deploy workaround based on the presence of
2529 X86_BUG_ITLB_MULTIHIT.
2533 If the software workaround is enabled for the host,
2534 guests do need not to enable it for nested guests.
2536 kvm.nx_huge_pages_recovery_ratio=
2537 [KVM] Controls how many 4KiB pages are periodically zapped
2538 back to huge pages. 0 disables the recovery, otherwise if
2539 the value is N KVM will zap 1/Nth of the 4KiB pages every
2540 period (see below). The default is 60.
2542 kvm.nx_huge_pages_recovery_period_ms=
2543 [KVM] Controls the time period at which KVM zaps 4KiB pages
2544 back to huge pages. If the value is a non-zero N, KVM will
2545 zap a portion (see ratio above) of the pages every N msecs.
2546 If the value is 0 (the default), KVM will pick a period based
2547 on the ratio, such that a page is zapped after 1 hour on average.
2549 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM.
2550 Default is 1 (enabled)
2552 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU)
2554 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode.
2557 [KVM,ARM] Select one of KVM/arm64's modes of operation.
2559 none: Forcefully disable KVM.
2561 nvhe: Standard nVHE-based mode, without support for
2564 protected: nVHE-based mode with support for guests whose
2565 state is kept private from the host.
2567 nested: VHE-based mode with support for nested
2568 virtualization. Requires at least ARMv8.3
2571 Defaults to VHE/nVHE based on hardware support. Setting
2572 mode to "protected" will disable kexec and hibernation
2573 for the host. "nested" is experimental and should be
2574 used with extreme caution.
2576 kvm-arm.vgic_v3_group0_trap=
2577 [KVM,ARM] Trap guest accesses to GICv3 group-0
2580 kvm-arm.vgic_v3_group1_trap=
2581 [KVM,ARM] Trap guest accesses to GICv3 group-1
2584 kvm-arm.vgic_v3_common_trap=
2585 [KVM,ARM] Trap guest accesses to GICv3 common
2588 kvm-arm.vgic_v4_enable=
2589 [KVM,ARM] Allow use of GICv4 for direct injection of
2592 kvm_cma_resv_ratio=n [PPC]
2593 Reserves given percentage from system memory area for
2594 contiguous memory allocation for KVM hash pagetable
2596 By default it reserves 5% of total system memory.
2600 kvm-intel.ept= [KVM,Intel] Disable extended page tables
2601 (virtualized MMU) support on capable Intel chips.
2602 Default is 1 (enabled)
2604 kvm-intel.emulate_invalid_guest_state=
2605 [KVM,Intel] Disable emulation of invalid guest state.
2606 Ignored if kvm-intel.enable_unrestricted_guest=1, as
2607 guest state is never invalid for unrestricted guests.
2608 This param doesn't apply to nested guests (L2), as KVM
2609 never emulates invalid L2 guest state.
2610 Default is 1 (enabled)
2612 kvm-intel.flexpriority=
2613 [KVM,Intel] Disable FlexPriority feature (TPR shadow).
2614 Default is 1 (enabled)
2617 [KVM,Intel] Enable VMX nesting (nVMX).
2618 Default is 0 (disabled)
2620 kvm-intel.unrestricted_guest=
2621 [KVM,Intel] Disable unrestricted guest feature
2622 (virtualized real and unpaged mode) on capable
2623 Intel chips. Default is 1 (enabled)
2625 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2628 Valid arguments: never, cond, always
2630 always: L1D cache flush on every VMENTER.
2631 cond: Flush L1D on VMENTER only when the code between
2632 VMEXIT and VMENTER can leak host memory.
2633 never: Disables the mitigation
2635 Default is cond (do L1 cache flush in specific instances)
2637 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification
2638 feature (tagged TLBs) on capable Intel chips.
2639 Default is 1 (enabled)
2641 l1d_flush= [X86,INTEL]
2642 Control mitigation for L1D based snooping vulnerability.
2644 Certain CPUs are vulnerable to an exploit against CPU
2645 internal buffers which can forward information to a
2646 disclosure gadget under certain conditions.
2648 In vulnerable processors, the speculatively
2649 forwarded data can be used in a cache side channel
2650 attack, to access data to which the attacker does
2651 not have direct access.
2653 This parameter controls the mitigation. The
2656 on - enable the interface for the mitigation
2658 l1tf= [X86] Control mitigation of the L1TF vulnerability on
2661 The kernel PTE inversion protection is unconditionally
2662 enabled and cannot be disabled.
2665 Provides all available mitigations for the
2666 L1TF vulnerability. Disables SMT and
2667 enables all mitigations in the
2668 hypervisors, i.e. unconditional L1D flush.
2670 SMT control and L1D flush control via the
2671 sysfs interface is still possible after
2672 boot. Hypervisors will issue a warning
2673 when the first VM is started in a
2674 potentially insecure configuration,
2675 i.e. SMT enabled or L1D flush disabled.
2678 Same as 'full', but disables SMT and L1D
2679 flush runtime control. Implies the
2680 'nosmt=force' command line option.
2681 (i.e. sysfs control of SMT is disabled.)
2684 Leaves SMT enabled and enables the default
2685 hypervisor mitigation, i.e. conditional
2688 SMT control and L1D flush control via the
2689 sysfs interface is still possible after
2690 boot. Hypervisors will issue a warning
2691 when the first VM is started in a
2692 potentially insecure configuration,
2693 i.e. SMT enabled or L1D flush disabled.
2697 Disables SMT and enables the default
2698 hypervisor mitigation.
2700 SMT control and L1D flush control via the
2701 sysfs interface is still possible after
2702 boot. Hypervisors will issue a warning
2703 when the first VM is started in a
2704 potentially insecure configuration,
2705 i.e. SMT enabled or L1D flush disabled.
2708 Same as 'flush', but hypervisors will not
2709 warn when a VM is started in a potentially
2710 insecure configuration.
2713 Disables hypervisor mitigations and doesn't
2715 It also drops the swap size and available
2716 RAM limit restriction on both hypervisor and
2721 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2727 lapic [X86-32,APIC] Enable the local APIC even if BIOS
2730 lapic= [X86,APIC] Do not use TSC deadline
2731 value for LAPIC timer one-shot implementation. Default
2732 back to the programmable timer unit in the LAPIC.
2733 Format: notscdeadline
2735 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
2738 libata.dma= [LIBATA] DMA control
2739 libata.dma=0 Disable all PATA and SATA DMA
2740 libata.dma=1 PATA and SATA Disk DMA only
2741 libata.dma=2 ATAPI (CDROM) DMA only
2742 libata.dma=4 Compact Flash DMA only
2743 Combinations also work, so libata.dma=3 enables DMA
2744 for disks and CDROMs, but not CFs.
2746 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2747 libata.ignore_hpa=0 keep BIOS limits (default)
2748 libata.ignore_hpa=1 ignore limits, using full disk
2750 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2754 libata.force= [LIBATA] Force configurations. The format is a comma-
2755 separated list of "[ID:]VAL" where ID is PORT[.DEVICE].
2756 PORT and DEVICE are decimal numbers matching port, link
2757 or device. Basically, it matches the ATA ID string
2758 printed on console by libata. If the whole ID part is
2759 omitted, the last PORT and DEVICE values are used. If
2760 ID hasn't been specified yet, the configuration applies
2761 to all ports, links and devices.
2763 If only DEVICE is omitted, the parameter applies to
2764 the port and all links and devices behind it. DEVICE
2765 number of 0 either selects the first device or the
2766 first fan-out link behind PMP device. It does not
2767 select the host link. DEVICE number of 15 selects the
2768 host link and device attached to it.
2770 The VAL specifies the configuration to force. As long
2771 as there is no ambiguity, shortcut notation is allowed.
2772 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2773 The following configurations can be forced.
2775 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2776 Any ID with matching PORT is used.
2778 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2780 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2781 udma[/][16,25,33,44,66,100,133] notation is also
2784 * nohrst, nosrst, norst: suppress hard, soft and both
2787 * rstonce: only attempt one reset during hot-unplug
2790 * [no]dbdelay: Enable or disable the extra 200ms delay
2791 before debouncing a link PHY and device presence
2794 * [no]ncq: Turn on or off NCQ.
2796 * [no]ncqtrim: Enable or disable queued DSM TRIM.
2798 * [no]ncqati: Enable or disable NCQ trim on ATI chipset.
2800 * [no]trim: Enable or disable (unqueued) TRIM.
2802 * trim_zero: Indicate that TRIM command zeroes data.
2804 * max_trim_128m: Set 128M maximum trim size limit.
2806 * [no]dma: Turn on or off DMA transfers.
2808 * atapi_dmadir: Enable ATAPI DMADIR bridge support.
2810 * atapi_mod16_dma: Enable the use of ATAPI DMA for
2811 commands that are not a multiple of 16 bytes.
2813 * [no]dmalog: Enable or disable the use of the
2814 READ LOG DMA EXT command to access logs.
2816 * [no]iddevlog: Enable or disable access to the
2817 identify device data log.
2819 * [no]logdir: Enable or disable access to the general
2820 purpose log directory.
2822 * max_sec_128: Set transfer size limit to 128 sectors.
2824 * max_sec_1024: Set or clear transfer size limit to
2827 * max_sec_lba48: Set or clear transfer size limit to
2830 * [no]lpm: Enable or disable link power management.
2832 * [no]setxfer: Indicate if transfer speed mode setting
2835 * [no]fua: Disable or enable FUA (Force Unit Access)
2836 support for devices supporting this feature.
2838 * dump_id: Dump IDENTIFY data.
2840 * disable: Disable this device.
2842 If there are multiple matching configurations changing
2843 the same attribute, the last one is used.
2845 load_ramdisk= [RAM] [Deprecated]
2847 lockd.nlm_grace_period=P [NFS] Assign grace period.
2850 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2853 lockd.nlm_timeout=T [NFS] Assign timeout value.
2856 lockd.nlm_udpport=M [NFS] Assign UDP port.
2859 lockdown= [SECURITY]
2860 { integrity | confidentiality }
2861 Enable the kernel lockdown feature. If set to
2862 integrity, kernel features that allow userland to
2863 modify the running kernel are disabled. If set to
2864 confidentiality, kernel features that allow userland
2865 to extract confidential information from the kernel
2868 locktorture.nreaders_stress= [KNL]
2869 Set the number of locking read-acquisition kthreads.
2870 Defaults to being automatically set based on the
2871 number of online CPUs.
2873 locktorture.nwriters_stress= [KNL]
2874 Set the number of locking write-acquisition kthreads.
2876 locktorture.onoff_holdoff= [KNL]
2877 Set time (s) after boot for CPU-hotplug testing.
2879 locktorture.onoff_interval= [KNL]
2880 Set time (s) between CPU-hotplug operations, or
2881 zero to disable CPU-hotplug testing.
2883 locktorture.shuffle_interval= [KNL]
2884 Set task-shuffle interval (jiffies). Shuffling
2885 tasks allows some CPUs to go into dyntick-idle
2886 mode during the locktorture test.
2888 locktorture.shutdown_secs= [KNL]
2889 Set time (s) after boot system shutdown. This
2890 is useful for hands-off automated testing.
2892 locktorture.stat_interval= [KNL]
2893 Time (s) between statistics printk()s.
2895 locktorture.stutter= [KNL]
2896 Time (s) to stutter testing, for example,
2897 specifying five seconds causes the test to run for
2898 five seconds, wait for five seconds, and so on.
2899 This tests the locking primitive's ability to
2900 transition abruptly to and from idle.
2902 locktorture.torture_type= [KNL]
2903 Specify the locking implementation to test.
2905 locktorture.verbose= [KNL]
2906 Enable additional printk() statements.
2908 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
2911 loglevel= All Kernel Messages with a loglevel smaller than the
2912 console loglevel will be printed to the console. It can
2913 also be changed with klogd or other programs. The
2914 loglevels are defined as follows:
2916 0 (KERN_EMERG) system is unusable
2917 1 (KERN_ALERT) action must be taken immediately
2918 2 (KERN_CRIT) critical conditions
2919 3 (KERN_ERR) error conditions
2920 4 (KERN_WARNING) warning conditions
2921 5 (KERN_NOTICE) normal but significant condition
2922 6 (KERN_INFO) informational
2923 7 (KERN_DEBUG) debug-level messages
2925 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
2926 in bytes. n must be a power of two and greater
2927 than the minimal size. The minimal size is defined
2928 by LOG_BUF_SHIFT kernel config parameter. There is
2929 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2930 that allows to increase the default size depending on
2931 the number of CPUs. See init/Kconfig for more details.
2933 logo.nologo [FB] Disables display of the built-in Linux logo.
2934 This may be used to provide more screen space for
2935 kernel log messages and is useful when debugging
2936 kernel boot problems.
2938 lp=0 [LP] Specify parallel ports to use, e.g,
2939 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2940 lp=reset first parallel port). 'lp=0' disables the
2941 lp=auto printer driver. 'lp=reset' (which can be
2942 specified in addition to the ports) causes
2943 attached printers to be reset. Using
2944 lp=port1,port2,... specifies the parallel ports
2945 to associate lp devices with, starting with
2946 lp0. A port specification may be 'none' to skip
2947 that lp device, or a parport name such as
2948 'parport0'. Specifying 'lp=auto' instead of a
2949 port specification list means that device IDs
2950 from each port should be examined, to see if
2951 an IEEE 1284-compliant printer is attached; if
2952 so, the driver will manage that printer.
2953 See also header of drivers/char/lp.c.
2956 Sets loops_per_jiffy to given constant, thus avoiding
2957 time-consuming boot-time autodetection (up to 250 ms per
2958 CPU). 0 enables autodetection (default). To determine
2959 the correct value for your kernel, boot with normal
2960 autodetection and see what value is printed. Note that
2961 on SMP systems the preset will be applied to all CPUs,
2962 which is likely to cause problems if your CPUs need
2963 significantly divergent settings. An incorrect value
2964 will cause delays in the kernel to be wrong, leading to
2965 unpredictable I/O errors and other breakage. Although
2966 unlikely, in the extreme case this might damage your
2970 Format: <io>,<irq>,<dma>
2972 lsm.debug [SECURITY] Enable LSM initialization debugging output.
2975 [SECURITY] Choose order of LSM initialization. This
2976 overrides CONFIG_LSM, and the "security=" parameter.
2978 machvec= [IA-64] Force the use of a particular machine-vector
2979 (machvec) in a generic kernel.
2980 Example: machvec=hpzx1
2982 machtype= [Loongson] Share the same kernel image file between
2983 different yeeloong laptops.
2984 Example: machtype=lemote-yeeloong-2f-7inch
2986 max_addr=nn[KMG] [KNL,BOOT,IA-64] All physical memory greater
2987 than or equal to this physical address is ignored.
2989 maxcpus= [SMP] Maximum number of processors that an SMP kernel
2990 will bring up during bootup. maxcpus=n : n >= 0 limits
2991 the kernel to bring up 'n' processors. Surely after
2992 bootup you can bring up the other plugged cpu by executing
2993 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
2994 only takes effect during system bootup.
2995 While n=0 is a special case, it is equivalent to "nosmp",
2996 which also disables the IO APIC.
2998 max_loop= [LOOP] The number of loop block devices that get
2999 (loop.max_loop) unconditionally pre-created at init time. The default
3000 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
3001 of statically allocating a predefined number, loop
3002 devices can be requested on-demand with the
3003 /dev/loop-control interface.
3005 mce [X86-32] Machine Check Exception
3007 mce=option [X86-64] See Documentation/arch/x86/x86_64/boot-options.rst
3009 md= [HW] RAID subsystems devices and level
3010 See Documentation/admin-guide/md.rst.
3013 Format: <first>,<last>
3014 Specifies range of consoles to be captured by the MDA.
3017 Control mitigation for the Micro-architectural Data
3018 Sampling (MDS) vulnerability.
3020 Certain CPUs are vulnerable to an exploit against CPU
3021 internal buffers which can forward information to a
3022 disclosure gadget under certain conditions.
3024 In vulnerable processors, the speculatively
3025 forwarded data can be used in a cache side channel
3026 attack, to access data to which the attacker does
3027 not have direct access.
3029 This parameter controls the MDS mitigation. The
3032 full - Enable MDS mitigation on vulnerable CPUs
3033 full,nosmt - Enable MDS mitigation and disable
3034 SMT on vulnerable CPUs
3035 off - Unconditionally disable MDS mitigation
3037 On TAA-affected machines, mds=off can be prevented by
3038 an active TAA mitigation as both vulnerabilities are
3039 mitigated with the same mechanism so in order to disable
3040 this mitigation, you need to specify tsx_async_abort=off
3043 Not specifying this option is equivalent to
3046 For details see: Documentation/admin-guide/hw-vuln/mds.rst
3048 mem=nn[KMG] [HEXAGON] Set the memory size.
3049 Must be specified, otherwise memory size will be 0.
3051 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
3052 Amount of memory to be used in cases as follows:
3055 2 when the kernel is not able to see the whole system memory;
3056 3 memory that lies after 'mem=' boundary is excluded from
3057 the hypervisor, then assigned to KVM guests.
3058 4 to limit the memory available for kdump kernel.
3060 [ARC,MICROBLAZE] - the limit applies only to low memory,
3061 high memory is not affected.
3063 [ARM64] - only limits memory covered by the linear
3064 mapping. The NOMAP regions are not affected.
3066 [X86] Work as limiting max address. Use together
3067 with memmap= to avoid physical address space collisions.
3068 Without memmap= PCI devices could be placed at addresses
3069 belonging to unused RAM.
3071 Note that this only takes effects during boot time since
3072 in above case 3, memory may need be hot added after boot
3073 if system memory of hypervisor is not sufficient.
3076 [ARM,MIPS] - override the memory layout reported by
3078 Define a memory region of size nn[KMG] starting at
3080 Multiple different regions can be specified with
3081 multiple mem= parameters on the command line.
3083 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
3086 memblock=debug [KNL] Enable memblock debug messages.
3089 [KNL,SH] Allow user to override the default size for
3090 per-device physically contiguous DMA buffers.
3092 memhp_default_state=online/offline
3093 [KNL] Set the initial state for the memory hotplug
3094 onlining policy. If not specified, the default value is
3095 set according to the
3096 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
3098 See Documentation/admin-guide/mm/memory-hotplug.rst.
3100 memmap=exactmap [KNL,X86] Enable setting of an exact
3101 E820 memory map, as specified by the user.
3102 Such memmap=exactmap lines can be constructed based on
3103 BIOS output or other requirements. See the memmap=nn@ss
3106 memmap=nn[KMG]@ss[KMG]
3107 [KNL, X86, MIPS, XTENSA] Force usage of a specific region of memory.
3108 Region of memory to be used is from ss to ss+nn.
3109 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
3110 which limits max address to nn[KMG].
3111 Multiple different regions can be specified,
3114 memmap=100M@2G,100M#3G,1G!1024G
3116 memmap=nn[KMG]#ss[KMG]
3117 [KNL,ACPI] Mark specific memory as ACPI data.
3118 Region of memory to be marked is from ss to ss+nn.
3120 memmap=nn[KMG]$ss[KMG]
3121 [KNL,ACPI] Mark specific memory as reserved.
3122 Region of memory to be reserved is from ss to ss+nn.
3123 Example: Exclude memory from 0x18690000-0x1869ffff
3124 memmap=64K$0x18690000
3126 memmap=0x10000$0x18690000
3127 Some bootloaders may need an escape character before '$',
3128 like Grub2, otherwise '$' and the following number
3131 memmap=nn[KMG]!ss[KMG]
3132 [KNL,X86] Mark specific memory as protected.
3133 Region of memory to be used, from ss to ss+nn.
3134 The memory region may be marked as e820 type 12 (0xc)
3135 and is NVDIMM or ADR memory.
3137 memmap=<size>%<offset>-<oldtype>+<newtype>
3138 [KNL,ACPI] Convert memory within the specified region
3139 from <oldtype> to <newtype>. If "-<oldtype>" is left
3140 out, the whole region will be marked as <newtype>,
3141 even if previously unavailable. If "+<newtype>" is left
3142 out, matching memory will be removed. Types are
3143 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
3144 3 = ACPI, 12 = PRAM.
3146 memory_corruption_check=0/1 [X86]
3147 Some BIOSes seem to corrupt the first 64k of
3148 memory when doing things like suspend/resume.
3149 Setting this option will scan the memory
3150 looking for corruption. Enabling this will
3151 both detect corruption and prevent the kernel
3152 from using the memory being corrupted.
3153 However, its intended as a diagnostic tool; if
3154 repeatable BIOS-originated corruption always
3155 affects the same memory, you can use memmap=
3156 to prevent the kernel from using that memory.
3158 memory_corruption_check_size=size [X86]
3159 By default it checks for corruption in the low
3160 64k, making this memory unavailable for normal
3161 use. Use this parameter to scan for
3162 corruption in more or less memory.
3164 memory_corruption_check_period=seconds [X86]
3165 By default it checks for corruption every 60
3166 seconds. Use this parameter to check at some
3167 other rate. 0 disables periodic checking.
3169 memory_hotplug.memmap_on_memory
3170 [KNL,X86,ARM] Boolean flag to enable this feature.
3171 Format: {on | off (default)}
3172 When enabled, runtime hotplugged memory will
3173 allocate its internal metadata (struct pages,
3174 those vmemmap pages cannot be optimized even
3175 if hugetlb_free_vmemmap is enabled) from the
3176 hotadded memory which will allow to hotadd a
3177 lot of memory without requiring additional
3179 This feature is disabled by default because it
3180 has some implication on large (e.g. GB)
3181 allocations in some configurations (e.g. small
3183 The state of the flag can be read in
3184 /sys/module/memory_hotplug/parameters/memmap_on_memory.
3185 Note that even when enabled, there are a few cases where
3186 the feature is not effective.
3188 memtest= [KNL,X86,ARM,M68K,PPC,RISCV] Enable memtest
3190 default : 0 <disable>
3191 Specifies the number of memtest passes to be
3192 performed. Each pass selects another test
3193 pattern from a given set of patterns. Memtest
3194 fills the memory with this pattern, validates
3195 memory contents and reserves bad memory
3196 regions that are detected.
3198 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
3199 Valid arguments: on, off
3200 Default (depends on kernel configuration option):
3201 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
3202 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
3203 mem_encrypt=on: Activate SME
3204 mem_encrypt=off: Do not activate SME
3206 Refer to Documentation/virt/kvm/x86/amd-memory-encryption.rst
3207 for details on when memory encryption can be activated.
3209 mem_sleep_default= [SUSPEND] Default system suspend mode:
3210 s2idle - Suspend-To-Idle
3211 shallow - Power-On Suspend or equivalent (if supported)
3212 deep - Suspend-To-RAM or equivalent (if supported)
3213 See Documentation/admin-guide/pm/sleep-states.rst.
3215 mfgpt_irq= [IA-32] Specify the IRQ to use for the
3216 Multi-Function General Purpose Timers on AMD Geode
3219 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
3220 the BIOS has incorrectly applied a workaround. TinyBIOS
3221 version 0.98 is known to be affected, 0.99 fixes the
3222 problem by letting the user disable the workaround.
3226 min_addr=nn[KMG] [KNL,BOOT,IA-64] All physical memory below this
3227 physical address is ignored.
3229 mini2440= [ARM,HW,KNL]
3230 Format:[0..2][b][c][t]
3232 MINI2440 configuration specification:
3233 0 - The attached screen is the 3.5" TFT
3234 1 - The attached screen is the 7" TFT
3235 2 - The VGA Shield is attached (1024x768)
3236 Leaving out the screen size parameter will not load
3237 the TFT driver, and the framebuffer will be left
3239 b - Enable backlight. The TFT backlight pin will be
3240 linked to the kernel VESA blanking code and a GPIO
3241 LED. This parameter is not necessary when using the
3243 c - Enable the s3c camera interface.
3244 t - Reserved for enabling touchscreen support. The
3245 touchscreen support is not enabled in the mainstream
3246 kernel as of 2.6.30, a preliminary port can be found
3247 in the "bleeding edge" mini2440 support kernel at
3248 https://repo.or.cz/w/linux-2.6/mini2440.git
3251 [X86,PPC,S390,ARM64] Control optional mitigations for
3252 CPU vulnerabilities. This is a set of curated,
3253 arch-independent options, each of which is an
3254 aggregation of existing arch-specific options.
3257 Disable all optional CPU mitigations. This
3258 improves system performance, but it may also
3259 expose users to several CPU vulnerabilities.
3260 Equivalent to: nopti [X86,PPC]
3261 if nokaslr then kpti=0 [ARM64]
3262 nospectre_v1 [X86,PPC]
3264 nospectre_v2 [X86,PPC,S390,ARM64]
3265 spectre_v2_user=off [X86]
3266 spec_store_bypass_disable=off [X86,PPC]
3267 ssbd=force-off [ARM64]
3268 nospectre_bhb [ARM64]
3271 tsx_async_abort=off [X86]
3272 kvm.nx_huge_pages=off [X86]
3273 srbds=off [X86,INTEL]
3274 no_entry_flush [PPC]
3275 no_uaccess_flush [PPC]
3276 mmio_stale_data=off [X86]
3280 This does not have any effect on
3281 kvm.nx_huge_pages when
3282 kvm.nx_huge_pages=force.
3285 Mitigate all CPU vulnerabilities, but leave SMT
3286 enabled, even if it's vulnerable. This is for
3287 users who don't want to be surprised by SMT
3288 getting disabled across kernel upgrades, or who
3289 have other ways of avoiding SMT-based attacks.
3290 Equivalent to: (default behavior)
3293 Mitigate all CPU vulnerabilities, disabling SMT
3294 if needed. This is for users who always want to
3295 be fully mitigated, even if it means losing SMT.
3296 Equivalent to: l1tf=flush,nosmt [X86]
3297 mds=full,nosmt [X86]
3298 tsx_async_abort=full,nosmt [X86]
3299 mmio_stale_data=full,nosmt [X86]
3300 retbleed=auto,nosmt [X86]
3303 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
3304 parameter allows control of the logging verbosity for
3305 the additional memory initialisation checks. A value
3306 of 0 disables mminit logging and a level of 4 will
3307 log everything. Information is printed at KERN_DEBUG
3308 so loglevel=8 may also need to be specified.
3311 [X86,INTEL] Control mitigation for the Processor
3312 MMIO Stale Data vulnerabilities.
3314 Processor MMIO Stale Data is a class of
3315 vulnerabilities that may expose data after an MMIO
3316 operation. Exposed data could originate or end in
3317 the same CPU buffers as affected by MDS and TAA.
3318 Therefore, similar to MDS and TAA, the mitigation
3319 is to clear the affected CPU buffers.
3321 This parameter controls the mitigation. The
3324 full - Enable mitigation on vulnerable CPUs
3326 full,nosmt - Enable mitigation and disable SMT on
3329 off - Unconditionally disable mitigation
3331 On MDS or TAA affected machines,
3332 mmio_stale_data=off can be prevented by an active
3333 MDS or TAA mitigation as these vulnerabilities are
3334 mitigated with the same mechanism so in order to
3335 disable this mitigation, you need to specify
3336 mds=off and tsx_async_abort=off too.
3338 Not specifying this option is equivalent to
3339 mmio_stale_data=full.
3342 Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst
3344 <module>.async_probe[=<bool>] [KNL]
3345 If no <bool> value is specified or if the value
3346 specified is not a valid <bool>, enable asynchronous
3347 probe on this module. Otherwise, enable/disable
3348 asynchronous probe on this module as indicated by the
3349 <bool> value. See also: module.async_probe
3351 module.async_probe=<bool>
3352 [KNL] When set to true, modules will use async probing
3353 by default. To enable/disable async probing for a
3354 specific module, use the module specific control that
3355 is documented under <module>.async_probe. When both
3356 module.async_probe and <module>.async_probe are
3357 specified, <module>.async_probe takes precedence for
3358 the specific module.
3360 module.enable_dups_trace
3361 [KNL] When CONFIG_MODULE_DEBUG_AUTOLOAD_DUPS is set,
3362 this means that duplicate request_module() calls will
3363 trigger a WARN_ON() instead of a pr_warn(). Note that
3364 if MODULE_DEBUG_AUTOLOAD_DUPS_TRACE is set, WARN_ON()s
3365 will always be issued and this option does nothing.
3367 [KNL] When CONFIG_MODULE_SIG is set, this means that
3368 modules without (valid) signatures will fail to load.
3369 Note that if CONFIG_MODULE_SIG_FORCE is set, that
3370 is always true, so this option does nothing.
3372 module_blacklist= [KNL] Do not load a comma-separated list of
3373 modules. Useful for debugging problem modules.
3376 [MOUSE] Maximum time between finger touching and
3377 leaving touchpad surface for touch to be considered
3378 a tap and be reported as a left button click (for
3379 touchpads working in absolute mode only).
3381 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
3382 reporting absolute coordinates, such as tablets
3383 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
3384 reporting absolute coordinates, such as tablets
3386 movablecore= [KNL,X86,IA-64,PPC]
3387 Format: nn[KMGTPE] | nn%
3388 This parameter is the complement to kernelcore=, it
3389 specifies the amount of memory used for migratable
3390 allocations. If both kernelcore and movablecore is
3391 specified, then kernelcore will be at *least* the
3392 specified value but may be more. If movablecore on its
3393 own is specified, the administrator must be careful
3394 that the amount of memory usable for all allocations
3397 movable_node [KNL] Boot-time switch to make hotplugable memory
3398 NUMA nodes to be movable. This means that the memory
3399 of such nodes will be usable only for movable
3400 allocations which rules out almost all kernel
3401 allocations. Use with caution!
3403 MTD_Partition= [MTD]
3404 Format: <name>,<region-number>,<size>,<offset>
3406 MTD_Region= [MTD] Format:
3407 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
3410 See drivers/mtd/parsers/cmdlinepart.c
3413 ARM/S3C2412 JIVE boot control
3415 See arch/arm/mach-s3c/mach-jive.c
3417 mtouchusb.raw_coordinates=
3418 [HW] Make the MicroTouch USB driver use raw coordinates
3419 ('y', default) or cooked coordinates ('n')
3421 mtrr_chunk_size=nn[KMG] [X86]
3422 used for mtrr cleanup. It is largest continuous chunk
3423 that could hold holes aka. UC entries.
3425 mtrr_gran_size=nn[KMG] [X86]
3426 Used for mtrr cleanup. It is granularity of mtrr block.
3428 Large value could prevent small alignment from
3431 mtrr_spare_reg_nr=n [X86]
3433 Range: 0,7 : spare reg number
3435 Used for mtrr cleanup. It is spare mtrr entries number.
3436 Set to 2 or more if your graphical card needs more.
3438 multitce=off [PPC] This parameter disables the use of the pSeries
3439 firmware feature for updating multiple TCE entries
3442 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
3444 netdev= [NET] Network devices parameters
3445 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
3446 Note that mem_start is often overloaded to mean
3447 something different and driver-specific.
3448 This usage is only documented in each driver source
3451 netpoll.carrier_timeout=
3452 [NET] Specifies amount of time (in seconds) that
3453 netpoll should wait for a carrier. By default netpoll
3457 [NETFILTER] Enable connection tracking flow accounting
3458 0 to disable accounting
3459 1 to enable accounting
3463 [NFS] sets the pathname to the program which is used
3464 to update the NFS client cache entries.
3466 nfs.cache_getent_timeout=
3467 [NFS] sets the timeout after which an attempt to
3468 update a cache entry is deemed to have failed.
3470 nfs.callback_nr_threads=
3471 [NFSv4] set the total number of threads that the
3472 NFS client will assign to service NFSv4 callback
3475 nfs.callback_tcpport=
3476 [NFS] set the TCP port on which the NFSv4 callback
3477 channel should listen.
3480 [NFS] enable 64-bit inode numbers.
3481 If zero, the NFS client will fake up a 32-bit inode
3482 number for the readdir() and stat() syscalls instead
3483 of returning the full 64-bit number.
3484 The default is to return 64-bit inode numbers.
3486 nfs.idmap_cache_timeout=
3487 [NFS] set the maximum lifetime for idmapper cache
3490 nfs.max_session_cb_slots=
3491 [NFSv4.1] Sets the maximum number of session
3492 slots the client will assign to the callback
3493 channel. This determines the maximum number of
3494 callbacks the client will process in parallel for
3495 a particular server.
3497 nfs.max_session_slots=
3498 [NFSv4.1] Sets the maximum number of session slots
3499 the client will attempt to negotiate with the server.
3500 This limits the number of simultaneous RPC requests
3501 that the client can send to the NFSv4.1 server.
3502 Note that there is little point in setting this
3503 value higher than the max_tcp_slot_table_limit.
3505 nfs.nfs4_disable_idmapping=
3506 [NFSv4] When set to the default of '1', this option
3507 ensures that both the RPC level authentication
3508 scheme and the NFS level operations agree to use
3509 numeric uids/gids if the mount is using the
3510 'sec=sys' security flavour. In effect it is
3511 disabling idmapping, which can make migration from
3512 legacy NFSv2/v3 systems to NFSv4 easier.
3513 Servers that do not support this mode of operation
3514 will be autodetected by the client, and it will fall
3515 back to using the idmapper.
3516 To turn off this behaviour, set the value to '0'.
3519 [NFS4] Specify an additional fixed unique ident-
3520 ification string that NFSv4 clients can insert into
3521 their nfs_client_id4 string. This is typically a
3522 UUID that is generated at system install time.
3524 nfs.recover_lost_locks=
3525 [NFSv4] Attempt to recover locks that were lost due
3526 to a lease timeout on the server. Please note that
3527 doing this risks data corruption, since there are
3528 no guarantees that the file will remain unchanged
3529 after the locks are lost.
3530 If you want to enable the kernel legacy behaviour of
3531 attempting to recover these locks, then set this
3533 The default parameter value of '0' causes the kernel
3534 not to attempt recovery of lost locks.
3536 nfs.send_implementation_id=
3537 [NFSv4.1] Send client implementation identification
3538 information in exchange_id requests.
3539 If zero, no implementation identification information
3541 The default is to send the implementation identification
3544 nfs4.layoutstats_timer=
3545 [NFSv4.2] Change the rate at which the kernel sends
3546 layoutstats to the pNFS metadata server.
3548 Setting this to value to 0 causes the kernel to use
3549 whatever value is the default set by the layout
3550 driver. A non-zero value sets the minimum interval
3551 in seconds between layoutstats transmissions.
3553 nfsd.inter_copy_offload_enable=
3554 [NFSv4.2] When set to 1, the server will support
3555 server-to-server copies for which this server is
3556 the destination of the copy.
3558 nfsd.nfs4_disable_idmapping=
3559 [NFSv4] When set to the default of '1', the NFSv4
3560 server will return only numeric uids and gids to
3561 clients using auth_sys, and will accept numeric uids
3562 and gids from such clients. This is intended to ease
3563 migration from NFSv2/v3.
3565 nfsd.nfsd4_ssc_umount_timeout=
3566 [NFSv4.2] When used as the destination of a
3567 server-to-server copy, knfsd temporarily mounts
3568 the source server. It caches the mount in case
3569 it will be needed again, and discards it if not
3570 used for the number of milliseconds specified by
3573 nfsaddrs= [NFS] Deprecated. Use ip= instead.
3574 See Documentation/admin-guide/nfs/nfsroot.rst.
3576 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
3577 See Documentation/admin-guide/nfs/nfsroot.rst.
3579 nfsrootdebug [NFS] enable nfsroot debugging messages.
3580 See Documentation/admin-guide/nfs/nfsroot.rst.
3582 nmi_backtrace.backtrace_idle [KNL]
3583 Dump stacks even of idle CPUs in response to an
3584 NMI stack-backtrace request.
3586 nmi_debug= [KNL,SH] Specify one or more actions to take
3587 when a NMI is triggered.
3588 Format: [state][,regs][,debounce][,die]
3590 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
3591 Format: [panic,][nopanic,][num]
3593 0 - turn hardlockup detector in nmi_watchdog off
3594 1 - turn hardlockup detector in nmi_watchdog on
3595 When panic is specified, panic when an NMI watchdog
3596 timeout occurs (or 'nopanic' to not panic on an NMI
3597 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
3598 To disable both hard and soft lockup detectors,
3599 please see 'nowatchdog'.
3600 This is useful when you use a panic=... timeout and
3601 need the box quickly up again.
3603 These settings can be accessed at runtime via
3604 the nmi_watchdog and hardlockup_panic sysctls.
3606 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3607 emulation library even if a 387 maths coprocessor
3610 no4lvl [RISCV] Disable 4-level and 5-level paging modes. Forces
3611 kernel to use 3-level paging instead.
3613 no5lvl [X86-64,RISCV] Disable 5-level paging mode. Forces
3614 kernel to use 4-level paging instead.
3616 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
3617 caches in the slab allocator. Saves per-node memory,
3618 but will impact performance.
3622 noaltinstr [S390] Disables alternative instructions patching
3623 (CPU alternatives feature).
3625 noapic [SMP,APIC] Tells the kernel to not make use of any
3626 IOAPICs that may be present in the system.
3628 noautogroup Disable scheduler automatic task group creation.
3633 [HW] Never suspend the console
3634 Disable suspending of consoles during suspend and
3635 hibernate operations. Once disabled, debugging
3636 messages can reach various consoles while the rest
3637 of the system is being put to sleep (ie, while
3638 debugging driver suspend/resume hooks). This may
3639 not work reliably with all consoles, but is known
3640 to work with serial and VGA consoles.
3641 To facilitate more flexible debugging, we also add
3642 console_suspend, a printk module parameter to control
3643 it. Users could use console_suspend (usually
3644 /sys/module/printk/parameters/console_suspend) to
3645 turn on/off it dynamically.
3648 [KNL] Disable object debugging
3650 nodsp [SH] Disable hardware DSP at boot time.
3652 noefi Disable EFI runtime services support.
3654 no_entry_flush [PPC] Don't flush the L1-D cache when entering the kernel.
3659 This affects only 32-bit executables.
3660 noexec32=on: enable non-executable mappings (default)
3661 read doesn't imply executable mappings
3662 noexec32=off: disable non-executable mappings
3663 read implies executable mappings
3665 no_file_caps Tells the kernel not to honor file capabilities. The
3666 only way then for a file to be executed with privilege
3667 is to be setuid root or executed by root.
3669 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3671 nofsgsbase [X86] Disables FSGSBASE instructions.
3673 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3674 register save and restore. The kernel will only save
3675 legacy floating-point registers on task switch.
3677 nohalt [IA-64] Tells the kernel not to use the power saving
3678 function PAL_HALT_LIGHT when idle. This increases
3679 power-consumption. On the positive side, it reduces
3680 interrupt wake-up latency, which may improve performance
3681 in certain environments such as networked servers or
3685 Force pointers printed to the console or buffers to be
3686 unhashed. By default, when a pointer is printed via %p
3687 format string, that pointer is "hashed", i.e. obscured
3688 by hashing the pointer value. This is a security feature
3689 that hides actual kernel addresses from unprivileged
3690 users, but it also makes debugging the kernel more
3691 difficult since unequal pointers can no longer be
3692 compared. However, if this command-line option is
3693 specified, then all normal pointers will have their true
3694 value printed. This option should only be specified when
3695 debugging the kernel. Please do not use on production
3698 nohibernate [HIBERNATION] Disable hibernation and resume.
3700 nohlt [ARM,ARM64,MICROBLAZE,SH] Forces the kernel to busy wait
3701 in do_idle() and not use the arch_cpu_idle()
3702 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
3703 to be effective. This is useful on platforms where the
3704 sleep(SH) or wfi(ARM,ARM64) instructions do not work
3705 correctly or when doing power measurements to evaluate
3706 the impact of the sleep instructions. This is also
3707 useful when using JTAG debugger.
3709 nohugeiomap [KNL,X86,PPC,ARM64] Disable kernel huge I/O mappings.
3711 nohugevmalloc [KNL,X86,PPC,ARM64] Disable kernel huge vmalloc mappings.
3713 nohz= [KNL] Boottime enable/disable dynamic ticks
3714 Valid arguments: on, off
3717 nohz_full= [KNL,BOOT,SMP,ISOL]
3718 The argument is a cpu list, as described above.
3719 In kernels built with CONFIG_NO_HZ_FULL=y, set
3720 the specified list of CPUs whose tick will be stopped
3721 whenever possible. The boot CPU will be forced outside
3722 the range to maintain the timekeeping. Any CPUs
3723 in this list will have their RCU callbacks offloaded,
3724 just as if they had also been called out in the
3725 rcu_nocbs= boot parameter.
3727 Note that this argument takes precedence over
3728 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option.
3730 noinitrd [RAM] Tells the kernel not to load any configured
3733 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
3735 [Deprecated - use intremap=off]
3739 noinvpcid [X86] Disable the INVPCID cpu feature.
3741 noiotrap [SH] Disables trapped I/O port accesses.
3743 noirqdebug [X86-32] Disables the code which attempts to detect and
3744 disable unhandled interrupt sources.
3746 noisapnp [ISAPNP] Disables ISA PnP code.
3748 nojitter [IA-64] Disables jitter checking for ITC timers.
3751 When CONFIG_RANDOMIZE_BASE is set, this disables
3752 kernel and module base offset ASLR (Address Space
3753 Layout Randomization).
3755 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
3758 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
3760 nolapic [X86-32,APIC] Do not enable or use the local APIC.
3762 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
3764 nomca [IA-64] Disable machine check abort handling
3766 nomce [X86-32] Disable Machine Check Exception
3768 nomfgpt [X86-32] Disable Multi-Function General Purpose
3769 Timer usage (for AMD Geode machines).
3771 nomodeset Disable kernel modesetting. Most systems' firmware
3772 sets up a display mode and provides framebuffer memory
3773 for output. With nomodeset, DRM and fbdev drivers will
3774 not load if they could possibly displace the pre-
3775 initialized output. Only the system framebuffer will
3776 be available for use. The respective drivers will not
3777 perform display-mode changes or accelerated rendering.
3779 Useful as error fallback, or for testing and debugging.
3781 nomodule Disable module load
3783 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3784 shutdown the other cpus. Instead use the REBOOT_VECTOR
3787 nopat [X86] Disable PAT (page attribute table extension of
3788 pagetables) support.
3790 nopcid [X86-64] Disable the PCID cpu feature.
3792 nopku [X86] Disable Memory Protection Keys CPU feature found
3796 Equivalent to pti=off
3798 nopv= [X86,XEN,KVM,HYPER_V,VMWARE]
3799 Disables the PV optimizations forcing the guest to run
3800 as generic guest with no PV drivers. Currently support
3801 XEN HVM, KVM, HYPER_V and VMWARE guest.
3803 nopvspin [X86,XEN,KVM]
3804 Disables the qspinlock slow path using PV optimizations
3805 which allow the hypervisor to 'idle' the guest on lock
3808 norandmaps Don't use address space randomization. Equivalent to
3809 echo 0 > /proc/sys/kernel/randomize_va_space
3811 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
3812 with UP alternatives
3814 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).
3823 nosgx [X86-64,SGX] Disables Intel SGX kernel support.
3826 Disable SMAP (Supervisor Mode Access Prevention)
3827 even if it is supported by processor.
3830 Disable SMEP (Supervisor Mode Execution Prevention)
3831 even if it is supported by processor.
3833 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
3834 and disable the IO APIC. legacy for "maxcpus=0".
3836 nosmt [KNL,S390] Disable symmetric multithreading (SMT).
3837 Equivalent to smt=1.
3839 [KNL,X86] Disable symmetric multithreading (SMT).
3840 nosmt=force: Force disable SMT, cannot be undone
3841 via the sysfs control file.
3843 nosoftlockup [KNL] Disable the soft-lockup detector.
3845 nospec_store_bypass_disable
3846 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
3848 nospectre_bhb [ARM64] Disable all mitigations for Spectre-BHB (branch
3849 history injection) vulnerability. System may allow data leaks
3852 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1
3853 (bounds check bypass). With this option data leaks are
3854 possible in the system.
3856 nospectre_v2 [X86,PPC_E500,ARM64] Disable all mitigations for
3857 the Spectre variant 2 (indirect branch prediction)
3858 vulnerability. System may allow data leaks with this
3861 no-steal-acc [X86,PV_OPS,ARM64,PPC/PSERIES] Disable paravirtualized
3862 steal time accounting. steal time is computed, but
3863 won't influence scheduler behaviour
3865 nosync [HW,M68K] Disables sync negotiation for all devices.
3867 no_timer_check [X86,APIC] Disables the code which tests for
3868 broken timer IRQ sources.
3871 [PPC] Don't flush the L1-D cache after accessing user data.
3873 novmcoredd [KNL,KDUMP]
3874 Disable device dump. Device dump allows drivers to
3875 append dump data to vmcore so you can collect driver
3876 specified debug info. Drivers can append the data
3877 without any limit and this data is stored in memory,
3878 so this may cause significant memory stress. Disabling
3879 device dump can help save memory but the driver debug
3880 data will be no longer available. This parameter
3881 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
3885 [X86,PV_OPS] Disable paravirtualized VMware scheduler
3886 clock and use the default one.
3888 nowatchdog [KNL] Disable both lockup detectors, i.e.
3889 soft-lockup and NMI watchdog (hard-lockup).
3893 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
3895 NOTE: this parameter will be ignored on systems with the
3896 LEGACY_XAPIC_DISABLED bit set in the
3897 IA32_XAPIC_DISABLE_STATUS MSR.
3899 noxsave [BUGS=X86] Disables x86 extended register state save
3900 and restore using xsave. The kernel will fallback to
3901 enabling legacy floating-point and sse state.
3903 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
3904 register states. The kernel will fall back to use
3905 xsave to save the states. By using this parameter,
3906 performance of saving the states is degraded because
3907 xsave doesn't support modified optimization while
3908 xsaveopt supports it on xsaveopt enabled systems.
3910 noxsaves [X86] Disables xsaves and xrstors used in saving and
3911 restoring x86 extended register state in compacted
3912 form of xsave area. The kernel will fall back to use
3913 xsaveopt and xrstor to save and restore the states
3914 in standard form of xsave area. By using this
3915 parameter, xsave area per process might occupy more
3916 memory on xsaves enabled systems.
3918 nps_mtm_hs_ctr= [KNL,ARC]
3919 This parameter sets the maximum duration, in
3920 cycles, each HW thread of the CTOP can run
3921 without interruptions, before HW switches it.
3922 The actual maximum duration is 16 times this
3924 Format: integer between 1 and 255
3927 nptcg= [IA-64] Override max number of concurrent global TLB
3928 purges which is reported from either PAL_VM_SUMMARY or
3931 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
3932 could support. nr_cpus=n : n >= 1 limits the kernel to
3933 support 'n' processors. It could be larger than the
3934 number of already plugged CPU during bootup, later in
3935 runtime you can physically add extra cpu until it reaches
3936 n. So during boot up some boot time memory for per-cpu
3937 variables need be pre-allocated for later physical cpu
3940 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
3942 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86] Disable NUMA, Only
3943 set up a single NUMA node spanning all memory.
3945 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic
3947 Allowed values are enable and disable
3949 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
3950 'node', 'default' can be specified
3951 This can be set from sysctl after boot.
3952 See Documentation/admin-guide/sysctl/vm.rst for details.
3954 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
3955 See Documentation/core-api/debugging-via-ohci1394.rst for more
3958 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
3959 Rather than timing out after 20 ms if an EC
3960 command is not properly ACKed, override the length
3961 of the timeout. We have interrupts disabled while
3962 waiting for the ACK, so if this is set too high
3963 interrupts *may* be lost!
3965 omap_mux= [OMAP] Override bootloader pin multiplexing.
3966 Format: <mux_mode0.mode_name=value>...
3967 For example, to override I2C bus2:
3968 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
3970 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
3972 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
3974 boundary - index of last SLC block on Flex-OneNAND.
3975 The remaining blocks are configured as MLC blocks.
3976 lock - Configure if Flex-OneNAND boundary should be locked.
3977 Once locked, the boundary cannot be changed.
3978 1 indicates lock status, 0 indicates unlock status.
3980 oops=panic Always panic on oopses. Default is to just kill the
3981 process, but there is a small probability of
3982 deadlocking the machine.
3983 This will also cause panics on machine check exceptions.
3984 Useful together with panic=30 to trigger a reboot.
3987 [KNL] Boolean flag to control whether the page allocator
3988 should randomize its free lists. The randomization may
3989 be automatically enabled if the kernel detects it is
3990 running on a platform with a direct-mapped memory-side
3991 cache, and this parameter can be used to
3992 override/disable that behavior. The state of the flag
3993 can be read from sysfs at:
3994 /sys/module/page_alloc/parameters/shuffle.
3996 page_owner= [KNL] Boot-time page_owner enabling option.
3997 Storage of the information about who allocated
3998 each page is disabled in default. With this switch,
4000 on: enable the feature
4002 page_poison= [KNL] Boot-time parameter changing the state of
4003 poisoning on the buddy allocator, available with
4004 CONFIG_PAGE_POISONING=y.
4005 off: turn off poisoning (default)
4006 on: turn on poisoning
4008 page_reporting.page_reporting_order=
4009 [KNL] Minimal page reporting order
4011 Adjust the minimal page reporting order. The page
4012 reporting is disabled when it exceeds MAX_ORDER.
4014 panic= [KNL] Kernel behaviour on panic: delay <timeout>
4015 timeout > 0: seconds before rebooting
4016 timeout = 0: wait forever
4017 timeout < 0: reboot immediately
4020 panic_print= Bitmask for printing system info when panic happens.
4021 User can chose combination of the following bits:
4022 bit 0: print all tasks info
4023 bit 1: print system memory info
4024 bit 2: print timer info
4025 bit 3: print locks info if CONFIG_LOCKDEP is on
4026 bit 4: print ftrace buffer
4027 bit 5: print all printk messages in buffer
4028 bit 6: print all CPUs backtrace (if available in the arch)
4029 *Be aware* that this option may print a _lot_ of lines,
4030 so there are risks of losing older messages in the log.
4031 Use this option carefully, maybe worth to setup a
4032 bigger log buffer with "log_buf_len" along with this.
4034 panic_on_taint= Bitmask for conditionally calling panic() in add_taint()
4035 Format: <hex>[,nousertaint]
4036 Hexadecimal bitmask representing the set of TAINT flags
4037 that will cause the kernel to panic when add_taint() is
4038 called with any of the flags in this set.
4039 The optional switch "nousertaint" can be utilized to
4040 prevent userspace forced crashes by writing to sysctl
4041 /proc/sys/kernel/tainted any flagset matching with the
4042 bitmask set on panic_on_taint.
4043 See Documentation/admin-guide/tainted-kernels.rst for
4044 extra details on the taint flags that users can pick
4045 to compose the bitmask to assign to panic_on_taint.
4047 panic_on_warn panic() instead of WARN(). Useful to cause kdump
4050 parkbd.port= [HW] Parallel port number the keyboard adapter is
4051 connected to, default is 0.
4053 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
4054 0 for XT, 1 for AT (default is AT).
4057 parport= [HW,PPT] Specify parallel ports. 0 disables.
4058 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
4059 Use 'auto' to force the driver to use any
4060 IRQ/DMA settings detected (the default is to
4061 ignore detected IRQ/DMA settings because of
4062 possible conflicts). You can specify the base
4063 address, IRQ, and DMA settings; IRQ and DMA
4064 should be numbers, or 'auto' (for using detected
4065 settings on that particular port), or 'nofifo'
4066 (to avoid using a FIFO even if it is detected).
4067 Parallel ports are assigned in the order they
4068 are specified on the command line, starting
4071 parport_init_mode= [HW,PPT]
4072 Configure VIA parallel port to operate in
4073 a specific mode. This is necessary on Pegasos
4074 computer where firmware has no options for setting
4075 up parallel port mode and sets it to spp.
4076 Currently this function knows 686a and 8231 chips.
4077 Format: [spp|ps2|epp|ecp|ecpepp]
4079 pata_legacy.all= [HW,LIBATA]
4081 Set to non-zero to probe primary and secondary ISA
4082 port ranges on PCI systems where no PCI PATA device
4083 has been found at either range. Disabled by default.
4085 pata_legacy.autospeed= [HW,LIBATA]
4087 Set to non-zero if a chip is present that snoops speed
4088 changes. Disabled by default.
4090 pata_legacy.ht6560a= [HW,LIBATA]
4092 Set to 1, 2, or 3 for HT 6560A on the primary channel,
4093 the secondary channel, or both channels respectively.
4094 Disabled by default.
4096 pata_legacy.ht6560b= [HW,LIBATA]
4098 Set to 1, 2, or 3 for HT 6560B on the primary channel,
4099 the secondary channel, or both channels respectively.
4100 Disabled by default.
4102 pata_legacy.iordy_mask= [HW,LIBATA]
4104 IORDY enable mask. Set individual bits to allow IORDY
4105 for the respective channel. Bit 0 is for the first
4106 legacy channel handled by this driver, bit 1 is for
4107 the second channel, and so on. The sequence will often
4108 correspond to the primary legacy channel, the secondary
4109 legacy channel, and so on, but the handling of a PCI
4110 bus and the use of other driver options may interfere
4111 with the sequence. By default IORDY is allowed across
4114 pata_legacy.opti82c46x= [HW,LIBATA]
4116 Set to 1, 2, or 3 for Opti 82c611A on the primary
4117 channel, the secondary channel, or both channels
4118 respectively. Disabled by default.
4120 pata_legacy.opti82c611a= [HW,LIBATA]
4122 Set to 1, 2, or 3 for Opti 82c465MV on the primary
4123 channel, the secondary channel, or both channels
4124 respectively. Disabled by default.
4126 pata_legacy.pio_mask= [HW,LIBATA]
4128 PIO mode mask for autospeed devices. Set individual
4129 bits to allow the use of the respective PIO modes.
4130 Bit 0 is for mode 0, bit 1 is for mode 1, and so on.
4131 All modes allowed by default.
4133 pata_legacy.probe_all= [HW,LIBATA]
4135 Set to non-zero to probe tertiary and further ISA
4136 port ranges on PCI systems. Disabled by default.
4138 pata_legacy.probe_mask= [HW,LIBATA]
4140 Probe mask for legacy ISA PATA ports. Depending on
4141 platform configuration and the use of other driver
4142 options up to 6 legacy ports are supported: 0x1f0,
4143 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing
4144 of individual ports can be disabled by setting the
4145 corresponding bits in the mask to 1. Bit 0 is for
4146 the first port in the list above (0x1f0), and so on.
4147 By default all supported ports are probed.
4149 pata_legacy.qdi= [HW,LIBATA]
4151 Set to non-zero to probe QDI controllers. By default
4152 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise.
4154 pata_legacy.winbond= [HW,LIBATA]
4156 Set to non-zero to probe Winbond controllers. Use
4157 the standard I/O port (0x130) if 1, otherwise the
4158 value given is the I/O port to use (typically 0x1b0).
4159 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE,
4162 pata_platform.pio_mask= [HW,LIBATA]
4164 Supported PIO mode mask. Set individual bits to allow
4165 the use of the respective PIO modes. Bit 0 is for
4166 mode 0, bit 1 is for mode 1, and so on. Mode 0 only
4170 Halt all CPUs after the first oops has been printed for
4171 the specified number of seconds. This is to be used if
4172 your oopses keep scrolling off the screen.
4176 pci=option[,option...] [PCI] various PCI subsystem options.
4178 Some options herein operate on a specific device
4179 or a set of devices (<pci_dev>). These are
4180 specified in one of the following formats:
4182 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
4183 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
4185 Note: the first format specifies a PCI
4186 bus/device/function address which may change
4187 if new hardware is inserted, if motherboard
4188 firmware changes, or due to changes caused
4189 by other kernel parameters. If the
4190 domain is left unspecified, it is
4191 taken to be zero. Optionally, a path
4192 to a device through multiple device/function
4193 addresses can be specified after the base
4194 address (this is more robust against
4195 renumbering issues). The second format
4196 selects devices using IDs from the
4197 configuration space which may match multiple
4198 devices in the system.
4200 earlydump dump PCI config space before the kernel
4202 off [X86] don't probe for the PCI bus
4203 bios [X86-32] force use of PCI BIOS, don't access
4204 the hardware directly. Use this if your machine
4205 has a non-standard PCI host bridge.
4206 nobios [X86-32] disallow use of PCI BIOS, only direct
4207 hardware access methods are allowed. Use this
4208 if you experience crashes upon bootup and you
4209 suspect they are caused by the BIOS.
4210 conf1 [X86] Force use of PCI Configuration Access
4211 Mechanism 1 (config address in IO port 0xCF8,
4212 data in IO port 0xCFC, both 32-bit).
4213 conf2 [X86] Force use of PCI Configuration Access
4214 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
4215 the function, IO port 0xCFA, also 8-bit, sets
4216 bus number. The config space is then accessed
4217 through ports 0xC000-0xCFFF).
4218 See http://wiki.osdev.org/PCI for more info
4219 on the configuration access mechanisms.
4220 noaer [PCIE] If the PCIEAER kernel config parameter is
4221 enabled, this kernel boot option can be used to
4222 disable the use of PCIE advanced error reporting.
4223 nodomains [PCI] Disable support for multiple PCI
4224 root domains (aka PCI segments, in ACPI-speak).
4225 nommconf [X86] Disable use of MMCONFIG for PCI
4227 check_enable_amd_mmconf [X86] check for and enable
4228 properly configured MMIO access to PCI
4229 config space on AMD family 10h CPU
4230 nomsi [MSI] If the PCI_MSI kernel config parameter is
4231 enabled, this kernel boot option can be used to
4232 disable the use of MSI interrupts system-wide.
4233 noioapicquirk [APIC] Disable all boot interrupt quirks.
4234 Safety option to keep boot IRQs enabled. This
4235 should never be necessary.
4236 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
4237 primary IO-APIC for bridges that cannot disable
4238 boot IRQs. This fixes a source of spurious IRQs
4239 when the system masks IRQs.
4240 noioapicreroute [APIC] Disable workaround that uses the
4241 boot IRQ equivalent of an IRQ that connects to
4242 a chipset where boot IRQs cannot be disabled.
4243 The opposite of ioapicreroute.
4244 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
4245 routing table. These calls are known to be buggy
4246 on several machines and they hang the machine
4247 when used, but on other computers it's the only
4248 way to get the interrupt routing table. Try
4249 this option if the kernel is unable to allocate
4250 IRQs or discover secondary PCI buses on your
4252 rom [X86] Assign address space to expansion ROMs.
4253 Use with caution as certain devices share
4254 address decoders between ROMs and other
4256 norom [X86] Do not assign address space to
4257 expansion ROMs that do not already have
4258 BIOS assigned address ranges.
4259 nobar [X86] Do not assign address space to the
4260 BARs that weren't assigned by the BIOS.
4261 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
4262 assigned automatically to PCI devices. You can
4263 make the kernel exclude IRQs of your ISA cards
4265 pirqaddr=0xAAAAA [X86] Specify the physical address
4266 of the PIRQ table (normally generated
4267 by the BIOS) if it is outside the
4268 F0000h-100000h range.
4269 lastbus=N [X86] Scan all buses thru bus #N. Can be
4270 useful if the kernel is unable to find your
4271 secondary buses and you want to tell it
4272 explicitly which ones they are.
4273 assign-busses [X86] Always assign all PCI bus
4274 numbers ourselves, overriding
4275 whatever the firmware may have done.
4276 usepirqmask [X86] Honor the possible IRQ mask stored
4277 in the BIOS $PIR table. This is needed on
4278 some systems with broken BIOSes, notably
4279 some HP Pavilion N5400 and Omnibook XE3
4280 notebooks. This will have no effect if ACPI
4281 IRQ routing is enabled.
4282 noacpi [X86] Do not use ACPI for IRQ routing
4283 or for PCI scanning.
4284 use_crs [X86] Use PCI host bridge window information
4285 from ACPI. On BIOSes from 2008 or later, this
4286 is enabled by default. If you need to use this,
4287 please report a bug.
4288 nocrs [X86] Ignore PCI host bridge windows from ACPI.
4289 If you need to use this, please report a bug.
4290 use_e820 [X86] Use E820 reservations to exclude parts of
4291 PCI host bridge windows. This is a workaround
4292 for BIOS defects in host bridge _CRS methods.
4293 If you need to use this, please report a bug to
4294 <linux-pci@vger.kernel.org>.
4295 no_e820 [X86] Ignore E820 reservations for PCI host
4296 bridge windows. This is the default on modern
4297 hardware. If you need to use this, please report
4298 a bug to <linux-pci@vger.kernel.org>.
4299 routeirq Do IRQ routing for all PCI devices.
4300 This is normally done in pci_enable_device(),
4301 so this option is a temporary workaround
4302 for broken drivers that don't call it.
4303 skip_isa_align [X86] do not align io start addr, so can
4304 handle more pci cards
4305 noearly [X86] Don't do any early type 1 scanning.
4306 This might help on some broken boards which
4307 machine check when some devices' config space
4308 is read. But various workarounds are disabled
4309 and some IOMMU drivers will not work.
4310 bfsort Sort PCI devices into breadth-first order.
4311 This sorting is done to get a device
4312 order compatible with older (<= 2.4) kernels.
4313 nobfsort Don't sort PCI devices into breadth-first order.
4314 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
4315 tuning and use the BIOS-configured MPS defaults.
4316 pcie_bus_safe Set every device's MPS to the largest value
4317 supported by all devices below the root complex.
4318 pcie_bus_perf Set device MPS to the largest allowable MPS
4319 based on its parent bus. Also set MRRS (Max
4320 Read Request Size) to the largest supported
4321 value (no larger than the MPS that the device
4322 or bus can support) for best performance.
4323 pcie_bus_peer2peer Set every device's MPS to 128B, which
4324 every device is guaranteed to support. This
4325 configuration allows peer-to-peer DMA between
4326 any pair of devices, possibly at the cost of
4327 reduced performance. This also guarantees
4328 that hot-added devices will work.
4329 cbiosize=nn[KMG] The fixed amount of bus space which is
4330 reserved for the CardBus bridge's IO window.
4331 The default value is 256 bytes.
4332 cbmemsize=nn[KMG] The fixed amount of bus space which is
4333 reserved for the CardBus bridge's memory
4334 window. The default value is 64 megabytes.
4337 [<order of align>@]<pci_dev>[; ...]
4338 Specifies alignment and device to reassign
4339 aligned memory resources. How to
4340 specify the device is described above.
4341 If <order of align> is not specified,
4342 PAGE_SIZE is used as alignment.
4343 A PCI-PCI bridge can be specified if resource
4344 windows need to be expanded.
4345 To specify the alignment for several
4346 instances of a device, the PCI vendor,
4347 device, subvendor, and subdevice may be
4348 specified, e.g., 12@pci:8086:9c22:103c:198f
4349 for 4096-byte alignment.
4350 ecrc= Enable/disable PCIe ECRC (transaction layer
4351 end-to-end CRC checking). Only effective if
4352 OS has native AER control (either granted by
4353 ACPI _OSC or forced via "pcie_ports=native")
4354 bios: Use BIOS/firmware settings. This is the
4358 hpiosize=nn[KMG] The fixed amount of bus space which is
4359 reserved for hotplug bridge's IO window.
4360 Default size is 256 bytes.
4361 hpmmiosize=nn[KMG] The fixed amount of bus space which is
4362 reserved for hotplug bridge's MMIO window.
4363 Default size is 2 megabytes.
4364 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
4365 reserved for hotplug bridge's MMIO_PREF window.
4366 Default size is 2 megabytes.
4367 hpmemsize=nn[KMG] The fixed amount of bus space which is
4368 reserved for hotplug bridge's MMIO and
4370 Default size is 2 megabytes.
4371 hpbussize=nn The minimum amount of additional bus numbers
4372 reserved for buses below a hotplug bridge.
4374 realloc= Enable/disable reallocating PCI bridge resources
4375 if allocations done by BIOS are too small to
4376 accommodate resources required by all child
4378 off: Turn realloc off
4380 realloc same as realloc=on
4381 noari do not use PCIe ARI.
4382 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
4383 do not use PCIe ATS (and IOMMU device IOTLB).
4384 pcie_scan_all Scan all possible PCIe devices. Otherwise we
4385 only look for one device below a PCIe downstream
4387 big_root_window Try to add a big 64bit memory window to the PCIe
4388 root complex on AMD CPUs. Some GFX hardware
4389 can resize a BAR to allow access to all VRAM.
4390 Adding the window is slightly risky (it may
4391 conflict with unreported devices), so this
4393 disable_acs_redir=<pci_dev>[; ...]
4394 Specify one or more PCI devices (in the format
4395 specified above) separated by semicolons.
4396 Each device specified will have the PCI ACS
4397 redirect capabilities forced off which will
4398 allow P2P traffic between devices through
4399 bridges without forcing it upstream. Note:
4400 this removes isolation between devices and
4401 may put more devices in an IOMMU group.
4402 force_floating [S390] Force usage of floating interrupts.
4403 nomio [S390] Do not use MIO instructions.
4404 norid [S390] ignore the RID field and force use of
4405 one PCI domain per PCI function
4407 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
4410 force Enable ASPM even on devices that claim not to support it.
4411 WARNING: Forcing ASPM on may cause system lockups.
4413 pcie_ports= [PCIE] PCIe port services handling:
4414 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
4415 even if the platform doesn't give the OS permission to
4416 use them. This may cause conflicts if the platform
4417 also tries to use these services.
4418 dpc-native Use native PCIe service for DPC only. May
4419 cause conflicts if firmware uses AER or DPC.
4420 compat Disable native PCIe services (PME, AER, DPC, PCIe
4423 pcie_port_pm= [PCIE] PCIe port power management handling:
4424 off Disable power management of all PCIe ports
4425 force Forcibly enable power management of all PCIe ports
4427 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
4428 nomsi Do not use MSI for native PCIe PME signaling (this makes
4429 all PCIe root ports use INTx for all services).
4431 pcmv= [HW,PCMCIA] BadgePAD 4
4435 Keep all power-domains already enabled by bootloader on,
4436 even if no driver has claimed them. This is useful
4437 for debug and development, but should not be
4438 needed on a platform with proper driver support.
4440 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
4443 See arch/parisc/kernel/pdc_chassis.c
4445 percpu_alloc= Select which percpu first chunk allocator to use.
4446 Currently supported values are "embed" and "page".
4447 Archs may support subset or none of the selections.
4448 See comments in mm/percpu.c for details on each
4449 allocator. This parameter is primarily for debugging
4450 and performance comparison.
4452 pirq= [SMP,APIC] Manual mp-table setup
4453 See Documentation/arch/x86/i386/IO-APIC.rst.
4455 plip= [PPT,NET] Parallel port network link
4456 Format: { parport<nr> | timid | 0 }
4457 See also Documentation/admin-guide/parport.rst.
4459 pmtmr= [X86] Manual setup of pmtmr I/O Port.
4460 Override pmtimer IOPort with a hex value.
4463 pmu_override= [PPC] Override the PMU.
4464 This option takes over the PMU facility, so it is no
4465 longer usable by perf. Setting this option starts the
4466 PMU counters by setting MMCR0 to 0 (the FC bit is
4467 cleared). If a number is given, then MMCR1 is set to
4468 that number, otherwise (e.g., 'pmu_override=on'), MMCR1
4471 pm_debug_messages [SUSPEND,KNL]
4472 Enable suspend/resume debug messages during boot up.
4475 Enable PNP debug messages (depends on the
4476 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
4477 via /sys/module/pnp/parameters/debug. We always show
4478 current resource usage; turning this on also shows
4479 possible settings and some assignment information.
4485 { on | off | curr | res | no-curr | no-res }
4488 [ISAPNP] Exclude IRQs for the autoconfiguration
4491 [ISAPNP] Exclude DMAs for the autoconfiguration
4493 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
4494 Ranges are in pairs (I/O port base and size).
4497 [ISAPNP] Exclude memory regions for the
4499 Ranges are in pairs (memory base and size).
4501 ports= [IP_VS_FTP] IPVS ftp helper module
4503 Up to 8 (IP_VS_APP_MAX_PORTS) ports
4505 Format: <port>,<port>....
4507 powersave=off [PPC] This option disables power saving features.
4508 It specifically disables cpuidle and sets the
4509 platform machine description specific power_save
4510 function to NULL. On Idle the CPU just reduces
4513 ppc_strict_facility_enable
4514 [PPC] This option catches any kernel floating point,
4515 Altivec, VSX and SPE outside of regions specifically
4516 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
4517 There is some performance impact when enabling this.
4521 Disable Hardware Transactional Memory
4524 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC
4525 none - Limited to cond_resched() calls
4526 voluntary - Limited to cond_resched() and might_sleep() calls
4527 full - Any section that isn't explicitly preempt disabled
4528 can be preempted anytime.
4530 print-fatal-signals=
4531 [KNL] debug: print fatal signals
4533 If enabled, warn about various signal handling
4534 related application anomalies: too many signals,
4535 too many POSIX.1 timers, fatal signals causing a
4538 If you hit the warning due to signal overflow,
4539 you might want to try "ulimit -i unlimited".
4543 printk.always_kmsg_dump=
4544 Trigger kmsg_dump for cases other than kernel oops or
4546 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4549 printk.console_no_auto_verbose=
4550 Disable console loglevel raise on oops, panic
4551 or lockdep-detected issues (only if lock debug is on).
4552 With an exception to setups with low baudrate on
4553 serial console, keeping this 0 is a good choice
4554 in order to provide more debug information.
4556 default: 0 (auto_verbose is enabled)
4558 printk.devkmsg={on,off,ratelimit}
4559 Control writing to /dev/kmsg.
4560 on - unlimited logging to /dev/kmsg from userspace
4561 off - logging to /dev/kmsg disabled
4562 ratelimit - ratelimit the logging
4565 printk.time= Show timing data prefixed to each printk message line
4566 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4568 processor.max_cstate= [HW,ACPI]
4569 Limit processor to maximum C-state
4570 max_cstate=9 overrides any DMI blacklist limit.
4572 processor.nocst [HW,ACPI]
4573 Ignore the _CST method to determine C-states,
4574 instead using the legacy FADT method
4576 profile= [KNL] Enable kernel profiling via /proc/profile
4577 Format: [<profiletype>,]<number>
4578 Param: <profiletype>: "schedule", "sleep", or "kvm"
4579 [defaults to kernel profiling]
4580 Param: "schedule" - profile schedule points.
4581 Param: "sleep" - profile D-state sleeping (millisecs).
4582 Requires CONFIG_SCHEDSTATS
4583 Param: "kvm" - profile VM exits.
4584 Param: <number> - step/bucket size as a power of 2 for
4585 statistical time based profiling.
4587 prompt_ramdisk= [RAM] [Deprecated]
4589 prot_virt= [S390] enable hosting protected virtual machines
4590 isolated from the hypervisor (if hardware supports
4594 psi= [KNL] Enable or disable pressure stall information
4598 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
4599 probe for; one of (bare|imps|exps|lifebook|any).
4600 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
4602 psmouse.resetafter= [HW,MOUSE]
4603 Try to reset the device after so many bad packets
4606 [HW,MOUSE] Set desired mouse resolution, in dpi.
4607 psmouse.smartscroll=
4608 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
4609 0 = disabled, 1 = enabled (default).
4611 pstore.backend= Specify the name of the pstore backend to use
4613 pti= [X86-64] Control Page Table Isolation of user and
4614 kernel address spaces. Disabling this feature
4615 removes hardening, but improves performance of
4616 system calls and interrupts.
4618 on - unconditionally enable
4619 off - unconditionally disable
4620 auto - kernel detects whether your CPU model is
4621 vulnerable to issues that PTI mitigates
4623 Not specifying this option is equivalent to pti=auto.
4626 [KNL] Number of legacy pty's. Overwrites compiled-in
4629 quiet [KNL] Disable most log messages
4633 radix_hcall_invalidate=on [PPC/PSERIES]
4634 Disable RADIX GTSE feature and use hcall for TLB
4638 See Documentation/admin-guide/md.rst.
4640 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
4641 See Documentation/admin-guide/blockdev/ramdisk.rst.
4643 ramdisk_start= [RAM] RAM disk image start address
4645 random.trust_cpu=off
4646 [KNL] Disable trusting the use of the CPU's
4647 random number generator (if available) to
4648 initialize the kernel's RNG.
4650 random.trust_bootloader=off
4651 [KNL] Disable trusting the use of the a seed
4652 passed by the bootloader (if available) to
4653 initialize the kernel's RNG.
4655 randomize_kstack_offset=
4656 [KNL] Enable or disable kernel stack offset
4657 randomization, which provides roughly 5 bits of
4658 entropy, frustrating memory corruption attacks
4659 that depend on stack address determinism or
4660 cross-syscall address exposures. This is only
4661 available on architectures that have defined
4662 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET.
4663 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4664 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT.
4666 ras=option[,option,...] [KNL] RAS-specific options
4669 Disable the Correctable Errors Collector,
4670 see CONFIG_RAS_CEC help text.
4672 rcu_nocbs[=cpu-list]
4673 [KNL] The optional argument is a cpu list,
4676 In kernels built with CONFIG_RCU_NOCB_CPU=y,
4677 enable the no-callback CPU mode, which prevents
4678 such CPUs' callbacks from being invoked in
4679 softirq context. Invocation of such CPUs' RCU
4680 callbacks will instead be offloaded to "rcuox/N"
4681 kthreads created for that purpose, where "x" is
4682 "p" for RCU-preempt, "s" for RCU-sched, and "g"
4683 for the kthreads that mediate grace periods; and
4684 "N" is the CPU number. This reduces OS jitter on
4685 the offloaded CPUs, which can be useful for HPC
4686 and real-time workloads. It can also improve
4687 energy efficiency for asymmetric multiprocessors.
4689 If a cpulist is passed as an argument, the specified
4690 list of CPUs is set to no-callback mode from boot.
4692 Otherwise, if the '=' sign and the cpulist
4693 arguments are omitted, no CPU will be set to
4694 no-callback mode from boot but the mode may be
4695 toggled at runtime via cpusets.
4697 Note that this argument takes precedence over
4698 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option.
4701 Rather than requiring that offloaded CPUs
4702 (specified by rcu_nocbs= above) explicitly
4703 awaken the corresponding "rcuoN" kthreads,
4704 make these kthreads poll for callbacks.
4705 This improves the real-time response for the
4706 offloaded CPUs by relieving them of the need to
4707 wake up the corresponding kthread, but degrades
4708 energy efficiency by requiring that the kthreads
4709 periodically wake up to do the polling.
4711 rcutree.blimit= [KNL]
4712 Set maximum number of finished RCU callbacks to
4713 process in one batch.
4715 rcutree.dump_tree= [KNL]
4716 Dump the structure of the rcu_node combining tree
4717 out at early boot. This is used for diagnostic
4718 purposes, to verify correct tree setup.
4720 rcutree.gp_cleanup_delay= [KNL]
4721 Set the number of jiffies to delay each step of
4722 RCU grace-period cleanup.
4724 rcutree.gp_init_delay= [KNL]
4725 Set the number of jiffies to delay each step of
4726 RCU grace-period initialization.
4728 rcutree.gp_preinit_delay= [KNL]
4729 Set the number of jiffies to delay each step of
4730 RCU grace-period pre-initialization, that is,
4731 the propagation of recent CPU-hotplug changes up
4732 the rcu_node combining tree.
4734 rcutree.jiffies_till_first_fqs= [KNL]
4735 Set delay from grace-period initialization to
4736 first attempt to force quiescent states.
4737 Units are jiffies, minimum value is zero,
4738 and maximum value is HZ.
4740 rcutree.jiffies_till_next_fqs= [KNL]
4741 Set delay between subsequent attempts to force
4742 quiescent states. Units are jiffies, minimum
4743 value is one, and maximum value is HZ.
4745 rcutree.jiffies_till_sched_qs= [KNL]
4746 Set required age in jiffies for a
4747 given grace period before RCU starts
4748 soliciting quiescent-state help from
4749 rcu_note_context_switch() and cond_resched().
4750 If not specified, the kernel will calculate
4751 a value based on the most recent settings
4752 of rcutree.jiffies_till_first_fqs
4753 and rcutree.jiffies_till_next_fqs.
4754 This calculated value may be viewed in
4755 rcutree.jiffies_to_sched_qs. Any attempt to set
4756 rcutree.jiffies_to_sched_qs will be cheerfully
4759 rcutree.kthread_prio= [KNL,BOOT]
4760 Set the SCHED_FIFO priority of the RCU per-CPU
4761 kthreads (rcuc/N). This value is also used for
4762 the priority of the RCU boost threads (rcub/N)
4763 and for the RCU grace-period kthreads (rcu_bh,
4764 rcu_preempt, and rcu_sched). If RCU_BOOST is
4765 set, valid values are 1-99 and the default is 1
4766 (the least-favored priority). Otherwise, when
4767 RCU_BOOST is not set, valid values are 0-99 and
4768 the default is zero (non-realtime operation).
4769 When RCU_NOCB_CPU is set, also adjust the
4770 priority of NOCB callback kthreads.
4772 rcutree.nocb_nobypass_lim_per_jiffy= [KNL]
4773 On callback-offloaded (rcu_nocbs) CPUs,
4774 RCU reduces the lock contention that would
4775 otherwise be caused by callback floods through
4776 use of the ->nocb_bypass list. However, in the
4777 common non-flooded case, RCU queues directly to
4778 the main ->cblist in order to avoid the extra
4779 overhead of the ->nocb_bypass list and its lock.
4780 But if there are too many callbacks queued during
4781 a single jiffy, RCU pre-queues the callbacks into
4782 the ->nocb_bypass queue. The definition of "too
4783 many" is supplied by this kernel boot parameter.
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_delay_page_cache_fill_msec= [KNL]
4803 Set the page-cache refill delay (in milliseconds)
4804 in response to low-memory conditions. The range
4805 of permitted values is in the range 0:100000.
4807 rcutree.rcu_divisor= [KNL]
4808 Set the shift-right count to use to compute
4809 the callback-invocation batch limit bl from
4810 the number of callbacks queued on this CPU.
4811 The result will be bounded below by the value of
4812 the rcutree.blimit kernel parameter. Every bl
4813 callbacks, the softirq handler will exit in
4814 order to allow the CPU to do other work.
4816 Please note that this callback-invocation batch
4817 limit applies only to non-offloaded callback
4818 invocation. Offloaded callbacks are instead
4819 invoked in the context of an rcuoc kthread, which
4820 scheduler will preempt as it does any other task.
4822 rcutree.rcu_fanout_exact= [KNL]
4823 Disable autobalancing of the rcu_node combining
4824 tree. This is used by rcutorture, and might
4825 possibly be useful for architectures having high
4826 cache-to-cache transfer latencies.
4828 rcutree.rcu_fanout_leaf= [KNL]
4829 Change the number of CPUs assigned to each
4830 leaf rcu_node structure. Useful for very
4831 large systems, which will choose the value 64,
4832 and for NUMA systems with large remote-access
4833 latencies, which will choose a value aligned
4834 with the appropriate hardware boundaries.
4836 rcutree.rcu_min_cached_objs= [KNL]
4837 Minimum number of objects which are cached and
4838 maintained per one CPU. Object size is equal
4839 to PAGE_SIZE. The cache allows to reduce the
4840 pressure to page allocator, also it makes the
4841 whole algorithm to behave better in low memory
4844 rcutree.rcu_nocb_gp_stride= [KNL]
4845 Set the number of NOCB callback kthreads in
4846 each group, which defaults to the square root
4847 of the number of CPUs. Larger numbers reduce
4848 the wakeup overhead on the global grace-period
4849 kthread, but increases that same overhead on
4850 each group's NOCB grace-period kthread.
4852 rcutree.rcu_kick_kthreads= [KNL]
4853 Cause the grace-period kthread to get an extra
4854 wake_up() if it sleeps three times longer than
4855 it should at force-quiescent-state time.
4856 This wake_up() will be accompanied by a
4857 WARN_ONCE() splat and an ftrace_dump().
4859 rcutree.rcu_resched_ns= [KNL]
4860 Limit the time spend invoking a batch of RCU
4861 callbacks to the specified number of nanoseconds.
4862 By default, this limit is checked only once
4863 every 32 callbacks in order to limit the pain
4864 inflicted by local_clock() overhead.
4866 rcutree.rcu_unlock_delay= [KNL]
4867 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels,
4868 this specifies an rcu_read_unlock()-time delay
4869 in microseconds. This defaults to zero.
4870 Larger delays increase the probability of
4871 catching RCU pointer leaks, that is, buggy use
4872 of RCU-protected pointers after the relevant
4873 rcu_read_unlock() has completed.
4875 rcutree.sysrq_rcu= [KNL]
4876 Commandeer a sysrq key to dump out Tree RCU's
4877 rcu_node tree with an eye towards determining
4878 why a new grace period has not yet started.
4880 rcutree.use_softirq= [KNL]
4881 If set to zero, move all RCU_SOFTIRQ processing to
4882 per-CPU rcuc kthreads. Defaults to a non-zero
4883 value, meaning that RCU_SOFTIRQ is used by default.
4884 Specify rcutree.use_softirq=0 to use rcuc kthreads.
4886 But note that CONFIG_PREEMPT_RT=y kernels disable
4887 this kernel boot parameter, forcibly setting it
4890 rcuscale.gp_async= [KNL]
4891 Measure performance of asynchronous
4892 grace-period primitives such as call_rcu().
4894 rcuscale.gp_async_max= [KNL]
4895 Specify the maximum number of outstanding
4896 callbacks per writer thread. When a writer
4897 thread exceeds this limit, it invokes the
4898 corresponding flavor of rcu_barrier() to allow
4899 previously posted callbacks to drain.
4901 rcuscale.gp_exp= [KNL]
4902 Measure performance of expedited synchronous
4903 grace-period primitives.
4905 rcuscale.holdoff= [KNL]
4906 Set test-start holdoff period. The purpose of
4907 this parameter is to delay the start of the
4908 test until boot completes in order to avoid
4911 rcuscale.kfree_rcu_test= [KNL]
4912 Set to measure performance of kfree_rcu() flooding.
4914 rcuscale.kfree_rcu_test_double= [KNL]
4915 Test the double-argument variant of kfree_rcu().
4916 If this parameter has the same value as
4917 rcuscale.kfree_rcu_test_single, both the single-
4918 and double-argument variants are tested.
4920 rcuscale.kfree_rcu_test_single= [KNL]
4921 Test the single-argument variant of kfree_rcu().
4922 If this parameter has the same value as
4923 rcuscale.kfree_rcu_test_double, both the single-
4924 and double-argument variants are tested.
4926 rcuscale.kfree_nthreads= [KNL]
4927 The number of threads running loops of kfree_rcu().
4929 rcuscale.kfree_alloc_num= [KNL]
4930 Number of allocations and frees done in an iteration.
4932 rcuscale.kfree_loops= [KNL]
4933 Number of loops doing rcuscale.kfree_alloc_num number
4934 of allocations and frees.
4936 rcuscale.nreaders= [KNL]
4937 Set number of RCU readers. The value -1 selects
4938 N, where N is the number of CPUs. A value
4939 "n" less than -1 selects N-n+1, where N is again
4940 the number of CPUs. For example, -2 selects N
4941 (the number of CPUs), -3 selects N+1, and so on.
4942 A value of "n" less than or equal to -N selects
4945 rcuscale.nwriters= [KNL]
4946 Set number of RCU writers. The values operate
4947 the same as for rcuscale.nreaders.
4948 N, where N is the number of CPUs
4950 rcuscale.perf_type= [KNL]
4951 Specify the RCU implementation to test.
4953 rcuscale.shutdown= [KNL]
4954 Shut the system down after performance tests
4955 complete. This is useful for hands-off automated
4958 rcuscale.verbose= [KNL]
4959 Enable additional printk() statements.
4961 rcuscale.writer_holdoff= [KNL]
4962 Write-side holdoff between grace periods,
4963 in microseconds. The default of zero says
4966 rcutorture.fqs_duration= [KNL]
4967 Set duration of force_quiescent_state bursts
4970 rcutorture.fqs_holdoff= [KNL]
4971 Set holdoff time within force_quiescent_state bursts
4974 rcutorture.fqs_stutter= [KNL]
4975 Set wait time between force_quiescent_state bursts
4978 rcutorture.fwd_progress= [KNL]
4979 Specifies the number of kthreads to be used
4980 for RCU grace-period forward-progress testing
4981 for the types of RCU supporting this notion.
4982 Defaults to 1 kthread, values less than zero or
4983 greater than the number of CPUs cause the number
4986 rcutorture.fwd_progress_div= [KNL]
4987 Specify the fraction of a CPU-stall-warning
4988 period to do tight-loop forward-progress testing.
4990 rcutorture.fwd_progress_holdoff= [KNL]
4991 Number of seconds to wait between successive
4992 forward-progress tests.
4994 rcutorture.fwd_progress_need_resched= [KNL]
4995 Enclose cond_resched() calls within checks for
4996 need_resched() during tight-loop forward-progress
4999 rcutorture.gp_cond= [KNL]
5000 Use conditional/asynchronous update-side
5001 primitives, if available.
5003 rcutorture.gp_exp= [KNL]
5004 Use expedited update-side primitives, if available.
5006 rcutorture.gp_normal= [KNL]
5007 Use normal (non-expedited) asynchronous
5008 update-side primitives, if available.
5010 rcutorture.gp_sync= [KNL]
5011 Use normal (non-expedited) synchronous
5012 update-side primitives, if available. If all
5013 of rcutorture.gp_cond=, rcutorture.gp_exp=,
5014 rcutorture.gp_normal=, and rcutorture.gp_sync=
5015 are zero, rcutorture acts as if is interpreted
5016 they are all non-zero.
5018 rcutorture.irqreader= [KNL]
5019 Run RCU readers from irq handlers, or, more
5020 accurately, from a timer handler. Not all RCU
5021 flavors take kindly to this sort of thing.
5023 rcutorture.leakpointer= [KNL]
5024 Leak an RCU-protected pointer out of the reader.
5025 This can of course result in splats, and is
5026 intended to test the ability of things like
5027 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect
5030 rcutorture.n_barrier_cbs= [KNL]
5031 Set callbacks/threads for rcu_barrier() testing.
5033 rcutorture.nfakewriters= [KNL]
5034 Set number of concurrent RCU writers. These just
5035 stress RCU, they don't participate in the actual
5036 test, hence the "fake".
5038 rcutorture.nocbs_nthreads= [KNL]
5039 Set number of RCU callback-offload togglers.
5040 Zero (the default) disables toggling.
5042 rcutorture.nocbs_toggle= [KNL]
5043 Set the delay in milliseconds between successive
5044 callback-offload toggling attempts.
5046 rcutorture.nreaders= [KNL]
5047 Set number of RCU readers. The value -1 selects
5048 N-1, where N is the number of CPUs. A value
5049 "n" less than -1 selects N-n-2, where N is again
5050 the number of CPUs. For example, -2 selects N
5051 (the number of CPUs), -3 selects N+1, and so on.
5053 rcutorture.object_debug= [KNL]
5054 Enable debug-object double-call_rcu() testing.
5056 rcutorture.onoff_holdoff= [KNL]
5057 Set time (s) after boot for CPU-hotplug testing.
5059 rcutorture.onoff_interval= [KNL]
5060 Set time (jiffies) between CPU-hotplug operations,
5061 or zero to disable CPU-hotplug testing.
5063 rcutorture.read_exit= [KNL]
5064 Set the number of read-then-exit kthreads used
5065 to test the interaction of RCU updaters and
5066 task-exit processing.
5068 rcutorture.read_exit_burst= [KNL]
5069 The number of times in a given read-then-exit
5070 episode that a set of read-then-exit kthreads
5073 rcutorture.read_exit_delay= [KNL]
5074 The delay, in seconds, between successive
5075 read-then-exit testing episodes.
5077 rcutorture.shuffle_interval= [KNL]
5078 Set task-shuffle interval (s). Shuffling tasks
5079 allows some CPUs to go into dyntick-idle mode
5080 during the rcutorture test.
5082 rcutorture.shutdown_secs= [KNL]
5083 Set time (s) after boot system shutdown. This
5084 is useful for hands-off automated testing.
5086 rcutorture.stall_cpu= [KNL]
5087 Duration of CPU stall (s) to test RCU CPU stall
5088 warnings, zero to disable.
5090 rcutorture.stall_cpu_block= [KNL]
5091 Sleep while stalling if set. This will result
5092 in warnings from preemptible RCU in addition to
5093 any other stall-related activity. Note that
5094 in kernels built with CONFIG_PREEMPTION=n and
5095 CONFIG_PREEMPT_COUNT=y, this parameter will
5096 cause the CPU to pass through a quiescent state.
5097 Given CONFIG_PREEMPTION=n, this will suppress
5098 RCU CPU stall warnings, but will instead result
5099 in scheduling-while-atomic splats.
5101 Use of this module parameter results in splats.
5104 rcutorture.stall_cpu_holdoff= [KNL]
5105 Time to wait (s) after boot before inducing stall.
5107 rcutorture.stall_cpu_irqsoff= [KNL]
5108 Disable interrupts while stalling if set.
5110 rcutorture.stall_gp_kthread= [KNL]
5111 Duration (s) of forced sleep within RCU
5112 grace-period kthread to test RCU CPU stall
5113 warnings, zero to disable. If both stall_cpu
5114 and stall_gp_kthread are specified, the
5115 kthread is starved first, then the CPU.
5117 rcutorture.stat_interval= [KNL]
5118 Time (s) between statistics printk()s.
5120 rcutorture.stutter= [KNL]
5121 Time (s) to stutter testing, for example, specifying
5122 five seconds causes the test to run for five seconds,
5123 wait for five seconds, and so on. This tests RCU's
5124 ability to transition abruptly to and from idle.
5126 rcutorture.test_boost= [KNL]
5127 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
5128 "Maybe" means test if the RCU implementation
5129 under test support RCU priority boosting.
5131 rcutorture.test_boost_duration= [KNL]
5132 Duration (s) of each individual boost test.
5134 rcutorture.test_boost_interval= [KNL]
5135 Interval (s) between each boost test.
5137 rcutorture.test_no_idle_hz= [KNL]
5138 Test RCU's dyntick-idle handling. See also the
5139 rcutorture.shuffle_interval parameter.
5141 rcutorture.torture_type= [KNL]
5142 Specify the RCU implementation to test.
5144 rcutorture.verbose= [KNL]
5145 Enable additional printk() statements.
5147 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
5148 Dump ftrace buffer after reporting RCU CPU
5151 rcupdate.rcu_cpu_stall_suppress= [KNL]
5152 Suppress RCU CPU stall warning messages.
5154 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL]
5155 Suppress RCU CPU stall warning messages and
5156 rcutorture writer stall warnings that occur
5157 during early boot, that is, during the time
5158 before the init task is spawned.
5160 rcupdate.rcu_cpu_stall_timeout= [KNL]
5161 Set timeout for RCU CPU stall warning messages.
5162 The value is in seconds and the maximum allowed
5163 value is 300 seconds.
5165 rcupdate.rcu_exp_cpu_stall_timeout= [KNL]
5166 Set timeout for expedited RCU CPU stall warning
5167 messages. The value is in milliseconds
5168 and the maximum allowed value is 21000
5169 milliseconds. Please note that this value is
5170 adjusted to an arch timer tick resolution.
5171 Setting this to zero causes the value from
5172 rcupdate.rcu_cpu_stall_timeout to be used (after
5173 conversion from seconds to milliseconds).
5175 rcupdate.rcu_cpu_stall_cputime= [KNL]
5176 Provide statistics on the cputime and count of
5177 interrupts and tasks during the sampling period. For
5178 multiple continuous RCU stalls, all sampling periods
5179 begin at half of the first RCU stall timeout.
5181 rcupdate.rcu_exp_stall_task_details= [KNL]
5182 Print stack dumps of any tasks blocking the
5183 current expedited RCU grace period during an
5184 expedited RCU CPU stall warning.
5186 rcupdate.rcu_expedited= [KNL]
5187 Use expedited grace-period primitives, for
5188 example, synchronize_rcu_expedited() instead
5189 of synchronize_rcu(). This reduces latency,
5190 but can increase CPU utilization, degrade
5191 real-time latency, and degrade energy efficiency.
5192 No effect on CONFIG_TINY_RCU kernels.
5194 rcupdate.rcu_normal= [KNL]
5195 Use only normal grace-period primitives,
5196 for example, synchronize_rcu() instead of
5197 synchronize_rcu_expedited(). This improves
5198 real-time latency, CPU utilization, and
5199 energy efficiency, but can expose users to
5200 increased grace-period latency. This parameter
5201 overrides rcupdate.rcu_expedited. No effect on
5202 CONFIG_TINY_RCU kernels.
5204 rcupdate.rcu_normal_after_boot= [KNL]
5205 Once boot has completed (that is, after
5206 rcu_end_inkernel_boot() has been invoked), use
5207 only normal grace-period primitives. No effect
5208 on CONFIG_TINY_RCU kernels.
5210 But note that CONFIG_PREEMPT_RT=y kernels enables
5211 this kernel boot parameter, forcibly setting
5212 it to the value one, that is, converting any
5213 post-boot attempt at an expedited RCU grace
5214 period to instead use normal non-expedited
5215 grace-period processing.
5217 rcupdate.rcu_task_collapse_lim= [KNL]
5218 Set the maximum number of callbacks present
5219 at the beginning of a grace period that allows
5220 the RCU Tasks flavors to collapse back to using
5221 a single callback queue. This switching only
5222 occurs when rcupdate.rcu_task_enqueue_lim is
5223 set to the default value of -1.
5225 rcupdate.rcu_task_contend_lim= [KNL]
5226 Set the minimum number of callback-queuing-time
5227 lock-contention events per jiffy required to
5228 cause the RCU Tasks flavors to switch to per-CPU
5229 callback queuing. This switching only occurs
5230 when rcupdate.rcu_task_enqueue_lim is set to
5231 the default value of -1.
5233 rcupdate.rcu_task_enqueue_lim= [KNL]
5234 Set the number of callback queues to use for the
5235 RCU Tasks family of RCU flavors. The default
5236 of -1 allows this to be automatically (and
5237 dynamically) adjusted. This parameter is intended
5240 rcupdate.rcu_task_ipi_delay= [KNL]
5241 Set time in jiffies during which RCU tasks will
5242 avoid sending IPIs, starting with the beginning
5243 of a given grace period. Setting a large
5244 number avoids disturbing real-time workloads,
5245 but lengthens grace periods.
5247 rcupdate.rcu_task_stall_info= [KNL]
5248 Set initial timeout in jiffies for RCU task stall
5249 informational messages, which give some indication
5250 of the problem for those not patient enough to
5251 wait for ten minutes. Informational messages are
5252 only printed prior to the stall-warning message
5253 for a given grace period. Disable with a value
5254 less than or equal to zero. Defaults to ten
5255 seconds. A change in value does not take effect
5256 until the beginning of the next grace period.
5258 rcupdate.rcu_task_stall_info_mult= [KNL]
5259 Multiplier for time interval between successive
5260 RCU task stall informational messages for a given
5261 RCU tasks grace period. This value is clamped
5262 to one through ten, inclusive. It defaults to
5263 the value three, so that the first informational
5264 message is printed 10 seconds into the grace
5265 period, the second at 40 seconds, the third at
5266 160 seconds, and then the stall warning at 600
5267 seconds would prevent a fourth at 640 seconds.
5269 rcupdate.rcu_task_stall_timeout= [KNL]
5270 Set timeout in jiffies for RCU task stall
5271 warning messages. Disable with a value less
5272 than or equal to zero. Defaults to ten minutes.
5273 A change in value does not take effect until
5274 the beginning of the next grace period.
5276 rcupdate.rcu_self_test= [KNL]
5277 Run the RCU early boot self tests
5281 Run specified binary instead of /init from the ramdisk,
5282 used for early userspace startup. See initrd.
5285 force - Override the decision by the kernel to hide the
5286 advertisement of RDRAND support (this affects
5287 certain AMD processors because of buggy BIOS
5288 support, specifically around the suspend/resume
5292 Turn on/off individual RDT features. List is:
5293 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
5295 E.g. to turn on cmt and turn off mba use:
5299 Format (x86 or x86_64):
5300 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] | d[efault] \
5302 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
5304 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
5305 (prefix with 'panic_' to set mode for panic
5307 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
5308 reboot_force is either force or not specified,
5309 reboot_cpu is s[mp]#### with #### being the processor
5310 to be used for rebooting.
5312 refscale.holdoff= [KNL]
5313 Set test-start holdoff period. The purpose of
5314 this parameter is to delay the start of the
5315 test until boot completes in order to avoid
5318 refscale.loops= [KNL]
5319 Set the number of loops over the synchronization
5320 primitive under test. Increasing this number
5321 reduces noise due to loop start/end overhead,
5322 but the default has already reduced the per-pass
5323 noise to a handful of picoseconds on ca. 2020
5326 refscale.nreaders= [KNL]
5327 Set number of readers. The default value of -1
5328 selects N, where N is roughly 75% of the number
5329 of CPUs. A value of zero is an interesting choice.
5331 refscale.nruns= [KNL]
5332 Set number of runs, each of which is dumped onto
5335 refscale.readdelay= [KNL]
5336 Set the read-side critical-section duration,
5337 measured in microseconds.
5339 refscale.scale_type= [KNL]
5340 Specify the read-protection implementation to test.
5342 refscale.shutdown= [KNL]
5343 Shut down the system at the end of the performance
5344 test. This defaults to 1 (shut it down) when
5345 refscale is built into the kernel and to 0 (leave
5346 it running) when refscale is built as a module.
5348 refscale.verbose= [KNL]
5349 Enable additional printk() statements.
5351 refscale.verbose_batched= [KNL]
5352 Batch the additional printk() statements. If zero
5353 (the default) or negative, print everything. Otherwise,
5354 print every Nth verbose statement, where N is the value
5358 [KNL, SMP] Set scheduler's default relax_domain_level.
5359 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
5361 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
5362 Format: <base1>,<size1>[,<base2>,<size2>,...]
5363 Reserve I/O ports or memory so the kernel won't use
5364 them. If <base> is less than 0x10000, the region
5365 is assumed to be I/O ports; otherwise it is memory.
5367 reservetop= [X86-32]
5369 Reserves a hole at the top of the kernel virtual
5372 reset_devices [KNL] Force drivers to reset the underlying device
5373 during initialization.
5376 Specify the partition device for software suspend
5378 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
5380 resume_offset= [SWSUSP]
5381 Specify the offset from the beginning of the partition
5382 given by "resume=" at which the swap header is located,
5383 in <PAGE_SIZE> units (needed only for swap files).
5384 See Documentation/power/swsusp-and-swap-files.rst
5386 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
5387 read the resume files
5389 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
5390 Useful for devices that are detected asynchronously
5391 (e.g. USB and MMC devices).
5393 retain_initrd [RAM] Keep initrd memory after extraction
5395 retbleed= [X86] Control mitigation of RETBleed (Arbitrary
5396 Speculative Code Execution with Return Instructions)
5399 AMD-based UNRET and IBPB mitigations alone do not stop
5400 sibling threads from influencing the predictions of other
5401 sibling threads. For that reason, STIBP is used on pro-
5402 cessors that support it, and mitigate SMT on processors
5406 auto - automatically select a migitation
5407 auto,nosmt - automatically select a mitigation,
5408 disabling SMT if necessary for
5409 the full mitigation (only on Zen1
5410 and older without STIBP).
5411 ibpb - On AMD, mitigate short speculation
5412 windows on basic block boundaries too.
5413 Safe, highest perf impact. It also
5414 enables STIBP if present. Not suitable
5416 ibpb,nosmt - Like "ibpb" above but will disable SMT
5417 when STIBP is not available. This is
5418 the alternative for systems which do not
5420 unret - Force enable untrained return thunks,
5421 only effective on AMD f15h-f17h based
5423 unret,nosmt - Like unret, but will disable SMT when STIBP
5424 is not available. This is the alternative for
5425 systems which do not have STIBP.
5427 Selecting 'auto' will choose a mitigation method at run
5428 time according to the CPU.
5430 Not specifying this option is equivalent to retbleed=auto.
5432 rfkill.default_state=
5433 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
5434 etc. communication is blocked by default.
5437 rfkill.master_switch_mode=
5438 0 The "airplane mode" button does nothing.
5439 1 The "airplane mode" button toggles between everything
5440 blocked and the previous configuration.
5441 2 The "airplane mode" button toggles between everything
5442 blocked and everything unblocked.
5444 rhash_entries= [KNL,NET]
5445 Set number of hash buckets for route cache
5448 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
5451 ro [KNL] Mount root device read-only on boot
5454 on Mark read-only kernel memory as read-only (default).
5455 off Leave read-only kernel memory writable for debugging.
5456 full Mark read-only kernel memory and aliases as read-only
5460 Enable the uart passthrough on the designated usb port
5461 on Rockchip SoCs. When active, the signals of the
5462 debug-uart get routed to the D+ and D- pins of the usb
5463 port and the regular usb controller gets disabled.
5465 root= [KNL] Root filesystem
5466 Usually this a a block device specifier of some kind,
5467 see the early_lookup_bdev comment in
5468 block/early-lookup.c for details.
5469 Alternatively this can be "ram" for the legacy initial
5470 ramdisk, "nfs" and "cifs" for root on a network file
5471 system, or "mtd" and "ubi" for mounting from raw flash.
5473 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
5474 mount the root filesystem
5476 rootflags= [KNL] Set root filesystem mount option string
5478 rootfstype= [KNL] Set root filesystem type
5480 rootwait [KNL] Wait (indefinitely) for root device to show up.
5481 Useful for devices that are detected asynchronously
5482 (e.g. USB and MMC devices).
5484 rproc_mem=nn[KMG][@address]
5485 [KNL,ARM,CMA] Remoteproc physical memory block.
5486 Memory area to be used by remote processor image,
5489 rw [KNL] Mount root device read-write on boot
5491 S [KNL] Run init in single mode
5493 s390_iommu= [HW,S390]
5494 Set s390 IOTLB flushing mode
5496 With strict flushing every unmap operation will result in
5497 an IOTLB flush. Default is lazy flushing before reuse,
5500 s390_iommu_aperture= [KNL,S390]
5501 Specifies the size of the per device DMA address space
5502 accessible through the DMA and IOMMU APIs as a decimal
5503 factor of the size of main memory.
5504 The default is 1 meaning that one can concurrently use
5505 as many DMA addresses as physical memory is installed,
5506 if supported by hardware, and thus map all of memory
5507 once. With a value of 2 one can map all of memory twice
5508 and so on. As a special case a factor of 0 imposes no
5509 restrictions other than those given by hardware at the
5510 cost of significant additional memory use for tables.
5513 See drivers/net/irda/sa1100_ir.c.
5515 sched_verbose [KNL] Enables verbose scheduler debug messages.
5517 schedstats= [KNL,X86] Enable or disable scheduled statistics.
5518 Allowed values are enable and disable. This feature
5519 incurs a small amount of overhead in the scheduler
5520 but is useful for debugging and performance tuning.
5522 sched_thermal_decay_shift=
5523 [KNL, SMP] Set a decay shift for scheduler thermal
5524 pressure signal. Thermal pressure signal follows the
5525 default decay period of other scheduler pelt
5526 signals(usually 32 ms but configurable). Setting
5527 sched_thermal_decay_shift will left shift the decay
5528 period for the thermal pressure signal by the shift
5530 i.e. with the default pelt decay period of 32 ms
5531 sched_thermal_decay_shift thermal pressure decay pr
5535 Format: integer between 0 and 10
5538 scftorture.holdoff= [KNL]
5539 Number of seconds to hold off before starting
5540 test. Defaults to zero for module insertion and
5541 to 10 seconds for built-in smp_call_function()
5544 scftorture.longwait= [KNL]
5545 Request ridiculously long waits randomly selected
5546 up to the chosen limit in seconds. Zero (the
5547 default) disables this feature. Please note
5548 that requesting even small non-zero numbers of
5549 seconds can result in RCU CPU stall warnings,
5550 softlockup complaints, and so on.
5552 scftorture.nthreads= [KNL]
5553 Number of kthreads to spawn to invoke the
5554 smp_call_function() family of functions.
5555 The default of -1 specifies a number of kthreads
5556 equal to the number of CPUs.
5558 scftorture.onoff_holdoff= [KNL]
5559 Number seconds to wait after the start of the
5560 test before initiating CPU-hotplug operations.
5562 scftorture.onoff_interval= [KNL]
5563 Number seconds to wait between successive
5564 CPU-hotplug operations. Specifying zero (which
5565 is the default) disables CPU-hotplug operations.
5567 scftorture.shutdown_secs= [KNL]
5568 The number of seconds following the start of the
5569 test after which to shut down the system. The
5570 default of zero avoids shutting down the system.
5571 Non-zero values are useful for automated tests.
5573 scftorture.stat_interval= [KNL]
5574 The number of seconds between outputting the
5575 current test statistics to the console. A value
5576 of zero disables statistics output.
5578 scftorture.stutter_cpus= [KNL]
5579 The number of jiffies to wait between each change
5580 to the set of CPUs under test.
5582 scftorture.use_cpus_read_lock= [KNL]
5583 Use use_cpus_read_lock() instead of the default
5584 preempt_disable() to disable CPU hotplug
5585 while invoking one of the smp_call_function*()
5588 scftorture.verbose= [KNL]
5589 Enable additional printk() statements.
5591 scftorture.weight_single= [KNL]
5592 The probability weighting to use for the
5593 smp_call_function_single() function with a zero
5594 "wait" parameter. A value of -1 selects the
5595 default if all other weights are -1. However,
5596 if at least one weight has some other value, a
5597 value of -1 will instead select a weight of zero.
5599 scftorture.weight_single_wait= [KNL]
5600 The probability weighting to use for the
5601 smp_call_function_single() function with a
5602 non-zero "wait" parameter. See weight_single.
5604 scftorture.weight_many= [KNL]
5605 The probability weighting to use for the
5606 smp_call_function_many() function with a zero
5607 "wait" parameter. See weight_single.
5608 Note well that setting a high probability for
5609 this weighting can place serious IPI load
5612 scftorture.weight_many_wait= [KNL]
5613 The probability weighting to use for the
5614 smp_call_function_many() function with a
5615 non-zero "wait" parameter. See weight_single
5618 scftorture.weight_all= [KNL]
5619 The probability weighting to use for the
5620 smp_call_function_all() function with a zero
5621 "wait" parameter. See weight_single and
5624 scftorture.weight_all_wait= [KNL]
5625 The probability weighting to use for the
5626 smp_call_function_all() function with a
5627 non-zero "wait" parameter. See weight_single
5630 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
5631 xtime_lock contention on larger systems, and/or RCU lock
5632 contention on all systems with CONFIG_MAXSMP set.
5633 Format: { "0" | "1" }
5634 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
5636 Note: increases power consumption, thus should only be
5637 enabled if running jitter sensitive (HPC/RT) workloads.
5639 security= [SECURITY] Choose a legacy "major" security module to
5640 enable at boot. This has been deprecated by the
5643 selinux= [SELINUX] Disable or enable SELinux at boot time.
5644 Format: { "0" | "1" }
5645 See security/selinux/Kconfig help text.
5650 serialnumber [BUGS=X86-32]
5652 sev=option[,option...] [X86-64] See Documentation/arch/x86/x86_64/boot-options.rst
5655 Maximal number of shapers.
5657 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
5658 Limit apic dumping. The parameter defines the maximal
5659 number of local apics being dumped. Also it is possible
5660 to set it to "all" by meaning -- no limit here.
5661 Format: { 1 (default) | 2 | ... | all }.
5662 The parameter valid if only apic=debug or
5663 apic=verbose is specified.
5664 Example: apic=debug show_lapic=all
5672 Enable merging of slabs with similar size when the
5673 kernel is built without CONFIG_SLAB_MERGE_DEFAULT.
5676 Disable merging of slabs with similar size. May be
5677 necessary if there is some reason to distinguish
5678 allocs to different slabs, especially in hardened
5679 environments where the risk of heap overflows and
5680 layout control by attackers can usually be
5681 frustrated by disabling merging. This will reduce
5682 most of the exposure of a heap attack to a single
5683 cache (risks via metadata attacks are mostly
5684 unchanged). Debug options disable merging on their
5686 For more information see Documentation/mm/slub.rst.
5688 slab_max_order= [MM, SLAB]
5689 Determines the maximum allowed order for slabs.
5690 A high setting may cause OOMs due to memory
5691 fragmentation. Defaults to 1 for systems with
5692 more than 32MB of RAM, 0 otherwise.
5694 slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB]
5695 Enabling slub_debug allows one to determine the
5696 culprit if slab objects become corrupted. Enabling
5697 slub_debug can create guard zones around objects and
5698 may poison objects when not in use. Also tracks the
5699 last alloc / free. For more information see
5700 Documentation/mm/slub.rst.
5702 slub_max_order= [MM, SLUB]
5703 Determines the maximum allowed order for slabs.
5704 A high setting may cause OOMs due to memory
5705 fragmentation. For more information see
5706 Documentation/mm/slub.rst.
5708 slub_min_objects= [MM, SLUB]
5709 The minimum number of objects per slab. SLUB will
5710 increase the slab order up to slub_max_order to
5711 generate a sufficiently large slab able to contain
5712 the number of objects indicated. The higher the number
5713 of objects the smaller the overhead of tracking slabs
5714 and the less frequently locks need to be acquired.
5715 For more information see Documentation/mm/slub.rst.
5717 slub_min_order= [MM, SLUB]
5718 Determines the minimum page order for slabs. Must be
5719 lower than slub_max_order.
5720 For more information see Documentation/mm/slub.rst.
5722 slub_merge [MM, SLUB]
5723 Same with slab_merge.
5725 slub_nomerge [MM, SLUB]
5726 Same with slab_nomerge. This is supported for legacy.
5727 See slab_nomerge for more information.
5730 Format: <io1>[,<io2>[,...,<io8>]]
5732 smp.csd_lock_timeout= [KNL]
5733 Specify the period of time in milliseconds
5734 that smp_call_function() and friends will wait
5735 for a CPU to release the CSD lock. This is
5736 useful when diagnosing bugs involving CPUs
5737 disabling interrupts for extended periods
5738 of time. Defaults to 5,000 milliseconds, and
5739 setting a value of zero disables this feature.
5740 This feature may be more efficiently disabled
5741 using the csdlock_debug- kernel parameter.
5743 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
5744 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
5745 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
5746 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
5747 smsc-ircc2.ircc_irq= [HW] IRQ line
5748 smsc-ircc2.ircc_dma= [HW] DMA channel
5749 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
5750 0: Toshiba Satellite 1800 (GP data pin select)
5751 1: Fast pin select (default)
5754 smt= [KNL,S390] Set the maximum number of threads (logical
5755 CPUs) to use per physical CPU on systems capable of
5756 symmetric multithreading (SMT). Will be capped to the
5757 actual hardware limit.
5759 Default: -1 (no limit)
5762 [KNL] Should the soft-lockup detector generate panics.
5765 A value of 1 instructs the soft-lockup detector
5766 to panic the machine when a soft-lockup occurs. It is
5767 also controlled by the kernel.softlockup_panic sysctl
5768 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
5769 respective build-time switch to that functionality.
5771 softlockup_all_cpu_backtrace=
5772 [KNL] Should the soft-lockup detector generate
5773 backtraces on all cpus.
5776 sonypi.*= [HW] Sony Programmable I/O Control Device driver
5777 See Documentation/admin-guide/laptops/sonypi.rst
5779 spectre_v2= [X86] Control mitigation of Spectre variant 2
5780 (indirect branch speculation) vulnerability.
5781 The default operation protects the kernel from
5784 on - unconditionally enable, implies
5786 off - unconditionally disable, implies
5788 auto - kernel detects whether your CPU model is
5791 Selecting 'on' will, and 'auto' may, choose a
5792 mitigation method at run time according to the
5793 CPU, the available microcode, the setting of the
5794 CONFIG_RETPOLINE configuration option, and the
5795 compiler with which the kernel was built.
5797 Selecting 'on' will also enable the mitigation
5798 against user space to user space task attacks.
5800 Selecting 'off' will disable both the kernel and
5801 the user space protections.
5803 Specific mitigations can also be selected manually:
5805 retpoline - replace indirect branches
5806 retpoline,generic - Retpolines
5807 retpoline,lfence - LFENCE; indirect branch
5808 retpoline,amd - alias for retpoline,lfence
5809 eibrs - Enhanced/Auto IBRS
5810 eibrs,retpoline - Enhanced/Auto IBRS + Retpolines
5811 eibrs,lfence - Enhanced/Auto IBRS + LFENCE
5812 ibrs - use IBRS to protect kernel
5814 Not specifying this option is equivalent to
5818 [X86] Control mitigation of Spectre variant 2
5819 (indirect branch speculation) vulnerability between
5822 on - Unconditionally enable mitigations. Is
5823 enforced by spectre_v2=on
5825 off - Unconditionally disable mitigations. Is
5826 enforced by spectre_v2=off
5828 prctl - Indirect branch speculation is enabled,
5829 but mitigation can be enabled via prctl
5830 per thread. The mitigation control state
5831 is inherited on fork.
5834 - Like "prctl" above, but only STIBP is
5835 controlled per thread. IBPB is issued
5836 always when switching between different user
5840 - Same as "prctl" above, but all seccomp
5841 threads will enable the mitigation unless
5842 they explicitly opt out.
5845 - Like "seccomp" above, but only STIBP is
5846 controlled per thread. IBPB is issued
5847 always when switching between different
5848 user space processes.
5850 auto - Kernel selects the mitigation depending on
5851 the available CPU features and vulnerability.
5853 Default mitigation: "prctl"
5855 Not specifying this option is equivalent to
5856 spectre_v2_user=auto.
5858 spec_store_bypass_disable=
5859 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
5860 (Speculative Store Bypass vulnerability)
5862 Certain CPUs are vulnerable to an exploit against a
5863 a common industry wide performance optimization known
5864 as "Speculative Store Bypass" in which recent stores
5865 to the same memory location may not be observed by
5866 later loads during speculative execution. The idea
5867 is that such stores are unlikely and that they can
5868 be detected prior to instruction retirement at the
5869 end of a particular speculation execution window.
5871 In vulnerable processors, the speculatively forwarded
5872 store can be used in a cache side channel attack, for
5873 example to read memory to which the attacker does not
5874 directly have access (e.g. inside sandboxed code).
5876 This parameter controls whether the Speculative Store
5877 Bypass optimization is used.
5879 On x86 the options are:
5881 on - Unconditionally disable Speculative Store Bypass
5882 off - Unconditionally enable Speculative Store Bypass
5883 auto - Kernel detects whether the CPU model contains an
5884 implementation of Speculative Store Bypass and
5885 picks the most appropriate mitigation. If the
5886 CPU is not vulnerable, "off" is selected. If the
5887 CPU is vulnerable the default mitigation is
5888 architecture and Kconfig dependent. See below.
5889 prctl - Control Speculative Store Bypass per thread
5890 via prctl. Speculative Store Bypass is enabled
5891 for a process by default. The state of the control
5892 is inherited on fork.
5893 seccomp - Same as "prctl" above, but all seccomp threads
5894 will disable SSB unless they explicitly opt out.
5896 Default mitigations:
5899 On powerpc the options are:
5901 on,auto - On Power8 and Power9 insert a store-forwarding
5902 barrier on kernel entry and exit. On Power7
5903 perform a software flush on kernel entry and
5907 Not specifying this option is equivalent to
5908 spec_store_bypass_disable=auto.
5910 spia_io_base= [HW,MTD]
5916 [X86] Enable split lock detection or bus lock detection
5918 When enabled (and if hardware support is present), atomic
5919 instructions that access data across cache line
5920 boundaries will result in an alignment check exception
5921 for split lock detection or a debug exception for
5926 warn - the kernel will emit rate-limited warnings
5927 about applications triggering the #AC
5928 exception or the #DB exception. This mode is
5929 the default on CPUs that support split lock
5930 detection or bus lock detection. Default
5931 behavior is by #AC if both features are
5932 enabled in hardware.
5934 fatal - the kernel will send SIGBUS to applications
5935 that trigger the #AC exception or the #DB
5936 exception. Default behavior is by #AC if
5937 both features are enabled in hardware.
5940 Set system wide rate limit to N bus locks
5941 per second for bus lock detection.
5944 N/A for split lock detection.
5947 If an #AC exception is hit in the kernel or in
5948 firmware (i.e. not while executing in user mode)
5949 the kernel will oops in either "warn" or "fatal"
5952 #DB exception for bus lock is triggered only when
5956 Control the Special Register Buffer Data Sampling
5959 Certain CPUs are vulnerable to an MDS-like
5960 exploit which can leak bits from the random
5963 By default, this issue is mitigated by
5964 microcode. However, the microcode fix can cause
5965 the RDRAND and RDSEED instructions to become
5966 much slower. Among other effects, this will
5967 result in reduced throughput from /dev/urandom.
5969 The microcode mitigation can be disabled with
5970 the following option:
5972 off: Disable mitigation and remove
5973 performance impact to RDRAND and RDSEED
5975 srcutree.big_cpu_lim [KNL]
5976 Specifies the number of CPUs constituting a
5977 large system, such that srcu_struct structures
5978 should immediately allocate an srcu_node array.
5979 This kernel-boot parameter defaults to 128,
5980 but takes effect only when the low-order four
5981 bits of srcutree.convert_to_big is equal to 3
5984 srcutree.convert_to_big [KNL]
5985 Specifies under what conditions an SRCU tree
5986 srcu_struct structure will be converted to big
5987 form, that is, with an rcu_node tree:
5990 1: At init_srcu_struct() time.
5991 2: When rcutorture decides to.
5992 3: Decide at boot time (default).
5993 0x1X: Above plus if high contention.
5995 Either way, the srcu_node tree will be sized based
5996 on the actual runtime number of CPUs (nr_cpu_ids)
5997 instead of the compile-time CONFIG_NR_CPUS.
5999 srcutree.counter_wrap_check [KNL]
6000 Specifies how frequently to check for
6001 grace-period sequence counter wrap for the
6002 srcu_data structure's ->srcu_gp_seq_needed field.
6003 The greater the number of bits set in this kernel
6004 parameter, the less frequently counter wrap will
6005 be checked for. Note that the bottom two bits
6008 srcutree.exp_holdoff [KNL]
6009 Specifies how many nanoseconds must elapse
6010 since the end of the last SRCU grace period for
6011 a given srcu_struct until the next normal SRCU
6012 grace period will be considered for automatic
6013 expediting. Set to zero to disable automatic
6016 srcutree.srcu_max_nodelay [KNL]
6017 Specifies the number of no-delay instances
6018 per jiffy for which the SRCU grace period
6019 worker thread will be rescheduled with zero
6020 delay. Beyond this limit, worker thread will
6021 be rescheduled with a sleep delay of one jiffy.
6023 srcutree.srcu_max_nodelay_phase [KNL]
6024 Specifies the per-grace-period phase, number of
6025 non-sleeping polls of readers. Beyond this limit,
6026 grace period worker thread will be rescheduled
6027 with a sleep delay of one jiffy, between each
6028 rescan of the readers, for a grace period phase.
6030 srcutree.srcu_retry_check_delay [KNL]
6031 Specifies number of microseconds of non-sleeping
6032 delay between each non-sleeping poll of readers.
6034 srcutree.small_contention_lim [KNL]
6035 Specifies the number of update-side contention
6036 events per jiffy will be tolerated before
6037 initiating a conversion of an srcu_struct
6038 structure to big form. Note that the value of
6039 srcutree.convert_to_big must have the 0x10 bit
6040 set for contention-based conversions to occur.
6043 Speculative Store Bypass Disable control
6045 On CPUs that are vulnerable to the Speculative
6046 Store Bypass vulnerability and offer a
6047 firmware based mitigation, this parameter
6048 indicates how the mitigation should be used:
6050 force-on: Unconditionally enable mitigation for
6051 for both kernel and userspace
6052 force-off: Unconditionally disable mitigation for
6053 for both kernel and userspace
6054 kernel: Always enable mitigation in the
6055 kernel, and offer a prctl interface
6056 to allow userspace to register its
6057 interest in being mitigated too.
6059 stack_guard_gap= [MM]
6060 override the default stack gap protection. The value
6061 is in page units and it defines how many pages prior
6062 to (for stacks growing down) resp. after (for stacks
6063 growing up) the main stack are reserved for no other
6064 mapping. Default value is 256 pages.
6066 stack_depot_disable= [KNL]
6067 Setting this to true through kernel command line will
6068 disable the stack depot thereby saving the static memory
6069 consumed by the stack hash table. By default this is set
6073 Enabled the stack tracer on boot up.
6075 stacktrace_filter=[function-list]
6076 [FTRACE] Limit the functions that the stack tracer
6077 will trace at boot up. function-list is a comma-separated
6078 list of functions. This list can be changed at run
6079 time by the stack_trace_filter file in the debugfs
6080 tracing directory. Note, this enables stack tracing
6081 and the stacktrace above is not needed.
6085 Set the STI (builtin display/keyboard on the HP-PARISC
6086 machines) console (graphic card) which should be used
6087 as the initial boot-console.
6088 See also comment in drivers/video/console/sticore.c.
6091 See comment in drivers/video/console/sticore.c.
6094 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
6099 Enable or disable strict sigaltstack size checks
6100 against the required signal frame size which
6101 depends on the supported FPU features. This can
6102 be used to filter out binaries which have
6103 not yet been made aware of AT_MINSIGSTKSZ.
6106 Limits the number of kernel HPT entries in the hash
6107 page table to increase the rate of hash page table
6108 faults on kernel addresses.
6111 Limits the number of kernel SLB entries, and flushes
6112 them frequently to increase the rate of SLB faults
6113 on kernel addresses.
6115 sunrpc.min_resvport=
6116 sunrpc.max_resvport=
6118 SunRPC servers often require that client requests
6119 originate from a privileged port (i.e. a port in the
6120 range 0 < portnr < 1024).
6121 An administrator who wishes to reserve some of these
6122 ports for other uses may adjust the range that the
6123 kernel's sunrpc client considers to be privileged
6124 using these two parameters to set the minimum and
6125 maximum port values.
6127 sunrpc.svc_rpc_per_connection_limit=
6129 Limit the number of requests that the server will
6130 process in parallel from a single connection.
6131 The default value is 0 (no limit).
6135 Control how the NFS server code allocates CPUs to
6136 service thread pools. Depending on how many NICs
6137 you have and where their interrupts are bound, this
6138 option will affect which CPUs will do NFS serving.
6139 Note: this parameter cannot be changed while the
6140 NFS server is running.
6142 auto the server chooses an appropriate mode
6143 automatically using heuristics
6144 global a single global pool contains all CPUs
6145 percpu one pool for each CPU
6146 pernode one pool for each NUMA node (equivalent
6147 to global on non-NUMA machines)
6149 sunrpc.tcp_slot_table_entries=
6150 sunrpc.udp_slot_table_entries=
6152 Sets the upper limit on the number of simultaneous
6153 RPC calls that can be sent from the client to a
6154 server. Increasing these values may allow you to
6155 improve throughput, but will also increase the
6156 amount of memory reserved for use by the client.
6158 suspend.pm_test_delay=
6160 Sets the number of seconds to remain in a suspend test
6161 mode before resuming the system (see
6162 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
6163 is set. Default value is 5.
6166 Format: { on | off | y | n | 1 | 0 }
6167 This parameter controls use of the Protected
6168 Execution Facility on pSeries.
6170 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
6171 Format: { <int> [,<int>] | force | noforce }
6172 <int> -- Number of I/O TLB slabs
6173 <int> -- Second integer after comma. Number of swiotlb
6174 areas with their own lock. Will be rounded up
6176 force -- force using of bounce buffers even if they
6177 wouldn't be automatically used by the kernel
6178 noforce -- Never use bounce buffers (for debugging)
6183 Set a sysctl parameter, right before loading the init
6184 process, as if the value was written to the respective
6185 /proc/sys/... file. Both '.' and '/' are recognized as
6186 separators. Unrecognized parameters and invalid values
6187 are reported in the kernel log. Sysctls registered
6188 later by a loaded module cannot be set this way.
6189 Example: sysctl.vm.swappiness=40
6191 sysrq_always_enabled
6193 Ignore sysrq setting - this boot parameter will
6194 neutralize any effect of /proc/sys/kernel/sysrq.
6195 Useful for debugging.
6197 tcpmhash_entries= [KNL,NET]
6198 Set the number of tcp_metrics_hash slots.
6199 Default value is 8192 or 16384 depending on total
6200 ram pages. This is used to specify the TCP metrics
6201 cache size. See Documentation/networking/ip-sysctl.rst
6202 "tcp_no_metrics_save" section for more details.
6206 test_suspend= [SUSPEND]
6207 Format: { "mem" | "standby" | "freeze" }[,N]
6208 Specify "mem" (for Suspend-to-RAM) or "standby" (for
6209 standby suspend) or "freeze" (for suspend type freeze)
6210 as the system sleep state during system startup with
6211 the optional capability to repeat N number of times.
6212 The system is woken from this state using a
6213 wakeup-capable RTC alarm.
6215 thash_entries= [KNL,NET]
6216 Set number of hash buckets for TCP connection
6218 thermal.act= [HW,ACPI]
6219 -1: disable all active trip points in all thermal zones
6220 <degrees C>: override all lowest active trip points
6222 thermal.crt= [HW,ACPI]
6223 -1: disable all critical trip points in all thermal zones
6224 <degrees C>: override all critical trip points
6226 thermal.nocrt= [HW,ACPI]
6227 Set to disable actions on ACPI thermal zone
6228 critical and hot trip points.
6230 thermal.off= [HW,ACPI]
6231 1: disable ACPI thermal control
6233 thermal.psv= [HW,ACPI]
6234 -1: disable all passive trip points
6235 <degrees C>: override all passive trip points to this
6238 thermal.tzp= [HW,ACPI]
6239 Specify global default ACPI thermal zone polling rate
6240 <deci-seconds>: poll all this frequency
6241 0: no polling (default)
6244 Force threading of all interrupt handlers except those
6245 marked explicitly IRQF_NO_THREAD.
6249 Specify if the kernel should make use of the cpu
6250 topology information if the hardware supports this.
6251 The scheduler will make use of this information and
6252 e.g. base its process migration decisions on it.
6255 topology_updates= [KNL, PPC, NUMA]
6257 Specify if the kernel should ignore (off)
6258 topology updates sent by the hypervisor to this
6261 torture.disable_onoff_at_boot= [KNL]
6262 Prevent the CPU-hotplug component of torturing
6263 until after init has spawned.
6265 torture.ftrace_dump_at_shutdown= [KNL]
6266 Dump the ftrace buffer at torture-test shutdown,
6267 even if there were no errors. This can be a
6268 very costly operation when many torture tests
6269 are running concurrently, especially on systems
6270 with rotating-rust storage.
6272 torture.verbose_sleep_frequency= [KNL]
6273 Specifies how many verbose printk()s should be
6274 emitted between each sleep. The default of zero
6275 disables verbose-printk() sleeping.
6277 torture.verbose_sleep_duration= [KNL]
6278 Duration of each verbose-printk() sleep in jiffies.
6282 tpm_suspend_pcr=[HW,TPM]
6283 Format: integer pcr id
6284 Specify that at suspend time, the tpm driver
6285 should extend the specified pcr with zeros,
6286 as a workaround for some chips which fail to
6287 flush the last written pcr on TPM_SaveState.
6288 This will guarantee that all the other pcrs
6292 Have the tracepoints sent to printk as well as the
6293 tracing ring buffer. This is useful for early boot up
6294 where the system hangs or reboots and does not give the
6295 option for reading the tracing buffer or performing a
6296 ftrace_dump_on_oops.
6298 To turn off having tracepoints sent to printk,
6299 echo 0 > /proc/sys/kernel/tracepoint_printk
6300 Note, echoing 1 into this file without the
6301 tracepoint_printk kernel cmdline option has no effect.
6303 The tp_printk_stop_on_boot (see below) can also be used
6304 to stop the printing of events to console at
6309 Having tracepoints sent to printk() and activating high
6310 frequency tracepoints such as irq or sched, can cause
6311 the system to live lock.
6313 tp_printk_stop_on_boot [FTRACE]
6314 When tp_printk (above) is set, it can cause a lot of noise
6315 on the console. It may be useful to only include the
6316 printing of events during boot up, as user space may
6317 make the system inoperable.
6319 This command line option will stop the printing of events
6320 to console at the late_initcall_sync() time frame.
6322 trace_buf_size=nn[KMG]
6323 [FTRACE] will set tracing buffer size on each cpu.
6325 trace_clock= [FTRACE] Set the clock used for tracing events
6327 local - Use the per CPU time stamp counter
6328 (converted into nanoseconds). Fast, but
6329 depending on the architecture, may not be
6330 in sync between CPUs.
6331 global - Event time stamps are synchronize across
6332 CPUs. May be slower than the local clock,
6333 but better for some race conditions.
6334 counter - Simple counting of events (1, 2, ..)
6335 note, some counts may be skipped due to the
6336 infrastructure grabbing the clock more than
6338 uptime - Use jiffies as the time stamp.
6339 perf - Use the same clock that perf uses.
6340 mono - Use ktime_get_mono_fast_ns() for time stamps.
6341 mono_raw - Use ktime_get_raw_fast_ns() for time
6343 boot - Use ktime_get_boot_fast_ns() for time stamps.
6344 Architectures may add more clocks. See
6345 Documentation/trace/ftrace.rst for more details.
6347 trace_event=[event-list]
6348 [FTRACE] Set and start specified trace events in order
6349 to facilitate early boot debugging. The event-list is a
6350 comma-separated list of trace events to enable. See
6351 also Documentation/trace/events.rst
6353 trace_instance=[instance-info]
6354 [FTRACE] Create a ring buffer instance early in boot up.
6355 This will be listed in:
6357 /sys/kernel/tracing/instances
6359 Events can be enabled at the time the instance is created
6362 trace_instance=<name>,<system1>:<event1>,<system2>:<event2>
6364 Note, the "<system*>:" portion is optional if the event is
6367 trace_instance=foo,sched:sched_switch,irq_handler_entry,initcall
6369 will enable the "sched_switch" event (note, the "sched:" is optional, and
6370 the same thing would happen if it was left off). The irq_handler_entry
6371 event, and all events under the "initcall" system.
6373 trace_options=[option-list]
6374 [FTRACE] Enable or disable tracer options at boot.
6375 The option-list is a comma delimited list of options
6376 that can be enabled or disabled just as if you were
6377 to echo the option name into
6379 /sys/kernel/tracing/trace_options
6381 For example, to enable stacktrace option (to dump the
6382 stack trace of each event), add to the command line:
6384 trace_options=stacktrace
6386 See also Documentation/trace/ftrace.rst "trace options"
6389 trace_trigger=[trigger-list]
6390 [FTRACE] Add a event trigger on specific events.
6391 Set a trigger on top of a specific event, with an optional
6394 The format is is "trace_trigger=<event>.<trigger>[ if <filter>],..."
6395 Where more than one trigger may be specified that are comma deliminated.
6399 trace_trigger="sched_switch.stacktrace if prev_state == 2"
6401 The above will enable the "stacktrace" trigger on the "sched_switch"
6402 event but only trigger it if the "prev_state" of the "sched_switch"
6403 event is "2" (TASK_UNINTERUPTIBLE).
6405 See also "Event triggers" in Documentation/trace/events.rst
6409 [FTRACE] enable this option to disable tracing when a
6410 warning is hit. This turns off "tracing_on". Tracing can
6411 be enabled again by echoing '1' into the "tracing_on"
6412 file located in /sys/kernel/tracing/
6414 This option is useful, as it disables the trace before
6415 the WARNING dump is called, which prevents the trace to
6416 be filled with content caused by the warning output.
6418 This option can also be set at run time via the sysctl
6419 option: kernel/traceoff_on_warning
6421 transparent_hugepage=
6423 Format: [always|madvise|never]
6424 Can be used to control the default behavior of the system
6425 with respect to transparent hugepages.
6426 See Documentation/admin-guide/mm/transhuge.rst
6429 trusted.source= [KEYS]
6431 This parameter identifies the trust source as a backend
6432 for trusted keys implementation. Supported trust
6437 If not specified then it defaults to iterating through
6438 the trust source list starting with TPM and assigns the
6439 first trust source as a backend which is initialized
6440 successfully during iteration.
6444 The RNG used to generate key material for trusted keys.
6447 - the same value as trusted.source: "tpm" or "tee"
6449 If not specified, "default" is used. In this case,
6450 the RNG's choice is left to each individual trust source.
6452 tsc= Disable clocksource stability checks for TSC.
6454 [x86] reliable: mark tsc clocksource as reliable, this
6455 disables clocksource verification at runtime, as well
6456 as the stability checks done at bootup. Used to enable
6457 high-resolution timer mode on older hardware, and in
6458 virtualized environment.
6459 [x86] noirqtime: Do not use TSC to do irq accounting.
6460 Used to run time disable IRQ_TIME_ACCOUNTING on any
6461 platforms where RDTSC is slow and this accounting
6463 [x86] unstable: mark the TSC clocksource as unstable, this
6464 marks the TSC unconditionally unstable at bootup and
6465 avoids any further wobbles once the TSC watchdog notices.
6466 [x86] nowatchdog: disable clocksource watchdog. Used
6467 in situations with strict latency requirements (where
6468 interruptions from clocksource watchdog are not
6470 [x86] recalibrate: force recalibration against a HW timer
6471 (HPET or PM timer) on systems whose TSC frequency was
6472 obtained from HW or FW using either an MSR or CPUID(0x15).
6473 Warn if the difference is more than 500 ppm.
6474 [x86] watchdog: Use TSC as the watchdog clocksource with
6475 which to check other HW timers (HPET or PM timer), but
6476 only on systems where TSC has been deemed trustworthy.
6477 This will be suppressed by an earlier tsc=nowatchdog and
6478 can be overridden by a later tsc=nowatchdog. A console
6479 message will flag any such suppression or overriding.
6481 tsc_early_khz= [X86] Skip early TSC calibration and use the given
6482 value instead. Useful when the early TSC frequency discovery
6483 procedure is not reliable, such as on overclocked systems
6484 with CPUID.16h support and partial CPUID.15h support.
6485 Format: <unsigned int>
6487 tsx= [X86] Control Transactional Synchronization
6488 Extensions (TSX) feature in Intel processors that
6489 support TSX control.
6491 This parameter controls the TSX feature. The options are:
6493 on - Enable TSX on the system. Although there are
6494 mitigations for all known security vulnerabilities,
6495 TSX has been known to be an accelerator for
6496 several previous speculation-related CVEs, and
6497 so there may be unknown security risks associated
6498 with leaving it enabled.
6500 off - Disable TSX on the system. (Note that this
6501 option takes effect only on newer CPUs which are
6502 not vulnerable to MDS, i.e., have
6503 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
6504 the new IA32_TSX_CTRL MSR through a microcode
6505 update. This new MSR allows for the reliable
6506 deactivation of the TSX functionality.)
6508 auto - Disable TSX if X86_BUG_TAA is present,
6509 otherwise enable TSX on the system.
6511 Not specifying this option is equivalent to tsx=off.
6513 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6516 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
6517 Abort (TAA) vulnerability.
6519 Similar to Micro-architectural Data Sampling (MDS)
6520 certain CPUs that support Transactional
6521 Synchronization Extensions (TSX) are vulnerable to an
6522 exploit against CPU internal buffers which can forward
6523 information to a disclosure gadget under certain
6526 In vulnerable processors, the speculatively forwarded
6527 data can be used in a cache side channel attack, to
6528 access data to which the attacker does not have direct
6531 This parameter controls the TAA mitigation. The
6534 full - Enable TAA mitigation on vulnerable CPUs
6537 full,nosmt - Enable TAA mitigation and disable SMT on
6538 vulnerable CPUs. If TSX is disabled, SMT
6539 is not disabled because CPU is not
6540 vulnerable to cross-thread TAA attacks.
6541 off - Unconditionally disable TAA mitigation
6543 On MDS-affected machines, tsx_async_abort=off can be
6544 prevented by an active MDS mitigation as both vulnerabilities
6545 are mitigated with the same mechanism so in order to disable
6546 this mitigation, you need to specify mds=off too.
6548 Not specifying this option is equivalent to
6549 tsx_async_abort=full. On CPUs which are MDS affected
6550 and deploy MDS mitigation, TAA mitigation is not
6551 required and doesn't provide any additional
6555 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6557 turbografx.map[2|3]= [HW,JOY]
6558 TurboGraFX parallel port interface
6560 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
6561 See also Documentation/input/devices/joystick-parport.rst
6563 udbg-immortal [PPC] When debugging early kernel crashes that
6564 happen after console_init() and before a proper
6565 console driver takes over, this boot options might
6566 help "seeing" what's going on.
6568 uhash_entries= [KNL,NET]
6569 Set number of hash buckets for UDP/UDP-Lite connections
6572 [USB] Ignore overcurrent events (default N).
6573 Some badly-designed motherboards generate lots of
6574 bogus events, for ports that aren't wired to
6575 anything. Set this parameter to avoid log spamming.
6576 Note that genuine overcurrent events won't be
6580 [X86] Cause panic on unknown NMI.
6582 usbcore.authorized_default=
6583 [USB] Default USB device authorization:
6584 (default -1 = authorized except for wireless USB,
6585 0 = not authorized, 1 = authorized, 2 = authorized
6586 if device connected to internal port)
6588 usbcore.autosuspend=
6589 [USB] The autosuspend time delay (in seconds) used
6590 for newly-detected USB devices (default 2). This
6591 is the time required before an idle device will be
6592 autosuspended. Devices for which the delay is set
6593 to a negative value won't be autosuspended at all.
6595 usbcore.usbfs_snoop=
6596 [USB] Set to log all usbfs traffic (default 0 = off).
6598 usbcore.usbfs_snoop_max=
6599 [USB] Maximum number of bytes to snoop in each URB
6602 usbcore.blinkenlights=
6603 [USB] Set to cycle leds on hubs (default 0 = off).
6605 usbcore.old_scheme_first=
6606 [USB] Start with the old device initialization
6607 scheme (default 0 = off).
6609 usbcore.usbfs_memory_mb=
6610 [USB] Memory limit (in MB) for buffers allocated by
6611 usbfs (default = 16, 0 = max = 2047).
6613 usbcore.use_both_schemes=
6614 [USB] Try the other device initialization scheme
6615 if the first one fails (default 1 = enabled).
6617 usbcore.initial_descriptor_timeout=
6618 [USB] Specifies timeout for the initial 64-byte
6619 USB_REQ_GET_DESCRIPTOR request in milliseconds
6620 (default 5000 = 5.0 seconds).
6622 usbcore.nousb [USB] Disable the USB subsystem
6625 [USB] A list of quirk entries to augment the built-in
6626 usb core quirk list. List entries are separated by
6627 commas. Each entry has the form
6628 VendorID:ProductID:Flags. The IDs are 4-digit hex
6629 numbers and Flags is a set of letters. Each letter
6630 will change the built-in quirk; setting it if it is
6631 clear and clearing it if it is set. The letters have
6632 the following meanings:
6633 a = USB_QUIRK_STRING_FETCH_255 (string
6634 descriptors must not be fetched using
6636 b = USB_QUIRK_RESET_RESUME (device can't resume
6637 correctly so reset it instead);
6638 c = USB_QUIRK_NO_SET_INTF (device can't handle
6639 Set-Interface requests);
6640 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
6641 handle its Configuration or Interface
6643 e = USB_QUIRK_RESET (device can't be reset
6644 (e.g morph devices), don't use reset);
6645 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
6646 more interface descriptions than the
6647 bNumInterfaces count, and can't handle
6648 talking to these interfaces);
6649 g = USB_QUIRK_DELAY_INIT (device needs a pause
6650 during initialization, after we read
6651 the device descriptor);
6652 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
6653 high speed and super speed interrupt
6654 endpoints, the USB 2.0 and USB 3.0 spec
6655 require the interval in microframes (1
6656 microframe = 125 microseconds) to be
6657 calculated as interval = 2 ^
6659 Devices with this quirk report their
6660 bInterval as the result of this
6661 calculation instead of the exponent
6662 variable used in the calculation);
6663 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
6664 handle device_qualifier descriptor
6666 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
6667 generates spurious wakeup, ignore
6668 remote wakeup capability);
6669 k = USB_QUIRK_NO_LPM (device can't handle Link
6671 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
6672 (Device reports its bInterval as linear
6673 frames instead of the USB 2.0
6675 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
6676 to be disconnected before suspend to
6677 prevent spurious wakeup);
6678 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
6679 pause after every control message);
6680 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
6681 delay after resetting its port);
6682 Example: quirks=0781:5580:bk,0a5c:5834:gij
6685 [USBHID] The interval which mice are to be polled at.
6688 [USBHID] The interval which joysticks are to be polled at.
6691 [USBHID] The interval which keyboards are to be polled at.
6693 usb-storage.delay_use=
6694 [UMS] The delay in seconds before a new device is
6695 scanned for Logical Units (default 1).
6698 [UMS] A list of quirks entries to supplement or
6699 override the built-in unusual_devs list. List
6700 entries are separated by commas. Each entry has
6701 the form VID:PID:Flags where VID and PID are Vendor
6702 and Product ID values (4-digit hex numbers) and
6703 Flags is a set of characters, each corresponding
6704 to a common usb-storage quirk flag as follows:
6705 a = SANE_SENSE (collect more than 18 bytes
6706 of sense data, not on uas);
6707 b = BAD_SENSE (don't collect more than 18
6708 bytes of sense data, not on uas);
6709 c = FIX_CAPACITY (decrease the reported
6710 device capacity by one sector);
6711 d = NO_READ_DISC_INFO (don't use
6712 READ_DISC_INFO command, not on uas);
6713 e = NO_READ_CAPACITY_16 (don't use
6714 READ_CAPACITY_16 command);
6715 f = NO_REPORT_OPCODES (don't use report opcodes
6717 g = MAX_SECTORS_240 (don't transfer more than
6718 240 sectors at a time, uas only);
6719 h = CAPACITY_HEURISTICS (decrease the
6720 reported device capacity by one
6721 sector if the number is odd);
6722 i = IGNORE_DEVICE (don't bind to this
6724 j = NO_REPORT_LUNS (don't use report luns
6726 k = NO_SAME (do not use WRITE_SAME, uas only)
6727 l = NOT_LOCKABLE (don't try to lock and
6728 unlock ejectable media, not on uas);
6729 m = MAX_SECTORS_64 (don't transfer more
6730 than 64 sectors = 32 KB at a time,
6732 n = INITIAL_READ10 (force a retry of the
6733 initial READ(10) command, not on uas);
6734 o = CAPACITY_OK (accept the capacity
6735 reported by the device, not on uas);
6736 p = WRITE_CACHE (the device cache is ON
6737 by default, not on uas);
6738 r = IGNORE_RESIDUE (the device reports
6739 bogus residue values, not on uas);
6740 s = SINGLE_LUN (the device has only one
6742 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
6743 commands, uas only);
6744 u = IGNORE_UAS (don't bind to the uas driver);
6745 w = NO_WP_DETECT (don't test whether the
6746 medium is write-protected).
6747 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
6748 even if the device claims no cache,
6750 Example: quirks=0419:aaf5:rl,0421:0433:rc
6752 user_debug= [KNL,ARM]
6754 See arch/arm/Kconfig.debug help text.
6755 1 - undefined instruction events
6757 4 - invalid data aborts
6760 Example: user_debug=31
6763 [X86] Flags controlling user PTE allocations.
6765 nohigh = do not allocate PTE pages in
6766 HIGHMEM regardless of setting
6769 vdso= [X86,SH,SPARC]
6770 On X86_32, this is an alias for vdso32=. Otherwise:
6772 vdso=1: enable VDSO (the default)
6773 vdso=0: disable VDSO mapping
6775 vdso32= [X86] Control the 32-bit vDSO
6776 vdso32=1: enable 32-bit VDSO
6777 vdso32=0 or vdso32=2: disable 32-bit VDSO
6779 See the help text for CONFIG_COMPAT_VDSO for more
6780 details. If CONFIG_COMPAT_VDSO is set, the default is
6781 vdso32=0; otherwise, the default is vdso32=1.
6783 For compatibility with older kernels, vdso32=2 is an
6786 Try vdso32=0 if you encounter an error that says:
6787 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
6790 vector=percpu: enable percpu vector domain
6792 video= [FB] Frame buffer configuration
6793 See Documentation/fb/modedb.rst.
6795 video.brightness_switch_enabled= [ACPI]
6797 If set to 1, on receiving an ACPI notify event
6798 generated by hotkey, video driver will adjust brightness
6799 level and then send out the event to user space through
6800 the allocated input device. If set to 0, video driver
6801 will only send out the event without touching backlight
6806 [VMMIO] Memory mapped virtio (platform) device.
6808 <size>@<baseaddr>:<irq>[:<id>]
6810 <size> := size (can use standard suffixes
6812 <baseaddr> := physical base address
6813 <irq> := interrupt number (as passed to
6815 <id> := (optional) platform device id
6817 virtio_mmio.device=1K@0x100b0000:48:7
6819 Can be used multiple times for multiple devices.
6821 vga= [BOOT,X86-32] Select a particular video mode
6822 See Documentation/arch/x86/boot.rst and
6823 Documentation/admin-guide/svga.rst.
6824 Use vga=ask for menu.
6825 This is actually a boot loader parameter; the value is
6826 passed to the kernel using a special protocol.
6828 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
6829 May slow down system boot speed, especially when
6830 enabled on systems with a large amount of memory.
6831 All options are enabled by default, and this
6832 interface is meant to allow for selectively
6833 enabling or disabling specific virtual memory
6836 Available options are:
6837 P Enable page structure init time poisoning
6838 - Disable all of the above options
6840 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
6841 size of <nn>. This can be used to increase the
6842 minimum size (128MB on x86). It can also be used to
6843 decrease the size and leave more room for directly
6846 vmcp_cma=nn[MG] [KNL,S390]
6847 Sets the memory size reserved for contiguous memory
6848 allocations for the vmcp device driver.
6850 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
6853 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
6856 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
6860 Controls the behavior of vsyscalls (i.e. calls to
6861 fixed addresses of 0xffffffffff600x00 from legacy
6862 code). Most statically-linked binaries and older
6863 versions of glibc use these calls. Because these
6864 functions are at fixed addresses, they make nice
6865 targets for exploits that can control RIP.
6867 emulate Vsyscalls turn into traps and are emulated
6868 reasonably safely. The vsyscall page is
6871 xonly [default] Vsyscalls turn into traps and are
6872 emulated reasonably safely. The vsyscall
6873 page is not readable.
6875 none Vsyscalls don't work at all. This makes
6876 them quite hard to use for exploits but
6877 might break your system.
6879 vt.color= [VT] Default text color.
6880 Format: 0xYX, X = foreground, Y = background.
6881 Default: 0x07 = light gray on black.
6883 vt.cur_default= [VT] Default cursor shape.
6884 Format: 0xCCBBAA, where AA, BB, and CC are the same as
6885 the parameters of the <Esc>[?A;B;Cc escape sequence;
6886 see VGA-softcursor.txt. Default: 2 = underline.
6888 vt.default_blu= [VT]
6889 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
6890 Change the default blue palette of the console.
6891 This is a 16-member array composed of values
6894 vt.default_grn= [VT]
6895 Format: <green0>,<green1>,<green2>,...,<green15>
6896 Change the default green palette of the console.
6897 This is a 16-member array composed of values
6900 vt.default_red= [VT]
6901 Format: <red0>,<red1>,<red2>,...,<red15>
6902 Change the default red palette of the console.
6903 This is a 16-member array composed of values
6909 Set system-wide default UTF-8 mode for all tty's.
6910 Default is 1, i.e. UTF-8 mode is enabled for all
6911 newly opened terminals.
6913 vt.global_cursor_default=
6916 Set system-wide default for whether a cursor
6917 is shown on new VTs. Default is -1,
6918 i.e. cursors will be created by default unless
6919 overridden by individual drivers. 0 will hide
6920 cursors, 1 will display them.
6922 vt.italic= [VT] Default color for italic text; 0-15.
6925 vt.underline= [VT] Default color for underlined text; 0-15.
6928 watchdog timers [HW,WDT] For information on watchdog timers,
6929 see Documentation/watchdog/watchdog-parameters.rst
6930 or other driver-specific files in the
6931 Documentation/watchdog/ directory.
6935 Set the hard lockup detector stall duration
6936 threshold in seconds. The soft lockup detector
6937 threshold is set to twice the value. A value of 0
6938 disables both lockup detectors. Default is 10
6941 workqueue.watchdog_thresh=
6942 If CONFIG_WQ_WATCHDOG is configured, workqueue can
6943 warn stall conditions and dump internal state to
6944 help debugging. 0 disables workqueue stall
6945 detection; otherwise, it's the stall threshold
6946 duration in seconds. The default value is 30 and
6947 it can be updated at runtime by writing to the
6948 corresponding sysfs file.
6950 workqueue.disable_numa
6951 By default, all work items queued to unbound
6952 workqueues are affine to the NUMA nodes they're
6953 issued on, which results in better behavior in
6954 general. If NUMA affinity needs to be disabled for
6955 whatever reason, this option can be used. Note
6956 that this also can be controlled per-workqueue for
6957 workqueues visible under /sys/bus/workqueue/.
6959 workqueue.power_efficient
6960 Per-cpu workqueues are generally preferred because
6961 they show better performance thanks to cache
6962 locality; unfortunately, per-cpu workqueues tend to
6963 be more power hungry than unbound workqueues.
6965 Enabling this makes the per-cpu workqueues which
6966 were observed to contribute significantly to power
6967 consumption unbound, leading to measurably lower
6968 power usage at the cost of small performance
6971 The default value of this parameter is determined by
6972 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
6974 workqueue.debug_force_rr_cpu
6975 Workqueue used to implicitly guarantee that work
6976 items queued without explicit CPU specified are put
6977 on the local CPU. This guarantee is no longer true
6978 and while local CPU is still preferred work items
6979 may be put on foreign CPUs. This debug option
6980 forces round-robin CPU selection to flush out
6981 usages which depend on the now broken guarantee.
6982 When enabled, memory and cache locality will be
6985 writecombine= [LOONGARCH] Control the MAT (Memory Access Type) of
6988 on - Enable writecombine, use WUC for ioremap_wc()
6989 off - Disable writecombine, use SUC for ioremap_wc()
6991 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
6992 default x2apic cluster mode on platforms
6995 xen_512gb_limit [KNL,X86-64,XEN]
6996 Restricts the kernel running paravirtualized under Xen
6997 to use only up to 512 GB of RAM. The reason to do so is
6998 crash analysis tools and Xen tools for doing domain
6999 save/restore/migration must be enabled to handle larger
7002 xen_emul_unplug= [HW,X86,XEN]
7003 Unplug Xen emulated devices
7004 Format: [unplug0,][unplug1]
7005 ide-disks -- unplug primary master IDE devices
7006 aux-ide-disks -- unplug non-primary-master IDE devices
7007 nics -- unplug network devices
7008 all -- unplug all emulated devices (NICs and IDE disks)
7009 unnecessary -- unplugging emulated devices is
7010 unnecessary even if the host did not respond to
7012 never -- do not unplug even if version check succeeds
7014 xen_legacy_crash [X86,XEN]
7015 Crash from Xen panic notifier, without executing late
7016 panic() code such as dumping handler.
7018 xen_msr_safe= [X86,XEN]
7020 Select whether to always use non-faulting (safe) MSR
7021 access functions when running as Xen PV guest. The
7022 default value is controlled by CONFIG_XEN_PV_MSR_SAFE.
7024 xen_nopvspin [X86,XEN]
7025 Disables the qspinlock slowpath using Xen PV optimizations.
7026 This parameter is obsoleted by "nopvspin" parameter, which
7027 has equivalent effect for XEN platform.
7030 Disables the PV optimizations forcing the HVM guest to
7031 run as generic HVM guest with no PV drivers.
7032 This option is obsoleted by the "nopv" option, which
7033 has equivalent effect for XEN platform.
7035 xen_no_vector_callback
7036 [KNL,X86,XEN] Disable the vector callback for Xen
7037 event channel interrupts.
7039 xen_scrub_pages= [XEN]
7040 Boolean option to control scrubbing pages before giving them back
7041 to Xen, for use by other domains. Can be also changed at runtime
7042 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
7043 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
7045 xen_timer_slop= [X86-64,XEN]
7046 Set the timer slop (in nanoseconds) for the virtual Xen
7047 timers (default is 100000). This adjusts the minimum
7048 delta of virtualized Xen timers, where lower values
7049 improve timer resolution at the expense of processing
7050 more timer interrupts.
7052 xen.balloon_boot_timeout= [XEN]
7053 The time (in seconds) to wait before giving up to boot
7054 in case initial ballooning fails to free enough memory.
7055 Applies only when running as HVM or PVH guest and
7056 started with less memory configured than allowed at
7057 max. Default is 180.
7059 xen.event_eoi_delay= [XEN]
7060 How long to delay EOI handling in case of event
7061 storms (jiffies). Default is 10.
7063 xen.event_loop_timeout= [XEN]
7064 After which time (jiffies) the event handling loop
7065 should start to delay EOI handling. Default is 2.
7067 xen.fifo_events= [XEN]
7068 Boolean parameter to disable using fifo event handling
7069 even if available. Normally fifo event handling is
7070 preferred over the 2-level event handling, as it is
7071 fairer and the number of possible event channels is
7072 much higher. Default is on (use fifo events).
7074 xirc2ps_cs= [NET,PCMCIA]
7076 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
7079 By default on POWER9 and above, the kernel will
7080 natively use the XIVE interrupt controller. This option
7081 allows the fallback firmware mode to be used:
7083 off Fallback to firmware control of XIVE interrupt
7084 controller on both pseries and powernv
7085 platforms. Only useful on POWER9 and above.
7087 xive.store-eoi=off [PPC]
7088 By default on POWER10 and above, the kernel will use
7089 stores for EOI handling when the XIVE interrupt mode
7090 is active. This option allows the XIVE driver to use
7091 loads instead, as on POWER9.
7093 xhci-hcd.quirks [USB,KNL]
7094 A hex value specifying bitmask with supplemental xhci
7095 host controller quirks. Meaning of each bit can be
7096 consulted in header drivers/usb/host/xhci.h.
7099 Format: { early | on | rw | ro | off }
7100 Controls if xmon debugger is enabled. Default is off.
7101 Passing only "xmon" is equivalent to "xmon=early".
7102 early Call xmon as early as possible on boot; xmon
7103 debugger is called from setup_arch().
7104 on xmon debugger hooks will be installed so xmon
7105 is only called on a kernel crash. Default mode,
7106 i.e. either "ro" or "rw" mode, is controlled
7107 with CONFIG_XMON_DEFAULT_RO_MODE.
7108 rw xmon debugger hooks will be installed so xmon
7109 is called only on a kernel crash, mode is write,
7110 meaning SPR registers, memory and, other data
7111 can be written using xmon commands.
7112 ro same as "rw" option above but SPR registers,
7113 memory, and other data can't be written using
7115 off xmon is disabled.