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.
326 irtcachedis - Disable Interrupt Remapping Table (IRT) caching.
328 amd_iommu_dump= [HW,X86-64]
329 Enable AMD IOMMU driver option to dump the ACPI table
330 for AMD IOMMU. With this option enabled, AMD IOMMU
331 driver will print ACPI tables for AMD IOMMU during
332 IOMMU initialization.
334 amd_iommu_intr= [HW,X86-64]
335 Specifies one of the following AMD IOMMU interrupt
337 legacy - Use legacy interrupt remapping mode.
338 vapic - Use virtual APIC mode, which allows IOMMU
339 to inject interrupts directly into guest.
340 This mode requires kvm-amd.avic=1.
341 (Default when IOMMU HW support is present.)
343 amijoy.map= [HW,JOY] Amiga joystick support
344 Map of devices attached to JOY0DAT and JOY1DAT
346 See also Documentation/input/joydev/joystick.rst
348 analog.map= [HW,JOY] Analog joystick and gamepad support
349 Specifies type or capabilities of an analog joystick
350 connected to one of 16 gameports
351 Format: <type1>,<type2>,..<type16>
354 Power management functions (SPARCstation-4/5 + deriv.)
356 Disable APC CPU standby support. SPARCstation-Fox does
357 not play well with APC CPU idle - disable it if you have
358 APC and your system crashes randomly.
360 apic= [APIC,X86] Advanced Programmable Interrupt Controller
361 Change the output verbosity while booting
362 Format: { quiet (default) | verbose | debug }
363 Change the amount of debugging information output
364 when initialising the APIC and IO-APIC components.
365 For X86-32, this can also be used to specify an APIC
367 Format: apic=driver_name
368 Examples: apic=bigsmp
370 apic_extnmi= [APIC,X86] External NMI delivery setting
371 Format: { bsp (default) | all | none }
372 bsp: External NMI is delivered only to CPU 0
373 all: External NMIs are broadcast to all CPUs as a
375 none: External NMI is masked for all CPUs. This is
376 useful so that a dump capture kernel won't be
380 See Documentation/networking/ipv6.rst.
382 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
383 Limit apic dumping. The parameter defines the maximal
384 number of local apics being dumped. Also it is possible
385 to set it to "all" by meaning -- no limit here.
386 Format: { 1 (default) | 2 | ... | all }.
387 The parameter valid if only apic=debug or
388 apic=verbose is specified.
389 Example: apic=debug show_lapic=all
391 apm= [APM] Advanced Power Management
392 See header of arch/x86/kernel/apm_32.c.
394 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
395 Format: <io>,<irq>,<nodeID>
397 arm64.nobti [ARM64] Unconditionally disable Branch Target
398 Identification support
400 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication
403 arm64.nomte [ARM64] Unconditionally disable Memory Tagging Extension
406 arm64.nosve [ARM64] Unconditionally disable Scalable Vector
409 arm64.nosme [ARM64] Unconditionally disable Scalable Matrix
414 atarimouse= [HW,MOUSE] Atari Mouse
416 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
417 EzKey and similar keyboards
419 atkbd.reset= [HW] Reset keyboard during initialization
421 atkbd.set= [HW] Select keyboard code set
422 Format: <int> (2 = AT (default), 3 = PS/2)
424 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
427 atkbd.softraw= [HW] Choose between synthetic and real raw mode
428 Format: <bool> (0 = real, 1 = synthetic (default))
430 atkbd.softrepeat= [HW]
431 Use software keyboard repeat
433 audit= [KNL] Enable the audit sub-system
434 Format: { "0" | "1" | "off" | "on" }
435 0 | off - kernel audit is disabled and can not be
436 enabled until the next reboot
437 unset - kernel audit is initialized but disabled and
438 will be fully enabled by the userspace auditd.
439 1 | on - kernel audit is initialized and partially
440 enabled, storing at most audit_backlog_limit
441 messages in RAM until it is fully enabled by the
445 audit_backlog_limit= [KNL] Set the audit queue size limit.
446 Format: <int> (must be >=0)
449 bau= [X86_UV] Enable the BAU on SGI UV. The default
450 behavior is to disable the BAU (i.e. bau=0).
451 Format: { "0" | "1" }
454 unset - Disable the BAU.
456 baycom_epp= [HW,AX25]
459 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
461 See header of drivers/net/hamradio/baycom_par.c.
463 baycom_ser_fdx= [HW,AX25]
464 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
465 Format: <io>,<irq>,<mode>[,<baud>]
466 See header of drivers/net/hamradio/baycom_ser_fdx.c.
468 baycom_ser_hdx= [HW,AX25]
469 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
470 Format: <io>,<irq>,<mode>
471 See header of drivers/net/hamradio/baycom_ser_hdx.c.
474 Disable BERT OS support on buggy BIOSes.
476 bgrt_disable [ACPI][X86]
477 Disable BGRT to avoid flickering OEM logo.
479 blkdevparts= Manual partition parsing of block device(s) for
480 embedded devices based on command line input.
481 See Documentation/block/cmdline-partition.rst
483 boot_delay= Milliseconds to delay each printk during boot.
484 Values larger than 10 seconds (10000) are changed to
489 Extended command line options can be added to an initrd
490 and this will cause the kernel to look for it.
492 See Documentation/admin-guide/bootconfig.rst
494 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
495 bttv.radio= Most important insmod options are available as
497 bttv.pll= See Documentation/admin-guide/media/bttv.rst
500 bulk_remove=off [PPC] This parameter disables the use of the pSeries
501 firmware feature for flushing multiple hpte entries
504 c101= [NET] Moxa C101 synchronous serial card
506 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
507 Sometimes CPU hardware bugs make them report the cache
508 size incorrectly. The kernel will attempt work arounds
509 to fix known problems, but for some CPUs it is not
510 possible to determine what the correct size should be.
511 This option provides an override for these situations.
514 [NET] Specifies amount of time (in seconds) that
515 the kernel should wait for a network carrier. By default
516 it waits 120 seconds.
518 ca_keys= [KEYS] This parameter identifies a specific key(s) on
519 the system trusted keyring to be used for certificate
521 format: { id:<keyid> | builtin }
523 cca= [MIPS] Override the kernel pages' cache coherency
524 algorithm. Accepted values range from 0 to 7
525 inclusive. See arch/mips/include/asm/pgtable-bits.h
526 for platform specific values (SB1, Loongson3 and
529 ccw_timeout_log [S390]
530 See Documentation/s390/common_io.rst for details.
532 cgroup_disable= [KNL] Disable a particular controller or optional feature
533 Format: {name of the controller(s) or feature(s) to disable}
534 The effects of cgroup_disable=foo are:
535 - foo isn't auto-mounted if you mount all cgroups in
537 - foo isn't visible as an individually mountable
539 - if foo is an optional feature then the feature is
540 disabled and corresponding cgroup files are not
542 {Currently only "memory" controller deal with this and
543 cut the overhead, others just disable the usage. So
544 only cgroup_disable=memory is actually worthy}
545 Specifying "pressure" disables per-cgroup pressure
546 stall information accounting feature
548 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
549 Format: { { controller | "all" | "named" }
550 [,{ controller | "all" | "named" }...] }
551 Like cgroup_disable, but only applies to cgroup v1;
552 the blacklisted controllers remain available in cgroup2.
553 "all" blacklists all controllers and "named" disables
554 named mounts. Specifying both "all" and "named" disables
557 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
559 nosocket -- Disable socket memory accounting.
560 nokmem -- Disable kernel memory accounting.
562 checkreqprot= [SELINUX] Set initial checkreqprot flag value.
563 Format: { "0" | "1" }
564 See security/selinux/Kconfig help text.
565 0 -- check protection applied by kernel (includes
566 any implied execute protection).
567 1 -- check protection requested by application.
568 Default value is set via a kernel config option.
569 Value can be changed at runtime via
570 /sys/fs/selinux/checkreqprot.
571 Setting checkreqprot to 1 is deprecated.
574 See Documentation/s390/common_io.rst for details.
576 clearcpuid=X[,X...] [X86]
577 Disable CPUID feature X for the kernel. See
578 arch/x86/include/asm/cpufeatures.h for the valid bit
579 numbers X. Note the Linux-specific bits are not necessarily
580 stable over kernel options, but the vendor-specific
582 X can also be a string as appearing in the flags: line
583 in /proc/cpuinfo which does not have the above
584 instability issue. However, not all features have names
586 Note that using this option will taint your kernel.
587 Also note that user programs calling CPUID directly
588 or using the feature without checking anything
589 will still see it. This just prevents it from
590 being used by the kernel or shown in /proc/cpuinfo.
591 Also note the kernel might malfunction if you disable
596 Prevents the clock framework from automatically gating
597 clocks that have not been explicitly enabled by a Linux
598 device driver but are enabled in hardware at reset or
599 by the bootloader/firmware. Note that this does not
600 force such clocks to be always-on nor does it reserve
601 those clocks in any way. This parameter is useful for
602 debug and development, but should not be needed on a
603 platform with proper driver support. For more
604 information, see Documentation/driver-api/clk.rst.
606 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
608 Forces specified clocksource (if available) to be used
609 when calculating gettimeofday(). If specified
610 clocksource is not available, it defaults to PIT.
611 Format: { pit | tsc | cyclone | pmtmr }
613 clocksource= Override the default clocksource
615 Override the default clocksource and use the clocksource
616 with the name specified.
617 Some clocksource names to choose from, depending on
619 [all] jiffies (this is the base, fallback clocksource)
621 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
622 pxa_timer,timer3,32k_counter,timer0_1
623 [X86-32] pit,hpet,tsc;
624 scx200_hrt on Geode; cyclone on IBM x440
632 clocksource.arm_arch_timer.evtstrm=
635 Enable/disable the eventstream feature of the ARM
636 architected timer so that code using WFE-based polling
637 loops can be debugged more effectively on production
640 clocksource.max_cswd_read_retries= [KNL]
641 Number of clocksource_watchdog() retries due to
642 external delays before the clock will be marked
643 unstable. Defaults to two retries, that is,
644 three attempts to read the clock under test.
646 clocksource.verify_n_cpus= [KNL]
647 Limit the number of CPUs checked for clocksources
648 marked with CLOCK_SOURCE_VERIFY_PERCPU that
649 are marked unstable due to excessive skew.
650 A negative value says to check all CPUs, while
651 zero says not to check any. Values larger than
652 nr_cpu_ids are silently truncated to nr_cpu_ids.
653 The actual CPUs are chosen randomly, with
654 no replacement if the same CPU is chosen twice.
656 clocksource-wdtest.holdoff= [KNL]
657 Set the time in seconds that the clocksource
658 watchdog test waits before commencing its tests.
659 Defaults to zero when built as a module and to
660 10 seconds when built into the kernel.
662 cma=nn[MG]@[start[MG][-end[MG]]]
664 Sets the size of kernel global memory area for
665 contiguous memory allocations and optionally the
666 placement constraint by the physical address range of
667 memory allocations. A value of 0 disables CMA
668 altogether. For more information, see
669 kernel/dma/contiguous.c
673 Sets the size of kernel per-numa memory area for
674 contiguous memory allocations. A value of 0 disables
675 per-numa CMA altogether. And If this option is not
676 specificed, the default value is 0.
677 With per-numa CMA enabled, DMA users on node nid will
678 first try to allocate buffer from the pernuma area
679 which is located in node nid, if the allocation fails,
680 they will fallback to the global default memory area.
682 cmo_free_hint= [PPC] Format: { yes | no }
683 Specify whether pages are marked as being inactive
684 when they are freed. This is used in CMO environments
685 to determine OS memory pressure for page stealing by
689 coherent_pool=nn[KMG] [ARM,KNL]
690 Sets the size of memory pool for coherent, atomic dma
691 allocations, by default set to 256K.
693 com20020= [HW,NET] ARCnet - COM20020 chipset
695 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
697 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
701 ARCnet - COM90xx chipset (memory-mapped buffers)
702 Format: <io>[,<irq>[,<memstart>]]
704 condev= [HW,S390] console device
707 console= [KNL] Output console device and options.
709 tty<n> Use the virtual console device <n>.
713 Use the specified serial port. The options are of
714 the form "bbbbpnf", where "bbbb" is the baud rate,
715 "p" is parity ("n", "o", or "e"), "n" is number of
716 bits, and "f" is flow control ("r" for RTS or
717 omit it). Default is "9600n8".
719 See Documentation/admin-guide/serial-console.rst for more
721 Documentation/networking/netconsole.rst for an
724 uart[8250],io,<addr>[,options]
725 uart[8250],mmio,<addr>[,options]
726 uart[8250],mmio16,<addr>[,options]
727 uart[8250],mmio32,<addr>[,options]
728 uart[8250],0x<addr>[,options]
729 Start an early, polled-mode console on the 8250/16550
730 UART at the specified I/O port or MMIO address,
731 switching to the matching ttyS device later.
732 MMIO inter-register address stride is either 8-bit
733 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
734 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
735 to be equivalent to 'mmio'. 'options' are specified in
736 the same format described for ttyS above; if unspecified,
737 the h/w is not re-initialized.
739 hvc<n> Use the hypervisor console device <n>. This is for
740 both Xen and PowerPC hypervisors.
743 Use to disable console output, i.e., to have kernel
744 console messages discarded.
745 This must be the only console= parameter used on the
748 If the device connected to the port is not a TTY but a braille
749 device, prepend "brl," before the device type, for instance
751 For now, only VisioBraille is supported.
754 [KNL] Change console messages format
756 By default we print messages on consoles in
757 "[time stamp] text\n" format (time stamp may not be
758 printed, depending on CONFIG_PRINTK_TIME or
759 `printk_time' param).
761 Switch to syslog format: "<%u>[time stamp] text\n"
762 IOW, each message will have a facility and loglevel
763 prefix. The format is similar to one used by syslog()
764 syscall, or to executing "dmesg -S --raw" or to reading
767 consoleblank= [KNL] The console blank (screen saver) timeout in
768 seconds. A value of 0 disables the blank timer.
772 [KNL] Change the default value for
773 /proc/<pid>/coredump_filter.
774 See also Documentation/filesystems/proc.rst.
776 coresight_cpu_debug.enable
779 Enable/disable the CPU sampling based debugging.
780 0: default value, disable debugging
781 1: enable debugging at boot time
783 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
785 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
787 cpu0_hotplug [X86] Turn on CPU0 hotplug feature when
788 CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off.
789 Some features depend on CPU0. Known dependencies are:
790 1. Resume from suspend/hibernate depends on CPU0.
791 Suspend/hibernate will fail if CPU0 is offline and you
792 need to online CPU0 before suspend/hibernate.
793 2. PIC interrupts also depend on CPU0. CPU0 can't be
794 removed if a PIC interrupt is detected.
795 It's said poweroff/reboot may depend on CPU0 on some
796 machines although I haven't seen such issues so far
797 after CPU0 is offline on a few tested machines.
798 If the dependencies are under your control, you can
799 turn on cpu0_hotplug.
801 cpuidle.off=1 [CPU_IDLE]
802 disable the cpuidle sub-system
805 [CPU_IDLE] Name of the cpuidle governor to use.
807 cpufreq.off=1 [CPU_FREQ]
808 disable the cpufreq sub-system
810 cpufreq.default_governor=
811 [CPU_FREQ] Name of the default cpufreq governor or
812 policy to use. This governor must be registered in the
813 kernel before the cpufreq driver probes.
816 [X86] Delay for N microsec between assert and de-assert
817 of APIC INIT to start processors. This delay occurs
818 on every CPU online, such as boot, and resume from suspend.
821 crash_kexec_post_notifiers
822 Run kdump after running panic-notifiers and dumping
823 kmsg. This only for the users who doubt kdump always
824 succeeds in any situation.
825 Note that this also increases risks of kdump failure,
826 because some panic notifiers can make the crashed
827 kernel more unstable.
829 crashkernel=size[KMG][@offset[KMG]]
830 [KNL] Using kexec, Linux can switch to a 'crash kernel'
831 upon panic. This parameter reserves the physical
832 memory region [offset, offset + size] for that kernel
833 image. If '@offset' is omitted, then a suitable offset
834 is selected automatically.
835 [KNL, X86-64] Select a region under 4G first, and
836 fall back to reserve region above 4G when '@offset'
837 hasn't been specified.
838 See Documentation/admin-guide/kdump/kdump.rst for further details.
840 crashkernel=range1:size1[,range2:size2,...][@offset]
841 [KNL] Same as above, but depends on the memory
842 in the running system. The syntax of range is
843 start-[end] where start and end are both
844 a memory unit (amount[KMG]). See also
845 Documentation/admin-guide/kdump/kdump.rst for an example.
847 crashkernel=size[KMG],high
848 [KNL, X86-64, ARM64] range could be above 4G. Allow kernel
849 to allocate physical memory region from top, so could
850 be above 4G if system have more than 4G ram installed.
851 Otherwise memory region will be allocated below 4G, if
853 It will be ignored if crashkernel=X is specified.
854 crashkernel=size[KMG],low
855 [KNL, X86-64] range under 4G. When crashkernel=X,high
856 is passed, kernel could allocate physical memory region
857 above 4G, that cause second kernel crash on system
858 that require some amount of low memory, e.g. swiotlb
859 requires at least 64M+32K low memory, also enough extra
860 low memory is needed to make sure DMA buffers for 32-bit
861 devices won't run out. Kernel would try to allocate
862 at least 256M below 4G automatically.
863 This one lets the user specify own low range under 4G
864 for second kernel instead.
865 0: to disable low allocation.
866 It will be ignored when crashkernel=X,high is not used
867 or memory reserved is below 4G.
869 [KNL, ARM64] range in low memory.
870 This one lets the user specify a low range in the
871 DMA zone for the crash dump kernel.
872 It will be ignored when crashkernel=X,high is not used
873 or memory reserved is located in the DMA zones.
876 [KNL] Disable crypto self-tests
881 cs89x0_media= [HW,NET]
882 Format: { rj45 | aui | bnc }
884 csdlock_debug= [KNL] Enable debug add-ons of cross-CPU function call
885 handling. When switched on, additional debug data is
886 printed to the console in case a hanging CPU is
887 detected, and that CPU is pinged again in order to try
888 to resolve the hang situation.
889 0: disable csdlock debugging (default)
890 1: enable basic csdlock debugging (minor impact)
891 ext: enable extended csdlock debugging (more impact,
895 See header of drivers/s390/block/dasd_devmap.c.
897 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
898 (one device per port)
899 Format: <port#>,<type>
900 See also Documentation/input/devices/joystick-parport.rst
902 debug [KNL] Enable kernel debugging (events log level).
905 [KNL] Enable printing [hashed] pointers early in the
906 boot sequence. If enabled, we use a weak hash instead
907 of siphash to hash pointers. Use this option if you are
908 seeing instances of '(___ptrval___)') and need to see a
909 value (hashed pointer) instead. Cryptographically
910 insecure, please do not use on production kernels.
913 [KNL] verbose locking self-tests
915 Print debugging info while doing the locking API
917 Bitmask for the various LOCKTYPE_ tests. Defaults to 0
918 (no extra messages), setting it to -1 (all bits set)
919 will print _a_lot_ more information - normally only
920 useful to lockdep developers.
922 debug_objects [KNL] Enable object debugging
925 [KNL] Disable object debugging
927 debug_guardpage_minorder=
928 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
929 parameter allows control of the order of pages that will
930 be intentionally kept free (and hence protected) by the
931 buddy allocator. Bigger value increase the probability
932 of catching random memory corruption, but reduce the
933 amount of memory for normal system use. The maximum
934 possible value is MAX_ORDER/2. Setting this parameter
935 to 1 or 2 should be enough to identify most random
936 memory corruption problems caused by bugs in kernel or
937 driver code when a CPU writes to (or reads from) a
938 random memory location. Note that there exists a class
939 of memory corruptions problems caused by buggy H/W or
940 F/W or by drivers badly programing DMA (basically when
941 memory is written at bus level and the CPU MMU is
942 bypassed) which are not detectable by
943 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
944 tracking down these problems.
947 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
948 enables the feature at boot time. By default, it is
949 disabled and the system will work mostly the same as a
950 kernel built without CONFIG_DEBUG_PAGEALLOC.
951 Note: to get most of debug_pagealloc error reports, it's
952 useful to also enable the page_owner functionality.
953 on: enable the feature
955 debugfs= [KNL] This parameter enables what is exposed to userspace
956 and debugfs internal clients.
957 Format: { on, no-mount, off }
958 on: All functions are enabled.
960 Filesystem is not registered but kernel clients can
961 access APIs and a crashkernel can be used to read
962 its content. There is nothing to mount.
963 off: Filesystem is not registered and clients
964 get a -EPERM as result when trying to register files
965 or directories within debugfs.
966 This is equivalent of the runtime functionality if
967 debugfs was not enabled in the kernel at all.
968 Default value is set in build-time with a kernel configuration.
970 debugpat [X86] Enable PAT debugging
973 [HW] The size of the default HugeTLB page. This is
974 the size represented by the legacy /proc/ hugepages
975 APIs. In addition, this is the default hugetlb size
976 used for shmget(), mmap() and mounting hugetlbfs
977 filesystems. If not specified, defaults to the
978 architecture's default huge page size. Huge page
979 sizes are architecture dependent. See also
980 Documentation/admin-guide/mm/hugetlbpage.rst.
983 deferred_probe_timeout=
984 [KNL] Debugging option to set a timeout in seconds for
985 deferred probe to give up waiting on dependencies to
986 probe. Only specific dependencies (subsystems or
987 drivers) that have opted in will be ignored. A timeout
988 of 0 will timeout at the end of initcalls. If the time
989 out hasn't expired, it'll be restarted by each
990 successful driver registration. This option will also
991 dump out devices still on the deferred probe list after
994 delayacct [KNL] Enable per-task delay accounting
996 dell_smm_hwmon.ignore_dmi=
997 [HW] Continue probing hardware even if DMI data
998 indicates that the driver is running on unsupported
1001 dell_smm_hwmon.force=
1002 [HW] Activate driver even if SMM BIOS signature does
1003 not match list of supported models and enable otherwise
1004 blacklisted features.
1006 dell_smm_hwmon.power_status=
1007 [HW] Report power status in /proc/i8k
1008 (disabled by default).
1010 dell_smm_hwmon.restricted=
1011 [HW] Allow controlling fans only if SYS_ADMIN
1014 dell_smm_hwmon.fan_mult=
1015 [HW] Factor to multiply fan speed with.
1017 dell_smm_hwmon.fan_max=
1018 [HW] Maximum configurable fan speed.
1021 Format: { on | off | def_only | inf_only | always }
1022 on: s390 zlib hardware support for compression on
1023 level 1 and decompression (default)
1024 off: No s390 zlib hardware support
1025 def_only: s390 zlib hardware support for deflate
1026 only (compression on level 1)
1027 inf_only: s390 zlib hardware support for inflate
1028 only (decompression)
1029 always: Same as 'on' but ignores the selected compression
1030 level always using hardware support (used for debugging)
1032 dhash_entries= [KNL]
1033 Set number of hash buckets for dentry cache.
1035 disable_1tb_segments [PPC]
1036 Disables the use of 1TB hash page table segments. This
1037 causes the kernel to fall back to 256MB segments which
1038 can be useful when debugging issues that require an SLB
1042 Limits the number of kernel SLB entries, and flushes
1043 them frequently to increase the rate of SLB faults
1044 on kernel addresses.
1047 See Documentation/networking/ipv6.rst.
1050 Disable RADIX MMU mode on POWER9
1052 radix_hcall_invalidate=on [PPC/PSERIES]
1053 Disable RADIX GTSE feature and use hcall for TLB
1057 Disable TLBIE instruction. Currently does not work
1058 with KVM, with HASH MMU, or with coherent accelerators.
1060 disable_cpu_apicid= [X86,APIC,SMP]
1062 The number of initial APIC ID for the
1063 corresponding CPU to be disabled at boot,
1064 mostly used for the kdump 2nd kernel to
1065 disable BSP to wake up multiple CPUs without
1066 causing system reset or hang due to sending
1067 INIT from AP to BSP.
1069 disable_ddw [PPC/PSERIES]
1070 Disable Dynamic DMA Window support. Use this
1071 to workaround buggy firmware.
1073 disable_ipv6= [IPV6]
1074 See Documentation/networking/ipv6.rst.
1076 disable_mtrr_cleanup [X86]
1077 The kernel tries to adjust MTRR layout from continuous
1078 to discrete, to make X server driver able to add WB
1079 entry later. This parameter disables that.
1081 disable_mtrr_trim [X86, Intel and AMD only]
1082 By default the kernel will trim any uncacheable
1083 memory out of your available memory pool based on
1084 MTRR settings. This parameter disables that behavior,
1085 possibly causing your machine to run very slowly.
1087 disable_timer_pin_1 [X86]
1088 Disable PIN 1 of APIC timer
1089 Can be useful to work around chipset bugs.
1091 dis_ucode_ldr [X86] Disable the microcode loader.
1093 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
1094 this option disables the debugging code at boot.
1096 dma_debug_entries=<number>
1097 This option allows to tune the number of preallocated
1098 entries for DMA-API debugging code. One entry is
1099 required per DMA-API allocation. Use this if the
1100 DMA-API debugging code disables itself because the
1101 architectural default is too low.
1103 dma_debug_driver=<driver_name>
1104 With this option the DMA-API debugging driver
1105 filter feature can be enabled at boot time. Just
1106 pass the driver to filter for as the parameter.
1107 The filter can be disabled or changed to another
1108 driver later using sysfs.
1110 reg_file_data_sampling=
1111 [X86] Controls mitigation for Register File Data
1112 Sampling (RFDS) vulnerability. RFDS is a CPU
1113 vulnerability which may allow userspace to infer
1114 kernel data values previously stored in floating point
1115 registers, vector registers, or integer registers.
1116 RFDS only affects Intel Atom processors.
1118 on: Turns ON the mitigation.
1119 off: Turns OFF the mitigation.
1121 This parameter overrides the compile time default set
1122 by CONFIG_MITIGATION_RFDS. Mitigation cannot be
1123 disabled when other VERW based mitigations (like MDS)
1124 are enabled. In order to disable RFDS mitigation all
1125 VERW based mitigations need to be disabled.
1128 Documentation/admin-guide/hw-vuln/reg-file-data-sampling.rst
1130 driver_async_probe= [KNL]
1131 List of driver names to be probed asynchronously. *
1132 matches with all driver names. If * is specified, the
1133 rest of the listed driver names are those that will NOT
1135 Format: <driver_name1>,<driver_name2>...
1137 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
1138 Broken monitors, graphic adapters, KVMs and EDIDless
1139 panels may send no or incorrect EDID data sets.
1140 This parameter allows to specify an EDID data sets
1141 in the /lib/firmware directory that are used instead.
1142 Generic built-in EDID data sets are used, if one of
1143 edid/1024x768.bin, edid/1280x1024.bin,
1144 edid/1680x1050.bin, or edid/1920x1080.bin is given
1145 and no file with the same name exists. Details and
1146 instructions how to build your own EDID data are
1147 available in Documentation/admin-guide/edid.rst. An EDID
1148 data set will only be used for a particular connector,
1149 if its name and a colon are prepended to the EDID
1150 name. Each connector may use a unique EDID data
1151 set by separating the files with a comma. An EDID
1152 data set with no connector name will be used for
1153 any connectors not explicitly specified.
1158 Format: {"off" | "known"}
1159 Control how the dt_cpu_ftrs device-tree binding is
1160 used for CPU feature discovery and setup (if it
1162 off: Do not use it, fall back to legacy cpu table.
1163 known: Do not pass through unknown features to guests
1164 or userspace, only those that the kernel is aware of.
1166 dump_apple_properties [X86]
1167 Dump name and content of EFI device properties on
1168 x86 Macs. Useful for driver authors to determine
1169 what data is available or for reverse-engineering.
1171 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
1172 <module>.dyndbg[="val"]
1173 Enable debug messages at boot time. See
1174 Documentation/admin-guide/dynamic-debug-howto.rst
1177 nopku [X86] Disable Memory Protection Keys CPU feature found
1180 <module>.async_probe[=<bool>] [KNL]
1181 If no <bool> value is specified or if the value
1182 specified is not a valid <bool>, enable asynchronous
1183 probe on this module. Otherwise, enable/disable
1184 asynchronous probe on this module as indicated by the
1185 <bool> value. See also: module.async_probe
1187 early_ioremap_debug [KNL]
1188 Enable debug messages in early_ioremap support. This
1189 is useful for tracking down temporary early mappings
1190 which are not unmapped.
1192 earlycon= [KNL] Output early console device and options.
1194 When used with no options, the early console is
1195 determined by stdout-path property in device tree's
1196 chosen node or the ACPI SPCR table if supported by
1199 cdns,<addr>[,options]
1200 Start an early, polled-mode console on a Cadence
1201 (xuartps) serial port at the specified address. Only
1202 supported option is baud rate. If baud rate is not
1203 specified, the serial port must already be setup and
1206 uart[8250],io,<addr>[,options]
1207 uart[8250],mmio,<addr>[,options]
1208 uart[8250],mmio32,<addr>[,options]
1209 uart[8250],mmio32be,<addr>[,options]
1210 uart[8250],0x<addr>[,options]
1211 Start an early, polled-mode console on the 8250/16550
1212 UART at the specified I/O port or MMIO address.
1213 MMIO inter-register address stride is either 8-bit
1214 (mmio) or 32-bit (mmio32 or mmio32be).
1215 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1216 to be equivalent to 'mmio'. 'options' are specified
1217 in the same format described for "console=ttyS<n>"; if
1218 unspecified, the h/w is not initialized.
1222 Start an early, polled-mode console on a pl011 serial
1223 port at the specified address. The pl011 serial port
1224 must already be setup and configured. Options are not
1225 yet supported. If 'mmio32' is specified, then only
1226 the driver will use only 32-bit accessors to read/write
1227 the device registers.
1230 Start an early console on a litex serial port at the
1231 specified address. The serial port must already be
1232 setup and configured. Options are not yet supported.
1235 Start an early, polled-mode console on a meson serial
1236 port at the specified address. The serial port must
1237 already be setup and configured. Options are not yet
1241 Start an early, polled-mode console on an msm serial
1242 port at the specified address. The serial port
1243 must already be setup and configured. Options are not
1246 msm_serial_dm,<addr>
1247 Start an early, polled-mode console on an msm serial
1248 dm port at the specified address. The serial port
1249 must already be setup and configured. Options are not
1253 Start an early, polled-mode console on a serial port
1254 of an Actions Semi SoC, such as S500 or S900, at the
1255 specified address. The serial port must already be
1256 setup and configured. Options are not yet supported.
1259 Start an early, polled-mode console on a serial port
1260 of an RDA Micro SoC, such as RDA8810PL, at the
1261 specified address. The serial port must already be
1262 setup and configured. Options are not yet supported.
1265 Use RISC-V SBI (Supervisor Binary Interface) for early
1268 smh Use ARM semihosting calls for early console.
1276 Use early console provided by serial driver available
1277 on Samsung SoCs, requires selecting proper type and
1278 a correct base address of the selected UART port. The
1279 serial port must already be setup and configured.
1280 Options are not yet supported.
1283 Start an early, polled-mode console on a lantiq serial
1284 (lqasc) port at the specified address. The serial port
1285 must already be setup and configured. Options are not
1290 Use early console provided by Freescale LP UART driver
1291 found on Freescale Vybrid and QorIQ LS1021A processors.
1292 A valid base address must be provided, and the serial
1293 port must already be setup and configured.
1297 Start an early, polled-mode, output-only console on the
1298 Freescale i.MX UART at the specified address. The UART
1299 must already be setup and configured.
1302 Start an early, polled-mode console on the
1303 Armada 3700 serial port at the specified
1304 address. The serial port must already be setup
1305 and configured. Options are not yet supported.
1308 Start an early, polled-mode console on a Qualcomm
1309 Generic Interface (GENI) based serial port at the
1310 specified address. The serial port must already be
1311 setup and configured. Options are not yet supported.
1314 Start an early, unaccelerated console on the EFI
1315 memory mapped framebuffer (if available). On cache
1316 coherent non-x86 systems that use system memory for
1317 the framebuffer, pass the 'ram' option so that it is
1318 mapped with the correct attributes.
1321 Use early console provided by Freescale LINFlexD UART
1322 serial driver for NXP S32V234 SoCs. A valid base
1323 address must be provided, and the serial port must
1324 already be setup and configured.
1326 earlyprintk= [X86,SH,ARM,M68k,S390]
1330 earlyprintk=serial[,ttySn[,baudrate]]
1331 earlyprintk=serial[,0x...[,baudrate]]
1332 earlyprintk=ttySn[,baudrate]
1333 earlyprintk=dbgp[debugController#]
1334 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1335 earlyprintk=xdbc[xhciController#]
1337 earlyprintk is useful when the kernel crashes before
1338 the normal console is initialized. It is not enabled by
1339 default because it has some cosmetic problems.
1341 Append ",keep" to not disable it when the real console
1344 Only one of vga, serial, or usb debug port can
1347 Currently only ttyS0 and ttyS1 may be specified by
1348 name. Other I/O ports may be explicitly specified
1349 on some architectures (x86 and arm at least) by
1350 replacing ttySn with an I/O port address, like this:
1351 earlyprintk=serial,0x1008,115200
1352 You can find the port for a given device in
1353 /proc/tty/driver/serial:
1354 2: uart:ST16650V2 port:00001008 irq:18 ...
1356 Interaction with the standard serial driver is not
1359 The VGA output is eventually overwritten by
1362 The xen option can only be used in Xen domains.
1364 The sclp output can only be used on s390.
1366 The optional "force" to "pciserial" enables use of a
1367 PCI device even when its classcode is not of the
1370 edac_report= [HW,EDAC] Control how to report EDAC event
1371 Format: {"on" | "off" | "force"}
1372 on: enable EDAC to report H/W event. May be overridden
1373 by other higher priority error reporting module.
1374 off: disable H/W event reporting through EDAC.
1375 force: enforce the use of EDAC to report H/W event.
1379 Format: {"off" | "on" | "skip[mbr]"}
1382 Format: { "debug", "disable_early_pci_dma",
1383 "nochunk", "noruntime", "nosoftreserve",
1384 "novamap", "no_disable_early_pci_dma" }
1385 debug: enable misc debug output.
1386 disable_early_pci_dma: disable the busmaster bit on all
1387 PCI bridges while in the EFI boot stub.
1388 nochunk: disable reading files in "chunks" in the EFI
1389 boot stub, as chunking can cause problems with some
1390 firmware implementations.
1391 noruntime : disable EFI runtime services support
1392 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1393 attribute may cause the kernel to reserve the
1394 memory range for a memory mapping driver to
1395 claim. Specify efi=nosoftreserve to disable this
1396 reservation and treat the memory by its base type
1397 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1398 novamap: do not call SetVirtualAddressMap().
1399 no_disable_early_pci_dma: Leave the busmaster bit set
1400 on all PCI bridges while in the EFI boot stub
1402 efi_no_storage_paranoia [EFI; X86]
1403 Using this parameter you can use more than 50% of
1404 your efi variable storage. Use this parameter only if
1405 you are really sure that your UEFI does sane gc and
1406 fulfills the spec otherwise your board may brick.
1408 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1409 Add arbitrary attribute to specific memory range by
1410 updating original EFI memory map.
1411 Region of memory which aa attribute is added to is
1414 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1415 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1416 attribute is added to range 0x100000000-0x180000000 and
1417 0x10a0000000-0x1120000000.
1419 If efi_fake_mem=8G@9G:0x40000 is specified, the
1420 EFI_MEMORY_SP(0x40000) attribute is added to
1421 range 0x240000000-0x43fffffff.
1423 Using this parameter you can do debugging of EFI memmap
1424 related features. For example, you can do debugging of
1425 Address Range Mirroring feature even if your box
1426 doesn't support it, or mark specific memory as
1429 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1430 that is to be dynamically loaded by Linux. If there are
1431 multiple variables with the same name but with different
1432 vendor GUIDs, all of them will be loaded. See
1433 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1436 eisa_irq_edge= [PARISC,HW]
1437 See header of drivers/parisc/eisa.c.
1439 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1442 This is designed to be used in conjunction with
1443 the boot argument: earlyprintk=vga
1445 This parameter works in place of the kgdboc parameter
1446 but can only be used if the backing tty is available
1447 very early in the boot process. For early debugging
1448 via a serial port see kgdboc_earlycon instead.
1451 See comment before function elanfreq_setup() in
1452 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1454 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1455 Specifies physical address of start of kernel core
1456 image elf header and optionally the size. Generally
1457 kexec loader will pass this option to capture kernel.
1458 See Documentation/admin-guide/kdump/kdump.rst for details.
1460 enable_mtrr_cleanup [X86]
1461 The kernel tries to adjust MTRR layout from continuous
1462 to discrete, to make X server driver able to add WB
1463 entry later. This parameter enables that.
1465 enable_timer_pin_1 [X86]
1466 Enable PIN 1 of APIC timer
1467 Can be useful to work around chipset bugs
1468 (in particular on some ATI chipsets).
1469 The kernel tries to set a reasonable default.
1471 enforcing= [SELINUX] Set initial enforcing status.
1473 See security/selinux/Kconfig help text.
1474 0 -- permissive (log only, no denials).
1475 1 -- enforcing (deny and log).
1477 Value can be changed at runtime via
1478 /sys/fs/selinux/enforce.
1481 Disable Error Record Serialization Table (ERST)
1484 ether= [HW,NET] Ethernet cards parameters
1485 This option is obsoleted by the "netdev=" option, which
1486 has equivalent usage. See its documentation for details.
1490 Permit 'security.evm' to be updated regardless of
1491 current integrity status.
1493 early_page_ext [KNL] Enforces page_ext initialization to earlier
1494 stages so cover more early boot allocations.
1495 Please note that as side effect some optimizations
1496 might be disabled to achieve that (e.g. parallelized
1497 memory initialization is disabled) so the boot process
1498 might take longer, especially on systems with a lot of
1499 memory. Available with CONFIG_PAGE_EXTENSION=y.
1504 fail_make_request=[KNL]
1505 General fault injection mechanism.
1506 Format: <interval>,<probability>,<space>,<times>
1507 See also Documentation/fault-injection/.
1510 Format: { initns | none }
1511 See Documentation/admin-guide/sysctl/net.rst for
1512 fb_tunnels_only_for_init_ns
1515 See Documentation/admin-guide/blockdev/floppy.rst.
1517 force_pal_cache_flush
1518 [IA-64] Avoid check_sal_cache_flush which may hang on
1519 buggy SAL_CACHE_FLUSH implementations. Using this
1520 parameter will force ia64_sal_cache_flush to call
1521 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1524 Forcefully enable Physical Address Extension (PAE).
1525 Many Pentium M systems disable PAE but may have a
1526 functionally usable PAE implementation.
1527 Warning: use of this parameter will taint the kernel
1528 and may cause unknown problems.
1531 [FTRACE] will set and start the specified tracer
1532 as early as possible in order to facilitate early
1535 ftrace_boot_snapshot
1536 [FTRACE] On boot up, a snapshot will be taken of the
1537 ftrace ring buffer that can be read at:
1538 /sys/kernel/tracing/snapshot.
1539 This is useful if you need tracing information from kernel
1540 boot up that is likely to be overridden by user space
1541 start up functionality.
1543 ftrace_dump_on_oops[=orig_cpu]
1544 [FTRACE] will dump the trace buffers on oops.
1545 If no parameter is passed, ftrace will dump
1546 buffers of all CPUs, but if you pass orig_cpu, it will
1547 dump only the buffer of the CPU that triggered the
1550 ftrace_filter=[function-list]
1551 [FTRACE] Limit the functions traced by the function
1552 tracer at boot up. function-list is a comma-separated
1553 list of functions. This list can be changed at run
1554 time by the set_ftrace_filter file in the debugfs
1557 ftrace_notrace=[function-list]
1558 [FTRACE] Do not trace the functions specified in
1559 function-list. This list can be changed at run time
1560 by the set_ftrace_notrace file in the debugfs
1563 ftrace_graph_filter=[function-list]
1564 [FTRACE] Limit the top level callers functions traced
1565 by the function graph tracer at boot up.
1566 function-list is a comma-separated list of functions
1567 that can be changed at run time by the
1568 set_graph_function file in the debugfs tracing directory.
1570 ftrace_graph_notrace=[function-list]
1571 [FTRACE] Do not trace from the functions specified in
1572 function-list. This list is a comma-separated list of
1573 functions that can be changed at run time by the
1574 set_graph_notrace file in the debugfs tracing directory.
1576 ftrace_graph_max_depth=<uint>
1577 [FTRACE] Used with the function graph tracer. This is
1578 the max depth it will trace into a function. This value
1579 can be changed at run time by the max_graph_depth file
1580 in the tracefs tracing directory. default: 0 (no limit)
1582 fw_devlink= [KNL] Create device links between consumer and supplier
1583 devices by scanning the firmware to infer the
1584 consumer/supplier relationships. This feature is
1585 especially useful when drivers are loaded as modules as
1586 it ensures proper ordering of tasks like device probing
1587 (suppliers first, then consumers), supplier boot state
1588 clean up (only after all consumers have probed),
1589 suspend/resume & runtime PM (consumers first, then
1591 Format: { off | permissive | on | rpm }
1592 off -- Don't create device links from firmware info.
1593 permissive -- Create device links from firmware info
1594 but use it only for ordering boot state clean
1595 up (sync_state() calls).
1596 on -- Create device links from firmware info and use it
1597 to enforce probe and suspend/resume ordering.
1598 rpm -- Like "on", but also use to order runtime PM.
1600 fw_devlink.strict=<bool>
1601 [KNL] Treat all inferred dependencies as mandatory
1602 dependencies. This only applies for fw_devlink=on|rpm.
1606 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1607 support via parallel port (up to 5 devices per port)
1608 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1609 See also Documentation/input/devices/joystick-parport.rst
1613 gart_fix_e820= [X86-64] disable the fix e820 for K8 GART
1617 gather_data_sampling=
1618 [X86,INTEL] Control the Gather Data Sampling (GDS)
1621 Gather Data Sampling is a hardware vulnerability which
1622 allows unprivileged speculative access to data which was
1623 previously stored in vector registers.
1625 This issue is mitigated by default in updated microcode.
1626 The mitigation may have a performance impact but can be
1627 disabled. On systems without the microcode mitigation
1628 disabling AVX serves as a mitigation.
1630 force: Disable AVX to mitigate systems without
1631 microcode mitigation. No effect if the microcode
1632 mitigation is present. Known to cause crashes in
1633 userspace with buggy AVX enumeration.
1635 off: Disable GDS mitigation.
1637 gcov_persist= [GCOV] When non-zero (default), profiling data for
1638 kernel modules is saved and remains accessible via
1639 debugfs, even when the module is unloaded/reloaded.
1640 When zero, profiling data is discarded and associated
1641 debugfs files are removed at module unload time.
1643 goldfish [X86] Enable the goldfish android emulator platform.
1644 Don't use this when you are not running on the
1647 gpio-mockup.gpio_mockup_ranges
1648 [HW] Sets the ranges of gpiochip of for this device.
1649 Format: <start1>,<end1>,<start2>,<end2>...
1650 gpio-mockup.gpio_mockup_named_lines
1651 [HW] Let the driver know GPIO lines should be named.
1653 gpt [EFI] Forces disk with valid GPT signature but
1654 invalid Protective MBR to be treated as GPT. If the
1655 primary GPT is corrupted, it enables the backup/alternate
1656 GPT to be used instead.
1658 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1659 the "Enable 0" bit of the configuration register.
1662 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1663 the "Enable 0" bit of the configuration register.
1666 grcan.select= [HW] Select which physical interface to use.
1669 grcan.txsize= [HW] Sets the size of the tx buffer.
1670 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1672 grcan.rxsize= [HW] Sets the size of the rx buffer.
1673 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1677 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
1678 hardening is enabled for this boot. Hardened
1679 usercopy checking is used to protect the kernel
1680 from reading or writing beyond known memory
1681 allocation boundaries as a proactive defense
1682 against bounds-checking flaws in the kernel's
1683 copy_to_user()/copy_from_user() interface.
1684 on Perform hardened usercopy checks (default).
1685 off Disable hardened usercopy checks.
1687 hardlockup_all_cpu_backtrace=
1688 [KNL] Should the hard-lockup detector generate
1689 backtraces on all cpus.
1692 hashdist= [KNL,NUMA] Large hashes allocated during boot
1693 are distributed across NUMA nodes. Defaults on
1694 for 64-bit NUMA, off otherwise.
1695 Format: 0 | 1 (for off | on)
1697 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1699 hd= [EIDE] (E)IDE hard drive subsystem geometry
1700 Format: <cyl>,<head>,<sect>
1703 Disable Hardware Error Source Table (HEST) support;
1704 corresponding firmware-first mode error processing
1705 logic will be disabled.
1707 hibernate= [HIBERNATION]
1708 noresume Don't check if there's a hibernation image
1709 present during boot.
1710 nocompress Don't compress/decompress hibernation images.
1711 no Disable hibernation and resume.
1712 protect_image Turn on image protection during restoration
1713 (that will set all pages holding image data
1714 during restoration read-only).
1716 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1717 size of <nn>. This works even on boxes that have no
1718 highmem otherwise. This also works to reduce highmem
1719 size on bigger boxes.
1721 highres= [KNL] Enable/disable high resolution timer mode.
1722 Valid parameters: "on", "off"
1727 hostname= [KNL] Set the hostname (aka UTS nodename).
1729 This allows setting the system's hostname during early
1730 startup. This sets the name returned by gethostname.
1731 Using this parameter to set the hostname makes it
1732 possible to ensure the hostname is correctly set before
1733 any userspace processes run, avoiding the possibility
1734 that a process may call gethostname before the hostname
1735 has been explicitly set, resulting in the calling
1736 process getting an incorrect result. The string must
1737 not exceed the maximum allowed hostname length (usually
1738 64 characters) and will be truncated otherwise.
1740 hpet= [X86-32,HPET] option to control HPET usage
1741 Format: { enable (default) | disable | force |
1743 disable: disable HPET and use PIT instead
1744 force: allow force enabled of undocumented chips (ICH4,
1746 verbose: show contents of HPET registers during setup
1748 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1749 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1751 hugepages= [HW] Number of HugeTLB pages to allocate at boot.
1752 If this follows hugepagesz (below), it specifies
1753 the number of pages of hugepagesz to be allocated.
1754 If this is the first HugeTLB parameter on the command
1755 line, it specifies the number of pages to allocate for
1756 the default huge page size. If using node format, the
1757 number of pages to allocate per-node can be specified.
1758 See also Documentation/admin-guide/mm/hugetlbpage.rst.
1759 Format: <integer> or (node format)
1760 <node>:<integer>[,<node>:<integer>]
1763 [HW] The size of the HugeTLB pages. This is used in
1764 conjunction with hugepages (above) to allocate huge
1765 pages of a specific size at boot. The pair
1766 hugepagesz=X hugepages=Y can be specified once for
1767 each supported huge page size. Huge page sizes are
1768 architecture dependent. See also
1769 Documentation/admin-guide/mm/hugetlbpage.rst.
1772 hugetlb_cma= [HW,CMA] The size of a CMA area used for allocation
1773 of gigantic hugepages. Or using node format, the size
1774 of a CMA area per node can be specified.
1775 Format: nn[KMGTPE] or (node format)
1776 <node>:nn[KMGTPE][,<node>:nn[KMGTPE]]
1778 Reserve a CMA area of given size and allocate gigantic
1779 hugepages using the CMA allocator. If enabled, the
1780 boot-time allocation of gigantic hugepages is skipped.
1782 hugetlb_free_vmemmap=
1783 [KNL] Reguires CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP
1785 Control if HugeTLB Vmemmap Optimization (HVO) is enabled.
1786 Allows heavy hugetlb users to free up some more
1787 memory (7 * PAGE_SIZE for each 2MB hugetlb page).
1788 Format: { on | off (default) }
1793 Built with CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP_DEFAULT_ON=y,
1796 Note that the vmemmap pages may be allocated from the added
1797 memory block itself when memory_hotplug.memmap_on_memory is
1798 enabled, those vmemmap pages cannot be optimized even if this
1799 feature is enabled. Other vmemmap pages not allocated from
1800 the added memory block itself do not be affected.
1803 [KNL] Should the hung task detector generate panics.
1806 A value of 1 instructs the kernel to panic when a
1807 hung task is detected. The default value is controlled
1808 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1809 option. The value selected by this boot parameter can
1810 be changed later by the kernel.hung_task_panic sysctl.
1812 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1813 terminal devices. Valid values: 0..8
1814 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1815 If specified, z/VM IUCV HVC accepts connections
1816 from listed z/VM user IDs only.
1818 hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations
1819 which allow the hypervisor to 'idle' the
1820 guest on lock contention.
1823 Do not unregister boot console at start. This is only
1824 useful for debugging when something happens in the window
1825 between unregistering the boot console and initializing
1828 i2c_bus= [HW] Override the default board specific I2C bus speed
1829 or register an additional I2C bus that is not
1830 registered from board initialization code.
1834 i8042.debug [HW] Toggle i8042 debug mode
1835 i8042.unmask_kbd_data
1836 [HW] Enable printing of interrupt data from the KBD port
1837 (disabled by default, and as a pre-condition
1838 requires that i8042.debug=1 be enabled)
1839 i8042.direct [HW] Put keyboard port into non-translated mode
1840 i8042.dumbkbd [HW] Pretend that controller can only read data from
1841 keyboard and cannot control its state
1842 (Don't attempt to blink the leds)
1843 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1844 i8042.nokbd [HW] Don't check/create keyboard port
1845 i8042.noloop [HW] Disable the AUX Loopback command while probing
1847 i8042.nomux [HW] Don't check presence of an active multiplexing
1849 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1851 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1852 i8042.reset [HW] Reset the controller during init, cleanup and
1853 suspend-to-ram transitions, only during s2r
1854 transitions, or never reset
1855 Format: { 1 | Y | y | 0 | N | n }
1856 1, Y, y: always reset controller
1857 0, N, n: don't ever reset controller
1858 Default: only on s2r transitions on x86; most other
1859 architectures force reset to be always executed
1860 i8042.unlock [HW] Unlock (ignore) the keylock
1861 i8042.kbdreset [HW] Reset device connected to KBD port
1863 [HW] Allow deferred probing upon i8042 probe errors
1867 i915.invert_brightness=
1868 [DRM] Invert the sense of the variable that is used to
1869 set the brightness of the panel backlight. Normally a
1870 brightness value of 0 indicates backlight switched off,
1871 and the maximum of the brightness value sets the backlight
1872 to maximum brightness. If this parameter is set to 0
1873 (default) and the machine requires it, or this parameter
1874 is set to 1, a brightness value of 0 sets the backlight
1875 to maximum brightness, and the maximum of the brightness
1876 value switches the backlight off.
1877 -1 -- never invert brightness
1878 0 -- machine default
1879 1 -- force brightness inversion
1882 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1886 Format: idle=poll, idle=halt, idle=nomwait
1887 Poll forces a polling idle loop that can slightly
1888 improve the performance of waking up a idle CPU, but
1889 will use a lot of power and make the system run hot.
1891 idle=halt: Halt is forced to be used for CPU idle.
1892 In such case C2/C3 won't be used again.
1893 idle=nomwait: Disable mwait for CPU C-states
1897 Allow force disabling of Shared Virtual Memory (SVA)
1898 support for the idxd driver. By default it is set to
1901 idxd.tc_override= [HW]
1903 Allow override of default traffic class configuration
1904 for the device. By default it is set to false (0).
1906 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1907 Format: { strict | legacy | 2008 | relaxed }
1910 Choose which programs will be accepted for execution
1911 based on the IEEE 754 NaN encoding(s) supported by
1912 the FPU and the NaN encoding requested with the value
1913 of an ELF file header flag individually set by each
1914 binary. Hardware implementations are permitted to
1915 support either or both of the legacy and the 2008 NaN
1918 Available settings are as follows:
1919 strict accept binaries that request a NaN encoding
1920 supported by the FPU
1921 legacy only accept legacy-NaN binaries, if supported
1923 2008 only accept 2008-NaN binaries, if supported
1925 relaxed accept any binaries regardless of whether
1926 supported by the FPU
1928 The FPU emulator is always able to support both NaN
1929 encodings, so if no FPU hardware is present or it has
1930 been disabled with 'nofpu', then the settings of
1931 'legacy' and '2008' strap the emulator accordingly,
1932 'relaxed' straps the emulator for both legacy-NaN and
1933 2008-NaN, whereas 'strict' enables legacy-NaN only on
1934 legacy processors and both NaN encodings on MIPS32 or
1937 The setting for ABS.fmt/NEG.fmt instruction execution
1938 mode generally follows that for the NaN encoding,
1939 except where unsupported by hardware.
1941 ignore_loglevel [KNL]
1942 Ignore loglevel setting - this will print /all/
1943 kernel messages to the console. Useful for debugging.
1944 We also add it as printk module parameter, so users
1945 could change it dynamically, usually by
1946 /sys/module/printk/parameters/ignore_loglevel.
1949 Ignore RLIMIT_DATA setting for data mappings,
1950 print warning at first misuse. Can be changed via
1951 /sys/module/kernel/parameters/ignore_rlimit_data.
1953 ihash_entries= [KNL]
1954 Set number of hash buckets for inode cache.
1956 ima_appraise= [IMA] appraise integrity measurements
1957 Format: { "off" | "enforce" | "fix" | "log" }
1960 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
1961 The builtin appraise policy appraises all files
1964 ima_canonical_fmt [IMA]
1965 Use the canonical format for the binary runtime
1966 measurements, instead of host native format.
1969 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1973 The list of supported hash algorithms is defined
1974 in crypto/hash_info.h.
1977 The builtin policies to load during IMA setup.
1978 Format: "tcb | appraise_tcb | secure_boot |
1979 fail_securely | critical_data"
1981 The "tcb" policy measures all programs exec'd, files
1982 mmap'd for exec, and all files opened with the read
1983 mode bit set by either the effective uid (euid=0) or
1986 The "appraise_tcb" policy appraises the integrity of
1987 all files owned by root.
1989 The "secure_boot" policy appraises the integrity
1990 of files (eg. kexec kernel image, kernel modules,
1991 firmware, policy, etc) based on file signatures.
1993 The "fail_securely" policy forces file signature
1994 verification failure also on privileged mounted
1995 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
1998 The "critical_data" policy measures kernel integrity
2001 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
2002 Load a policy which meets the needs of the Trusted
2003 Computing Base. This means IMA will measure all
2004 programs exec'd, files mmap'd for exec, and all files
2005 opened for read by uid=0.
2008 Select one of defined IMA measurements template formats.
2009 Formats: { "ima" | "ima-ng" | "ima-ngv2" | "ima-sig" |
2014 [IMA] Define a custom template format.
2015 Format: { "field1|...|fieldN" }
2017 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
2018 Format: <min_file_size>
2019 Set the minimal file size for using asynchronous hash.
2020 If left unspecified, ahash usage is disabled.
2022 ahash performance varies for different data sizes on
2023 different crypto accelerators. This option can be used
2024 to achieve the best performance for a particular HW.
2026 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
2028 Set hashing buffer size. Default: 4k.
2030 ahash performance varies for different chunk sizes on
2031 different crypto accelerators. This option can be used
2032 to achieve best performance for particular HW.
2036 Run specified binary instead of /sbin/init as init
2039 initcall_debug [KNL] Trace initcalls as they are executed. Useful
2040 for working out where the kernel is dying during
2043 initcall_blacklist= [KNL] Do not execute a comma-separated list of
2044 initcall functions. Useful for debugging built-in
2045 modules and initcalls.
2047 initramfs_async= [KNL]
2050 This parameter controls whether the initramfs
2051 image is unpacked asynchronously, concurrently
2052 with devices being probed and
2053 initialized. This should normally just work,
2054 but as a debugging aid, one can get the
2055 historical behaviour of the initramfs
2056 unpacking being completed before device_ and
2059 initrd= [BOOT] Specify the location of the initial ramdisk
2061 initrdmem= [KNL] Specify a physical address and size from which to
2062 load the initrd. If an initrd is compiled in or
2063 specified in the bootparams, it takes priority over this
2065 Format: ss[KMG],nn[KMG]
2068 init_on_alloc= [MM] Fill newly allocated pages and heap objects with
2071 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
2073 init_on_free= [MM] Fill freed pages and heap objects with zeroes.
2075 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
2077 init_pkru= [X86] Specify the default memory protection keys rights
2078 register contents for all processes. 0x55555554 by
2079 default (disallow access to all but pkey 0). Can
2080 override in debugfs after boot.
2082 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
2085 int_pln_enable [X86] Enable power limit notification interrupt
2087 integrity_audit=[IMA]
2088 Format: { "0" | "1" }
2089 0 -- basic integrity auditing messages. (Default)
2090 1 -- additional integrity auditing messages.
2092 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
2094 Enable intel iommu driver.
2096 Disable intel iommu driver.
2097 igfx_off [Default Off]
2098 By default, gfx is mapped as normal device. If a gfx
2099 device has a dedicated DMAR unit, the DMAR unit is
2100 bypassed by not enabling DMAR with this option. In
2101 this case, gfx device will use physical address for
2103 strict [Default Off]
2104 Deprecated, equivalent to iommu.strict=1.
2105 sp_off [Default Off]
2106 By default, super page will be supported if Intel IOMMU
2107 has the capability. With this option, super page will
2110 Enable the Intel IOMMU scalable mode if the hardware
2111 advertises that it has support for the scalable mode
2114 Disallow use of the Intel IOMMU scalable mode.
2115 tboot_noforce [Default Off]
2116 Do not force the Intel IOMMU enabled under tboot.
2117 By default, tboot will force Intel IOMMU on, which
2118 could harm performance of some high-throughput
2119 devices like 40GBit network cards, even if identity
2121 Note that using this option lowers the security
2122 provided by tboot because it makes the system
2123 vulnerable to DMA attacks.
2125 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
2126 0 disables intel_idle and fall back on acpi_idle.
2127 1 to 9 specify maximum depth of C-state.
2131 Do not enable intel_pstate as the default
2132 scaling driver for the supported processors
2134 Use intel_pstate as a scaling driver, but configure it
2135 to work with generic cpufreq governors (instead of
2136 enabling its internal governor). This mode cannot be
2137 used along with the hardware-managed P-states (HWP)
2140 Enable intel_pstate on systems that prohibit it by default
2141 in favor of acpi-cpufreq. Forcing the intel_pstate driver
2142 instead of acpi-cpufreq may disable platform features, such
2143 as thermal controls and power capping, that rely on ACPI
2144 P-States information being indicated to OSPM and therefore
2145 should be used with caution. This option does not work with
2146 processors that aren't supported by the intel_pstate driver
2147 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
2149 Do not enable hardware P state control (HWP)
2152 Only load intel_pstate on systems which support
2153 hardware P state control (HWP) if available.
2155 Enforce ACPI _PPC performance limits. If the Fixed ACPI
2156 Description Table, specifies preferred power management
2157 profile as "Enterprise Server" or "Performance Server",
2158 then this feature is turned on by default.
2160 Allow per-logical-CPU P-State performance control limits using
2161 cpufreq sysfs interface
2163 intremap= [X86-64, Intel-IOMMU]
2164 on enable Interrupt Remapping (default)
2165 off disable Interrupt Remapping
2166 nosid disable Source ID checking
2168 BIOS x2APIC opt-out request will be ignored
2169 nopost disable Interrupt Posting
2171 iomem= Disable strict checking of access to MMIO memory
2172 strict regions from userspace.
2187 nobypass [PPC/POWERNV]
2188 Disable IOMMU bypass, using IOMMU for PCI devices.
2190 iommu.forcedac= [ARM64, X86] Control IOVA allocation for PCI devices.
2191 Format: { "0" | "1" }
2192 0 - Try to allocate a 32-bit DMA address first, before
2193 falling back to the full range if needed.
2194 1 - Allocate directly from the full usable range,
2195 forcing Dual Address Cycle for PCI cards supporting
2196 greater than 32-bit addressing.
2198 iommu.strict= [ARM64, X86] Configure TLB invalidation behaviour
2199 Format: { "0" | "1" }
2201 Request that DMA unmap operations use deferred
2202 invalidation of hardware TLBs, for increased
2203 throughput at the cost of reduced device isolation.
2204 Will fall back to strict mode if not supported by
2205 the relevant IOMMU driver.
2207 DMA unmap operations invalidate IOMMU hardware TLBs
2209 unset - Use value of CONFIG_IOMMU_DEFAULT_DMA_{LAZY,STRICT}.
2210 Note: on x86, strict mode specified via one of the
2211 legacy driver-specific options takes precedence.
2214 [ARM64, X86] Configure DMA to bypass the IOMMU by default.
2215 Format: { "0" | "1" }
2216 0 - Use IOMMU translation for DMA.
2217 1 - Bypass the IOMMU for DMA.
2218 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
2220 io7= [HW] IO7 for Marvel-based Alpha systems
2221 See comment before marvel_specify_io7 in
2222 arch/alpha/kernel/core_marvel.c.
2224 io_delay= [X86] I/O delay method
2226 Standard port 0x80 based delay
2228 Alternate port 0xed based delay (needed on some systems)
2230 Simple two microseconds delay
2235 See Documentation/admin-guide/nfs/nfsroot.rst.
2237 ipcmni_extend [KNL] Extend the maximum number of unique System V
2238 IPC identifiers from 32,768 to 16,777,216.
2240 irqaffinity= [SMP] Set the default irq affinity mask
2241 The argument is a cpu list, as described above.
2243 irqchip.gicv2_force_probe=
2246 Force the kernel to look for the second 4kB page
2247 of a GICv2 controller even if the memory range
2248 exposed by the device tree is too small.
2250 irqchip.gicv3_nolpi=
2252 Force the kernel to ignore the availability of
2253 LPIs (and by consequence ITSs). Intended for system
2254 that use the kernel as a bootloader, and thus want
2255 to let secondary kernels in charge of setting up
2258 irqchip.gicv3_pseudo_nmi= [ARM64]
2259 Enables support for pseudo-NMIs in the kernel. This
2260 requires the kernel to be built with
2261 CONFIG_ARM64_PSEUDO_NMI.
2264 When an interrupt is not handled search all handlers
2265 for it. Intended to get systems with badly broken
2269 When an interrupt is not handled search all handlers
2270 for it. Also check all handlers each timer
2271 interrupt. Intended to get systems with badly broken
2275 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
2277 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
2278 [Deprecated - use cpusets instead]
2279 Format: [flag-list,]<cpu-list>
2281 Specify one or more CPUs to isolate from disturbances
2282 specified in the flag list (default: domain):
2285 Disable the tick when a single task runs.
2287 A residual 1Hz tick is offloaded to workqueues, which you
2288 need to affine to housekeeping through the global
2289 workqueue's affinity configured via the
2290 /sys/devices/virtual/workqueue/cpumask sysfs file, or
2291 by using the 'domain' flag described below.
2293 NOTE: by default the global workqueue runs on all CPUs,
2294 so to protect individual CPUs the 'cpumask' file has to
2295 be configured manually after bootup.
2298 Isolate from the general SMP balancing and scheduling
2299 algorithms. Note that performing domain isolation this way
2300 is irreversible: it's not possible to bring back a CPU to
2301 the domains once isolated through isolcpus. It's strongly
2302 advised to use cpusets instead to disable scheduler load
2303 balancing through the "cpuset.sched_load_balance" file.
2304 It offers a much more flexible interface where CPUs can
2305 move in and out of an isolated set anytime.
2307 You can move a process onto or off an "isolated" CPU via
2308 the CPU affinity syscalls or cpuset.
2309 <cpu number> begins at 0 and the maximum value is
2310 "number of CPUs in system - 1".
2314 Isolate from being targeted by managed interrupts
2315 which have an interrupt mask containing isolated
2316 CPUs. The affinity of managed interrupts is
2317 handled by the kernel and cannot be changed via
2318 the /proc/irq/* interfaces.
2320 This isolation is best effort and only effective
2321 if the automatically assigned interrupt mask of a
2322 device queue contains isolated and housekeeping
2323 CPUs. If housekeeping CPUs are online then such
2324 interrupts are directed to the housekeeping CPU
2325 so that IO submitted on the housekeeping CPU
2326 cannot disturb the isolated CPU.
2328 If a queue's affinity mask contains only isolated
2329 CPUs then this parameter has no effect on the
2330 interrupt routing decision, though interrupts are
2331 only delivered when tasks running on those
2332 isolated CPUs submit IO. IO submitted on
2333 housekeeping CPUs has no influence on those
2336 The format of <cpu-list> is described above.
2340 ivrs_ioapic [HW,X86-64]
2341 Provide an override to the IOAPIC-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 IOAPIC-ID decimal 10 to
2346 PCI segment 0x1 and PCI device 00:14.0,
2347 write the parameter as:
2348 ivrs_ioapic=10@0001:00:14.0
2351 * To map IOAPIC-ID decimal 10 to PCI device 00:14.0
2352 write the parameter as:
2353 ivrs_ioapic[10]=00:14.0
2354 * To map IOAPIC-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_hpet [HW,X86-64]
2359 Provide an override to the HPET-ID<->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 HPET-ID decimal 10 to
2364 PCI segment 0x1 and PCI device 00:14.0,
2365 write the parameter as:
2366 ivrs_hpet=10@0001:00:14.0
2369 * To map HPET-ID decimal 0 to PCI device 00:14.0
2370 write the parameter as:
2371 ivrs_hpet[0]=00:14.0
2372 * To map HPET-ID decimal 10 to PCI segment 0x1 and
2373 PCI device 00:14.0 write the parameter as:
2374 ivrs_ioapic[10]=0001:00:14.0
2376 ivrs_acpihid [HW,X86-64]
2377 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2378 mapping provided in the IVRS ACPI table.
2379 By default, PCI segment is 0, and can be omitted.
2381 For example, to map UART-HID:UID AMD0020:0 to
2382 PCI segment 0x1 and PCI device ID 00:14.5,
2383 write the parameter as:
2384 ivrs_acpihid=AMD0020:0@0001:00:14.5
2387 * To map UART-HID:UID AMD0020:0 to PCI segment is 0,
2388 PCI device ID 00:14.5, write the parameter as:
2389 ivrs_acpihid[00:14.5]=AMD0020:0
2390 * To map UART-HID:UID AMD0020:0 to PCI segment 0x1 and
2391 PCI device ID 00:14.5, write the parameter as:
2392 ivrs_acpihid[0001:00:14.5]=AMD0020:0
2394 js= [HW,JOY] Analog joystick
2395 See Documentation/input/joydev/joystick.rst.
2398 When CONFIG_RANDOMIZE_BASE is set, this disables
2399 kernel and module base offset ASLR (Address Space
2400 Layout Randomization).
2403 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2404 report on every invalid memory access. Without this
2405 parameter KASAN will print report only for the first
2410 kernelcore= [KNL,X86,IA-64,PPC]
2411 Format: nn[KMGTPE] | nn% | "mirror"
2412 This parameter specifies the amount of memory usable by
2413 the kernel for non-movable allocations. The requested
2414 amount is spread evenly throughout all nodes in the
2415 system as ZONE_NORMAL. The remaining memory is used for
2416 movable memory in its own zone, ZONE_MOVABLE. In the
2417 event, a node is too small to have both ZONE_NORMAL and
2418 ZONE_MOVABLE, kernelcore memory will take priority and
2419 other nodes will have a larger ZONE_MOVABLE.
2421 ZONE_MOVABLE is used for the allocation of pages that
2422 may be reclaimed or moved by the page migration
2423 subsystem. Note that allocations like PTEs-from-HighMem
2424 still use the HighMem zone if it exists, and the Normal
2425 zone if it does not.
2427 It is possible to specify the exact amount of memory in
2428 the form of "nn[KMGTPE]", a percentage of total system
2429 memory in the form of "nn%", or "mirror". If "mirror"
2430 option is specified, mirrored (reliable) memory is used
2431 for non-movable allocations and remaining memory is used
2432 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2433 are exclusive, so you cannot specify multiple forms.
2435 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
2436 Format: <Controller#>[,poll interval]
2437 The controller # is the number of the ehci usb debug
2438 port as it is probed via PCI. The poll interval is
2439 optional and is the number seconds in between
2440 each poll cycle to the debug port in case you need
2441 the functionality for interrupting the kernel with
2442 gdb or control-c on the dbgp connection. When
2443 not using this parameter you use sysrq-g to break into
2444 the kernel debugger.
2446 kgdboc= [KGDB,HW] kgdb over consoles.
2447 Requires a tty driver that supports console polling,
2448 or a supported polling keyboard driver (non-usb).
2449 Serial only format: <serial_device>[,baud]
2450 keyboard only format: kbd
2451 keyboard and serial format: kbd,<serial_device>[,baud]
2452 Optional Kernel mode setting:
2453 kms, kbd format: kms,kbd
2454 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2456 kgdboc_earlycon= [KGDB,HW]
2457 If the boot console provides the ability to read
2458 characters and can work in polling mode, you can use
2459 this parameter to tell kgdb to use it as a backend
2460 until the normal console is registered. Intended to
2461 be used together with the kgdboc parameter which
2462 specifies the normal console to transition to.
2464 The name of the early console should be specified
2465 as the value of this parameter. Note that the name of
2466 the early console might be different than the tty
2467 name passed to kgdboc. It's OK to leave the value
2468 blank and the first boot console that implements
2469 read() will be picked.
2471 kgdbwait [KGDB] Stop kernel execution and enter the
2472 kernel debugger at the earliest opportunity.
2474 kmac= [MIPS] Korina ethernet MAC address.
2475 Configure the RouterBoard 532 series on-chip
2476 Ethernet adapter MAC address.
2478 kmemleak= [KNL] Boot-time kmemleak enable/disable
2479 Valid arguments: on, off
2481 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2484 kprobe_event=[probe-list]
2485 [FTRACE] Add kprobe events and enable at boot time.
2486 The probe-list is a semicolon delimited list of probe
2487 definitions. Each definition is same as kprobe_events
2488 interface, but the parameters are comma delimited.
2489 For example, to add a kprobe event on vfs_read with
2490 arg1 and arg2, add to the command line;
2492 kprobe_event=p,vfs_read,$arg1,$arg2
2494 See also Documentation/trace/kprobetrace.rst "Kernel
2495 Boot Parameter" section.
2497 kpti= [ARM64] Control page table isolation of user
2498 and kernel address spaces.
2499 Default: enabled on cores which need mitigation.
2503 kunit.enable= [KUNIT] Enable executing KUnit tests. Requires
2504 CONFIG_KUNIT to be set to be fully enabled. The
2505 default value can be overridden via
2506 KUNIT_DEFAULT_ENABLED.
2507 Default is 1 (enabled)
2509 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2510 Default is 0 (don't ignore, but inject #GP)
2512 kvm.eager_page_split=
2513 [KVM,X86] Controls whether or not KVM will try to
2514 proactively split all huge pages during dirty logging.
2515 Eager page splitting reduces interruptions to vCPU
2516 execution by eliminating the write-protection faults
2517 and MMU lock contention that would otherwise be
2518 required to split huge pages lazily.
2520 VM workloads that rarely perform writes or that write
2521 only to a small region of VM memory may benefit from
2522 disabling eager page splitting to allow huge pages to
2523 still be used for reads.
2525 The behavior of eager page splitting depends on whether
2526 KVM_DIRTY_LOG_INITIALLY_SET is enabled or disabled. If
2527 disabled, all huge pages in a memslot will be eagerly
2528 split when dirty logging is enabled on that memslot. If
2529 enabled, eager page splitting will be performed during
2530 the KVM_CLEAR_DIRTY ioctl, and only for the pages being
2533 Eager page splitting is only supported when kvm.tdp_mmu=Y.
2537 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2538 Default is false (don't support).
2541 [KVM] Controls the software workaround for the
2542 X86_BUG_ITLB_MULTIHIT bug.
2543 force : Always deploy workaround.
2544 off : Never deploy workaround.
2545 auto : Deploy workaround based on the presence of
2546 X86_BUG_ITLB_MULTIHIT.
2550 If the software workaround is enabled for the host,
2551 guests do need not to enable it for nested guests.
2553 kvm.nx_huge_pages_recovery_ratio=
2554 [KVM] Controls how many 4KiB pages are periodically zapped
2555 back to huge pages. 0 disables the recovery, otherwise if
2556 the value is N KVM will zap 1/Nth of the 4KiB pages every
2557 period (see below). The default is 60.
2559 kvm.nx_huge_pages_recovery_period_ms=
2560 [KVM] Controls the time period at which KVM zaps 4KiB pages
2561 back to huge pages. If the value is a non-zero N, KVM will
2562 zap a portion (see ratio above) of the pages every N msecs.
2563 If the value is 0 (the default), KVM will pick a period based
2564 on the ratio, such that a page is zapped after 1 hour on average.
2566 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM.
2567 Default is 1 (enabled)
2569 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU)
2571 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode.
2574 [KVM,ARM] Select one of KVM/arm64's modes of operation.
2576 none: Forcefully disable KVM.
2578 nvhe: Standard nVHE-based mode, without support for
2581 protected: nVHE-based mode with support for guests whose
2582 state is kept private from the host.
2584 Defaults to VHE/nVHE based on hardware support. Setting
2585 mode to "protected" will disable kexec and hibernation
2588 kvm-arm.vgic_v3_group0_trap=
2589 [KVM,ARM] Trap guest accesses to GICv3 group-0
2592 kvm-arm.vgic_v3_group1_trap=
2593 [KVM,ARM] Trap guest accesses to GICv3 group-1
2596 kvm-arm.vgic_v3_common_trap=
2597 [KVM,ARM] Trap guest accesses to GICv3 common
2600 kvm-arm.vgic_v4_enable=
2601 [KVM,ARM] Allow use of GICv4 for direct injection of
2604 kvm_cma_resv_ratio=n [PPC]
2605 Reserves given percentage from system memory area for
2606 contiguous memory allocation for KVM hash pagetable
2608 By default it reserves 5% of total system memory.
2612 kvm-intel.ept= [KVM,Intel] Disable extended page tables
2613 (virtualized MMU) support on capable Intel chips.
2614 Default is 1 (enabled)
2616 kvm-intel.emulate_invalid_guest_state=
2617 [KVM,Intel] Disable emulation of invalid guest state.
2618 Ignored if kvm-intel.enable_unrestricted_guest=1, as
2619 guest state is never invalid for unrestricted guests.
2620 This param doesn't apply to nested guests (L2), as KVM
2621 never emulates invalid L2 guest state.
2622 Default is 1 (enabled)
2624 kvm-intel.flexpriority=
2625 [KVM,Intel] Disable FlexPriority feature (TPR shadow).
2626 Default is 1 (enabled)
2629 [KVM,Intel] Enable VMX nesting (nVMX).
2630 Default is 0 (disabled)
2632 kvm-intel.unrestricted_guest=
2633 [KVM,Intel] Disable unrestricted guest feature
2634 (virtualized real and unpaged mode) on capable
2635 Intel chips. Default is 1 (enabled)
2637 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2640 Valid arguments: never, cond, always
2642 always: L1D cache flush on every VMENTER.
2643 cond: Flush L1D on VMENTER only when the code between
2644 VMEXIT and VMENTER can leak host memory.
2645 never: Disables the mitigation
2647 Default is cond (do L1 cache flush in specific instances)
2649 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification
2650 feature (tagged TLBs) on capable Intel chips.
2651 Default is 1 (enabled)
2653 l1d_flush= [X86,INTEL]
2654 Control mitigation for L1D based snooping vulnerability.
2656 Certain CPUs are vulnerable to an exploit against CPU
2657 internal buffers which can forward information to a
2658 disclosure gadget under certain conditions.
2660 In vulnerable processors, the speculatively
2661 forwarded data can be used in a cache side channel
2662 attack, to access data to which the attacker does
2663 not have direct access.
2665 This parameter controls the mitigation. The
2668 on - enable the interface for the mitigation
2670 l1tf= [X86] Control mitigation of the L1TF vulnerability on
2673 The kernel PTE inversion protection is unconditionally
2674 enabled and cannot be disabled.
2677 Provides all available mitigations for the
2678 L1TF vulnerability. Disables SMT and
2679 enables all mitigations in the
2680 hypervisors, i.e. unconditional L1D flush.
2682 SMT control and L1D flush control via the
2683 sysfs interface is still possible after
2684 boot. Hypervisors will issue a warning
2685 when the first VM is started in a
2686 potentially insecure configuration,
2687 i.e. SMT enabled or L1D flush disabled.
2690 Same as 'full', but disables SMT and L1D
2691 flush runtime control. Implies the
2692 'nosmt=force' command line option.
2693 (i.e. sysfs control of SMT is disabled.)
2696 Leaves SMT enabled and enables the default
2697 hypervisor mitigation, i.e. conditional
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.
2709 Disables SMT and enables the default
2710 hypervisor mitigation.
2712 SMT control and L1D flush control via the
2713 sysfs interface is still possible after
2714 boot. Hypervisors will issue a warning
2715 when the first VM is started in a
2716 potentially insecure configuration,
2717 i.e. SMT enabled or L1D flush disabled.
2720 Same as 'flush', but hypervisors will not
2721 warn when a VM is started in a potentially
2722 insecure configuration.
2725 Disables hypervisor mitigations and doesn't
2727 It also drops the swap size and available
2728 RAM limit restriction on both hypervisor and
2733 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2739 lapic [X86-32,APIC] Enable the local APIC even if BIOS
2742 lapic= [X86,APIC] Do not use TSC deadline
2743 value for LAPIC timer one-shot implementation. Default
2744 back to the programmable timer unit in the LAPIC.
2745 Format: notscdeadline
2747 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
2750 libata.dma= [LIBATA] DMA control
2751 libata.dma=0 Disable all PATA and SATA DMA
2752 libata.dma=1 PATA and SATA Disk DMA only
2753 libata.dma=2 ATAPI (CDROM) DMA only
2754 libata.dma=4 Compact Flash DMA only
2755 Combinations also work, so libata.dma=3 enables DMA
2756 for disks and CDROMs, but not CFs.
2758 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2759 libata.ignore_hpa=0 keep BIOS limits (default)
2760 libata.ignore_hpa=1 ignore limits, using full disk
2762 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2766 libata.force= [LIBATA] Force configurations. The format is a comma-
2767 separated list of "[ID:]VAL" where ID is PORT[.DEVICE].
2768 PORT and DEVICE are decimal numbers matching port, link
2769 or device. Basically, it matches the ATA ID string
2770 printed on console by libata. If the whole ID part is
2771 omitted, the last PORT and DEVICE values are used. If
2772 ID hasn't been specified yet, the configuration applies
2773 to all ports, links and devices.
2775 If only DEVICE is omitted, the parameter applies to
2776 the port and all links and devices behind it. DEVICE
2777 number of 0 either selects the first device or the
2778 first fan-out link behind PMP device. It does not
2779 select the host link. DEVICE number of 15 selects the
2780 host link and device attached to it.
2782 The VAL specifies the configuration to force. As long
2783 as there is no ambiguity, shortcut notation is allowed.
2784 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2785 The following configurations can be forced.
2787 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2788 Any ID with matching PORT is used.
2790 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2792 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2793 udma[/][16,25,33,44,66,100,133] notation is also
2796 * nohrst, nosrst, norst: suppress hard, soft and both
2799 * rstonce: only attempt one reset during hot-unplug
2802 * [no]dbdelay: Enable or disable the extra 200ms delay
2803 before debouncing a link PHY and device presence
2806 * [no]ncq: Turn on or off NCQ.
2808 * [no]ncqtrim: Enable or disable queued DSM TRIM.
2810 * [no]ncqati: Enable or disable NCQ trim on ATI chipset.
2812 * [no]trim: Enable or disable (unqueued) TRIM.
2814 * trim_zero: Indicate that TRIM command zeroes data.
2816 * max_trim_128m: Set 128M maximum trim size limit.
2818 * [no]dma: Turn on or off DMA transfers.
2820 * atapi_dmadir: Enable ATAPI DMADIR bridge support.
2822 * atapi_mod16_dma: Enable the use of ATAPI DMA for
2823 commands that are not a multiple of 16 bytes.
2825 * [no]dmalog: Enable or disable the use of the
2826 READ LOG DMA EXT command to access logs.
2828 * [no]iddevlog: Enable or disable access to the
2829 identify device data log.
2831 * [no]logdir: Enable or disable access to the general
2832 purpose log directory.
2834 * max_sec_128: Set transfer size limit to 128 sectors.
2836 * max_sec_1024: Set or clear transfer size limit to
2839 * max_sec_lba48: Set or clear transfer size limit to
2842 * [no]lpm: Enable or disable link power management.
2844 * [no]setxfer: Indicate if transfer speed mode setting
2847 * dump_id: Dump IDENTIFY data.
2849 * disable: Disable this device.
2851 If there are multiple matching configurations changing
2852 the same attribute, the last one is used.
2854 load_ramdisk= [RAM] [Deprecated]
2856 lockd.nlm_grace_period=P [NFS] Assign grace period.
2859 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2862 lockd.nlm_timeout=T [NFS] Assign timeout value.
2865 lockd.nlm_udpport=M [NFS] Assign UDP port.
2868 lockdown= [SECURITY]
2869 { integrity | confidentiality }
2870 Enable the kernel lockdown feature. If set to
2871 integrity, kernel features that allow userland to
2872 modify the running kernel are disabled. If set to
2873 confidentiality, kernel features that allow userland
2874 to extract confidential information from the kernel
2877 locktorture.nreaders_stress= [KNL]
2878 Set the number of locking read-acquisition kthreads.
2879 Defaults to being automatically set based on the
2880 number of online CPUs.
2882 locktorture.nwriters_stress= [KNL]
2883 Set the number of locking write-acquisition kthreads.
2885 locktorture.onoff_holdoff= [KNL]
2886 Set time (s) after boot for CPU-hotplug testing.
2888 locktorture.onoff_interval= [KNL]
2889 Set time (s) between CPU-hotplug operations, or
2890 zero to disable CPU-hotplug testing.
2892 locktorture.shuffle_interval= [KNL]
2893 Set task-shuffle interval (jiffies). Shuffling
2894 tasks allows some CPUs to go into dyntick-idle
2895 mode during the locktorture test.
2897 locktorture.shutdown_secs= [KNL]
2898 Set time (s) after boot system shutdown. This
2899 is useful for hands-off automated testing.
2901 locktorture.stat_interval= [KNL]
2902 Time (s) between statistics printk()s.
2904 locktorture.stutter= [KNL]
2905 Time (s) to stutter testing, for example,
2906 specifying five seconds causes the test to run for
2907 five seconds, wait for five seconds, and so on.
2908 This tests the locking primitive's ability to
2909 transition abruptly to and from idle.
2911 locktorture.torture_type= [KNL]
2912 Specify the locking implementation to test.
2914 locktorture.verbose= [KNL]
2915 Enable additional printk() statements.
2917 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
2920 loglevel= All Kernel Messages with a loglevel smaller than the
2921 console loglevel will be printed to the console. It can
2922 also be changed with klogd or other programs. The
2923 loglevels are defined as follows:
2925 0 (KERN_EMERG) system is unusable
2926 1 (KERN_ALERT) action must be taken immediately
2927 2 (KERN_CRIT) critical conditions
2928 3 (KERN_ERR) error conditions
2929 4 (KERN_WARNING) warning conditions
2930 5 (KERN_NOTICE) normal but significant condition
2931 6 (KERN_INFO) informational
2932 7 (KERN_DEBUG) debug-level messages
2934 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
2935 in bytes. n must be a power of two and greater
2936 than the minimal size. The minimal size is defined
2937 by LOG_BUF_SHIFT kernel config parameter. There is
2938 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2939 that allows to increase the default size depending on
2940 the number of CPUs. See init/Kconfig for more details.
2942 logo.nologo [FB] Disables display of the built-in Linux logo.
2943 This may be used to provide more screen space for
2944 kernel log messages and is useful when debugging
2945 kernel boot problems.
2947 lp=0 [LP] Specify parallel ports to use, e.g,
2948 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2949 lp=reset first parallel port). 'lp=0' disables the
2950 lp=auto printer driver. 'lp=reset' (which can be
2951 specified in addition to the ports) causes
2952 attached printers to be reset. Using
2953 lp=port1,port2,... specifies the parallel ports
2954 to associate lp devices with, starting with
2955 lp0. A port specification may be 'none' to skip
2956 that lp device, or a parport name such as
2957 'parport0'. Specifying 'lp=auto' instead of a
2958 port specification list means that device IDs
2959 from each port should be examined, to see if
2960 an IEEE 1284-compliant printer is attached; if
2961 so, the driver will manage that printer.
2962 See also header of drivers/char/lp.c.
2965 Sets loops_per_jiffy to given constant, thus avoiding
2966 time-consuming boot-time autodetection (up to 250 ms per
2967 CPU). 0 enables autodetection (default). To determine
2968 the correct value for your kernel, boot with normal
2969 autodetection and see what value is printed. Note that
2970 on SMP systems the preset will be applied to all CPUs,
2971 which is likely to cause problems if your CPUs need
2972 significantly divergent settings. An incorrect value
2973 will cause delays in the kernel to be wrong, leading to
2974 unpredictable I/O errors and other breakage. Although
2975 unlikely, in the extreme case this might damage your
2979 Format: <io>,<irq>,<dma>
2981 lsm.debug [SECURITY] Enable LSM initialization debugging output.
2984 [SECURITY] Choose order of LSM initialization. This
2985 overrides CONFIG_LSM, and the "security=" parameter.
2987 machvec= [IA-64] Force the use of a particular machine-vector
2988 (machvec) in a generic kernel.
2989 Example: machvec=hpzx1
2991 machtype= [Loongson] Share the same kernel image file between
2992 different yeeloong laptops.
2993 Example: machtype=lemote-yeeloong-2f-7inch
2995 max_addr=nn[KMG] [KNL,BOOT,IA-64] All physical memory greater
2996 than or equal to this physical address is ignored.
2998 maxcpus= [SMP] Maximum number of processors that an SMP kernel
2999 will bring up during bootup. maxcpus=n : n >= 0 limits
3000 the kernel to bring up 'n' processors. Surely after
3001 bootup you can bring up the other plugged cpu by executing
3002 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
3003 only takes effect during system bootup.
3004 While n=0 is a special case, it is equivalent to "nosmp",
3005 which also disables the IO APIC.
3007 max_loop= [LOOP] The number of loop block devices that get
3008 (loop.max_loop) unconditionally pre-created at init time. The default
3009 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
3010 of statically allocating a predefined number, loop
3011 devices can be requested on-demand with the
3012 /dev/loop-control interface.
3014 mce [X86-32] Machine Check Exception
3016 mce=option [X86-64] See Documentation/x86/x86_64/boot-options.rst
3018 md= [HW] RAID subsystems devices and level
3019 See Documentation/admin-guide/md.rst.
3022 Format: <first>,<last>
3023 Specifies range of consoles to be captured by the MDA.
3026 Control mitigation for the Micro-architectural Data
3027 Sampling (MDS) vulnerability.
3029 Certain CPUs are vulnerable to an exploit against CPU
3030 internal buffers which can forward information to a
3031 disclosure gadget under certain conditions.
3033 In vulnerable processors, the speculatively
3034 forwarded data can be used in a cache side channel
3035 attack, to access data to which the attacker does
3036 not have direct access.
3038 This parameter controls the MDS mitigation. The
3041 full - Enable MDS mitigation on vulnerable CPUs
3042 full,nosmt - Enable MDS mitigation and disable
3043 SMT on vulnerable CPUs
3044 off - Unconditionally disable MDS mitigation
3046 On TAA-affected machines, mds=off can be prevented by
3047 an active TAA mitigation as both vulnerabilities are
3048 mitigated with the same mechanism so in order to disable
3049 this mitigation, you need to specify tsx_async_abort=off
3052 Not specifying this option is equivalent to
3055 For details see: Documentation/admin-guide/hw-vuln/mds.rst
3057 mem=nn[KMG] [HEXAGON] Set the memory size.
3058 Must be specified, otherwise memory size will be 0.
3060 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
3061 Amount of memory to be used in cases as follows:
3064 2 when the kernel is not able to see the whole system memory;
3065 3 memory that lies after 'mem=' boundary is excluded from
3066 the hypervisor, then assigned to KVM guests.
3067 4 to limit the memory available for kdump kernel.
3069 [ARC,MICROBLAZE] - the limit applies only to low memory,
3070 high memory is not affected.
3072 [ARM64] - only limits memory covered by the linear
3073 mapping. The NOMAP regions are not affected.
3075 [X86] Work as limiting max address. Use together
3076 with memmap= to avoid physical address space collisions.
3077 Without memmap= PCI devices could be placed at addresses
3078 belonging to unused RAM.
3080 Note that this only takes effects during boot time since
3081 in above case 3, memory may need be hot added after boot
3082 if system memory of hypervisor is not sufficient.
3085 [ARM,MIPS] - override the memory layout reported by
3087 Define a memory region of size nn[KMG] starting at
3089 Multiple different regions can be specified with
3090 multiple mem= parameters on the command line.
3092 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
3095 memblock=debug [KNL] Enable memblock debug messages.
3098 [KNL,SH] Allow user to override the default size for
3099 per-device physically contiguous DMA buffers.
3101 memhp_default_state=online/offline
3102 [KNL] Set the initial state for the memory hotplug
3103 onlining policy. If not specified, the default value is
3104 set according to the
3105 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
3107 See Documentation/admin-guide/mm/memory-hotplug.rst.
3109 memmap=exactmap [KNL,X86] Enable setting of an exact
3110 E820 memory map, as specified by the user.
3111 Such memmap=exactmap lines can be constructed based on
3112 BIOS output or other requirements. See the memmap=nn@ss
3115 memmap=nn[KMG]@ss[KMG]
3116 [KNL, X86, MIPS, XTENSA] Force usage of a specific region of memory.
3117 Region of memory to be used is from ss to ss+nn.
3118 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
3119 which limits max address to nn[KMG].
3120 Multiple different regions can be specified,
3123 memmap=100M@2G,100M#3G,1G!1024G
3125 memmap=nn[KMG]#ss[KMG]
3126 [KNL,ACPI] Mark specific memory as ACPI data.
3127 Region of memory to be marked is from ss to ss+nn.
3129 memmap=nn[KMG]$ss[KMG]
3130 [KNL,ACPI] Mark specific memory as reserved.
3131 Region of memory to be reserved is from ss to ss+nn.
3132 Example: Exclude memory from 0x18690000-0x1869ffff
3133 memmap=64K$0x18690000
3135 memmap=0x10000$0x18690000
3136 Some bootloaders may need an escape character before '$',
3137 like Grub2, otherwise '$' and the following number
3140 memmap=nn[KMG]!ss[KMG]
3141 [KNL,X86] Mark specific memory as protected.
3142 Region of memory to be used, from ss to ss+nn.
3143 The memory region may be marked as e820 type 12 (0xc)
3144 and is NVDIMM or ADR memory.
3146 memmap=<size>%<offset>-<oldtype>+<newtype>
3147 [KNL,ACPI] Convert memory within the specified region
3148 from <oldtype> to <newtype>. If "-<oldtype>" is left
3149 out, the whole region will be marked as <newtype>,
3150 even if previously unavailable. If "+<newtype>" is left
3151 out, matching memory will be removed. Types are
3152 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
3153 3 = ACPI, 12 = PRAM.
3155 memory_corruption_check=0/1 [X86]
3156 Some BIOSes seem to corrupt the first 64k of
3157 memory when doing things like suspend/resume.
3158 Setting this option will scan the memory
3159 looking for corruption. Enabling this will
3160 both detect corruption and prevent the kernel
3161 from using the memory being corrupted.
3162 However, its intended as a diagnostic tool; if
3163 repeatable BIOS-originated corruption always
3164 affects the same memory, you can use memmap=
3165 to prevent the kernel from using that memory.
3167 memory_corruption_check_size=size [X86]
3168 By default it checks for corruption in the low
3169 64k, making this memory unavailable for normal
3170 use. Use this parameter to scan for
3171 corruption in more or less memory.
3173 memory_corruption_check_period=seconds [X86]
3174 By default it checks for corruption every 60
3175 seconds. Use this parameter to check at some
3176 other rate. 0 disables periodic checking.
3178 memory_hotplug.memmap_on_memory
3179 [KNL,X86,ARM] Boolean flag to enable this feature.
3180 Format: {on | off (default)}
3181 When enabled, runtime hotplugged memory will
3182 allocate its internal metadata (struct pages,
3183 those vmemmap pages cannot be optimized even
3184 if hugetlb_free_vmemmap is enabled) from the
3185 hotadded memory which will allow to hotadd a
3186 lot of memory without requiring additional
3188 This feature is disabled by default because it
3189 has some implication on large (e.g. GB)
3190 allocations in some configurations (e.g. small
3192 The state of the flag can be read in
3193 /sys/module/memory_hotplug/parameters/memmap_on_memory.
3194 Note that even when enabled, there are a few cases where
3195 the feature is not effective.
3197 memtest= [KNL,X86,ARM,M68K,PPC,RISCV] Enable memtest
3199 default : 0 <disable>
3200 Specifies the number of memtest passes to be
3201 performed. Each pass selects another test
3202 pattern from a given set of patterns. Memtest
3203 fills the memory with this pattern, validates
3204 memory contents and reserves bad memory
3205 regions that are detected.
3207 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
3208 Valid arguments: on, off
3210 mem_encrypt=on: Activate SME
3211 mem_encrypt=off: Do not activate SME
3213 Refer to Documentation/virt/kvm/x86/amd-memory-encryption.rst
3214 for details on when memory encryption can be activated.
3216 mem_sleep_default= [SUSPEND] Default system suspend mode:
3217 s2idle - Suspend-To-Idle
3218 shallow - Power-On Suspend or equivalent (if supported)
3219 deep - Suspend-To-RAM or equivalent (if supported)
3220 See Documentation/admin-guide/pm/sleep-states.rst.
3222 meye.*= [HW] Set MotionEye Camera parameters
3223 See Documentation/admin-guide/media/meye.rst.
3225 mfgpt_irq= [IA-32] Specify the IRQ to use for the
3226 Multi-Function General Purpose Timers on AMD Geode
3229 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
3230 the BIOS has incorrectly applied a workaround. TinyBIOS
3231 version 0.98 is known to be affected, 0.99 fixes the
3232 problem by letting the user disable the workaround.
3236 min_addr=nn[KMG] [KNL,BOOT,IA-64] All physical memory below this
3237 physical address is ignored.
3239 mini2440= [ARM,HW,KNL]
3240 Format:[0..2][b][c][t]
3242 MINI2440 configuration specification:
3243 0 - The attached screen is the 3.5" TFT
3244 1 - The attached screen is the 7" TFT
3245 2 - The VGA Shield is attached (1024x768)
3246 Leaving out the screen size parameter will not load
3247 the TFT driver, and the framebuffer will be left
3249 b - Enable backlight. The TFT backlight pin will be
3250 linked to the kernel VESA blanking code and a GPIO
3251 LED. This parameter is not necessary when using the
3253 c - Enable the s3c camera interface.
3254 t - Reserved for enabling touchscreen support. The
3255 touchscreen support is not enabled in the mainstream
3256 kernel as of 2.6.30, a preliminary port can be found
3257 in the "bleeding edge" mini2440 support kernel at
3258 https://repo.or.cz/w/linux-2.6/mini2440.git
3261 [X86,PPC,S390,ARM64] Control optional mitigations for
3262 CPU vulnerabilities. This is a set of curated,
3263 arch-independent options, each of which is an
3264 aggregation of existing arch-specific options.
3267 Disable all optional CPU mitigations. This
3268 improves system performance, but it may also
3269 expose users to several CPU vulnerabilities.
3270 Equivalent to: if nokaslr then kpti=0 [ARM64]
3271 gather_data_sampling=off [X86]
3272 kvm.nx_huge_pages=off [X86]
3275 mmio_stale_data=off [X86]
3276 no_entry_flush [PPC]
3277 no_uaccess_flush [PPC]
3280 nospectre_bhb [ARM64]
3281 nospectre_v1 [X86,PPC]
3282 nospectre_v2 [X86,PPC,S390,ARM64]
3283 reg_file_data_sampling=off [X86]
3285 spec_store_bypass_disable=off [X86,PPC]
3286 spectre_bhi=off [X86]
3287 spectre_v2_user=off [X86]
3288 srbds=off [X86,INTEL]
3289 ssbd=force-off [ARM64]
3290 tsx_async_abort=off [X86]
3293 This does not have any effect on
3294 kvm.nx_huge_pages when
3295 kvm.nx_huge_pages=force.
3298 Mitigate all CPU vulnerabilities, but leave SMT
3299 enabled, even if it's vulnerable. This is for
3300 users who don't want to be surprised by SMT
3301 getting disabled across kernel upgrades, or who
3302 have other ways of avoiding SMT-based attacks.
3303 Equivalent to: (default behavior)
3306 Mitigate all CPU vulnerabilities, disabling SMT
3307 if needed. This is for users who always want to
3308 be fully mitigated, even if it means losing SMT.
3309 Equivalent to: l1tf=flush,nosmt [X86]
3310 mds=full,nosmt [X86]
3311 tsx_async_abort=full,nosmt [X86]
3312 mmio_stale_data=full,nosmt [X86]
3313 retbleed=auto,nosmt [X86]
3316 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
3317 parameter allows control of the logging verbosity for
3318 the additional memory initialisation checks. A value
3319 of 0 disables mminit logging and a level of 4 will
3320 log everything. Information is printed at KERN_DEBUG
3321 so loglevel=8 may also need to be specified.
3324 [X86,INTEL] Control mitigation for the Processor
3325 MMIO Stale Data vulnerabilities.
3327 Processor MMIO Stale Data is a class of
3328 vulnerabilities that may expose data after an MMIO
3329 operation. Exposed data could originate or end in
3330 the same CPU buffers as affected by MDS and TAA.
3331 Therefore, similar to MDS and TAA, the mitigation
3332 is to clear the affected CPU buffers.
3334 This parameter controls the mitigation. The
3337 full - Enable mitigation on vulnerable CPUs
3339 full,nosmt - Enable mitigation and disable SMT on
3342 off - Unconditionally disable mitigation
3344 On MDS or TAA affected machines,
3345 mmio_stale_data=off can be prevented by an active
3346 MDS or TAA mitigation as these vulnerabilities are
3347 mitigated with the same mechanism so in order to
3348 disable this mitigation, you need to specify
3349 mds=off and tsx_async_abort=off too.
3351 Not specifying this option is equivalent to
3352 mmio_stale_data=full.
3355 Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst
3357 module.async_probe=<bool>
3358 [KNL] When set to true, modules will use async probing
3359 by default. To enable/disable async probing for a
3360 specific module, use the module specific control that
3361 is documented under <module>.async_probe. When both
3362 module.async_probe and <module>.async_probe are
3363 specified, <module>.async_probe takes precedence for
3364 the specific module.
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
3462 nfsaddrs= [NFS] Deprecated. Use ip= instead.
3463 See Documentation/admin-guide/nfs/nfsroot.rst.
3465 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
3466 See Documentation/admin-guide/nfs/nfsroot.rst.
3468 nfsrootdebug [NFS] enable nfsroot debugging messages.
3469 See Documentation/admin-guide/nfs/nfsroot.rst.
3471 nfs.callback_nr_threads=
3472 [NFSv4] set the total number of threads that the
3473 NFS client will assign to service NFSv4 callback
3476 nfs.callback_tcpport=
3477 [NFS] set the TCP port on which the NFSv4 callback
3478 channel should listen.
3481 [NFS] sets the pathname to the program which is used
3482 to update the NFS client cache entries.
3484 nfs.cache_getent_timeout=
3485 [NFS] sets the timeout after which an attempt to
3486 update a cache entry is deemed to have failed.
3488 nfs.idmap_cache_timeout=
3489 [NFS] set the maximum lifetime for idmapper cache
3493 [NFS] enable 64-bit inode numbers.
3494 If zero, the NFS client will fake up a 32-bit inode
3495 number for the readdir() and stat() syscalls instead
3496 of returning the full 64-bit number.
3497 The default is to return 64-bit inode numbers.
3499 nfs.max_session_cb_slots=
3500 [NFSv4.1] Sets the maximum number of session
3501 slots the client will assign to the callback
3502 channel. This determines the maximum number of
3503 callbacks the client will process in parallel for
3504 a particular server.
3506 nfs.max_session_slots=
3507 [NFSv4.1] Sets the maximum number of session slots
3508 the client will attempt to negotiate with the server.
3509 This limits the number of simultaneous RPC requests
3510 that the client can send to the NFSv4.1 server.
3511 Note that there is little point in setting this
3512 value higher than the max_tcp_slot_table_limit.
3514 nfs.nfs4_disable_idmapping=
3515 [NFSv4] When set to the default of '1', this option
3516 ensures that both the RPC level authentication
3517 scheme and the NFS level operations agree to use
3518 numeric uids/gids if the mount is using the
3519 'sec=sys' security flavour. In effect it is
3520 disabling idmapping, which can make migration from
3521 legacy NFSv2/v3 systems to NFSv4 easier.
3522 Servers that do not support this mode of operation
3523 will be autodetected by the client, and it will fall
3524 back to using the idmapper.
3525 To turn off this behaviour, set the value to '0'.
3527 [NFS4] Specify an additional fixed unique ident-
3528 ification string that NFSv4 clients can insert into
3529 their nfs_client_id4 string. This is typically a
3530 UUID that is generated at system install time.
3532 nfs.send_implementation_id =
3533 [NFSv4.1] Send client implementation identification
3534 information in exchange_id requests.
3535 If zero, no implementation identification information
3537 The default is to send the implementation identification
3540 nfs.recover_lost_locks =
3541 [NFSv4] Attempt to recover locks that were lost due
3542 to a lease timeout on the server. Please note that
3543 doing this risks data corruption, since there are
3544 no guarantees that the file will remain unchanged
3545 after the locks are lost.
3546 If you want to enable the kernel legacy behaviour of
3547 attempting to recover these locks, then set this
3549 The default parameter value of '0' causes the kernel
3550 not to attempt recovery of lost locks.
3552 nfs4.layoutstats_timer =
3553 [NFSv4.2] Change the rate at which the kernel sends
3554 layoutstats to the pNFS metadata server.
3556 Setting this to value to 0 causes the kernel to use
3557 whatever value is the default set by the layout
3558 driver. A non-zero value sets the minimum interval
3559 in seconds between layoutstats transmissions.
3561 nfsd.inter_copy_offload_enable =
3562 [NFSv4.2] When set to 1, the server will support
3563 server-to-server copies for which this server is
3564 the destination of the copy.
3566 nfsd.nfsd4_ssc_umount_timeout =
3567 [NFSv4.2] When used as the destination of a
3568 server-to-server copy, knfsd temporarily mounts
3569 the source server. It caches the mount in case
3570 it will be needed again, and discards it if not
3571 used for the number of milliseconds specified by
3574 nfsd.nfs4_disable_idmapping=
3575 [NFSv4] When set to the default of '1', the NFSv4
3576 server will return only numeric uids and gids to
3577 clients using auth_sys, and will accept numeric uids
3578 and gids from such clients. This is intended to ease
3579 migration from NFSv2/v3.
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 no5lvl [X86-64] Disable 5-level paging mode. Forces
3611 kernel to use 4-level paging instead.
3613 nofsgsbase [X86] Disables FSGSBASE instructions.
3616 [HW] Never suspend the console
3617 Disable suspending of consoles during suspend and
3618 hibernate operations. Once disabled, debugging
3619 messages can reach various consoles while the rest
3620 of the system is being put to sleep (ie, while
3621 debugging driver suspend/resume hooks). This may
3622 not work reliably with all consoles, but is known
3623 to work with serial and VGA consoles.
3624 To facilitate more flexible debugging, we also add
3625 console_suspend, a printk module parameter to control
3626 it. Users could use console_suspend (usually
3627 /sys/module/printk/parameters/console_suspend) to
3628 turn on/off it dynamically.
3630 novmcoredd [KNL,KDUMP]
3631 Disable device dump. Device dump allows drivers to
3632 append dump data to vmcore so you can collect driver
3633 specified debug info. Drivers can append the data
3634 without any limit and this data is stored in memory,
3635 so this may cause significant memory stress. Disabling
3636 device dump can help save memory but the driver debug
3637 data will be no longer available. This parameter
3638 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
3641 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
3642 caches in the slab allocator. Saves per-node memory,
3643 but will impact performance.
3647 noaltinstr [S390] Disables alternative instructions patching
3648 (CPU alternatives feature).
3650 noapic [SMP,APIC] Tells the kernel to not make use of any
3651 IOAPICs that may be present in the system.
3653 noautogroup Disable scheduler automatic task group creation.
3657 nodsp [SH] Disable hardware DSP at boot time.
3659 noefi Disable EFI runtime services support.
3661 no_entry_flush [PPC] Don't flush the L1-D cache when entering the kernel.
3666 Disable SMAP (Supervisor Mode Access Prevention)
3667 even if it is supported by processor.
3670 Disable SMEP (Supervisor Mode Execution Prevention)
3671 even if it is supported by processor.
3674 This affects only 32-bit executables.
3675 noexec32=on: enable non-executable mappings (default)
3676 read doesn't imply executable mappings
3677 noexec32=off: disable non-executable mappings
3678 read implies executable mappings
3680 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3682 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3683 register save and restore. The kernel will only save
3684 legacy floating-point registers on task switch.
3686 nohugeiomap [KNL,X86,PPC,ARM64] Disable kernel huge I/O mappings.
3688 nohugevmalloc [KNL,X86,PPC,ARM64] Disable kernel huge vmalloc mappings.
3690 nosmt [KNL,S390] Disable symmetric multithreading (SMT).
3691 Equivalent to smt=1.
3693 [KNL,X86] Disable symmetric multithreading (SMT).
3694 nosmt=force: Force disable SMT, cannot be undone
3695 via the sysfs control file.
3697 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1
3698 (bounds check bypass). With this option data leaks are
3699 possible in the system.
3701 nospectre_v2 [X86,PPC_E500,ARM64] Disable all mitigations for
3702 the Spectre variant 2 (indirect branch prediction)
3703 vulnerability. System may allow data leaks with this
3706 nospectre_bhb [ARM64] Disable all mitigations for Spectre-BHB (branch
3707 history injection) vulnerability. System may allow data leaks
3710 nospec_store_bypass_disable
3711 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
3714 [PPC] Don't flush the L1-D cache after accessing user data.
3716 noxsave [BUGS=X86] Disables x86 extended register state save
3717 and restore using xsave. The kernel will fallback to
3718 enabling legacy floating-point and sse state.
3720 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
3721 register states. The kernel will fall back to use
3722 xsave to save the states. By using this parameter,
3723 performance of saving the states is degraded because
3724 xsave doesn't support modified optimization while
3725 xsaveopt supports it on xsaveopt enabled systems.
3727 noxsaves [X86] Disables xsaves and xrstors used in saving and
3728 restoring x86 extended register state in compacted
3729 form of xsave area. The kernel will fall back to use
3730 xsaveopt and xrstor to save and restore the states
3731 in standard form of xsave area. By using this
3732 parameter, xsave area per process might occupy more
3733 memory on xsaves enabled systems.
3735 nohlt [ARM,ARM64,MICROBLAZE,SH] Forces the kernel to busy wait
3736 in do_idle() and not use the arch_cpu_idle()
3737 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
3738 to be effective. This is useful on platforms where the
3739 sleep(SH) or wfi(ARM,ARM64) instructions do not work
3740 correctly or when doing power measurements to evalute
3741 the impact of the sleep instructions. This is also
3742 useful when using JTAG debugger.
3744 no_file_caps Tells the kernel not to honor file capabilities. The
3745 only way then for a file to be executed with privilege
3746 is to be setuid root or executed by root.
3748 nohalt [IA-64] Tells the kernel not to use the power saving
3749 function PAL_HALT_LIGHT when idle. This increases
3750 power-consumption. On the positive side, it reduces
3751 interrupt wake-up latency, which may improve performance
3752 in certain environments such as networked servers or
3756 Force pointers printed to the console or buffers to be
3757 unhashed. By default, when a pointer is printed via %p
3758 format string, that pointer is "hashed", i.e. obscured
3759 by hashing the pointer value. This is a security feature
3760 that hides actual kernel addresses from unprivileged
3761 users, but it also makes debugging the kernel more
3762 difficult since unequal pointers can no longer be
3763 compared. However, if this command-line option is
3764 specified, then all normal pointers will have their true
3765 value printed. This option should only be specified when
3766 debugging the kernel. Please do not use on production
3769 nohibernate [HIBERNATION] Disable hibernation and resume.
3771 nohz= [KNL] Boottime enable/disable dynamic ticks
3772 Valid arguments: on, off
3775 nohz_full= [KNL,BOOT,SMP,ISOL]
3776 The argument is a cpu list, as described above.
3777 In kernels built with CONFIG_NO_HZ_FULL=y, set
3778 the specified list of CPUs whose tick will be stopped
3779 whenever possible. The boot CPU will be forced outside
3780 the range to maintain the timekeeping. Any CPUs
3781 in this list will have their RCU callbacks offloaded,
3782 just as if they had also been called out in the
3783 rcu_nocbs= boot parameter.
3785 Note that this argument takes precedence over
3786 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option.
3788 noiotrap [SH] Disables trapped I/O port accesses.
3790 noirqdebug [X86-32] Disables the code which attempts to detect and
3791 disable unhandled interrupt sources.
3793 no_timer_check [X86,APIC] Disables the code which tests for
3794 broken timer IRQ sources.
3796 noisapnp [ISAPNP] Disables ISA PnP code.
3798 noinitrd [RAM] Tells the kernel not to load any configured
3801 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
3803 [Deprecated - use intremap=off]
3807 noinvpcid [X86] Disable the INVPCID cpu feature.
3809 nojitter [IA-64] Disables jitter checking for ITC timers.
3811 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
3813 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
3817 [X86,PV_OPS] Disable paravirtualized VMware scheduler
3818 clock and use the default one.
3820 no-steal-acc [X86,PV_OPS,ARM64,PPC/PSERIES] Disable paravirtualized
3821 steal time accounting. steal time is computed, but
3822 won't influence scheduler behaviour
3824 nolapic [X86-32,APIC] Do not enable or use the local APIC.
3826 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
3828 nomca [IA-64] Disable machine check abort handling
3830 nomce [X86-32] Disable Machine Check Exception
3832 nomfgpt [X86-32] Disable Multi-Function General Purpose
3833 Timer usage (for AMD Geode machines).
3835 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3836 shutdown the other cpus. Instead use the REBOOT_VECTOR
3839 nomodeset Disable kernel modesetting. DRM drivers will not perform
3840 display-mode changes or accelerated rendering. Only the
3841 system framebuffer will be available for use if this was
3842 set-up by the firmware or boot loader.
3844 Useful as fallback, or for testing and debugging.
3846 nomodule Disable module load
3848 nopat [X86] Disable PAT (page attribute table extension of
3849 pagetables) support.
3851 nopcid [X86-64] Disable the PCID cpu feature.
3853 norandmaps Don't use address space randomization. Equivalent to
3854 echo 0 > /proc/sys/kernel/randomize_va_space
3856 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
3857 with UP alternatives
3859 noresume [SWSUSP] Disables resume and restores original swap
3862 no-scroll [VGA] Disables scrollback.
3863 This is required for the Braillex ib80-piezo Braille
3864 reader made by F.H. Papenmeier (Germany).
3868 nosgx [X86-64,SGX] Disables Intel SGX kernel support.
3870 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
3871 and disable the IO APIC. legacy for "maxcpus=0".
3873 nosoftlockup [KNL] Disable the soft-lockup detector.
3875 nosync [HW,M68K] Disables sync negotiation for all devices.
3877 nowatchdog [KNL] Disable both lockup detectors, i.e.
3878 soft-lockup and NMI watchdog (hard-lockup).
3882 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
3884 NOTE: this parameter will be ignored on systems with the
3885 LEGACY_XAPIC_DISABLED bit set in the
3886 IA32_XAPIC_DISABLE_STATUS MSR.
3888 nps_mtm_hs_ctr= [KNL,ARC]
3889 This parameter sets the maximum duration, in
3890 cycles, each HW thread of the CTOP can run
3891 without interruptions, before HW switches it.
3892 The actual maximum duration is 16 times this
3894 Format: integer between 1 and 255
3897 nptcg= [IA-64] Override max number of concurrent global TLB
3898 purges which is reported from either PAL_VM_SUMMARY or
3901 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
3902 could support. nr_cpus=n : n >= 1 limits the kernel to
3903 support 'n' processors. It could be larger than the
3904 number of already plugged CPU during bootup, later in
3905 runtime you can physically add extra cpu until it reaches
3906 n. So during boot up some boot time memory for per-cpu
3907 variables need be pre-allocated for later physical cpu
3910 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
3912 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86] Disable NUMA, Only
3913 set up a single NUMA node spanning all memory.
3915 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic
3917 Allowed values are enable and disable
3919 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
3920 'node', 'default' can be specified
3921 This can be set from sysctl after boot.
3922 See Documentation/admin-guide/sysctl/vm.rst for details.
3924 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
3925 See Documentation/core-api/debugging-via-ohci1394.rst for more
3928 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
3929 Rather than timing out after 20 ms if an EC
3930 command is not properly ACKed, override the length
3931 of the timeout. We have interrupts disabled while
3932 waiting for the ACK, so if this is set too high
3933 interrupts *may* be lost!
3935 omap_mux= [OMAP] Override bootloader pin multiplexing.
3936 Format: <mux_mode0.mode_name=value>...
3937 For example, to override I2C bus2:
3938 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
3940 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
3942 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
3944 boundary - index of last SLC block on Flex-OneNAND.
3945 The remaining blocks are configured as MLC blocks.
3946 lock - Configure if Flex-OneNAND boundary should be locked.
3947 Once locked, the boundary cannot be changed.
3948 1 indicates lock status, 0 indicates unlock status.
3950 oops=panic Always panic on oopses. Default is to just kill the
3951 process, but there is a small probability of
3952 deadlocking the machine.
3953 This will also cause panics on machine check exceptions.
3954 Useful together with panic=30 to trigger a reboot.
3957 [KNL] Boolean flag to control whether the page allocator
3958 should randomize its free lists. The randomization may
3959 be automatically enabled if the kernel detects it is
3960 running on a platform with a direct-mapped memory-side
3961 cache, and this parameter can be used to
3962 override/disable that behavior. The state of the flag
3963 can be read from sysfs at:
3964 /sys/module/page_alloc/parameters/shuffle.
3966 page_owner= [KNL] Boot-time page_owner enabling option.
3967 Storage of the information about who allocated
3968 each page is disabled in default. With this switch,
3970 on: enable the feature
3972 page_poison= [KNL] Boot-time parameter changing the state of
3973 poisoning on the buddy allocator, available with
3974 CONFIG_PAGE_POISONING=y.
3975 off: turn off poisoning (default)
3976 on: turn on poisoning
3978 page_reporting.page_reporting_order=
3979 [KNL] Minimal page reporting order
3981 Adjust the minimal page reporting order. The page
3982 reporting is disabled when it exceeds (MAX_ORDER-1).
3984 panic= [KNL] Kernel behaviour on panic: delay <timeout>
3985 timeout > 0: seconds before rebooting
3986 timeout = 0: wait forever
3987 timeout < 0: reboot immediately
3990 panic_print= Bitmask for printing system info when panic happens.
3991 User can chose combination of the following bits:
3992 bit 0: print all tasks info
3993 bit 1: print system memory info
3994 bit 2: print timer info
3995 bit 3: print locks info if CONFIG_LOCKDEP is on
3996 bit 4: print ftrace buffer
3997 bit 5: print all printk messages in buffer
3998 bit 6: print all CPUs backtrace (if available in the arch)
3999 *Be aware* that this option may print a _lot_ of lines,
4000 so there are risks of losing older messages in the log.
4001 Use this option carefully, maybe worth to setup a
4002 bigger log buffer with "log_buf_len" along with this.
4004 panic_on_taint= Bitmask for conditionally calling panic() in add_taint()
4005 Format: <hex>[,nousertaint]
4006 Hexadecimal bitmask representing the set of TAINT flags
4007 that will cause the kernel to panic when add_taint() is
4008 called with any of the flags in this set.
4009 The optional switch "nousertaint" can be utilized to
4010 prevent userspace forced crashes by writing to sysctl
4011 /proc/sys/kernel/tainted any flagset matching with the
4012 bitmask set on panic_on_taint.
4013 See Documentation/admin-guide/tainted-kernels.rst for
4014 extra details on the taint flags that users can pick
4015 to compose the bitmask to assign to panic_on_taint.
4017 panic_on_warn panic() instead of WARN(). Useful to cause kdump
4020 parkbd.port= [HW] Parallel port number the keyboard adapter is
4021 connected to, default is 0.
4023 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
4024 0 for XT, 1 for AT (default is AT).
4027 parport= [HW,PPT] Specify parallel ports. 0 disables.
4028 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
4029 Use 'auto' to force the driver to use any
4030 IRQ/DMA settings detected (the default is to
4031 ignore detected IRQ/DMA settings because of
4032 possible conflicts). You can specify the base
4033 address, IRQ, and DMA settings; IRQ and DMA
4034 should be numbers, or 'auto' (for using detected
4035 settings on that particular port), or 'nofifo'
4036 (to avoid using a FIFO even if it is detected).
4037 Parallel ports are assigned in the order they
4038 are specified on the command line, starting
4041 parport_init_mode= [HW,PPT]
4042 Configure VIA parallel port to operate in
4043 a specific mode. This is necessary on Pegasos
4044 computer where firmware has no options for setting
4045 up parallel port mode and sets it to spp.
4046 Currently this function knows 686a and 8231 chips.
4047 Format: [spp|ps2|epp|ecp|ecpepp]
4049 pata_legacy.all= [HW,LIBATA]
4051 Set to non-zero to probe primary and secondary ISA
4052 port ranges on PCI systems where no PCI PATA device
4053 has been found at either range. Disabled by default.
4055 pata_legacy.autospeed= [HW,LIBATA]
4057 Set to non-zero if a chip is present that snoops speed
4058 changes. Disabled by default.
4060 pata_legacy.ht6560a= [HW,LIBATA]
4062 Set to 1, 2, or 3 for HT 6560A on the primary channel,
4063 the secondary channel, or both channels respectively.
4064 Disabled by default.
4066 pata_legacy.ht6560b= [HW,LIBATA]
4068 Set to 1, 2, or 3 for HT 6560B on the primary channel,
4069 the secondary channel, or both channels respectively.
4070 Disabled by default.
4072 pata_legacy.iordy_mask= [HW,LIBATA]
4074 IORDY enable mask. Set individual bits to allow IORDY
4075 for the respective channel. Bit 0 is for the first
4076 legacy channel handled by this driver, bit 1 is for
4077 the second channel, and so on. The sequence will often
4078 correspond to the primary legacy channel, the secondary
4079 legacy channel, and so on, but the handling of a PCI
4080 bus and the use of other driver options may interfere
4081 with the sequence. By default IORDY is allowed across
4084 pata_legacy.opti82c46x= [HW,LIBATA]
4086 Set to 1, 2, or 3 for Opti 82c611A on the primary
4087 channel, the secondary channel, or both channels
4088 respectively. Disabled by default.
4090 pata_legacy.opti82c611a= [HW,LIBATA]
4092 Set to 1, 2, or 3 for Opti 82c465MV on the primary
4093 channel, the secondary channel, or both channels
4094 respectively. Disabled by default.
4096 pata_legacy.pio_mask= [HW,LIBATA]
4098 PIO mode mask for autospeed devices. Set individual
4099 bits to allow the use of the respective PIO modes.
4100 Bit 0 is for mode 0, bit 1 is for mode 1, and so on.
4101 All modes allowed by default.
4103 pata_legacy.probe_all= [HW,LIBATA]
4105 Set to non-zero to probe tertiary and further ISA
4106 port ranges on PCI systems. Disabled by default.
4108 pata_legacy.probe_mask= [HW,LIBATA]
4110 Probe mask for legacy ISA PATA ports. Depending on
4111 platform configuration and the use of other driver
4112 options up to 6 legacy ports are supported: 0x1f0,
4113 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing
4114 of individual ports can be disabled by setting the
4115 corresponding bits in the mask to 1. Bit 0 is for
4116 the first port in the list above (0x1f0), and so on.
4117 By default all supported ports are probed.
4119 pata_legacy.qdi= [HW,LIBATA]
4121 Set to non-zero to probe QDI controllers. By default
4122 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise.
4124 pata_legacy.winbond= [HW,LIBATA]
4126 Set to non-zero to probe Winbond controllers. Use
4127 the standard I/O port (0x130) if 1, otherwise the
4128 value given is the I/O port to use (typically 0x1b0).
4129 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE,
4132 pata_platform.pio_mask= [HW,LIBATA]
4134 Supported PIO mode mask. Set individual bits to allow
4135 the use of the respective PIO modes. Bit 0 is for
4136 mode 0, bit 1 is for mode 1, and so on. Mode 0 only
4140 Halt all CPUs after the first oops has been printed for
4141 the specified number of seconds. This is to be used if
4142 your oopses keep scrolling off the screen.
4147 See header of drivers/block/paride/pcd.c.
4148 See also Documentation/admin-guide/blockdev/paride.rst.
4150 pci=option[,option...] [PCI] various PCI subsystem options.
4152 Some options herein operate on a specific device
4153 or a set of devices (<pci_dev>). These are
4154 specified in one of the following formats:
4156 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
4157 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
4159 Note: the first format specifies a PCI
4160 bus/device/function address which may change
4161 if new hardware is inserted, if motherboard
4162 firmware changes, or due to changes caused
4163 by other kernel parameters. If the
4164 domain is left unspecified, it is
4165 taken to be zero. Optionally, a path
4166 to a device through multiple device/function
4167 addresses can be specified after the base
4168 address (this is more robust against
4169 renumbering issues). The second format
4170 selects devices using IDs from the
4171 configuration space which may match multiple
4172 devices in the system.
4174 earlydump dump PCI config space before the kernel
4176 off [X86] don't probe for the PCI bus
4177 bios [X86-32] force use of PCI BIOS, don't access
4178 the hardware directly. Use this if your machine
4179 has a non-standard PCI host bridge.
4180 nobios [X86-32] disallow use of PCI BIOS, only direct
4181 hardware access methods are allowed. Use this
4182 if you experience crashes upon bootup and you
4183 suspect they are caused by the BIOS.
4184 conf1 [X86] Force use of PCI Configuration Access
4185 Mechanism 1 (config address in IO port 0xCF8,
4186 data in IO port 0xCFC, both 32-bit).
4187 conf2 [X86] Force use of PCI Configuration Access
4188 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
4189 the function, IO port 0xCFA, also 8-bit, sets
4190 bus number. The config space is then accessed
4191 through ports 0xC000-0xCFFF).
4192 See http://wiki.osdev.org/PCI for more info
4193 on the configuration access mechanisms.
4194 noaer [PCIE] If the PCIEAER kernel config parameter is
4195 enabled, this kernel boot option can be used to
4196 disable the use of PCIE advanced error reporting.
4197 nodomains [PCI] Disable support for multiple PCI
4198 root domains (aka PCI segments, in ACPI-speak).
4199 nommconf [X86] Disable use of MMCONFIG for PCI
4201 check_enable_amd_mmconf [X86] check for and enable
4202 properly configured MMIO access to PCI
4203 config space on AMD family 10h CPU
4204 nomsi [MSI] If the PCI_MSI kernel config parameter is
4205 enabled, this kernel boot option can be used to
4206 disable the use of MSI interrupts system-wide.
4207 noioapicquirk [APIC] Disable all boot interrupt quirks.
4208 Safety option to keep boot IRQs enabled. This
4209 should never be necessary.
4210 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
4211 primary IO-APIC for bridges that cannot disable
4212 boot IRQs. This fixes a source of spurious IRQs
4213 when the system masks IRQs.
4214 noioapicreroute [APIC] Disable workaround that uses the
4215 boot IRQ equivalent of an IRQ that connects to
4216 a chipset where boot IRQs cannot be disabled.
4217 The opposite of ioapicreroute.
4218 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
4219 routing table. These calls are known to be buggy
4220 on several machines and they hang the machine
4221 when used, but on other computers it's the only
4222 way to get the interrupt routing table. Try
4223 this option if the kernel is unable to allocate
4224 IRQs or discover secondary PCI buses on your
4226 rom [X86] Assign address space to expansion ROMs.
4227 Use with caution as certain devices share
4228 address decoders between ROMs and other
4230 norom [X86] Do not assign address space to
4231 expansion ROMs that do not already have
4232 BIOS assigned address ranges.
4233 nobar [X86] Do not assign address space to the
4234 BARs that weren't assigned by the BIOS.
4235 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
4236 assigned automatically to PCI devices. You can
4237 make the kernel exclude IRQs of your ISA cards
4239 pirqaddr=0xAAAAA [X86] Specify the physical address
4240 of the PIRQ table (normally generated
4241 by the BIOS) if it is outside the
4242 F0000h-100000h range.
4243 lastbus=N [X86] Scan all buses thru bus #N. Can be
4244 useful if the kernel is unable to find your
4245 secondary buses and you want to tell it
4246 explicitly which ones they are.
4247 assign-busses [X86] Always assign all PCI bus
4248 numbers ourselves, overriding
4249 whatever the firmware may have done.
4250 usepirqmask [X86] Honor the possible IRQ mask stored
4251 in the BIOS $PIR table. This is needed on
4252 some systems with broken BIOSes, notably
4253 some HP Pavilion N5400 and Omnibook XE3
4254 notebooks. This will have no effect if ACPI
4255 IRQ routing is enabled.
4256 noacpi [X86] Do not use ACPI for IRQ routing
4257 or for PCI scanning.
4258 use_crs [X86] Use PCI host bridge window information
4259 from ACPI. On BIOSes from 2008 or later, this
4260 is enabled by default. If you need to use this,
4261 please report a bug.
4262 nocrs [X86] Ignore PCI host bridge windows from ACPI.
4263 If you need to use this, please report a bug.
4264 use_e820 [X86] Use E820 reservations to exclude parts of
4265 PCI host bridge windows. This is a workaround
4266 for BIOS defects in host bridge _CRS methods.
4267 If you need to use this, please report a bug to
4268 <linux-pci@vger.kernel.org>.
4269 no_e820 [X86] Ignore E820 reservations for PCI host
4270 bridge windows. This is the default on modern
4271 hardware. If you need to use this, please report
4272 a bug to <linux-pci@vger.kernel.org>.
4273 routeirq Do IRQ routing for all PCI devices.
4274 This is normally done in pci_enable_device(),
4275 so this option is a temporary workaround
4276 for broken drivers that don't call it.
4277 skip_isa_align [X86] do not align io start addr, so can
4278 handle more pci cards
4279 noearly [X86] Don't do any early type 1 scanning.
4280 This might help on some broken boards which
4281 machine check when some devices' config space
4282 is read. But various workarounds are disabled
4283 and some IOMMU drivers will not work.
4284 bfsort Sort PCI devices into breadth-first order.
4285 This sorting is done to get a device
4286 order compatible with older (<= 2.4) kernels.
4287 nobfsort Don't sort PCI devices into breadth-first order.
4288 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
4289 tuning and use the BIOS-configured MPS defaults.
4290 pcie_bus_safe Set every device's MPS to the largest value
4291 supported by all devices below the root complex.
4292 pcie_bus_perf Set device MPS to the largest allowable MPS
4293 based on its parent bus. Also set MRRS (Max
4294 Read Request Size) to the largest supported
4295 value (no larger than the MPS that the device
4296 or bus can support) for best performance.
4297 pcie_bus_peer2peer Set every device's MPS to 128B, which
4298 every device is guaranteed to support. This
4299 configuration allows peer-to-peer DMA between
4300 any pair of devices, possibly at the cost of
4301 reduced performance. This also guarantees
4302 that hot-added devices will work.
4303 cbiosize=nn[KMG] The fixed amount of bus space which is
4304 reserved for the CardBus bridge's IO window.
4305 The default value is 256 bytes.
4306 cbmemsize=nn[KMG] The fixed amount of bus space which is
4307 reserved for the CardBus bridge's memory
4308 window. The default value is 64 megabytes.
4311 [<order of align>@]<pci_dev>[; ...]
4312 Specifies alignment and device to reassign
4313 aligned memory resources. How to
4314 specify the device is described above.
4315 If <order of align> is not specified,
4316 PAGE_SIZE is used as alignment.
4317 A PCI-PCI bridge can be specified if resource
4318 windows need to be expanded.
4319 To specify the alignment for several
4320 instances of a device, the PCI vendor,
4321 device, subvendor, and subdevice may be
4322 specified, e.g., 12@pci:8086:9c22:103c:198f
4323 for 4096-byte alignment.
4324 ecrc= Enable/disable PCIe ECRC (transaction layer
4325 end-to-end CRC checking).
4326 bios: Use BIOS/firmware settings. This is the
4330 hpiosize=nn[KMG] The fixed amount of bus space which is
4331 reserved for hotplug bridge's IO window.
4332 Default size is 256 bytes.
4333 hpmmiosize=nn[KMG] The fixed amount of bus space which is
4334 reserved for hotplug bridge's MMIO window.
4335 Default size is 2 megabytes.
4336 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
4337 reserved for hotplug bridge's MMIO_PREF window.
4338 Default size is 2 megabytes.
4339 hpmemsize=nn[KMG] The fixed amount of bus space which is
4340 reserved for hotplug bridge's MMIO and
4342 Default size is 2 megabytes.
4343 hpbussize=nn The minimum amount of additional bus numbers
4344 reserved for buses below a hotplug bridge.
4346 realloc= Enable/disable reallocating PCI bridge resources
4347 if allocations done by BIOS are too small to
4348 accommodate resources required by all child
4350 off: Turn realloc off
4352 realloc same as realloc=on
4353 noari do not use PCIe ARI.
4354 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
4355 do not use PCIe ATS (and IOMMU device IOTLB).
4356 pcie_scan_all Scan all possible PCIe devices. Otherwise we
4357 only look for one device below a PCIe downstream
4359 big_root_window Try to add a big 64bit memory window to the PCIe
4360 root complex on AMD CPUs. Some GFX hardware
4361 can resize a BAR to allow access to all VRAM.
4362 Adding the window is slightly risky (it may
4363 conflict with unreported devices), so this
4365 disable_acs_redir=<pci_dev>[; ...]
4366 Specify one or more PCI devices (in the format
4367 specified above) separated by semicolons.
4368 Each device specified will have the PCI ACS
4369 redirect capabilities forced off which will
4370 allow P2P traffic between devices through
4371 bridges without forcing it upstream. Note:
4372 this removes isolation between devices and
4373 may put more devices in an IOMMU group.
4374 force_floating [S390] Force usage of floating interrupts.
4375 nomio [S390] Do not use MIO instructions.
4376 norid [S390] ignore the RID field and force use of
4377 one PCI domain per PCI function
4379 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
4382 force Enable ASPM even on devices that claim not to support it.
4383 WARNING: Forcing ASPM on may cause system lockups.
4385 pcie_ports= [PCIE] PCIe port services handling:
4386 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
4387 even if the platform doesn't give the OS permission to
4388 use them. This may cause conflicts if the platform
4389 also tries to use these services.
4390 dpc-native Use native PCIe service for DPC only. May
4391 cause conflicts if firmware uses AER or DPC.
4392 compat Disable native PCIe services (PME, AER, DPC, PCIe
4395 pcie_port_pm= [PCIE] PCIe port power management handling:
4396 off Disable power management of all PCIe ports
4397 force Forcibly enable power management of all PCIe ports
4399 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
4400 nomsi Do not use MSI for native PCIe PME signaling (this makes
4401 all PCIe root ports use INTx for all services).
4403 pcmv= [HW,PCMCIA] BadgePAD 4
4407 Keep all power-domains already enabled by bootloader on,
4408 even if no driver has claimed them. This is useful
4409 for debug and development, but should not be
4410 needed on a platform with proper driver support.
4413 See Documentation/admin-guide/blockdev/paride.rst.
4415 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
4418 See arch/parisc/kernel/pdc_chassis.c
4420 percpu_alloc= Select which percpu first chunk allocator to use.
4421 Currently supported values are "embed" and "page".
4422 Archs may support subset or none of the selections.
4423 See comments in mm/percpu.c for details on each
4424 allocator. This parameter is primarily for debugging
4425 and performance comparison.
4428 See Documentation/admin-guide/blockdev/paride.rst.
4431 See Documentation/admin-guide/blockdev/paride.rst.
4433 pirq= [SMP,APIC] Manual mp-table setup
4434 See Documentation/x86/i386/IO-APIC.rst.
4436 plip= [PPT,NET] Parallel port network link
4437 Format: { parport<nr> | timid | 0 }
4438 See also Documentation/admin-guide/parport.rst.
4440 pmtmr= [X86] Manual setup of pmtmr I/O Port.
4441 Override pmtimer IOPort with a hex value.
4444 pmu_override= [PPC] Override the PMU.
4445 This option takes over the PMU facility, so it is no
4446 longer usable by perf. Setting this option starts the
4447 PMU counters by setting MMCR0 to 0 (the FC bit is
4448 cleared). If a number is given, then MMCR1 is set to
4449 that number, otherwise (e.g., 'pmu_override=on'), MMCR1
4452 pm_debug_messages [SUSPEND,KNL]
4453 Enable suspend/resume debug messages during boot up.
4456 Enable PNP debug messages (depends on the
4457 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
4458 via /sys/module/pnp/parameters/debug. We always show
4459 current resource usage; turning this on also shows
4460 possible settings and some assignment information.
4466 { on | off | curr | res | no-curr | no-res }
4469 [ISAPNP] Exclude IRQs for the autoconfiguration
4472 [ISAPNP] Exclude DMAs for the autoconfiguration
4474 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
4475 Ranges are in pairs (I/O port base and size).
4478 [ISAPNP] Exclude memory regions for the
4480 Ranges are in pairs (memory base and size).
4482 ports= [IP_VS_FTP] IPVS ftp helper module
4484 Up to 8 (IP_VS_APP_MAX_PORTS) ports
4486 Format: <port>,<port>....
4488 powersave=off [PPC] This option disables power saving features.
4489 It specifically disables cpuidle and sets the
4490 platform machine description specific power_save
4491 function to NULL. On Idle the CPU just reduces
4494 ppc_strict_facility_enable
4495 [PPC] This option catches any kernel floating point,
4496 Altivec, VSX and SPE outside of regions specifically
4497 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
4498 There is some performance impact when enabling this.
4502 Disable Hardware Transactional Memory
4505 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC
4506 none - Limited to cond_resched() calls
4507 voluntary - Limited to cond_resched() and might_sleep() calls
4508 full - Any section that isn't explicitly preempt disabled
4509 can be preempted anytime.
4511 print-fatal-signals=
4512 [KNL] debug: print fatal signals
4514 If enabled, warn about various signal handling
4515 related application anomalies: too many signals,
4516 too many POSIX.1 timers, fatal signals causing a
4519 If you hit the warning due to signal overflow,
4520 you might want to try "ulimit -i unlimited".
4524 printk.always_kmsg_dump=
4525 Trigger kmsg_dump for cases other than kernel oops or
4527 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4530 printk.console_no_auto_verbose=
4531 Disable console loglevel raise on oops, panic
4532 or lockdep-detected issues (only if lock debug is on).
4533 With an exception to setups with low baudrate on
4534 serial console, keeping this 0 is a good choice
4535 in order to provide more debug information.
4537 default: 0 (auto_verbose is enabled)
4539 printk.devkmsg={on,off,ratelimit}
4540 Control writing to /dev/kmsg.
4541 on - unlimited logging to /dev/kmsg from userspace
4542 off - logging to /dev/kmsg disabled
4543 ratelimit - ratelimit the logging
4546 printk.time= Show timing data prefixed to each printk message line
4547 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4549 processor.max_cstate= [HW,ACPI]
4550 Limit processor to maximum C-state
4551 max_cstate=9 overrides any DMI blacklist limit.
4553 processor.nocst [HW,ACPI]
4554 Ignore the _CST method to determine C-states,
4555 instead using the legacy FADT method
4557 profile= [KNL] Enable kernel profiling via /proc/profile
4558 Format: [<profiletype>,]<number>
4559 Param: <profiletype>: "schedule", "sleep", or "kvm"
4560 [defaults to kernel profiling]
4561 Param: "schedule" - profile schedule points.
4562 Param: "sleep" - profile D-state sleeping (millisecs).
4563 Requires CONFIG_SCHEDSTATS
4564 Param: "kvm" - profile VM exits.
4565 Param: <number> - step/bucket size as a power of 2 for
4566 statistical time based profiling.
4568 prompt_ramdisk= [RAM] [Deprecated]
4570 prot_virt= [S390] enable hosting protected virtual machines
4571 isolated from the hypervisor (if hardware supports
4575 psi= [KNL] Enable or disable pressure stall information
4579 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
4580 probe for; one of (bare|imps|exps|lifebook|any).
4581 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
4583 psmouse.resetafter= [HW,MOUSE]
4584 Try to reset the device after so many bad packets
4587 [HW,MOUSE] Set desired mouse resolution, in dpi.
4588 psmouse.smartscroll=
4589 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
4590 0 = disabled, 1 = enabled (default).
4592 pstore.backend= Specify the name of the pstore backend to use
4595 See Documentation/admin-guide/blockdev/paride.rst.
4597 pti= [X86-64] Control Page Table Isolation of user and
4598 kernel address spaces. Disabling this feature
4599 removes hardening, but improves performance of
4600 system calls and interrupts.
4602 on - unconditionally enable
4603 off - unconditionally disable
4604 auto - kernel detects whether your CPU model is
4605 vulnerable to issues that PTI mitigates
4607 Not specifying this option is equivalent to pti=auto.
4610 Equivalent to pti=off
4613 [KNL] Number of legacy pty's. Overwrites compiled-in
4616 quiet [KNL] Disable most log messages
4621 See Documentation/admin-guide/md.rst.
4623 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
4624 See Documentation/admin-guide/blockdev/ramdisk.rst.
4626 ramdisk_start= [RAM] RAM disk image start address
4628 random.trust_cpu={on,off}
4629 [KNL] Enable or disable trusting the use of the
4630 CPU's random number generator (if available) to
4631 fully seed the kernel's CRNG. Default is controlled
4632 by CONFIG_RANDOM_TRUST_CPU.
4634 random.trust_bootloader={on,off}
4635 [KNL] Enable or disable trusting the use of a
4636 seed passed by the bootloader (if available) to
4637 fully seed the kernel's CRNG. Default is controlled
4638 by CONFIG_RANDOM_TRUST_BOOTLOADER.
4640 randomize_kstack_offset=
4641 [KNL] Enable or disable kernel stack offset
4642 randomization, which provides roughly 5 bits of
4643 entropy, frustrating memory corruption attacks
4644 that depend on stack address determinism or
4645 cross-syscall address exposures. This is only
4646 available on architectures that have defined
4647 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET.
4648 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4649 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT.
4651 ras=option[,option,...] [KNL] RAS-specific options
4654 Disable the Correctable Errors Collector,
4655 see CONFIG_RAS_CEC help text.
4657 rcu_nocbs[=cpu-list]
4658 [KNL] The optional argument is a cpu list,
4661 In kernels built with CONFIG_RCU_NOCB_CPU=y,
4662 enable the no-callback CPU mode, which prevents
4663 such CPUs' callbacks from being invoked in
4664 softirq context. Invocation of such CPUs' RCU
4665 callbacks will instead be offloaded to "rcuox/N"
4666 kthreads created for that purpose, where "x" is
4667 "p" for RCU-preempt, "s" for RCU-sched, and "g"
4668 for the kthreads that mediate grace periods; and
4669 "N" is the CPU number. This reduces OS jitter on
4670 the offloaded CPUs, which can be useful for HPC
4671 and real-time workloads. It can also improve
4672 energy efficiency for asymmetric multiprocessors.
4674 If a cpulist is passed as an argument, the specified
4675 list of CPUs is set to no-callback mode from boot.
4677 Otherwise, if the '=' sign and the cpulist
4678 arguments are omitted, no CPU will be set to
4679 no-callback mode from boot but the mode may be
4680 toggled at runtime via cpusets.
4682 Note that this argument takes precedence over
4683 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option.
4686 Rather than requiring that offloaded CPUs
4687 (specified by rcu_nocbs= above) explicitly
4688 awaken the corresponding "rcuoN" kthreads,
4689 make these kthreads poll for callbacks.
4690 This improves the real-time response for the
4691 offloaded CPUs by relieving them of the need to
4692 wake up the corresponding kthread, but degrades
4693 energy efficiency by requiring that the kthreads
4694 periodically wake up to do the polling.
4696 rcutree.blimit= [KNL]
4697 Set maximum number of finished RCU callbacks to
4698 process in one batch.
4700 rcutree.dump_tree= [KNL]
4701 Dump the structure of the rcu_node combining tree
4702 out at early boot. This is used for diagnostic
4703 purposes, to verify correct tree setup.
4705 rcutree.gp_cleanup_delay= [KNL]
4706 Set the number of jiffies to delay each step of
4707 RCU grace-period cleanup.
4709 rcutree.gp_init_delay= [KNL]
4710 Set the number of jiffies to delay each step of
4711 RCU grace-period initialization.
4713 rcutree.gp_preinit_delay= [KNL]
4714 Set the number of jiffies to delay each step of
4715 RCU grace-period pre-initialization, that is,
4716 the propagation of recent CPU-hotplug changes up
4717 the rcu_node combining tree.
4719 rcutree.use_softirq= [KNL]
4720 If set to zero, move all RCU_SOFTIRQ processing to
4721 per-CPU rcuc kthreads. Defaults to a non-zero
4722 value, meaning that RCU_SOFTIRQ is used by default.
4723 Specify rcutree.use_softirq=0 to use rcuc kthreads.
4725 But note that CONFIG_PREEMPT_RT=y kernels disable
4726 this kernel boot parameter, forcibly setting it
4729 rcutree.rcu_fanout_exact= [KNL]
4730 Disable autobalancing of the rcu_node combining
4731 tree. This is used by rcutorture, and might
4732 possibly be useful for architectures having high
4733 cache-to-cache transfer latencies.
4735 rcutree.rcu_fanout_leaf= [KNL]
4736 Change the number of CPUs assigned to each
4737 leaf rcu_node structure. Useful for very
4738 large systems, which will choose the value 64,
4739 and for NUMA systems with large remote-access
4740 latencies, which will choose a value aligned
4741 with the appropriate hardware boundaries.
4743 rcutree.rcu_min_cached_objs= [KNL]
4744 Minimum number of objects which are cached and
4745 maintained per one CPU. Object size is equal
4746 to PAGE_SIZE. The cache allows to reduce the
4747 pressure to page allocator, also it makes the
4748 whole algorithm to behave better in low memory
4751 rcutree.rcu_delay_page_cache_fill_msec= [KNL]
4752 Set the page-cache refill delay (in milliseconds)
4753 in response to low-memory conditions. The range
4754 of permitted values is in the range 0:100000.
4756 rcutree.jiffies_till_first_fqs= [KNL]
4757 Set delay from grace-period initialization to
4758 first attempt to force quiescent states.
4759 Units are jiffies, minimum value is zero,
4760 and maximum value is HZ.
4762 rcutree.jiffies_till_next_fqs= [KNL]
4763 Set delay between subsequent attempts to force
4764 quiescent states. Units are jiffies, minimum
4765 value is one, and maximum value is HZ.
4767 rcutree.jiffies_till_sched_qs= [KNL]
4768 Set required age in jiffies for a
4769 given grace period before RCU starts
4770 soliciting quiescent-state help from
4771 rcu_note_context_switch() and cond_resched().
4772 If not specified, the kernel will calculate
4773 a value based on the most recent settings
4774 of rcutree.jiffies_till_first_fqs
4775 and rcutree.jiffies_till_next_fqs.
4776 This calculated value may be viewed in
4777 rcutree.jiffies_to_sched_qs. Any attempt to set
4778 rcutree.jiffies_to_sched_qs will be cheerfully
4781 rcutree.kthread_prio= [KNL,BOOT]
4782 Set the SCHED_FIFO priority of the RCU per-CPU
4783 kthreads (rcuc/N). This value is also used for
4784 the priority of the RCU boost threads (rcub/N)
4785 and for the RCU grace-period kthreads (rcu_bh,
4786 rcu_preempt, and rcu_sched). If RCU_BOOST is
4787 set, valid values are 1-99 and the default is 1
4788 (the least-favored priority). Otherwise, when
4789 RCU_BOOST is not set, valid values are 0-99 and
4790 the default is zero (non-realtime operation).
4791 When RCU_NOCB_CPU is set, also adjust the
4792 priority of NOCB callback kthreads.
4794 rcutree.rcu_divisor= [KNL]
4795 Set the shift-right count to use to compute
4796 the callback-invocation batch limit bl from
4797 the number of callbacks queued on this CPU.
4798 The result will be bounded below by the value of
4799 the rcutree.blimit kernel parameter. Every bl
4800 callbacks, the softirq handler will exit in
4801 order to allow the CPU to do other work.
4803 Please note that this callback-invocation batch
4804 limit applies only to non-offloaded callback
4805 invocation. Offloaded callbacks are instead
4806 invoked in the context of an rcuoc kthread, which
4807 scheduler will preempt as it does any other task.
4809 rcutree.nocb_nobypass_lim_per_jiffy= [KNL]
4810 On callback-offloaded (rcu_nocbs) CPUs,
4811 RCU reduces the lock contention that would
4812 otherwise be caused by callback floods through
4813 use of the ->nocb_bypass list. However, in the
4814 common non-flooded case, RCU queues directly to
4815 the main ->cblist in order to avoid the extra
4816 overhead of the ->nocb_bypass list and its lock.
4817 But if there are too many callbacks queued during
4818 a single jiffy, RCU pre-queues the callbacks into
4819 the ->nocb_bypass queue. The definition of "too
4820 many" is supplied by this kernel boot parameter.
4822 rcutree.rcu_nocb_gp_stride= [KNL]
4823 Set the number of NOCB callback kthreads in
4824 each group, which defaults to the square root
4825 of the number of CPUs. Larger numbers reduce
4826 the wakeup overhead on the global grace-period
4827 kthread, but increases that same overhead on
4828 each group's NOCB grace-period kthread.
4830 rcutree.qhimark= [KNL]
4831 Set threshold of queued RCU callbacks beyond which
4832 batch limiting is disabled.
4834 rcutree.qlowmark= [KNL]
4835 Set threshold of queued RCU callbacks below which
4836 batch limiting is re-enabled.
4838 rcutree.qovld= [KNL]
4839 Set threshold of queued RCU callbacks beyond which
4840 RCU's force-quiescent-state scan will aggressively
4841 enlist help from cond_resched() and sched IPIs to
4842 help CPUs more quickly reach quiescent states.
4843 Set to less than zero to make this be set based
4844 on rcutree.qhimark at boot time and to zero to
4845 disable more aggressive help enlistment.
4847 rcutree.rcu_kick_kthreads= [KNL]
4848 Cause the grace-period kthread to get an extra
4849 wake_up() if it sleeps three times longer than
4850 it should at force-quiescent-state time.
4851 This wake_up() will be accompanied by a
4852 WARN_ONCE() splat and an ftrace_dump().
4854 rcutree.rcu_unlock_delay= [KNL]
4855 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels,
4856 this specifies an rcu_read_unlock()-time delay
4857 in microseconds. This defaults to zero.
4858 Larger delays increase the probability of
4859 catching RCU pointer leaks, that is, buggy use
4860 of RCU-protected pointers after the relevant
4861 rcu_read_unlock() has completed.
4863 rcutree.sysrq_rcu= [KNL]
4864 Commandeer a sysrq key to dump out Tree RCU's
4865 rcu_node tree with an eye towards determining
4866 why a new grace period has not yet started.
4868 rcuscale.gp_async= [KNL]
4869 Measure performance of asynchronous
4870 grace-period primitives such as call_rcu().
4872 rcuscale.gp_async_max= [KNL]
4873 Specify the maximum number of outstanding
4874 callbacks per writer thread. When a writer
4875 thread exceeds this limit, it invokes the
4876 corresponding flavor of rcu_barrier() to allow
4877 previously posted callbacks to drain.
4879 rcuscale.gp_exp= [KNL]
4880 Measure performance of expedited synchronous
4881 grace-period primitives.
4883 rcuscale.holdoff= [KNL]
4884 Set test-start holdoff period. The purpose of
4885 this parameter is to delay the start of the
4886 test until boot completes in order to avoid
4889 rcuscale.kfree_rcu_test= [KNL]
4890 Set to measure performance of kfree_rcu() flooding.
4892 rcuscale.kfree_rcu_test_double= [KNL]
4893 Test the double-argument variant of kfree_rcu().
4894 If this parameter has the same value as
4895 rcuscale.kfree_rcu_test_single, both the single-
4896 and double-argument variants are tested.
4898 rcuscale.kfree_rcu_test_single= [KNL]
4899 Test the single-argument variant of kfree_rcu().
4900 If this parameter has the same value as
4901 rcuscale.kfree_rcu_test_double, both the single-
4902 and double-argument variants are tested.
4904 rcuscale.kfree_nthreads= [KNL]
4905 The number of threads running loops of kfree_rcu().
4907 rcuscale.kfree_alloc_num= [KNL]
4908 Number of allocations and frees done in an iteration.
4910 rcuscale.kfree_loops= [KNL]
4911 Number of loops doing rcuscale.kfree_alloc_num number
4912 of allocations and frees.
4914 rcuscale.nreaders= [KNL]
4915 Set number of RCU readers. The value -1 selects
4916 N, where N is the number of CPUs. A value
4917 "n" less than -1 selects N-n+1, where N is again
4918 the number of CPUs. For example, -2 selects N
4919 (the number of CPUs), -3 selects N+1, and so on.
4920 A value of "n" less than or equal to -N selects
4923 rcuscale.nwriters= [KNL]
4924 Set number of RCU writers. The values operate
4925 the same as for rcuscale.nreaders.
4926 N, where N is the number of CPUs
4928 rcuscale.perf_type= [KNL]
4929 Specify the RCU implementation to test.
4931 rcuscale.shutdown= [KNL]
4932 Shut the system down after performance tests
4933 complete. This is useful for hands-off automated
4936 rcuscale.verbose= [KNL]
4937 Enable additional printk() statements.
4939 rcuscale.writer_holdoff= [KNL]
4940 Write-side holdoff between grace periods,
4941 in microseconds. The default of zero says
4944 rcutorture.fqs_duration= [KNL]
4945 Set duration of force_quiescent_state bursts
4948 rcutorture.fqs_holdoff= [KNL]
4949 Set holdoff time within force_quiescent_state bursts
4952 rcutorture.fqs_stutter= [KNL]
4953 Set wait time between force_quiescent_state bursts
4956 rcutorture.fwd_progress= [KNL]
4957 Specifies the number of kthreads to be used
4958 for RCU grace-period forward-progress testing
4959 for the types of RCU supporting this notion.
4960 Defaults to 1 kthread, values less than zero or
4961 greater than the number of CPUs cause the number
4964 rcutorture.fwd_progress_div= [KNL]
4965 Specify the fraction of a CPU-stall-warning
4966 period to do tight-loop forward-progress testing.
4968 rcutorture.fwd_progress_holdoff= [KNL]
4969 Number of seconds to wait between successive
4970 forward-progress tests.
4972 rcutorture.fwd_progress_need_resched= [KNL]
4973 Enclose cond_resched() calls within checks for
4974 need_resched() during tight-loop forward-progress
4977 rcutorture.gp_cond= [KNL]
4978 Use conditional/asynchronous update-side
4979 primitives, if available.
4981 rcutorture.gp_exp= [KNL]
4982 Use expedited update-side primitives, if available.
4984 rcutorture.gp_normal= [KNL]
4985 Use normal (non-expedited) asynchronous
4986 update-side primitives, if available.
4988 rcutorture.gp_sync= [KNL]
4989 Use normal (non-expedited) synchronous
4990 update-side primitives, if available. If all
4991 of rcutorture.gp_cond=, rcutorture.gp_exp=,
4992 rcutorture.gp_normal=, and rcutorture.gp_sync=
4993 are zero, rcutorture acts as if is interpreted
4994 they are all non-zero.
4996 rcutorture.irqreader= [KNL]
4997 Run RCU readers from irq handlers, or, more
4998 accurately, from a timer handler. Not all RCU
4999 flavors take kindly to this sort of thing.
5001 rcutorture.leakpointer= [KNL]
5002 Leak an RCU-protected pointer out of the reader.
5003 This can of course result in splats, and is
5004 intended to test the ability of things like
5005 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect
5008 rcutorture.n_barrier_cbs= [KNL]
5009 Set callbacks/threads for rcu_barrier() testing.
5011 rcutorture.nfakewriters= [KNL]
5012 Set number of concurrent RCU writers. These just
5013 stress RCU, they don't participate in the actual
5014 test, hence the "fake".
5016 rcutorture.nocbs_nthreads= [KNL]
5017 Set number of RCU callback-offload togglers.
5018 Zero (the default) disables toggling.
5020 rcutorture.nocbs_toggle= [KNL]
5021 Set the delay in milliseconds between successive
5022 callback-offload toggling attempts.
5024 rcutorture.nreaders= [KNL]
5025 Set number of RCU readers. The value -1 selects
5026 N-1, where N is the number of CPUs. A value
5027 "n" less than -1 selects N-n-2, where N is again
5028 the number of CPUs. For example, -2 selects N
5029 (the number of CPUs), -3 selects N+1, and so on.
5031 rcutorture.object_debug= [KNL]
5032 Enable debug-object double-call_rcu() testing.
5034 rcutorture.onoff_holdoff= [KNL]
5035 Set time (s) after boot for CPU-hotplug testing.
5037 rcutorture.onoff_interval= [KNL]
5038 Set time (jiffies) between CPU-hotplug operations,
5039 or zero to disable CPU-hotplug testing.
5041 rcutorture.read_exit= [KNL]
5042 Set the number of read-then-exit kthreads used
5043 to test the interaction of RCU updaters and
5044 task-exit processing.
5046 rcutorture.read_exit_burst= [KNL]
5047 The number of times in a given read-then-exit
5048 episode that a set of read-then-exit kthreads
5051 rcutorture.read_exit_delay= [KNL]
5052 The delay, in seconds, between successive
5053 read-then-exit testing episodes.
5055 rcutorture.shuffle_interval= [KNL]
5056 Set task-shuffle interval (s). Shuffling tasks
5057 allows some CPUs to go into dyntick-idle mode
5058 during the rcutorture test.
5060 rcutorture.shutdown_secs= [KNL]
5061 Set time (s) after boot system shutdown. This
5062 is useful for hands-off automated testing.
5064 rcutorture.stall_cpu= [KNL]
5065 Duration of CPU stall (s) to test RCU CPU stall
5066 warnings, zero to disable.
5068 rcutorture.stall_cpu_block= [KNL]
5069 Sleep while stalling if set. This will result
5070 in warnings from preemptible RCU in addition
5071 to any other stall-related activity.
5073 rcutorture.stall_cpu_holdoff= [KNL]
5074 Time to wait (s) after boot before inducing stall.
5076 rcutorture.stall_cpu_irqsoff= [KNL]
5077 Disable interrupts while stalling if set.
5079 rcutorture.stall_gp_kthread= [KNL]
5080 Duration (s) of forced sleep within RCU
5081 grace-period kthread to test RCU CPU stall
5082 warnings, zero to disable. If both stall_cpu
5083 and stall_gp_kthread are specified, the
5084 kthread is starved first, then the CPU.
5086 rcutorture.stat_interval= [KNL]
5087 Time (s) between statistics printk()s.
5089 rcutorture.stutter= [KNL]
5090 Time (s) to stutter testing, for example, specifying
5091 five seconds causes the test to run for five seconds,
5092 wait for five seconds, and so on. This tests RCU's
5093 ability to transition abruptly to and from idle.
5095 rcutorture.test_boost= [KNL]
5096 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
5097 "Maybe" means test if the RCU implementation
5098 under test support RCU priority boosting.
5100 rcutorture.test_boost_duration= [KNL]
5101 Duration (s) of each individual boost test.
5103 rcutorture.test_boost_interval= [KNL]
5104 Interval (s) between each boost test.
5106 rcutorture.test_no_idle_hz= [KNL]
5107 Test RCU's dyntick-idle handling. See also the
5108 rcutorture.shuffle_interval parameter.
5110 rcutorture.torture_type= [KNL]
5111 Specify the RCU implementation to test.
5113 rcutorture.verbose= [KNL]
5114 Enable additional printk() statements.
5116 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
5117 Dump ftrace buffer after reporting RCU CPU
5120 rcupdate.rcu_cpu_stall_suppress= [KNL]
5121 Suppress RCU CPU stall warning messages.
5123 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL]
5124 Suppress RCU CPU stall warning messages and
5125 rcutorture writer stall warnings that occur
5126 during early boot, that is, during the time
5127 before the init task is spawned.
5129 rcupdate.rcu_cpu_stall_timeout= [KNL]
5130 Set timeout for RCU CPU stall warning messages.
5131 The value is in seconds and the maximum allowed
5132 value is 300 seconds.
5134 rcupdate.rcu_exp_cpu_stall_timeout= [KNL]
5135 Set timeout for expedited RCU CPU stall warning
5136 messages. The value is in milliseconds
5137 and the maximum allowed value is 21000
5138 milliseconds. Please note that this value is
5139 adjusted to an arch timer tick resolution.
5140 Setting this to zero causes the value from
5141 rcupdate.rcu_cpu_stall_timeout to be used (after
5142 conversion from seconds to milliseconds).
5144 rcupdate.rcu_expedited= [KNL]
5145 Use expedited grace-period primitives, for
5146 example, synchronize_rcu_expedited() instead
5147 of synchronize_rcu(). This reduces latency,
5148 but can increase CPU utilization, degrade
5149 real-time latency, and degrade energy efficiency.
5150 No effect on CONFIG_TINY_RCU kernels.
5152 rcupdate.rcu_normal= [KNL]
5153 Use only normal grace-period primitives,
5154 for example, synchronize_rcu() instead of
5155 synchronize_rcu_expedited(). This improves
5156 real-time latency, CPU utilization, and
5157 energy efficiency, but can expose users to
5158 increased grace-period latency. This parameter
5159 overrides rcupdate.rcu_expedited. No effect on
5160 CONFIG_TINY_RCU kernels.
5162 rcupdate.rcu_normal_after_boot= [KNL]
5163 Once boot has completed (that is, after
5164 rcu_end_inkernel_boot() has been invoked), use
5165 only normal grace-period primitives. No effect
5166 on CONFIG_TINY_RCU kernels.
5168 But note that CONFIG_PREEMPT_RT=y kernels enables
5169 this kernel boot parameter, forcibly setting
5170 it to the value one, that is, converting any
5171 post-boot attempt at an expedited RCU grace
5172 period to instead use normal non-expedited
5173 grace-period processing.
5175 rcupdate.rcu_task_collapse_lim= [KNL]
5176 Set the maximum number of callbacks present
5177 at the beginning of a grace period that allows
5178 the RCU Tasks flavors to collapse back to using
5179 a single callback queue. This switching only
5180 occurs when rcupdate.rcu_task_enqueue_lim is
5181 set to the default value of -1.
5183 rcupdate.rcu_task_contend_lim= [KNL]
5184 Set the minimum number of callback-queuing-time
5185 lock-contention events per jiffy required to
5186 cause the RCU Tasks flavors to switch to per-CPU
5187 callback queuing. This switching only occurs
5188 when rcupdate.rcu_task_enqueue_lim is set to
5189 the default value of -1.
5191 rcupdate.rcu_task_enqueue_lim= [KNL]
5192 Set the number of callback queues to use for the
5193 RCU Tasks family of RCU flavors. The default
5194 of -1 allows this to be automatically (and
5195 dynamically) adjusted. This parameter is intended
5198 rcupdate.rcu_task_ipi_delay= [KNL]
5199 Set time in jiffies during which RCU tasks will
5200 avoid sending IPIs, starting with the beginning
5201 of a given grace period. Setting a large
5202 number avoids disturbing real-time workloads,
5203 but lengthens grace periods.
5205 rcupdate.rcu_task_stall_info= [KNL]
5206 Set initial timeout in jiffies for RCU task stall
5207 informational messages, which give some indication
5208 of the problem for those not patient enough to
5209 wait for ten minutes. Informational messages are
5210 only printed prior to the stall-warning message
5211 for a given grace period. Disable with a value
5212 less than or equal to zero. Defaults to ten
5213 seconds. A change in value does not take effect
5214 until the beginning of the next grace period.
5216 rcupdate.rcu_task_stall_info_mult= [KNL]
5217 Multiplier for time interval between successive
5218 RCU task stall informational messages for a given
5219 RCU tasks grace period. This value is clamped
5220 to one through ten, inclusive. It defaults to
5221 the value three, so that the first informational
5222 message is printed 10 seconds into the grace
5223 period, the second at 40 seconds, the third at
5224 160 seconds, and then the stall warning at 600
5225 seconds would prevent a fourth at 640 seconds.
5227 rcupdate.rcu_task_stall_timeout= [KNL]
5228 Set timeout in jiffies for RCU task stall
5229 warning messages. Disable with a value less
5230 than or equal to zero. Defaults to ten minutes.
5231 A change in value does not take effect until
5232 the beginning of the next grace period.
5234 rcupdate.rcu_self_test= [KNL]
5235 Run the RCU early boot self tests
5239 Run specified binary instead of /init from the ramdisk,
5240 used for early userspace startup. See initrd.
5243 force - Override the decision by the kernel to hide the
5244 advertisement of RDRAND support (this affects
5245 certain AMD processors because of buggy BIOS
5246 support, specifically around the suspend/resume
5250 Turn on/off individual RDT features. List is:
5251 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
5253 E.g. to turn on cmt and turn off mba use:
5257 Format (x86 or x86_64):
5258 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] | d[efault] \
5260 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
5262 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
5263 (prefix with 'panic_' to set mode for panic
5265 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
5266 reboot_force is either force or not specified,
5267 reboot_cpu is s[mp]#### with #### being the processor
5268 to be used for rebooting.
5270 refscale.holdoff= [KNL]
5271 Set test-start holdoff period. The purpose of
5272 this parameter is to delay the start of the
5273 test until boot completes in order to avoid
5276 refscale.loops= [KNL]
5277 Set the number of loops over the synchronization
5278 primitive under test. Increasing this number
5279 reduces noise due to loop start/end overhead,
5280 but the default has already reduced the per-pass
5281 noise to a handful of picoseconds on ca. 2020
5284 refscale.nreaders= [KNL]
5285 Set number of readers. The default value of -1
5286 selects N, where N is roughly 75% of the number
5287 of CPUs. A value of zero is an interesting choice.
5289 refscale.nruns= [KNL]
5290 Set number of runs, each of which is dumped onto
5293 refscale.readdelay= [KNL]
5294 Set the read-side critical-section duration,
5295 measured in microseconds.
5297 refscale.scale_type= [KNL]
5298 Specify the read-protection implementation to test.
5300 refscale.shutdown= [KNL]
5301 Shut down the system at the end of the performance
5302 test. This defaults to 1 (shut it down) when
5303 refscale is built into the kernel and to 0 (leave
5304 it running) when refscale is built as a module.
5306 refscale.verbose= [KNL]
5307 Enable additional printk() statements.
5309 refscale.verbose_batched= [KNL]
5310 Batch the additional printk() statements. If zero
5311 (the default) or negative, print everything. Otherwise,
5312 print every Nth verbose statement, where N is the value
5316 [KNL, SMP] Set scheduler's default relax_domain_level.
5317 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
5319 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
5320 Format: <base1>,<size1>[,<base2>,<size2>,...]
5321 Reserve I/O ports or memory so the kernel won't use
5322 them. If <base> is less than 0x10000, the region
5323 is assumed to be I/O ports; otherwise it is memory.
5325 reservetop= [X86-32]
5327 Reserves a hole at the top of the kernel virtual
5330 reset_devices [KNL] Force drivers to reset the underlying device
5331 during initialization.
5334 Specify the partition device for software suspend
5336 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
5338 resume_offset= [SWSUSP]
5339 Specify the offset from the beginning of the partition
5340 given by "resume=" at which the swap header is located,
5341 in <PAGE_SIZE> units (needed only for swap files).
5342 See Documentation/power/swsusp-and-swap-files.rst
5344 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
5345 read the resume files
5347 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
5348 Useful for devices that are detected asynchronously
5349 (e.g. USB and MMC devices).
5351 retain_initrd [RAM] Keep initrd memory after extraction
5353 retbleed= [X86] Control mitigation of RETBleed (Arbitrary
5354 Speculative Code Execution with Return Instructions)
5357 AMD-based UNRET and IBPB mitigations alone do not stop
5358 sibling threads from influencing the predictions of other
5359 sibling threads. For that reason, STIBP is used on pro-
5360 cessors that support it, and mitigate SMT on processors
5364 auto - automatically select a migitation
5365 auto,nosmt - automatically select a mitigation,
5366 disabling SMT if necessary for
5367 the full mitigation (only on Zen1
5368 and older without STIBP).
5369 ibpb - On AMD, mitigate short speculation
5370 windows on basic block boundaries too.
5371 Safe, highest perf impact. It also
5372 enables STIBP if present. Not suitable
5374 ibpb,nosmt - Like "ibpb" above but will disable SMT
5375 when STIBP is not available. This is
5376 the alternative for systems which do not
5378 unret - Force enable untrained return thunks,
5379 only effective on AMD f15h-f17h based
5381 unret,nosmt - Like unret, but will disable SMT when STIBP
5382 is not available. This is the alternative for
5383 systems which do not have STIBP.
5385 Selecting 'auto' will choose a mitigation method at run
5386 time according to the CPU.
5388 Not specifying this option is equivalent to retbleed=auto.
5390 rfkill.default_state=
5391 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
5392 etc. communication is blocked by default.
5395 rfkill.master_switch_mode=
5396 0 The "airplane mode" button does nothing.
5397 1 The "airplane mode" button toggles between everything
5398 blocked and the previous configuration.
5399 2 The "airplane mode" button toggles between everything
5400 blocked and everything unblocked.
5402 rhash_entries= [KNL,NET]
5403 Set number of hash buckets for route cache
5406 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
5409 ro [KNL] Mount root device read-only on boot
5412 on Mark read-only kernel memory as read-only (default).
5413 off Leave read-only kernel memory writable for debugging.
5414 full Mark read-only kernel memory and aliases as read-only
5418 Enable the uart passthrough on the designated usb port
5419 on Rockchip SoCs. When active, the signals of the
5420 debug-uart get routed to the D+ and D- pins of the usb
5421 port and the regular usb controller gets disabled.
5423 root= [KNL] Root filesystem
5424 See name_to_dev_t comment in init/do_mounts.c.
5426 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
5427 mount the root filesystem
5429 rootflags= [KNL] Set root filesystem mount option string
5431 rootfstype= [KNL] Set root filesystem type
5433 rootwait [KNL] Wait (indefinitely) for root device to show up.
5434 Useful for devices that are detected asynchronously
5435 (e.g. USB and MMC devices).
5437 rproc_mem=nn[KMG][@address]
5438 [KNL,ARM,CMA] Remoteproc physical memory block.
5439 Memory area to be used by remote processor image,
5442 rw [KNL] Mount root device read-write on boot
5444 S [KNL] Run init in single mode
5446 s390_iommu= [HW,S390]
5447 Set s390 IOTLB flushing mode
5449 With strict flushing every unmap operation will result in
5450 an IOTLB flush. Default is lazy flushing before reuse,
5453 s390_iommu_aperture= [KNL,S390]
5454 Specifies the size of the per device DMA address space
5455 accessible through the DMA and IOMMU APIs as a decimal
5456 factor of the size of main memory.
5457 The default is 1 meaning that one can concurrently use
5458 as many DMA addresses as physical memory is installed,
5459 if supported by hardware, and thus map all of memory
5460 once. With a value of 2 one can map all of memory twice
5461 and so on. As a special case a factor of 0 imposes no
5462 restrictions other than those given by hardware at the
5463 cost of significant additional memory use for tables.
5466 See drivers/net/irda/sa1100_ir.c.
5468 sched_verbose [KNL] Enables verbose scheduler debug messages.
5470 schedstats= [KNL,X86] Enable or disable scheduled statistics.
5471 Allowed values are enable and disable. This feature
5472 incurs a small amount of overhead in the scheduler
5473 but is useful for debugging and performance tuning.
5475 sched_thermal_decay_shift=
5476 [KNL, SMP] Set a decay shift for scheduler thermal
5477 pressure signal. Thermal pressure signal follows the
5478 default decay period of other scheduler pelt
5479 signals(usually 32 ms but configurable). Setting
5480 sched_thermal_decay_shift will left shift the decay
5481 period for the thermal pressure signal by the shift
5483 i.e. with the default pelt decay period of 32 ms
5484 sched_thermal_decay_shift thermal pressure decay pr
5488 Format: integer between 0 and 10
5491 scftorture.holdoff= [KNL]
5492 Number of seconds to hold off before starting
5493 test. Defaults to zero for module insertion and
5494 to 10 seconds for built-in smp_call_function()
5497 scftorture.longwait= [KNL]
5498 Request ridiculously long waits randomly selected
5499 up to the chosen limit in seconds. Zero (the
5500 default) disables this feature. Please note
5501 that requesting even small non-zero numbers of
5502 seconds can result in RCU CPU stall warnings,
5503 softlockup complaints, and so on.
5505 scftorture.nthreads= [KNL]
5506 Number of kthreads to spawn to invoke the
5507 smp_call_function() family of functions.
5508 The default of -1 specifies a number of kthreads
5509 equal to the number of CPUs.
5511 scftorture.onoff_holdoff= [KNL]
5512 Number seconds to wait after the start of the
5513 test before initiating CPU-hotplug operations.
5515 scftorture.onoff_interval= [KNL]
5516 Number seconds to wait between successive
5517 CPU-hotplug operations. Specifying zero (which
5518 is the default) disables CPU-hotplug operations.
5520 scftorture.shutdown_secs= [KNL]
5521 The number of seconds following the start of the
5522 test after which to shut down the system. The
5523 default of zero avoids shutting down the system.
5524 Non-zero values are useful for automated tests.
5526 scftorture.stat_interval= [KNL]
5527 The number of seconds between outputting the
5528 current test statistics to the console. A value
5529 of zero disables statistics output.
5531 scftorture.stutter_cpus= [KNL]
5532 The number of jiffies to wait between each change
5533 to the set of CPUs under test.
5535 scftorture.use_cpus_read_lock= [KNL]
5536 Use use_cpus_read_lock() instead of the default
5537 preempt_disable() to disable CPU hotplug
5538 while invoking one of the smp_call_function*()
5541 scftorture.verbose= [KNL]
5542 Enable additional printk() statements.
5544 scftorture.weight_single= [KNL]
5545 The probability weighting to use for the
5546 smp_call_function_single() function with a zero
5547 "wait" parameter. A value of -1 selects the
5548 default if all other weights are -1. However,
5549 if at least one weight has some other value, a
5550 value of -1 will instead select a weight of zero.
5552 scftorture.weight_single_wait= [KNL]
5553 The probability weighting to use for the
5554 smp_call_function_single() function with a
5555 non-zero "wait" parameter. See weight_single.
5557 scftorture.weight_many= [KNL]
5558 The probability weighting to use for the
5559 smp_call_function_many() function with a zero
5560 "wait" parameter. See weight_single.
5561 Note well that setting a high probability for
5562 this weighting can place serious IPI load
5565 scftorture.weight_many_wait= [KNL]
5566 The probability weighting to use for the
5567 smp_call_function_many() function with a
5568 non-zero "wait" parameter. See weight_single
5571 scftorture.weight_all= [KNL]
5572 The probability weighting to use for the
5573 smp_call_function_all() function with a zero
5574 "wait" parameter. See weight_single and
5577 scftorture.weight_all_wait= [KNL]
5578 The probability weighting to use for the
5579 smp_call_function_all() function with a
5580 non-zero "wait" parameter. See weight_single
5583 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
5584 xtime_lock contention on larger systems, and/or RCU lock
5585 contention on all systems with CONFIG_MAXSMP set.
5586 Format: { "0" | "1" }
5587 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
5589 Note: increases power consumption, thus should only be
5590 enabled if running jitter sensitive (HPC/RT) workloads.
5592 security= [SECURITY] Choose a legacy "major" security module to
5593 enable at boot. This has been deprecated by the
5596 selinux= [SELINUX] Disable or enable SELinux at boot time.
5597 Format: { "0" | "1" }
5598 See security/selinux/Kconfig help text.
5603 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
5604 Format: { "0" | "1" }
5605 See security/apparmor/Kconfig help text
5608 Default value is set via kernel config option.
5610 serialnumber [BUGS=X86-32]
5612 sev=option[,option...] [X86-64] See Documentation/x86/x86_64/boot-options.rst
5615 Maximal number of shapers.
5623 Enable merging of slabs with similar size when the
5624 kernel is built without CONFIG_SLAB_MERGE_DEFAULT.
5627 Disable merging of slabs with similar size. May be
5628 necessary if there is some reason to distinguish
5629 allocs to different slabs, especially in hardened
5630 environments where the risk of heap overflows and
5631 layout control by attackers can usually be
5632 frustrated by disabling merging. This will reduce
5633 most of the exposure of a heap attack to a single
5634 cache (risks via metadata attacks are mostly
5635 unchanged). Debug options disable merging on their
5637 For more information see Documentation/mm/slub.rst.
5639 slab_max_order= [MM, SLAB]
5640 Determines the maximum allowed order for slabs.
5641 A high setting may cause OOMs due to memory
5642 fragmentation. Defaults to 1 for systems with
5643 more than 32MB of RAM, 0 otherwise.
5645 slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB]
5646 Enabling slub_debug allows one to determine the
5647 culprit if slab objects become corrupted. Enabling
5648 slub_debug can create guard zones around objects and
5649 may poison objects when not in use. Also tracks the
5650 last alloc / free. For more information see
5651 Documentation/mm/slub.rst.
5653 slub_max_order= [MM, SLUB]
5654 Determines the maximum allowed order for slabs.
5655 A high setting may cause OOMs due to memory
5656 fragmentation. For more information see
5657 Documentation/mm/slub.rst.
5659 slub_min_objects= [MM, SLUB]
5660 The minimum number of objects per slab. SLUB will
5661 increase the slab order up to slub_max_order to
5662 generate a sufficiently large slab able to contain
5663 the number of objects indicated. The higher the number
5664 of objects the smaller the overhead of tracking slabs
5665 and the less frequently locks need to be acquired.
5666 For more information see Documentation/mm/slub.rst.
5668 slub_min_order= [MM, SLUB]
5669 Determines the minimum page order for slabs. Must be
5670 lower than slub_max_order.
5671 For more information see Documentation/mm/slub.rst.
5673 slub_merge [MM, SLUB]
5674 Same with slab_merge.
5676 slub_nomerge [MM, SLUB]
5677 Same with slab_nomerge. This is supported for legacy.
5678 See slab_nomerge for more information.
5681 Format: <io1>[,<io2>[,...,<io8>]]
5683 smp.csd_lock_timeout= [KNL]
5684 Specify the period of time in milliseconds
5685 that smp_call_function() and friends will wait
5686 for a CPU to release the CSD lock. This is
5687 useful when diagnosing bugs involving CPUs
5688 disabling interrupts for extended periods
5689 of time. Defaults to 5,000 milliseconds, and
5690 setting a value of zero disables this feature.
5691 This feature may be more efficiently disabled
5692 using the csdlock_debug- kernel parameter.
5694 smp.panic_on_ipistall= [KNL]
5695 If a csd_lock_timeout extends for more than
5696 the specified number of milliseconds, panic the
5697 system. By default, let CSD-lock acquisition
5698 take as long as they take. Specifying 300,000
5699 for this value provides a 5-minute timeout.
5701 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
5702 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
5703 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
5704 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
5705 smsc-ircc2.ircc_irq= [HW] IRQ line
5706 smsc-ircc2.ircc_dma= [HW] DMA channel
5707 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
5708 0: Toshiba Satellite 1800 (GP data pin select)
5709 1: Fast pin select (default)
5712 smt= [KNL,S390] Set the maximum number of threads (logical
5713 CPUs) to use per physical CPU on systems capable of
5714 symmetric multithreading (SMT). Will be capped to the
5715 actual hardware limit.
5717 Default: -1 (no limit)
5720 [KNL] Should the soft-lockup detector generate panics.
5723 A value of 1 instructs the soft-lockup detector
5724 to panic the machine when a soft-lockup occurs. It is
5725 also controlled by the kernel.softlockup_panic sysctl
5726 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
5727 respective build-time switch to that functionality.
5729 softlockup_all_cpu_backtrace=
5730 [KNL] Should the soft-lockup detector generate
5731 backtraces on all cpus.
5734 sonypi.*= [HW] Sony Programmable I/O Control Device driver
5735 See Documentation/admin-guide/laptops/sonypi.rst
5737 spectre_bhi= [X86] Control mitigation of Branch History Injection
5738 (BHI) vulnerability. This setting affects the
5739 deployment of the HW BHI control and the SW BHB
5742 on - (default) Enable the HW or SW mitigation
5744 off - Disable the mitigation.
5746 spectre_v2= [X86] Control mitigation of Spectre variant 2
5747 (indirect branch speculation) vulnerability.
5748 The default operation protects the kernel from
5751 on - unconditionally enable, implies
5753 off - unconditionally disable, implies
5755 auto - kernel detects whether your CPU model is
5758 Selecting 'on' will, and 'auto' may, choose a
5759 mitigation method at run time according to the
5760 CPU, the available microcode, the setting of the
5761 CONFIG_RETPOLINE configuration option, and the
5762 compiler with which the kernel was built.
5764 Selecting 'on' will also enable the mitigation
5765 against user space to user space task attacks.
5767 Selecting 'off' will disable both the kernel and
5768 the user space protections.
5770 Specific mitigations can also be selected manually:
5772 retpoline - replace indirect branches
5773 retpoline,generic - Retpolines
5774 retpoline,lfence - LFENCE; indirect branch
5775 retpoline,amd - alias for retpoline,lfence
5776 eibrs - Enhanced/Auto IBRS
5777 eibrs,retpoline - Enhanced/Auto IBRS + Retpolines
5778 eibrs,lfence - Enhanced/Auto IBRS + LFENCE
5779 ibrs - use IBRS to protect kernel
5781 Not specifying this option is equivalent to
5785 [X86] Control mitigation of Spectre variant 2
5786 (indirect branch speculation) vulnerability between
5789 on - Unconditionally enable mitigations. Is
5790 enforced by spectre_v2=on
5792 off - Unconditionally disable mitigations. Is
5793 enforced by spectre_v2=off
5795 prctl - Indirect branch speculation is enabled,
5796 but mitigation can be enabled via prctl
5797 per thread. The mitigation control state
5798 is inherited on fork.
5801 - Like "prctl" above, but only STIBP is
5802 controlled per thread. IBPB is issued
5803 always when switching between different user
5807 - Same as "prctl" above, but all seccomp
5808 threads will enable the mitigation unless
5809 they explicitly opt out.
5812 - Like "seccomp" above, but only STIBP is
5813 controlled per thread. IBPB is issued
5814 always when switching between different
5815 user space processes.
5817 auto - Kernel selects the mitigation depending on
5818 the available CPU features and vulnerability.
5820 Default mitigation: "prctl"
5822 Not specifying this option is equivalent to
5823 spectre_v2_user=auto.
5825 spec_rstack_overflow=
5826 [X86] Control RAS overflow mitigation on AMD Zen CPUs
5828 off - Disable mitigation
5829 microcode - Enable microcode mitigation only
5830 safe-ret - Enable sw-only safe RET mitigation (default)
5831 ibpb - Enable mitigation by issuing IBPB on
5833 ibpb-vmexit - Issue IBPB only on VMEXIT
5834 (cloud-specific mitigation)
5836 spec_store_bypass_disable=
5837 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
5838 (Speculative Store Bypass vulnerability)
5840 Certain CPUs are vulnerable to an exploit against a
5841 a common industry wide performance optimization known
5842 as "Speculative Store Bypass" in which recent stores
5843 to the same memory location may not be observed by
5844 later loads during speculative execution. The idea
5845 is that such stores are unlikely and that they can
5846 be detected prior to instruction retirement at the
5847 end of a particular speculation execution window.
5849 In vulnerable processors, the speculatively forwarded
5850 store can be used in a cache side channel attack, for
5851 example to read memory to which the attacker does not
5852 directly have access (e.g. inside sandboxed code).
5854 This parameter controls whether the Speculative Store
5855 Bypass optimization is used.
5857 On x86 the options are:
5859 on - Unconditionally disable Speculative Store Bypass
5860 off - Unconditionally enable Speculative Store Bypass
5861 auto - Kernel detects whether the CPU model contains an
5862 implementation of Speculative Store Bypass and
5863 picks the most appropriate mitigation. If the
5864 CPU is not vulnerable, "off" is selected. If the
5865 CPU is vulnerable the default mitigation is
5866 architecture and Kconfig dependent. See below.
5867 prctl - Control Speculative Store Bypass per thread
5868 via prctl. Speculative Store Bypass is enabled
5869 for a process by default. The state of the control
5870 is inherited on fork.
5871 seccomp - Same as "prctl" above, but all seccomp threads
5872 will disable SSB unless they explicitly opt out.
5874 Default mitigations:
5877 On powerpc the options are:
5879 on,auto - On Power8 and Power9 insert a store-forwarding
5880 barrier on kernel entry and exit. On Power7
5881 perform a software flush on kernel entry and
5885 Not specifying this option is equivalent to
5886 spec_store_bypass_disable=auto.
5888 spia_io_base= [HW,MTD]
5894 [X86] Enable split lock detection or bus lock detection
5896 When enabled (and if hardware support is present), atomic
5897 instructions that access data across cache line
5898 boundaries will result in an alignment check exception
5899 for split lock detection or a debug exception for
5904 warn - the kernel will emit rate-limited warnings
5905 about applications triggering the #AC
5906 exception or the #DB exception. This mode is
5907 the default on CPUs that support split lock
5908 detection or bus lock detection. Default
5909 behavior is by #AC if both features are
5910 enabled in hardware.
5912 fatal - the kernel will send SIGBUS to applications
5913 that trigger the #AC exception or the #DB
5914 exception. Default behavior is by #AC if
5915 both features are enabled in hardware.
5918 Set system wide rate limit to N bus locks
5919 per second for bus lock detection.
5922 N/A for split lock detection.
5925 If an #AC exception is hit in the kernel or in
5926 firmware (i.e. not while executing in user mode)
5927 the kernel will oops in either "warn" or "fatal"
5930 #DB exception for bus lock is triggered only when
5934 Control the Special Register Buffer Data Sampling
5937 Certain CPUs are vulnerable to an MDS-like
5938 exploit which can leak bits from the random
5941 By default, this issue is mitigated by
5942 microcode. However, the microcode fix can cause
5943 the RDRAND and RDSEED instructions to become
5944 much slower. Among other effects, this will
5945 result in reduced throughput from /dev/urandom.
5947 The microcode mitigation can be disabled with
5948 the following option:
5950 off: Disable mitigation and remove
5951 performance impact to RDRAND and RDSEED
5953 srcutree.big_cpu_lim [KNL]
5954 Specifies the number of CPUs constituting a
5955 large system, such that srcu_struct structures
5956 should immediately allocate an srcu_node array.
5957 This kernel-boot parameter defaults to 128,
5958 but takes effect only when the low-order four
5959 bits of srcutree.convert_to_big is equal to 3
5962 srcutree.convert_to_big [KNL]
5963 Specifies under what conditions an SRCU tree
5964 srcu_struct structure will be converted to big
5965 form, that is, with an rcu_node tree:
5968 1: At init_srcu_struct() time.
5969 2: When rcutorture decides to.
5970 3: Decide at boot time (default).
5971 0x1X: Above plus if high contention.
5973 Either way, the srcu_node tree will be sized based
5974 on the actual runtime number of CPUs (nr_cpu_ids)
5975 instead of the compile-time CONFIG_NR_CPUS.
5977 srcutree.counter_wrap_check [KNL]
5978 Specifies how frequently to check for
5979 grace-period sequence counter wrap for the
5980 srcu_data structure's ->srcu_gp_seq_needed field.
5981 The greater the number of bits set in this kernel
5982 parameter, the less frequently counter wrap will
5983 be checked for. Note that the bottom two bits
5986 srcutree.exp_holdoff [KNL]
5987 Specifies how many nanoseconds must elapse
5988 since the end of the last SRCU grace period for
5989 a given srcu_struct until the next normal SRCU
5990 grace period will be considered for automatic
5991 expediting. Set to zero to disable automatic
5994 srcutree.srcu_max_nodelay [KNL]
5995 Specifies the number of no-delay instances
5996 per jiffy for which the SRCU grace period
5997 worker thread will be rescheduled with zero
5998 delay. Beyond this limit, worker thread will
5999 be rescheduled with a sleep delay of one jiffy.
6001 srcutree.srcu_max_nodelay_phase [KNL]
6002 Specifies the per-grace-period phase, number of
6003 non-sleeping polls of readers. Beyond this limit,
6004 grace period worker thread will be rescheduled
6005 with a sleep delay of one jiffy, between each
6006 rescan of the readers, for a grace period phase.
6008 srcutree.srcu_retry_check_delay [KNL]
6009 Specifies number of microseconds of non-sleeping
6010 delay between each non-sleeping poll of readers.
6012 srcutree.small_contention_lim [KNL]
6013 Specifies the number of update-side contention
6014 events per jiffy will be tolerated before
6015 initiating a conversion of an srcu_struct
6016 structure to big form. Note that the value of
6017 srcutree.convert_to_big must have the 0x10 bit
6018 set for contention-based conversions to occur.
6021 Speculative Store Bypass Disable control
6023 On CPUs that are vulnerable to the Speculative
6024 Store Bypass vulnerability and offer a
6025 firmware based mitigation, this parameter
6026 indicates how the mitigation should be used:
6028 force-on: Unconditionally enable mitigation for
6029 for both kernel and userspace
6030 force-off: Unconditionally disable mitigation for
6031 for both kernel and userspace
6032 kernel: Always enable mitigation in the
6033 kernel, and offer a prctl interface
6034 to allow userspace to register its
6035 interest in being mitigated too.
6037 stack_guard_gap= [MM]
6038 override the default stack gap protection. The value
6039 is in page units and it defines how many pages prior
6040 to (for stacks growing down) resp. after (for stacks
6041 growing up) the main stack are reserved for no other
6042 mapping. Default value is 256 pages.
6044 stack_depot_disable= [KNL]
6045 Setting this to true through kernel command line will
6046 disable the stack depot thereby saving the static memory
6047 consumed by the stack hash table. By default this is set
6051 Enabled the stack tracer on boot up.
6053 stacktrace_filter=[function-list]
6054 [FTRACE] Limit the functions that the stack tracer
6055 will trace at boot up. function-list is a comma-separated
6056 list of functions. This list can be changed at run
6057 time by the stack_trace_filter file in the debugfs
6058 tracing directory. Note, this enables stack tracing
6059 and the stacktrace above is not needed.
6063 Set the STI (builtin display/keyboard on the HP-PARISC
6064 machines) console (graphic card) which should be used
6065 as the initial boot-console.
6066 See also comment in drivers/video/console/sticore.c.
6069 See comment in drivers/video/console/sticore.c.
6072 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
6077 Enable or disable strict sigaltstack size checks
6078 against the required signal frame size which
6079 depends on the supported FPU features. This can
6080 be used to filter out binaries which have
6081 not yet been made aware of AT_MINSIGSTKSZ.
6083 sunrpc.min_resvport=
6084 sunrpc.max_resvport=
6086 SunRPC servers often require that client requests
6087 originate from a privileged port (i.e. a port in the
6088 range 0 < portnr < 1024).
6089 An administrator who wishes to reserve some of these
6090 ports for other uses may adjust the range that the
6091 kernel's sunrpc client considers to be privileged
6092 using these two parameters to set the minimum and
6093 maximum port values.
6095 sunrpc.svc_rpc_per_connection_limit=
6097 Limit the number of requests that the server will
6098 process in parallel from a single connection.
6099 The default value is 0 (no limit).
6103 Control how the NFS server code allocates CPUs to
6104 service thread pools. Depending on how many NICs
6105 you have and where their interrupts are bound, this
6106 option will affect which CPUs will do NFS serving.
6107 Note: this parameter cannot be changed while the
6108 NFS server is running.
6110 auto the server chooses an appropriate mode
6111 automatically using heuristics
6112 global a single global pool contains all CPUs
6113 percpu one pool for each CPU
6114 pernode one pool for each NUMA node (equivalent
6115 to global on non-NUMA machines)
6117 sunrpc.tcp_slot_table_entries=
6118 sunrpc.udp_slot_table_entries=
6120 Sets the upper limit on the number of simultaneous
6121 RPC calls that can be sent from the client to a
6122 server. Increasing these values may allow you to
6123 improve throughput, but will also increase the
6124 amount of memory reserved for use by the client.
6126 suspend.pm_test_delay=
6128 Sets the number of seconds to remain in a suspend test
6129 mode before resuming the system (see
6130 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
6131 is set. Default value is 5.
6134 Format: { on | off | y | n | 1 | 0 }
6135 This parameter controls use of the Protected
6136 Execution Facility on pSeries.
6138 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
6139 Format: { <int> [,<int>] | force | noforce }
6140 <int> -- Number of I/O TLB slabs
6141 <int> -- Second integer after comma. Number of swiotlb
6142 areas with their own lock. Will be rounded up
6144 force -- force using of bounce buffers even if they
6145 wouldn't be automatically used by the kernel
6146 noforce -- Never use bounce buffers (for debugging)
6151 Set a sysctl parameter, right before loading the init
6152 process, as if the value was written to the respective
6153 /proc/sys/... file. Both '.' and '/' are recognized as
6154 separators. Unrecognized parameters and invalid values
6155 are reported in the kernel log. Sysctls registered
6156 later by a loaded module cannot be set this way.
6157 Example: sysctl.vm.swappiness=40
6159 sysfs.deprecated=0|1 [KNL]
6160 Enable/disable old style sysfs layout for old udev
6161 on older distributions. When this option is enabled
6162 very new udev will not work anymore. When this option
6163 is disabled (or CONFIG_SYSFS_DEPRECATED not compiled)
6164 in older udev will not work anymore.
6165 Default depends on CONFIG_SYSFS_DEPRECATED_V2 set in
6166 the kernel configuration.
6168 sysrq_always_enabled
6170 Ignore sysrq setting - this boot parameter will
6171 neutralize any effect of /proc/sys/kernel/sysrq.
6172 Useful for debugging.
6174 tcpmhash_entries= [KNL,NET]
6175 Set the number of tcp_metrics_hash slots.
6176 Default value is 8192 or 16384 depending on total
6177 ram pages. This is used to specify the TCP metrics
6178 cache size. See Documentation/networking/ip-sysctl.rst
6179 "tcp_no_metrics_save" section for more details.
6183 test_suspend= [SUSPEND]
6184 Format: { "mem" | "standby" | "freeze" }[,N]
6185 Specify "mem" (for Suspend-to-RAM) or "standby" (for
6186 standby suspend) or "freeze" (for suspend type freeze)
6187 as the system sleep state during system startup with
6188 the optional capability to repeat N number of times.
6189 The system is woken from this state using a
6190 wakeup-capable RTC alarm.
6192 thash_entries= [KNL,NET]
6193 Set number of hash buckets for TCP connection
6195 thermal.act= [HW,ACPI]
6196 -1: disable all active trip points in all thermal zones
6197 <degrees C>: override all lowest active trip points
6199 thermal.crt= [HW,ACPI]
6200 -1: disable all critical trip points in all thermal zones
6201 <degrees C>: override all critical trip points
6203 thermal.off= [HW,ACPI]
6204 1: disable ACPI thermal control
6206 thermal.psv= [HW,ACPI]
6207 -1: disable all passive trip points
6208 <degrees C>: override all passive trip points to this
6211 thermal.tzp= [HW,ACPI]
6212 Specify global default ACPI thermal zone polling rate
6213 <deci-seconds>: poll all this frequency
6214 0: no polling (default)
6217 Force threading of all interrupt handlers except those
6218 marked explicitly IRQF_NO_THREAD.
6222 Specify if the kernel should make use of the cpu
6223 topology information if the hardware supports this.
6224 The scheduler will make use of this information and
6225 e.g. base its process migration decisions on it.
6228 topology_updates= [KNL, PPC, NUMA]
6230 Specify if the kernel should ignore (off)
6231 topology updates sent by the hypervisor to this
6234 torture.disable_onoff_at_boot= [KNL]
6235 Prevent the CPU-hotplug component of torturing
6236 until after init has spawned.
6238 torture.ftrace_dump_at_shutdown= [KNL]
6239 Dump the ftrace buffer at torture-test shutdown,
6240 even if there were no errors. This can be a
6241 very costly operation when many torture tests
6242 are running concurrently, especially on systems
6243 with rotating-rust storage.
6245 torture.verbose_sleep_frequency= [KNL]
6246 Specifies how many verbose printk()s should be
6247 emitted between each sleep. The default of zero
6248 disables verbose-printk() sleeping.
6250 torture.verbose_sleep_duration= [KNL]
6251 Duration of each verbose-printk() sleep in jiffies.
6255 tpm_suspend_pcr=[HW,TPM]
6256 Format: integer pcr id
6257 Specify that at suspend time, the tpm driver
6258 should extend the specified pcr with zeros,
6259 as a workaround for some chips which fail to
6260 flush the last written pcr on TPM_SaveState.
6261 This will guarantee that all the other pcrs
6265 Have the tracepoints sent to printk as well as the
6266 tracing ring buffer. This is useful for early boot up
6267 where the system hangs or reboots and does not give the
6268 option for reading the tracing buffer or performing a
6269 ftrace_dump_on_oops.
6271 To turn off having tracepoints sent to printk,
6272 echo 0 > /proc/sys/kernel/tracepoint_printk
6273 Note, echoing 1 into this file without the
6274 tracepoint_printk kernel cmdline option has no effect.
6276 The tp_printk_stop_on_boot (see below) can also be used
6277 to stop the printing of events to console at
6282 Having tracepoints sent to printk() and activating high
6283 frequency tracepoints such as irq or sched, can cause
6284 the system to live lock.
6286 tp_printk_stop_on_boot [FTRACE]
6287 When tp_printk (above) is set, it can cause a lot of noise
6288 on the console. It may be useful to only include the
6289 printing of events during boot up, as user space may
6290 make the system inoperable.
6292 This command line option will stop the printing of events
6293 to console at the late_initcall_sync() time frame.
6295 trace_buf_size=nn[KMG]
6296 [FTRACE] will set tracing buffer size on each cpu.
6298 trace_clock= [FTRACE] Set the clock used for tracing events
6300 local - Use the per CPU time stamp counter
6301 (converted into nanoseconds). Fast, but
6302 depending on the architecture, may not be
6303 in sync between CPUs.
6304 global - Event time stamps are synchronize across
6305 CPUs. May be slower than the local clock,
6306 but better for some race conditions.
6307 counter - Simple counting of events (1, 2, ..)
6308 note, some counts may be skipped due to the
6309 infrastructure grabbing the clock more than
6311 uptime - Use jiffies as the time stamp.
6312 perf - Use the same clock that perf uses.
6313 mono - Use ktime_get_mono_fast_ns() for time stamps.
6314 mono_raw - Use ktime_get_raw_fast_ns() for time
6316 boot - Use ktime_get_boot_fast_ns() for time stamps.
6317 Architectures may add more clocks. See
6318 Documentation/trace/ftrace.rst for more details.
6320 trace_event=[event-list]
6321 [FTRACE] Set and start specified trace events in order
6322 to facilitate early boot debugging. The event-list is a
6323 comma-separated list of trace events to enable. See
6324 also Documentation/trace/events.rst
6326 trace_options=[option-list]
6327 [FTRACE] Enable or disable tracer options at boot.
6328 The option-list is a comma delimited list of options
6329 that can be enabled or disabled just as if you were
6330 to echo the option name into
6332 /sys/kernel/debug/tracing/trace_options
6334 For example, to enable stacktrace option (to dump the
6335 stack trace of each event), add to the command line:
6337 trace_options=stacktrace
6339 See also Documentation/trace/ftrace.rst "trace options"
6343 [FTRACE] enable this option to disable tracing when a
6344 warning is hit. This turns off "tracing_on". Tracing can
6345 be enabled again by echoing '1' into the "tracing_on"
6346 file located in /sys/kernel/debug/tracing/
6348 This option is useful, as it disables the trace before
6349 the WARNING dump is called, which prevents the trace to
6350 be filled with content caused by the warning output.
6352 This option can also be set at run time via the sysctl
6353 option: kernel/traceoff_on_warning
6355 transparent_hugepage=
6357 Format: [always|madvise|never]
6358 Can be used to control the default behavior of the system
6359 with respect to transparent hugepages.
6360 See Documentation/admin-guide/mm/transhuge.rst
6363 trusted.source= [KEYS]
6365 This parameter identifies the trust source as a backend
6366 for trusted keys implementation. Supported trust
6371 If not specified then it defaults to iterating through
6372 the trust source list starting with TPM and assigns the
6373 first trust source as a backend which is initialized
6374 successfully during iteration.
6378 The RNG used to generate key material for trusted keys.
6381 - the same value as trusted.source: "tpm" or "tee"
6383 If not specified, "default" is used. In this case,
6384 the RNG's choice is left to each individual trust source.
6386 tsc= Disable clocksource stability checks for TSC.
6388 [x86] reliable: mark tsc clocksource as reliable, this
6389 disables clocksource verification at runtime, as well
6390 as the stability checks done at bootup. Used to enable
6391 high-resolution timer mode on older hardware, and in
6392 virtualized environment.
6393 [x86] noirqtime: Do not use TSC to do irq accounting.
6394 Used to run time disable IRQ_TIME_ACCOUNTING on any
6395 platforms where RDTSC is slow and this accounting
6397 [x86] unstable: mark the TSC clocksource as unstable, this
6398 marks the TSC unconditionally unstable at bootup and
6399 avoids any further wobbles once the TSC watchdog notices.
6400 [x86] nowatchdog: disable clocksource watchdog. Used
6401 in situations with strict latency requirements (where
6402 interruptions from clocksource watchdog are not
6405 tsc_early_khz= [X86] Skip early TSC calibration and use the given
6406 value instead. Useful when the early TSC frequency discovery
6407 procedure is not reliable, such as on overclocked systems
6408 with CPUID.16h support and partial CPUID.15h support.
6409 Format: <unsigned int>
6411 tsx= [X86] Control Transactional Synchronization
6412 Extensions (TSX) feature in Intel processors that
6413 support TSX control.
6415 This parameter controls the TSX feature. The options are:
6417 on - Enable TSX on the system. Although there are
6418 mitigations for all known security vulnerabilities,
6419 TSX has been known to be an accelerator for
6420 several previous speculation-related CVEs, and
6421 so there may be unknown security risks associated
6422 with leaving it enabled.
6424 off - Disable TSX on the system. (Note that this
6425 option takes effect only on newer CPUs which are
6426 not vulnerable to MDS, i.e., have
6427 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
6428 the new IA32_TSX_CTRL MSR through a microcode
6429 update. This new MSR allows for the reliable
6430 deactivation of the TSX functionality.)
6432 auto - Disable TSX if X86_BUG_TAA is present,
6433 otherwise enable TSX on the system.
6435 Not specifying this option is equivalent to tsx=off.
6437 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6440 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
6441 Abort (TAA) vulnerability.
6443 Similar to Micro-architectural Data Sampling (MDS)
6444 certain CPUs that support Transactional
6445 Synchronization Extensions (TSX) are vulnerable to an
6446 exploit against CPU internal buffers which can forward
6447 information to a disclosure gadget under certain
6450 In vulnerable processors, the speculatively forwarded
6451 data can be used in a cache side channel attack, to
6452 access data to which the attacker does not have direct
6455 This parameter controls the TAA mitigation. The
6458 full - Enable TAA mitigation on vulnerable CPUs
6461 full,nosmt - Enable TAA mitigation and disable SMT on
6462 vulnerable CPUs. If TSX is disabled, SMT
6463 is not disabled because CPU is not
6464 vulnerable to cross-thread TAA attacks.
6465 off - Unconditionally disable TAA mitigation
6467 On MDS-affected machines, tsx_async_abort=off can be
6468 prevented by an active MDS mitigation as both vulnerabilities
6469 are mitigated with the same mechanism so in order to disable
6470 this mitigation, you need to specify mds=off too.
6472 Not specifying this option is equivalent to
6473 tsx_async_abort=full. On CPUs which are MDS affected
6474 and deploy MDS mitigation, TAA mitigation is not
6475 required and doesn't provide any additional
6479 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6481 turbografx.map[2|3]= [HW,JOY]
6482 TurboGraFX parallel port interface
6484 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
6485 See also Documentation/input/devices/joystick-parport.rst
6487 udbg-immortal [PPC] When debugging early kernel crashes that
6488 happen after console_init() and before a proper
6489 console driver takes over, this boot options might
6490 help "seeing" what's going on.
6492 uhash_entries= [KNL,NET]
6493 Set number of hash buckets for UDP/UDP-Lite connections
6496 [USB] Ignore overcurrent events (default N).
6497 Some badly-designed motherboards generate lots of
6498 bogus events, for ports that aren't wired to
6499 anything. Set this parameter to avoid log spamming.
6500 Note that genuine overcurrent events won't be
6504 [X86] Cause panic on unknown NMI.
6506 usbcore.authorized_default=
6507 [USB] Default USB device authorization:
6508 (default -1 = authorized except for wireless USB,
6509 0 = not authorized, 1 = authorized, 2 = authorized
6510 if device connected to internal port)
6512 usbcore.autosuspend=
6513 [USB] The autosuspend time delay (in seconds) used
6514 for newly-detected USB devices (default 2). This
6515 is the time required before an idle device will be
6516 autosuspended. Devices for which the delay is set
6517 to a negative value won't be autosuspended at all.
6519 usbcore.usbfs_snoop=
6520 [USB] Set to log all usbfs traffic (default 0 = off).
6522 usbcore.usbfs_snoop_max=
6523 [USB] Maximum number of bytes to snoop in each URB
6526 usbcore.blinkenlights=
6527 [USB] Set to cycle leds on hubs (default 0 = off).
6529 usbcore.old_scheme_first=
6530 [USB] Start with the old device initialization
6531 scheme (default 0 = off).
6533 usbcore.usbfs_memory_mb=
6534 [USB] Memory limit (in MB) for buffers allocated by
6535 usbfs (default = 16, 0 = max = 2047).
6537 usbcore.use_both_schemes=
6538 [USB] Try the other device initialization scheme
6539 if the first one fails (default 1 = enabled).
6541 usbcore.initial_descriptor_timeout=
6542 [USB] Specifies timeout for the initial 64-byte
6543 USB_REQ_GET_DESCRIPTOR request in milliseconds
6544 (default 5000 = 5.0 seconds).
6546 usbcore.nousb [USB] Disable the USB subsystem
6549 [USB] A list of quirk entries to augment the built-in
6550 usb core quirk list. List entries are separated by
6551 commas. Each entry has the form
6552 VendorID:ProductID:Flags. The IDs are 4-digit hex
6553 numbers and Flags is a set of letters. Each letter
6554 will change the built-in quirk; setting it if it is
6555 clear and clearing it if it is set. The letters have
6556 the following meanings:
6557 a = USB_QUIRK_STRING_FETCH_255 (string
6558 descriptors must not be fetched using
6560 b = USB_QUIRK_RESET_RESUME (device can't resume
6561 correctly so reset it instead);
6562 c = USB_QUIRK_NO_SET_INTF (device can't handle
6563 Set-Interface requests);
6564 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
6565 handle its Configuration or Interface
6567 e = USB_QUIRK_RESET (device can't be reset
6568 (e.g morph devices), don't use reset);
6569 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
6570 more interface descriptions than the
6571 bNumInterfaces count, and can't handle
6572 talking to these interfaces);
6573 g = USB_QUIRK_DELAY_INIT (device needs a pause
6574 during initialization, after we read
6575 the device descriptor);
6576 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
6577 high speed and super speed interrupt
6578 endpoints, the USB 2.0 and USB 3.0 spec
6579 require the interval in microframes (1
6580 microframe = 125 microseconds) to be
6581 calculated as interval = 2 ^
6583 Devices with this quirk report their
6584 bInterval as the result of this
6585 calculation instead of the exponent
6586 variable used in the calculation);
6587 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
6588 handle device_qualifier descriptor
6590 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
6591 generates spurious wakeup, ignore
6592 remote wakeup capability);
6593 k = USB_QUIRK_NO_LPM (device can't handle Link
6595 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
6596 (Device reports its bInterval as linear
6597 frames instead of the USB 2.0
6599 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
6600 to be disconnected before suspend to
6601 prevent spurious wakeup);
6602 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
6603 pause after every control message);
6604 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
6605 delay after resetting its port);
6606 p = USB_QUIRK_SHORT_SET_ADDRESS_REQ_TIMEOUT
6607 (Reduce timeout of the SET_ADDRESS
6608 request from 5000 ms to 500 ms);
6609 Example: quirks=0781:5580:bk,0a5c:5834:gij
6612 [USBHID] The interval which mice are to be polled at.
6615 [USBHID] The interval which joysticks are to be polled at.
6618 [USBHID] The interval which keyboards are to be polled at.
6620 usb-storage.delay_use=
6621 [UMS] The delay in seconds before a new device is
6622 scanned for Logical Units (default 1).
6625 [UMS] A list of quirks entries to supplement or
6626 override the built-in unusual_devs list. List
6627 entries are separated by commas. Each entry has
6628 the form VID:PID:Flags where VID and PID are Vendor
6629 and Product ID values (4-digit hex numbers) and
6630 Flags is a set of characters, each corresponding
6631 to a common usb-storage quirk flag as follows:
6632 a = SANE_SENSE (collect more than 18 bytes
6633 of sense data, not on uas);
6634 b = BAD_SENSE (don't collect more than 18
6635 bytes of sense data, not on uas);
6636 c = FIX_CAPACITY (decrease the reported
6637 device capacity by one sector);
6638 d = NO_READ_DISC_INFO (don't use
6639 READ_DISC_INFO command, not on uas);
6640 e = NO_READ_CAPACITY_16 (don't use
6641 READ_CAPACITY_16 command);
6642 f = NO_REPORT_OPCODES (don't use report opcodes
6644 g = MAX_SECTORS_240 (don't transfer more than
6645 240 sectors at a time, uas only);
6646 h = CAPACITY_HEURISTICS (decrease the
6647 reported device capacity by one
6648 sector if the number is odd);
6649 i = IGNORE_DEVICE (don't bind to this
6651 j = NO_REPORT_LUNS (don't use report luns
6653 k = NO_SAME (do not use WRITE_SAME, uas only)
6654 l = NOT_LOCKABLE (don't try to lock and
6655 unlock ejectable media, not on uas);
6656 m = MAX_SECTORS_64 (don't transfer more
6657 than 64 sectors = 32 KB at a time,
6659 n = INITIAL_READ10 (force a retry of the
6660 initial READ(10) command, not on uas);
6661 o = CAPACITY_OK (accept the capacity
6662 reported by the device, not on uas);
6663 p = WRITE_CACHE (the device cache is ON
6664 by default, not on uas);
6665 r = IGNORE_RESIDUE (the device reports
6666 bogus residue values, not on uas);
6667 s = SINGLE_LUN (the device has only one
6669 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
6670 commands, uas only);
6671 u = IGNORE_UAS (don't bind to the uas driver);
6672 w = NO_WP_DETECT (don't test whether the
6673 medium is write-protected).
6674 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
6675 even if the device claims no cache,
6677 Example: quirks=0419:aaf5:rl,0421:0433:rc
6679 user_debug= [KNL,ARM]
6681 See arch/arm/Kconfig.debug help text.
6682 1 - undefined instruction events
6684 4 - invalid data aborts
6687 Example: user_debug=31
6690 [X86] Flags controlling user PTE allocations.
6692 nohigh = do not allocate PTE pages in
6693 HIGHMEM regardless of setting
6696 vdso= [X86,SH,SPARC]
6697 On X86_32, this is an alias for vdso32=. Otherwise:
6699 vdso=1: enable VDSO (the default)
6700 vdso=0: disable VDSO mapping
6702 vdso32= [X86] Control the 32-bit vDSO
6703 vdso32=1: enable 32-bit VDSO
6704 vdso32=0 or vdso32=2: disable 32-bit VDSO
6706 See the help text for CONFIG_COMPAT_VDSO for more
6707 details. If CONFIG_COMPAT_VDSO is set, the default is
6708 vdso32=0; otherwise, the default is vdso32=1.
6710 For compatibility with older kernels, vdso32=2 is an
6713 Try vdso32=0 if you encounter an error that says:
6714 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
6717 vector=percpu: enable percpu vector domain
6719 video= [FB] Frame buffer configuration
6720 See Documentation/fb/modedb.rst.
6722 video.brightness_switch_enabled= [ACPI]
6724 If set to 1, on receiving an ACPI notify event
6725 generated by hotkey, video driver will adjust brightness
6726 level and then send out the event to user space through
6727 the allocated input device. If set to 0, video driver
6728 will only send out the event without touching backlight
6733 [VMMIO] Memory mapped virtio (platform) device.
6735 <size>@<baseaddr>:<irq>[:<id>]
6737 <size> := size (can use standard suffixes
6739 <baseaddr> := physical base address
6740 <irq> := interrupt number (as passed to
6742 <id> := (optional) platform device id
6744 virtio_mmio.device=1K@0x100b0000:48:7
6746 Can be used multiple times for multiple devices.
6748 vga= [BOOT,X86-32] Select a particular video mode
6749 See Documentation/x86/boot.rst and
6750 Documentation/admin-guide/svga.rst.
6751 Use vga=ask for menu.
6752 This is actually a boot loader parameter; the value is
6753 passed to the kernel using a special protocol.
6755 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
6756 May slow down system boot speed, especially when
6757 enabled on systems with a large amount of memory.
6758 All options are enabled by default, and this
6759 interface is meant to allow for selectively
6760 enabling or disabling specific virtual memory
6763 Available options are:
6764 P Enable page structure init time poisoning
6765 - Disable all of the above options
6767 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
6768 size of <nn>. This can be used to increase the
6769 minimum size (128MB on x86). It can also be used to
6770 decrease the size and leave more room for directly
6773 vmcp_cma=nn[MG] [KNL,S390]
6774 Sets the memory size reserved for contiguous memory
6775 allocations for the vmcp device driver.
6777 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
6780 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
6783 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
6787 Controls the behavior of vsyscalls (i.e. calls to
6788 fixed addresses of 0xffffffffff600x00 from legacy
6789 code). Most statically-linked binaries and older
6790 versions of glibc use these calls. Because these
6791 functions are at fixed addresses, they make nice
6792 targets for exploits that can control RIP.
6794 emulate [default] Vsyscalls turn into traps and are
6795 emulated reasonably safely. The vsyscall
6798 xonly Vsyscalls turn into traps and are
6799 emulated reasonably safely. The vsyscall
6800 page is not readable.
6802 none Vsyscalls don't work at all. This makes
6803 them quite hard to use for exploits but
6804 might break your system.
6806 vt.color= [VT] Default text color.
6807 Format: 0xYX, X = foreground, Y = background.
6808 Default: 0x07 = light gray on black.
6810 vt.cur_default= [VT] Default cursor shape.
6811 Format: 0xCCBBAA, where AA, BB, and CC are the same as
6812 the parameters of the <Esc>[?A;B;Cc escape sequence;
6813 see VGA-softcursor.txt. Default: 2 = underline.
6815 vt.default_blu= [VT]
6816 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
6817 Change the default blue palette of the console.
6818 This is a 16-member array composed of values
6821 vt.default_grn= [VT]
6822 Format: <green0>,<green1>,<green2>,...,<green15>
6823 Change the default green palette of the console.
6824 This is a 16-member array composed of values
6827 vt.default_red= [VT]
6828 Format: <red0>,<red1>,<red2>,...,<red15>
6829 Change the default red palette of the console.
6830 This is a 16-member array composed of values
6836 Set system-wide default UTF-8 mode for all tty's.
6837 Default is 1, i.e. UTF-8 mode is enabled for all
6838 newly opened terminals.
6840 vt.global_cursor_default=
6843 Set system-wide default for whether a cursor
6844 is shown on new VTs. Default is -1,
6845 i.e. cursors will be created by default unless
6846 overridden by individual drivers. 0 will hide
6847 cursors, 1 will display them.
6849 vt.italic= [VT] Default color for italic text; 0-15.
6852 vt.underline= [VT] Default color for underlined text; 0-15.
6855 watchdog timers [HW,WDT] For information on watchdog timers,
6856 see Documentation/watchdog/watchdog-parameters.rst
6857 or other driver-specific files in the
6858 Documentation/watchdog/ directory.
6862 Set the hard lockup detector stall duration
6863 threshold in seconds. The soft lockup detector
6864 threshold is set to twice the value. A value of 0
6865 disables both lockup detectors. Default is 10
6868 workqueue.watchdog_thresh=
6869 If CONFIG_WQ_WATCHDOG is configured, workqueue can
6870 warn stall conditions and dump internal state to
6871 help debugging. 0 disables workqueue stall
6872 detection; otherwise, it's the stall threshold
6873 duration in seconds. The default value is 30 and
6874 it can be updated at runtime by writing to the
6875 corresponding sysfs file.
6877 workqueue.disable_numa
6878 By default, all work items queued to unbound
6879 workqueues are affine to the NUMA nodes they're
6880 issued on, which results in better behavior in
6881 general. If NUMA affinity needs to be disabled for
6882 whatever reason, this option can be used. Note
6883 that this also can be controlled per-workqueue for
6884 workqueues visible under /sys/bus/workqueue/.
6886 workqueue.power_efficient
6887 Per-cpu workqueues are generally preferred because
6888 they show better performance thanks to cache
6889 locality; unfortunately, per-cpu workqueues tend to
6890 be more power hungry than unbound workqueues.
6892 Enabling this makes the per-cpu workqueues which
6893 were observed to contribute significantly to power
6894 consumption unbound, leading to measurably lower
6895 power usage at the cost of small performance
6898 The default value of this parameter is determined by
6899 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
6901 workqueue.debug_force_rr_cpu
6902 Workqueue used to implicitly guarantee that work
6903 items queued without explicit CPU specified are put
6904 on the local CPU. This guarantee is no longer true
6905 and while local CPU is still preferred work items
6906 may be put on foreign CPUs. This debug option
6907 forces round-robin CPU selection to flush out
6908 usages which depend on the now broken guarantee.
6909 When enabled, memory and cache locality will be
6912 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
6913 default x2apic cluster mode on platforms
6916 xen_512gb_limit [KNL,X86-64,XEN]
6917 Restricts the kernel running paravirtualized under Xen
6918 to use only up to 512 GB of RAM. The reason to do so is
6919 crash analysis tools and Xen tools for doing domain
6920 save/restore/migration must be enabled to handle larger
6923 xen_emul_unplug= [HW,X86,XEN]
6924 Unplug Xen emulated devices
6925 Format: [unplug0,][unplug1]
6926 ide-disks -- unplug primary master IDE devices
6927 aux-ide-disks -- unplug non-primary-master IDE devices
6928 nics -- unplug network devices
6929 all -- unplug all emulated devices (NICs and IDE disks)
6930 unnecessary -- unplugging emulated devices is
6931 unnecessary even if the host did not respond to
6933 never -- do not unplug even if version check succeeds
6935 xen_legacy_crash [X86,XEN]
6936 Crash from Xen panic notifier, without executing late
6937 panic() code such as dumping handler.
6939 xen_msr_safe= [X86,XEN]
6941 Select whether to always use non-faulting (safe) MSR
6942 access functions when running as Xen PV guest. The
6943 default value is controlled by CONFIG_XEN_PV_MSR_SAFE.
6945 xen_nopvspin [X86,XEN]
6946 Disables the qspinlock slowpath using Xen PV optimizations.
6947 This parameter is obsoleted by "nopvspin" parameter, which
6948 has equivalent effect for XEN platform.
6951 Disables the PV optimizations forcing the HVM guest to
6952 run as generic HVM guest with no PV drivers.
6953 This option is obsoleted by the "nopv" option, which
6954 has equivalent effect for XEN platform.
6956 xen_no_vector_callback
6957 [KNL,X86,XEN] Disable the vector callback for Xen
6958 event channel interrupts.
6960 xen_scrub_pages= [XEN]
6961 Boolean option to control scrubbing pages before giving them back
6962 to Xen, for use by other domains. Can be also changed at runtime
6963 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
6964 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
6966 xen_timer_slop= [X86-64,XEN]
6967 Set the timer slop (in nanoseconds) for the virtual Xen
6968 timers (default is 100000). This adjusts the minimum
6969 delta of virtualized Xen timers, where lower values
6970 improve timer resolution at the expense of processing
6971 more timer interrupts.
6973 xen.balloon_boot_timeout= [XEN]
6974 The time (in seconds) to wait before giving up to boot
6975 in case initial ballooning fails to free enough memory.
6976 Applies only when running as HVM or PVH guest and
6977 started with less memory configured than allowed at
6978 max. Default is 180.
6980 xen.event_eoi_delay= [XEN]
6981 How long to delay EOI handling in case of event
6982 storms (jiffies). Default is 10.
6984 xen.event_loop_timeout= [XEN]
6985 After which time (jiffies) the event handling loop
6986 should start to delay EOI handling. Default is 2.
6988 xen.fifo_events= [XEN]
6989 Boolean parameter to disable using fifo event handling
6990 even if available. Normally fifo event handling is
6991 preferred over the 2-level event handling, as it is
6992 fairer and the number of possible event channels is
6993 much higher. Default is on (use fifo events).
6995 nopv= [X86,XEN,KVM,HYPER_V,VMWARE]
6996 Disables the PV optimizations forcing the guest to run
6997 as generic guest with no PV drivers. Currently support
6998 XEN HVM, KVM, HYPER_V and VMWARE guest.
7000 nopvspin [X86,XEN,KVM]
7001 Disables the qspinlock slow path using PV optimizations
7002 which allow the hypervisor to 'idle' the guest on lock
7005 xirc2ps_cs= [NET,PCMCIA]
7007 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
7010 By default on POWER9 and above, the kernel will
7011 natively use the XIVE interrupt controller. This option
7012 allows the fallback firmware mode to be used:
7014 off Fallback to firmware control of XIVE interrupt
7015 controller on both pseries and powernv
7016 platforms. Only useful on POWER9 and above.
7018 xive.store-eoi=off [PPC]
7019 By default on POWER10 and above, the kernel will use
7020 stores for EOI handling when the XIVE interrupt mode
7021 is active. This option allows the XIVE driver to use
7022 loads instead, as on POWER9.
7024 xhci-hcd.quirks [USB,KNL]
7025 A hex value specifying bitmask with supplemental xhci
7026 host controller quirks. Meaning of each bit can be
7027 consulted in header drivers/usb/host/xhci.h.
7030 Format: { early | on | rw | ro | off }
7031 Controls if xmon debugger is enabled. Default is off.
7032 Passing only "xmon" is equivalent to "xmon=early".
7033 early Call xmon as early as possible on boot; xmon
7034 debugger is called from setup_arch().
7035 on xmon debugger hooks will be installed so xmon
7036 is only called on a kernel crash. Default mode,
7037 i.e. either "ro" or "rw" mode, is controlled
7038 with CONFIG_XMON_DEFAULT_RO_MODE.
7039 rw xmon debugger hooks will be installed so xmon
7040 is called only on a kernel crash, mode is write,
7041 meaning SPR registers, memory and, other data
7042 can be written using xmon commands.
7043 ro same as "rw" option above but SPR registers,
7044 memory, and other data can't be written using
7046 off xmon is disabled.
7050 Do not enable amd_pstate as the default
7051 scaling driver for the supported processors
7053 Use amd_pstate as a scaling driver, driver requests a
7054 desired performance on this abstract scale and the power
7055 management firmware translates the requests into actual
7056 hardware states (core frequency, data fabric and memory