1 acpi= [HW,ACPI,X86,ARM64,RISCV64]
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,riscv64]
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 and RISCV64, ONLY "acpi=off", "acpi=on" or
14 "acpi=force" are available
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/arch/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.)
345 Do not enable amd_pstate as the default
346 scaling driver for the supported processors
348 Use amd_pstate with passive mode as a scaling driver.
349 In this mode autonomous selection is disabled.
350 Driver requests a desired performance level and platform
351 tries to match the same performance level if it is
352 satisfied by guaranteed performance level.
354 Use amd_pstate_epp driver instance as the scaling driver,
355 driver provides a hint to the hardware if software wants
356 to bias toward performance (0x0) or energy efficiency (0xff)
357 to the CPPC firmware. then CPPC power algorithm will
358 calculate the runtime workload and adjust the realtime cores
361 Activate guided autonomous mode. Driver requests minimum and
362 maximum performance level and the platform autonomously
363 selects a performance level in this range and appropriate
364 to the current workload.
366 amijoy.map= [HW,JOY] Amiga joystick support
367 Map of devices attached to JOY0DAT and JOY1DAT
369 See also Documentation/input/joydev/joystick.rst
371 analog.map= [HW,JOY] Analog joystick and gamepad support
372 Specifies type or capabilities of an analog joystick
373 connected to one of 16 gameports
374 Format: <type1>,<type2>,..<type16>
377 Power management functions (SPARCstation-4/5 + deriv.)
379 Disable APC CPU standby support. SPARCstation-Fox does
380 not play well with APC CPU idle - disable it if you have
381 APC and your system crashes randomly.
383 apic= [APIC,X86] Advanced Programmable Interrupt Controller
384 Change the output verbosity while booting
385 Format: { quiet (default) | verbose | debug }
386 Change the amount of debugging information output
387 when initialising the APIC and IO-APIC components.
388 For X86-32, this can also be used to specify an APIC
390 Format: apic=driver_name
391 Examples: apic=bigsmp
393 apic_extnmi= [APIC,X86] External NMI delivery setting
394 Format: { bsp (default) | all | none }
395 bsp: External NMI is delivered only to CPU 0
396 all: External NMIs are broadcast to all CPUs as a
398 none: External NMI is masked for all CPUs. This is
399 useful so that a dump capture kernel won't be
403 See Documentation/networking/ipv6.rst.
405 apm= [APM] Advanced Power Management
406 See header of arch/x86/kernel/apm_32.c.
408 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
409 Format: { "0" | "1" }
410 See security/apparmor/Kconfig help text
413 Default value is set via kernel config option.
415 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
416 Format: <io>,<irq>,<nodeID>
418 arm64.nobti [ARM64] Unconditionally disable Branch Target
419 Identification support
421 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication
424 arm64.nomte [ARM64] Unconditionally disable Memory Tagging Extension
427 arm64.nosve [ARM64] Unconditionally disable Scalable Vector
430 arm64.nosme [ARM64] Unconditionally disable Scalable Matrix
433 arm64.nomops [ARM64] Unconditionally disable Memory Copy and Memory
434 Set instructions support
438 atarimouse= [HW,MOUSE] Atari Mouse
440 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
441 EzKey and similar keyboards
443 atkbd.reset= [HW] Reset keyboard during initialization
445 atkbd.set= [HW] Select keyboard code set
446 Format: <int> (2 = AT (default), 3 = PS/2)
448 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
451 atkbd.softraw= [HW] Choose between synthetic and real raw mode
452 Format: <bool> (0 = real, 1 = synthetic (default))
454 atkbd.softrepeat= [HW]
455 Use software keyboard repeat
457 audit= [KNL] Enable the audit sub-system
458 Format: { "0" | "1" | "off" | "on" }
459 0 | off - kernel audit is disabled and can not be
460 enabled until the next reboot
461 unset - kernel audit is initialized but disabled and
462 will be fully enabled by the userspace auditd.
463 1 | on - kernel audit is initialized and partially
464 enabled, storing at most audit_backlog_limit
465 messages in RAM until it is fully enabled by the
469 audit_backlog_limit= [KNL] Set the audit queue size limit.
470 Format: <int> (must be >=0)
473 bau= [X86_UV] Enable the BAU on SGI UV. The default
474 behavior is to disable the BAU (i.e. bau=0).
475 Format: { "0" | "1" }
478 unset - Disable the BAU.
480 baycom_epp= [HW,AX25]
483 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
485 See header of drivers/net/hamradio/baycom_par.c.
487 baycom_ser_fdx= [HW,AX25]
488 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
489 Format: <io>,<irq>,<mode>[,<baud>]
490 See header of drivers/net/hamradio/baycom_ser_fdx.c.
492 baycom_ser_hdx= [HW,AX25]
493 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
494 Format: <io>,<irq>,<mode>
495 See header of drivers/net/hamradio/baycom_ser_hdx.c.
498 Disable BERT OS support on buggy BIOSes.
500 bgrt_disable [ACPI][X86]
501 Disable BGRT to avoid flickering OEM logo.
503 blkdevparts= Manual partition parsing of block device(s) for
504 embedded devices based on command line input.
505 See Documentation/block/cmdline-partition.rst
507 boot_delay= Milliseconds to delay each printk during boot.
508 Only works if CONFIG_BOOT_PRINTK_DELAY is enabled,
509 and you may also have to specify "lpj=". Boot_delay
510 values larger than 10 seconds (10000) are assumed
511 erroneous and ignored.
515 Extended command line options can be added to an initrd
516 and this will cause the kernel to look for it.
518 See Documentation/admin-guide/bootconfig.rst
520 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
521 bttv.radio= Most important insmod options are available as
523 bttv.pll= See Documentation/admin-guide/media/bttv.rst
526 bulk_remove=off [PPC] This parameter disables the use of the pSeries
527 firmware feature for flushing multiple hpte entries
530 c101= [NET] Moxa C101 synchronous serial card
532 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
533 Sometimes CPU hardware bugs make them report the cache
534 size incorrectly. The kernel will attempt work arounds
535 to fix known problems, but for some CPUs it is not
536 possible to determine what the correct size should be.
537 This option provides an override for these situations.
540 [NET] Specifies amount of time (in seconds) that
541 the kernel should wait for a network carrier. By default
542 it waits 120 seconds.
544 ca_keys= [KEYS] This parameter identifies a specific key(s) on
545 the system trusted keyring to be used for certificate
547 format: { id:<keyid> | builtin }
549 cca= [MIPS] Override the kernel pages' cache coherency
550 algorithm. Accepted values range from 0 to 7
551 inclusive. See arch/mips/include/asm/pgtable-bits.h
552 for platform specific values (SB1, Loongson3 and
555 ccw_timeout_log [S390]
556 See Documentation/arch/s390/common_io.rst for details.
558 cgroup_disable= [KNL] Disable a particular controller or optional feature
559 Format: {name of the controller(s) or feature(s) to disable}
560 The effects of cgroup_disable=foo are:
561 - foo isn't auto-mounted if you mount all cgroups in
563 - foo isn't visible as an individually mountable
565 - if foo is an optional feature then the feature is
566 disabled and corresponding cgroup files are not
568 {Currently only "memory" controller deal with this and
569 cut the overhead, others just disable the usage. So
570 only cgroup_disable=memory is actually worthy}
571 Specifying "pressure" disables per-cgroup pressure
572 stall information accounting feature
574 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
575 Format: { { controller | "all" | "named" }
576 [,{ controller | "all" | "named" }...] }
577 Like cgroup_disable, but only applies to cgroup v1;
578 the blacklisted controllers remain available in cgroup2.
579 "all" blacklists all controllers and "named" disables
580 named mounts. Specifying both "all" and "named" disables
583 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
585 nosocket -- Disable socket memory accounting.
586 nokmem -- Disable kernel memory accounting.
587 nobpf -- Disable BPF memory accounting.
589 checkreqprot= [SELINUX] Set initial checkreqprot flag value.
590 Format: { "0" | "1" }
591 See security/selinux/Kconfig help text.
592 0 -- check protection applied by kernel (includes
593 any implied execute protection).
594 1 -- check protection requested by application.
595 Default value is set via a kernel config option.
596 Value can be changed at runtime via
597 /sys/fs/selinux/checkreqprot.
598 Setting checkreqprot to 1 is deprecated.
601 See Documentation/arch/s390/common_io.rst for details.
603 clearcpuid=X[,X...] [X86]
604 Disable CPUID feature X for the kernel. See
605 arch/x86/include/asm/cpufeatures.h for the valid bit
606 numbers X. Note the Linux-specific bits are not necessarily
607 stable over kernel options, but the vendor-specific
609 X can also be a string as appearing in the flags: line
610 in /proc/cpuinfo which does not have the above
611 instability issue. However, not all features have names
613 Note that using this option will taint your kernel.
614 Also note that user programs calling CPUID directly
615 or using the feature without checking anything
616 will still see it. This just prevents it from
617 being used by the kernel or shown in /proc/cpuinfo.
618 Also note the kernel might malfunction if you disable
623 Prevents the clock framework from automatically gating
624 clocks that have not been explicitly enabled by a Linux
625 device driver but are enabled in hardware at reset or
626 by the bootloader/firmware. Note that this does not
627 force such clocks to be always-on nor does it reserve
628 those clocks in any way. This parameter is useful for
629 debug and development, but should not be needed on a
630 platform with proper driver support. For more
631 information, see Documentation/driver-api/clk.rst.
633 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
635 Forces specified clocksource (if available) to be used
636 when calculating gettimeofday(). If specified
637 clocksource is not available, it defaults to PIT.
638 Format: { pit | tsc | cyclone | pmtmr }
640 clocksource= Override the default clocksource
642 Override the default clocksource and use the clocksource
643 with the name specified.
644 Some clocksource names to choose from, depending on
646 [all] jiffies (this is the base, fallback clocksource)
648 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
649 pxa_timer,timer3,32k_counter,timer0_1
650 [X86-32] pit,hpet,tsc;
651 scx200_hrt on Geode; cyclone on IBM x440
659 clocksource.arm_arch_timer.evtstrm=
662 Enable/disable the eventstream feature of the ARM
663 architected timer so that code using WFE-based polling
664 loops can be debugged more effectively on production
667 clocksource.max_cswd_read_retries= [KNL]
668 Number of clocksource_watchdog() retries due to
669 external delays before the clock will be marked
670 unstable. Defaults to two retries, that is,
671 three attempts to read the clock under test.
673 clocksource.verify_n_cpus= [KNL]
674 Limit the number of CPUs checked for clocksources
675 marked with CLOCK_SOURCE_VERIFY_PERCPU that
676 are marked unstable due to excessive skew.
677 A negative value says to check all CPUs, while
678 zero says not to check any. Values larger than
679 nr_cpu_ids are silently truncated to nr_cpu_ids.
680 The actual CPUs are chosen randomly, with
681 no replacement if the same CPU is chosen twice.
683 clocksource-wdtest.holdoff= [KNL]
684 Set the time in seconds that the clocksource
685 watchdog test waits before commencing its tests.
686 Defaults to zero when built as a module and to
687 10 seconds when built into the kernel.
689 cma=nn[MG]@[start[MG][-end[MG]]]
691 Sets the size of kernel global memory area for
692 contiguous memory allocations and optionally the
693 placement constraint by the physical address range of
694 memory allocations. A value of 0 disables CMA
695 altogether. For more information, see
696 kernel/dma/contiguous.c
700 Sets the size of kernel per-numa memory area for
701 contiguous memory allocations. A value of 0 disables
702 per-numa CMA altogether. And If this option is not
703 specified, the default value is 0.
704 With per-numa CMA enabled, DMA users on node nid will
705 first try to allocate buffer from the pernuma area
706 which is located in node nid, if the allocation fails,
707 they will fallback to the global default memory area.
709 cmo_free_hint= [PPC] Format: { yes | no }
710 Specify whether pages are marked as being inactive
711 when they are freed. This is used in CMO environments
712 to determine OS memory pressure for page stealing by
716 coherent_pool=nn[KMG] [ARM,KNL]
717 Sets the size of memory pool for coherent, atomic dma
718 allocations, by default set to 256K.
720 com20020= [HW,NET] ARCnet - COM20020 chipset
722 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
724 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
728 ARCnet - COM90xx chipset (memory-mapped buffers)
729 Format: <io>[,<irq>[,<memstart>]]
731 condev= [HW,S390] console device
734 con3215_drop= [S390] 3215 console drop mode.
736 When set to true, drop data on the 3215 console when
737 the console buffer is full. In this case the
738 operator using a 3270 terminal emulator (for example
739 x3270) does not have to enter the clear key for the
740 console output to advance and the kernel to continue.
741 This leads to a much faster boot time when a 3270
742 terminal emulator is active. If no 3270 terminal
743 emulator is used, this parameter has no effect.
745 console= [KNL] Output console device and options.
747 tty<n> Use the virtual console device <n>.
751 Use the specified serial port. The options are of
752 the form "bbbbpnf", where "bbbb" is the baud rate,
753 "p" is parity ("n", "o", or "e"), "n" is number of
754 bits, and "f" is flow control ("r" for RTS or
755 omit it). Default is "9600n8".
757 See Documentation/admin-guide/serial-console.rst for more
759 Documentation/networking/netconsole.rst for an
762 uart[8250],io,<addr>[,options]
763 uart[8250],mmio,<addr>[,options]
764 uart[8250],mmio16,<addr>[,options]
765 uart[8250],mmio32,<addr>[,options]
766 uart[8250],0x<addr>[,options]
767 Start an early, polled-mode console on the 8250/16550
768 UART at the specified I/O port or MMIO address,
769 switching to the matching ttyS device later.
770 MMIO inter-register address stride is either 8-bit
771 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
772 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
773 to be equivalent to 'mmio'. 'options' are specified in
774 the same format described for ttyS above; if unspecified,
775 the h/w is not re-initialized.
777 hvc<n> Use the hypervisor console device <n>. This is for
778 both Xen and PowerPC hypervisors.
781 Use to disable console output, i.e., to have kernel
782 console messages discarded.
783 This must be the only console= parameter used on the
786 If the device connected to the port is not a TTY but a braille
787 device, prepend "brl," before the device type, for instance
789 For now, only VisioBraille is supported.
792 [KNL] Change console messages format
794 By default we print messages on consoles in
795 "[time stamp] text\n" format (time stamp may not be
796 printed, depending on CONFIG_PRINTK_TIME or
797 `printk_time' param).
799 Switch to syslog format: "<%u>[time stamp] text\n"
800 IOW, each message will have a facility and loglevel
801 prefix. The format is similar to one used by syslog()
802 syscall, or to executing "dmesg -S --raw" or to reading
805 consoleblank= [KNL] The console blank (screen saver) timeout in
806 seconds. A value of 0 disables the blank timer.
810 [KNL] Change the default value for
811 /proc/<pid>/coredump_filter.
812 See also Documentation/filesystems/proc.rst.
814 coresight_cpu_debug.enable
817 Enable/disable the CPU sampling based debugging.
818 0: default value, disable debugging
819 1: enable debugging at boot time
821 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
823 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
825 cpuidle.off=1 [CPU_IDLE]
826 disable the cpuidle sub-system
829 [CPU_IDLE] Name of the cpuidle governor to use.
831 cpufreq.off=1 [CPU_FREQ]
832 disable the cpufreq sub-system
834 cpufreq.default_governor=
835 [CPU_FREQ] Name of the default cpufreq governor or
836 policy to use. This governor must be registered in the
837 kernel before the cpufreq driver probes.
840 [X86] Delay for N microsec between assert and de-assert
841 of APIC INIT to start processors. This delay occurs
842 on every CPU online, such as boot, and resume from suspend.
846 [SMP] Enable/disable parallel bringup of secondary CPUs
848 Default is enabled if CONFIG_HOTPLUG_PARALLEL=y. Otherwise
849 the parameter has no effect.
851 crash_kexec_post_notifiers
852 Run kdump after running panic-notifiers and dumping
853 kmsg. This only for the users who doubt kdump always
854 succeeds in any situation.
855 Note that this also increases risks of kdump failure,
856 because some panic notifiers can make the crashed
857 kernel more unstable.
859 crashkernel=size[KMG][@offset[KMG]]
860 [KNL] Using kexec, Linux can switch to a 'crash kernel'
861 upon panic. This parameter reserves the physical
862 memory region [offset, offset + size] for that kernel
863 image. If '@offset' is omitted, then a suitable offset
864 is selected automatically.
865 [KNL, X86-64, ARM64] Select a region under 4G first, and
866 fall back to reserve region above 4G when '@offset'
867 hasn't been specified.
868 See Documentation/admin-guide/kdump/kdump.rst for further details.
870 crashkernel=range1:size1[,range2:size2,...][@offset]
871 [KNL] Same as above, but depends on the memory
872 in the running system. The syntax of range is
873 start-[end] where start and end are both
874 a memory unit (amount[KMG]). See also
875 Documentation/admin-guide/kdump/kdump.rst for an example.
877 crashkernel=size[KMG],high
878 [KNL, X86-64, ARM64] range could be above 4G. Allow kernel
879 to allocate physical memory region from top, so could
880 be above 4G if system have more than 4G ram installed.
881 Otherwise memory region will be allocated below 4G, if
883 It will be ignored if crashkernel=X is specified.
884 crashkernel=size[KMG],low
885 [KNL, X86-64, ARM64] range under 4G. When crashkernel=X,high
886 is passed, kernel could allocate physical memory region
887 above 4G, that cause second kernel crash on system
888 that require some amount of low memory, e.g. swiotlb
889 requires at least 64M+32K low memory, also enough extra
890 low memory is needed to make sure DMA buffers for 32-bit
891 devices won't run out. Kernel would try to allocate
892 default size of memory below 4G automatically. The default
893 size is platform dependent.
894 --> x86: max(swiotlb_size_or_default() + 8MiB, 256MiB)
896 This one lets the user specify own low range under 4G
897 for second kernel instead.
898 0: to disable low allocation.
899 It will be ignored when crashkernel=X,high is not used
900 or memory reserved is below 4G.
903 [KNL] Disable crypto self-tests
908 cs89x0_media= [HW,NET]
909 Format: { rj45 | aui | bnc }
911 csdlock_debug= [KNL] Enable or disable debug add-ons of cross-CPU
912 function call handling. When switched on,
913 additional debug data is printed to the console
914 in case a hanging CPU is detected, and that
915 CPU is pinged again in order to try to resolve
916 the hang situation. The default value of this
917 option depends on the CSD_LOCK_WAIT_DEBUG_DEFAULT
921 See header of drivers/s390/block/dasd_devmap.c.
923 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
924 (one device per port)
925 Format: <port#>,<type>
926 See also Documentation/input/devices/joystick-parport.rst
928 debug [KNL] Enable kernel debugging (events log level).
931 [KNL] Enable printing [hashed] pointers early in the
932 boot sequence. If enabled, we use a weak hash instead
933 of siphash to hash pointers. Use this option if you are
934 seeing instances of '(___ptrval___)') and need to see a
935 value (hashed pointer) instead. Cryptographically
936 insecure, please do not use on production kernels.
939 [KNL] verbose locking self-tests
941 Print debugging info while doing the locking API
943 Bitmask for the various LOCKTYPE_ tests. Defaults to 0
944 (no extra messages), setting it to -1 (all bits set)
945 will print _a_lot_ more information - normally only
946 useful to lockdep developers.
948 debug_objects [KNL] Enable object debugging
950 debug_guardpage_minorder=
951 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
952 parameter allows control of the order of pages that will
953 be intentionally kept free (and hence protected) by the
954 buddy allocator. Bigger value increase the probability
955 of catching random memory corruption, but reduce the
956 amount of memory for normal system use. The maximum
957 possible value is MAX_ORDER/2. Setting this parameter
958 to 1 or 2 should be enough to identify most random
959 memory corruption problems caused by bugs in kernel or
960 driver code when a CPU writes to (or reads from) a
961 random memory location. Note that there exists a class
962 of memory corruptions problems caused by buggy H/W or
963 F/W or by drivers badly programming DMA (basically when
964 memory is written at bus level and the CPU MMU is
965 bypassed) which are not detectable by
966 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
967 tracking down these problems.
970 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
971 enables the feature at boot time. By default, it is
972 disabled and the system will work mostly the same as a
973 kernel built without CONFIG_DEBUG_PAGEALLOC.
974 Note: to get most of debug_pagealloc error reports, it's
975 useful to also enable the page_owner functionality.
976 on: enable the feature
978 debugfs= [KNL] This parameter enables what is exposed to userspace
979 and debugfs internal clients.
980 Format: { on, no-mount, off }
981 on: All functions are enabled.
983 Filesystem is not registered but kernel clients can
984 access APIs and a crashkernel can be used to read
985 its content. There is nothing to mount.
986 off: Filesystem is not registered and clients
987 get a -EPERM as result when trying to register files
988 or directories within debugfs.
989 This is equivalent of the runtime functionality if
990 debugfs was not enabled in the kernel at all.
991 Default value is set in build-time with a kernel configuration.
993 debugpat [X86] Enable PAT debugging
996 [HW] The size of the default HugeTLB page. This is
997 the size represented by the legacy /proc/ hugepages
998 APIs. In addition, this is the default hugetlb size
999 used for shmget(), mmap() and mounting hugetlbfs
1000 filesystems. If not specified, defaults to the
1001 architecture's default huge page size. Huge page
1002 sizes are architecture dependent. See also
1003 Documentation/admin-guide/mm/hugetlbpage.rst.
1006 deferred_probe_timeout=
1007 [KNL] Debugging option to set a timeout in seconds for
1008 deferred probe to give up waiting on dependencies to
1009 probe. Only specific dependencies (subsystems or
1010 drivers) that have opted in will be ignored. A timeout
1011 of 0 will timeout at the end of initcalls. If the time
1012 out hasn't expired, it'll be restarted by each
1013 successful driver registration. This option will also
1014 dump out devices still on the deferred probe list after
1017 delayacct [KNL] Enable per-task delay accounting
1019 dell_smm_hwmon.ignore_dmi=
1020 [HW] Continue probing hardware even if DMI data
1021 indicates that the driver is running on unsupported
1024 dell_smm_hwmon.force=
1025 [HW] Activate driver even if SMM BIOS signature does
1026 not match list of supported models and enable otherwise
1027 blacklisted features.
1029 dell_smm_hwmon.power_status=
1030 [HW] Report power status in /proc/i8k
1031 (disabled by default).
1033 dell_smm_hwmon.restricted=
1034 [HW] Allow controlling fans only if SYS_ADMIN
1037 dell_smm_hwmon.fan_mult=
1038 [HW] Factor to multiply fan speed with.
1040 dell_smm_hwmon.fan_max=
1041 [HW] Maximum configurable fan speed.
1044 Format: { on | off | def_only | inf_only | always }
1045 on: s390 zlib hardware support for compression on
1046 level 1 and decompression (default)
1047 off: No s390 zlib hardware support
1048 def_only: s390 zlib hardware support for deflate
1049 only (compression on level 1)
1050 inf_only: s390 zlib hardware support for inflate
1051 only (decompression)
1052 always: Same as 'on' but ignores the selected compression
1053 level always using hardware support (used for debugging)
1055 dhash_entries= [KNL]
1056 Set number of hash buckets for dentry cache.
1058 disable_1tb_segments [PPC]
1059 Disables the use of 1TB hash page table segments. This
1060 causes the kernel to fall back to 256MB segments which
1061 can be useful when debugging issues that require an SLB
1065 See Documentation/networking/ipv6.rst.
1068 Disable RADIX MMU mode on POWER9
1071 Disable TLBIE instruction. Currently does not work
1072 with KVM, with HASH MMU, or with coherent accelerators.
1074 disable_cpu_apicid= [X86,APIC,SMP]
1076 The number of initial APIC ID for the
1077 corresponding CPU to be disabled at boot,
1078 mostly used for the kdump 2nd kernel to
1079 disable BSP to wake up multiple CPUs without
1080 causing system reset or hang due to sending
1081 INIT from AP to BSP.
1083 disable_ddw [PPC/PSERIES]
1084 Disable Dynamic DMA Window support. Use this
1085 to workaround buggy firmware.
1087 disable_ipv6= [IPV6]
1088 See Documentation/networking/ipv6.rst.
1090 disable_mtrr_cleanup [X86]
1091 The kernel tries to adjust MTRR layout from continuous
1092 to discrete, to make X server driver able to add WB
1093 entry later. This parameter disables that.
1095 disable_mtrr_trim [X86, Intel and AMD only]
1096 By default the kernel will trim any uncacheable
1097 memory out of your available memory pool based on
1098 MTRR settings. This parameter disables that behavior,
1099 possibly causing your machine to run very slowly.
1101 disable_timer_pin_1 [X86]
1102 Disable PIN 1 of APIC timer
1103 Can be useful to work around chipset bugs.
1105 dis_ucode_ldr [X86] Disable the microcode loader.
1107 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
1108 this option disables the debugging code at boot.
1110 dma_debug_entries=<number>
1111 This option allows to tune the number of preallocated
1112 entries for DMA-API debugging code. One entry is
1113 required per DMA-API allocation. Use this if the
1114 DMA-API debugging code disables itself because the
1115 architectural default is too low.
1117 dma_debug_driver=<driver_name>
1118 With this option the DMA-API debugging driver
1119 filter feature can be enabled at boot time. Just
1120 pass the driver to filter for as the parameter.
1121 The filter can be disabled or changed to another
1122 driver later using sysfs.
1124 driver_async_probe= [KNL]
1125 List of driver names to be probed asynchronously. *
1126 matches with all driver names. If * is specified, the
1127 rest of the listed driver names are those that will NOT
1129 Format: <driver_name1>,<driver_name2>...
1131 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
1132 Broken monitors, graphic adapters, KVMs and EDIDless
1133 panels may send no or incorrect EDID data sets.
1134 This parameter allows to specify an EDID data sets
1135 in the /lib/firmware directory that are used instead.
1136 Generic built-in EDID data sets are used, if one of
1137 edid/1024x768.bin, edid/1280x1024.bin,
1138 edid/1680x1050.bin, or edid/1920x1080.bin is given
1139 and no file with the same name exists. Details and
1140 instructions how to build your own EDID data are
1141 available in Documentation/admin-guide/edid.rst. An EDID
1142 data set will only be used for a particular connector,
1143 if its name and a colon are prepended to the EDID
1144 name. Each connector may use a unique EDID data
1145 set by separating the files with a comma. An EDID
1146 data set with no connector name will be used for
1147 any connectors not explicitly specified.
1152 Format: {"off" | "known"}
1153 Control how the dt_cpu_ftrs device-tree binding is
1154 used for CPU feature discovery and setup (if it
1156 off: Do not use it, fall back to legacy cpu table.
1157 known: Do not pass through unknown features to guests
1158 or userspace, only those that the kernel is aware of.
1160 dump_apple_properties [X86]
1161 Dump name and content of EFI device properties on
1162 x86 Macs. Useful for driver authors to determine
1163 what data is available or for reverse-engineering.
1165 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
1166 <module>.dyndbg[="val"]
1167 Enable debug messages at boot time. See
1168 Documentation/admin-guide/dynamic-debug-howto.rst
1171 early_ioremap_debug [KNL]
1172 Enable debug messages in early_ioremap support. This
1173 is useful for tracking down temporary early mappings
1174 which are not unmapped.
1176 earlycon= [KNL] Output early console device and options.
1178 When used with no options, the early console is
1179 determined by stdout-path property in device tree's
1180 chosen node or the ACPI SPCR table if supported by
1183 cdns,<addr>[,options]
1184 Start an early, polled-mode console on a Cadence
1185 (xuartps) serial port at the specified address. Only
1186 supported option is baud rate. If baud rate is not
1187 specified, the serial port must already be setup and
1190 uart[8250],io,<addr>[,options[,uartclk]]
1191 uart[8250],mmio,<addr>[,options[,uartclk]]
1192 uart[8250],mmio32,<addr>[,options[,uartclk]]
1193 uart[8250],mmio32be,<addr>[,options[,uartclk]]
1194 uart[8250],0x<addr>[,options]
1195 Start an early, polled-mode console on the 8250/16550
1196 UART at the specified I/O port or MMIO address.
1197 MMIO inter-register address stride is either 8-bit
1198 (mmio) or 32-bit (mmio32 or mmio32be).
1199 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1200 to be equivalent to 'mmio'. 'options' are specified
1201 in the same format described for "console=ttyS<n>"; if
1202 unspecified, the h/w is not initialized. 'uartclk' is
1203 the uart clock frequency; if unspecified, it is set
1204 to 'BASE_BAUD' * 16.
1208 Start an early, polled-mode console on a pl011 serial
1209 port at the specified address. The pl011 serial port
1210 must already be setup and configured. Options are not
1211 yet supported. If 'mmio32' is specified, then only
1212 the driver will use only 32-bit accessors to read/write
1213 the device registers.
1216 Start an early console on a litex serial port at the
1217 specified address. The serial port must already be
1218 setup and configured. Options are not yet supported.
1221 Start an early, polled-mode console on a meson serial
1222 port at the specified address. The serial port must
1223 already be setup and configured. Options are not yet
1227 Start an early, polled-mode console on an msm serial
1228 port at the specified address. The serial port
1229 must already be setup and configured. Options are not
1232 msm_serial_dm,<addr>
1233 Start an early, polled-mode console on an msm serial
1234 dm port at the specified address. The serial port
1235 must already be setup and configured. Options are not
1239 Start an early, polled-mode console on a serial port
1240 of an Actions Semi SoC, such as S500 or S900, at the
1241 specified address. The serial port must already be
1242 setup and configured. Options are not yet supported.
1245 Start an early, polled-mode console on a serial port
1246 of an RDA Micro SoC, such as RDA8810PL, at the
1247 specified address. The serial port must already be
1248 setup and configured. Options are not yet supported.
1251 Use RISC-V SBI (Supervisor Binary Interface) for early
1254 smh Use ARM semihosting calls for early console.
1262 Use early console provided by serial driver available
1263 on Samsung SoCs, requires selecting proper type and
1264 a correct base address of the selected UART port. The
1265 serial port must already be setup and configured.
1266 Options are not yet supported.
1269 Start an early, polled-mode console on a lantiq serial
1270 (lqasc) port at the specified address. The serial port
1271 must already be setup and configured. Options are not
1276 Use early console provided by Freescale LP UART driver
1277 found on Freescale Vybrid and QorIQ LS1021A processors.
1278 A valid base address must be provided, and the serial
1279 port must already be setup and configured.
1283 Start an early, polled-mode, output-only console on the
1284 Freescale i.MX UART at the specified address. The UART
1285 must already be setup and configured.
1288 Start an early, polled-mode console on the
1289 Armada 3700 serial port at the specified
1290 address. The serial port must already be setup
1291 and configured. Options are not yet supported.
1294 Start an early, polled-mode console on a Qualcomm
1295 Generic Interface (GENI) based serial port at the
1296 specified address. The serial port must already be
1297 setup and configured. Options are not yet supported.
1300 Start an early, unaccelerated console on the EFI
1301 memory mapped framebuffer (if available). On cache
1302 coherent non-x86 systems that use system memory for
1303 the framebuffer, pass the 'ram' option so that it is
1304 mapped with the correct attributes.
1307 Use early console provided by Freescale LINFlexD UART
1308 serial driver for NXP S32V234 SoCs. A valid base
1309 address must be provided, and the serial port must
1310 already be setup and configured.
1312 earlyprintk= [X86,SH,ARM,M68k,S390]
1316 earlyprintk=serial[,ttySn[,baudrate]]
1317 earlyprintk=serial[,0x...[,baudrate]]
1318 earlyprintk=ttySn[,baudrate]
1319 earlyprintk=dbgp[debugController#]
1320 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1321 earlyprintk=xdbc[xhciController#]
1323 earlyprintk is useful when the kernel crashes before
1324 the normal console is initialized. It is not enabled by
1325 default because it has some cosmetic problems.
1327 Append ",keep" to not disable it when the real console
1330 Only one of vga, serial, or usb debug port can
1333 Currently only ttyS0 and ttyS1 may be specified by
1334 name. Other I/O ports may be explicitly specified
1335 on some architectures (x86 and arm at least) by
1336 replacing ttySn with an I/O port address, like this:
1337 earlyprintk=serial,0x1008,115200
1338 You can find the port for a given device in
1339 /proc/tty/driver/serial:
1340 2: uart:ST16650V2 port:00001008 irq:18 ...
1342 Interaction with the standard serial driver is not
1345 The VGA output is eventually overwritten by
1348 The xen option can only be used in Xen domains.
1350 The sclp output can only be used on s390.
1352 The optional "force" to "pciserial" enables use of a
1353 PCI device even when its classcode is not of the
1356 edac_report= [HW,EDAC] Control how to report EDAC event
1357 Format: {"on" | "off" | "force"}
1358 on: enable EDAC to report H/W event. May be overridden
1359 by other higher priority error reporting module.
1360 off: disable H/W event reporting through EDAC.
1361 force: enforce the use of EDAC to report H/W event.
1365 Format: {"off" | "on" | "skip[mbr]"}
1368 Format: { "debug", "disable_early_pci_dma",
1369 "nochunk", "noruntime", "nosoftreserve",
1370 "novamap", "no_disable_early_pci_dma" }
1371 debug: enable misc debug output.
1372 disable_early_pci_dma: disable the busmaster bit on all
1373 PCI bridges while in the EFI boot stub.
1374 nochunk: disable reading files in "chunks" in the EFI
1375 boot stub, as chunking can cause problems with some
1376 firmware implementations.
1377 noruntime : disable EFI runtime services support
1378 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1379 attribute may cause the kernel to reserve the
1380 memory range for a memory mapping driver to
1381 claim. Specify efi=nosoftreserve to disable this
1382 reservation and treat the memory by its base type
1383 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1384 novamap: do not call SetVirtualAddressMap().
1385 no_disable_early_pci_dma: Leave the busmaster bit set
1386 on all PCI bridges while in the EFI boot stub
1388 efi_no_storage_paranoia [EFI; X86]
1389 Using this parameter you can use more than 50% of
1390 your efi variable storage. Use this parameter only if
1391 you are really sure that your UEFI does sane gc and
1392 fulfills the spec otherwise your board may brick.
1394 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1395 Add arbitrary attribute to specific memory range by
1396 updating original EFI memory map.
1397 Region of memory which aa attribute is added to is
1400 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1401 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1402 attribute is added to range 0x100000000-0x180000000 and
1403 0x10a0000000-0x1120000000.
1405 If efi_fake_mem=8G@9G:0x40000 is specified, the
1406 EFI_MEMORY_SP(0x40000) attribute is added to
1407 range 0x240000000-0x43fffffff.
1409 Using this parameter you can do debugging of EFI memmap
1410 related features. For example, you can do debugging of
1411 Address Range Mirroring feature even if your box
1412 doesn't support it, or mark specific memory as
1415 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1416 that is to be dynamically loaded by Linux. If there are
1417 multiple variables with the same name but with different
1418 vendor GUIDs, all of them will be loaded. See
1419 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1422 eisa_irq_edge= [PARISC,HW]
1423 See header of drivers/parisc/eisa.c.
1425 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1428 This is designed to be used in conjunction with
1429 the boot argument: earlyprintk=vga
1431 This parameter works in place of the kgdboc parameter
1432 but can only be used if the backing tty is available
1433 very early in the boot process. For early debugging
1434 via a serial port see kgdboc_earlycon instead.
1437 See comment before function elanfreq_setup() in
1438 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1440 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1441 Specifies physical address of start of kernel core
1442 image elf header and optionally the size. Generally
1443 kexec loader will pass this option to capture kernel.
1444 See Documentation/admin-guide/kdump/kdump.rst for details.
1446 enable_mtrr_cleanup [X86]
1447 The kernel tries to adjust MTRR layout from continuous
1448 to discrete, to make X server driver able to add WB
1449 entry later. This parameter enables that.
1451 enable_timer_pin_1 [X86]
1452 Enable PIN 1 of APIC timer
1453 Can be useful to work around chipset bugs
1454 (in particular on some ATI chipsets).
1455 The kernel tries to set a reasonable default.
1457 enforcing= [SELINUX] Set initial enforcing status.
1459 See security/selinux/Kconfig help text.
1460 0 -- permissive (log only, no denials).
1461 1 -- enforcing (deny and log).
1463 Value can be changed at runtime via
1464 /sys/fs/selinux/enforce.
1467 Disable Error Record Serialization Table (ERST)
1470 ether= [HW,NET] Ethernet cards parameters
1471 This option is obsoleted by the "netdev=" option, which
1472 has equivalent usage. See its documentation for details.
1476 Permit 'security.evm' to be updated regardless of
1477 current integrity status.
1479 early_page_ext [KNL] Enforces page_ext initialization to earlier
1480 stages so cover more early boot allocations.
1481 Please note that as side effect some optimizations
1482 might be disabled to achieve that (e.g. parallelized
1483 memory initialization is disabled) so the boot process
1484 might take longer, especially on systems with a lot of
1485 memory. Available with CONFIG_PAGE_EXTENSION=y.
1490 fail_make_request=[KNL]
1491 General fault injection mechanism.
1492 Format: <interval>,<probability>,<space>,<times>
1493 See also Documentation/fault-injection/.
1496 Format: { initns | none }
1497 See Documentation/admin-guide/sysctl/net.rst for
1498 fb_tunnels_only_for_init_ns
1501 See Documentation/admin-guide/blockdev/floppy.rst.
1503 force_pal_cache_flush
1504 [IA-64] Avoid check_sal_cache_flush which may hang on
1505 buggy SAL_CACHE_FLUSH implementations. Using this
1506 parameter will force ia64_sal_cache_flush to call
1507 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1510 Forcefully enable Physical Address Extension (PAE).
1511 Many Pentium M systems disable PAE but may have a
1512 functionally usable PAE implementation.
1513 Warning: use of this parameter will taint the kernel
1514 and may cause unknown problems.
1517 [FTRACE] will set and start the specified tracer
1518 as early as possible in order to facilitate early
1521 ftrace_boot_snapshot
1522 [FTRACE] On boot up, a snapshot will be taken of the
1523 ftrace ring buffer that can be read at:
1524 /sys/kernel/tracing/snapshot.
1525 This is useful if you need tracing information from kernel
1526 boot up that is likely to be overridden by user space
1527 start up functionality.
1529 Optionally, the snapshot can also be defined for a tracing
1530 instance that was created by the trace_instance= command
1533 trace_instance=foo,sched_switch ftrace_boot_snapshot=foo
1535 The above will cause the "foo" tracing instance to trigger
1536 a snapshot at the end of boot up.
1538 ftrace_dump_on_oops[=orig_cpu]
1539 [FTRACE] will dump the trace buffers on oops.
1540 If no parameter is passed, ftrace will dump
1541 buffers of all CPUs, but if you pass orig_cpu, it will
1542 dump only the buffer of the CPU that triggered the
1545 ftrace_filter=[function-list]
1546 [FTRACE] Limit the functions traced by the function
1547 tracer at boot up. function-list is a comma-separated
1548 list of functions. This list can be changed at run
1549 time by the set_ftrace_filter file in the debugfs
1552 ftrace_notrace=[function-list]
1553 [FTRACE] Do not trace the functions specified in
1554 function-list. This list can be changed at run time
1555 by the set_ftrace_notrace file in the debugfs
1558 ftrace_graph_filter=[function-list]
1559 [FTRACE] Limit the top level callers functions traced
1560 by the function graph tracer at boot up.
1561 function-list is a comma-separated list of functions
1562 that can be changed at run time by the
1563 set_graph_function file in the debugfs tracing directory.
1565 ftrace_graph_notrace=[function-list]
1566 [FTRACE] Do not trace from the functions specified in
1567 function-list. This list is a comma-separated list of
1568 functions that can be changed at run time by the
1569 set_graph_notrace file in the debugfs tracing directory.
1571 ftrace_graph_max_depth=<uint>
1572 [FTRACE] Used with the function graph tracer. This is
1573 the max depth it will trace into a function. This value
1574 can be changed at run time by the max_graph_depth file
1575 in the tracefs tracing directory. default: 0 (no limit)
1577 fw_devlink= [KNL] Create device links between consumer and supplier
1578 devices by scanning the firmware to infer the
1579 consumer/supplier relationships. This feature is
1580 especially useful when drivers are loaded as modules as
1581 it ensures proper ordering of tasks like device probing
1582 (suppliers first, then consumers), supplier boot state
1583 clean up (only after all consumers have probed),
1584 suspend/resume & runtime PM (consumers first, then
1586 Format: { off | permissive | on | rpm }
1587 off -- Don't create device links from firmware info.
1588 permissive -- Create device links from firmware info
1589 but use it only for ordering boot state clean
1590 up (sync_state() calls).
1591 on -- Create device links from firmware info and use it
1592 to enforce probe and suspend/resume ordering.
1593 rpm -- Like "on", but also use to order runtime PM.
1595 fw_devlink.strict=<bool>
1596 [KNL] Treat all inferred dependencies as mandatory
1597 dependencies. This only applies for fw_devlink=on|rpm.
1600 fw_devlink.sync_state =
1601 [KNL] When all devices that could probe have finished
1602 probing, this parameter controls what to do with
1603 devices that haven't yet received their sync_state()
1605 Format: { strict | timeout }
1606 strict -- Default. Continue waiting on consumers to
1608 timeout -- Give up waiting on consumers and call
1609 sync_state() on any devices that haven't yet
1610 received their sync_state() calls after
1611 deferred_probe_timeout has expired or by
1612 late_initcall() if !CONFIG_MODULES.
1615 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1616 support via parallel port (up to 5 devices per port)
1617 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1618 See also Documentation/input/devices/joystick-parport.rst
1622 gart_fix_e820= [X86-64] disable the fix e820 for K8 GART
1626 gather_data_sampling=
1627 [X86,INTEL] Control the Gather Data Sampling (GDS)
1630 Gather Data Sampling is a hardware vulnerability which
1631 allows unprivileged speculative access to data which was
1632 previously stored in vector registers.
1634 This issue is mitigated by default in updated microcode.
1635 The mitigation may have a performance impact but can be
1636 disabled. On systems without the microcode mitigation
1637 disabling AVX serves as a mitigation.
1639 force: Disable AVX to mitigate systems without
1640 microcode mitigation. No effect if the microcode
1641 mitigation is present. Known to cause crashes in
1642 userspace with buggy AVX enumeration.
1644 off: Disable GDS mitigation.
1646 gcov_persist= [GCOV] When non-zero (default), profiling data for
1647 kernel modules is saved and remains accessible via
1648 debugfs, even when the module is unloaded/reloaded.
1649 When zero, profiling data is discarded and associated
1650 debugfs files are removed at module unload time.
1652 goldfish [X86] Enable the goldfish android emulator platform.
1653 Don't use this when you are not running on the
1656 gpio-mockup.gpio_mockup_ranges
1657 [HW] Sets the ranges of gpiochip of for this device.
1658 Format: <start1>,<end1>,<start2>,<end2>...
1659 gpio-mockup.gpio_mockup_named_lines
1660 [HW] Let the driver know GPIO lines should be named.
1662 gpt [EFI] Forces disk with valid GPT signature but
1663 invalid Protective MBR to be treated as GPT. If the
1664 primary GPT is corrupted, it enables the backup/alternate
1665 GPT to be used instead.
1667 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1668 the "Enable 0" bit of the configuration register.
1671 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1672 the "Enable 0" bit of the configuration register.
1675 grcan.select= [HW] Select which physical interface to use.
1678 grcan.txsize= [HW] Sets the size of the tx buffer.
1679 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1681 grcan.rxsize= [HW] Sets the size of the rx buffer.
1682 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1686 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
1687 hardening is enabled for this boot. Hardened
1688 usercopy checking is used to protect the kernel
1689 from reading or writing beyond known memory
1690 allocation boundaries as a proactive defense
1691 against bounds-checking flaws in the kernel's
1692 copy_to_user()/copy_from_user() interface.
1693 on Perform hardened usercopy checks (default).
1694 off Disable hardened usercopy checks.
1696 hardlockup_all_cpu_backtrace=
1697 [KNL] Should the hard-lockup detector generate
1698 backtraces on all cpus.
1701 hashdist= [KNL,NUMA] Large hashes allocated during boot
1702 are distributed across NUMA nodes. Defaults on
1703 for 64-bit NUMA, off otherwise.
1704 Format: 0 | 1 (for off | on)
1706 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1708 hd= [EIDE] (E)IDE hard drive subsystem geometry
1709 Format: <cyl>,<head>,<sect>
1712 Disable Hardware Error Source Table (HEST) support;
1713 corresponding firmware-first mode error processing
1714 logic will be disabled.
1716 hibernate= [HIBERNATION]
1717 noresume Don't check if there's a hibernation image
1718 present during boot.
1719 nocompress Don't compress/decompress hibernation images.
1720 no Disable hibernation and resume.
1721 protect_image Turn on image protection during restoration
1722 (that will set all pages holding image data
1723 during restoration read-only).
1725 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1726 size of <nn>. This works even on boxes that have no
1727 highmem otherwise. This also works to reduce highmem
1728 size on bigger boxes.
1730 highres= [KNL] Enable/disable high resolution timer mode.
1731 Valid parameters: "on", "off"
1736 hostname= [KNL] Set the hostname (aka UTS nodename).
1738 This allows setting the system's hostname during early
1739 startup. This sets the name returned by gethostname.
1740 Using this parameter to set the hostname makes it
1741 possible to ensure the hostname is correctly set before
1742 any userspace processes run, avoiding the possibility
1743 that a process may call gethostname before the hostname
1744 has been explicitly set, resulting in the calling
1745 process getting an incorrect result. The string must
1746 not exceed the maximum allowed hostname length (usually
1747 64 characters) and will be truncated otherwise.
1749 hpet= [X86-32,HPET] option to control HPET usage
1750 Format: { enable (default) | disable | force |
1752 disable: disable HPET and use PIT instead
1753 force: allow force enabled of undocumented chips (ICH4,
1755 verbose: show contents of HPET registers during setup
1757 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1758 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1760 hugepages= [HW] Number of HugeTLB pages to allocate at boot.
1761 If this follows hugepagesz (below), it specifies
1762 the number of pages of hugepagesz to be allocated.
1763 If this is the first HugeTLB parameter on the command
1764 line, it specifies the number of pages to allocate for
1765 the default huge page size. If using node format, the
1766 number of pages to allocate per-node can be specified.
1767 See also Documentation/admin-guide/mm/hugetlbpage.rst.
1768 Format: <integer> or (node format)
1769 <node>:<integer>[,<node>:<integer>]
1772 [HW] The size of the HugeTLB pages. This is used in
1773 conjunction with hugepages (above) to allocate huge
1774 pages of a specific size at boot. The pair
1775 hugepagesz=X hugepages=Y can be specified once for
1776 each supported huge page size. Huge page sizes are
1777 architecture dependent. See also
1778 Documentation/admin-guide/mm/hugetlbpage.rst.
1781 hugetlb_cma= [HW,CMA] The size of a CMA area used for allocation
1782 of gigantic hugepages. Or using node format, the size
1783 of a CMA area per node can be specified.
1784 Format: nn[KMGTPE] or (node format)
1785 <node>:nn[KMGTPE][,<node>:nn[KMGTPE]]
1787 Reserve a CMA area of given size and allocate gigantic
1788 hugepages using the CMA allocator. If enabled, the
1789 boot-time allocation of gigantic hugepages is skipped.
1791 hugetlb_free_vmemmap=
1792 [KNL] Requires CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP
1794 Control if HugeTLB Vmemmap Optimization (HVO) is enabled.
1795 Allows heavy hugetlb users to free up some more
1796 memory (7 * PAGE_SIZE for each 2MB hugetlb page).
1797 Format: { on | off (default) }
1802 Built with CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP_DEFAULT_ON=y,
1805 Note that the vmemmap pages may be allocated from the added
1806 memory block itself when memory_hotplug.memmap_on_memory is
1807 enabled, those vmemmap pages cannot be optimized even if this
1808 feature is enabled. Other vmemmap pages not allocated from
1809 the added memory block itself do not be affected.
1812 [KNL] Should the hung task detector generate panics.
1815 A value of 1 instructs the kernel to panic when a
1816 hung task is detected. The default value is controlled
1817 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1818 option. The value selected by this boot parameter can
1819 be changed later by the kernel.hung_task_panic sysctl.
1821 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1822 terminal devices. Valid values: 0..8
1823 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1824 If specified, z/VM IUCV HVC accepts connections
1825 from listed z/VM user IDs only.
1827 hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations
1828 which allow the hypervisor to 'idle' the
1829 guest on lock contention.
1831 i2c_bus= [HW] Override the default board specific I2C bus speed
1832 or register an additional I2C bus that is not
1833 registered from board initialization code.
1837 i8042.debug [HW] Toggle i8042 debug mode
1838 i8042.unmask_kbd_data
1839 [HW] Enable printing of interrupt data from the KBD port
1840 (disabled by default, and as a pre-condition
1841 requires that i8042.debug=1 be enabled)
1842 i8042.direct [HW] Put keyboard port into non-translated mode
1843 i8042.dumbkbd [HW] Pretend that controller can only read data from
1844 keyboard and cannot control its state
1845 (Don't attempt to blink the leds)
1846 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1847 i8042.nokbd [HW] Don't check/create keyboard port
1848 i8042.noloop [HW] Disable the AUX Loopback command while probing
1850 i8042.nomux [HW] Don't check presence of an active multiplexing
1852 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1854 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1855 i8042.reset [HW] Reset the controller during init, cleanup and
1856 suspend-to-ram transitions, only during s2r
1857 transitions, or never reset
1858 Format: { 1 | Y | y | 0 | N | n }
1859 1, Y, y: always reset controller
1860 0, N, n: don't ever reset controller
1861 Default: only on s2r transitions on x86; most other
1862 architectures force reset to be always executed
1863 i8042.unlock [HW] Unlock (ignore) the keylock
1864 i8042.kbdreset [HW] Reset device connected to KBD port
1866 [HW] Allow deferred probing upon i8042 probe errors
1870 i915.invert_brightness=
1871 [DRM] Invert the sense of the variable that is used to
1872 set the brightness of the panel backlight. Normally a
1873 brightness value of 0 indicates backlight switched off,
1874 and the maximum of the brightness value sets the backlight
1875 to maximum brightness. If this parameter is set to 0
1876 (default) and the machine requires it, or this parameter
1877 is set to 1, a brightness value of 0 sets the backlight
1878 to maximum brightness, and the maximum of the brightness
1879 value switches the backlight off.
1880 -1 -- never invert brightness
1881 0 -- machine default
1882 1 -- force brightness inversion
1885 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1889 Format: idle=poll, idle=halt, idle=nomwait
1890 Poll forces a polling idle loop that can slightly
1891 improve the performance of waking up a idle CPU, but
1892 will use a lot of power and make the system run hot.
1894 idle=halt: Halt is forced to be used for CPU idle.
1895 In such case C2/C3 won't be used again.
1896 idle=nomwait: Disable mwait for CPU C-states
1900 Allow force disabling of Shared Virtual Memory (SVA)
1901 support for the idxd driver. By default it is set to
1904 idxd.tc_override= [HW]
1906 Allow override of default traffic class configuration
1907 for the device. By default it is set to false (0).
1909 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1910 Format: { strict | legacy | 2008 | relaxed }
1913 Choose which programs will be accepted for execution
1914 based on the IEEE 754 NaN encoding(s) supported by
1915 the FPU and the NaN encoding requested with the value
1916 of an ELF file header flag individually set by each
1917 binary. Hardware implementations are permitted to
1918 support either or both of the legacy and the 2008 NaN
1921 Available settings are as follows:
1922 strict accept binaries that request a NaN encoding
1923 supported by the FPU
1924 legacy only accept legacy-NaN binaries, if supported
1926 2008 only accept 2008-NaN binaries, if supported
1928 relaxed accept any binaries regardless of whether
1929 supported by the FPU
1931 The FPU emulator is always able to support both NaN
1932 encodings, so if no FPU hardware is present or it has
1933 been disabled with 'nofpu', then the settings of
1934 'legacy' and '2008' strap the emulator accordingly,
1935 'relaxed' straps the emulator for both legacy-NaN and
1936 2008-NaN, whereas 'strict' enables legacy-NaN only on
1937 legacy processors and both NaN encodings on MIPS32 or
1940 The setting for ABS.fmt/NEG.fmt instruction execution
1941 mode generally follows that for the NaN encoding,
1942 except where unsupported by hardware.
1944 ignore_loglevel [KNL]
1945 Ignore loglevel setting - this will print /all/
1946 kernel messages to the console. Useful for debugging.
1947 We also add it as printk module parameter, so users
1948 could change it dynamically, usually by
1949 /sys/module/printk/parameters/ignore_loglevel.
1952 Ignore RLIMIT_DATA setting for data mappings,
1953 print warning at first misuse. Can be changed via
1954 /sys/module/kernel/parameters/ignore_rlimit_data.
1956 ihash_entries= [KNL]
1957 Set number of hash buckets for inode cache.
1959 ima_appraise= [IMA] appraise integrity measurements
1960 Format: { "off" | "enforce" | "fix" | "log" }
1963 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
1964 The builtin appraise policy appraises all files
1967 ima_canonical_fmt [IMA]
1968 Use the canonical format for the binary runtime
1969 measurements, instead of host native format.
1972 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1976 The list of supported hash algorithms is defined
1977 in crypto/hash_info.h.
1980 The builtin policies to load during IMA setup.
1981 Format: "tcb | appraise_tcb | secure_boot |
1982 fail_securely | critical_data"
1984 The "tcb" policy measures all programs exec'd, files
1985 mmap'd for exec, and all files opened with the read
1986 mode bit set by either the effective uid (euid=0) or
1989 The "appraise_tcb" policy appraises the integrity of
1990 all files owned by root.
1992 The "secure_boot" policy appraises the integrity
1993 of files (eg. kexec kernel image, kernel modules,
1994 firmware, policy, etc) based on file signatures.
1996 The "fail_securely" policy forces file signature
1997 verification failure also on privileged mounted
1998 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
2001 The "critical_data" policy measures kernel integrity
2004 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
2005 Load a policy which meets the needs of the Trusted
2006 Computing Base. This means IMA will measure all
2007 programs exec'd, files mmap'd for exec, and all files
2008 opened for read by uid=0.
2011 Select one of defined IMA measurements template formats.
2012 Formats: { "ima" | "ima-ng" | "ima-ngv2" | "ima-sig" |
2017 [IMA] Define a custom template format.
2018 Format: { "field1|...|fieldN" }
2020 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
2021 Format: <min_file_size>
2022 Set the minimal file size for using asynchronous hash.
2023 If left unspecified, ahash usage is disabled.
2025 ahash performance varies for different data sizes on
2026 different crypto accelerators. This option can be used
2027 to achieve the best performance for a particular HW.
2029 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
2031 Set hashing buffer size. Default: 4k.
2033 ahash performance varies for different chunk sizes on
2034 different crypto accelerators. This option can be used
2035 to achieve best performance for particular HW.
2039 Run specified binary instead of /sbin/init as init
2042 initcall_debug [KNL] Trace initcalls as they are executed. Useful
2043 for working out where the kernel is dying during
2046 initcall_blacklist= [KNL] Do not execute a comma-separated list of
2047 initcall functions. Useful for debugging built-in
2048 modules and initcalls.
2050 initramfs_async= [KNL]
2053 This parameter controls whether the initramfs
2054 image is unpacked asynchronously, concurrently
2055 with devices being probed and
2056 initialized. This should normally just work,
2057 but as a debugging aid, one can get the
2058 historical behaviour of the initramfs
2059 unpacking being completed before device_ and
2062 initrd= [BOOT] Specify the location of the initial ramdisk
2064 initrdmem= [KNL] Specify a physical address and size from which to
2065 load the initrd. If an initrd is compiled in or
2066 specified in the bootparams, it takes priority over this
2068 Format: ss[KMG],nn[KMG]
2071 init_on_alloc= [MM] Fill newly allocated pages and heap objects with
2074 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
2076 init_on_free= [MM] Fill freed pages and heap objects with zeroes.
2078 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
2080 init_pkru= [X86] Specify the default memory protection keys rights
2081 register contents for all processes. 0x55555554 by
2082 default (disallow access to all but pkey 0). Can
2083 override in debugfs after boot.
2085 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
2088 int_pln_enable [X86] Enable power limit notification interrupt
2090 integrity_audit=[IMA]
2091 Format: { "0" | "1" }
2092 0 -- basic integrity auditing messages. (Default)
2093 1 -- additional integrity auditing messages.
2095 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
2097 Enable intel iommu driver.
2099 Disable intel iommu driver.
2100 igfx_off [Default Off]
2101 By default, gfx is mapped as normal device. If a gfx
2102 device has a dedicated DMAR unit, the DMAR unit is
2103 bypassed by not enabling DMAR with this option. In
2104 this case, gfx device will use physical address for
2106 strict [Default Off]
2107 Deprecated, equivalent to iommu.strict=1.
2108 sp_off [Default Off]
2109 By default, super page will be supported if Intel IOMMU
2110 has the capability. With this option, super page will
2113 Enable the Intel IOMMU scalable mode if the hardware
2114 advertises that it has support for the scalable mode
2117 Disallow use of the Intel IOMMU scalable mode.
2118 tboot_noforce [Default Off]
2119 Do not force the Intel IOMMU enabled under tboot.
2120 By default, tboot will force Intel IOMMU on, which
2121 could harm performance of some high-throughput
2122 devices like 40GBit network cards, even if identity
2124 Note that using this option lowers the security
2125 provided by tboot because it makes the system
2126 vulnerable to DMA attacks.
2128 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
2129 0 disables intel_idle and fall back on acpi_idle.
2130 1 to 9 specify maximum depth of C-state.
2134 Do not enable intel_pstate as the default
2135 scaling driver for the supported processors
2137 Use intel_pstate driver to bypass the scaling
2138 governors layer of cpufreq and provides it own
2139 algorithms for p-state selection. There are two
2140 P-state selection algorithms provided by
2141 intel_pstate in the active mode: powersave and
2142 performance. The way they both operate depends
2143 on whether or not the hardware managed P-states
2144 (HWP) feature has been enabled in the processor
2145 and possibly on the processor model.
2147 Use intel_pstate as a scaling driver, but configure it
2148 to work with generic cpufreq governors (instead of
2149 enabling its internal governor). This mode cannot be
2150 used along with the hardware-managed P-states (HWP)
2153 Enable intel_pstate on systems that prohibit it by default
2154 in favor of acpi-cpufreq. Forcing the intel_pstate driver
2155 instead of acpi-cpufreq may disable platform features, such
2156 as thermal controls and power capping, that rely on ACPI
2157 P-States information being indicated to OSPM and therefore
2158 should be used with caution. This option does not work with
2159 processors that aren't supported by the intel_pstate driver
2160 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
2162 Do not enable hardware P state control (HWP)
2165 Only load intel_pstate on systems which support
2166 hardware P state control (HWP) if available.
2168 Enforce ACPI _PPC performance limits. If the Fixed ACPI
2169 Description Table, specifies preferred power management
2170 profile as "Enterprise Server" or "Performance Server",
2171 then this feature is turned on by default.
2173 Allow per-logical-CPU P-State performance control limits using
2174 cpufreq sysfs interface
2176 intremap= [X86-64, Intel-IOMMU]
2177 on enable Interrupt Remapping (default)
2178 off disable Interrupt Remapping
2179 nosid disable Source ID checking
2181 BIOS x2APIC opt-out request will be ignored
2182 nopost disable Interrupt Posting
2184 iomem= Disable strict checking of access to MMIO memory
2185 strict regions from userspace.
2200 nobypass [PPC/POWERNV]
2201 Disable IOMMU bypass, using IOMMU for PCI devices.
2203 iommu.forcedac= [ARM64, X86] Control IOVA allocation for PCI devices.
2204 Format: { "0" | "1" }
2205 0 - Try to allocate a 32-bit DMA address first, before
2206 falling back to the full range if needed.
2207 1 - Allocate directly from the full usable range,
2208 forcing Dual Address Cycle for PCI cards supporting
2209 greater than 32-bit addressing.
2211 iommu.strict= [ARM64, X86] Configure TLB invalidation behaviour
2212 Format: { "0" | "1" }
2214 Request that DMA unmap operations use deferred
2215 invalidation of hardware TLBs, for increased
2216 throughput at the cost of reduced device isolation.
2217 Will fall back to strict mode if not supported by
2218 the relevant IOMMU driver.
2220 DMA unmap operations invalidate IOMMU hardware TLBs
2222 unset - Use value of CONFIG_IOMMU_DEFAULT_DMA_{LAZY,STRICT}.
2223 Note: on x86, strict mode specified via one of the
2224 legacy driver-specific options takes precedence.
2227 [ARM64, X86] Configure DMA to bypass the IOMMU by default.
2228 Format: { "0" | "1" }
2229 0 - Use IOMMU translation for DMA.
2230 1 - Bypass the IOMMU for DMA.
2231 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
2233 io7= [HW] IO7 for Marvel-based Alpha systems
2234 See comment before marvel_specify_io7 in
2235 arch/alpha/kernel/core_marvel.c.
2237 io_delay= [X86] I/O delay method
2239 Standard port 0x80 based delay
2241 Alternate port 0xed based delay (needed on some systems)
2243 Simple two microseconds delay
2248 See Documentation/admin-guide/nfs/nfsroot.rst.
2250 ipcmni_extend [KNL] Extend the maximum number of unique System V
2251 IPC identifiers from 32,768 to 16,777,216.
2253 irqaffinity= [SMP] Set the default irq affinity mask
2254 The argument is a cpu list, as described above.
2256 irqchip.gicv2_force_probe=
2259 Force the kernel to look for the second 4kB page
2260 of a GICv2 controller even if the memory range
2261 exposed by the device tree is too small.
2263 irqchip.gicv3_nolpi=
2265 Force the kernel to ignore the availability of
2266 LPIs (and by consequence ITSs). Intended for system
2267 that use the kernel as a bootloader, and thus want
2268 to let secondary kernels in charge of setting up
2271 irqchip.gicv3_pseudo_nmi= [ARM64]
2272 Enables support for pseudo-NMIs in the kernel. This
2273 requires the kernel to be built with
2274 CONFIG_ARM64_PSEUDO_NMI.
2277 When an interrupt is not handled search all handlers
2278 for it. Intended to get systems with badly broken
2282 When an interrupt is not handled search all handlers
2283 for it. Also check all handlers each timer
2284 interrupt. Intended to get systems with badly broken
2288 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
2290 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
2291 [Deprecated - use cpusets instead]
2292 Format: [flag-list,]<cpu-list>
2294 Specify one or more CPUs to isolate from disturbances
2295 specified in the flag list (default: domain):
2298 Disable the tick when a single task runs.
2300 A residual 1Hz tick is offloaded to workqueues, which you
2301 need to affine to housekeeping through the global
2302 workqueue's affinity configured via the
2303 /sys/devices/virtual/workqueue/cpumask sysfs file, or
2304 by using the 'domain' flag described below.
2306 NOTE: by default the global workqueue runs on all CPUs,
2307 so to protect individual CPUs the 'cpumask' file has to
2308 be configured manually after bootup.
2311 Isolate from the general SMP balancing and scheduling
2312 algorithms. Note that performing domain isolation this way
2313 is irreversible: it's not possible to bring back a CPU to
2314 the domains once isolated through isolcpus. It's strongly
2315 advised to use cpusets instead to disable scheduler load
2316 balancing through the "cpuset.sched_load_balance" file.
2317 It offers a much more flexible interface where CPUs can
2318 move in and out of an isolated set anytime.
2320 You can move a process onto or off an "isolated" CPU via
2321 the CPU affinity syscalls or cpuset.
2322 <cpu number> begins at 0 and the maximum value is
2323 "number of CPUs in system - 1".
2327 Isolate from being targeted by managed interrupts
2328 which have an interrupt mask containing isolated
2329 CPUs. The affinity of managed interrupts is
2330 handled by the kernel and cannot be changed via
2331 the /proc/irq/* interfaces.
2333 This isolation is best effort and only effective
2334 if the automatically assigned interrupt mask of a
2335 device queue contains isolated and housekeeping
2336 CPUs. If housekeeping CPUs are online then such
2337 interrupts are directed to the housekeeping CPU
2338 so that IO submitted on the housekeeping CPU
2339 cannot disturb the isolated CPU.
2341 If a queue's affinity mask contains only isolated
2342 CPUs then this parameter has no effect on the
2343 interrupt routing decision, though interrupts are
2344 only delivered when tasks running on those
2345 isolated CPUs submit IO. IO submitted on
2346 housekeeping CPUs has no influence on those
2349 The format of <cpu-list> is described above.
2353 ivrs_ioapic [HW,X86-64]
2354 Provide an override to the IOAPIC-ID<->DEVICE-ID
2355 mapping provided in the IVRS ACPI table.
2356 By default, PCI segment is 0, and can be omitted.
2358 For example, to map IOAPIC-ID decimal 10 to
2359 PCI segment 0x1 and PCI device 00:14.0,
2360 write the parameter as:
2361 ivrs_ioapic=10@0001:00:14.0
2364 * To map IOAPIC-ID decimal 10 to PCI device 00:14.0
2365 write the parameter as:
2366 ivrs_ioapic[10]=00:14.0
2367 * To map IOAPIC-ID decimal 10 to PCI segment 0x1 and
2368 PCI device 00:14.0 write the parameter as:
2369 ivrs_ioapic[10]=0001:00:14.0
2371 ivrs_hpet [HW,X86-64]
2372 Provide an override to the HPET-ID<->DEVICE-ID
2373 mapping provided in the IVRS ACPI table.
2374 By default, PCI segment is 0, and can be omitted.
2376 For example, to map HPET-ID decimal 10 to
2377 PCI segment 0x1 and PCI device 00:14.0,
2378 write the parameter as:
2379 ivrs_hpet=10@0001:00:14.0
2382 * To map HPET-ID decimal 0 to PCI device 00:14.0
2383 write the parameter as:
2384 ivrs_hpet[0]=00:14.0
2385 * To map HPET-ID decimal 10 to PCI segment 0x1 and
2386 PCI device 00:14.0 write the parameter as:
2387 ivrs_ioapic[10]=0001:00:14.0
2389 ivrs_acpihid [HW,X86-64]
2390 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2391 mapping provided in the IVRS ACPI table.
2392 By default, PCI segment is 0, and can be omitted.
2394 For example, to map UART-HID:UID AMD0020:0 to
2395 PCI segment 0x1 and PCI device ID 00:14.5,
2396 write the parameter as:
2397 ivrs_acpihid=AMD0020:0@0001:00:14.5
2400 * To map UART-HID:UID AMD0020:0 to PCI segment is 0,
2401 PCI device ID 00:14.5, write the parameter as:
2402 ivrs_acpihid[00:14.5]=AMD0020:0
2403 * To map UART-HID:UID AMD0020:0 to PCI segment 0x1 and
2404 PCI device ID 00:14.5, write the parameter as:
2405 ivrs_acpihid[0001:00:14.5]=AMD0020:0
2407 js= [HW,JOY] Analog joystick
2408 See Documentation/input/joydev/joystick.rst.
2411 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2412 report on every invalid memory access. Without this
2413 parameter KASAN will print report only for the first
2417 Do not unregister boot console at start. This is only
2418 useful for debugging when something happens in the window
2419 between unregistering the boot console and initializing
2424 kernelcore= [KNL,X86,IA-64,PPC]
2425 Format: nn[KMGTPE] | nn% | "mirror"
2426 This parameter specifies the amount of memory usable by
2427 the kernel for non-movable allocations. The requested
2428 amount is spread evenly throughout all nodes in the
2429 system as ZONE_NORMAL. The remaining memory is used for
2430 movable memory in its own zone, ZONE_MOVABLE. In the
2431 event, a node is too small to have both ZONE_NORMAL and
2432 ZONE_MOVABLE, kernelcore memory will take priority and
2433 other nodes will have a larger ZONE_MOVABLE.
2435 ZONE_MOVABLE is used for the allocation of pages that
2436 may be reclaimed or moved by the page migration
2437 subsystem. Note that allocations like PTEs-from-HighMem
2438 still use the HighMem zone if it exists, and the Normal
2439 zone if it does not.
2441 It is possible to specify the exact amount of memory in
2442 the form of "nn[KMGTPE]", a percentage of total system
2443 memory in the form of "nn%", or "mirror". If "mirror"
2444 option is specified, mirrored (reliable) memory is used
2445 for non-movable allocations and remaining memory is used
2446 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2447 are exclusive, so you cannot specify multiple forms.
2449 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
2450 Format: <Controller#>[,poll interval]
2451 The controller # is the number of the ehci usb debug
2452 port as it is probed via PCI. The poll interval is
2453 optional and is the number seconds in between
2454 each poll cycle to the debug port in case you need
2455 the functionality for interrupting the kernel with
2456 gdb or control-c on the dbgp connection. When
2457 not using this parameter you use sysrq-g to break into
2458 the kernel debugger.
2460 kgdboc= [KGDB,HW] kgdb over consoles.
2461 Requires a tty driver that supports console polling,
2462 or a supported polling keyboard driver (non-usb).
2463 Serial only format: <serial_device>[,baud]
2464 keyboard only format: kbd
2465 keyboard and serial format: kbd,<serial_device>[,baud]
2466 Optional Kernel mode setting:
2467 kms, kbd format: kms,kbd
2468 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2470 kgdboc_earlycon= [KGDB,HW]
2471 If the boot console provides the ability to read
2472 characters and can work in polling mode, you can use
2473 this parameter to tell kgdb to use it as a backend
2474 until the normal console is registered. Intended to
2475 be used together with the kgdboc parameter which
2476 specifies the normal console to transition to.
2478 The name of the early console should be specified
2479 as the value of this parameter. Note that the name of
2480 the early console might be different than the tty
2481 name passed to kgdboc. It's OK to leave the value
2482 blank and the first boot console that implements
2483 read() will be picked.
2485 kgdbwait [KGDB] Stop kernel execution and enter the
2486 kernel debugger at the earliest opportunity.
2488 kmac= [MIPS] Korina ethernet MAC address.
2489 Configure the RouterBoard 532 series on-chip
2490 Ethernet adapter MAC address.
2492 kmemleak= [KNL] Boot-time kmemleak enable/disable
2493 Valid arguments: on, off
2495 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2498 kprobe_event=[probe-list]
2499 [FTRACE] Add kprobe events and enable at boot time.
2500 The probe-list is a semicolon delimited list of probe
2501 definitions. Each definition is same as kprobe_events
2502 interface, but the parameters are comma delimited.
2503 For example, to add a kprobe event on vfs_read with
2504 arg1 and arg2, add to the command line;
2506 kprobe_event=p,vfs_read,$arg1,$arg2
2508 See also Documentation/trace/kprobetrace.rst "Kernel
2509 Boot Parameter" section.
2511 kpti= [ARM64] Control page table isolation of user
2512 and kernel address spaces.
2513 Default: enabled on cores which need mitigation.
2517 kunit.enable= [KUNIT] Enable executing KUnit tests. Requires
2518 CONFIG_KUNIT to be set to be fully enabled. The
2519 default value can be overridden via
2520 KUNIT_DEFAULT_ENABLED.
2521 Default is 1 (enabled)
2523 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2524 Default is 0 (don't ignore, but inject #GP)
2526 kvm.eager_page_split=
2527 [KVM,X86] Controls whether or not KVM will try to
2528 proactively split all huge pages during dirty logging.
2529 Eager page splitting reduces interruptions to vCPU
2530 execution by eliminating the write-protection faults
2531 and MMU lock contention that would otherwise be
2532 required to split huge pages lazily.
2534 VM workloads that rarely perform writes or that write
2535 only to a small region of VM memory may benefit from
2536 disabling eager page splitting to allow huge pages to
2537 still be used for reads.
2539 The behavior of eager page splitting depends on whether
2540 KVM_DIRTY_LOG_INITIALLY_SET is enabled or disabled. If
2541 disabled, all huge pages in a memslot will be eagerly
2542 split when dirty logging is enabled on that memslot. If
2543 enabled, eager page splitting will be performed during
2544 the KVM_CLEAR_DIRTY ioctl, and only for the pages being
2547 Eager page splitting is only supported when kvm.tdp_mmu=Y.
2551 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2552 Default is false (don't support).
2555 [KVM] Controls the software workaround for the
2556 X86_BUG_ITLB_MULTIHIT bug.
2557 force : Always deploy workaround.
2558 off : Never deploy workaround.
2559 auto : Deploy workaround based on the presence of
2560 X86_BUG_ITLB_MULTIHIT.
2564 If the software workaround is enabled for the host,
2565 guests do need not to enable it for nested guests.
2567 kvm.nx_huge_pages_recovery_ratio=
2568 [KVM] Controls how many 4KiB pages are periodically zapped
2569 back to huge pages. 0 disables the recovery, otherwise if
2570 the value is N KVM will zap 1/Nth of the 4KiB pages every
2571 period (see below). The default is 60.
2573 kvm.nx_huge_pages_recovery_period_ms=
2574 [KVM] Controls the time period at which KVM zaps 4KiB pages
2575 back to huge pages. If the value is a non-zero N, KVM will
2576 zap a portion (see ratio above) of the pages every N msecs.
2577 If the value is 0 (the default), KVM will pick a period based
2578 on the ratio, such that a page is zapped after 1 hour on average.
2580 kvm-amd.nested= [KVM,AMD] Control nested virtualization feature in
2581 KVM/SVM. Default is 1 (enabled).
2583 kvm-amd.npt= [KVM,AMD] Control KVM's use of Nested Page Tables,
2584 a.k.a. Two-Dimensional Page Tables. Default is 1
2585 (enabled). Disable by KVM if hardware lacks support
2589 [KVM,ARM] Select one of KVM/arm64's modes of operation.
2591 none: Forcefully disable KVM.
2593 nvhe: Standard nVHE-based mode, without support for
2596 protected: nVHE-based mode with support for guests whose
2597 state is kept private from the host.
2599 nested: VHE-based mode with support for nested
2600 virtualization. Requires at least ARMv8.3
2603 Defaults to VHE/nVHE based on hardware support. Setting
2604 mode to "protected" will disable kexec and hibernation
2605 for the host. "nested" is experimental and should be
2606 used with extreme caution.
2608 kvm-arm.vgic_v3_group0_trap=
2609 [KVM,ARM] Trap guest accesses to GICv3 group-0
2612 kvm-arm.vgic_v3_group1_trap=
2613 [KVM,ARM] Trap guest accesses to GICv3 group-1
2616 kvm-arm.vgic_v3_common_trap=
2617 [KVM,ARM] Trap guest accesses to GICv3 common
2620 kvm-arm.vgic_v4_enable=
2621 [KVM,ARM] Allow use of GICv4 for direct injection of
2624 kvm_cma_resv_ratio=n [PPC]
2625 Reserves given percentage from system memory area for
2626 contiguous memory allocation for KVM hash pagetable
2628 By default it reserves 5% of total system memory.
2632 kvm-intel.ept= [KVM,Intel] Control KVM's use of Extended Page Tables,
2633 a.k.a. Two-Dimensional Page Tables. Default is 1
2634 (enabled). Disable by KVM if hardware lacks support
2637 kvm-intel.emulate_invalid_guest_state=
2638 [KVM,Intel] Control whether to emulate invalid guest
2639 state. Ignored if kvm-intel.enable_unrestricted_guest=1,
2640 as guest state is never invalid for unrestricted
2641 guests. This param doesn't apply to nested guests (L2),
2642 as KVM never emulates invalid L2 guest state.
2643 Default is 1 (enabled).
2645 kvm-intel.flexpriority=
2646 [KVM,Intel] Control KVM's use of FlexPriority feature
2647 (TPR shadow). Default is 1 (enabled). Disalbe by KVM if
2648 hardware lacks support for it.
2651 [KVM,Intel] Control nested virtualization feature in
2652 KVM/VMX. Default is 1 (enabled).
2654 kvm-intel.unrestricted_guest=
2655 [KVM,Intel] Control KVM's use of unrestricted guest
2656 feature (virtualized real and unpaged mode). Default
2657 is 1 (enabled). Disable by KVM if EPT is disabled or
2658 hardware lacks support for it.
2660 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2663 Valid arguments: never, cond, always
2665 always: L1D cache flush on every VMENTER.
2666 cond: Flush L1D on VMENTER only when the code between
2667 VMEXIT and VMENTER can leak host memory.
2668 never: Disables the mitigation
2670 Default is cond (do L1 cache flush in specific instances)
2672 kvm-intel.vpid= [KVM,Intel] Control KVM's use of Virtual Processor
2673 Identification feature (tagged TLBs). Default is 1
2674 (enabled). Disable by KVM if hardware lacks support
2677 l1d_flush= [X86,INTEL]
2678 Control mitigation for L1D based snooping vulnerability.
2680 Certain CPUs are vulnerable to an exploit against CPU
2681 internal buffers which can forward information to a
2682 disclosure gadget under certain conditions.
2684 In vulnerable processors, the speculatively
2685 forwarded data can be used in a cache side channel
2686 attack, to access data to which the attacker does
2687 not have direct access.
2689 This parameter controls the mitigation. The
2692 on - enable the interface for the mitigation
2694 l1tf= [X86] Control mitigation of the L1TF vulnerability on
2697 The kernel PTE inversion protection is unconditionally
2698 enabled and cannot be disabled.
2701 Provides all available mitigations for the
2702 L1TF vulnerability. Disables SMT and
2703 enables all mitigations in the
2704 hypervisors, i.e. unconditional L1D flush.
2706 SMT control and L1D flush control via the
2707 sysfs interface is still possible after
2708 boot. Hypervisors will issue a warning
2709 when the first VM is started in a
2710 potentially insecure configuration,
2711 i.e. SMT enabled or L1D flush disabled.
2714 Same as 'full', but disables SMT and L1D
2715 flush runtime control. Implies the
2716 'nosmt=force' command line option.
2717 (i.e. sysfs control of SMT is disabled.)
2720 Leaves SMT enabled and enables the default
2721 hypervisor mitigation, i.e. conditional
2724 SMT control and L1D flush control via the
2725 sysfs interface is still possible after
2726 boot. Hypervisors will issue a warning
2727 when the first VM is started in a
2728 potentially insecure configuration,
2729 i.e. SMT enabled or L1D flush disabled.
2733 Disables SMT and enables the default
2734 hypervisor mitigation.
2736 SMT control and L1D flush control via the
2737 sysfs interface is still possible after
2738 boot. Hypervisors will issue a warning
2739 when the first VM is started in a
2740 potentially insecure configuration,
2741 i.e. SMT enabled or L1D flush disabled.
2744 Same as 'flush', but hypervisors will not
2745 warn when a VM is started in a potentially
2746 insecure configuration.
2749 Disables hypervisor mitigations and doesn't
2751 It also drops the swap size and available
2752 RAM limit restriction on both hypervisor and
2757 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2763 lapic [X86-32,APIC] Enable the local APIC even if BIOS
2766 lapic= [X86,APIC] Do not use TSC deadline
2767 value for LAPIC timer one-shot implementation. Default
2768 back to the programmable timer unit in the LAPIC.
2769 Format: notscdeadline
2771 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
2774 libata.dma= [LIBATA] DMA control
2775 libata.dma=0 Disable all PATA and SATA DMA
2776 libata.dma=1 PATA and SATA Disk DMA only
2777 libata.dma=2 ATAPI (CDROM) DMA only
2778 libata.dma=4 Compact Flash DMA only
2779 Combinations also work, so libata.dma=3 enables DMA
2780 for disks and CDROMs, but not CFs.
2782 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2783 libata.ignore_hpa=0 keep BIOS limits (default)
2784 libata.ignore_hpa=1 ignore limits, using full disk
2786 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2790 libata.force= [LIBATA] Force configurations. The format is a comma-
2791 separated list of "[ID:]VAL" where ID is PORT[.DEVICE].
2792 PORT and DEVICE are decimal numbers matching port, link
2793 or device. Basically, it matches the ATA ID string
2794 printed on console by libata. If the whole ID part is
2795 omitted, the last PORT and DEVICE values are used. If
2796 ID hasn't been specified yet, the configuration applies
2797 to all ports, links and devices.
2799 If only DEVICE is omitted, the parameter applies to
2800 the port and all links and devices behind it. DEVICE
2801 number of 0 either selects the first device or the
2802 first fan-out link behind PMP device. It does not
2803 select the host link. DEVICE number of 15 selects the
2804 host link and device attached to it.
2806 The VAL specifies the configuration to force. As long
2807 as there is no ambiguity, shortcut notation is allowed.
2808 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2809 The following configurations can be forced.
2811 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2812 Any ID with matching PORT is used.
2814 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2816 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2817 udma[/][16,25,33,44,66,100,133] notation is also
2820 * nohrst, nosrst, norst: suppress hard, soft and both
2823 * rstonce: only attempt one reset during hot-unplug
2826 * [no]dbdelay: Enable or disable the extra 200ms delay
2827 before debouncing a link PHY and device presence
2830 * [no]ncq: Turn on or off NCQ.
2832 * [no]ncqtrim: Enable or disable queued DSM TRIM.
2834 * [no]ncqati: Enable or disable NCQ trim on ATI chipset.
2836 * [no]trim: Enable or disable (unqueued) TRIM.
2838 * trim_zero: Indicate that TRIM command zeroes data.
2840 * max_trim_128m: Set 128M maximum trim size limit.
2842 * [no]dma: Turn on or off DMA transfers.
2844 * atapi_dmadir: Enable ATAPI DMADIR bridge support.
2846 * atapi_mod16_dma: Enable the use of ATAPI DMA for
2847 commands that are not a multiple of 16 bytes.
2849 * [no]dmalog: Enable or disable the use of the
2850 READ LOG DMA EXT command to access logs.
2852 * [no]iddevlog: Enable or disable access to the
2853 identify device data log.
2855 * [no]logdir: Enable or disable access to the general
2856 purpose log directory.
2858 * max_sec_128: Set transfer size limit to 128 sectors.
2860 * max_sec_1024: Set or clear transfer size limit to
2863 * max_sec_lba48: Set or clear transfer size limit to
2866 * [no]lpm: Enable or disable link power management.
2868 * [no]setxfer: Indicate if transfer speed mode setting
2871 * [no]fua: Disable or enable FUA (Force Unit Access)
2872 support for devices supporting this feature.
2874 * dump_id: Dump IDENTIFY data.
2876 * disable: Disable this device.
2878 If there are multiple matching configurations changing
2879 the same attribute, the last one is used.
2881 load_ramdisk= [RAM] [Deprecated]
2883 lockd.nlm_grace_period=P [NFS] Assign grace period.
2886 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2889 lockd.nlm_timeout=T [NFS] Assign timeout value.
2892 lockd.nlm_udpport=M [NFS] Assign UDP port.
2895 lockdown= [SECURITY]
2896 { integrity | confidentiality }
2897 Enable the kernel lockdown feature. If set to
2898 integrity, kernel features that allow userland to
2899 modify the running kernel are disabled. If set to
2900 confidentiality, kernel features that allow userland
2901 to extract confidential information from the kernel
2904 locktorture.nreaders_stress= [KNL]
2905 Set the number of locking read-acquisition kthreads.
2906 Defaults to being automatically set based on the
2907 number of online CPUs.
2909 locktorture.nwriters_stress= [KNL]
2910 Set the number of locking write-acquisition kthreads.
2912 locktorture.onoff_holdoff= [KNL]
2913 Set time (s) after boot for CPU-hotplug testing.
2915 locktorture.onoff_interval= [KNL]
2916 Set time (s) between CPU-hotplug operations, or
2917 zero to disable CPU-hotplug testing.
2919 locktorture.shuffle_interval= [KNL]
2920 Set task-shuffle interval (jiffies). Shuffling
2921 tasks allows some CPUs to go into dyntick-idle
2922 mode during the locktorture test.
2924 locktorture.shutdown_secs= [KNL]
2925 Set time (s) after boot system shutdown. This
2926 is useful for hands-off automated testing.
2928 locktorture.stat_interval= [KNL]
2929 Time (s) between statistics printk()s.
2931 locktorture.stutter= [KNL]
2932 Time (s) to stutter testing, for example,
2933 specifying five seconds causes the test to run for
2934 five seconds, wait for five seconds, and so on.
2935 This tests the locking primitive's ability to
2936 transition abruptly to and from idle.
2938 locktorture.torture_type= [KNL]
2939 Specify the locking implementation to test.
2941 locktorture.writer_fifo= [KNL]
2942 Run the write-side locktorture kthreads at
2943 sched_set_fifo() real-time priority.
2945 locktorture.verbose= [KNL]
2946 Enable additional printk() statements.
2948 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
2951 loglevel= All Kernel Messages with a loglevel smaller than the
2952 console loglevel will be printed to the console. It can
2953 also be changed with klogd or other programs. The
2954 loglevels are defined as follows:
2956 0 (KERN_EMERG) system is unusable
2957 1 (KERN_ALERT) action must be taken immediately
2958 2 (KERN_CRIT) critical conditions
2959 3 (KERN_ERR) error conditions
2960 4 (KERN_WARNING) warning conditions
2961 5 (KERN_NOTICE) normal but significant condition
2962 6 (KERN_INFO) informational
2963 7 (KERN_DEBUG) debug-level messages
2965 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
2966 in bytes. n must be a power of two and greater
2967 than the minimal size. The minimal size is defined
2968 by LOG_BUF_SHIFT kernel config parameter. There is
2969 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2970 that allows to increase the default size depending on
2971 the number of CPUs. See init/Kconfig for more details.
2973 logo.nologo [FB] Disables display of the built-in Linux logo.
2974 This may be used to provide more screen space for
2975 kernel log messages and is useful when debugging
2976 kernel boot problems.
2978 lp=0 [LP] Specify parallel ports to use, e.g,
2979 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2980 lp=reset first parallel port). 'lp=0' disables the
2981 lp=auto printer driver. 'lp=reset' (which can be
2982 specified in addition to the ports) causes
2983 attached printers to be reset. Using
2984 lp=port1,port2,... specifies the parallel ports
2985 to associate lp devices with, starting with
2986 lp0. A port specification may be 'none' to skip
2987 that lp device, or a parport name such as
2988 'parport0'. Specifying 'lp=auto' instead of a
2989 port specification list means that device IDs
2990 from each port should be examined, to see if
2991 an IEEE 1284-compliant printer is attached; if
2992 so, the driver will manage that printer.
2993 See also header of drivers/char/lp.c.
2996 Sets loops_per_jiffy to given constant, thus avoiding
2997 time-consuming boot-time autodetection (up to 250 ms per
2998 CPU). 0 enables autodetection (default). To determine
2999 the correct value for your kernel, boot with normal
3000 autodetection and see what value is printed. Note that
3001 on SMP systems the preset will be applied to all CPUs,
3002 which is likely to cause problems if your CPUs need
3003 significantly divergent settings. An incorrect value
3004 will cause delays in the kernel to be wrong, leading to
3005 unpredictable I/O errors and other breakage. Although
3006 unlikely, in the extreme case this might damage your
3010 Format: <io>,<irq>,<dma>
3012 lsm.debug [SECURITY] Enable LSM initialization debugging output.
3015 [SECURITY] Choose order of LSM initialization. This
3016 overrides CONFIG_LSM, and the "security=" parameter.
3018 machvec= [IA-64] Force the use of a particular machine-vector
3019 (machvec) in a generic kernel.
3020 Example: machvec=hpzx1
3022 machtype= [Loongson] Share the same kernel image file between
3023 different yeeloong laptops.
3024 Example: machtype=lemote-yeeloong-2f-7inch
3026 max_addr=nn[KMG] [KNL,BOOT,IA-64] All physical memory greater
3027 than or equal to this physical address is ignored.
3029 maxcpus= [SMP] Maximum number of processors that an SMP kernel
3030 will bring up during bootup. maxcpus=n : n >= 0 limits
3031 the kernel to bring up 'n' processors. Surely after
3032 bootup you can bring up the other plugged cpu by executing
3033 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
3034 only takes effect during system bootup.
3035 While n=0 is a special case, it is equivalent to "nosmp",
3036 which also disables the IO APIC.
3038 max_loop= [LOOP] The number of loop block devices that get
3039 (loop.max_loop) unconditionally pre-created at init time. The default
3040 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
3041 of statically allocating a predefined number, loop
3042 devices can be requested on-demand with the
3043 /dev/loop-control interface.
3045 mce [X86-32] Machine Check Exception
3047 mce=option [X86-64] See Documentation/arch/x86/x86_64/boot-options.rst
3049 md= [HW] RAID subsystems devices and level
3050 See Documentation/admin-guide/md.rst.
3053 Format: <first>,<last>
3054 Specifies range of consoles to be captured by the MDA.
3057 Control mitigation for the Micro-architectural Data
3058 Sampling (MDS) vulnerability.
3060 Certain CPUs are vulnerable to an exploit against CPU
3061 internal buffers which can forward information to a
3062 disclosure gadget under certain conditions.
3064 In vulnerable processors, the speculatively
3065 forwarded data can be used in a cache side channel
3066 attack, to access data to which the attacker does
3067 not have direct access.
3069 This parameter controls the MDS mitigation. The
3072 full - Enable MDS mitigation on vulnerable CPUs
3073 full,nosmt - Enable MDS mitigation and disable
3074 SMT on vulnerable CPUs
3075 off - Unconditionally disable MDS mitigation
3077 On TAA-affected machines, mds=off can be prevented by
3078 an active TAA mitigation as both vulnerabilities are
3079 mitigated with the same mechanism so in order to disable
3080 this mitigation, you need to specify tsx_async_abort=off
3083 Not specifying this option is equivalent to
3086 For details see: Documentation/admin-guide/hw-vuln/mds.rst
3088 mem=nn[KMG] [HEXAGON] Set the memory size.
3089 Must be specified, otherwise memory size will be 0.
3091 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
3092 Amount of memory to be used in cases as follows:
3095 2 when the kernel is not able to see the whole system memory;
3096 3 memory that lies after 'mem=' boundary is excluded from
3097 the hypervisor, then assigned to KVM guests.
3098 4 to limit the memory available for kdump kernel.
3100 [ARC,MICROBLAZE] - the limit applies only to low memory,
3101 high memory is not affected.
3103 [ARM64] - only limits memory covered by the linear
3104 mapping. The NOMAP regions are not affected.
3106 [X86] Work as limiting max address. Use together
3107 with memmap= to avoid physical address space collisions.
3108 Without memmap= PCI devices could be placed at addresses
3109 belonging to unused RAM.
3111 Note that this only takes effects during boot time since
3112 in above case 3, memory may need be hot added after boot
3113 if system memory of hypervisor is not sufficient.
3116 [ARM,MIPS] - override the memory layout reported by
3118 Define a memory region of size nn[KMG] starting at
3120 Multiple different regions can be specified with
3121 multiple mem= parameters on the command line.
3123 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
3126 memblock=debug [KNL] Enable memblock debug messages.
3129 [KNL,SH] Allow user to override the default size for
3130 per-device physically contiguous DMA buffers.
3132 memhp_default_state=online/offline
3133 [KNL] Set the initial state for the memory hotplug
3134 onlining policy. If not specified, the default value is
3135 set according to the
3136 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
3138 See Documentation/admin-guide/mm/memory-hotplug.rst.
3140 memmap=exactmap [KNL,X86] Enable setting of an exact
3141 E820 memory map, as specified by the user.
3142 Such memmap=exactmap lines can be constructed based on
3143 BIOS output or other requirements. See the memmap=nn@ss
3146 memmap=nn[KMG]@ss[KMG]
3147 [KNL, X86, MIPS, XTENSA] Force usage of a specific region of memory.
3148 Region of memory to be used is from ss to ss+nn.
3149 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
3150 which limits max address to nn[KMG].
3151 Multiple different regions can be specified,
3154 memmap=100M@2G,100M#3G,1G!1024G
3156 memmap=nn[KMG]#ss[KMG]
3157 [KNL,ACPI] Mark specific memory as ACPI data.
3158 Region of memory to be marked is from ss to ss+nn.
3160 memmap=nn[KMG]$ss[KMG]
3161 [KNL,ACPI] Mark specific memory as reserved.
3162 Region of memory to be reserved is from ss to ss+nn.
3163 Example: Exclude memory from 0x18690000-0x1869ffff
3164 memmap=64K$0x18690000
3166 memmap=0x10000$0x18690000
3167 Some bootloaders may need an escape character before '$',
3168 like Grub2, otherwise '$' and the following number
3171 memmap=nn[KMG]!ss[KMG]
3172 [KNL,X86] Mark specific memory as protected.
3173 Region of memory to be used, from ss to ss+nn.
3174 The memory region may be marked as e820 type 12 (0xc)
3175 and is NVDIMM or ADR memory.
3177 memmap=<size>%<offset>-<oldtype>+<newtype>
3178 [KNL,ACPI] Convert memory within the specified region
3179 from <oldtype> to <newtype>. If "-<oldtype>" is left
3180 out, the whole region will be marked as <newtype>,
3181 even if previously unavailable. If "+<newtype>" is left
3182 out, matching memory will be removed. Types are
3183 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
3184 3 = ACPI, 12 = PRAM.
3186 memory_corruption_check=0/1 [X86]
3187 Some BIOSes seem to corrupt the first 64k of
3188 memory when doing things like suspend/resume.
3189 Setting this option will scan the memory
3190 looking for corruption. Enabling this will
3191 both detect corruption and prevent the kernel
3192 from using the memory being corrupted.
3193 However, its intended as a diagnostic tool; if
3194 repeatable BIOS-originated corruption always
3195 affects the same memory, you can use memmap=
3196 to prevent the kernel from using that memory.
3198 memory_corruption_check_size=size [X86]
3199 By default it checks for corruption in the low
3200 64k, making this memory unavailable for normal
3201 use. Use this parameter to scan for
3202 corruption in more or less memory.
3204 memory_corruption_check_period=seconds [X86]
3205 By default it checks for corruption every 60
3206 seconds. Use this parameter to check at some
3207 other rate. 0 disables periodic checking.
3209 memory_hotplug.memmap_on_memory
3210 [KNL,X86,ARM] Boolean flag to enable this feature.
3211 Format: {on | off (default)}
3212 When enabled, runtime hotplugged memory will
3213 allocate its internal metadata (struct pages,
3214 those vmemmap pages cannot be optimized even
3215 if hugetlb_free_vmemmap is enabled) from the
3216 hotadded memory which will allow to hotadd a
3217 lot of memory without requiring additional
3219 This feature is disabled by default because it
3220 has some implication on large (e.g. GB)
3221 allocations in some configurations (e.g. small
3223 The state of the flag can be read in
3224 /sys/module/memory_hotplug/parameters/memmap_on_memory.
3225 Note that even when enabled, there are a few cases where
3226 the feature is not effective.
3228 memtest= [KNL,X86,ARM,M68K,PPC,RISCV] Enable memtest
3230 default : 0 <disable>
3231 Specifies the number of memtest passes to be
3232 performed. Each pass selects another test
3233 pattern from a given set of patterns. Memtest
3234 fills the memory with this pattern, validates
3235 memory contents and reserves bad memory
3236 regions that are detected.
3238 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
3239 Valid arguments: on, off
3240 Default (depends on kernel configuration option):
3241 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
3242 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
3243 mem_encrypt=on: Activate SME
3244 mem_encrypt=off: Do not activate SME
3246 Refer to Documentation/virt/kvm/x86/amd-memory-encryption.rst
3247 for details on when memory encryption can be activated.
3249 mem_sleep_default= [SUSPEND] Default system suspend mode:
3250 s2idle - Suspend-To-Idle
3251 shallow - Power-On Suspend or equivalent (if supported)
3252 deep - Suspend-To-RAM or equivalent (if supported)
3253 See Documentation/admin-guide/pm/sleep-states.rst.
3255 mfgpt_irq= [IA-32] Specify the IRQ to use for the
3256 Multi-Function General Purpose Timers on AMD Geode
3259 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
3260 the BIOS has incorrectly applied a workaround. TinyBIOS
3261 version 0.98 is known to be affected, 0.99 fixes the
3262 problem by letting the user disable the workaround.
3266 min_addr=nn[KMG] [KNL,BOOT,IA-64] All physical memory below this
3267 physical address is ignored.
3269 mini2440= [ARM,HW,KNL]
3270 Format:[0..2][b][c][t]
3272 MINI2440 configuration specification:
3273 0 - The attached screen is the 3.5" TFT
3274 1 - The attached screen is the 7" TFT
3275 2 - The VGA Shield is attached (1024x768)
3276 Leaving out the screen size parameter will not load
3277 the TFT driver, and the framebuffer will be left
3279 b - Enable backlight. The TFT backlight pin will be
3280 linked to the kernel VESA blanking code and a GPIO
3281 LED. This parameter is not necessary when using the
3283 c - Enable the s3c camera interface.
3284 t - Reserved for enabling touchscreen support. The
3285 touchscreen support is not enabled in the mainstream
3286 kernel as of 2.6.30, a preliminary port can be found
3287 in the "bleeding edge" mini2440 support kernel at
3288 https://repo.or.cz/w/linux-2.6/mini2440.git
3291 [X86,PPC,S390,ARM64] Control optional mitigations for
3292 CPU vulnerabilities. This is a set of curated,
3293 arch-independent options, each of which is an
3294 aggregation of existing arch-specific options.
3297 Disable all optional CPU mitigations. This
3298 improves system performance, but it may also
3299 expose users to several CPU vulnerabilities.
3300 Equivalent to: if nokaslr then kpti=0 [ARM64]
3301 gather_data_sampling=off [X86]
3302 kvm.nx_huge_pages=off [X86]
3305 mmio_stale_data=off [X86]
3306 no_entry_flush [PPC]
3307 no_uaccess_flush [PPC]
3310 nospectre_bhb [ARM64]
3311 nospectre_v1 [X86,PPC]
3312 nospectre_v2 [X86,PPC,S390,ARM64]
3314 spec_store_bypass_disable=off [X86,PPC]
3315 spectre_v2_user=off [X86]
3316 srbds=off [X86,INTEL]
3317 ssbd=force-off [ARM64]
3318 tsx_async_abort=off [X86]
3321 This does not have any effect on
3322 kvm.nx_huge_pages when
3323 kvm.nx_huge_pages=force.
3326 Mitigate all CPU vulnerabilities, but leave SMT
3327 enabled, even if it's vulnerable. This is for
3328 users who don't want to be surprised by SMT
3329 getting disabled across kernel upgrades, or who
3330 have other ways of avoiding SMT-based attacks.
3331 Equivalent to: (default behavior)
3334 Mitigate all CPU vulnerabilities, disabling SMT
3335 if needed. This is for users who always want to
3336 be fully mitigated, even if it means losing SMT.
3337 Equivalent to: l1tf=flush,nosmt [X86]
3338 mds=full,nosmt [X86]
3339 tsx_async_abort=full,nosmt [X86]
3340 mmio_stale_data=full,nosmt [X86]
3341 retbleed=auto,nosmt [X86]
3344 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
3345 parameter allows control of the logging verbosity for
3346 the additional memory initialisation checks. A value
3347 of 0 disables mminit logging and a level of 4 will
3348 log everything. Information is printed at KERN_DEBUG
3349 so loglevel=8 may also need to be specified.
3352 [X86,INTEL] Control mitigation for the Processor
3353 MMIO Stale Data vulnerabilities.
3355 Processor MMIO Stale Data is a class of
3356 vulnerabilities that may expose data after an MMIO
3357 operation. Exposed data could originate or end in
3358 the same CPU buffers as affected by MDS and TAA.
3359 Therefore, similar to MDS and TAA, the mitigation
3360 is to clear the affected CPU buffers.
3362 This parameter controls the mitigation. The
3365 full - Enable mitigation on vulnerable CPUs
3367 full,nosmt - Enable mitigation and disable SMT on
3370 off - Unconditionally disable mitigation
3372 On MDS or TAA affected machines,
3373 mmio_stale_data=off can be prevented by an active
3374 MDS or TAA mitigation as these vulnerabilities are
3375 mitigated with the same mechanism so in order to
3376 disable this mitigation, you need to specify
3377 mds=off and tsx_async_abort=off too.
3379 Not specifying this option is equivalent to
3380 mmio_stale_data=full.
3383 Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst
3385 <module>.async_probe[=<bool>] [KNL]
3386 If no <bool> value is specified or if the value
3387 specified is not a valid <bool>, enable asynchronous
3388 probe on this module. Otherwise, enable/disable
3389 asynchronous probe on this module as indicated by the
3390 <bool> value. See also: module.async_probe
3392 module.async_probe=<bool>
3393 [KNL] When set to true, modules will use async probing
3394 by default. To enable/disable async probing for a
3395 specific module, use the module specific control that
3396 is documented under <module>.async_probe. When both
3397 module.async_probe and <module>.async_probe are
3398 specified, <module>.async_probe takes precedence for
3399 the specific module.
3401 module.enable_dups_trace
3402 [KNL] When CONFIG_MODULE_DEBUG_AUTOLOAD_DUPS is set,
3403 this means that duplicate request_module() calls will
3404 trigger a WARN_ON() instead of a pr_warn(). Note that
3405 if MODULE_DEBUG_AUTOLOAD_DUPS_TRACE is set, WARN_ON()s
3406 will always be issued and this option does nothing.
3408 [KNL] When CONFIG_MODULE_SIG is set, this means that
3409 modules without (valid) signatures will fail to load.
3410 Note that if CONFIG_MODULE_SIG_FORCE is set, that
3411 is always true, so this option does nothing.
3413 module_blacklist= [KNL] Do not load a comma-separated list of
3414 modules. Useful for debugging problem modules.
3417 [MOUSE] Maximum time between finger touching and
3418 leaving touchpad surface for touch to be considered
3419 a tap and be reported as a left button click (for
3420 touchpads working in absolute mode only).
3422 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
3423 reporting absolute coordinates, such as tablets
3424 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
3425 reporting absolute coordinates, such as tablets
3427 movablecore= [KNL,X86,IA-64,PPC]
3428 Format: nn[KMGTPE] | nn%
3429 This parameter is the complement to kernelcore=, it
3430 specifies the amount of memory used for migratable
3431 allocations. If both kernelcore and movablecore is
3432 specified, then kernelcore will be at *least* the
3433 specified value but may be more. If movablecore on its
3434 own is specified, the administrator must be careful
3435 that the amount of memory usable for all allocations
3438 movable_node [KNL] Boot-time switch to make hotplugable memory
3439 NUMA nodes to be movable. This means that the memory
3440 of such nodes will be usable only for movable
3441 allocations which rules out almost all kernel
3442 allocations. Use with caution!
3444 MTD_Partition= [MTD]
3445 Format: <name>,<region-number>,<size>,<offset>
3447 MTD_Region= [MTD] Format:
3448 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
3451 See drivers/mtd/parsers/cmdlinepart.c
3454 ARM/S3C2412 JIVE boot control
3456 See arch/arm/mach-s3c/mach-jive.c
3458 mtouchusb.raw_coordinates=
3459 [HW] Make the MicroTouch USB driver use raw coordinates
3460 ('y', default) or cooked coordinates ('n')
3463 Enable printing debug information related to MTRR
3464 registers at boot time.
3466 mtrr_chunk_size=nn[KMG] [X86]
3467 used for mtrr cleanup. It is largest continuous chunk
3468 that could hold holes aka. UC entries.
3470 mtrr_gran_size=nn[KMG] [X86]
3471 Used for mtrr cleanup. It is granularity of mtrr block.
3473 Large value could prevent small alignment from
3476 mtrr_spare_reg_nr=n [X86]
3478 Range: 0,7 : spare reg number
3480 Used for mtrr cleanup. It is spare mtrr entries number.
3481 Set to 2 or more if your graphical card needs more.
3483 multitce=off [PPC] This parameter disables the use of the pSeries
3484 firmware feature for updating multiple TCE entries
3487 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
3489 netdev= [NET] Network devices parameters
3490 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
3491 Note that mem_start is often overloaded to mean
3492 something different and driver-specific.
3493 This usage is only documented in each driver source
3496 netpoll.carrier_timeout=
3497 [NET] Specifies amount of time (in seconds) that
3498 netpoll should wait for a carrier. By default netpoll
3502 [NETFILTER] Enable connection tracking flow accounting
3503 0 to disable accounting
3504 1 to enable accounting
3508 [NFS] sets the pathname to the program which is used
3509 to update the NFS client cache entries.
3511 nfs.cache_getent_timeout=
3512 [NFS] sets the timeout after which an attempt to
3513 update a cache entry is deemed to have failed.
3515 nfs.callback_nr_threads=
3516 [NFSv4] set the total number of threads that the
3517 NFS client will assign to service NFSv4 callback
3520 nfs.callback_tcpport=
3521 [NFS] set the TCP port on which the NFSv4 callback
3522 channel should listen.
3525 [NFS] enable 64-bit inode numbers.
3526 If zero, the NFS client will fake up a 32-bit inode
3527 number for the readdir() and stat() syscalls instead
3528 of returning the full 64-bit number.
3529 The default is to return 64-bit inode numbers.
3531 nfs.idmap_cache_timeout=
3532 [NFS] set the maximum lifetime for idmapper cache
3535 nfs.max_session_cb_slots=
3536 [NFSv4.1] Sets the maximum number of session
3537 slots the client will assign to the callback
3538 channel. This determines the maximum number of
3539 callbacks the client will process in parallel for
3540 a particular server.
3542 nfs.max_session_slots=
3543 [NFSv4.1] Sets the maximum number of session slots
3544 the client will attempt to negotiate with the server.
3545 This limits the number of simultaneous RPC requests
3546 that the client can send to the NFSv4.1 server.
3547 Note that there is little point in setting this
3548 value higher than the max_tcp_slot_table_limit.
3550 nfs.nfs4_disable_idmapping=
3551 [NFSv4] When set to the default of '1', this option
3552 ensures that both the RPC level authentication
3553 scheme and the NFS level operations agree to use
3554 numeric uids/gids if the mount is using the
3555 'sec=sys' security flavour. In effect it is
3556 disabling idmapping, which can make migration from
3557 legacy NFSv2/v3 systems to NFSv4 easier.
3558 Servers that do not support this mode of operation
3559 will be autodetected by the client, and it will fall
3560 back to using the idmapper.
3561 To turn off this behaviour, set the value to '0'.
3564 [NFS4] Specify an additional fixed unique ident-
3565 ification string that NFSv4 clients can insert into
3566 their nfs_client_id4 string. This is typically a
3567 UUID that is generated at system install time.
3569 nfs.recover_lost_locks=
3570 [NFSv4] Attempt to recover locks that were lost due
3571 to a lease timeout on the server. Please note that
3572 doing this risks data corruption, since there are
3573 no guarantees that the file will remain unchanged
3574 after the locks are lost.
3575 If you want to enable the kernel legacy behaviour of
3576 attempting to recover these locks, then set this
3578 The default parameter value of '0' causes the kernel
3579 not to attempt recovery of lost locks.
3581 nfs.send_implementation_id=
3582 [NFSv4.1] Send client implementation identification
3583 information in exchange_id requests.
3584 If zero, no implementation identification information
3586 The default is to send the implementation identification
3589 nfs4.layoutstats_timer=
3590 [NFSv4.2] Change the rate at which the kernel sends
3591 layoutstats to the pNFS metadata server.
3593 Setting this to value to 0 causes the kernel to use
3594 whatever value is the default set by the layout
3595 driver. A non-zero value sets the minimum interval
3596 in seconds between layoutstats transmissions.
3598 nfsd.inter_copy_offload_enable=
3599 [NFSv4.2] When set to 1, the server will support
3600 server-to-server copies for which this server is
3601 the destination of the copy.
3603 nfsd.nfs4_disable_idmapping=
3604 [NFSv4] When set to the default of '1', the NFSv4
3605 server will return only numeric uids and gids to
3606 clients using auth_sys, and will accept numeric uids
3607 and gids from such clients. This is intended to ease
3608 migration from NFSv2/v3.
3610 nfsd.nfsd4_ssc_umount_timeout=
3611 [NFSv4.2] When used as the destination of a
3612 server-to-server copy, knfsd temporarily mounts
3613 the source server. It caches the mount in case
3614 it will be needed again, and discards it if not
3615 used for the number of milliseconds specified by
3618 nfsaddrs= [NFS] Deprecated. Use ip= instead.
3619 See Documentation/admin-guide/nfs/nfsroot.rst.
3621 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
3622 See Documentation/admin-guide/nfs/nfsroot.rst.
3624 nfsrootdebug [NFS] enable nfsroot debugging messages.
3625 See Documentation/admin-guide/nfs/nfsroot.rst.
3627 nmi_backtrace.backtrace_idle [KNL]
3628 Dump stacks even of idle CPUs in response to an
3629 NMI stack-backtrace request.
3631 nmi_debug= [KNL,SH] Specify one or more actions to take
3632 when a NMI is triggered.
3633 Format: [state][,regs][,debounce][,die]
3635 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
3636 Format: [panic,][nopanic,][num]
3638 0 - turn hardlockup detector in nmi_watchdog off
3639 1 - turn hardlockup detector in nmi_watchdog on
3640 When panic is specified, panic when an NMI watchdog
3641 timeout occurs (or 'nopanic' to not panic on an NMI
3642 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
3643 To disable both hard and soft lockup detectors,
3644 please see 'nowatchdog'.
3645 This is useful when you use a panic=... timeout and
3646 need the box quickly up again.
3648 These settings can be accessed at runtime via
3649 the nmi_watchdog and hardlockup_panic sysctls.
3651 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3652 emulation library even if a 387 maths coprocessor
3655 no4lvl [RISCV] Disable 4-level and 5-level paging modes. Forces
3656 kernel to use 3-level paging instead.
3658 no5lvl [X86-64,RISCV] Disable 5-level paging mode. Forces
3659 kernel to use 4-level paging instead.
3661 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
3662 caches in the slab allocator. Saves per-node memory,
3663 but will impact performance.
3667 noaltinstr [S390] Disables alternative instructions patching
3668 (CPU alternatives feature).
3670 noapic [SMP,APIC] Tells the kernel to not make use of any
3671 IOAPICs that may be present in the system.
3673 noautogroup Disable scheduler automatic task group creation.
3678 [HW] Never suspend the console
3679 Disable suspending of consoles during suspend and
3680 hibernate operations. Once disabled, debugging
3681 messages can reach various consoles while the rest
3682 of the system is being put to sleep (ie, while
3683 debugging driver suspend/resume hooks). This may
3684 not work reliably with all consoles, but is known
3685 to work with serial and VGA consoles.
3686 To facilitate more flexible debugging, we also add
3687 console_suspend, a printk module parameter to control
3688 it. Users could use console_suspend (usually
3689 /sys/module/printk/parameters/console_suspend) to
3690 turn on/off it dynamically.
3693 [KNL] Disable object debugging
3695 nodsp [SH] Disable hardware DSP at boot time.
3697 noefi Disable EFI runtime services support.
3699 no_entry_flush [PPC] Don't flush the L1-D cache when entering the kernel.
3704 This affects only 32-bit executables.
3705 noexec32=on: enable non-executable mappings (default)
3706 read doesn't imply executable mappings
3707 noexec32=off: disable non-executable mappings
3708 read implies executable mappings
3710 no_file_caps Tells the kernel not to honor file capabilities. The
3711 only way then for a file to be executed with privilege
3712 is to be setuid root or executed by root.
3714 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3716 nofsgsbase [X86] Disables FSGSBASE instructions.
3718 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3719 register save and restore. The kernel will only save
3720 legacy floating-point registers on task switch.
3722 nohalt [IA-64] Tells the kernel not to use the power saving
3723 function PAL_HALT_LIGHT when idle. This increases
3724 power-consumption. On the positive side, it reduces
3725 interrupt wake-up latency, which may improve performance
3726 in certain environments such as networked servers or
3730 Force pointers printed to the console or buffers to be
3731 unhashed. By default, when a pointer is printed via %p
3732 format string, that pointer is "hashed", i.e. obscured
3733 by hashing the pointer value. This is a security feature
3734 that hides actual kernel addresses from unprivileged
3735 users, but it also makes debugging the kernel more
3736 difficult since unequal pointers can no longer be
3737 compared. However, if this command-line option is
3738 specified, then all normal pointers will have their true
3739 value printed. This option should only be specified when
3740 debugging the kernel. Please do not use on production
3743 nohibernate [HIBERNATION] Disable hibernation and resume.
3745 nohlt [ARM,ARM64,MICROBLAZE,MIPS,SH] Forces the kernel to
3746 busy wait in do_idle() and not use the arch_cpu_idle()
3747 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
3748 to be effective. This is useful on platforms where the
3749 sleep(SH) or wfi(ARM,ARM64) instructions do not work
3750 correctly or when doing power measurements to evaluate
3751 the impact of the sleep instructions. This is also
3752 useful when using JTAG debugger.
3754 nohugeiomap [KNL,X86,PPC,ARM64] Disable kernel huge I/O mappings.
3756 nohugevmalloc [KNL,X86,PPC,ARM64] Disable kernel huge vmalloc mappings.
3758 nohz= [KNL] Boottime enable/disable dynamic ticks
3759 Valid arguments: on, off
3762 nohz_full= [KNL,BOOT,SMP,ISOL]
3763 The argument is a cpu list, as described above.
3764 In kernels built with CONFIG_NO_HZ_FULL=y, set
3765 the specified list of CPUs whose tick will be stopped
3766 whenever possible. The boot CPU will be forced outside
3767 the range to maintain the timekeeping. Any CPUs
3768 in this list will have their RCU callbacks offloaded,
3769 just as if they had also been called out in the
3770 rcu_nocbs= boot parameter.
3772 Note that this argument takes precedence over
3773 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option.
3775 noinitrd [RAM] Tells the kernel not to load any configured
3778 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
3780 [Deprecated - use intremap=off]
3784 noinvpcid [X86] Disable the INVPCID cpu feature.
3786 noiotrap [SH] Disables trapped I/O port accesses.
3788 noirqdebug [X86-32] Disables the code which attempts to detect and
3789 disable unhandled interrupt sources.
3791 noisapnp [ISAPNP] Disables ISA PnP code.
3793 nojitter [IA-64] Disables jitter checking for ITC timers.
3796 When CONFIG_RANDOMIZE_BASE is set, this disables
3797 kernel and module base offset ASLR (Address Space
3798 Layout Randomization).
3800 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
3803 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
3805 nolapic [X86-32,APIC] Do not enable or use the local APIC.
3807 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
3809 nomca [IA-64] Disable machine check abort handling
3811 nomce [X86-32] Disable Machine Check Exception
3813 nomfgpt [X86-32] Disable Multi-Function General Purpose
3814 Timer usage (for AMD Geode machines).
3816 nomodeset Disable kernel modesetting. Most systems' firmware
3817 sets up a display mode and provides framebuffer memory
3818 for output. With nomodeset, DRM and fbdev drivers will
3819 not load if they could possibly displace the pre-
3820 initialized output. Only the system framebuffer will
3821 be available for use. The respective drivers will not
3822 perform display-mode changes or accelerated rendering.
3824 Useful as error fallback, or for testing and debugging.
3826 nomodule Disable module load
3828 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3829 shutdown the other cpus. Instead use the REBOOT_VECTOR
3832 nopat [X86] Disable PAT (page attribute table extension of
3833 pagetables) support.
3835 nopcid [X86-64] Disable the PCID cpu feature.
3837 nopku [X86] Disable Memory Protection Keys CPU feature found
3841 Equivalent to pti=off
3843 nopv= [X86,XEN,KVM,HYPER_V,VMWARE]
3844 Disables the PV optimizations forcing the guest to run
3845 as generic guest with no PV drivers. Currently support
3846 XEN HVM, KVM, HYPER_V and VMWARE guest.
3848 nopvspin [X86,XEN,KVM]
3849 Disables the qspinlock slow path using PV optimizations
3850 which allow the hypervisor to 'idle' the guest on lock
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
3864 no-scroll [VGA] Disables scrollback.
3865 This is required for the Braillex ib80-piezo Braille
3866 reader made by F.H. Papenmeier (Germany).
3868 nosgx [X86-64,SGX] Disables Intel SGX kernel support.
3871 Disable SMAP (Supervisor Mode Access Prevention)
3872 even if it is supported by processor.
3875 Disable SMEP (Supervisor Mode Execution Prevention)
3876 even if it is supported by processor.
3878 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
3879 and disable the IO APIC. legacy for "maxcpus=0".
3881 nosmt [KNL,MIPS,S390] Disable symmetric multithreading (SMT).
3882 Equivalent to smt=1.
3884 [KNL,X86] Disable symmetric multithreading (SMT).
3885 nosmt=force: Force disable SMT, cannot be undone
3886 via the sysfs control file.
3888 nosoftlockup [KNL] Disable the soft-lockup detector.
3890 nospec_store_bypass_disable
3891 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
3893 nospectre_bhb [ARM64] Disable all mitigations for Spectre-BHB (branch
3894 history injection) vulnerability. System may allow data leaks
3897 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1
3898 (bounds check bypass). With this option data leaks are
3899 possible in the system.
3901 nospectre_v2 [X86,PPC_E500,ARM64] Disable all mitigations for
3902 the Spectre variant 2 (indirect branch prediction)
3903 vulnerability. System may allow data leaks with this
3906 no-steal-acc [X86,PV_OPS,ARM64,PPC/PSERIES] Disable paravirtualized
3907 steal time accounting. steal time is computed, but
3908 won't influence scheduler behaviour
3910 nosync [HW,M68K] Disables sync negotiation for all devices.
3912 no_timer_check [X86,APIC] Disables the code which tests for
3913 broken timer IRQ sources.
3916 [PPC] Don't flush the L1-D cache after accessing user data.
3918 novmcoredd [KNL,KDUMP]
3919 Disable device dump. Device dump allows drivers to
3920 append dump data to vmcore so you can collect driver
3921 specified debug info. Drivers can append the data
3922 without any limit and this data is stored in memory,
3923 so this may cause significant memory stress. Disabling
3924 device dump can help save memory but the driver debug
3925 data will be no longer available. This parameter
3926 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
3930 [X86,PV_OPS] Disable paravirtualized VMware scheduler
3931 clock and use the default one.
3933 nowatchdog [KNL] Disable both lockup detectors, i.e.
3934 soft-lockup and NMI watchdog (hard-lockup).
3938 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
3940 NOTE: this parameter will be ignored on systems with the
3941 LEGACY_XAPIC_DISABLED bit set in the
3942 IA32_XAPIC_DISABLE_STATUS MSR.
3944 noxsave [BUGS=X86] Disables x86 extended register state save
3945 and restore using xsave. The kernel will fallback to
3946 enabling legacy floating-point and sse state.
3948 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
3949 register states. The kernel will fall back to use
3950 xsave to save the states. By using this parameter,
3951 performance of saving the states is degraded because
3952 xsave doesn't support modified optimization while
3953 xsaveopt supports it on xsaveopt enabled systems.
3955 noxsaves [X86] Disables xsaves and xrstors used in saving and
3956 restoring x86 extended register state in compacted
3957 form of xsave area. The kernel will fall back to use
3958 xsaveopt and xrstor to save and restore the states
3959 in standard form of xsave area. By using this
3960 parameter, xsave area per process might occupy more
3961 memory on xsaves enabled systems.
3963 nps_mtm_hs_ctr= [KNL,ARC]
3964 This parameter sets the maximum duration, in
3965 cycles, each HW thread of the CTOP can run
3966 without interruptions, before HW switches it.
3967 The actual maximum duration is 16 times this
3969 Format: integer between 1 and 255
3972 nptcg= [IA-64] Override max number of concurrent global TLB
3973 purges which is reported from either PAL_VM_SUMMARY or
3976 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
3977 could support. nr_cpus=n : n >= 1 limits the kernel to
3978 support 'n' processors. It could be larger than the
3979 number of already plugged CPU during bootup, later in
3980 runtime you can physically add extra cpu until it reaches
3981 n. So during boot up some boot time memory for per-cpu
3982 variables need be pre-allocated for later physical cpu
3985 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
3987 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86] Disable NUMA, Only
3988 set up a single NUMA node spanning all memory.
3990 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic
3992 Allowed values are enable and disable
3994 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
3995 'node', 'default' can be specified
3996 This can be set from sysctl after boot.
3997 See Documentation/admin-guide/sysctl/vm.rst for details.
3999 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
4000 See Documentation/core-api/debugging-via-ohci1394.rst for more
4003 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
4004 Rather than timing out after 20 ms if an EC
4005 command is not properly ACKed, override the length
4006 of the timeout. We have interrupts disabled while
4007 waiting for the ACK, so if this is set too high
4008 interrupts *may* be lost!
4010 omap_mux= [OMAP] Override bootloader pin multiplexing.
4011 Format: <mux_mode0.mode_name=value>...
4012 For example, to override I2C bus2:
4013 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
4015 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
4017 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
4019 boundary - index of last SLC block on Flex-OneNAND.
4020 The remaining blocks are configured as MLC blocks.
4021 lock - Configure if Flex-OneNAND boundary should be locked.
4022 Once locked, the boundary cannot be changed.
4023 1 indicates lock status, 0 indicates unlock status.
4025 oops=panic Always panic on oopses. Default is to just kill the
4026 process, but there is a small probability of
4027 deadlocking the machine.
4028 This will also cause panics on machine check exceptions.
4029 Useful together with panic=30 to trigger a reboot.
4032 [KNL] Boolean flag to control whether the page allocator
4033 should randomize its free lists. The randomization may
4034 be automatically enabled if the kernel detects it is
4035 running on a platform with a direct-mapped memory-side
4036 cache, and this parameter can be used to
4037 override/disable that behavior. The state of the flag
4038 can be read from sysfs at:
4039 /sys/module/page_alloc/parameters/shuffle.
4041 page_owner= [KNL] Boot-time page_owner enabling option.
4042 Storage of the information about who allocated
4043 each page is disabled in default. With this switch,
4045 on: enable the feature
4047 page_poison= [KNL] Boot-time parameter changing the state of
4048 poisoning on the buddy allocator, available with
4049 CONFIG_PAGE_POISONING=y.
4050 off: turn off poisoning (default)
4051 on: turn on poisoning
4053 page_reporting.page_reporting_order=
4054 [KNL] Minimal page reporting order
4056 Adjust the minimal page reporting order. The page
4057 reporting is disabled when it exceeds MAX_ORDER.
4059 panic= [KNL] Kernel behaviour on panic: delay <timeout>
4060 timeout > 0: seconds before rebooting
4061 timeout = 0: wait forever
4062 timeout < 0: reboot immediately
4065 panic_print= Bitmask for printing system info when panic happens.
4066 User can chose combination of the following bits:
4067 bit 0: print all tasks info
4068 bit 1: print system memory info
4069 bit 2: print timer info
4070 bit 3: print locks info if CONFIG_LOCKDEP is on
4071 bit 4: print ftrace buffer
4072 bit 5: print all printk messages in buffer
4073 bit 6: print all CPUs backtrace (if available in the arch)
4074 *Be aware* that this option may print a _lot_ of lines,
4075 so there are risks of losing older messages in the log.
4076 Use this option carefully, maybe worth to setup a
4077 bigger log buffer with "log_buf_len" along with this.
4079 panic_on_taint= Bitmask for conditionally calling panic() in add_taint()
4080 Format: <hex>[,nousertaint]
4081 Hexadecimal bitmask representing the set of TAINT flags
4082 that will cause the kernel to panic when add_taint() is
4083 called with any of the flags in this set.
4084 The optional switch "nousertaint" can be utilized to
4085 prevent userspace forced crashes by writing to sysctl
4086 /proc/sys/kernel/tainted any flagset matching with the
4087 bitmask set on panic_on_taint.
4088 See Documentation/admin-guide/tainted-kernels.rst for
4089 extra details on the taint flags that users can pick
4090 to compose the bitmask to assign to panic_on_taint.
4092 panic_on_warn=1 panic() instead of WARN(). Useful to cause kdump
4095 parkbd.port= [HW] Parallel port number the keyboard adapter is
4096 connected to, default is 0.
4098 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
4099 0 for XT, 1 for AT (default is AT).
4102 parport= [HW,PPT] Specify parallel ports. 0 disables.
4103 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
4104 Use 'auto' to force the driver to use any
4105 IRQ/DMA settings detected (the default is to
4106 ignore detected IRQ/DMA settings because of
4107 possible conflicts). You can specify the base
4108 address, IRQ, and DMA settings; IRQ and DMA
4109 should be numbers, or 'auto' (for using detected
4110 settings on that particular port), or 'nofifo'
4111 (to avoid using a FIFO even if it is detected).
4112 Parallel ports are assigned in the order they
4113 are specified on the command line, starting
4116 parport_init_mode= [HW,PPT]
4117 Configure VIA parallel port to operate in
4118 a specific mode. This is necessary on Pegasos
4119 computer where firmware has no options for setting
4120 up parallel port mode and sets it to spp.
4121 Currently this function knows 686a and 8231 chips.
4122 Format: [spp|ps2|epp|ecp|ecpepp]
4124 pata_legacy.all= [HW,LIBATA]
4126 Set to non-zero to probe primary and secondary ISA
4127 port ranges on PCI systems where no PCI PATA device
4128 has been found at either range. Disabled by default.
4130 pata_legacy.autospeed= [HW,LIBATA]
4132 Set to non-zero if a chip is present that snoops speed
4133 changes. Disabled by default.
4135 pata_legacy.ht6560a= [HW,LIBATA]
4137 Set to 1, 2, or 3 for HT 6560A on the primary channel,
4138 the secondary channel, or both channels respectively.
4139 Disabled by default.
4141 pata_legacy.ht6560b= [HW,LIBATA]
4143 Set to 1, 2, or 3 for HT 6560B on the primary channel,
4144 the secondary channel, or both channels respectively.
4145 Disabled by default.
4147 pata_legacy.iordy_mask= [HW,LIBATA]
4149 IORDY enable mask. Set individual bits to allow IORDY
4150 for the respective channel. Bit 0 is for the first
4151 legacy channel handled by this driver, bit 1 is for
4152 the second channel, and so on. The sequence will often
4153 correspond to the primary legacy channel, the secondary
4154 legacy channel, and so on, but the handling of a PCI
4155 bus and the use of other driver options may interfere
4156 with the sequence. By default IORDY is allowed across
4159 pata_legacy.opti82c46x= [HW,LIBATA]
4161 Set to 1, 2, or 3 for Opti 82c611A on the primary
4162 channel, the secondary channel, or both channels
4163 respectively. Disabled by default.
4165 pata_legacy.opti82c611a= [HW,LIBATA]
4167 Set to 1, 2, or 3 for Opti 82c465MV on the primary
4168 channel, the secondary channel, or both channels
4169 respectively. Disabled by default.
4171 pata_legacy.pio_mask= [HW,LIBATA]
4173 PIO mode mask for autospeed devices. Set individual
4174 bits to allow the use of the respective PIO modes.
4175 Bit 0 is for mode 0, bit 1 is for mode 1, and so on.
4176 All modes allowed by default.
4178 pata_legacy.probe_all= [HW,LIBATA]
4180 Set to non-zero to probe tertiary and further ISA
4181 port ranges on PCI systems. Disabled by default.
4183 pata_legacy.probe_mask= [HW,LIBATA]
4185 Probe mask for legacy ISA PATA ports. Depending on
4186 platform configuration and the use of other driver
4187 options up to 6 legacy ports are supported: 0x1f0,
4188 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing
4189 of individual ports can be disabled by setting the
4190 corresponding bits in the mask to 1. Bit 0 is for
4191 the first port in the list above (0x1f0), and so on.
4192 By default all supported ports are probed.
4194 pata_legacy.qdi= [HW,LIBATA]
4196 Set to non-zero to probe QDI controllers. By default
4197 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise.
4199 pata_legacy.winbond= [HW,LIBATA]
4201 Set to non-zero to probe Winbond controllers. Use
4202 the standard I/O port (0x130) if 1, otherwise the
4203 value given is the I/O port to use (typically 0x1b0).
4204 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE,
4207 pata_platform.pio_mask= [HW,LIBATA]
4209 Supported PIO mode mask. Set individual bits to allow
4210 the use of the respective PIO modes. Bit 0 is for
4211 mode 0, bit 1 is for mode 1, and so on. Mode 0 only
4215 Halt all CPUs after the first oops has been printed for
4216 the specified number of seconds. This is to be used if
4217 your oopses keep scrolling off the screen.
4221 pci=option[,option...] [PCI] various PCI subsystem options.
4223 Some options herein operate on a specific device
4224 or a set of devices (<pci_dev>). These are
4225 specified in one of the following formats:
4227 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
4228 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
4230 Note: the first format specifies a PCI
4231 bus/device/function address which may change
4232 if new hardware is inserted, if motherboard
4233 firmware changes, or due to changes caused
4234 by other kernel parameters. If the
4235 domain is left unspecified, it is
4236 taken to be zero. Optionally, a path
4237 to a device through multiple device/function
4238 addresses can be specified after the base
4239 address (this is more robust against
4240 renumbering issues). The second format
4241 selects devices using IDs from the
4242 configuration space which may match multiple
4243 devices in the system.
4245 earlydump dump PCI config space before the kernel
4247 off [X86] don't probe for the PCI bus
4248 bios [X86-32] force use of PCI BIOS, don't access
4249 the hardware directly. Use this if your machine
4250 has a non-standard PCI host bridge.
4251 nobios [X86-32] disallow use of PCI BIOS, only direct
4252 hardware access methods are allowed. Use this
4253 if you experience crashes upon bootup and you
4254 suspect they are caused by the BIOS.
4255 conf1 [X86] Force use of PCI Configuration Access
4256 Mechanism 1 (config address in IO port 0xCF8,
4257 data in IO port 0xCFC, both 32-bit).
4258 conf2 [X86] Force use of PCI Configuration Access
4259 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
4260 the function, IO port 0xCFA, also 8-bit, sets
4261 bus number. The config space is then accessed
4262 through ports 0xC000-0xCFFF).
4263 See http://wiki.osdev.org/PCI for more info
4264 on the configuration access mechanisms.
4265 noaer [PCIE] If the PCIEAER kernel config parameter is
4266 enabled, this kernel boot option can be used to
4267 disable the use of PCIE advanced error reporting.
4268 nodomains [PCI] Disable support for multiple PCI
4269 root domains (aka PCI segments, in ACPI-speak).
4270 nommconf [X86] Disable use of MMCONFIG for PCI
4272 check_enable_amd_mmconf [X86] check for and enable
4273 properly configured MMIO access to PCI
4274 config space on AMD family 10h CPU
4275 nomsi [MSI] If the PCI_MSI kernel config parameter is
4276 enabled, this kernel boot option can be used to
4277 disable the use of MSI interrupts system-wide.
4278 noioapicquirk [APIC] Disable all boot interrupt quirks.
4279 Safety option to keep boot IRQs enabled. This
4280 should never be necessary.
4281 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
4282 primary IO-APIC for bridges that cannot disable
4283 boot IRQs. This fixes a source of spurious IRQs
4284 when the system masks IRQs.
4285 noioapicreroute [APIC] Disable workaround that uses the
4286 boot IRQ equivalent of an IRQ that connects to
4287 a chipset where boot IRQs cannot be disabled.
4288 The opposite of ioapicreroute.
4289 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
4290 routing table. These calls are known to be buggy
4291 on several machines and they hang the machine
4292 when used, but on other computers it's the only
4293 way to get the interrupt routing table. Try
4294 this option if the kernel is unable to allocate
4295 IRQs or discover secondary PCI buses on your
4297 rom [X86] Assign address space to expansion ROMs.
4298 Use with caution as certain devices share
4299 address decoders between ROMs and other
4301 norom [X86] Do not assign address space to
4302 expansion ROMs that do not already have
4303 BIOS assigned address ranges.
4304 nobar [X86] Do not assign address space to the
4305 BARs that weren't assigned by the BIOS.
4306 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
4307 assigned automatically to PCI devices. You can
4308 make the kernel exclude IRQs of your ISA cards
4310 pirqaddr=0xAAAAA [X86] Specify the physical address
4311 of the PIRQ table (normally generated
4312 by the BIOS) if it is outside the
4313 F0000h-100000h range.
4314 lastbus=N [X86] Scan all buses thru bus #N. Can be
4315 useful if the kernel is unable to find your
4316 secondary buses and you want to tell it
4317 explicitly which ones they are.
4318 assign-busses [X86] Always assign all PCI bus
4319 numbers ourselves, overriding
4320 whatever the firmware may have done.
4321 usepirqmask [X86] Honor the possible IRQ mask stored
4322 in the BIOS $PIR table. This is needed on
4323 some systems with broken BIOSes, notably
4324 some HP Pavilion N5400 and Omnibook XE3
4325 notebooks. This will have no effect if ACPI
4326 IRQ routing is enabled.
4327 noacpi [X86] Do not use ACPI for IRQ routing
4328 or for PCI scanning.
4329 use_crs [X86] Use PCI host bridge window information
4330 from ACPI. On BIOSes from 2008 or later, this
4331 is enabled by default. If you need to use this,
4332 please report a bug.
4333 nocrs [X86] Ignore PCI host bridge windows from ACPI.
4334 If you need to use this, please report a bug.
4335 use_e820 [X86] Use E820 reservations to exclude parts of
4336 PCI host bridge windows. This is a workaround
4337 for BIOS defects in host bridge _CRS methods.
4338 If you need to use this, please report a bug to
4339 <linux-pci@vger.kernel.org>.
4340 no_e820 [X86] Ignore E820 reservations for PCI host
4341 bridge windows. This is the default on modern
4342 hardware. If you need to use this, please report
4343 a bug to <linux-pci@vger.kernel.org>.
4344 routeirq Do IRQ routing for all PCI devices.
4345 This is normally done in pci_enable_device(),
4346 so this option is a temporary workaround
4347 for broken drivers that don't call it.
4348 skip_isa_align [X86] do not align io start addr, so can
4349 handle more pci cards
4350 noearly [X86] Don't do any early type 1 scanning.
4351 This might help on some broken boards which
4352 machine check when some devices' config space
4353 is read. But various workarounds are disabled
4354 and some IOMMU drivers will not work.
4355 bfsort Sort PCI devices into breadth-first order.
4356 This sorting is done to get a device
4357 order compatible with older (<= 2.4) kernels.
4358 nobfsort Don't sort PCI devices into breadth-first order.
4359 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
4360 tuning and use the BIOS-configured MPS defaults.
4361 pcie_bus_safe Set every device's MPS to the largest value
4362 supported by all devices below the root complex.
4363 pcie_bus_perf Set device MPS to the largest allowable MPS
4364 based on its parent bus. Also set MRRS (Max
4365 Read Request Size) to the largest supported
4366 value (no larger than the MPS that the device
4367 or bus can support) for best performance.
4368 pcie_bus_peer2peer Set every device's MPS to 128B, which
4369 every device is guaranteed to support. This
4370 configuration allows peer-to-peer DMA between
4371 any pair of devices, possibly at the cost of
4372 reduced performance. This also guarantees
4373 that hot-added devices will work.
4374 cbiosize=nn[KMG] The fixed amount of bus space which is
4375 reserved for the CardBus bridge's IO window.
4376 The default value is 256 bytes.
4377 cbmemsize=nn[KMG] The fixed amount of bus space which is
4378 reserved for the CardBus bridge's memory
4379 window. The default value is 64 megabytes.
4382 [<order of align>@]<pci_dev>[; ...]
4383 Specifies alignment and device to reassign
4384 aligned memory resources. How to
4385 specify the device is described above.
4386 If <order of align> is not specified,
4387 PAGE_SIZE is used as alignment.
4388 A PCI-PCI bridge can be specified if resource
4389 windows need to be expanded.
4390 To specify the alignment for several
4391 instances of a device, the PCI vendor,
4392 device, subvendor, and subdevice may be
4393 specified, e.g., 12@pci:8086:9c22:103c:198f
4394 for 4096-byte alignment.
4395 ecrc= Enable/disable PCIe ECRC (transaction layer
4396 end-to-end CRC checking). Only effective if
4397 OS has native AER control (either granted by
4398 ACPI _OSC or forced via "pcie_ports=native")
4399 bios: Use BIOS/firmware settings. This is the
4403 hpiosize=nn[KMG] The fixed amount of bus space which is
4404 reserved for hotplug bridge's IO window.
4405 Default size is 256 bytes.
4406 hpmmiosize=nn[KMG] The fixed amount of bus space which is
4407 reserved for hotplug bridge's MMIO window.
4408 Default size is 2 megabytes.
4409 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
4410 reserved for hotplug bridge's MMIO_PREF window.
4411 Default size is 2 megabytes.
4412 hpmemsize=nn[KMG] The fixed amount of bus space which is
4413 reserved for hotplug bridge's MMIO and
4415 Default size is 2 megabytes.
4416 hpbussize=nn The minimum amount of additional bus numbers
4417 reserved for buses below a hotplug bridge.
4419 realloc= Enable/disable reallocating PCI bridge resources
4420 if allocations done by BIOS are too small to
4421 accommodate resources required by all child
4423 off: Turn realloc off
4425 realloc same as realloc=on
4426 noari do not use PCIe ARI.
4427 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
4428 do not use PCIe ATS (and IOMMU device IOTLB).
4429 pcie_scan_all Scan all possible PCIe devices. Otherwise we
4430 only look for one device below a PCIe downstream
4432 big_root_window Try to add a big 64bit memory window to the PCIe
4433 root complex on AMD CPUs. Some GFX hardware
4434 can resize a BAR to allow access to all VRAM.
4435 Adding the window is slightly risky (it may
4436 conflict with unreported devices), so this
4438 disable_acs_redir=<pci_dev>[; ...]
4439 Specify one or more PCI devices (in the format
4440 specified above) separated by semicolons.
4441 Each device specified will have the PCI ACS
4442 redirect capabilities forced off which will
4443 allow P2P traffic between devices through
4444 bridges without forcing it upstream. Note:
4445 this removes isolation between devices and
4446 may put more devices in an IOMMU group.
4447 force_floating [S390] Force usage of floating interrupts.
4448 nomio [S390] Do not use MIO instructions.
4449 norid [S390] ignore the RID field and force use of
4450 one PCI domain per PCI function
4452 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
4455 force Enable ASPM even on devices that claim not to support it.
4456 WARNING: Forcing ASPM on may cause system lockups.
4458 pcie_ports= [PCIE] PCIe port services handling:
4459 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
4460 even if the platform doesn't give the OS permission to
4461 use them. This may cause conflicts if the platform
4462 also tries to use these services.
4463 dpc-native Use native PCIe service for DPC only. May
4464 cause conflicts if firmware uses AER or DPC.
4465 compat Disable native PCIe services (PME, AER, DPC, PCIe
4468 pcie_port_pm= [PCIE] PCIe port power management handling:
4469 off Disable power management of all PCIe ports
4470 force Forcibly enable power management of all PCIe ports
4472 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
4473 nomsi Do not use MSI for native PCIe PME signaling (this makes
4474 all PCIe root ports use INTx for all services).
4476 pcmv= [HW,PCMCIA] BadgePAD 4
4480 Keep all power-domains already enabled by bootloader on,
4481 even if no driver has claimed them. This is useful
4482 for debug and development, but should not be
4483 needed on a platform with proper driver support.
4485 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
4488 See arch/parisc/kernel/pdc_chassis.c
4490 percpu_alloc= Select which percpu first chunk allocator to use.
4491 Currently supported values are "embed" and "page".
4492 Archs may support subset or none of the selections.
4493 See comments in mm/percpu.c for details on each
4494 allocator. This parameter is primarily for debugging
4495 and performance comparison.
4497 pirq= [SMP,APIC] Manual mp-table setup
4498 See Documentation/arch/x86/i386/IO-APIC.rst.
4500 plip= [PPT,NET] Parallel port network link
4501 Format: { parport<nr> | timid | 0 }
4502 See also Documentation/admin-guide/parport.rst.
4504 pmtmr= [X86] Manual setup of pmtmr I/O Port.
4505 Override pmtimer IOPort with a hex value.
4508 pmu_override= [PPC] Override the PMU.
4509 This option takes over the PMU facility, so it is no
4510 longer usable by perf. Setting this option starts the
4511 PMU counters by setting MMCR0 to 0 (the FC bit is
4512 cleared). If a number is given, then MMCR1 is set to
4513 that number, otherwise (e.g., 'pmu_override=on'), MMCR1
4516 pm_debug_messages [SUSPEND,KNL]
4517 Enable suspend/resume debug messages during boot up.
4520 Enable PNP debug messages (depends on the
4521 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
4522 via /sys/module/pnp/parameters/debug. We always show
4523 current resource usage; turning this on also shows
4524 possible settings and some assignment information.
4530 { on | off | curr | res | no-curr | no-res }
4533 [ISAPNP] Exclude IRQs for the autoconfiguration
4536 [ISAPNP] Exclude DMAs for the autoconfiguration
4538 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
4539 Ranges are in pairs (I/O port base and size).
4542 [ISAPNP] Exclude memory regions for the
4544 Ranges are in pairs (memory base and size).
4546 ports= [IP_VS_FTP] IPVS ftp helper module
4548 Up to 8 (IP_VS_APP_MAX_PORTS) ports
4550 Format: <port>,<port>....
4552 powersave=off [PPC] This option disables power saving features.
4553 It specifically disables cpuidle and sets the
4554 platform machine description specific power_save
4555 function to NULL. On Idle the CPU just reduces
4558 ppc_strict_facility_enable
4559 [PPC] This option catches any kernel floating point,
4560 Altivec, VSX and SPE outside of regions specifically
4561 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
4562 There is some performance impact when enabling this.
4566 Disable Hardware Transactional Memory
4569 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC
4570 none - Limited to cond_resched() calls
4571 voluntary - Limited to cond_resched() and might_sleep() calls
4572 full - Any section that isn't explicitly preempt disabled
4573 can be preempted anytime.
4575 print-fatal-signals=
4576 [KNL] debug: print fatal signals
4578 If enabled, warn about various signal handling
4579 related application anomalies: too many signals,
4580 too many POSIX.1 timers, fatal signals causing a
4583 If you hit the warning due to signal overflow,
4584 you might want to try "ulimit -i unlimited".
4588 printk.always_kmsg_dump=
4589 Trigger kmsg_dump for cases other than kernel oops or
4591 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4594 printk.console_no_auto_verbose=
4595 Disable console loglevel raise on oops, panic
4596 or lockdep-detected issues (only if lock debug is on).
4597 With an exception to setups with low baudrate on
4598 serial console, keeping this 0 is a good choice
4599 in order to provide more debug information.
4601 default: 0 (auto_verbose is enabled)
4603 printk.devkmsg={on,off,ratelimit}
4604 Control writing to /dev/kmsg.
4605 on - unlimited logging to /dev/kmsg from userspace
4606 off - logging to /dev/kmsg disabled
4607 ratelimit - ratelimit the logging
4610 printk.time= Show timing data prefixed to each printk message line
4611 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4613 processor.max_cstate= [HW,ACPI]
4614 Limit processor to maximum C-state
4615 max_cstate=9 overrides any DMI blacklist limit.
4617 processor.nocst [HW,ACPI]
4618 Ignore the _CST method to determine C-states,
4619 instead using the legacy FADT method
4621 profile= [KNL] Enable kernel profiling via /proc/profile
4622 Format: [<profiletype>,]<number>
4623 Param: <profiletype>: "schedule", "sleep", or "kvm"
4624 [defaults to kernel profiling]
4625 Param: "schedule" - profile schedule points.
4626 Param: "sleep" - profile D-state sleeping (millisecs).
4627 Requires CONFIG_SCHEDSTATS
4628 Param: "kvm" - profile VM exits.
4629 Param: <number> - step/bucket size as a power of 2 for
4630 statistical time based profiling.
4632 prompt_ramdisk= [RAM] [Deprecated]
4634 prot_virt= [S390] enable hosting protected virtual machines
4635 isolated from the hypervisor (if hardware supports
4639 psi= [KNL] Enable or disable pressure stall information
4643 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
4644 probe for; one of (bare|imps|exps|lifebook|any).
4645 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
4647 psmouse.resetafter= [HW,MOUSE]
4648 Try to reset the device after so many bad packets
4651 [HW,MOUSE] Set desired mouse resolution, in dpi.
4652 psmouse.smartscroll=
4653 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
4654 0 = disabled, 1 = enabled (default).
4656 pstore.backend= Specify the name of the pstore backend to use
4658 pti= [X86-64] Control Page Table Isolation of user and
4659 kernel address spaces. Disabling this feature
4660 removes hardening, but improves performance of
4661 system calls and interrupts.
4663 on - unconditionally enable
4664 off - unconditionally disable
4665 auto - kernel detects whether your CPU model is
4666 vulnerable to issues that PTI mitigates
4668 Not specifying this option is equivalent to pti=auto.
4671 [KNL] Number of legacy pty's. Overwrites compiled-in
4674 quiet [KNL] Disable most log messages
4678 radix_hcall_invalidate=on [PPC/PSERIES]
4679 Disable RADIX GTSE feature and use hcall for TLB
4683 See Documentation/admin-guide/md.rst.
4685 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
4686 See Documentation/admin-guide/blockdev/ramdisk.rst.
4688 ramdisk_start= [RAM] RAM disk image start address
4690 random.trust_cpu=off
4691 [KNL] Disable trusting the use of the CPU's
4692 random number generator (if available) to
4693 initialize the kernel's RNG.
4695 random.trust_bootloader=off
4696 [KNL] Disable trusting the use of the a seed
4697 passed by the bootloader (if available) to
4698 initialize the kernel's RNG.
4700 randomize_kstack_offset=
4701 [KNL] Enable or disable kernel stack offset
4702 randomization, which provides roughly 5 bits of
4703 entropy, frustrating memory corruption attacks
4704 that depend on stack address determinism or
4705 cross-syscall address exposures. This is only
4706 available on architectures that have defined
4707 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET.
4708 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4709 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT.
4711 ras=option[,option,...] [KNL] RAS-specific options
4714 Disable the Correctable Errors Collector,
4715 see CONFIG_RAS_CEC help text.
4717 rcu_nocbs[=cpu-list]
4718 [KNL] The optional argument is a cpu list,
4721 In kernels built with CONFIG_RCU_NOCB_CPU=y,
4722 enable the no-callback CPU mode, which prevents
4723 such CPUs' callbacks from being invoked in
4724 softirq context. Invocation of such CPUs' RCU
4725 callbacks will instead be offloaded to "rcuox/N"
4726 kthreads created for that purpose, where "x" is
4727 "p" for RCU-preempt, "s" for RCU-sched, and "g"
4728 for the kthreads that mediate grace periods; and
4729 "N" is the CPU number. This reduces OS jitter on
4730 the offloaded CPUs, which can be useful for HPC
4731 and real-time workloads. It can also improve
4732 energy efficiency for asymmetric multiprocessors.
4734 If a cpulist is passed as an argument, the specified
4735 list of CPUs is set to no-callback mode from boot.
4737 Otherwise, if the '=' sign and the cpulist
4738 arguments are omitted, no CPU will be set to
4739 no-callback mode from boot but the mode may be
4740 toggled at runtime via cpusets.
4742 Note that this argument takes precedence over
4743 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option.
4746 Rather than requiring that offloaded CPUs
4747 (specified by rcu_nocbs= above) explicitly
4748 awaken the corresponding "rcuoN" kthreads,
4749 make these kthreads poll for callbacks.
4750 This improves the real-time response for the
4751 offloaded CPUs by relieving them of the need to
4752 wake up the corresponding kthread, but degrades
4753 energy efficiency by requiring that the kthreads
4754 periodically wake up to do the polling.
4756 rcutree.blimit= [KNL]
4757 Set maximum number of finished RCU callbacks to
4758 process in one batch.
4760 rcutree.dump_tree= [KNL]
4761 Dump the structure of the rcu_node combining tree
4762 out at early boot. This is used for diagnostic
4763 purposes, to verify correct tree setup.
4765 rcutree.gp_cleanup_delay= [KNL]
4766 Set the number of jiffies to delay each step of
4767 RCU grace-period cleanup.
4769 rcutree.gp_init_delay= [KNL]
4770 Set the number of jiffies to delay each step of
4771 RCU grace-period initialization.
4773 rcutree.gp_preinit_delay= [KNL]
4774 Set the number of jiffies to delay each step of
4775 RCU grace-period pre-initialization, that is,
4776 the propagation of recent CPU-hotplug changes up
4777 the rcu_node combining tree.
4779 rcutree.jiffies_till_first_fqs= [KNL]
4780 Set delay from grace-period initialization to
4781 first attempt to force quiescent states.
4782 Units are jiffies, minimum value is zero,
4783 and maximum value is HZ.
4785 rcutree.jiffies_till_next_fqs= [KNL]
4786 Set delay between subsequent attempts to force
4787 quiescent states. Units are jiffies, minimum
4788 value is one, and maximum value is HZ.
4790 rcutree.jiffies_till_sched_qs= [KNL]
4791 Set required age in jiffies for a
4792 given grace period before RCU starts
4793 soliciting quiescent-state help from
4794 rcu_note_context_switch() and cond_resched().
4795 If not specified, the kernel will calculate
4796 a value based on the most recent settings
4797 of rcutree.jiffies_till_first_fqs
4798 and rcutree.jiffies_till_next_fqs.
4799 This calculated value may be viewed in
4800 rcutree.jiffies_to_sched_qs. Any attempt to set
4801 rcutree.jiffies_to_sched_qs will be cheerfully
4804 rcutree.kthread_prio= [KNL,BOOT]
4805 Set the SCHED_FIFO priority of the RCU per-CPU
4806 kthreads (rcuc/N). This value is also used for
4807 the priority of the RCU boost threads (rcub/N)
4808 and for the RCU grace-period kthreads (rcu_bh,
4809 rcu_preempt, and rcu_sched). If RCU_BOOST is
4810 set, valid values are 1-99 and the default is 1
4811 (the least-favored priority). Otherwise, when
4812 RCU_BOOST is not set, valid values are 0-99 and
4813 the default is zero (non-realtime operation).
4814 When RCU_NOCB_CPU is set, also adjust the
4815 priority of NOCB callback kthreads.
4817 rcutree.nocb_nobypass_lim_per_jiffy= [KNL]
4818 On callback-offloaded (rcu_nocbs) CPUs,
4819 RCU reduces the lock contention that would
4820 otherwise be caused by callback floods through
4821 use of the ->nocb_bypass list. However, in the
4822 common non-flooded case, RCU queues directly to
4823 the main ->cblist in order to avoid the extra
4824 overhead of the ->nocb_bypass list and its lock.
4825 But if there are too many callbacks queued during
4826 a single jiffy, RCU pre-queues the callbacks into
4827 the ->nocb_bypass queue. The definition of "too
4828 many" is supplied by this kernel boot parameter.
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_delay_page_cache_fill_msec= [KNL]
4848 Set the page-cache refill delay (in milliseconds)
4849 in response to low-memory conditions. The range
4850 of permitted values is in the range 0:100000.
4852 rcutree.rcu_divisor= [KNL]
4853 Set the shift-right count to use to compute
4854 the callback-invocation batch limit bl from
4855 the number of callbacks queued on this CPU.
4856 The result will be bounded below by the value of
4857 the rcutree.blimit kernel parameter. Every bl
4858 callbacks, the softirq handler will exit in
4859 order to allow the CPU to do other work.
4861 Please note that this callback-invocation batch
4862 limit applies only to non-offloaded callback
4863 invocation. Offloaded callbacks are instead
4864 invoked in the context of an rcuoc kthread, which
4865 scheduler will preempt as it does any other task.
4867 rcutree.rcu_fanout_exact= [KNL]
4868 Disable autobalancing of the rcu_node combining
4869 tree. This is used by rcutorture, and might
4870 possibly be useful for architectures having high
4871 cache-to-cache transfer latencies.
4873 rcutree.rcu_fanout_leaf= [KNL]
4874 Change the number of CPUs assigned to each
4875 leaf rcu_node structure. Useful for very
4876 large systems, which will choose the value 64,
4877 and for NUMA systems with large remote-access
4878 latencies, which will choose a value aligned
4879 with the appropriate hardware boundaries.
4881 rcutree.rcu_min_cached_objs= [KNL]
4882 Minimum number of objects which are cached and
4883 maintained per one CPU. Object size is equal
4884 to PAGE_SIZE. The cache allows to reduce the
4885 pressure to page allocator, also it makes the
4886 whole algorithm to behave better in low memory
4889 rcutree.rcu_nocb_gp_stride= [KNL]
4890 Set the number of NOCB callback kthreads in
4891 each group, which defaults to the square root
4892 of the number of CPUs. Larger numbers reduce
4893 the wakeup overhead on the global grace-period
4894 kthread, but increases that same overhead on
4895 each group's NOCB grace-period kthread.
4897 rcutree.rcu_kick_kthreads= [KNL]
4898 Cause the grace-period kthread to get an extra
4899 wake_up() if it sleeps three times longer than
4900 it should at force-quiescent-state time.
4901 This wake_up() will be accompanied by a
4902 WARN_ONCE() splat and an ftrace_dump().
4904 rcutree.rcu_resched_ns= [KNL]
4905 Limit the time spend invoking a batch of RCU
4906 callbacks to the specified number of nanoseconds.
4907 By default, this limit is checked only once
4908 every 32 callbacks in order to limit the pain
4909 inflicted by local_clock() overhead.
4911 rcutree.rcu_unlock_delay= [KNL]
4912 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels,
4913 this specifies an rcu_read_unlock()-time delay
4914 in microseconds. This defaults to zero.
4915 Larger delays increase the probability of
4916 catching RCU pointer leaks, that is, buggy use
4917 of RCU-protected pointers after the relevant
4918 rcu_read_unlock() has completed.
4920 rcutree.sysrq_rcu= [KNL]
4921 Commandeer a sysrq key to dump out Tree RCU's
4922 rcu_node tree with an eye towards determining
4923 why a new grace period has not yet started.
4925 rcutree.use_softirq= [KNL]
4926 If set to zero, move all RCU_SOFTIRQ processing to
4927 per-CPU rcuc kthreads. Defaults to a non-zero
4928 value, meaning that RCU_SOFTIRQ is used by default.
4929 Specify rcutree.use_softirq=0 to use rcuc kthreads.
4931 But note that CONFIG_PREEMPT_RT=y kernels disable
4932 this kernel boot parameter, forcibly setting it
4935 rcuscale.gp_async= [KNL]
4936 Measure performance of asynchronous
4937 grace-period primitives such as call_rcu().
4939 rcuscale.gp_async_max= [KNL]
4940 Specify the maximum number of outstanding
4941 callbacks per writer thread. When a writer
4942 thread exceeds this limit, it invokes the
4943 corresponding flavor of rcu_barrier() to allow
4944 previously posted callbacks to drain.
4946 rcuscale.gp_exp= [KNL]
4947 Measure performance of expedited synchronous
4948 grace-period primitives.
4950 rcuscale.holdoff= [KNL]
4951 Set test-start holdoff period. The purpose of
4952 this parameter is to delay the start of the
4953 test until boot completes in order to avoid
4956 rcuscale.kfree_by_call_rcu= [KNL]
4957 In kernels built with CONFIG_RCU_LAZY=y, test
4958 call_rcu() instead of kfree_rcu().
4960 rcuscale.kfree_mult= [KNL]
4961 Instead of allocating an object of size kfree_obj,
4962 allocate one of kfree_mult * sizeof(kfree_obj).
4965 rcuscale.kfree_rcu_test= [KNL]
4966 Set to measure performance of kfree_rcu() flooding.
4968 rcuscale.kfree_rcu_test_double= [KNL]
4969 Test the double-argument variant of kfree_rcu().
4970 If this parameter has the same value as
4971 rcuscale.kfree_rcu_test_single, both the single-
4972 and double-argument variants are tested.
4974 rcuscale.kfree_rcu_test_single= [KNL]
4975 Test the single-argument variant of kfree_rcu().
4976 If this parameter has the same value as
4977 rcuscale.kfree_rcu_test_double, both the single-
4978 and double-argument variants are tested.
4980 rcuscale.kfree_nthreads= [KNL]
4981 The number of threads running loops of kfree_rcu().
4983 rcuscale.kfree_alloc_num= [KNL]
4984 Number of allocations and frees done in an iteration.
4986 rcuscale.kfree_loops= [KNL]
4987 Number of loops doing rcuscale.kfree_alloc_num number
4988 of allocations and frees.
4990 rcuscale.minruntime= [KNL]
4991 Set the minimum test run time in seconds. This
4992 does not affect the data-collection interval,
4993 but instead allows better measurement of things
4994 like CPU consumption.
4996 rcuscale.nreaders= [KNL]
4997 Set number of RCU readers. The value -1 selects
4998 N, where N is the number of CPUs. A value
4999 "n" less than -1 selects N-n+1, where N is again
5000 the number of CPUs. For example, -2 selects N
5001 (the number of CPUs), -3 selects N+1, and so on.
5002 A value of "n" less than or equal to -N selects
5005 rcuscale.nwriters= [KNL]
5006 Set number of RCU writers. The values operate
5007 the same as for rcuscale.nreaders.
5008 N, where N is the number of CPUs
5010 rcuscale.scale_type= [KNL]
5011 Specify the RCU implementation to test.
5013 rcuscale.shutdown= [KNL]
5014 Shut the system down after performance tests
5015 complete. This is useful for hands-off automated
5018 rcuscale.verbose= [KNL]
5019 Enable additional printk() statements.
5021 rcuscale.writer_holdoff= [KNL]
5022 Write-side holdoff between grace periods,
5023 in microseconds. The default of zero says
5026 rcuscale.writer_holdoff_jiffies= [KNL]
5027 Additional write-side holdoff between grace
5028 periods, but in jiffies. The default of zero
5031 rcutorture.fqs_duration= [KNL]
5032 Set duration of force_quiescent_state bursts
5035 rcutorture.fqs_holdoff= [KNL]
5036 Set holdoff time within force_quiescent_state bursts
5039 rcutorture.fqs_stutter= [KNL]
5040 Set wait time between force_quiescent_state bursts
5043 rcutorture.fwd_progress= [KNL]
5044 Specifies the number of kthreads to be used
5045 for RCU grace-period forward-progress testing
5046 for the types of RCU supporting this notion.
5047 Defaults to 1 kthread, values less than zero or
5048 greater than the number of CPUs cause the number
5051 rcutorture.fwd_progress_div= [KNL]
5052 Specify the fraction of a CPU-stall-warning
5053 period to do tight-loop forward-progress testing.
5055 rcutorture.fwd_progress_holdoff= [KNL]
5056 Number of seconds to wait between successive
5057 forward-progress tests.
5059 rcutorture.fwd_progress_need_resched= [KNL]
5060 Enclose cond_resched() calls within checks for
5061 need_resched() during tight-loop forward-progress
5064 rcutorture.gp_cond= [KNL]
5065 Use conditional/asynchronous update-side
5066 primitives, if available.
5068 rcutorture.gp_exp= [KNL]
5069 Use expedited update-side primitives, if available.
5071 rcutorture.gp_normal= [KNL]
5072 Use normal (non-expedited) asynchronous
5073 update-side primitives, if available.
5075 rcutorture.gp_sync= [KNL]
5076 Use normal (non-expedited) synchronous
5077 update-side primitives, if available. If all
5078 of rcutorture.gp_cond=, rcutorture.gp_exp=,
5079 rcutorture.gp_normal=, and rcutorture.gp_sync=
5080 are zero, rcutorture acts as if is interpreted
5081 they are all non-zero.
5083 rcutorture.irqreader= [KNL]
5084 Run RCU readers from irq handlers, or, more
5085 accurately, from a timer handler. Not all RCU
5086 flavors take kindly to this sort of thing.
5088 rcutorture.leakpointer= [KNL]
5089 Leak an RCU-protected pointer out of the reader.
5090 This can of course result in splats, and is
5091 intended to test the ability of things like
5092 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect
5095 rcutorture.n_barrier_cbs= [KNL]
5096 Set callbacks/threads for rcu_barrier() testing.
5098 rcutorture.nfakewriters= [KNL]
5099 Set number of concurrent RCU writers. These just
5100 stress RCU, they don't participate in the actual
5101 test, hence the "fake".
5103 rcutorture.nocbs_nthreads= [KNL]
5104 Set number of RCU callback-offload togglers.
5105 Zero (the default) disables toggling.
5107 rcutorture.nocbs_toggle= [KNL]
5108 Set the delay in milliseconds between successive
5109 callback-offload toggling attempts.
5111 rcutorture.nreaders= [KNL]
5112 Set number of RCU readers. The value -1 selects
5113 N-1, where N is the number of CPUs. A value
5114 "n" less than -1 selects N-n-2, where N is again
5115 the number of CPUs. For example, -2 selects N
5116 (the number of CPUs), -3 selects N+1, and so on.
5118 rcutorture.object_debug= [KNL]
5119 Enable debug-object double-call_rcu() testing.
5121 rcutorture.onoff_holdoff= [KNL]
5122 Set time (s) after boot for CPU-hotplug testing.
5124 rcutorture.onoff_interval= [KNL]
5125 Set time (jiffies) between CPU-hotplug operations,
5126 or zero to disable CPU-hotplug testing.
5128 rcutorture.read_exit= [KNL]
5129 Set the number of read-then-exit kthreads used
5130 to test the interaction of RCU updaters and
5131 task-exit processing.
5133 rcutorture.read_exit_burst= [KNL]
5134 The number of times in a given read-then-exit
5135 episode that a set of read-then-exit kthreads
5138 rcutorture.read_exit_delay= [KNL]
5139 The delay, in seconds, between successive
5140 read-then-exit testing episodes.
5142 rcutorture.shuffle_interval= [KNL]
5143 Set task-shuffle interval (s). Shuffling tasks
5144 allows some CPUs to go into dyntick-idle mode
5145 during the rcutorture test.
5147 rcutorture.shutdown_secs= [KNL]
5148 Set time (s) after boot system shutdown. This
5149 is useful for hands-off automated testing.
5151 rcutorture.stall_cpu= [KNL]
5152 Duration of CPU stall (s) to test RCU CPU stall
5153 warnings, zero to disable.
5155 rcutorture.stall_cpu_block= [KNL]
5156 Sleep while stalling if set. This will result
5157 in warnings from preemptible RCU in addition to
5158 any other stall-related activity. Note that
5159 in kernels built with CONFIG_PREEMPTION=n and
5160 CONFIG_PREEMPT_COUNT=y, this parameter will
5161 cause the CPU to pass through a quiescent state.
5162 Given CONFIG_PREEMPTION=n, this will suppress
5163 RCU CPU stall warnings, but will instead result
5164 in scheduling-while-atomic splats.
5166 Use of this module parameter results in splats.
5169 rcutorture.stall_cpu_holdoff= [KNL]
5170 Time to wait (s) after boot before inducing stall.
5172 rcutorture.stall_cpu_irqsoff= [KNL]
5173 Disable interrupts while stalling if set.
5175 rcutorture.stall_gp_kthread= [KNL]
5176 Duration (s) of forced sleep within RCU
5177 grace-period kthread to test RCU CPU stall
5178 warnings, zero to disable. If both stall_cpu
5179 and stall_gp_kthread are specified, the
5180 kthread is starved first, then the CPU.
5182 rcutorture.stat_interval= [KNL]
5183 Time (s) between statistics printk()s.
5185 rcutorture.stutter= [KNL]
5186 Time (s) to stutter testing, for example, specifying
5187 five seconds causes the test to run for five seconds,
5188 wait for five seconds, and so on. This tests RCU's
5189 ability to transition abruptly to and from idle.
5191 rcutorture.test_boost= [KNL]
5192 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
5193 "Maybe" means test if the RCU implementation
5194 under test support RCU priority boosting.
5196 rcutorture.test_boost_duration= [KNL]
5197 Duration (s) of each individual boost test.
5199 rcutorture.test_boost_interval= [KNL]
5200 Interval (s) between each boost test.
5202 rcutorture.test_no_idle_hz= [KNL]
5203 Test RCU's dyntick-idle handling. See also the
5204 rcutorture.shuffle_interval parameter.
5206 rcutorture.torture_type= [KNL]
5207 Specify the RCU implementation to test.
5209 rcutorture.verbose= [KNL]
5210 Enable additional printk() statements.
5212 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
5213 Dump ftrace buffer after reporting RCU CPU
5216 rcupdate.rcu_cpu_stall_suppress= [KNL]
5217 Suppress RCU CPU stall warning messages.
5219 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL]
5220 Suppress RCU CPU stall warning messages and
5221 rcutorture writer stall warnings that occur
5222 during early boot, that is, during the time
5223 before the init task is spawned.
5225 rcupdate.rcu_cpu_stall_timeout= [KNL]
5226 Set timeout for RCU CPU stall warning messages.
5227 The value is in seconds and the maximum allowed
5228 value is 300 seconds.
5230 rcupdate.rcu_exp_cpu_stall_timeout= [KNL]
5231 Set timeout for expedited RCU CPU stall warning
5232 messages. The value is in milliseconds
5233 and the maximum allowed value is 21000
5234 milliseconds. Please note that this value is
5235 adjusted to an arch timer tick resolution.
5236 Setting this to zero causes the value from
5237 rcupdate.rcu_cpu_stall_timeout to be used (after
5238 conversion from seconds to milliseconds).
5240 rcupdate.rcu_cpu_stall_cputime= [KNL]
5241 Provide statistics on the cputime and count of
5242 interrupts and tasks during the sampling period. For
5243 multiple continuous RCU stalls, all sampling periods
5244 begin at half of the first RCU stall timeout.
5246 rcupdate.rcu_exp_stall_task_details= [KNL]
5247 Print stack dumps of any tasks blocking the
5248 current expedited RCU grace period during an
5249 expedited RCU CPU stall warning.
5251 rcupdate.rcu_expedited= [KNL]
5252 Use expedited grace-period primitives, for
5253 example, synchronize_rcu_expedited() instead
5254 of synchronize_rcu(). This reduces latency,
5255 but can increase CPU utilization, degrade
5256 real-time latency, and degrade energy efficiency.
5257 No effect on CONFIG_TINY_RCU kernels.
5259 rcupdate.rcu_normal= [KNL]
5260 Use only normal grace-period primitives,
5261 for example, synchronize_rcu() instead of
5262 synchronize_rcu_expedited(). This improves
5263 real-time latency, CPU utilization, and
5264 energy efficiency, but can expose users to
5265 increased grace-period latency. This parameter
5266 overrides rcupdate.rcu_expedited. No effect on
5267 CONFIG_TINY_RCU kernels.
5269 rcupdate.rcu_normal_after_boot= [KNL]
5270 Once boot has completed (that is, after
5271 rcu_end_inkernel_boot() has been invoked), use
5272 only normal grace-period primitives. No effect
5273 on CONFIG_TINY_RCU kernels.
5275 But note that CONFIG_PREEMPT_RT=y kernels enables
5276 this kernel boot parameter, forcibly setting
5277 it to the value one, that is, converting any
5278 post-boot attempt at an expedited RCU grace
5279 period to instead use normal non-expedited
5280 grace-period processing.
5282 rcupdate.rcu_task_collapse_lim= [KNL]
5283 Set the maximum number of callbacks present
5284 at the beginning of a grace period that allows
5285 the RCU Tasks flavors to collapse back to using
5286 a single callback queue. This switching only
5287 occurs when rcupdate.rcu_task_enqueue_lim is
5288 set to the default value of -1.
5290 rcupdate.rcu_task_contend_lim= [KNL]
5291 Set the minimum number of callback-queuing-time
5292 lock-contention events per jiffy required to
5293 cause the RCU Tasks flavors to switch to per-CPU
5294 callback queuing. This switching only occurs
5295 when rcupdate.rcu_task_enqueue_lim is set to
5296 the default value of -1.
5298 rcupdate.rcu_task_enqueue_lim= [KNL]
5299 Set the number of callback queues to use for the
5300 RCU Tasks family of RCU flavors. The default
5301 of -1 allows this to be automatically (and
5302 dynamically) adjusted. This parameter is intended
5305 rcupdate.rcu_task_ipi_delay= [KNL]
5306 Set time in jiffies during which RCU tasks will
5307 avoid sending IPIs, starting with the beginning
5308 of a given grace period. Setting a large
5309 number avoids disturbing real-time workloads,
5310 but lengthens grace periods.
5312 rcupdate.rcu_task_lazy_lim= [KNL]
5313 Number of callbacks on a given CPU that will
5314 cancel laziness on that CPU. Use -1 to disable
5315 cancellation of laziness, but be advised that
5316 doing so increases the danger of OOM due to
5319 rcupdate.rcu_task_stall_info= [KNL]
5320 Set initial timeout in jiffies for RCU task stall
5321 informational messages, which give some indication
5322 of the problem for those not patient enough to
5323 wait for ten minutes. Informational messages are
5324 only printed prior to the stall-warning message
5325 for a given grace period. Disable with a value
5326 less than or equal to zero. Defaults to ten
5327 seconds. A change in value does not take effect
5328 until the beginning of the next grace period.
5330 rcupdate.rcu_task_stall_info_mult= [KNL]
5331 Multiplier for time interval between successive
5332 RCU task stall informational messages for a given
5333 RCU tasks grace period. This value is clamped
5334 to one through ten, inclusive. It defaults to
5335 the value three, so that the first informational
5336 message is printed 10 seconds into the grace
5337 period, the second at 40 seconds, the third at
5338 160 seconds, and then the stall warning at 600
5339 seconds would prevent a fourth at 640 seconds.
5341 rcupdate.rcu_task_stall_timeout= [KNL]
5342 Set timeout in jiffies for RCU task stall
5343 warning messages. Disable with a value less
5344 than or equal to zero. Defaults to ten minutes.
5345 A change in value does not take effect until
5346 the beginning of the next grace period.
5348 rcupdate.rcu_tasks_lazy_ms= [KNL]
5349 Set timeout in milliseconds RCU Tasks asynchronous
5350 callback batching for call_rcu_tasks().
5351 A negative value will take the default. A value
5352 of zero will disable batching. Batching is
5353 always disabled for synchronize_rcu_tasks().
5355 rcupdate.rcu_tasks_rude_lazy_ms= [KNL]
5356 Set timeout in milliseconds RCU Tasks
5357 Rude asynchronous callback batching for
5358 call_rcu_tasks_rude(). A negative value
5359 will take the default. A value of zero will
5360 disable batching. Batching is always disabled
5361 for synchronize_rcu_tasks_rude().
5363 rcupdate.rcu_tasks_trace_lazy_ms= [KNL]
5364 Set timeout in milliseconds RCU Tasks
5365 Trace asynchronous callback batching for
5366 call_rcu_tasks_trace(). A negative value
5367 will take the default. A value of zero will
5368 disable batching. Batching is always disabled
5369 for synchronize_rcu_tasks_trace().
5371 rcupdate.rcu_self_test= [KNL]
5372 Run the RCU early boot self tests
5376 Run specified binary instead of /init from the ramdisk,
5377 used for early userspace startup. See initrd.
5380 force - Override the decision by the kernel to hide the
5381 advertisement of RDRAND support (this affects
5382 certain AMD processors because of buggy BIOS
5383 support, specifically around the suspend/resume
5387 Turn on/off individual RDT features. List is:
5388 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
5390 E.g. to turn on cmt and turn off mba use:
5394 Format (x86 or x86_64):
5395 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] | d[efault] \
5397 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
5399 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
5400 (prefix with 'panic_' to set mode for panic
5402 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
5403 reboot_force is either force or not specified,
5404 reboot_cpu is s[mp]#### with #### being the processor
5405 to be used for rebooting.
5407 refscale.holdoff= [KNL]
5408 Set test-start holdoff period. The purpose of
5409 this parameter is to delay the start of the
5410 test until boot completes in order to avoid
5413 refscale.loops= [KNL]
5414 Set the number of loops over the synchronization
5415 primitive under test. Increasing this number
5416 reduces noise due to loop start/end overhead,
5417 but the default has already reduced the per-pass
5418 noise to a handful of picoseconds on ca. 2020
5421 refscale.nreaders= [KNL]
5422 Set number of readers. The default value of -1
5423 selects N, where N is roughly 75% of the number
5424 of CPUs. A value of zero is an interesting choice.
5426 refscale.nruns= [KNL]
5427 Set number of runs, each of which is dumped onto
5430 refscale.readdelay= [KNL]
5431 Set the read-side critical-section duration,
5432 measured in microseconds.
5434 refscale.scale_type= [KNL]
5435 Specify the read-protection implementation to test.
5437 refscale.shutdown= [KNL]
5438 Shut down the system at the end of the performance
5439 test. This defaults to 1 (shut it down) when
5440 refscale is built into the kernel and to 0 (leave
5441 it running) when refscale is built as a module.
5443 refscale.verbose= [KNL]
5444 Enable additional printk() statements.
5446 refscale.verbose_batched= [KNL]
5447 Batch the additional printk() statements. If zero
5448 (the default) or negative, print everything. Otherwise,
5449 print every Nth verbose statement, where N is the value
5453 [KNL, SMP] Set scheduler's default relax_domain_level.
5454 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
5456 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
5457 Format: <base1>,<size1>[,<base2>,<size2>,...]
5458 Reserve I/O ports or memory so the kernel won't use
5459 them. If <base> is less than 0x10000, the region
5460 is assumed to be I/O ports; otherwise it is memory.
5462 reservetop= [X86-32]
5464 Reserves a hole at the top of the kernel virtual
5467 reset_devices [KNL] Force drivers to reset the underlying device
5468 during initialization.
5471 Specify the partition device for software suspend
5473 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
5475 resume_offset= [SWSUSP]
5476 Specify the offset from the beginning of the partition
5477 given by "resume=" at which the swap header is located,
5478 in <PAGE_SIZE> units (needed only for swap files).
5479 See Documentation/power/swsusp-and-swap-files.rst
5481 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
5482 read the resume files
5484 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
5485 Useful for devices that are detected asynchronously
5486 (e.g. USB and MMC devices).
5488 retain_initrd [RAM] Keep initrd memory after extraction
5490 retbleed= [X86] Control mitigation of RETBleed (Arbitrary
5491 Speculative Code Execution with Return Instructions)
5494 AMD-based UNRET and IBPB mitigations alone do not stop
5495 sibling threads from influencing the predictions of other
5496 sibling threads. For that reason, STIBP is used on pro-
5497 cessors that support it, and mitigate SMT on processors
5501 auto - automatically select a migitation
5502 auto,nosmt - automatically select a mitigation,
5503 disabling SMT if necessary for
5504 the full mitigation (only on Zen1
5505 and older without STIBP).
5506 ibpb - On AMD, mitigate short speculation
5507 windows on basic block boundaries too.
5508 Safe, highest perf impact. It also
5509 enables STIBP if present. Not suitable
5511 ibpb,nosmt - Like "ibpb" above but will disable SMT
5512 when STIBP is not available. This is
5513 the alternative for systems which do not
5515 unret - Force enable untrained return thunks,
5516 only effective on AMD f15h-f17h based
5518 unret,nosmt - Like unret, but will disable SMT when STIBP
5519 is not available. This is the alternative for
5520 systems which do not have STIBP.
5522 Selecting 'auto' will choose a mitigation method at run
5523 time according to the CPU.
5525 Not specifying this option is equivalent to retbleed=auto.
5527 rfkill.default_state=
5528 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
5529 etc. communication is blocked by default.
5532 rfkill.master_switch_mode=
5533 0 The "airplane mode" button does nothing.
5534 1 The "airplane mode" button toggles between everything
5535 blocked and the previous configuration.
5536 2 The "airplane mode" button toggles between everything
5537 blocked and everything unblocked.
5539 rhash_entries= [KNL,NET]
5540 Set number of hash buckets for route cache
5543 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
5546 ro [KNL] Mount root device read-only on boot
5549 on Mark read-only kernel memory as read-only (default).
5550 off Leave read-only kernel memory writable for debugging.
5551 full Mark read-only kernel memory and aliases as read-only
5555 Enable the uart passthrough on the designated usb port
5556 on Rockchip SoCs. When active, the signals of the
5557 debug-uart get routed to the D+ and D- pins of the usb
5558 port and the regular usb controller gets disabled.
5560 root= [KNL] Root filesystem
5561 Usually this a a block device specifier of some kind,
5562 see the early_lookup_bdev comment in
5563 block/early-lookup.c for details.
5564 Alternatively this can be "ram" for the legacy initial
5565 ramdisk, "nfs" and "cifs" for root on a network file
5566 system, or "mtd" and "ubi" for mounting from raw flash.
5568 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
5569 mount the root filesystem
5571 rootflags= [KNL] Set root filesystem mount option string
5573 rootfstype= [KNL] Set root filesystem type
5575 rootwait [KNL] Wait (indefinitely) for root device to show up.
5576 Useful for devices that are detected asynchronously
5577 (e.g. USB and MMC devices).
5579 rootwait= [KNL] Maximum time (in seconds) to wait for root device
5580 to show up before attempting to mount the root
5583 rproc_mem=nn[KMG][@address]
5584 [KNL,ARM,CMA] Remoteproc physical memory block.
5585 Memory area to be used by remote processor image,
5588 rw [KNL] Mount root device read-write on boot
5590 S [KNL] Run init in single mode
5592 s390_iommu= [HW,S390]
5593 Set s390 IOTLB flushing mode
5595 With strict flushing every unmap operation will result in
5596 an IOTLB flush. Default is lazy flushing before reuse,
5599 s390_iommu_aperture= [KNL,S390]
5600 Specifies the size of the per device DMA address space
5601 accessible through the DMA and IOMMU APIs as a decimal
5602 factor of the size of main memory.
5603 The default is 1 meaning that one can concurrently use
5604 as many DMA addresses as physical memory is installed,
5605 if supported by hardware, and thus map all of memory
5606 once. With a value of 2 one can map all of memory twice
5607 and so on. As a special case a factor of 0 imposes no
5608 restrictions other than those given by hardware at the
5609 cost of significant additional memory use for tables.
5612 See drivers/net/irda/sa1100_ir.c.
5614 sched_verbose [KNL] Enables verbose scheduler debug messages.
5616 schedstats= [KNL,X86] Enable or disable scheduled statistics.
5617 Allowed values are enable and disable. This feature
5618 incurs a small amount of overhead in the scheduler
5619 but is useful for debugging and performance tuning.
5621 sched_thermal_decay_shift=
5622 [KNL, SMP] Set a decay shift for scheduler thermal
5623 pressure signal. Thermal pressure signal follows the
5624 default decay period of other scheduler pelt
5625 signals(usually 32 ms but configurable). Setting
5626 sched_thermal_decay_shift will left shift the decay
5627 period for the thermal pressure signal by the shift
5629 i.e. with the default pelt decay period of 32 ms
5630 sched_thermal_decay_shift thermal pressure decay pr
5634 Format: integer between 0 and 10
5637 scftorture.holdoff= [KNL]
5638 Number of seconds to hold off before starting
5639 test. Defaults to zero for module insertion and
5640 to 10 seconds for built-in smp_call_function()
5643 scftorture.longwait= [KNL]
5644 Request ridiculously long waits randomly selected
5645 up to the chosen limit in seconds. Zero (the
5646 default) disables this feature. Please note
5647 that requesting even small non-zero numbers of
5648 seconds can result in RCU CPU stall warnings,
5649 softlockup complaints, and so on.
5651 scftorture.nthreads= [KNL]
5652 Number of kthreads to spawn to invoke the
5653 smp_call_function() family of functions.
5654 The default of -1 specifies a number of kthreads
5655 equal to the number of CPUs.
5657 scftorture.onoff_holdoff= [KNL]
5658 Number seconds to wait after the start of the
5659 test before initiating CPU-hotplug operations.
5661 scftorture.onoff_interval= [KNL]
5662 Number seconds to wait between successive
5663 CPU-hotplug operations. Specifying zero (which
5664 is the default) disables CPU-hotplug operations.
5666 scftorture.shutdown_secs= [KNL]
5667 The number of seconds following the start of the
5668 test after which to shut down the system. The
5669 default of zero avoids shutting down the system.
5670 Non-zero values are useful for automated tests.
5672 scftorture.stat_interval= [KNL]
5673 The number of seconds between outputting the
5674 current test statistics to the console. A value
5675 of zero disables statistics output.
5677 scftorture.stutter_cpus= [KNL]
5678 The number of jiffies to wait between each change
5679 to the set of CPUs under test.
5681 scftorture.use_cpus_read_lock= [KNL]
5682 Use use_cpus_read_lock() instead of the default
5683 preempt_disable() to disable CPU hotplug
5684 while invoking one of the smp_call_function*()
5687 scftorture.verbose= [KNL]
5688 Enable additional printk() statements.
5690 scftorture.weight_single= [KNL]
5691 The probability weighting to use for the
5692 smp_call_function_single() function with a zero
5693 "wait" parameter. A value of -1 selects the
5694 default if all other weights are -1. However,
5695 if at least one weight has some other value, a
5696 value of -1 will instead select a weight of zero.
5698 scftorture.weight_single_wait= [KNL]
5699 The probability weighting to use for the
5700 smp_call_function_single() function with a
5701 non-zero "wait" parameter. See weight_single.
5703 scftorture.weight_many= [KNL]
5704 The probability weighting to use for the
5705 smp_call_function_many() function with a zero
5706 "wait" parameter. See weight_single.
5707 Note well that setting a high probability for
5708 this weighting can place serious IPI load
5711 scftorture.weight_many_wait= [KNL]
5712 The probability weighting to use for the
5713 smp_call_function_many() function with a
5714 non-zero "wait" parameter. See weight_single
5717 scftorture.weight_all= [KNL]
5718 The probability weighting to use for the
5719 smp_call_function_all() function with a zero
5720 "wait" parameter. See weight_single and
5723 scftorture.weight_all_wait= [KNL]
5724 The probability weighting to use for the
5725 smp_call_function_all() function with a
5726 non-zero "wait" parameter. See weight_single
5729 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
5730 xtime_lock contention on larger systems, and/or RCU lock
5731 contention on all systems with CONFIG_MAXSMP set.
5732 Format: { "0" | "1" }
5733 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
5735 Note: increases power consumption, thus should only be
5736 enabled if running jitter sensitive (HPC/RT) workloads.
5738 security= [SECURITY] Choose a legacy "major" security module to
5739 enable at boot. This has been deprecated by the
5742 selinux= [SELINUX] Disable or enable SELinux at boot time.
5743 Format: { "0" | "1" }
5744 See security/selinux/Kconfig help text.
5749 serialnumber [BUGS=X86-32]
5751 sev=option[,option...] [X86-64] See Documentation/arch/x86/x86_64/boot-options.rst
5754 Maximal number of shapers.
5756 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
5757 Limit apic dumping. The parameter defines the maximal
5758 number of local apics being dumped. Also it is possible
5759 to set it to "all" by meaning -- no limit here.
5760 Format: { 1 (default) | 2 | ... | all }.
5761 The parameter valid if only apic=debug or
5762 apic=verbose is specified.
5763 Example: apic=debug show_lapic=all
5771 Enable merging of slabs with similar size when the
5772 kernel is built without CONFIG_SLAB_MERGE_DEFAULT.
5775 Disable merging of slabs with similar size. May be
5776 necessary if there is some reason to distinguish
5777 allocs to different slabs, especially in hardened
5778 environments where the risk of heap overflows and
5779 layout control by attackers can usually be
5780 frustrated by disabling merging. This will reduce
5781 most of the exposure of a heap attack to a single
5782 cache (risks via metadata attacks are mostly
5783 unchanged). Debug options disable merging on their
5785 For more information see Documentation/mm/slub.rst.
5787 slab_max_order= [MM, SLAB]
5788 Determines the maximum allowed order for slabs.
5789 A high setting may cause OOMs due to memory
5790 fragmentation. Defaults to 1 for systems with
5791 more than 32MB of RAM, 0 otherwise.
5793 slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB]
5794 Enabling slub_debug allows one to determine the
5795 culprit if slab objects become corrupted. Enabling
5796 slub_debug can create guard zones around objects and
5797 may poison objects when not in use. Also tracks the
5798 last alloc / free. For more information see
5799 Documentation/mm/slub.rst.
5801 slub_max_order= [MM, SLUB]
5802 Determines the maximum allowed order for slabs.
5803 A high setting may cause OOMs due to memory
5804 fragmentation. For more information see
5805 Documentation/mm/slub.rst.
5807 slub_min_objects= [MM, SLUB]
5808 The minimum number of objects per slab. SLUB will
5809 increase the slab order up to slub_max_order to
5810 generate a sufficiently large slab able to contain
5811 the number of objects indicated. The higher the number
5812 of objects the smaller the overhead of tracking slabs
5813 and the less frequently locks need to be acquired.
5814 For more information see Documentation/mm/slub.rst.
5816 slub_min_order= [MM, SLUB]
5817 Determines the minimum page order for slabs. Must be
5818 lower than slub_max_order.
5819 For more information see Documentation/mm/slub.rst.
5821 slub_merge [MM, SLUB]
5822 Same with slab_merge.
5824 slub_nomerge [MM, SLUB]
5825 Same with slab_nomerge. This is supported for legacy.
5826 See slab_nomerge for more information.
5829 Format: <io1>[,<io2>[,...,<io8>]]
5831 smp.csd_lock_timeout= [KNL]
5832 Specify the period of time in milliseconds
5833 that smp_call_function() and friends will wait
5834 for a CPU to release the CSD lock. This is
5835 useful when diagnosing bugs involving CPUs
5836 disabling interrupts for extended periods
5837 of time. Defaults to 5,000 milliseconds, and
5838 setting a value of zero disables this feature.
5839 This feature may be more efficiently disabled
5840 using the csdlock_debug- kernel parameter.
5842 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
5843 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
5844 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
5845 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
5846 smsc-ircc2.ircc_irq= [HW] IRQ line
5847 smsc-ircc2.ircc_dma= [HW] DMA channel
5848 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
5849 0: Toshiba Satellite 1800 (GP data pin select)
5850 1: Fast pin select (default)
5853 smt= [KNL,MIPS,S390] Set the maximum number of threads (logical
5854 CPUs) to use per physical CPU on systems capable of
5855 symmetric multithreading (SMT). Will be capped to the
5856 actual hardware limit.
5858 Default: -1 (no limit)
5861 [KNL] Should the soft-lockup detector generate panics.
5864 A value of 1 instructs the soft-lockup detector
5865 to panic the machine when a soft-lockup occurs. It is
5866 also controlled by the kernel.softlockup_panic sysctl
5867 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
5868 respective build-time switch to that functionality.
5870 softlockup_all_cpu_backtrace=
5871 [KNL] Should the soft-lockup detector generate
5872 backtraces on all cpus.
5875 sonypi.*= [HW] Sony Programmable I/O Control Device driver
5876 See Documentation/admin-guide/laptops/sonypi.rst
5878 spectre_v2= [X86] Control mitigation of Spectre variant 2
5879 (indirect branch speculation) vulnerability.
5880 The default operation protects the kernel from
5883 on - unconditionally enable, implies
5885 off - unconditionally disable, implies
5887 auto - kernel detects whether your CPU model is
5890 Selecting 'on' will, and 'auto' may, choose a
5891 mitigation method at run time according to the
5892 CPU, the available microcode, the setting of the
5893 CONFIG_RETPOLINE configuration option, and the
5894 compiler with which the kernel was built.
5896 Selecting 'on' will also enable the mitigation
5897 against user space to user space task attacks.
5899 Selecting 'off' will disable both the kernel and
5900 the user space protections.
5902 Specific mitigations can also be selected manually:
5904 retpoline - replace indirect branches
5905 retpoline,generic - Retpolines
5906 retpoline,lfence - LFENCE; indirect branch
5907 retpoline,amd - alias for retpoline,lfence
5908 eibrs - Enhanced/Auto IBRS
5909 eibrs,retpoline - Enhanced/Auto IBRS + Retpolines
5910 eibrs,lfence - Enhanced/Auto IBRS + LFENCE
5911 ibrs - use IBRS to protect kernel
5913 Not specifying this option is equivalent to
5917 [X86] Control mitigation of Spectre variant 2
5918 (indirect branch speculation) vulnerability between
5921 on - Unconditionally enable mitigations. Is
5922 enforced by spectre_v2=on
5924 off - Unconditionally disable mitigations. Is
5925 enforced by spectre_v2=off
5927 prctl - Indirect branch speculation is enabled,
5928 but mitigation can be enabled via prctl
5929 per thread. The mitigation control state
5930 is inherited on fork.
5933 - Like "prctl" above, but only STIBP is
5934 controlled per thread. IBPB is issued
5935 always when switching between different user
5939 - Same as "prctl" above, but all seccomp
5940 threads will enable the mitigation unless
5941 they explicitly opt out.
5944 - Like "seccomp" above, but only STIBP is
5945 controlled per thread. IBPB is issued
5946 always when switching between different
5947 user space processes.
5949 auto - Kernel selects the mitigation depending on
5950 the available CPU features and vulnerability.
5952 Default mitigation: "prctl"
5954 Not specifying this option is equivalent to
5955 spectre_v2_user=auto.
5957 spec_rstack_overflow=
5958 [X86] Control RAS overflow mitigation on AMD Zen CPUs
5960 off - Disable mitigation
5961 microcode - Enable microcode mitigation only
5962 safe-ret - Enable sw-only safe RET mitigation (default)
5963 ibpb - Enable mitigation by issuing IBPB on
5965 ibpb-vmexit - Issue IBPB only on VMEXIT
5966 (cloud-specific mitigation)
5968 spec_store_bypass_disable=
5969 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
5970 (Speculative Store Bypass vulnerability)
5972 Certain CPUs are vulnerable to an exploit against a
5973 a common industry wide performance optimization known
5974 as "Speculative Store Bypass" in which recent stores
5975 to the same memory location may not be observed by
5976 later loads during speculative execution. The idea
5977 is that such stores are unlikely and that they can
5978 be detected prior to instruction retirement at the
5979 end of a particular speculation execution window.
5981 In vulnerable processors, the speculatively forwarded
5982 store can be used in a cache side channel attack, for
5983 example to read memory to which the attacker does not
5984 directly have access (e.g. inside sandboxed code).
5986 This parameter controls whether the Speculative Store
5987 Bypass optimization is used.
5989 On x86 the options are:
5991 on - Unconditionally disable Speculative Store Bypass
5992 off - Unconditionally enable Speculative Store Bypass
5993 auto - Kernel detects whether the CPU model contains an
5994 implementation of Speculative Store Bypass and
5995 picks the most appropriate mitigation. If the
5996 CPU is not vulnerable, "off" is selected. If the
5997 CPU is vulnerable the default mitigation is
5998 architecture and Kconfig dependent. See below.
5999 prctl - Control Speculative Store Bypass per thread
6000 via prctl. Speculative Store Bypass is enabled
6001 for a process by default. The state of the control
6002 is inherited on fork.
6003 seccomp - Same as "prctl" above, but all seccomp threads
6004 will disable SSB unless they explicitly opt out.
6006 Default mitigations:
6009 On powerpc the options are:
6011 on,auto - On Power8 and Power9 insert a store-forwarding
6012 barrier on kernel entry and exit. On Power7
6013 perform a software flush on kernel entry and
6017 Not specifying this option is equivalent to
6018 spec_store_bypass_disable=auto.
6020 spia_io_base= [HW,MTD]
6026 [X86] Enable split lock detection or bus lock detection
6028 When enabled (and if hardware support is present), atomic
6029 instructions that access data across cache line
6030 boundaries will result in an alignment check exception
6031 for split lock detection or a debug exception for
6036 warn - the kernel will emit rate-limited warnings
6037 about applications triggering the #AC
6038 exception or the #DB exception. This mode is
6039 the default on CPUs that support split lock
6040 detection or bus lock detection. Default
6041 behavior is by #AC if both features are
6042 enabled in hardware.
6044 fatal - the kernel will send SIGBUS to applications
6045 that trigger the #AC exception or the #DB
6046 exception. Default behavior is by #AC if
6047 both features are enabled in hardware.
6050 Set system wide rate limit to N bus locks
6051 per second for bus lock detection.
6054 N/A for split lock detection.
6057 If an #AC exception is hit in the kernel or in
6058 firmware (i.e. not while executing in user mode)
6059 the kernel will oops in either "warn" or "fatal"
6062 #DB exception for bus lock is triggered only when
6066 Control the Special Register Buffer Data Sampling
6069 Certain CPUs are vulnerable to an MDS-like
6070 exploit which can leak bits from the random
6073 By default, this issue is mitigated by
6074 microcode. However, the microcode fix can cause
6075 the RDRAND and RDSEED instructions to become
6076 much slower. Among other effects, this will
6077 result in reduced throughput from /dev/urandom.
6079 The microcode mitigation can be disabled with
6080 the following option:
6082 off: Disable mitigation and remove
6083 performance impact to RDRAND and RDSEED
6085 srcutree.big_cpu_lim [KNL]
6086 Specifies the number of CPUs constituting a
6087 large system, such that srcu_struct structures
6088 should immediately allocate an srcu_node array.
6089 This kernel-boot parameter defaults to 128,
6090 but takes effect only when the low-order four
6091 bits of srcutree.convert_to_big is equal to 3
6094 srcutree.convert_to_big [KNL]
6095 Specifies under what conditions an SRCU tree
6096 srcu_struct structure will be converted to big
6097 form, that is, with an rcu_node tree:
6100 1: At init_srcu_struct() time.
6101 2: When rcutorture decides to.
6102 3: Decide at boot time (default).
6103 0x1X: Above plus if high contention.
6105 Either way, the srcu_node tree will be sized based
6106 on the actual runtime number of CPUs (nr_cpu_ids)
6107 instead of the compile-time CONFIG_NR_CPUS.
6109 srcutree.counter_wrap_check [KNL]
6110 Specifies how frequently to check for
6111 grace-period sequence counter wrap for the
6112 srcu_data structure's ->srcu_gp_seq_needed field.
6113 The greater the number of bits set in this kernel
6114 parameter, the less frequently counter wrap will
6115 be checked for. Note that the bottom two bits
6118 srcutree.exp_holdoff [KNL]
6119 Specifies how many nanoseconds must elapse
6120 since the end of the last SRCU grace period for
6121 a given srcu_struct until the next normal SRCU
6122 grace period will be considered for automatic
6123 expediting. Set to zero to disable automatic
6126 srcutree.srcu_max_nodelay [KNL]
6127 Specifies the number of no-delay instances
6128 per jiffy for which the SRCU grace period
6129 worker thread will be rescheduled with zero
6130 delay. Beyond this limit, worker thread will
6131 be rescheduled with a sleep delay of one jiffy.
6133 srcutree.srcu_max_nodelay_phase [KNL]
6134 Specifies the per-grace-period phase, number of
6135 non-sleeping polls of readers. Beyond this limit,
6136 grace period worker thread will be rescheduled
6137 with a sleep delay of one jiffy, between each
6138 rescan of the readers, for a grace period phase.
6140 srcutree.srcu_retry_check_delay [KNL]
6141 Specifies number of microseconds of non-sleeping
6142 delay between each non-sleeping poll of readers.
6144 srcutree.small_contention_lim [KNL]
6145 Specifies the number of update-side contention
6146 events per jiffy will be tolerated before
6147 initiating a conversion of an srcu_struct
6148 structure to big form. Note that the value of
6149 srcutree.convert_to_big must have the 0x10 bit
6150 set for contention-based conversions to occur.
6153 Speculative Store Bypass Disable control
6155 On CPUs that are vulnerable to the Speculative
6156 Store Bypass vulnerability and offer a
6157 firmware based mitigation, this parameter
6158 indicates how the mitigation should be used:
6160 force-on: Unconditionally enable mitigation for
6161 for both kernel and userspace
6162 force-off: Unconditionally disable mitigation for
6163 for both kernel and userspace
6164 kernel: Always enable mitigation in the
6165 kernel, and offer a prctl interface
6166 to allow userspace to register its
6167 interest in being mitigated too.
6169 stack_guard_gap= [MM]
6170 override the default stack gap protection. The value
6171 is in page units and it defines how many pages prior
6172 to (for stacks growing down) resp. after (for stacks
6173 growing up) the main stack are reserved for no other
6174 mapping. Default value is 256 pages.
6176 stack_depot_disable= [KNL]
6177 Setting this to true through kernel command line will
6178 disable the stack depot thereby saving the static memory
6179 consumed by the stack hash table. By default this is set
6183 Enabled the stack tracer on boot up.
6185 stacktrace_filter=[function-list]
6186 [FTRACE] Limit the functions that the stack tracer
6187 will trace at boot up. function-list is a comma-separated
6188 list of functions. This list can be changed at run
6189 time by the stack_trace_filter file in the debugfs
6190 tracing directory. Note, this enables stack tracing
6191 and the stacktrace above is not needed.
6195 Set the STI (builtin display/keyboard on the HP-PARISC
6196 machines) console (graphic card) which should be used
6197 as the initial boot-console.
6198 See also comment in drivers/video/console/sticore.c.
6201 See comment in drivers/video/console/sticore.c.
6204 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
6209 Enable or disable strict sigaltstack size checks
6210 against the required signal frame size which
6211 depends on the supported FPU features. This can
6212 be used to filter out binaries which have
6213 not yet been made aware of AT_MINSIGSTKSZ.
6216 Limits the number of kernel HPT entries in the hash
6217 page table to increase the rate of hash page table
6218 faults on kernel addresses.
6221 Limits the number of kernel SLB entries, and flushes
6222 them frequently to increase the rate of SLB faults
6223 on kernel addresses.
6225 sunrpc.min_resvport=
6226 sunrpc.max_resvport=
6228 SunRPC servers often require that client requests
6229 originate from a privileged port (i.e. a port in the
6230 range 0 < portnr < 1024).
6231 An administrator who wishes to reserve some of these
6232 ports for other uses may adjust the range that the
6233 kernel's sunrpc client considers to be privileged
6234 using these two parameters to set the minimum and
6235 maximum port values.
6237 sunrpc.svc_rpc_per_connection_limit=
6239 Limit the number of requests that the server will
6240 process in parallel from a single connection.
6241 The default value is 0 (no limit).
6245 Control how the NFS server code allocates CPUs to
6246 service thread pools. Depending on how many NICs
6247 you have and where their interrupts are bound, this
6248 option will affect which CPUs will do NFS serving.
6249 Note: this parameter cannot be changed while the
6250 NFS server is running.
6252 auto the server chooses an appropriate mode
6253 automatically using heuristics
6254 global a single global pool contains all CPUs
6255 percpu one pool for each CPU
6256 pernode one pool for each NUMA node (equivalent
6257 to global on non-NUMA machines)
6259 sunrpc.tcp_slot_table_entries=
6260 sunrpc.udp_slot_table_entries=
6262 Sets the upper limit on the number of simultaneous
6263 RPC calls that can be sent from the client to a
6264 server. Increasing these values may allow you to
6265 improve throughput, but will also increase the
6266 amount of memory reserved for use by the client.
6268 suspend.pm_test_delay=
6270 Sets the number of seconds to remain in a suspend test
6271 mode before resuming the system (see
6272 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
6273 is set. Default value is 5.
6276 Format: { on | off | y | n | 1 | 0 }
6277 This parameter controls use of the Protected
6278 Execution Facility on pSeries.
6280 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
6281 Format: { <int> [,<int>] | force | noforce }
6282 <int> -- Number of I/O TLB slabs
6283 <int> -- Second integer after comma. Number of swiotlb
6284 areas with their own lock. Will be rounded up
6286 force -- force using of bounce buffers even if they
6287 wouldn't be automatically used by the kernel
6288 noforce -- Never use bounce buffers (for debugging)
6293 Set a sysctl parameter, right before loading the init
6294 process, as if the value was written to the respective
6295 /proc/sys/... file. Both '.' and '/' are recognized as
6296 separators. Unrecognized parameters and invalid values
6297 are reported in the kernel log. Sysctls registered
6298 later by a loaded module cannot be set this way.
6299 Example: sysctl.vm.swappiness=40
6301 sysrq_always_enabled
6303 Ignore sysrq setting - this boot parameter will
6304 neutralize any effect of /proc/sys/kernel/sysrq.
6305 Useful for debugging.
6307 tcpmhash_entries= [KNL,NET]
6308 Set the number of tcp_metrics_hash slots.
6309 Default value is 8192 or 16384 depending on total
6310 ram pages. This is used to specify the TCP metrics
6311 cache size. See Documentation/networking/ip-sysctl.rst
6312 "tcp_no_metrics_save" section for more details.
6316 test_suspend= [SUSPEND]
6317 Format: { "mem" | "standby" | "freeze" }[,N]
6318 Specify "mem" (for Suspend-to-RAM) or "standby" (for
6319 standby suspend) or "freeze" (for suspend type freeze)
6320 as the system sleep state during system startup with
6321 the optional capability to repeat N number of times.
6322 The system is woken from this state using a
6323 wakeup-capable RTC alarm.
6325 thash_entries= [KNL,NET]
6326 Set number of hash buckets for TCP connection
6328 thermal.act= [HW,ACPI]
6329 -1: disable all active trip points in all thermal zones
6330 <degrees C>: override all lowest active trip points
6332 thermal.crt= [HW,ACPI]
6333 -1: disable all critical trip points in all thermal zones
6334 <degrees C>: override all critical trip points
6336 thermal.off= [HW,ACPI]
6337 1: disable ACPI thermal control
6339 thermal.psv= [HW,ACPI]
6340 -1: disable all passive trip points
6341 <degrees C>: override all passive trip points to this
6344 thermal.tzp= [HW,ACPI]
6345 Specify global default ACPI thermal zone polling rate
6346 <deci-seconds>: poll all this frequency
6347 0: no polling (default)
6350 Force threading of all interrupt handlers except those
6351 marked explicitly IRQF_NO_THREAD.
6355 Specify if the kernel should make use of the cpu
6356 topology information if the hardware supports this.
6357 The scheduler will make use of this information and
6358 e.g. base its process migration decisions on it.
6361 topology_updates= [KNL, PPC, NUMA]
6363 Specify if the kernel should ignore (off)
6364 topology updates sent by the hypervisor to this
6367 torture.disable_onoff_at_boot= [KNL]
6368 Prevent the CPU-hotplug component of torturing
6369 until after init has spawned.
6371 torture.ftrace_dump_at_shutdown= [KNL]
6372 Dump the ftrace buffer at torture-test shutdown,
6373 even if there were no errors. This can be a
6374 very costly operation when many torture tests
6375 are running concurrently, especially on systems
6376 with rotating-rust storage.
6378 torture.verbose_sleep_frequency= [KNL]
6379 Specifies how many verbose printk()s should be
6380 emitted between each sleep. The default of zero
6381 disables verbose-printk() sleeping.
6383 torture.verbose_sleep_duration= [KNL]
6384 Duration of each verbose-printk() sleep in jiffies.
6388 tpm_suspend_pcr=[HW,TPM]
6389 Format: integer pcr id
6390 Specify that at suspend time, the tpm driver
6391 should extend the specified pcr with zeros,
6392 as a workaround for some chips which fail to
6393 flush the last written pcr on TPM_SaveState.
6394 This will guarantee that all the other pcrs
6397 tpm_tis.interrupts= [HW,TPM]
6398 Enable interrupts for the MMIO based physical layer
6399 for the FIFO interface. By default it is set to false
6400 (0). For more information about TPM hardware interfaces
6401 defined by Trusted Computing Group (TCG) see
6402 https://trustedcomputinggroup.org/resource/pc-client-platform-tpm-profile-ptp-specification/
6405 Have the tracepoints sent to printk as well as the
6406 tracing ring buffer. This is useful for early boot up
6407 where the system hangs or reboots and does not give the
6408 option for reading the tracing buffer or performing a
6409 ftrace_dump_on_oops.
6411 To turn off having tracepoints sent to printk,
6412 echo 0 > /proc/sys/kernel/tracepoint_printk
6413 Note, echoing 1 into this file without the
6414 tracepoint_printk kernel cmdline option has no effect.
6416 The tp_printk_stop_on_boot (see below) can also be used
6417 to stop the printing of events to console at
6422 Having tracepoints sent to printk() and activating high
6423 frequency tracepoints such as irq or sched, can cause
6424 the system to live lock.
6426 tp_printk_stop_on_boot [FTRACE]
6427 When tp_printk (above) is set, it can cause a lot of noise
6428 on the console. It may be useful to only include the
6429 printing of events during boot up, as user space may
6430 make the system inoperable.
6432 This command line option will stop the printing of events
6433 to console at the late_initcall_sync() time frame.
6435 trace_buf_size=nn[KMG]
6436 [FTRACE] will set tracing buffer size on each cpu.
6438 trace_clock= [FTRACE] Set the clock used for tracing events
6440 local - Use the per CPU time stamp counter
6441 (converted into nanoseconds). Fast, but
6442 depending on the architecture, may not be
6443 in sync between CPUs.
6444 global - Event time stamps are synchronize across
6445 CPUs. May be slower than the local clock,
6446 but better for some race conditions.
6447 counter - Simple counting of events (1, 2, ..)
6448 note, some counts may be skipped due to the
6449 infrastructure grabbing the clock more than
6451 uptime - Use jiffies as the time stamp.
6452 perf - Use the same clock that perf uses.
6453 mono - Use ktime_get_mono_fast_ns() for time stamps.
6454 mono_raw - Use ktime_get_raw_fast_ns() for time
6456 boot - Use ktime_get_boot_fast_ns() for time stamps.
6457 Architectures may add more clocks. See
6458 Documentation/trace/ftrace.rst for more details.
6460 trace_event=[event-list]
6461 [FTRACE] Set and start specified trace events in order
6462 to facilitate early boot debugging. The event-list is a
6463 comma-separated list of trace events to enable. See
6464 also Documentation/trace/events.rst
6466 trace_instance=[instance-info]
6467 [FTRACE] Create a ring buffer instance early in boot up.
6468 This will be listed in:
6470 /sys/kernel/tracing/instances
6472 Events can be enabled at the time the instance is created
6475 trace_instance=<name>,<system1>:<event1>,<system2>:<event2>
6477 Note, the "<system*>:" portion is optional if the event is
6480 trace_instance=foo,sched:sched_switch,irq_handler_entry,initcall
6482 will enable the "sched_switch" event (note, the "sched:" is optional, and
6483 the same thing would happen if it was left off). The irq_handler_entry
6484 event, and all events under the "initcall" system.
6486 trace_options=[option-list]
6487 [FTRACE] Enable or disable tracer options at boot.
6488 The option-list is a comma delimited list of options
6489 that can be enabled or disabled just as if you were
6490 to echo the option name into
6492 /sys/kernel/tracing/trace_options
6494 For example, to enable stacktrace option (to dump the
6495 stack trace of each event), add to the command line:
6497 trace_options=stacktrace
6499 See also Documentation/trace/ftrace.rst "trace options"
6502 trace_trigger=[trigger-list]
6503 [FTRACE] Add a event trigger on specific events.
6504 Set a trigger on top of a specific event, with an optional
6507 The format is is "trace_trigger=<event>.<trigger>[ if <filter>],..."
6508 Where more than one trigger may be specified that are comma deliminated.
6512 trace_trigger="sched_switch.stacktrace if prev_state == 2"
6514 The above will enable the "stacktrace" trigger on the "sched_switch"
6515 event but only trigger it if the "prev_state" of the "sched_switch"
6516 event is "2" (TASK_UNINTERUPTIBLE).
6518 See also "Event triggers" in Documentation/trace/events.rst
6522 [FTRACE] enable this option to disable tracing when a
6523 warning is hit. This turns off "tracing_on". Tracing can
6524 be enabled again by echoing '1' into the "tracing_on"
6525 file located in /sys/kernel/tracing/
6527 This option is useful, as it disables the trace before
6528 the WARNING dump is called, which prevents the trace to
6529 be filled with content caused by the warning output.
6531 This option can also be set at run time via the sysctl
6532 option: kernel/traceoff_on_warning
6534 transparent_hugepage=
6536 Format: [always|madvise|never]
6537 Can be used to control the default behavior of the system
6538 with respect to transparent hugepages.
6539 See Documentation/admin-guide/mm/transhuge.rst
6542 trusted.source= [KEYS]
6544 This parameter identifies the trust source as a backend
6545 for trusted keys implementation. Supported trust
6550 If not specified then it defaults to iterating through
6551 the trust source list starting with TPM and assigns the
6552 first trust source as a backend which is initialized
6553 successfully during iteration.
6557 The RNG used to generate key material for trusted keys.
6560 - the same value as trusted.source: "tpm" or "tee"
6562 If not specified, "default" is used. In this case,
6563 the RNG's choice is left to each individual trust source.
6565 tsc= Disable clocksource stability checks for TSC.
6567 [x86] reliable: mark tsc clocksource as reliable, this
6568 disables clocksource verification at runtime, as well
6569 as the stability checks done at bootup. Used to enable
6570 high-resolution timer mode on older hardware, and in
6571 virtualized environment.
6572 [x86] noirqtime: Do not use TSC to do irq accounting.
6573 Used to run time disable IRQ_TIME_ACCOUNTING on any
6574 platforms where RDTSC is slow and this accounting
6576 [x86] unstable: mark the TSC clocksource as unstable, this
6577 marks the TSC unconditionally unstable at bootup and
6578 avoids any further wobbles once the TSC watchdog notices.
6579 [x86] nowatchdog: disable clocksource watchdog. Used
6580 in situations with strict latency requirements (where
6581 interruptions from clocksource watchdog are not
6583 [x86] recalibrate: force recalibration against a HW timer
6584 (HPET or PM timer) on systems whose TSC frequency was
6585 obtained from HW or FW using either an MSR or CPUID(0x15).
6586 Warn if the difference is more than 500 ppm.
6587 [x86] watchdog: Use TSC as the watchdog clocksource with
6588 which to check other HW timers (HPET or PM timer), but
6589 only on systems where TSC has been deemed trustworthy.
6590 This will be suppressed by an earlier tsc=nowatchdog and
6591 can be overridden by a later tsc=nowatchdog. A console
6592 message will flag any such suppression or overriding.
6594 tsc_early_khz= [X86] Skip early TSC calibration and use the given
6595 value instead. Useful when the early TSC frequency discovery
6596 procedure is not reliable, such as on overclocked systems
6597 with CPUID.16h support and partial CPUID.15h support.
6598 Format: <unsigned int>
6600 tsx= [X86] Control Transactional Synchronization
6601 Extensions (TSX) feature in Intel processors that
6602 support TSX control.
6604 This parameter controls the TSX feature. The options are:
6606 on - Enable TSX on the system. Although there are
6607 mitigations for all known security vulnerabilities,
6608 TSX has been known to be an accelerator for
6609 several previous speculation-related CVEs, and
6610 so there may be unknown security risks associated
6611 with leaving it enabled.
6613 off - Disable TSX on the system. (Note that this
6614 option takes effect only on newer CPUs which are
6615 not vulnerable to MDS, i.e., have
6616 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
6617 the new IA32_TSX_CTRL MSR through a microcode
6618 update. This new MSR allows for the reliable
6619 deactivation of the TSX functionality.)
6621 auto - Disable TSX if X86_BUG_TAA is present,
6622 otherwise enable TSX on the system.
6624 Not specifying this option is equivalent to tsx=off.
6626 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6629 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
6630 Abort (TAA) vulnerability.
6632 Similar to Micro-architectural Data Sampling (MDS)
6633 certain CPUs that support Transactional
6634 Synchronization Extensions (TSX) are vulnerable to an
6635 exploit against CPU internal buffers which can forward
6636 information to a disclosure gadget under certain
6639 In vulnerable processors, the speculatively forwarded
6640 data can be used in a cache side channel attack, to
6641 access data to which the attacker does not have direct
6644 This parameter controls the TAA mitigation. The
6647 full - Enable TAA mitigation on vulnerable CPUs
6650 full,nosmt - Enable TAA mitigation and disable SMT on
6651 vulnerable CPUs. If TSX is disabled, SMT
6652 is not disabled because CPU is not
6653 vulnerable to cross-thread TAA attacks.
6654 off - Unconditionally disable TAA mitigation
6656 On MDS-affected machines, tsx_async_abort=off can be
6657 prevented by an active MDS mitigation as both vulnerabilities
6658 are mitigated with the same mechanism so in order to disable
6659 this mitigation, you need to specify mds=off too.
6661 Not specifying this option is equivalent to
6662 tsx_async_abort=full. On CPUs which are MDS affected
6663 and deploy MDS mitigation, TAA mitigation is not
6664 required and doesn't provide any additional
6668 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6670 turbografx.map[2|3]= [HW,JOY]
6671 TurboGraFX parallel port interface
6673 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
6674 See also Documentation/input/devices/joystick-parport.rst
6676 udbg-immortal [PPC] When debugging early kernel crashes that
6677 happen after console_init() and before a proper
6678 console driver takes over, this boot options might
6679 help "seeing" what's going on.
6681 uhash_entries= [KNL,NET]
6682 Set number of hash buckets for UDP/UDP-Lite connections
6685 [USB] Ignore overcurrent events (default N).
6686 Some badly-designed motherboards generate lots of
6687 bogus events, for ports that aren't wired to
6688 anything. Set this parameter to avoid log spamming.
6689 Note that genuine overcurrent events won't be
6693 [X86] Cause panic on unknown NMI.
6695 unwind_debug [X86-64]
6696 Enable unwinder debug output. This can be
6697 useful for debugging certain unwinder error
6698 conditions, including corrupt stacks and
6699 bad/missing unwinder metadata.
6701 usbcore.authorized_default=
6702 [USB] Default USB device authorization:
6703 (default -1 = authorized except for wireless USB,
6704 0 = not authorized, 1 = authorized, 2 = authorized
6705 if device connected to internal port)
6707 usbcore.autosuspend=
6708 [USB] The autosuspend time delay (in seconds) used
6709 for newly-detected USB devices (default 2). This
6710 is the time required before an idle device will be
6711 autosuspended. Devices for which the delay is set
6712 to a negative value won't be autosuspended at all.
6714 usbcore.usbfs_snoop=
6715 [USB] Set to log all usbfs traffic (default 0 = off).
6717 usbcore.usbfs_snoop_max=
6718 [USB] Maximum number of bytes to snoop in each URB
6721 usbcore.blinkenlights=
6722 [USB] Set to cycle leds on hubs (default 0 = off).
6724 usbcore.old_scheme_first=
6725 [USB] Start with the old device initialization
6726 scheme (default 0 = off).
6728 usbcore.usbfs_memory_mb=
6729 [USB] Memory limit (in MB) for buffers allocated by
6730 usbfs (default = 16, 0 = max = 2047).
6732 usbcore.use_both_schemes=
6733 [USB] Try the other device initialization scheme
6734 if the first one fails (default 1 = enabled).
6736 usbcore.initial_descriptor_timeout=
6737 [USB] Specifies timeout for the initial 64-byte
6738 USB_REQ_GET_DESCRIPTOR request in milliseconds
6739 (default 5000 = 5.0 seconds).
6741 usbcore.nousb [USB] Disable the USB subsystem
6744 [USB] A list of quirk entries to augment the built-in
6745 usb core quirk list. List entries are separated by
6746 commas. Each entry has the form
6747 VendorID:ProductID:Flags. The IDs are 4-digit hex
6748 numbers and Flags is a set of letters. Each letter
6749 will change the built-in quirk; setting it if it is
6750 clear and clearing it if it is set. The letters have
6751 the following meanings:
6752 a = USB_QUIRK_STRING_FETCH_255 (string
6753 descriptors must not be fetched using
6755 b = USB_QUIRK_RESET_RESUME (device can't resume
6756 correctly so reset it instead);
6757 c = USB_QUIRK_NO_SET_INTF (device can't handle
6758 Set-Interface requests);
6759 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
6760 handle its Configuration or Interface
6762 e = USB_QUIRK_RESET (device can't be reset
6763 (e.g morph devices), don't use reset);
6764 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
6765 more interface descriptions than the
6766 bNumInterfaces count, and can't handle
6767 talking to these interfaces);
6768 g = USB_QUIRK_DELAY_INIT (device needs a pause
6769 during initialization, after we read
6770 the device descriptor);
6771 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
6772 high speed and super speed interrupt
6773 endpoints, the USB 2.0 and USB 3.0 spec
6774 require the interval in microframes (1
6775 microframe = 125 microseconds) to be
6776 calculated as interval = 2 ^
6778 Devices with this quirk report their
6779 bInterval as the result of this
6780 calculation instead of the exponent
6781 variable used in the calculation);
6782 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
6783 handle device_qualifier descriptor
6785 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
6786 generates spurious wakeup, ignore
6787 remote wakeup capability);
6788 k = USB_QUIRK_NO_LPM (device can't handle Link
6790 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
6791 (Device reports its bInterval as linear
6792 frames instead of the USB 2.0
6794 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
6795 to be disconnected before suspend to
6796 prevent spurious wakeup);
6797 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
6798 pause after every control message);
6799 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
6800 delay after resetting its port);
6801 Example: quirks=0781:5580:bk,0a5c:5834:gij
6804 [USBHID] The interval which mice are to be polled at.
6807 [USBHID] The interval which joysticks are to be polled at.
6810 [USBHID] The interval which keyboards are to be polled at.
6812 usb-storage.delay_use=
6813 [UMS] The delay in seconds before a new device is
6814 scanned for Logical Units (default 1).
6817 [UMS] A list of quirks entries to supplement or
6818 override the built-in unusual_devs list. List
6819 entries are separated by commas. Each entry has
6820 the form VID:PID:Flags where VID and PID are Vendor
6821 and Product ID values (4-digit hex numbers) and
6822 Flags is a set of characters, each corresponding
6823 to a common usb-storage quirk flag as follows:
6824 a = SANE_SENSE (collect more than 18 bytes
6825 of sense data, not on uas);
6826 b = BAD_SENSE (don't collect more than 18
6827 bytes of sense data, not on uas);
6828 c = FIX_CAPACITY (decrease the reported
6829 device capacity by one sector);
6830 d = NO_READ_DISC_INFO (don't use
6831 READ_DISC_INFO command, not on uas);
6832 e = NO_READ_CAPACITY_16 (don't use
6833 READ_CAPACITY_16 command);
6834 f = NO_REPORT_OPCODES (don't use report opcodes
6836 g = MAX_SECTORS_240 (don't transfer more than
6837 240 sectors at a time, uas only);
6838 h = CAPACITY_HEURISTICS (decrease the
6839 reported device capacity by one
6840 sector if the number is odd);
6841 i = IGNORE_DEVICE (don't bind to this
6843 j = NO_REPORT_LUNS (don't use report luns
6845 k = NO_SAME (do not use WRITE_SAME, uas only)
6846 l = NOT_LOCKABLE (don't try to lock and
6847 unlock ejectable media, not on uas);
6848 m = MAX_SECTORS_64 (don't transfer more
6849 than 64 sectors = 32 KB at a time,
6851 n = INITIAL_READ10 (force a retry of the
6852 initial READ(10) command, not on uas);
6853 o = CAPACITY_OK (accept the capacity
6854 reported by the device, not on uas);
6855 p = WRITE_CACHE (the device cache is ON
6856 by default, not on uas);
6857 r = IGNORE_RESIDUE (the device reports
6858 bogus residue values, not on uas);
6859 s = SINGLE_LUN (the device has only one
6861 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
6862 commands, uas only);
6863 u = IGNORE_UAS (don't bind to the uas driver);
6864 w = NO_WP_DETECT (don't test whether the
6865 medium is write-protected).
6866 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
6867 even if the device claims no cache,
6869 Example: quirks=0419:aaf5:rl,0421:0433:rc
6871 user_debug= [KNL,ARM]
6873 See arch/arm/Kconfig.debug help text.
6874 1 - undefined instruction events
6876 4 - invalid data aborts
6879 Example: user_debug=31
6882 [X86] Flags controlling user PTE allocations.
6884 nohigh = do not allocate PTE pages in
6885 HIGHMEM regardless of setting
6888 vdso= [X86,SH,SPARC]
6889 On X86_32, this is an alias for vdso32=. Otherwise:
6891 vdso=1: enable VDSO (the default)
6892 vdso=0: disable VDSO mapping
6894 vdso32= [X86] Control the 32-bit vDSO
6895 vdso32=1: enable 32-bit VDSO
6896 vdso32=0 or vdso32=2: disable 32-bit VDSO
6898 See the help text for CONFIG_COMPAT_VDSO for more
6899 details. If CONFIG_COMPAT_VDSO is set, the default is
6900 vdso32=0; otherwise, the default is vdso32=1.
6902 For compatibility with older kernels, vdso32=2 is an
6905 Try vdso32=0 if you encounter an error that says:
6906 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
6909 vector=percpu: enable percpu vector domain
6911 video= [FB] Frame buffer configuration
6912 See Documentation/fb/modedb.rst.
6914 video.brightness_switch_enabled= [ACPI]
6916 If set to 1, on receiving an ACPI notify event
6917 generated by hotkey, video driver will adjust brightness
6918 level and then send out the event to user space through
6919 the allocated input device. If set to 0, video driver
6920 will only send out the event without touching backlight
6925 [VMMIO] Memory mapped virtio (platform) device.
6927 <size>@<baseaddr>:<irq>[:<id>]
6929 <size> := size (can use standard suffixes
6931 <baseaddr> := physical base address
6932 <irq> := interrupt number (as passed to
6934 <id> := (optional) platform device id
6936 virtio_mmio.device=1K@0x100b0000:48:7
6938 Can be used multiple times for multiple devices.
6940 vga= [BOOT,X86-32] Select a particular video mode
6941 See Documentation/arch/x86/boot.rst and
6942 Documentation/admin-guide/svga.rst.
6943 Use vga=ask for menu.
6944 This is actually a boot loader parameter; the value is
6945 passed to the kernel using a special protocol.
6947 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
6948 May slow down system boot speed, especially when
6949 enabled on systems with a large amount of memory.
6950 All options are enabled by default, and this
6951 interface is meant to allow for selectively
6952 enabling or disabling specific virtual memory
6955 Available options are:
6956 P Enable page structure init time poisoning
6957 - Disable all of the above options
6959 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
6960 size of <nn>. This can be used to increase the
6961 minimum size (128MB on x86). It can also be used to
6962 decrease the size and leave more room for directly
6965 vmcp_cma=nn[MG] [KNL,S390]
6966 Sets the memory size reserved for contiguous memory
6967 allocations for the vmcp device driver.
6969 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
6972 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
6975 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
6979 Controls the behavior of vsyscalls (i.e. calls to
6980 fixed addresses of 0xffffffffff600x00 from legacy
6981 code). Most statically-linked binaries and older
6982 versions of glibc use these calls. Because these
6983 functions are at fixed addresses, they make nice
6984 targets for exploits that can control RIP.
6986 emulate Vsyscalls turn into traps and are emulated
6987 reasonably safely. The vsyscall page is
6990 xonly [default] Vsyscalls turn into traps and are
6991 emulated reasonably safely. The vsyscall
6992 page is not readable.
6994 none Vsyscalls don't work at all. This makes
6995 them quite hard to use for exploits but
6996 might break your system.
6998 vt.color= [VT] Default text color.
6999 Format: 0xYX, X = foreground, Y = background.
7000 Default: 0x07 = light gray on black.
7002 vt.cur_default= [VT] Default cursor shape.
7003 Format: 0xCCBBAA, where AA, BB, and CC are the same as
7004 the parameters of the <Esc>[?A;B;Cc escape sequence;
7005 see VGA-softcursor.txt. Default: 2 = underline.
7007 vt.default_blu= [VT]
7008 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
7009 Change the default blue palette of the console.
7010 This is a 16-member array composed of values
7013 vt.default_grn= [VT]
7014 Format: <green0>,<green1>,<green2>,...,<green15>
7015 Change the default green palette of the console.
7016 This is a 16-member array composed of values
7019 vt.default_red= [VT]
7020 Format: <red0>,<red1>,<red2>,...,<red15>
7021 Change the default red palette of the console.
7022 This is a 16-member array composed of values
7028 Set system-wide default UTF-8 mode for all tty's.
7029 Default is 1, i.e. UTF-8 mode is enabled for all
7030 newly opened terminals.
7032 vt.global_cursor_default=
7035 Set system-wide default for whether a cursor
7036 is shown on new VTs. Default is -1,
7037 i.e. cursors will be created by default unless
7038 overridden by individual drivers. 0 will hide
7039 cursors, 1 will display them.
7041 vt.italic= [VT] Default color for italic text; 0-15.
7044 vt.underline= [VT] Default color for underlined text; 0-15.
7047 watchdog timers [HW,WDT] For information on watchdog timers,
7048 see Documentation/watchdog/watchdog-parameters.rst
7049 or other driver-specific files in the
7050 Documentation/watchdog/ directory.
7054 Set the hard lockup detector stall duration
7055 threshold in seconds. The soft lockup detector
7056 threshold is set to twice the value. A value of 0
7057 disables both lockup detectors. Default is 10
7060 workqueue.watchdog_thresh=
7061 If CONFIG_WQ_WATCHDOG is configured, workqueue can
7062 warn stall conditions and dump internal state to
7063 help debugging. 0 disables workqueue stall
7064 detection; otherwise, it's the stall threshold
7065 duration in seconds. The default value is 30 and
7066 it can be updated at runtime by writing to the
7067 corresponding sysfs file.
7069 workqueue.cpu_intensive_thresh_us=
7070 Per-cpu work items which run for longer than this
7071 threshold are automatically considered CPU intensive
7072 and excluded from concurrency management to prevent
7073 them from noticeably delaying other per-cpu work
7074 items. Default is 10000 (10ms).
7076 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel
7077 will report the work functions which violate this
7078 threshold repeatedly. They are likely good
7079 candidates for using WQ_UNBOUND workqueues instead.
7081 workqueue.disable_numa
7082 By default, all work items queued to unbound
7083 workqueues are affine to the NUMA nodes they're
7084 issued on, which results in better behavior in
7085 general. If NUMA affinity needs to be disabled for
7086 whatever reason, this option can be used. Note
7087 that this also can be controlled per-workqueue for
7088 workqueues visible under /sys/bus/workqueue/.
7090 workqueue.power_efficient
7091 Per-cpu workqueues are generally preferred because
7092 they show better performance thanks to cache
7093 locality; unfortunately, per-cpu workqueues tend to
7094 be more power hungry than unbound workqueues.
7096 Enabling this makes the per-cpu workqueues which
7097 were observed to contribute significantly to power
7098 consumption unbound, leading to measurably lower
7099 power usage at the cost of small performance
7102 The default value of this parameter is determined by
7103 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
7105 workqueue.debug_force_rr_cpu
7106 Workqueue used to implicitly guarantee that work
7107 items queued without explicit CPU specified are put
7108 on the local CPU. This guarantee is no longer true
7109 and while local CPU is still preferred work items
7110 may be put on foreign CPUs. This debug option
7111 forces round-robin CPU selection to flush out
7112 usages which depend on the now broken guarantee.
7113 When enabled, memory and cache locality will be
7116 writecombine= [LOONGARCH] Control the MAT (Memory Access Type) of
7119 on - Enable writecombine, use WUC for ioremap_wc()
7120 off - Disable writecombine, use SUC for ioremap_wc()
7122 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
7123 default x2apic cluster mode on platforms
7126 xen_512gb_limit [KNL,X86-64,XEN]
7127 Restricts the kernel running paravirtualized under Xen
7128 to use only up to 512 GB of RAM. The reason to do so is
7129 crash analysis tools and Xen tools for doing domain
7130 save/restore/migration must be enabled to handle larger
7133 xen_emul_unplug= [HW,X86,XEN]
7134 Unplug Xen emulated devices
7135 Format: [unplug0,][unplug1]
7136 ide-disks -- unplug primary master IDE devices
7137 aux-ide-disks -- unplug non-primary-master IDE devices
7138 nics -- unplug network devices
7139 all -- unplug all emulated devices (NICs and IDE disks)
7140 unnecessary -- unplugging emulated devices is
7141 unnecessary even if the host did not respond to
7143 never -- do not unplug even if version check succeeds
7145 xen_legacy_crash [X86,XEN]
7146 Crash from Xen panic notifier, without executing late
7147 panic() code such as dumping handler.
7149 xen_msr_safe= [X86,XEN]
7151 Select whether to always use non-faulting (safe) MSR
7152 access functions when running as Xen PV guest. The
7153 default value is controlled by CONFIG_XEN_PV_MSR_SAFE.
7155 xen_nopvspin [X86,XEN]
7156 Disables the qspinlock slowpath using Xen PV optimizations.
7157 This parameter is obsoleted by "nopvspin" parameter, which
7158 has equivalent effect for XEN platform.
7161 Disables the PV optimizations forcing the HVM guest to
7162 run as generic HVM guest with no PV drivers.
7163 This option is obsoleted by the "nopv" option, which
7164 has equivalent effect for XEN platform.
7166 xen_no_vector_callback
7167 [KNL,X86,XEN] Disable the vector callback for Xen
7168 event channel interrupts.
7170 xen_scrub_pages= [XEN]
7171 Boolean option to control scrubbing pages before giving them back
7172 to Xen, for use by other domains. Can be also changed at runtime
7173 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
7174 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
7176 xen_timer_slop= [X86-64,XEN]
7177 Set the timer slop (in nanoseconds) for the virtual Xen
7178 timers (default is 100000). This adjusts the minimum
7179 delta of virtualized Xen timers, where lower values
7180 improve timer resolution at the expense of processing
7181 more timer interrupts.
7183 xen.balloon_boot_timeout= [XEN]
7184 The time (in seconds) to wait before giving up to boot
7185 in case initial ballooning fails to free enough memory.
7186 Applies only when running as HVM or PVH guest and
7187 started with less memory configured than allowed at
7188 max. Default is 180.
7190 xen.event_eoi_delay= [XEN]
7191 How long to delay EOI handling in case of event
7192 storms (jiffies). Default is 10.
7194 xen.event_loop_timeout= [XEN]
7195 After which time (jiffies) the event handling loop
7196 should start to delay EOI handling. Default is 2.
7198 xen.fifo_events= [XEN]
7199 Boolean parameter to disable using fifo event handling
7200 even if available. Normally fifo event handling is
7201 preferred over the 2-level event handling, as it is
7202 fairer and the number of possible event channels is
7203 much higher. Default is on (use fifo events).
7205 xirc2ps_cs= [NET,PCMCIA]
7207 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
7210 By default on POWER9 and above, the kernel will
7211 natively use the XIVE interrupt controller. This option
7212 allows the fallback firmware mode to be used:
7214 off Fallback to firmware control of XIVE interrupt
7215 controller on both pseries and powernv
7216 platforms. Only useful on POWER9 and above.
7218 xive.store-eoi=off [PPC]
7219 By default on POWER10 and above, the kernel will use
7220 stores for EOI handling when the XIVE interrupt mode
7221 is active. This option allows the XIVE driver to use
7222 loads instead, as on POWER9.
7224 xhci-hcd.quirks [USB,KNL]
7225 A hex value specifying bitmask with supplemental xhci
7226 host controller quirks. Meaning of each bit can be
7227 consulted in header drivers/usb/host/xhci.h.
7230 Format: { early | on | rw | ro | off }
7231 Controls if xmon debugger is enabled. Default is off.
7232 Passing only "xmon" is equivalent to "xmon=early".
7233 early Call xmon as early as possible on boot; xmon
7234 debugger is called from setup_arch().
7235 on xmon debugger hooks will be installed so xmon
7236 is only called on a kernel crash. Default mode,
7237 i.e. either "ro" or "rw" mode, is controlled
7238 with CONFIG_XMON_DEFAULT_RO_MODE.
7239 rw xmon debugger hooks will be installed so xmon
7240 is called only on a kernel crash, mode is write,
7241 meaning SPR registers, memory and, other data
7242 can be written using xmon commands.
7243 ro same as "rw" option above but SPR registers,
7244 memory, and other data can't be written using
7246 off xmon is disabled.