2 Format: { eager | lazy }
4 By default, unaccepted memory is accepted lazily to
5 avoid prolonged boot times. The lazy option will add
6 some runtime overhead until all memory is eventually
7 accepted. In most cases the overhead is negligible.
8 For some workloads or for debugging purposes
9 accept_memory=eager can be used to accept all memory
12 acpi= [HW,ACPI,X86,ARM64,RISCV64]
13 Advanced Configuration and Power Interface
14 Format: { force | on | off | strict | noirq | rsdt |
16 force -- enable ACPI if default was off
17 on -- enable ACPI but allow fallback to DT [arm64,riscv64]
18 off -- disable ACPI if default was on
19 noirq -- do not use ACPI for IRQ routing
20 strict -- Be less tolerant of platforms that are not
21 strictly ACPI specification compliant.
22 rsdt -- prefer RSDT over (default) XSDT
23 copy_dsdt -- copy DSDT to memory
24 For ARM64 and RISCV64, ONLY "acpi=off", "acpi=on" or
25 "acpi=force" are available
27 See also Documentation/power/runtime_pm.rst, pci=noacpi
29 acpi_apic_instance= [ACPI, IOAPIC]
31 2: use 2nd APIC table, if available
32 1,0: use 1st APIC table
35 acpi_backlight= [HW,ACPI]
36 { vendor | video | native | none }
37 If set to vendor, prefer vendor-specific driver
38 (e.g. thinkpad_acpi, sony_acpi, etc.) instead
39 of the ACPI video.ko driver.
40 If set to video, use the ACPI video.ko driver.
41 If set to native, use the device's native backlight mode.
42 If set to none, disable the ACPI backlight interface.
44 acpi_force_32bit_fadt_addr
45 force FADT to use 32 bit addresses rather than the
46 64 bit X_* addresses. Some firmware have broken 64
47 bit addresses for force ACPI ignore these and use
48 the older legacy 32 bit addresses.
50 acpica_no_return_repair [HW, ACPI]
51 Disable AML predefined validation mechanism
52 This mechanism can repair the evaluation result to make
53 the return objects more ACPI specification compliant.
54 This option is useful for developers to identify the
55 root cause of an AML interpreter issue when the issue
56 has something to do with the repair mechanism.
58 acpi.debug_layer= [HW,ACPI,ACPI_DEBUG]
59 acpi.debug_level= [HW,ACPI,ACPI_DEBUG]
61 CONFIG_ACPI_DEBUG must be enabled to produce any ACPI
62 debug output. Bits in debug_layer correspond to a
63 _COMPONENT in an ACPI source file, e.g.,
64 #define _COMPONENT ACPI_EVENTS
65 Bits in debug_level correspond to a level in
66 ACPI_DEBUG_PRINT statements, e.g.,
67 ACPI_DEBUG_PRINT((ACPI_DB_INFO, ...
68 The debug_level mask defaults to "info". See
69 Documentation/firmware-guide/acpi/debug.rst for more information about
70 debug layers and levels.
72 Enable processor driver info messages:
73 acpi.debug_layer=0x20000000
74 Enable AML "Debug" output, i.e., stores to the Debug
75 object while interpreting AML:
76 acpi.debug_layer=0xffffffff acpi.debug_level=0x2
77 Enable all messages related to ACPI hardware:
78 acpi.debug_layer=0x2 acpi.debug_level=0xffffffff
80 Some values produce so much output that the system is
81 unusable. The "log_buf_len" parameter may be useful
82 if you need to capture more output.
84 acpi_enforce_resources= [ACPI]
86 Check for resource conflicts between native drivers
87 and ACPI OperationRegions (SystemIO and SystemMemory
88 only). IO ports and memory declared in ACPI might be
89 used by the ACPI subsystem in arbitrary AML code and
90 can interfere with legacy drivers.
91 strict (default): access to resources claimed by ACPI
92 is denied; legacy drivers trying to access reserved
93 resources will fail to bind to device using them.
94 lax: access to resources claimed by ACPI is allowed;
95 legacy drivers trying to access reserved resources
96 will bind successfully but a warning message is logged.
97 no: ACPI OperationRegions are not marked as reserved,
98 no further checks are performed.
100 acpi_force_table_verification [HW,ACPI]
101 Enable table checksum verification during early stage.
102 By default, this is disabled due to x86 early mapping
105 acpi_irq_balance [HW,ACPI]
106 ACPI will balance active IRQs
109 acpi_irq_nobalance [HW,ACPI]
110 ACPI will not move active IRQs (default)
113 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA
114 Format: <irq>,<irq>...
116 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for
118 Format: <irq>,<irq>...
120 acpi_mask_gpe= [HW,ACPI]
121 Due to the existence of _Lxx/_Exx, some GPEs triggered
122 by unsupported hardware/firmware features can result in
123 GPE floodings that cannot be automatically disabled by
125 This facility can be used to prevent such uncontrolled
127 Format: <byte> or <bitmap-list>
129 acpi_no_auto_serialize [HW,ACPI]
130 Disable auto-serialization of AML methods
131 AML control methods that contain the opcodes to create
132 named objects will be marked as "Serialized" by the
133 auto-serialization feature.
134 This feature is enabled by default.
135 This option allows to turn off the feature.
137 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump
140 acpi_no_static_ssdt [HW,ACPI]
141 Disable installation of static SSDTs at early boot time
142 By default, SSDTs contained in the RSDT/XSDT will be
143 installed automatically and they will appear under
144 /sys/firmware/acpi/tables.
145 This option turns off this feature.
146 Note that specifying this option does not affect
147 dynamic table installation which will install SSDT
148 tables to /sys/firmware/acpi/tables/dynamic.
150 acpi_no_watchdog [HW,ACPI,WDT]
151 Ignore the ACPI-based watchdog interface (WDAT) and let
152 a native driver control the watchdog device instead.
154 acpi_rsdp= [ACPI,EFI,KEXEC]
155 Pass the RSDP address to the kernel, mostly used
156 on machines running EFI runtime service to boot the
157 second kernel for kdump.
159 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS
160 Format: To spoof as Windows 98: ="Microsoft Windows"
162 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead
163 of 2 which is mandated by ACPI 6) as the supported ACPI
164 specification revision (when using this switch, it may
165 be necessary to carry out a cold reboot _twice_ in a
166 row to make it take effect on the platform firmware).
168 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings
169 acpi_osi="string1" # add string1
170 acpi_osi="!string2" # remove string2
171 acpi_osi=!* # remove all strings
172 acpi_osi=! # disable all built-in OS vendor
174 acpi_osi=!! # enable all built-in OS vendor
176 acpi_osi= # disable all strings
178 'acpi_osi=!' can be used in combination with single or
179 multiple 'acpi_osi="string1"' to support specific OS
180 vendor string(s). Note that such command can only
181 affect the default state of the OS vendor strings, thus
182 it cannot affect the default state of the feature group
183 strings and the current state of the OS vendor strings,
184 specifying it multiple times through kernel command line
185 is meaningless. This command is useful when one do not
186 care about the state of the feature group strings which
187 should be controlled by the OSPM.
189 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent
190 to 'acpi_osi="Windows 2000" acpi_osi=!', they all
191 can make '_OSI("Windows 2000")' TRUE.
193 'acpi_osi=' cannot be used in combination with other
194 'acpi_osi=' command lines, the _OSI method will not
195 exist in the ACPI namespace. NOTE that such command can
196 only affect the _OSI support state, thus specifying it
197 multiple times through kernel command line is also
200 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)'
203 'acpi_osi=!*' can be used in combination with single or
204 multiple 'acpi_osi="string1"' to support specific
205 string(s). Note that such command can affect the
206 current state of both the OS vendor strings and the
207 feature group strings, thus specifying it multiple times
208 through kernel command line is meaningful. But it may
209 still not able to affect the final state of a string if
210 there are quirks related to this string. This command
211 is useful when one want to control the state of the
212 feature group strings to debug BIOS issues related to
215 1. 'acpi_osi="Module Device" acpi_osi=!*' can make
216 '_OSI("Module Device")' FALSE.
217 2. 'acpi_osi=!* acpi_osi="Module Device"' can make
218 '_OSI("Module Device")' TRUE.
219 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is
221 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"'
223 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!',
224 they all will make '_OSI("Windows 2000")' TRUE.
227 Override the pmtimer bug detection: force the kernel
228 to assume that this machine's pmtimer latches its value
229 and always returns good values.
231 acpi_sci= [HW,ACPI] ACPI System Control Interrupt trigger mode
232 Format: { level | edge | high | low }
234 acpi_skip_timer_override [HW,ACPI]
235 Recognize and ignore IRQ0/pin2 Interrupt Override.
236 For broken nForce2 BIOS resulting in XT-PIC timer.
238 acpi_sleep= [HW,ACPI] Sleep options
239 Format: { s3_bios, s3_mode, s3_beep, s4_hwsig,
240 s4_nohwsig, old_ordering, nonvs,
241 sci_force_enable, nobl }
242 See Documentation/power/video.rst for information on
244 s3_beep is for debugging; it makes the PC's speaker beep
245 as soon as the kernel's real-mode entry point is called.
246 s4_hwsig causes the kernel to check the ACPI hardware
247 signature during resume from hibernation, and gracefully
248 refuse to resume if it has changed. This complies with
249 the ACPI specification but not with reality, since
250 Windows does not do this and many laptops do change it
251 on docking. So the default behaviour is to allow resume
252 and simply warn when the signature changes, unless the
253 s4_hwsig option is enabled.
254 s4_nohwsig prevents ACPI hardware signature from being
255 used (or even warned about) during resume.
256 old_ordering causes the ACPI 1.0 ordering of the _PTS
257 control method, with respect to putting devices into
258 low power states, to be enforced (the ACPI 2.0 ordering
259 of _PTS is used by default).
260 nonvs prevents the kernel from saving/restoring the
261 ACPI NVS memory during suspend/hibernation and resume.
262 sci_force_enable causes the kernel to set SCI_EN directly
263 on resume from S1/S3 (which is against the ACPI spec,
264 but some broken systems don't work without it).
265 nobl causes the internal blacklist of systems known to
266 behave incorrectly in some ways with respect to system
267 suspend and resume to be ignored (use wisely).
269 acpi_use_timer_override [HW,ACPI]
270 Use timer override. For some broken Nvidia NF5 boards
271 that require a timer override, but don't have HPET
273 add_efi_memmap [EFI; X86] Include EFI memory map in
274 kernel's map of available physical RAM.
277 { off | try_unsupported }
278 off: disable AGP support
279 try_unsupported: try to drive unsupported chipsets
280 (may crash computer or cause data corruption)
283 See Documentation/sound/alsa-configuration.rst
286 Allow the default userspace alignment fault handler
287 behaviour to be specified. Bit 0 enables warnings,
288 bit 1 enables fixups, and bit 2 sends a segfault.
290 align_va_addr= [X86-64]
291 Align virtual addresses by clearing slice [14:12] when
292 allocating a VMA at process creation time. This option
293 gives you up to 3% performance improvement on AMD F15h
294 machines (where it is enabled by default) for a
295 CPU-intensive style benchmark, and it can vary highly in
296 a microbenchmark depending on workload and compiler.
298 32: only for 32-bit processes
299 64: only for 64-bit processes
300 on: enable for both 32- and 64-bit processes
301 off: disable for both 32- and 64-bit processes
303 alloc_snapshot [FTRACE]
304 Allocate the ftrace snapshot buffer on boot up when the
305 main buffer is allocated. This is handy if debugging
306 and you need to use tracing_snapshot() on boot up, and
307 do not want to use tracing_snapshot_alloc() as it needs
308 to be done where GFP_KERNEL allocations are allowed.
310 allow_mismatched_32bit_el0 [ARM64]
311 Allow execve() of 32-bit applications and setting of the
312 PER_LINUX32 personality on systems where only a strict
313 subset of the CPUs support 32-bit EL0. When this
314 parameter is present, the set of CPUs supporting 32-bit
315 EL0 is indicated by /sys/devices/system/cpu/aarch32_el0
316 and hot-unplug operations may be restricted.
318 See Documentation/arch/arm64/asymmetric-32bit.rst for more
321 amd_iommu= [HW,X86-64]
322 Pass parameters to the AMD IOMMU driver in the system.
324 fullflush - Deprecated, equivalent to iommu.strict=1
325 off - do not initialize any AMD IOMMU found in
327 force_isolation - Force device isolation for all
328 devices. The IOMMU driver is not
329 allowed anymore to lift isolation
330 requirements as needed. This option
331 does not override iommu=pt
332 force_enable - Force enable the IOMMU on platforms known
333 to be buggy with IOMMU enabled. Use this
335 pgtbl_v1 - Use v1 page table for DMA-API (Default).
336 pgtbl_v2 - Use v2 page table for DMA-API.
337 irtcachedis - Disable Interrupt Remapping Table (IRT) caching.
339 amd_iommu_dump= [HW,X86-64]
340 Enable AMD IOMMU driver option to dump the ACPI table
341 for AMD IOMMU. With this option enabled, AMD IOMMU
342 driver will print ACPI tables for AMD IOMMU during
343 IOMMU initialization.
345 amd_iommu_intr= [HW,X86-64]
346 Specifies one of the following AMD IOMMU interrupt
348 legacy - Use legacy interrupt remapping mode.
349 vapic - Use virtual APIC mode, which allows IOMMU
350 to inject interrupts directly into guest.
351 This mode requires kvm-amd.avic=1.
352 (Default when IOMMU HW support is present.)
356 Do not enable amd_pstate as the default
357 scaling driver for the supported processors
359 Use amd_pstate with passive mode as a scaling driver.
360 In this mode autonomous selection is disabled.
361 Driver requests a desired performance level and platform
362 tries to match the same performance level if it is
363 satisfied by guaranteed performance level.
365 Use amd_pstate_epp driver instance as the scaling driver,
366 driver provides a hint to the hardware if software wants
367 to bias toward performance (0x0) or energy efficiency (0xff)
368 to the CPPC firmware. then CPPC power algorithm will
369 calculate the runtime workload and adjust the realtime cores
372 Activate guided autonomous mode. Driver requests minimum and
373 maximum performance level and the platform autonomously
374 selects a performance level in this range and appropriate
375 to the current workload.
377 amijoy.map= [HW,JOY] Amiga joystick support
378 Map of devices attached to JOY0DAT and JOY1DAT
380 See also Documentation/input/joydev/joystick.rst
382 analog.map= [HW,JOY] Analog joystick and gamepad support
383 Specifies type or capabilities of an analog joystick
384 connected to one of 16 gameports
385 Format: <type1>,<type2>,..<type16>
388 Power management functions (SPARCstation-4/5 + deriv.)
390 Disable APC CPU standby support. SPARCstation-Fox does
391 not play well with APC CPU idle - disable it if you have
392 APC and your system crashes randomly.
394 apic= [APIC,X86] Advanced Programmable Interrupt Controller
395 Change the output verbosity while booting
396 Format: { quiet (default) | verbose | debug }
397 Change the amount of debugging information output
398 when initialising the APIC and IO-APIC components.
399 For X86-32, this can also be used to specify an APIC
401 Format: apic=driver_name
402 Examples: apic=bigsmp
404 apic_extnmi= [APIC,X86] External NMI delivery setting
405 Format: { bsp (default) | all | none }
406 bsp: External NMI is delivered only to CPU 0
407 all: External NMIs are broadcast to all CPUs as a
409 none: External NMI is masked for all CPUs. This is
410 useful so that a dump capture kernel won't be
414 See Documentation/networking/ipv6.rst.
416 apm= [APM] Advanced Power Management
417 See header of arch/x86/kernel/apm_32.c.
419 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
420 Format: { "0" | "1" }
421 See security/apparmor/Kconfig help text
424 Default value is set via kernel config option.
426 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
427 Format: <io>,<irq>,<nodeID>
429 arm64.nobti [ARM64] Unconditionally disable Branch Target
430 Identification support
432 arm64.nomops [ARM64] Unconditionally disable Memory Copy and Memory
433 Set instructions support
435 arm64.nomte [ARM64] Unconditionally disable Memory Tagging Extension
438 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication
441 arm64.nosme [ARM64] Unconditionally disable Scalable Matrix
444 arm64.nosve [ARM64] Unconditionally disable Scalable Vector
449 atarimouse= [HW,MOUSE] Atari Mouse
451 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
452 EzKey and similar keyboards
454 atkbd.reset= [HW] Reset keyboard during initialization
456 atkbd.set= [HW] Select keyboard code set
457 Format: <int> (2 = AT (default), 3 = PS/2)
459 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
462 atkbd.softraw= [HW] Choose between synthetic and real raw mode
463 Format: <bool> (0 = real, 1 = synthetic (default))
465 atkbd.softrepeat= [HW]
466 Use software keyboard repeat
468 audit= [KNL] Enable the audit sub-system
469 Format: { "0" | "1" | "off" | "on" }
470 0 | off - kernel audit is disabled and can not be
471 enabled until the next reboot
472 unset - kernel audit is initialized but disabled and
473 will be fully enabled by the userspace auditd.
474 1 | on - kernel audit is initialized and partially
475 enabled, storing at most audit_backlog_limit
476 messages in RAM until it is fully enabled by the
480 audit_backlog_limit= [KNL] Set the audit queue size limit.
481 Format: <int> (must be >=0)
484 bau= [X86_UV] Enable the BAU on SGI UV. The default
485 behavior is to disable the BAU (i.e. bau=0).
486 Format: { "0" | "1" }
489 unset - Disable the BAU.
491 baycom_epp= [HW,AX25]
494 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
496 See header of drivers/net/hamradio/baycom_par.c.
498 baycom_ser_fdx= [HW,AX25]
499 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
500 Format: <io>,<irq>,<mode>[,<baud>]
501 See header of drivers/net/hamradio/baycom_ser_fdx.c.
503 baycom_ser_hdx= [HW,AX25]
504 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
505 Format: <io>,<irq>,<mode>
506 See header of drivers/net/hamradio/baycom_ser_hdx.c.
509 Disable BERT OS support on buggy BIOSes.
511 bgrt_disable [ACPI][X86]
512 Disable BGRT to avoid flickering OEM logo.
514 blkdevparts= Manual partition parsing of block device(s) for
515 embedded devices based on command line input.
516 See Documentation/block/cmdline-partition.rst
518 boot_delay= Milliseconds to delay each printk during boot.
519 Only works if CONFIG_BOOT_PRINTK_DELAY is enabled,
520 and you may also have to specify "lpj=". Boot_delay
521 values larger than 10 seconds (10000) are assumed
522 erroneous and ignored.
526 Extended command line options can be added to an initrd
527 and this will cause the kernel to look for it.
529 See Documentation/admin-guide/bootconfig.rst
531 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
532 bttv.radio= Most important insmod options are available as
534 bttv.pll= See Documentation/admin-guide/media/bttv.rst
537 bulk_remove=off [PPC] This parameter disables the use of the pSeries
538 firmware feature for flushing multiple hpte entries
541 c101= [NET] Moxa C101 synchronous serial card
543 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
544 Sometimes CPU hardware bugs make them report the cache
545 size incorrectly. The kernel will attempt work arounds
546 to fix known problems, but for some CPUs it is not
547 possible to determine what the correct size should be.
548 This option provides an override for these situations.
551 [NET] Specifies amount of time (in seconds) that
552 the kernel should wait for a network carrier. By default
553 it waits 120 seconds.
555 ca_keys= [KEYS] This parameter identifies a specific key(s) on
556 the system trusted keyring to be used for certificate
558 format: { id:<keyid> | builtin }
560 cca= [MIPS] Override the kernel pages' cache coherency
561 algorithm. Accepted values range from 0 to 7
562 inclusive. See arch/mips/include/asm/pgtable-bits.h
563 for platform specific values (SB1, Loongson3 and
566 ccw_timeout_log [S390]
567 See Documentation/arch/s390/common_io.rst for details.
569 cgroup_disable= [KNL] Disable a particular controller or optional feature
570 Format: {name of the controller(s) or feature(s) to disable}
571 The effects of cgroup_disable=foo are:
572 - foo isn't auto-mounted if you mount all cgroups in
574 - foo isn't visible as an individually mountable
576 - if foo is an optional feature then the feature is
577 disabled and corresponding cgroup files are not
579 {Currently only "memory" controller deal with this and
580 cut the overhead, others just disable the usage. So
581 only cgroup_disable=memory is actually worthy}
582 Specifying "pressure" disables per-cgroup pressure
583 stall information accounting feature
585 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
586 Format: { { controller | "all" | "named" }
587 [,{ controller | "all" | "named" }...] }
588 Like cgroup_disable, but only applies to cgroup v1;
589 the blacklisted controllers remain available in cgroup2.
590 "all" blacklists all controllers and "named" disables
591 named mounts. Specifying both "all" and "named" disables
594 cgroup_favordynmods= [KNL] Enable or Disable favordynmods.
595 Format: { "true" | "false" }
596 Defaults to the value of CONFIG_CGROUP_FAVOR_DYNMODS.
598 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
600 nosocket -- Disable socket memory accounting.
601 nokmem -- Disable kernel memory accounting.
602 nobpf -- Disable BPF memory accounting.
604 checkreqprot= [SELINUX] Set initial checkreqprot flag value.
605 Format: { "0" | "1" }
606 See security/selinux/Kconfig help text.
607 0 -- check protection applied by kernel (includes
608 any implied execute protection).
609 1 -- check protection requested by application.
610 Default value is set via a kernel config option.
611 Value can be changed at runtime via
612 /sys/fs/selinux/checkreqprot.
613 Setting checkreqprot to 1 is deprecated.
616 See Documentation/arch/s390/common_io.rst for details.
618 clearcpuid=X[,X...] [X86]
619 Disable CPUID feature X for the kernel. See
620 arch/x86/include/asm/cpufeatures.h for the valid bit
621 numbers X. Note the Linux-specific bits are not necessarily
622 stable over kernel options, but the vendor-specific
624 X can also be a string as appearing in the flags: line
625 in /proc/cpuinfo which does not have the above
626 instability issue. However, not all features have names
628 Note that using this option will taint your kernel.
629 Also note that user programs calling CPUID directly
630 or using the feature without checking anything
631 will still see it. This just prevents it from
632 being used by the kernel or shown in /proc/cpuinfo.
633 Also note the kernel might malfunction if you disable
638 Prevents the clock framework from automatically gating
639 clocks that have not been explicitly enabled by a Linux
640 device driver but are enabled in hardware at reset or
641 by the bootloader/firmware. Note that this does not
642 force such clocks to be always-on nor does it reserve
643 those clocks in any way. This parameter is useful for
644 debug and development, but should not be needed on a
645 platform with proper driver support. For more
646 information, see Documentation/driver-api/clk.rst.
648 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
650 Forces specified clocksource (if available) to be used
651 when calculating gettimeofday(). If specified
652 clocksource is not available, it defaults to PIT.
653 Format: { pit | tsc | cyclone | pmtmr }
655 clocksource= Override the default clocksource
657 Override the default clocksource and use the clocksource
658 with the name specified.
659 Some clocksource names to choose from, depending on
661 [all] jiffies (this is the base, fallback clocksource)
663 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
664 pxa_timer,timer3,32k_counter,timer0_1
665 [X86-32] pit,hpet,tsc;
666 scx200_hrt on Geode; cyclone on IBM x440
674 clocksource.arm_arch_timer.evtstrm=
677 Enable/disable the eventstream feature of the ARM
678 architected timer so that code using WFE-based polling
679 loops can be debugged more effectively on production
682 clocksource.verify_n_cpus= [KNL]
683 Limit the number of CPUs checked for clocksources
684 marked with CLOCK_SOURCE_VERIFY_PERCPU that
685 are marked unstable due to excessive skew.
686 A negative value says to check all CPUs, while
687 zero says not to check any. Values larger than
688 nr_cpu_ids are silently truncated to nr_cpu_ids.
689 The actual CPUs are chosen randomly, with
690 no replacement if the same CPU is chosen twice.
692 clocksource-wdtest.holdoff= [KNL]
693 Set the time in seconds that the clocksource
694 watchdog test waits before commencing its tests.
695 Defaults to zero when built as a module and to
696 10 seconds when built into the kernel.
698 cma=nn[MG]@[start[MG][-end[MG]]]
700 Sets the size of kernel global memory area for
701 contiguous memory allocations and optionally the
702 placement constraint by the physical address range of
703 memory allocations. A value of 0 disables CMA
704 altogether. For more information, see
705 kernel/dma/contiguous.c
709 Sets the size of kernel per-numa memory area for
710 contiguous memory allocations. A value of 0 disables
711 per-numa CMA altogether. And If this option is not
712 specified, the default value is 0.
713 With per-numa CMA enabled, DMA users on node nid will
714 first try to allocate buffer from the pernuma area
715 which is located in node nid, if the allocation fails,
716 they will fallback to the global default memory area.
718 numa_cma=<node>:nn[MG][,<node>:nn[MG]]
720 Sets the size of kernel numa memory area for
721 contiguous memory allocations. It will reserve CMA
722 area for the specified node.
724 With numa CMA enabled, DMA users on node nid will
725 first try to allocate buffer from the numa area
726 which is located in node nid, if the allocation fails,
727 they will fallback to the global default memory area.
729 cmo_free_hint= [PPC] Format: { yes | no }
730 Specify whether pages are marked as being inactive
731 when they are freed. This is used in CMO environments
732 to determine OS memory pressure for page stealing by
736 coherent_pool=nn[KMG] [ARM,KNL]
737 Sets the size of memory pool for coherent, atomic dma
738 allocations, by default set to 256K.
740 com20020= [HW,NET] ARCnet - COM20020 chipset
742 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
744 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
748 ARCnet - COM90xx chipset (memory-mapped buffers)
749 Format: <io>[,<irq>[,<memstart>]]
751 condev= [HW,S390] console device
754 con3215_drop= [S390] 3215 console drop mode.
756 When set to true, drop data on the 3215 console when
757 the console buffer is full. In this case the
758 operator using a 3270 terminal emulator (for example
759 x3270) does not have to enter the clear key for the
760 console output to advance and the kernel to continue.
761 This leads to a much faster boot time when a 3270
762 terminal emulator is active. If no 3270 terminal
763 emulator is used, this parameter has no effect.
765 console= [KNL] Output console device and options.
767 tty<n> Use the virtual console device <n>.
771 Use the specified serial port. The options are of
772 the form "bbbbpnf", where "bbbb" is the baud rate,
773 "p" is parity ("n", "o", or "e"), "n" is number of
774 bits, and "f" is flow control ("r" for RTS or
775 omit it). Default is "9600n8".
777 See Documentation/admin-guide/serial-console.rst for more
779 Documentation/networking/netconsole.rst for an
782 uart[8250],io,<addr>[,options]
783 uart[8250],mmio,<addr>[,options]
784 uart[8250],mmio16,<addr>[,options]
785 uart[8250],mmio32,<addr>[,options]
786 uart[8250],0x<addr>[,options]
787 Start an early, polled-mode console on the 8250/16550
788 UART at the specified I/O port or MMIO address,
789 switching to the matching ttyS device later.
790 MMIO inter-register address stride is either 8-bit
791 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
792 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
793 to be equivalent to 'mmio'. 'options' are specified in
794 the same format described for ttyS above; if unspecified,
795 the h/w is not re-initialized.
797 hvc<n> Use the hypervisor console device <n>. This is for
798 both Xen and PowerPC hypervisors.
801 Use to disable console output, i.e., to have kernel
802 console messages discarded.
803 This must be the only console= parameter used on the
806 If the device connected to the port is not a TTY but a braille
807 device, prepend "brl," before the device type, for instance
809 For now, only VisioBraille is supported.
812 [KNL] Change console messages format
814 By default we print messages on consoles in
815 "[time stamp] text\n" format (time stamp may not be
816 printed, depending on CONFIG_PRINTK_TIME or
817 `printk_time' param).
819 Switch to syslog format: "<%u>[time stamp] text\n"
820 IOW, each message will have a facility and loglevel
821 prefix. The format is similar to one used by syslog()
822 syscall, or to executing "dmesg -S --raw" or to reading
825 consoleblank= [KNL] The console blank (screen saver) timeout in
826 seconds. A value of 0 disables the blank timer.
830 [KNL] Change the default value for
831 /proc/<pid>/coredump_filter.
832 See also Documentation/filesystems/proc.rst.
834 coresight_cpu_debug.enable
837 Enable/disable the CPU sampling based debugging.
838 0: default value, disable debugging
839 1: enable debugging at boot time
841 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
843 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
845 cpuidle.off=1 [CPU_IDLE]
846 disable the cpuidle sub-system
849 [CPU_IDLE] Name of the cpuidle governor to use.
851 cpufreq.off=1 [CPU_FREQ]
852 disable the cpufreq sub-system
854 cpufreq.default_governor=
855 [CPU_FREQ] Name of the default cpufreq governor or
856 policy to use. This governor must be registered in the
857 kernel before the cpufreq driver probes.
860 [X86] Delay for N microsec between assert and de-assert
861 of APIC INIT to start processors. This delay occurs
862 on every CPU online, such as boot, and resume from suspend.
866 [SMP] Enable/disable parallel bringup of secondary CPUs
868 Default is enabled if CONFIG_HOTPLUG_PARALLEL=y. Otherwise
869 the parameter has no effect.
871 crash_kexec_post_notifiers
872 Run kdump after running panic-notifiers and dumping
873 kmsg. This only for the users who doubt kdump always
874 succeeds in any situation.
875 Note that this also increases risks of kdump failure,
876 because some panic notifiers can make the crashed
877 kernel more unstable.
879 crashkernel=size[KMG][@offset[KMG]]
880 [KNL] Using kexec, Linux can switch to a 'crash kernel'
881 upon panic. This parameter reserves the physical
882 memory region [offset, offset + size] for that kernel
883 image. If '@offset' is omitted, then a suitable offset
884 is selected automatically.
885 [KNL, X86-64, ARM64, RISCV, LoongArch] Select a region
886 under 4G first, and fall back to reserve region above
887 4G when '@offset' hasn't been specified.
888 See Documentation/admin-guide/kdump/kdump.rst for further details.
890 crashkernel=range1:size1[,range2:size2,...][@offset]
891 [KNL] Same as above, but depends on the memory
892 in the running system. The syntax of range is
893 start-[end] where start and end are both
894 a memory unit (amount[KMG]). See also
895 Documentation/admin-guide/kdump/kdump.rst for an example.
897 crashkernel=size[KMG],high
898 [KNL, X86-64, ARM64, RISCV, LoongArch] range could be
900 Allow kernel to allocate physical memory region from top,
901 so could be above 4G if system have more than 4G ram
902 installed. Otherwise memory region will be allocated
903 below 4G, if available.
904 It will be ignored if crashkernel=X is specified.
905 crashkernel=size[KMG],low
906 [KNL, X86-64, ARM64, RISCV, LoongArch] range under 4G.
907 When crashkernel=X,high is passed, kernel could allocate
908 physical memory region above 4G, that cause second kernel
909 crash on system that require some amount of low memory,
910 e.g. swiotlb requires at least 64M+32K low memory, also
911 enough extra low memory is needed to make sure DMA buffers
912 for 32-bit devices won't run out. Kernel would try to allocate
913 default size of memory below 4G automatically. The default
914 size is platform dependent.
915 --> x86: max(swiotlb_size_or_default() + 8MiB, 256MiB)
918 --> loongarch: 128MiB
919 This one lets the user specify own low range under 4G
920 for second kernel instead.
921 0: to disable low allocation.
922 It will be ignored when crashkernel=X,high is not used
923 or memory reserved is below 4G.
926 [KNL] Disable crypto self-tests
931 cs89x0_media= [HW,NET]
932 Format: { rj45 | aui | bnc }
934 csdlock_debug= [KNL] Enable or disable debug add-ons of cross-CPU
935 function call handling. When switched on,
936 additional debug data is printed to the console
937 in case a hanging CPU is detected, and that
938 CPU is pinged again in order to try to resolve
939 the hang situation. The default value of this
940 option depends on the CSD_LOCK_WAIT_DEBUG_DEFAULT
944 See header of drivers/s390/block/dasd_devmap.c.
946 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
947 (one device per port)
948 Format: <port#>,<type>
949 See also Documentation/input/devices/joystick-parport.rst
951 debug [KNL] Enable kernel debugging (events log level).
954 [KNL] Enable printing [hashed] pointers early in the
955 boot sequence. If enabled, we use a weak hash instead
956 of siphash to hash pointers. Use this option if you are
957 seeing instances of '(___ptrval___)') and need to see a
958 value (hashed pointer) instead. Cryptographically
959 insecure, please do not use on production kernels.
962 [KNL] verbose locking self-tests
964 Print debugging info while doing the locking API
966 Bitmask for the various LOCKTYPE_ tests. Defaults to 0
967 (no extra messages), setting it to -1 (all bits set)
968 will print _a_lot_ more information - normally only
969 useful to lockdep developers.
971 debug_objects [KNL] Enable object debugging
973 debug_guardpage_minorder=
974 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
975 parameter allows control of the order of pages that will
976 be intentionally kept free (and hence protected) by the
977 buddy allocator. Bigger value increase the probability
978 of catching random memory corruption, but reduce the
979 amount of memory for normal system use. The maximum
980 possible value is MAX_PAGE_ORDER/2. Setting this
981 parameter to 1 or 2 should be enough to identify most
982 random memory corruption problems caused by bugs in
983 kernel or driver code when a CPU writes to (or reads
984 from) a random memory location. Note that there exists
985 a class of memory corruptions problems caused by buggy
986 H/W or F/W or by drivers badly programming DMA
987 (basically when memory is written at bus level and the
988 CPU MMU is bypassed) which are not detectable by
989 CONFIG_DEBUG_PAGEALLOC, hence this option will not
990 help tracking down these problems.
993 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
994 enables the feature at boot time. By default, it is
995 disabled and the system will work mostly the same as a
996 kernel built without CONFIG_DEBUG_PAGEALLOC.
997 Note: to get most of debug_pagealloc error reports, it's
998 useful to also enable the page_owner functionality.
999 on: enable the feature
1001 debugfs= [KNL] This parameter enables what is exposed to userspace
1002 and debugfs internal clients.
1003 Format: { on, no-mount, off }
1004 on: All functions are enabled.
1006 Filesystem is not registered but kernel clients can
1007 access APIs and a crashkernel can be used to read
1008 its content. There is nothing to mount.
1009 off: Filesystem is not registered and clients
1010 get a -EPERM as result when trying to register files
1011 or directories within debugfs.
1012 This is equivalent of the runtime functionality if
1013 debugfs was not enabled in the kernel at all.
1014 Default value is set in build-time with a kernel configuration.
1016 debugpat [X86] Enable PAT debugging
1019 [HW] The size of the default HugeTLB page. This is
1020 the size represented by the legacy /proc/ hugepages
1021 APIs. In addition, this is the default hugetlb size
1022 used for shmget(), mmap() and mounting hugetlbfs
1023 filesystems. If not specified, defaults to the
1024 architecture's default huge page size. Huge page
1025 sizes are architecture dependent. See also
1026 Documentation/admin-guide/mm/hugetlbpage.rst.
1029 deferred_probe_timeout=
1030 [KNL] Debugging option to set a timeout in seconds for
1031 deferred probe to give up waiting on dependencies to
1032 probe. Only specific dependencies (subsystems or
1033 drivers) that have opted in will be ignored. A timeout
1034 of 0 will timeout at the end of initcalls. If the time
1035 out hasn't expired, it'll be restarted by each
1036 successful driver registration. This option will also
1037 dump out devices still on the deferred probe list after
1040 delayacct [KNL] Enable per-task delay accounting
1042 dell_smm_hwmon.ignore_dmi=
1043 [HW] Continue probing hardware even if DMI data
1044 indicates that the driver is running on unsupported
1047 dell_smm_hwmon.force=
1048 [HW] Activate driver even if SMM BIOS signature does
1049 not match list of supported models and enable otherwise
1050 blacklisted features.
1052 dell_smm_hwmon.power_status=
1053 [HW] Report power status in /proc/i8k
1054 (disabled by default).
1056 dell_smm_hwmon.restricted=
1057 [HW] Allow controlling fans only if SYS_ADMIN
1060 dell_smm_hwmon.fan_mult=
1061 [HW] Factor to multiply fan speed with.
1063 dell_smm_hwmon.fan_max=
1064 [HW] Maximum configurable fan speed.
1067 Format: { on | off | def_only | inf_only | always }
1068 on: s390 zlib hardware support for compression on
1069 level 1 and decompression (default)
1070 off: No s390 zlib hardware support
1071 def_only: s390 zlib hardware support for deflate
1072 only (compression on level 1)
1073 inf_only: s390 zlib hardware support for inflate
1074 only (decompression)
1075 always: Same as 'on' but ignores the selected compression
1076 level always using hardware support (used for debugging)
1078 dhash_entries= [KNL]
1079 Set number of hash buckets for dentry cache.
1081 disable_1tb_segments [PPC]
1082 Disables the use of 1TB hash page table segments. This
1083 causes the kernel to fall back to 256MB segments which
1084 can be useful when debugging issues that require an SLB
1088 See Documentation/networking/ipv6.rst.
1091 Disable RADIX MMU mode on POWER9
1094 Disable TLBIE instruction. Currently does not work
1095 with KVM, with HASH MMU, or with coherent accelerators.
1097 disable_cpu_apicid= [X86,APIC,SMP]
1099 The number of initial APIC ID for the
1100 corresponding CPU to be disabled at boot,
1101 mostly used for the kdump 2nd kernel to
1102 disable BSP to wake up multiple CPUs without
1103 causing system reset or hang due to sending
1104 INIT from AP to BSP.
1106 disable_ddw [PPC/PSERIES]
1107 Disable Dynamic DMA Window support. Use this
1108 to workaround buggy firmware.
1110 disable_ipv6= [IPV6]
1111 See Documentation/networking/ipv6.rst.
1113 disable_mtrr_cleanup [X86]
1114 The kernel tries to adjust MTRR layout from continuous
1115 to discrete, to make X server driver able to add WB
1116 entry later. This parameter disables that.
1118 disable_mtrr_trim [X86, Intel and AMD only]
1119 By default the kernel will trim any uncacheable
1120 memory out of your available memory pool based on
1121 MTRR settings. This parameter disables that behavior,
1122 possibly causing your machine to run very slowly.
1124 disable_timer_pin_1 [X86]
1125 Disable PIN 1 of APIC timer
1126 Can be useful to work around chipset bugs.
1128 dis_ucode_ldr [X86] Disable the microcode loader.
1130 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
1131 this option disables the debugging code at boot.
1133 dma_debug_entries=<number>
1134 This option allows to tune the number of preallocated
1135 entries for DMA-API debugging code. One entry is
1136 required per DMA-API allocation. Use this if the
1137 DMA-API debugging code disables itself because the
1138 architectural default is too low.
1140 dma_debug_driver=<driver_name>
1141 With this option the DMA-API debugging driver
1142 filter feature can be enabled at boot time. Just
1143 pass the driver to filter for as the parameter.
1144 The filter can be disabled or changed to another
1145 driver later using sysfs.
1147 driver_async_probe= [KNL]
1148 List of driver names to be probed asynchronously. *
1149 matches with all driver names. If * is specified, the
1150 rest of the listed driver names are those that will NOT
1152 Format: <driver_name1>,<driver_name2>...
1154 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
1155 Broken monitors, graphic adapters, KVMs and EDIDless
1156 panels may send no or incorrect EDID data sets.
1157 This parameter allows to specify an EDID data sets
1158 in the /lib/firmware directory that are used instead.
1159 Generic built-in EDID data sets are used, if one of
1160 edid/1024x768.bin, edid/1280x1024.bin,
1161 edid/1680x1050.bin, or edid/1920x1080.bin is given
1162 and no file with the same name exists. Details and
1163 instructions how to build your own EDID data are
1164 available in Documentation/admin-guide/edid.rst. An EDID
1165 data set will only be used for a particular connector,
1166 if its name and a colon are prepended to the EDID
1167 name. Each connector may use a unique EDID data
1168 set by separating the files with a comma. An EDID
1169 data set with no connector name will be used for
1170 any connectors not explicitly specified.
1175 Format: {"off" | "known"}
1176 Control how the dt_cpu_ftrs device-tree binding is
1177 used for CPU feature discovery and setup (if it
1179 off: Do not use it, fall back to legacy cpu table.
1180 known: Do not pass through unknown features to guests
1181 or userspace, only those that the kernel is aware of.
1183 dump_apple_properties [X86]
1184 Dump name and content of EFI device properties on
1185 x86 Macs. Useful for driver authors to determine
1186 what data is available or for reverse-engineering.
1188 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
1189 <module>.dyndbg[="val"]
1190 Enable debug messages at boot time. See
1191 Documentation/admin-guide/dynamic-debug-howto.rst
1194 early_ioremap_debug [KNL]
1195 Enable debug messages in early_ioremap support. This
1196 is useful for tracking down temporary early mappings
1197 which are not unmapped.
1199 earlycon= [KNL] Output early console device and options.
1201 When used with no options, the early console is
1202 determined by stdout-path property in device tree's
1203 chosen node or the ACPI SPCR table if supported by
1206 cdns,<addr>[,options]
1207 Start an early, polled-mode console on a Cadence
1208 (xuartps) serial port at the specified address. Only
1209 supported option is baud rate. If baud rate is not
1210 specified, the serial port must already be setup and
1213 uart[8250],io,<addr>[,options[,uartclk]]
1214 uart[8250],mmio,<addr>[,options[,uartclk]]
1215 uart[8250],mmio32,<addr>[,options[,uartclk]]
1216 uart[8250],mmio32be,<addr>[,options[,uartclk]]
1217 uart[8250],0x<addr>[,options]
1218 Start an early, polled-mode console on the 8250/16550
1219 UART at the specified I/O port or MMIO address.
1220 MMIO inter-register address stride is either 8-bit
1221 (mmio) or 32-bit (mmio32 or mmio32be).
1222 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1223 to be equivalent to 'mmio'. 'options' are specified
1224 in the same format described for "console=ttyS<n>"; if
1225 unspecified, the h/w is not initialized. 'uartclk' is
1226 the uart clock frequency; if unspecified, it is set
1227 to 'BASE_BAUD' * 16.
1231 Start an early, polled-mode console on a pl011 serial
1232 port at the specified address. The pl011 serial port
1233 must already be setup and configured. Options are not
1234 yet supported. If 'mmio32' is specified, then only
1235 the driver will use only 32-bit accessors to read/write
1236 the device registers.
1239 Start an early console on a litex serial port at the
1240 specified address. The serial port must already be
1241 setup and configured. Options are not yet supported.
1244 Start an early, polled-mode console on a meson serial
1245 port at the specified address. The serial port must
1246 already be setup and configured. Options are not yet
1250 Start an early, polled-mode console on an msm serial
1251 port at the specified address. The serial port
1252 must already be setup and configured. Options are not
1255 msm_serial_dm,<addr>
1256 Start an early, polled-mode console on an msm serial
1257 dm port at the specified address. The serial port
1258 must already be setup and configured. Options are not
1262 Start an early, polled-mode console on a serial port
1263 of an Actions Semi SoC, such as S500 or S900, at the
1264 specified address. The serial port must already be
1265 setup and configured. Options are not yet supported.
1268 Start an early, polled-mode console on a serial port
1269 of an RDA Micro SoC, such as RDA8810PL, at the
1270 specified address. The serial port must already be
1271 setup and configured. Options are not yet supported.
1274 Use RISC-V SBI (Supervisor Binary Interface) for early
1277 smh Use ARM semihosting calls for early console.
1285 Use early console provided by serial driver available
1286 on Samsung SoCs, requires selecting proper type and
1287 a correct base address of the selected UART port. The
1288 serial port must already be setup and configured.
1289 Options are not yet supported.
1292 Start an early, polled-mode console on a lantiq serial
1293 (lqasc) port at the specified address. The serial port
1294 must already be setup and configured. Options are not
1299 Use early console provided by Freescale LP UART driver
1300 found on Freescale Vybrid and QorIQ LS1021A processors.
1301 A valid base address must be provided, and the serial
1302 port must already be setup and configured.
1306 Start an early, polled-mode, output-only console on the
1307 Freescale i.MX UART at the specified address. The UART
1308 must already be setup and configured.
1311 Start an early, polled-mode console on the
1312 Armada 3700 serial port at the specified
1313 address. The serial port must already be setup
1314 and configured. Options are not yet supported.
1317 Start an early, polled-mode console on a Qualcomm
1318 Generic Interface (GENI) based serial port at the
1319 specified address. The serial port must already be
1320 setup and configured. Options are not yet supported.
1323 Start an early, unaccelerated console on the EFI
1324 memory mapped framebuffer (if available). On cache
1325 coherent non-x86 systems that use system memory for
1326 the framebuffer, pass the 'ram' option so that it is
1327 mapped with the correct attributes.
1330 Use early console provided by Freescale LINFlexD UART
1331 serial driver for NXP S32V234 SoCs. A valid base
1332 address must be provided, and the serial port must
1333 already be setup and configured.
1335 earlyprintk= [X86,SH,ARM,M68k,S390]
1339 earlyprintk=serial[,ttySn[,baudrate]]
1340 earlyprintk=serial[,0x...[,baudrate]]
1341 earlyprintk=ttySn[,baudrate]
1342 earlyprintk=dbgp[debugController#]
1343 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1344 earlyprintk=xdbc[xhciController#]
1347 earlyprintk is useful when the kernel crashes before
1348 the normal console is initialized. It is not enabled by
1349 default because it has some cosmetic problems.
1351 Append ",keep" to not disable it when the real console
1354 Only one of vga, serial, or usb debug port can
1357 Currently only ttyS0 and ttyS1 may be specified by
1358 name. Other I/O ports may be explicitly specified
1359 on some architectures (x86 and arm at least) by
1360 replacing ttySn with an I/O port address, like this:
1361 earlyprintk=serial,0x1008,115200
1362 You can find the port for a given device in
1363 /proc/tty/driver/serial:
1364 2: uart:ST16650V2 port:00001008 irq:18 ...
1366 Interaction with the standard serial driver is not
1369 The VGA output is eventually overwritten by
1372 The xen option can only be used in Xen domains.
1374 The sclp output can only be used on s390.
1376 The bios output can only be used on SuperH.
1378 The optional "force" to "pciserial" enables use of a
1379 PCI device even when its classcode is not of the
1382 edac_report= [HW,EDAC] Control how to report EDAC event
1383 Format: {"on" | "off" | "force"}
1384 on: enable EDAC to report H/W event. May be overridden
1385 by other higher priority error reporting module.
1386 off: disable H/W event reporting through EDAC.
1387 force: enforce the use of EDAC to report H/W event.
1391 Format: {"off" | "on" | "skip[mbr]"}
1394 Format: { "debug", "disable_early_pci_dma",
1395 "nochunk", "noruntime", "nosoftreserve",
1396 "novamap", "no_disable_early_pci_dma" }
1397 debug: enable misc debug output.
1398 disable_early_pci_dma: disable the busmaster bit on all
1399 PCI bridges while in the EFI boot stub.
1400 nochunk: disable reading files in "chunks" in the EFI
1401 boot stub, as chunking can cause problems with some
1402 firmware implementations.
1403 noruntime : disable EFI runtime services support
1404 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1405 attribute may cause the kernel to reserve the
1406 memory range for a memory mapping driver to
1407 claim. Specify efi=nosoftreserve to disable this
1408 reservation and treat the memory by its base type
1409 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1410 novamap: do not call SetVirtualAddressMap().
1411 no_disable_early_pci_dma: Leave the busmaster bit set
1412 on all PCI bridges while in the EFI boot stub
1414 efi_no_storage_paranoia [EFI; X86]
1415 Using this parameter you can use more than 50% of
1416 your efi variable storage. Use this parameter only if
1417 you are really sure that your UEFI does sane gc and
1418 fulfills the spec otherwise your board may brick.
1420 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1421 Add arbitrary attribute to specific memory range by
1422 updating original EFI memory map.
1423 Region of memory which aa attribute is added to is
1426 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1427 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1428 attribute is added to range 0x100000000-0x180000000 and
1429 0x10a0000000-0x1120000000.
1431 If efi_fake_mem=8G@9G:0x40000 is specified, the
1432 EFI_MEMORY_SP(0x40000) attribute is added to
1433 range 0x240000000-0x43fffffff.
1435 Using this parameter you can do debugging of EFI memmap
1436 related features. For example, you can do debugging of
1437 Address Range Mirroring feature even if your box
1438 doesn't support it, or mark specific memory as
1441 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1442 that is to be dynamically loaded by Linux. If there are
1443 multiple variables with the same name but with different
1444 vendor GUIDs, all of them will be loaded. See
1445 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1448 eisa_irq_edge= [PARISC,HW]
1449 See header of drivers/parisc/eisa.c.
1451 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1454 This is designed to be used in conjunction with
1455 the boot argument: earlyprintk=vga
1457 This parameter works in place of the kgdboc parameter
1458 but can only be used if the backing tty is available
1459 very early in the boot process. For early debugging
1460 via a serial port see kgdboc_earlycon instead.
1463 See comment before function elanfreq_setup() in
1464 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1466 elfcorehdr=[size[KMG]@]offset[KMG] [PPC,SH,X86,S390]
1467 Specifies physical address of start of kernel core
1468 image elf header and optionally the size. Generally
1469 kexec loader will pass this option to capture kernel.
1470 See Documentation/admin-guide/kdump/kdump.rst for details.
1472 enable_mtrr_cleanup [X86]
1473 The kernel tries to adjust MTRR layout from continuous
1474 to discrete, to make X server driver able to add WB
1475 entry later. This parameter enables that.
1477 enable_timer_pin_1 [X86]
1478 Enable PIN 1 of APIC timer
1479 Can be useful to work around chipset bugs
1480 (in particular on some ATI chipsets).
1481 The kernel tries to set a reasonable default.
1483 enforcing= [SELINUX] Set initial enforcing status.
1485 See security/selinux/Kconfig help text.
1486 0 -- permissive (log only, no denials).
1487 1 -- enforcing (deny and log).
1489 Value can be changed at runtime via
1490 /sys/fs/selinux/enforce.
1493 Disable Error Record Serialization Table (ERST)
1496 ether= [HW,NET] Ethernet cards parameters
1497 This option is obsoleted by the "netdev=" option, which
1498 has equivalent usage. See its documentation for details.
1502 Permit 'security.evm' to be updated regardless of
1503 current integrity status.
1505 early_page_ext [KNL] Enforces page_ext initialization to earlier
1506 stages so cover more early boot allocations.
1507 Please note that as side effect some optimizations
1508 might be disabled to achieve that (e.g. parallelized
1509 memory initialization is disabled) so the boot process
1510 might take longer, especially on systems with a lot of
1511 memory. Available with CONFIG_PAGE_EXTENSION=y.
1516 fail_make_request=[KNL]
1517 General fault injection mechanism.
1518 Format: <interval>,<probability>,<space>,<times>
1519 See also Documentation/fault-injection/.
1522 Format: { initns | none }
1523 See Documentation/admin-guide/sysctl/net.rst for
1524 fb_tunnels_only_for_init_ns
1527 See Documentation/admin-guide/blockdev/floppy.rst.
1530 Forcefully enable Physical Address Extension (PAE).
1531 Many Pentium M systems disable PAE but may have a
1532 functionally usable PAE implementation.
1533 Warning: use of this parameter will taint the kernel
1534 and may cause unknown problems.
1537 [FTRACE] will set and start the specified tracer
1538 as early as possible in order to facilitate early
1541 ftrace_boot_snapshot
1542 [FTRACE] On boot up, a snapshot will be taken of the
1543 ftrace ring buffer that can be read at:
1544 /sys/kernel/tracing/snapshot.
1545 This is useful if you need tracing information from kernel
1546 boot up that is likely to be overridden by user space
1547 start up functionality.
1549 Optionally, the snapshot can also be defined for a tracing
1550 instance that was created by the trace_instance= command
1553 trace_instance=foo,sched_switch ftrace_boot_snapshot=foo
1555 The above will cause the "foo" tracing instance to trigger
1556 a snapshot at the end of boot up.
1558 ftrace_dump_on_oops[=orig_cpu]
1559 [FTRACE] will dump the trace buffers on oops.
1560 If no parameter is passed, ftrace will dump
1561 buffers of all CPUs, but if you pass orig_cpu, it will
1562 dump only the buffer of the CPU that triggered the
1565 ftrace_filter=[function-list]
1566 [FTRACE] Limit the functions traced by the function
1567 tracer at boot up. function-list is a comma-separated
1568 list of functions. This list can be changed at run
1569 time by the set_ftrace_filter file in the debugfs
1572 ftrace_notrace=[function-list]
1573 [FTRACE] Do not trace the functions specified in
1574 function-list. This list can be changed at run time
1575 by the set_ftrace_notrace file in the debugfs
1578 ftrace_graph_filter=[function-list]
1579 [FTRACE] Limit the top level callers functions traced
1580 by the function graph tracer at boot up.
1581 function-list is a comma-separated list of functions
1582 that can be changed at run time by the
1583 set_graph_function file in the debugfs tracing directory.
1585 ftrace_graph_notrace=[function-list]
1586 [FTRACE] Do not trace from the functions specified in
1587 function-list. This list is a comma-separated list of
1588 functions that can be changed at run time by the
1589 set_graph_notrace file in the debugfs tracing directory.
1591 ftrace_graph_max_depth=<uint>
1592 [FTRACE] Used with the function graph tracer. This is
1593 the max depth it will trace into a function. This value
1594 can be changed at run time by the max_graph_depth file
1595 in the tracefs tracing directory. default: 0 (no limit)
1597 fw_devlink= [KNL] Create device links between consumer and supplier
1598 devices by scanning the firmware to infer the
1599 consumer/supplier relationships. This feature is
1600 especially useful when drivers are loaded as modules as
1601 it ensures proper ordering of tasks like device probing
1602 (suppliers first, then consumers), supplier boot state
1603 clean up (only after all consumers have probed),
1604 suspend/resume & runtime PM (consumers first, then
1606 Format: { off | permissive | on | rpm }
1607 off -- Don't create device links from firmware info.
1608 permissive -- Create device links from firmware info
1609 but use it only for ordering boot state clean
1610 up (sync_state() calls).
1611 on -- Create device links from firmware info and use it
1612 to enforce probe and suspend/resume ordering.
1613 rpm -- Like "on", but also use to order runtime PM.
1615 fw_devlink.strict=<bool>
1616 [KNL] Treat all inferred dependencies as mandatory
1617 dependencies. This only applies for fw_devlink=on|rpm.
1620 fw_devlink.sync_state =
1621 [KNL] When all devices that could probe have finished
1622 probing, this parameter controls what to do with
1623 devices that haven't yet received their sync_state()
1625 Format: { strict | timeout }
1626 strict -- Default. Continue waiting on consumers to
1628 timeout -- Give up waiting on consumers and call
1629 sync_state() on any devices that haven't yet
1630 received their sync_state() calls after
1631 deferred_probe_timeout has expired or by
1632 late_initcall() if !CONFIG_MODULES.
1635 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1636 support via parallel port (up to 5 devices per port)
1637 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1638 See also Documentation/input/devices/joystick-parport.rst
1642 gart_fix_e820= [X86-64] disable the fix e820 for K8 GART
1646 gather_data_sampling=
1647 [X86,INTEL] Control the Gather Data Sampling (GDS)
1650 Gather Data Sampling is a hardware vulnerability which
1651 allows unprivileged speculative access to data which was
1652 previously stored in vector registers.
1654 This issue is mitigated by default in updated microcode.
1655 The mitigation may have a performance impact but can be
1656 disabled. On systems without the microcode mitigation
1657 disabling AVX serves as a mitigation.
1659 force: Disable AVX to mitigate systems without
1660 microcode mitigation. No effect if the microcode
1661 mitigation is present. Known to cause crashes in
1662 userspace with buggy AVX enumeration.
1664 off: Disable GDS mitigation.
1666 gcov_persist= [GCOV] When non-zero (default), profiling data for
1667 kernel modules is saved and remains accessible via
1668 debugfs, even when the module is unloaded/reloaded.
1669 When zero, profiling data is discarded and associated
1670 debugfs files are removed at module unload time.
1672 goldfish [X86] Enable the goldfish android emulator platform.
1673 Don't use this when you are not running on the
1676 gpio-mockup.gpio_mockup_ranges
1677 [HW] Sets the ranges of gpiochip of for this device.
1678 Format: <start1>,<end1>,<start2>,<end2>...
1679 gpio-mockup.gpio_mockup_named_lines
1680 [HW] Let the driver know GPIO lines should be named.
1682 gpt [EFI] Forces disk with valid GPT signature but
1683 invalid Protective MBR to be treated as GPT. If the
1684 primary GPT is corrupted, it enables the backup/alternate
1685 GPT to be used instead.
1687 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1688 the "Enable 0" bit of the configuration register.
1691 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1692 the "Enable 0" bit of the configuration register.
1695 grcan.select= [HW] Select which physical interface to use.
1698 grcan.txsize= [HW] Sets the size of the tx buffer.
1699 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1701 grcan.rxsize= [HW] Sets the size of the rx buffer.
1702 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1706 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
1707 hardening is enabled for this boot. Hardened
1708 usercopy checking is used to protect the kernel
1709 from reading or writing beyond known memory
1710 allocation boundaries as a proactive defense
1711 against bounds-checking flaws in the kernel's
1712 copy_to_user()/copy_from_user() interface.
1713 on Perform hardened usercopy checks (default).
1714 off Disable hardened usercopy checks.
1716 hardlockup_all_cpu_backtrace=
1717 [KNL] Should the hard-lockup detector generate
1718 backtraces on all cpus.
1721 hashdist= [KNL,NUMA] Large hashes allocated during boot
1722 are distributed across NUMA nodes. Defaults on
1723 for 64-bit NUMA, off otherwise.
1724 Format: 0 | 1 (for off | on)
1726 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1728 hd= [EIDE] (E)IDE hard drive subsystem geometry
1729 Format: <cyl>,<head>,<sect>
1732 Disable Hardware Error Source Table (HEST) support;
1733 corresponding firmware-first mode error processing
1734 logic will be disabled.
1736 hibernate= [HIBERNATION]
1737 noresume Don't check if there's a hibernation image
1738 present during boot.
1739 nocompress Don't compress/decompress hibernation images.
1740 no Disable hibernation and resume.
1741 protect_image Turn on image protection during restoration
1742 (that will set all pages holding image data
1743 during restoration read-only).
1745 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1746 size of <nn>. This works even on boxes that have no
1747 highmem otherwise. This also works to reduce highmem
1748 size on bigger boxes.
1750 highres= [KNL] Enable/disable high resolution timer mode.
1751 Valid parameters: "on", "off"
1756 hostname= [KNL] Set the hostname (aka UTS nodename).
1758 This allows setting the system's hostname during early
1759 startup. This sets the name returned by gethostname.
1760 Using this parameter to set the hostname makes it
1761 possible to ensure the hostname is correctly set before
1762 any userspace processes run, avoiding the possibility
1763 that a process may call gethostname before the hostname
1764 has been explicitly set, resulting in the calling
1765 process getting an incorrect result. The string must
1766 not exceed the maximum allowed hostname length (usually
1767 64 characters) and will be truncated otherwise.
1769 hpet= [X86-32,HPET] option to control HPET usage
1770 Format: { enable (default) | disable | force |
1772 disable: disable HPET and use PIT instead
1773 force: allow force enabled of undocumented chips (ICH4,
1775 verbose: show contents of HPET registers during setup
1777 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1778 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1780 hugepages= [HW] Number of HugeTLB pages to allocate at boot.
1781 If this follows hugepagesz (below), it specifies
1782 the number of pages of hugepagesz to be allocated.
1783 If this is the first HugeTLB parameter on the command
1784 line, it specifies the number of pages to allocate for
1785 the default huge page size. If using node format, the
1786 number of pages to allocate per-node can be specified.
1787 See also Documentation/admin-guide/mm/hugetlbpage.rst.
1788 Format: <integer> or (node format)
1789 <node>:<integer>[,<node>:<integer>]
1792 [HW] The size of the HugeTLB pages. This is used in
1793 conjunction with hugepages (above) to allocate huge
1794 pages of a specific size at boot. The pair
1795 hugepagesz=X hugepages=Y can be specified once for
1796 each supported huge page size. Huge page sizes are
1797 architecture dependent. See also
1798 Documentation/admin-guide/mm/hugetlbpage.rst.
1801 hugetlb_cma= [HW,CMA] The size of a CMA area used for allocation
1802 of gigantic hugepages. Or using node format, the size
1803 of a CMA area per node can be specified.
1804 Format: nn[KMGTPE] or (node format)
1805 <node>:nn[KMGTPE][,<node>:nn[KMGTPE]]
1807 Reserve a CMA area of given size and allocate gigantic
1808 hugepages using the CMA allocator. If enabled, the
1809 boot-time allocation of gigantic hugepages is skipped.
1811 hugetlb_free_vmemmap=
1812 [KNL] Requires CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP
1814 Control if HugeTLB Vmemmap Optimization (HVO) is enabled.
1815 Allows heavy hugetlb users to free up some more
1816 memory (7 * PAGE_SIZE for each 2MB hugetlb page).
1817 Format: { on | off (default) }
1822 Built with CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP_DEFAULT_ON=y,
1825 Note that the vmemmap pages may be allocated from the added
1826 memory block itself when memory_hotplug.memmap_on_memory is
1827 enabled, those vmemmap pages cannot be optimized even if this
1828 feature is enabled. Other vmemmap pages not allocated from
1829 the added memory block itself do not be affected.
1832 [KNL] Should the hung task detector generate panics.
1835 A value of 1 instructs the kernel to panic when a
1836 hung task is detected. The default value is controlled
1837 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1838 option. The value selected by this boot parameter can
1839 be changed later by the kernel.hung_task_panic sysctl.
1841 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1842 terminal devices. Valid values: 0..8
1843 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1844 If specified, z/VM IUCV HVC accepts connections
1845 from listed z/VM user IDs only.
1847 hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations
1848 which allow the hypervisor to 'idle' the
1849 guest on lock contention.
1851 i2c_bus= [HW] Override the default board specific I2C bus speed
1852 or register an additional I2C bus that is not
1853 registered from board initialization code.
1857 i8042.debug [HW] Toggle i8042 debug mode
1858 i8042.unmask_kbd_data
1859 [HW] Enable printing of interrupt data from the KBD port
1860 (disabled by default, and as a pre-condition
1861 requires that i8042.debug=1 be enabled)
1862 i8042.direct [HW] Put keyboard port into non-translated mode
1863 i8042.dumbkbd [HW] Pretend that controller can only read data from
1864 keyboard and cannot control its state
1865 (Don't attempt to blink the leds)
1866 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1867 i8042.nokbd [HW] Don't check/create keyboard port
1868 i8042.noloop [HW] Disable the AUX Loopback command while probing
1870 i8042.nomux [HW] Don't check presence of an active multiplexing
1872 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1874 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1875 i8042.reset [HW] Reset the controller during init, cleanup and
1876 suspend-to-ram transitions, only during s2r
1877 transitions, or never reset
1878 Format: { 1 | Y | y | 0 | N | n }
1879 1, Y, y: always reset controller
1880 0, N, n: don't ever reset controller
1881 Default: only on s2r transitions on x86; most other
1882 architectures force reset to be always executed
1883 i8042.unlock [HW] Unlock (ignore) the keylock
1884 i8042.kbdreset [HW] Reset device connected to KBD port
1886 [HW] Allow deferred probing upon i8042 probe errors
1890 i915.invert_brightness=
1891 [DRM] Invert the sense of the variable that is used to
1892 set the brightness of the panel backlight. Normally a
1893 brightness value of 0 indicates backlight switched off,
1894 and the maximum of the brightness value sets the backlight
1895 to maximum brightness. If this parameter is set to 0
1896 (default) and the machine requires it, or this parameter
1897 is set to 1, a brightness value of 0 sets the backlight
1898 to maximum brightness, and the maximum of the brightness
1899 value switches the backlight off.
1900 -1 -- never invert brightness
1901 0 -- machine default
1902 1 -- force brightness inversion
1904 ia32_emulation= [X86-64]
1906 When true, allows loading 32-bit programs and executing 32-bit
1907 syscalls, essentially overriding IA32_EMULATION_DEFAULT_DISABLED at
1908 boot time. When false, unconditionally disables IA32 emulation.
1911 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1915 Format: idle=poll, idle=halt, idle=nomwait
1916 Poll forces a polling idle loop that can slightly
1917 improve the performance of waking up a idle CPU, but
1918 will use a lot of power and make the system run hot.
1920 idle=halt: Halt is forced to be used for CPU idle.
1921 In such case C2/C3 won't be used again.
1922 idle=nomwait: Disable mwait for CPU C-states
1926 Allow force disabling of Shared Virtual Memory (SVA)
1927 support for the idxd driver. By default it is set to
1930 idxd.tc_override= [HW]
1932 Allow override of default traffic class configuration
1933 for the device. By default it is set to false (0).
1935 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1936 Format: { strict | legacy | 2008 | relaxed }
1939 Choose which programs will be accepted for execution
1940 based on the IEEE 754 NaN encoding(s) supported by
1941 the FPU and the NaN encoding requested with the value
1942 of an ELF file header flag individually set by each
1943 binary. Hardware implementations are permitted to
1944 support either or both of the legacy and the 2008 NaN
1947 Available settings are as follows:
1948 strict accept binaries that request a NaN encoding
1949 supported by the FPU
1950 legacy only accept legacy-NaN binaries, if supported
1952 2008 only accept 2008-NaN binaries, if supported
1954 relaxed accept any binaries regardless of whether
1955 supported by the FPU
1957 The FPU emulator is always able to support both NaN
1958 encodings, so if no FPU hardware is present or it has
1959 been disabled with 'nofpu', then the settings of
1960 'legacy' and '2008' strap the emulator accordingly,
1961 'relaxed' straps the emulator for both legacy-NaN and
1962 2008-NaN, whereas 'strict' enables legacy-NaN only on
1963 legacy processors and both NaN encodings on MIPS32 or
1966 The setting for ABS.fmt/NEG.fmt instruction execution
1967 mode generally follows that for the NaN encoding,
1968 except where unsupported by hardware.
1970 ignore_loglevel [KNL]
1971 Ignore loglevel setting - this will print /all/
1972 kernel messages to the console. Useful for debugging.
1973 We also add it as printk module parameter, so users
1974 could change it dynamically, usually by
1975 /sys/module/printk/parameters/ignore_loglevel.
1978 Ignore RLIMIT_DATA setting for data mappings,
1979 print warning at first misuse. Can be changed via
1980 /sys/module/kernel/parameters/ignore_rlimit_data.
1982 ihash_entries= [KNL]
1983 Set number of hash buckets for inode cache.
1985 ima_appraise= [IMA] appraise integrity measurements
1986 Format: { "off" | "enforce" | "fix" | "log" }
1989 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
1990 The builtin appraise policy appraises all files
1993 ima_canonical_fmt [IMA]
1994 Use the canonical format for the binary runtime
1995 measurements, instead of host native format.
1998 Format: { md5 | sha1 | rmd160 | sha256 | sha384
2002 The list of supported hash algorithms is defined
2003 in crypto/hash_info.h.
2006 The builtin policies to load during IMA setup.
2007 Format: "tcb | appraise_tcb | secure_boot |
2008 fail_securely | critical_data"
2010 The "tcb" policy measures all programs exec'd, files
2011 mmap'd for exec, and all files opened with the read
2012 mode bit set by either the effective uid (euid=0) or
2015 The "appraise_tcb" policy appraises the integrity of
2016 all files owned by root.
2018 The "secure_boot" policy appraises the integrity
2019 of files (eg. kexec kernel image, kernel modules,
2020 firmware, policy, etc) based on file signatures.
2022 The "fail_securely" policy forces file signature
2023 verification failure also on privileged mounted
2024 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
2027 The "critical_data" policy measures kernel integrity
2030 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
2031 Load a policy which meets the needs of the Trusted
2032 Computing Base. This means IMA will measure all
2033 programs exec'd, files mmap'd for exec, and all files
2034 opened for read by uid=0.
2037 Select one of defined IMA measurements template formats.
2038 Formats: { "ima" | "ima-ng" | "ima-ngv2" | "ima-sig" |
2043 [IMA] Define a custom template format.
2044 Format: { "field1|...|fieldN" }
2046 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
2047 Format: <min_file_size>
2048 Set the minimal file size for using asynchronous hash.
2049 If left unspecified, ahash usage is disabled.
2051 ahash performance varies for different data sizes on
2052 different crypto accelerators. This option can be used
2053 to achieve the best performance for a particular HW.
2055 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
2057 Set hashing buffer size. Default: 4k.
2059 ahash performance varies for different chunk sizes on
2060 different crypto accelerators. This option can be used
2061 to achieve best performance for particular HW.
2065 Run specified binary instead of /sbin/init as init
2068 initcall_debug [KNL] Trace initcalls as they are executed. Useful
2069 for working out where the kernel is dying during
2072 initcall_blacklist= [KNL] Do not execute a comma-separated list of
2073 initcall functions. Useful for debugging built-in
2074 modules and initcalls.
2076 initramfs_async= [KNL]
2079 This parameter controls whether the initramfs
2080 image is unpacked asynchronously, concurrently
2081 with devices being probed and
2082 initialized. This should normally just work,
2083 but as a debugging aid, one can get the
2084 historical behaviour of the initramfs
2085 unpacking being completed before device_ and
2088 initrd= [BOOT] Specify the location of the initial ramdisk
2090 initrdmem= [KNL] Specify a physical address and size from which to
2091 load the initrd. If an initrd is compiled in or
2092 specified in the bootparams, it takes priority over this
2094 Format: ss[KMG],nn[KMG]
2097 init_on_alloc= [MM] Fill newly allocated pages and heap objects with
2100 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
2102 init_on_free= [MM] Fill freed pages and heap objects with zeroes.
2104 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
2106 init_pkru= [X86] Specify the default memory protection keys rights
2107 register contents for all processes. 0x55555554 by
2108 default (disallow access to all but pkey 0). Can
2109 override in debugfs after boot.
2111 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
2114 int_pln_enable [X86] Enable power limit notification interrupt
2116 integrity_audit=[IMA]
2117 Format: { "0" | "1" }
2118 0 -- basic integrity auditing messages. (Default)
2119 1 -- additional integrity auditing messages.
2121 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
2123 Enable intel iommu driver.
2125 Disable intel iommu driver.
2126 igfx_off [Default Off]
2127 By default, gfx is mapped as normal device. If a gfx
2128 device has a dedicated DMAR unit, the DMAR unit is
2129 bypassed by not enabling DMAR with this option. In
2130 this case, gfx device will use physical address for
2132 strict [Default Off]
2133 Deprecated, equivalent to iommu.strict=1.
2134 sp_off [Default Off]
2135 By default, super page will be supported if Intel IOMMU
2136 has the capability. With this option, super page will
2139 Enable the Intel IOMMU scalable mode if the hardware
2140 advertises that it has support for the scalable mode
2143 Disallow use of the Intel IOMMU scalable mode.
2144 tboot_noforce [Default Off]
2145 Do not force the Intel IOMMU enabled under tboot.
2146 By default, tboot will force Intel IOMMU on, which
2147 could harm performance of some high-throughput
2148 devices like 40GBit network cards, even if identity
2150 Note that using this option lowers the security
2151 provided by tboot because it makes the system
2152 vulnerable to DMA attacks.
2154 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
2155 0 disables intel_idle and fall back on acpi_idle.
2156 1 to 9 specify maximum depth of C-state.
2160 Do not enable intel_pstate as the default
2161 scaling driver for the supported processors
2163 Use intel_pstate driver to bypass the scaling
2164 governors layer of cpufreq and provides it own
2165 algorithms for p-state selection. There are two
2166 P-state selection algorithms provided by
2167 intel_pstate in the active mode: powersave and
2168 performance. The way they both operate depends
2169 on whether or not the hardware managed P-states
2170 (HWP) feature has been enabled in the processor
2171 and possibly on the processor model.
2173 Use intel_pstate as a scaling driver, but configure it
2174 to work with generic cpufreq governors (instead of
2175 enabling its internal governor). This mode cannot be
2176 used along with the hardware-managed P-states (HWP)
2179 Enable intel_pstate on systems that prohibit it by default
2180 in favor of acpi-cpufreq. Forcing the intel_pstate driver
2181 instead of acpi-cpufreq may disable platform features, such
2182 as thermal controls and power capping, that rely on ACPI
2183 P-States information being indicated to OSPM and therefore
2184 should be used with caution. This option does not work with
2185 processors that aren't supported by the intel_pstate driver
2186 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
2188 Do not enable hardware P state control (HWP)
2191 Only load intel_pstate on systems which support
2192 hardware P state control (HWP) if available.
2194 Enforce ACPI _PPC performance limits. If the Fixed ACPI
2195 Description Table, specifies preferred power management
2196 profile as "Enterprise Server" or "Performance Server",
2197 then this feature is turned on by default.
2199 Allow per-logical-CPU P-State performance control limits using
2200 cpufreq sysfs interface
2202 intremap= [X86-64, Intel-IOMMU]
2203 on enable Interrupt Remapping (default)
2204 off disable Interrupt Remapping
2205 nosid disable Source ID checking
2207 BIOS x2APIC opt-out request will be ignored
2208 nopost disable Interrupt Posting
2210 iomem= Disable strict checking of access to MMIO memory
2211 strict regions from userspace.
2226 nobypass [PPC/POWERNV]
2227 Disable IOMMU bypass, using IOMMU for PCI devices.
2229 iommu.forcedac= [ARM64, X86] Control IOVA allocation for PCI devices.
2230 Format: { "0" | "1" }
2231 0 - Try to allocate a 32-bit DMA address first, before
2232 falling back to the full range if needed.
2233 1 - Allocate directly from the full usable range,
2234 forcing Dual Address Cycle for PCI cards supporting
2235 greater than 32-bit addressing.
2237 iommu.strict= [ARM64, X86, S390] Configure TLB invalidation behaviour
2238 Format: { "0" | "1" }
2240 Request that DMA unmap operations use deferred
2241 invalidation of hardware TLBs, for increased
2242 throughput at the cost of reduced device isolation.
2243 Will fall back to strict mode if not supported by
2244 the relevant IOMMU driver.
2246 DMA unmap operations invalidate IOMMU hardware TLBs
2248 unset - Use value of CONFIG_IOMMU_DEFAULT_DMA_{LAZY,STRICT}.
2249 Note: on x86, strict mode specified via one of the
2250 legacy driver-specific options takes precedence.
2253 [ARM64, X86] Configure DMA to bypass the IOMMU by default.
2254 Format: { "0" | "1" }
2255 0 - Use IOMMU translation for DMA.
2256 1 - Bypass the IOMMU for DMA.
2257 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
2259 io7= [HW] IO7 for Marvel-based Alpha systems
2260 See comment before marvel_specify_io7 in
2261 arch/alpha/kernel/core_marvel.c.
2263 io_delay= [X86] I/O delay method
2265 Standard port 0x80 based delay
2267 Alternate port 0xed based delay (needed on some systems)
2269 Simple two microseconds delay
2274 See Documentation/admin-guide/nfs/nfsroot.rst.
2276 ipcmni_extend [KNL] Extend the maximum number of unique System V
2277 IPC identifiers from 32,768 to 16,777,216.
2279 irqaffinity= [SMP] Set the default irq affinity mask
2280 The argument is a cpu list, as described above.
2282 irqchip.gicv2_force_probe=
2285 Force the kernel to look for the second 4kB page
2286 of a GICv2 controller even if the memory range
2287 exposed by the device tree is too small.
2289 irqchip.gicv3_nolpi=
2291 Force the kernel to ignore the availability of
2292 LPIs (and by consequence ITSs). Intended for system
2293 that use the kernel as a bootloader, and thus want
2294 to let secondary kernels in charge of setting up
2297 irqchip.gicv3_pseudo_nmi= [ARM64]
2298 Enables support for pseudo-NMIs in the kernel. This
2299 requires the kernel to be built with
2300 CONFIG_ARM64_PSEUDO_NMI.
2303 When an interrupt is not handled search all handlers
2304 for it. Intended to get systems with badly broken
2308 When an interrupt is not handled search all handlers
2309 for it. Also check all handlers each timer
2310 interrupt. Intended to get systems with badly broken
2314 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
2316 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
2317 [Deprecated - use cpusets instead]
2318 Format: [flag-list,]<cpu-list>
2320 Specify one or more CPUs to isolate from disturbances
2321 specified in the flag list (default: domain):
2324 Disable the tick when a single task runs.
2326 A residual 1Hz tick is offloaded to workqueues, which you
2327 need to affine to housekeeping through the global
2328 workqueue's affinity configured via the
2329 /sys/devices/virtual/workqueue/cpumask sysfs file, or
2330 by using the 'domain' flag described below.
2332 NOTE: by default the global workqueue runs on all CPUs,
2333 so to protect individual CPUs the 'cpumask' file has to
2334 be configured manually after bootup.
2337 Isolate from the general SMP balancing and scheduling
2338 algorithms. Note that performing domain isolation this way
2339 is irreversible: it's not possible to bring back a CPU to
2340 the domains once isolated through isolcpus. It's strongly
2341 advised to use cpusets instead to disable scheduler load
2342 balancing through the "cpuset.sched_load_balance" file.
2343 It offers a much more flexible interface where CPUs can
2344 move in and out of an isolated set anytime.
2346 You can move a process onto or off an "isolated" CPU via
2347 the CPU affinity syscalls or cpuset.
2348 <cpu number> begins at 0 and the maximum value is
2349 "number of CPUs in system - 1".
2353 Isolate from being targeted by managed interrupts
2354 which have an interrupt mask containing isolated
2355 CPUs. The affinity of managed interrupts is
2356 handled by the kernel and cannot be changed via
2357 the /proc/irq/* interfaces.
2359 This isolation is best effort and only effective
2360 if the automatically assigned interrupt mask of a
2361 device queue contains isolated and housekeeping
2362 CPUs. If housekeeping CPUs are online then such
2363 interrupts are directed to the housekeeping CPU
2364 so that IO submitted on the housekeeping CPU
2365 cannot disturb the isolated CPU.
2367 If a queue's affinity mask contains only isolated
2368 CPUs then this parameter has no effect on the
2369 interrupt routing decision, though interrupts are
2370 only delivered when tasks running on those
2371 isolated CPUs submit IO. IO submitted on
2372 housekeeping CPUs has no influence on those
2375 The format of <cpu-list> is described above.
2379 ivrs_ioapic [HW,X86-64]
2380 Provide an override to the IOAPIC-ID<->DEVICE-ID
2381 mapping provided in the IVRS ACPI table.
2382 By default, PCI segment is 0, and can be omitted.
2384 For example, to map IOAPIC-ID decimal 10 to
2385 PCI segment 0x1 and PCI device 00:14.0,
2386 write the parameter as:
2387 ivrs_ioapic=10@0001:00:14.0
2390 * To map IOAPIC-ID decimal 10 to PCI device 00:14.0
2391 write the parameter as:
2392 ivrs_ioapic[10]=00:14.0
2393 * To map IOAPIC-ID decimal 10 to PCI segment 0x1 and
2394 PCI device 00:14.0 write the parameter as:
2395 ivrs_ioapic[10]=0001:00:14.0
2397 ivrs_hpet [HW,X86-64]
2398 Provide an override to the HPET-ID<->DEVICE-ID
2399 mapping provided in the IVRS ACPI table.
2400 By default, PCI segment is 0, and can be omitted.
2402 For example, to map HPET-ID decimal 10 to
2403 PCI segment 0x1 and PCI device 00:14.0,
2404 write the parameter as:
2405 ivrs_hpet=10@0001:00:14.0
2408 * To map HPET-ID decimal 0 to PCI device 00:14.0
2409 write the parameter as:
2410 ivrs_hpet[0]=00:14.0
2411 * To map HPET-ID decimal 10 to PCI segment 0x1 and
2412 PCI device 00:14.0 write the parameter as:
2413 ivrs_ioapic[10]=0001:00:14.0
2415 ivrs_acpihid [HW,X86-64]
2416 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2417 mapping provided in the IVRS ACPI table.
2418 By default, PCI segment is 0, and can be omitted.
2420 For example, to map UART-HID:UID AMD0020:0 to
2421 PCI segment 0x1 and PCI device ID 00:14.5,
2422 write the parameter as:
2423 ivrs_acpihid=AMD0020:0@0001:00:14.5
2426 * To map UART-HID:UID AMD0020:0 to PCI segment is 0,
2427 PCI device ID 00:14.5, write the parameter as:
2428 ivrs_acpihid[00:14.5]=AMD0020:0
2429 * To map UART-HID:UID AMD0020:0 to PCI segment 0x1 and
2430 PCI device ID 00:14.5, write the parameter as:
2431 ivrs_acpihid[0001:00:14.5]=AMD0020:0
2433 js= [HW,JOY] Analog joystick
2434 See Documentation/input/joydev/joystick.rst.
2437 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2438 report on every invalid memory access. Without this
2439 parameter KASAN will print report only for the first
2443 Do not unregister boot console at start. This is only
2444 useful for debugging when something happens in the window
2445 between unregistering the boot console and initializing
2448 keepinitrd [HW,ARM] See retain_initrd.
2450 kernelcore= [KNL,X86,IA-64,PPC]
2451 Format: nn[KMGTPE] | nn% | "mirror"
2452 This parameter specifies the amount of memory usable by
2453 the kernel for non-movable allocations. The requested
2454 amount is spread evenly throughout all nodes in the
2455 system as ZONE_NORMAL. The remaining memory is used for
2456 movable memory in its own zone, ZONE_MOVABLE. In the
2457 event, a node is too small to have both ZONE_NORMAL and
2458 ZONE_MOVABLE, kernelcore memory will take priority and
2459 other nodes will have a larger ZONE_MOVABLE.
2461 ZONE_MOVABLE is used for the allocation of pages that
2462 may be reclaimed or moved by the page migration
2463 subsystem. Note that allocations like PTEs-from-HighMem
2464 still use the HighMem zone if it exists, and the Normal
2465 zone if it does not.
2467 It is possible to specify the exact amount of memory in
2468 the form of "nn[KMGTPE]", a percentage of total system
2469 memory in the form of "nn%", or "mirror". If "mirror"
2470 option is specified, mirrored (reliable) memory is used
2471 for non-movable allocations and remaining memory is used
2472 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2473 are exclusive, so you cannot specify multiple forms.
2475 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
2476 Format: <Controller#>[,poll interval]
2477 The controller # is the number of the ehci usb debug
2478 port as it is probed via PCI. The poll interval is
2479 optional and is the number seconds in between
2480 each poll cycle to the debug port in case you need
2481 the functionality for interrupting the kernel with
2482 gdb or control-c on the dbgp connection. When
2483 not using this parameter you use sysrq-g to break into
2484 the kernel debugger.
2486 kgdboc= [KGDB,HW] kgdb over consoles.
2487 Requires a tty driver that supports console polling,
2488 or a supported polling keyboard driver (non-usb).
2489 Serial only format: <serial_device>[,baud]
2490 keyboard only format: kbd
2491 keyboard and serial format: kbd,<serial_device>[,baud]
2492 Optional Kernel mode setting:
2493 kms, kbd format: kms,kbd
2494 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2496 kgdboc_earlycon= [KGDB,HW]
2497 If the boot console provides the ability to read
2498 characters and can work in polling mode, you can use
2499 this parameter to tell kgdb to use it as a backend
2500 until the normal console is registered. Intended to
2501 be used together with the kgdboc parameter which
2502 specifies the normal console to transition to.
2504 The name of the early console should be specified
2505 as the value of this parameter. Note that the name of
2506 the early console might be different than the tty
2507 name passed to kgdboc. It's OK to leave the value
2508 blank and the first boot console that implements
2509 read() will be picked.
2511 kgdbwait [KGDB] Stop kernel execution and enter the
2512 kernel debugger at the earliest opportunity.
2514 kmac= [MIPS] Korina ethernet MAC address.
2515 Configure the RouterBoard 532 series on-chip
2516 Ethernet adapter MAC address.
2518 kmemleak= [KNL] Boot-time kmemleak enable/disable
2519 Valid arguments: on, off
2521 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2524 kprobe_event=[probe-list]
2525 [FTRACE] Add kprobe events and enable at boot time.
2526 The probe-list is a semicolon delimited list of probe
2527 definitions. Each definition is same as kprobe_events
2528 interface, but the parameters are comma delimited.
2529 For example, to add a kprobe event on vfs_read with
2530 arg1 and arg2, add to the command line;
2532 kprobe_event=p,vfs_read,$arg1,$arg2
2534 See also Documentation/trace/kprobetrace.rst "Kernel
2535 Boot Parameter" section.
2537 kpti= [ARM64] Control page table isolation of user
2538 and kernel address spaces.
2539 Default: enabled on cores which need mitigation.
2543 kunit.enable= [KUNIT] Enable executing KUnit tests. Requires
2544 CONFIG_KUNIT to be set to be fully enabled. The
2545 default value can be overridden via
2546 KUNIT_DEFAULT_ENABLED.
2547 Default is 1 (enabled)
2549 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2550 Default is 0 (don't ignore, but inject #GP)
2552 kvm.eager_page_split=
2553 [KVM,X86] Controls whether or not KVM will try to
2554 proactively split all huge pages during dirty logging.
2555 Eager page splitting reduces interruptions to vCPU
2556 execution by eliminating the write-protection faults
2557 and MMU lock contention that would otherwise be
2558 required to split huge pages lazily.
2560 VM workloads that rarely perform writes or that write
2561 only to a small region of VM memory may benefit from
2562 disabling eager page splitting to allow huge pages to
2563 still be used for reads.
2565 The behavior of eager page splitting depends on whether
2566 KVM_DIRTY_LOG_INITIALLY_SET is enabled or disabled. If
2567 disabled, all huge pages in a memslot will be eagerly
2568 split when dirty logging is enabled on that memslot. If
2569 enabled, eager page splitting will be performed during
2570 the KVM_CLEAR_DIRTY ioctl, and only for the pages being
2573 Eager page splitting is only supported when kvm.tdp_mmu=Y.
2577 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2578 Default is false (don't support).
2581 [KVM] Controls the software workaround for the
2582 X86_BUG_ITLB_MULTIHIT bug.
2583 force : Always deploy workaround.
2584 off : Never deploy workaround.
2585 auto : Deploy workaround based on the presence of
2586 X86_BUG_ITLB_MULTIHIT.
2590 If the software workaround is enabled for the host,
2591 guests do need not to enable it for nested guests.
2593 kvm.nx_huge_pages_recovery_ratio=
2594 [KVM] Controls how many 4KiB pages are periodically zapped
2595 back to huge pages. 0 disables the recovery, otherwise if
2596 the value is N KVM will zap 1/Nth of the 4KiB pages every
2597 period (see below). The default is 60.
2599 kvm.nx_huge_pages_recovery_period_ms=
2600 [KVM] Controls the time period at which KVM zaps 4KiB pages
2601 back to huge pages. If the value is a non-zero N, KVM will
2602 zap a portion (see ratio above) of the pages every N msecs.
2603 If the value is 0 (the default), KVM will pick a period based
2604 on the ratio, such that a page is zapped after 1 hour on average.
2606 kvm-amd.nested= [KVM,AMD] Control nested virtualization feature in
2607 KVM/SVM. Default is 1 (enabled).
2609 kvm-amd.npt= [KVM,AMD] Control KVM's use of Nested Page Tables,
2610 a.k.a. Two-Dimensional Page Tables. Default is 1
2611 (enabled). Disable by KVM if hardware lacks support
2615 [KVM,ARM] Select one of KVM/arm64's modes of operation.
2617 none: Forcefully disable KVM.
2619 nvhe: Standard nVHE-based mode, without support for
2622 protected: nVHE-based mode with support for guests whose
2623 state is kept private from the host.
2625 nested: VHE-based mode with support for nested
2626 virtualization. Requires at least ARMv8.3
2629 Defaults to VHE/nVHE based on hardware support. Setting
2630 mode to "protected" will disable kexec and hibernation
2631 for the host. "nested" is experimental and should be
2632 used with extreme caution.
2634 kvm-arm.vgic_v3_group0_trap=
2635 [KVM,ARM] Trap guest accesses to GICv3 group-0
2638 kvm-arm.vgic_v3_group1_trap=
2639 [KVM,ARM] Trap guest accesses to GICv3 group-1
2642 kvm-arm.vgic_v3_common_trap=
2643 [KVM,ARM] Trap guest accesses to GICv3 common
2646 kvm-arm.vgic_v4_enable=
2647 [KVM,ARM] Allow use of GICv4 for direct injection of
2650 kvm_cma_resv_ratio=n [PPC]
2651 Reserves given percentage from system memory area for
2652 contiguous memory allocation for KVM hash pagetable
2654 By default it reserves 5% of total system memory.
2658 kvm-intel.ept= [KVM,Intel] Control KVM's use of Extended Page Tables,
2659 a.k.a. Two-Dimensional Page Tables. Default is 1
2660 (enabled). Disable by KVM if hardware lacks support
2663 kvm-intel.emulate_invalid_guest_state=
2664 [KVM,Intel] Control whether to emulate invalid guest
2665 state. Ignored if kvm-intel.enable_unrestricted_guest=1,
2666 as guest state is never invalid for unrestricted
2667 guests. This param doesn't apply to nested guests (L2),
2668 as KVM never emulates invalid L2 guest state.
2669 Default is 1 (enabled).
2671 kvm-intel.flexpriority=
2672 [KVM,Intel] Control KVM's use of FlexPriority feature
2673 (TPR shadow). Default is 1 (enabled). Disable by KVM if
2674 hardware lacks support for it.
2677 [KVM,Intel] Control nested virtualization feature in
2678 KVM/VMX. Default is 1 (enabled).
2680 kvm-intel.unrestricted_guest=
2681 [KVM,Intel] Control KVM's use of unrestricted guest
2682 feature (virtualized real and unpaged mode). Default
2683 is 1 (enabled). Disable by KVM if EPT is disabled or
2684 hardware lacks support for it.
2686 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2689 Valid arguments: never, cond, always
2691 always: L1D cache flush on every VMENTER.
2692 cond: Flush L1D on VMENTER only when the code between
2693 VMEXIT and VMENTER can leak host memory.
2694 never: Disables the mitigation
2696 Default is cond (do L1 cache flush in specific instances)
2698 kvm-intel.vpid= [KVM,Intel] Control KVM's use of Virtual Processor
2699 Identification feature (tagged TLBs). Default is 1
2700 (enabled). Disable by KVM if hardware lacks support
2703 l1d_flush= [X86,INTEL]
2704 Control mitigation for L1D based snooping vulnerability.
2706 Certain CPUs are vulnerable to an exploit against CPU
2707 internal buffers which can forward information to a
2708 disclosure gadget under certain conditions.
2710 In vulnerable processors, the speculatively
2711 forwarded data can be used in a cache side channel
2712 attack, to access data to which the attacker does
2713 not have direct access.
2715 This parameter controls the mitigation. The
2718 on - enable the interface for the mitigation
2720 l1tf= [X86] Control mitigation of the L1TF vulnerability on
2723 The kernel PTE inversion protection is unconditionally
2724 enabled and cannot be disabled.
2727 Provides all available mitigations for the
2728 L1TF vulnerability. Disables SMT and
2729 enables all mitigations in the
2730 hypervisors, i.e. unconditional L1D flush.
2732 SMT control and L1D flush control via the
2733 sysfs interface is still possible after
2734 boot. Hypervisors will issue a warning
2735 when the first VM is started in a
2736 potentially insecure configuration,
2737 i.e. SMT enabled or L1D flush disabled.
2740 Same as 'full', but disables SMT and L1D
2741 flush runtime control. Implies the
2742 'nosmt=force' command line option.
2743 (i.e. sysfs control of SMT is disabled.)
2746 Leaves SMT enabled and enables the default
2747 hypervisor mitigation, i.e. conditional
2750 SMT control and L1D flush control via the
2751 sysfs interface is still possible after
2752 boot. Hypervisors will issue a warning
2753 when the first VM is started in a
2754 potentially insecure configuration,
2755 i.e. SMT enabled or L1D flush disabled.
2759 Disables SMT and enables the default
2760 hypervisor mitigation.
2762 SMT control and L1D flush control via the
2763 sysfs interface is still possible after
2764 boot. Hypervisors will issue a warning
2765 when the first VM is started in a
2766 potentially insecure configuration,
2767 i.e. SMT enabled or L1D flush disabled.
2770 Same as 'flush', but hypervisors will not
2771 warn when a VM is started in a potentially
2772 insecure configuration.
2775 Disables hypervisor mitigations and doesn't
2777 It also drops the swap size and available
2778 RAM limit restriction on both hypervisor and
2783 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2789 lapic [X86-32,APIC] Enable the local APIC even if BIOS
2792 lapic= [X86,APIC] Do not use TSC deadline
2793 value for LAPIC timer one-shot implementation. Default
2794 back to the programmable timer unit in the LAPIC.
2795 Format: notscdeadline
2797 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
2800 libata.dma= [LIBATA] DMA control
2801 libata.dma=0 Disable all PATA and SATA DMA
2802 libata.dma=1 PATA and SATA Disk DMA only
2803 libata.dma=2 ATAPI (CDROM) DMA only
2804 libata.dma=4 Compact Flash DMA only
2805 Combinations also work, so libata.dma=3 enables DMA
2806 for disks and CDROMs, but not CFs.
2808 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2809 libata.ignore_hpa=0 keep BIOS limits (default)
2810 libata.ignore_hpa=1 ignore limits, using full disk
2812 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2816 libata.force= [LIBATA] Force configurations. The format is a comma-
2817 separated list of "[ID:]VAL" where ID is PORT[.DEVICE].
2818 PORT and DEVICE are decimal numbers matching port, link
2819 or device. Basically, it matches the ATA ID string
2820 printed on console by libata. If the whole ID part is
2821 omitted, the last PORT and DEVICE values are used. If
2822 ID hasn't been specified yet, the configuration applies
2823 to all ports, links and devices.
2825 If only DEVICE is omitted, the parameter applies to
2826 the port and all links and devices behind it. DEVICE
2827 number of 0 either selects the first device or the
2828 first fan-out link behind PMP device. It does not
2829 select the host link. DEVICE number of 15 selects the
2830 host link and device attached to it.
2832 The VAL specifies the configuration to force. As long
2833 as there is no ambiguity, shortcut notation is allowed.
2834 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2835 The following configurations can be forced.
2837 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2838 Any ID with matching PORT is used.
2840 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2842 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2843 udma[/][16,25,33,44,66,100,133] notation is also
2846 * nohrst, nosrst, norst: suppress hard, soft and both
2849 * rstonce: only attempt one reset during hot-unplug
2852 * [no]dbdelay: Enable or disable the extra 200ms delay
2853 before debouncing a link PHY and device presence
2856 * [no]ncq: Turn on or off NCQ.
2858 * [no]ncqtrim: Enable or disable queued DSM TRIM.
2860 * [no]ncqati: Enable or disable NCQ trim on ATI chipset.
2862 * [no]trim: Enable or disable (unqueued) TRIM.
2864 * trim_zero: Indicate that TRIM command zeroes data.
2866 * max_trim_128m: Set 128M maximum trim size limit.
2868 * [no]dma: Turn on or off DMA transfers.
2870 * atapi_dmadir: Enable ATAPI DMADIR bridge support.
2872 * atapi_mod16_dma: Enable the use of ATAPI DMA for
2873 commands that are not a multiple of 16 bytes.
2875 * [no]dmalog: Enable or disable the use of the
2876 READ LOG DMA EXT command to access logs.
2878 * [no]iddevlog: Enable or disable access to the
2879 identify device data log.
2881 * [no]logdir: Enable or disable access to the general
2882 purpose log directory.
2884 * max_sec_128: Set transfer size limit to 128 sectors.
2886 * max_sec_1024: Set or clear transfer size limit to
2889 * max_sec_lba48: Set or clear transfer size limit to
2892 * [no]lpm: Enable or disable link power management.
2894 * [no]setxfer: Indicate if transfer speed mode setting
2897 * [no]fua: Disable or enable FUA (Force Unit Access)
2898 support for devices supporting this feature.
2900 * dump_id: Dump IDENTIFY data.
2902 * disable: Disable this device.
2904 If there are multiple matching configurations changing
2905 the same attribute, the last one is used.
2907 load_ramdisk= [RAM] [Deprecated]
2909 lockd.nlm_grace_period=P [NFS] Assign grace period.
2912 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2915 lockd.nlm_timeout=T [NFS] Assign timeout value.
2918 lockd.nlm_udpport=M [NFS] Assign UDP port.
2921 lockdown= [SECURITY]
2922 { integrity | confidentiality }
2923 Enable the kernel lockdown feature. If set to
2924 integrity, kernel features that allow userland to
2925 modify the running kernel are disabled. If set to
2926 confidentiality, kernel features that allow userland
2927 to extract confidential information from the kernel
2930 locktorture.acq_writer_lim= [KNL]
2931 Set the time limit in jiffies for a lock
2932 acquisition. Acquisitions exceeding this limit
2933 will result in a splat once they do complete.
2935 locktorture.bind_readers= [KNL]
2936 Specify the list of CPUs to which the readers are
2939 locktorture.bind_writers= [KNL]
2940 Specify the list of CPUs to which the writers are
2943 locktorture.call_rcu_chains= [KNL]
2944 Specify the number of self-propagating call_rcu()
2945 chains to set up. These are used to ensure that
2946 there is a high probability of an RCU grace period
2947 in progress at any given time. Defaults to 0,
2948 which disables these call_rcu() chains.
2950 locktorture.long_hold= [KNL]
2951 Specify the duration in milliseconds for the
2952 occasional long-duration lock hold time. Defaults
2953 to 100 milliseconds. Select 0 to disable.
2955 locktorture.nested_locks= [KNL]
2956 Specify the maximum lock nesting depth that
2957 locktorture is to exercise, up to a limit of 8
2958 (MAX_NESTED_LOCKS). Specify zero to disable.
2959 Note that this parameter is ineffective on types
2960 of locks that do not support nested acquisition.
2962 locktorture.nreaders_stress= [KNL]
2963 Set the number of locking read-acquisition kthreads.
2964 Defaults to being automatically set based on the
2965 number of online CPUs.
2967 locktorture.nwriters_stress= [KNL]
2968 Set the number of locking write-acquisition kthreads.
2970 locktorture.onoff_holdoff= [KNL]
2971 Set time (s) after boot for CPU-hotplug testing.
2973 locktorture.onoff_interval= [KNL]
2974 Set time (s) between CPU-hotplug operations, or
2975 zero to disable CPU-hotplug testing.
2977 locktorture.rt_boost= [KNL]
2978 Do periodic testing of real-time lock priority
2979 boosting. Select 0 to disable, 1 to boost
2980 only rt_mutex, and 2 to boost unconditionally.
2981 Defaults to 2, which might seem to be an
2982 odd choice, but which should be harmless for
2983 non-real-time spinlocks, due to their disabling
2984 of preemption. Note that non-realtime mutexes
2987 locktorture.rt_boost_factor= [KNL]
2988 Number that determines how often and for how
2989 long priority boosting is exercised. This is
2990 scaled down by the number of writers, so that the
2991 number of boosts per unit time remains roughly
2992 constant as the number of writers increases.
2993 On the other hand, the duration of each boost
2994 increases with the number of writers.
2996 locktorture.shuffle_interval= [KNL]
2997 Set task-shuffle interval (jiffies). Shuffling
2998 tasks allows some CPUs to go into dyntick-idle
2999 mode during the locktorture test.
3001 locktorture.shutdown_secs= [KNL]
3002 Set time (s) after boot system shutdown. This
3003 is useful for hands-off automated testing.
3005 locktorture.stat_interval= [KNL]
3006 Time (s) between statistics printk()s.
3008 locktorture.stutter= [KNL]
3009 Time (s) to stutter testing, for example,
3010 specifying five seconds causes the test to run for
3011 five seconds, wait for five seconds, and so on.
3012 This tests the locking primitive's ability to
3013 transition abruptly to and from idle.
3015 locktorture.torture_type= [KNL]
3016 Specify the locking implementation to test.
3018 locktorture.verbose= [KNL]
3019 Enable additional printk() statements.
3021 locktorture.writer_fifo= [KNL]
3022 Run the write-side locktorture kthreads at
3023 sched_set_fifo() real-time priority.
3025 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
3028 loglevel= All Kernel Messages with a loglevel smaller than the
3029 console loglevel will be printed to the console. It can
3030 also be changed with klogd or other programs. The
3031 loglevels are defined as follows:
3033 0 (KERN_EMERG) system is unusable
3034 1 (KERN_ALERT) action must be taken immediately
3035 2 (KERN_CRIT) critical conditions
3036 3 (KERN_ERR) error conditions
3037 4 (KERN_WARNING) warning conditions
3038 5 (KERN_NOTICE) normal but significant condition
3039 6 (KERN_INFO) informational
3040 7 (KERN_DEBUG) debug-level messages
3042 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
3043 in bytes. n must be a power of two and greater
3044 than the minimal size. The minimal size is defined
3045 by LOG_BUF_SHIFT kernel config parameter. There is
3046 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
3047 that allows to increase the default size depending on
3048 the number of CPUs. See init/Kconfig for more details.
3050 logo.nologo [FB] Disables display of the built-in Linux logo.
3051 This may be used to provide more screen space for
3052 kernel log messages and is useful when debugging
3053 kernel boot problems.
3055 lp=0 [LP] Specify parallel ports to use, e.g,
3056 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
3057 lp=reset first parallel port). 'lp=0' disables the
3058 lp=auto printer driver. 'lp=reset' (which can be
3059 specified in addition to the ports) causes
3060 attached printers to be reset. Using
3061 lp=port1,port2,... specifies the parallel ports
3062 to associate lp devices with, starting with
3063 lp0. A port specification may be 'none' to skip
3064 that lp device, or a parport name such as
3065 'parport0'. Specifying 'lp=auto' instead of a
3066 port specification list means that device IDs
3067 from each port should be examined, to see if
3068 an IEEE 1284-compliant printer is attached; if
3069 so, the driver will manage that printer.
3070 See also header of drivers/char/lp.c.
3073 Sets loops_per_jiffy to given constant, thus avoiding
3074 time-consuming boot-time autodetection (up to 250 ms per
3075 CPU). 0 enables autodetection (default). To determine
3076 the correct value for your kernel, boot with normal
3077 autodetection and see what value is printed. Note that
3078 on SMP systems the preset will be applied to all CPUs,
3079 which is likely to cause problems if your CPUs need
3080 significantly divergent settings. An incorrect value
3081 will cause delays in the kernel to be wrong, leading to
3082 unpredictable I/O errors and other breakage. Although
3083 unlikely, in the extreme case this might damage your
3087 Format: <io>,<irq>,<dma>
3089 lsm.debug [SECURITY] Enable LSM initialization debugging output.
3092 [SECURITY] Choose order of LSM initialization. This
3093 overrides CONFIG_LSM, and the "security=" parameter.
3095 machvec= [IA-64] Force the use of a particular machine-vector
3096 (machvec) in a generic kernel.
3097 Example: machvec=hpzx1
3099 machtype= [Loongson] Share the same kernel image file between
3100 different yeeloong laptops.
3101 Example: machtype=lemote-yeeloong-2f-7inch
3103 max_addr=nn[KMG] [KNL,BOOT,IA-64] All physical memory greater
3104 than or equal to this physical address is ignored.
3106 maxcpus= [SMP] Maximum number of processors that an SMP kernel
3107 will bring up during bootup. maxcpus=n : n >= 0 limits
3108 the kernel to bring up 'n' processors. Surely after
3109 bootup you can bring up the other plugged cpu by executing
3110 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
3111 only takes effect during system bootup.
3112 While n=0 is a special case, it is equivalent to "nosmp",
3113 which also disables the IO APIC.
3115 max_loop= [LOOP] The number of loop block devices that get
3116 (loop.max_loop) unconditionally pre-created at init time. The default
3117 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
3118 of statically allocating a predefined number, loop
3119 devices can be requested on-demand with the
3120 /dev/loop-control interface.
3122 mce [X86-32] Machine Check Exception
3124 mce=option [X86-64] See Documentation/arch/x86/x86_64/boot-options.rst
3126 md= [HW] RAID subsystems devices and level
3127 See Documentation/admin-guide/md.rst.
3130 Format: <first>,<last>
3131 Specifies range of consoles to be captured by the MDA.
3134 Control mitigation for the Micro-architectural Data
3135 Sampling (MDS) vulnerability.
3137 Certain CPUs are vulnerable to an exploit against CPU
3138 internal buffers which can forward information to a
3139 disclosure gadget under certain conditions.
3141 In vulnerable processors, the speculatively
3142 forwarded data can be used in a cache side channel
3143 attack, to access data to which the attacker does
3144 not have direct access.
3146 This parameter controls the MDS mitigation. The
3149 full - Enable MDS mitigation on vulnerable CPUs
3150 full,nosmt - Enable MDS mitigation and disable
3151 SMT on vulnerable CPUs
3152 off - Unconditionally disable MDS mitigation
3154 On TAA-affected machines, mds=off can be prevented by
3155 an active TAA mitigation as both vulnerabilities are
3156 mitigated with the same mechanism so in order to disable
3157 this mitigation, you need to specify tsx_async_abort=off
3160 Not specifying this option is equivalent to
3163 For details see: Documentation/admin-guide/hw-vuln/mds.rst
3165 mem=nn[KMG] [HEXAGON] Set the memory size.
3166 Must be specified, otherwise memory size will be 0.
3168 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
3169 Amount of memory to be used in cases as follows:
3172 2 when the kernel is not able to see the whole system memory;
3173 3 memory that lies after 'mem=' boundary is excluded from
3174 the hypervisor, then assigned to KVM guests.
3175 4 to limit the memory available for kdump kernel.
3177 [ARC,MICROBLAZE] - the limit applies only to low memory,
3178 high memory is not affected.
3180 [ARM64] - only limits memory covered by the linear
3181 mapping. The NOMAP regions are not affected.
3183 [X86] Work as limiting max address. Use together
3184 with memmap= to avoid physical address space collisions.
3185 Without memmap= PCI devices could be placed at addresses
3186 belonging to unused RAM.
3188 Note that this only takes effects during boot time since
3189 in above case 3, memory may need be hot added after boot
3190 if system memory of hypervisor is not sufficient.
3193 [ARM,MIPS] - override the memory layout reported by
3195 Define a memory region of size nn[KMG] starting at
3197 Multiple different regions can be specified with
3198 multiple mem= parameters on the command line.
3200 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
3203 memblock=debug [KNL] Enable memblock debug messages.
3206 [KNL,SH] Allow user to override the default size for
3207 per-device physically contiguous DMA buffers.
3209 memhp_default_state=online/offline/online_kernel/online_movable
3210 [KNL] Set the initial state for the memory hotplug
3211 onlining policy. If not specified, the default value is
3212 set according to the
3213 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
3215 See Documentation/admin-guide/mm/memory-hotplug.rst.
3217 memmap=exactmap [KNL,X86] Enable setting of an exact
3218 E820 memory map, as specified by the user.
3219 Such memmap=exactmap lines can be constructed based on
3220 BIOS output or other requirements. See the memmap=nn@ss
3223 memmap=nn[KMG]@ss[KMG]
3224 [KNL, X86, MIPS, XTENSA] Force usage of a specific region of memory.
3225 Region of memory to be used is from ss to ss+nn.
3226 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
3227 which limits max address to nn[KMG].
3228 Multiple different regions can be specified,
3231 memmap=100M@2G,100M#3G,1G!1024G
3233 memmap=nn[KMG]#ss[KMG]
3234 [KNL,ACPI] Mark specific memory as ACPI data.
3235 Region of memory to be marked is from ss to ss+nn.
3237 memmap=nn[KMG]$ss[KMG]
3238 [KNL,ACPI] Mark specific memory as reserved.
3239 Region of memory to be reserved is from ss to ss+nn.
3240 Example: Exclude memory from 0x18690000-0x1869ffff
3241 memmap=64K$0x18690000
3243 memmap=0x10000$0x18690000
3244 Some bootloaders may need an escape character before '$',
3245 like Grub2, otherwise '$' and the following number
3248 memmap=nn[KMG]!ss[KMG]
3249 [KNL,X86] Mark specific memory as protected.
3250 Region of memory to be used, from ss to ss+nn.
3251 The memory region may be marked as e820 type 12 (0xc)
3252 and is NVDIMM or ADR memory.
3254 memmap=<size>%<offset>-<oldtype>+<newtype>
3255 [KNL,ACPI] Convert memory within the specified region
3256 from <oldtype> to <newtype>. If "-<oldtype>" is left
3257 out, the whole region will be marked as <newtype>,
3258 even if previously unavailable. If "+<newtype>" is left
3259 out, matching memory will be removed. Types are
3260 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
3261 3 = ACPI, 12 = PRAM.
3263 memory_corruption_check=0/1 [X86]
3264 Some BIOSes seem to corrupt the first 64k of
3265 memory when doing things like suspend/resume.
3266 Setting this option will scan the memory
3267 looking for corruption. Enabling this will
3268 both detect corruption and prevent the kernel
3269 from using the memory being corrupted.
3270 However, its intended as a diagnostic tool; if
3271 repeatable BIOS-originated corruption always
3272 affects the same memory, you can use memmap=
3273 to prevent the kernel from using that memory.
3275 memory_corruption_check_size=size [X86]
3276 By default it checks for corruption in the low
3277 64k, making this memory unavailable for normal
3278 use. Use this parameter to scan for
3279 corruption in more or less memory.
3281 memory_corruption_check_period=seconds [X86]
3282 By default it checks for corruption every 60
3283 seconds. Use this parameter to check at some
3284 other rate. 0 disables periodic checking.
3286 memory_hotplug.memmap_on_memory
3287 [KNL,X86,ARM] Boolean flag to enable this feature.
3288 Format: {on | off (default)}
3289 When enabled, runtime hotplugged memory will
3290 allocate its internal metadata (struct pages,
3291 those vmemmap pages cannot be optimized even
3292 if hugetlb_free_vmemmap is enabled) from the
3293 hotadded memory which will allow to hotadd a
3294 lot of memory without requiring additional
3296 This feature is disabled by default because it
3297 has some implication on large (e.g. GB)
3298 allocations in some configurations (e.g. small
3300 The state of the flag can be read in
3301 /sys/module/memory_hotplug/parameters/memmap_on_memory.
3302 Note that even when enabled, there are a few cases where
3303 the feature is not effective.
3305 memtest= [KNL,X86,ARM,M68K,PPC,RISCV] Enable memtest
3307 default : 0 <disable>
3308 Specifies the number of memtest passes to be
3309 performed. Each pass selects another test
3310 pattern from a given set of patterns. Memtest
3311 fills the memory with this pattern, validates
3312 memory contents and reserves bad memory
3313 regions that are detected.
3315 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
3316 Valid arguments: on, off
3317 Default (depends on kernel configuration option):
3318 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
3319 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
3320 mem_encrypt=on: Activate SME
3321 mem_encrypt=off: Do not activate SME
3323 Refer to Documentation/virt/kvm/x86/amd-memory-encryption.rst
3324 for details on when memory encryption can be activated.
3326 mem_sleep_default= [SUSPEND] Default system suspend mode:
3327 s2idle - Suspend-To-Idle
3328 shallow - Power-On Suspend or equivalent (if supported)
3329 deep - Suspend-To-RAM or equivalent (if supported)
3330 See Documentation/admin-guide/pm/sleep-states.rst.
3332 mfgpt_irq= [IA-32] Specify the IRQ to use for the
3333 Multi-Function General Purpose Timers on AMD Geode
3336 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
3337 the BIOS has incorrectly applied a workaround. TinyBIOS
3338 version 0.98 is known to be affected, 0.99 fixes the
3339 problem by letting the user disable the workaround.
3343 microcode.force_minrev= [X86]
3345 Enable or disable the microcode minimal revision
3346 enforcement for the runtime microcode loader.
3348 min_addr=nn[KMG] [KNL,BOOT,IA-64] All physical memory below this
3349 physical address is ignored.
3351 mini2440= [ARM,HW,KNL]
3352 Format:[0..2][b][c][t]
3354 MINI2440 configuration specification:
3355 0 - The attached screen is the 3.5" TFT
3356 1 - The attached screen is the 7" TFT
3357 2 - The VGA Shield is attached (1024x768)
3358 Leaving out the screen size parameter will not load
3359 the TFT driver, and the framebuffer will be left
3361 b - Enable backlight. The TFT backlight pin will be
3362 linked to the kernel VESA blanking code and a GPIO
3363 LED. This parameter is not necessary when using the
3365 c - Enable the s3c camera interface.
3366 t - Reserved for enabling touchscreen support. The
3367 touchscreen support is not enabled in the mainstream
3368 kernel as of 2.6.30, a preliminary port can be found
3369 in the "bleeding edge" mini2440 support kernel at
3370 https://repo.or.cz/w/linux-2.6/mini2440.git
3373 [X86,PPC,S390,ARM64] Control optional mitigations for
3374 CPU vulnerabilities. This is a set of curated,
3375 arch-independent options, each of which is an
3376 aggregation of existing arch-specific options.
3379 Disable all optional CPU mitigations. This
3380 improves system performance, but it may also
3381 expose users to several CPU vulnerabilities.
3382 Equivalent to: if nokaslr then kpti=0 [ARM64]
3383 gather_data_sampling=off [X86]
3384 kvm.nx_huge_pages=off [X86]
3387 mmio_stale_data=off [X86]
3388 no_entry_flush [PPC]
3389 no_uaccess_flush [PPC]
3392 nospectre_bhb [ARM64]
3393 nospectre_v1 [X86,PPC]
3394 nospectre_v2 [X86,PPC,S390,ARM64]
3396 spec_store_bypass_disable=off [X86,PPC]
3397 spectre_v2_user=off [X86]
3398 srbds=off [X86,INTEL]
3399 ssbd=force-off [ARM64]
3400 tsx_async_abort=off [X86]
3403 This does not have any effect on
3404 kvm.nx_huge_pages when
3405 kvm.nx_huge_pages=force.
3408 Mitigate all CPU vulnerabilities, but leave SMT
3409 enabled, even if it's vulnerable. This is for
3410 users who don't want to be surprised by SMT
3411 getting disabled across kernel upgrades, or who
3412 have other ways of avoiding SMT-based attacks.
3413 Equivalent to: (default behavior)
3416 Mitigate all CPU vulnerabilities, disabling SMT
3417 if needed. This is for users who always want to
3418 be fully mitigated, even if it means losing SMT.
3419 Equivalent to: l1tf=flush,nosmt [X86]
3420 mds=full,nosmt [X86]
3421 tsx_async_abort=full,nosmt [X86]
3422 mmio_stale_data=full,nosmt [X86]
3423 retbleed=auto,nosmt [X86]
3426 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
3427 parameter allows control of the logging verbosity for
3428 the additional memory initialisation checks. A value
3429 of 0 disables mminit logging and a level of 4 will
3430 log everything. Information is printed at KERN_DEBUG
3431 so loglevel=8 may also need to be specified.
3434 [X86,INTEL] Control mitigation for the Processor
3435 MMIO Stale Data vulnerabilities.
3437 Processor MMIO Stale Data is a class of
3438 vulnerabilities that may expose data after an MMIO
3439 operation. Exposed data could originate or end in
3440 the same CPU buffers as affected by MDS and TAA.
3441 Therefore, similar to MDS and TAA, the mitigation
3442 is to clear the affected CPU buffers.
3444 This parameter controls the mitigation. The
3447 full - Enable mitigation on vulnerable CPUs
3449 full,nosmt - Enable mitigation and disable SMT on
3452 off - Unconditionally disable mitigation
3454 On MDS or TAA affected machines,
3455 mmio_stale_data=off can be prevented by an active
3456 MDS or TAA mitigation as these vulnerabilities are
3457 mitigated with the same mechanism so in order to
3458 disable this mitigation, you need to specify
3459 mds=off and tsx_async_abort=off too.
3461 Not specifying this option is equivalent to
3462 mmio_stale_data=full.
3465 Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst
3467 <module>.async_probe[=<bool>] [KNL]
3468 If no <bool> value is specified or if the value
3469 specified is not a valid <bool>, enable asynchronous
3470 probe on this module. Otherwise, enable/disable
3471 asynchronous probe on this module as indicated by the
3472 <bool> value. See also: module.async_probe
3474 module.async_probe=<bool>
3475 [KNL] When set to true, modules will use async probing
3476 by default. To enable/disable async probing for a
3477 specific module, use the module specific control that
3478 is documented under <module>.async_probe. When both
3479 module.async_probe and <module>.async_probe are
3480 specified, <module>.async_probe takes precedence for
3481 the specific module.
3483 module.enable_dups_trace
3484 [KNL] When CONFIG_MODULE_DEBUG_AUTOLOAD_DUPS is set,
3485 this means that duplicate request_module() calls will
3486 trigger a WARN_ON() instead of a pr_warn(). Note that
3487 if MODULE_DEBUG_AUTOLOAD_DUPS_TRACE is set, WARN_ON()s
3488 will always be issued and this option does nothing.
3490 [KNL] When CONFIG_MODULE_SIG is set, this means that
3491 modules without (valid) signatures will fail to load.
3492 Note that if CONFIG_MODULE_SIG_FORCE is set, that
3493 is always true, so this option does nothing.
3495 module_blacklist= [KNL] Do not load a comma-separated list of
3496 modules. Useful for debugging problem modules.
3499 [MOUSE] Maximum time between finger touching and
3500 leaving touchpad surface for touch to be considered
3501 a tap and be reported as a left button click (for
3502 touchpads working in absolute mode only).
3504 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
3505 reporting absolute coordinates, such as tablets
3506 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
3507 reporting absolute coordinates, such as tablets
3509 movablecore= [KNL,X86,IA-64,PPC]
3510 Format: nn[KMGTPE] | nn%
3511 This parameter is the complement to kernelcore=, it
3512 specifies the amount of memory used for migratable
3513 allocations. If both kernelcore and movablecore is
3514 specified, then kernelcore will be at *least* the
3515 specified value but may be more. If movablecore on its
3516 own is specified, the administrator must be careful
3517 that the amount of memory usable for all allocations
3520 movable_node [KNL] Boot-time switch to make hotplugable memory
3521 NUMA nodes to be movable. This means that the memory
3522 of such nodes will be usable only for movable
3523 allocations which rules out almost all kernel
3524 allocations. Use with caution!
3526 MTD_Partition= [MTD]
3527 Format: <name>,<region-number>,<size>,<offset>
3529 MTD_Region= [MTD] Format:
3530 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
3533 See drivers/mtd/parsers/cmdlinepart.c
3536 ARM/S3C2412 JIVE boot control
3538 See arch/arm/mach-s3c/mach-jive.c
3540 mtouchusb.raw_coordinates=
3541 [HW] Make the MicroTouch USB driver use raw coordinates
3542 ('y', default) or cooked coordinates ('n')
3545 Enable printing debug information related to MTRR
3546 registers at boot time.
3548 mtrr_chunk_size=nn[KMG] [X86]
3549 used for mtrr cleanup. It is largest continuous chunk
3550 that could hold holes aka. UC entries.
3552 mtrr_gran_size=nn[KMG] [X86]
3553 Used for mtrr cleanup. It is granularity of mtrr block.
3555 Large value could prevent small alignment from
3558 mtrr_spare_reg_nr=n [X86]
3560 Range: 0,7 : spare reg number
3562 Used for mtrr cleanup. It is spare mtrr entries number.
3563 Set to 2 or more if your graphical card needs more.
3565 multitce=off [PPC] This parameter disables the use of the pSeries
3566 firmware feature for updating multiple TCE entries
3569 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
3571 netdev= [NET] Network devices parameters
3572 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
3573 Note that mem_start is often overloaded to mean
3574 something different and driver-specific.
3575 This usage is only documented in each driver source
3578 netpoll.carrier_timeout=
3579 [NET] Specifies amount of time (in seconds) that
3580 netpoll should wait for a carrier. By default netpoll
3584 [NETFILTER] Enable connection tracking flow accounting
3585 0 to disable accounting
3586 1 to enable accounting
3590 [NFS] sets the pathname to the program which is used
3591 to update the NFS client cache entries.
3593 nfs.cache_getent_timeout=
3594 [NFS] sets the timeout after which an attempt to
3595 update a cache entry is deemed to have failed.
3597 nfs.callback_nr_threads=
3598 [NFSv4] set the total number of threads that the
3599 NFS client will assign to service NFSv4 callback
3602 nfs.callback_tcpport=
3603 [NFS] set the TCP port on which the NFSv4 callback
3604 channel should listen.
3607 [NFS] specifies the number of times the NFSv4 client
3608 retries the request before returning an EAGAIN error,
3609 after a reply of NFS4ERR_DELAY from the server.
3610 Only applies if the softerr mount option is enabled,
3611 and the specified value is >= 0.
3614 [NFS] enable 64-bit inode numbers.
3615 If zero, the NFS client will fake up a 32-bit inode
3616 number for the readdir() and stat() syscalls instead
3617 of returning the full 64-bit number.
3618 The default is to return 64-bit inode numbers.
3620 nfs.idmap_cache_timeout=
3621 [NFS] set the maximum lifetime for idmapper cache
3624 nfs.max_session_cb_slots=
3625 [NFSv4.1] Sets the maximum number of session
3626 slots the client will assign to the callback
3627 channel. This determines the maximum number of
3628 callbacks the client will process in parallel for
3629 a particular server.
3631 nfs.max_session_slots=
3632 [NFSv4.1] Sets the maximum number of session slots
3633 the client will attempt to negotiate with the server.
3634 This limits the number of simultaneous RPC requests
3635 that the client can send to the NFSv4.1 server.
3636 Note that there is little point in setting this
3637 value higher than the max_tcp_slot_table_limit.
3639 nfs.nfs4_disable_idmapping=
3640 [NFSv4] When set to the default of '1', this option
3641 ensures that both the RPC level authentication
3642 scheme and the NFS level operations agree to use
3643 numeric uids/gids if the mount is using the
3644 'sec=sys' security flavour. In effect it is
3645 disabling idmapping, which can make migration from
3646 legacy NFSv2/v3 systems to NFSv4 easier.
3647 Servers that do not support this mode of operation
3648 will be autodetected by the client, and it will fall
3649 back to using the idmapper.
3650 To turn off this behaviour, set the value to '0'.
3653 [NFS4] Specify an additional fixed unique ident-
3654 ification string that NFSv4 clients can insert into
3655 their nfs_client_id4 string. This is typically a
3656 UUID that is generated at system install time.
3658 nfs.recover_lost_locks=
3659 [NFSv4] Attempt to recover locks that were lost due
3660 to a lease timeout on the server. Please note that
3661 doing this risks data corruption, since there are
3662 no guarantees that the file will remain unchanged
3663 after the locks are lost.
3664 If you want to enable the kernel legacy behaviour of
3665 attempting to recover these locks, then set this
3667 The default parameter value of '0' causes the kernel
3668 not to attempt recovery of lost locks.
3670 nfs.send_implementation_id=
3671 [NFSv4.1] Send client implementation identification
3672 information in exchange_id requests.
3673 If zero, no implementation identification information
3675 The default is to send the implementation identification
3678 nfs4.layoutstats_timer=
3679 [NFSv4.2] Change the rate at which the kernel sends
3680 layoutstats to the pNFS metadata server.
3682 Setting this to value to 0 causes the kernel to use
3683 whatever value is the default set by the layout
3684 driver. A non-zero value sets the minimum interval
3685 in seconds between layoutstats transmissions.
3687 nfsd.inter_copy_offload_enable=
3688 [NFSv4.2] When set to 1, the server will support
3689 server-to-server copies for which this server is
3690 the destination of the copy.
3692 nfsd.nfs4_disable_idmapping=
3693 [NFSv4] When set to the default of '1', the NFSv4
3694 server will return only numeric uids and gids to
3695 clients using auth_sys, and will accept numeric uids
3696 and gids from such clients. This is intended to ease
3697 migration from NFSv2/v3.
3699 nfsd.nfsd4_ssc_umount_timeout=
3700 [NFSv4.2] When used as the destination of a
3701 server-to-server copy, knfsd temporarily mounts
3702 the source server. It caches the mount in case
3703 it will be needed again, and discards it if not
3704 used for the number of milliseconds specified by
3707 nfsaddrs= [NFS] Deprecated. Use ip= instead.
3708 See Documentation/admin-guide/nfs/nfsroot.rst.
3710 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
3711 See Documentation/admin-guide/nfs/nfsroot.rst.
3713 nfsrootdebug [NFS] enable nfsroot debugging messages.
3714 See Documentation/admin-guide/nfs/nfsroot.rst.
3716 nmi_backtrace.backtrace_idle [KNL]
3717 Dump stacks even of idle CPUs in response to an
3718 NMI stack-backtrace request.
3720 nmi_debug= [KNL,SH] Specify one or more actions to take
3721 when a NMI is triggered.
3722 Format: [state][,regs][,debounce][,die]
3724 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
3725 Format: [panic,][nopanic,][num]
3727 0 - turn hardlockup detector in nmi_watchdog off
3728 1 - turn hardlockup detector in nmi_watchdog on
3729 When panic is specified, panic when an NMI watchdog
3730 timeout occurs (or 'nopanic' to not panic on an NMI
3731 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
3732 To disable both hard and soft lockup detectors,
3733 please see 'nowatchdog'.
3734 This is useful when you use a panic=... timeout and
3735 need the box quickly up again.
3737 These settings can be accessed at runtime via
3738 the nmi_watchdog and hardlockup_panic sysctls.
3740 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3741 emulation library even if a 387 maths coprocessor
3744 no4lvl [RISCV] Disable 4-level and 5-level paging modes. Forces
3745 kernel to use 3-level paging instead.
3747 no5lvl [X86-64,RISCV] Disable 5-level paging mode. Forces
3748 kernel to use 4-level paging instead.
3750 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
3751 caches in the slab allocator. Saves per-node memory,
3752 but will impact performance.
3756 noaltinstr [S390] Disables alternative instructions patching
3757 (CPU alternatives feature).
3759 noapic [SMP,APIC] Tells the kernel to not make use of any
3760 IOAPICs that may be present in the system.
3762 noautogroup Disable scheduler automatic task group creation.
3767 [HW] Never suspend the console
3768 Disable suspending of consoles during suspend and
3769 hibernate operations. Once disabled, debugging
3770 messages can reach various consoles while the rest
3771 of the system is being put to sleep (ie, while
3772 debugging driver suspend/resume hooks). This may
3773 not work reliably with all consoles, but is known
3774 to work with serial and VGA consoles.
3775 To facilitate more flexible debugging, we also add
3776 console_suspend, a printk module parameter to control
3777 it. Users could use console_suspend (usually
3778 /sys/module/printk/parameters/console_suspend) to
3779 turn on/off it dynamically.
3782 [KNL] Disable object debugging
3784 nodsp [SH] Disable hardware DSP at boot time.
3786 noefi Disable EFI runtime services support.
3788 no_entry_flush [PPC] Don't flush the L1-D cache when entering the kernel.
3793 This affects only 32-bit executables.
3794 noexec32=on: enable non-executable mappings (default)
3795 read doesn't imply executable mappings
3796 noexec32=off: disable non-executable mappings
3797 read implies executable mappings
3799 no_file_caps Tells the kernel not to honor file capabilities. The
3800 only way then for a file to be executed with privilege
3801 is to be setuid root or executed by root.
3803 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3805 nofsgsbase [X86] Disables FSGSBASE instructions.
3807 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3808 register save and restore. The kernel will only save
3809 legacy floating-point registers on task switch.
3811 nohalt [IA-64] Tells the kernel not to use the power saving
3812 function PAL_HALT_LIGHT when idle. This increases
3813 power-consumption. On the positive side, it reduces
3814 interrupt wake-up latency, which may improve performance
3815 in certain environments such as networked servers or
3819 Force pointers printed to the console or buffers to be
3820 unhashed. By default, when a pointer is printed via %p
3821 format string, that pointer is "hashed", i.e. obscured
3822 by hashing the pointer value. This is a security feature
3823 that hides actual kernel addresses from unprivileged
3824 users, but it also makes debugging the kernel more
3825 difficult since unequal pointers can no longer be
3826 compared. However, if this command-line option is
3827 specified, then all normal pointers will have their true
3828 value printed. This option should only be specified when
3829 debugging the kernel. Please do not use on production
3832 nohibernate [HIBERNATION] Disable hibernation and resume.
3834 nohlt [ARM,ARM64,MICROBLAZE,MIPS,PPC,SH] Forces the kernel to
3835 busy wait in do_idle() and not use the arch_cpu_idle()
3836 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
3837 to be effective. This is useful on platforms where the
3838 sleep(SH) or wfi(ARM,ARM64) instructions do not work
3839 correctly or when doing power measurements to evaluate
3840 the impact of the sleep instructions. This is also
3841 useful when using JTAG debugger.
3843 nohugeiomap [KNL,X86,PPC,ARM64] Disable kernel huge I/O mappings.
3845 nohugevmalloc [KNL,X86,PPC,ARM64] Disable kernel huge vmalloc mappings.
3847 nohz= [KNL] Boottime enable/disable dynamic ticks
3848 Valid arguments: on, off
3851 nohz_full= [KNL,BOOT,SMP,ISOL]
3852 The argument is a cpu list, as described above.
3853 In kernels built with CONFIG_NO_HZ_FULL=y, set
3854 the specified list of CPUs whose tick will be stopped
3855 whenever possible. The boot CPU will be forced outside
3856 the range to maintain the timekeeping. Any CPUs
3857 in this list will have their RCU callbacks offloaded,
3858 just as if they had also been called out in the
3859 rcu_nocbs= boot parameter.
3861 Note that this argument takes precedence over
3862 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option.
3864 noinitrd [RAM] Tells the kernel not to load any configured
3867 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
3869 [Deprecated - use intremap=off]
3873 noinvpcid [X86] Disable the INVPCID cpu feature.
3875 noiotrap [SH] Disables trapped I/O port accesses.
3877 noirqdebug [X86-32] Disables the code which attempts to detect and
3878 disable unhandled interrupt sources.
3880 noisapnp [ISAPNP] Disables ISA PnP code.
3882 nojitter [IA-64] Disables jitter checking for ITC timers.
3885 When CONFIG_RANDOMIZE_BASE is set, this disables
3886 kernel and module base offset ASLR (Address Space
3887 Layout Randomization).
3889 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
3892 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
3894 nolapic [X86-32,APIC] Do not enable or use the local APIC.
3896 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
3898 nomca [IA-64] Disable machine check abort handling
3900 nomce [X86-32] Disable Machine Check Exception
3902 nomfgpt [X86-32] Disable Multi-Function General Purpose
3903 Timer usage (for AMD Geode machines).
3905 nomodeset Disable kernel modesetting. Most systems' firmware
3906 sets up a display mode and provides framebuffer memory
3907 for output. With nomodeset, DRM and fbdev drivers will
3908 not load if they could possibly displace the pre-
3909 initialized output. Only the system framebuffer will
3910 be available for use. The respective drivers will not
3911 perform display-mode changes or accelerated rendering.
3913 Useful as error fallback, or for testing and debugging.
3915 nomodule Disable module load
3917 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3918 shutdown the other cpus. Instead use the REBOOT_VECTOR
3921 nopat [X86] Disable PAT (page attribute table extension of
3922 pagetables) support.
3924 nopcid [X86-64] Disable the PCID cpu feature.
3926 nopku [X86] Disable Memory Protection Keys CPU feature found
3930 Equivalent to pti=off
3932 nopv= [X86,XEN,KVM,HYPER_V,VMWARE]
3933 Disables the PV optimizations forcing the guest to run
3934 as generic guest with no PV drivers. Currently support
3935 XEN HVM, KVM, HYPER_V and VMWARE guest.
3937 nopvspin [X86,XEN,KVM]
3938 Disables the qspinlock slow path using PV optimizations
3939 which allow the hypervisor to 'idle' the guest on lock
3942 norandmaps Don't use address space randomization. Equivalent to
3943 echo 0 > /proc/sys/kernel/randomize_va_space
3945 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
3946 with UP alternatives
3948 noresume [SWSUSP] Disables resume and restores original swap
3953 no-scroll [VGA] Disables scrollback.
3954 This is required for the Braillex ib80-piezo Braille
3955 reader made by F.H. Papenmeier (Germany).
3957 nosgx [X86-64,SGX] Disables Intel SGX kernel support.
3960 Disable SMAP (Supervisor Mode Access Prevention)
3961 even if it is supported by processor.
3964 Disable SMEP (Supervisor Mode Execution Prevention)
3965 even if it is supported by processor.
3967 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
3968 and disable the IO APIC. legacy for "maxcpus=0".
3970 nosmt [KNL,MIPS,PPC,S390] Disable symmetric multithreading (SMT).
3971 Equivalent to smt=1.
3973 [KNL,X86,PPC] Disable symmetric multithreading (SMT).
3974 nosmt=force: Force disable SMT, cannot be undone
3975 via the sysfs control file.
3977 nosoftlockup [KNL] Disable the soft-lockup detector.
3979 nospec_store_bypass_disable
3980 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
3982 nospectre_bhb [ARM64] Disable all mitigations for Spectre-BHB (branch
3983 history injection) vulnerability. System may allow data leaks
3986 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1
3987 (bounds check bypass). With this option data leaks are
3988 possible in the system.
3990 nospectre_v2 [X86,PPC_E500,ARM64] Disable all mitigations for
3991 the Spectre variant 2 (indirect branch prediction)
3992 vulnerability. System may allow data leaks with this
3995 no-steal-acc [X86,PV_OPS,ARM64,PPC/PSERIES,RISCV] Disable
3996 paravirtualized steal time accounting. steal time is
3997 computed, but won't influence scheduler behaviour
3999 nosync [HW,M68K] Disables sync negotiation for all devices.
4001 no_timer_check [X86,APIC] Disables the code which tests for
4002 broken timer IRQ sources.
4005 [PPC] Don't flush the L1-D cache after accessing user data.
4007 novmcoredd [KNL,KDUMP]
4008 Disable device dump. Device dump allows drivers to
4009 append dump data to vmcore so you can collect driver
4010 specified debug info. Drivers can append the data
4011 without any limit and this data is stored in memory,
4012 so this may cause significant memory stress. Disabling
4013 device dump can help save memory but the driver debug
4014 data will be no longer available. This parameter
4015 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
4019 [X86,PV_OPS] Disable paravirtualized VMware scheduler
4020 clock and use the default one.
4022 nowatchdog [KNL] Disable both lockup detectors, i.e.
4023 soft-lockup and NMI watchdog (hard-lockup).
4027 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
4029 NOTE: this parameter will be ignored on systems with the
4030 LEGACY_XAPIC_DISABLED bit set in the
4031 IA32_XAPIC_DISABLE_STATUS MSR.
4033 noxsave [BUGS=X86] Disables x86 extended register state save
4034 and restore using xsave. The kernel will fallback to
4035 enabling legacy floating-point and sse state.
4037 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
4038 register states. The kernel will fall back to use
4039 xsave to save the states. By using this parameter,
4040 performance of saving the states is degraded because
4041 xsave doesn't support modified optimization while
4042 xsaveopt supports it on xsaveopt enabled systems.
4044 noxsaves [X86] Disables xsaves and xrstors used in saving and
4045 restoring x86 extended register state in compacted
4046 form of xsave area. The kernel will fall back to use
4047 xsaveopt and xrstor to save and restore the states
4048 in standard form of xsave area. By using this
4049 parameter, xsave area per process might occupy more
4050 memory on xsaves enabled systems.
4052 nps_mtm_hs_ctr= [KNL,ARC]
4053 This parameter sets the maximum duration, in
4054 cycles, each HW thread of the CTOP can run
4055 without interruptions, before HW switches it.
4056 The actual maximum duration is 16 times this
4058 Format: integer between 1 and 255
4061 nptcg= [IA-64] Override max number of concurrent global TLB
4062 purges which is reported from either PAL_VM_SUMMARY or
4065 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
4066 could support. nr_cpus=n : n >= 1 limits the kernel to
4067 support 'n' processors. It could be larger than the
4068 number of already plugged CPU during bootup, later in
4069 runtime you can physically add extra cpu until it reaches
4070 n. So during boot up some boot time memory for per-cpu
4071 variables need be pre-allocated for later physical cpu
4074 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
4076 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86] Disable NUMA, Only
4077 set up a single NUMA node spanning all memory.
4079 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic
4081 Allowed values are enable and disable
4083 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
4084 'node', 'default' can be specified
4085 This can be set from sysctl after boot.
4086 See Documentation/admin-guide/sysctl/vm.rst for details.
4088 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
4089 See Documentation/core-api/debugging-via-ohci1394.rst for more
4092 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
4093 Rather than timing out after 20 ms if an EC
4094 command is not properly ACKed, override the length
4095 of the timeout. We have interrupts disabled while
4096 waiting for the ACK, so if this is set too high
4097 interrupts *may* be lost!
4099 omap_mux= [OMAP] Override bootloader pin multiplexing.
4100 Format: <mux_mode0.mode_name=value>...
4101 For example, to override I2C bus2:
4102 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
4104 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
4106 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
4108 boundary - index of last SLC block on Flex-OneNAND.
4109 The remaining blocks are configured as MLC blocks.
4110 lock - Configure if Flex-OneNAND boundary should be locked.
4111 Once locked, the boundary cannot be changed.
4112 1 indicates lock status, 0 indicates unlock status.
4114 oops=panic Always panic on oopses. Default is to just kill the
4115 process, but there is a small probability of
4116 deadlocking the machine.
4117 This will also cause panics on machine check exceptions.
4118 Useful together with panic=30 to trigger a reboot.
4121 [KNL] Boolean flag to control whether the page allocator
4122 should randomize its free lists. The randomization may
4123 be automatically enabled if the kernel detects it is
4124 running on a platform with a direct-mapped memory-side
4125 cache, and this parameter can be used to
4126 override/disable that behavior. The state of the flag
4127 can be read from sysfs at:
4128 /sys/module/page_alloc/parameters/shuffle.
4130 page_owner= [KNL] Boot-time page_owner enabling option.
4131 Storage of the information about who allocated
4132 each page is disabled in default. With this switch,
4134 on: enable the feature
4136 page_poison= [KNL] Boot-time parameter changing the state of
4137 poisoning on the buddy allocator, available with
4138 CONFIG_PAGE_POISONING=y.
4139 off: turn off poisoning (default)
4140 on: turn on poisoning
4142 page_reporting.page_reporting_order=
4143 [KNL] Minimal page reporting order
4145 Adjust the minimal page reporting order. The page
4146 reporting is disabled when it exceeds MAX_PAGE_ORDER.
4148 panic= [KNL] Kernel behaviour on panic: delay <timeout>
4149 timeout > 0: seconds before rebooting
4150 timeout = 0: wait forever
4151 timeout < 0: reboot immediately
4154 panic_on_taint= Bitmask for conditionally calling panic() in add_taint()
4155 Format: <hex>[,nousertaint]
4156 Hexadecimal bitmask representing the set of TAINT flags
4157 that will cause the kernel to panic when add_taint() is
4158 called with any of the flags in this set.
4159 The optional switch "nousertaint" can be utilized to
4160 prevent userspace forced crashes by writing to sysctl
4161 /proc/sys/kernel/tainted any flagset matching with the
4162 bitmask set on panic_on_taint.
4163 See Documentation/admin-guide/tainted-kernels.rst for
4164 extra details on the taint flags that users can pick
4165 to compose the bitmask to assign to panic_on_taint.
4167 panic_on_warn=1 panic() instead of WARN(). Useful to cause kdump
4170 panic_print= Bitmask for printing system info when panic happens.
4171 User can chose combination of the following bits:
4172 bit 0: print all tasks info
4173 bit 1: print system memory info
4174 bit 2: print timer info
4175 bit 3: print locks info if CONFIG_LOCKDEP is on
4176 bit 4: print ftrace buffer
4177 bit 5: print all printk messages in buffer
4178 bit 6: print all CPUs backtrace (if available in the arch)
4179 *Be aware* that this option may print a _lot_ of lines,
4180 so there are risks of losing older messages in the log.
4181 Use this option carefully, maybe worth to setup a
4182 bigger log buffer with "log_buf_len" along with this.
4184 parkbd.port= [HW] Parallel port number the keyboard adapter is
4185 connected to, default is 0.
4187 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
4188 0 for XT, 1 for AT (default is AT).
4191 parport= [HW,PPT] Specify parallel ports. 0 disables.
4192 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
4193 Use 'auto' to force the driver to use any
4194 IRQ/DMA settings detected (the default is to
4195 ignore detected IRQ/DMA settings because of
4196 possible conflicts). You can specify the base
4197 address, IRQ, and DMA settings; IRQ and DMA
4198 should be numbers, or 'auto' (for using detected
4199 settings on that particular port), or 'nofifo'
4200 (to avoid using a FIFO even if it is detected).
4201 Parallel ports are assigned in the order they
4202 are specified on the command line, starting
4205 parport_init_mode= [HW,PPT]
4206 Configure VIA parallel port to operate in
4207 a specific mode. This is necessary on Pegasos
4208 computer where firmware has no options for setting
4209 up parallel port mode and sets it to spp.
4210 Currently this function knows 686a and 8231 chips.
4211 Format: [spp|ps2|epp|ecp|ecpepp]
4213 pata_legacy.all= [HW,LIBATA]
4215 Set to non-zero to probe primary and secondary ISA
4216 port ranges on PCI systems where no PCI PATA device
4217 has been found at either range. Disabled by default.
4219 pata_legacy.autospeed= [HW,LIBATA]
4221 Set to non-zero if a chip is present that snoops speed
4222 changes. Disabled by default.
4224 pata_legacy.ht6560a= [HW,LIBATA]
4226 Set to 1, 2, or 3 for HT 6560A on the primary channel,
4227 the secondary channel, or both channels respectively.
4228 Disabled by default.
4230 pata_legacy.ht6560b= [HW,LIBATA]
4232 Set to 1, 2, or 3 for HT 6560B on the primary channel,
4233 the secondary channel, or both channels respectively.
4234 Disabled by default.
4236 pata_legacy.iordy_mask= [HW,LIBATA]
4238 IORDY enable mask. Set individual bits to allow IORDY
4239 for the respective channel. Bit 0 is for the first
4240 legacy channel handled by this driver, bit 1 is for
4241 the second channel, and so on. The sequence will often
4242 correspond to the primary legacy channel, the secondary
4243 legacy channel, and so on, but the handling of a PCI
4244 bus and the use of other driver options may interfere
4245 with the sequence. By default IORDY is allowed across
4248 pata_legacy.opti82c46x= [HW,LIBATA]
4250 Set to 1, 2, or 3 for Opti 82c611A on the primary
4251 channel, the secondary channel, or both channels
4252 respectively. Disabled by default.
4254 pata_legacy.opti82c611a= [HW,LIBATA]
4256 Set to 1, 2, or 3 for Opti 82c465MV on the primary
4257 channel, the secondary channel, or both channels
4258 respectively. Disabled by default.
4260 pata_legacy.pio_mask= [HW,LIBATA]
4262 PIO mode mask for autospeed devices. Set individual
4263 bits to allow the use of the respective PIO modes.
4264 Bit 0 is for mode 0, bit 1 is for mode 1, and so on.
4265 All modes allowed by default.
4267 pata_legacy.probe_all= [HW,LIBATA]
4269 Set to non-zero to probe tertiary and further ISA
4270 port ranges on PCI systems. Disabled by default.
4272 pata_legacy.probe_mask= [HW,LIBATA]
4274 Probe mask for legacy ISA PATA ports. Depending on
4275 platform configuration and the use of other driver
4276 options up to 6 legacy ports are supported: 0x1f0,
4277 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing
4278 of individual ports can be disabled by setting the
4279 corresponding bits in the mask to 1. Bit 0 is for
4280 the first port in the list above (0x1f0), and so on.
4281 By default all supported ports are probed.
4283 pata_legacy.qdi= [HW,LIBATA]
4285 Set to non-zero to probe QDI controllers. By default
4286 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise.
4288 pata_legacy.winbond= [HW,LIBATA]
4290 Set to non-zero to probe Winbond controllers. Use
4291 the standard I/O port (0x130) if 1, otherwise the
4292 value given is the I/O port to use (typically 0x1b0).
4293 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE,
4296 pata_platform.pio_mask= [HW,LIBATA]
4298 Supported PIO mode mask. Set individual bits to allow
4299 the use of the respective PIO modes. Bit 0 is for
4300 mode 0, bit 1 is for mode 1, and so on. Mode 0 only
4304 Halt all CPUs after the first oops has been printed for
4305 the specified number of seconds. This is to be used if
4306 your oopses keep scrolling off the screen.
4310 pci=option[,option...] [PCI] various PCI subsystem options.
4312 Some options herein operate on a specific device
4313 or a set of devices (<pci_dev>). These are
4314 specified in one of the following formats:
4316 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
4317 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
4319 Note: the first format specifies a PCI
4320 bus/device/function address which may change
4321 if new hardware is inserted, if motherboard
4322 firmware changes, or due to changes caused
4323 by other kernel parameters. If the
4324 domain is left unspecified, it is
4325 taken to be zero. Optionally, a path
4326 to a device through multiple device/function
4327 addresses can be specified after the base
4328 address (this is more robust against
4329 renumbering issues). The second format
4330 selects devices using IDs from the
4331 configuration space which may match multiple
4332 devices in the system.
4334 earlydump dump PCI config space before the kernel
4336 off [X86] don't probe for the PCI bus
4337 bios [X86-32] force use of PCI BIOS, don't access
4338 the hardware directly. Use this if your machine
4339 has a non-standard PCI host bridge.
4340 nobios [X86-32] disallow use of PCI BIOS, only direct
4341 hardware access methods are allowed. Use this
4342 if you experience crashes upon bootup and you
4343 suspect they are caused by the BIOS.
4344 conf1 [X86] Force use of PCI Configuration Access
4345 Mechanism 1 (config address in IO port 0xCF8,
4346 data in IO port 0xCFC, both 32-bit).
4347 conf2 [X86] Force use of PCI Configuration Access
4348 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
4349 the function, IO port 0xCFA, also 8-bit, sets
4350 bus number. The config space is then accessed
4351 through ports 0xC000-0xCFFF).
4352 See http://wiki.osdev.org/PCI for more info
4353 on the configuration access mechanisms.
4354 noaer [PCIE] If the PCIEAER kernel config parameter is
4355 enabled, this kernel boot option can be used to
4356 disable the use of PCIE advanced error reporting.
4357 nodomains [PCI] Disable support for multiple PCI
4358 root domains (aka PCI segments, in ACPI-speak).
4359 nommconf [X86] Disable use of MMCONFIG for PCI
4361 check_enable_amd_mmconf [X86] check for and enable
4362 properly configured MMIO access to PCI
4363 config space on AMD family 10h CPU
4364 nomsi [MSI] If the PCI_MSI kernel config parameter is
4365 enabled, this kernel boot option can be used to
4366 disable the use of MSI interrupts system-wide.
4367 noioapicquirk [APIC] Disable all boot interrupt quirks.
4368 Safety option to keep boot IRQs enabled. This
4369 should never be necessary.
4370 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
4371 primary IO-APIC for bridges that cannot disable
4372 boot IRQs. This fixes a source of spurious IRQs
4373 when the system masks IRQs.
4374 noioapicreroute [APIC] Disable workaround that uses the
4375 boot IRQ equivalent of an IRQ that connects to
4376 a chipset where boot IRQs cannot be disabled.
4377 The opposite of ioapicreroute.
4378 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
4379 routing table. These calls are known to be buggy
4380 on several machines and they hang the machine
4381 when used, but on other computers it's the only
4382 way to get the interrupt routing table. Try
4383 this option if the kernel is unable to allocate
4384 IRQs or discover secondary PCI buses on your
4386 rom [X86] Assign address space to expansion ROMs.
4387 Use with caution as certain devices share
4388 address decoders between ROMs and other
4390 norom [X86] Do not assign address space to
4391 expansion ROMs that do not already have
4392 BIOS assigned address ranges.
4393 nobar [X86] Do not assign address space to the
4394 BARs that weren't assigned by the BIOS.
4395 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
4396 assigned automatically to PCI devices. You can
4397 make the kernel exclude IRQs of your ISA cards
4399 pirqaddr=0xAAAAA [X86] Specify the physical address
4400 of the PIRQ table (normally generated
4401 by the BIOS) if it is outside the
4402 F0000h-100000h range.
4403 lastbus=N [X86] Scan all buses thru bus #N. Can be
4404 useful if the kernel is unable to find your
4405 secondary buses and you want to tell it
4406 explicitly which ones they are.
4407 assign-busses [X86] Always assign all PCI bus
4408 numbers ourselves, overriding
4409 whatever the firmware may have done.
4410 usepirqmask [X86] Honor the possible IRQ mask stored
4411 in the BIOS $PIR table. This is needed on
4412 some systems with broken BIOSes, notably
4413 some HP Pavilion N5400 and Omnibook XE3
4414 notebooks. This will have no effect if ACPI
4415 IRQ routing is enabled.
4416 noacpi [X86] Do not use ACPI for IRQ routing
4417 or for PCI scanning.
4418 use_crs [X86] Use PCI host bridge window information
4419 from ACPI. On BIOSes from 2008 or later, this
4420 is enabled by default. If you need to use this,
4421 please report a bug.
4422 nocrs [X86] Ignore PCI host bridge windows from ACPI.
4423 If you need to use this, please report a bug.
4424 use_e820 [X86] Use E820 reservations to exclude parts of
4425 PCI host bridge windows. This is a workaround
4426 for BIOS defects in host bridge _CRS methods.
4427 If you need to use this, please report a bug to
4428 <linux-pci@vger.kernel.org>.
4429 no_e820 [X86] Ignore E820 reservations for PCI host
4430 bridge windows. This is the default on modern
4431 hardware. If you need to use this, please report
4432 a bug to <linux-pci@vger.kernel.org>.
4433 routeirq Do IRQ routing for all PCI devices.
4434 This is normally done in pci_enable_device(),
4435 so this option is a temporary workaround
4436 for broken drivers that don't call it.
4437 skip_isa_align [X86] do not align io start addr, so can
4438 handle more pci cards
4439 noearly [X86] Don't do any early type 1 scanning.
4440 This might help on some broken boards which
4441 machine check when some devices' config space
4442 is read. But various workarounds are disabled
4443 and some IOMMU drivers will not work.
4444 bfsort Sort PCI devices into breadth-first order.
4445 This sorting is done to get a device
4446 order compatible with older (<= 2.4) kernels.
4447 nobfsort Don't sort PCI devices into breadth-first order.
4448 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
4449 tuning and use the BIOS-configured MPS defaults.
4450 pcie_bus_safe Set every device's MPS to the largest value
4451 supported by all devices below the root complex.
4452 pcie_bus_perf Set device MPS to the largest allowable MPS
4453 based on its parent bus. Also set MRRS (Max
4454 Read Request Size) to the largest supported
4455 value (no larger than the MPS that the device
4456 or bus can support) for best performance.
4457 pcie_bus_peer2peer Set every device's MPS to 128B, which
4458 every device is guaranteed to support. This
4459 configuration allows peer-to-peer DMA between
4460 any pair of devices, possibly at the cost of
4461 reduced performance. This also guarantees
4462 that hot-added devices will work.
4463 cbiosize=nn[KMG] The fixed amount of bus space which is
4464 reserved for the CardBus bridge's IO window.
4465 The default value is 256 bytes.
4466 cbmemsize=nn[KMG] The fixed amount of bus space which is
4467 reserved for the CardBus bridge's memory
4468 window. The default value is 64 megabytes.
4471 [<order of align>@]<pci_dev>[; ...]
4472 Specifies alignment and device to reassign
4473 aligned memory resources. How to
4474 specify the device is described above.
4475 If <order of align> is not specified,
4476 PAGE_SIZE is used as alignment.
4477 A PCI-PCI bridge can be specified if resource
4478 windows need to be expanded.
4479 To specify the alignment for several
4480 instances of a device, the PCI vendor,
4481 device, subvendor, and subdevice may be
4482 specified, e.g., 12@pci:8086:9c22:103c:198f
4483 for 4096-byte alignment.
4484 ecrc= Enable/disable PCIe ECRC (transaction layer
4485 end-to-end CRC checking). Only effective if
4486 OS has native AER control (either granted by
4487 ACPI _OSC or forced via "pcie_ports=native")
4488 bios: Use BIOS/firmware settings. This is the
4492 hpiosize=nn[KMG] The fixed amount of bus space which is
4493 reserved for hotplug bridge's IO window.
4494 Default size is 256 bytes.
4495 hpmmiosize=nn[KMG] The fixed amount of bus space which is
4496 reserved for hotplug bridge's MMIO window.
4497 Default size is 2 megabytes.
4498 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
4499 reserved for hotplug bridge's MMIO_PREF window.
4500 Default size is 2 megabytes.
4501 hpmemsize=nn[KMG] The fixed amount of bus space which is
4502 reserved for hotplug bridge's MMIO and
4504 Default size is 2 megabytes.
4505 hpbussize=nn The minimum amount of additional bus numbers
4506 reserved for buses below a hotplug bridge.
4508 realloc= Enable/disable reallocating PCI bridge resources
4509 if allocations done by BIOS are too small to
4510 accommodate resources required by all child
4512 off: Turn realloc off
4514 realloc same as realloc=on
4515 noari do not use PCIe ARI.
4516 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
4517 do not use PCIe ATS (and IOMMU device IOTLB).
4518 pcie_scan_all Scan all possible PCIe devices. Otherwise we
4519 only look for one device below a PCIe downstream
4521 big_root_window Try to add a big 64bit memory window to the PCIe
4522 root complex on AMD CPUs. Some GFX hardware
4523 can resize a BAR to allow access to all VRAM.
4524 Adding the window is slightly risky (it may
4525 conflict with unreported devices), so this
4527 disable_acs_redir=<pci_dev>[; ...]
4528 Specify one or more PCI devices (in the format
4529 specified above) separated by semicolons.
4530 Each device specified will have the PCI ACS
4531 redirect capabilities forced off which will
4532 allow P2P traffic between devices through
4533 bridges without forcing it upstream. Note:
4534 this removes isolation between devices and
4535 may put more devices in an IOMMU group.
4536 force_floating [S390] Force usage of floating interrupts.
4537 nomio [S390] Do not use MIO instructions.
4538 norid [S390] ignore the RID field and force use of
4539 one PCI domain per PCI function
4541 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
4544 force Enable ASPM even on devices that claim not to support it.
4545 WARNING: Forcing ASPM on may cause system lockups.
4547 pcie_ports= [PCIE] PCIe port services handling:
4548 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
4549 even if the platform doesn't give the OS permission to
4550 use them. This may cause conflicts if the platform
4551 also tries to use these services.
4552 dpc-native Use native PCIe service for DPC only. May
4553 cause conflicts if firmware uses AER or DPC.
4554 compat Disable native PCIe services (PME, AER, DPC, PCIe
4557 pcie_port_pm= [PCIE] PCIe port power management handling:
4558 off Disable power management of all PCIe ports
4559 force Forcibly enable power management of all PCIe ports
4561 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
4562 nomsi Do not use MSI for native PCIe PME signaling (this makes
4563 all PCIe root ports use INTx for all services).
4565 pcmv= [HW,PCMCIA] BadgePAD 4
4569 Keep all power-domains already enabled by bootloader on,
4570 even if no driver has claimed them. This is useful
4571 for debug and development, but should not be
4572 needed on a platform with proper driver support.
4574 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
4577 See arch/parisc/kernel/pdc_chassis.c
4579 percpu_alloc= Select which percpu first chunk allocator to use.
4580 Currently supported values are "embed" and "page".
4581 Archs may support subset or none of the selections.
4582 See comments in mm/percpu.c for details on each
4583 allocator. This parameter is primarily for debugging
4584 and performance comparison.
4586 pirq= [SMP,APIC] Manual mp-table setup
4587 See Documentation/arch/x86/i386/IO-APIC.rst.
4589 plip= [PPT,NET] Parallel port network link
4590 Format: { parport<nr> | timid | 0 }
4591 See also Documentation/admin-guide/parport.rst.
4593 pmtmr= [X86] Manual setup of pmtmr I/O Port.
4594 Override pmtimer IOPort with a hex value.
4597 pmu_override= [PPC] Override the PMU.
4598 This option takes over the PMU facility, so it is no
4599 longer usable by perf. Setting this option starts the
4600 PMU counters by setting MMCR0 to 0 (the FC bit is
4601 cleared). If a number is given, then MMCR1 is set to
4602 that number, otherwise (e.g., 'pmu_override=on'), MMCR1
4605 pm_debug_messages [SUSPEND,KNL]
4606 Enable suspend/resume debug messages during boot up.
4609 Enable PNP debug messages (depends on the
4610 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
4611 via /sys/module/pnp/parameters/debug. We always show
4612 current resource usage; turning this on also shows
4613 possible settings and some assignment information.
4619 { on | off | curr | res | no-curr | no-res }
4622 [ISAPNP] Exclude IRQs for the autoconfiguration
4625 [ISAPNP] Exclude DMAs for the autoconfiguration
4627 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
4628 Ranges are in pairs (I/O port base and size).
4631 [ISAPNP] Exclude memory regions for the
4633 Ranges are in pairs (memory base and size).
4635 ports= [IP_VS_FTP] IPVS ftp helper module
4637 Up to 8 (IP_VS_APP_MAX_PORTS) ports
4639 Format: <port>,<port>....
4641 powersave=off [PPC] This option disables power saving features.
4642 It specifically disables cpuidle and sets the
4643 platform machine description specific power_save
4644 function to NULL. On Idle the CPU just reduces
4647 ppc_strict_facility_enable
4648 [PPC] This option catches any kernel floating point,
4649 Altivec, VSX and SPE outside of regions specifically
4650 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
4651 There is some performance impact when enabling this.
4655 Disable Hardware Transactional Memory
4658 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC
4659 none - Limited to cond_resched() calls
4660 voluntary - Limited to cond_resched() and might_sleep() calls
4661 full - Any section that isn't explicitly preempt disabled
4662 can be preempted anytime.
4664 print-fatal-signals=
4665 [KNL] debug: print fatal signals
4667 If enabled, warn about various signal handling
4668 related application anomalies: too many signals,
4669 too many POSIX.1 timers, fatal signals causing a
4672 If you hit the warning due to signal overflow,
4673 you might want to try "ulimit -i unlimited".
4677 printk.always_kmsg_dump=
4678 Trigger kmsg_dump for cases other than kernel oops or
4680 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4683 printk.console_no_auto_verbose=
4684 Disable console loglevel raise on oops, panic
4685 or lockdep-detected issues (only if lock debug is on).
4686 With an exception to setups with low baudrate on
4687 serial console, keeping this 0 is a good choice
4688 in order to provide more debug information.
4690 default: 0 (auto_verbose is enabled)
4692 printk.devkmsg={on,off,ratelimit}
4693 Control writing to /dev/kmsg.
4694 on - unlimited logging to /dev/kmsg from userspace
4695 off - logging to /dev/kmsg disabled
4696 ratelimit - ratelimit the logging
4699 printk.time= Show timing data prefixed to each printk message line
4700 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4702 processor.max_cstate= [HW,ACPI]
4703 Limit processor to maximum C-state
4704 max_cstate=9 overrides any DMI blacklist limit.
4706 processor.nocst [HW,ACPI]
4707 Ignore the _CST method to determine C-states,
4708 instead using the legacy FADT method
4710 profile= [KNL] Enable kernel profiling via /proc/profile
4711 Format: [<profiletype>,]<number>
4712 Param: <profiletype>: "schedule", "sleep", or "kvm"
4713 [defaults to kernel profiling]
4714 Param: "schedule" - profile schedule points.
4715 Param: "sleep" - profile D-state sleeping (millisecs).
4716 Requires CONFIG_SCHEDSTATS
4717 Param: "kvm" - profile VM exits.
4718 Param: <number> - step/bucket size as a power of 2 for
4719 statistical time based profiling.
4721 prompt_ramdisk= [RAM] [Deprecated]
4723 prot_virt= [S390] enable hosting protected virtual machines
4724 isolated from the hypervisor (if hardware supports
4728 psi= [KNL] Enable or disable pressure stall information
4732 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
4733 probe for; one of (bare|imps|exps|lifebook|any).
4734 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
4736 psmouse.resetafter= [HW,MOUSE]
4737 Try to reset the device after so many bad packets
4740 [HW,MOUSE] Set desired mouse resolution, in dpi.
4741 psmouse.smartscroll=
4742 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
4743 0 = disabled, 1 = enabled (default).
4745 pstore.backend= Specify the name of the pstore backend to use
4747 pti= [X86-64] Control Page Table Isolation of user and
4748 kernel address spaces. Disabling this feature
4749 removes hardening, but improves performance of
4750 system calls and interrupts.
4752 on - unconditionally enable
4753 off - unconditionally disable
4754 auto - kernel detects whether your CPU model is
4755 vulnerable to issues that PTI mitigates
4757 Not specifying this option is equivalent to pti=auto.
4760 [KNL] Number of legacy pty's. Overwrites compiled-in
4763 quiet [KNL] Disable most log messages
4767 radix_hcall_invalidate=on [PPC/PSERIES]
4768 Disable RADIX GTSE feature and use hcall for TLB
4772 See Documentation/admin-guide/md.rst.
4774 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
4775 See Documentation/admin-guide/blockdev/ramdisk.rst.
4777 ramdisk_start= [RAM] RAM disk image start address
4779 random.trust_cpu=off
4780 [KNL] Disable trusting the use of the CPU's
4781 random number generator (if available) to
4782 initialize the kernel's RNG.
4784 random.trust_bootloader=off
4785 [KNL] Disable trusting the use of the a seed
4786 passed by the bootloader (if available) to
4787 initialize the kernel's RNG.
4789 randomize_kstack_offset=
4790 [KNL] Enable or disable kernel stack offset
4791 randomization, which provides roughly 5 bits of
4792 entropy, frustrating memory corruption attacks
4793 that depend on stack address determinism or
4794 cross-syscall address exposures. This is only
4795 available on architectures that have defined
4796 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET.
4797 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4798 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT.
4800 ras=option[,option,...] [KNL] RAS-specific options
4803 Disable the Correctable Errors Collector,
4804 see CONFIG_RAS_CEC help text.
4806 rcu_nocbs[=cpu-list]
4807 [KNL] The optional argument is a cpu list,
4810 In kernels built with CONFIG_RCU_NOCB_CPU=y,
4811 enable the no-callback CPU mode, which prevents
4812 such CPUs' callbacks from being invoked in
4813 softirq context. Invocation of such CPUs' RCU
4814 callbacks will instead be offloaded to "rcuox/N"
4815 kthreads created for that purpose, where "x" is
4816 "p" for RCU-preempt, "s" for RCU-sched, and "g"
4817 for the kthreads that mediate grace periods; and
4818 "N" is the CPU number. This reduces OS jitter on
4819 the offloaded CPUs, which can be useful for HPC
4820 and real-time workloads. It can also improve
4821 energy efficiency for asymmetric multiprocessors.
4823 If a cpulist is passed as an argument, the specified
4824 list of CPUs is set to no-callback mode from boot.
4826 Otherwise, if the '=' sign and the cpulist
4827 arguments are omitted, no CPU will be set to
4828 no-callback mode from boot but the mode may be
4829 toggled at runtime via cpusets.
4831 Note that this argument takes precedence over
4832 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option.
4835 Rather than requiring that offloaded CPUs
4836 (specified by rcu_nocbs= above) explicitly
4837 awaken the corresponding "rcuoN" kthreads,
4838 make these kthreads poll for callbacks.
4839 This improves the real-time response for the
4840 offloaded CPUs by relieving them of the need to
4841 wake up the corresponding kthread, but degrades
4842 energy efficiency by requiring that the kthreads
4843 periodically wake up to do the polling.
4845 rcutree.blimit= [KNL]
4846 Set maximum number of finished RCU callbacks to
4847 process in one batch.
4849 rcutree.do_rcu_barrier= [KNL]
4850 Request a call to rcu_barrier(). This is
4851 throttled so that userspace tests can safely
4852 hammer on the sysfs variable if they so choose.
4853 If triggered before the RCU grace-period machinery
4854 is fully active, this will error out with EAGAIN.
4856 rcutree.dump_tree= [KNL]
4857 Dump the structure of the rcu_node combining tree
4858 out at early boot. This is used for diagnostic
4859 purposes, to verify correct tree setup.
4861 rcutree.gp_cleanup_delay= [KNL]
4862 Set the number of jiffies to delay each step of
4863 RCU grace-period cleanup.
4865 rcutree.gp_init_delay= [KNL]
4866 Set the number of jiffies to delay each step of
4867 RCU grace-period initialization.
4869 rcutree.gp_preinit_delay= [KNL]
4870 Set the number of jiffies to delay each step of
4871 RCU grace-period pre-initialization, that is,
4872 the propagation of recent CPU-hotplug changes up
4873 the rcu_node combining tree.
4875 rcutree.jiffies_till_first_fqs= [KNL]
4876 Set delay from grace-period initialization to
4877 first attempt to force quiescent states.
4878 Units are jiffies, minimum value is zero,
4879 and maximum value is HZ.
4881 rcutree.jiffies_till_next_fqs= [KNL]
4882 Set delay between subsequent attempts to force
4883 quiescent states. Units are jiffies, minimum
4884 value is one, and maximum value is HZ.
4886 rcutree.jiffies_till_sched_qs= [KNL]
4887 Set required age in jiffies for a
4888 given grace period before RCU starts
4889 soliciting quiescent-state help from
4890 rcu_note_context_switch() and cond_resched().
4891 If not specified, the kernel will calculate
4892 a value based on the most recent settings
4893 of rcutree.jiffies_till_first_fqs
4894 and rcutree.jiffies_till_next_fqs.
4895 This calculated value may be viewed in
4896 rcutree.jiffies_to_sched_qs. Any attempt to set
4897 rcutree.jiffies_to_sched_qs will be cheerfully
4900 rcutree.kthread_prio= [KNL,BOOT]
4901 Set the SCHED_FIFO priority of the RCU per-CPU
4902 kthreads (rcuc/N). This value is also used for
4903 the priority of the RCU boost threads (rcub/N)
4904 and for the RCU grace-period kthreads (rcu_bh,
4905 rcu_preempt, and rcu_sched). If RCU_BOOST is
4906 set, valid values are 1-99 and the default is 1
4907 (the least-favored priority). Otherwise, when
4908 RCU_BOOST is not set, valid values are 0-99 and
4909 the default is zero (non-realtime operation).
4910 When RCU_NOCB_CPU is set, also adjust the
4911 priority of NOCB callback kthreads.
4913 rcutree.nocb_nobypass_lim_per_jiffy= [KNL]
4914 On callback-offloaded (rcu_nocbs) CPUs,
4915 RCU reduces the lock contention that would
4916 otherwise be caused by callback floods through
4917 use of the ->nocb_bypass list. However, in the
4918 common non-flooded case, RCU queues directly to
4919 the main ->cblist in order to avoid the extra
4920 overhead of the ->nocb_bypass list and its lock.
4921 But if there are too many callbacks queued during
4922 a single jiffy, RCU pre-queues the callbacks into
4923 the ->nocb_bypass queue. The definition of "too
4924 many" is supplied by this kernel boot parameter.
4926 rcutree.qhimark= [KNL]
4927 Set threshold of queued RCU callbacks beyond which
4928 batch limiting is disabled.
4930 rcutree.qlowmark= [KNL]
4931 Set threshold of queued RCU callbacks below which
4932 batch limiting is re-enabled.
4934 rcutree.qovld= [KNL]
4935 Set threshold of queued RCU callbacks beyond which
4936 RCU's force-quiescent-state scan will aggressively
4937 enlist help from cond_resched() and sched IPIs to
4938 help CPUs more quickly reach quiescent states.
4939 Set to less than zero to make this be set based
4940 on rcutree.qhimark at boot time and to zero to
4941 disable more aggressive help enlistment.
4943 rcutree.rcu_delay_page_cache_fill_msec= [KNL]
4944 Set the page-cache refill delay (in milliseconds)
4945 in response to low-memory conditions. The range
4946 of permitted values is in the range 0:100000.
4948 rcutree.rcu_divisor= [KNL]
4949 Set the shift-right count to use to compute
4950 the callback-invocation batch limit bl from
4951 the number of callbacks queued on this CPU.
4952 The result will be bounded below by the value of
4953 the rcutree.blimit kernel parameter. Every bl
4954 callbacks, the softirq handler will exit in
4955 order to allow the CPU to do other work.
4957 Please note that this callback-invocation batch
4958 limit applies only to non-offloaded callback
4959 invocation. Offloaded callbacks are instead
4960 invoked in the context of an rcuoc kthread, which
4961 scheduler will preempt as it does any other task.
4963 rcutree.rcu_fanout_exact= [KNL]
4964 Disable autobalancing of the rcu_node combining
4965 tree. This is used by rcutorture, and might
4966 possibly be useful for architectures having high
4967 cache-to-cache transfer latencies.
4969 rcutree.rcu_fanout_leaf= [KNL]
4970 Change the number of CPUs assigned to each
4971 leaf rcu_node structure. Useful for very
4972 large systems, which will choose the value 64,
4973 and for NUMA systems with large remote-access
4974 latencies, which will choose a value aligned
4975 with the appropriate hardware boundaries.
4977 rcutree.rcu_min_cached_objs= [KNL]
4978 Minimum number of objects which are cached and
4979 maintained per one CPU. Object size is equal
4980 to PAGE_SIZE. The cache allows to reduce the
4981 pressure to page allocator, also it makes the
4982 whole algorithm to behave better in low memory
4985 rcutree.rcu_nocb_gp_stride= [KNL]
4986 Set the number of NOCB callback kthreads in
4987 each group, which defaults to the square root
4988 of the number of CPUs. Larger numbers reduce
4989 the wakeup overhead on the global grace-period
4990 kthread, but increases that same overhead on
4991 each group's NOCB grace-period kthread.
4993 rcutree.rcu_kick_kthreads= [KNL]
4994 Cause the grace-period kthread to get an extra
4995 wake_up() if it sleeps three times longer than
4996 it should at force-quiescent-state time.
4997 This wake_up() will be accompanied by a
4998 WARN_ONCE() splat and an ftrace_dump().
5000 rcutree.rcu_resched_ns= [KNL]
5001 Limit the time spend invoking a batch of RCU
5002 callbacks to the specified number of nanoseconds.
5003 By default, this limit is checked only once
5004 every 32 callbacks in order to limit the pain
5005 inflicted by local_clock() overhead.
5007 rcutree.rcu_unlock_delay= [KNL]
5008 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels,
5009 this specifies an rcu_read_unlock()-time delay
5010 in microseconds. This defaults to zero.
5011 Larger delays increase the probability of
5012 catching RCU pointer leaks, that is, buggy use
5013 of RCU-protected pointers after the relevant
5014 rcu_read_unlock() has completed.
5016 rcutree.sysrq_rcu= [KNL]
5017 Commandeer a sysrq key to dump out Tree RCU's
5018 rcu_node tree with an eye towards determining
5019 why a new grace period has not yet started.
5021 rcutree.use_softirq= [KNL]
5022 If set to zero, move all RCU_SOFTIRQ processing to
5023 per-CPU rcuc kthreads. Defaults to a non-zero
5024 value, meaning that RCU_SOFTIRQ is used by default.
5025 Specify rcutree.use_softirq=0 to use rcuc kthreads.
5027 But note that CONFIG_PREEMPT_RT=y kernels disable
5028 this kernel boot parameter, forcibly setting it
5031 rcuscale.gp_async= [KNL]
5032 Measure performance of asynchronous
5033 grace-period primitives such as call_rcu().
5035 rcuscale.gp_async_max= [KNL]
5036 Specify the maximum number of outstanding
5037 callbacks per writer thread. When a writer
5038 thread exceeds this limit, it invokes the
5039 corresponding flavor of rcu_barrier() to allow
5040 previously posted callbacks to drain.
5042 rcuscale.gp_exp= [KNL]
5043 Measure performance of expedited synchronous
5044 grace-period primitives.
5046 rcuscale.holdoff= [KNL]
5047 Set test-start holdoff period. The purpose of
5048 this parameter is to delay the start of the
5049 test until boot completes in order to avoid
5052 rcuscale.kfree_by_call_rcu= [KNL]
5053 In kernels built with CONFIG_RCU_LAZY=y, test
5054 call_rcu() instead of kfree_rcu().
5056 rcuscale.kfree_mult= [KNL]
5057 Instead of allocating an object of size kfree_obj,
5058 allocate one of kfree_mult * sizeof(kfree_obj).
5061 rcuscale.kfree_rcu_test= [KNL]
5062 Set to measure performance of kfree_rcu() flooding.
5064 rcuscale.kfree_rcu_test_double= [KNL]
5065 Test the double-argument variant of kfree_rcu().
5066 If this parameter has the same value as
5067 rcuscale.kfree_rcu_test_single, both the single-
5068 and double-argument variants are tested.
5070 rcuscale.kfree_rcu_test_single= [KNL]
5071 Test the single-argument variant of kfree_rcu().
5072 If this parameter has the same value as
5073 rcuscale.kfree_rcu_test_double, both the single-
5074 and double-argument variants are tested.
5076 rcuscale.kfree_nthreads= [KNL]
5077 The number of threads running loops of kfree_rcu().
5079 rcuscale.kfree_alloc_num= [KNL]
5080 Number of allocations and frees done in an iteration.
5082 rcuscale.kfree_loops= [KNL]
5083 Number of loops doing rcuscale.kfree_alloc_num number
5084 of allocations and frees.
5086 rcuscale.minruntime= [KNL]
5087 Set the minimum test run time in seconds. This
5088 does not affect the data-collection interval,
5089 but instead allows better measurement of things
5090 like CPU consumption.
5092 rcuscale.nreaders= [KNL]
5093 Set number of RCU readers. The value -1 selects
5094 N, where N is the number of CPUs. A value
5095 "n" less than -1 selects N-n+1, where N is again
5096 the number of CPUs. For example, -2 selects N
5097 (the number of CPUs), -3 selects N+1, and so on.
5098 A value of "n" less than or equal to -N selects
5101 rcuscale.nwriters= [KNL]
5102 Set number of RCU writers. The values operate
5103 the same as for rcuscale.nreaders.
5104 N, where N is the number of CPUs
5106 rcuscale.scale_type= [KNL]
5107 Specify the RCU implementation to test.
5109 rcuscale.shutdown= [KNL]
5110 Shut the system down after performance tests
5111 complete. This is useful for hands-off automated
5114 rcuscale.verbose= [KNL]
5115 Enable additional printk() statements.
5117 rcuscale.writer_holdoff= [KNL]
5118 Write-side holdoff between grace periods,
5119 in microseconds. The default of zero says
5122 rcuscale.writer_holdoff_jiffies= [KNL]
5123 Additional write-side holdoff between grace
5124 periods, but in jiffies. The default of zero
5127 rcutorture.fqs_duration= [KNL]
5128 Set duration of force_quiescent_state bursts
5131 rcutorture.fqs_holdoff= [KNL]
5132 Set holdoff time within force_quiescent_state bursts
5135 rcutorture.fqs_stutter= [KNL]
5136 Set wait time between force_quiescent_state bursts
5139 rcutorture.fwd_progress= [KNL]
5140 Specifies the number of kthreads to be used
5141 for RCU grace-period forward-progress testing
5142 for the types of RCU supporting this notion.
5143 Defaults to 1 kthread, values less than zero or
5144 greater than the number of CPUs cause the number
5147 rcutorture.fwd_progress_div= [KNL]
5148 Specify the fraction of a CPU-stall-warning
5149 period to do tight-loop forward-progress testing.
5151 rcutorture.fwd_progress_holdoff= [KNL]
5152 Number of seconds to wait between successive
5153 forward-progress tests.
5155 rcutorture.fwd_progress_need_resched= [KNL]
5156 Enclose cond_resched() calls within checks for
5157 need_resched() during tight-loop forward-progress
5160 rcutorture.gp_cond= [KNL]
5161 Use conditional/asynchronous update-side
5162 primitives, if available.
5164 rcutorture.gp_exp= [KNL]
5165 Use expedited update-side primitives, if available.
5167 rcutorture.gp_normal= [KNL]
5168 Use normal (non-expedited) asynchronous
5169 update-side primitives, if available.
5171 rcutorture.gp_sync= [KNL]
5172 Use normal (non-expedited) synchronous
5173 update-side primitives, if available. If all
5174 of rcutorture.gp_cond=, rcutorture.gp_exp=,
5175 rcutorture.gp_normal=, and rcutorture.gp_sync=
5176 are zero, rcutorture acts as if is interpreted
5177 they are all non-zero.
5179 rcutorture.irqreader= [KNL]
5180 Run RCU readers from irq handlers, or, more
5181 accurately, from a timer handler. Not all RCU
5182 flavors take kindly to this sort of thing.
5184 rcutorture.leakpointer= [KNL]
5185 Leak an RCU-protected pointer out of the reader.
5186 This can of course result in splats, and is
5187 intended to test the ability of things like
5188 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect
5191 rcutorture.n_barrier_cbs= [KNL]
5192 Set callbacks/threads for rcu_barrier() testing.
5194 rcutorture.nfakewriters= [KNL]
5195 Set number of concurrent RCU writers. These just
5196 stress RCU, they don't participate in the actual
5197 test, hence the "fake".
5199 rcutorture.nocbs_nthreads= [KNL]
5200 Set number of RCU callback-offload togglers.
5201 Zero (the default) disables toggling.
5203 rcutorture.nocbs_toggle= [KNL]
5204 Set the delay in milliseconds between successive
5205 callback-offload toggling attempts.
5207 rcutorture.nreaders= [KNL]
5208 Set number of RCU readers. The value -1 selects
5209 N-1, where N is the number of CPUs. A value
5210 "n" less than -1 selects N-n-2, where N is again
5211 the number of CPUs. For example, -2 selects N
5212 (the number of CPUs), -3 selects N+1, and so on.
5214 rcutorture.object_debug= [KNL]
5215 Enable debug-object double-call_rcu() testing.
5217 rcutorture.onoff_holdoff= [KNL]
5218 Set time (s) after boot for CPU-hotplug testing.
5220 rcutorture.onoff_interval= [KNL]
5221 Set time (jiffies) between CPU-hotplug operations,
5222 or zero to disable CPU-hotplug testing.
5224 rcutorture.read_exit= [KNL]
5225 Set the number of read-then-exit kthreads used
5226 to test the interaction of RCU updaters and
5227 task-exit processing.
5229 rcutorture.read_exit_burst= [KNL]
5230 The number of times in a given read-then-exit
5231 episode that a set of read-then-exit kthreads
5234 rcutorture.read_exit_delay= [KNL]
5235 The delay, in seconds, between successive
5236 read-then-exit testing episodes.
5238 rcutorture.shuffle_interval= [KNL]
5239 Set task-shuffle interval (s). Shuffling tasks
5240 allows some CPUs to go into dyntick-idle mode
5241 during the rcutorture test.
5243 rcutorture.shutdown_secs= [KNL]
5244 Set time (s) after boot system shutdown. This
5245 is useful for hands-off automated testing.
5247 rcutorture.stall_cpu= [KNL]
5248 Duration of CPU stall (s) to test RCU CPU stall
5249 warnings, zero to disable.
5251 rcutorture.stall_cpu_block= [KNL]
5252 Sleep while stalling if set. This will result
5253 in warnings from preemptible RCU in addition to
5254 any other stall-related activity. Note that
5255 in kernels built with CONFIG_PREEMPTION=n and
5256 CONFIG_PREEMPT_COUNT=y, this parameter will
5257 cause the CPU to pass through a quiescent state.
5258 Given CONFIG_PREEMPTION=n, this will suppress
5259 RCU CPU stall warnings, but will instead result
5260 in scheduling-while-atomic splats.
5262 Use of this module parameter results in splats.
5265 rcutorture.stall_cpu_holdoff= [KNL]
5266 Time to wait (s) after boot before inducing stall.
5268 rcutorture.stall_cpu_irqsoff= [KNL]
5269 Disable interrupts while stalling if set.
5271 rcutorture.stall_gp_kthread= [KNL]
5272 Duration (s) of forced sleep within RCU
5273 grace-period kthread to test RCU CPU stall
5274 warnings, zero to disable. If both stall_cpu
5275 and stall_gp_kthread are specified, the
5276 kthread is starved first, then the CPU.
5278 rcutorture.stat_interval= [KNL]
5279 Time (s) between statistics printk()s.
5281 rcutorture.stutter= [KNL]
5282 Time (s) to stutter testing, for example, specifying
5283 five seconds causes the test to run for five seconds,
5284 wait for five seconds, and so on. This tests RCU's
5285 ability to transition abruptly to and from idle.
5287 rcutorture.test_boost= [KNL]
5288 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
5289 "Maybe" means test if the RCU implementation
5290 under test support RCU priority boosting.
5292 rcutorture.test_boost_duration= [KNL]
5293 Duration (s) of each individual boost test.
5295 rcutorture.test_boost_interval= [KNL]
5296 Interval (s) between each boost test.
5298 rcutorture.test_no_idle_hz= [KNL]
5299 Test RCU's dyntick-idle handling. See also the
5300 rcutorture.shuffle_interval parameter.
5302 rcutorture.torture_type= [KNL]
5303 Specify the RCU implementation to test.
5305 rcutorture.verbose= [KNL]
5306 Enable additional printk() statements.
5308 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
5309 Dump ftrace buffer after reporting RCU CPU
5312 rcupdate.rcu_cpu_stall_notifiers= [KNL]
5313 Provide RCU CPU stall notifiers, but see the
5314 warnings in the RCU_CPU_STALL_NOTIFIER Kconfig
5315 option's help text. TL;DR: You almost certainly
5316 do not want rcupdate.rcu_cpu_stall_notifiers.
5318 rcupdate.rcu_cpu_stall_suppress= [KNL]
5319 Suppress RCU CPU stall warning messages.
5321 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL]
5322 Suppress RCU CPU stall warning messages and
5323 rcutorture writer stall warnings that occur
5324 during early boot, that is, during the time
5325 before the init task is spawned.
5327 rcupdate.rcu_cpu_stall_timeout= [KNL]
5328 Set timeout for RCU CPU stall warning messages.
5329 The value is in seconds and the maximum allowed
5330 value is 300 seconds.
5332 rcupdate.rcu_exp_cpu_stall_timeout= [KNL]
5333 Set timeout for expedited RCU CPU stall warning
5334 messages. The value is in milliseconds
5335 and the maximum allowed value is 21000
5336 milliseconds. Please note that this value is
5337 adjusted to an arch timer tick resolution.
5338 Setting this to zero causes the value from
5339 rcupdate.rcu_cpu_stall_timeout to be used (after
5340 conversion from seconds to milliseconds).
5342 rcupdate.rcu_cpu_stall_cputime= [KNL]
5343 Provide statistics on the cputime and count of
5344 interrupts and tasks during the sampling period. For
5345 multiple continuous RCU stalls, all sampling periods
5346 begin at half of the first RCU stall timeout.
5348 rcupdate.rcu_exp_stall_task_details= [KNL]
5349 Print stack dumps of any tasks blocking the
5350 current expedited RCU grace period during an
5351 expedited RCU CPU stall warning.
5353 rcupdate.rcu_expedited= [KNL]
5354 Use expedited grace-period primitives, for
5355 example, synchronize_rcu_expedited() instead
5356 of synchronize_rcu(). This reduces latency,
5357 but can increase CPU utilization, degrade
5358 real-time latency, and degrade energy efficiency.
5359 No effect on CONFIG_TINY_RCU kernels.
5361 rcupdate.rcu_normal= [KNL]
5362 Use only normal grace-period primitives,
5363 for example, synchronize_rcu() instead of
5364 synchronize_rcu_expedited(). This improves
5365 real-time latency, CPU utilization, and
5366 energy efficiency, but can expose users to
5367 increased grace-period latency. This parameter
5368 overrides rcupdate.rcu_expedited. No effect on
5369 CONFIG_TINY_RCU kernels.
5371 rcupdate.rcu_normal_after_boot= [KNL]
5372 Once boot has completed (that is, after
5373 rcu_end_inkernel_boot() has been invoked), use
5374 only normal grace-period primitives. No effect
5375 on CONFIG_TINY_RCU kernels.
5377 But note that CONFIG_PREEMPT_RT=y kernels enables
5378 this kernel boot parameter, forcibly setting
5379 it to the value one, that is, converting any
5380 post-boot attempt at an expedited RCU grace
5381 period to instead use normal non-expedited
5382 grace-period processing.
5384 rcupdate.rcu_task_collapse_lim= [KNL]
5385 Set the maximum number of callbacks present
5386 at the beginning of a grace period that allows
5387 the RCU Tasks flavors to collapse back to using
5388 a single callback queue. This switching only
5389 occurs when rcupdate.rcu_task_enqueue_lim is
5390 set to the default value of -1.
5392 rcupdate.rcu_task_contend_lim= [KNL]
5393 Set the minimum number of callback-queuing-time
5394 lock-contention events per jiffy required to
5395 cause the RCU Tasks flavors to switch to per-CPU
5396 callback queuing. This switching only occurs
5397 when rcupdate.rcu_task_enqueue_lim is set to
5398 the default value of -1.
5400 rcupdate.rcu_task_enqueue_lim= [KNL]
5401 Set the number of callback queues to use for the
5402 RCU Tasks family of RCU flavors. The default
5403 of -1 allows this to be automatically (and
5404 dynamically) adjusted. This parameter is intended
5407 rcupdate.rcu_task_ipi_delay= [KNL]
5408 Set time in jiffies during which RCU tasks will
5409 avoid sending IPIs, starting with the beginning
5410 of a given grace period. Setting a large
5411 number avoids disturbing real-time workloads,
5412 but lengthens grace periods.
5414 rcupdate.rcu_task_lazy_lim= [KNL]
5415 Number of callbacks on a given CPU that will
5416 cancel laziness on that CPU. Use -1 to disable
5417 cancellation of laziness, but be advised that
5418 doing so increases the danger of OOM due to
5421 rcupdate.rcu_task_stall_info= [KNL]
5422 Set initial timeout in jiffies for RCU task stall
5423 informational messages, which give some indication
5424 of the problem for those not patient enough to
5425 wait for ten minutes. Informational messages are
5426 only printed prior to the stall-warning message
5427 for a given grace period. Disable with a value
5428 less than or equal to zero. Defaults to ten
5429 seconds. A change in value does not take effect
5430 until the beginning of the next grace period.
5432 rcupdate.rcu_task_stall_info_mult= [KNL]
5433 Multiplier for time interval between successive
5434 RCU task stall informational messages for a given
5435 RCU tasks grace period. This value is clamped
5436 to one through ten, inclusive. It defaults to
5437 the value three, so that the first informational
5438 message is printed 10 seconds into the grace
5439 period, the second at 40 seconds, the third at
5440 160 seconds, and then the stall warning at 600
5441 seconds would prevent a fourth at 640 seconds.
5443 rcupdate.rcu_task_stall_timeout= [KNL]
5444 Set timeout in jiffies for RCU task stall
5445 warning messages. Disable with a value less
5446 than or equal to zero. Defaults to ten minutes.
5447 A change in value does not take effect until
5448 the beginning of the next grace period.
5450 rcupdate.rcu_tasks_lazy_ms= [KNL]
5451 Set timeout in milliseconds RCU Tasks asynchronous
5452 callback batching for call_rcu_tasks().
5453 A negative value will take the default. A value
5454 of zero will disable batching. Batching is
5455 always disabled for synchronize_rcu_tasks().
5457 rcupdate.rcu_tasks_rude_lazy_ms= [KNL]
5458 Set timeout in milliseconds RCU Tasks
5459 Rude asynchronous callback batching for
5460 call_rcu_tasks_rude(). A negative value
5461 will take the default. A value of zero will
5462 disable batching. Batching is always disabled
5463 for synchronize_rcu_tasks_rude().
5465 rcupdate.rcu_tasks_trace_lazy_ms= [KNL]
5466 Set timeout in milliseconds RCU Tasks
5467 Trace asynchronous callback batching for
5468 call_rcu_tasks_trace(). A negative value
5469 will take the default. A value of zero will
5470 disable batching. Batching is always disabled
5471 for synchronize_rcu_tasks_trace().
5473 rcupdate.rcu_self_test= [KNL]
5474 Run the RCU early boot self tests
5478 Run specified binary instead of /init from the ramdisk,
5479 used for early userspace startup. See initrd.
5482 force - Override the decision by the kernel to hide the
5483 advertisement of RDRAND support (this affects
5484 certain AMD processors because of buggy BIOS
5485 support, specifically around the suspend/resume
5489 Turn on/off individual RDT features. List is:
5490 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
5492 E.g. to turn on cmt and turn off mba use:
5496 Format (x86 or x86_64):
5497 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] | d[efault] \
5499 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
5501 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
5502 (prefix with 'panic_' to set mode for panic
5504 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
5505 reboot_force is either force or not specified,
5506 reboot_cpu is s[mp]#### with #### being the processor
5507 to be used for rebooting.
5509 refscale.holdoff= [KNL]
5510 Set test-start holdoff period. The purpose of
5511 this parameter is to delay the start of the
5512 test until boot completes in order to avoid
5515 refscale.lookup_instances= [KNL]
5516 Number of data elements to use for the forms of
5517 SLAB_TYPESAFE_BY_RCU testing. A negative number
5518 is negated and multiplied by nr_cpu_ids, while
5519 zero specifies nr_cpu_ids.
5521 refscale.loops= [KNL]
5522 Set the number of loops over the synchronization
5523 primitive under test. Increasing this number
5524 reduces noise due to loop start/end overhead,
5525 but the default has already reduced the per-pass
5526 noise to a handful of picoseconds on ca. 2020
5529 refscale.nreaders= [KNL]
5530 Set number of readers. The default value of -1
5531 selects N, where N is roughly 75% of the number
5532 of CPUs. A value of zero is an interesting choice.
5534 refscale.nruns= [KNL]
5535 Set number of runs, each of which is dumped onto
5538 refscale.readdelay= [KNL]
5539 Set the read-side critical-section duration,
5540 measured in microseconds.
5542 refscale.scale_type= [KNL]
5543 Specify the read-protection implementation to test.
5545 refscale.shutdown= [KNL]
5546 Shut down the system at the end of the performance
5547 test. This defaults to 1 (shut it down) when
5548 refscale is built into the kernel and to 0 (leave
5549 it running) when refscale is built as a module.
5551 refscale.verbose= [KNL]
5552 Enable additional printk() statements.
5554 refscale.verbose_batched= [KNL]
5555 Batch the additional printk() statements. If zero
5556 (the default) or negative, print everything. Otherwise,
5557 print every Nth verbose statement, where N is the value
5560 regulator_ignore_unused
5562 Prevents regulator framework from disabling regulators
5563 that are unused, due no driver claiming them. This may
5564 be useful for debug and development, but should not be
5565 needed on a platform with proper driver support.
5568 [KNL, SMP] Set scheduler's default relax_domain_level.
5569 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
5571 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
5572 Format: <base1>,<size1>[,<base2>,<size2>,...]
5573 Reserve I/O ports or memory so the kernel won't use
5574 them. If <base> is less than 0x10000, the region
5575 is assumed to be I/O ports; otherwise it is memory.
5577 reservetop= [X86-32]
5579 Reserves a hole at the top of the kernel virtual
5582 reset_devices [KNL] Force drivers to reset the underlying device
5583 during initialization.
5586 Specify the partition device for software suspend
5588 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
5590 resume_offset= [SWSUSP]
5591 Specify the offset from the beginning of the partition
5592 given by "resume=" at which the swap header is located,
5593 in <PAGE_SIZE> units (needed only for swap files).
5594 See Documentation/power/swsusp-and-swap-files.rst
5596 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
5597 read the resume files
5599 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
5600 Useful for devices that are detected asynchronously
5601 (e.g. USB and MMC devices).
5603 retain_initrd [RAM] Keep initrd memory after extraction. After boot, it will
5604 be accessible via /sys/firmware/initrd.
5606 retbleed= [X86] Control mitigation of RETBleed (Arbitrary
5607 Speculative Code Execution with Return Instructions)
5610 AMD-based UNRET and IBPB mitigations alone do not stop
5611 sibling threads from influencing the predictions of other
5612 sibling threads. For that reason, STIBP is used on pro-
5613 cessors that support it, and mitigate SMT on processors
5617 auto - automatically select a migitation
5618 auto,nosmt - automatically select a mitigation,
5619 disabling SMT if necessary for
5620 the full mitigation (only on Zen1
5621 and older without STIBP).
5622 ibpb - On AMD, mitigate short speculation
5623 windows on basic block boundaries too.
5624 Safe, highest perf impact. It also
5625 enables STIBP if present. Not suitable
5627 ibpb,nosmt - Like "ibpb" above but will disable SMT
5628 when STIBP is not available. This is
5629 the alternative for systems which do not
5631 unret - Force enable untrained return thunks,
5632 only effective on AMD f15h-f17h based
5634 unret,nosmt - Like unret, but will disable SMT when STIBP
5635 is not available. This is the alternative for
5636 systems which do not have STIBP.
5638 Selecting 'auto' will choose a mitigation method at run
5639 time according to the CPU.
5641 Not specifying this option is equivalent to retbleed=auto.
5643 rfkill.default_state=
5644 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
5645 etc. communication is blocked by default.
5648 rfkill.master_switch_mode=
5649 0 The "airplane mode" button does nothing.
5650 1 The "airplane mode" button toggles between everything
5651 blocked and the previous configuration.
5652 2 The "airplane mode" button toggles between everything
5653 blocked and everything unblocked.
5655 rhash_entries= [KNL,NET]
5656 Set number of hash buckets for route cache
5659 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
5662 riscv_isa_fallback [RISCV]
5663 When CONFIG_RISCV_ISA_FALLBACK is not enabled, permit
5664 falling back to detecting extension support by parsing
5665 "riscv,isa" property on devicetree systems when the
5666 replacement properties are not found. See the Kconfig
5667 entry for RISCV_ISA_FALLBACK.
5669 ro [KNL] Mount root device read-only on boot
5672 on Mark read-only kernel memory as read-only (default).
5673 off Leave read-only kernel memory writable for debugging.
5674 full Mark read-only kernel memory and aliases as read-only
5678 Enable the uart passthrough on the designated usb port
5679 on Rockchip SoCs. When active, the signals of the
5680 debug-uart get routed to the D+ and D- pins of the usb
5681 port and the regular usb controller gets disabled.
5683 root= [KNL] Root filesystem
5684 Usually this a a block device specifier of some kind,
5685 see the early_lookup_bdev comment in
5686 block/early-lookup.c for details.
5687 Alternatively this can be "ram" for the legacy initial
5688 ramdisk, "nfs" and "cifs" for root on a network file
5689 system, or "mtd" and "ubi" for mounting from raw flash.
5691 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
5692 mount the root filesystem
5694 rootflags= [KNL] Set root filesystem mount option string
5696 rootfstype= [KNL] Set root filesystem type
5698 rootwait [KNL] Wait (indefinitely) for root device to show up.
5699 Useful for devices that are detected asynchronously
5700 (e.g. USB and MMC devices).
5702 rootwait= [KNL] Maximum time (in seconds) to wait for root device
5703 to show up before attempting to mount the root
5706 rproc_mem=nn[KMG][@address]
5707 [KNL,ARM,CMA] Remoteproc physical memory block.
5708 Memory area to be used by remote processor image,
5711 rw [KNL] Mount root device read-write on boot
5713 S [KNL] Run init in single mode
5715 s390_iommu= [HW,S390]
5716 Set s390 IOTLB flushing mode
5718 With strict flushing every unmap operation will result
5719 in an IOTLB flush. Default is lazy flushing before
5720 reuse, which is faster. Deprecated, equivalent to
5723 s390_iommu_aperture= [KNL,S390]
5724 Specifies the size of the per device DMA address space
5725 accessible through the DMA and IOMMU APIs as a decimal
5726 factor of the size of main memory.
5727 The default is 1 meaning that one can concurrently use
5728 as many DMA addresses as physical memory is installed,
5729 if supported by hardware, and thus map all of memory
5730 once. With a value of 2 one can map all of memory twice
5731 and so on. As a special case a factor of 0 imposes no
5732 restrictions other than those given by hardware at the
5733 cost of significant additional memory use for tables.
5736 See drivers/net/irda/sa1100_ir.c.
5738 sched_verbose [KNL] Enables verbose scheduler debug messages.
5740 schedstats= [KNL,X86] Enable or disable scheduled statistics.
5741 Allowed values are enable and disable. This feature
5742 incurs a small amount of overhead in the scheduler
5743 but is useful for debugging and performance tuning.
5745 sched_thermal_decay_shift=
5746 [KNL, SMP] Set a decay shift for scheduler thermal
5747 pressure signal. Thermal pressure signal follows the
5748 default decay period of other scheduler pelt
5749 signals(usually 32 ms but configurable). Setting
5750 sched_thermal_decay_shift will left shift the decay
5751 period for the thermal pressure signal by the shift
5753 i.e. with the default pelt decay period of 32 ms
5754 sched_thermal_decay_shift thermal pressure decay pr
5758 Format: integer between 0 and 10
5761 scftorture.holdoff= [KNL]
5762 Number of seconds to hold off before starting
5763 test. Defaults to zero for module insertion and
5764 to 10 seconds for built-in smp_call_function()
5767 scftorture.longwait= [KNL]
5768 Request ridiculously long waits randomly selected
5769 up to the chosen limit in seconds. Zero (the
5770 default) disables this feature. Please note
5771 that requesting even small non-zero numbers of
5772 seconds can result in RCU CPU stall warnings,
5773 softlockup complaints, and so on.
5775 scftorture.nthreads= [KNL]
5776 Number of kthreads to spawn to invoke the
5777 smp_call_function() family of functions.
5778 The default of -1 specifies a number of kthreads
5779 equal to the number of CPUs.
5781 scftorture.onoff_holdoff= [KNL]
5782 Number seconds to wait after the start of the
5783 test before initiating CPU-hotplug operations.
5785 scftorture.onoff_interval= [KNL]
5786 Number seconds to wait between successive
5787 CPU-hotplug operations. Specifying zero (which
5788 is the default) disables CPU-hotplug operations.
5790 scftorture.shutdown_secs= [KNL]
5791 The number of seconds following the start of the
5792 test after which to shut down the system. The
5793 default of zero avoids shutting down the system.
5794 Non-zero values are useful for automated tests.
5796 scftorture.stat_interval= [KNL]
5797 The number of seconds between outputting the
5798 current test statistics to the console. A value
5799 of zero disables statistics output.
5801 scftorture.stutter_cpus= [KNL]
5802 The number of jiffies to wait between each change
5803 to the set of CPUs under test.
5805 scftorture.use_cpus_read_lock= [KNL]
5806 Use use_cpus_read_lock() instead of the default
5807 preempt_disable() to disable CPU hotplug
5808 while invoking one of the smp_call_function*()
5811 scftorture.verbose= [KNL]
5812 Enable additional printk() statements.
5814 scftorture.weight_single= [KNL]
5815 The probability weighting to use for the
5816 smp_call_function_single() function with a zero
5817 "wait" parameter. A value of -1 selects the
5818 default if all other weights are -1. However,
5819 if at least one weight has some other value, a
5820 value of -1 will instead select a weight of zero.
5822 scftorture.weight_single_wait= [KNL]
5823 The probability weighting to use for the
5824 smp_call_function_single() function with a
5825 non-zero "wait" parameter. See weight_single.
5827 scftorture.weight_many= [KNL]
5828 The probability weighting to use for the
5829 smp_call_function_many() function with a zero
5830 "wait" parameter. See weight_single.
5831 Note well that setting a high probability for
5832 this weighting can place serious IPI load
5835 scftorture.weight_many_wait= [KNL]
5836 The probability weighting to use for the
5837 smp_call_function_many() function with a
5838 non-zero "wait" parameter. See weight_single
5841 scftorture.weight_all= [KNL]
5842 The probability weighting to use for the
5843 smp_call_function_all() function with a zero
5844 "wait" parameter. See weight_single and
5847 scftorture.weight_all_wait= [KNL]
5848 The probability weighting to use for the
5849 smp_call_function_all() function with a
5850 non-zero "wait" parameter. See weight_single
5853 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
5854 xtime_lock contention on larger systems, and/or RCU lock
5855 contention on all systems with CONFIG_MAXSMP set.
5856 Format: { "0" | "1" }
5857 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
5859 Note: increases power consumption, thus should only be
5860 enabled if running jitter sensitive (HPC/RT) workloads.
5862 security= [SECURITY] Choose a legacy "major" security module to
5863 enable at boot. This has been deprecated by the
5866 selinux= [SELINUX] Disable or enable SELinux at boot time.
5867 Format: { "0" | "1" }
5868 See security/selinux/Kconfig help text.
5873 serialnumber [BUGS=X86-32]
5875 sev=option[,option...] [X86-64] See Documentation/arch/x86/x86_64/boot-options.rst
5878 Maximal number of shapers.
5880 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
5881 Limit apic dumping. The parameter defines the maximal
5882 number of local apics being dumped. Also it is possible
5883 to set it to "all" by meaning -- no limit here.
5884 Format: { 1 (default) | 2 | ... | all }.
5885 The parameter valid if only apic=debug or
5886 apic=verbose is specified.
5887 Example: apic=debug show_lapic=all
5895 Enable merging of slabs with similar size when the
5896 kernel is built without CONFIG_SLAB_MERGE_DEFAULT.
5899 Disable merging of slabs with similar size. May be
5900 necessary if there is some reason to distinguish
5901 allocs to different slabs, especially in hardened
5902 environments where the risk of heap overflows and
5903 layout control by attackers can usually be
5904 frustrated by disabling merging. This will reduce
5905 most of the exposure of a heap attack to a single
5906 cache (risks via metadata attacks are mostly
5907 unchanged). Debug options disable merging on their
5909 For more information see Documentation/mm/slub.rst.
5911 slab_max_order= [MM, SLAB]
5912 Determines the maximum allowed order for slabs.
5913 A high setting may cause OOMs due to memory
5914 fragmentation. Defaults to 1 for systems with
5915 more than 32MB of RAM, 0 otherwise.
5917 slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB]
5918 Enabling slub_debug allows one to determine the
5919 culprit if slab objects become corrupted. Enabling
5920 slub_debug can create guard zones around objects and
5921 may poison objects when not in use. Also tracks the
5922 last alloc / free. For more information see
5923 Documentation/mm/slub.rst.
5925 slub_max_order= [MM, SLUB]
5926 Determines the maximum allowed order for slabs.
5927 A high setting may cause OOMs due to memory
5928 fragmentation. For more information see
5929 Documentation/mm/slub.rst.
5931 slub_min_objects= [MM, SLUB]
5932 The minimum number of objects per slab. SLUB will
5933 increase the slab order up to slub_max_order to
5934 generate a sufficiently large slab able to contain
5935 the number of objects indicated. The higher the number
5936 of objects the smaller the overhead of tracking slabs
5937 and the less frequently locks need to be acquired.
5938 For more information see Documentation/mm/slub.rst.
5940 slub_min_order= [MM, SLUB]
5941 Determines the minimum page order for slabs. Must be
5942 lower than slub_max_order.
5943 For more information see Documentation/mm/slub.rst.
5945 slub_merge [MM, SLUB]
5946 Same with slab_merge.
5948 slub_nomerge [MM, SLUB]
5949 Same with slab_nomerge. This is supported for legacy.
5950 See slab_nomerge for more information.
5953 Format: <io1>[,<io2>[,...,<io8>]]
5955 smp.csd_lock_timeout= [KNL]
5956 Specify the period of time in milliseconds
5957 that smp_call_function() and friends will wait
5958 for a CPU to release the CSD lock. This is
5959 useful when diagnosing bugs involving CPUs
5960 disabling interrupts for extended periods
5961 of time. Defaults to 5,000 milliseconds, and
5962 setting a value of zero disables this feature.
5963 This feature may be more efficiently disabled
5964 using the csdlock_debug- kernel parameter.
5966 smp.panic_on_ipistall= [KNL]
5967 If a csd_lock_timeout extends for more than
5968 the specified number of milliseconds, panic the
5969 system. By default, let CSD-lock acquisition
5970 take as long as they take. Specifying 300,000
5971 for this value provides a 5-minute timeout.
5973 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
5974 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
5975 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
5976 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
5977 smsc-ircc2.ircc_irq= [HW] IRQ line
5978 smsc-ircc2.ircc_dma= [HW] DMA channel
5979 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
5980 0: Toshiba Satellite 1800 (GP data pin select)
5981 1: Fast pin select (default)
5984 smt= [KNL,MIPS,S390] Set the maximum number of threads (logical
5985 CPUs) to use per physical CPU on systems capable of
5986 symmetric multithreading (SMT). Will be capped to the
5987 actual hardware limit.
5989 Default: -1 (no limit)
5992 [KNL] Should the soft-lockup detector generate panics.
5995 A value of 1 instructs the soft-lockup detector
5996 to panic the machine when a soft-lockup occurs. It is
5997 also controlled by the kernel.softlockup_panic sysctl
5998 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
5999 respective build-time switch to that functionality.
6001 softlockup_all_cpu_backtrace=
6002 [KNL] Should the soft-lockup detector generate
6003 backtraces on all cpus.
6006 sonypi.*= [HW] Sony Programmable I/O Control Device driver
6007 See Documentation/admin-guide/laptops/sonypi.rst
6009 spectre_v2= [X86] Control mitigation of Spectre variant 2
6010 (indirect branch speculation) vulnerability.
6011 The default operation protects the kernel from
6014 on - unconditionally enable, implies
6016 off - unconditionally disable, implies
6018 auto - kernel detects whether your CPU model is
6021 Selecting 'on' will, and 'auto' may, choose a
6022 mitigation method at run time according to the
6023 CPU, the available microcode, the setting of the
6024 CONFIG_RETPOLINE configuration option, and the
6025 compiler with which the kernel was built.
6027 Selecting 'on' will also enable the mitigation
6028 against user space to user space task attacks.
6030 Selecting 'off' will disable both the kernel and
6031 the user space protections.
6033 Specific mitigations can also be selected manually:
6035 retpoline - replace indirect branches
6036 retpoline,generic - Retpolines
6037 retpoline,lfence - LFENCE; indirect branch
6038 retpoline,amd - alias for retpoline,lfence
6039 eibrs - Enhanced/Auto IBRS
6040 eibrs,retpoline - Enhanced/Auto IBRS + Retpolines
6041 eibrs,lfence - Enhanced/Auto IBRS + LFENCE
6042 ibrs - use IBRS to protect kernel
6044 Not specifying this option is equivalent to
6048 [X86] Control mitigation of Spectre variant 2
6049 (indirect branch speculation) vulnerability between
6052 on - Unconditionally enable mitigations. Is
6053 enforced by spectre_v2=on
6055 off - Unconditionally disable mitigations. Is
6056 enforced by spectre_v2=off
6058 prctl - Indirect branch speculation is enabled,
6059 but mitigation can be enabled via prctl
6060 per thread. The mitigation control state
6061 is inherited on fork.
6064 - Like "prctl" above, but only STIBP is
6065 controlled per thread. IBPB is issued
6066 always when switching between different user
6070 - Same as "prctl" above, but all seccomp
6071 threads will enable the mitigation unless
6072 they explicitly opt out.
6075 - Like "seccomp" above, but only STIBP is
6076 controlled per thread. IBPB is issued
6077 always when switching between different
6078 user space processes.
6080 auto - Kernel selects the mitigation depending on
6081 the available CPU features and vulnerability.
6083 Default mitigation: "prctl"
6085 Not specifying this option is equivalent to
6086 spectre_v2_user=auto.
6088 spec_rstack_overflow=
6089 [X86] Control RAS overflow mitigation on AMD Zen CPUs
6091 off - Disable mitigation
6092 microcode - Enable microcode mitigation only
6093 safe-ret - Enable sw-only safe RET mitigation (default)
6094 ibpb - Enable mitigation by issuing IBPB on
6096 ibpb-vmexit - Issue IBPB only on VMEXIT
6097 (cloud-specific mitigation)
6099 spec_store_bypass_disable=
6100 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
6101 (Speculative Store Bypass vulnerability)
6103 Certain CPUs are vulnerable to an exploit against a
6104 a common industry wide performance optimization known
6105 as "Speculative Store Bypass" in which recent stores
6106 to the same memory location may not be observed by
6107 later loads during speculative execution. The idea
6108 is that such stores are unlikely and that they can
6109 be detected prior to instruction retirement at the
6110 end of a particular speculation execution window.
6112 In vulnerable processors, the speculatively forwarded
6113 store can be used in a cache side channel attack, for
6114 example to read memory to which the attacker does not
6115 directly have access (e.g. inside sandboxed code).
6117 This parameter controls whether the Speculative Store
6118 Bypass optimization is used.
6120 On x86 the options are:
6122 on - Unconditionally disable Speculative Store Bypass
6123 off - Unconditionally enable Speculative Store Bypass
6124 auto - Kernel detects whether the CPU model contains an
6125 implementation of Speculative Store Bypass and
6126 picks the most appropriate mitigation. If the
6127 CPU is not vulnerable, "off" is selected. If the
6128 CPU is vulnerable the default mitigation is
6129 architecture and Kconfig dependent. See below.
6130 prctl - Control Speculative Store Bypass per thread
6131 via prctl. Speculative Store Bypass is enabled
6132 for a process by default. The state of the control
6133 is inherited on fork.
6134 seccomp - Same as "prctl" above, but all seccomp threads
6135 will disable SSB unless they explicitly opt out.
6137 Default mitigations:
6140 On powerpc the options are:
6142 on,auto - On Power8 and Power9 insert a store-forwarding
6143 barrier on kernel entry and exit. On Power7
6144 perform a software flush on kernel entry and
6148 Not specifying this option is equivalent to
6149 spec_store_bypass_disable=auto.
6151 spia_io_base= [HW,MTD]
6157 [X86] Enable split lock detection or bus lock detection
6159 When enabled (and if hardware support is present), atomic
6160 instructions that access data across cache line
6161 boundaries will result in an alignment check exception
6162 for split lock detection or a debug exception for
6167 warn - the kernel will emit rate-limited warnings
6168 about applications triggering the #AC
6169 exception or the #DB exception. This mode is
6170 the default on CPUs that support split lock
6171 detection or bus lock detection. Default
6172 behavior is by #AC if both features are
6173 enabled in hardware.
6175 fatal - the kernel will send SIGBUS to applications
6176 that trigger the #AC exception or the #DB
6177 exception. Default behavior is by #AC if
6178 both features are enabled in hardware.
6181 Set system wide rate limit to N bus locks
6182 per second for bus lock detection.
6185 N/A for split lock detection.
6188 If an #AC exception is hit in the kernel or in
6189 firmware (i.e. not while executing in user mode)
6190 the kernel will oops in either "warn" or "fatal"
6193 #DB exception for bus lock is triggered only when
6197 Control the Special Register Buffer Data Sampling
6200 Certain CPUs are vulnerable to an MDS-like
6201 exploit which can leak bits from the random
6204 By default, this issue is mitigated by
6205 microcode. However, the microcode fix can cause
6206 the RDRAND and RDSEED instructions to become
6207 much slower. Among other effects, this will
6208 result in reduced throughput from /dev/urandom.
6210 The microcode mitigation can be disabled with
6211 the following option:
6213 off: Disable mitigation and remove
6214 performance impact to RDRAND and RDSEED
6216 srcutree.big_cpu_lim [KNL]
6217 Specifies the number of CPUs constituting a
6218 large system, such that srcu_struct structures
6219 should immediately allocate an srcu_node array.
6220 This kernel-boot parameter defaults to 128,
6221 but takes effect only when the low-order four
6222 bits of srcutree.convert_to_big is equal to 3
6225 srcutree.convert_to_big [KNL]
6226 Specifies under what conditions an SRCU tree
6227 srcu_struct structure will be converted to big
6228 form, that is, with an rcu_node tree:
6231 1: At init_srcu_struct() time.
6232 2: When rcutorture decides to.
6233 3: Decide at boot time (default).
6234 0x1X: Above plus if high contention.
6236 Either way, the srcu_node tree will be sized based
6237 on the actual runtime number of CPUs (nr_cpu_ids)
6238 instead of the compile-time CONFIG_NR_CPUS.
6240 srcutree.counter_wrap_check [KNL]
6241 Specifies how frequently to check for
6242 grace-period sequence counter wrap for the
6243 srcu_data structure's ->srcu_gp_seq_needed field.
6244 The greater the number of bits set in this kernel
6245 parameter, the less frequently counter wrap will
6246 be checked for. Note that the bottom two bits
6249 srcutree.exp_holdoff [KNL]
6250 Specifies how many nanoseconds must elapse
6251 since the end of the last SRCU grace period for
6252 a given srcu_struct until the next normal SRCU
6253 grace period will be considered for automatic
6254 expediting. Set to zero to disable automatic
6257 srcutree.srcu_max_nodelay [KNL]
6258 Specifies the number of no-delay instances
6259 per jiffy for which the SRCU grace period
6260 worker thread will be rescheduled with zero
6261 delay. Beyond this limit, worker thread will
6262 be rescheduled with a sleep delay of one jiffy.
6264 srcutree.srcu_max_nodelay_phase [KNL]
6265 Specifies the per-grace-period phase, number of
6266 non-sleeping polls of readers. Beyond this limit,
6267 grace period worker thread will be rescheduled
6268 with a sleep delay of one jiffy, between each
6269 rescan of the readers, for a grace period phase.
6271 srcutree.srcu_retry_check_delay [KNL]
6272 Specifies number of microseconds of non-sleeping
6273 delay between each non-sleeping poll of readers.
6275 srcutree.small_contention_lim [KNL]
6276 Specifies the number of update-side contention
6277 events per jiffy will be tolerated before
6278 initiating a conversion of an srcu_struct
6279 structure to big form. Note that the value of
6280 srcutree.convert_to_big must have the 0x10 bit
6281 set for contention-based conversions to occur.
6284 Speculative Store Bypass Disable control
6286 On CPUs that are vulnerable to the Speculative
6287 Store Bypass vulnerability and offer a
6288 firmware based mitigation, this parameter
6289 indicates how the mitigation should be used:
6291 force-on: Unconditionally enable mitigation for
6292 for both kernel and userspace
6293 force-off: Unconditionally disable mitigation for
6294 for both kernel and userspace
6295 kernel: Always enable mitigation in the
6296 kernel, and offer a prctl interface
6297 to allow userspace to register its
6298 interest in being mitigated too.
6300 stack_guard_gap= [MM]
6301 override the default stack gap protection. The value
6302 is in page units and it defines how many pages prior
6303 to (for stacks growing down) resp. after (for stacks
6304 growing up) the main stack are reserved for no other
6305 mapping. Default value is 256 pages.
6307 stack_depot_disable= [KNL]
6308 Setting this to true through kernel command line will
6309 disable the stack depot thereby saving the static memory
6310 consumed by the stack hash table. By default this is set
6314 Enabled the stack tracer on boot up.
6316 stacktrace_filter=[function-list]
6317 [FTRACE] Limit the functions that the stack tracer
6318 will trace at boot up. function-list is a comma-separated
6319 list of functions. This list can be changed at run
6320 time by the stack_trace_filter file in the debugfs
6321 tracing directory. Note, this enables stack tracing
6322 and the stacktrace above is not needed.
6326 Set the STI (builtin display/keyboard on the HP-PARISC
6327 machines) console (graphic card) which should be used
6328 as the initial boot-console.
6329 See also comment in drivers/video/console/sticore.c.
6332 See comment in drivers/video/console/sticore.c.
6335 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
6340 Enable or disable strict sigaltstack size checks
6341 against the required signal frame size which
6342 depends on the supported FPU features. This can
6343 be used to filter out binaries which have
6344 not yet been made aware of AT_MINSIGSTKSZ.
6347 Limits the number of kernel HPT entries in the hash
6348 page table to increase the rate of hash page table
6349 faults on kernel addresses.
6352 Limits the number of kernel SLB entries, and flushes
6353 them frequently to increase the rate of SLB faults
6354 on kernel addresses.
6356 sunrpc.min_resvport=
6357 sunrpc.max_resvport=
6359 SunRPC servers often require that client requests
6360 originate from a privileged port (i.e. a port in the
6361 range 0 < portnr < 1024).
6362 An administrator who wishes to reserve some of these
6363 ports for other uses may adjust the range that the
6364 kernel's sunrpc client considers to be privileged
6365 using these two parameters to set the minimum and
6366 maximum port values.
6368 sunrpc.svc_rpc_per_connection_limit=
6370 Limit the number of requests that the server will
6371 process in parallel from a single connection.
6372 The default value is 0 (no limit).
6376 Control how the NFS server code allocates CPUs to
6377 service thread pools. Depending on how many NICs
6378 you have and where their interrupts are bound, this
6379 option will affect which CPUs will do NFS serving.
6380 Note: this parameter cannot be changed while the
6381 NFS server is running.
6383 auto the server chooses an appropriate mode
6384 automatically using heuristics
6385 global a single global pool contains all CPUs
6386 percpu one pool for each CPU
6387 pernode one pool for each NUMA node (equivalent
6388 to global on non-NUMA machines)
6390 sunrpc.tcp_slot_table_entries=
6391 sunrpc.udp_slot_table_entries=
6393 Sets the upper limit on the number of simultaneous
6394 RPC calls that can be sent from the client to a
6395 server. Increasing these values may allow you to
6396 improve throughput, but will also increase the
6397 amount of memory reserved for use by the client.
6399 suspend.pm_test_delay=
6401 Sets the number of seconds to remain in a suspend test
6402 mode before resuming the system (see
6403 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
6404 is set. Default value is 5.
6407 Format: { on | off | y | n | 1 | 0 }
6408 This parameter controls use of the Protected
6409 Execution Facility on pSeries.
6411 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
6412 Format: { <int> [,<int>] | force | noforce }
6413 <int> -- Number of I/O TLB slabs
6414 <int> -- Second integer after comma. Number of swiotlb
6415 areas with their own lock. Will be rounded up
6417 force -- force using of bounce buffers even if they
6418 wouldn't be automatically used by the kernel
6419 noforce -- Never use bounce buffers (for debugging)
6424 Set a sysctl parameter, right before loading the init
6425 process, as if the value was written to the respective
6426 /proc/sys/... file. Both '.' and '/' are recognized as
6427 separators. Unrecognized parameters and invalid values
6428 are reported in the kernel log. Sysctls registered
6429 later by a loaded module cannot be set this way.
6430 Example: sysctl.vm.swappiness=40
6432 sysrq_always_enabled
6434 Ignore sysrq setting - this boot parameter will
6435 neutralize any effect of /proc/sys/kernel/sysrq.
6436 Useful for debugging.
6438 tcpmhash_entries= [KNL,NET]
6439 Set the number of tcp_metrics_hash slots.
6440 Default value is 8192 or 16384 depending on total
6441 ram pages. This is used to specify the TCP metrics
6442 cache size. See Documentation/networking/ip-sysctl.rst
6443 "tcp_no_metrics_save" section for more details.
6447 test_suspend= [SUSPEND]
6448 Format: { "mem" | "standby" | "freeze" }[,N]
6449 Specify "mem" (for Suspend-to-RAM) or "standby" (for
6450 standby suspend) or "freeze" (for suspend type freeze)
6451 as the system sleep state during system startup with
6452 the optional capability to repeat N number of times.
6453 The system is woken from this state using a
6454 wakeup-capable RTC alarm.
6456 thash_entries= [KNL,NET]
6457 Set number of hash buckets for TCP connection
6459 thermal.act= [HW,ACPI]
6460 -1: disable all active trip points in all thermal zones
6461 <degrees C>: override all lowest active trip points
6463 thermal.crt= [HW,ACPI]
6464 -1: disable all critical trip points in all thermal zones
6465 <degrees C>: override all critical trip points
6467 thermal.off= [HW,ACPI]
6468 1: disable ACPI thermal control
6470 thermal.psv= [HW,ACPI]
6471 -1: disable all passive trip points
6472 <degrees C>: override all passive trip points to this
6475 thermal.tzp= [HW,ACPI]
6476 Specify global default ACPI thermal zone polling rate
6477 <deci-seconds>: poll all this frequency
6478 0: no polling (default)
6481 Force threading of all interrupt handlers except those
6482 marked explicitly IRQF_NO_THREAD.
6486 Specify if the kernel should make use of the cpu
6487 topology information if the hardware supports this.
6488 The scheduler will make use of this information and
6489 e.g. base its process migration decisions on it.
6492 topology_updates= [KNL, PPC, NUMA]
6494 Specify if the kernel should ignore (off)
6495 topology updates sent by the hypervisor to this
6498 torture.disable_onoff_at_boot= [KNL]
6499 Prevent the CPU-hotplug component of torturing
6500 until after init has spawned.
6502 torture.ftrace_dump_at_shutdown= [KNL]
6503 Dump the ftrace buffer at torture-test shutdown,
6504 even if there were no errors. This can be a
6505 very costly operation when many torture tests
6506 are running concurrently, especially on systems
6507 with rotating-rust storage.
6509 torture.verbose_sleep_frequency= [KNL]
6510 Specifies how many verbose printk()s should be
6511 emitted between each sleep. The default of zero
6512 disables verbose-printk() sleeping.
6514 torture.verbose_sleep_duration= [KNL]
6515 Duration of each verbose-printk() sleep in jiffies.
6519 tpm_suspend_pcr=[HW,TPM]
6520 Format: integer pcr id
6521 Specify that at suspend time, the tpm driver
6522 should extend the specified pcr with zeros,
6523 as a workaround for some chips which fail to
6524 flush the last written pcr on TPM_SaveState.
6525 This will guarantee that all the other pcrs
6528 tpm_tis.interrupts= [HW,TPM]
6529 Enable interrupts for the MMIO based physical layer
6530 for the FIFO interface. By default it is set to false
6531 (0). For more information about TPM hardware interfaces
6532 defined by Trusted Computing Group (TCG) see
6533 https://trustedcomputinggroup.org/resource/pc-client-platform-tpm-profile-ptp-specification/
6536 Have the tracepoints sent to printk as well as the
6537 tracing ring buffer. This is useful for early boot up
6538 where the system hangs or reboots and does not give the
6539 option for reading the tracing buffer or performing a
6540 ftrace_dump_on_oops.
6542 To turn off having tracepoints sent to printk,
6543 echo 0 > /proc/sys/kernel/tracepoint_printk
6544 Note, echoing 1 into this file without the
6545 tracepoint_printk kernel cmdline option has no effect.
6547 The tp_printk_stop_on_boot (see below) can also be used
6548 to stop the printing of events to console at
6553 Having tracepoints sent to printk() and activating high
6554 frequency tracepoints such as irq or sched, can cause
6555 the system to live lock.
6557 tp_printk_stop_on_boot [FTRACE]
6558 When tp_printk (above) is set, it can cause a lot of noise
6559 on the console. It may be useful to only include the
6560 printing of events during boot up, as user space may
6561 make the system inoperable.
6563 This command line option will stop the printing of events
6564 to console at the late_initcall_sync() time frame.
6566 trace_buf_size=nn[KMG]
6567 [FTRACE] will set tracing buffer size on each cpu.
6569 trace_clock= [FTRACE] Set the clock used for tracing events
6571 local - Use the per CPU time stamp counter
6572 (converted into nanoseconds). Fast, but
6573 depending on the architecture, may not be
6574 in sync between CPUs.
6575 global - Event time stamps are synchronize across
6576 CPUs. May be slower than the local clock,
6577 but better for some race conditions.
6578 counter - Simple counting of events (1, 2, ..)
6579 note, some counts may be skipped due to the
6580 infrastructure grabbing the clock more than
6582 uptime - Use jiffies as the time stamp.
6583 perf - Use the same clock that perf uses.
6584 mono - Use ktime_get_mono_fast_ns() for time stamps.
6585 mono_raw - Use ktime_get_raw_fast_ns() for time
6587 boot - Use ktime_get_boot_fast_ns() for time stamps.
6588 Architectures may add more clocks. See
6589 Documentation/trace/ftrace.rst for more details.
6591 trace_event=[event-list]
6592 [FTRACE] Set and start specified trace events in order
6593 to facilitate early boot debugging. The event-list is a
6594 comma-separated list of trace events to enable. See
6595 also Documentation/trace/events.rst
6597 trace_instance=[instance-info]
6598 [FTRACE] Create a ring buffer instance early in boot up.
6599 This will be listed in:
6601 /sys/kernel/tracing/instances
6603 Events can be enabled at the time the instance is created
6606 trace_instance=<name>,<system1>:<event1>,<system2>:<event2>
6608 Note, the "<system*>:" portion is optional if the event is
6611 trace_instance=foo,sched:sched_switch,irq_handler_entry,initcall
6613 will enable the "sched_switch" event (note, the "sched:" is optional, and
6614 the same thing would happen if it was left off). The irq_handler_entry
6615 event, and all events under the "initcall" system.
6617 trace_options=[option-list]
6618 [FTRACE] Enable or disable tracer options at boot.
6619 The option-list is a comma delimited list of options
6620 that can be enabled or disabled just as if you were
6621 to echo the option name into
6623 /sys/kernel/tracing/trace_options
6625 For example, to enable stacktrace option (to dump the
6626 stack trace of each event), add to the command line:
6628 trace_options=stacktrace
6630 See also Documentation/trace/ftrace.rst "trace options"
6633 trace_trigger=[trigger-list]
6634 [FTRACE] Add a event trigger on specific events.
6635 Set a trigger on top of a specific event, with an optional
6638 The format is is "trace_trigger=<event>.<trigger>[ if <filter>],..."
6639 Where more than one trigger may be specified that are comma deliminated.
6643 trace_trigger="sched_switch.stacktrace if prev_state == 2"
6645 The above will enable the "stacktrace" trigger on the "sched_switch"
6646 event but only trigger it if the "prev_state" of the "sched_switch"
6647 event is "2" (TASK_UNINTERUPTIBLE).
6649 See also "Event triggers" in Documentation/trace/events.rst
6653 [FTRACE] enable this option to disable tracing when a
6654 warning is hit. This turns off "tracing_on". Tracing can
6655 be enabled again by echoing '1' into the "tracing_on"
6656 file located in /sys/kernel/tracing/
6658 This option is useful, as it disables the trace before
6659 the WARNING dump is called, which prevents the trace to
6660 be filled with content caused by the warning output.
6662 This option can also be set at run time via the sysctl
6663 option: kernel/traceoff_on_warning
6665 transparent_hugepage=
6667 Format: [always|madvise|never]
6668 Can be used to control the default behavior of the system
6669 with respect to transparent hugepages.
6670 See Documentation/admin-guide/mm/transhuge.rst
6673 trusted.source= [KEYS]
6675 This parameter identifies the trust source as a backend
6676 for trusted keys implementation. Supported trust
6681 If not specified then it defaults to iterating through
6682 the trust source list starting with TPM and assigns the
6683 first trust source as a backend which is initialized
6684 successfully during iteration.
6688 The RNG used to generate key material for trusted keys.
6691 - the same value as trusted.source: "tpm" or "tee"
6693 If not specified, "default" is used. In this case,
6694 the RNG's choice is left to each individual trust source.
6696 tsc= Disable clocksource stability checks for TSC.
6698 [x86] reliable: mark tsc clocksource as reliable, this
6699 disables clocksource verification at runtime, as well
6700 as the stability checks done at bootup. Used to enable
6701 high-resolution timer mode on older hardware, and in
6702 virtualized environment.
6703 [x86] noirqtime: Do not use TSC to do irq accounting.
6704 Used to run time disable IRQ_TIME_ACCOUNTING on any
6705 platforms where RDTSC is slow and this accounting
6707 [x86] unstable: mark the TSC clocksource as unstable, this
6708 marks the TSC unconditionally unstable at bootup and
6709 avoids any further wobbles once the TSC watchdog notices.
6710 [x86] nowatchdog: disable clocksource watchdog. Used
6711 in situations with strict latency requirements (where
6712 interruptions from clocksource watchdog are not
6714 [x86] recalibrate: force recalibration against a HW timer
6715 (HPET or PM timer) on systems whose TSC frequency was
6716 obtained from HW or FW using either an MSR or CPUID(0x15).
6717 Warn if the difference is more than 500 ppm.
6718 [x86] watchdog: Use TSC as the watchdog clocksource with
6719 which to check other HW timers (HPET or PM timer), but
6720 only on systems where TSC has been deemed trustworthy.
6721 This will be suppressed by an earlier tsc=nowatchdog and
6722 can be overridden by a later tsc=nowatchdog. A console
6723 message will flag any such suppression or overriding.
6725 tsc_early_khz= [X86] Skip early TSC calibration and use the given
6726 value instead. Useful when the early TSC frequency discovery
6727 procedure is not reliable, such as on overclocked systems
6728 with CPUID.16h support and partial CPUID.15h support.
6729 Format: <unsigned int>
6731 tsx= [X86] Control Transactional Synchronization
6732 Extensions (TSX) feature in Intel processors that
6733 support TSX control.
6735 This parameter controls the TSX feature. The options are:
6737 on - Enable TSX on the system. Although there are
6738 mitigations for all known security vulnerabilities,
6739 TSX has been known to be an accelerator for
6740 several previous speculation-related CVEs, and
6741 so there may be unknown security risks associated
6742 with leaving it enabled.
6744 off - Disable TSX on the system. (Note that this
6745 option takes effect only on newer CPUs which are
6746 not vulnerable to MDS, i.e., have
6747 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
6748 the new IA32_TSX_CTRL MSR through a microcode
6749 update. This new MSR allows for the reliable
6750 deactivation of the TSX functionality.)
6752 auto - Disable TSX if X86_BUG_TAA is present,
6753 otherwise enable TSX on the system.
6755 Not specifying this option is equivalent to tsx=off.
6757 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6760 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
6761 Abort (TAA) vulnerability.
6763 Similar to Micro-architectural Data Sampling (MDS)
6764 certain CPUs that support Transactional
6765 Synchronization Extensions (TSX) are vulnerable to an
6766 exploit against CPU internal buffers which can forward
6767 information to a disclosure gadget under certain
6770 In vulnerable processors, the speculatively forwarded
6771 data can be used in a cache side channel attack, to
6772 access data to which the attacker does not have direct
6775 This parameter controls the TAA mitigation. The
6778 full - Enable TAA mitigation on vulnerable CPUs
6781 full,nosmt - Enable TAA mitigation and disable SMT on
6782 vulnerable CPUs. If TSX is disabled, SMT
6783 is not disabled because CPU is not
6784 vulnerable to cross-thread TAA attacks.
6785 off - Unconditionally disable TAA mitigation
6787 On MDS-affected machines, tsx_async_abort=off can be
6788 prevented by an active MDS mitigation as both vulnerabilities
6789 are mitigated with the same mechanism so in order to disable
6790 this mitigation, you need to specify mds=off too.
6792 Not specifying this option is equivalent to
6793 tsx_async_abort=full. On CPUs which are MDS affected
6794 and deploy MDS mitigation, TAA mitigation is not
6795 required and doesn't provide any additional
6799 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6801 turbografx.map[2|3]= [HW,JOY]
6802 TurboGraFX parallel port interface
6804 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
6805 See also Documentation/input/devices/joystick-parport.rst
6807 udbg-immortal [PPC] When debugging early kernel crashes that
6808 happen after console_init() and before a proper
6809 console driver takes over, this boot options might
6810 help "seeing" what's going on.
6812 uhash_entries= [KNL,NET]
6813 Set number of hash buckets for UDP/UDP-Lite connections
6816 [USB] Ignore overcurrent events (default N).
6817 Some badly-designed motherboards generate lots of
6818 bogus events, for ports that aren't wired to
6819 anything. Set this parameter to avoid log spamming.
6820 Note that genuine overcurrent events won't be
6824 [X86] Cause panic on unknown NMI.
6826 unwind_debug [X86-64]
6827 Enable unwinder debug output. This can be
6828 useful for debugging certain unwinder error
6829 conditions, including corrupt stacks and
6830 bad/missing unwinder metadata.
6832 usbcore.authorized_default=
6833 [USB] Default USB device authorization:
6834 (default -1 = authorized (same as 1),
6835 0 = not authorized, 1 = authorized, 2 = authorized
6836 if device connected to internal port)
6838 usbcore.autosuspend=
6839 [USB] The autosuspend time delay (in seconds) used
6840 for newly-detected USB devices (default 2). This
6841 is the time required before an idle device will be
6842 autosuspended. Devices for which the delay is set
6843 to a negative value won't be autosuspended at all.
6845 usbcore.usbfs_snoop=
6846 [USB] Set to log all usbfs traffic (default 0 = off).
6848 usbcore.usbfs_snoop_max=
6849 [USB] Maximum number of bytes to snoop in each URB
6852 usbcore.blinkenlights=
6853 [USB] Set to cycle leds on hubs (default 0 = off).
6855 usbcore.old_scheme_first=
6856 [USB] Start with the old device initialization
6857 scheme (default 0 = off).
6859 usbcore.usbfs_memory_mb=
6860 [USB] Memory limit (in MB) for buffers allocated by
6861 usbfs (default = 16, 0 = max = 2047).
6863 usbcore.use_both_schemes=
6864 [USB] Try the other device initialization scheme
6865 if the first one fails (default 1 = enabled).
6867 usbcore.initial_descriptor_timeout=
6868 [USB] Specifies timeout for the initial 64-byte
6869 USB_REQ_GET_DESCRIPTOR request in milliseconds
6870 (default 5000 = 5.0 seconds).
6872 usbcore.nousb [USB] Disable the USB subsystem
6875 [USB] A list of quirk entries to augment the built-in
6876 usb core quirk list. List entries are separated by
6877 commas. Each entry has the form
6878 VendorID:ProductID:Flags. The IDs are 4-digit hex
6879 numbers and Flags is a set of letters. Each letter
6880 will change the built-in quirk; setting it if it is
6881 clear and clearing it if it is set. The letters have
6882 the following meanings:
6883 a = USB_QUIRK_STRING_FETCH_255 (string
6884 descriptors must not be fetched using
6886 b = USB_QUIRK_RESET_RESUME (device can't resume
6887 correctly so reset it instead);
6888 c = USB_QUIRK_NO_SET_INTF (device can't handle
6889 Set-Interface requests);
6890 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
6891 handle its Configuration or Interface
6893 e = USB_QUIRK_RESET (device can't be reset
6894 (e.g morph devices), don't use reset);
6895 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
6896 more interface descriptions than the
6897 bNumInterfaces count, and can't handle
6898 talking to these interfaces);
6899 g = USB_QUIRK_DELAY_INIT (device needs a pause
6900 during initialization, after we read
6901 the device descriptor);
6902 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
6903 high speed and super speed interrupt
6904 endpoints, the USB 2.0 and USB 3.0 spec
6905 require the interval in microframes (1
6906 microframe = 125 microseconds) to be
6907 calculated as interval = 2 ^
6909 Devices with this quirk report their
6910 bInterval as the result of this
6911 calculation instead of the exponent
6912 variable used in the calculation);
6913 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
6914 handle device_qualifier descriptor
6916 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
6917 generates spurious wakeup, ignore
6918 remote wakeup capability);
6919 k = USB_QUIRK_NO_LPM (device can't handle Link
6921 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
6922 (Device reports its bInterval as linear
6923 frames instead of the USB 2.0
6925 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
6926 to be disconnected before suspend to
6927 prevent spurious wakeup);
6928 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
6929 pause after every control message);
6930 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
6931 delay after resetting its port);
6932 p = USB_QUIRK_SHORT_SET_ADDRESS_REQ_TIMEOUT
6933 (Reduce timeout of the SET_ADDRESS
6934 request from 5000 ms to 500 ms);
6935 Example: quirks=0781:5580:bk,0a5c:5834:gij
6938 [USBHID] The interval which mice are to be polled at.
6941 [USBHID] The interval which joysticks are to be polled at.
6944 [USBHID] The interval which keyboards are to be polled at.
6946 usb-storage.delay_use=
6947 [UMS] The delay in seconds before a new device is
6948 scanned for Logical Units (default 1).
6951 [UMS] A list of quirks entries to supplement or
6952 override the built-in unusual_devs list. List
6953 entries are separated by commas. Each entry has
6954 the form VID:PID:Flags where VID and PID are Vendor
6955 and Product ID values (4-digit hex numbers) and
6956 Flags is a set of characters, each corresponding
6957 to a common usb-storage quirk flag as follows:
6958 a = SANE_SENSE (collect more than 18 bytes
6959 of sense data, not on uas);
6960 b = BAD_SENSE (don't collect more than 18
6961 bytes of sense data, not on uas);
6962 c = FIX_CAPACITY (decrease the reported
6963 device capacity by one sector);
6964 d = NO_READ_DISC_INFO (don't use
6965 READ_DISC_INFO command, not on uas);
6966 e = NO_READ_CAPACITY_16 (don't use
6967 READ_CAPACITY_16 command);
6968 f = NO_REPORT_OPCODES (don't use report opcodes
6970 g = MAX_SECTORS_240 (don't transfer more than
6971 240 sectors at a time, uas only);
6972 h = CAPACITY_HEURISTICS (decrease the
6973 reported device capacity by one
6974 sector if the number is odd);
6975 i = IGNORE_DEVICE (don't bind to this
6977 j = NO_REPORT_LUNS (don't use report luns
6979 k = NO_SAME (do not use WRITE_SAME, uas only)
6980 l = NOT_LOCKABLE (don't try to lock and
6981 unlock ejectable media, not on uas);
6982 m = MAX_SECTORS_64 (don't transfer more
6983 than 64 sectors = 32 KB at a time,
6985 n = INITIAL_READ10 (force a retry of the
6986 initial READ(10) command, not on uas);
6987 o = CAPACITY_OK (accept the capacity
6988 reported by the device, not on uas);
6989 p = WRITE_CACHE (the device cache is ON
6990 by default, not on uas);
6991 r = IGNORE_RESIDUE (the device reports
6992 bogus residue values, not on uas);
6993 s = SINGLE_LUN (the device has only one
6995 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
6996 commands, uas only);
6997 u = IGNORE_UAS (don't bind to the uas driver);
6998 w = NO_WP_DETECT (don't test whether the
6999 medium is write-protected).
7000 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
7001 even if the device claims no cache,
7003 Example: quirks=0419:aaf5:rl,0421:0433:rc
7005 user_debug= [KNL,ARM]
7007 See arch/arm/Kconfig.debug help text.
7008 1 - undefined instruction events
7010 4 - invalid data aborts
7013 Example: user_debug=31
7016 [X86] Flags controlling user PTE allocations.
7018 nohigh = do not allocate PTE pages in
7019 HIGHMEM regardless of setting
7022 vdso= [X86,SH,SPARC]
7023 On X86_32, this is an alias for vdso32=. Otherwise:
7025 vdso=1: enable VDSO (the default)
7026 vdso=0: disable VDSO mapping
7028 vdso32= [X86] Control the 32-bit vDSO
7029 vdso32=1: enable 32-bit VDSO
7030 vdso32=0 or vdso32=2: disable 32-bit VDSO
7032 See the help text for CONFIG_COMPAT_VDSO for more
7033 details. If CONFIG_COMPAT_VDSO is set, the default is
7034 vdso32=0; otherwise, the default is vdso32=1.
7036 For compatibility with older kernels, vdso32=2 is an
7039 Try vdso32=0 if you encounter an error that says:
7040 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
7043 vector=percpu: enable percpu vector domain
7045 video= [FB] Frame buffer configuration
7046 See Documentation/fb/modedb.rst.
7048 video.brightness_switch_enabled= [ACPI]
7050 If set to 1, on receiving an ACPI notify event
7051 generated by hotkey, video driver will adjust brightness
7052 level and then send out the event to user space through
7053 the allocated input device. If set to 0, video driver
7054 will only send out the event without touching backlight
7059 [VMMIO] Memory mapped virtio (platform) device.
7061 <size>@<baseaddr>:<irq>[:<id>]
7063 <size> := size (can use standard suffixes
7065 <baseaddr> := physical base address
7066 <irq> := interrupt number (as passed to
7068 <id> := (optional) platform device id
7070 virtio_mmio.device=1K@0x100b0000:48:7
7072 Can be used multiple times for multiple devices.
7074 vga= [BOOT,X86-32] Select a particular video mode
7075 See Documentation/arch/x86/boot.rst and
7076 Documentation/admin-guide/svga.rst.
7077 Use vga=ask for menu.
7078 This is actually a boot loader parameter; the value is
7079 passed to the kernel using a special protocol.
7081 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
7082 May slow down system boot speed, especially when
7083 enabled on systems with a large amount of memory.
7084 All options are enabled by default, and this
7085 interface is meant to allow for selectively
7086 enabling or disabling specific virtual memory
7089 Available options are:
7090 P Enable page structure init time poisoning
7091 - Disable all of the above options
7093 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
7094 size of <nn>. This can be used to increase the
7095 minimum size (128MB on x86). It can also be used to
7096 decrease the size and leave more room for directly
7099 vmcp_cma=nn[MG] [KNL,S390]
7100 Sets the memory size reserved for contiguous memory
7101 allocations for the vmcp device driver.
7103 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
7106 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
7109 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
7113 Controls the behavior of vsyscalls (i.e. calls to
7114 fixed addresses of 0xffffffffff600x00 from legacy
7115 code). Most statically-linked binaries and older
7116 versions of glibc use these calls. Because these
7117 functions are at fixed addresses, they make nice
7118 targets for exploits that can control RIP.
7120 emulate Vsyscalls turn into traps and are emulated
7121 reasonably safely. The vsyscall page is
7124 xonly [default] Vsyscalls turn into traps and are
7125 emulated reasonably safely. The vsyscall
7126 page is not readable.
7128 none Vsyscalls don't work at all. This makes
7129 them quite hard to use for exploits but
7130 might break your system.
7132 vt.color= [VT] Default text color.
7133 Format: 0xYX, X = foreground, Y = background.
7134 Default: 0x07 = light gray on black.
7136 vt.cur_default= [VT] Default cursor shape.
7137 Format: 0xCCBBAA, where AA, BB, and CC are the same as
7138 the parameters of the <Esc>[?A;B;Cc escape sequence;
7139 see VGA-softcursor.txt. Default: 2 = underline.
7141 vt.default_blu= [VT]
7142 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
7143 Change the default blue palette of the console.
7144 This is a 16-member array composed of values
7147 vt.default_grn= [VT]
7148 Format: <green0>,<green1>,<green2>,...,<green15>
7149 Change the default green palette of the console.
7150 This is a 16-member array composed of values
7153 vt.default_red= [VT]
7154 Format: <red0>,<red1>,<red2>,...,<red15>
7155 Change the default red palette of the console.
7156 This is a 16-member array composed of values
7162 Set system-wide default UTF-8 mode for all tty's.
7163 Default is 1, i.e. UTF-8 mode is enabled for all
7164 newly opened terminals.
7166 vt.global_cursor_default=
7169 Set system-wide default for whether a cursor
7170 is shown on new VTs. Default is -1,
7171 i.e. cursors will be created by default unless
7172 overridden by individual drivers. 0 will hide
7173 cursors, 1 will display them.
7175 vt.italic= [VT] Default color for italic text; 0-15.
7178 vt.underline= [VT] Default color for underlined text; 0-15.
7181 watchdog timers [HW,WDT] For information on watchdog timers,
7182 see Documentation/watchdog/watchdog-parameters.rst
7183 or other driver-specific files in the
7184 Documentation/watchdog/ directory.
7188 Set the hard lockup detector stall duration
7189 threshold in seconds. The soft lockup detector
7190 threshold is set to twice the value. A value of 0
7191 disables both lockup detectors. Default is 10
7194 workqueue.unbound_cpus=
7195 [KNL,SMP] Specify to constrain one or some CPUs
7196 to use in unbound workqueues.
7198 By default, all online CPUs are available for
7201 workqueue.watchdog_thresh=
7202 If CONFIG_WQ_WATCHDOG is configured, workqueue can
7203 warn stall conditions and dump internal state to
7204 help debugging. 0 disables workqueue stall
7205 detection; otherwise, it's the stall threshold
7206 duration in seconds. The default value is 30 and
7207 it can be updated at runtime by writing to the
7208 corresponding sysfs file.
7210 workqueue.cpu_intensive_thresh_us=
7211 Per-cpu work items which run for longer than this
7212 threshold are automatically considered CPU intensive
7213 and excluded from concurrency management to prevent
7214 them from noticeably delaying other per-cpu work
7215 items. Default is 10000 (10ms).
7217 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel
7218 will report the work functions which violate this
7219 threshold repeatedly. They are likely good
7220 candidates for using WQ_UNBOUND workqueues instead.
7222 workqueue.power_efficient
7223 Per-cpu workqueues are generally preferred because
7224 they show better performance thanks to cache
7225 locality; unfortunately, per-cpu workqueues tend to
7226 be more power hungry than unbound workqueues.
7228 Enabling this makes the per-cpu workqueues which
7229 were observed to contribute significantly to power
7230 consumption unbound, leading to measurably lower
7231 power usage at the cost of small performance
7234 The default value of this parameter is determined by
7235 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
7237 workqueue.default_affinity_scope=
7238 Select the default affinity scope to use for unbound
7239 workqueues. Can be one of "cpu", "smt", "cache",
7240 "numa" and "system". Default is "cache". For more
7241 information, see the Affinity Scopes section in
7242 Documentation/core-api/workqueue.rst.
7244 This can be changed after boot by writing to the
7245 matching /sys/module/workqueue/parameters file. All
7246 workqueues with the "default" affinity scope will be
7247 updated accordignly.
7249 workqueue.debug_force_rr_cpu
7250 Workqueue used to implicitly guarantee that work
7251 items queued without explicit CPU specified are put
7252 on the local CPU. This guarantee is no longer true
7253 and while local CPU is still preferred work items
7254 may be put on foreign CPUs. This debug option
7255 forces round-robin CPU selection to flush out
7256 usages which depend on the now broken guarantee.
7257 When enabled, memory and cache locality will be
7260 writecombine= [LOONGARCH] Control the MAT (Memory Access Type) of
7263 on - Enable writecombine, use WUC for ioremap_wc()
7264 off - Disable writecombine, use SUC for ioremap_wc()
7266 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
7267 default x2apic cluster mode on platforms
7270 xen_512gb_limit [KNL,X86-64,XEN]
7271 Restricts the kernel running paravirtualized under Xen
7272 to use only up to 512 GB of RAM. The reason to do so is
7273 crash analysis tools and Xen tools for doing domain
7274 save/restore/migration must be enabled to handle larger
7277 xen_emul_unplug= [HW,X86,XEN]
7278 Unplug Xen emulated devices
7279 Format: [unplug0,][unplug1]
7280 ide-disks -- unplug primary master IDE devices
7281 aux-ide-disks -- unplug non-primary-master IDE devices
7282 nics -- unplug network devices
7283 all -- unplug all emulated devices (NICs and IDE disks)
7284 unnecessary -- unplugging emulated devices is
7285 unnecessary even if the host did not respond to
7287 never -- do not unplug even if version check succeeds
7289 xen_legacy_crash [X86,XEN]
7290 Crash from Xen panic notifier, without executing late
7291 panic() code such as dumping handler.
7293 xen_msr_safe= [X86,XEN]
7295 Select whether to always use non-faulting (safe) MSR
7296 access functions when running as Xen PV guest. The
7297 default value is controlled by CONFIG_XEN_PV_MSR_SAFE.
7299 xen_nopvspin [X86,XEN]
7300 Disables the qspinlock slowpath using Xen PV optimizations.
7301 This parameter is obsoleted by "nopvspin" parameter, which
7302 has equivalent effect for XEN platform.
7305 Disables the PV optimizations forcing the HVM guest to
7306 run as generic HVM guest with no PV drivers.
7307 This option is obsoleted by the "nopv" option, which
7308 has equivalent effect for XEN platform.
7310 xen_no_vector_callback
7311 [KNL,X86,XEN] Disable the vector callback for Xen
7312 event channel interrupts.
7314 xen_scrub_pages= [XEN]
7315 Boolean option to control scrubbing pages before giving them back
7316 to Xen, for use by other domains. Can be also changed at runtime
7317 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
7318 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
7320 xen_timer_slop= [X86-64,XEN]
7321 Set the timer slop (in nanoseconds) for the virtual Xen
7322 timers (default is 100000). This adjusts the minimum
7323 delta of virtualized Xen timers, where lower values
7324 improve timer resolution at the expense of processing
7325 more timer interrupts.
7327 xen.balloon_boot_timeout= [XEN]
7328 The time (in seconds) to wait before giving up to boot
7329 in case initial ballooning fails to free enough memory.
7330 Applies only when running as HVM or PVH guest and
7331 started with less memory configured than allowed at
7332 max. Default is 180.
7334 xen.event_eoi_delay= [XEN]
7335 How long to delay EOI handling in case of event
7336 storms (jiffies). Default is 10.
7338 xen.event_loop_timeout= [XEN]
7339 After which time (jiffies) the event handling loop
7340 should start to delay EOI handling. Default is 2.
7342 xen.fifo_events= [XEN]
7343 Boolean parameter to disable using fifo event handling
7344 even if available. Normally fifo event handling is
7345 preferred over the 2-level event handling, as it is
7346 fairer and the number of possible event channels is
7347 much higher. Default is on (use fifo events).
7349 xirc2ps_cs= [NET,PCMCIA]
7351 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
7354 By default on POWER9 and above, the kernel will
7355 natively use the XIVE interrupt controller. This option
7356 allows the fallback firmware mode to be used:
7358 off Fallback to firmware control of XIVE interrupt
7359 controller on both pseries and powernv
7360 platforms. Only useful on POWER9 and above.
7362 xive.store-eoi=off [PPC]
7363 By default on POWER10 and above, the kernel will use
7364 stores for EOI handling when the XIVE interrupt mode
7365 is active. This option allows the XIVE driver to use
7366 loads instead, as on POWER9.
7368 xhci-hcd.quirks [USB,KNL]
7369 A hex value specifying bitmask with supplemental xhci
7370 host controller quirks. Meaning of each bit can be
7371 consulted in header drivers/usb/host/xhci.h.
7374 Format: { early | on | rw | ro | off }
7375 Controls if xmon debugger is enabled. Default is off.
7376 Passing only "xmon" is equivalent to "xmon=early".
7377 early Call xmon as early as possible on boot; xmon
7378 debugger is called from setup_arch().
7379 on xmon debugger hooks will be installed so xmon
7380 is only called on a kernel crash. Default mode,
7381 i.e. either "ro" or "rw" mode, is controlled
7382 with CONFIG_XMON_DEFAULT_RO_MODE.
7383 rw xmon debugger hooks will be installed so xmon
7384 is called only on a kernel crash, mode is write,
7385 meaning SPR registers, memory and, other data
7386 can be written using xmon commands.
7387 ro same as "rw" option above but SPR registers,
7388 memory, and other data can't be written using
7390 off xmon is disabled.