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,EARLY]
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,EARLY]
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 [ACPI,EARLY]
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,EARLY]
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,EARLY]
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,EARLY]
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,EARLY] ACPI System Control Interrupt trigger mode
232 Format: { level | edge | high | low }
234 acpi_skip_timer_override [HW,ACPI,EARLY]
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,EARLY]
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,EARLY] 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,EARLY]
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.)
354 amd_pstate= [X86,EARLY]
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.
380 Disable amd-pstate preferred core.
382 amijoy.map= [HW,JOY] Amiga joystick support
383 Map of devices attached to JOY0DAT and JOY1DAT
385 See also Documentation/input/joydev/joystick.rst
387 analog.map= [HW,JOY] Analog joystick and gamepad support
388 Specifies type or capabilities of an analog joystick
389 connected to one of 16 gameports
390 Format: <type1>,<type2>,..<type16>
393 Power management functions (SPARCstation-4/5 + deriv.)
395 Disable APC CPU standby support. SPARCstation-Fox does
396 not play well with APC CPU idle - disable it if you have
397 APC and your system crashes randomly.
399 apic= [APIC,X86,EARLY] Advanced Programmable Interrupt Controller
400 Change the output verbosity while booting
401 Format: { quiet (default) | verbose | debug }
402 Change the amount of debugging information output
403 when initialising the APIC and IO-APIC components.
404 For X86-32, this can also be used to specify an APIC
406 Format: apic=driver_name
407 Examples: apic=bigsmp
409 apic_extnmi= [APIC,X86,EARLY] External NMI delivery setting
410 Format: { bsp (default) | all | none }
411 bsp: External NMI is delivered only to CPU 0
412 all: External NMIs are broadcast to all CPUs as a
414 none: External NMI is masked for all CPUs. This is
415 useful so that a dump capture kernel won't be
419 See Documentation/networking/ipv6.rst.
421 apm= [APM] Advanced Power Management
422 See header of arch/x86/kernel/apm_32.c.
424 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
425 Format: { "0" | "1" }
426 See security/apparmor/Kconfig help text
429 Default value is set via kernel config option.
431 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
432 Format: <io>,<irq>,<nodeID>
434 arm64.nobti [ARM64] Unconditionally disable Branch Target
435 Identification support
437 arm64.nomops [ARM64] Unconditionally disable Memory Copy and Memory
438 Set instructions support
440 arm64.nomte [ARM64] Unconditionally disable Memory Tagging Extension
443 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication
446 arm64.nosme [ARM64] Unconditionally disable Scalable Matrix
449 arm64.nosve [ARM64] Unconditionally disable Scalable Vector
454 atarimouse= [HW,MOUSE] Atari Mouse
456 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
457 EzKey and similar keyboards
459 atkbd.reset= [HW] Reset keyboard during initialization
461 atkbd.set= [HW] Select keyboard code set
462 Format: <int> (2 = AT (default), 3 = PS/2)
464 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
467 atkbd.softraw= [HW] Choose between synthetic and real raw mode
468 Format: <bool> (0 = real, 1 = synthetic (default))
470 atkbd.softrepeat= [HW]
471 Use software keyboard repeat
473 audit= [KNL] Enable the audit sub-system
474 Format: { "0" | "1" | "off" | "on" }
475 0 | off - kernel audit is disabled and can not be
476 enabled until the next reboot
477 unset - kernel audit is initialized but disabled and
478 will be fully enabled by the userspace auditd.
479 1 | on - kernel audit is initialized and partially
480 enabled, storing at most audit_backlog_limit
481 messages in RAM until it is fully enabled by the
485 audit_backlog_limit= [KNL] Set the audit queue size limit.
486 Format: <int> (must be >=0)
489 bau= [X86_UV] Enable the BAU on SGI UV. The default
490 behavior is to disable the BAU (i.e. bau=0).
491 Format: { "0" | "1" }
494 unset - Disable the BAU.
496 baycom_epp= [HW,AX25]
499 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
501 See header of drivers/net/hamradio/baycom_par.c.
503 baycom_ser_fdx= [HW,AX25]
504 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
505 Format: <io>,<irq>,<mode>[,<baud>]
506 See header of drivers/net/hamradio/baycom_ser_fdx.c.
508 baycom_ser_hdx= [HW,AX25]
509 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
510 Format: <io>,<irq>,<mode>
511 See header of drivers/net/hamradio/baycom_ser_hdx.c.
514 Disable BERT OS support on buggy BIOSes.
516 bgrt_disable [ACPI,X86,EARLY]
517 Disable BGRT to avoid flickering OEM logo.
519 blkdevparts= Manual partition parsing of block device(s) for
520 embedded devices based on command line input.
521 See Documentation/block/cmdline-partition.rst
523 boot_delay= [KNL,EARLY]
524 Milliseconds to delay each printk during boot.
525 Only works if CONFIG_BOOT_PRINTK_DELAY is enabled,
526 and you may also have to specify "lpj=". Boot_delay
527 values larger than 10 seconds (10000) are assumed
528 erroneous and ignored.
531 bootconfig [KNL,EARLY]
532 Extended command line options can be added to an initrd
533 and this will cause the kernel to look for it.
535 See Documentation/admin-guide/bootconfig.rst
537 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
538 bttv.radio= Most important insmod options are available as
540 bttv.pll= See Documentation/admin-guide/media/bttv.rst
543 bulk_remove=off [PPC] This parameter disables the use of the pSeries
544 firmware feature for flushing multiple hpte entries
547 c101= [NET] Moxa C101 synchronous serial card
549 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
550 Sometimes CPU hardware bugs make them report the cache
551 size incorrectly. The kernel will attempt work arounds
552 to fix known problems, but for some CPUs it is not
553 possible to determine what the correct size should be.
554 This option provides an override for these situations.
557 [NET] Specifies amount of time (in seconds) that
558 the kernel should wait for a network carrier. By default
559 it waits 120 seconds.
561 ca_keys= [KEYS] This parameter identifies a specific key(s) on
562 the system trusted keyring to be used for certificate
564 format: { id:<keyid> | builtin }
566 cca= [MIPS,EARLY] Override the kernel pages' cache coherency
567 algorithm. Accepted values range from 0 to 7
568 inclusive. See arch/mips/include/asm/pgtable-bits.h
569 for platform specific values (SB1, Loongson3 and
572 ccw_timeout_log [S390]
573 See Documentation/arch/s390/common_io.rst for details.
575 cgroup_disable= [KNL] Disable a particular controller or optional feature
576 Format: {name of the controller(s) or feature(s) to disable}
577 The effects of cgroup_disable=foo are:
578 - foo isn't auto-mounted if you mount all cgroups in
580 - foo isn't visible as an individually mountable
582 - if foo is an optional feature then the feature is
583 disabled and corresponding cgroup files are not
585 {Currently only "memory" controller deal with this and
586 cut the overhead, others just disable the usage. So
587 only cgroup_disable=memory is actually worthy}
588 Specifying "pressure" disables per-cgroup pressure
589 stall information accounting feature
591 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
592 Format: { { controller | "all" | "named" }
593 [,{ controller | "all" | "named" }...] }
594 Like cgroup_disable, but only applies to cgroup v1;
595 the blacklisted controllers remain available in cgroup2.
596 "all" blacklists all controllers and "named" disables
597 named mounts. Specifying both "all" and "named" disables
600 cgroup_favordynmods= [KNL] Enable or Disable favordynmods.
601 Format: { "true" | "false" }
602 Defaults to the value of CONFIG_CGROUP_FAVOR_DYNMODS.
604 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
606 nosocket -- Disable socket memory accounting.
607 nokmem -- Disable kernel memory accounting.
608 nobpf -- Disable BPF memory accounting.
610 checkreqprot= [SELINUX] Set initial checkreqprot flag value.
611 Format: { "0" | "1" }
612 See security/selinux/Kconfig help text.
613 0 -- check protection applied by kernel (includes
614 any implied execute protection).
615 1 -- check protection requested by application.
616 Default value is set via a kernel config option.
617 Value can be changed at runtime via
618 /sys/fs/selinux/checkreqprot.
619 Setting checkreqprot to 1 is deprecated.
622 See Documentation/arch/s390/common_io.rst for details.
624 clearcpuid=X[,X...] [X86]
625 Disable CPUID feature X for the kernel. See
626 arch/x86/include/asm/cpufeatures.h for the valid bit
627 numbers X. Note the Linux-specific bits are not necessarily
628 stable over kernel options, but the vendor-specific
630 X can also be a string as appearing in the flags: line
631 in /proc/cpuinfo which does not have the above
632 instability issue. However, not all features have names
634 Note that using this option will taint your kernel.
635 Also note that user programs calling CPUID directly
636 or using the feature without checking anything
637 will still see it. This just prevents it from
638 being used by the kernel or shown in /proc/cpuinfo.
639 Also note the kernel might malfunction if you disable
644 Prevents the clock framework from automatically gating
645 clocks that have not been explicitly enabled by a Linux
646 device driver but are enabled in hardware at reset or
647 by the bootloader/firmware. Note that this does not
648 force such clocks to be always-on nor does it reserve
649 those clocks in any way. This parameter is useful for
650 debug and development, but should not be needed on a
651 platform with proper driver support. For more
652 information, see Documentation/driver-api/clk.rst.
654 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
656 Forces specified clocksource (if available) to be used
657 when calculating gettimeofday(). If specified
658 clocksource is not available, it defaults to PIT.
659 Format: { pit | tsc | cyclone | pmtmr }
661 clocksource= Override the default clocksource
663 Override the default clocksource and use the clocksource
664 with the name specified.
665 Some clocksource names to choose from, depending on
667 [all] jiffies (this is the base, fallback clocksource)
669 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
670 pxa_timer,timer3,32k_counter,timer0_1
671 [X86-32] pit,hpet,tsc;
672 scx200_hrt on Geode; cyclone on IBM x440
680 clocksource.arm_arch_timer.evtstrm=
683 Enable/disable the eventstream feature of the ARM
684 architected timer so that code using WFE-based polling
685 loops can be debugged more effectively on production
688 clocksource.verify_n_cpus= [KNL]
689 Limit the number of CPUs checked for clocksources
690 marked with CLOCK_SOURCE_VERIFY_PERCPU that
691 are marked unstable due to excessive skew.
692 A negative value says to check all CPUs, while
693 zero says not to check any. Values larger than
694 nr_cpu_ids are silently truncated to nr_cpu_ids.
695 The actual CPUs are chosen randomly, with
696 no replacement if the same CPU is chosen twice.
698 clocksource-wdtest.holdoff= [KNL]
699 Set the time in seconds that the clocksource
700 watchdog test waits before commencing its tests.
701 Defaults to zero when built as a module and to
702 10 seconds when built into the kernel.
704 cma=nn[MG]@[start[MG][-end[MG]]]
706 Sets the size of kernel global memory area for
707 contiguous memory allocations and optionally the
708 placement constraint by the physical address range of
709 memory allocations. A value of 0 disables CMA
710 altogether. For more information, see
711 kernel/dma/contiguous.c
715 Sets the size of kernel per-numa memory area for
716 contiguous memory allocations. A value of 0 disables
717 per-numa CMA altogether. And If this option is not
718 specified, the default value is 0.
719 With per-numa CMA enabled, DMA users on node nid will
720 first try to allocate buffer from the pernuma area
721 which is located in node nid, if the allocation fails,
722 they will fallback to the global default memory area.
724 numa_cma=<node>:nn[MG][,<node>:nn[MG]]
726 Sets the size of kernel numa memory area for
727 contiguous memory allocations. It will reserve CMA
728 area for the specified node.
730 With numa CMA enabled, DMA users on node nid will
731 first try to allocate buffer from the numa area
732 which is located in node nid, if the allocation fails,
733 they will fallback to the global default memory area.
735 cmo_free_hint= [PPC] Format: { yes | no }
736 Specify whether pages are marked as being inactive
737 when they are freed. This is used in CMO environments
738 to determine OS memory pressure for page stealing by
742 coherent_pool=nn[KMG] [ARM,KNL,EARLY]
743 Sets the size of memory pool for coherent, atomic dma
744 allocations, by default set to 256K.
746 com20020= [HW,NET] ARCnet - COM20020 chipset
748 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
750 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
754 ARCnet - COM90xx chipset (memory-mapped buffers)
755 Format: <io>[,<irq>[,<memstart>]]
757 condev= [HW,S390] console device
760 con3215_drop= [S390,EARLY] 3215 console drop mode.
762 When set to true, drop data on the 3215 console when
763 the console buffer is full. In this case the
764 operator using a 3270 terminal emulator (for example
765 x3270) does not have to enter the clear key for the
766 console output to advance and the kernel to continue.
767 This leads to a much faster boot time when a 3270
768 terminal emulator is active. If no 3270 terminal
769 emulator is used, this parameter has no effect.
771 console= [KNL] Output console device and options.
773 tty<n> Use the virtual console device <n>.
777 Use the specified serial port. The options are of
778 the form "bbbbpnf", where "bbbb" is the baud rate,
779 "p" is parity ("n", "o", or "e"), "n" is number of
780 bits, and "f" is flow control ("r" for RTS or
781 omit it). Default is "9600n8".
783 See Documentation/admin-guide/serial-console.rst for more
785 Documentation/networking/netconsole.rst for an
788 uart[8250],io,<addr>[,options]
789 uart[8250],mmio,<addr>[,options]
790 uart[8250],mmio16,<addr>[,options]
791 uart[8250],mmio32,<addr>[,options]
792 uart[8250],0x<addr>[,options]
793 Start an early, polled-mode console on the 8250/16550
794 UART at the specified I/O port or MMIO address,
795 switching to the matching ttyS device later.
796 MMIO inter-register address stride is either 8-bit
797 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
798 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
799 to be equivalent to 'mmio'. 'options' are specified in
800 the same format described for ttyS above; if unspecified,
801 the h/w is not re-initialized.
803 hvc<n> Use the hypervisor console device <n>. This is for
804 both Xen and PowerPC hypervisors.
807 Use to disable console output, i.e., to have kernel
808 console messages discarded.
809 This must be the only console= parameter used on the
812 If the device connected to the port is not a TTY but a braille
813 device, prepend "brl," before the device type, for instance
815 For now, only VisioBraille is supported.
818 [KNL] Change console messages format
820 By default we print messages on consoles in
821 "[time stamp] text\n" format (time stamp may not be
822 printed, depending on CONFIG_PRINTK_TIME or
823 `printk_time' param).
825 Switch to syslog format: "<%u>[time stamp] text\n"
826 IOW, each message will have a facility and loglevel
827 prefix. The format is similar to one used by syslog()
828 syscall, or to executing "dmesg -S --raw" or to reading
831 consoleblank= [KNL] The console blank (screen saver) timeout in
832 seconds. A value of 0 disables the blank timer.
836 [KNL] Change the default value for
837 /proc/<pid>/coredump_filter.
838 See also Documentation/filesystems/proc.rst.
840 coresight_cpu_debug.enable
843 Enable/disable the CPU sampling based debugging.
844 0: default value, disable debugging
845 1: enable debugging at boot time
847 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
849 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
851 cpuidle.off=1 [CPU_IDLE]
852 disable the cpuidle sub-system
855 [CPU_IDLE] Name of the cpuidle governor to use.
857 cpufreq.off=1 [CPU_FREQ]
858 disable the cpufreq sub-system
860 cpufreq.default_governor=
861 [CPU_FREQ] Name of the default cpufreq governor or
862 policy to use. This governor must be registered in the
863 kernel before the cpufreq driver probes.
866 [X86,EARLY] Delay for N microsec between assert and de-assert
867 of APIC INIT to start processors. This delay occurs
868 on every CPU online, such as boot, and resume from suspend.
872 [SMP] Enable/disable parallel bringup of secondary CPUs
874 Default is enabled if CONFIG_HOTPLUG_PARALLEL=y. Otherwise
875 the parameter has no effect.
877 crash_kexec_post_notifiers
878 Run kdump after running panic-notifiers and dumping
879 kmsg. This only for the users who doubt kdump always
880 succeeds in any situation.
881 Note that this also increases risks of kdump failure,
882 because some panic notifiers can make the crashed
883 kernel more unstable.
885 crashkernel=size[KMG][@offset[KMG]]
886 [KNL,EARLY] Using kexec, Linux can switch to a 'crash kernel'
887 upon panic. This parameter reserves the physical
888 memory region [offset, offset + size] for that kernel
889 image. If '@offset' is omitted, then a suitable offset
890 is selected automatically.
891 [KNL, X86-64, ARM64, RISCV, LoongArch] Select a region
892 under 4G first, and fall back to reserve region above
893 4G when '@offset' hasn't been specified.
894 See Documentation/admin-guide/kdump/kdump.rst for further details.
896 crashkernel=range1:size1[,range2:size2,...][@offset]
897 [KNL] Same as above, but depends on the memory
898 in the running system. The syntax of range is
899 start-[end] where start and end are both
900 a memory unit (amount[KMG]). See also
901 Documentation/admin-guide/kdump/kdump.rst for an example.
903 crashkernel=size[KMG],high
904 [KNL, X86-64, ARM64, RISCV, LoongArch] range could be
906 Allow kernel to allocate physical memory region from top,
907 so could be above 4G if system have more than 4G ram
908 installed. Otherwise memory region will be allocated
909 below 4G, if available.
910 It will be ignored if crashkernel=X is specified.
911 crashkernel=size[KMG],low
912 [KNL, X86-64, ARM64, RISCV, LoongArch] range under 4G.
913 When crashkernel=X,high is passed, kernel could allocate
914 physical memory region above 4G, that cause second kernel
915 crash on system that require some amount of low memory,
916 e.g. swiotlb requires at least 64M+32K low memory, also
917 enough extra low memory is needed to make sure DMA buffers
918 for 32-bit devices won't run out. Kernel would try to allocate
919 default size of memory below 4G automatically. The default
920 size is platform dependent.
921 --> x86: max(swiotlb_size_or_default() + 8MiB, 256MiB)
924 --> loongarch: 128MiB
925 This one lets the user specify own low range under 4G
926 for second kernel instead.
927 0: to disable low allocation.
928 It will be ignored when crashkernel=X,high is not used
929 or memory reserved is below 4G.
932 [KNL] Disable crypto self-tests
937 cs89x0_media= [HW,NET]
938 Format: { rj45 | aui | bnc }
940 csdlock_debug= [KNL] Enable or disable debug add-ons of cross-CPU
941 function call handling. When switched on,
942 additional debug data is printed to the console
943 in case a hanging CPU is detected, and that
944 CPU is pinged again in order to try to resolve
945 the hang situation. The default value of this
946 option depends on the CSD_LOCK_WAIT_DEBUG_DEFAULT
950 See header of drivers/s390/block/dasd_devmap.c.
952 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
953 (one device per port)
954 Format: <port#>,<type>
955 See also Documentation/input/devices/joystick-parport.rst
957 debug [KNL,EARLY] Enable kernel debugging (events log level).
960 [KNL,EARLY] Enable printing [hashed] pointers early in the
961 boot sequence. If enabled, we use a weak hash instead
962 of siphash to hash pointers. Use this option if you are
963 seeing instances of '(___ptrval___)') and need to see a
964 value (hashed pointer) instead. Cryptographically
965 insecure, please do not use on production kernels.
968 [KNL] verbose locking self-tests
970 Print debugging info while doing the locking API
972 Bitmask for the various LOCKTYPE_ tests. Defaults to 0
973 (no extra messages), setting it to -1 (all bits set)
974 will print _a_lot_ more information - normally only
975 useful to lockdep developers.
977 debug_objects [KNL,EARLY] Enable object debugging
979 debug_guardpage_minorder=
980 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this
981 parameter allows control of the order of pages that will
982 be intentionally kept free (and hence protected) by the
983 buddy allocator. Bigger value increase the probability
984 of catching random memory corruption, but reduce the
985 amount of memory for normal system use. The maximum
986 possible value is MAX_PAGE_ORDER/2. Setting this
987 parameter to 1 or 2 should be enough to identify most
988 random memory corruption problems caused by bugs in
989 kernel or driver code when a CPU writes to (or reads
990 from) a random memory location. Note that there exists
991 a class of memory corruptions problems caused by buggy
992 H/W or F/W or by drivers badly programming DMA
993 (basically when memory is written at bus level and the
994 CPU MMU is bypassed) which are not detectable by
995 CONFIG_DEBUG_PAGEALLOC, hence this option will not
996 help tracking down these problems.
999 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
1000 enables the feature at boot time. By default, it is
1001 disabled and the system will work mostly the same as a
1002 kernel built without CONFIG_DEBUG_PAGEALLOC.
1003 Note: to get most of debug_pagealloc error reports, it's
1004 useful to also enable the page_owner functionality.
1005 on: enable the feature
1007 debugfs= [KNL,EARLY] This parameter enables what is exposed to
1008 userspace and debugfs internal clients.
1009 Format: { on, no-mount, off }
1010 on: All functions are enabled.
1012 Filesystem is not registered but kernel clients can
1013 access APIs and a crashkernel can be used to read
1014 its content. There is nothing to mount.
1015 off: Filesystem is not registered and clients
1016 get a -EPERM as result when trying to register files
1017 or directories within debugfs.
1018 This is equivalent of the runtime functionality if
1019 debugfs was not enabled in the kernel at all.
1020 Default value is set in build-time with a kernel configuration.
1022 debugpat [X86] Enable PAT debugging
1025 [HW] The size of the default HugeTLB page. This is
1026 the size represented by the legacy /proc/ hugepages
1027 APIs. In addition, this is the default hugetlb size
1028 used for shmget(), mmap() and mounting hugetlbfs
1029 filesystems. If not specified, defaults to the
1030 architecture's default huge page size. Huge page
1031 sizes are architecture dependent. See also
1032 Documentation/admin-guide/mm/hugetlbpage.rst.
1035 deferred_probe_timeout=
1036 [KNL] Debugging option to set a timeout in seconds for
1037 deferred probe to give up waiting on dependencies to
1038 probe. Only specific dependencies (subsystems or
1039 drivers) that have opted in will be ignored. A timeout
1040 of 0 will timeout at the end of initcalls. If the time
1041 out hasn't expired, it'll be restarted by each
1042 successful driver registration. This option will also
1043 dump out devices still on the deferred probe list after
1046 delayacct [KNL] Enable per-task delay accounting
1048 dell_smm_hwmon.ignore_dmi=
1049 [HW] Continue probing hardware even if DMI data
1050 indicates that the driver is running on unsupported
1053 dell_smm_hwmon.force=
1054 [HW] Activate driver even if SMM BIOS signature does
1055 not match list of supported models and enable otherwise
1056 blacklisted features.
1058 dell_smm_hwmon.power_status=
1059 [HW] Report power status in /proc/i8k
1060 (disabled by default).
1062 dell_smm_hwmon.restricted=
1063 [HW] Allow controlling fans only if SYS_ADMIN
1066 dell_smm_hwmon.fan_mult=
1067 [HW] Factor to multiply fan speed with.
1069 dell_smm_hwmon.fan_max=
1070 [HW] Maximum configurable fan speed.
1073 Format: { on | off | def_only | inf_only | always }
1074 on: s390 zlib hardware support for compression on
1075 level 1 and decompression (default)
1076 off: No s390 zlib hardware support
1077 def_only: s390 zlib hardware support for deflate
1078 only (compression on level 1)
1079 inf_only: s390 zlib hardware support for inflate
1080 only (decompression)
1081 always: Same as 'on' but ignores the selected compression
1082 level always using hardware support (used for debugging)
1084 dhash_entries= [KNL]
1085 Set number of hash buckets for dentry cache.
1087 disable_1tb_segments [PPC,EARLY]
1088 Disables the use of 1TB hash page table segments. This
1089 causes the kernel to fall back to 256MB segments which
1090 can be useful when debugging issues that require an SLB
1094 See Documentation/networking/ipv6.rst.
1096 disable_radix [PPC,EARLY]
1097 Disable RADIX MMU mode on POWER9
1100 Disable TLBIE instruction. Currently does not work
1101 with KVM, with HASH MMU, or with coherent accelerators.
1103 disable_ddw [PPC/PSERIES,EARLY]
1104 Disable Dynamic DMA Window support. Use this
1105 to workaround buggy firmware.
1107 disable_ipv6= [IPV6]
1108 See Documentation/networking/ipv6.rst.
1110 disable_mtrr_cleanup [X86,EARLY]
1111 The kernel tries to adjust MTRR layout from continuous
1112 to discrete, to make X server driver able to add WB
1113 entry later. This parameter disables that.
1115 disable_mtrr_trim [X86, Intel and AMD only,EARLY]
1116 By default the kernel will trim any uncacheable
1117 memory out of your available memory pool based on
1118 MTRR settings. This parameter disables that behavior,
1119 possibly causing your machine to run very slowly.
1121 disable_timer_pin_1 [X86,EARLY]
1122 Disable PIN 1 of APIC timer
1123 Can be useful to work around chipset bugs.
1125 dis_ucode_ldr [X86] Disable the microcode loader.
1127 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
1128 this option disables the debugging code at boot.
1130 dma_debug_entries=<number>
1131 This option allows to tune the number of preallocated
1132 entries for DMA-API debugging code. One entry is
1133 required per DMA-API allocation. Use this if the
1134 DMA-API debugging code disables itself because the
1135 architectural default is too low.
1137 dma_debug_driver=<driver_name>
1138 With this option the DMA-API debugging driver
1139 filter feature can be enabled at boot time. Just
1140 pass the driver to filter for as the parameter.
1141 The filter can be disabled or changed to another
1142 driver later using sysfs.
1144 reg_file_data_sampling=
1145 [X86] Controls mitigation for Register File Data
1146 Sampling (RFDS) vulnerability. RFDS is a CPU
1147 vulnerability which may allow userspace to infer
1148 kernel data values previously stored in floating point
1149 registers, vector registers, or integer registers.
1150 RFDS only affects Intel Atom processors.
1152 on: Turns ON the mitigation.
1153 off: Turns OFF the mitigation.
1155 This parameter overrides the compile time default set
1156 by CONFIG_MITIGATION_RFDS. Mitigation cannot be
1157 disabled when other VERW based mitigations (like MDS)
1158 are enabled. In order to disable RFDS mitigation all
1159 VERW based mitigations need to be disabled.
1162 Documentation/admin-guide/hw-vuln/reg-file-data-sampling.rst
1164 driver_async_probe= [KNL]
1165 List of driver names to be probed asynchronously. *
1166 matches with all driver names. If * is specified, the
1167 rest of the listed driver names are those that will NOT
1169 Format: <driver_name1>,<driver_name2>...
1171 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
1172 Broken monitors, graphic adapters, KVMs and EDIDless
1173 panels may send no or incorrect EDID data sets.
1174 This parameter allows to specify an EDID data sets
1175 in the /lib/firmware directory that are used instead.
1176 An EDID data set will only be used for a particular
1177 connector, if its name and a colon are prepended to
1178 the EDID name. Each connector may use a unique EDID
1179 data set by separating the files with a comma. An EDID
1180 data set with no connector name will be used for
1181 any connectors not explicitly specified.
1185 dt_cpu_ftrs= [PPC,EARLY]
1186 Format: {"off" | "known"}
1187 Control how the dt_cpu_ftrs device-tree binding is
1188 used for CPU feature discovery and setup (if it
1190 off: Do not use it, fall back to legacy cpu table.
1191 known: Do not pass through unknown features to guests
1192 or userspace, only those that the kernel is aware of.
1194 dump_apple_properties [X86]
1195 Dump name and content of EFI device properties on
1196 x86 Macs. Useful for driver authors to determine
1197 what data is available or for reverse-engineering.
1199 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
1200 <module>.dyndbg[="val"]
1201 Enable debug messages at boot time. See
1202 Documentation/admin-guide/dynamic-debug-howto.rst
1205 early_ioremap_debug [KNL,EARLY]
1206 Enable debug messages in early_ioremap support. This
1207 is useful for tracking down temporary early mappings
1208 which are not unmapped.
1210 earlycon= [KNL,EARLY] Output early console device and options.
1212 When used with no options, the early console is
1213 determined by stdout-path property in device tree's
1214 chosen node or the ACPI SPCR table if supported by
1217 cdns,<addr>[,options]
1218 Start an early, polled-mode console on a Cadence
1219 (xuartps) serial port at the specified address. Only
1220 supported option is baud rate. If baud rate is not
1221 specified, the serial port must already be setup and
1224 uart[8250],io,<addr>[,options[,uartclk]]
1225 uart[8250],mmio,<addr>[,options[,uartclk]]
1226 uart[8250],mmio32,<addr>[,options[,uartclk]]
1227 uart[8250],mmio32be,<addr>[,options[,uartclk]]
1228 uart[8250],0x<addr>[,options]
1229 Start an early, polled-mode console on the 8250/16550
1230 UART at the specified I/O port or MMIO address.
1231 MMIO inter-register address stride is either 8-bit
1232 (mmio) or 32-bit (mmio32 or mmio32be).
1233 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1234 to be equivalent to 'mmio'. 'options' are specified
1235 in the same format described for "console=ttyS<n>"; if
1236 unspecified, the h/w is not initialized. 'uartclk' is
1237 the uart clock frequency; if unspecified, it is set
1238 to 'BASE_BAUD' * 16.
1242 Start an early, polled-mode console on a pl011 serial
1243 port at the specified address. The pl011 serial port
1244 must already be setup and configured. Options are not
1245 yet supported. If 'mmio32' is specified, then only
1246 the driver will use only 32-bit accessors to read/write
1247 the device registers.
1250 Start an early console on a litex serial port at the
1251 specified address. The serial port must already be
1252 setup and configured. Options are not yet supported.
1255 Start an early, polled-mode console on a meson serial
1256 port at the specified address. The serial port must
1257 already be setup and configured. Options are not yet
1261 Start an early, polled-mode console on an msm serial
1262 port at the specified address. The serial port
1263 must already be setup and configured. Options are not
1266 msm_serial_dm,<addr>
1267 Start an early, polled-mode console on an msm serial
1268 dm port at the specified address. The serial port
1269 must already be setup and configured. Options are not
1273 Start an early, polled-mode console on a serial port
1274 of an Actions Semi SoC, such as S500 or S900, at the
1275 specified address. The serial port must already be
1276 setup and configured. Options are not yet supported.
1279 Start an early, polled-mode console on a serial port
1280 of an RDA Micro SoC, such as RDA8810PL, at the
1281 specified address. The serial port must already be
1282 setup and configured. Options are not yet supported.
1285 Use RISC-V SBI (Supervisor Binary Interface) for early
1288 smh Use ARM semihosting calls for early console.
1296 Use early console provided by serial driver available
1297 on Samsung SoCs, requires selecting proper type and
1298 a correct base address of the selected UART port. The
1299 serial port must already be setup and configured.
1300 Options are not yet supported.
1303 Start an early, polled-mode console on a lantiq serial
1304 (lqasc) port at the specified address. The serial port
1305 must already be setup and configured. Options are not
1310 Use early console provided by Freescale LP UART driver
1311 found on Freescale Vybrid and QorIQ LS1021A processors.
1312 A valid base address must be provided, and the serial
1313 port must already be setup and configured.
1317 Start an early, polled-mode, output-only console on the
1318 Freescale i.MX UART at the specified address. The UART
1319 must already be setup and configured.
1322 Start an early, polled-mode console on the
1323 Armada 3700 serial port at the specified
1324 address. The serial port must already be setup
1325 and configured. Options are not yet supported.
1328 Start an early, polled-mode console on a Qualcomm
1329 Generic Interface (GENI) based serial port at the
1330 specified address. The serial port must already be
1331 setup and configured. Options are not yet supported.
1334 Start an early, unaccelerated console on the EFI
1335 memory mapped framebuffer (if available). On cache
1336 coherent non-x86 systems that use system memory for
1337 the framebuffer, pass the 'ram' option so that it is
1338 mapped with the correct attributes.
1341 Use early console provided by Freescale LINFlexD UART
1342 serial driver for NXP S32V234 SoCs. A valid base
1343 address must be provided, and the serial port must
1344 already be setup and configured.
1346 earlyprintk= [X86,SH,ARM,M68k,S390,UM,EARLY]
1350 earlyprintk=serial[,ttySn[,baudrate]]
1351 earlyprintk=serial[,0x...[,baudrate]]
1352 earlyprintk=ttySn[,baudrate]
1353 earlyprintk=dbgp[debugController#]
1354 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1355 earlyprintk=xdbc[xhciController#]
1358 earlyprintk is useful when the kernel crashes before
1359 the normal console is initialized. It is not enabled by
1360 default because it has some cosmetic problems.
1362 Append ",keep" to not disable it when the real console
1365 Only one of vga, serial, or usb debug port can
1368 Currently only ttyS0 and ttyS1 may be specified by
1369 name. Other I/O ports may be explicitly specified
1370 on some architectures (x86 and arm at least) by
1371 replacing ttySn with an I/O port address, like this:
1372 earlyprintk=serial,0x1008,115200
1373 You can find the port for a given device in
1374 /proc/tty/driver/serial:
1375 2: uart:ST16650V2 port:00001008 irq:18 ...
1377 Interaction with the standard serial driver is not
1380 The VGA output is eventually overwritten by
1383 The xen option can only be used in Xen domains.
1385 The sclp output can only be used on s390.
1387 The bios output can only be used on SuperH.
1389 The optional "force" to "pciserial" enables use of a
1390 PCI device even when its classcode is not of the
1393 edac_report= [HW,EDAC] Control how to report EDAC event
1394 Format: {"on" | "off" | "force"}
1395 on: enable EDAC to report H/W event. May be overridden
1396 by other higher priority error reporting module.
1397 off: disable H/W event reporting through EDAC.
1398 force: enforce the use of EDAC to report H/W event.
1402 Format: {"off" | "on" | "skip[mbr]"}
1405 Format: { "debug", "disable_early_pci_dma",
1406 "nochunk", "noruntime", "nosoftreserve",
1407 "novamap", "no_disable_early_pci_dma" }
1408 debug: enable misc debug output.
1409 disable_early_pci_dma: disable the busmaster bit on all
1410 PCI bridges while in the EFI boot stub.
1411 nochunk: disable reading files in "chunks" in the EFI
1412 boot stub, as chunking can cause problems with some
1413 firmware implementations.
1414 noruntime : disable EFI runtime services support
1415 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1416 attribute may cause the kernel to reserve the
1417 memory range for a memory mapping driver to
1418 claim. Specify efi=nosoftreserve to disable this
1419 reservation and treat the memory by its base type
1420 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1421 novamap: do not call SetVirtualAddressMap().
1422 no_disable_early_pci_dma: Leave the busmaster bit set
1423 on all PCI bridges while in the EFI boot stub
1425 efi_no_storage_paranoia [EFI,X86,EARLY]
1426 Using this parameter you can use more than 50% of
1427 your efi variable storage. Use this parameter only if
1428 you are really sure that your UEFI does sane gc and
1429 fulfills the spec otherwise your board may brick.
1431 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI,X86,EARLY]
1432 Add arbitrary attribute to specific memory range by
1433 updating original EFI memory map.
1434 Region of memory which aa attribute is added to is
1437 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1438 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1439 attribute is added to range 0x100000000-0x180000000 and
1440 0x10a0000000-0x1120000000.
1442 If efi_fake_mem=8G@9G:0x40000 is specified, the
1443 EFI_MEMORY_SP(0x40000) attribute is added to
1444 range 0x240000000-0x43fffffff.
1446 Using this parameter you can do debugging of EFI memmap
1447 related features. For example, you can do debugging of
1448 Address Range Mirroring feature even if your box
1449 doesn't support it, or mark specific memory as
1452 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1453 that is to be dynamically loaded by Linux. If there are
1454 multiple variables with the same name but with different
1455 vendor GUIDs, all of them will be loaded. See
1456 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1459 eisa_irq_edge= [PARISC,HW]
1460 See header of drivers/parisc/eisa.c.
1462 ekgdboc= [X86,KGDB,EARLY] Allow early kernel console debugging
1465 This is designed to be used in conjunction with
1466 the boot argument: earlyprintk=vga
1468 This parameter works in place of the kgdboc parameter
1469 but can only be used if the backing tty is available
1470 very early in the boot process. For early debugging
1471 via a serial port see kgdboc_earlycon instead.
1474 See comment before function elanfreq_setup() in
1475 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1477 elfcorehdr=[size[KMG]@]offset[KMG] [PPC,SH,X86,S390,EARLY]
1478 Specifies physical address of start of kernel core
1479 image elf header and optionally the size. Generally
1480 kexec loader will pass this option to capture kernel.
1481 See Documentation/admin-guide/kdump/kdump.rst for details.
1483 enable_mtrr_cleanup [X86,EARLY]
1484 The kernel tries to adjust MTRR layout from continuous
1485 to discrete, to make X server driver able to add WB
1486 entry later. This parameter enables that.
1488 enable_timer_pin_1 [X86]
1489 Enable PIN 1 of APIC timer
1490 Can be useful to work around chipset bugs
1491 (in particular on some ATI chipsets).
1492 The kernel tries to set a reasonable default.
1494 enforcing= [SELINUX] Set initial enforcing status.
1496 See security/selinux/Kconfig help text.
1497 0 -- permissive (log only, no denials).
1498 1 -- enforcing (deny and log).
1500 Value can be changed at runtime via
1501 /sys/fs/selinux/enforce.
1504 Disable Error Record Serialization Table (ERST)
1507 ether= [HW,NET] Ethernet cards parameters
1508 This option is obsoleted by the "netdev=" option, which
1509 has equivalent usage. See its documentation for details.
1513 Permit 'security.evm' to be updated regardless of
1514 current integrity status.
1516 early_page_ext [KNL,EARLY] Enforces page_ext initialization to earlier
1517 stages so cover more early boot allocations.
1518 Please note that as side effect some optimizations
1519 might be disabled to achieve that (e.g. parallelized
1520 memory initialization is disabled) so the boot process
1521 might take longer, especially on systems with a lot of
1522 memory. Available with CONFIG_PAGE_EXTENSION=y.
1527 fail_make_request=[KNL]
1528 General fault injection mechanism.
1529 Format: <interval>,<probability>,<space>,<times>
1530 See also Documentation/fault-injection/.
1533 Format: { initns | none }
1534 See Documentation/admin-guide/sysctl/net.rst for
1535 fb_tunnels_only_for_init_ns
1538 See Documentation/admin-guide/blockdev/floppy.rst.
1541 Forcefully enable Physical Address Extension (PAE).
1542 Many Pentium M systems disable PAE but may have a
1543 functionally usable PAE implementation.
1544 Warning: use of this parameter will taint the kernel
1545 and may cause unknown problems.
1548 Enable/disable Flexible Return and Event Delivery.
1549 Format: { on | off }
1550 on: enable FRED when it's present.
1551 off: disable FRED, the default setting.
1554 [FTRACE] will set and start the specified tracer
1555 as early as possible in order to facilitate early
1558 ftrace_boot_snapshot
1559 [FTRACE] On boot up, a snapshot will be taken of the
1560 ftrace ring buffer that can be read at:
1561 /sys/kernel/tracing/snapshot.
1562 This is useful if you need tracing information from kernel
1563 boot up that is likely to be overridden by user space
1564 start up functionality.
1566 Optionally, the snapshot can also be defined for a tracing
1567 instance that was created by the trace_instance= command
1570 trace_instance=foo,sched_switch ftrace_boot_snapshot=foo
1572 The above will cause the "foo" tracing instance to trigger
1573 a snapshot at the end of boot up.
1575 ftrace_dump_on_oops[=orig_cpu]
1576 [FTRACE] will dump the trace buffers on oops.
1577 If no parameter is passed, ftrace will dump
1578 buffers of all CPUs, but if you pass orig_cpu, it will
1579 dump only the buffer of the CPU that triggered the
1582 ftrace_filter=[function-list]
1583 [FTRACE] Limit the functions traced by the function
1584 tracer at boot up. function-list is a comma-separated
1585 list of functions. This list can be changed at run
1586 time by the set_ftrace_filter file in the debugfs
1589 ftrace_notrace=[function-list]
1590 [FTRACE] Do not trace the functions specified in
1591 function-list. This list can be changed at run time
1592 by the set_ftrace_notrace file in the debugfs
1595 ftrace_graph_filter=[function-list]
1596 [FTRACE] Limit the top level callers functions traced
1597 by the function graph tracer at boot up.
1598 function-list is a comma-separated list of functions
1599 that can be changed at run time by the
1600 set_graph_function file in the debugfs tracing directory.
1602 ftrace_graph_notrace=[function-list]
1603 [FTRACE] Do not trace from the functions specified in
1604 function-list. This list is a comma-separated list of
1605 functions that can be changed at run time by the
1606 set_graph_notrace file in the debugfs tracing directory.
1608 ftrace_graph_max_depth=<uint>
1609 [FTRACE] Used with the function graph tracer. This is
1610 the max depth it will trace into a function. This value
1611 can be changed at run time by the max_graph_depth file
1612 in the tracefs tracing directory. default: 0 (no limit)
1614 fw_devlink= [KNL,EARLY] Create device links between consumer and supplier
1615 devices by scanning the firmware to infer the
1616 consumer/supplier relationships. This feature is
1617 especially useful when drivers are loaded as modules as
1618 it ensures proper ordering of tasks like device probing
1619 (suppliers first, then consumers), supplier boot state
1620 clean up (only after all consumers have probed),
1621 suspend/resume & runtime PM (consumers first, then
1623 Format: { off | permissive | on | rpm }
1624 off -- Don't create device links from firmware info.
1625 permissive -- Create device links from firmware info
1626 but use it only for ordering boot state clean
1627 up (sync_state() calls).
1628 on -- Create device links from firmware info and use it
1629 to enforce probe and suspend/resume ordering.
1630 rpm -- Like "on", but also use to order runtime PM.
1632 fw_devlink.strict=<bool>
1633 [KNL,EARLY] Treat all inferred dependencies as mandatory
1634 dependencies. This only applies for fw_devlink=on|rpm.
1637 fw_devlink.sync_state =
1638 [KNL,EARLY] When all devices that could probe have finished
1639 probing, this parameter controls what to do with
1640 devices that haven't yet received their sync_state()
1642 Format: { strict | timeout }
1643 strict -- Default. Continue waiting on consumers to
1645 timeout -- Give up waiting on consumers and call
1646 sync_state() on any devices that haven't yet
1647 received their sync_state() calls after
1648 deferred_probe_timeout has expired or by
1649 late_initcall() if !CONFIG_MODULES.
1652 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1653 support via parallel port (up to 5 devices per port)
1654 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1655 See also Documentation/input/devices/joystick-parport.rst
1659 gart_fix_e820= [X86-64,EARLY] disable the fix e820 for K8 GART
1663 gather_data_sampling=
1664 [X86,INTEL,EARLY] Control the Gather Data Sampling (GDS)
1667 Gather Data Sampling is a hardware vulnerability which
1668 allows unprivileged speculative access to data which was
1669 previously stored in vector registers.
1671 This issue is mitigated by default in updated microcode.
1672 The mitigation may have a performance impact but can be
1673 disabled. On systems without the microcode mitigation
1674 disabling AVX serves as a mitigation.
1676 force: Disable AVX to mitigate systems without
1677 microcode mitigation. No effect if the microcode
1678 mitigation is present. Known to cause crashes in
1679 userspace with buggy AVX enumeration.
1681 off: Disable GDS mitigation.
1683 gcov_persist= [GCOV] When non-zero (default), profiling data for
1684 kernel modules is saved and remains accessible via
1685 debugfs, even when the module is unloaded/reloaded.
1686 When zero, profiling data is discarded and associated
1687 debugfs files are removed at module unload time.
1689 goldfish [X86] Enable the goldfish android emulator platform.
1690 Don't use this when you are not running on the
1693 gpio-mockup.gpio_mockup_ranges
1694 [HW] Sets the ranges of gpiochip of for this device.
1695 Format: <start1>,<end1>,<start2>,<end2>...
1696 gpio-mockup.gpio_mockup_named_lines
1697 [HW] Let the driver know GPIO lines should be named.
1699 gpt [EFI] Forces disk with valid GPT signature but
1700 invalid Protective MBR to be treated as GPT. If the
1701 primary GPT is corrupted, it enables the backup/alternate
1702 GPT to be used instead.
1704 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1705 the "Enable 0" bit of the configuration register.
1708 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1709 the "Enable 0" bit of the configuration register.
1712 grcan.select= [HW] Select which physical interface to use.
1715 grcan.txsize= [HW] Sets the size of the tx buffer.
1716 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1718 grcan.rxsize= [HW] Sets the size of the rx buffer.
1719 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1723 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
1724 hardening is enabled for this boot. Hardened
1725 usercopy checking is used to protect the kernel
1726 from reading or writing beyond known memory
1727 allocation boundaries as a proactive defense
1728 against bounds-checking flaws in the kernel's
1729 copy_to_user()/copy_from_user() interface.
1730 on Perform hardened usercopy checks (default).
1731 off Disable hardened usercopy checks.
1733 hardlockup_all_cpu_backtrace=
1734 [KNL] Should the hard-lockup detector generate
1735 backtraces on all cpus.
1738 hashdist= [KNL,NUMA] Large hashes allocated during boot
1739 are distributed across NUMA nodes. Defaults on
1740 for 64-bit NUMA, off otherwise.
1741 Format: 0 | 1 (for off | on)
1743 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1745 hd= [EIDE] (E)IDE hard drive subsystem geometry
1746 Format: <cyl>,<head>,<sect>
1749 Disable Hardware Error Source Table (HEST) support;
1750 corresponding firmware-first mode error processing
1751 logic will be disabled.
1753 hibernate= [HIBERNATION]
1754 noresume Don't check if there's a hibernation image
1755 present during boot.
1756 nocompress Don't compress/decompress hibernation images.
1757 no Disable hibernation and resume.
1758 protect_image Turn on image protection during restoration
1759 (that will set all pages holding image data
1760 during restoration read-only).
1762 hibernate.compressor= [HIBERNATION] Compression algorithm to be
1763 used with hibernation.
1764 Format: { lzo | lz4 }
1767 lzo: Select LZO compression algorithm to
1768 compress/decompress hibernation image.
1770 lz4: Select LZ4 compression algorithm to
1771 compress/decompress hibernation image.
1773 highmem=nn[KMG] [KNL,BOOT,EARLY] forces the highmem zone to have an exact
1774 size of <nn>. This works even on boxes that have no
1775 highmem otherwise. This also works to reduce highmem
1776 size on bigger boxes.
1778 highres= [KNL] Enable/disable high resolution timer mode.
1779 Valid parameters: "on", "off"
1784 hostname= [KNL,EARLY] Set the hostname (aka UTS nodename).
1786 This allows setting the system's hostname during early
1787 startup. This sets the name returned by gethostname.
1788 Using this parameter to set the hostname makes it
1789 possible to ensure the hostname is correctly set before
1790 any userspace processes run, avoiding the possibility
1791 that a process may call gethostname before the hostname
1792 has been explicitly set, resulting in the calling
1793 process getting an incorrect result. The string must
1794 not exceed the maximum allowed hostname length (usually
1795 64 characters) and will be truncated otherwise.
1797 hpet= [X86-32,HPET] option to control HPET usage
1798 Format: { enable (default) | disable | force |
1800 disable: disable HPET and use PIT instead
1801 force: allow force enabled of undocumented chips (ICH4,
1803 verbose: show contents of HPET registers during setup
1805 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1806 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1808 hugepages= [HW] Number of HugeTLB pages to allocate at boot.
1809 If this follows hugepagesz (below), it specifies
1810 the number of pages of hugepagesz to be allocated.
1811 If this is the first HugeTLB parameter on the command
1812 line, it specifies the number of pages to allocate for
1813 the default huge page size. If using node format, the
1814 number of pages to allocate per-node can be specified.
1815 See also Documentation/admin-guide/mm/hugetlbpage.rst.
1816 Format: <integer> or (node format)
1817 <node>:<integer>[,<node>:<integer>]
1820 [HW] The size of the HugeTLB pages. This is used in
1821 conjunction with hugepages (above) to allocate huge
1822 pages of a specific size at boot. The pair
1823 hugepagesz=X hugepages=Y can be specified once for
1824 each supported huge page size. Huge page sizes are
1825 architecture dependent. See also
1826 Documentation/admin-guide/mm/hugetlbpage.rst.
1829 hugetlb_cma= [HW,CMA,EARLY] The size of a CMA area used for allocation
1830 of gigantic hugepages. Or using node format, the size
1831 of a CMA area per node can be specified.
1832 Format: nn[KMGTPE] or (node format)
1833 <node>:nn[KMGTPE][,<node>:nn[KMGTPE]]
1835 Reserve a CMA area of given size and allocate gigantic
1836 hugepages using the CMA allocator. If enabled, the
1837 boot-time allocation of gigantic hugepages is skipped.
1839 hugetlb_free_vmemmap=
1840 [KNL] Requires CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP
1842 Control if HugeTLB Vmemmap Optimization (HVO) is enabled.
1843 Allows heavy hugetlb users to free up some more
1844 memory (7 * PAGE_SIZE for each 2MB hugetlb page).
1845 Format: { on | off (default) }
1850 Built with CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP_DEFAULT_ON=y,
1853 Note that the vmemmap pages may be allocated from the added
1854 memory block itself when memory_hotplug.memmap_on_memory is
1855 enabled, those vmemmap pages cannot be optimized even if this
1856 feature is enabled. Other vmemmap pages not allocated from
1857 the added memory block itself do not be affected.
1860 [KNL] Should the hung task detector generate panics.
1863 A value of 1 instructs the kernel to panic when a
1864 hung task is detected. The default value is controlled
1865 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1866 option. The value selected by this boot parameter can
1867 be changed later by the kernel.hung_task_panic sysctl.
1869 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1870 terminal devices. Valid values: 0..8
1871 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1872 If specified, z/VM IUCV HVC accepts connections
1873 from listed z/VM user IDs only.
1875 hv_nopvspin [X86,HYPER_V,EARLY]
1876 Disables the paravirt spinlock optimizations
1877 which allow the hypervisor to 'idle' the guest
1880 i2c_bus= [HW] Override the default board specific I2C bus speed
1881 or register an additional I2C bus that is not
1882 registered from board initialization code.
1886 i8042.debug [HW] Toggle i8042 debug mode
1887 i8042.unmask_kbd_data
1888 [HW] Enable printing of interrupt data from the KBD port
1889 (disabled by default, and as a pre-condition
1890 requires that i8042.debug=1 be enabled)
1891 i8042.direct [HW] Put keyboard port into non-translated mode
1892 i8042.dumbkbd [HW] Pretend that controller can only read data from
1893 keyboard and cannot control its state
1894 (Don't attempt to blink the leds)
1895 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1896 i8042.nokbd [HW] Don't check/create keyboard port
1897 i8042.noloop [HW] Disable the AUX Loopback command while probing
1899 i8042.nomux [HW] Don't check presence of an active multiplexing
1901 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1903 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1904 i8042.reset [HW] Reset the controller during init, cleanup and
1905 suspend-to-ram transitions, only during s2r
1906 transitions, or never reset
1907 Format: { 1 | Y | y | 0 | N | n }
1908 1, Y, y: always reset controller
1909 0, N, n: don't ever reset controller
1910 Default: only on s2r transitions on x86; most other
1911 architectures force reset to be always executed
1912 i8042.unlock [HW] Unlock (ignore) the keylock
1913 i8042.kbdreset [HW] Reset device connected to KBD port
1915 [HW] Allow deferred probing upon i8042 probe errors
1919 i915.invert_brightness=
1920 [DRM] Invert the sense of the variable that is used to
1921 set the brightness of the panel backlight. Normally a
1922 brightness value of 0 indicates backlight switched off,
1923 and the maximum of the brightness value sets the backlight
1924 to maximum brightness. If this parameter is set to 0
1925 (default) and the machine requires it, or this parameter
1926 is set to 1, a brightness value of 0 sets the backlight
1927 to maximum brightness, and the maximum of the brightness
1928 value switches the backlight off.
1929 -1 -- never invert brightness
1930 0 -- machine default
1931 1 -- force brightness inversion
1933 ia32_emulation= [X86-64]
1935 When true, allows loading 32-bit programs and executing 32-bit
1936 syscalls, essentially overriding IA32_EMULATION_DEFAULT_DISABLED at
1937 boot time. When false, unconditionally disables IA32 emulation.
1940 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1944 Format: idle=poll, idle=halt, idle=nomwait
1945 Poll forces a polling idle loop that can slightly
1946 improve the performance of waking up a idle CPU, but
1947 will use a lot of power and make the system run hot.
1949 idle=halt: Halt is forced to be used for CPU idle.
1950 In such case C2/C3 won't be used again.
1951 idle=nomwait: Disable mwait for CPU C-states
1955 Allow force disabling of Shared Virtual Memory (SVA)
1956 support for the idxd driver. By default it is set to
1959 idxd.tc_override= [HW]
1961 Allow override of default traffic class configuration
1962 for the device. By default it is set to false (0).
1964 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1965 Format: { strict | legacy | 2008 | relaxed }
1968 Choose which programs will be accepted for execution
1969 based on the IEEE 754 NaN encoding(s) supported by
1970 the FPU and the NaN encoding requested with the value
1971 of an ELF file header flag individually set by each
1972 binary. Hardware implementations are permitted to
1973 support either or both of the legacy and the 2008 NaN
1976 Available settings are as follows:
1977 strict accept binaries that request a NaN encoding
1978 supported by the FPU
1979 legacy only accept legacy-NaN binaries, if supported
1981 2008 only accept 2008-NaN binaries, if supported
1983 relaxed accept any binaries regardless of whether
1984 supported by the FPU
1986 The FPU emulator is always able to support both NaN
1987 encodings, so if no FPU hardware is present or it has
1988 been disabled with 'nofpu', then the settings of
1989 'legacy' and '2008' strap the emulator accordingly,
1990 'relaxed' straps the emulator for both legacy-NaN and
1991 2008-NaN, whereas 'strict' enables legacy-NaN only on
1992 legacy processors and both NaN encodings on MIPS32 or
1995 The setting for ABS.fmt/NEG.fmt instruction execution
1996 mode generally follows that for the NaN encoding,
1997 except where unsupported by hardware.
1999 ignore_loglevel [KNL,EARLY]
2000 Ignore loglevel setting - this will print /all/
2001 kernel messages to the console. Useful for debugging.
2002 We also add it as printk module parameter, so users
2003 could change it dynamically, usually by
2004 /sys/module/printk/parameters/ignore_loglevel.
2007 Ignore RLIMIT_DATA setting for data mappings,
2008 print warning at first misuse. Can be changed via
2009 /sys/module/kernel/parameters/ignore_rlimit_data.
2011 ihash_entries= [KNL]
2012 Set number of hash buckets for inode cache.
2014 ima_appraise= [IMA] appraise integrity measurements
2015 Format: { "off" | "enforce" | "fix" | "log" }
2018 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
2019 The builtin appraise policy appraises all files
2022 ima_canonical_fmt [IMA]
2023 Use the canonical format for the binary runtime
2024 measurements, instead of host native format.
2027 Format: { md5 | sha1 | rmd160 | sha256 | sha384
2031 The list of supported hash algorithms is defined
2032 in crypto/hash_info.h.
2035 The builtin policies to load during IMA setup.
2036 Format: "tcb | appraise_tcb | secure_boot |
2037 fail_securely | critical_data"
2039 The "tcb" policy measures all programs exec'd, files
2040 mmap'd for exec, and all files opened with the read
2041 mode bit set by either the effective uid (euid=0) or
2044 The "appraise_tcb" policy appraises the integrity of
2045 all files owned by root.
2047 The "secure_boot" policy appraises the integrity
2048 of files (eg. kexec kernel image, kernel modules,
2049 firmware, policy, etc) based on file signatures.
2051 The "fail_securely" policy forces file signature
2052 verification failure also on privileged mounted
2053 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
2056 The "critical_data" policy measures kernel integrity
2059 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
2060 Load a policy which meets the needs of the Trusted
2061 Computing Base. This means IMA will measure all
2062 programs exec'd, files mmap'd for exec, and all files
2063 opened for read by uid=0.
2066 Select one of defined IMA measurements template formats.
2067 Formats: { "ima" | "ima-ng" | "ima-ngv2" | "ima-sig" |
2072 [IMA] Define a custom template format.
2073 Format: { "field1|...|fieldN" }
2075 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
2076 Format: <min_file_size>
2077 Set the minimal file size for using asynchronous hash.
2078 If left unspecified, ahash usage is disabled.
2080 ahash performance varies for different data sizes on
2081 different crypto accelerators. This option can be used
2082 to achieve the best performance for a particular HW.
2084 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
2086 Set hashing buffer size. Default: 4k.
2088 ahash performance varies for different chunk sizes on
2089 different crypto accelerators. This option can be used
2090 to achieve best performance for particular HW.
2094 Run specified binary instead of /sbin/init as init
2097 initcall_debug [KNL] Trace initcalls as they are executed. Useful
2098 for working out where the kernel is dying during
2101 initcall_blacklist= [KNL] Do not execute a comma-separated list of
2102 initcall functions. Useful for debugging built-in
2103 modules and initcalls.
2105 initramfs_async= [KNL]
2108 This parameter controls whether the initramfs
2109 image is unpacked asynchronously, concurrently
2110 with devices being probed and
2111 initialized. This should normally just work,
2112 but as a debugging aid, one can get the
2113 historical behaviour of the initramfs
2114 unpacking being completed before device_ and
2117 initrd= [BOOT,EARLY] Specify the location of the initial ramdisk
2119 initrdmem= [KNL,EARLY] Specify a physical address and size from which to
2120 load the initrd. If an initrd is compiled in or
2121 specified in the bootparams, it takes priority over this
2123 Format: ss[KMG],nn[KMG]
2126 init_on_alloc= [MM,EARLY] Fill newly allocated pages and heap objects with
2129 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
2131 init_on_free= [MM,EARLY] Fill freed pages and heap objects with zeroes.
2133 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
2135 init_pkru= [X86] Specify the default memory protection keys rights
2136 register contents for all processes. 0x55555554 by
2137 default (disallow access to all but pkey 0). Can
2138 override in debugfs after boot.
2140 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
2143 int_pln_enable [X86] Enable power limit notification interrupt
2145 integrity_audit=[IMA]
2146 Format: { "0" | "1" }
2147 0 -- basic integrity auditing messages. (Default)
2148 1 -- additional integrity auditing messages.
2150 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
2152 Enable intel iommu driver.
2154 Disable intel iommu driver.
2155 igfx_off [Default Off]
2156 By default, gfx is mapped as normal device. If a gfx
2157 device has a dedicated DMAR unit, the DMAR unit is
2158 bypassed by not enabling DMAR with this option. In
2159 this case, gfx device will use physical address for
2161 strict [Default Off]
2162 Deprecated, equivalent to iommu.strict=1.
2163 sp_off [Default Off]
2164 By default, super page will be supported if Intel IOMMU
2165 has the capability. With this option, super page will
2168 Enable the Intel IOMMU scalable mode if the hardware
2169 advertises that it has support for the scalable mode
2172 Disallow use of the Intel IOMMU scalable mode.
2173 tboot_noforce [Default Off]
2174 Do not force the Intel IOMMU enabled under tboot.
2175 By default, tboot will force Intel IOMMU on, which
2176 could harm performance of some high-throughput
2177 devices like 40GBit network cards, even if identity
2179 Note that using this option lowers the security
2180 provided by tboot because it makes the system
2181 vulnerable to DMA attacks.
2183 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
2184 0 disables intel_idle and fall back on acpi_idle.
2185 1 to 9 specify maximum depth of C-state.
2187 intel_pstate= [X86,EARLY]
2189 Do not enable intel_pstate as the default
2190 scaling driver for the supported processors
2192 Use intel_pstate driver to bypass the scaling
2193 governors layer of cpufreq and provides it own
2194 algorithms for p-state selection. There are two
2195 P-state selection algorithms provided by
2196 intel_pstate in the active mode: powersave and
2197 performance. The way they both operate depends
2198 on whether or not the hardware managed P-states
2199 (HWP) feature has been enabled in the processor
2200 and possibly on the processor model.
2202 Use intel_pstate as a scaling driver, but configure it
2203 to work with generic cpufreq governors (instead of
2204 enabling its internal governor). This mode cannot be
2205 used along with the hardware-managed P-states (HWP)
2208 Enable intel_pstate on systems that prohibit it by default
2209 in favor of acpi-cpufreq. Forcing the intel_pstate driver
2210 instead of acpi-cpufreq may disable platform features, such
2211 as thermal controls and power capping, that rely on ACPI
2212 P-States information being indicated to OSPM and therefore
2213 should be used with caution. This option does not work with
2214 processors that aren't supported by the intel_pstate driver
2215 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
2217 Do not enable hardware P state control (HWP)
2220 Only load intel_pstate on systems which support
2221 hardware P state control (HWP) if available.
2223 Enforce ACPI _PPC performance limits. If the Fixed ACPI
2224 Description Table, specifies preferred power management
2225 profile as "Enterprise Server" or "Performance Server",
2226 then this feature is turned on by default.
2228 Allow per-logical-CPU P-State performance control limits using
2229 cpufreq sysfs interface
2231 intremap= [X86-64,Intel-IOMMU,EARLY]
2232 on enable Interrupt Remapping (default)
2233 off disable Interrupt Remapping
2234 nosid disable Source ID checking
2236 BIOS x2APIC opt-out request will be ignored
2237 nopost disable Interrupt Posting
2239 iomem= Disable strict checking of access to MMIO memory
2240 strict regions from userspace.
2255 nobypass [PPC/POWERNV]
2256 Disable IOMMU bypass, using IOMMU for PCI devices.
2258 iommu.forcedac= [ARM64,X86,EARLY] Control IOVA allocation for PCI devices.
2259 Format: { "0" | "1" }
2260 0 - Try to allocate a 32-bit DMA address first, before
2261 falling back to the full range if needed.
2262 1 - Allocate directly from the full usable range,
2263 forcing Dual Address Cycle for PCI cards supporting
2264 greater than 32-bit addressing.
2266 iommu.strict= [ARM64,X86,S390,EARLY] Configure TLB invalidation behaviour
2267 Format: { "0" | "1" }
2269 Request that DMA unmap operations use deferred
2270 invalidation of hardware TLBs, for increased
2271 throughput at the cost of reduced device isolation.
2272 Will fall back to strict mode if not supported by
2273 the relevant IOMMU driver.
2275 DMA unmap operations invalidate IOMMU hardware TLBs
2277 unset - Use value of CONFIG_IOMMU_DEFAULT_DMA_{LAZY,STRICT}.
2278 Note: on x86, strict mode specified via one of the
2279 legacy driver-specific options takes precedence.
2282 [ARM64,X86,EARLY] Configure DMA to bypass the IOMMU by default.
2283 Format: { "0" | "1" }
2284 0 - Use IOMMU translation for DMA.
2285 1 - Bypass the IOMMU for DMA.
2286 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
2288 io7= [HW] IO7 for Marvel-based Alpha systems
2289 See comment before marvel_specify_io7 in
2290 arch/alpha/kernel/core_marvel.c.
2292 io_delay= [X86,EARLY] I/O delay method
2294 Standard port 0x80 based delay
2296 Alternate port 0xed based delay (needed on some systems)
2298 Simple two microseconds delay
2303 See Documentation/admin-guide/nfs/nfsroot.rst.
2305 ipcmni_extend [KNL,EARLY] Extend the maximum number of unique System V
2306 IPC identifiers from 32,768 to 16,777,216.
2308 irqaffinity= [SMP] Set the default irq affinity mask
2309 The argument is a cpu list, as described above.
2311 irqchip.gicv2_force_probe=
2314 Force the kernel to look for the second 4kB page
2315 of a GICv2 controller even if the memory range
2316 exposed by the device tree is too small.
2318 irqchip.gicv3_nolpi=
2320 Force the kernel to ignore the availability of
2321 LPIs (and by consequence ITSs). Intended for system
2322 that use the kernel as a bootloader, and thus want
2323 to let secondary kernels in charge of setting up
2326 irqchip.gicv3_pseudo_nmi= [ARM64,EARLY]
2327 Enables support for pseudo-NMIs in the kernel. This
2328 requires the kernel to be built with
2329 CONFIG_ARM64_PSEUDO_NMI.
2332 When an interrupt is not handled search all handlers
2333 for it. Intended to get systems with badly broken
2337 When an interrupt is not handled search all handlers
2338 for it. Also check all handlers each timer
2339 interrupt. Intended to get systems with badly broken
2343 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
2345 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
2346 [Deprecated - use cpusets instead]
2347 Format: [flag-list,]<cpu-list>
2349 Specify one or more CPUs to isolate from disturbances
2350 specified in the flag list (default: domain):
2353 Disable the tick when a single task runs.
2355 A residual 1Hz tick is offloaded to workqueues, which you
2356 need to affine to housekeeping through the global
2357 workqueue's affinity configured via the
2358 /sys/devices/virtual/workqueue/cpumask sysfs file, or
2359 by using the 'domain' flag described below.
2361 NOTE: by default the global workqueue runs on all CPUs,
2362 so to protect individual CPUs the 'cpumask' file has to
2363 be configured manually after bootup.
2366 Isolate from the general SMP balancing and scheduling
2367 algorithms. Note that performing domain isolation this way
2368 is irreversible: it's not possible to bring back a CPU to
2369 the domains once isolated through isolcpus. It's strongly
2370 advised to use cpusets instead to disable scheduler load
2371 balancing through the "cpuset.sched_load_balance" file.
2372 It offers a much more flexible interface where CPUs can
2373 move in and out of an isolated set anytime.
2375 You can move a process onto or off an "isolated" CPU via
2376 the CPU affinity syscalls or cpuset.
2377 <cpu number> begins at 0 and the maximum value is
2378 "number of CPUs in system - 1".
2382 Isolate from being targeted by managed interrupts
2383 which have an interrupt mask containing isolated
2384 CPUs. The affinity of managed interrupts is
2385 handled by the kernel and cannot be changed via
2386 the /proc/irq/* interfaces.
2388 This isolation is best effort and only effective
2389 if the automatically assigned interrupt mask of a
2390 device queue contains isolated and housekeeping
2391 CPUs. If housekeeping CPUs are online then such
2392 interrupts are directed to the housekeeping CPU
2393 so that IO submitted on the housekeeping CPU
2394 cannot disturb the isolated CPU.
2396 If a queue's affinity mask contains only isolated
2397 CPUs then this parameter has no effect on the
2398 interrupt routing decision, though interrupts are
2399 only delivered when tasks running on those
2400 isolated CPUs submit IO. IO submitted on
2401 housekeeping CPUs has no influence on those
2404 The format of <cpu-list> is described above.
2408 ivrs_ioapic [HW,X86-64]
2409 Provide an override to the IOAPIC-ID<->DEVICE-ID
2410 mapping provided in the IVRS ACPI table.
2411 By default, PCI segment is 0, and can be omitted.
2413 For example, to map IOAPIC-ID decimal 10 to
2414 PCI segment 0x1 and PCI device 00:14.0,
2415 write the parameter as:
2416 ivrs_ioapic=10@0001:00:14.0
2419 * To map IOAPIC-ID decimal 10 to PCI device 00:14.0
2420 write the parameter as:
2421 ivrs_ioapic[10]=00:14.0
2422 * To map IOAPIC-ID decimal 10 to PCI segment 0x1 and
2423 PCI device 00:14.0 write the parameter as:
2424 ivrs_ioapic[10]=0001:00:14.0
2426 ivrs_hpet [HW,X86-64]
2427 Provide an override to the HPET-ID<->DEVICE-ID
2428 mapping provided in the IVRS ACPI table.
2429 By default, PCI segment is 0, and can be omitted.
2431 For example, to map HPET-ID decimal 10 to
2432 PCI segment 0x1 and PCI device 00:14.0,
2433 write the parameter as:
2434 ivrs_hpet=10@0001:00:14.0
2437 * To map HPET-ID decimal 0 to PCI device 00:14.0
2438 write the parameter as:
2439 ivrs_hpet[0]=00:14.0
2440 * To map HPET-ID decimal 10 to PCI segment 0x1 and
2441 PCI device 00:14.0 write the parameter as:
2442 ivrs_ioapic[10]=0001:00:14.0
2444 ivrs_acpihid [HW,X86-64]
2445 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2446 mapping provided in the IVRS ACPI table.
2447 By default, PCI segment is 0, and can be omitted.
2449 For example, to map UART-HID:UID AMD0020:0 to
2450 PCI segment 0x1 and PCI device ID 00:14.5,
2451 write the parameter as:
2452 ivrs_acpihid=AMD0020:0@0001:00:14.5
2455 * To map UART-HID:UID AMD0020:0 to PCI segment is 0,
2456 PCI device ID 00:14.5, write the parameter as:
2457 ivrs_acpihid[00:14.5]=AMD0020:0
2458 * To map UART-HID:UID AMD0020:0 to PCI segment 0x1 and
2459 PCI device ID 00:14.5, write the parameter as:
2460 ivrs_acpihid[0001:00:14.5]=AMD0020:0
2462 js= [HW,JOY] Analog joystick
2463 See Documentation/input/joydev/joystick.rst.
2466 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2467 report on every invalid memory access. Without this
2468 parameter KASAN will print report only for the first
2471 keep_bootcon [KNL,EARLY]
2472 Do not unregister boot console at start. This is only
2473 useful for debugging when something happens in the window
2474 between unregistering the boot console and initializing
2477 keepinitrd [HW,ARM] See retain_initrd.
2479 kernelcore= [KNL,X86,IA-64,PPC,EARLY]
2480 Format: nn[KMGTPE] | nn% | "mirror"
2481 This parameter specifies the amount of memory usable by
2482 the kernel for non-movable allocations. The requested
2483 amount is spread evenly throughout all nodes in the
2484 system as ZONE_NORMAL. The remaining memory is used for
2485 movable memory in its own zone, ZONE_MOVABLE. In the
2486 event, a node is too small to have both ZONE_NORMAL and
2487 ZONE_MOVABLE, kernelcore memory will take priority and
2488 other nodes will have a larger ZONE_MOVABLE.
2490 ZONE_MOVABLE is used for the allocation of pages that
2491 may be reclaimed or moved by the page migration
2492 subsystem. Note that allocations like PTEs-from-HighMem
2493 still use the HighMem zone if it exists, and the Normal
2494 zone if it does not.
2496 It is possible to specify the exact amount of memory in
2497 the form of "nn[KMGTPE]", a percentage of total system
2498 memory in the form of "nn%", or "mirror". If "mirror"
2499 option is specified, mirrored (reliable) memory is used
2500 for non-movable allocations and remaining memory is used
2501 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2502 are exclusive, so you cannot specify multiple forms.
2504 kgdbdbgp= [KGDB,HW,EARLY] kgdb over EHCI usb debug port.
2505 Format: <Controller#>[,poll interval]
2506 The controller # is the number of the ehci usb debug
2507 port as it is probed via PCI. The poll interval is
2508 optional and is the number seconds in between
2509 each poll cycle to the debug port in case you need
2510 the functionality for interrupting the kernel with
2511 gdb or control-c on the dbgp connection. When
2512 not using this parameter you use sysrq-g to break into
2513 the kernel debugger.
2515 kgdboc= [KGDB,HW] kgdb over consoles.
2516 Requires a tty driver that supports console polling,
2517 or a supported polling keyboard driver (non-usb).
2518 Serial only format: <serial_device>[,baud]
2519 keyboard only format: kbd
2520 keyboard and serial format: kbd,<serial_device>[,baud]
2521 Optional Kernel mode setting:
2522 kms, kbd format: kms,kbd
2523 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2525 kgdboc_earlycon= [KGDB,HW,EARLY]
2526 If the boot console provides the ability to read
2527 characters and can work in polling mode, you can use
2528 this parameter to tell kgdb to use it as a backend
2529 until the normal console is registered. Intended to
2530 be used together with the kgdboc parameter which
2531 specifies the normal console to transition to.
2533 The name of the early console should be specified
2534 as the value of this parameter. Note that the name of
2535 the early console might be different than the tty
2536 name passed to kgdboc. It's OK to leave the value
2537 blank and the first boot console that implements
2538 read() will be picked.
2540 kgdbwait [KGDB,EARLY] Stop kernel execution and enter the
2541 kernel debugger at the earliest opportunity.
2543 kmac= [MIPS] Korina ethernet MAC address.
2544 Configure the RouterBoard 532 series on-chip
2545 Ethernet adapter MAC address.
2547 kmemleak= [KNL,EARLY] Boot-time kmemleak enable/disable
2548 Valid arguments: on, off
2550 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2553 kprobe_event=[probe-list]
2554 [FTRACE] Add kprobe events and enable at boot time.
2555 The probe-list is a semicolon delimited list of probe
2556 definitions. Each definition is same as kprobe_events
2557 interface, but the parameters are comma delimited.
2558 For example, to add a kprobe event on vfs_read with
2559 arg1 and arg2, add to the command line;
2561 kprobe_event=p,vfs_read,$arg1,$arg2
2563 See also Documentation/trace/kprobetrace.rst "Kernel
2564 Boot Parameter" section.
2566 kpti= [ARM64,EARLY] Control page table isolation of
2567 user and kernel address spaces.
2568 Default: enabled on cores which need mitigation.
2572 kunit.enable= [KUNIT] Enable executing KUnit tests. Requires
2573 CONFIG_KUNIT to be set to be fully enabled. The
2574 default value can be overridden via
2575 KUNIT_DEFAULT_ENABLED.
2576 Default is 1 (enabled)
2578 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2579 Default is 0 (don't ignore, but inject #GP)
2581 kvm.eager_page_split=
2582 [KVM,X86] Controls whether or not KVM will try to
2583 proactively split all huge pages during dirty logging.
2584 Eager page splitting reduces interruptions to vCPU
2585 execution by eliminating the write-protection faults
2586 and MMU lock contention that would otherwise be
2587 required to split huge pages lazily.
2589 VM workloads that rarely perform writes or that write
2590 only to a small region of VM memory may benefit from
2591 disabling eager page splitting to allow huge pages to
2592 still be used for reads.
2594 The behavior of eager page splitting depends on whether
2595 KVM_DIRTY_LOG_INITIALLY_SET is enabled or disabled. If
2596 disabled, all huge pages in a memslot will be eagerly
2597 split when dirty logging is enabled on that memslot. If
2598 enabled, eager page splitting will be performed during
2599 the KVM_CLEAR_DIRTY ioctl, and only for the pages being
2602 Eager page splitting is only supported when kvm.tdp_mmu=Y.
2606 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2607 Default is false (don't support).
2610 [KVM] Controls the software workaround for the
2611 X86_BUG_ITLB_MULTIHIT bug.
2612 force : Always deploy workaround.
2613 off : Never deploy workaround.
2614 auto : Deploy workaround based on the presence of
2615 X86_BUG_ITLB_MULTIHIT.
2619 If the software workaround is enabled for the host,
2620 guests do need not to enable it for nested guests.
2622 kvm.nx_huge_pages_recovery_ratio=
2623 [KVM] Controls how many 4KiB pages are periodically zapped
2624 back to huge pages. 0 disables the recovery, otherwise if
2625 the value is N KVM will zap 1/Nth of the 4KiB pages every
2626 period (see below). The default is 60.
2628 kvm.nx_huge_pages_recovery_period_ms=
2629 [KVM] Controls the time period at which KVM zaps 4KiB pages
2630 back to huge pages. If the value is a non-zero N, KVM will
2631 zap a portion (see ratio above) of the pages every N msecs.
2632 If the value is 0 (the default), KVM will pick a period based
2633 on the ratio, such that a page is zapped after 1 hour on average.
2635 kvm-amd.nested= [KVM,AMD] Control nested virtualization feature in
2636 KVM/SVM. Default is 1 (enabled).
2638 kvm-amd.npt= [KVM,AMD] Control KVM's use of Nested Page Tables,
2639 a.k.a. Two-Dimensional Page Tables. Default is 1
2640 (enabled). Disable by KVM if hardware lacks support
2644 [KVM,ARM,EARLY] Select one of KVM/arm64's modes of
2647 none: Forcefully disable KVM.
2649 nvhe: Standard nVHE-based mode, without support for
2652 protected: nVHE-based mode with support for guests whose
2653 state is kept private from the host.
2655 nested: VHE-based mode with support for nested
2656 virtualization. Requires at least ARMv8.3
2659 Defaults to VHE/nVHE based on hardware support. Setting
2660 mode to "protected" will disable kexec and hibernation
2661 for the host. "nested" is experimental and should be
2662 used with extreme caution.
2664 kvm-arm.vgic_v3_group0_trap=
2665 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-0
2668 kvm-arm.vgic_v3_group1_trap=
2669 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-1
2672 kvm-arm.vgic_v3_common_trap=
2673 [KVM,ARM,EARLY] Trap guest accesses to GICv3 common
2676 kvm-arm.vgic_v4_enable=
2677 [KVM,ARM,EARLY] Allow use of GICv4 for direct
2680 kvm_cma_resv_ratio=n [PPC,EARLY]
2681 Reserves given percentage from system memory area for
2682 contiguous memory allocation for KVM hash pagetable
2684 By default it reserves 5% of total system memory.
2688 kvm-intel.ept= [KVM,Intel] Control KVM's use of Extended Page Tables,
2689 a.k.a. Two-Dimensional Page Tables. Default is 1
2690 (enabled). Disable by KVM if hardware lacks support
2693 kvm-intel.emulate_invalid_guest_state=
2694 [KVM,Intel] Control whether to emulate invalid guest
2695 state. Ignored if kvm-intel.enable_unrestricted_guest=1,
2696 as guest state is never invalid for unrestricted
2697 guests. This param doesn't apply to nested guests (L2),
2698 as KVM never emulates invalid L2 guest state.
2699 Default is 1 (enabled).
2701 kvm-intel.flexpriority=
2702 [KVM,Intel] Control KVM's use of FlexPriority feature
2703 (TPR shadow). Default is 1 (enabled). Disable by KVM if
2704 hardware lacks support for it.
2707 [KVM,Intel] Control nested virtualization feature in
2708 KVM/VMX. Default is 1 (enabled).
2710 kvm-intel.unrestricted_guest=
2711 [KVM,Intel] Control KVM's use of unrestricted guest
2712 feature (virtualized real and unpaged mode). Default
2713 is 1 (enabled). Disable by KVM if EPT is disabled or
2714 hardware lacks support for it.
2716 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2719 Valid arguments: never, cond, always
2721 always: L1D cache flush on every VMENTER.
2722 cond: Flush L1D on VMENTER only when the code between
2723 VMEXIT and VMENTER can leak host memory.
2724 never: Disables the mitigation
2726 Default is cond (do L1 cache flush in specific instances)
2728 kvm-intel.vpid= [KVM,Intel] Control KVM's use of Virtual Processor
2729 Identification feature (tagged TLBs). Default is 1
2730 (enabled). Disable by KVM if hardware lacks support
2733 l1d_flush= [X86,INTEL,EARLY]
2734 Control mitigation for L1D based snooping vulnerability.
2736 Certain CPUs are vulnerable to an exploit against CPU
2737 internal buffers which can forward information to a
2738 disclosure gadget under certain conditions.
2740 In vulnerable processors, the speculatively
2741 forwarded data can be used in a cache side channel
2742 attack, to access data to which the attacker does
2743 not have direct access.
2745 This parameter controls the mitigation. The
2748 on - enable the interface for the mitigation
2750 l1tf= [X86,EARLY] Control mitigation of the L1TF vulnerability on
2753 The kernel PTE inversion protection is unconditionally
2754 enabled and cannot be disabled.
2757 Provides all available mitigations for the
2758 L1TF vulnerability. Disables SMT and
2759 enables all mitigations in the
2760 hypervisors, i.e. unconditional L1D flush.
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 'full', but disables SMT and L1D
2771 flush runtime control. Implies the
2772 'nosmt=force' command line option.
2773 (i.e. sysfs control of SMT is disabled.)
2776 Leaves SMT enabled and enables the default
2777 hypervisor mitigation, i.e. conditional
2780 SMT control and L1D flush control via the
2781 sysfs interface is still possible after
2782 boot. Hypervisors will issue a warning
2783 when the first VM is started in a
2784 potentially insecure configuration,
2785 i.e. SMT enabled or L1D flush disabled.
2789 Disables SMT and enables the default
2790 hypervisor mitigation.
2792 SMT control and L1D flush control via the
2793 sysfs interface is still possible after
2794 boot. Hypervisors will issue a warning
2795 when the first VM is started in a
2796 potentially insecure configuration,
2797 i.e. SMT enabled or L1D flush disabled.
2800 Same as 'flush', but hypervisors will not
2801 warn when a VM is started in a potentially
2802 insecure configuration.
2805 Disables hypervisor mitigations and doesn't
2807 It also drops the swap size and available
2808 RAM limit restriction on both hypervisor and
2813 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2819 lapic [X86-32,APIC,EARLY] Enable the local APIC even if BIOS
2822 lapic= [X86,APIC] Do not use TSC deadline
2823 value for LAPIC timer one-shot implementation. Default
2824 back to the programmable timer unit in the LAPIC.
2825 Format: notscdeadline
2827 lapic_timer_c2_ok [X86,APIC,EARLY] trust the local apic timer
2830 libata.dma= [LIBATA] DMA control
2831 libata.dma=0 Disable all PATA and SATA DMA
2832 libata.dma=1 PATA and SATA Disk DMA only
2833 libata.dma=2 ATAPI (CDROM) DMA only
2834 libata.dma=4 Compact Flash DMA only
2835 Combinations also work, so libata.dma=3 enables DMA
2836 for disks and CDROMs, but not CFs.
2838 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2839 libata.ignore_hpa=0 keep BIOS limits (default)
2840 libata.ignore_hpa=1 ignore limits, using full disk
2842 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2846 libata.force= [LIBATA] Force configurations. The format is a comma-
2847 separated list of "[ID:]VAL" where ID is PORT[.DEVICE].
2848 PORT and DEVICE are decimal numbers matching port, link
2849 or device. Basically, it matches the ATA ID string
2850 printed on console by libata. If the whole ID part is
2851 omitted, the last PORT and DEVICE values are used. If
2852 ID hasn't been specified yet, the configuration applies
2853 to all ports, links and devices.
2855 If only DEVICE is omitted, the parameter applies to
2856 the port and all links and devices behind it. DEVICE
2857 number of 0 either selects the first device or the
2858 first fan-out link behind PMP device. It does not
2859 select the host link. DEVICE number of 15 selects the
2860 host link and device attached to it.
2862 The VAL specifies the configuration to force. As long
2863 as there is no ambiguity, shortcut notation is allowed.
2864 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2865 The following configurations can be forced.
2867 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2868 Any ID with matching PORT is used.
2870 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2872 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2873 udma[/][16,25,33,44,66,100,133] notation is also
2876 * nohrst, nosrst, norst: suppress hard, soft and both
2879 * rstonce: only attempt one reset during hot-unplug
2882 * [no]dbdelay: Enable or disable the extra 200ms delay
2883 before debouncing a link PHY and device presence
2886 * [no]ncq: Turn on or off NCQ.
2888 * [no]ncqtrim: Enable or disable queued DSM TRIM.
2890 * [no]ncqati: Enable or disable NCQ trim on ATI chipset.
2892 * [no]trim: Enable or disable (unqueued) TRIM.
2894 * trim_zero: Indicate that TRIM command zeroes data.
2896 * max_trim_128m: Set 128M maximum trim size limit.
2898 * [no]dma: Turn on or off DMA transfers.
2900 * atapi_dmadir: Enable ATAPI DMADIR bridge support.
2902 * atapi_mod16_dma: Enable the use of ATAPI DMA for
2903 commands that are not a multiple of 16 bytes.
2905 * [no]dmalog: Enable or disable the use of the
2906 READ LOG DMA EXT command to access logs.
2908 * [no]iddevlog: Enable or disable access to the
2909 identify device data log.
2911 * [no]logdir: Enable or disable access to the general
2912 purpose log directory.
2914 * max_sec_128: Set transfer size limit to 128 sectors.
2916 * max_sec_1024: Set or clear transfer size limit to
2919 * max_sec_lba48: Set or clear transfer size limit to
2922 * [no]lpm: Enable or disable link power management.
2924 * [no]setxfer: Indicate if transfer speed mode setting
2927 * [no]fua: Disable or enable FUA (Force Unit Access)
2928 support for devices supporting this feature.
2930 * dump_id: Dump IDENTIFY data.
2932 * disable: Disable this device.
2934 If there are multiple matching configurations changing
2935 the same attribute, the last one is used.
2937 load_ramdisk= [RAM] [Deprecated]
2939 lockd.nlm_grace_period=P [NFS] Assign grace period.
2942 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2945 lockd.nlm_timeout=T [NFS] Assign timeout value.
2948 lockd.nlm_udpport=M [NFS] Assign UDP port.
2951 lockdown= [SECURITY,EARLY]
2952 { integrity | confidentiality }
2953 Enable the kernel lockdown feature. If set to
2954 integrity, kernel features that allow userland to
2955 modify the running kernel are disabled. If set to
2956 confidentiality, kernel features that allow userland
2957 to extract confidential information from the kernel
2960 locktorture.acq_writer_lim= [KNL]
2961 Set the time limit in jiffies for a lock
2962 acquisition. Acquisitions exceeding this limit
2963 will result in a splat once they do complete.
2965 locktorture.bind_readers= [KNL]
2966 Specify the list of CPUs to which the readers are
2969 locktorture.bind_writers= [KNL]
2970 Specify the list of CPUs to which the writers are
2973 locktorture.call_rcu_chains= [KNL]
2974 Specify the number of self-propagating call_rcu()
2975 chains to set up. These are used to ensure that
2976 there is a high probability of an RCU grace period
2977 in progress at any given time. Defaults to 0,
2978 which disables these call_rcu() chains.
2980 locktorture.long_hold= [KNL]
2981 Specify the duration in milliseconds for the
2982 occasional long-duration lock hold time. Defaults
2983 to 100 milliseconds. Select 0 to disable.
2985 locktorture.nested_locks= [KNL]
2986 Specify the maximum lock nesting depth that
2987 locktorture is to exercise, up to a limit of 8
2988 (MAX_NESTED_LOCKS). Specify zero to disable.
2989 Note that this parameter is ineffective on types
2990 of locks that do not support nested acquisition.
2992 locktorture.nreaders_stress= [KNL]
2993 Set the number of locking read-acquisition kthreads.
2994 Defaults to being automatically set based on the
2995 number of online CPUs.
2997 locktorture.nwriters_stress= [KNL]
2998 Set the number of locking write-acquisition kthreads.
3000 locktorture.onoff_holdoff= [KNL]
3001 Set time (s) after boot for CPU-hotplug testing.
3003 locktorture.onoff_interval= [KNL]
3004 Set time (s) between CPU-hotplug operations, or
3005 zero to disable CPU-hotplug testing.
3007 locktorture.rt_boost= [KNL]
3008 Do periodic testing of real-time lock priority
3009 boosting. Select 0 to disable, 1 to boost
3010 only rt_mutex, and 2 to boost unconditionally.
3011 Defaults to 2, which might seem to be an
3012 odd choice, but which should be harmless for
3013 non-real-time spinlocks, due to their disabling
3014 of preemption. Note that non-realtime mutexes
3017 locktorture.rt_boost_factor= [KNL]
3018 Number that determines how often and for how
3019 long priority boosting is exercised. This is
3020 scaled down by the number of writers, so that the
3021 number of boosts per unit time remains roughly
3022 constant as the number of writers increases.
3023 On the other hand, the duration of each boost
3024 increases with the number of writers.
3026 locktorture.shuffle_interval= [KNL]
3027 Set task-shuffle interval (jiffies). Shuffling
3028 tasks allows some CPUs to go into dyntick-idle
3029 mode during the locktorture test.
3031 locktorture.shutdown_secs= [KNL]
3032 Set time (s) after boot system shutdown. This
3033 is useful for hands-off automated testing.
3035 locktorture.stat_interval= [KNL]
3036 Time (s) between statistics printk()s.
3038 locktorture.stutter= [KNL]
3039 Time (s) to stutter testing, for example,
3040 specifying five seconds causes the test to run for
3041 five seconds, wait for five seconds, and so on.
3042 This tests the locking primitive's ability to
3043 transition abruptly to and from idle.
3045 locktorture.torture_type= [KNL]
3046 Specify the locking implementation to test.
3048 locktorture.verbose= [KNL]
3049 Enable additional printk() statements.
3051 locktorture.writer_fifo= [KNL]
3052 Run the write-side locktorture kthreads at
3053 sched_set_fifo() real-time priority.
3055 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
3058 loglevel= [KNL,EARLY]
3059 All Kernel Messages with a loglevel smaller than the
3060 console loglevel will be printed to the console. It can
3061 also be changed with klogd or other programs. The
3062 loglevels are defined as follows:
3064 0 (KERN_EMERG) system is unusable
3065 1 (KERN_ALERT) action must be taken immediately
3066 2 (KERN_CRIT) critical conditions
3067 3 (KERN_ERR) error conditions
3068 4 (KERN_WARNING) warning conditions
3069 5 (KERN_NOTICE) normal but significant condition
3070 6 (KERN_INFO) informational
3071 7 (KERN_DEBUG) debug-level messages
3073 log_buf_len=n[KMG] [KNL,EARLY]
3074 Sets the size of the printk ring buffer, in bytes.
3075 n must be a power of two and greater than the
3076 minimal size. The minimal size is defined by
3077 LOG_BUF_SHIFT kernel config parameter. There
3078 is also CONFIG_LOG_CPU_MAX_BUF_SHIFT config
3079 parameter that allows to increase the default size
3080 depending on the number of CPUs. See init/Kconfig
3083 logo.nologo [FB] Disables display of the built-in Linux logo.
3084 This may be used to provide more screen space for
3085 kernel log messages and is useful when debugging
3086 kernel boot problems.
3088 lp=0 [LP] Specify parallel ports to use, e.g,
3089 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
3090 lp=reset first parallel port). 'lp=0' disables the
3091 lp=auto printer driver. 'lp=reset' (which can be
3092 specified in addition to the ports) causes
3093 attached printers to be reset. Using
3094 lp=port1,port2,... specifies the parallel ports
3095 to associate lp devices with, starting with
3096 lp0. A port specification may be 'none' to skip
3097 that lp device, or a parport name such as
3098 'parport0'. Specifying 'lp=auto' instead of a
3099 port specification list means that device IDs
3100 from each port should be examined, to see if
3101 an IEEE 1284-compliant printer is attached; if
3102 so, the driver will manage that printer.
3103 See also header of drivers/char/lp.c.
3106 Sets loops_per_jiffy to given constant, thus avoiding
3107 time-consuming boot-time autodetection (up to 250 ms per
3108 CPU). 0 enables autodetection (default). To determine
3109 the correct value for your kernel, boot with normal
3110 autodetection and see what value is printed. Note that
3111 on SMP systems the preset will be applied to all CPUs,
3112 which is likely to cause problems if your CPUs need
3113 significantly divergent settings. An incorrect value
3114 will cause delays in the kernel to be wrong, leading to
3115 unpredictable I/O errors and other breakage. Although
3116 unlikely, in the extreme case this might damage your
3120 Format: <io>,<irq>,<dma>
3122 lsm.debug [SECURITY] Enable LSM initialization debugging output.
3125 [SECURITY] Choose order of LSM initialization. This
3126 overrides CONFIG_LSM, and the "security=" parameter.
3128 machvec= [IA-64] Force the use of a particular machine-vector
3129 (machvec) in a generic kernel.
3130 Example: machvec=hpzx1
3132 machtype= [Loongson] Share the same kernel image file between
3133 different yeeloong laptops.
3134 Example: machtype=lemote-yeeloong-2f-7inch
3136 max_addr=nn[KMG] [KNL,BOOT,IA-64] All physical memory greater
3137 than or equal to this physical address is ignored.
3139 maxcpus= [SMP,EARLY] Maximum number of processors that an SMP kernel
3140 will bring up during bootup. maxcpus=n : n >= 0 limits
3141 the kernel to bring up 'n' processors. Surely after
3142 bootup you can bring up the other plugged cpu by executing
3143 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
3144 only takes effect during system bootup.
3145 While n=0 is a special case, it is equivalent to "nosmp",
3146 which also disables the IO APIC.
3148 max_loop= [LOOP] The number of loop block devices that get
3149 (loop.max_loop) unconditionally pre-created at init time. The default
3150 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
3151 of statically allocating a predefined number, loop
3152 devices can be requested on-demand with the
3153 /dev/loop-control interface.
3155 mce [X86-32] Machine Check Exception
3157 mce=option [X86-64] See Documentation/arch/x86/x86_64/boot-options.rst
3159 md= [HW] RAID subsystems devices and level
3160 See Documentation/admin-guide/md.rst.
3163 Format: <first>,<last>
3164 Specifies range of consoles to be captured by the MDA.
3166 mds= [X86,INTEL,EARLY]
3167 Control mitigation for the Micro-architectural Data
3168 Sampling (MDS) vulnerability.
3170 Certain CPUs are vulnerable to an exploit against CPU
3171 internal buffers which can forward information to a
3172 disclosure gadget under certain conditions.
3174 In vulnerable processors, the speculatively
3175 forwarded data can be used in a cache side channel
3176 attack, to access data to which the attacker does
3177 not have direct access.
3179 This parameter controls the MDS mitigation. The
3182 full - Enable MDS mitigation on vulnerable CPUs
3183 full,nosmt - Enable MDS mitigation and disable
3184 SMT on vulnerable CPUs
3185 off - Unconditionally disable MDS mitigation
3187 On TAA-affected machines, mds=off can be prevented by
3188 an active TAA mitigation as both vulnerabilities are
3189 mitigated with the same mechanism so in order to disable
3190 this mitigation, you need to specify tsx_async_abort=off
3193 Not specifying this option is equivalent to
3196 For details see: Documentation/admin-guide/hw-vuln/mds.rst
3198 mem=nn[KMG] [HEXAGON,EARLY] Set the memory size.
3199 Must be specified, otherwise memory size will be 0.
3201 mem=nn[KMG] [KNL,BOOT,EARLY] Force usage of a specific amount
3202 of memory Amount of memory to be used in cases
3206 2 when the kernel is not able to see the whole system memory;
3207 3 memory that lies after 'mem=' boundary is excluded from
3208 the hypervisor, then assigned to KVM guests.
3209 4 to limit the memory available for kdump kernel.
3211 [ARC,MICROBLAZE] - the limit applies only to low memory,
3212 high memory is not affected.
3214 [ARM64] - only limits memory covered by the linear
3215 mapping. The NOMAP regions are not affected.
3217 [X86] Work as limiting max address. Use together
3218 with memmap= to avoid physical address space collisions.
3219 Without memmap= PCI devices could be placed at addresses
3220 belonging to unused RAM.
3222 Note that this only takes effects during boot time since
3223 in above case 3, memory may need be hot added after boot
3224 if system memory of hypervisor is not sufficient.
3227 [ARM,MIPS,EARLY] - override the memory layout
3228 reported by firmware.
3229 Define a memory region of size nn[KMG] starting at
3231 Multiple different regions can be specified with
3232 multiple mem= parameters on the command line.
3234 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
3237 memblock=debug [KNL,EARLY] Enable memblock debug messages.
3240 [KNL,SH] Allow user to override the default size for
3241 per-device physically contiguous DMA buffers.
3243 memhp_default_state=online/offline/online_kernel/online_movable
3244 [KNL] Set the initial state for the memory hotplug
3245 onlining policy. If not specified, the default value is
3246 set according to the
3247 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
3249 See Documentation/admin-guide/mm/memory-hotplug.rst.
3251 memmap=exactmap [KNL,X86,EARLY] Enable setting of an exact
3252 E820 memory map, as specified by the user.
3253 Such memmap=exactmap lines can be constructed based on
3254 BIOS output or other requirements. See the memmap=nn@ss
3257 memmap=nn[KMG]@ss[KMG]
3258 [KNL, X86,MIPS,XTENSA,EARLY] Force usage of a specific region of memory.
3259 Region of memory to be used is from ss to ss+nn.
3260 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
3261 which limits max address to nn[KMG].
3262 Multiple different regions can be specified,
3265 memmap=100M@2G,100M#3G,1G!1024G
3267 memmap=nn[KMG]#ss[KMG]
3268 [KNL,ACPI,EARLY] Mark specific memory as ACPI data.
3269 Region of memory to be marked is from ss to ss+nn.
3271 memmap=nn[KMG]$ss[KMG]
3272 [KNL,ACPI,EARLY] Mark specific memory as reserved.
3273 Region of memory to be reserved is from ss to ss+nn.
3274 Example: Exclude memory from 0x18690000-0x1869ffff
3275 memmap=64K$0x18690000
3277 memmap=0x10000$0x18690000
3278 Some bootloaders may need an escape character before '$',
3279 like Grub2, otherwise '$' and the following number
3282 memmap=nn[KMG]!ss[KMG,EARLY]
3283 [KNL,X86] Mark specific memory as protected.
3284 Region of memory to be used, from ss to ss+nn.
3285 The memory region may be marked as e820 type 12 (0xc)
3286 and is NVDIMM or ADR memory.
3288 memmap=<size>%<offset>-<oldtype>+<newtype>
3289 [KNL,ACPI,EARLY] Convert memory within the specified region
3290 from <oldtype> to <newtype>. If "-<oldtype>" is left
3291 out, the whole region will be marked as <newtype>,
3292 even if previously unavailable. If "+<newtype>" is left
3293 out, matching memory will be removed. Types are
3294 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
3295 3 = ACPI, 12 = PRAM.
3297 memory_corruption_check=0/1 [X86,EARLY]
3298 Some BIOSes seem to corrupt the first 64k of
3299 memory when doing things like suspend/resume.
3300 Setting this option will scan the memory
3301 looking for corruption. Enabling this will
3302 both detect corruption and prevent the kernel
3303 from using the memory being corrupted.
3304 However, its intended as a diagnostic tool; if
3305 repeatable BIOS-originated corruption always
3306 affects the same memory, you can use memmap=
3307 to prevent the kernel from using that memory.
3309 memory_corruption_check_size=size [X86,EARLY]
3310 By default it checks for corruption in the low
3311 64k, making this memory unavailable for normal
3312 use. Use this parameter to scan for
3313 corruption in more or less memory.
3315 memory_corruption_check_period=seconds [X86,EARLY]
3316 By default it checks for corruption every 60
3317 seconds. Use this parameter to check at some
3318 other rate. 0 disables periodic checking.
3320 memory_hotplug.memmap_on_memory
3321 [KNL,X86,ARM] Boolean flag to enable this feature.
3322 Format: {on | off (default)}
3323 When enabled, runtime hotplugged memory will
3324 allocate its internal metadata (struct pages,
3325 those vmemmap pages cannot be optimized even
3326 if hugetlb_free_vmemmap is enabled) from the
3327 hotadded memory which will allow to hotadd a
3328 lot of memory without requiring additional
3330 This feature is disabled by default because it
3331 has some implication on large (e.g. GB)
3332 allocations in some configurations (e.g. small
3334 The state of the flag can be read in
3335 /sys/module/memory_hotplug/parameters/memmap_on_memory.
3336 Note that even when enabled, there are a few cases where
3337 the feature is not effective.
3339 memtest= [KNL,X86,ARM,M68K,PPC,RISCV,EARLY] Enable memtest
3341 default : 0 <disable>
3342 Specifies the number of memtest passes to be
3343 performed. Each pass selects another test
3344 pattern from a given set of patterns. Memtest
3345 fills the memory with this pattern, validates
3346 memory contents and reserves bad memory
3347 regions that are detected.
3349 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
3350 Valid arguments: on, off
3352 mem_encrypt=on: Activate SME
3353 mem_encrypt=off: Do not activate SME
3355 Refer to Documentation/virt/kvm/x86/amd-memory-encryption.rst
3356 for details on when memory encryption can be activated.
3358 mem_sleep_default= [SUSPEND] Default system suspend mode:
3359 s2idle - Suspend-To-Idle
3360 shallow - Power-On Suspend or equivalent (if supported)
3361 deep - Suspend-To-RAM or equivalent (if supported)
3362 See Documentation/admin-guide/pm/sleep-states.rst.
3364 mfgpt_irq= [IA-32] Specify the IRQ to use for the
3365 Multi-Function General Purpose Timers on AMD Geode
3368 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
3369 the BIOS has incorrectly applied a workaround. TinyBIOS
3370 version 0.98 is known to be affected, 0.99 fixes the
3371 problem by letting the user disable the workaround.
3375 microcode.force_minrev= [X86]
3377 Enable or disable the microcode minimal revision
3378 enforcement for the runtime microcode loader.
3380 min_addr=nn[KMG] [KNL,BOOT,IA-64] All physical memory below this
3381 physical address is ignored.
3383 mini2440= [ARM,HW,KNL]
3384 Format:[0..2][b][c][t]
3386 MINI2440 configuration specification:
3387 0 - The attached screen is the 3.5" TFT
3388 1 - The attached screen is the 7" TFT
3389 2 - The VGA Shield is attached (1024x768)
3390 Leaving out the screen size parameter will not load
3391 the TFT driver, and the framebuffer will be left
3393 b - Enable backlight. The TFT backlight pin will be
3394 linked to the kernel VESA blanking code and a GPIO
3395 LED. This parameter is not necessary when using the
3397 c - Enable the s3c camera interface.
3398 t - Reserved for enabling touchscreen support. The
3399 touchscreen support is not enabled in the mainstream
3400 kernel as of 2.6.30, a preliminary port can be found
3401 in the "bleeding edge" mini2440 support kernel at
3402 https://repo.or.cz/w/linux-2.6/mini2440.git
3405 [X86,PPC,S390,ARM64,EARLY] Control optional mitigations for
3406 CPU vulnerabilities. This is a set of curated,
3407 arch-independent options, each of which is an
3408 aggregation of existing arch-specific options.
3411 Disable all optional CPU mitigations. This
3412 improves system performance, but it may also
3413 expose users to several CPU vulnerabilities.
3414 Equivalent to: if nokaslr then kpti=0 [ARM64]
3415 gather_data_sampling=off [X86]
3416 kvm.nx_huge_pages=off [X86]
3419 mmio_stale_data=off [X86]
3420 no_entry_flush [PPC]
3421 no_uaccess_flush [PPC]
3424 nospectre_bhb [ARM64]
3425 nospectre_v1 [X86,PPC]
3426 nospectre_v2 [X86,PPC,S390,ARM64]
3427 reg_file_data_sampling=off [X86]
3429 spec_rstack_overflow=off [X86]
3430 spec_store_bypass_disable=off [X86,PPC]
3431 spectre_v2_user=off [X86]
3432 srbds=off [X86,INTEL]
3433 ssbd=force-off [ARM64]
3434 tsx_async_abort=off [X86]
3437 This does not have any effect on
3438 kvm.nx_huge_pages when
3439 kvm.nx_huge_pages=force.
3442 Mitigate all CPU vulnerabilities, but leave SMT
3443 enabled, even if it's vulnerable. This is for
3444 users who don't want to be surprised by SMT
3445 getting disabled across kernel upgrades, or who
3446 have other ways of avoiding SMT-based attacks.
3447 Equivalent to: (default behavior)
3450 Mitigate all CPU vulnerabilities, disabling SMT
3451 if needed. This is for users who always want to
3452 be fully mitigated, even if it means losing SMT.
3453 Equivalent to: l1tf=flush,nosmt [X86]
3454 mds=full,nosmt [X86]
3455 tsx_async_abort=full,nosmt [X86]
3456 mmio_stale_data=full,nosmt [X86]
3457 retbleed=auto,nosmt [X86]
3460 [KNL,EARLY] When CONFIG_DEBUG_MEMORY_INIT is set, this
3461 parameter allows control of the logging verbosity for
3462 the additional memory initialisation checks. A value
3463 of 0 disables mminit logging and a level of 4 will
3464 log everything. Information is printed at KERN_DEBUG
3465 so loglevel=8 may also need to be specified.
3468 [X86,INTEL,EARLY] Control mitigation for the Processor
3469 MMIO Stale Data vulnerabilities.
3471 Processor MMIO Stale Data is a class of
3472 vulnerabilities that may expose data after an MMIO
3473 operation. Exposed data could originate or end in
3474 the same CPU buffers as affected by MDS and TAA.
3475 Therefore, similar to MDS and TAA, the mitigation
3476 is to clear the affected CPU buffers.
3478 This parameter controls the mitigation. The
3481 full - Enable mitigation on vulnerable CPUs
3483 full,nosmt - Enable mitigation and disable SMT on
3486 off - Unconditionally disable mitigation
3488 On MDS or TAA affected machines,
3489 mmio_stale_data=off can be prevented by an active
3490 MDS or TAA mitigation as these vulnerabilities are
3491 mitigated with the same mechanism so in order to
3492 disable this mitigation, you need to specify
3493 mds=off and tsx_async_abort=off too.
3495 Not specifying this option is equivalent to
3496 mmio_stale_data=full.
3499 Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst
3501 <module>.async_probe[=<bool>] [KNL]
3502 If no <bool> value is specified or if the value
3503 specified is not a valid <bool>, enable asynchronous
3504 probe on this module. Otherwise, enable/disable
3505 asynchronous probe on this module as indicated by the
3506 <bool> value. See also: module.async_probe
3508 module.async_probe=<bool>
3509 [KNL] When set to true, modules will use async probing
3510 by default. To enable/disable async probing for a
3511 specific module, use the module specific control that
3512 is documented under <module>.async_probe. When both
3513 module.async_probe and <module>.async_probe are
3514 specified, <module>.async_probe takes precedence for
3515 the specific module.
3517 module.enable_dups_trace
3518 [KNL] When CONFIG_MODULE_DEBUG_AUTOLOAD_DUPS is set,
3519 this means that duplicate request_module() calls will
3520 trigger a WARN_ON() instead of a pr_warn(). Note that
3521 if MODULE_DEBUG_AUTOLOAD_DUPS_TRACE is set, WARN_ON()s
3522 will always be issued and this option does nothing.
3524 [KNL] When CONFIG_MODULE_SIG is set, this means that
3525 modules without (valid) signatures will fail to load.
3526 Note that if CONFIG_MODULE_SIG_FORCE is set, that
3527 is always true, so this option does nothing.
3529 module_blacklist= [KNL] Do not load a comma-separated list of
3530 modules. Useful for debugging problem modules.
3533 [MOUSE] Maximum time between finger touching and
3534 leaving touchpad surface for touch to be considered
3535 a tap and be reported as a left button click (for
3536 touchpads working in absolute mode only).
3538 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
3539 reporting absolute coordinates, such as tablets
3540 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
3541 reporting absolute coordinates, such as tablets
3543 movablecore= [KNL,X86,IA-64,PPC,EARLY]
3544 Format: nn[KMGTPE] | nn%
3545 This parameter is the complement to kernelcore=, it
3546 specifies the amount of memory used for migratable
3547 allocations. If both kernelcore and movablecore is
3548 specified, then kernelcore will be at *least* the
3549 specified value but may be more. If movablecore on its
3550 own is specified, the administrator must be careful
3551 that the amount of memory usable for all allocations
3554 movable_node [KNL,EARLY] Boot-time switch to make hotplugable memory
3555 NUMA nodes to be movable. This means that the memory
3556 of such nodes will be usable only for movable
3557 allocations which rules out almost all kernel
3558 allocations. Use with caution!
3560 MTD_Partition= [MTD]
3561 Format: <name>,<region-number>,<size>,<offset>
3563 MTD_Region= [MTD] Format:
3564 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
3567 See drivers/mtd/parsers/cmdlinepart.c
3570 ARM/S3C2412 JIVE boot control
3572 See arch/arm/mach-s3c/mach-jive.c
3574 mtouchusb.raw_coordinates=
3575 [HW] Make the MicroTouch USB driver use raw coordinates
3576 ('y', default) or cooked coordinates ('n')
3578 mtrr=debug [X86,EARLY]
3579 Enable printing debug information related to MTRR
3580 registers at boot time.
3582 mtrr_chunk_size=nn[KMG,X86,EARLY]
3583 used for mtrr cleanup. It is largest continuous chunk
3584 that could hold holes aka. UC entries.
3586 mtrr_gran_size=nn[KMG,X86,EARLY]
3587 Used for mtrr cleanup. It is granularity of mtrr block.
3589 Large value could prevent small alignment from
3592 mtrr_spare_reg_nr=n [X86,EARLY]
3594 Range: 0,7 : spare reg number
3596 Used for mtrr cleanup. It is spare mtrr entries number.
3597 Set to 2 or more if your graphical card needs more.
3599 multitce=off [PPC] This parameter disables the use of the pSeries
3600 firmware feature for updating multiple TCE entries
3603 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
3605 netdev= [NET] Network devices parameters
3606 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
3607 Note that mem_start is often overloaded to mean
3608 something different and driver-specific.
3609 This usage is only documented in each driver source
3612 netpoll.carrier_timeout=
3613 [NET] Specifies amount of time (in seconds) that
3614 netpoll should wait for a carrier. By default netpoll
3618 [NETFILTER] Enable connection tracking flow accounting
3619 0 to disable accounting
3620 1 to enable accounting
3624 [NFS] sets the pathname to the program which is used
3625 to update the NFS client cache entries.
3627 nfs.cache_getent_timeout=
3628 [NFS] sets the timeout after which an attempt to
3629 update a cache entry is deemed to have failed.
3631 nfs.callback_nr_threads=
3632 [NFSv4] set the total number of threads that the
3633 NFS client will assign to service NFSv4 callback
3636 nfs.callback_tcpport=
3637 [NFS] set the TCP port on which the NFSv4 callback
3638 channel should listen.
3641 [NFS] specifies the number of times the NFSv4 client
3642 retries the request before returning an EAGAIN error,
3643 after a reply of NFS4ERR_DELAY from the server.
3644 Only applies if the softerr mount option is enabled,
3645 and the specified value is >= 0.
3648 [NFS] enable 64-bit inode numbers.
3649 If zero, the NFS client will fake up a 32-bit inode
3650 number for the readdir() and stat() syscalls instead
3651 of returning the full 64-bit number.
3652 The default is to return 64-bit inode numbers.
3654 nfs.idmap_cache_timeout=
3655 [NFS] set the maximum lifetime for idmapper cache
3658 nfs.max_session_cb_slots=
3659 [NFSv4.1] Sets the maximum number of session
3660 slots the client will assign to the callback
3661 channel. This determines the maximum number of
3662 callbacks the client will process in parallel for
3663 a particular server.
3665 nfs.max_session_slots=
3666 [NFSv4.1] Sets the maximum number of session slots
3667 the client will attempt to negotiate with the server.
3668 This limits the number of simultaneous RPC requests
3669 that the client can send to the NFSv4.1 server.
3670 Note that there is little point in setting this
3671 value higher than the max_tcp_slot_table_limit.
3673 nfs.nfs4_disable_idmapping=
3674 [NFSv4] When set to the default of '1', this option
3675 ensures that both the RPC level authentication
3676 scheme and the NFS level operations agree to use
3677 numeric uids/gids if the mount is using the
3678 'sec=sys' security flavour. In effect it is
3679 disabling idmapping, which can make migration from
3680 legacy NFSv2/v3 systems to NFSv4 easier.
3681 Servers that do not support this mode of operation
3682 will be autodetected by the client, and it will fall
3683 back to using the idmapper.
3684 To turn off this behaviour, set the value to '0'.
3687 [NFS4] Specify an additional fixed unique ident-
3688 ification string that NFSv4 clients can insert into
3689 their nfs_client_id4 string. This is typically a
3690 UUID that is generated at system install time.
3692 nfs.recover_lost_locks=
3693 [NFSv4] Attempt to recover locks that were lost due
3694 to a lease timeout on the server. Please note that
3695 doing this risks data corruption, since there are
3696 no guarantees that the file will remain unchanged
3697 after the locks are lost.
3698 If you want to enable the kernel legacy behaviour of
3699 attempting to recover these locks, then set this
3701 The default parameter value of '0' causes the kernel
3702 not to attempt recovery of lost locks.
3704 nfs.send_implementation_id=
3705 [NFSv4.1] Send client implementation identification
3706 information in exchange_id requests.
3707 If zero, no implementation identification information
3709 The default is to send the implementation identification
3712 nfs4.layoutstats_timer=
3713 [NFSv4.2] Change the rate at which the kernel sends
3714 layoutstats to the pNFS metadata server.
3716 Setting this to value to 0 causes the kernel to use
3717 whatever value is the default set by the layout
3718 driver. A non-zero value sets the minimum interval
3719 in seconds between layoutstats transmissions.
3721 nfsd.inter_copy_offload_enable=
3722 [NFSv4.2] When set to 1, the server will support
3723 server-to-server copies for which this server is
3724 the destination of the copy.
3726 nfsd.nfs4_disable_idmapping=
3727 [NFSv4] When set to the default of '1', the NFSv4
3728 server will return only numeric uids and gids to
3729 clients using auth_sys, and will accept numeric uids
3730 and gids from such clients. This is intended to ease
3731 migration from NFSv2/v3.
3733 nfsd.nfsd4_ssc_umount_timeout=
3734 [NFSv4.2] When used as the destination of a
3735 server-to-server copy, knfsd temporarily mounts
3736 the source server. It caches the mount in case
3737 it will be needed again, and discards it if not
3738 used for the number of milliseconds specified by
3741 nfsaddrs= [NFS] Deprecated. Use ip= instead.
3742 See Documentation/admin-guide/nfs/nfsroot.rst.
3744 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
3745 See Documentation/admin-guide/nfs/nfsroot.rst.
3747 nfsrootdebug [NFS] enable nfsroot debugging messages.
3748 See Documentation/admin-guide/nfs/nfsroot.rst.
3750 nmi_backtrace.backtrace_idle [KNL]
3751 Dump stacks even of idle CPUs in response to an
3752 NMI stack-backtrace request.
3754 nmi_debug= [KNL,SH] Specify one or more actions to take
3755 when a NMI is triggered.
3756 Format: [state][,regs][,debounce][,die]
3758 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
3759 Format: [panic,][nopanic,][num]
3761 0 - turn hardlockup detector in nmi_watchdog off
3762 1 - turn hardlockup detector in nmi_watchdog on
3763 When panic is specified, panic when an NMI watchdog
3764 timeout occurs (or 'nopanic' to not panic on an NMI
3765 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
3766 To disable both hard and soft lockup detectors,
3767 please see 'nowatchdog'.
3768 This is useful when you use a panic=... timeout and
3769 need the box quickly up again.
3771 These settings can be accessed at runtime via
3772 the nmi_watchdog and hardlockup_panic sysctls.
3774 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3775 emulation library even if a 387 maths coprocessor
3778 no4lvl [RISCV,EARLY] Disable 4-level and 5-level paging modes.
3779 Forces kernel to use 3-level paging instead.
3781 no5lvl [X86-64,RISCV,EARLY] Disable 5-level paging mode. Forces
3782 kernel to use 4-level paging instead.
3786 noaltinstr [S390,EARLY] Disables alternative instructions
3787 patching (CPU alternatives feature).
3789 noapic [SMP,APIC,EARLY] Tells the kernel to not make use of any
3790 IOAPICs that may be present in the system.
3792 noautogroup Disable scheduler automatic task group creation.
3797 [HW] Never suspend the console
3798 Disable suspending of consoles during suspend and
3799 hibernate operations. Once disabled, debugging
3800 messages can reach various consoles while the rest
3801 of the system is being put to sleep (ie, while
3802 debugging driver suspend/resume hooks). This may
3803 not work reliably with all consoles, but is known
3804 to work with serial and VGA consoles.
3805 To facilitate more flexible debugging, we also add
3806 console_suspend, a printk module parameter to control
3807 it. Users could use console_suspend (usually
3808 /sys/module/printk/parameters/console_suspend) to
3809 turn on/off it dynamically.
3812 [KNL,EARLY] Disable object debugging
3814 nodsp [SH] Disable hardware DSP at boot time.
3816 noefi [EFI,EARLY] Disable EFI runtime services support.
3818 no_entry_flush [PPC,EARLY] Don't flush the L1-D cache when entering the kernel.
3823 This affects only 32-bit executables.
3824 noexec32=on: enable non-executable mappings (default)
3825 read doesn't imply executable mappings
3826 noexec32=off: disable non-executable mappings
3827 read implies executable mappings
3829 no_file_caps Tells the kernel not to honor file capabilities. The
3830 only way then for a file to be executed with privilege
3831 is to be setuid root or executed by root.
3833 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3835 nofsgsbase [X86] Disables FSGSBASE instructions.
3837 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3838 register save and restore. The kernel will only save
3839 legacy floating-point registers on task switch.
3841 nohalt [IA-64] Tells the kernel not to use the power saving
3842 function PAL_HALT_LIGHT when idle. This increases
3843 power-consumption. On the positive side, it reduces
3844 interrupt wake-up latency, which may improve performance
3845 in certain environments such as networked servers or
3850 Force pointers printed to the console or buffers to be
3851 unhashed. By default, when a pointer is printed via %p
3852 format string, that pointer is "hashed", i.e. obscured
3853 by hashing the pointer value. This is a security feature
3854 that hides actual kernel addresses from unprivileged
3855 users, but it also makes debugging the kernel more
3856 difficult since unequal pointers can no longer be
3857 compared. However, if this command-line option is
3858 specified, then all normal pointers will have their true
3859 value printed. This option should only be specified when
3860 debugging the kernel. Please do not use on production
3863 nohibernate [HIBERNATION] Disable hibernation and resume.
3865 nohlt [ARM,ARM64,MICROBLAZE,MIPS,PPC,SH] Forces the kernel to
3866 busy wait in do_idle() and not use the arch_cpu_idle()
3867 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
3868 to be effective. This is useful on platforms where the
3869 sleep(SH) or wfi(ARM,ARM64) instructions do not work
3870 correctly or when doing power measurements to evaluate
3871 the impact of the sleep instructions. This is also
3872 useful when using JTAG debugger.
3874 nohugeiomap [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge I/O mappings.
3876 nohugevmalloc [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge vmalloc mappings.
3878 nohz= [KNL] Boottime enable/disable dynamic ticks
3879 Valid arguments: on, off
3882 nohz_full= [KNL,BOOT,SMP,ISOL]
3883 The argument is a cpu list, as described above.
3884 In kernels built with CONFIG_NO_HZ_FULL=y, set
3885 the specified list of CPUs whose tick will be stopped
3886 whenever possible. The boot CPU will be forced outside
3887 the range to maintain the timekeeping. Any CPUs
3888 in this list will have their RCU callbacks offloaded,
3889 just as if they had also been called out in the
3890 rcu_nocbs= boot parameter.
3892 Note that this argument takes precedence over
3893 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option.
3895 noinitrd [RAM] Tells the kernel not to load any configured
3898 nointremap [X86-64,Intel-IOMMU,EARLY] Do not enable interrupt
3900 [Deprecated - use intremap=off]
3904 noinvpcid [X86,EARLY] Disable the INVPCID cpu feature.
3906 noiotrap [SH] Disables trapped I/O port accesses.
3908 noirqdebug [X86-32] Disables the code which attempts to detect and
3909 disable unhandled interrupt sources.
3911 noisapnp [ISAPNP] Disables ISA PnP code.
3913 nojitter [IA-64] Disables jitter checking for ITC timers.
3916 When CONFIG_RANDOMIZE_BASE is set, this disables
3917 kernel and module base offset ASLR (Address Space
3918 Layout Randomization).
3920 no-kvmapf [X86,KVM,EARLY] Disable paravirtualized asynchronous page
3923 no-kvmclock [X86,KVM,EARLY] Disable paravirtualized KVM clock driver
3925 nolapic [X86-32,APIC,EARLY] Do not enable or use the local APIC.
3927 nolapic_timer [X86-32,APIC,EARLY] Do not use the local APIC timer.
3929 nomca [IA-64] Disable machine check abort handling
3931 nomce [X86-32] Disable Machine Check Exception
3933 nomfgpt [X86-32] Disable Multi-Function General Purpose
3934 Timer usage (for AMD Geode machines).
3936 nomodeset Disable kernel modesetting. Most systems' firmware
3937 sets up a display mode and provides framebuffer memory
3938 for output. With nomodeset, DRM and fbdev drivers will
3939 not load if they could possibly displace the pre-
3940 initialized output. Only the system framebuffer will
3941 be available for use. The respective drivers will not
3942 perform display-mode changes or accelerated rendering.
3944 Useful as error fallback, or for testing and debugging.
3946 nomodule Disable module load
3948 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3949 shutdown the other cpus. Instead use the REBOOT_VECTOR
3952 nopat [X86,EARLY] Disable PAT (page attribute table extension of
3953 pagetables) support.
3955 nopcid [X86-64,EARLY] Disable the PCID cpu feature.
3957 nopku [X86] Disable Memory Protection Keys CPU feature found
3960 nopti [X86-64,EARLY]
3961 Equivalent to pti=off
3963 nopv= [X86,XEN,KVM,HYPER_V,VMWARE,EARLY]
3964 Disables the PV optimizations forcing the guest to run
3965 as generic guest with no PV drivers. Currently support
3966 XEN HVM, KVM, HYPER_V and VMWARE guest.
3968 nopvspin [X86,XEN,KVM,EARLY]
3969 Disables the qspinlock slow path using PV optimizations
3970 which allow the hypervisor to 'idle' the guest on lock
3973 norandmaps Don't use address space randomization. Equivalent to
3974 echo 0 > /proc/sys/kernel/randomize_va_space
3976 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
3977 with UP alternatives
3979 noresume [SWSUSP] Disables resume and restores original swap
3984 no-scroll [VGA] Disables scrollback.
3985 This is required for the Braillex ib80-piezo Braille
3986 reader made by F.H. Papenmeier (Germany).
3988 nosgx [X86-64,SGX,EARLY] Disables Intel SGX kernel support.
3991 Disable SMAP (Supervisor Mode Access Prevention)
3992 even if it is supported by processor.
3994 nosmep [PPC64s,EARLY]
3995 Disable SMEP (Supervisor Mode Execution Prevention)
3996 even if it is supported by processor.
3998 nosmp [SMP,EARLY] Tells an SMP kernel to act as a UP kernel,
3999 and disable the IO APIC. legacy for "maxcpus=0".
4001 nosmt [KNL,MIPS,PPC,S390,EARLY] Disable symmetric multithreading (SMT).
4002 Equivalent to smt=1.
4004 [KNL,X86,PPC] Disable symmetric multithreading (SMT).
4005 nosmt=force: Force disable SMT, cannot be undone
4006 via the sysfs control file.
4008 nosoftlockup [KNL] Disable the soft-lockup detector.
4010 nospec_store_bypass_disable
4011 [HW,EARLY] Disable all mitigations for the Speculative
4012 Store Bypass vulnerability
4014 nospectre_bhb [ARM64,EARLY] Disable all mitigations for Spectre-BHB (branch
4015 history injection) vulnerability. System may allow data leaks
4018 nospectre_v1 [X86,PPC,EARLY] Disable mitigations for Spectre Variant 1
4019 (bounds check bypass). With this option data leaks are
4020 possible in the system.
4022 nospectre_v2 [X86,PPC_E500,ARM64,EARLY] Disable all mitigations
4023 for the Spectre variant 2 (indirect branch
4024 prediction) vulnerability. System may allow data
4025 leaks with this option.
4027 no-steal-acc [X86,PV_OPS,ARM64,PPC/PSERIES,RISCV,EARLY] Disable
4028 paravirtualized steal time accounting. steal time is
4029 computed, but won't influence scheduler behaviour
4031 nosync [HW,M68K] Disables sync negotiation for all devices.
4033 no_timer_check [X86,APIC] Disables the code which tests for
4034 broken timer IRQ sources.
4037 [PPC,EARLY] Don't flush the L1-D cache after accessing user data.
4039 novmcoredd [KNL,KDUMP]
4040 Disable device dump. Device dump allows drivers to
4041 append dump data to vmcore so you can collect driver
4042 specified debug info. Drivers can append the data
4043 without any limit and this data is stored in memory,
4044 so this may cause significant memory stress. Disabling
4045 device dump can help save memory but the driver debug
4046 data will be no longer available. This parameter
4047 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
4051 [X86,PV_OPS,EARLY] Disable paravirtualized VMware
4052 scheduler clock and use the default one.
4054 nowatchdog [KNL] Disable both lockup detectors, i.e.
4055 soft-lockup and NMI watchdog (hard-lockup).
4059 nox2apic [X86-64,APIC,EARLY] Do not enable x2APIC mode.
4061 NOTE: this parameter will be ignored on systems with the
4062 LEGACY_XAPIC_DISABLED bit set in the
4063 IA32_XAPIC_DISABLE_STATUS MSR.
4065 noxsave [BUGS=X86] Disables x86 extended register state save
4066 and restore using xsave. The kernel will fallback to
4067 enabling legacy floating-point and sse state.
4069 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
4070 register states. The kernel will fall back to use
4071 xsave to save the states. By using this parameter,
4072 performance of saving the states is degraded because
4073 xsave doesn't support modified optimization while
4074 xsaveopt supports it on xsaveopt enabled systems.
4076 noxsaves [X86] Disables xsaves and xrstors used in saving and
4077 restoring x86 extended register state in compacted
4078 form of xsave area. The kernel will fall back to use
4079 xsaveopt and xrstor to save and restore the states
4080 in standard form of xsave area. By using this
4081 parameter, xsave area per process might occupy more
4082 memory on xsaves enabled systems.
4084 nps_mtm_hs_ctr= [KNL,ARC]
4085 This parameter sets the maximum duration, in
4086 cycles, each HW thread of the CTOP can run
4087 without interruptions, before HW switches it.
4088 The actual maximum duration is 16 times this
4090 Format: integer between 1 and 255
4093 nptcg= [IA-64] Override max number of concurrent global TLB
4094 purges which is reported from either PAL_VM_SUMMARY or
4097 nr_cpus= [SMP,EARLY] Maximum number of processors that an SMP kernel
4098 could support. nr_cpus=n : n >= 1 limits the kernel to
4099 support 'n' processors. It could be larger than the
4100 number of already plugged CPU during bootup, later in
4101 runtime you can physically add extra cpu until it reaches
4102 n. So during boot up some boot time memory for per-cpu
4103 variables need be pre-allocated for later physical cpu
4106 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
4108 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86, EARLY]
4109 Disable NUMA, Only set up a single NUMA node
4110 spanning all memory.
4112 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic
4114 Allowed values are enable and disable
4116 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
4117 'node', 'default' can be specified
4118 This can be set from sysctl after boot.
4119 See Documentation/admin-guide/sysctl/vm.rst for details.
4121 ohci1394_dma=early [HW,EARLY] enable debugging via the ohci1394 driver.
4122 See Documentation/core-api/debugging-via-ohci1394.rst for more
4125 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
4126 Rather than timing out after 20 ms if an EC
4127 command is not properly ACKed, override the length
4128 of the timeout. We have interrupts disabled while
4129 waiting for the ACK, so if this is set too high
4130 interrupts *may* be lost!
4132 omap_mux= [OMAP] Override bootloader pin multiplexing.
4133 Format: <mux_mode0.mode_name=value>...
4134 For example, to override I2C bus2:
4135 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
4137 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
4139 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
4141 boundary - index of last SLC block on Flex-OneNAND.
4142 The remaining blocks are configured as MLC blocks.
4143 lock - Configure if Flex-OneNAND boundary should be locked.
4144 Once locked, the boundary cannot be changed.
4145 1 indicates lock status, 0 indicates unlock status.
4147 oops=panic [KNL,EARLY]
4148 Always panic on oopses. Default is to just kill the
4149 process, but there is a small probability of
4150 deadlocking the machine.
4151 This will also cause panics on machine check exceptions.
4152 Useful together with panic=30 to trigger a reboot.
4155 [KNL] Boolean flag to control whether the page allocator
4156 should randomize its free lists. The randomization may
4157 be automatically enabled if the kernel detects it is
4158 running on a platform with a direct-mapped memory-side
4159 cache, and this parameter can be used to
4160 override/disable that behavior. The state of the flag
4161 can be read from sysfs at:
4162 /sys/module/page_alloc/parameters/shuffle.
4164 page_owner= [KNL,EARLY] Boot-time page_owner enabling option.
4165 Storage of the information about who allocated
4166 each page is disabled in default. With this switch,
4168 on: enable the feature
4170 page_poison= [KNL,EARLY] Boot-time parameter changing the state of
4171 poisoning on the buddy allocator, available with
4172 CONFIG_PAGE_POISONING=y.
4173 off: turn off poisoning (default)
4174 on: turn on poisoning
4176 page_reporting.page_reporting_order=
4177 [KNL] Minimal page reporting order
4179 Adjust the minimal page reporting order. The page
4180 reporting is disabled when it exceeds MAX_PAGE_ORDER.
4182 panic= [KNL] Kernel behaviour on panic: delay <timeout>
4183 timeout > 0: seconds before rebooting
4184 timeout = 0: wait forever
4185 timeout < 0: reboot immediately
4188 panic_on_taint= [KNL,EARLY]
4189 Bitmask for conditionally calling panic() in add_taint()
4190 Format: <hex>[,nousertaint]
4191 Hexadecimal bitmask representing the set of TAINT flags
4192 that will cause the kernel to panic when add_taint() is
4193 called with any of the flags in this set.
4194 The optional switch "nousertaint" can be utilized to
4195 prevent userspace forced crashes by writing to sysctl
4196 /proc/sys/kernel/tainted any flagset matching with the
4197 bitmask set on panic_on_taint.
4198 See Documentation/admin-guide/tainted-kernels.rst for
4199 extra details on the taint flags that users can pick
4200 to compose the bitmask to assign to panic_on_taint.
4202 panic_on_warn=1 panic() instead of WARN(). Useful to cause kdump
4205 panic_print= Bitmask for printing system info when panic happens.
4206 User can chose combination of the following bits:
4207 bit 0: print all tasks info
4208 bit 1: print system memory info
4209 bit 2: print timer info
4210 bit 3: print locks info if CONFIG_LOCKDEP is on
4211 bit 4: print ftrace buffer
4212 bit 5: print all printk messages in buffer
4213 bit 6: print all CPUs backtrace (if available in the arch)
4214 bit 7: print only tasks in uninterruptible (blocked) state
4215 *Be aware* that this option may print a _lot_ of lines,
4216 so there are risks of losing older messages in the log.
4217 Use this option carefully, maybe worth to setup a
4218 bigger log buffer with "log_buf_len" along with this.
4220 parkbd.port= [HW] Parallel port number the keyboard adapter is
4221 connected to, default is 0.
4223 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
4224 0 for XT, 1 for AT (default is AT).
4227 parport= [HW,PPT] Specify parallel ports. 0 disables.
4228 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
4229 Use 'auto' to force the driver to use any
4230 IRQ/DMA settings detected (the default is to
4231 ignore detected IRQ/DMA settings because of
4232 possible conflicts). You can specify the base
4233 address, IRQ, and DMA settings; IRQ and DMA
4234 should be numbers, or 'auto' (for using detected
4235 settings on that particular port), or 'nofifo'
4236 (to avoid using a FIFO even if it is detected).
4237 Parallel ports are assigned in the order they
4238 are specified on the command line, starting
4241 parport_init_mode= [HW,PPT]
4242 Configure VIA parallel port to operate in
4243 a specific mode. This is necessary on Pegasos
4244 computer where firmware has no options for setting
4245 up parallel port mode and sets it to spp.
4246 Currently this function knows 686a and 8231 chips.
4247 Format: [spp|ps2|epp|ecp|ecpepp]
4249 pata_legacy.all= [HW,LIBATA]
4251 Set to non-zero to probe primary and secondary ISA
4252 port ranges on PCI systems where no PCI PATA device
4253 has been found at either range. Disabled by default.
4255 pata_legacy.autospeed= [HW,LIBATA]
4257 Set to non-zero if a chip is present that snoops speed
4258 changes. Disabled by default.
4260 pata_legacy.ht6560a= [HW,LIBATA]
4262 Set to 1, 2, or 3 for HT 6560A on the primary channel,
4263 the secondary channel, or both channels respectively.
4264 Disabled by default.
4266 pata_legacy.ht6560b= [HW,LIBATA]
4268 Set to 1, 2, or 3 for HT 6560B on the primary channel,
4269 the secondary channel, or both channels respectively.
4270 Disabled by default.
4272 pata_legacy.iordy_mask= [HW,LIBATA]
4274 IORDY enable mask. Set individual bits to allow IORDY
4275 for the respective channel. Bit 0 is for the first
4276 legacy channel handled by this driver, bit 1 is for
4277 the second channel, and so on. The sequence will often
4278 correspond to the primary legacy channel, the secondary
4279 legacy channel, and so on, but the handling of a PCI
4280 bus and the use of other driver options may interfere
4281 with the sequence. By default IORDY is allowed across
4284 pata_legacy.opti82c46x= [HW,LIBATA]
4286 Set to 1, 2, or 3 for Opti 82c611A on the primary
4287 channel, the secondary channel, or both channels
4288 respectively. Disabled by default.
4290 pata_legacy.opti82c611a= [HW,LIBATA]
4292 Set to 1, 2, or 3 for Opti 82c465MV on the primary
4293 channel, the secondary channel, or both channels
4294 respectively. Disabled by default.
4296 pata_legacy.pio_mask= [HW,LIBATA]
4298 PIO mode mask for autospeed devices. Set individual
4299 bits to allow the use of the respective PIO modes.
4300 Bit 0 is for mode 0, bit 1 is for mode 1, and so on.
4301 All modes allowed by default.
4303 pata_legacy.probe_all= [HW,LIBATA]
4305 Set to non-zero to probe tertiary and further ISA
4306 port ranges on PCI systems. Disabled by default.
4308 pata_legacy.probe_mask= [HW,LIBATA]
4310 Probe mask for legacy ISA PATA ports. Depending on
4311 platform configuration and the use of other driver
4312 options up to 6 legacy ports are supported: 0x1f0,
4313 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing
4314 of individual ports can be disabled by setting the
4315 corresponding bits in the mask to 1. Bit 0 is for
4316 the first port in the list above (0x1f0), and so on.
4317 By default all supported ports are probed.
4319 pata_legacy.qdi= [HW,LIBATA]
4321 Set to non-zero to probe QDI controllers. By default
4322 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise.
4324 pata_legacy.winbond= [HW,LIBATA]
4326 Set to non-zero to probe Winbond controllers. Use
4327 the standard I/O port (0x130) if 1, otherwise the
4328 value given is the I/O port to use (typically 0x1b0).
4329 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE,
4332 pata_platform.pio_mask= [HW,LIBATA]
4334 Supported PIO mode mask. Set individual bits to allow
4335 the use of the respective PIO modes. Bit 0 is for
4336 mode 0, bit 1 is for mode 1, and so on. Mode 0 only
4340 Halt all CPUs after the first oops has been printed for
4341 the specified number of seconds. This is to be used if
4342 your oopses keep scrolling off the screen.
4346 pci=option[,option...] [PCI,EARLY] various PCI subsystem options.
4348 Some options herein operate on a specific device
4349 or a set of devices (<pci_dev>). These are
4350 specified in one of the following formats:
4352 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
4353 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
4355 Note: the first format specifies a PCI
4356 bus/device/function address which may change
4357 if new hardware is inserted, if motherboard
4358 firmware changes, or due to changes caused
4359 by other kernel parameters. If the
4360 domain is left unspecified, it is
4361 taken to be zero. Optionally, a path
4362 to a device through multiple device/function
4363 addresses can be specified after the base
4364 address (this is more robust against
4365 renumbering issues). The second format
4366 selects devices using IDs from the
4367 configuration space which may match multiple
4368 devices in the system.
4370 earlydump dump PCI config space before the kernel
4372 off [X86] don't probe for the PCI bus
4373 bios [X86-32] force use of PCI BIOS, don't access
4374 the hardware directly. Use this if your machine
4375 has a non-standard PCI host bridge.
4376 nobios [X86-32] disallow use of PCI BIOS, only direct
4377 hardware access methods are allowed. Use this
4378 if you experience crashes upon bootup and you
4379 suspect they are caused by the BIOS.
4380 conf1 [X86] Force use of PCI Configuration Access
4381 Mechanism 1 (config address in IO port 0xCF8,
4382 data in IO port 0xCFC, both 32-bit).
4383 conf2 [X86] Force use of PCI Configuration Access
4384 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
4385 the function, IO port 0xCFA, also 8-bit, sets
4386 bus number. The config space is then accessed
4387 through ports 0xC000-0xCFFF).
4388 See http://wiki.osdev.org/PCI for more info
4389 on the configuration access mechanisms.
4390 noaer [PCIE] If the PCIEAER kernel config parameter is
4391 enabled, this kernel boot option can be used to
4392 disable the use of PCIE advanced error reporting.
4393 nodomains [PCI] Disable support for multiple PCI
4394 root domains (aka PCI segments, in ACPI-speak).
4395 nommconf [X86] Disable use of MMCONFIG for PCI
4397 check_enable_amd_mmconf [X86] check for and enable
4398 properly configured MMIO access to PCI
4399 config space on AMD family 10h CPU
4400 nomsi [MSI] If the PCI_MSI kernel config parameter is
4401 enabled, this kernel boot option can be used to
4402 disable the use of MSI interrupts system-wide.
4403 noioapicquirk [APIC] Disable all boot interrupt quirks.
4404 Safety option to keep boot IRQs enabled. This
4405 should never be necessary.
4406 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
4407 primary IO-APIC for bridges that cannot disable
4408 boot IRQs. This fixes a source of spurious IRQs
4409 when the system masks IRQs.
4410 noioapicreroute [APIC] Disable workaround that uses the
4411 boot IRQ equivalent of an IRQ that connects to
4412 a chipset where boot IRQs cannot be disabled.
4413 The opposite of ioapicreroute.
4414 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
4415 routing table. These calls are known to be buggy
4416 on several machines and they hang the machine
4417 when used, but on other computers it's the only
4418 way to get the interrupt routing table. Try
4419 this option if the kernel is unable to allocate
4420 IRQs or discover secondary PCI buses on your
4422 rom [X86] Assign address space to expansion ROMs.
4423 Use with caution as certain devices share
4424 address decoders between ROMs and other
4426 norom [X86] Do not assign address space to
4427 expansion ROMs that do not already have
4428 BIOS assigned address ranges.
4429 nobar [X86] Do not assign address space to the
4430 BARs that weren't assigned by the BIOS.
4431 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
4432 assigned automatically to PCI devices. You can
4433 make the kernel exclude IRQs of your ISA cards
4435 pirqaddr=0xAAAAA [X86] Specify the physical address
4436 of the PIRQ table (normally generated
4437 by the BIOS) if it is outside the
4438 F0000h-100000h range.
4439 lastbus=N [X86] Scan all buses thru bus #N. Can be
4440 useful if the kernel is unable to find your
4441 secondary buses and you want to tell it
4442 explicitly which ones they are.
4443 assign-busses [X86] Always assign all PCI bus
4444 numbers ourselves, overriding
4445 whatever the firmware may have done.
4446 usepirqmask [X86] Honor the possible IRQ mask stored
4447 in the BIOS $PIR table. This is needed on
4448 some systems with broken BIOSes, notably
4449 some HP Pavilion N5400 and Omnibook XE3
4450 notebooks. This will have no effect if ACPI
4451 IRQ routing is enabled.
4452 noacpi [X86] Do not use ACPI for IRQ routing
4453 or for PCI scanning.
4454 use_crs [X86] Use PCI host bridge window information
4455 from ACPI. On BIOSes from 2008 or later, this
4456 is enabled by default. If you need to use this,
4457 please report a bug.
4458 nocrs [X86] Ignore PCI host bridge windows from ACPI.
4459 If you need to use this, please report a bug.
4460 use_e820 [X86] Use E820 reservations to exclude parts of
4461 PCI host bridge windows. This is a workaround
4462 for BIOS defects in host bridge _CRS methods.
4463 If you need to use this, please report a bug to
4464 <linux-pci@vger.kernel.org>.
4465 no_e820 [X86] Ignore E820 reservations for PCI host
4466 bridge windows. This is the default on modern
4467 hardware. If you need to use this, please report
4468 a bug to <linux-pci@vger.kernel.org>.
4469 routeirq Do IRQ routing for all PCI devices.
4470 This is normally done in pci_enable_device(),
4471 so this option is a temporary workaround
4472 for broken drivers that don't call it.
4473 skip_isa_align [X86] do not align io start addr, so can
4474 handle more pci cards
4475 noearly [X86] Don't do any early type 1 scanning.
4476 This might help on some broken boards which
4477 machine check when some devices' config space
4478 is read. But various workarounds are disabled
4479 and some IOMMU drivers will not work.
4480 bfsort Sort PCI devices into breadth-first order.
4481 This sorting is done to get a device
4482 order compatible with older (<= 2.4) kernels.
4483 nobfsort Don't sort PCI devices into breadth-first order.
4484 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
4485 tuning and use the BIOS-configured MPS defaults.
4486 pcie_bus_safe Set every device's MPS to the largest value
4487 supported by all devices below the root complex.
4488 pcie_bus_perf Set device MPS to the largest allowable MPS
4489 based on its parent bus. Also set MRRS (Max
4490 Read Request Size) to the largest supported
4491 value (no larger than the MPS that the device
4492 or bus can support) for best performance.
4493 pcie_bus_peer2peer Set every device's MPS to 128B, which
4494 every device is guaranteed to support. This
4495 configuration allows peer-to-peer DMA between
4496 any pair of devices, possibly at the cost of
4497 reduced performance. This also guarantees
4498 that hot-added devices will work.
4499 cbiosize=nn[KMG] The fixed amount of bus space which is
4500 reserved for the CardBus bridge's IO window.
4501 The default value is 256 bytes.
4502 cbmemsize=nn[KMG] The fixed amount of bus space which is
4503 reserved for the CardBus bridge's memory
4504 window. The default value is 64 megabytes.
4507 [<order of align>@]<pci_dev>[; ...]
4508 Specifies alignment and device to reassign
4509 aligned memory resources. How to
4510 specify the device is described above.
4511 If <order of align> is not specified,
4512 PAGE_SIZE is used as alignment.
4513 A PCI-PCI bridge can be specified if resource
4514 windows need to be expanded.
4515 To specify the alignment for several
4516 instances of a device, the PCI vendor,
4517 device, subvendor, and subdevice may be
4518 specified, e.g., 12@pci:8086:9c22:103c:198f
4519 for 4096-byte alignment.
4520 ecrc= Enable/disable PCIe ECRC (transaction layer
4521 end-to-end CRC checking). Only effective if
4522 OS has native AER control (either granted by
4523 ACPI _OSC or forced via "pcie_ports=native")
4524 bios: Use BIOS/firmware settings. This is the
4528 hpiosize=nn[KMG] The fixed amount of bus space which is
4529 reserved for hotplug bridge's IO window.
4530 Default size is 256 bytes.
4531 hpmmiosize=nn[KMG] The fixed amount of bus space which is
4532 reserved for hotplug bridge's MMIO window.
4533 Default size is 2 megabytes.
4534 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
4535 reserved for hotplug bridge's MMIO_PREF window.
4536 Default size is 2 megabytes.
4537 hpmemsize=nn[KMG] The fixed amount of bus space which is
4538 reserved for hotplug bridge's MMIO and
4540 Default size is 2 megabytes.
4541 hpbussize=nn The minimum amount of additional bus numbers
4542 reserved for buses below a hotplug bridge.
4544 realloc= Enable/disable reallocating PCI bridge resources
4545 if allocations done by BIOS are too small to
4546 accommodate resources required by all child
4548 off: Turn realloc off
4550 realloc same as realloc=on
4551 noari do not use PCIe ARI.
4552 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
4553 do not use PCIe ATS (and IOMMU device IOTLB).
4554 pcie_scan_all Scan all possible PCIe devices. Otherwise we
4555 only look for one device below a PCIe downstream
4557 big_root_window Try to add a big 64bit memory window to the PCIe
4558 root complex on AMD CPUs. Some GFX hardware
4559 can resize a BAR to allow access to all VRAM.
4560 Adding the window is slightly risky (it may
4561 conflict with unreported devices), so this
4563 disable_acs_redir=<pci_dev>[; ...]
4564 Specify one or more PCI devices (in the format
4565 specified above) separated by semicolons.
4566 Each device specified will have the PCI ACS
4567 redirect capabilities forced off which will
4568 allow P2P traffic between devices through
4569 bridges without forcing it upstream. Note:
4570 this removes isolation between devices and
4571 may put more devices in an IOMMU group.
4572 force_floating [S390] Force usage of floating interrupts.
4573 nomio [S390] Do not use MIO instructions.
4574 norid [S390] ignore the RID field and force use of
4575 one PCI domain per PCI function
4577 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
4580 force Enable ASPM even on devices that claim not to support it.
4581 WARNING: Forcing ASPM on may cause system lockups.
4583 pcie_ports= [PCIE] PCIe port services handling:
4584 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
4585 even if the platform doesn't give the OS permission to
4586 use them. This may cause conflicts if the platform
4587 also tries to use these services.
4588 dpc-native Use native PCIe service for DPC only. May
4589 cause conflicts if firmware uses AER or DPC.
4590 compat Disable native PCIe services (PME, AER, DPC, PCIe
4593 pcie_port_pm= [PCIE] PCIe port power management handling:
4594 off Disable power management of all PCIe ports
4595 force Forcibly enable power management of all PCIe ports
4597 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
4598 nomsi Do not use MSI for native PCIe PME signaling (this makes
4599 all PCIe root ports use INTx for all services).
4601 pcmv= [HW,PCMCIA] BadgePAD 4
4605 Keep all power-domains already enabled by bootloader on,
4606 even if no driver has claimed them. This is useful
4607 for debug and development, but should not be
4608 needed on a platform with proper driver support.
4610 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
4613 See arch/parisc/kernel/pdc_chassis.c
4615 percpu_alloc= [MM,EARLY]
4616 Select which percpu first chunk allocator to use.
4617 Currently supported values are "embed" and "page".
4618 Archs may support subset or none of the selections.
4619 See comments in mm/percpu.c for details on each
4620 allocator. This parameter is primarily for debugging
4621 and performance comparison.
4623 pirq= [SMP,APIC] Manual mp-table setup
4624 See Documentation/arch/x86/i386/IO-APIC.rst.
4626 plip= [PPT,NET] Parallel port network link
4627 Format: { parport<nr> | timid | 0 }
4628 See also Documentation/admin-guide/parport.rst.
4630 pmtmr= [X86] Manual setup of pmtmr I/O Port.
4631 Override pmtimer IOPort with a hex value.
4634 pmu_override= [PPC] Override the PMU.
4635 This option takes over the PMU facility, so it is no
4636 longer usable by perf. Setting this option starts the
4637 PMU counters by setting MMCR0 to 0 (the FC bit is
4638 cleared). If a number is given, then MMCR1 is set to
4639 that number, otherwise (e.g., 'pmu_override=on'), MMCR1
4642 pm_debug_messages [SUSPEND,KNL]
4643 Enable suspend/resume debug messages during boot up.
4646 Enable PNP debug messages (depends on the
4647 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
4648 via /sys/module/pnp/parameters/debug. We always show
4649 current resource usage; turning this on also shows
4650 possible settings and some assignment information.
4656 { on | off | curr | res | no-curr | no-res }
4659 [ISAPNP] Exclude IRQs for the autoconfiguration
4662 [ISAPNP] Exclude DMAs for the autoconfiguration
4664 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
4665 Ranges are in pairs (I/O port base and size).
4668 [ISAPNP] Exclude memory regions for the
4670 Ranges are in pairs (memory base and size).
4672 ports= [IP_VS_FTP] IPVS ftp helper module
4674 Up to 8 (IP_VS_APP_MAX_PORTS) ports
4676 Format: <port>,<port>....
4678 possible_cpus= [SMP,S390,X86]
4679 Format: <unsigned int>
4680 Set the number of possible CPUs, overriding the
4681 regular discovery mechanisms (such as ACPI/FW, etc).
4683 powersave=off [PPC] This option disables power saving features.
4684 It specifically disables cpuidle and sets the
4685 platform machine description specific power_save
4686 function to NULL. On Idle the CPU just reduces
4689 ppc_strict_facility_enable
4690 [PPC,ENABLE] This option catches any kernel floating point,
4691 Altivec, VSX and SPE outside of regions specifically
4692 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
4693 There is some performance impact when enabling this.
4697 Disable Hardware Transactional Memory
4700 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC
4701 none - Limited to cond_resched() calls
4702 voluntary - Limited to cond_resched() and might_sleep() calls
4703 full - Any section that isn't explicitly preempt disabled
4704 can be preempted anytime.
4706 print-fatal-signals=
4707 [KNL] debug: print fatal signals
4709 If enabled, warn about various signal handling
4710 related application anomalies: too many signals,
4711 too many POSIX.1 timers, fatal signals causing a
4714 If you hit the warning due to signal overflow,
4715 you might want to try "ulimit -i unlimited".
4719 printk.always_kmsg_dump=
4720 Trigger kmsg_dump for cases other than kernel oops or
4722 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4725 printk.console_no_auto_verbose=
4726 Disable console loglevel raise on oops, panic
4727 or lockdep-detected issues (only if lock debug is on).
4728 With an exception to setups with low baudrate on
4729 serial console, keeping this 0 is a good choice
4730 in order to provide more debug information.
4732 default: 0 (auto_verbose is enabled)
4734 printk.devkmsg={on,off,ratelimit}
4735 Control writing to /dev/kmsg.
4736 on - unlimited logging to /dev/kmsg from userspace
4737 off - logging to /dev/kmsg disabled
4738 ratelimit - ratelimit the logging
4741 printk.time= Show timing data prefixed to each printk message line
4742 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4744 processor.max_cstate= [HW,ACPI]
4745 Limit processor to maximum C-state
4746 max_cstate=9 overrides any DMI blacklist limit.
4748 processor.nocst [HW,ACPI]
4749 Ignore the _CST method to determine C-states,
4750 instead using the legacy FADT method
4752 profile= [KNL] Enable kernel profiling via /proc/profile
4753 Format: [<profiletype>,]<number>
4754 Param: <profiletype>: "schedule", "sleep", or "kvm"
4755 [defaults to kernel profiling]
4756 Param: "schedule" - profile schedule points.
4757 Param: "sleep" - profile D-state sleeping (millisecs).
4758 Requires CONFIG_SCHEDSTATS
4759 Param: "kvm" - profile VM exits.
4760 Param: <number> - step/bucket size as a power of 2 for
4761 statistical time based profiling.
4763 prompt_ramdisk= [RAM] [Deprecated]
4765 prot_virt= [S390] enable hosting protected virtual machines
4766 isolated from the hypervisor (if hardware supports
4770 psi= [KNL] Enable or disable pressure stall information
4774 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
4775 probe for; one of (bare|imps|exps|lifebook|any).
4776 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
4778 psmouse.resetafter= [HW,MOUSE]
4779 Try to reset the device after so many bad packets
4782 [HW,MOUSE] Set desired mouse resolution, in dpi.
4783 psmouse.smartscroll=
4784 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
4785 0 = disabled, 1 = enabled (default).
4787 pstore.backend= Specify the name of the pstore backend to use
4789 pti= [X86-64] Control Page Table Isolation of user and
4790 kernel address spaces. Disabling this feature
4791 removes hardening, but improves performance of
4792 system calls and interrupts.
4794 on - unconditionally enable
4795 off - unconditionally disable
4796 auto - kernel detects whether your CPU model is
4797 vulnerable to issues that PTI mitigates
4799 Not specifying this option is equivalent to pti=auto.
4802 [KNL] Number of legacy pty's. Overwrites compiled-in
4805 quiet [KNL,EARLY] Disable most log messages
4809 radix_hcall_invalidate=on [PPC/PSERIES]
4810 Disable RADIX GTSE feature and use hcall for TLB
4814 See Documentation/admin-guide/md.rst.
4816 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
4817 See Documentation/admin-guide/blockdev/ramdisk.rst.
4819 ramdisk_start= [RAM] RAM disk image start address
4821 random.trust_cpu=off
4822 [KNL,EARLY] Disable trusting the use of the CPU's
4823 random number generator (if available) to
4824 initialize the kernel's RNG.
4826 random.trust_bootloader=off
4827 [KNL,EARLY] Disable trusting the use of the a seed
4828 passed by the bootloader (if available) to
4829 initialize the kernel's RNG.
4831 randomize_kstack_offset=
4832 [KNL,EARLY] Enable or disable kernel stack offset
4833 randomization, which provides roughly 5 bits of
4834 entropy, frustrating memory corruption attacks
4835 that depend on stack address determinism or
4836 cross-syscall address exposures. This is only
4837 available on architectures that have defined
4838 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET.
4839 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4840 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT.
4842 ras=option[,option,...] [KNL] RAS-specific options
4845 Disable the Correctable Errors Collector,
4846 see CONFIG_RAS_CEC help text.
4848 rcu_nocbs[=cpu-list]
4849 [KNL] The optional argument is a cpu list,
4852 In kernels built with CONFIG_RCU_NOCB_CPU=y,
4853 enable the no-callback CPU mode, which prevents
4854 such CPUs' callbacks from being invoked in
4855 softirq context. Invocation of such CPUs' RCU
4856 callbacks will instead be offloaded to "rcuox/N"
4857 kthreads created for that purpose, where "x" is
4858 "p" for RCU-preempt, "s" for RCU-sched, and "g"
4859 for the kthreads that mediate grace periods; and
4860 "N" is the CPU number. This reduces OS jitter on
4861 the offloaded CPUs, which can be useful for HPC
4862 and real-time workloads. It can also improve
4863 energy efficiency for asymmetric multiprocessors.
4865 If a cpulist is passed as an argument, the specified
4866 list of CPUs is set to no-callback mode from boot.
4868 Otherwise, if the '=' sign and the cpulist
4869 arguments are omitted, no CPU will be set to
4870 no-callback mode from boot but the mode may be
4871 toggled at runtime via cpusets.
4873 Note that this argument takes precedence over
4874 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option.
4877 Rather than requiring that offloaded CPUs
4878 (specified by rcu_nocbs= above) explicitly
4879 awaken the corresponding "rcuoN" kthreads,
4880 make these kthreads poll for callbacks.
4881 This improves the real-time response for the
4882 offloaded CPUs by relieving them of the need to
4883 wake up the corresponding kthread, but degrades
4884 energy efficiency by requiring that the kthreads
4885 periodically wake up to do the polling.
4887 rcutree.blimit= [KNL]
4888 Set maximum number of finished RCU callbacks to
4889 process in one batch.
4891 rcutree.do_rcu_barrier= [KNL]
4892 Request a call to rcu_barrier(). This is
4893 throttled so that userspace tests can safely
4894 hammer on the sysfs variable if they so choose.
4895 If triggered before the RCU grace-period machinery
4896 is fully active, this will error out with EAGAIN.
4898 rcutree.dump_tree= [KNL]
4899 Dump the structure of the rcu_node combining tree
4900 out at early boot. This is used for diagnostic
4901 purposes, to verify correct tree setup.
4903 rcutree.gp_cleanup_delay= [KNL]
4904 Set the number of jiffies to delay each step of
4905 RCU grace-period cleanup.
4907 rcutree.gp_init_delay= [KNL]
4908 Set the number of jiffies to delay each step of
4909 RCU grace-period initialization.
4911 rcutree.gp_preinit_delay= [KNL]
4912 Set the number of jiffies to delay each step of
4913 RCU grace-period pre-initialization, that is,
4914 the propagation of recent CPU-hotplug changes up
4915 the rcu_node combining tree.
4917 rcutree.jiffies_till_first_fqs= [KNL]
4918 Set delay from grace-period initialization to
4919 first attempt to force quiescent states.
4920 Units are jiffies, minimum value is zero,
4921 and maximum value is HZ.
4923 rcutree.jiffies_till_next_fqs= [KNL]
4924 Set delay between subsequent attempts to force
4925 quiescent states. Units are jiffies, minimum
4926 value is one, and maximum value is HZ.
4928 rcutree.jiffies_till_sched_qs= [KNL]
4929 Set required age in jiffies for a
4930 given grace period before RCU starts
4931 soliciting quiescent-state help from
4932 rcu_note_context_switch() and cond_resched().
4933 If not specified, the kernel will calculate
4934 a value based on the most recent settings
4935 of rcutree.jiffies_till_first_fqs
4936 and rcutree.jiffies_till_next_fqs.
4937 This calculated value may be viewed in
4938 rcutree.jiffies_to_sched_qs. Any attempt to set
4939 rcutree.jiffies_to_sched_qs will be cheerfully
4942 rcutree.kthread_prio= [KNL,BOOT]
4943 Set the SCHED_FIFO priority of the RCU per-CPU
4944 kthreads (rcuc/N). This value is also used for
4945 the priority of the RCU boost threads (rcub/N)
4946 and for the RCU grace-period kthreads (rcu_bh,
4947 rcu_preempt, and rcu_sched). If RCU_BOOST is
4948 set, valid values are 1-99 and the default is 1
4949 (the least-favored priority). Otherwise, when
4950 RCU_BOOST is not set, valid values are 0-99 and
4951 the default is zero (non-realtime operation).
4952 When RCU_NOCB_CPU is set, also adjust the
4953 priority of NOCB callback kthreads.
4955 rcutree.nocb_nobypass_lim_per_jiffy= [KNL]
4956 On callback-offloaded (rcu_nocbs) CPUs,
4957 RCU reduces the lock contention that would
4958 otherwise be caused by callback floods through
4959 use of the ->nocb_bypass list. However, in the
4960 common non-flooded case, RCU queues directly to
4961 the main ->cblist in order to avoid the extra
4962 overhead of the ->nocb_bypass list and its lock.
4963 But if there are too many callbacks queued during
4964 a single jiffy, RCU pre-queues the callbacks into
4965 the ->nocb_bypass queue. The definition of "too
4966 many" is supplied by this kernel boot parameter.
4968 rcutree.qhimark= [KNL]
4969 Set threshold of queued RCU callbacks beyond which
4970 batch limiting is disabled.
4972 rcutree.qlowmark= [KNL]
4973 Set threshold of queued RCU callbacks below which
4974 batch limiting is re-enabled.
4976 rcutree.qovld= [KNL]
4977 Set threshold of queued RCU callbacks beyond which
4978 RCU's force-quiescent-state scan will aggressively
4979 enlist help from cond_resched() and sched IPIs to
4980 help CPUs more quickly reach quiescent states.
4981 Set to less than zero to make this be set based
4982 on rcutree.qhimark at boot time and to zero to
4983 disable more aggressive help enlistment.
4985 rcutree.rcu_delay_page_cache_fill_msec= [KNL]
4986 Set the page-cache refill delay (in milliseconds)
4987 in response to low-memory conditions. The range
4988 of permitted values is in the range 0:100000.
4990 rcutree.rcu_divisor= [KNL]
4991 Set the shift-right count to use to compute
4992 the callback-invocation batch limit bl from
4993 the number of callbacks queued on this CPU.
4994 The result will be bounded below by the value of
4995 the rcutree.blimit kernel parameter. Every bl
4996 callbacks, the softirq handler will exit in
4997 order to allow the CPU to do other work.
4999 Please note that this callback-invocation batch
5000 limit applies only to non-offloaded callback
5001 invocation. Offloaded callbacks are instead
5002 invoked in the context of an rcuoc kthread, which
5003 scheduler will preempt as it does any other task.
5005 rcutree.rcu_fanout_exact= [KNL]
5006 Disable autobalancing of the rcu_node combining
5007 tree. This is used by rcutorture, and might
5008 possibly be useful for architectures having high
5009 cache-to-cache transfer latencies.
5011 rcutree.rcu_fanout_leaf= [KNL]
5012 Change the number of CPUs assigned to each
5013 leaf rcu_node structure. Useful for very
5014 large systems, which will choose the value 64,
5015 and for NUMA systems with large remote-access
5016 latencies, which will choose a value aligned
5017 with the appropriate hardware boundaries.
5019 rcutree.rcu_min_cached_objs= [KNL]
5020 Minimum number of objects which are cached and
5021 maintained per one CPU. Object size is equal
5022 to PAGE_SIZE. The cache allows to reduce the
5023 pressure to page allocator, also it makes the
5024 whole algorithm to behave better in low memory
5027 rcutree.rcu_nocb_gp_stride= [KNL]
5028 Set the number of NOCB callback kthreads in
5029 each group, which defaults to the square root
5030 of the number of CPUs. Larger numbers reduce
5031 the wakeup overhead on the global grace-period
5032 kthread, but increases that same overhead on
5033 each group's NOCB grace-period kthread.
5035 rcutree.rcu_kick_kthreads= [KNL]
5036 Cause the grace-period kthread to get an extra
5037 wake_up() if it sleeps three times longer than
5038 it should at force-quiescent-state time.
5039 This wake_up() will be accompanied by a
5040 WARN_ONCE() splat and an ftrace_dump().
5042 rcutree.rcu_resched_ns= [KNL]
5043 Limit the time spend invoking a batch of RCU
5044 callbacks to the specified number of nanoseconds.
5045 By default, this limit is checked only once
5046 every 32 callbacks in order to limit the pain
5047 inflicted by local_clock() overhead.
5049 rcutree.rcu_unlock_delay= [KNL]
5050 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels,
5051 this specifies an rcu_read_unlock()-time delay
5052 in microseconds. This defaults to zero.
5053 Larger delays increase the probability of
5054 catching RCU pointer leaks, that is, buggy use
5055 of RCU-protected pointers after the relevant
5056 rcu_read_unlock() has completed.
5058 rcutree.sysrq_rcu= [KNL]
5059 Commandeer a sysrq key to dump out Tree RCU's
5060 rcu_node tree with an eye towards determining
5061 why a new grace period has not yet started.
5063 rcutree.use_softirq= [KNL]
5064 If set to zero, move all RCU_SOFTIRQ processing to
5065 per-CPU rcuc kthreads. Defaults to a non-zero
5066 value, meaning that RCU_SOFTIRQ is used by default.
5067 Specify rcutree.use_softirq=0 to use rcuc kthreads.
5069 But note that CONFIG_PREEMPT_RT=y kernels disable
5070 this kernel boot parameter, forcibly setting it
5073 rcutree.enable_rcu_lazy= [KNL]
5074 To save power, batch RCU callbacks and flush after
5075 delay, memory pressure or callback list growing too
5078 rcuscale.gp_async= [KNL]
5079 Measure performance of asynchronous
5080 grace-period primitives such as call_rcu().
5082 rcuscale.gp_async_max= [KNL]
5083 Specify the maximum number of outstanding
5084 callbacks per writer thread. When a writer
5085 thread exceeds this limit, it invokes the
5086 corresponding flavor of rcu_barrier() to allow
5087 previously posted callbacks to drain.
5089 rcuscale.gp_exp= [KNL]
5090 Measure performance of expedited synchronous
5091 grace-period primitives.
5093 rcuscale.holdoff= [KNL]
5094 Set test-start holdoff period. The purpose of
5095 this parameter is to delay the start of the
5096 test until boot completes in order to avoid
5099 rcuscale.kfree_by_call_rcu= [KNL]
5100 In kernels built with CONFIG_RCU_LAZY=y, test
5101 call_rcu() instead of kfree_rcu().
5103 rcuscale.kfree_mult= [KNL]
5104 Instead of allocating an object of size kfree_obj,
5105 allocate one of kfree_mult * sizeof(kfree_obj).
5108 rcuscale.kfree_rcu_test= [KNL]
5109 Set to measure performance of kfree_rcu() flooding.
5111 rcuscale.kfree_rcu_test_double= [KNL]
5112 Test the double-argument variant of kfree_rcu().
5113 If this parameter has the same value as
5114 rcuscale.kfree_rcu_test_single, both the single-
5115 and double-argument variants are tested.
5117 rcuscale.kfree_rcu_test_single= [KNL]
5118 Test the single-argument variant of kfree_rcu().
5119 If this parameter has the same value as
5120 rcuscale.kfree_rcu_test_double, both the single-
5121 and double-argument variants are tested.
5123 rcuscale.kfree_nthreads= [KNL]
5124 The number of threads running loops of kfree_rcu().
5126 rcuscale.kfree_alloc_num= [KNL]
5127 Number of allocations and frees done in an iteration.
5129 rcuscale.kfree_loops= [KNL]
5130 Number of loops doing rcuscale.kfree_alloc_num number
5131 of allocations and frees.
5133 rcuscale.minruntime= [KNL]
5134 Set the minimum test run time in seconds. This
5135 does not affect the data-collection interval,
5136 but instead allows better measurement of things
5137 like CPU consumption.
5139 rcuscale.nreaders= [KNL]
5140 Set number of RCU readers. The value -1 selects
5141 N, where N is the number of CPUs. A value
5142 "n" less than -1 selects N-n+1, where N is again
5143 the number of CPUs. For example, -2 selects N
5144 (the number of CPUs), -3 selects N+1, and so on.
5145 A value of "n" less than or equal to -N selects
5148 rcuscale.nwriters= [KNL]
5149 Set number of RCU writers. The values operate
5150 the same as for rcuscale.nreaders.
5151 N, where N is the number of CPUs
5153 rcuscale.scale_type= [KNL]
5154 Specify the RCU implementation to test.
5156 rcuscale.shutdown= [KNL]
5157 Shut the system down after performance tests
5158 complete. This is useful for hands-off automated
5161 rcuscale.verbose= [KNL]
5162 Enable additional printk() statements.
5164 rcuscale.writer_holdoff= [KNL]
5165 Write-side holdoff between grace periods,
5166 in microseconds. The default of zero says
5169 rcuscale.writer_holdoff_jiffies= [KNL]
5170 Additional write-side holdoff between grace
5171 periods, but in jiffies. The default of zero
5174 rcutorture.fqs_duration= [KNL]
5175 Set duration of force_quiescent_state bursts
5178 rcutorture.fqs_holdoff= [KNL]
5179 Set holdoff time within force_quiescent_state bursts
5182 rcutorture.fqs_stutter= [KNL]
5183 Set wait time between force_quiescent_state bursts
5186 rcutorture.fwd_progress= [KNL]
5187 Specifies the number of kthreads to be used
5188 for RCU grace-period forward-progress testing
5189 for the types of RCU supporting this notion.
5190 Defaults to 1 kthread, values less than zero or
5191 greater than the number of CPUs cause the number
5194 rcutorture.fwd_progress_div= [KNL]
5195 Specify the fraction of a CPU-stall-warning
5196 period to do tight-loop forward-progress testing.
5198 rcutorture.fwd_progress_holdoff= [KNL]
5199 Number of seconds to wait between successive
5200 forward-progress tests.
5202 rcutorture.fwd_progress_need_resched= [KNL]
5203 Enclose cond_resched() calls within checks for
5204 need_resched() during tight-loop forward-progress
5207 rcutorture.gp_cond= [KNL]
5208 Use conditional/asynchronous update-side
5209 primitives, if available.
5211 rcutorture.gp_exp= [KNL]
5212 Use expedited update-side primitives, if available.
5214 rcutorture.gp_normal= [KNL]
5215 Use normal (non-expedited) asynchronous
5216 update-side primitives, if available.
5218 rcutorture.gp_sync= [KNL]
5219 Use normal (non-expedited) synchronous
5220 update-side primitives, if available. If all
5221 of rcutorture.gp_cond=, rcutorture.gp_exp=,
5222 rcutorture.gp_normal=, and rcutorture.gp_sync=
5223 are zero, rcutorture acts as if is interpreted
5224 they are all non-zero.
5226 rcutorture.irqreader= [KNL]
5227 Run RCU readers from irq handlers, or, more
5228 accurately, from a timer handler. Not all RCU
5229 flavors take kindly to this sort of thing.
5231 rcutorture.leakpointer= [KNL]
5232 Leak an RCU-protected pointer out of the reader.
5233 This can of course result in splats, and is
5234 intended to test the ability of things like
5235 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect
5238 rcutorture.n_barrier_cbs= [KNL]
5239 Set callbacks/threads for rcu_barrier() testing.
5241 rcutorture.nfakewriters= [KNL]
5242 Set number of concurrent RCU writers. These just
5243 stress RCU, they don't participate in the actual
5244 test, hence the "fake".
5246 rcutorture.nocbs_nthreads= [KNL]
5247 Set number of RCU callback-offload togglers.
5248 Zero (the default) disables toggling.
5250 rcutorture.nocbs_toggle= [KNL]
5251 Set the delay in milliseconds between successive
5252 callback-offload toggling attempts.
5254 rcutorture.nreaders= [KNL]
5255 Set number of RCU readers. The value -1 selects
5256 N-1, where N is the number of CPUs. A value
5257 "n" less than -1 selects N-n-2, where N is again
5258 the number of CPUs. For example, -2 selects N
5259 (the number of CPUs), -3 selects N+1, and so on.
5261 rcutorture.object_debug= [KNL]
5262 Enable debug-object double-call_rcu() testing.
5264 rcutorture.onoff_holdoff= [KNL]
5265 Set time (s) after boot for CPU-hotplug testing.
5267 rcutorture.onoff_interval= [KNL]
5268 Set time (jiffies) between CPU-hotplug operations,
5269 or zero to disable CPU-hotplug testing.
5271 rcutorture.read_exit= [KNL]
5272 Set the number of read-then-exit kthreads used
5273 to test the interaction of RCU updaters and
5274 task-exit processing.
5276 rcutorture.read_exit_burst= [KNL]
5277 The number of times in a given read-then-exit
5278 episode that a set of read-then-exit kthreads
5281 rcutorture.read_exit_delay= [KNL]
5282 The delay, in seconds, between successive
5283 read-then-exit testing episodes.
5285 rcutorture.shuffle_interval= [KNL]
5286 Set task-shuffle interval (s). Shuffling tasks
5287 allows some CPUs to go into dyntick-idle mode
5288 during the rcutorture test.
5290 rcutorture.shutdown_secs= [KNL]
5291 Set time (s) after boot system shutdown. This
5292 is useful for hands-off automated testing.
5294 rcutorture.stall_cpu= [KNL]
5295 Duration of CPU stall (s) to test RCU CPU stall
5296 warnings, zero to disable.
5298 rcutorture.stall_cpu_block= [KNL]
5299 Sleep while stalling if set. This will result
5300 in warnings from preemptible RCU in addition to
5301 any other stall-related activity. Note that
5302 in kernels built with CONFIG_PREEMPTION=n and
5303 CONFIG_PREEMPT_COUNT=y, this parameter will
5304 cause the CPU to pass through a quiescent state.
5305 Given CONFIG_PREEMPTION=n, this will suppress
5306 RCU CPU stall warnings, but will instead result
5307 in scheduling-while-atomic splats.
5309 Use of this module parameter results in splats.
5312 rcutorture.stall_cpu_holdoff= [KNL]
5313 Time to wait (s) after boot before inducing stall.
5315 rcutorture.stall_cpu_irqsoff= [KNL]
5316 Disable interrupts while stalling if set.
5318 rcutorture.stall_gp_kthread= [KNL]
5319 Duration (s) of forced sleep within RCU
5320 grace-period kthread to test RCU CPU stall
5321 warnings, zero to disable. If both stall_cpu
5322 and stall_gp_kthread are specified, the
5323 kthread is starved first, then the CPU.
5325 rcutorture.stat_interval= [KNL]
5326 Time (s) between statistics printk()s.
5328 rcutorture.stutter= [KNL]
5329 Time (s) to stutter testing, for example, specifying
5330 five seconds causes the test to run for five seconds,
5331 wait for five seconds, and so on. This tests RCU's
5332 ability to transition abruptly to and from idle.
5334 rcutorture.test_boost= [KNL]
5335 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
5336 "Maybe" means test if the RCU implementation
5337 under test support RCU priority boosting.
5339 rcutorture.test_boost_duration= [KNL]
5340 Duration (s) of each individual boost test.
5342 rcutorture.test_boost_interval= [KNL]
5343 Interval (s) between each boost test.
5345 rcutorture.test_no_idle_hz= [KNL]
5346 Test RCU's dyntick-idle handling. See also the
5347 rcutorture.shuffle_interval parameter.
5349 rcutorture.torture_type= [KNL]
5350 Specify the RCU implementation to test.
5352 rcutorture.verbose= [KNL]
5353 Enable additional printk() statements.
5355 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
5356 Dump ftrace buffer after reporting RCU CPU
5359 rcupdate.rcu_cpu_stall_notifiers= [KNL]
5360 Provide RCU CPU stall notifiers, but see the
5361 warnings in the RCU_CPU_STALL_NOTIFIER Kconfig
5362 option's help text. TL;DR: You almost certainly
5363 do not want rcupdate.rcu_cpu_stall_notifiers.
5365 rcupdate.rcu_cpu_stall_suppress= [KNL]
5366 Suppress RCU CPU stall warning messages.
5368 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL]
5369 Suppress RCU CPU stall warning messages and
5370 rcutorture writer stall warnings that occur
5371 during early boot, that is, during the time
5372 before the init task is spawned.
5374 rcupdate.rcu_cpu_stall_timeout= [KNL]
5375 Set timeout for RCU CPU stall warning messages.
5376 The value is in seconds and the maximum allowed
5377 value is 300 seconds.
5379 rcupdate.rcu_exp_cpu_stall_timeout= [KNL]
5380 Set timeout for expedited RCU CPU stall warning
5381 messages. The value is in milliseconds
5382 and the maximum allowed value is 21000
5383 milliseconds. Please note that this value is
5384 adjusted to an arch timer tick resolution.
5385 Setting this to zero causes the value from
5386 rcupdate.rcu_cpu_stall_timeout to be used (after
5387 conversion from seconds to milliseconds).
5389 rcupdate.rcu_cpu_stall_cputime= [KNL]
5390 Provide statistics on the cputime and count of
5391 interrupts and tasks during the sampling period. For
5392 multiple continuous RCU stalls, all sampling periods
5393 begin at half of the first RCU stall timeout.
5395 rcupdate.rcu_exp_stall_task_details= [KNL]
5396 Print stack dumps of any tasks blocking the
5397 current expedited RCU grace period during an
5398 expedited RCU CPU stall warning.
5400 rcupdate.rcu_expedited= [KNL]
5401 Use expedited grace-period primitives, for
5402 example, synchronize_rcu_expedited() instead
5403 of synchronize_rcu(). This reduces latency,
5404 but can increase CPU utilization, degrade
5405 real-time latency, and degrade energy efficiency.
5406 No effect on CONFIG_TINY_RCU kernels.
5408 rcupdate.rcu_normal= [KNL]
5409 Use only normal grace-period primitives,
5410 for example, synchronize_rcu() instead of
5411 synchronize_rcu_expedited(). This improves
5412 real-time latency, CPU utilization, and
5413 energy efficiency, but can expose users to
5414 increased grace-period latency. This parameter
5415 overrides rcupdate.rcu_expedited. No effect on
5416 CONFIG_TINY_RCU kernels.
5418 rcupdate.rcu_normal_after_boot= [KNL]
5419 Once boot has completed (that is, after
5420 rcu_end_inkernel_boot() has been invoked), use
5421 only normal grace-period primitives. No effect
5422 on CONFIG_TINY_RCU kernels.
5424 But note that CONFIG_PREEMPT_RT=y kernels enables
5425 this kernel boot parameter, forcibly setting
5426 it to the value one, that is, converting any
5427 post-boot attempt at an expedited RCU grace
5428 period to instead use normal non-expedited
5429 grace-period processing.
5431 rcupdate.rcu_task_collapse_lim= [KNL]
5432 Set the maximum number of callbacks present
5433 at the beginning of a grace period that allows
5434 the RCU Tasks flavors to collapse back to using
5435 a single callback queue. This switching only
5436 occurs when rcupdate.rcu_task_enqueue_lim is
5437 set to the default value of -1.
5439 rcupdate.rcu_task_contend_lim= [KNL]
5440 Set the minimum number of callback-queuing-time
5441 lock-contention events per jiffy required to
5442 cause the RCU Tasks flavors to switch to per-CPU
5443 callback queuing. This switching only occurs
5444 when rcupdate.rcu_task_enqueue_lim is set to
5445 the default value of -1.
5447 rcupdate.rcu_task_enqueue_lim= [KNL]
5448 Set the number of callback queues to use for the
5449 RCU Tasks family of RCU flavors. The default
5450 of -1 allows this to be automatically (and
5451 dynamically) adjusted. This parameter is intended
5454 rcupdate.rcu_task_ipi_delay= [KNL]
5455 Set time in jiffies during which RCU tasks will
5456 avoid sending IPIs, starting with the beginning
5457 of a given grace period. Setting a large
5458 number avoids disturbing real-time workloads,
5459 but lengthens grace periods.
5461 rcupdate.rcu_task_lazy_lim= [KNL]
5462 Number of callbacks on a given CPU that will
5463 cancel laziness on that CPU. Use -1 to disable
5464 cancellation of laziness, but be advised that
5465 doing so increases the danger of OOM due to
5468 rcupdate.rcu_task_stall_info= [KNL]
5469 Set initial timeout in jiffies for RCU task stall
5470 informational messages, which give some indication
5471 of the problem for those not patient enough to
5472 wait for ten minutes. Informational messages are
5473 only printed prior to the stall-warning message
5474 for a given grace period. Disable with a value
5475 less than or equal to zero. Defaults to ten
5476 seconds. A change in value does not take effect
5477 until the beginning of the next grace period.
5479 rcupdate.rcu_task_stall_info_mult= [KNL]
5480 Multiplier for time interval between successive
5481 RCU task stall informational messages for a given
5482 RCU tasks grace period. This value is clamped
5483 to one through ten, inclusive. It defaults to
5484 the value three, so that the first informational
5485 message is printed 10 seconds into the grace
5486 period, the second at 40 seconds, the third at
5487 160 seconds, and then the stall warning at 600
5488 seconds would prevent a fourth at 640 seconds.
5490 rcupdate.rcu_task_stall_timeout= [KNL]
5491 Set timeout in jiffies for RCU task stall
5492 warning messages. Disable with a value less
5493 than or equal to zero. Defaults to ten minutes.
5494 A change in value does not take effect until
5495 the beginning of the next grace period.
5497 rcupdate.rcu_tasks_lazy_ms= [KNL]
5498 Set timeout in milliseconds RCU Tasks asynchronous
5499 callback batching for call_rcu_tasks().
5500 A negative value will take the default. A value
5501 of zero will disable batching. Batching is
5502 always disabled for synchronize_rcu_tasks().
5504 rcupdate.rcu_tasks_rude_lazy_ms= [KNL]
5505 Set timeout in milliseconds RCU Tasks
5506 Rude asynchronous callback batching for
5507 call_rcu_tasks_rude(). A negative value
5508 will take the default. A value of zero will
5509 disable batching. Batching is always disabled
5510 for synchronize_rcu_tasks_rude().
5512 rcupdate.rcu_tasks_trace_lazy_ms= [KNL]
5513 Set timeout in milliseconds RCU Tasks
5514 Trace asynchronous callback batching for
5515 call_rcu_tasks_trace(). A negative value
5516 will take the default. A value of zero will
5517 disable batching. Batching is always disabled
5518 for synchronize_rcu_tasks_trace().
5520 rcupdate.rcu_self_test= [KNL]
5521 Run the RCU early boot self tests
5525 Run specified binary instead of /init from the ramdisk,
5526 used for early userspace startup. See initrd.
5529 force - Override the decision by the kernel to hide the
5530 advertisement of RDRAND support (this affects
5531 certain AMD processors because of buggy BIOS
5532 support, specifically around the suspend/resume
5536 Turn on/off individual RDT features. List is:
5537 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
5539 E.g. to turn on cmt and turn off mba use:
5543 Format (x86 or x86_64):
5544 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] | d[efault] \
5546 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
5548 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
5549 (prefix with 'panic_' to set mode for panic
5551 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
5552 reboot_force is either force or not specified,
5553 reboot_cpu is s[mp]#### with #### being the processor
5554 to be used for rebooting.
5556 refscale.holdoff= [KNL]
5557 Set test-start holdoff period. The purpose of
5558 this parameter is to delay the start of the
5559 test until boot completes in order to avoid
5562 refscale.lookup_instances= [KNL]
5563 Number of data elements to use for the forms of
5564 SLAB_TYPESAFE_BY_RCU testing. A negative number
5565 is negated and multiplied by nr_cpu_ids, while
5566 zero specifies nr_cpu_ids.
5568 refscale.loops= [KNL]
5569 Set the number of loops over the synchronization
5570 primitive under test. Increasing this number
5571 reduces noise due to loop start/end overhead,
5572 but the default has already reduced the per-pass
5573 noise to a handful of picoseconds on ca. 2020
5576 refscale.nreaders= [KNL]
5577 Set number of readers. The default value of -1
5578 selects N, where N is roughly 75% of the number
5579 of CPUs. A value of zero is an interesting choice.
5581 refscale.nruns= [KNL]
5582 Set number of runs, each of which is dumped onto
5585 refscale.readdelay= [KNL]
5586 Set the read-side critical-section duration,
5587 measured in microseconds.
5589 refscale.scale_type= [KNL]
5590 Specify the read-protection implementation to test.
5592 refscale.shutdown= [KNL]
5593 Shut down the system at the end of the performance
5594 test. This defaults to 1 (shut it down) when
5595 refscale is built into the kernel and to 0 (leave
5596 it running) when refscale is built as a module.
5598 refscale.verbose= [KNL]
5599 Enable additional printk() statements.
5601 refscale.verbose_batched= [KNL]
5602 Batch the additional printk() statements. If zero
5603 (the default) or negative, print everything. Otherwise,
5604 print every Nth verbose statement, where N is the value
5607 regulator_ignore_unused
5609 Prevents regulator framework from disabling regulators
5610 that are unused, due no driver claiming them. This may
5611 be useful for debug and development, but should not be
5612 needed on a platform with proper driver support.
5615 [KNL, SMP] Set scheduler's default relax_domain_level.
5616 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
5618 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
5619 Format: <base1>,<size1>[,<base2>,<size2>,...]
5620 Reserve I/O ports or memory so the kernel won't use
5621 them. If <base> is less than 0x10000, the region
5622 is assumed to be I/O ports; otherwise it is memory.
5624 reservetop= [X86-32,EARLY]
5626 Reserves a hole at the top of the kernel virtual
5629 reset_devices [KNL] Force drivers to reset the underlying device
5630 during initialization.
5633 Specify the partition device for software suspend
5635 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
5637 resume_offset= [SWSUSP]
5638 Specify the offset from the beginning of the partition
5639 given by "resume=" at which the swap header is located,
5640 in <PAGE_SIZE> units (needed only for swap files).
5641 See Documentation/power/swsusp-and-swap-files.rst
5643 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
5644 read the resume files
5646 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
5647 Useful for devices that are detected asynchronously
5648 (e.g. USB and MMC devices).
5650 retain_initrd [RAM] Keep initrd memory after extraction. After boot, it will
5651 be accessible via /sys/firmware/initrd.
5653 retbleed= [X86] Control mitigation of RETBleed (Arbitrary
5654 Speculative Code Execution with Return Instructions)
5657 AMD-based UNRET and IBPB mitigations alone do not stop
5658 sibling threads from influencing the predictions of other
5659 sibling threads. For that reason, STIBP is used on pro-
5660 cessors that support it, and mitigate SMT on processors
5664 auto - automatically select a migitation
5665 auto,nosmt - automatically select a mitigation,
5666 disabling SMT if necessary for
5667 the full mitigation (only on Zen1
5668 and older without STIBP).
5669 ibpb - On AMD, mitigate short speculation
5670 windows on basic block boundaries too.
5671 Safe, highest perf impact. It also
5672 enables STIBP if present. Not suitable
5674 ibpb,nosmt - Like "ibpb" above but will disable SMT
5675 when STIBP is not available. This is
5676 the alternative for systems which do not
5678 unret - Force enable untrained return thunks,
5679 only effective on AMD f15h-f17h based
5681 unret,nosmt - Like unret, but will disable SMT when STIBP
5682 is not available. This is the alternative for
5683 systems which do not have STIBP.
5685 Selecting 'auto' will choose a mitigation method at run
5686 time according to the CPU.
5688 Not specifying this option is equivalent to retbleed=auto.
5690 rfkill.default_state=
5691 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
5692 etc. communication is blocked by default.
5695 rfkill.master_switch_mode=
5696 0 The "airplane mode" button does nothing.
5697 1 The "airplane mode" button toggles between everything
5698 blocked and the previous configuration.
5699 2 The "airplane mode" button toggles between everything
5700 blocked and everything unblocked.
5702 rhash_entries= [KNL,NET]
5703 Set number of hash buckets for route cache
5706 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
5709 riscv_isa_fallback [RISCV,EARLY]
5710 When CONFIG_RISCV_ISA_FALLBACK is not enabled, permit
5711 falling back to detecting extension support by parsing
5712 "riscv,isa" property on devicetree systems when the
5713 replacement properties are not found. See the Kconfig
5714 entry for RISCV_ISA_FALLBACK.
5716 ro [KNL] Mount root device read-only on boot
5719 on Mark read-only kernel memory as read-only (default).
5720 off Leave read-only kernel memory writable for debugging.
5721 full Mark read-only kernel memory and aliases as read-only
5726 Enable the uart passthrough on the designated usb port
5727 on Rockchip SoCs. When active, the signals of the
5728 debug-uart get routed to the D+ and D- pins of the usb
5729 port and the regular usb controller gets disabled.
5731 root= [KNL] Root filesystem
5732 Usually this a a block device specifier of some kind,
5733 see the early_lookup_bdev comment in
5734 block/early-lookup.c for details.
5735 Alternatively this can be "ram" for the legacy initial
5736 ramdisk, "nfs" and "cifs" for root on a network file
5737 system, or "mtd" and "ubi" for mounting from raw flash.
5739 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
5740 mount the root filesystem
5742 rootflags= [KNL] Set root filesystem mount option string
5744 rootfstype= [KNL] Set root filesystem type
5746 rootwait [KNL] Wait (indefinitely) for root device to show up.
5747 Useful for devices that are detected asynchronously
5748 (e.g. USB and MMC devices).
5750 rootwait= [KNL] Maximum time (in seconds) to wait for root device
5751 to show up before attempting to mount the root
5754 rproc_mem=nn[KMG][@address]
5755 [KNL,ARM,CMA] Remoteproc physical memory block.
5756 Memory area to be used by remote processor image,
5759 rw [KNL] Mount root device read-write on boot
5761 S [KNL] Run init in single mode
5763 s390_iommu= [HW,S390]
5764 Set s390 IOTLB flushing mode
5766 With strict flushing every unmap operation will result
5767 in an IOTLB flush. Default is lazy flushing before
5768 reuse, which is faster. Deprecated, equivalent to
5771 s390_iommu_aperture= [KNL,S390]
5772 Specifies the size of the per device DMA address space
5773 accessible through the DMA and IOMMU APIs as a decimal
5774 factor of the size of main memory.
5775 The default is 1 meaning that one can concurrently use
5776 as many DMA addresses as physical memory is installed,
5777 if supported by hardware, and thus map all of memory
5778 once. With a value of 2 one can map all of memory twice
5779 and so on. As a special case a factor of 0 imposes no
5780 restrictions other than those given by hardware at the
5781 cost of significant additional memory use for tables.
5784 See drivers/net/irda/sa1100_ir.c.
5786 sched_verbose [KNL,EARLY] Enables verbose scheduler debug messages.
5788 schedstats= [KNL,X86] Enable or disable scheduled statistics.
5789 Allowed values are enable and disable. This feature
5790 incurs a small amount of overhead in the scheduler
5791 but is useful for debugging and performance tuning.
5793 sched_thermal_decay_shift=
5794 [KNL, SMP] Set a decay shift for scheduler thermal
5795 pressure signal. Thermal pressure signal follows the
5796 default decay period of other scheduler pelt
5797 signals(usually 32 ms but configurable). Setting
5798 sched_thermal_decay_shift will left shift the decay
5799 period for the thermal pressure signal by the shift
5801 i.e. with the default pelt decay period of 32 ms
5802 sched_thermal_decay_shift thermal pressure decay pr
5806 Format: integer between 0 and 10
5809 scftorture.holdoff= [KNL]
5810 Number of seconds to hold off before starting
5811 test. Defaults to zero for module insertion and
5812 to 10 seconds for built-in smp_call_function()
5815 scftorture.longwait= [KNL]
5816 Request ridiculously long waits randomly selected
5817 up to the chosen limit in seconds. Zero (the
5818 default) disables this feature. Please note
5819 that requesting even small non-zero numbers of
5820 seconds can result in RCU CPU stall warnings,
5821 softlockup complaints, and so on.
5823 scftorture.nthreads= [KNL]
5824 Number of kthreads to spawn to invoke the
5825 smp_call_function() family of functions.
5826 The default of -1 specifies a number of kthreads
5827 equal to the number of CPUs.
5829 scftorture.onoff_holdoff= [KNL]
5830 Number seconds to wait after the start of the
5831 test before initiating CPU-hotplug operations.
5833 scftorture.onoff_interval= [KNL]
5834 Number seconds to wait between successive
5835 CPU-hotplug operations. Specifying zero (which
5836 is the default) disables CPU-hotplug operations.
5838 scftorture.shutdown_secs= [KNL]
5839 The number of seconds following the start of the
5840 test after which to shut down the system. The
5841 default of zero avoids shutting down the system.
5842 Non-zero values are useful for automated tests.
5844 scftorture.stat_interval= [KNL]
5845 The number of seconds between outputting the
5846 current test statistics to the console. A value
5847 of zero disables statistics output.
5849 scftorture.stutter_cpus= [KNL]
5850 The number of jiffies to wait between each change
5851 to the set of CPUs under test.
5853 scftorture.use_cpus_read_lock= [KNL]
5854 Use use_cpus_read_lock() instead of the default
5855 preempt_disable() to disable CPU hotplug
5856 while invoking one of the smp_call_function*()
5859 scftorture.verbose= [KNL]
5860 Enable additional printk() statements.
5862 scftorture.weight_single= [KNL]
5863 The probability weighting to use for the
5864 smp_call_function_single() function with a zero
5865 "wait" parameter. A value of -1 selects the
5866 default if all other weights are -1. However,
5867 if at least one weight has some other value, a
5868 value of -1 will instead select a weight of zero.
5870 scftorture.weight_single_wait= [KNL]
5871 The probability weighting to use for the
5872 smp_call_function_single() function with a
5873 non-zero "wait" parameter. See weight_single.
5875 scftorture.weight_many= [KNL]
5876 The probability weighting to use for the
5877 smp_call_function_many() function with a zero
5878 "wait" parameter. See weight_single.
5879 Note well that setting a high probability for
5880 this weighting can place serious IPI load
5883 scftorture.weight_many_wait= [KNL]
5884 The probability weighting to use for the
5885 smp_call_function_many() function with a
5886 non-zero "wait" parameter. See weight_single
5889 scftorture.weight_all= [KNL]
5890 The probability weighting to use for the
5891 smp_call_function_all() function with a zero
5892 "wait" parameter. See weight_single and
5895 scftorture.weight_all_wait= [KNL]
5896 The probability weighting to use for the
5897 smp_call_function_all() function with a
5898 non-zero "wait" parameter. See weight_single
5901 skew_tick= [KNL,EARLY] Offset the periodic timer tick per cpu to mitigate
5902 xtime_lock contention on larger systems, and/or RCU lock
5903 contention on all systems with CONFIG_MAXSMP set.
5904 Format: { "0" | "1" }
5905 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
5907 Note: increases power consumption, thus should only be
5908 enabled if running jitter sensitive (HPC/RT) workloads.
5910 security= [SECURITY] Choose a legacy "major" security module to
5911 enable at boot. This has been deprecated by the
5914 selinux= [SELINUX] Disable or enable SELinux at boot time.
5915 Format: { "0" | "1" }
5916 See security/selinux/Kconfig help text.
5921 serialnumber [BUGS=X86-32]
5923 sev=option[,option...] [X86-64] See Documentation/arch/x86/x86_64/boot-options.rst
5926 Maximal number of shapers.
5928 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
5929 Limit apic dumping. The parameter defines the maximal
5930 number of local apics being dumped. Also it is possible
5931 to set it to "all" by meaning -- no limit here.
5932 Format: { 1 (default) | 2 | ... | all }.
5933 The parameter valid if only apic=debug or
5934 apic=verbose is specified.
5935 Example: apic=debug show_lapic=all
5940 slab_debug[=options[,slabs][;[options[,slabs]]...] [MM]
5941 Enabling slab_debug allows one to determine the
5942 culprit if slab objects become corrupted. Enabling
5943 slab_debug can create guard zones around objects and
5944 may poison objects when not in use. Also tracks the
5945 last alloc / free. For more information see
5946 Documentation/mm/slub.rst.
5947 (slub_debug legacy name also accepted for now)
5949 slab_max_order= [MM]
5950 Determines the maximum allowed order for slabs.
5951 A high setting may cause OOMs due to memory
5952 fragmentation. For more information see
5953 Documentation/mm/slub.rst.
5954 (slub_max_order legacy name also accepted for now)
5957 Enable merging of slabs with similar size when the
5958 kernel is built without CONFIG_SLAB_MERGE_DEFAULT.
5959 (slub_merge legacy name also accepted for now)
5961 slab_min_objects= [MM]
5962 The minimum number of objects per slab. SLUB will
5963 increase the slab order up to slab_max_order to
5964 generate a sufficiently large slab able to contain
5965 the number of objects indicated. The higher the number
5966 of objects the smaller the overhead of tracking slabs
5967 and the less frequently locks need to be acquired.
5968 For more information see Documentation/mm/slub.rst.
5969 (slub_min_objects legacy name also accepted for now)
5971 slab_min_order= [MM]
5972 Determines the minimum page order for slabs. Must be
5973 lower or equal to slab_max_order. For more information see
5974 Documentation/mm/slub.rst.
5975 (slub_min_order legacy name also accepted for now)
5978 Disable merging of slabs with similar size. May be
5979 necessary if there is some reason to distinguish
5980 allocs to different slabs, especially in hardened
5981 environments where the risk of heap overflows and
5982 layout control by attackers can usually be
5983 frustrated by disabling merging. This will reduce
5984 most of the exposure of a heap attack to a single
5985 cache (risks via metadata attacks are mostly
5986 unchanged). Debug options disable merging on their
5988 For more information see Documentation/mm/slub.rst.
5989 (slub_nomerge legacy name also accepted for now)
5994 Format: <io1>[,<io2>[,...,<io8>]]
5996 smp.csd_lock_timeout= [KNL]
5997 Specify the period of time in milliseconds
5998 that smp_call_function() and friends will wait
5999 for a CPU to release the CSD lock. This is
6000 useful when diagnosing bugs involving CPUs
6001 disabling interrupts for extended periods
6002 of time. Defaults to 5,000 milliseconds, and
6003 setting a value of zero disables this feature.
6004 This feature may be more efficiently disabled
6005 using the csdlock_debug- kernel parameter.
6007 smp.panic_on_ipistall= [KNL]
6008 If a csd_lock_timeout extends for more than
6009 the specified number of milliseconds, panic the
6010 system. By default, let CSD-lock acquisition
6011 take as long as they take. Specifying 300,000
6012 for this value provides a 5-minute timeout.
6014 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
6015 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
6016 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
6017 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
6018 smsc-ircc2.ircc_irq= [HW] IRQ line
6019 smsc-ircc2.ircc_dma= [HW] DMA channel
6020 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
6021 0: Toshiba Satellite 1800 (GP data pin select)
6022 1: Fast pin select (default)
6025 smt= [KNL,MIPS,S390,EARLY] Set the maximum number of threads
6026 (logical CPUs) to use per physical CPU on systems
6027 capable of symmetric multithreading (SMT). Will
6028 be capped to the actual hardware limit.
6030 Default: -1 (no limit)
6033 [KNL] Should the soft-lockup detector generate panics.
6036 A value of 1 instructs the soft-lockup detector
6037 to panic the machine when a soft-lockup occurs. It is
6038 also controlled by the kernel.softlockup_panic sysctl
6039 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
6040 respective build-time switch to that functionality.
6042 softlockup_all_cpu_backtrace=
6043 [KNL] Should the soft-lockup detector generate
6044 backtraces on all cpus.
6047 sonypi.*= [HW] Sony Programmable I/O Control Device driver
6048 See Documentation/admin-guide/laptops/sonypi.rst
6050 spectre_v2= [X86,EARLY] Control mitigation of Spectre variant 2
6051 (indirect branch speculation) vulnerability.
6052 The default operation protects the kernel from
6055 on - unconditionally enable, implies
6057 off - unconditionally disable, implies
6059 auto - kernel detects whether your CPU model is
6062 Selecting 'on' will, and 'auto' may, choose a
6063 mitigation method at run time according to the
6064 CPU, the available microcode, the setting of the
6065 CONFIG_MITIGATION_RETPOLINE configuration option,
6066 and the compiler with which the kernel was built.
6068 Selecting 'on' will also enable the mitigation
6069 against user space to user space task attacks.
6071 Selecting 'off' will disable both the kernel and
6072 the user space protections.
6074 Specific mitigations can also be selected manually:
6076 retpoline - replace indirect branches
6077 retpoline,generic - Retpolines
6078 retpoline,lfence - LFENCE; indirect branch
6079 retpoline,amd - alias for retpoline,lfence
6080 eibrs - Enhanced/Auto IBRS
6081 eibrs,retpoline - Enhanced/Auto IBRS + Retpolines
6082 eibrs,lfence - Enhanced/Auto IBRS + LFENCE
6083 ibrs - use IBRS to protect kernel
6085 Not specifying this option is equivalent to
6089 [X86] Control mitigation of Spectre variant 2
6090 (indirect branch speculation) vulnerability between
6093 on - Unconditionally enable mitigations. Is
6094 enforced by spectre_v2=on
6096 off - Unconditionally disable mitigations. Is
6097 enforced by spectre_v2=off
6099 prctl - Indirect branch speculation is enabled,
6100 but mitigation can be enabled via prctl
6101 per thread. The mitigation control state
6102 is inherited on fork.
6105 - Like "prctl" above, but only STIBP is
6106 controlled per thread. IBPB is issued
6107 always when switching between different user
6111 - Same as "prctl" above, but all seccomp
6112 threads will enable the mitigation unless
6113 they explicitly opt out.
6116 - Like "seccomp" above, but only STIBP is
6117 controlled per thread. IBPB is issued
6118 always when switching between different
6119 user space processes.
6121 auto - Kernel selects the mitigation depending on
6122 the available CPU features and vulnerability.
6124 Default mitigation: "prctl"
6126 Not specifying this option is equivalent to
6127 spectre_v2_user=auto.
6129 spec_rstack_overflow=
6130 [X86,EARLY] Control RAS overflow mitigation on AMD Zen CPUs
6132 off - Disable mitigation
6133 microcode - Enable microcode mitigation only
6134 safe-ret - Enable sw-only safe RET mitigation (default)
6135 ibpb - Enable mitigation by issuing IBPB on
6137 ibpb-vmexit - Issue IBPB only on VMEXIT
6138 (cloud-specific mitigation)
6140 spec_store_bypass_disable=
6141 [HW,EARLY] Control Speculative Store Bypass (SSB) Disable mitigation
6142 (Speculative Store Bypass vulnerability)
6144 Certain CPUs are vulnerable to an exploit against a
6145 a common industry wide performance optimization known
6146 as "Speculative Store Bypass" in which recent stores
6147 to the same memory location may not be observed by
6148 later loads during speculative execution. The idea
6149 is that such stores are unlikely and that they can
6150 be detected prior to instruction retirement at the
6151 end of a particular speculation execution window.
6153 In vulnerable processors, the speculatively forwarded
6154 store can be used in a cache side channel attack, for
6155 example to read memory to which the attacker does not
6156 directly have access (e.g. inside sandboxed code).
6158 This parameter controls whether the Speculative Store
6159 Bypass optimization is used.
6161 On x86 the options are:
6163 on - Unconditionally disable Speculative Store Bypass
6164 off - Unconditionally enable Speculative Store Bypass
6165 auto - Kernel detects whether the CPU model contains an
6166 implementation of Speculative Store Bypass and
6167 picks the most appropriate mitigation. If the
6168 CPU is not vulnerable, "off" is selected. If the
6169 CPU is vulnerable the default mitigation is
6170 architecture and Kconfig dependent. See below.
6171 prctl - Control Speculative Store Bypass per thread
6172 via prctl. Speculative Store Bypass is enabled
6173 for a process by default. The state of the control
6174 is inherited on fork.
6175 seccomp - Same as "prctl" above, but all seccomp threads
6176 will disable SSB unless they explicitly opt out.
6178 Default mitigations:
6181 On powerpc the options are:
6183 on,auto - On Power8 and Power9 insert a store-forwarding
6184 barrier on kernel entry and exit. On Power7
6185 perform a software flush on kernel entry and
6189 Not specifying this option is equivalent to
6190 spec_store_bypass_disable=auto.
6192 spia_io_base= [HW,MTD]
6198 [X86] Enable split lock detection or bus lock detection
6200 When enabled (and if hardware support is present), atomic
6201 instructions that access data across cache line
6202 boundaries will result in an alignment check exception
6203 for split lock detection or a debug exception for
6208 warn - the kernel will emit rate-limited warnings
6209 about applications triggering the #AC
6210 exception or the #DB exception. This mode is
6211 the default on CPUs that support split lock
6212 detection or bus lock detection. Default
6213 behavior is by #AC if both features are
6214 enabled in hardware.
6216 fatal - the kernel will send SIGBUS to applications
6217 that trigger the #AC exception or the #DB
6218 exception. Default behavior is by #AC if
6219 both features are enabled in hardware.
6222 Set system wide rate limit to N bus locks
6223 per second for bus lock detection.
6226 N/A for split lock detection.
6229 If an #AC exception is hit in the kernel or in
6230 firmware (i.e. not while executing in user mode)
6231 the kernel will oops in either "warn" or "fatal"
6234 #DB exception for bus lock is triggered only when
6237 srbds= [X86,INTEL,EARLY]
6238 Control the Special Register Buffer Data Sampling
6241 Certain CPUs are vulnerable to an MDS-like
6242 exploit which can leak bits from the random
6245 By default, this issue is mitigated by
6246 microcode. However, the microcode fix can cause
6247 the RDRAND and RDSEED instructions to become
6248 much slower. Among other effects, this will
6249 result in reduced throughput from /dev/urandom.
6251 The microcode mitigation can be disabled with
6252 the following option:
6254 off: Disable mitigation and remove
6255 performance impact to RDRAND and RDSEED
6257 srcutree.big_cpu_lim [KNL]
6258 Specifies the number of CPUs constituting a
6259 large system, such that srcu_struct structures
6260 should immediately allocate an srcu_node array.
6261 This kernel-boot parameter defaults to 128,
6262 but takes effect only when the low-order four
6263 bits of srcutree.convert_to_big is equal to 3
6266 srcutree.convert_to_big [KNL]
6267 Specifies under what conditions an SRCU tree
6268 srcu_struct structure will be converted to big
6269 form, that is, with an rcu_node tree:
6272 1: At init_srcu_struct() time.
6273 2: When rcutorture decides to.
6274 3: Decide at boot time (default).
6275 0x1X: Above plus if high contention.
6277 Either way, the srcu_node tree will be sized based
6278 on the actual runtime number of CPUs (nr_cpu_ids)
6279 instead of the compile-time CONFIG_NR_CPUS.
6281 srcutree.counter_wrap_check [KNL]
6282 Specifies how frequently to check for
6283 grace-period sequence counter wrap for the
6284 srcu_data structure's ->srcu_gp_seq_needed field.
6285 The greater the number of bits set in this kernel
6286 parameter, the less frequently counter wrap will
6287 be checked for. Note that the bottom two bits
6290 srcutree.exp_holdoff [KNL]
6291 Specifies how many nanoseconds must elapse
6292 since the end of the last SRCU grace period for
6293 a given srcu_struct until the next normal SRCU
6294 grace period will be considered for automatic
6295 expediting. Set to zero to disable automatic
6298 srcutree.srcu_max_nodelay [KNL]
6299 Specifies the number of no-delay instances
6300 per jiffy for which the SRCU grace period
6301 worker thread will be rescheduled with zero
6302 delay. Beyond this limit, worker thread will
6303 be rescheduled with a sleep delay of one jiffy.
6305 srcutree.srcu_max_nodelay_phase [KNL]
6306 Specifies the per-grace-period phase, number of
6307 non-sleeping polls of readers. Beyond this limit,
6308 grace period worker thread will be rescheduled
6309 with a sleep delay of one jiffy, between each
6310 rescan of the readers, for a grace period phase.
6312 srcutree.srcu_retry_check_delay [KNL]
6313 Specifies number of microseconds of non-sleeping
6314 delay between each non-sleeping poll of readers.
6316 srcutree.small_contention_lim [KNL]
6317 Specifies the number of update-side contention
6318 events per jiffy will be tolerated before
6319 initiating a conversion of an srcu_struct
6320 structure to big form. Note that the value of
6321 srcutree.convert_to_big must have the 0x10 bit
6322 set for contention-based conversions to occur.
6324 ssbd= [ARM64,HW,EARLY]
6325 Speculative Store Bypass Disable control
6327 On CPUs that are vulnerable to the Speculative
6328 Store Bypass vulnerability and offer a
6329 firmware based mitigation, this parameter
6330 indicates how the mitigation should be used:
6332 force-on: Unconditionally enable mitigation for
6333 for both kernel and userspace
6334 force-off: Unconditionally disable mitigation for
6335 for both kernel and userspace
6336 kernel: Always enable mitigation in the
6337 kernel, and offer a prctl interface
6338 to allow userspace to register its
6339 interest in being mitigated too.
6341 stack_guard_gap= [MM]
6342 override the default stack gap protection. The value
6343 is in page units and it defines how many pages prior
6344 to (for stacks growing down) resp. after (for stacks
6345 growing up) the main stack are reserved for no other
6346 mapping. Default value is 256 pages.
6348 stack_depot_disable= [KNL,EARLY]
6349 Setting this to true through kernel command line will
6350 disable the stack depot thereby saving the static memory
6351 consumed by the stack hash table. By default this is set
6355 Enabled the stack tracer on boot up.
6357 stacktrace_filter=[function-list]
6358 [FTRACE] Limit the functions that the stack tracer
6359 will trace at boot up. function-list is a comma-separated
6360 list of functions. This list can be changed at run
6361 time by the stack_trace_filter file in the debugfs
6362 tracing directory. Note, this enables stack tracing
6363 and the stacktrace above is not needed.
6367 Set the STI (builtin display/keyboard on the HP-PARISC
6368 machines) console (graphic card) which should be used
6369 as the initial boot-console.
6370 See also comment in drivers/video/console/sticore.c.
6373 See comment in drivers/video/console/sticore.c.
6376 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
6381 Enable or disable strict sigaltstack size checks
6382 against the required signal frame size which
6383 depends on the supported FPU features. This can
6384 be used to filter out binaries which have
6385 not yet been made aware of AT_MINSIGSTKSZ.
6387 stress_hpt [PPC,EARLY]
6388 Limits the number of kernel HPT entries in the hash
6389 page table to increase the rate of hash page table
6390 faults on kernel addresses.
6392 stress_slb [PPC,EARLY]
6393 Limits the number of kernel SLB entries, and flushes
6394 them frequently to increase the rate of SLB faults
6395 on kernel addresses.
6397 sunrpc.min_resvport=
6398 sunrpc.max_resvport=
6400 SunRPC servers often require that client requests
6401 originate from a privileged port (i.e. a port in the
6402 range 0 < portnr < 1024).
6403 An administrator who wishes to reserve some of these
6404 ports for other uses may adjust the range that the
6405 kernel's sunrpc client considers to be privileged
6406 using these two parameters to set the minimum and
6407 maximum port values.
6409 sunrpc.svc_rpc_per_connection_limit=
6411 Limit the number of requests that the server will
6412 process in parallel from a single connection.
6413 The default value is 0 (no limit).
6417 Control how the NFS server code allocates CPUs to
6418 service thread pools. Depending on how many NICs
6419 you have and where their interrupts are bound, this
6420 option will affect which CPUs will do NFS serving.
6421 Note: this parameter cannot be changed while the
6422 NFS server is running.
6424 auto the server chooses an appropriate mode
6425 automatically using heuristics
6426 global a single global pool contains all CPUs
6427 percpu one pool for each CPU
6428 pernode one pool for each NUMA node (equivalent
6429 to global on non-NUMA machines)
6431 sunrpc.tcp_slot_table_entries=
6432 sunrpc.udp_slot_table_entries=
6434 Sets the upper limit on the number of simultaneous
6435 RPC calls that can be sent from the client to a
6436 server. Increasing these values may allow you to
6437 improve throughput, but will also increase the
6438 amount of memory reserved for use by the client.
6440 suspend.pm_test_delay=
6442 Sets the number of seconds to remain in a suspend test
6443 mode before resuming the system (see
6444 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
6445 is set. Default value is 5.
6448 Format: { on | off | y | n | 1 | 0 }
6449 This parameter controls use of the Protected
6450 Execution Facility on pSeries.
6452 swiotlb= [ARM,IA-64,PPC,MIPS,X86,EARLY]
6453 Format: { <int> [,<int>] | force | noforce }
6454 <int> -- Number of I/O TLB slabs
6455 <int> -- Second integer after comma. Number of swiotlb
6456 areas with their own lock. Will be rounded up
6458 force -- force using of bounce buffers even if they
6459 wouldn't be automatically used by the kernel
6460 noforce -- Never use bounce buffers (for debugging)
6462 switches= [HW,M68k,EARLY]
6465 Set a sysctl parameter, right before loading the init
6466 process, as if the value was written to the respective
6467 /proc/sys/... file. Both '.' and '/' are recognized as
6468 separators. Unrecognized parameters and invalid values
6469 are reported in the kernel log. Sysctls registered
6470 later by a loaded module cannot be set this way.
6471 Example: sysctl.vm.swappiness=40
6473 sysrq_always_enabled
6475 Ignore sysrq setting - this boot parameter will
6476 neutralize any effect of /proc/sys/kernel/sysrq.
6477 Useful for debugging.
6479 tcpmhash_entries= [KNL,NET]
6480 Set the number of tcp_metrics_hash slots.
6481 Default value is 8192 or 16384 depending on total
6482 ram pages. This is used to specify the TCP metrics
6483 cache size. See Documentation/networking/ip-sysctl.rst
6484 "tcp_no_metrics_save" section for more details.
6488 test_suspend= [SUSPEND]
6489 Format: { "mem" | "standby" | "freeze" }[,N]
6490 Specify "mem" (for Suspend-to-RAM) or "standby" (for
6491 standby suspend) or "freeze" (for suspend type freeze)
6492 as the system sleep state during system startup with
6493 the optional capability to repeat N number of times.
6494 The system is woken from this state using a
6495 wakeup-capable RTC alarm.
6497 thash_entries= [KNL,NET]
6498 Set number of hash buckets for TCP connection
6500 thermal.act= [HW,ACPI]
6501 -1: disable all active trip points in all thermal zones
6502 <degrees C>: override all lowest active trip points
6504 thermal.crt= [HW,ACPI]
6505 -1: disable all critical trip points in all thermal zones
6506 <degrees C>: override all critical trip points
6508 thermal.off= [HW,ACPI]
6509 1: disable ACPI thermal control
6511 thermal.psv= [HW,ACPI]
6512 -1: disable all passive trip points
6513 <degrees C>: override all passive trip points to this
6516 thermal.tzp= [HW,ACPI]
6517 Specify global default ACPI thermal zone polling rate
6518 <deci-seconds>: poll all this frequency
6519 0: no polling (default)
6521 threadirqs [KNL,EARLY]
6522 Force threading of all interrupt handlers except those
6523 marked explicitly IRQF_NO_THREAD.
6525 topology= [S390,EARLY]
6527 Specify if the kernel should make use of the cpu
6528 topology information if the hardware supports this.
6529 The scheduler will make use of this information and
6530 e.g. base its process migration decisions on it.
6533 topology_updates= [KNL, PPC, NUMA]
6535 Specify if the kernel should ignore (off)
6536 topology updates sent by the hypervisor to this
6539 torture.disable_onoff_at_boot= [KNL]
6540 Prevent the CPU-hotplug component of torturing
6541 until after init has spawned.
6543 torture.ftrace_dump_at_shutdown= [KNL]
6544 Dump the ftrace buffer at torture-test shutdown,
6545 even if there were no errors. This can be a
6546 very costly operation when many torture tests
6547 are running concurrently, especially on systems
6548 with rotating-rust storage.
6550 torture.verbose_sleep_frequency= [KNL]
6551 Specifies how many verbose printk()s should be
6552 emitted between each sleep. The default of zero
6553 disables verbose-printk() sleeping.
6555 torture.verbose_sleep_duration= [KNL]
6556 Duration of each verbose-printk() sleep in jiffies.
6560 tpm_suspend_pcr=[HW,TPM]
6561 Format: integer pcr id
6562 Specify that at suspend time, the tpm driver
6563 should extend the specified pcr with zeros,
6564 as a workaround for some chips which fail to
6565 flush the last written pcr on TPM_SaveState.
6566 This will guarantee that all the other pcrs
6569 tpm_tis.interrupts= [HW,TPM]
6570 Enable interrupts for the MMIO based physical layer
6571 for the FIFO interface. By default it is set to false
6572 (0). For more information about TPM hardware interfaces
6573 defined by Trusted Computing Group (TCG) see
6574 https://trustedcomputinggroup.org/resource/pc-client-platform-tpm-profile-ptp-specification/
6577 Have the tracepoints sent to printk as well as the
6578 tracing ring buffer. This is useful for early boot up
6579 where the system hangs or reboots and does not give the
6580 option for reading the tracing buffer or performing a
6581 ftrace_dump_on_oops.
6583 To turn off having tracepoints sent to printk,
6584 echo 0 > /proc/sys/kernel/tracepoint_printk
6585 Note, echoing 1 into this file without the
6586 tracepoint_printk kernel cmdline option has no effect.
6588 The tp_printk_stop_on_boot (see below) can also be used
6589 to stop the printing of events to console at
6594 Having tracepoints sent to printk() and activating high
6595 frequency tracepoints such as irq or sched, can cause
6596 the system to live lock.
6598 tp_printk_stop_on_boot [FTRACE]
6599 When tp_printk (above) is set, it can cause a lot of noise
6600 on the console. It may be useful to only include the
6601 printing of events during boot up, as user space may
6602 make the system inoperable.
6604 This command line option will stop the printing of events
6605 to console at the late_initcall_sync() time frame.
6607 trace_buf_size=nn[KMG]
6608 [FTRACE] will set tracing buffer size on each cpu.
6610 trace_clock= [FTRACE] Set the clock used for tracing events
6612 local - Use the per CPU time stamp counter
6613 (converted into nanoseconds). Fast, but
6614 depending on the architecture, may not be
6615 in sync between CPUs.
6616 global - Event time stamps are synchronize across
6617 CPUs. May be slower than the local clock,
6618 but better for some race conditions.
6619 counter - Simple counting of events (1, 2, ..)
6620 note, some counts may be skipped due to the
6621 infrastructure grabbing the clock more than
6623 uptime - Use jiffies as the time stamp.
6624 perf - Use the same clock that perf uses.
6625 mono - Use ktime_get_mono_fast_ns() for time stamps.
6626 mono_raw - Use ktime_get_raw_fast_ns() for time
6628 boot - Use ktime_get_boot_fast_ns() for time stamps.
6629 Architectures may add more clocks. See
6630 Documentation/trace/ftrace.rst for more details.
6632 trace_event=[event-list]
6633 [FTRACE] Set and start specified trace events in order
6634 to facilitate early boot debugging. The event-list is a
6635 comma-separated list of trace events to enable. See
6636 also Documentation/trace/events.rst
6638 trace_instance=[instance-info]
6639 [FTRACE] Create a ring buffer instance early in boot up.
6640 This will be listed in:
6642 /sys/kernel/tracing/instances
6644 Events can be enabled at the time the instance is created
6647 trace_instance=<name>,<system1>:<event1>,<system2>:<event2>
6649 Note, the "<system*>:" portion is optional if the event is
6652 trace_instance=foo,sched:sched_switch,irq_handler_entry,initcall
6654 will enable the "sched_switch" event (note, the "sched:" is optional, and
6655 the same thing would happen if it was left off). The irq_handler_entry
6656 event, and all events under the "initcall" system.
6658 trace_options=[option-list]
6659 [FTRACE] Enable or disable tracer options at boot.
6660 The option-list is a comma delimited list of options
6661 that can be enabled or disabled just as if you were
6662 to echo the option name into
6664 /sys/kernel/tracing/trace_options
6666 For example, to enable stacktrace option (to dump the
6667 stack trace of each event), add to the command line:
6669 trace_options=stacktrace
6671 See also Documentation/trace/ftrace.rst "trace options"
6674 trace_trigger=[trigger-list]
6675 [FTRACE] Add a event trigger on specific events.
6676 Set a trigger on top of a specific event, with an optional
6679 The format is is "trace_trigger=<event>.<trigger>[ if <filter>],..."
6680 Where more than one trigger may be specified that are comma deliminated.
6684 trace_trigger="sched_switch.stacktrace if prev_state == 2"
6686 The above will enable the "stacktrace" trigger on the "sched_switch"
6687 event but only trigger it if the "prev_state" of the "sched_switch"
6688 event is "2" (TASK_UNINTERUPTIBLE).
6690 See also "Event triggers" in Documentation/trace/events.rst
6694 [FTRACE] enable this option to disable tracing when a
6695 warning is hit. This turns off "tracing_on". Tracing can
6696 be enabled again by echoing '1' into the "tracing_on"
6697 file located in /sys/kernel/tracing/
6699 This option is useful, as it disables the trace before
6700 the WARNING dump is called, which prevents the trace to
6701 be filled with content caused by the warning output.
6703 This option can also be set at run time via the sysctl
6704 option: kernel/traceoff_on_warning
6706 transparent_hugepage=
6708 Format: [always|madvise|never]
6709 Can be used to control the default behavior of the system
6710 with respect to transparent hugepages.
6711 See Documentation/admin-guide/mm/transhuge.rst
6714 trusted.source= [KEYS]
6716 This parameter identifies the trust source as a backend
6717 for trusted keys implementation. Supported trust
6722 If not specified then it defaults to iterating through
6723 the trust source list starting with TPM and assigns the
6724 first trust source as a backend which is initialized
6725 successfully during iteration.
6729 The RNG used to generate key material for trusted keys.
6732 - the same value as trusted.source: "tpm" or "tee"
6734 If not specified, "default" is used. In this case,
6735 the RNG's choice is left to each individual trust source.
6737 tsc= Disable clocksource stability checks for TSC.
6739 [x86] reliable: mark tsc clocksource as reliable, this
6740 disables clocksource verification at runtime, as well
6741 as the stability checks done at bootup. Used to enable
6742 high-resolution timer mode on older hardware, and in
6743 virtualized environment.
6744 [x86] noirqtime: Do not use TSC to do irq accounting.
6745 Used to run time disable IRQ_TIME_ACCOUNTING on any
6746 platforms where RDTSC is slow and this accounting
6748 [x86] unstable: mark the TSC clocksource as unstable, this
6749 marks the TSC unconditionally unstable at bootup and
6750 avoids any further wobbles once the TSC watchdog notices.
6751 [x86] nowatchdog: disable clocksource watchdog. Used
6752 in situations with strict latency requirements (where
6753 interruptions from clocksource watchdog are not
6755 [x86] recalibrate: force recalibration against a HW timer
6756 (HPET or PM timer) on systems whose TSC frequency was
6757 obtained from HW or FW using either an MSR or CPUID(0x15).
6758 Warn if the difference is more than 500 ppm.
6759 [x86] watchdog: Use TSC as the watchdog clocksource with
6760 which to check other HW timers (HPET or PM timer), but
6761 only on systems where TSC has been deemed trustworthy.
6762 This will be suppressed by an earlier tsc=nowatchdog and
6763 can be overridden by a later tsc=nowatchdog. A console
6764 message will flag any such suppression or overriding.
6766 tsc_early_khz= [X86,EARLY] Skip early TSC calibration and use the given
6767 value instead. Useful when the early TSC frequency discovery
6768 procedure is not reliable, such as on overclocked systems
6769 with CPUID.16h support and partial CPUID.15h support.
6770 Format: <unsigned int>
6772 tsx= [X86] Control Transactional Synchronization
6773 Extensions (TSX) feature in Intel processors that
6774 support TSX control.
6776 This parameter controls the TSX feature. The options are:
6778 on - Enable TSX on the system. Although there are
6779 mitigations for all known security vulnerabilities,
6780 TSX has been known to be an accelerator for
6781 several previous speculation-related CVEs, and
6782 so there may be unknown security risks associated
6783 with leaving it enabled.
6785 off - Disable TSX on the system. (Note that this
6786 option takes effect only on newer CPUs which are
6787 not vulnerable to MDS, i.e., have
6788 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
6789 the new IA32_TSX_CTRL MSR through a microcode
6790 update. This new MSR allows for the reliable
6791 deactivation of the TSX functionality.)
6793 auto - Disable TSX if X86_BUG_TAA is present,
6794 otherwise enable TSX on the system.
6796 Not specifying this option is equivalent to tsx=off.
6798 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6801 tsx_async_abort= [X86,INTEL,EARLY] Control mitigation for the TSX Async
6802 Abort (TAA) vulnerability.
6804 Similar to Micro-architectural Data Sampling (MDS)
6805 certain CPUs that support Transactional
6806 Synchronization Extensions (TSX) are vulnerable to an
6807 exploit against CPU internal buffers which can forward
6808 information to a disclosure gadget under certain
6811 In vulnerable processors, the speculatively forwarded
6812 data can be used in a cache side channel attack, to
6813 access data to which the attacker does not have direct
6816 This parameter controls the TAA mitigation. The
6819 full - Enable TAA mitigation on vulnerable CPUs
6822 full,nosmt - Enable TAA mitigation and disable SMT on
6823 vulnerable CPUs. If TSX is disabled, SMT
6824 is not disabled because CPU is not
6825 vulnerable to cross-thread TAA attacks.
6826 off - Unconditionally disable TAA mitigation
6828 On MDS-affected machines, tsx_async_abort=off can be
6829 prevented by an active MDS mitigation as both vulnerabilities
6830 are mitigated with the same mechanism so in order to disable
6831 this mitigation, you need to specify mds=off too.
6833 Not specifying this option is equivalent to
6834 tsx_async_abort=full. On CPUs which are MDS affected
6835 and deploy MDS mitigation, TAA mitigation is not
6836 required and doesn't provide any additional
6840 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6842 turbografx.map[2|3]= [HW,JOY]
6843 TurboGraFX parallel port interface
6845 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
6846 See also Documentation/input/devices/joystick-parport.rst
6848 udbg-immortal [PPC] When debugging early kernel crashes that
6849 happen after console_init() and before a proper
6850 console driver takes over, this boot options might
6851 help "seeing" what's going on.
6853 uhash_entries= [KNL,NET]
6854 Set number of hash buckets for UDP/UDP-Lite connections
6857 [USB] Ignore overcurrent events (default N).
6858 Some badly-designed motherboards generate lots of
6859 bogus events, for ports that aren't wired to
6860 anything. Set this parameter to avoid log spamming.
6861 Note that genuine overcurrent events won't be
6865 [X86] Cause panic on unknown NMI.
6867 unwind_debug [X86-64,EARLY]
6868 Enable unwinder debug output. This can be
6869 useful for debugging certain unwinder error
6870 conditions, including corrupt stacks and
6871 bad/missing unwinder metadata.
6873 usbcore.authorized_default=
6874 [USB] Default USB device authorization:
6875 (default -1 = authorized (same as 1),
6876 0 = not authorized, 1 = authorized, 2 = authorized
6877 if device connected to internal port)
6879 usbcore.autosuspend=
6880 [USB] The autosuspend time delay (in seconds) used
6881 for newly-detected USB devices (default 2). This
6882 is the time required before an idle device will be
6883 autosuspended. Devices for which the delay is set
6884 to a negative value won't be autosuspended at all.
6886 usbcore.usbfs_snoop=
6887 [USB] Set to log all usbfs traffic (default 0 = off).
6889 usbcore.usbfs_snoop_max=
6890 [USB] Maximum number of bytes to snoop in each URB
6893 usbcore.blinkenlights=
6894 [USB] Set to cycle leds on hubs (default 0 = off).
6896 usbcore.old_scheme_first=
6897 [USB] Start with the old device initialization
6898 scheme (default 0 = off).
6900 usbcore.usbfs_memory_mb=
6901 [USB] Memory limit (in MB) for buffers allocated by
6902 usbfs (default = 16, 0 = max = 2047).
6904 usbcore.use_both_schemes=
6905 [USB] Try the other device initialization scheme
6906 if the first one fails (default 1 = enabled).
6908 usbcore.initial_descriptor_timeout=
6909 [USB] Specifies timeout for the initial 64-byte
6910 USB_REQ_GET_DESCRIPTOR request in milliseconds
6911 (default 5000 = 5.0 seconds).
6913 usbcore.nousb [USB] Disable the USB subsystem
6916 [USB] A list of quirk entries to augment the built-in
6917 usb core quirk list. List entries are separated by
6918 commas. Each entry has the form
6919 VendorID:ProductID:Flags. The IDs are 4-digit hex
6920 numbers and Flags is a set of letters. Each letter
6921 will change the built-in quirk; setting it if it is
6922 clear and clearing it if it is set. The letters have
6923 the following meanings:
6924 a = USB_QUIRK_STRING_FETCH_255 (string
6925 descriptors must not be fetched using
6927 b = USB_QUIRK_RESET_RESUME (device can't resume
6928 correctly so reset it instead);
6929 c = USB_QUIRK_NO_SET_INTF (device can't handle
6930 Set-Interface requests);
6931 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
6932 handle its Configuration or Interface
6934 e = USB_QUIRK_RESET (device can't be reset
6935 (e.g morph devices), don't use reset);
6936 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
6937 more interface descriptions than the
6938 bNumInterfaces count, and can't handle
6939 talking to these interfaces);
6940 g = USB_QUIRK_DELAY_INIT (device needs a pause
6941 during initialization, after we read
6942 the device descriptor);
6943 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
6944 high speed and super speed interrupt
6945 endpoints, the USB 2.0 and USB 3.0 spec
6946 require the interval in microframes (1
6947 microframe = 125 microseconds) to be
6948 calculated as interval = 2 ^
6950 Devices with this quirk report their
6951 bInterval as the result of this
6952 calculation instead of the exponent
6953 variable used in the calculation);
6954 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
6955 handle device_qualifier descriptor
6957 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
6958 generates spurious wakeup, ignore
6959 remote wakeup capability);
6960 k = USB_QUIRK_NO_LPM (device can't handle Link
6962 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
6963 (Device reports its bInterval as linear
6964 frames instead of the USB 2.0
6966 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
6967 to be disconnected before suspend to
6968 prevent spurious wakeup);
6969 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
6970 pause after every control message);
6971 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
6972 delay after resetting its port);
6973 p = USB_QUIRK_SHORT_SET_ADDRESS_REQ_TIMEOUT
6974 (Reduce timeout of the SET_ADDRESS
6975 request from 5000 ms to 500 ms);
6976 Example: quirks=0781:5580:bk,0a5c:5834:gij
6979 [USBHID] The interval which mice are to be polled at.
6982 [USBHID] The interval which joysticks are to be polled at.
6985 [USBHID] The interval which keyboards are to be polled at.
6987 usb-storage.delay_use=
6988 [UMS] The delay in seconds before a new device is
6989 scanned for Logical Units (default 1).
6992 [UMS] A list of quirks entries to supplement or
6993 override the built-in unusual_devs list. List
6994 entries are separated by commas. Each entry has
6995 the form VID:PID:Flags where VID and PID are Vendor
6996 and Product ID values (4-digit hex numbers) and
6997 Flags is a set of characters, each corresponding
6998 to a common usb-storage quirk flag as follows:
6999 a = SANE_SENSE (collect more than 18 bytes
7000 of sense data, not on uas);
7001 b = BAD_SENSE (don't collect more than 18
7002 bytes of sense data, not on uas);
7003 c = FIX_CAPACITY (decrease the reported
7004 device capacity by one sector);
7005 d = NO_READ_DISC_INFO (don't use
7006 READ_DISC_INFO command, not on uas);
7007 e = NO_READ_CAPACITY_16 (don't use
7008 READ_CAPACITY_16 command);
7009 f = NO_REPORT_OPCODES (don't use report opcodes
7011 g = MAX_SECTORS_240 (don't transfer more than
7012 240 sectors at a time, uas only);
7013 h = CAPACITY_HEURISTICS (decrease the
7014 reported device capacity by one
7015 sector if the number is odd);
7016 i = IGNORE_DEVICE (don't bind to this
7018 j = NO_REPORT_LUNS (don't use report luns
7020 k = NO_SAME (do not use WRITE_SAME, uas only)
7021 l = NOT_LOCKABLE (don't try to lock and
7022 unlock ejectable media, not on uas);
7023 m = MAX_SECTORS_64 (don't transfer more
7024 than 64 sectors = 32 KB at a time,
7026 n = INITIAL_READ10 (force a retry of the
7027 initial READ(10) command, not on uas);
7028 o = CAPACITY_OK (accept the capacity
7029 reported by the device, not on uas);
7030 p = WRITE_CACHE (the device cache is ON
7031 by default, not on uas);
7032 r = IGNORE_RESIDUE (the device reports
7033 bogus residue values, not on uas);
7034 s = SINGLE_LUN (the device has only one
7036 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
7037 commands, uas only);
7038 u = IGNORE_UAS (don't bind to the uas driver);
7039 w = NO_WP_DETECT (don't test whether the
7040 medium is write-protected).
7041 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
7042 even if the device claims no cache,
7044 Example: quirks=0419:aaf5:rl,0421:0433:rc
7046 user_debug= [KNL,ARM]
7048 See arch/arm/Kconfig.debug help text.
7049 1 - undefined instruction events
7051 4 - invalid data aborts
7054 Example: user_debug=31
7057 [X86,EARLY] Flags controlling user PTE allocations.
7059 nohigh = do not allocate PTE pages in
7060 HIGHMEM regardless of setting
7063 vdso= [X86,SH,SPARC]
7064 On X86_32, this is an alias for vdso32=. Otherwise:
7066 vdso=1: enable VDSO (the default)
7067 vdso=0: disable VDSO mapping
7069 vdso32= [X86] Control the 32-bit vDSO
7070 vdso32=1: enable 32-bit VDSO
7071 vdso32=0 or vdso32=2: disable 32-bit VDSO
7073 See the help text for CONFIG_COMPAT_VDSO for more
7074 details. If CONFIG_COMPAT_VDSO is set, the default is
7075 vdso32=0; otherwise, the default is vdso32=1.
7077 For compatibility with older kernels, vdso32=2 is an
7080 Try vdso32=0 if you encounter an error that says:
7081 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
7084 vector=percpu: enable percpu vector domain
7086 video= [FB,EARLY] Frame buffer configuration
7087 See Documentation/fb/modedb.rst.
7089 video.brightness_switch_enabled= [ACPI]
7091 If set to 1, on receiving an ACPI notify event
7092 generated by hotkey, video driver will adjust brightness
7093 level and then send out the event to user space through
7094 the allocated input device. If set to 0, video driver
7095 will only send out the event without touching backlight
7100 [VMMIO] Memory mapped virtio (platform) device.
7102 <size>@<baseaddr>:<irq>[:<id>]
7104 <size> := size (can use standard suffixes
7106 <baseaddr> := physical base address
7107 <irq> := interrupt number (as passed to
7109 <id> := (optional) platform device id
7111 virtio_mmio.device=1K@0x100b0000:48:7
7113 Can be used multiple times for multiple devices.
7115 vga= [BOOT,X86-32] Select a particular video mode
7116 See Documentation/arch/x86/boot.rst and
7117 Documentation/admin-guide/svga.rst.
7118 Use vga=ask for menu.
7119 This is actually a boot loader parameter; the value is
7120 passed to the kernel using a special protocol.
7122 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
7123 May slow down system boot speed, especially when
7124 enabled on systems with a large amount of memory.
7125 All options are enabled by default, and this
7126 interface is meant to allow for selectively
7127 enabling or disabling specific virtual memory
7130 Available options are:
7131 P Enable page structure init time poisoning
7132 - Disable all of the above options
7134 vmalloc=nn[KMG] [KNL,BOOT,EARLY] Forces the vmalloc area to have an
7135 exact size of <nn>. This can be used to increase
7136 the minimum size (128MB on x86). It can also be
7137 used to decrease the size and leave more room
7138 for directly mapped kernel RAM.
7140 vmcp_cma=nn[MG] [KNL,S390,EARLY]
7141 Sets the memory size reserved for contiguous memory
7142 allocations for the vmcp device driver.
7144 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
7147 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
7150 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
7153 vsyscall= [X86-64,EARLY]
7154 Controls the behavior of vsyscalls (i.e. calls to
7155 fixed addresses of 0xffffffffff600x00 from legacy
7156 code). Most statically-linked binaries and older
7157 versions of glibc use these calls. Because these
7158 functions are at fixed addresses, they make nice
7159 targets for exploits that can control RIP.
7161 emulate Vsyscalls turn into traps and are emulated
7162 reasonably safely. The vsyscall page is
7165 xonly [default] Vsyscalls turn into traps and are
7166 emulated reasonably safely. The vsyscall
7167 page is not readable.
7169 none Vsyscalls don't work at all. This makes
7170 them quite hard to use for exploits but
7171 might break your system.
7173 vt.color= [VT] Default text color.
7174 Format: 0xYX, X = foreground, Y = background.
7175 Default: 0x07 = light gray on black.
7177 vt.cur_default= [VT] Default cursor shape.
7178 Format: 0xCCBBAA, where AA, BB, and CC are the same as
7179 the parameters of the <Esc>[?A;B;Cc escape sequence;
7180 see VGA-softcursor.txt. Default: 2 = underline.
7182 vt.default_blu= [VT]
7183 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
7184 Change the default blue palette of the console.
7185 This is a 16-member array composed of values
7188 vt.default_grn= [VT]
7189 Format: <green0>,<green1>,<green2>,...,<green15>
7190 Change the default green palette of the console.
7191 This is a 16-member array composed of values
7194 vt.default_red= [VT]
7195 Format: <red0>,<red1>,<red2>,...,<red15>
7196 Change the default red palette of the console.
7197 This is a 16-member array composed of values
7203 Set system-wide default UTF-8 mode for all tty's.
7204 Default is 1, i.e. UTF-8 mode is enabled for all
7205 newly opened terminals.
7207 vt.global_cursor_default=
7210 Set system-wide default for whether a cursor
7211 is shown on new VTs. Default is -1,
7212 i.e. cursors will be created by default unless
7213 overridden by individual drivers. 0 will hide
7214 cursors, 1 will display them.
7216 vt.italic= [VT] Default color for italic text; 0-15.
7219 vt.underline= [VT] Default color for underlined text; 0-15.
7222 watchdog timers [HW,WDT] For information on watchdog timers,
7223 see Documentation/watchdog/watchdog-parameters.rst
7224 or other driver-specific files in the
7225 Documentation/watchdog/ directory.
7229 Set the hard lockup detector stall duration
7230 threshold in seconds. The soft lockup detector
7231 threshold is set to twice the value. A value of 0
7232 disables both lockup detectors. Default is 10
7235 workqueue.unbound_cpus=
7236 [KNL,SMP] Specify to constrain one or some CPUs
7237 to use in unbound workqueues.
7239 By default, all online CPUs are available for
7242 workqueue.watchdog_thresh=
7243 If CONFIG_WQ_WATCHDOG is configured, workqueue can
7244 warn stall conditions and dump internal state to
7245 help debugging. 0 disables workqueue stall
7246 detection; otherwise, it's the stall threshold
7247 duration in seconds. The default value is 30 and
7248 it can be updated at runtime by writing to the
7249 corresponding sysfs file.
7251 workqueue.cpu_intensive_thresh_us=
7252 Per-cpu work items which run for longer than this
7253 threshold are automatically considered CPU intensive
7254 and excluded from concurrency management to prevent
7255 them from noticeably delaying other per-cpu work
7256 items. Default is 10000 (10ms).
7258 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel
7259 will report the work functions which violate this
7260 threshold repeatedly. They are likely good
7261 candidates for using WQ_UNBOUND workqueues instead.
7263 workqueue.cpu_intensive_warning_thresh=<uint>
7264 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel
7265 will report the work functions which violate the
7266 intensive_threshold_us repeatedly. In order to prevent
7267 spurious warnings, start printing only after a work
7268 function has violated this threshold number of times.
7270 The default is 4 times. 0 disables the warning.
7272 workqueue.power_efficient
7273 Per-cpu workqueues are generally preferred because
7274 they show better performance thanks to cache
7275 locality; unfortunately, per-cpu workqueues tend to
7276 be more power hungry than unbound workqueues.
7278 Enabling this makes the per-cpu workqueues which
7279 were observed to contribute significantly to power
7280 consumption unbound, leading to measurably lower
7281 power usage at the cost of small performance
7284 The default value of this parameter is determined by
7285 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
7287 workqueue.default_affinity_scope=
7288 Select the default affinity scope to use for unbound
7289 workqueues. Can be one of "cpu", "smt", "cache",
7290 "numa" and "system". Default is "cache". For more
7291 information, see the Affinity Scopes section in
7292 Documentation/core-api/workqueue.rst.
7294 This can be changed after boot by writing to the
7295 matching /sys/module/workqueue/parameters file. All
7296 workqueues with the "default" affinity scope will be
7297 updated accordignly.
7299 workqueue.debug_force_rr_cpu
7300 Workqueue used to implicitly guarantee that work
7301 items queued without explicit CPU specified are put
7302 on the local CPU. This guarantee is no longer true
7303 and while local CPU is still preferred work items
7304 may be put on foreign CPUs. This debug option
7305 forces round-robin CPU selection to flush out
7306 usages which depend on the now broken guarantee.
7307 When enabled, memory and cache locality will be
7310 writecombine= [LOONGARCH,EARLY] Control the MAT (Memory Access
7311 Type) of ioremap_wc().
7313 on - Enable writecombine, use WUC for ioremap_wc()
7314 off - Disable writecombine, use SUC for ioremap_wc()
7316 x2apic_phys [X86-64,APIC,EARLY] Use x2apic physical mode instead of
7317 default x2apic cluster mode on platforms
7320 xen_512gb_limit [KNL,X86-64,XEN]
7321 Restricts the kernel running paravirtualized under Xen
7322 to use only up to 512 GB of RAM. The reason to do so is
7323 crash analysis tools and Xen tools for doing domain
7324 save/restore/migration must be enabled to handle larger
7327 xen_emul_unplug= [HW,X86,XEN,EARLY]
7328 Unplug Xen emulated devices
7329 Format: [unplug0,][unplug1]
7330 ide-disks -- unplug primary master IDE devices
7331 aux-ide-disks -- unplug non-primary-master IDE devices
7332 nics -- unplug network devices
7333 all -- unplug all emulated devices (NICs and IDE disks)
7334 unnecessary -- unplugging emulated devices is
7335 unnecessary even if the host did not respond to
7337 never -- do not unplug even if version check succeeds
7339 xen_legacy_crash [X86,XEN,EARLY]
7340 Crash from Xen panic notifier, without executing late
7341 panic() code such as dumping handler.
7343 xen_msr_safe= [X86,XEN,EARLY]
7345 Select whether to always use non-faulting (safe) MSR
7346 access functions when running as Xen PV guest. The
7347 default value is controlled by CONFIG_XEN_PV_MSR_SAFE.
7349 xen_nopvspin [X86,XEN,EARLY]
7350 Disables the qspinlock slowpath using Xen PV optimizations.
7351 This parameter is obsoleted by "nopvspin" parameter, which
7352 has equivalent effect for XEN platform.
7355 Disables the PV optimizations forcing the HVM guest to
7356 run as generic HVM guest with no PV drivers.
7357 This option is obsoleted by the "nopv" option, which
7358 has equivalent effect for XEN platform.
7360 xen_no_vector_callback
7361 [KNL,X86,XEN,EARLY] Disable the vector callback for Xen
7362 event channel interrupts.
7364 xen_scrub_pages= [XEN]
7365 Boolean option to control scrubbing pages before giving them back
7366 to Xen, for use by other domains. Can be also changed at runtime
7367 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
7368 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
7370 xen_timer_slop= [X86-64,XEN,EARLY]
7371 Set the timer slop (in nanoseconds) for the virtual Xen
7372 timers (default is 100000). This adjusts the minimum
7373 delta of virtualized Xen timers, where lower values
7374 improve timer resolution at the expense of processing
7375 more timer interrupts.
7377 xen.balloon_boot_timeout= [XEN]
7378 The time (in seconds) to wait before giving up to boot
7379 in case initial ballooning fails to free enough memory.
7380 Applies only when running as HVM or PVH guest and
7381 started with less memory configured than allowed at
7382 max. Default is 180.
7384 xen.event_eoi_delay= [XEN]
7385 How long to delay EOI handling in case of event
7386 storms (jiffies). Default is 10.
7388 xen.event_loop_timeout= [XEN]
7389 After which time (jiffies) the event handling loop
7390 should start to delay EOI handling. Default is 2.
7392 xen.fifo_events= [XEN]
7393 Boolean parameter to disable using fifo event handling
7394 even if available. Normally fifo event handling is
7395 preferred over the 2-level event handling, as it is
7396 fairer and the number of possible event channels is
7397 much higher. Default is on (use fifo events).
7399 xirc2ps_cs= [NET,PCMCIA]
7401 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
7404 By default on POWER9 and above, the kernel will
7405 natively use the XIVE interrupt controller. This option
7406 allows the fallback firmware mode to be used:
7408 off Fallback to firmware control of XIVE interrupt
7409 controller on both pseries and powernv
7410 platforms. Only useful on POWER9 and above.
7412 xive.store-eoi=off [PPC]
7413 By default on POWER10 and above, the kernel will use
7414 stores for EOI handling when the XIVE interrupt mode
7415 is active. This option allows the XIVE driver to use
7416 loads instead, as on POWER9.
7418 xhci-hcd.quirks [USB,KNL]
7419 A hex value specifying bitmask with supplemental xhci
7420 host controller quirks. Meaning of each bit can be
7421 consulted in header drivers/usb/host/xhci.h.
7424 Format: { early | on | rw | ro | off }
7425 Controls if xmon debugger is enabled. Default is off.
7426 Passing only "xmon" is equivalent to "xmon=early".
7427 early Call xmon as early as possible on boot; xmon
7428 debugger is called from setup_arch().
7429 on xmon debugger hooks will be installed so xmon
7430 is only called on a kernel crash. Default mode,
7431 i.e. either "ro" or "rw" mode, is controlled
7432 with CONFIG_XMON_DEFAULT_RO_MODE.
7433 rw xmon debugger hooks will be installed so xmon
7434 is called only on a kernel crash, mode is write,
7435 meaning SPR registers, memory and, other data
7436 can be written using xmon commands.
7437 ro same as "rw" option above but SPR registers,
7438 memory, and other data can't be written using
7440 off xmon is disabled.