1 acpi= [HW,ACPI,X86,ARM64]
2 Advanced Configuration and Power Interface
3 Format: { force | on | off | strict | noirq | rsdt |
5 force -- enable ACPI if default was off
6 on -- enable ACPI but allow fallback to DT [arm64]
7 off -- disable ACPI if default was on
8 noirq -- do not use ACPI for IRQ routing
9 strict -- Be less tolerant of platforms that are not
10 strictly ACPI specification compliant.
11 rsdt -- prefer RSDT over (default) XSDT
12 copy_dsdt -- copy DSDT to memory
13 For ARM64, ONLY "acpi=off", "acpi=on" or "acpi=force"
16 See also Documentation/power/runtime_pm.rst, pci=noacpi
18 acpi_apic_instance= [ACPI, IOAPIC]
20 2: use 2nd APIC table, if available
21 1,0: use 1st APIC table
24 acpi_backlight= [HW,ACPI]
25 { vendor | video | native | none }
26 If set to vendor, prefer vendor-specific driver
27 (e.g. thinkpad_acpi, sony_acpi, etc.) instead
28 of the ACPI video.ko driver.
29 If set to video, use the ACPI video.ko driver.
30 If set to native, use the device's native backlight mode.
31 If set to none, disable the ACPI backlight interface.
33 acpi_force_32bit_fadt_addr
34 force FADT to use 32 bit addresses rather than the
35 64 bit X_* addresses. Some firmware have broken 64
36 bit addresses for force ACPI ignore these and use
37 the older legacy 32 bit addresses.
39 acpica_no_return_repair [HW, ACPI]
40 Disable AML predefined validation mechanism
41 This mechanism can repair the evaluation result to make
42 the return objects more ACPI specification compliant.
43 This option is useful for developers to identify the
44 root cause of an AML interpreter issue when the issue
45 has something to do with the repair mechanism.
47 acpi.debug_layer= [HW,ACPI,ACPI_DEBUG]
48 acpi.debug_level= [HW,ACPI,ACPI_DEBUG]
50 CONFIG_ACPI_DEBUG must be enabled to produce any ACPI
51 debug output. Bits in debug_layer correspond to a
52 _COMPONENT in an ACPI source file, e.g.,
53 #define _COMPONENT ACPI_EVENTS
54 Bits in debug_level correspond to a level in
55 ACPI_DEBUG_PRINT statements, e.g.,
56 ACPI_DEBUG_PRINT((ACPI_DB_INFO, ...
57 The debug_level mask defaults to "info". See
58 Documentation/firmware-guide/acpi/debug.rst for more information about
59 debug layers and levels.
61 Enable processor driver info messages:
62 acpi.debug_layer=0x20000000
63 Enable AML "Debug" output, i.e., stores to the Debug
64 object while interpreting AML:
65 acpi.debug_layer=0xffffffff acpi.debug_level=0x2
66 Enable all messages related to ACPI hardware:
67 acpi.debug_layer=0x2 acpi.debug_level=0xffffffff
69 Some values produce so much output that the system is
70 unusable. The "log_buf_len" parameter may be useful
71 if you need to capture more output.
73 acpi_enforce_resources= [ACPI]
75 Check for resource conflicts between native drivers
76 and ACPI OperationRegions (SystemIO and SystemMemory
77 only). IO ports and memory declared in ACPI might be
78 used by the ACPI subsystem in arbitrary AML code and
79 can interfere with legacy drivers.
80 strict (default): access to resources claimed by ACPI
81 is denied; legacy drivers trying to access reserved
82 resources will fail to bind to device using them.
83 lax: access to resources claimed by ACPI is allowed;
84 legacy drivers trying to access reserved resources
85 will bind successfully but a warning message is logged.
86 no: ACPI OperationRegions are not marked as reserved,
87 no further checks are performed.
89 acpi_force_table_verification [HW,ACPI]
90 Enable table checksum verification during early stage.
91 By default, this is disabled due to x86 early mapping
94 acpi_irq_balance [HW,ACPI]
95 ACPI will balance active IRQs
98 acpi_irq_nobalance [HW,ACPI]
99 ACPI will not move active IRQs (default)
102 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA
103 Format: <irq>,<irq>...
105 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for
107 Format: <irq>,<irq>...
109 acpi_mask_gpe= [HW,ACPI]
110 Due to the existence of _Lxx/_Exx, some GPEs triggered
111 by unsupported hardware/firmware features can result in
112 GPE floodings that cannot be automatically disabled by
114 This facility can be used to prevent such uncontrolled
116 Format: <byte> or <bitmap-list>
118 acpi_no_auto_serialize [HW,ACPI]
119 Disable auto-serialization of AML methods
120 AML control methods that contain the opcodes to create
121 named objects will be marked as "Serialized" by the
122 auto-serialization feature.
123 This feature is enabled by default.
124 This option allows to turn off the feature.
126 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump
129 acpi_no_static_ssdt [HW,ACPI]
130 Disable installation of static SSDTs at early boot time
131 By default, SSDTs contained in the RSDT/XSDT will be
132 installed automatically and they will appear under
133 /sys/firmware/acpi/tables.
134 This option turns off this feature.
135 Note that specifying this option does not affect
136 dynamic table installation which will install SSDT
137 tables to /sys/firmware/acpi/tables/dynamic.
139 acpi_no_watchdog [HW,ACPI,WDT]
140 Ignore the ACPI-based watchdog interface (WDAT) and let
141 a native driver control the watchdog device instead.
143 acpi_rsdp= [ACPI,EFI,KEXEC]
144 Pass the RSDP address to the kernel, mostly used
145 on machines running EFI runtime service to boot the
146 second kernel for kdump.
148 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS
149 Format: To spoof as Windows 98: ="Microsoft Windows"
151 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead
152 of 2 which is mandated by ACPI 6) as the supported ACPI
153 specification revision (when using this switch, it may
154 be necessary to carry out a cold reboot _twice_ in a
155 row to make it take effect on the platform firmware).
157 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings
158 acpi_osi="string1" # add string1
159 acpi_osi="!string2" # remove string2
160 acpi_osi=!* # remove all strings
161 acpi_osi=! # disable all built-in OS vendor
163 acpi_osi=!! # enable all built-in OS vendor
165 acpi_osi= # disable all strings
167 'acpi_osi=!' can be used in combination with single or
168 multiple 'acpi_osi="string1"' to support specific OS
169 vendor string(s). Note that such command can only
170 affect the default state of the OS vendor strings, thus
171 it cannot affect the default state of the feature group
172 strings and the current state of the OS vendor strings,
173 specifying it multiple times through kernel command line
174 is meaningless. This command is useful when one do not
175 care about the state of the feature group strings which
176 should be controlled by the OSPM.
178 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent
179 to 'acpi_osi="Windows 2000" acpi_osi=!', they all
180 can make '_OSI("Windows 2000")' TRUE.
182 'acpi_osi=' cannot be used in combination with other
183 'acpi_osi=' command lines, the _OSI method will not
184 exist in the ACPI namespace. NOTE that such command can
185 only affect the _OSI support state, thus specifying it
186 multiple times through kernel command line is also
189 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)'
192 'acpi_osi=!*' can be used in combination with single or
193 multiple 'acpi_osi="string1"' to support specific
194 string(s). Note that such command can affect the
195 current state of both the OS vendor strings and the
196 feature group strings, thus specifying it multiple times
197 through kernel command line is meaningful. But it may
198 still not able to affect the final state of a string if
199 there are quirks related to this string. This command
200 is useful when one want to control the state of the
201 feature group strings to debug BIOS issues related to
204 1. 'acpi_osi="Module Device" acpi_osi=!*' can make
205 '_OSI("Module Device")' FALSE.
206 2. 'acpi_osi=!* acpi_osi="Module Device"' can make
207 '_OSI("Module Device")' TRUE.
208 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is
210 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"'
212 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!',
213 they all will make '_OSI("Windows 2000")' TRUE.
216 Override the pmtimer bug detection: force the kernel
217 to assume that this machine's pmtimer latches its value
218 and always returns good values.
220 acpi_sci= [HW,ACPI] ACPI System Control Interrupt trigger mode
221 Format: { level | edge | high | low }
223 acpi_skip_timer_override [HW,ACPI]
224 Recognize and ignore IRQ0/pin2 Interrupt Override.
225 For broken nForce2 BIOS resulting in XT-PIC timer.
227 acpi_sleep= [HW,ACPI] Sleep options
228 Format: { s3_bios, s3_mode, s3_beep, s4_hwsig,
229 s4_nohwsig, old_ordering, nonvs,
230 sci_force_enable, nobl }
231 See Documentation/power/video.rst for information on
233 s3_beep is for debugging; it makes the PC's speaker beep
234 as soon as the kernel's real-mode entry point is called.
235 s4_hwsig causes the kernel to check the ACPI hardware
236 signature during resume from hibernation, and gracefully
237 refuse to resume if it has changed. This complies with
238 the ACPI specification but not with reality, since
239 Windows does not do this and many laptops do change it
240 on docking. So the default behaviour is to allow resume
241 and simply warn when the signature changes, unless the
242 s4_hwsig option is enabled.
243 s4_nohwsig prevents ACPI hardware signature from being
244 used (or even warned about) during resume.
245 old_ordering causes the ACPI 1.0 ordering of the _PTS
246 control method, with respect to putting devices into
247 low power states, to be enforced (the ACPI 2.0 ordering
248 of _PTS is used by default).
249 nonvs prevents the kernel from saving/restoring the
250 ACPI NVS memory during suspend/hibernation and resume.
251 sci_force_enable causes the kernel to set SCI_EN directly
252 on resume from S1/S3 (which is against the ACPI spec,
253 but some broken systems don't work without it).
254 nobl causes the internal blacklist of systems known to
255 behave incorrectly in some ways with respect to system
256 suspend and resume to be ignored (use wisely).
258 acpi_use_timer_override [HW,ACPI]
259 Use timer override. For some broken Nvidia NF5 boards
260 that require a timer override, but don't have HPET
262 add_efi_memmap [EFI; X86] Include EFI memory map in
263 kernel's map of available physical RAM.
266 { off | try_unsupported }
267 off: disable AGP support
268 try_unsupported: try to drive unsupported chipsets
269 (may crash computer or cause data corruption)
272 See Documentation/sound/alsa-configuration.rst
275 Allow the default userspace alignment fault handler
276 behaviour to be specified. Bit 0 enables warnings,
277 bit 1 enables fixups, and bit 2 sends a segfault.
279 align_va_addr= [X86-64]
280 Align virtual addresses by clearing slice [14:12] when
281 allocating a VMA at process creation time. This option
282 gives you up to 3% performance improvement on AMD F15h
283 machines (where it is enabled by default) for a
284 CPU-intensive style benchmark, and it can vary highly in
285 a microbenchmark depending on workload and compiler.
287 32: only for 32-bit processes
288 64: only for 64-bit processes
289 on: enable for both 32- and 64-bit processes
290 off: disable for both 32- and 64-bit processes
292 alloc_snapshot [FTRACE]
293 Allocate the ftrace snapshot buffer on boot up when the
294 main buffer is allocated. This is handy if debugging
295 and you need to use tracing_snapshot() on boot up, and
296 do not want to use tracing_snapshot_alloc() as it needs
297 to be done where GFP_KERNEL allocations are allowed.
299 allow_mismatched_32bit_el0 [ARM64]
300 Allow execve() of 32-bit applications and setting of the
301 PER_LINUX32 personality on systems where only a strict
302 subset of the CPUs support 32-bit EL0. When this
303 parameter is present, the set of CPUs supporting 32-bit
304 EL0 is indicated by /sys/devices/system/cpu/aarch32_el0
305 and hot-unplug operations may be restricted.
307 See Documentation/arm64/asymmetric-32bit.rst for more
310 amd_iommu= [HW,X86-64]
311 Pass parameters to the AMD IOMMU driver in the system.
313 fullflush - Deprecated, equivalent to iommu.strict=1
314 off - do not initialize any AMD IOMMU found in
316 force_isolation - Force device isolation for all
317 devices. The IOMMU driver is not
318 allowed anymore to lift isolation
319 requirements as needed. This option
320 does not override iommu=pt
321 force_enable - Force enable the IOMMU on platforms known
322 to be buggy with IOMMU enabled. Use this
325 amd_iommu_dump= [HW,X86-64]
326 Enable AMD IOMMU driver option to dump the ACPI table
327 for AMD IOMMU. With this option enabled, AMD IOMMU
328 driver will print ACPI tables for AMD IOMMU during
329 IOMMU initialization.
331 amd_iommu_intr= [HW,X86-64]
332 Specifies one of the following AMD IOMMU interrupt
334 legacy - Use legacy interrupt remapping mode.
335 vapic - Use virtual APIC mode, which allows IOMMU
336 to inject interrupts directly into guest.
337 This mode requires kvm-amd.avic=1.
338 (Default when IOMMU HW support is present.)
340 amijoy.map= [HW,JOY] Amiga joystick support
341 Map of devices attached to JOY0DAT and JOY1DAT
343 See also Documentation/input/joydev/joystick.rst
345 analog.map= [HW,JOY] Analog joystick and gamepad support
346 Specifies type or capabilities of an analog joystick
347 connected to one of 16 gameports
348 Format: <type1>,<type2>,..<type16>
351 Power management functions (SPARCstation-4/5 + deriv.)
353 Disable APC CPU standby support. SPARCstation-Fox does
354 not play well with APC CPU idle - disable it if you have
355 APC and your system crashes randomly.
357 apic= [APIC,X86] Advanced Programmable Interrupt Controller
358 Change the output verbosity while booting
359 Format: { quiet (default) | verbose | debug }
360 Change the amount of debugging information output
361 when initialising the APIC and IO-APIC components.
362 For X86-32, this can also be used to specify an APIC
364 Format: apic=driver_name
365 Examples: apic=bigsmp
367 apic_extnmi= [APIC,X86] External NMI delivery setting
368 Format: { bsp (default) | all | none }
369 bsp: External NMI is delivered only to CPU 0
370 all: External NMIs are broadcast to all CPUs as a
372 none: External NMI is masked for all CPUs. This is
373 useful so that a dump capture kernel won't be
377 See Documentation/networking/ipv6.rst.
379 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
380 Limit apic dumping. The parameter defines the maximal
381 number of local apics being dumped. Also it is possible
382 to set it to "all" by meaning -- no limit here.
383 Format: { 1 (default) | 2 | ... | all }.
384 The parameter valid if only apic=debug or
385 apic=verbose is specified.
386 Example: apic=debug show_lapic=all
388 apm= [APM] Advanced Power Management
389 See header of arch/x86/kernel/apm_32.c.
391 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
392 Format: <io>,<irq>,<nodeID>
394 arm64.nobti [ARM64] Unconditionally disable Branch Target
395 Identification support
397 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication
400 arm64.nomte [ARM64] Unconditionally disable Memory Tagging Extension
405 atarimouse= [HW,MOUSE] Atari Mouse
407 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
408 EzKey and similar keyboards
410 atkbd.reset= [HW] Reset keyboard during initialization
412 atkbd.set= [HW] Select keyboard code set
413 Format: <int> (2 = AT (default), 3 = PS/2)
415 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
418 atkbd.softraw= [HW] Choose between synthetic and real raw mode
419 Format: <bool> (0 = real, 1 = synthetic (default))
421 atkbd.softrepeat= [HW]
422 Use software keyboard repeat
424 audit= [KNL] Enable the audit sub-system
425 Format: { "0" | "1" | "off" | "on" }
426 0 | off - kernel audit is disabled and can not be
427 enabled until the next reboot
428 unset - kernel audit is initialized but disabled and
429 will be fully enabled by the userspace auditd.
430 1 | on - kernel audit is initialized and partially
431 enabled, storing at most audit_backlog_limit
432 messages in RAM until it is fully enabled by the
436 audit_backlog_limit= [KNL] Set the audit queue size limit.
437 Format: <int> (must be >=0)
440 bau= [X86_UV] Enable the BAU on SGI UV. The default
441 behavior is to disable the BAU (i.e. bau=0).
442 Format: { "0" | "1" }
445 unset - Disable the BAU.
447 baycom_epp= [HW,AX25]
450 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
452 See header of drivers/net/hamradio/baycom_par.c.
454 baycom_ser_fdx= [HW,AX25]
455 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
456 Format: <io>,<irq>,<mode>[,<baud>]
457 See header of drivers/net/hamradio/baycom_ser_fdx.c.
459 baycom_ser_hdx= [HW,AX25]
460 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
461 Format: <io>,<irq>,<mode>
462 See header of drivers/net/hamradio/baycom_ser_hdx.c.
464 blkdevparts= Manual partition parsing of block device(s) for
465 embedded devices based on command line input.
466 See Documentation/block/cmdline-partition.rst
468 boot_delay= Milliseconds to delay each printk during boot.
469 Values larger than 10 seconds (10000) are changed to
474 Extended command line options can be added to an initrd
475 and this will cause the kernel to look for it.
477 See Documentation/admin-guide/bootconfig.rst
480 Disable BERT OS support on buggy BIOSes.
482 bgrt_disable [ACPI][X86]
483 Disable BGRT to avoid flickering OEM logo.
485 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
486 bttv.radio= Most important insmod options are available as
488 bttv.pll= See Documentation/admin-guide/media/bttv.rst
491 bulk_remove=off [PPC] This parameter disables the use of the pSeries
492 firmware feature for flushing multiple hpte entries
495 c101= [NET] Moxa C101 synchronous serial card
497 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
498 Sometimes CPU hardware bugs make them report the cache
499 size incorrectly. The kernel will attempt work arounds
500 to fix known problems, but for some CPUs it is not
501 possible to determine what the correct size should be.
502 This option provides an override for these situations.
505 [NET] Specifies amount of time (in seconds) that
506 the kernel should wait for a network carrier. By default
507 it waits 120 seconds.
509 ca_keys= [KEYS] This parameter identifies a specific key(s) on
510 the system trusted keyring to be used for certificate
512 format: { id:<keyid> | builtin }
514 cca= [MIPS] Override the kernel pages' cache coherency
515 algorithm. Accepted values range from 0 to 7
516 inclusive. See arch/mips/include/asm/pgtable-bits.h
517 for platform specific values (SB1, Loongson3 and
520 ccw_timeout_log [S390]
521 See Documentation/s390/common_io.rst for details.
523 cgroup_disable= [KNL] Disable a particular controller or optional feature
524 Format: {name of the controller(s) or feature(s) to disable}
525 The effects of cgroup_disable=foo are:
526 - foo isn't auto-mounted if you mount all cgroups in
528 - foo isn't visible as an individually mountable
530 - if foo is an optional feature then the feature is
531 disabled and corresponding cgroup files are not
533 {Currently only "memory" controller deal with this and
534 cut the overhead, others just disable the usage. So
535 only cgroup_disable=memory is actually worthy}
536 Specifying "pressure" disables per-cgroup pressure
537 stall information accounting feature
539 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
540 Format: { { controller | "all" | "named" }
541 [,{ controller | "all" | "named" }...] }
542 Like cgroup_disable, but only applies to cgroup v1;
543 the blacklisted controllers remain available in cgroup2.
544 "all" blacklists all controllers and "named" disables
545 named mounts. Specifying both "all" and "named" disables
548 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
550 nosocket -- Disable socket memory accounting.
551 nokmem -- Disable kernel memory accounting.
553 checkreqprot [SELINUX] Set initial checkreqprot flag value.
554 Format: { "0" | "1" }
555 See security/selinux/Kconfig help text.
556 0 -- check protection applied by kernel (includes
557 any implied execute protection).
558 1 -- check protection requested by application.
559 Default value is set via a kernel config option.
560 Value can be changed at runtime via
561 /sys/fs/selinux/checkreqprot.
562 Setting checkreqprot to 1 is deprecated.
565 See Documentation/s390/common_io.rst for details.
568 Prevents the clock framework from automatically gating
569 clocks that have not been explicitly enabled by a Linux
570 device driver but are enabled in hardware at reset or
571 by the bootloader/firmware. Note that this does not
572 force such clocks to be always-on nor does it reserve
573 those clocks in any way. This parameter is useful for
574 debug and development, but should not be needed on a
575 platform with proper driver support. For more
576 information, see Documentation/driver-api/clk.rst.
578 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
580 Forces specified clocksource (if available) to be used
581 when calculating gettimeofday(). If specified
582 clocksource is not available, it defaults to PIT.
583 Format: { pit | tsc | cyclone | pmtmr }
585 clocksource= Override the default clocksource
587 Override the default clocksource and use the clocksource
588 with the name specified.
589 Some clocksource names to choose from, depending on
591 [all] jiffies (this is the base, fallback clocksource)
593 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
594 pxa_timer,timer3,32k_counter,timer0_1
595 [X86-32] pit,hpet,tsc;
596 scx200_hrt on Geode; cyclone on IBM x440
604 clocksource.arm_arch_timer.evtstrm=
607 Enable/disable the eventstream feature of the ARM
608 architected timer so that code using WFE-based polling
609 loops can be debugged more effectively on production
612 clocksource.max_cswd_read_retries= [KNL]
613 Number of clocksource_watchdog() retries due to
614 external delays before the clock will be marked
615 unstable. Defaults to two retries, that is,
616 three attempts to read the clock under test.
618 clocksource.verify_n_cpus= [KNL]
619 Limit the number of CPUs checked for clocksources
620 marked with CLOCK_SOURCE_VERIFY_PERCPU that
621 are marked unstable due to excessive skew.
622 A negative value says to check all CPUs, while
623 zero says not to check any. Values larger than
624 nr_cpu_ids are silently truncated to nr_cpu_ids.
625 The actual CPUs are chosen randomly, with
626 no replacement if the same CPU is chosen twice.
628 clocksource-wdtest.holdoff= [KNL]
629 Set the time in seconds that the clocksource
630 watchdog test waits before commencing its tests.
631 Defaults to zero when built as a module and to
632 10 seconds when built into the kernel.
634 clearcpuid=BITNUM[,BITNUM...] [X86]
635 Disable CPUID feature X for the kernel. See
636 arch/x86/include/asm/cpufeatures.h for the valid bit
637 numbers. Note the Linux specific bits are not necessarily
638 stable over kernel options, but the vendor specific
640 Also note that user programs calling CPUID directly
641 or using the feature without checking anything
642 will still see it. This just prevents it from
643 being used by the kernel or shown in /proc/cpuinfo.
644 Also note the kernel might malfunction if you disable
647 cma=nn[MG]@[start[MG][-end[MG]]]
649 Sets the size of kernel global memory area for
650 contiguous memory allocations and optionally the
651 placement constraint by the physical address range of
652 memory allocations. A value of 0 disables CMA
653 altogether. For more information, see
654 kernel/dma/contiguous.c
658 Sets the size of kernel per-numa memory area for
659 contiguous memory allocations. A value of 0 disables
660 per-numa CMA altogether. And If this option is not
661 specificed, the default value is 0.
662 With per-numa CMA enabled, DMA users on node nid will
663 first try to allocate buffer from the pernuma area
664 which is located in node nid, if the allocation fails,
665 they will fallback to the global default memory area.
667 cmo_free_hint= [PPC] Format: { yes | no }
668 Specify whether pages are marked as being inactive
669 when they are freed. This is used in CMO environments
670 to determine OS memory pressure for page stealing by
674 coherent_pool=nn[KMG] [ARM,KNL]
675 Sets the size of memory pool for coherent, atomic dma
676 allocations, by default set to 256K.
678 com20020= [HW,NET] ARCnet - COM20020 chipset
680 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
682 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
686 ARCnet - COM90xx chipset (memory-mapped buffers)
687 Format: <io>[,<irq>[,<memstart>]]
689 condev= [HW,S390] console device
692 console= [KNL] Output console device and options.
694 tty<n> Use the virtual console device <n>.
698 Use the specified serial port. The options are of
699 the form "bbbbpnf", where "bbbb" is the baud rate,
700 "p" is parity ("n", "o", or "e"), "n" is number of
701 bits, and "f" is flow control ("r" for RTS or
702 omit it). Default is "9600n8".
704 See Documentation/admin-guide/serial-console.rst for more
706 Documentation/networking/netconsole.rst for an
709 uart[8250],io,<addr>[,options]
710 uart[8250],mmio,<addr>[,options]
711 uart[8250],mmio16,<addr>[,options]
712 uart[8250],mmio32,<addr>[,options]
713 uart[8250],0x<addr>[,options]
714 Start an early, polled-mode console on the 8250/16550
715 UART at the specified I/O port or MMIO address,
716 switching to the matching ttyS device later.
717 MMIO inter-register address stride is either 8-bit
718 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
719 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
720 to be equivalent to 'mmio'. 'options' are specified in
721 the same format described for ttyS above; if unspecified,
722 the h/w is not re-initialized.
724 hvc<n> Use the hypervisor console device <n>. This is for
725 both Xen and PowerPC hypervisors.
727 If the device connected to the port is not a TTY but a braille
728 device, prepend "brl," before the device type, for instance
730 For now, only VisioBraille is supported.
733 [KNL] Change console messages format
735 By default we print messages on consoles in
736 "[time stamp] text\n" format (time stamp may not be
737 printed, depending on CONFIG_PRINTK_TIME or
738 `printk_time' param).
740 Switch to syslog format: "<%u>[time stamp] text\n"
741 IOW, each message will have a facility and loglevel
742 prefix. The format is similar to one used by syslog()
743 syscall, or to executing "dmesg -S --raw" or to reading
746 consoleblank= [KNL] The console blank (screen saver) timeout in
747 seconds. A value of 0 disables the blank timer.
751 [KNL] Change the default value for
752 /proc/<pid>/coredump_filter.
753 See also Documentation/filesystems/proc.rst.
755 coresight_cpu_debug.enable
758 Enable/disable the CPU sampling based debugging.
759 0: default value, disable debugging
760 1: enable debugging at boot time
762 cpuidle.off=1 [CPU_IDLE]
763 disable the cpuidle sub-system
766 [CPU_IDLE] Name of the cpuidle governor to use.
768 cpufreq.off=1 [CPU_FREQ]
769 disable the cpufreq sub-system
771 cpufreq.default_governor=
772 [CPU_FREQ] Name of the default cpufreq governor or
773 policy to use. This governor must be registered in the
774 kernel before the cpufreq driver probes.
777 [X86] Delay for N microsec between assert and de-assert
778 of APIC INIT to start processors. This delay occurs
779 on every CPU online, such as boot, and resume from suspend.
782 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
784 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
786 crashkernel=size[KMG][@offset[KMG]]
787 [KNL] Using kexec, Linux can switch to a 'crash kernel'
788 upon panic. This parameter reserves the physical
789 memory region [offset, offset + size] for that kernel
790 image. If '@offset' is omitted, then a suitable offset
791 is selected automatically.
792 [KNL, X86-64] Select a region under 4G first, and
793 fall back to reserve region above 4G when '@offset'
794 hasn't been specified.
795 See Documentation/admin-guide/kdump/kdump.rst for further details.
797 crashkernel=range1:size1[,range2:size2,...][@offset]
798 [KNL] Same as above, but depends on the memory
799 in the running system. The syntax of range is
800 start-[end] where start and end are both
801 a memory unit (amount[KMG]). See also
802 Documentation/admin-guide/kdump/kdump.rst for an example.
804 crashkernel=size[KMG],high
805 [KNL, X86-64] range could be above 4G. Allow kernel
806 to allocate physical memory region from top, so could
807 be above 4G if system have more than 4G ram installed.
808 Otherwise memory region will be allocated below 4G, if
810 It will be ignored if crashkernel=X is specified.
811 crashkernel=size[KMG],low
812 [KNL, X86-64] range under 4G. When crashkernel=X,high
813 is passed, kernel could allocate physical memory region
814 above 4G, that cause second kernel crash on system
815 that require some amount of low memory, e.g. swiotlb
816 requires at least 64M+32K low memory, also enough extra
817 low memory is needed to make sure DMA buffers for 32-bit
818 devices won't run out. Kernel would try to allocate at
819 at least 256M below 4G automatically.
820 This one let user to specify own low range under 4G
821 for second kernel instead.
822 0: to disable low allocation.
823 It will be ignored when crashkernel=X,high is not used
824 or memory reserved is below 4G.
827 [KNL] Disable crypto self-tests
832 cs89x0_media= [HW,NET]
833 Format: { rj45 | aui | bnc }
835 csdlock_debug= [KNL] Enable debug add-ons of cross-CPU function call
836 handling. When switched on, additional debug data is
837 printed to the console in case a hanging CPU is
838 detected, and that CPU is pinged again in order to try
839 to resolve the hang situation.
840 0: disable csdlock debugging (default)
841 1: enable basic csdlock debugging (minor impact)
842 ext: enable extended csdlock debugging (more impact,
846 See header of drivers/s390/block/dasd_devmap.c.
848 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
849 (one device per port)
850 Format: <port#>,<type>
851 See also Documentation/input/devices/joystick-parport.rst
853 debug [KNL] Enable kernel debugging (events log level).
856 [KNL] Enable printing [hashed] pointers early in the
857 boot sequence. If enabled, we use a weak hash instead
858 of siphash to hash pointers. Use this option if you are
859 seeing instances of '(___ptrval___)') and need to see a
860 value (hashed pointer) instead. Cryptographically
861 insecure, please do not use on production kernels.
864 [KNL] verbose locking self-tests
866 Print debugging info while doing the locking API
868 Bitmask for the various LOCKTYPE_ tests. Defaults to 0
869 (no extra messages), setting it to -1 (all bits set)
870 will print _a_lot_ more information - normally only
871 useful to lockdep developers.
873 debug_objects [KNL] Enable object debugging
876 [KNL] Disable object debugging
878 debug_guardpage_minorder=
879 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
880 parameter allows control of the order of pages that will
881 be intentionally kept free (and hence protected) by the
882 buddy allocator. Bigger value increase the probability
883 of catching random memory corruption, but reduce the
884 amount of memory for normal system use. The maximum
885 possible value is MAX_ORDER/2. Setting this parameter
886 to 1 or 2 should be enough to identify most random
887 memory corruption problems caused by bugs in kernel or
888 driver code when a CPU writes to (or reads from) a
889 random memory location. Note that there exists a class
890 of memory corruptions problems caused by buggy H/W or
891 F/W or by drivers badly programing DMA (basically when
892 memory is written at bus level and the CPU MMU is
893 bypassed) which are not detectable by
894 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
895 tracking down these problems.
898 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
899 enables the feature at boot time. By default, it is
900 disabled and the system will work mostly the same as a
901 kernel built without CONFIG_DEBUG_PAGEALLOC.
902 Note: to get most of debug_pagealloc error reports, it's
903 useful to also enable the page_owner functionality.
904 on: enable the feature
906 debugfs= [KNL] This parameter enables what is exposed to userspace
907 and debugfs internal clients.
908 Format: { on, no-mount, off }
909 on: All functions are enabled.
911 Filesystem is not registered but kernel clients can
912 access APIs and a crashkernel can be used to read
913 its content. There is nothing to mount.
914 off: Filesystem is not registered and clients
915 get a -EPERM as result when trying to register files
916 or directories within debugfs.
917 This is equivalent of the runtime functionality if
918 debugfs was not enabled in the kernel at all.
919 Default value is set in build-time with a kernel configuration.
921 debugpat [X86] Enable PAT debugging
923 decnet.addr= [HW,NET]
924 Format: <area>[,<node>]
925 See also Documentation/networking/decnet.rst.
928 [HW] The size of the default HugeTLB page. This is
929 the size represented by the legacy /proc/ hugepages
930 APIs. In addition, this is the default hugetlb size
931 used for shmget(), mmap() and mounting hugetlbfs
932 filesystems. If not specified, defaults to the
933 architecture's default huge page size. Huge page
934 sizes are architecture dependent. See also
935 Documentation/admin-guide/mm/hugetlbpage.rst.
938 deferred_probe_timeout=
939 [KNL] Debugging option to set a timeout in seconds for
940 deferred probe to give up waiting on dependencies to
941 probe. Only specific dependencies (subsystems or
942 drivers) that have opted in will be ignored. A timeout of 0
943 will timeout at the end of initcalls. This option will also
944 dump out devices still on the deferred probe list after
948 Format: { on | off | def_only | inf_only | always }
949 on: s390 zlib hardware support for compression on
950 level 1 and decompression (default)
951 off: No s390 zlib hardware support
952 def_only: s390 zlib hardware support for deflate
953 only (compression on level 1)
954 inf_only: s390 zlib hardware support for inflate
956 always: Same as 'on' but ignores the selected compression
957 level always using hardware support (used for debugging)
960 Set number of hash buckets for dentry cache.
962 disable_1tb_segments [PPC]
963 Disables the use of 1TB hash page table segments. This
964 causes the kernel to fall back to 256MB segments which
965 can be useful when debugging issues that require an SLB
969 Limits the number of kernel SLB entries, and flushes
970 them frequently to increase the rate of SLB faults
974 See Documentation/networking/ipv6.rst.
977 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
978 hardening is enabled for this boot. Hardened
979 usercopy checking is used to protect the kernel
980 from reading or writing beyond known memory
981 allocation boundaries as a proactive defense
982 against bounds-checking flaws in the kernel's
983 copy_to_user()/copy_from_user() interface.
984 on Perform hardened usercopy checks (default).
985 off Disable hardened usercopy checks.
988 Disable RADIX MMU mode on POWER9
990 radix_hcall_invalidate=on [PPC/PSERIES]
991 Disable RADIX GTSE feature and use hcall for TLB
995 Disable TLBIE instruction. Currently does not work
996 with KVM, with HASH MMU, or with coherent accelerators.
998 disable_cpu_apicid= [X86,APIC,SMP]
1000 The number of initial APIC ID for the
1001 corresponding CPU to be disabled at boot,
1002 mostly used for the kdump 2nd kernel to
1003 disable BSP to wake up multiple CPUs without
1004 causing system reset or hang due to sending
1005 INIT from AP to BSP.
1007 disable_ddw [PPC/PSERIES]
1008 Disable Dynamic DMA Window support. Use this
1009 to workaround buggy firmware.
1011 disable_ipv6= [IPV6]
1012 See Documentation/networking/ipv6.rst.
1014 disable_mtrr_cleanup [X86]
1015 The kernel tries to adjust MTRR layout from continuous
1016 to discrete, to make X server driver able to add WB
1017 entry later. This parameter disables that.
1019 disable_mtrr_trim [X86, Intel and AMD only]
1020 By default the kernel will trim any uncacheable
1021 memory out of your available memory pool based on
1022 MTRR settings. This parameter disables that behavior,
1023 possibly causing your machine to run very slowly.
1025 disable_timer_pin_1 [X86]
1026 Disable PIN 1 of APIC timer
1027 Can be useful to work around chipset bugs.
1029 dis_ucode_ldr [X86] Disable the microcode loader.
1031 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
1032 this option disables the debugging code at boot.
1034 dma_debug_entries=<number>
1035 This option allows to tune the number of preallocated
1036 entries for DMA-API debugging code. One entry is
1037 required per DMA-API allocation. Use this if the
1038 DMA-API debugging code disables itself because the
1039 architectural default is too low.
1041 dma_debug_driver=<driver_name>
1042 With this option the DMA-API debugging driver
1043 filter feature can be enabled at boot time. Just
1044 pass the driver to filter for as the parameter.
1045 The filter can be disabled or changed to another
1046 driver later using sysfs.
1048 driver_async_probe= [KNL]
1049 List of driver names to be probed asynchronously.
1050 Format: <driver_name1>,<driver_name2>...
1052 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
1053 Broken monitors, graphic adapters, KVMs and EDIDless
1054 panels may send no or incorrect EDID data sets.
1055 This parameter allows to specify an EDID data sets
1056 in the /lib/firmware directory that are used instead.
1057 Generic built-in EDID data sets are used, if one of
1058 edid/1024x768.bin, edid/1280x1024.bin,
1059 edid/1680x1050.bin, or edid/1920x1080.bin is given
1060 and no file with the same name exists. Details and
1061 instructions how to build your own EDID data are
1062 available in Documentation/admin-guide/edid.rst. An EDID
1063 data set will only be used for a particular connector,
1064 if its name and a colon are prepended to the EDID
1065 name. Each connector may use a unique EDID data
1066 set by separating the files with a comma. An EDID
1067 data set with no connector name will be used for
1068 any connectors not explicitly specified.
1073 Format: {"off" | "known"}
1074 Control how the dt_cpu_ftrs device-tree binding is
1075 used for CPU feature discovery and setup (if it
1077 off: Do not use it, fall back to legacy cpu table.
1078 known: Do not pass through unknown features to guests
1079 or userspace, only those that the kernel is aware of.
1081 dump_apple_properties [X86]
1082 Dump name and content of EFI device properties on
1083 x86 Macs. Useful for driver authors to determine
1084 what data is available or for reverse-engineering.
1086 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
1087 <module>.dyndbg[="val"]
1088 Enable debug messages at boot time. See
1089 Documentation/admin-guide/dynamic-debug-howto.rst
1092 nopku [X86] Disable Memory Protection Keys CPU feature found
1095 <module>.async_probe [KNL]
1096 Enable asynchronous probe on this module.
1098 early_ioremap_debug [KNL]
1099 Enable debug messages in early_ioremap support. This
1100 is useful for tracking down temporary early mappings
1101 which are not unmapped.
1103 earlycon= [KNL] Output early console device and options.
1105 When used with no options, the early console is
1106 determined by stdout-path property in device tree's
1107 chosen node or the ACPI SPCR table if supported by
1110 cdns,<addr>[,options]
1111 Start an early, polled-mode console on a Cadence
1112 (xuartps) serial port at the specified address. Only
1113 supported option is baud rate. If baud rate is not
1114 specified, the serial port must already be setup and
1117 uart[8250],io,<addr>[,options]
1118 uart[8250],mmio,<addr>[,options]
1119 uart[8250],mmio32,<addr>[,options]
1120 uart[8250],mmio32be,<addr>[,options]
1121 uart[8250],0x<addr>[,options]
1122 Start an early, polled-mode console on the 8250/16550
1123 UART at the specified I/O port or MMIO address.
1124 MMIO inter-register address stride is either 8-bit
1125 (mmio) or 32-bit (mmio32 or mmio32be).
1126 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1127 to be equivalent to 'mmio'. 'options' are specified
1128 in the same format described for "console=ttyS<n>"; if
1129 unspecified, the h/w is not initialized.
1133 Start an early, polled-mode console on a pl011 serial
1134 port at the specified address. The pl011 serial port
1135 must already be setup and configured. Options are not
1136 yet supported. If 'mmio32' is specified, then only
1137 the driver will use only 32-bit accessors to read/write
1138 the device registers.
1141 Start an early console on a litex serial port at the
1142 specified address. The serial port must already be
1143 setup and configured. Options are not yet supported.
1146 Start an early, polled-mode console on a meson serial
1147 port at the specified address. The serial port must
1148 already be setup and configured. Options are not yet
1152 Start an early, polled-mode console on an msm serial
1153 port at the specified address. The serial port
1154 must already be setup and configured. Options are not
1157 msm_serial_dm,<addr>
1158 Start an early, polled-mode console on an msm serial
1159 dm port at the specified address. The serial port
1160 must already be setup and configured. Options are not
1164 Start an early, polled-mode console on a serial port
1165 of an Actions Semi SoC, such as S500 or S900, at the
1166 specified address. The serial port must already be
1167 setup and configured. Options are not yet supported.
1170 Start an early, polled-mode console on a serial port
1171 of an RDA Micro SoC, such as RDA8810PL, at the
1172 specified address. The serial port must already be
1173 setup and configured. Options are not yet supported.
1176 Use RISC-V SBI (Supervisor Binary Interface) for early
1179 smh Use ARM semihosting calls for early console.
1187 Use early console provided by serial driver available
1188 on Samsung SoCs, requires selecting proper type and
1189 a correct base address of the selected UART port. The
1190 serial port must already be setup and configured.
1191 Options are not yet supported.
1194 Start an early, polled-mode console on a lantiq serial
1195 (lqasc) port at the specified address. The serial port
1196 must already be setup and configured. Options are not
1201 Use early console provided by Freescale LP UART driver
1202 found on Freescale Vybrid and QorIQ LS1021A processors.
1203 A valid base address must be provided, and the serial
1204 port must already be setup and configured.
1208 Start an early, polled-mode, output-only console on the
1209 Freescale i.MX UART at the specified address. The UART
1210 must already be setup and configured.
1213 Start an early, polled-mode console on the
1214 Armada 3700 serial port at the specified
1215 address. The serial port must already be setup
1216 and configured. Options are not yet supported.
1219 Start an early, polled-mode console on a Qualcomm
1220 Generic Interface (GENI) based serial port at the
1221 specified address. The serial port must already be
1222 setup and configured. Options are not yet supported.
1225 Start an early, unaccelerated console on the EFI
1226 memory mapped framebuffer (if available). On cache
1227 coherent non-x86 systems that use system memory for
1228 the framebuffer, pass the 'ram' option so that it is
1229 mapped with the correct attributes.
1232 Use early console provided by Freescale LINFlexD UART
1233 serial driver for NXP S32V234 SoCs. A valid base
1234 address must be provided, and the serial port must
1235 already be setup and configured.
1237 earlyprintk= [X86,SH,ARM,M68k,S390]
1241 earlyprintk=serial[,ttySn[,baudrate]]
1242 earlyprintk=serial[,0x...[,baudrate]]
1243 earlyprintk=ttySn[,baudrate]
1244 earlyprintk=dbgp[debugController#]
1245 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1246 earlyprintk=xdbc[xhciController#]
1248 earlyprintk is useful when the kernel crashes before
1249 the normal console is initialized. It is not enabled by
1250 default because it has some cosmetic problems.
1252 Append ",keep" to not disable it when the real console
1255 Only one of vga, efi, serial, or usb debug port can
1258 Currently only ttyS0 and ttyS1 may be specified by
1259 name. Other I/O ports may be explicitly specified
1260 on some architectures (x86 and arm at least) by
1261 replacing ttySn with an I/O port address, like this:
1262 earlyprintk=serial,0x1008,115200
1263 You can find the port for a given device in
1264 /proc/tty/driver/serial:
1265 2: uart:ST16650V2 port:00001008 irq:18 ...
1267 Interaction with the standard serial driver is not
1270 The VGA and EFI output is eventually overwritten by
1273 The xen option can only be used in Xen domains.
1275 The sclp output can only be used on s390.
1277 The optional "force" to "pciserial" enables use of a
1278 PCI device even when its classcode is not of the
1281 edac_report= [HW,EDAC] Control how to report EDAC event
1282 Format: {"on" | "off" | "force"}
1283 on: enable EDAC to report H/W event. May be overridden
1284 by other higher priority error reporting module.
1285 off: disable H/W event reporting through EDAC.
1286 force: enforce the use of EDAC to report H/W event.
1289 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1292 This is designed to be used in conjunction with
1293 the boot argument: earlyprintk=vga
1295 This parameter works in place of the kgdboc parameter
1296 but can only be used if the backing tty is available
1297 very early in the boot process. For early debugging
1298 via a serial port see kgdboc_earlycon instead.
1301 Format: {"off" | "on" | "skip[mbr]"}
1304 Format: { "debug", "disable_early_pci_dma",
1305 "nochunk", "noruntime", "nosoftreserve",
1306 "novamap", "no_disable_early_pci_dma" }
1307 debug: enable misc debug output.
1308 disable_early_pci_dma: disable the busmaster bit on all
1309 PCI bridges while in the EFI boot stub.
1310 nochunk: disable reading files in "chunks" in the EFI
1311 boot stub, as chunking can cause problems with some
1312 firmware implementations.
1313 noruntime : disable EFI runtime services support
1314 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1315 attribute may cause the kernel to reserve the
1316 memory range for a memory mapping driver to
1317 claim. Specify efi=nosoftreserve to disable this
1318 reservation and treat the memory by its base type
1319 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1320 novamap: do not call SetVirtualAddressMap().
1321 no_disable_early_pci_dma: Leave the busmaster bit set
1322 on all PCI bridges while in the EFI boot stub
1324 efi_no_storage_paranoia [EFI; X86]
1325 Using this parameter you can use more than 50% of
1326 your efi variable storage. Use this parameter only if
1327 you are really sure that your UEFI does sane gc and
1328 fulfills the spec otherwise your board may brick.
1330 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1331 Add arbitrary attribute to specific memory range by
1332 updating original EFI memory map.
1333 Region of memory which aa attribute is added to is
1336 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1337 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1338 attribute is added to range 0x100000000-0x180000000 and
1339 0x10a0000000-0x1120000000.
1341 If efi_fake_mem=8G@9G:0x40000 is specified, the
1342 EFI_MEMORY_SP(0x40000) attribute is added to
1343 range 0x240000000-0x43fffffff.
1345 Using this parameter you can do debugging of EFI memmap
1346 related features. For example, you can do debugging of
1347 Address Range Mirroring feature even if your box
1348 doesn't support it, or mark specific memory as
1351 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1352 that is to be dynamically loaded by Linux. If there are
1353 multiple variables with the same name but with different
1354 vendor GUIDs, all of them will be loaded. See
1355 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1358 eisa_irq_edge= [PARISC,HW]
1359 See header of drivers/parisc/eisa.c.
1362 See comment before function elanfreq_setup() in
1363 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1365 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1366 Specifies physical address of start of kernel core
1367 image elf header and optionally the size. Generally
1368 kexec loader will pass this option to capture kernel.
1369 See Documentation/admin-guide/kdump/kdump.rst for details.
1371 enable_mtrr_cleanup [X86]
1372 The kernel tries to adjust MTRR layout from continuous
1373 to discrete, to make X server driver able to add WB
1374 entry later. This parameter enables that.
1376 enable_timer_pin_1 [X86]
1377 Enable PIN 1 of APIC timer
1378 Can be useful to work around chipset bugs
1379 (in particular on some ATI chipsets).
1380 The kernel tries to set a reasonable default.
1382 enforcing [SELINUX] Set initial enforcing status.
1384 See security/selinux/Kconfig help text.
1385 0 -- permissive (log only, no denials).
1386 1 -- enforcing (deny and log).
1388 Value can be changed at runtime via
1389 /sys/fs/selinux/enforce.
1392 Disable Error Record Serialization Table (ERST)
1395 ether= [HW,NET] Ethernet cards parameters
1396 This option is obsoleted by the "netdev=" option, which
1397 has equivalent usage. See its documentation for details.
1401 Permit 'security.evm' to be updated regardless of
1402 current integrity status.
1407 fail_make_request=[KNL]
1408 General fault injection mechanism.
1409 Format: <interval>,<probability>,<space>,<times>
1410 See also Documentation/fault-injection/.
1413 Format: { initns | none }
1414 See Documentation/admin-guide/sysctl/net.rst for
1415 fb_tunnels_only_for_init_ns
1418 See Documentation/admin-guide/blockdev/floppy.rst.
1420 force_pal_cache_flush
1421 [IA-64] Avoid check_sal_cache_flush which may hang on
1422 buggy SAL_CACHE_FLUSH implementations. Using this
1423 parameter will force ia64_sal_cache_flush to call
1424 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1427 Forcefully enable Physical Address Extension (PAE).
1428 Many Pentium M systems disable PAE but may have a
1429 functionally usable PAE implementation.
1430 Warning: use of this parameter will taint the kernel
1431 and may cause unknown problems.
1434 [FTRACE] will set and start the specified tracer
1435 as early as possible in order to facilitate early
1438 ftrace_dump_on_oops[=orig_cpu]
1439 [FTRACE] will dump the trace buffers on oops.
1440 If no parameter is passed, ftrace will dump
1441 buffers of all CPUs, but if you pass orig_cpu, it will
1442 dump only the buffer of the CPU that triggered the
1445 ftrace_filter=[function-list]
1446 [FTRACE] Limit the functions traced by the function
1447 tracer at boot up. function-list is a comma-separated
1448 list of functions. This list can be changed at run
1449 time by the set_ftrace_filter file in the debugfs
1452 ftrace_notrace=[function-list]
1453 [FTRACE] Do not trace the functions specified in
1454 function-list. This list can be changed at run time
1455 by the set_ftrace_notrace file in the debugfs
1458 ftrace_graph_filter=[function-list]
1459 [FTRACE] Limit the top level callers functions traced
1460 by the function graph tracer at boot up.
1461 function-list is a comma-separated list of functions
1462 that can be changed at run time by the
1463 set_graph_function file in the debugfs tracing directory.
1465 ftrace_graph_notrace=[function-list]
1466 [FTRACE] Do not trace from the functions specified in
1467 function-list. This list is a comma-separated list of
1468 functions that can be changed at run time by the
1469 set_graph_notrace file in the debugfs tracing directory.
1471 ftrace_graph_max_depth=<uint>
1472 [FTRACE] Used with the function graph tracer. This is
1473 the max depth it will trace into a function. This value
1474 can be changed at run time by the max_graph_depth file
1475 in the tracefs tracing directory. default: 0 (no limit)
1477 fw_devlink= [KNL] Create device links between consumer and supplier
1478 devices by scanning the firmware to infer the
1479 consumer/supplier relationships. This feature is
1480 especially useful when drivers are loaded as modules as
1481 it ensures proper ordering of tasks like device probing
1482 (suppliers first, then consumers), supplier boot state
1483 clean up (only after all consumers have probed),
1484 suspend/resume & runtime PM (consumers first, then
1486 Format: { off | permissive | on | rpm }
1487 off -- Don't create device links from firmware info.
1488 permissive -- Create device links from firmware info
1489 but use it only for ordering boot state clean
1490 up (sync_state() calls).
1491 on -- Create device links from firmware info and use it
1492 to enforce probe and suspend/resume ordering.
1493 rpm -- Like "on", but also use to order runtime PM.
1495 fw_devlink.strict=<bool>
1496 [KNL] Treat all inferred dependencies as mandatory
1497 dependencies. This only applies for fw_devlink=on|rpm.
1501 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1502 support via parallel port (up to 5 devices per port)
1503 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1504 See also Documentation/input/devices/joystick-parport.rst
1508 gart_fix_e820= [X86-64] disable the fix e820 for K8 GART
1512 gcov_persist= [GCOV] When non-zero (default), profiling data for
1513 kernel modules is saved and remains accessible via
1514 debugfs, even when the module is unloaded/reloaded.
1515 When zero, profiling data is discarded and associated
1516 debugfs files are removed at module unload time.
1518 goldfish [X86] Enable the goldfish android emulator platform.
1519 Don't use this when you are not running on the
1522 gpio-mockup.gpio_mockup_ranges
1523 [HW] Sets the ranges of gpiochip of for this device.
1524 Format: <start1>,<end1>,<start2>,<end2>...
1525 gpio-mockup.gpio_mockup_named_lines
1526 [HW] Let the driver know GPIO lines should be named.
1528 gpt [EFI] Forces disk with valid GPT signature but
1529 invalid Protective MBR to be treated as GPT. If the
1530 primary GPT is corrupted, it enables the backup/alternate
1531 GPT to be used instead.
1533 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1534 the "Enable 0" bit of the configuration register.
1537 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1538 the "Enable 0" bit of the configuration register.
1541 grcan.select= [HW] Select which physical interface to use.
1544 grcan.txsize= [HW] Sets the size of the tx buffer.
1545 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1547 grcan.rxsize= [HW] Sets the size of the rx buffer.
1548 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1551 hardlockup_all_cpu_backtrace=
1552 [KNL] Should the hard-lockup detector generate
1553 backtraces on all cpus.
1556 hashdist= [KNL,NUMA] Large hashes allocated during boot
1557 are distributed across NUMA nodes. Defaults on
1558 for 64-bit NUMA, off otherwise.
1559 Format: 0 | 1 (for off | on)
1561 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1563 hd= [EIDE] (E)IDE hard drive subsystem geometry
1564 Format: <cyl>,<head>,<sect>
1567 Disable Hardware Error Source Table (HEST) support;
1568 corresponding firmware-first mode error processing
1569 logic will be disabled.
1571 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1572 size of <nn>. This works even on boxes that have no
1573 highmem otherwise. This also works to reduce highmem
1574 size on bigger boxes.
1576 highres= [KNL] Enable/disable high resolution timer mode.
1577 Valid parameters: "on", "off"
1582 hpet= [X86-32,HPET] option to control HPET usage
1583 Format: { enable (default) | disable | force |
1585 disable: disable HPET and use PIT instead
1586 force: allow force enabled of undocumented chips (ICH4,
1588 verbose: show contents of HPET registers during setup
1590 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1591 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1593 hugetlb_cma= [HW,CMA] The size of a CMA area used for allocation
1594 of gigantic hugepages. Or using node format, the size
1595 of a CMA area per node can be specified.
1596 Format: nn[KMGTPE] or (node format)
1597 <node>:nn[KMGTPE][,<node>:nn[KMGTPE]]
1599 Reserve a CMA area of given size and allocate gigantic
1600 hugepages using the CMA allocator. If enabled, the
1601 boot-time allocation of gigantic hugepages is skipped.
1603 hugepages= [HW] Number of HugeTLB pages to allocate at boot.
1604 If this follows hugepagesz (below), it specifies
1605 the number of pages of hugepagesz to be allocated.
1606 If this is the first HugeTLB parameter on the command
1607 line, it specifies the number of pages to allocate for
1608 the default huge page size. If using node format, the
1609 number of pages to allocate per-node can be specified.
1610 See also Documentation/admin-guide/mm/hugetlbpage.rst.
1611 Format: <integer> or (node format)
1612 <node>:<integer>[,<node>:<integer>]
1615 [HW] The size of the HugeTLB pages. This is used in
1616 conjunction with hugepages (above) to allocate huge
1617 pages of a specific size at boot. The pair
1618 hugepagesz=X hugepages=Y can be specified once for
1619 each supported huge page size. Huge page sizes are
1620 architecture dependent. See also
1621 Documentation/admin-guide/mm/hugetlbpage.rst.
1624 hugetlb_free_vmemmap=
1625 [KNL] Reguires CONFIG_HUGETLB_PAGE_FREE_VMEMMAP
1627 Allows heavy hugetlb users to free up some more
1628 memory (7 * PAGE_SIZE for each 2MB hugetlb page).
1629 Format: { on | off (default) }
1631 on: enable the feature
1632 off: disable the feature
1634 Built with CONFIG_HUGETLB_PAGE_FREE_VMEMMAP_DEFAULT_ON=y,
1637 This is not compatible with memory_hotplug.memmap_on_memory.
1638 If both parameters are enabled, hugetlb_free_vmemmap takes
1639 precedence over memory_hotplug.memmap_on_memory.
1642 [KNL] Should the hung task detector generate panics.
1645 A value of 1 instructs the kernel to panic when a
1646 hung task is detected. The default value is controlled
1647 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1648 option. The value selected by this boot parameter can
1649 be changed later by the kernel.hung_task_panic sysctl.
1651 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1652 terminal devices. Valid values: 0..8
1653 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1654 If specified, z/VM IUCV HVC accepts connections
1655 from listed z/VM user IDs only.
1657 hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations
1658 which allow the hypervisor to 'idle' the
1659 guest on lock contention.
1662 Do not unregister boot console at start. This is only
1663 useful for debugging when something happens in the window
1664 between unregistering the boot console and initializing
1667 i2c_bus= [HW] Override the default board specific I2C bus speed
1668 or register an additional I2C bus that is not
1669 registered from board initialization code.
1673 i8042.debug [HW] Toggle i8042 debug mode
1674 i8042.unmask_kbd_data
1675 [HW] Enable printing of interrupt data from the KBD port
1676 (disabled by default, and as a pre-condition
1677 requires that i8042.debug=1 be enabled)
1678 i8042.direct [HW] Put keyboard port into non-translated mode
1679 i8042.dumbkbd [HW] Pretend that controller can only read data from
1680 keyboard and cannot control its state
1681 (Don't attempt to blink the leds)
1682 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1683 i8042.nokbd [HW] Don't check/create keyboard port
1684 i8042.noloop [HW] Disable the AUX Loopback command while probing
1686 i8042.nomux [HW] Don't check presence of an active multiplexing
1688 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1690 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1691 i8042.reset [HW] Reset the controller during init, cleanup and
1692 suspend-to-ram transitions, only during s2r
1693 transitions, or never reset
1694 Format: { 1 | Y | y | 0 | N | n }
1695 1, Y, y: always reset controller
1696 0, N, n: don't ever reset controller
1697 Default: only on s2r transitions on x86; most other
1698 architectures force reset to be always executed
1699 i8042.unlock [HW] Unlock (ignore) the keylock
1700 i8042.kbdreset [HW] Reset device connected to KBD port
1702 [HW] Allow deferred probing upon i8042 probe errors
1706 i8k.ignore_dmi [HW] Continue probing hardware even if DMI data
1707 indicates that the driver is running on unsupported
1709 i8k.force [HW] Activate i8k driver even if SMM BIOS signature
1710 does not match list of supported models.
1712 [HW] Report power status in /proc/i8k
1713 (disabled by default)
1714 i8k.restricted [HW] Allow controlling fans only if SYS_ADMIN
1717 i915.invert_brightness=
1718 [DRM] Invert the sense of the variable that is used to
1719 set the brightness of the panel backlight. Normally a
1720 brightness value of 0 indicates backlight switched off,
1721 and the maximum of the brightness value sets the backlight
1722 to maximum brightness. If this parameter is set to 0
1723 (default) and the machine requires it, or this parameter
1724 is set to 1, a brightness value of 0 sets the backlight
1725 to maximum brightness, and the maximum of the brightness
1726 value switches the backlight off.
1727 -1 -- never invert brightness
1728 0 -- machine default
1729 1 -- force brightness inversion
1732 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1734 ide-core.nodma= [HW] (E)IDE subsystem
1735 Format: =0.0 to prevent dma on hda, =0.1 hdb =1.0 hdc
1736 .vlb_clock .pci_clock .noflush .nohpa .noprobe .nowerr
1737 .cdrom .chs .ignore_cable are additional options
1738 See Documentation/ide/ide.rst.
1740 ide-generic.probe-mask= [HW] (E)IDE subsystem
1742 Probe mask for legacy ISA IDE ports. Depending on
1743 platform up to 6 ports are supported, enabled by
1744 setting corresponding bits in the mask to 1. The
1745 default value is 0x0, which has a special meaning.
1746 On systems that have PCI, it triggers scanning the
1747 PCI bus for the first and the second port, which
1748 are then probed. On systems without PCI the value
1749 of 0x0 enables probing the two first ports as if it
1752 ide-pci-generic.all-generic-ide [HW] (E)IDE subsystem
1753 Claim all unknown PCI IDE storage controllers.
1756 Format: idle=poll, idle=halt, idle=nomwait
1757 Poll forces a polling idle loop that can slightly
1758 improve the performance of waking up a idle CPU, but
1759 will use a lot of power and make the system run hot.
1761 idle=halt: Halt is forced to be used for CPU idle.
1762 In such case C2/C3 won't be used again.
1763 idle=nomwait: Disable mwait for CPU C-states
1767 Allow force disabling of Shared Virtual Memory (SVA)
1768 support for the idxd driver. By default it is set to
1771 idxd.tc_override= [HW]
1773 Allow override of default traffic class configuration
1774 for the device. By default it is set to false (0).
1776 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1777 Format: { strict | legacy | 2008 | relaxed }
1780 Choose which programs will be accepted for execution
1781 based on the IEEE 754 NaN encoding(s) supported by
1782 the FPU and the NaN encoding requested with the value
1783 of an ELF file header flag individually set by each
1784 binary. Hardware implementations are permitted to
1785 support either or both of the legacy and the 2008 NaN
1788 Available settings are as follows:
1789 strict accept binaries that request a NaN encoding
1790 supported by the FPU
1791 legacy only accept legacy-NaN binaries, if supported
1793 2008 only accept 2008-NaN binaries, if supported
1795 relaxed accept any binaries regardless of whether
1796 supported by the FPU
1798 The FPU emulator is always able to support both NaN
1799 encodings, so if no FPU hardware is present or it has
1800 been disabled with 'nofpu', then the settings of
1801 'legacy' and '2008' strap the emulator accordingly,
1802 'relaxed' straps the emulator for both legacy-NaN and
1803 2008-NaN, whereas 'strict' enables legacy-NaN only on
1804 legacy processors and both NaN encodings on MIPS32 or
1807 The setting for ABS.fmt/NEG.fmt instruction execution
1808 mode generally follows that for the NaN encoding,
1809 except where unsupported by hardware.
1811 ignore_loglevel [KNL]
1812 Ignore loglevel setting - this will print /all/
1813 kernel messages to the console. Useful for debugging.
1814 We also add it as printk module parameter, so users
1815 could change it dynamically, usually by
1816 /sys/module/printk/parameters/ignore_loglevel.
1819 Ignore RLIMIT_DATA setting for data mappings,
1820 print warning at first misuse. Can be changed via
1821 /sys/module/kernel/parameters/ignore_rlimit_data.
1823 ihash_entries= [KNL]
1824 Set number of hash buckets for inode cache.
1826 ima_appraise= [IMA] appraise integrity measurements
1827 Format: { "off" | "enforce" | "fix" | "log" }
1830 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
1831 The builtin appraise policy appraises all files
1834 ima_canonical_fmt [IMA]
1835 Use the canonical format for the binary runtime
1836 measurements, instead of host native format.
1839 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1843 The list of supported hash algorithms is defined
1844 in crypto/hash_info.h.
1847 The builtin policies to load during IMA setup.
1848 Format: "tcb | appraise_tcb | secure_boot |
1849 fail_securely | critical_data"
1851 The "tcb" policy measures all programs exec'd, files
1852 mmap'd for exec, and all files opened with the read
1853 mode bit set by either the effective uid (euid=0) or
1856 The "appraise_tcb" policy appraises the integrity of
1857 all files owned by root.
1859 The "secure_boot" policy appraises the integrity
1860 of files (eg. kexec kernel image, kernel modules,
1861 firmware, policy, etc) based on file signatures.
1863 The "fail_securely" policy forces file signature
1864 verification failure also on privileged mounted
1865 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
1868 The "critical_data" policy measures kernel integrity
1871 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
1872 Load a policy which meets the needs of the Trusted
1873 Computing Base. This means IMA will measure all
1874 programs exec'd, files mmap'd for exec, and all files
1875 opened for read by uid=0.
1878 Select one of defined IMA measurements template formats.
1879 Formats: { "ima" | "ima-ng" | "ima-sig" }
1883 [IMA] Define a custom template format.
1884 Format: { "field1|...|fieldN" }
1886 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
1887 Format: <min_file_size>
1888 Set the minimal file size for using asynchronous hash.
1889 If left unspecified, ahash usage is disabled.
1891 ahash performance varies for different data sizes on
1892 different crypto accelerators. This option can be used
1893 to achieve the best performance for a particular HW.
1895 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
1897 Set hashing buffer size. Default: 4k.
1899 ahash performance varies for different chunk sizes on
1900 different crypto accelerators. This option can be used
1901 to achieve best performance for particular HW.
1905 Run specified binary instead of /sbin/init as init
1908 initcall_debug [KNL] Trace initcalls as they are executed. Useful
1909 for working out where the kernel is dying during
1912 initcall_blacklist= [KNL] Do not execute a comma-separated list of
1913 initcall functions. Useful for debugging built-in
1914 modules and initcalls.
1916 initramfs_async= [KNL]
1919 This parameter controls whether the initramfs
1920 image is unpacked asynchronously, concurrently
1921 with devices being probed and
1922 initialized. This should normally just work,
1923 but as a debugging aid, one can get the
1924 historical behaviour of the initramfs
1925 unpacking being completed before device_ and
1928 initrd= [BOOT] Specify the location of the initial ramdisk
1930 initrdmem= [KNL] Specify a physical address and size from which to
1931 load the initrd. If an initrd is compiled in or
1932 specified in the bootparams, it takes priority over this
1934 Format: ss[KMG],nn[KMG]
1937 init_on_alloc= [MM] Fill newly allocated pages and heap objects with
1940 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
1942 init_on_free= [MM] Fill freed pages and heap objects with zeroes.
1944 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
1946 init_pkru= [X86] Specify the default memory protection keys rights
1947 register contents for all processes. 0x55555554 by
1948 default (disallow access to all but pkey 0). Can
1949 override in debugfs after boot.
1951 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
1954 int_pln_enable [X86] Enable power limit notification interrupt
1956 integrity_audit=[IMA]
1957 Format: { "0" | "1" }
1958 0 -- basic integrity auditing messages. (Default)
1959 1 -- additional integrity auditing messages.
1961 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
1963 Enable intel iommu driver.
1965 Disable intel iommu driver.
1966 igfx_off [Default Off]
1967 By default, gfx is mapped as normal device. If a gfx
1968 device has a dedicated DMAR unit, the DMAR unit is
1969 bypassed by not enabling DMAR with this option. In
1970 this case, gfx device will use physical address for
1972 strict [Default Off]
1973 Deprecated, equivalent to iommu.strict=1.
1974 sp_off [Default Off]
1975 By default, super page will be supported if Intel IOMMU
1976 has the capability. With this option, super page will
1979 Enable the Intel IOMMU scalable mode if the hardware
1980 advertises that it has support for the scalable mode
1983 Disallow use of the Intel IOMMU scalable mode.
1984 tboot_noforce [Default Off]
1985 Do not force the Intel IOMMU enabled under tboot.
1986 By default, tboot will force Intel IOMMU on, which
1987 could harm performance of some high-throughput
1988 devices like 40GBit network cards, even if identity
1990 Note that using this option lowers the security
1991 provided by tboot because it makes the system
1992 vulnerable to DMA attacks.
1994 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
1995 0 disables intel_idle and fall back on acpi_idle.
1996 1 to 9 specify maximum depth of C-state.
2000 Do not enable intel_pstate as the default
2001 scaling driver for the supported processors
2003 Use intel_pstate as a scaling driver, but configure it
2004 to work with generic cpufreq governors (instead of
2005 enabling its internal governor). This mode cannot be
2006 used along with the hardware-managed P-states (HWP)
2009 Enable intel_pstate on systems that prohibit it by default
2010 in favor of acpi-cpufreq. Forcing the intel_pstate driver
2011 instead of acpi-cpufreq may disable platform features, such
2012 as thermal controls and power capping, that rely on ACPI
2013 P-States information being indicated to OSPM and therefore
2014 should be used with caution. This option does not work with
2015 processors that aren't supported by the intel_pstate driver
2016 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
2018 Do not enable hardware P state control (HWP)
2021 Only load intel_pstate on systems which support
2022 hardware P state control (HWP) if available.
2024 Enforce ACPI _PPC performance limits. If the Fixed ACPI
2025 Description Table, specifies preferred power management
2026 profile as "Enterprise Server" or "Performance Server",
2027 then this feature is turned on by default.
2029 Allow per-logical-CPU P-State performance control limits using
2030 cpufreq sysfs interface
2032 intremap= [X86-64, Intel-IOMMU]
2033 on enable Interrupt Remapping (default)
2034 off disable Interrupt Remapping
2035 nosid disable Source ID checking
2037 BIOS x2APIC opt-out request will be ignored
2038 nopost disable Interrupt Posting
2040 iomem= Disable strict checking of access to MMIO memory
2041 strict regions from userspace.
2056 nobypass [PPC/POWERNV]
2057 Disable IOMMU bypass, using IOMMU for PCI devices.
2059 iommu.forcedac= [ARM64, X86] Control IOVA allocation for PCI devices.
2060 Format: { "0" | "1" }
2061 0 - Try to allocate a 32-bit DMA address first, before
2062 falling back to the full range if needed.
2063 1 - Allocate directly from the full usable range,
2064 forcing Dual Address Cycle for PCI cards supporting
2065 greater than 32-bit addressing.
2067 iommu.strict= [ARM64, X86] Configure TLB invalidation behaviour
2068 Format: { "0" | "1" }
2070 Request that DMA unmap operations use deferred
2071 invalidation of hardware TLBs, for increased
2072 throughput at the cost of reduced device isolation.
2073 Will fall back to strict mode if not supported by
2074 the relevant IOMMU driver.
2076 DMA unmap operations invalidate IOMMU hardware TLBs
2078 unset - Use value of CONFIG_IOMMU_DEFAULT_DMA_{LAZY,STRICT}.
2079 Note: on x86, strict mode specified via one of the
2080 legacy driver-specific options takes precedence.
2083 [ARM64, X86] Configure DMA to bypass the IOMMU by default.
2084 Format: { "0" | "1" }
2085 0 - Use IOMMU translation for DMA.
2086 1 - Bypass the IOMMU for DMA.
2087 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
2089 io7= [HW] IO7 for Marvel-based Alpha systems
2090 See comment before marvel_specify_io7 in
2091 arch/alpha/kernel/core_marvel.c.
2093 io_delay= [X86] I/O delay method
2095 Standard port 0x80 based delay
2097 Alternate port 0xed based delay (needed on some systems)
2099 Simple two microseconds delay
2104 See Documentation/admin-guide/nfs/nfsroot.rst.
2106 ipcmni_extend [KNL] Extend the maximum number of unique System V
2107 IPC identifiers from 32,768 to 16,777,216.
2109 irqaffinity= [SMP] Set the default irq affinity mask
2110 The argument is a cpu list, as described above.
2112 irqchip.gicv2_force_probe=
2115 Force the kernel to look for the second 4kB page
2116 of a GICv2 controller even if the memory range
2117 exposed by the device tree is too small.
2119 irqchip.gicv3_nolpi=
2121 Force the kernel to ignore the availability of
2122 LPIs (and by consequence ITSs). Intended for system
2123 that use the kernel as a bootloader, and thus want
2124 to let secondary kernels in charge of setting up
2127 irqchip.gicv3_pseudo_nmi= [ARM64]
2128 Enables support for pseudo-NMIs in the kernel. This
2129 requires the kernel to be built with
2130 CONFIG_ARM64_PSEUDO_NMI.
2133 When an interrupt is not handled search all handlers
2134 for it. Intended to get systems with badly broken
2138 When an interrupt is not handled search all handlers
2139 for it. Also check all handlers each timer
2140 interrupt. Intended to get systems with badly broken
2144 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
2146 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
2147 [Deprecated - use cpusets instead]
2148 Format: [flag-list,]<cpu-list>
2150 Specify one or more CPUs to isolate from disturbances
2151 specified in the flag list (default: domain):
2154 Disable the tick when a single task runs.
2156 A residual 1Hz tick is offloaded to workqueues, which you
2157 need to affine to housekeeping through the global
2158 workqueue's affinity configured via the
2159 /sys/devices/virtual/workqueue/cpumask sysfs file, or
2160 by using the 'domain' flag described below.
2162 NOTE: by default the global workqueue runs on all CPUs,
2163 so to protect individual CPUs the 'cpumask' file has to
2164 be configured manually after bootup.
2167 Isolate from the general SMP balancing and scheduling
2168 algorithms. Note that performing domain isolation this way
2169 is irreversible: it's not possible to bring back a CPU to
2170 the domains once isolated through isolcpus. It's strongly
2171 advised to use cpusets instead to disable scheduler load
2172 balancing through the "cpuset.sched_load_balance" file.
2173 It offers a much more flexible interface where CPUs can
2174 move in and out of an isolated set anytime.
2176 You can move a process onto or off an "isolated" CPU via
2177 the CPU affinity syscalls or cpuset.
2178 <cpu number> begins at 0 and the maximum value is
2179 "number of CPUs in system - 1".
2183 Isolate from being targeted by managed interrupts
2184 which have an interrupt mask containing isolated
2185 CPUs. The affinity of managed interrupts is
2186 handled by the kernel and cannot be changed via
2187 the /proc/irq/* interfaces.
2189 This isolation is best effort and only effective
2190 if the automatically assigned interrupt mask of a
2191 device queue contains isolated and housekeeping
2192 CPUs. If housekeeping CPUs are online then such
2193 interrupts are directed to the housekeeping CPU
2194 so that IO submitted on the housekeeping CPU
2195 cannot disturb the isolated CPU.
2197 If a queue's affinity mask contains only isolated
2198 CPUs then this parameter has no effect on the
2199 interrupt routing decision, though interrupts are
2200 only delivered when tasks running on those
2201 isolated CPUs submit IO. IO submitted on
2202 housekeeping CPUs has no influence on those
2205 The format of <cpu-list> is described above.
2209 ivrs_ioapic [HW,X86-64]
2210 Provide an override to the IOAPIC-ID<->DEVICE-ID
2211 mapping provided in the IVRS ACPI table. For
2212 example, to map IOAPIC-ID decimal 10 to
2213 PCI device 00:14.0 write the parameter as:
2214 ivrs_ioapic[10]=00:14.0
2216 ivrs_hpet [HW,X86-64]
2217 Provide an override to the HPET-ID<->DEVICE-ID
2218 mapping provided in the IVRS ACPI table. For
2219 example, to map HPET-ID decimal 0 to
2220 PCI device 00:14.0 write the parameter as:
2221 ivrs_hpet[0]=00:14.0
2223 ivrs_acpihid [HW,X86-64]
2224 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2225 mapping provided in the IVRS ACPI table. For
2226 example, to map UART-HID:UID AMD0020:0 to
2227 PCI device 00:14.5 write the parameter as:
2228 ivrs_acpihid[00:14.5]=AMD0020:0
2230 js= [HW,JOY] Analog joystick
2231 See Documentation/input/joydev/joystick.rst.
2234 When CONFIG_RANDOMIZE_BASE is set, this disables
2235 kernel and module base offset ASLR (Address Space
2236 Layout Randomization).
2239 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2240 report on every invalid memory access. Without this
2241 parameter KASAN will print report only for the first
2246 kernelcore= [KNL,X86,IA-64,PPC]
2247 Format: nn[KMGTPE] | nn% | "mirror"
2248 This parameter specifies the amount of memory usable by
2249 the kernel for non-movable allocations. The requested
2250 amount is spread evenly throughout all nodes in the
2251 system as ZONE_NORMAL. The remaining memory is used for
2252 movable memory in its own zone, ZONE_MOVABLE. In the
2253 event, a node is too small to have both ZONE_NORMAL and
2254 ZONE_MOVABLE, kernelcore memory will take priority and
2255 other nodes will have a larger ZONE_MOVABLE.
2257 ZONE_MOVABLE is used for the allocation of pages that
2258 may be reclaimed or moved by the page migration
2259 subsystem. Note that allocations like PTEs-from-HighMem
2260 still use the HighMem zone if it exists, and the Normal
2261 zone if it does not.
2263 It is possible to specify the exact amount of memory in
2264 the form of "nn[KMGTPE]", a percentage of total system
2265 memory in the form of "nn%", or "mirror". If "mirror"
2266 option is specified, mirrored (reliable) memory is used
2267 for non-movable allocations and remaining memory is used
2268 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2269 are exclusive, so you cannot specify multiple forms.
2271 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
2272 Format: <Controller#>[,poll interval]
2273 The controller # is the number of the ehci usb debug
2274 port as it is probed via PCI. The poll interval is
2275 optional and is the number seconds in between
2276 each poll cycle to the debug port in case you need
2277 the functionality for interrupting the kernel with
2278 gdb or control-c on the dbgp connection. When
2279 not using this parameter you use sysrq-g to break into
2280 the kernel debugger.
2282 kgdboc= [KGDB,HW] kgdb over consoles.
2283 Requires a tty driver that supports console polling,
2284 or a supported polling keyboard driver (non-usb).
2285 Serial only format: <serial_device>[,baud]
2286 keyboard only format: kbd
2287 keyboard and serial format: kbd,<serial_device>[,baud]
2288 Optional Kernel mode setting:
2289 kms, kbd format: kms,kbd
2290 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2292 kgdboc_earlycon= [KGDB,HW]
2293 If the boot console provides the ability to read
2294 characters and can work in polling mode, you can use
2295 this parameter to tell kgdb to use it as a backend
2296 until the normal console is registered. Intended to
2297 be used together with the kgdboc parameter which
2298 specifies the normal console to transition to.
2300 The name of the early console should be specified
2301 as the value of this parameter. Note that the name of
2302 the early console might be different than the tty
2303 name passed to kgdboc. It's OK to leave the value
2304 blank and the first boot console that implements
2305 read() will be picked.
2307 kgdbwait [KGDB] Stop kernel execution and enter the
2308 kernel debugger at the earliest opportunity.
2310 kmac= [MIPS] Korina ethernet MAC address.
2311 Configure the RouterBoard 532 series on-chip
2312 Ethernet adapter MAC address.
2314 kmemleak= [KNL] Boot-time kmemleak enable/disable
2315 Valid arguments: on, off
2317 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2320 kprobe_event=[probe-list]
2321 [FTRACE] Add kprobe events and enable at boot time.
2322 The probe-list is a semicolon delimited list of probe
2323 definitions. Each definition is same as kprobe_events
2324 interface, but the parameters are comma delimited.
2325 For example, to add a kprobe event on vfs_read with
2326 arg1 and arg2, add to the command line;
2328 kprobe_event=p,vfs_read,$arg1,$arg2
2330 See also Documentation/trace/kprobetrace.rst "Kernel
2331 Boot Parameter" section.
2333 kpti= [ARM64] Control page table isolation of user
2334 and kernel address spaces.
2335 Default: enabled on cores which need mitigation.
2339 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2340 Default is 0 (don't ignore, but inject #GP)
2342 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2343 Default is false (don't support).
2345 kvm.mmu_audit= [KVM] This is a R/W parameter which allows audit
2350 [KVM] Controls the software workaround for the
2351 X86_BUG_ITLB_MULTIHIT bug.
2352 force : Always deploy workaround.
2353 off : Never deploy workaround.
2354 auto : Deploy workaround based on the presence of
2355 X86_BUG_ITLB_MULTIHIT.
2359 If the software workaround is enabled for the host,
2360 guests do need not to enable it for nested guests.
2362 kvm.nx_huge_pages_recovery_ratio=
2363 [KVM] Controls how many 4KiB pages are periodically zapped
2364 back to huge pages. 0 disables the recovery, otherwise if
2365 the value is N KVM will zap 1/Nth of the 4KiB pages every
2366 period (see below). The default is 60.
2368 kvm.nx_huge_pages_recovery_period_ms=
2369 [KVM] Controls the time period at which KVM zaps 4KiB pages
2370 back to huge pages. If the value is a non-zero N, KVM will
2371 zap a portion (see ratio above) of the pages every N msecs.
2372 If the value is 0 (the default), KVM will pick a period based
2373 on the ratio, such that a page is zapped after 1 hour on average.
2375 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM.
2376 Default is 1 (enabled)
2378 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU)
2380 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode.
2383 [KVM,ARM] Select one of KVM/arm64's modes of operation.
2385 none: Forcefully disable KVM.
2387 nvhe: Standard nVHE-based mode, without support for
2390 protected: nVHE-based mode with support for guests whose
2391 state is kept private from the host.
2392 Not valid if the kernel is running in EL2.
2394 Defaults to VHE/nVHE based on hardware support. Setting
2395 mode to "protected" will disable kexec and hibernation
2398 kvm-arm.vgic_v3_group0_trap=
2399 [KVM,ARM] Trap guest accesses to GICv3 group-0
2402 kvm-arm.vgic_v3_group1_trap=
2403 [KVM,ARM] Trap guest accesses to GICv3 group-1
2406 kvm-arm.vgic_v3_common_trap=
2407 [KVM,ARM] Trap guest accesses to GICv3 common
2410 kvm-arm.vgic_v4_enable=
2411 [KVM,ARM] Allow use of GICv4 for direct injection of
2414 kvm_cma_resv_ratio=n [PPC]
2415 Reserves given percentage from system memory area for
2416 contiguous memory allocation for KVM hash pagetable
2418 By default it reserves 5% of total system memory.
2422 kvm-intel.ept= [KVM,Intel] Disable extended page tables
2423 (virtualized MMU) support on capable Intel chips.
2424 Default is 1 (enabled)
2426 kvm-intel.emulate_invalid_guest_state=
2427 [KVM,Intel] Disable emulation of invalid guest state.
2428 Ignored if kvm-intel.enable_unrestricted_guest=1, as
2429 guest state is never invalid for unrestricted guests.
2430 This param doesn't apply to nested guests (L2), as KVM
2431 never emulates invalid L2 guest state.
2432 Default is 1 (enabled)
2434 kvm-intel.flexpriority=
2435 [KVM,Intel] Disable FlexPriority feature (TPR shadow).
2436 Default is 1 (enabled)
2439 [KVM,Intel] Enable VMX nesting (nVMX).
2440 Default is 0 (disabled)
2442 kvm-intel.unrestricted_guest=
2443 [KVM,Intel] Disable unrestricted guest feature
2444 (virtualized real and unpaged mode) on capable
2445 Intel chips. Default is 1 (enabled)
2447 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2450 Valid arguments: never, cond, always
2452 always: L1D cache flush on every VMENTER.
2453 cond: Flush L1D on VMENTER only when the code between
2454 VMEXIT and VMENTER can leak host memory.
2455 never: Disables the mitigation
2457 Default is cond (do L1 cache flush in specific instances)
2459 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification
2460 feature (tagged TLBs) on capable Intel chips.
2461 Default is 1 (enabled)
2463 l1d_flush= [X86,INTEL]
2464 Control mitigation for L1D based snooping vulnerability.
2466 Certain CPUs are vulnerable to an exploit against CPU
2467 internal buffers which can forward information to a
2468 disclosure gadget under certain conditions.
2470 In vulnerable processors, the speculatively
2471 forwarded data can be used in a cache side channel
2472 attack, to access data to which the attacker does
2473 not have direct access.
2475 This parameter controls the mitigation. The
2478 on - enable the interface for the mitigation
2480 l1tf= [X86] Control mitigation of the L1TF vulnerability on
2483 The kernel PTE inversion protection is unconditionally
2484 enabled and cannot be disabled.
2487 Provides all available mitigations for the
2488 L1TF vulnerability. Disables SMT and
2489 enables all mitigations in the
2490 hypervisors, i.e. unconditional L1D flush.
2492 SMT control and L1D flush control via the
2493 sysfs interface is still possible after
2494 boot. Hypervisors will issue a warning
2495 when the first VM is started in a
2496 potentially insecure configuration,
2497 i.e. SMT enabled or L1D flush disabled.
2500 Same as 'full', but disables SMT and L1D
2501 flush runtime control. Implies the
2502 'nosmt=force' command line option.
2503 (i.e. sysfs control of SMT is disabled.)
2506 Leaves SMT enabled and enables the default
2507 hypervisor mitigation, i.e. conditional
2510 SMT control and L1D flush control via the
2511 sysfs interface is still possible after
2512 boot. Hypervisors will issue a warning
2513 when the first VM is started in a
2514 potentially insecure configuration,
2515 i.e. SMT enabled or L1D flush disabled.
2519 Disables SMT and enables the default
2520 hypervisor mitigation.
2522 SMT control and L1D flush control via the
2523 sysfs interface is still possible after
2524 boot. Hypervisors will issue a warning
2525 when the first VM is started in a
2526 potentially insecure configuration,
2527 i.e. SMT enabled or L1D flush disabled.
2530 Same as 'flush', but hypervisors will not
2531 warn when a VM is started in a potentially
2532 insecure configuration.
2535 Disables hypervisor mitigations and doesn't
2537 It also drops the swap size and available
2538 RAM limit restriction on both hypervisor and
2543 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2549 lapic [X86-32,APIC] Enable the local APIC even if BIOS
2552 lapic= [X86,APIC] Do not use TSC deadline
2553 value for LAPIC timer one-shot implementation. Default
2554 back to the programmable timer unit in the LAPIC.
2555 Format: notscdeadline
2557 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
2560 libata.dma= [LIBATA] DMA control
2561 libata.dma=0 Disable all PATA and SATA DMA
2562 libata.dma=1 PATA and SATA Disk DMA only
2563 libata.dma=2 ATAPI (CDROM) DMA only
2564 libata.dma=4 Compact Flash DMA only
2565 Combinations also work, so libata.dma=3 enables DMA
2566 for disks and CDROMs, but not CFs.
2568 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2569 libata.ignore_hpa=0 keep BIOS limits (default)
2570 libata.ignore_hpa=1 ignore limits, using full disk
2572 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2576 libata.force= [LIBATA] Force configurations. The format is comma-
2577 separated list of "[ID:]VAL" where ID is
2578 PORT[.DEVICE]. PORT and DEVICE are decimal numbers
2579 matching port, link or device. Basically, it matches
2580 the ATA ID string printed on console by libata. If
2581 the whole ID part is omitted, the last PORT and DEVICE
2582 values are used. If ID hasn't been specified yet, the
2583 configuration applies to all ports, links and devices.
2585 If only DEVICE is omitted, the parameter applies to
2586 the port and all links and devices behind it. DEVICE
2587 number of 0 either selects the first device or the
2588 first fan-out link behind PMP device. It does not
2589 select the host link. DEVICE number of 15 selects the
2590 host link and device attached to it.
2592 The VAL specifies the configuration to force. As long
2593 as there's no ambiguity shortcut notation is allowed.
2594 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2595 The following configurations can be forced.
2597 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2598 Any ID with matching PORT is used.
2600 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2602 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2603 udma[/][16,25,33,44,66,100,133] notation is also
2606 * [no]ncq: Turn on or off NCQ.
2608 * [no]ncqtrim: Turn off queued DSM TRIM.
2610 * nohrst, nosrst, norst: suppress hard, soft
2613 * rstonce: only attempt one reset during
2614 hot-unplug link recovery
2616 * dump_id: dump IDENTIFY data.
2618 * atapi_dmadir: Enable ATAPI DMADIR bridge support
2620 * disable: Disable this device.
2622 If there are multiple matching configurations changing
2623 the same attribute, the last one is used.
2625 memblock=debug [KNL] Enable memblock debug messages.
2627 load_ramdisk= [RAM] [Deprecated]
2629 lockd.nlm_grace_period=P [NFS] Assign grace period.
2632 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2635 lockd.nlm_timeout=T [NFS] Assign timeout value.
2638 lockd.nlm_udpport=M [NFS] Assign UDP port.
2641 lockdown= [SECURITY]
2642 { integrity | confidentiality }
2643 Enable the kernel lockdown feature. If set to
2644 integrity, kernel features that allow userland to
2645 modify the running kernel are disabled. If set to
2646 confidentiality, kernel features that allow userland
2647 to extract confidential information from the kernel
2650 locktorture.nreaders_stress= [KNL]
2651 Set the number of locking read-acquisition kthreads.
2652 Defaults to being automatically set based on the
2653 number of online CPUs.
2655 locktorture.nwriters_stress= [KNL]
2656 Set the number of locking write-acquisition kthreads.
2658 locktorture.onoff_holdoff= [KNL]
2659 Set time (s) after boot for CPU-hotplug testing.
2661 locktorture.onoff_interval= [KNL]
2662 Set time (s) between CPU-hotplug operations, or
2663 zero to disable CPU-hotplug testing.
2665 locktorture.shuffle_interval= [KNL]
2666 Set task-shuffle interval (jiffies). Shuffling
2667 tasks allows some CPUs to go into dyntick-idle
2668 mode during the locktorture test.
2670 locktorture.shutdown_secs= [KNL]
2671 Set time (s) after boot system shutdown. This
2672 is useful for hands-off automated testing.
2674 locktorture.stat_interval= [KNL]
2675 Time (s) between statistics printk()s.
2677 locktorture.stutter= [KNL]
2678 Time (s) to stutter testing, for example,
2679 specifying five seconds causes the test to run for
2680 five seconds, wait for five seconds, and so on.
2681 This tests the locking primitive's ability to
2682 transition abruptly to and from idle.
2684 locktorture.torture_type= [KNL]
2685 Specify the locking implementation to test.
2687 locktorture.verbose= [KNL]
2688 Enable additional printk() statements.
2690 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
2693 loglevel= All Kernel Messages with a loglevel smaller than the
2694 console loglevel will be printed to the console. It can
2695 also be changed with klogd or other programs. The
2696 loglevels are defined as follows:
2698 0 (KERN_EMERG) system is unusable
2699 1 (KERN_ALERT) action must be taken immediately
2700 2 (KERN_CRIT) critical conditions
2701 3 (KERN_ERR) error conditions
2702 4 (KERN_WARNING) warning conditions
2703 5 (KERN_NOTICE) normal but significant condition
2704 6 (KERN_INFO) informational
2705 7 (KERN_DEBUG) debug-level messages
2707 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
2708 in bytes. n must be a power of two and greater
2709 than the minimal size. The minimal size is defined
2710 by LOG_BUF_SHIFT kernel config parameter. There is
2711 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2712 that allows to increase the default size depending on
2713 the number of CPUs. See init/Kconfig for more details.
2715 logo.nologo [FB] Disables display of the built-in Linux logo.
2716 This may be used to provide more screen space for
2717 kernel log messages and is useful when debugging
2718 kernel boot problems.
2720 lp=0 [LP] Specify parallel ports to use, e.g,
2721 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2722 lp=reset first parallel port). 'lp=0' disables the
2723 lp=auto printer driver. 'lp=reset' (which can be
2724 specified in addition to the ports) causes
2725 attached printers to be reset. Using
2726 lp=port1,port2,... specifies the parallel ports
2727 to associate lp devices with, starting with
2728 lp0. A port specification may be 'none' to skip
2729 that lp device, or a parport name such as
2730 'parport0'. Specifying 'lp=auto' instead of a
2731 port specification list means that device IDs
2732 from each port should be examined, to see if
2733 an IEEE 1284-compliant printer is attached; if
2734 so, the driver will manage that printer.
2735 See also header of drivers/char/lp.c.
2738 Sets loops_per_jiffy to given constant, thus avoiding
2739 time-consuming boot-time autodetection (up to 250 ms per
2740 CPU). 0 enables autodetection (default). To determine
2741 the correct value for your kernel, boot with normal
2742 autodetection and see what value is printed. Note that
2743 on SMP systems the preset will be applied to all CPUs,
2744 which is likely to cause problems if your CPUs need
2745 significantly divergent settings. An incorrect value
2746 will cause delays in the kernel to be wrong, leading to
2747 unpredictable I/O errors and other breakage. Although
2748 unlikely, in the extreme case this might damage your
2752 Format: <io>,<irq>,<dma>
2754 lsm.debug [SECURITY] Enable LSM initialization debugging output.
2757 [SECURITY] Choose order of LSM initialization. This
2758 overrides CONFIG_LSM, and the "security=" parameter.
2760 machvec= [IA-64] Force the use of a particular machine-vector
2761 (machvec) in a generic kernel.
2762 Example: machvec=hpzx1
2764 machtype= [Loongson] Share the same kernel image file between
2765 different yeeloong laptops.
2766 Example: machtype=lemote-yeeloong-2f-7inch
2768 max_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory greater
2769 than or equal to this physical address is ignored.
2771 maxcpus= [SMP] Maximum number of processors that an SMP kernel
2772 will bring up during bootup. maxcpus=n : n >= 0 limits
2773 the kernel to bring up 'n' processors. Surely after
2774 bootup you can bring up the other plugged cpu by executing
2775 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
2776 only takes effect during system bootup.
2777 While n=0 is a special case, it is equivalent to "nosmp",
2778 which also disables the IO APIC.
2780 max_loop= [LOOP] The number of loop block devices that get
2781 (loop.max_loop) unconditionally pre-created at init time. The default
2782 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
2783 of statically allocating a predefined number, loop
2784 devices can be requested on-demand with the
2785 /dev/loop-control interface.
2787 mce [X86-32] Machine Check Exception
2789 mce=option [X86-64] See Documentation/x86/x86_64/boot-options.rst
2791 md= [HW] RAID subsystems devices and level
2792 See Documentation/admin-guide/md.rst.
2795 Format: <first>,<last>
2796 Specifies range of consoles to be captured by the MDA.
2799 Control mitigation for the Micro-architectural Data
2800 Sampling (MDS) vulnerability.
2802 Certain CPUs are vulnerable to an exploit against CPU
2803 internal buffers which can forward information to a
2804 disclosure gadget under certain conditions.
2806 In vulnerable processors, the speculatively
2807 forwarded data can be used in a cache side channel
2808 attack, to access data to which the attacker does
2809 not have direct access.
2811 This parameter controls the MDS mitigation. The
2814 full - Enable MDS mitigation on vulnerable CPUs
2815 full,nosmt - Enable MDS mitigation and disable
2816 SMT on vulnerable CPUs
2817 off - Unconditionally disable MDS mitigation
2819 On TAA-affected machines, mds=off can be prevented by
2820 an active TAA mitigation as both vulnerabilities are
2821 mitigated with the same mechanism so in order to disable
2822 this mitigation, you need to specify tsx_async_abort=off
2825 Not specifying this option is equivalent to
2828 For details see: Documentation/admin-guide/hw-vuln/mds.rst
2830 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
2831 Amount of memory to be used in cases as follows:
2834 2 when the kernel is not able to see the whole system memory;
2835 3 memory that lies after 'mem=' boundary is excluded from
2836 the hypervisor, then assigned to KVM guests.
2838 [X86] Work as limiting max address. Use together
2839 with memmap= to avoid physical address space collisions.
2840 Without memmap= PCI devices could be placed at addresses
2841 belonging to unused RAM.
2843 Note that this only takes effects during boot time since
2844 in above case 3, memory may need be hot added after boot
2845 if system memory of hypervisor is not sufficient.
2847 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
2851 [KNL,SH] Allow user to override the default size for
2852 per-device physically contiguous DMA buffers.
2854 memhp_default_state=online/offline
2855 [KNL] Set the initial state for the memory hotplug
2856 onlining policy. If not specified, the default value is
2857 set according to the
2858 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
2860 See Documentation/admin-guide/mm/memory-hotplug.rst.
2862 memmap=exactmap [KNL,X86] Enable setting of an exact
2863 E820 memory map, as specified by the user.
2864 Such memmap=exactmap lines can be constructed based on
2865 BIOS output or other requirements. See the memmap=nn@ss
2868 memmap=nn[KMG]@ss[KMG]
2869 [KNL, X86, MIPS, XTENSA] Force usage of a specific region of memory.
2870 Region of memory to be used is from ss to ss+nn.
2871 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
2872 which limits max address to nn[KMG].
2873 Multiple different regions can be specified,
2876 memmap=100M@2G,100M#3G,1G!1024G
2878 memmap=nn[KMG]#ss[KMG]
2879 [KNL,ACPI] Mark specific memory as ACPI data.
2880 Region of memory to be marked is from ss to ss+nn.
2882 memmap=nn[KMG]$ss[KMG]
2883 [KNL,ACPI] Mark specific memory as reserved.
2884 Region of memory to be reserved is from ss to ss+nn.
2885 Example: Exclude memory from 0x18690000-0x1869ffff
2886 memmap=64K$0x18690000
2888 memmap=0x10000$0x18690000
2889 Some bootloaders may need an escape character before '$',
2890 like Grub2, otherwise '$' and the following number
2893 memmap=nn[KMG]!ss[KMG]
2894 [KNL,X86] Mark specific memory as protected.
2895 Region of memory to be used, from ss to ss+nn.
2896 The memory region may be marked as e820 type 12 (0xc)
2897 and is NVDIMM or ADR memory.
2899 memmap=<size>%<offset>-<oldtype>+<newtype>
2900 [KNL,ACPI] Convert memory within the specified region
2901 from <oldtype> to <newtype>. If "-<oldtype>" is left
2902 out, the whole region will be marked as <newtype>,
2903 even if previously unavailable. If "+<newtype>" is left
2904 out, matching memory will be removed. Types are
2905 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
2906 3 = ACPI, 12 = PRAM.
2908 memory_corruption_check=0/1 [X86]
2909 Some BIOSes seem to corrupt the first 64k of
2910 memory when doing things like suspend/resume.
2911 Setting this option will scan the memory
2912 looking for corruption. Enabling this will
2913 both detect corruption and prevent the kernel
2914 from using the memory being corrupted.
2915 However, its intended as a diagnostic tool; if
2916 repeatable BIOS-originated corruption always
2917 affects the same memory, you can use memmap=
2918 to prevent the kernel from using that memory.
2920 memory_corruption_check_size=size [X86]
2921 By default it checks for corruption in the low
2922 64k, making this memory unavailable for normal
2923 use. Use this parameter to scan for
2924 corruption in more or less memory.
2926 memory_corruption_check_period=seconds [X86]
2927 By default it checks for corruption every 60
2928 seconds. Use this parameter to check at some
2929 other rate. 0 disables periodic checking.
2931 memory_hotplug.memmap_on_memory
2932 [KNL,X86,ARM] Boolean flag to enable this feature.
2933 Format: {on | off (default)}
2934 When enabled, runtime hotplugged memory will
2935 allocate its internal metadata (struct pages)
2936 from the hotadded memory which will allow to
2937 hotadd a lot of memory without requiring
2938 additional memory to do so.
2939 This feature is disabled by default because it
2940 has some implication on large (e.g. GB)
2941 allocations in some configurations (e.g. small
2943 The state of the flag can be read in
2944 /sys/module/memory_hotplug/parameters/memmap_on_memory.
2945 Note that even when enabled, there are a few cases where
2946 the feature is not effective.
2948 This is not compatible with hugetlb_free_vmemmap. If
2949 both parameters are enabled, hugetlb_free_vmemmap takes
2950 precedence over memory_hotplug.memmap_on_memory.
2952 memtest= [KNL,X86,ARM,M68K,PPC,RISCV] Enable memtest
2954 default : 0 <disable>
2955 Specifies the number of memtest passes to be
2956 performed. Each pass selects another test
2957 pattern from a given set of patterns. Memtest
2958 fills the memory with this pattern, validates
2959 memory contents and reserves bad memory
2960 regions that are detected.
2962 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
2963 Valid arguments: on, off
2964 Default (depends on kernel configuration option):
2965 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
2966 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
2967 mem_encrypt=on: Activate SME
2968 mem_encrypt=off: Do not activate SME
2970 Refer to Documentation/virt/kvm/amd-memory-encryption.rst
2971 for details on when memory encryption can be activated.
2973 mem_sleep_default= [SUSPEND] Default system suspend mode:
2974 s2idle - Suspend-To-Idle
2975 shallow - Power-On Suspend or equivalent (if supported)
2976 deep - Suspend-To-RAM or equivalent (if supported)
2977 See Documentation/admin-guide/pm/sleep-states.rst.
2979 meye.*= [HW] Set MotionEye Camera parameters
2980 See Documentation/admin-guide/media/meye.rst.
2982 mfgpt_irq= [IA-32] Specify the IRQ to use for the
2983 Multi-Function General Purpose Timers on AMD Geode
2986 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
2987 the BIOS has incorrectly applied a workaround. TinyBIOS
2988 version 0.98 is known to be affected, 0.99 fixes the
2989 problem by letting the user disable the workaround.
2993 min_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory below this
2994 physical address is ignored.
2996 mini2440= [ARM,HW,KNL]
2997 Format:[0..2][b][c][t]
2999 MINI2440 configuration specification:
3000 0 - The attached screen is the 3.5" TFT
3001 1 - The attached screen is the 7" TFT
3002 2 - The VGA Shield is attached (1024x768)
3003 Leaving out the screen size parameter will not load
3004 the TFT driver, and the framebuffer will be left
3006 b - Enable backlight. The TFT backlight pin will be
3007 linked to the kernel VESA blanking code and a GPIO
3008 LED. This parameter is not necessary when using the
3010 c - Enable the s3c camera interface.
3011 t - Reserved for enabling touchscreen support. The
3012 touchscreen support is not enabled in the mainstream
3013 kernel as of 2.6.30, a preliminary port can be found
3014 in the "bleeding edge" mini2440 support kernel at
3015 https://repo.or.cz/w/linux-2.6/mini2440.git
3018 [X86,PPC,S390,ARM64] Control optional mitigations for
3019 CPU vulnerabilities. This is a set of curated,
3020 arch-independent options, each of which is an
3021 aggregation of existing arch-specific options.
3024 Disable all optional CPU mitigations. This
3025 improves system performance, but it may also
3026 expose users to several CPU vulnerabilities.
3027 Equivalent to: nopti [X86,PPC]
3029 nospectre_v1 [X86,PPC]
3031 nospectre_v2 [X86,PPC,S390,ARM64]
3032 spectre_v2_user=off [X86]
3033 spec_store_bypass_disable=off [X86,PPC]
3034 ssbd=force-off [ARM64]
3037 tsx_async_abort=off [X86]
3038 kvm.nx_huge_pages=off [X86]
3039 no_entry_flush [PPC]
3040 no_uaccess_flush [PPC]
3043 This does not have any effect on
3044 kvm.nx_huge_pages when
3045 kvm.nx_huge_pages=force.
3048 Mitigate all CPU vulnerabilities, but leave SMT
3049 enabled, even if it's vulnerable. This is for
3050 users who don't want to be surprised by SMT
3051 getting disabled across kernel upgrades, or who
3052 have other ways of avoiding SMT-based attacks.
3053 Equivalent to: (default behavior)
3056 Mitigate all CPU vulnerabilities, disabling SMT
3057 if needed. This is for users who always want to
3058 be fully mitigated, even if it means losing SMT.
3059 Equivalent to: l1tf=flush,nosmt [X86]
3060 mds=full,nosmt [X86]
3061 tsx_async_abort=full,nosmt [X86]
3064 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
3065 parameter allows control of the logging verbosity for
3066 the additional memory initialisation checks. A value
3067 of 0 disables mminit logging and a level of 4 will
3068 log everything. Information is printed at KERN_DEBUG
3069 so loglevel=8 may also need to be specified.
3072 [KNL] When CONFIG_MODULE_SIG is set, this means that
3073 modules without (valid) signatures will fail to load.
3074 Note that if CONFIG_MODULE_SIG_FORCE is set, that
3075 is always true, so this option does nothing.
3077 module_blacklist= [KNL] Do not load a comma-separated list of
3078 modules. Useful for debugging problem modules.
3081 [MOUSE] Maximum time between finger touching and
3082 leaving touchpad surface for touch to be considered
3083 a tap and be reported as a left button click (for
3084 touchpads working in absolute mode only).
3086 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
3087 reporting absolute coordinates, such as tablets
3088 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
3089 reporting absolute coordinates, such as tablets
3091 movablecore= [KNL,X86,IA-64,PPC]
3092 Format: nn[KMGTPE] | nn%
3093 This parameter is the complement to kernelcore=, it
3094 specifies the amount of memory used for migratable
3095 allocations. If both kernelcore and movablecore is
3096 specified, then kernelcore will be at *least* the
3097 specified value but may be more. If movablecore on its
3098 own is specified, the administrator must be careful
3099 that the amount of memory usable for all allocations
3102 movable_node [KNL] Boot-time switch to make hotplugable memory
3103 NUMA nodes to be movable. This means that the memory
3104 of such nodes will be usable only for movable
3105 allocations which rules out almost all kernel
3106 allocations. Use with caution!
3108 MTD_Partition= [MTD]
3109 Format: <name>,<region-number>,<size>,<offset>
3111 MTD_Region= [MTD] Format:
3112 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
3115 See drivers/mtd/parsers/cmdlinepart.c
3117 multitce=off [PPC] This parameter disables the use of the pSeries
3118 firmware feature for updating multiple TCE entries
3121 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
3123 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
3125 boundary - index of last SLC block on Flex-OneNAND.
3126 The remaining blocks are configured as MLC blocks.
3127 lock - Configure if Flex-OneNAND boundary should be locked.
3128 Once locked, the boundary cannot be changed.
3129 1 indicates lock status, 0 indicates unlock status.
3132 ARM/S3C2412 JIVE boot control
3134 See arch/arm/mach-s3c/mach-jive.c
3136 mtouchusb.raw_coordinates=
3137 [HW] Make the MicroTouch USB driver use raw coordinates
3138 ('y', default) or cooked coordinates ('n')
3140 mtrr_chunk_size=nn[KMG] [X86]
3141 used for mtrr cleanup. It is largest continuous chunk
3142 that could hold holes aka. UC entries.
3144 mtrr_gran_size=nn[KMG] [X86]
3145 Used for mtrr cleanup. It is granularity of mtrr block.
3147 Large value could prevent small alignment from
3150 mtrr_spare_reg_nr=n [X86]
3152 Range: 0,7 : spare reg number
3154 Used for mtrr cleanup. It is spare mtrr entries number.
3155 Set to 2 or more if your graphical card needs more.
3157 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
3159 netdev= [NET] Network devices parameters
3160 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
3161 Note that mem_start is often overloaded to mean
3162 something different and driver-specific.
3163 This usage is only documented in each driver source
3167 [NETFILTER] Enable connection tracking flow accounting
3168 0 to disable accounting
3169 1 to enable accounting
3172 nfsaddrs= [NFS] Deprecated. Use ip= instead.
3173 See Documentation/admin-guide/nfs/nfsroot.rst.
3175 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
3176 See Documentation/admin-guide/nfs/nfsroot.rst.
3178 nfsrootdebug [NFS] enable nfsroot debugging messages.
3179 See Documentation/admin-guide/nfs/nfsroot.rst.
3181 nfs.callback_nr_threads=
3182 [NFSv4] set the total number of threads that the
3183 NFS client will assign to service NFSv4 callback
3186 nfs.callback_tcpport=
3187 [NFS] set the TCP port on which the NFSv4 callback
3188 channel should listen.
3191 [NFS] sets the pathname to the program which is used
3192 to update the NFS client cache entries.
3194 nfs.cache_getent_timeout=
3195 [NFS] sets the timeout after which an attempt to
3196 update a cache entry is deemed to have failed.
3198 nfs.idmap_cache_timeout=
3199 [NFS] set the maximum lifetime for idmapper cache
3203 [NFS] enable 64-bit inode numbers.
3204 If zero, the NFS client will fake up a 32-bit inode
3205 number for the readdir() and stat() syscalls instead
3206 of returning the full 64-bit number.
3207 The default is to return 64-bit inode numbers.
3209 nfs.max_session_cb_slots=
3210 [NFSv4.1] Sets the maximum number of session
3211 slots the client will assign to the callback
3212 channel. This determines the maximum number of
3213 callbacks the client will process in parallel for
3214 a particular server.
3216 nfs.max_session_slots=
3217 [NFSv4.1] Sets the maximum number of session slots
3218 the client will attempt to negotiate with the server.
3219 This limits the number of simultaneous RPC requests
3220 that the client can send to the NFSv4.1 server.
3221 Note that there is little point in setting this
3222 value higher than the max_tcp_slot_table_limit.
3224 nfs.nfs4_disable_idmapping=
3225 [NFSv4] When set to the default of '1', this option
3226 ensures that both the RPC level authentication
3227 scheme and the NFS level operations agree to use
3228 numeric uids/gids if the mount is using the
3229 'sec=sys' security flavour. In effect it is
3230 disabling idmapping, which can make migration from
3231 legacy NFSv2/v3 systems to NFSv4 easier.
3232 Servers that do not support this mode of operation
3233 will be autodetected by the client, and it will fall
3234 back to using the idmapper.
3235 To turn off this behaviour, set the value to '0'.
3237 [NFS4] Specify an additional fixed unique ident-
3238 ification string that NFSv4 clients can insert into
3239 their nfs_client_id4 string. This is typically a
3240 UUID that is generated at system install time.
3242 nfs.send_implementation_id =
3243 [NFSv4.1] Send client implementation identification
3244 information in exchange_id requests.
3245 If zero, no implementation identification information
3247 The default is to send the implementation identification
3250 nfs.recover_lost_locks =
3251 [NFSv4] Attempt to recover locks that were lost due
3252 to a lease timeout on the server. Please note that
3253 doing this risks data corruption, since there are
3254 no guarantees that the file will remain unchanged
3255 after the locks are lost.
3256 If you want to enable the kernel legacy behaviour of
3257 attempting to recover these locks, then set this
3259 The default parameter value of '0' causes the kernel
3260 not to attempt recovery of lost locks.
3262 nfs4.layoutstats_timer =
3263 [NFSv4.2] Change the rate at which the kernel sends
3264 layoutstats to the pNFS metadata server.
3266 Setting this to value to 0 causes the kernel to use
3267 whatever value is the default set by the layout
3268 driver. A non-zero value sets the minimum interval
3269 in seconds between layoutstats transmissions.
3271 nfsd.inter_copy_offload_enable =
3272 [NFSv4.2] When set to 1, the server will support
3273 server-to-server copies for which this server is
3274 the destination of the copy.
3276 nfsd.nfsd4_ssc_umount_timeout =
3277 [NFSv4.2] When used as the destination of a
3278 server-to-server copy, knfsd temporarily mounts
3279 the source server. It caches the mount in case
3280 it will be needed again, and discards it if not
3281 used for the number of milliseconds specified by
3284 nfsd.nfs4_disable_idmapping=
3285 [NFSv4] When set to the default of '1', the NFSv4
3286 server will return only numeric uids and gids to
3287 clients using auth_sys, and will accept numeric uids
3288 and gids from such clients. This is intended to ease
3289 migration from NFSv2/v3.
3292 nmi_backtrace.backtrace_idle [KNL]
3293 Dump stacks even of idle CPUs in response to an
3294 NMI stack-backtrace request.
3296 nmi_debug= [KNL,SH] Specify one or more actions to take
3297 when a NMI is triggered.
3298 Format: [state][,regs][,debounce][,die]
3300 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
3301 Format: [panic,][nopanic,][num]
3303 0 - turn hardlockup detector in nmi_watchdog off
3304 1 - turn hardlockup detector in nmi_watchdog on
3305 When panic is specified, panic when an NMI watchdog
3306 timeout occurs (or 'nopanic' to not panic on an NMI
3307 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
3308 To disable both hard and soft lockup detectors,
3309 please see 'nowatchdog'.
3310 This is useful when you use a panic=... timeout and
3311 need the box quickly up again.
3313 These settings can be accessed at runtime via
3314 the nmi_watchdog and hardlockup_panic sysctls.
3316 netpoll.carrier_timeout=
3317 [NET] Specifies amount of time (in seconds) that
3318 netpoll should wait for a carrier. By default netpoll
3321 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3322 emulation library even if a 387 maths coprocessor
3325 no5lvl [X86-64] Disable 5-level paging mode. Forces
3326 kernel to use 4-level paging instead.
3328 nofsgsbase [X86] Disables FSGSBASE instructions.
3331 [HW] Never suspend the console
3332 Disable suspending of consoles during suspend and
3333 hibernate operations. Once disabled, debugging
3334 messages can reach various consoles while the rest
3335 of the system is being put to sleep (ie, while
3336 debugging driver suspend/resume hooks). This may
3337 not work reliably with all consoles, but is known
3338 to work with serial and VGA consoles.
3339 To facilitate more flexible debugging, we also add
3340 console_suspend, a printk module parameter to control
3341 it. Users could use console_suspend (usually
3342 /sys/module/printk/parameters/console_suspend) to
3343 turn on/off it dynamically.
3345 novmcoredd [KNL,KDUMP]
3346 Disable device dump. Device dump allows drivers to
3347 append dump data to vmcore so you can collect driver
3348 specified debug info. Drivers can append the data
3349 without any limit and this data is stored in memory,
3350 so this may cause significant memory stress. Disabling
3351 device dump can help save memory but the driver debug
3352 data will be no longer available. This parameter
3353 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
3356 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
3357 caches in the slab allocator. Saves per-node memory,
3358 but will impact performance.
3362 noaltinstr [S390] Disables alternative instructions patching
3363 (CPU alternatives feature).
3365 noapic [SMP,APIC] Tells the kernel to not make use of any
3366 IOAPICs that may be present in the system.
3368 noautogroup Disable scheduler automatic task group creation.
3370 nobats [PPC] Do not use BATs for mapping kernel lowmem
3371 on "Classic" PPC cores.
3375 noclflush [BUGS=X86] Don't use the CLFLUSH instruction
3377 delayacct [KNL] Enable per-task delay accounting
3379 nodsp [SH] Disable hardware DSP at boot time.
3381 noefi Disable EFI runtime services support.
3383 no_entry_flush [PPC] Don't flush the L1-D cache when entering the kernel.
3388 On X86-32 available only on PAE configured kernels.
3389 noexec=on: enable non-executable mappings (default)
3390 noexec=off: disable non-executable mappings
3393 Disable SMAP (Supervisor Mode Access Prevention)
3394 even if it is supported by processor.
3397 Disable SMEP (Supervisor Mode Execution Prevention)
3398 even if it is supported by processor.
3401 This affects only 32-bit executables.
3402 noexec32=on: enable non-executable mappings (default)
3403 read doesn't imply executable mappings
3404 noexec32=off: disable non-executable mappings
3405 read implies executable mappings
3407 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3409 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3410 register save and restore. The kernel will only save
3411 legacy floating-point registers on task switch.
3413 nohugeiomap [KNL,X86,PPC,ARM64] Disable kernel huge I/O mappings.
3415 nohugevmalloc [PPC] Disable kernel huge vmalloc mappings.
3417 nosmt [KNL,S390] Disable symmetric multithreading (SMT).
3418 Equivalent to smt=1.
3420 [KNL,X86] Disable symmetric multithreading (SMT).
3421 nosmt=force: Force disable SMT, cannot be undone
3422 via the sysfs control file.
3424 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1
3425 (bounds check bypass). With this option data leaks are
3426 possible in the system.
3428 nospectre_v2 [X86,PPC_FSL_BOOK3E,ARM64] Disable all mitigations for
3429 the Spectre variant 2 (indirect branch prediction)
3430 vulnerability. System may allow data leaks with this
3433 nospec_store_bypass_disable
3434 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
3437 [PPC] Don't flush the L1-D cache after accessing user data.
3439 noxsave [BUGS=X86] Disables x86 extended register state save
3440 and restore using xsave. The kernel will fallback to
3441 enabling legacy floating-point and sse state.
3443 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
3444 register states. The kernel will fall back to use
3445 xsave to save the states. By using this parameter,
3446 performance of saving the states is degraded because
3447 xsave doesn't support modified optimization while
3448 xsaveopt supports it on xsaveopt enabled systems.
3450 noxsaves [X86] Disables xsaves and xrstors used in saving and
3451 restoring x86 extended register state in compacted
3452 form of xsave area. The kernel will fall back to use
3453 xsaveopt and xrstor to save and restore the states
3454 in standard form of xsave area. By using this
3455 parameter, xsave area per process might occupy more
3456 memory on xsaves enabled systems.
3458 nohlt [ARM,ARM64,MICROBLAZE,SH] Forces the kernel to busy wait
3459 in do_idle() and not use the arch_cpu_idle()
3460 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
3461 to be effective. This is useful on platforms where the
3462 sleep(SH) or wfi(ARM,ARM64) instructions do not work
3463 correctly or when doing power measurements to evalute
3464 the impact of the sleep instructions. This is also
3465 useful when using JTAG debugger.
3467 no_file_caps Tells the kernel not to honor file capabilities. The
3468 only way then for a file to be executed with privilege
3469 is to be setuid root or executed by root.
3471 nohalt [IA-64] Tells the kernel not to use the power saving
3472 function PAL_HALT_LIGHT when idle. This increases
3473 power-consumption. On the positive side, it reduces
3474 interrupt wake-up latency, which may improve performance
3475 in certain environments such as networked servers or
3479 Force pointers printed to the console or buffers to be
3480 unhashed. By default, when a pointer is printed via %p
3481 format string, that pointer is "hashed", i.e. obscured
3482 by hashing the pointer value. This is a security feature
3483 that hides actual kernel addresses from unprivileged
3484 users, but it also makes debugging the kernel more
3485 difficult since unequal pointers can no longer be
3486 compared. However, if this command-line option is
3487 specified, then all normal pointers will have their true
3488 value printed. Pointers printed via %pK may still be
3489 hashed. This option should only be specified when
3490 debugging the kernel. Please do not use on production
3493 nohibernate [HIBERNATION] Disable hibernation and resume.
3495 nohz= [KNL] Boottime enable/disable dynamic ticks
3496 Valid arguments: on, off
3499 nohz_full= [KNL,BOOT,SMP,ISOL]
3500 The argument is a cpu list, as described above.
3501 In kernels built with CONFIG_NO_HZ_FULL=y, set
3502 the specified list of CPUs whose tick will be stopped
3503 whenever possible. The boot CPU will be forced outside
3504 the range to maintain the timekeeping. Any CPUs
3505 in this list will have their RCU callbacks offloaded,
3506 just as if they had also been called out in the
3507 rcu_nocbs= boot parameter.
3509 noiotrap [SH] Disables trapped I/O port accesses.
3511 noirqdebug [X86-32] Disables the code which attempts to detect and
3512 disable unhandled interrupt sources.
3514 no_timer_check [X86,APIC] Disables the code which tests for
3515 broken timer IRQ sources.
3517 noisapnp [ISAPNP] Disables ISA PnP code.
3519 noinitrd [RAM] Tells the kernel not to load any configured
3522 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
3524 [Deprecated - use intremap=off]
3528 noinvpcid [X86] Disable the INVPCID cpu feature.
3530 nojitter [IA-64] Disables jitter checking for ITC timers.
3532 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
3534 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
3538 [X86,PV_OPS] Disable paravirtualized VMware scheduler
3539 clock and use the default one.
3541 no-steal-acc [X86,PV_OPS,ARM64] Disable paravirtualized steal time
3542 accounting. steal time is computed, but won't
3543 influence scheduler behaviour
3545 nolapic [X86-32,APIC] Do not enable or use the local APIC.
3547 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
3549 noltlbs [PPC] Do not use large page/tlb entries for kernel
3550 lowmem mapping on PPC40x and PPC8xx
3552 nomca [IA-64] Disable machine check abort handling
3554 nomce [X86-32] Disable Machine Check Exception
3556 nomfgpt [X86-32] Disable Multi-Function General Purpose
3557 Timer usage (for AMD Geode machines).
3559 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3560 shutdown the other cpus. Instead use the REBOOT_VECTOR
3563 nomodeset Disable kernel modesetting. DRM drivers will not perform
3564 display-mode changes or accelerated rendering. Only the
3565 system framebuffer will be available for use if this was
3566 set-up by the firmware or boot loader.
3568 Useful as fallback, or for testing and debugging.
3570 nomodule Disable module load
3572 nopat [X86] Disable PAT (page attribute table extension of
3573 pagetables) support.
3575 nopcid [X86-64] Disable the PCID cpu feature.
3577 norandmaps Don't use address space randomization. Equivalent to
3578 echo 0 > /proc/sys/kernel/randomize_va_space
3580 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
3581 with UP alternatives
3583 nordrand [X86] Disable kernel use of the RDRAND and
3584 RDSEED instructions even if they are supported
3585 by the processor. RDRAND and RDSEED are still
3586 available to user space applications.
3588 noresume [SWSUSP] Disables resume and restores original swap
3591 no-scroll [VGA] Disables scrollback.
3592 This is required for the Braillex ib80-piezo Braille
3593 reader made by F.H. Papenmeier (Germany).
3597 nosep [BUGS=X86-32] Disables x86 SYSENTER/SYSEXIT support.
3599 nosgx [X86-64,SGX] Disables Intel SGX kernel support.
3601 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
3602 and disable the IO APIC. legacy for "maxcpus=0".
3604 nosoftlockup [KNL] Disable the soft-lockup detector.
3606 nosync [HW,M68K] Disables sync negotiation for all devices.
3608 nowatchdog [KNL] Disable both lockup detectors, i.e.
3609 soft-lockup and NMI watchdog (hard-lockup).
3613 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
3615 cpu0_hotplug [X86] Turn on CPU0 hotplug feature when
3616 CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off.
3617 Some features depend on CPU0. Known dependencies are:
3618 1. Resume from suspend/hibernate depends on CPU0.
3619 Suspend/hibernate will fail if CPU0 is offline and you
3620 need to online CPU0 before suspend/hibernate.
3621 2. PIC interrupts also depend on CPU0. CPU0 can't be
3622 removed if a PIC interrupt is detected.
3623 It's said poweroff/reboot may depend on CPU0 on some
3624 machines although I haven't seen such issues so far
3625 after CPU0 is offline on a few tested machines.
3626 If the dependencies are under your control, you can
3627 turn on cpu0_hotplug.
3629 nps_mtm_hs_ctr= [KNL,ARC]
3630 This parameter sets the maximum duration, in
3631 cycles, each HW thread of the CTOP can run
3632 without interruptions, before HW switches it.
3633 The actual maximum duration is 16 times this
3635 Format: integer between 1 and 255
3638 nptcg= [IA-64] Override max number of concurrent global TLB
3639 purges which is reported from either PAL_VM_SUMMARY or
3642 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
3643 could support. nr_cpus=n : n >= 1 limits the kernel to
3644 support 'n' processors. It could be larger than the
3645 number of already plugged CPU during bootup, later in
3646 runtime you can physically add extra cpu until it reaches
3647 n. So during boot up some boot time memory for per-cpu
3648 variables need be pre-allocated for later physical cpu
3651 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
3653 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86] Disable NUMA, Only
3654 set up a single NUMA node spanning all memory.
3656 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic
3658 Allowed values are enable and disable
3660 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
3661 'node', 'default' can be specified
3662 This can be set from sysctl after boot.
3663 See Documentation/admin-guide/sysctl/vm.rst for details.
3665 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
3666 See Documentation/core-api/debugging-via-ohci1394.rst for more
3669 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
3670 Rather than timing out after 20 ms if an EC
3671 command is not properly ACKed, override the length
3672 of the timeout. We have interrupts disabled while
3673 waiting for the ACK, so if this is set too high
3674 interrupts *may* be lost!
3676 omap_mux= [OMAP] Override bootloader pin multiplexing.
3677 Format: <mux_mode0.mode_name=value>...
3678 For example, to override I2C bus2:
3679 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
3681 oops=panic Always panic on oopses. Default is to just kill the
3682 process, but there is a small probability of
3683 deadlocking the machine.
3684 This will also cause panics on machine check exceptions.
3685 Useful together with panic=30 to trigger a reboot.
3688 [KNL] Boolean flag to control whether the page allocator
3689 should randomize its free lists. The randomization may
3690 be automatically enabled if the kernel detects it is
3691 running on a platform with a direct-mapped memory-side
3692 cache, and this parameter can be used to
3693 override/disable that behavior. The state of the flag
3694 can be read from sysfs at:
3695 /sys/module/page_alloc/parameters/shuffle.
3697 page_owner= [KNL] Boot-time page_owner enabling option.
3698 Storage of the information about who allocated
3699 each page is disabled in default. With this switch,
3701 on: enable the feature
3703 page_poison= [KNL] Boot-time parameter changing the state of
3704 poisoning on the buddy allocator, available with
3705 CONFIG_PAGE_POISONING=y.
3706 off: turn off poisoning (default)
3707 on: turn on poisoning
3709 page_reporting.page_reporting_order=
3710 [KNL] Minimal page reporting order
3712 Adjust the minimal page reporting order. The page
3713 reporting is disabled when it exceeds (MAX_ORDER-1).
3715 panic= [KNL] Kernel behaviour on panic: delay <timeout>
3716 timeout > 0: seconds before rebooting
3717 timeout = 0: wait forever
3718 timeout < 0: reboot immediately
3721 panic_print= Bitmask for printing system info when panic happens.
3722 User can chose combination of the following bits:
3723 bit 0: print all tasks info
3724 bit 1: print system memory info
3725 bit 2: print timer info
3726 bit 3: print locks info if CONFIG_LOCKDEP is on
3727 bit 4: print ftrace buffer
3728 bit 5: print all printk messages in buffer
3730 panic_on_taint= Bitmask for conditionally calling panic() in add_taint()
3731 Format: <hex>[,nousertaint]
3732 Hexadecimal bitmask representing the set of TAINT flags
3733 that will cause the kernel to panic when add_taint() is
3734 called with any of the flags in this set.
3735 The optional switch "nousertaint" can be utilized to
3736 prevent userspace forced crashes by writing to sysctl
3737 /proc/sys/kernel/tainted any flagset matching with the
3738 bitmask set on panic_on_taint.
3739 See Documentation/admin-guide/tainted-kernels.rst for
3740 extra details on the taint flags that users can pick
3741 to compose the bitmask to assign to panic_on_taint.
3743 panic_on_warn panic() instead of WARN(). Useful to cause kdump
3746 crash_kexec_post_notifiers
3747 Run kdump after running panic-notifiers and dumping
3748 kmsg. This only for the users who doubt kdump always
3749 succeeds in any situation.
3750 Note that this also increases risks of kdump failure,
3751 because some panic notifiers can make the crashed
3752 kernel more unstable.
3754 parkbd.port= [HW] Parallel port number the keyboard adapter is
3755 connected to, default is 0.
3757 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
3758 0 for XT, 1 for AT (default is AT).
3761 parport= [HW,PPT] Specify parallel ports. 0 disables.
3762 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
3763 Use 'auto' to force the driver to use any
3764 IRQ/DMA settings detected (the default is to
3765 ignore detected IRQ/DMA settings because of
3766 possible conflicts). You can specify the base
3767 address, IRQ, and DMA settings; IRQ and DMA
3768 should be numbers, or 'auto' (for using detected
3769 settings on that particular port), or 'nofifo'
3770 (to avoid using a FIFO even if it is detected).
3771 Parallel ports are assigned in the order they
3772 are specified on the command line, starting
3775 parport_init_mode= [HW,PPT]
3776 Configure VIA parallel port to operate in
3777 a specific mode. This is necessary on Pegasos
3778 computer where firmware has no options for setting
3779 up parallel port mode and sets it to spp.
3780 Currently this function knows 686a and 8231 chips.
3781 Format: [spp|ps2|epp|ecp|ecpepp]
3783 pata_legacy.all= [HW,LIBATA]
3785 Set to non-zero to probe primary and secondary ISA
3786 port ranges on PCI systems where no PCI PATA device
3787 has been found at either range. Disabled by default.
3789 pata_legacy.autospeed= [HW,LIBATA]
3791 Set to non-zero if a chip is present that snoops speed
3792 changes. Disabled by default.
3794 pata_legacy.ht6560a= [HW,LIBATA]
3796 Set to 1, 2, or 3 for HT 6560A on the primary channel,
3797 the secondary channel, or both channels respectively.
3798 Disabled by default.
3800 pata_legacy.ht6560b= [HW,LIBATA]
3802 Set to 1, 2, or 3 for HT 6560B on the primary channel,
3803 the secondary channel, or both channels respectively.
3804 Disabled by default.
3806 pata_legacy.iordy_mask= [HW,LIBATA]
3808 IORDY enable mask. Set individual bits to allow IORDY
3809 for the respective channel. Bit 0 is for the first
3810 legacy channel handled by this driver, bit 1 is for
3811 the second channel, and so on. The sequence will often
3812 correspond to the primary legacy channel, the secondary
3813 legacy channel, and so on, but the handling of a PCI
3814 bus and the use of other driver options may interfere
3815 with the sequence. By default IORDY is allowed across
3818 pata_legacy.opti82c46x= [HW,LIBATA]
3820 Set to 1, 2, or 3 for Opti 82c611A on the primary
3821 channel, the secondary channel, or both channels
3822 respectively. Disabled by default.
3824 pata_legacy.opti82c611a= [HW,LIBATA]
3826 Set to 1, 2, or 3 for Opti 82c465MV on the primary
3827 channel, the secondary channel, or both channels
3828 respectively. Disabled by default.
3830 pata_legacy.pio_mask= [HW,LIBATA]
3832 PIO mode mask for autospeed devices. Set individual
3833 bits to allow the use of the respective PIO modes.
3834 Bit 0 is for mode 0, bit 1 is for mode 1, and so on.
3835 All modes allowed by default.
3837 pata_legacy.probe_all= [HW,LIBATA]
3839 Set to non-zero to probe tertiary and further ISA
3840 port ranges on PCI systems. Disabled by default.
3842 pata_legacy.probe_mask= [HW,LIBATA]
3844 Probe mask for legacy ISA PATA ports. Depending on
3845 platform configuration and the use of other driver
3846 options up to 6 legacy ports are supported: 0x1f0,
3847 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing
3848 of individual ports can be disabled by setting the
3849 corresponding bits in the mask to 1. Bit 0 is for
3850 the first port in the list above (0x1f0), and so on.
3851 By default all supported ports are probed.
3853 pata_legacy.qdi= [HW,LIBATA]
3855 Set to non-zero to probe QDI controllers. By default
3856 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise.
3858 pata_legacy.winbond= [HW,LIBATA]
3860 Set to non-zero to probe Winbond controllers. Use
3861 the standard I/O port (0x130) if 1, otherwise the
3862 value given is the I/O port to use (typically 0x1b0).
3863 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE,
3866 pata_platform.pio_mask= [HW,LIBATA]
3868 Supported PIO mode mask. Set individual bits to allow
3869 the use of the respective PIO modes. Bit 0 is for
3870 mode 0, bit 1 is for mode 1, and so on. Mode 0 only
3874 Halt all CPUs after the first oops has been printed for
3875 the specified number of seconds. This is to be used if
3876 your oopses keep scrolling off the screen.
3881 See header of drivers/block/paride/pcd.c.
3882 See also Documentation/admin-guide/blockdev/paride.rst.
3884 pci=option[,option...] [PCI] various PCI subsystem options.
3886 Some options herein operate on a specific device
3887 or a set of devices (<pci_dev>). These are
3888 specified in one of the following formats:
3890 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
3891 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
3893 Note: the first format specifies a PCI
3894 bus/device/function address which may change
3895 if new hardware is inserted, if motherboard
3896 firmware changes, or due to changes caused
3897 by other kernel parameters. If the
3898 domain is left unspecified, it is
3899 taken to be zero. Optionally, a path
3900 to a device through multiple device/function
3901 addresses can be specified after the base
3902 address (this is more robust against
3903 renumbering issues). The second format
3904 selects devices using IDs from the
3905 configuration space which may match multiple
3906 devices in the system.
3908 earlydump dump PCI config space before the kernel
3910 off [X86] don't probe for the PCI bus
3911 bios [X86-32] force use of PCI BIOS, don't access
3912 the hardware directly. Use this if your machine
3913 has a non-standard PCI host bridge.
3914 nobios [X86-32] disallow use of PCI BIOS, only direct
3915 hardware access methods are allowed. Use this
3916 if you experience crashes upon bootup and you
3917 suspect they are caused by the BIOS.
3918 conf1 [X86] Force use of PCI Configuration Access
3919 Mechanism 1 (config address in IO port 0xCF8,
3920 data in IO port 0xCFC, both 32-bit).
3921 conf2 [X86] Force use of PCI Configuration Access
3922 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
3923 the function, IO port 0xCFA, also 8-bit, sets
3924 bus number. The config space is then accessed
3925 through ports 0xC000-0xCFFF).
3926 See http://wiki.osdev.org/PCI for more info
3927 on the configuration access mechanisms.
3928 noaer [PCIE] If the PCIEAER kernel config parameter is
3929 enabled, this kernel boot option can be used to
3930 disable the use of PCIE advanced error reporting.
3931 nodomains [PCI] Disable support for multiple PCI
3932 root domains (aka PCI segments, in ACPI-speak).
3933 nommconf [X86] Disable use of MMCONFIG for PCI
3935 check_enable_amd_mmconf [X86] check for and enable
3936 properly configured MMIO access to PCI
3937 config space on AMD family 10h CPU
3938 nomsi [MSI] If the PCI_MSI kernel config parameter is
3939 enabled, this kernel boot option can be used to
3940 disable the use of MSI interrupts system-wide.
3941 noioapicquirk [APIC] Disable all boot interrupt quirks.
3942 Safety option to keep boot IRQs enabled. This
3943 should never be necessary.
3944 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
3945 primary IO-APIC for bridges that cannot disable
3946 boot IRQs. This fixes a source of spurious IRQs
3947 when the system masks IRQs.
3948 noioapicreroute [APIC] Disable workaround that uses the
3949 boot IRQ equivalent of an IRQ that connects to
3950 a chipset where boot IRQs cannot be disabled.
3951 The opposite of ioapicreroute.
3952 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
3953 routing table. These calls are known to be buggy
3954 on several machines and they hang the machine
3955 when used, but on other computers it's the only
3956 way to get the interrupt routing table. Try
3957 this option if the kernel is unable to allocate
3958 IRQs or discover secondary PCI buses on your
3960 rom [X86] Assign address space to expansion ROMs.
3961 Use with caution as certain devices share
3962 address decoders between ROMs and other
3964 norom [X86] Do not assign address space to
3965 expansion ROMs that do not already have
3966 BIOS assigned address ranges.
3967 nobar [X86] Do not assign address space to the
3968 BARs that weren't assigned by the BIOS.
3969 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
3970 assigned automatically to PCI devices. You can
3971 make the kernel exclude IRQs of your ISA cards
3973 pirqaddr=0xAAAAA [X86] Specify the physical address
3974 of the PIRQ table (normally generated
3975 by the BIOS) if it is outside the
3976 F0000h-100000h range.
3977 lastbus=N [X86] Scan all buses thru bus #N. Can be
3978 useful if the kernel is unable to find your
3979 secondary buses and you want to tell it
3980 explicitly which ones they are.
3981 assign-busses [X86] Always assign all PCI bus
3982 numbers ourselves, overriding
3983 whatever the firmware may have done.
3984 usepirqmask [X86] Honor the possible IRQ mask stored
3985 in the BIOS $PIR table. This is needed on
3986 some systems with broken BIOSes, notably
3987 some HP Pavilion N5400 and Omnibook XE3
3988 notebooks. This will have no effect if ACPI
3989 IRQ routing is enabled.
3990 noacpi [X86] Do not use ACPI for IRQ routing
3991 or for PCI scanning.
3992 use_crs [X86] Use PCI host bridge window information
3993 from ACPI. On BIOSes from 2008 or later, this
3994 is enabled by default. If you need to use this,
3995 please report a bug.
3996 nocrs [X86] Ignore PCI host bridge windows from ACPI.
3997 If you need to use this, please report a bug.
3998 routeirq Do IRQ routing for all PCI devices.
3999 This is normally done in pci_enable_device(),
4000 so this option is a temporary workaround
4001 for broken drivers that don't call it.
4002 skip_isa_align [X86] do not align io start addr, so can
4003 handle more pci cards
4004 noearly [X86] Don't do any early type 1 scanning.
4005 This might help on some broken boards which
4006 machine check when some devices' config space
4007 is read. But various workarounds are disabled
4008 and some IOMMU drivers will not work.
4009 bfsort Sort PCI devices into breadth-first order.
4010 This sorting is done to get a device
4011 order compatible with older (<= 2.4) kernels.
4012 nobfsort Don't sort PCI devices into breadth-first order.
4013 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
4014 tuning and use the BIOS-configured MPS defaults.
4015 pcie_bus_safe Set every device's MPS to the largest value
4016 supported by all devices below the root complex.
4017 pcie_bus_perf Set device MPS to the largest allowable MPS
4018 based on its parent bus. Also set MRRS (Max
4019 Read Request Size) to the largest supported
4020 value (no larger than the MPS that the device
4021 or bus can support) for best performance.
4022 pcie_bus_peer2peer Set every device's MPS to 128B, which
4023 every device is guaranteed to support. This
4024 configuration allows peer-to-peer DMA between
4025 any pair of devices, possibly at the cost of
4026 reduced performance. This also guarantees
4027 that hot-added devices will work.
4028 cbiosize=nn[KMG] The fixed amount of bus space which is
4029 reserved for the CardBus bridge's IO window.
4030 The default value is 256 bytes.
4031 cbmemsize=nn[KMG] The fixed amount of bus space which is
4032 reserved for the CardBus bridge's memory
4033 window. The default value is 64 megabytes.
4036 [<order of align>@]<pci_dev>[; ...]
4037 Specifies alignment and device to reassign
4038 aligned memory resources. How to
4039 specify the device is described above.
4040 If <order of align> is not specified,
4041 PAGE_SIZE is used as alignment.
4042 A PCI-PCI bridge can be specified if resource
4043 windows need to be expanded.
4044 To specify the alignment for several
4045 instances of a device, the PCI vendor,
4046 device, subvendor, and subdevice may be
4047 specified, e.g., 12@pci:8086:9c22:103c:198f
4048 for 4096-byte alignment.
4049 ecrc= Enable/disable PCIe ECRC (transaction layer
4050 end-to-end CRC checking).
4051 bios: Use BIOS/firmware settings. This is the
4055 hpiosize=nn[KMG] The fixed amount of bus space which is
4056 reserved for hotplug bridge's IO window.
4057 Default size is 256 bytes.
4058 hpmmiosize=nn[KMG] The fixed amount of bus space which is
4059 reserved for hotplug bridge's MMIO window.
4060 Default size is 2 megabytes.
4061 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
4062 reserved for hotplug bridge's MMIO_PREF window.
4063 Default size is 2 megabytes.
4064 hpmemsize=nn[KMG] The fixed amount of bus space which is
4065 reserved for hotplug bridge's MMIO and
4067 Default size is 2 megabytes.
4068 hpbussize=nn The minimum amount of additional bus numbers
4069 reserved for buses below a hotplug bridge.
4071 realloc= Enable/disable reallocating PCI bridge resources
4072 if allocations done by BIOS are too small to
4073 accommodate resources required by all child
4075 off: Turn realloc off
4077 realloc same as realloc=on
4078 noari do not use PCIe ARI.
4079 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
4080 do not use PCIe ATS (and IOMMU device IOTLB).
4081 pcie_scan_all Scan all possible PCIe devices. Otherwise we
4082 only look for one device below a PCIe downstream
4084 big_root_window Try to add a big 64bit memory window to the PCIe
4085 root complex on AMD CPUs. Some GFX hardware
4086 can resize a BAR to allow access to all VRAM.
4087 Adding the window is slightly risky (it may
4088 conflict with unreported devices), so this
4090 disable_acs_redir=<pci_dev>[; ...]
4091 Specify one or more PCI devices (in the format
4092 specified above) separated by semicolons.
4093 Each device specified will have the PCI ACS
4094 redirect capabilities forced off which will
4095 allow P2P traffic between devices through
4096 bridges without forcing it upstream. Note:
4097 this removes isolation between devices and
4098 may put more devices in an IOMMU group.
4099 force_floating [S390] Force usage of floating interrupts.
4100 nomio [S390] Do not use MIO instructions.
4101 norid [S390] ignore the RID field and force use of
4102 one PCI domain per PCI function
4104 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
4107 force Enable ASPM even on devices that claim not to support it.
4108 WARNING: Forcing ASPM on may cause system lockups.
4110 pcie_ports= [PCIE] PCIe port services handling:
4111 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
4112 even if the platform doesn't give the OS permission to
4113 use them. This may cause conflicts if the platform
4114 also tries to use these services.
4115 dpc-native Use native PCIe service for DPC only. May
4116 cause conflicts if firmware uses AER or DPC.
4117 compat Disable native PCIe services (PME, AER, DPC, PCIe
4120 pcie_port_pm= [PCIE] PCIe port power management handling:
4121 off Disable power management of all PCIe ports
4122 force Forcibly enable power management of all PCIe ports
4124 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
4125 nomsi Do not use MSI for native PCIe PME signaling (this makes
4126 all PCIe root ports use INTx for all services).
4128 pcmv= [HW,PCMCIA] BadgePAD 4
4132 Keep all power-domains already enabled by bootloader on,
4133 even if no driver has claimed them. This is useful
4134 for debug and development, but should not be
4135 needed on a platform with proper driver support.
4138 See Documentation/admin-guide/blockdev/paride.rst.
4140 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
4143 See arch/parisc/kernel/pdc_chassis.c
4145 percpu_alloc= Select which percpu first chunk allocator to use.
4146 Currently supported values are "embed" and "page".
4147 Archs may support subset or none of the selections.
4148 See comments in mm/percpu.c for details on each
4149 allocator. This parameter is primarily for debugging
4150 and performance comparison.
4153 See Documentation/admin-guide/blockdev/paride.rst.
4156 See Documentation/admin-guide/blockdev/paride.rst.
4158 pirq= [SMP,APIC] Manual mp-table setup
4159 See Documentation/x86/i386/IO-APIC.rst.
4161 plip= [PPT,NET] Parallel port network link
4162 Format: { parport<nr> | timid | 0 }
4163 See also Documentation/admin-guide/parport.rst.
4165 pmtmr= [X86] Manual setup of pmtmr I/O Port.
4166 Override pmtimer IOPort with a hex value.
4169 pmu_override= [PPC] Override the PMU.
4170 This option takes over the PMU facility, so it is no
4171 longer usable by perf. Setting this option starts the
4172 PMU counters by setting MMCR0 to 0 (the FC bit is
4173 cleared). If a number is given, then MMCR1 is set to
4174 that number, otherwise (e.g., 'pmu_override=on'), MMCR1
4177 pm_debug_messages [SUSPEND,KNL]
4178 Enable suspend/resume debug messages during boot up.
4181 Enable PNP debug messages (depends on the
4182 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
4183 via /sys/module/pnp/parameters/debug. We always show
4184 current resource usage; turning this on also shows
4185 possible settings and some assignment information.
4191 { on | off | curr | res | no-curr | no-res }
4194 [ISAPNP] Exclude IRQs for the autoconfiguration
4197 [ISAPNP] Exclude DMAs for the autoconfiguration
4199 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
4200 Ranges are in pairs (I/O port base and size).
4203 [ISAPNP] Exclude memory regions for the
4205 Ranges are in pairs (memory base and size).
4207 ports= [IP_VS_FTP] IPVS ftp helper module
4209 Up to 8 (IP_VS_APP_MAX_PORTS) ports
4211 Format: <port>,<port>....
4213 powersave=off [PPC] This option disables power saving features.
4214 It specifically disables cpuidle and sets the
4215 platform machine description specific power_save
4216 function to NULL. On Idle the CPU just reduces
4219 ppc_strict_facility_enable
4220 [PPC] This option catches any kernel floating point,
4221 Altivec, VSX and SPE outside of regions specifically
4222 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
4223 There is some performance impact when enabling this.
4227 Disable Hardware Transactional Memory
4230 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC
4231 none - Limited to cond_resched() calls
4232 voluntary - Limited to cond_resched() and might_sleep() calls
4233 full - Any section that isn't explicitly preempt disabled
4234 can be preempted anytime.
4236 print-fatal-signals=
4237 [KNL] debug: print fatal signals
4239 If enabled, warn about various signal handling
4240 related application anomalies: too many signals,
4241 too many POSIX.1 timers, fatal signals causing a
4244 If you hit the warning due to signal overflow,
4245 you might want to try "ulimit -i unlimited".
4249 printk.always_kmsg_dump=
4250 Trigger kmsg_dump for cases other than kernel oops or
4252 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4255 printk.console_no_auto_verbose=
4256 Disable console loglevel raise on oops, panic
4257 or lockdep-detected issues (only if lock debug is on).
4258 With an exception to setups with low baudrate on
4259 serial console, keeping this 0 is a good choice
4260 in order to provide more debug information.
4262 default: 0 (auto_verbose is enabled)
4264 printk.devkmsg={on,off,ratelimit}
4265 Control writing to /dev/kmsg.
4266 on - unlimited logging to /dev/kmsg from userspace
4267 off - logging to /dev/kmsg disabled
4268 ratelimit - ratelimit the logging
4271 printk.time= Show timing data prefixed to each printk message line
4272 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4274 processor.max_cstate= [HW,ACPI]
4275 Limit processor to maximum C-state
4276 max_cstate=9 overrides any DMI blacklist limit.
4278 processor.nocst [HW,ACPI]
4279 Ignore the _CST method to determine C-states,
4280 instead using the legacy FADT method
4282 profile= [KNL] Enable kernel profiling via /proc/profile
4283 Format: [<profiletype>,]<number>
4284 Param: <profiletype>: "schedule", "sleep", or "kvm"
4285 [defaults to kernel profiling]
4286 Param: "schedule" - profile schedule points.
4287 Param: "sleep" - profile D-state sleeping (millisecs).
4288 Requires CONFIG_SCHEDSTATS
4289 Param: "kvm" - profile VM exits.
4290 Param: <number> - step/bucket size as a power of 2 for
4291 statistical time based profiling.
4293 prompt_ramdisk= [RAM] [Deprecated]
4295 prot_virt= [S390] enable hosting protected virtual machines
4296 isolated from the hypervisor (if hardware supports
4300 psi= [KNL] Enable or disable pressure stall information
4304 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
4305 probe for; one of (bare|imps|exps|lifebook|any).
4306 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
4308 psmouse.resetafter= [HW,MOUSE]
4309 Try to reset the device after so many bad packets
4312 [HW,MOUSE] Set desired mouse resolution, in dpi.
4313 psmouse.smartscroll=
4314 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
4315 0 = disabled, 1 = enabled (default).
4317 pstore.backend= Specify the name of the pstore backend to use
4320 See Documentation/admin-guide/blockdev/paride.rst.
4322 pti= [X86-64] Control Page Table Isolation of user and
4323 kernel address spaces. Disabling this feature
4324 removes hardening, but improves performance of
4325 system calls and interrupts.
4327 on - unconditionally enable
4328 off - unconditionally disable
4329 auto - kernel detects whether your CPU model is
4330 vulnerable to issues that PTI mitigates
4332 Not specifying this option is equivalent to pti=auto.
4335 Equivalent to pti=off
4338 [KNL] Number of legacy pty's. Overwrites compiled-in
4341 quiet [KNL] Disable most log messages
4346 See Documentation/admin-guide/md.rst.
4348 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
4349 See Documentation/admin-guide/blockdev/ramdisk.rst.
4351 ramdisk_start= [RAM] RAM disk image start address
4353 random.trust_cpu={on,off}
4354 [KNL] Enable or disable trusting the use of the
4355 CPU's random number generator (if available) to
4356 fully seed the kernel's CRNG. Default is controlled
4357 by CONFIG_RANDOM_TRUST_CPU.
4359 randomize_kstack_offset=
4360 [KNL] Enable or disable kernel stack offset
4361 randomization, which provides roughly 5 bits of
4362 entropy, frustrating memory corruption attacks
4363 that depend on stack address determinism or
4364 cross-syscall address exposures. This is only
4365 available on architectures that have defined
4366 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET.
4367 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4368 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT.
4370 ras=option[,option,...] [KNL] RAS-specific options
4373 Disable the Correctable Errors Collector,
4374 see CONFIG_RAS_CEC help text.
4376 rcu_nocbs[=cpu-list]
4377 [KNL] The optional argument is a cpu list,
4380 In kernels built with CONFIG_RCU_NOCB_CPU=y,
4381 enable the no-callback CPU mode, which prevents
4382 such CPUs' callbacks from being invoked in
4383 softirq context. Invocation of such CPUs' RCU
4384 callbacks will instead be offloaded to "rcuox/N"
4385 kthreads created for that purpose, where "x" is
4386 "p" for RCU-preempt, "s" for RCU-sched, and "g"
4387 for the kthreads that mediate grace periods; and
4388 "N" is the CPU number. This reduces OS jitter on
4389 the offloaded CPUs, which can be useful for HPC
4390 and real-time workloads. It can also improve
4391 energy efficiency for asymmetric multiprocessors.
4393 If a cpulist is passed as an argument, the specified
4394 list of CPUs is set to no-callback mode from boot.
4396 Otherwise, if the '=' sign and the cpulist
4397 arguments are omitted, no CPU will be set to
4398 no-callback mode from boot but the mode may be
4399 toggled at runtime via cpusets.
4402 Rather than requiring that offloaded CPUs
4403 (specified by rcu_nocbs= above) explicitly
4404 awaken the corresponding "rcuoN" kthreads,
4405 make these kthreads poll for callbacks.
4406 This improves the real-time response for the
4407 offloaded CPUs by relieving them of the need to
4408 wake up the corresponding kthread, but degrades
4409 energy efficiency by requiring that the kthreads
4410 periodically wake up to do the polling.
4412 rcutree.blimit= [KNL]
4413 Set maximum number of finished RCU callbacks to
4414 process in one batch.
4416 rcutree.dump_tree= [KNL]
4417 Dump the structure of the rcu_node combining tree
4418 out at early boot. This is used for diagnostic
4419 purposes, to verify correct tree setup.
4421 rcutree.gp_cleanup_delay= [KNL]
4422 Set the number of jiffies to delay each step of
4423 RCU grace-period cleanup.
4425 rcutree.gp_init_delay= [KNL]
4426 Set the number of jiffies to delay each step of
4427 RCU grace-period initialization.
4429 rcutree.gp_preinit_delay= [KNL]
4430 Set the number of jiffies to delay each step of
4431 RCU grace-period pre-initialization, that is,
4432 the propagation of recent CPU-hotplug changes up
4433 the rcu_node combining tree.
4435 rcutree.use_softirq= [KNL]
4436 If set to zero, move all RCU_SOFTIRQ processing to
4437 per-CPU rcuc kthreads. Defaults to a non-zero
4438 value, meaning that RCU_SOFTIRQ is used by default.
4439 Specify rcutree.use_softirq=0 to use rcuc kthreads.
4441 But note that CONFIG_PREEMPT_RT=y kernels disable
4442 this kernel boot parameter, forcibly setting it
4445 rcutree.rcu_fanout_exact= [KNL]
4446 Disable autobalancing of the rcu_node combining
4447 tree. This is used by rcutorture, and might
4448 possibly be useful for architectures having high
4449 cache-to-cache transfer latencies.
4451 rcutree.rcu_fanout_leaf= [KNL]
4452 Change the number of CPUs assigned to each
4453 leaf rcu_node structure. Useful for very
4454 large systems, which will choose the value 64,
4455 and for NUMA systems with large remote-access
4456 latencies, which will choose a value aligned
4457 with the appropriate hardware boundaries.
4459 rcutree.rcu_min_cached_objs= [KNL]
4460 Minimum number of objects which are cached and
4461 maintained per one CPU. Object size is equal
4462 to PAGE_SIZE. The cache allows to reduce the
4463 pressure to page allocator, also it makes the
4464 whole algorithm to behave better in low memory
4467 rcutree.rcu_delay_page_cache_fill_msec= [KNL]
4468 Set the page-cache refill delay (in milliseconds)
4469 in response to low-memory conditions. The range
4470 of permitted values is in the range 0:100000.
4472 rcutree.jiffies_till_first_fqs= [KNL]
4473 Set delay from grace-period initialization to
4474 first attempt to force quiescent states.
4475 Units are jiffies, minimum value is zero,
4476 and maximum value is HZ.
4478 rcutree.jiffies_till_next_fqs= [KNL]
4479 Set delay between subsequent attempts to force
4480 quiescent states. Units are jiffies, minimum
4481 value is one, and maximum value is HZ.
4483 rcutree.jiffies_till_sched_qs= [KNL]
4484 Set required age in jiffies for a
4485 given grace period before RCU starts
4486 soliciting quiescent-state help from
4487 rcu_note_context_switch() and cond_resched().
4488 If not specified, the kernel will calculate
4489 a value based on the most recent settings
4490 of rcutree.jiffies_till_first_fqs
4491 and rcutree.jiffies_till_next_fqs.
4492 This calculated value may be viewed in
4493 rcutree.jiffies_to_sched_qs. Any attempt to set
4494 rcutree.jiffies_to_sched_qs will be cheerfully
4497 rcutree.kthread_prio= [KNL,BOOT]
4498 Set the SCHED_FIFO priority of the RCU per-CPU
4499 kthreads (rcuc/N). This value is also used for
4500 the priority of the RCU boost threads (rcub/N)
4501 and for the RCU grace-period kthreads (rcu_bh,
4502 rcu_preempt, and rcu_sched). If RCU_BOOST is
4503 set, valid values are 1-99 and the default is 1
4504 (the least-favored priority). Otherwise, when
4505 RCU_BOOST is not set, valid values are 0-99 and
4506 the default is zero (non-realtime operation).
4508 rcutree.rcu_nocb_gp_stride= [KNL]
4509 Set the number of NOCB callback kthreads in
4510 each group, which defaults to the square root
4511 of the number of CPUs. Larger numbers reduce
4512 the wakeup overhead on the global grace-period
4513 kthread, but increases that same overhead on
4514 each group's NOCB grace-period kthread.
4516 rcutree.qhimark= [KNL]
4517 Set threshold of queued RCU callbacks beyond which
4518 batch limiting is disabled.
4520 rcutree.qlowmark= [KNL]
4521 Set threshold of queued RCU callbacks below which
4522 batch limiting is re-enabled.
4524 rcutree.qovld= [KNL]
4525 Set threshold of queued RCU callbacks beyond which
4526 RCU's force-quiescent-state scan will aggressively
4527 enlist help from cond_resched() and sched IPIs to
4528 help CPUs more quickly reach quiescent states.
4529 Set to less than zero to make this be set based
4530 on rcutree.qhimark at boot time and to zero to
4531 disable more aggressive help enlistment.
4533 rcutree.rcu_kick_kthreads= [KNL]
4534 Cause the grace-period kthread to get an extra
4535 wake_up() if it sleeps three times longer than
4536 it should at force-quiescent-state time.
4537 This wake_up() will be accompanied by a
4538 WARN_ONCE() splat and an ftrace_dump().
4540 rcutree.rcu_unlock_delay= [KNL]
4541 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels,
4542 this specifies an rcu_read_unlock()-time delay
4543 in microseconds. This defaults to zero.
4544 Larger delays increase the probability of
4545 catching RCU pointer leaks, that is, buggy use
4546 of RCU-protected pointers after the relevant
4547 rcu_read_unlock() has completed.
4549 rcutree.sysrq_rcu= [KNL]
4550 Commandeer a sysrq key to dump out Tree RCU's
4551 rcu_node tree with an eye towards determining
4552 why a new grace period has not yet started.
4554 rcuscale.gp_async= [KNL]
4555 Measure performance of asynchronous
4556 grace-period primitives such as call_rcu().
4558 rcuscale.gp_async_max= [KNL]
4559 Specify the maximum number of outstanding
4560 callbacks per writer thread. When a writer
4561 thread exceeds this limit, it invokes the
4562 corresponding flavor of rcu_barrier() to allow
4563 previously posted callbacks to drain.
4565 rcuscale.gp_exp= [KNL]
4566 Measure performance of expedited synchronous
4567 grace-period primitives.
4569 rcuscale.holdoff= [KNL]
4570 Set test-start holdoff period. The purpose of
4571 this parameter is to delay the start of the
4572 test until boot completes in order to avoid
4575 rcuscale.kfree_rcu_test= [KNL]
4576 Set to measure performance of kfree_rcu() flooding.
4578 rcuscale.kfree_rcu_test_double= [KNL]
4579 Test the double-argument variant of kfree_rcu().
4580 If this parameter has the same value as
4581 rcuscale.kfree_rcu_test_single, both the single-
4582 and double-argument variants are tested.
4584 rcuscale.kfree_rcu_test_single= [KNL]
4585 Test the single-argument variant of kfree_rcu().
4586 If this parameter has the same value as
4587 rcuscale.kfree_rcu_test_double, both the single-
4588 and double-argument variants are tested.
4590 rcuscale.kfree_nthreads= [KNL]
4591 The number of threads running loops of kfree_rcu().
4593 rcuscale.kfree_alloc_num= [KNL]
4594 Number of allocations and frees done in an iteration.
4596 rcuscale.kfree_loops= [KNL]
4597 Number of loops doing rcuscale.kfree_alloc_num number
4598 of allocations and frees.
4600 rcuscale.nreaders= [KNL]
4601 Set number of RCU readers. The value -1 selects
4602 N, where N is the number of CPUs. A value
4603 "n" less than -1 selects N-n+1, where N is again
4604 the number of CPUs. For example, -2 selects N
4605 (the number of CPUs), -3 selects N+1, and so on.
4606 A value of "n" less than or equal to -N selects
4609 rcuscale.nwriters= [KNL]
4610 Set number of RCU writers. The values operate
4611 the same as for rcuscale.nreaders.
4612 N, where N is the number of CPUs
4614 rcuscale.perf_type= [KNL]
4615 Specify the RCU implementation to test.
4617 rcuscale.shutdown= [KNL]
4618 Shut the system down after performance tests
4619 complete. This is useful for hands-off automated
4622 rcuscale.verbose= [KNL]
4623 Enable additional printk() statements.
4625 rcuscale.writer_holdoff= [KNL]
4626 Write-side holdoff between grace periods,
4627 in microseconds. The default of zero says
4630 rcutorture.fqs_duration= [KNL]
4631 Set duration of force_quiescent_state bursts
4634 rcutorture.fqs_holdoff= [KNL]
4635 Set holdoff time within force_quiescent_state bursts
4638 rcutorture.fqs_stutter= [KNL]
4639 Set wait time between force_quiescent_state bursts
4642 rcutorture.fwd_progress= [KNL]
4643 Specifies the number of kthreads to be used
4644 for RCU grace-period forward-progress testing
4645 for the types of RCU supporting this notion.
4646 Defaults to 1 kthread, values less than zero or
4647 greater than the number of CPUs cause the number
4650 rcutorture.fwd_progress_div= [KNL]
4651 Specify the fraction of a CPU-stall-warning
4652 period to do tight-loop forward-progress testing.
4654 rcutorture.fwd_progress_holdoff= [KNL]
4655 Number of seconds to wait between successive
4656 forward-progress tests.
4658 rcutorture.fwd_progress_need_resched= [KNL]
4659 Enclose cond_resched() calls within checks for
4660 need_resched() during tight-loop forward-progress
4663 rcutorture.gp_cond= [KNL]
4664 Use conditional/asynchronous update-side
4665 primitives, if available.
4667 rcutorture.gp_exp= [KNL]
4668 Use expedited update-side primitives, if available.
4670 rcutorture.gp_normal= [KNL]
4671 Use normal (non-expedited) asynchronous
4672 update-side primitives, if available.
4674 rcutorture.gp_sync= [KNL]
4675 Use normal (non-expedited) synchronous
4676 update-side primitives, if available. If all
4677 of rcutorture.gp_cond=, rcutorture.gp_exp=,
4678 rcutorture.gp_normal=, and rcutorture.gp_sync=
4679 are zero, rcutorture acts as if is interpreted
4680 they are all non-zero.
4682 rcutorture.irqreader= [KNL]
4683 Run RCU readers from irq handlers, or, more
4684 accurately, from a timer handler. Not all RCU
4685 flavors take kindly to this sort of thing.
4687 rcutorture.leakpointer= [KNL]
4688 Leak an RCU-protected pointer out of the reader.
4689 This can of course result in splats, and is
4690 intended to test the ability of things like
4691 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect
4694 rcutorture.n_barrier_cbs= [KNL]
4695 Set callbacks/threads for rcu_barrier() testing.
4697 rcutorture.nfakewriters= [KNL]
4698 Set number of concurrent RCU writers. These just
4699 stress RCU, they don't participate in the actual
4700 test, hence the "fake".
4702 rcutorture.nocbs_nthreads= [KNL]
4703 Set number of RCU callback-offload togglers.
4704 Zero (the default) disables toggling.
4706 rcutorture.nocbs_toggle= [KNL]
4707 Set the delay in milliseconds between successive
4708 callback-offload toggling attempts.
4710 rcutorture.nreaders= [KNL]
4711 Set number of RCU readers. The value -1 selects
4712 N-1, where N is the number of CPUs. A value
4713 "n" less than -1 selects N-n-2, where N is again
4714 the number of CPUs. For example, -2 selects N
4715 (the number of CPUs), -3 selects N+1, and so on.
4717 rcutorture.object_debug= [KNL]
4718 Enable debug-object double-call_rcu() testing.
4720 rcutorture.onoff_holdoff= [KNL]
4721 Set time (s) after boot for CPU-hotplug testing.
4723 rcutorture.onoff_interval= [KNL]
4724 Set time (jiffies) between CPU-hotplug operations,
4725 or zero to disable CPU-hotplug testing.
4727 rcutorture.read_exit= [KNL]
4728 Set the number of read-then-exit kthreads used
4729 to test the interaction of RCU updaters and
4730 task-exit processing.
4732 rcutorture.read_exit_burst= [KNL]
4733 The number of times in a given read-then-exit
4734 episode that a set of read-then-exit kthreads
4737 rcutorture.read_exit_delay= [KNL]
4738 The delay, in seconds, between successive
4739 read-then-exit testing episodes.
4741 rcutorture.shuffle_interval= [KNL]
4742 Set task-shuffle interval (s). Shuffling tasks
4743 allows some CPUs to go into dyntick-idle mode
4744 during the rcutorture test.
4746 rcutorture.shutdown_secs= [KNL]
4747 Set time (s) after boot system shutdown. This
4748 is useful for hands-off automated testing.
4750 rcutorture.stall_cpu= [KNL]
4751 Duration of CPU stall (s) to test RCU CPU stall
4752 warnings, zero to disable.
4754 rcutorture.stall_cpu_block= [KNL]
4755 Sleep while stalling if set. This will result
4756 in warnings from preemptible RCU in addition
4757 to any other stall-related activity.
4759 rcutorture.stall_cpu_holdoff= [KNL]
4760 Time to wait (s) after boot before inducing stall.
4762 rcutorture.stall_cpu_irqsoff= [KNL]
4763 Disable interrupts while stalling if set.
4765 rcutorture.stall_gp_kthread= [KNL]
4766 Duration (s) of forced sleep within RCU
4767 grace-period kthread to test RCU CPU stall
4768 warnings, zero to disable. If both stall_cpu
4769 and stall_gp_kthread are specified, the
4770 kthread is starved first, then the CPU.
4772 rcutorture.stat_interval= [KNL]
4773 Time (s) between statistics printk()s.
4775 rcutorture.stutter= [KNL]
4776 Time (s) to stutter testing, for example, specifying
4777 five seconds causes the test to run for five seconds,
4778 wait for five seconds, and so on. This tests RCU's
4779 ability to transition abruptly to and from idle.
4781 rcutorture.test_boost= [KNL]
4782 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
4783 "Maybe" means test if the RCU implementation
4784 under test support RCU priority boosting.
4786 rcutorture.test_boost_duration= [KNL]
4787 Duration (s) of each individual boost test.
4789 rcutorture.test_boost_interval= [KNL]
4790 Interval (s) between each boost test.
4792 rcutorture.test_no_idle_hz= [KNL]
4793 Test RCU's dyntick-idle handling. See also the
4794 rcutorture.shuffle_interval parameter.
4796 rcutorture.torture_type= [KNL]
4797 Specify the RCU implementation to test.
4799 rcutorture.verbose= [KNL]
4800 Enable additional printk() statements.
4802 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
4803 Dump ftrace buffer after reporting RCU CPU
4806 rcupdate.rcu_cpu_stall_suppress= [KNL]
4807 Suppress RCU CPU stall warning messages.
4809 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL]
4810 Suppress RCU CPU stall warning messages and
4811 rcutorture writer stall warnings that occur
4812 during early boot, that is, during the time
4813 before the init task is spawned.
4815 rcupdate.rcu_cpu_stall_timeout= [KNL]
4816 Set timeout for RCU CPU stall warning messages.
4818 rcupdate.rcu_expedited= [KNL]
4819 Use expedited grace-period primitives, for
4820 example, synchronize_rcu_expedited() instead
4821 of synchronize_rcu(). This reduces latency,
4822 but can increase CPU utilization, degrade
4823 real-time latency, and degrade energy efficiency.
4824 No effect on CONFIG_TINY_RCU kernels.
4826 rcupdate.rcu_normal= [KNL]
4827 Use only normal grace-period primitives,
4828 for example, synchronize_rcu() instead of
4829 synchronize_rcu_expedited(). This improves
4830 real-time latency, CPU utilization, and
4831 energy efficiency, but can expose users to
4832 increased grace-period latency. This parameter
4833 overrides rcupdate.rcu_expedited. No effect on
4834 CONFIG_TINY_RCU kernels.
4836 rcupdate.rcu_normal_after_boot= [KNL]
4837 Once boot has completed (that is, after
4838 rcu_end_inkernel_boot() has been invoked), use
4839 only normal grace-period primitives. No effect
4840 on CONFIG_TINY_RCU kernels.
4842 But note that CONFIG_PREEMPT_RT=y kernels enables
4843 this kernel boot parameter, forcibly setting
4844 it to the value one, that is, converting any
4845 post-boot attempt at an expedited RCU grace
4846 period to instead use normal non-expedited
4847 grace-period processing.
4849 rcupdate.rcu_task_collapse_lim= [KNL]
4850 Set the maximum number of callbacks present
4851 at the beginning of a grace period that allows
4852 the RCU Tasks flavors to collapse back to using
4853 a single callback queue. This switching only
4854 occurs when rcupdate.rcu_task_enqueue_lim is
4855 set to the default value of -1.
4857 rcupdate.rcu_task_contend_lim= [KNL]
4858 Set the minimum number of callback-queuing-time
4859 lock-contention events per jiffy required to
4860 cause the RCU Tasks flavors to switch to per-CPU
4861 callback queuing. This switching only occurs
4862 when rcupdate.rcu_task_enqueue_lim is set to
4863 the default value of -1.
4865 rcupdate.rcu_task_enqueue_lim= [KNL]
4866 Set the number of callback queues to use for the
4867 RCU Tasks family of RCU flavors. The default
4868 of -1 allows this to be automatically (and
4869 dynamically) adjusted. This parameter is intended
4872 rcupdate.rcu_task_ipi_delay= [KNL]
4873 Set time in jiffies during which RCU tasks will
4874 avoid sending IPIs, starting with the beginning
4875 of a given grace period. Setting a large
4876 number avoids disturbing real-time workloads,
4877 but lengthens grace periods.
4879 rcupdate.rcu_task_stall_timeout= [KNL]
4880 Set timeout in jiffies for RCU task stall warning
4881 messages. Disable with a value less than or equal
4884 rcupdate.rcu_self_test= [KNL]
4885 Run the RCU early boot self tests
4889 Run specified binary instead of /init from the ramdisk,
4890 used for early userspace startup. See initrd.
4893 force - Override the decision by the kernel to hide the
4894 advertisement of RDRAND support (this affects
4895 certain AMD processors because of buggy BIOS
4896 support, specifically around the suspend/resume
4900 Turn on/off individual RDT features. List is:
4901 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
4903 E.g. to turn on cmt and turn off mba use:
4907 Format (x86 or x86_64):
4908 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] | d[efault] \
4910 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
4912 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
4913 (prefix with 'panic_' to set mode for panic
4915 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
4916 reboot_force is either force or not specified,
4917 reboot_cpu is s[mp]#### with #### being the processor
4918 to be used for rebooting.
4920 refscale.holdoff= [KNL]
4921 Set test-start holdoff period. The purpose of
4922 this parameter is to delay the start of the
4923 test until boot completes in order to avoid
4926 refscale.loops= [KNL]
4927 Set the number of loops over the synchronization
4928 primitive under test. Increasing this number
4929 reduces noise due to loop start/end overhead,
4930 but the default has already reduced the per-pass
4931 noise to a handful of picoseconds on ca. 2020
4934 refscale.nreaders= [KNL]
4935 Set number of readers. The default value of -1
4936 selects N, where N is roughly 75% of the number
4937 of CPUs. A value of zero is an interesting choice.
4939 refscale.nruns= [KNL]
4940 Set number of runs, each of which is dumped onto
4943 refscale.readdelay= [KNL]
4944 Set the read-side critical-section duration,
4945 measured in microseconds.
4947 refscale.scale_type= [KNL]
4948 Specify the read-protection implementation to test.
4950 refscale.shutdown= [KNL]
4951 Shut down the system at the end of the performance
4952 test. This defaults to 1 (shut it down) when
4953 refscale is built into the kernel and to 0 (leave
4954 it running) when refscale is built as a module.
4956 refscale.verbose= [KNL]
4957 Enable additional printk() statements.
4959 refscale.verbose_batched= [KNL]
4960 Batch the additional printk() statements. If zero
4961 (the default) or negative, print everything. Otherwise,
4962 print every Nth verbose statement, where N is the value
4966 [KNL, SMP] Set scheduler's default relax_domain_level.
4967 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
4969 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
4970 Format: <base1>,<size1>[,<base2>,<size2>,...]
4971 Reserve I/O ports or memory so the kernel won't use
4972 them. If <base> is less than 0x10000, the region
4973 is assumed to be I/O ports; otherwise it is memory.
4975 reservetop= [X86-32]
4977 Reserves a hole at the top of the kernel virtual
4980 reset_devices [KNL] Force drivers to reset the underlying device
4981 during initialization.
4984 Specify the partition device for software suspend
4986 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
4988 resume_offset= [SWSUSP]
4989 Specify the offset from the beginning of the partition
4990 given by "resume=" at which the swap header is located,
4991 in <PAGE_SIZE> units (needed only for swap files).
4992 See Documentation/power/swsusp-and-swap-files.rst
4994 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
4995 read the resume files
4997 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
4998 Useful for devices that are detected asynchronously
4999 (e.g. USB and MMC devices).
5001 hibernate= [HIBERNATION]
5002 noresume Don't check if there's a hibernation image
5003 present during boot.
5004 nocompress Don't compress/decompress hibernation images.
5005 no Disable hibernation and resume.
5006 protect_image Turn on image protection during restoration
5007 (that will set all pages holding image data
5008 during restoration read-only).
5010 retain_initrd [RAM] Keep initrd memory after extraction
5012 rfkill.default_state=
5013 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
5014 etc. communication is blocked by default.
5017 rfkill.master_switch_mode=
5018 0 The "airplane mode" button does nothing.
5019 1 The "airplane mode" button toggles between everything
5020 blocked and the previous configuration.
5021 2 The "airplane mode" button toggles between everything
5022 blocked and everything unblocked.
5024 rhash_entries= [KNL,NET]
5025 Set number of hash buckets for route cache
5028 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
5031 ro [KNL] Mount root device read-only on boot
5034 on Mark read-only kernel memory as read-only (default).
5035 off Leave read-only kernel memory writable for debugging.
5038 Enable the uart passthrough on the designated usb port
5039 on Rockchip SoCs. When active, the signals of the
5040 debug-uart get routed to the D+ and D- pins of the usb
5041 port and the regular usb controller gets disabled.
5043 root= [KNL] Root filesystem
5044 See name_to_dev_t comment in init/do_mounts.c.
5046 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
5047 mount the root filesystem
5049 rootflags= [KNL] Set root filesystem mount option string
5051 rootfstype= [KNL] Set root filesystem type
5053 rootwait [KNL] Wait (indefinitely) for root device to show up.
5054 Useful for devices that are detected asynchronously
5055 (e.g. USB and MMC devices).
5057 rproc_mem=nn[KMG][@address]
5058 [KNL,ARM,CMA] Remoteproc physical memory block.
5059 Memory area to be used by remote processor image,
5062 rw [KNL] Mount root device read-write on boot
5064 S [KNL] Run init in single mode
5066 s390_iommu= [HW,S390]
5067 Set s390 IOTLB flushing mode
5069 With strict flushing every unmap operation will result in
5070 an IOTLB flush. Default is lazy flushing before reuse,
5073 s390_iommu_aperture= [KNL,S390]
5074 Specifies the size of the per device DMA address space
5075 accessible through the DMA and IOMMU APIs as a decimal
5076 factor of the size of main memory.
5077 The default is 1 meaning that one can concurrently use
5078 as many DMA addresses as physical memory is installed,
5079 if supported by hardware, and thus map all of memory
5080 once. With a value of 2 one can map all of memory twice
5081 and so on. As a special case a factor of 0 imposes no
5082 restrictions other than those given by hardware at the
5083 cost of significant additional memory use for tables.
5086 See drivers/net/irda/sa1100_ir.c.
5088 sched_verbose [KNL] Enables verbose scheduler debug messages.
5090 schedstats= [KNL,X86] Enable or disable scheduled statistics.
5091 Allowed values are enable and disable. This feature
5092 incurs a small amount of overhead in the scheduler
5093 but is useful for debugging and performance tuning.
5095 sched_thermal_decay_shift=
5096 [KNL, SMP] Set a decay shift for scheduler thermal
5097 pressure signal. Thermal pressure signal follows the
5098 default decay period of other scheduler pelt
5099 signals(usually 32 ms but configurable). Setting
5100 sched_thermal_decay_shift will left shift the decay
5101 period for the thermal pressure signal by the shift
5103 i.e. with the default pelt decay period of 32 ms
5104 sched_thermal_decay_shift thermal pressure decay pr
5108 Format: integer between 0 and 10
5111 scftorture.holdoff= [KNL]
5112 Number of seconds to hold off before starting
5113 test. Defaults to zero for module insertion and
5114 to 10 seconds for built-in smp_call_function()
5117 scftorture.longwait= [KNL]
5118 Request ridiculously long waits randomly selected
5119 up to the chosen limit in seconds. Zero (the
5120 default) disables this feature. Please note
5121 that requesting even small non-zero numbers of
5122 seconds can result in RCU CPU stall warnings,
5123 softlockup complaints, and so on.
5125 scftorture.nthreads= [KNL]
5126 Number of kthreads to spawn to invoke the
5127 smp_call_function() family of functions.
5128 The default of -1 specifies a number of kthreads
5129 equal to the number of CPUs.
5131 scftorture.onoff_holdoff= [KNL]
5132 Number seconds to wait after the start of the
5133 test before initiating CPU-hotplug operations.
5135 scftorture.onoff_interval= [KNL]
5136 Number seconds to wait between successive
5137 CPU-hotplug operations. Specifying zero (which
5138 is the default) disables CPU-hotplug operations.
5140 scftorture.shutdown_secs= [KNL]
5141 The number of seconds following the start of the
5142 test after which to shut down the system. The
5143 default of zero avoids shutting down the system.
5144 Non-zero values are useful for automated tests.
5146 scftorture.stat_interval= [KNL]
5147 The number of seconds between outputting the
5148 current test statistics to the console. A value
5149 of zero disables statistics output.
5151 scftorture.stutter_cpus= [KNL]
5152 The number of jiffies to wait between each change
5153 to the set of CPUs under test.
5155 scftorture.use_cpus_read_lock= [KNL]
5156 Use use_cpus_read_lock() instead of the default
5157 preempt_disable() to disable CPU hotplug
5158 while invoking one of the smp_call_function*()
5161 scftorture.verbose= [KNL]
5162 Enable additional printk() statements.
5164 scftorture.weight_single= [KNL]
5165 The probability weighting to use for the
5166 smp_call_function_single() function with a zero
5167 "wait" parameter. A value of -1 selects the
5168 default if all other weights are -1. However,
5169 if at least one weight has some other value, a
5170 value of -1 will instead select a weight of zero.
5172 scftorture.weight_single_wait= [KNL]
5173 The probability weighting to use for the
5174 smp_call_function_single() function with a
5175 non-zero "wait" parameter. See weight_single.
5177 scftorture.weight_many= [KNL]
5178 The probability weighting to use for the
5179 smp_call_function_many() function with a zero
5180 "wait" parameter. See weight_single.
5181 Note well that setting a high probability for
5182 this weighting can place serious IPI load
5185 scftorture.weight_many_wait= [KNL]
5186 The probability weighting to use for the
5187 smp_call_function_many() function with a
5188 non-zero "wait" parameter. See weight_single
5191 scftorture.weight_all= [KNL]
5192 The probability weighting to use for the
5193 smp_call_function_all() function with a zero
5194 "wait" parameter. See weight_single and
5197 scftorture.weight_all_wait= [KNL]
5198 The probability weighting to use for the
5199 smp_call_function_all() function with a
5200 non-zero "wait" parameter. See weight_single
5203 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
5204 xtime_lock contention on larger systems, and/or RCU lock
5205 contention on all systems with CONFIG_MAXSMP set.
5206 Format: { "0" | "1" }
5207 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
5209 Note: increases power consumption, thus should only be
5210 enabled if running jitter sensitive (HPC/RT) workloads.
5212 security= [SECURITY] Choose a legacy "major" security module to
5213 enable at boot. This has been deprecated by the
5216 selinux= [SELINUX] Disable or enable SELinux at boot time.
5217 Format: { "0" | "1" }
5218 See security/selinux/Kconfig help text.
5223 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
5224 Format: { "0" | "1" }
5225 See security/apparmor/Kconfig help text
5228 Default value is set via kernel config option.
5230 serialnumber [BUGS=X86-32]
5233 Maximal number of shapers.
5241 Enable merging of slabs with similar size when the
5242 kernel is built without CONFIG_SLAB_MERGE_DEFAULT.
5245 Disable merging of slabs with similar size. May be
5246 necessary if there is some reason to distinguish
5247 allocs to different slabs, especially in hardened
5248 environments where the risk of heap overflows and
5249 layout control by attackers can usually be
5250 frustrated by disabling merging. This will reduce
5251 most of the exposure of a heap attack to a single
5252 cache (risks via metadata attacks are mostly
5253 unchanged). Debug options disable merging on their
5255 For more information see Documentation/vm/slub.rst.
5257 slab_max_order= [MM, SLAB]
5258 Determines the maximum allowed order for slabs.
5259 A high setting may cause OOMs due to memory
5260 fragmentation. Defaults to 1 for systems with
5261 more than 32MB of RAM, 0 otherwise.
5263 slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB]
5264 Enabling slub_debug allows one to determine the
5265 culprit if slab objects become corrupted. Enabling
5266 slub_debug can create guard zones around objects and
5267 may poison objects when not in use. Also tracks the
5268 last alloc / free. For more information see
5269 Documentation/vm/slub.rst.
5271 slub_max_order= [MM, SLUB]
5272 Determines the maximum allowed order for slabs.
5273 A high setting may cause OOMs due to memory
5274 fragmentation. For more information see
5275 Documentation/vm/slub.rst.
5277 slub_min_objects= [MM, SLUB]
5278 The minimum number of objects per slab. SLUB will
5279 increase the slab order up to slub_max_order to
5280 generate a sufficiently large slab able to contain
5281 the number of objects indicated. The higher the number
5282 of objects the smaller the overhead of tracking slabs
5283 and the less frequently locks need to be acquired.
5284 For more information see Documentation/vm/slub.rst.
5286 slub_min_order= [MM, SLUB]
5287 Determines the minimum page order for slabs. Must be
5288 lower than slub_max_order.
5289 For more information see Documentation/vm/slub.rst.
5291 slub_merge [MM, SLUB]
5292 Same with slab_merge.
5294 slub_nomerge [MM, SLUB]
5295 Same with slab_nomerge. This is supported for legacy.
5296 See slab_nomerge for more information.
5299 Format: <io1>[,<io2>[,...,<io8>]]
5301 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
5302 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
5303 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
5304 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
5305 smsc-ircc2.ircc_irq= [HW] IRQ line
5306 smsc-ircc2.ircc_dma= [HW] DMA channel
5307 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
5308 0: Toshiba Satellite 1800 (GP data pin select)
5309 1: Fast pin select (default)
5312 smt [KNL,S390] Set the maximum number of threads (logical
5313 CPUs) to use per physical CPU on systems capable of
5314 symmetric multithreading (SMT). Will be capped to the
5315 actual hardware limit.
5317 Default: -1 (no limit)
5320 [KNL] Should the soft-lockup detector generate panics.
5323 A value of 1 instructs the soft-lockup detector
5324 to panic the machine when a soft-lockup occurs. It is
5325 also controlled by the kernel.softlockup_panic sysctl
5326 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
5327 respective build-time switch to that functionality.
5329 softlockup_all_cpu_backtrace=
5330 [KNL] Should the soft-lockup detector generate
5331 backtraces on all cpus.
5334 sonypi.*= [HW] Sony Programmable I/O Control Device driver
5335 See Documentation/admin-guide/laptops/sonypi.rst
5337 spectre_v2= [X86] Control mitigation of Spectre variant 2
5338 (indirect branch speculation) vulnerability.
5339 The default operation protects the kernel from
5342 on - unconditionally enable, implies
5344 off - unconditionally disable, implies
5346 auto - kernel detects whether your CPU model is
5349 Selecting 'on' will, and 'auto' may, choose a
5350 mitigation method at run time according to the
5351 CPU, the available microcode, the setting of the
5352 CONFIG_RETPOLINE configuration option, and the
5353 compiler with which the kernel was built.
5355 Selecting 'on' will also enable the mitigation
5356 against user space to user space task attacks.
5358 Selecting 'off' will disable both the kernel and
5359 the user space protections.
5361 Specific mitigations can also be selected manually:
5363 retpoline - replace indirect branches
5364 retpoline,generic - Retpolines
5365 retpoline,lfence - LFENCE; indirect branch
5366 retpoline,amd - alias for retpoline,lfence
5367 eibrs - enhanced IBRS
5368 eibrs,retpoline - enhanced IBRS + Retpolines
5369 eibrs,lfence - enhanced IBRS + LFENCE
5371 Not specifying this option is equivalent to
5375 [X86] Control mitigation of Spectre variant 2
5376 (indirect branch speculation) vulnerability between
5379 on - Unconditionally enable mitigations. Is
5380 enforced by spectre_v2=on
5382 off - Unconditionally disable mitigations. Is
5383 enforced by spectre_v2=off
5385 prctl - Indirect branch speculation is enabled,
5386 but mitigation can be enabled via prctl
5387 per thread. The mitigation control state
5388 is inherited on fork.
5391 - Like "prctl" above, but only STIBP is
5392 controlled per thread. IBPB is issued
5393 always when switching between different user
5397 - Same as "prctl" above, but all seccomp
5398 threads will enable the mitigation unless
5399 they explicitly opt out.
5402 - Like "seccomp" above, but only STIBP is
5403 controlled per thread. IBPB is issued
5404 always when switching between different
5405 user space processes.
5407 auto - Kernel selects the mitigation depending on
5408 the available CPU features and vulnerability.
5410 Default mitigation: "prctl"
5412 Not specifying this option is equivalent to
5413 spectre_v2_user=auto.
5415 spec_store_bypass_disable=
5416 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
5417 (Speculative Store Bypass vulnerability)
5419 Certain CPUs are vulnerable to an exploit against a
5420 a common industry wide performance optimization known
5421 as "Speculative Store Bypass" in which recent stores
5422 to the same memory location may not be observed by
5423 later loads during speculative execution. The idea
5424 is that such stores are unlikely and that they can
5425 be detected prior to instruction retirement at the
5426 end of a particular speculation execution window.
5428 In vulnerable processors, the speculatively forwarded
5429 store can be used in a cache side channel attack, for
5430 example to read memory to which the attacker does not
5431 directly have access (e.g. inside sandboxed code).
5433 This parameter controls whether the Speculative Store
5434 Bypass optimization is used.
5436 On x86 the options are:
5438 on - Unconditionally disable Speculative Store Bypass
5439 off - Unconditionally enable Speculative Store Bypass
5440 auto - Kernel detects whether the CPU model contains an
5441 implementation of Speculative Store Bypass and
5442 picks the most appropriate mitigation. If the
5443 CPU is not vulnerable, "off" is selected. If the
5444 CPU is vulnerable the default mitigation is
5445 architecture and Kconfig dependent. See below.
5446 prctl - Control Speculative Store Bypass per thread
5447 via prctl. Speculative Store Bypass is enabled
5448 for a process by default. The state of the control
5449 is inherited on fork.
5450 seccomp - Same as "prctl" above, but all seccomp threads
5451 will disable SSB unless they explicitly opt out.
5453 Default mitigations:
5456 On powerpc the options are:
5458 on,auto - On Power8 and Power9 insert a store-forwarding
5459 barrier on kernel entry and exit. On Power7
5460 perform a software flush on kernel entry and
5464 Not specifying this option is equivalent to
5465 spec_store_bypass_disable=auto.
5467 spia_io_base= [HW,MTD]
5473 [X86] Enable split lock detection or bus lock detection
5475 When enabled (and if hardware support is present), atomic
5476 instructions that access data across cache line
5477 boundaries will result in an alignment check exception
5478 for split lock detection or a debug exception for
5483 warn - the kernel will emit rate-limited warnings
5484 about applications triggering the #AC
5485 exception or the #DB exception. This mode is
5486 the default on CPUs that support split lock
5487 detection or bus lock detection. Default
5488 behavior is by #AC if both features are
5489 enabled in hardware.
5491 fatal - the kernel will send SIGBUS to applications
5492 that trigger the #AC exception or the #DB
5493 exception. Default behavior is by #AC if
5494 both features are enabled in hardware.
5497 Set system wide rate limit to N bus locks
5498 per second for bus lock detection.
5501 N/A for split lock detection.
5504 If an #AC exception is hit in the kernel or in
5505 firmware (i.e. not while executing in user mode)
5506 the kernel will oops in either "warn" or "fatal"
5509 #DB exception for bus lock is triggered only when
5513 Control the Special Register Buffer Data Sampling
5516 Certain CPUs are vulnerable to an MDS-like
5517 exploit which can leak bits from the random
5520 By default, this issue is mitigated by
5521 microcode. However, the microcode fix can cause
5522 the RDRAND and RDSEED instructions to become
5523 much slower. Among other effects, this will
5524 result in reduced throughput from /dev/urandom.
5526 The microcode mitigation can be disabled with
5527 the following option:
5529 off: Disable mitigation and remove
5530 performance impact to RDRAND and RDSEED
5532 srcutree.counter_wrap_check [KNL]
5533 Specifies how frequently to check for
5534 grace-period sequence counter wrap for the
5535 srcu_data structure's ->srcu_gp_seq_needed field.
5536 The greater the number of bits set in this kernel
5537 parameter, the less frequently counter wrap will
5538 be checked for. Note that the bottom two bits
5541 srcutree.exp_holdoff [KNL]
5542 Specifies how many nanoseconds must elapse
5543 since the end of the last SRCU grace period for
5544 a given srcu_struct until the next normal SRCU
5545 grace period will be considered for automatic
5546 expediting. Set to zero to disable automatic
5550 Speculative Store Bypass Disable control
5552 On CPUs that are vulnerable to the Speculative
5553 Store Bypass vulnerability and offer a
5554 firmware based mitigation, this parameter
5555 indicates how the mitigation should be used:
5557 force-on: Unconditionally enable mitigation for
5558 for both kernel and userspace
5559 force-off: Unconditionally disable mitigation for
5560 for both kernel and userspace
5561 kernel: Always enable mitigation in the
5562 kernel, and offer a prctl interface
5563 to allow userspace to register its
5564 interest in being mitigated too.
5566 stack_guard_gap= [MM]
5567 override the default stack gap protection. The value
5568 is in page units and it defines how many pages prior
5569 to (for stacks growing down) resp. after (for stacks
5570 growing up) the main stack are reserved for no other
5571 mapping. Default value is 256 pages.
5573 stack_depot_disable= [KNL]
5574 Setting this to true through kernel command line will
5575 disable the stack depot thereby saving the static memory
5576 consumed by the stack hash table. By default this is set
5580 Enabled the stack tracer on boot up.
5582 stacktrace_filter=[function-list]
5583 [FTRACE] Limit the functions that the stack tracer
5584 will trace at boot up. function-list is a comma-separated
5585 list of functions. This list can be changed at run
5586 time by the stack_trace_filter file in the debugfs
5587 tracing directory. Note, this enables stack tracing
5588 and the stacktrace above is not needed.
5592 Set the STI (builtin display/keyboard on the HP-PARISC
5593 machines) console (graphic card) which should be used
5594 as the initial boot-console.
5595 See also comment in drivers/video/console/sticore.c.
5598 See comment in drivers/video/console/sticore.c.
5601 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
5606 Enable or disable strict sigaltstack size checks
5607 against the required signal frame size which
5608 depends on the supported FPU features. This can
5609 be used to filter out binaries which have
5610 not yet been made aware of AT_MINSIGSTKSZ.
5612 sunrpc.min_resvport=
5613 sunrpc.max_resvport=
5615 SunRPC servers often require that client requests
5616 originate from a privileged port (i.e. a port in the
5617 range 0 < portnr < 1024).
5618 An administrator who wishes to reserve some of these
5619 ports for other uses may adjust the range that the
5620 kernel's sunrpc client considers to be privileged
5621 using these two parameters to set the minimum and
5622 maximum port values.
5624 sunrpc.svc_rpc_per_connection_limit=
5626 Limit the number of requests that the server will
5627 process in parallel from a single connection.
5628 The default value is 0 (no limit).
5632 Control how the NFS server code allocates CPUs to
5633 service thread pools. Depending on how many NICs
5634 you have and where their interrupts are bound, this
5635 option will affect which CPUs will do NFS serving.
5636 Note: this parameter cannot be changed while the
5637 NFS server is running.
5639 auto the server chooses an appropriate mode
5640 automatically using heuristics
5641 global a single global pool contains all CPUs
5642 percpu one pool for each CPU
5643 pernode one pool for each NUMA node (equivalent
5644 to global on non-NUMA machines)
5646 sunrpc.tcp_slot_table_entries=
5647 sunrpc.udp_slot_table_entries=
5649 Sets the upper limit on the number of simultaneous
5650 RPC calls that can be sent from the client to a
5651 server. Increasing these values may allow you to
5652 improve throughput, but will also increase the
5653 amount of memory reserved for use by the client.
5655 suspend.pm_test_delay=
5657 Sets the number of seconds to remain in a suspend test
5658 mode before resuming the system (see
5659 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
5660 is set. Default value is 5.
5663 Format: { on | off | y | n | 1 | 0 }
5664 This parameter controls use of the Protected
5665 Execution Facility on pSeries.
5668 [KNL] Enable accounting of swap in memory resource
5669 controller if no parameter or 1 is given or disable
5670 it if 0 is given (See Documentation/admin-guide/cgroup-v1/memory.rst)
5672 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
5673 Format: { <int> | force | noforce }
5674 <int> -- Number of I/O TLB slabs
5675 force -- force using of bounce buffers even if they
5676 wouldn't be automatically used by the kernel
5677 noforce -- Never use bounce buffers (for debugging)
5682 Set a sysctl parameter, right before loading the init
5683 process, as if the value was written to the respective
5684 /proc/sys/... file. Both '.' and '/' are recognized as
5685 separators. Unrecognized parameters and invalid values
5686 are reported in the kernel log. Sysctls registered
5687 later by a loaded module cannot be set this way.
5688 Example: sysctl.vm.swappiness=40
5690 sysfs.deprecated=0|1 [KNL]
5691 Enable/disable old style sysfs layout for old udev
5692 on older distributions. When this option is enabled
5693 very new udev will not work anymore. When this option
5694 is disabled (or CONFIG_SYSFS_DEPRECATED not compiled)
5695 in older udev will not work anymore.
5696 Default depends on CONFIG_SYSFS_DEPRECATED_V2 set in
5697 the kernel configuration.
5699 sysrq_always_enabled
5701 Ignore sysrq setting - this boot parameter will
5702 neutralize any effect of /proc/sys/kernel/sysrq.
5703 Useful for debugging.
5705 tcpmhash_entries= [KNL,NET]
5706 Set the number of tcp_metrics_hash slots.
5707 Default value is 8192 or 16384 depending on total
5708 ram pages. This is used to specify the TCP metrics
5709 cache size. See Documentation/networking/ip-sysctl.rst
5710 "tcp_no_metrics_save" section for more details.
5714 test_suspend= [SUSPEND][,N]
5715 Specify "mem" (for Suspend-to-RAM) or "standby" (for
5716 standby suspend) or "freeze" (for suspend type freeze)
5717 as the system sleep state during system startup with
5718 the optional capability to repeat N number of times.
5719 The system is woken from this state using a
5720 wakeup-capable RTC alarm.
5722 thash_entries= [KNL,NET]
5723 Set number of hash buckets for TCP connection
5725 thermal.act= [HW,ACPI]
5726 -1: disable all active trip points in all thermal zones
5727 <degrees C>: override all lowest active trip points
5729 thermal.crt= [HW,ACPI]
5730 -1: disable all critical trip points in all thermal zones
5731 <degrees C>: override all critical trip points
5733 thermal.nocrt= [HW,ACPI]
5734 Set to disable actions on ACPI thermal zone
5735 critical and hot trip points.
5737 thermal.off= [HW,ACPI]
5738 1: disable ACPI thermal control
5740 thermal.psv= [HW,ACPI]
5741 -1: disable all passive trip points
5742 <degrees C>: override all passive trip points to this
5745 thermal.tzp= [HW,ACPI]
5746 Specify global default ACPI thermal zone polling rate
5747 <deci-seconds>: poll all this frequency
5748 0: no polling (default)
5751 Force threading of all interrupt handlers except those
5752 marked explicitly IRQF_NO_THREAD.
5756 Specify if the kernel should make use of the cpu
5757 topology information if the hardware supports this.
5758 The scheduler will make use of this information and
5759 e.g. base its process migration decisions on it.
5762 topology_updates= [KNL, PPC, NUMA]
5764 Specify if the kernel should ignore (off)
5765 topology updates sent by the hypervisor to this
5768 torture.disable_onoff_at_boot= [KNL]
5769 Prevent the CPU-hotplug component of torturing
5770 until after init has spawned.
5772 torture.ftrace_dump_at_shutdown= [KNL]
5773 Dump the ftrace buffer at torture-test shutdown,
5774 even if there were no errors. This can be a
5775 very costly operation when many torture tests
5776 are running concurrently, especially on systems
5777 with rotating-rust storage.
5779 torture.verbose_sleep_frequency= [KNL]
5780 Specifies how many verbose printk()s should be
5781 emitted between each sleep. The default of zero
5782 disables verbose-printk() sleeping.
5784 torture.verbose_sleep_duration= [KNL]
5785 Duration of each verbose-printk() sleep in jiffies.
5789 tpm_suspend_pcr=[HW,TPM]
5790 Format: integer pcr id
5791 Specify that at suspend time, the tpm driver
5792 should extend the specified pcr with zeros,
5793 as a workaround for some chips which fail to
5794 flush the last written pcr on TPM_SaveState.
5795 This will guarantee that all the other pcrs
5798 trace_buf_size=nn[KMG]
5799 [FTRACE] will set tracing buffer size on each cpu.
5801 trace_event=[event-list]
5802 [FTRACE] Set and start specified trace events in order
5803 to facilitate early boot debugging. The event-list is a
5804 comma-separated list of trace events to enable. See
5805 also Documentation/trace/events.rst
5807 trace_options=[option-list]
5808 [FTRACE] Enable or disable tracer options at boot.
5809 The option-list is a comma delimited list of options
5810 that can be enabled or disabled just as if you were
5811 to echo the option name into
5813 /sys/kernel/debug/tracing/trace_options
5815 For example, to enable stacktrace option (to dump the
5816 stack trace of each event), add to the command line:
5818 trace_options=stacktrace
5820 See also Documentation/trace/ftrace.rst "trace options"
5824 Have the tracepoints sent to printk as well as the
5825 tracing ring buffer. This is useful for early boot up
5826 where the system hangs or reboots and does not give the
5827 option for reading the tracing buffer or performing a
5828 ftrace_dump_on_oops.
5830 To turn off having tracepoints sent to printk,
5831 echo 0 > /proc/sys/kernel/tracepoint_printk
5832 Note, echoing 1 into this file without the
5833 tracepoint_printk kernel cmdline option has no effect.
5835 The tp_printk_stop_on_boot (see below) can also be used
5836 to stop the printing of events to console at
5841 Having tracepoints sent to printk() and activating high
5842 frequency tracepoints such as irq or sched, can cause
5843 the system to live lock.
5845 tp_printk_stop_on_boot[FTRACE]
5846 When tp_printk (above) is set, it can cause a lot of noise
5847 on the console. It may be useful to only include the
5848 printing of events during boot up, as user space may
5849 make the system inoperable.
5851 This command line option will stop the printing of events
5852 to console at the late_initcall_sync() time frame.
5855 [FTRACE] enable this option to disable tracing when a
5856 warning is hit. This turns off "tracing_on". Tracing can
5857 be enabled again by echoing '1' into the "tracing_on"
5858 file located in /sys/kernel/debug/tracing/
5860 This option is useful, as it disables the trace before
5861 the WARNING dump is called, which prevents the trace to
5862 be filled with content caused by the warning output.
5864 This option can also be set at run time via the sysctl
5865 option: kernel/traceoff_on_warning
5867 transparent_hugepage=
5869 Format: [always|madvise|never]
5870 Can be used to control the default behavior of the system
5871 with respect to transparent hugepages.
5872 See Documentation/admin-guide/mm/transhuge.rst
5875 trusted.source= [KEYS]
5877 This parameter identifies the trust source as a backend
5878 for trusted keys implementation. Supported trust
5882 If not specified then it defaults to iterating through
5883 the trust source list starting with TPM and assigns the
5884 first trust source as a backend which is initialized
5885 successfully during iteration.
5887 tsc= Disable clocksource stability checks for TSC.
5889 [x86] reliable: mark tsc clocksource as reliable, this
5890 disables clocksource verification at runtime, as well
5891 as the stability checks done at bootup. Used to enable
5892 high-resolution timer mode on older hardware, and in
5893 virtualized environment.
5894 [x86] noirqtime: Do not use TSC to do irq accounting.
5895 Used to run time disable IRQ_TIME_ACCOUNTING on any
5896 platforms where RDTSC is slow and this accounting
5898 [x86] unstable: mark the TSC clocksource as unstable, this
5899 marks the TSC unconditionally unstable at bootup and
5900 avoids any further wobbles once the TSC watchdog notices.
5901 [x86] nowatchdog: disable clocksource watchdog. Used
5902 in situations with strict latency requirements (where
5903 interruptions from clocksource watchdog are not
5906 tsc_early_khz= [X86] Skip early TSC calibration and use the given
5907 value instead. Useful when the early TSC frequency discovery
5908 procedure is not reliable, such as on overclocked systems
5909 with CPUID.16h support and partial CPUID.15h support.
5910 Format: <unsigned int>
5912 tsx= [X86] Control Transactional Synchronization
5913 Extensions (TSX) feature in Intel processors that
5914 support TSX control.
5916 This parameter controls the TSX feature. The options are:
5918 on - Enable TSX on the system. Although there are
5919 mitigations for all known security vulnerabilities,
5920 TSX has been known to be an accelerator for
5921 several previous speculation-related CVEs, and
5922 so there may be unknown security risks associated
5923 with leaving it enabled.
5925 off - Disable TSX on the system. (Note that this
5926 option takes effect only on newer CPUs which are
5927 not vulnerable to MDS, i.e., have
5928 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
5929 the new IA32_TSX_CTRL MSR through a microcode
5930 update. This new MSR allows for the reliable
5931 deactivation of the TSX functionality.)
5933 auto - Disable TSX if X86_BUG_TAA is present,
5934 otherwise enable TSX on the system.
5936 Not specifying this option is equivalent to tsx=off.
5938 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5941 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
5942 Abort (TAA) vulnerability.
5944 Similar to Micro-architectural Data Sampling (MDS)
5945 certain CPUs that support Transactional
5946 Synchronization Extensions (TSX) are vulnerable to an
5947 exploit against CPU internal buffers which can forward
5948 information to a disclosure gadget under certain
5951 In vulnerable processors, the speculatively forwarded
5952 data can be used in a cache side channel attack, to
5953 access data to which the attacker does not have direct
5956 This parameter controls the TAA mitigation. The
5959 full - Enable TAA mitigation on vulnerable CPUs
5962 full,nosmt - Enable TAA mitigation and disable SMT on
5963 vulnerable CPUs. If TSX is disabled, SMT
5964 is not disabled because CPU is not
5965 vulnerable to cross-thread TAA attacks.
5966 off - Unconditionally disable TAA mitigation
5968 On MDS-affected machines, tsx_async_abort=off can be
5969 prevented by an active MDS mitigation as both vulnerabilities
5970 are mitigated with the same mechanism so in order to disable
5971 this mitigation, you need to specify mds=off too.
5973 Not specifying this option is equivalent to
5974 tsx_async_abort=full. On CPUs which are MDS affected
5975 and deploy MDS mitigation, TAA mitigation is not
5976 required and doesn't provide any additional
5980 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5982 turbografx.map[2|3]= [HW,JOY]
5983 TurboGraFX parallel port interface
5985 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
5986 See also Documentation/input/devices/joystick-parport.rst
5988 udbg-immortal [PPC] When debugging early kernel crashes that
5989 happen after console_init() and before a proper
5990 console driver takes over, this boot options might
5991 help "seeing" what's going on.
5993 uhash_entries= [KNL,NET]
5994 Set number of hash buckets for UDP/UDP-Lite connections
5997 [USB] Ignore overcurrent events (default N).
5998 Some badly-designed motherboards generate lots of
5999 bogus events, for ports that aren't wired to
6000 anything. Set this parameter to avoid log spamming.
6001 Note that genuine overcurrent events won't be
6005 [X86] Cause panic on unknown NMI.
6007 usbcore.authorized_default=
6008 [USB] Default USB device authorization:
6009 (default -1 = authorized except for wireless USB,
6010 0 = not authorized, 1 = authorized, 2 = authorized
6011 if device connected to internal port)
6013 usbcore.autosuspend=
6014 [USB] The autosuspend time delay (in seconds) used
6015 for newly-detected USB devices (default 2). This
6016 is the time required before an idle device will be
6017 autosuspended. Devices for which the delay is set
6018 to a negative value won't be autosuspended at all.
6020 usbcore.usbfs_snoop=
6021 [USB] Set to log all usbfs traffic (default 0 = off).
6023 usbcore.usbfs_snoop_max=
6024 [USB] Maximum number of bytes to snoop in each URB
6027 usbcore.blinkenlights=
6028 [USB] Set to cycle leds on hubs (default 0 = off).
6030 usbcore.old_scheme_first=
6031 [USB] Start with the old device initialization
6032 scheme (default 0 = off).
6034 usbcore.usbfs_memory_mb=
6035 [USB] Memory limit (in MB) for buffers allocated by
6036 usbfs (default = 16, 0 = max = 2047).
6038 usbcore.use_both_schemes=
6039 [USB] Try the other device initialization scheme
6040 if the first one fails (default 1 = enabled).
6042 usbcore.initial_descriptor_timeout=
6043 [USB] Specifies timeout for the initial 64-byte
6044 USB_REQ_GET_DESCRIPTOR request in milliseconds
6045 (default 5000 = 5.0 seconds).
6047 usbcore.nousb [USB] Disable the USB subsystem
6050 [USB] A list of quirk entries to augment the built-in
6051 usb core quirk list. List entries are separated by
6052 commas. Each entry has the form
6053 VendorID:ProductID:Flags. The IDs are 4-digit hex
6054 numbers and Flags is a set of letters. Each letter
6055 will change the built-in quirk; setting it if it is
6056 clear and clearing it if it is set. The letters have
6057 the following meanings:
6058 a = USB_QUIRK_STRING_FETCH_255 (string
6059 descriptors must not be fetched using
6061 b = USB_QUIRK_RESET_RESUME (device can't resume
6062 correctly so reset it instead);
6063 c = USB_QUIRK_NO_SET_INTF (device can't handle
6064 Set-Interface requests);
6065 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
6066 handle its Configuration or Interface
6068 e = USB_QUIRK_RESET (device can't be reset
6069 (e.g morph devices), don't use reset);
6070 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
6071 more interface descriptions than the
6072 bNumInterfaces count, and can't handle
6073 talking to these interfaces);
6074 g = USB_QUIRK_DELAY_INIT (device needs a pause
6075 during initialization, after we read
6076 the device descriptor);
6077 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
6078 high speed and super speed interrupt
6079 endpoints, the USB 2.0 and USB 3.0 spec
6080 require the interval in microframes (1
6081 microframe = 125 microseconds) to be
6082 calculated as interval = 2 ^
6084 Devices with this quirk report their
6085 bInterval as the result of this
6086 calculation instead of the exponent
6087 variable used in the calculation);
6088 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
6089 handle device_qualifier descriptor
6091 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
6092 generates spurious wakeup, ignore
6093 remote wakeup capability);
6094 k = USB_QUIRK_NO_LPM (device can't handle Link
6096 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
6097 (Device reports its bInterval as linear
6098 frames instead of the USB 2.0
6100 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
6101 to be disconnected before suspend to
6102 prevent spurious wakeup);
6103 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
6104 pause after every control message);
6105 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
6106 delay after resetting its port);
6107 Example: quirks=0781:5580:bk,0a5c:5834:gij
6110 [USBHID] The interval which mice are to be polled at.
6113 [USBHID] The interval which joysticks are to be polled at.
6116 [USBHID] The interval which keyboards are to be polled at.
6118 usb-storage.delay_use=
6119 [UMS] The delay in seconds before a new device is
6120 scanned for Logical Units (default 1).
6123 [UMS] A list of quirks entries to supplement or
6124 override the built-in unusual_devs list. List
6125 entries are separated by commas. Each entry has
6126 the form VID:PID:Flags where VID and PID are Vendor
6127 and Product ID values (4-digit hex numbers) and
6128 Flags is a set of characters, each corresponding
6129 to a common usb-storage quirk flag as follows:
6130 a = SANE_SENSE (collect more than 18 bytes
6131 of sense data, not on uas);
6132 b = BAD_SENSE (don't collect more than 18
6133 bytes of sense data, not on uas);
6134 c = FIX_CAPACITY (decrease the reported
6135 device capacity by one sector);
6136 d = NO_READ_DISC_INFO (don't use
6137 READ_DISC_INFO command, not on uas);
6138 e = NO_READ_CAPACITY_16 (don't use
6139 READ_CAPACITY_16 command);
6140 f = NO_REPORT_OPCODES (don't use report opcodes
6142 g = MAX_SECTORS_240 (don't transfer more than
6143 240 sectors at a time, uas only);
6144 h = CAPACITY_HEURISTICS (decrease the
6145 reported device capacity by one
6146 sector if the number is odd);
6147 i = IGNORE_DEVICE (don't bind to this
6149 j = NO_REPORT_LUNS (don't use report luns
6151 k = NO_SAME (do not use WRITE_SAME, uas only)
6152 l = NOT_LOCKABLE (don't try to lock and
6153 unlock ejectable media, not on uas);
6154 m = MAX_SECTORS_64 (don't transfer more
6155 than 64 sectors = 32 KB at a time,
6157 n = INITIAL_READ10 (force a retry of the
6158 initial READ(10) command, not on uas);
6159 o = CAPACITY_OK (accept the capacity
6160 reported by the device, not on uas);
6161 p = WRITE_CACHE (the device cache is ON
6162 by default, not on uas);
6163 r = IGNORE_RESIDUE (the device reports
6164 bogus residue values, not on uas);
6165 s = SINGLE_LUN (the device has only one
6167 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
6168 commands, uas only);
6169 u = IGNORE_UAS (don't bind to the uas driver);
6170 w = NO_WP_DETECT (don't test whether the
6171 medium is write-protected).
6172 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
6173 even if the device claims no cache,
6175 Example: quirks=0419:aaf5:rl,0421:0433:rc
6177 user_debug= [KNL,ARM]
6179 See arch/arm/Kconfig.debug help text.
6180 1 - undefined instruction events
6182 4 - invalid data aborts
6185 Example: user_debug=31
6188 [X86] Flags controlling user PTE allocations.
6190 nohigh = do not allocate PTE pages in
6191 HIGHMEM regardless of setting
6195 On X86_32, this is an alias for vdso32=. Otherwise:
6197 vdso=1: enable VDSO (the default)
6198 vdso=0: disable VDSO mapping
6200 vdso32= [X86] Control the 32-bit vDSO
6201 vdso32=1: enable 32-bit VDSO
6202 vdso32=0 or vdso32=2: disable 32-bit VDSO
6204 See the help text for CONFIG_COMPAT_VDSO for more
6205 details. If CONFIG_COMPAT_VDSO is set, the default is
6206 vdso32=0; otherwise, the default is vdso32=1.
6208 For compatibility with older kernels, vdso32=2 is an
6211 Try vdso32=0 if you encounter an error that says:
6212 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
6215 vector=percpu: enable percpu vector domain
6217 video= [FB] Frame buffer configuration
6218 See Documentation/fb/modedb.rst.
6220 video.brightness_switch_enabled= [0,1]
6221 If set to 1, on receiving an ACPI notify event
6222 generated by hotkey, video driver will adjust brightness
6223 level and then send out the event to user space through
6224 the allocated input device; If set to 0, video driver
6225 will only send out the event without touching backlight
6230 [VMMIO] Memory mapped virtio (platform) device.
6232 <size>@<baseaddr>:<irq>[:<id>]
6234 <size> := size (can use standard suffixes
6236 <baseaddr> := physical base address
6237 <irq> := interrupt number (as passed to
6239 <id> := (optional) platform device id
6241 virtio_mmio.device=1K@0x100b0000:48:7
6243 Can be used multiple times for multiple devices.
6245 vga= [BOOT,X86-32] Select a particular video mode
6246 See Documentation/x86/boot.rst and
6247 Documentation/admin-guide/svga.rst.
6248 Use vga=ask for menu.
6249 This is actually a boot loader parameter; the value is
6250 passed to the kernel using a special protocol.
6252 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
6253 May slow down system boot speed, especially when
6254 enabled on systems with a large amount of memory.
6255 All options are enabled by default, and this
6256 interface is meant to allow for selectively
6257 enabling or disabling specific virtual memory
6260 Available options are:
6261 P Enable page structure init time poisoning
6262 - Disable all of the above options
6264 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
6265 size of <nn>. This can be used to increase the
6266 minimum size (128MB on x86). It can also be used to
6267 decrease the size and leave more room for directly
6270 vmcp_cma=nn[MG] [KNL,S390]
6271 Sets the memory size reserved for contiguous memory
6272 allocations for the vmcp device driver.
6274 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
6277 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
6280 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
6284 Controls the behavior of vsyscalls (i.e. calls to
6285 fixed addresses of 0xffffffffff600x00 from legacy
6286 code). Most statically-linked binaries and older
6287 versions of glibc use these calls. Because these
6288 functions are at fixed addresses, they make nice
6289 targets for exploits that can control RIP.
6291 emulate [default] Vsyscalls turn into traps and are
6292 emulated reasonably safely. The vsyscall
6295 xonly Vsyscalls turn into traps and are
6296 emulated reasonably safely. The vsyscall
6297 page is not readable.
6299 none Vsyscalls don't work at all. This makes
6300 them quite hard to use for exploits but
6301 might break your system.
6303 vt.color= [VT] Default text color.
6304 Format: 0xYX, X = foreground, Y = background.
6305 Default: 0x07 = light gray on black.
6307 vt.cur_default= [VT] Default cursor shape.
6308 Format: 0xCCBBAA, where AA, BB, and CC are the same as
6309 the parameters of the <Esc>[?A;B;Cc escape sequence;
6310 see VGA-softcursor.txt. Default: 2 = underline.
6312 vt.default_blu= [VT]
6313 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
6314 Change the default blue palette of the console.
6315 This is a 16-member array composed of values
6318 vt.default_grn= [VT]
6319 Format: <green0>,<green1>,<green2>,...,<green15>
6320 Change the default green palette of the console.
6321 This is a 16-member array composed of values
6324 vt.default_red= [VT]
6325 Format: <red0>,<red1>,<red2>,...,<red15>
6326 Change the default red palette of the console.
6327 This is a 16-member array composed of values
6333 Set system-wide default UTF-8 mode for all tty's.
6334 Default is 1, i.e. UTF-8 mode is enabled for all
6335 newly opened terminals.
6337 vt.global_cursor_default=
6340 Set system-wide default for whether a cursor
6341 is shown on new VTs. Default is -1,
6342 i.e. cursors will be created by default unless
6343 overridden by individual drivers. 0 will hide
6344 cursors, 1 will display them.
6346 vt.italic= [VT] Default color for italic text; 0-15.
6349 vt.underline= [VT] Default color for underlined text; 0-15.
6352 watchdog timers [HW,WDT] For information on watchdog timers,
6353 see Documentation/watchdog/watchdog-parameters.rst
6354 or other driver-specific files in the
6355 Documentation/watchdog/ directory.
6359 Set the hard lockup detector stall duration
6360 threshold in seconds. The soft lockup detector
6361 threshold is set to twice the value. A value of 0
6362 disables both lockup detectors. Default is 10
6365 workqueue.watchdog_thresh=
6366 If CONFIG_WQ_WATCHDOG is configured, workqueue can
6367 warn stall conditions and dump internal state to
6368 help debugging. 0 disables workqueue stall
6369 detection; otherwise, it's the stall threshold
6370 duration in seconds. The default value is 30 and
6371 it can be updated at runtime by writing to the
6372 corresponding sysfs file.
6374 workqueue.disable_numa
6375 By default, all work items queued to unbound
6376 workqueues are affine to the NUMA nodes they're
6377 issued on, which results in better behavior in
6378 general. If NUMA affinity needs to be disabled for
6379 whatever reason, this option can be used. Note
6380 that this also can be controlled per-workqueue for
6381 workqueues visible under /sys/bus/workqueue/.
6383 workqueue.power_efficient
6384 Per-cpu workqueues are generally preferred because
6385 they show better performance thanks to cache
6386 locality; unfortunately, per-cpu workqueues tend to
6387 be more power hungry than unbound workqueues.
6389 Enabling this makes the per-cpu workqueues which
6390 were observed to contribute significantly to power
6391 consumption unbound, leading to measurably lower
6392 power usage at the cost of small performance
6395 The default value of this parameter is determined by
6396 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
6398 workqueue.debug_force_rr_cpu
6399 Workqueue used to implicitly guarantee that work
6400 items queued without explicit CPU specified are put
6401 on the local CPU. This guarantee is no longer true
6402 and while local CPU is still preferred work items
6403 may be put on foreign CPUs. This debug option
6404 forces round-robin CPU selection to flush out
6405 usages which depend on the now broken guarantee.
6406 When enabled, memory and cache locality will be
6409 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
6410 default x2apic cluster mode on platforms
6413 xen_512gb_limit [KNL,X86-64,XEN]
6414 Restricts the kernel running paravirtualized under Xen
6415 to use only up to 512 GB of RAM. The reason to do so is
6416 crash analysis tools and Xen tools for doing domain
6417 save/restore/migration must be enabled to handle larger
6420 xen_emul_unplug= [HW,X86,XEN]
6421 Unplug Xen emulated devices
6422 Format: [unplug0,][unplug1]
6423 ide-disks -- unplug primary master IDE devices
6424 aux-ide-disks -- unplug non-primary-master IDE devices
6425 nics -- unplug network devices
6426 all -- unplug all emulated devices (NICs and IDE disks)
6427 unnecessary -- unplugging emulated devices is
6428 unnecessary even if the host did not respond to
6430 never -- do not unplug even if version check succeeds
6432 xen_legacy_crash [X86,XEN]
6433 Crash from Xen panic notifier, without executing late
6434 panic() code such as dumping handler.
6436 xen_nopvspin [X86,XEN]
6437 Disables the qspinlock slowpath using Xen PV optimizations.
6438 This parameter is obsoleted by "nopvspin" parameter, which
6439 has equivalent effect for XEN platform.
6442 Disables the PV optimizations forcing the HVM guest to
6443 run as generic HVM guest with no PV drivers.
6444 This option is obsoleted by the "nopv" option, which
6445 has equivalent effect for XEN platform.
6447 xen_no_vector_callback
6448 [KNL,X86,XEN] Disable the vector callback for Xen
6449 event channel interrupts.
6451 xen_scrub_pages= [XEN]
6452 Boolean option to control scrubbing pages before giving them back
6453 to Xen, for use by other domains. Can be also changed at runtime
6454 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
6455 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
6457 xen_timer_slop= [X86-64,XEN]
6458 Set the timer slop (in nanoseconds) for the virtual Xen
6459 timers (default is 100000). This adjusts the minimum
6460 delta of virtualized Xen timers, where lower values
6461 improve timer resolution at the expense of processing
6462 more timer interrupts.
6464 xen.balloon_boot_timeout= [XEN]
6465 The time (in seconds) to wait before giving up to boot
6466 in case initial ballooning fails to free enough memory.
6467 Applies only when running as HVM or PVH guest and
6468 started with less memory configured than allowed at
6469 max. Default is 180.
6471 xen.event_eoi_delay= [XEN]
6472 How long to delay EOI handling in case of event
6473 storms (jiffies). Default is 10.
6475 xen.event_loop_timeout= [XEN]
6476 After which time (jiffies) the event handling loop
6477 should start to delay EOI handling. Default is 2.
6479 xen.fifo_events= [XEN]
6480 Boolean parameter to disable using fifo event handling
6481 even if available. Normally fifo event handling is
6482 preferred over the 2-level event handling, as it is
6483 fairer and the number of possible event channels is
6484 much higher. Default is on (use fifo events).
6486 nopv= [X86,XEN,KVM,HYPER_V,VMWARE]
6487 Disables the PV optimizations forcing the guest to run
6488 as generic guest with no PV drivers. Currently support
6489 XEN HVM, KVM, HYPER_V and VMWARE guest.
6491 nopvspin [X86,XEN,KVM]
6492 Disables the qspinlock slow path using PV optimizations
6493 which allow the hypervisor to 'idle' the guest on lock
6496 xirc2ps_cs= [NET,PCMCIA]
6498 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
6501 By default on POWER9 and above, the kernel will
6502 natively use the XIVE interrupt controller. This option
6503 allows the fallback firmware mode to be used:
6505 off Fallback to firmware control of XIVE interrupt
6506 controller on both pseries and powernv
6507 platforms. Only useful on POWER9 and above.
6509 xive.store-eoi=off [PPC]
6510 By default on POWER10 and above, the kernel will use
6511 stores for EOI handling when the XIVE interrupt mode
6512 is active. This option allows the XIVE driver to use
6513 loads instead, as on POWER9.
6515 xhci-hcd.quirks [USB,KNL]
6516 A hex value specifying bitmask with supplemental xhci
6517 host controller quirks. Meaning of each bit can be
6518 consulted in header drivers/usb/host/xhci.h.
6521 Format: { early | on | rw | ro | off }
6522 Controls if xmon debugger is enabled. Default is off.
6523 Passing only "xmon" is equivalent to "xmon=early".
6524 early Call xmon as early as possible on boot; xmon
6525 debugger is called from setup_arch().
6526 on xmon debugger hooks will be installed so xmon
6527 is only called on a kernel crash. Default mode,
6528 i.e. either "ro" or "rw" mode, is controlled
6529 with CONFIG_XMON_DEFAULT_RO_MODE.
6530 rw xmon debugger hooks will be installed so xmon
6531 is called only on a kernel crash, mode is write,
6532 meaning SPR registers, memory and, other data
6533 can be written using xmon commands.
6534 ro same as "rw" option above but SPR registers,
6535 memory, and other data can't be written using
6537 off xmon is disabled.