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_nohwsig,
229 old_ordering, nonvs, sci_force_enable, nobl }
230 See Documentation/power/video.rst for information on
232 s3_beep is for debugging; it makes the PC's speaker beep
233 as soon as the kernel's real-mode entry point is called.
234 s4_nohwsig prevents ACPI hardware signature from being
235 used during resume from hibernation.
236 old_ordering causes the ACPI 1.0 ordering of the _PTS
237 control method, with respect to putting devices into
238 low power states, to be enforced (the ACPI 2.0 ordering
239 of _PTS is used by default).
240 nonvs prevents the kernel from saving/restoring the
241 ACPI NVS memory during suspend/hibernation and resume.
242 sci_force_enable causes the kernel to set SCI_EN directly
243 on resume from S1/S3 (which is against the ACPI spec,
244 but some broken systems don't work without it).
245 nobl causes the internal blacklist of systems known to
246 behave incorrectly in some ways with respect to system
247 suspend and resume to be ignored (use wisely).
249 acpi_use_timer_override [HW,ACPI]
250 Use timer override. For some broken Nvidia NF5 boards
251 that require a timer override, but don't have HPET
253 add_efi_memmap [EFI; X86] Include EFI memory map in
254 kernel's map of available physical RAM.
257 { off | try_unsupported }
258 off: disable AGP support
259 try_unsupported: try to drive unsupported chipsets
260 (may crash computer or cause data corruption)
263 See Documentation/sound/alsa-configuration.rst
266 Allow the default userspace alignment fault handler
267 behaviour to be specified. Bit 0 enables warnings,
268 bit 1 enables fixups, and bit 2 sends a segfault.
270 align_va_addr= [X86-64]
271 Align virtual addresses by clearing slice [14:12] when
272 allocating a VMA at process creation time. This option
273 gives you up to 3% performance improvement on AMD F15h
274 machines (where it is enabled by default) for a
275 CPU-intensive style benchmark, and it can vary highly in
276 a microbenchmark depending on workload and compiler.
278 32: only for 32-bit processes
279 64: only for 64-bit processes
280 on: enable for both 32- and 64-bit processes
281 off: disable for both 32- and 64-bit processes
283 alloc_snapshot [FTRACE]
284 Allocate the ftrace snapshot buffer on boot up when the
285 main buffer is allocated. This is handy if debugging
286 and you need to use tracing_snapshot() on boot up, and
287 do not want to use tracing_snapshot_alloc() as it needs
288 to be done where GFP_KERNEL allocations are allowed.
290 amd_iommu= [HW,X86-64]
291 Pass parameters to the AMD IOMMU driver in the system.
293 fullflush - enable flushing of IO/TLB entries when
294 they are unmapped. Otherwise they are
295 flushed before they will be reused, which
297 off - do not initialize any AMD IOMMU found in
299 force_isolation - Force device isolation for all
300 devices. The IOMMU driver is not
301 allowed anymore to lift isolation
302 requirements as needed. This option
303 does not override iommu=pt
304 force_enable - Force enable the IOMMU on platforms known
305 to be buggy with IOMMU enabled. Use this
308 amd_iommu_dump= [HW,X86-64]
309 Enable AMD IOMMU driver option to dump the ACPI table
310 for AMD IOMMU. With this option enabled, AMD IOMMU
311 driver will print ACPI tables for AMD IOMMU during
312 IOMMU initialization.
314 amd_iommu_intr= [HW,X86-64]
315 Specifies one of the following AMD IOMMU interrupt
317 legacy - Use legacy interrupt remapping mode.
318 vapic - Use virtual APIC mode, which allows IOMMU
319 to inject interrupts directly into guest.
320 This mode requires kvm-amd.avic=1.
321 (Default when IOMMU HW support is present.)
323 amijoy.map= [HW,JOY] Amiga joystick support
324 Map of devices attached to JOY0DAT and JOY1DAT
326 See also Documentation/input/joydev/joystick.rst
328 analog.map= [HW,JOY] Analog joystick and gamepad support
329 Specifies type or capabilities of an analog joystick
330 connected to one of 16 gameports
331 Format: <type1>,<type2>,..<type16>
334 Power management functions (SPARCstation-4/5 + deriv.)
336 Disable APC CPU standby support. SPARCstation-Fox does
337 not play well with APC CPU idle - disable it if you have
338 APC and your system crashes randomly.
340 apic= [APIC,X86] Advanced Programmable Interrupt Controller
341 Change the output verbosity while booting
342 Format: { quiet (default) | verbose | debug }
343 Change the amount of debugging information output
344 when initialising the APIC and IO-APIC components.
345 For X86-32, this can also be used to specify an APIC
347 Format: apic=driver_name
348 Examples: apic=bigsmp
350 apic_extnmi= [APIC,X86] External NMI delivery setting
351 Format: { bsp (default) | all | none }
352 bsp: External NMI is delivered only to CPU 0
353 all: External NMIs are broadcast to all CPUs as a
355 none: External NMI is masked for all CPUs. This is
356 useful so that a dump capture kernel won't be
360 See Documentation/networking/ipv6.rst.
362 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
363 Limit apic dumping. The parameter defines the maximal
364 number of local apics being dumped. Also it is possible
365 to set it to "all" by meaning -- no limit here.
366 Format: { 1 (default) | 2 | ... | all }.
367 The parameter valid if only apic=debug or
368 apic=verbose is specified.
369 Example: apic=debug show_lapic=all
371 apm= [APM] Advanced Power Management
372 See header of arch/x86/kernel/apm_32.c.
374 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
375 Format: <io>,<irq>,<nodeID>
377 arm64.nobti [ARM64] Unconditionally disable Branch Target
378 Identification support
380 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication
385 atarimouse= [HW,MOUSE] Atari Mouse
387 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
388 EzKey and similar keyboards
390 atkbd.reset= [HW] Reset keyboard during initialization
392 atkbd.set= [HW] Select keyboard code set
393 Format: <int> (2 = AT (default), 3 = PS/2)
395 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
398 atkbd.softraw= [HW] Choose between synthetic and real raw mode
399 Format: <bool> (0 = real, 1 = synthetic (default))
401 atkbd.softrepeat= [HW]
402 Use software keyboard repeat
404 audit= [KNL] Enable the audit sub-system
405 Format: { "0" | "1" | "off" | "on" }
406 0 | off - kernel audit is disabled and can not be
407 enabled until the next reboot
408 unset - kernel audit is initialized but disabled and
409 will be fully enabled by the userspace auditd.
410 1 | on - kernel audit is initialized and partially
411 enabled, storing at most audit_backlog_limit
412 messages in RAM until it is fully enabled by the
416 audit_backlog_limit= [KNL] Set the audit queue size limit.
417 Format: <int> (must be >=0)
420 bau= [X86_UV] Enable the BAU on SGI UV. The default
421 behavior is to disable the BAU (i.e. bau=0).
422 Format: { "0" | "1" }
425 unset - Disable the BAU.
427 baycom_epp= [HW,AX25]
430 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
432 See header of drivers/net/hamradio/baycom_par.c.
434 baycom_ser_fdx= [HW,AX25]
435 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
436 Format: <io>,<irq>,<mode>[,<baud>]
437 See header of drivers/net/hamradio/baycom_ser_fdx.c.
439 baycom_ser_hdx= [HW,AX25]
440 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
441 Format: <io>,<irq>,<mode>
442 See header of drivers/net/hamradio/baycom_ser_hdx.c.
444 blkdevparts= Manual partition parsing of block device(s) for
445 embedded devices based on command line input.
446 See Documentation/block/cmdline-partition.rst
448 boot_delay= Milliseconds to delay each printk during boot.
449 Values larger than 10 seconds (10000) are changed to
454 Extended command line options can be added to an initrd
455 and this will cause the kernel to look for it.
457 See Documentation/admin-guide/bootconfig.rst
460 Disable BERT OS support on buggy BIOSes.
462 bgrt_disable [ACPI][X86]
463 Disable BGRT to avoid flickering OEM logo.
465 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
466 bttv.radio= Most important insmod options are available as
468 bttv.pll= See Documentation/admin-guide/media/bttv.rst
471 bulk_remove=off [PPC] This parameter disables the use of the pSeries
472 firmware feature for flushing multiple hpte entries
475 c101= [NET] Moxa C101 synchronous serial card
477 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
478 Sometimes CPU hardware bugs make them report the cache
479 size incorrectly. The kernel will attempt work arounds
480 to fix known problems, but for some CPUs it is not
481 possible to determine what the correct size should be.
482 This option provides an override for these situations.
485 [NET] Specifies amount of time (in seconds) that
486 the kernel should wait for a network carrier. By default
487 it waits 120 seconds.
489 ca_keys= [KEYS] This parameter identifies a specific key(s) on
490 the system trusted keyring to be used for certificate
492 format: { id:<keyid> | builtin }
494 cca= [MIPS] Override the kernel pages' cache coherency
495 algorithm. Accepted values range from 0 to 7
496 inclusive. See arch/mips/include/asm/pgtable-bits.h
497 for platform specific values (SB1, Loongson3 and
500 ccw_timeout_log [S390]
501 See Documentation/s390/common_io.rst for details.
503 cgroup_disable= [KNL] Disable a particular controller or optional feature
504 Format: {name of the controller(s) or feature(s) to disable}
505 The effects of cgroup_disable=foo are:
506 - foo isn't auto-mounted if you mount all cgroups in
508 - foo isn't visible as an individually mountable
510 - if foo is an optional feature then the feature is
511 disabled and corresponding cgroup files are not
513 {Currently only "memory" controller deal with this and
514 cut the overhead, others just disable the usage. So
515 only cgroup_disable=memory is actually worthy}
516 Specifying "pressure" disables per-cgroup pressure
517 stall information accounting feature
519 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
520 Format: { { controller | "all" | "named" }
521 [,{ controller | "all" | "named" }...] }
522 Like cgroup_disable, but only applies to cgroup v1;
523 the blacklisted controllers remain available in cgroup2.
524 "all" blacklists all controllers and "named" disables
525 named mounts. Specifying both "all" and "named" disables
528 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
530 nosocket -- Disable socket memory accounting.
531 nokmem -- Disable kernel memory accounting.
533 checkreqprot [SELINUX] Set initial checkreqprot flag value.
534 Format: { "0" | "1" }
535 See security/selinux/Kconfig help text.
536 0 -- check protection applied by kernel (includes
537 any implied execute protection).
538 1 -- check protection requested by application.
539 Default value is set via a kernel config option.
540 Value can be changed at runtime via
541 /sys/fs/selinux/checkreqprot.
542 Setting checkreqprot to 1 is deprecated.
545 See Documentation/s390/common_io.rst for details.
548 Prevents the clock framework from automatically gating
549 clocks that have not been explicitly enabled by a Linux
550 device driver but are enabled in hardware at reset or
551 by the bootloader/firmware. Note that this does not
552 force such clocks to be always-on nor does it reserve
553 those clocks in any way. This parameter is useful for
554 debug and development, but should not be needed on a
555 platform with proper driver support. For more
556 information, see Documentation/driver-api/clk.rst.
558 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
560 Forces specified clocksource (if available) to be used
561 when calculating gettimeofday(). If specified
562 clocksource is not available, it defaults to PIT.
563 Format: { pit | tsc | cyclone | pmtmr }
565 clocksource= Override the default clocksource
567 Override the default clocksource and use the clocksource
568 with the name specified.
569 Some clocksource names to choose from, depending on
571 [all] jiffies (this is the base, fallback clocksource)
573 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
574 pxa_timer,timer3,32k_counter,timer0_1
575 [X86-32] pit,hpet,tsc;
576 scx200_hrt on Geode; cyclone on IBM x440
584 clocksource.arm_arch_timer.evtstrm=
587 Enable/disable the eventstream feature of the ARM
588 architected timer so that code using WFE-based polling
589 loops can be debugged more effectively on production
592 clocksource.max_cswd_read_retries= [KNL]
593 Number of clocksource_watchdog() retries due to
594 external delays before the clock will be marked
595 unstable. Defaults to three retries, that is,
596 four attempts to read the clock under test.
598 clocksource.verify_n_cpus= [KNL]
599 Limit the number of CPUs checked for clocksources
600 marked with CLOCK_SOURCE_VERIFY_PERCPU that
601 are marked unstable due to excessive skew.
602 A negative value says to check all CPUs, while
603 zero says not to check any. Values larger than
604 nr_cpu_ids are silently truncated to nr_cpu_ids.
605 The actual CPUs are chosen randomly, with
606 no replacement if the same CPU is chosen twice.
608 clocksource-wdtest.holdoff= [KNL]
609 Set the time in seconds that the clocksource
610 watchdog test waits before commencing its tests.
611 Defaults to zero when built as a module and to
612 10 seconds when built into the kernel.
614 clearcpuid=BITNUM[,BITNUM...] [X86]
615 Disable CPUID feature X for the kernel. See
616 arch/x86/include/asm/cpufeatures.h for the valid bit
617 numbers. Note the Linux specific bits are not necessarily
618 stable over kernel options, but the vendor specific
620 Also note that user programs calling CPUID directly
621 or using the feature without checking anything
622 will still see it. This just prevents it from
623 being used by the kernel or shown in /proc/cpuinfo.
624 Also note the kernel might malfunction if you disable
627 cma=nn[MG]@[start[MG][-end[MG]]]
629 Sets the size of kernel global memory area for
630 contiguous memory allocations and optionally the
631 placement constraint by the physical address range of
632 memory allocations. A value of 0 disables CMA
633 altogether. For more information, see
634 kernel/dma/contiguous.c
638 Sets the size of kernel per-numa memory area for
639 contiguous memory allocations. A value of 0 disables
640 per-numa CMA altogether. And If this option is not
641 specificed, the default value is 0.
642 With per-numa CMA enabled, DMA users on node nid will
643 first try to allocate buffer from the pernuma area
644 which is located in node nid, if the allocation fails,
645 they will fallback to the global default memory area.
647 cmo_free_hint= [PPC] Format: { yes | no }
648 Specify whether pages are marked as being inactive
649 when they are freed. This is used in CMO environments
650 to determine OS memory pressure for page stealing by
654 coherent_pool=nn[KMG] [ARM,KNL]
655 Sets the size of memory pool for coherent, atomic dma
656 allocations, by default set to 256K.
658 com20020= [HW,NET] ARCnet - COM20020 chipset
660 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
662 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
666 ARCnet - COM90xx chipset (memory-mapped buffers)
667 Format: <io>[,<irq>[,<memstart>]]
669 condev= [HW,S390] console device
672 console= [KNL] Output console device and options.
674 tty<n> Use the virtual console device <n>.
678 Use the specified serial port. The options are of
679 the form "bbbbpnf", where "bbbb" is the baud rate,
680 "p" is parity ("n", "o", or "e"), "n" is number of
681 bits, and "f" is flow control ("r" for RTS or
682 omit it). Default is "9600n8".
684 See Documentation/admin-guide/serial-console.rst for more
686 Documentation/networking/netconsole.rst for an
689 uart[8250],io,<addr>[,options]
690 uart[8250],mmio,<addr>[,options]
691 uart[8250],mmio16,<addr>[,options]
692 uart[8250],mmio32,<addr>[,options]
693 uart[8250],0x<addr>[,options]
694 Start an early, polled-mode console on the 8250/16550
695 UART at the specified I/O port or MMIO address,
696 switching to the matching ttyS device later.
697 MMIO inter-register address stride is either 8-bit
698 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
699 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
700 to be equivalent to 'mmio'. 'options' are specified in
701 the same format described for ttyS above; if unspecified,
702 the h/w is not re-initialized.
704 hvc<n> Use the hypervisor console device <n>. This is for
705 both Xen and PowerPC hypervisors.
707 If the device connected to the port is not a TTY but a braille
708 device, prepend "brl," before the device type, for instance
710 For now, only VisioBraille is supported.
713 [KNL] Change console messages format
715 By default we print messages on consoles in
716 "[time stamp] text\n" format (time stamp may not be
717 printed, depending on CONFIG_PRINTK_TIME or
718 `printk_time' param).
720 Switch to syslog format: "<%u>[time stamp] text\n"
721 IOW, each message will have a facility and loglevel
722 prefix. The format is similar to one used by syslog()
723 syscall, or to executing "dmesg -S --raw" or to reading
726 consoleblank= [KNL] The console blank (screen saver) timeout in
727 seconds. A value of 0 disables the blank timer.
731 [KNL] Change the default value for
732 /proc/<pid>/coredump_filter.
733 See also Documentation/filesystems/proc.rst.
735 coresight_cpu_debug.enable
738 Enable/disable the CPU sampling based debugging.
739 0: default value, disable debugging
740 1: enable debugging at boot time
742 cpuidle.off=1 [CPU_IDLE]
743 disable the cpuidle sub-system
746 [CPU_IDLE] Name of the cpuidle governor to use.
748 cpufreq.off=1 [CPU_FREQ]
749 disable the cpufreq sub-system
751 cpufreq.default_governor=
752 [CPU_FREQ] Name of the default cpufreq governor or
753 policy to use. This governor must be registered in the
754 kernel before the cpufreq driver probes.
757 [X86] Delay for N microsec between assert and de-assert
758 of APIC INIT to start processors. This delay occurs
759 on every CPU online, such as boot, and resume from suspend.
762 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
764 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
766 crashkernel=size[KMG][@offset[KMG]]
767 [KNL] Using kexec, Linux can switch to a 'crash kernel'
768 upon panic. This parameter reserves the physical
769 memory region [offset, offset + size] for that kernel
770 image. If '@offset' is omitted, then a suitable offset
771 is selected automatically.
772 [KNL, X86-64] Select a region under 4G first, and
773 fall back to reserve region above 4G when '@offset'
774 hasn't been specified.
775 See Documentation/admin-guide/kdump/kdump.rst for further details.
777 crashkernel=range1:size1[,range2:size2,...][@offset]
778 [KNL] Same as above, but depends on the memory
779 in the running system. The syntax of range is
780 start-[end] where start and end are both
781 a memory unit (amount[KMG]). See also
782 Documentation/admin-guide/kdump/kdump.rst for an example.
784 crashkernel=size[KMG],high
785 [KNL, X86-64] range could be above 4G. Allow kernel
786 to allocate physical memory region from top, so could
787 be above 4G if system have more than 4G ram installed.
788 Otherwise memory region will be allocated below 4G, if
790 It will be ignored if crashkernel=X is specified.
791 crashkernel=size[KMG],low
792 [KNL, X86-64] range under 4G. When crashkernel=X,high
793 is passed, kernel could allocate physical memory region
794 above 4G, that cause second kernel crash on system
795 that require some amount of low memory, e.g. swiotlb
796 requires at least 64M+32K low memory, also enough extra
797 low memory is needed to make sure DMA buffers for 32-bit
798 devices won't run out. Kernel would try to allocate at
799 at least 256M below 4G automatically.
800 This one let user to specify own low range under 4G
801 for second kernel instead.
802 0: to disable low allocation.
803 It will be ignored when crashkernel=X,high is not used
804 or memory reserved is below 4G.
807 [KNL] Disable crypto self-tests
812 cs89x0_media= [HW,NET]
813 Format: { rj45 | aui | bnc }
815 csdlock_debug= [KNL] Enable debug add-ons of cross-CPU function call
816 handling. When switched on, additional debug data is
817 printed to the console in case a hanging CPU is
818 detected, and that CPU is pinged again in order to try
819 to resolve the hang situation.
820 0: disable csdlock debugging (default)
821 1: enable basic csdlock debugging (minor impact)
822 ext: enable extended csdlock debugging (more impact,
826 See header of drivers/s390/block/dasd_devmap.c.
828 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
829 (one device per port)
830 Format: <port#>,<type>
831 See also Documentation/input/devices/joystick-parport.rst
833 ddebug_query= [KNL,DYNAMIC_DEBUG] Enable debug messages at early boot
835 Documentation/admin-guide/dynamic-debug-howto.rst for
836 details. Deprecated, see dyndbg.
838 debug [KNL] Enable kernel debugging (events log level).
841 [KNL] Enable printing [hashed] pointers early in the
842 boot sequence. If enabled, we use a weak hash instead
843 of siphash to hash pointers. Use this option if you are
844 seeing instances of '(___ptrval___)') and need to see a
845 value (hashed pointer) instead. Cryptographically
846 insecure, please do not use on production kernels.
849 [KNL] verbose locking self-tests
851 Print debugging info while doing the locking API
853 Bitmask for the various LOCKTYPE_ tests. Defaults to 0
854 (no extra messages), setting it to -1 (all bits set)
855 will print _a_lot_ more information - normally only
856 useful to lockdep developers.
858 debug_objects [KNL] Enable object debugging
861 [KNL] Disable object debugging
863 debug_guardpage_minorder=
864 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
865 parameter allows control of the order of pages that will
866 be intentionally kept free (and hence protected) by the
867 buddy allocator. Bigger value increase the probability
868 of catching random memory corruption, but reduce the
869 amount of memory for normal system use. The maximum
870 possible value is MAX_ORDER/2. Setting this parameter
871 to 1 or 2 should be enough to identify most random
872 memory corruption problems caused by bugs in kernel or
873 driver code when a CPU writes to (or reads from) a
874 random memory location. Note that there exists a class
875 of memory corruptions problems caused by buggy H/W or
876 F/W or by drivers badly programing DMA (basically when
877 memory is written at bus level and the CPU MMU is
878 bypassed) which are not detectable by
879 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
880 tracking down these problems.
883 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
884 enables the feature at boot time. By default, it is
885 disabled and the system will work mostly the same as a
886 kernel built without CONFIG_DEBUG_PAGEALLOC.
887 Note: to get most of debug_pagealloc error reports, it's
888 useful to also enable the page_owner functionality.
889 on: enable the feature
891 debugfs= [KNL] This parameter enables what is exposed to userspace
892 and debugfs internal clients.
893 Format: { on, no-mount, off }
894 on: All functions are enabled.
896 Filesystem is not registered but kernel clients can
897 access APIs and a crashkernel can be used to read
898 its content. There is nothing to mount.
899 off: Filesystem is not registered and clients
900 get a -EPERM as result when trying to register files
901 or directories within debugfs.
902 This is equivalent of the runtime functionality if
903 debugfs was not enabled in the kernel at all.
904 Default value is set in build-time with a kernel configuration.
906 debugpat [X86] Enable PAT debugging
908 decnet.addr= [HW,NET]
909 Format: <area>[,<node>]
910 See also Documentation/networking/decnet.rst.
913 [HW] The size of the default HugeTLB page. This is
914 the size represented by the legacy /proc/ hugepages
915 APIs. In addition, this is the default hugetlb size
916 used for shmget(), mmap() and mounting hugetlbfs
917 filesystems. If not specified, defaults to the
918 architecture's default huge page size. Huge page
919 sizes are architecture dependent. See also
920 Documentation/admin-guide/mm/hugetlbpage.rst.
923 deferred_probe_timeout=
924 [KNL] Debugging option to set a timeout in seconds for
925 deferred probe to give up waiting on dependencies to
926 probe. Only specific dependencies (subsystems or
927 drivers) that have opted in will be ignored. A timeout of 0
928 will timeout at the end of initcalls. This option will also
929 dump out devices still on the deferred probe list after
933 Format: { on | off | def_only | inf_only | always }
934 on: s390 zlib hardware support for compression on
935 level 1 and decompression (default)
936 off: No s390 zlib hardware support
937 def_only: s390 zlib hardware support for deflate
938 only (compression on level 1)
939 inf_only: s390 zlib hardware support for inflate
941 always: Same as 'on' but ignores the selected compression
942 level always using hardware support (used for debugging)
945 Set number of hash buckets for dentry cache.
947 disable_1tb_segments [PPC]
948 Disables the use of 1TB hash page table segments. This
949 causes the kernel to fall back to 256MB segments which
950 can be useful when debugging issues that require an SLB
954 Limits the number of kernel SLB entries, and flushes
955 them frequently to increase the rate of SLB faults
959 See Documentation/networking/ipv6.rst.
962 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
963 hardening is enabled for this boot. Hardened
964 usercopy checking is used to protect the kernel
965 from reading or writing beyond known memory
966 allocation boundaries as a proactive defense
967 against bounds-checking flaws in the kernel's
968 copy_to_user()/copy_from_user() interface.
969 on Perform hardened usercopy checks (default).
970 off Disable hardened usercopy checks.
973 Disable RADIX MMU mode on POWER9
975 radix_hcall_invalidate=on [PPC/PSERIES]
976 Disable RADIX GTSE feature and use hcall for TLB
980 Disable TLBIE instruction. Currently does not work
981 with KVM, with HASH MMU, or with coherent accelerators.
983 disable_cpu_apicid= [X86,APIC,SMP]
985 The number of initial APIC ID for the
986 corresponding CPU to be disabled at boot,
987 mostly used for the kdump 2nd kernel to
988 disable BSP to wake up multiple CPUs without
989 causing system reset or hang due to sending
992 disable_ddw [PPC/PSERIES]
993 Disable Dynamic DMA Window support. Use this
994 to workaround buggy firmware.
997 See Documentation/networking/ipv6.rst.
999 disable_mtrr_cleanup [X86]
1000 The kernel tries to adjust MTRR layout from continuous
1001 to discrete, to make X server driver able to add WB
1002 entry later. This parameter disables that.
1004 disable_mtrr_trim [X86, Intel and AMD only]
1005 By default the kernel will trim any uncacheable
1006 memory out of your available memory pool based on
1007 MTRR settings. This parameter disables that behavior,
1008 possibly causing your machine to run very slowly.
1010 disable_timer_pin_1 [X86]
1011 Disable PIN 1 of APIC timer
1012 Can be useful to work around chipset bugs.
1014 dis_ucode_ldr [X86] Disable the microcode loader.
1016 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
1017 this option disables the debugging code at boot.
1019 dma_debug_entries=<number>
1020 This option allows to tune the number of preallocated
1021 entries for DMA-API debugging code. One entry is
1022 required per DMA-API allocation. Use this if the
1023 DMA-API debugging code disables itself because the
1024 architectural default is too low.
1026 dma_debug_driver=<driver_name>
1027 With this option the DMA-API debugging driver
1028 filter feature can be enabled at boot time. Just
1029 pass the driver to filter for as the parameter.
1030 The filter can be disabled or changed to another
1031 driver later using sysfs.
1033 driver_async_probe= [KNL]
1034 List of driver names to be probed asynchronously.
1035 Format: <driver_name1>,<driver_name2>...
1037 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
1038 Broken monitors, graphic adapters, KVMs and EDIDless
1039 panels may send no or incorrect EDID data sets.
1040 This parameter allows to specify an EDID data sets
1041 in the /lib/firmware directory that are used instead.
1042 Generic built-in EDID data sets are used, if one of
1043 edid/1024x768.bin, edid/1280x1024.bin,
1044 edid/1680x1050.bin, or edid/1920x1080.bin is given
1045 and no file with the same name exists. Details and
1046 instructions how to build your own EDID data are
1047 available in Documentation/admin-guide/edid.rst. An EDID
1048 data set will only be used for a particular connector,
1049 if its name and a colon are prepended to the EDID
1050 name. Each connector may use a unique EDID data
1051 set by separating the files with a comma. An EDID
1052 data set with no connector name will be used for
1053 any connectors not explicitly specified.
1058 Format: {"off" | "known"}
1059 Control how the dt_cpu_ftrs device-tree binding is
1060 used for CPU feature discovery and setup (if it
1062 off: Do not use it, fall back to legacy cpu table.
1063 known: Do not pass through unknown features to guests
1064 or userspace, only those that the kernel is aware of.
1066 dump_apple_properties [X86]
1067 Dump name and content of EFI device properties on
1068 x86 Macs. Useful for driver authors to determine
1069 what data is available or for reverse-engineering.
1071 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
1072 <module>.dyndbg[="val"]
1073 Enable debug messages at boot time. See
1074 Documentation/admin-guide/dynamic-debug-howto.rst
1077 nopku [X86] Disable Memory Protection Keys CPU feature found
1080 <module>.async_probe [KNL]
1081 Enable asynchronous probe on this module.
1083 early_ioremap_debug [KNL]
1084 Enable debug messages in early_ioremap support. This
1085 is useful for tracking down temporary early mappings
1086 which are not unmapped.
1088 earlycon= [KNL] Output early console device and options.
1090 When used with no options, the early console is
1091 determined by stdout-path property in device tree's
1092 chosen node or the ACPI SPCR table if supported by
1095 cdns,<addr>[,options]
1096 Start an early, polled-mode console on a Cadence
1097 (xuartps) serial port at the specified address. Only
1098 supported option is baud rate. If baud rate is not
1099 specified, the serial port must already be setup and
1102 uart[8250],io,<addr>[,options]
1103 uart[8250],mmio,<addr>[,options]
1104 uart[8250],mmio32,<addr>[,options]
1105 uart[8250],mmio32be,<addr>[,options]
1106 uart[8250],0x<addr>[,options]
1107 Start an early, polled-mode console on the 8250/16550
1108 UART at the specified I/O port or MMIO address.
1109 MMIO inter-register address stride is either 8-bit
1110 (mmio) or 32-bit (mmio32 or mmio32be).
1111 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1112 to be equivalent to 'mmio'. 'options' are specified
1113 in the same format described for "console=ttyS<n>"; if
1114 unspecified, the h/w is not initialized.
1118 Start an early, polled-mode console on a pl011 serial
1119 port at the specified address. The pl011 serial port
1120 must already be setup and configured. Options are not
1121 yet supported. If 'mmio32' is specified, then only
1122 the driver will use only 32-bit accessors to read/write
1123 the device registers.
1126 Start an early, polled-mode console on a meson serial
1127 port at the specified address. The serial port must
1128 already be setup and configured. Options are not yet
1132 Start an early, polled-mode console on an msm serial
1133 port at the specified address. The serial port
1134 must already be setup and configured. Options are not
1137 msm_serial_dm,<addr>
1138 Start an early, polled-mode console on an msm serial
1139 dm port at the specified address. The serial port
1140 must already be setup and configured. Options are not
1144 Start an early, polled-mode console on a serial port
1145 of an Actions Semi SoC, such as S500 or S900, at the
1146 specified address. The serial port must already be
1147 setup and configured. Options are not yet supported.
1150 Start an early, polled-mode console on a serial port
1151 of an RDA Micro SoC, such as RDA8810PL, at the
1152 specified address. The serial port must already be
1153 setup and configured. Options are not yet supported.
1156 Use RISC-V SBI (Supervisor Binary Interface) for early
1159 smh Use ARM semihosting calls for early console.
1167 Use early console provided by serial driver available
1168 on Samsung SoCs, requires selecting proper type and
1169 a correct base address of the selected UART port. The
1170 serial port must already be setup and configured.
1171 Options are not yet supported.
1174 Start an early, polled-mode console on a lantiq serial
1175 (lqasc) port at the specified address. The serial port
1176 must already be setup and configured. Options are not
1181 Use early console provided by Freescale LP UART driver
1182 found on Freescale Vybrid and QorIQ LS1021A processors.
1183 A valid base address must be provided, and the serial
1184 port must already be setup and configured.
1188 Start an early, polled-mode, output-only console on the
1189 Freescale i.MX UART at the specified address. The UART
1190 must already be setup and configured.
1193 Start an early, polled-mode console on the
1194 Armada 3700 serial port at the specified
1195 address. The serial port must already be setup
1196 and configured. Options are not yet supported.
1199 Start an early, polled-mode console on a Qualcomm
1200 Generic Interface (GENI) based serial port at the
1201 specified address. The serial port must already be
1202 setup and configured. Options are not yet supported.
1205 Start an early, unaccelerated console on the EFI
1206 memory mapped framebuffer (if available). On cache
1207 coherent non-x86 systems that use system memory for
1208 the framebuffer, pass the 'ram' option so that it is
1209 mapped with the correct attributes.
1212 Use early console provided by Freescale LINFlexD UART
1213 serial driver for NXP S32V234 SoCs. A valid base
1214 address must be provided, and the serial port must
1215 already be setup and configured.
1217 earlyprintk= [X86,SH,ARM,M68k,S390]
1221 earlyprintk=serial[,ttySn[,baudrate]]
1222 earlyprintk=serial[,0x...[,baudrate]]
1223 earlyprintk=ttySn[,baudrate]
1224 earlyprintk=dbgp[debugController#]
1225 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1226 earlyprintk=xdbc[xhciController#]
1228 earlyprintk is useful when the kernel crashes before
1229 the normal console is initialized. It is not enabled by
1230 default because it has some cosmetic problems.
1232 Append ",keep" to not disable it when the real console
1235 Only one of vga, efi, serial, or usb debug port can
1238 Currently only ttyS0 and ttyS1 may be specified by
1239 name. Other I/O ports may be explicitly specified
1240 on some architectures (x86 and arm at least) by
1241 replacing ttySn with an I/O port address, like this:
1242 earlyprintk=serial,0x1008,115200
1243 You can find the port for a given device in
1244 /proc/tty/driver/serial:
1245 2: uart:ST16650V2 port:00001008 irq:18 ...
1247 Interaction with the standard serial driver is not
1250 The VGA and EFI output is eventually overwritten by
1253 The xen output can only be used by Xen PV guests.
1255 The sclp output can only be used on s390.
1257 The optional "force" to "pciserial" enables use of a
1258 PCI device even when its classcode is not of the
1261 edac_report= [HW,EDAC] Control how to report EDAC event
1262 Format: {"on" | "off" | "force"}
1263 on: enable EDAC to report H/W event. May be overridden
1264 by other higher priority error reporting module.
1265 off: disable H/W event reporting through EDAC.
1266 force: enforce the use of EDAC to report H/W event.
1269 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1272 This is designed to be used in conjunction with
1273 the boot argument: earlyprintk=vga
1275 This parameter works in place of the kgdboc parameter
1276 but can only be used if the backing tty is available
1277 very early in the boot process. For early debugging
1278 via a serial port see kgdboc_earlycon instead.
1281 Format: {"off" | "on" | "skip[mbr]"}
1284 Format: { "debug", "disable_early_pci_dma",
1285 "nochunk", "noruntime", "nosoftreserve",
1286 "novamap", "no_disable_early_pci_dma" }
1287 debug: enable misc debug output.
1288 disable_early_pci_dma: disable the busmaster bit on all
1289 PCI bridges while in the EFI boot stub.
1290 nochunk: disable reading files in "chunks" in the EFI
1291 boot stub, as chunking can cause problems with some
1292 firmware implementations.
1293 noruntime : disable EFI runtime services support
1294 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1295 attribute may cause the kernel to reserve the
1296 memory range for a memory mapping driver to
1297 claim. Specify efi=nosoftreserve to disable this
1298 reservation and treat the memory by its base type
1299 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1300 novamap: do not call SetVirtualAddressMap().
1301 no_disable_early_pci_dma: Leave the busmaster bit set
1302 on all PCI bridges while in the EFI boot stub
1304 efi_no_storage_paranoia [EFI; X86]
1305 Using this parameter you can use more than 50% of
1306 your efi variable storage. Use this parameter only if
1307 you are really sure that your UEFI does sane gc and
1308 fulfills the spec otherwise your board may brick.
1310 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1311 Add arbitrary attribute to specific memory range by
1312 updating original EFI memory map.
1313 Region of memory which aa attribute is added to is
1316 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1317 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1318 attribute is added to range 0x100000000-0x180000000 and
1319 0x10a0000000-0x1120000000.
1321 If efi_fake_mem=8G@9G:0x40000 is specified, the
1322 EFI_MEMORY_SP(0x40000) attribute is added to
1323 range 0x240000000-0x43fffffff.
1325 Using this parameter you can do debugging of EFI memmap
1326 related features. For example, you can do debugging of
1327 Address Range Mirroring feature even if your box
1328 doesn't support it, or mark specific memory as
1331 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1332 that is to be dynamically loaded by Linux. If there are
1333 multiple variables with the same name but with different
1334 vendor GUIDs, all of them will be loaded. See
1335 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1338 eisa_irq_edge= [PARISC,HW]
1339 See header of drivers/parisc/eisa.c.
1342 See comment before function elanfreq_setup() in
1343 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1345 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1346 Specifies physical address of start of kernel core
1347 image elf header and optionally the size. Generally
1348 kexec loader will pass this option to capture kernel.
1349 See Documentation/admin-guide/kdump/kdump.rst for details.
1351 enable_mtrr_cleanup [X86]
1352 The kernel tries to adjust MTRR layout from continuous
1353 to discrete, to make X server driver able to add WB
1354 entry later. This parameter enables that.
1356 enable_timer_pin_1 [X86]
1357 Enable PIN 1 of APIC timer
1358 Can be useful to work around chipset bugs
1359 (in particular on some ATI chipsets).
1360 The kernel tries to set a reasonable default.
1362 enforcing [SELINUX] Set initial enforcing status.
1364 See security/selinux/Kconfig help text.
1365 0 -- permissive (log only, no denials).
1366 1 -- enforcing (deny and log).
1368 Value can be changed at runtime via
1369 /sys/fs/selinux/enforce.
1372 Disable Error Record Serialization Table (ERST)
1375 ether= [HW,NET] Ethernet cards parameters
1376 This option is obsoleted by the "netdev=" option, which
1377 has equivalent usage. See its documentation for details.
1381 Permit 'security.evm' to be updated regardless of
1382 current integrity status.
1387 fail_make_request=[KNL]
1388 General fault injection mechanism.
1389 Format: <interval>,<probability>,<space>,<times>
1390 See also Documentation/fault-injection/.
1393 Format: { initns | none }
1394 See Documentation/admin-guide/sysctl/net.rst for
1395 fb_tunnels_only_for_init_ns
1398 See Documentation/admin-guide/blockdev/floppy.rst.
1400 force_pal_cache_flush
1401 [IA-64] Avoid check_sal_cache_flush which may hang on
1402 buggy SAL_CACHE_FLUSH implementations. Using this
1403 parameter will force ia64_sal_cache_flush to call
1404 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1407 Forcefully enable Physical Address Extension (PAE).
1408 Many Pentium M systems disable PAE but may have a
1409 functionally usable PAE implementation.
1410 Warning: use of this parameter will taint the kernel
1411 and may cause unknown problems.
1414 [FTRACE] will set and start the specified tracer
1415 as early as possible in order to facilitate early
1418 ftrace_dump_on_oops[=orig_cpu]
1419 [FTRACE] will dump the trace buffers on oops.
1420 If no parameter is passed, ftrace will dump
1421 buffers of all CPUs, but if you pass orig_cpu, it will
1422 dump only the buffer of the CPU that triggered the
1425 ftrace_filter=[function-list]
1426 [FTRACE] Limit the functions traced by the function
1427 tracer at boot up. function-list is a comma-separated
1428 list of functions. This list can be changed at run
1429 time by the set_ftrace_filter file in the debugfs
1432 ftrace_notrace=[function-list]
1433 [FTRACE] Do not trace the functions specified in
1434 function-list. This list can be changed at run time
1435 by the set_ftrace_notrace file in the debugfs
1438 ftrace_graph_filter=[function-list]
1439 [FTRACE] Limit the top level callers functions traced
1440 by the function graph tracer at boot up.
1441 function-list is a comma-separated list of functions
1442 that can be changed at run time by the
1443 set_graph_function file in the debugfs tracing directory.
1445 ftrace_graph_notrace=[function-list]
1446 [FTRACE] Do not trace from the functions specified in
1447 function-list. This list is a comma-separated list of
1448 functions that can be changed at run time by the
1449 set_graph_notrace file in the debugfs tracing directory.
1451 ftrace_graph_max_depth=<uint>
1452 [FTRACE] Used with the function graph tracer. This is
1453 the max depth it will trace into a function. This value
1454 can be changed at run time by the max_graph_depth file
1455 in the tracefs tracing directory. default: 0 (no limit)
1457 fw_devlink= [KNL] Create device links between consumer and supplier
1458 devices by scanning the firmware to infer the
1459 consumer/supplier relationships. This feature is
1460 especially useful when drivers are loaded as modules as
1461 it ensures proper ordering of tasks like device probing
1462 (suppliers first, then consumers), supplier boot state
1463 clean up (only after all consumers have probed),
1464 suspend/resume & runtime PM (consumers first, then
1466 Format: { off | permissive | on | rpm }
1467 off -- Don't create device links from firmware info.
1468 permissive -- Create device links from firmware info
1469 but use it only for ordering boot state clean
1470 up (sync_state() calls).
1471 on -- Create device links from firmware info and use it
1472 to enforce probe and suspend/resume ordering.
1473 rpm -- Like "on", but also use to order runtime PM.
1475 fw_devlink.strict=<bool>
1476 [KNL] Treat all inferred dependencies as mandatory
1477 dependencies. This only applies for fw_devlink=on|rpm.
1481 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1482 support via parallel port (up to 5 devices per port)
1483 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1484 See also Documentation/input/devices/joystick-parport.rst
1488 gart_fix_e820= [X86-64] disable the fix e820 for K8 GART
1492 gcov_persist= [GCOV] When non-zero (default), profiling data for
1493 kernel modules is saved and remains accessible via
1494 debugfs, even when the module is unloaded/reloaded.
1495 When zero, profiling data is discarded and associated
1496 debugfs files are removed at module unload time.
1498 goldfish [X86] Enable the goldfish android emulator platform.
1499 Don't use this when you are not running on the
1502 gpio-mockup.gpio_mockup_ranges
1503 [HW] Sets the ranges of gpiochip of for this device.
1504 Format: <start1>,<end1>,<start2>,<end2>...
1505 gpio-mockup.gpio_mockup_named_lines
1506 [HW] Let the driver know GPIO lines should be named.
1508 gpt [EFI] Forces disk with valid GPT signature but
1509 invalid Protective MBR to be treated as GPT. If the
1510 primary GPT is corrupted, it enables the backup/alternate
1511 GPT to be used instead.
1513 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1514 the "Enable 0" bit of the configuration register.
1517 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1518 the "Enable 0" bit of the configuration register.
1521 grcan.select= [HW] Select which physical interface to use.
1524 grcan.txsize= [HW] Sets the size of the tx buffer.
1525 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1527 grcan.rxsize= [HW] Sets the size of the rx buffer.
1528 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1531 hardlockup_all_cpu_backtrace=
1532 [KNL] Should the hard-lockup detector generate
1533 backtraces on all cpus.
1536 hashdist= [KNL,NUMA] Large hashes allocated during boot
1537 are distributed across NUMA nodes. Defaults on
1538 for 64-bit NUMA, off otherwise.
1539 Format: 0 | 1 (for off | on)
1541 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1543 hd= [EIDE] (E)IDE hard drive subsystem geometry
1544 Format: <cyl>,<head>,<sect>
1547 Disable Hardware Error Source Table (HEST) support;
1548 corresponding firmware-first mode error processing
1549 logic will be disabled.
1551 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1552 size of <nn>. This works even on boxes that have no
1553 highmem otherwise. This also works to reduce highmem
1554 size on bigger boxes.
1556 highres= [KNL] Enable/disable high resolution timer mode.
1557 Valid parameters: "on", "off"
1562 hpet= [X86-32,HPET] option to control HPET usage
1563 Format: { enable (default) | disable | force |
1565 disable: disable HPET and use PIT instead
1566 force: allow force enabled of undocumented chips (ICH4,
1568 verbose: show contents of HPET registers during setup
1570 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1571 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1573 hugetlb_cma= [HW,CMA] The size of a CMA area used for allocation
1574 of gigantic hugepages.
1577 Reserve a CMA area of given size and allocate gigantic
1578 hugepages using the CMA allocator. If enabled, the
1579 boot-time allocation of gigantic hugepages is skipped.
1581 hugepages= [HW] Number of HugeTLB pages to allocate at boot.
1582 If this follows hugepagesz (below), it specifies
1583 the number of pages of hugepagesz to be allocated.
1584 If this is the first HugeTLB parameter on the command
1585 line, it specifies the number of pages to allocate for
1586 the default huge page size. See also
1587 Documentation/admin-guide/mm/hugetlbpage.rst.
1591 [HW] The size of the HugeTLB pages. This is used in
1592 conjunction with hugepages (above) to allocate huge
1593 pages of a specific size at boot. The pair
1594 hugepagesz=X hugepages=Y can be specified once for
1595 each supported huge page size. Huge page sizes are
1596 architecture dependent. See also
1597 Documentation/admin-guide/mm/hugetlbpage.rst.
1600 hugetlb_free_vmemmap=
1601 [KNL] Reguires CONFIG_HUGETLB_PAGE_FREE_VMEMMAP
1603 Allows heavy hugetlb users to free up some more
1604 memory (6 * PAGE_SIZE for each 2MB hugetlb page).
1605 Format: { on | off (default) }
1607 on: enable the feature
1608 off: disable the feature
1610 Built with CONFIG_HUGETLB_PAGE_FREE_VMEMMAP_DEFAULT_ON=y,
1613 This is not compatible with memory_hotplug.memmap_on_memory.
1614 If both parameters are enabled, hugetlb_free_vmemmap takes
1615 precedence over memory_hotplug.memmap_on_memory.
1618 [KNL] Should the hung task detector generate panics.
1621 A value of 1 instructs the kernel to panic when a
1622 hung task is detected. The default value is controlled
1623 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1624 option. The value selected by this boot parameter can
1625 be changed later by the kernel.hung_task_panic sysctl.
1627 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1628 terminal devices. Valid values: 0..8
1629 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1630 If specified, z/VM IUCV HVC accepts connections
1631 from listed z/VM user IDs only.
1633 hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations
1634 which allow the hypervisor to 'idle' the
1635 guest on lock contention.
1638 Do not unregister boot console at start. This is only
1639 useful for debugging when something happens in the window
1640 between unregistering the boot console and initializing
1643 i2c_bus= [HW] Override the default board specific I2C bus speed
1644 or register an additional I2C bus that is not
1645 registered from board initialization code.
1649 i8042.debug [HW] Toggle i8042 debug mode
1650 i8042.unmask_kbd_data
1651 [HW] Enable printing of interrupt data from the KBD port
1652 (disabled by default, and as a pre-condition
1653 requires that i8042.debug=1 be enabled)
1654 i8042.direct [HW] Put keyboard port into non-translated mode
1655 i8042.dumbkbd [HW] Pretend that controller can only read data from
1656 keyboard and cannot control its state
1657 (Don't attempt to blink the leds)
1658 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1659 i8042.nokbd [HW] Don't check/create keyboard port
1660 i8042.noloop [HW] Disable the AUX Loopback command while probing
1662 i8042.nomux [HW] Don't check presence of an active multiplexing
1664 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1666 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1667 i8042.reset [HW] Reset the controller during init, cleanup and
1668 suspend-to-ram transitions, only during s2r
1669 transitions, or never reset
1670 Format: { 1 | Y | y | 0 | N | n }
1671 1, Y, y: always reset controller
1672 0, N, n: don't ever reset controller
1673 Default: only on s2r transitions on x86; most other
1674 architectures force reset to be always executed
1675 i8042.unlock [HW] Unlock (ignore) the keylock
1676 i8042.kbdreset [HW] Reset device connected to KBD port
1680 i8k.ignore_dmi [HW] Continue probing hardware even if DMI data
1681 indicates that the driver is running on unsupported
1683 i8k.force [HW] Activate i8k driver even if SMM BIOS signature
1684 does not match list of supported models.
1686 [HW] Report power status in /proc/i8k
1687 (disabled by default)
1688 i8k.restricted [HW] Allow controlling fans only if SYS_ADMIN
1691 i915.invert_brightness=
1692 [DRM] Invert the sense of the variable that is used to
1693 set the brightness of the panel backlight. Normally a
1694 brightness value of 0 indicates backlight switched off,
1695 and the maximum of the brightness value sets the backlight
1696 to maximum brightness. If this parameter is set to 0
1697 (default) and the machine requires it, or this parameter
1698 is set to 1, a brightness value of 0 sets the backlight
1699 to maximum brightness, and the maximum of the brightness
1700 value switches the backlight off.
1701 -1 -- never invert brightness
1702 0 -- machine default
1703 1 -- force brightness inversion
1706 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1708 ide-core.nodma= [HW] (E)IDE subsystem
1709 Format: =0.0 to prevent dma on hda, =0.1 hdb =1.0 hdc
1710 .vlb_clock .pci_clock .noflush .nohpa .noprobe .nowerr
1711 .cdrom .chs .ignore_cable are additional options
1712 See Documentation/ide/ide.rst.
1714 ide-generic.probe-mask= [HW] (E)IDE subsystem
1716 Probe mask for legacy ISA IDE ports. Depending on
1717 platform up to 6 ports are supported, enabled by
1718 setting corresponding bits in the mask to 1. The
1719 default value is 0x0, which has a special meaning.
1720 On systems that have PCI, it triggers scanning the
1721 PCI bus for the first and the second port, which
1722 are then probed. On systems without PCI the value
1723 of 0x0 enables probing the two first ports as if it
1726 ide-pci-generic.all-generic-ide [HW] (E)IDE subsystem
1727 Claim all unknown PCI IDE storage controllers.
1730 Format: idle=poll, idle=halt, idle=nomwait
1731 Poll forces a polling idle loop that can slightly
1732 improve the performance of waking up a idle CPU, but
1733 will use a lot of power and make the system run hot.
1735 idle=halt: Halt is forced to be used for CPU idle.
1736 In such case C2/C3 won't be used again.
1737 idle=nomwait: Disable mwait for CPU C-states
1741 Allow force disabling of Shared Virtual Memory (SVA)
1742 support for the idxd driver. By default it is set to
1745 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1746 Format: { strict | legacy | 2008 | relaxed }
1749 Choose which programs will be accepted for execution
1750 based on the IEEE 754 NaN encoding(s) supported by
1751 the FPU and the NaN encoding requested with the value
1752 of an ELF file header flag individually set by each
1753 binary. Hardware implementations are permitted to
1754 support either or both of the legacy and the 2008 NaN
1757 Available settings are as follows:
1758 strict accept binaries that request a NaN encoding
1759 supported by the FPU
1760 legacy only accept legacy-NaN binaries, if supported
1762 2008 only accept 2008-NaN binaries, if supported
1764 relaxed accept any binaries regardless of whether
1765 supported by the FPU
1767 The FPU emulator is always able to support both NaN
1768 encodings, so if no FPU hardware is present or it has
1769 been disabled with 'nofpu', then the settings of
1770 'legacy' and '2008' strap the emulator accordingly,
1771 'relaxed' straps the emulator for both legacy-NaN and
1772 2008-NaN, whereas 'strict' enables legacy-NaN only on
1773 legacy processors and both NaN encodings on MIPS32 or
1776 The setting for ABS.fmt/NEG.fmt instruction execution
1777 mode generally follows that for the NaN encoding,
1778 except where unsupported by hardware.
1780 ignore_loglevel [KNL]
1781 Ignore loglevel setting - this will print /all/
1782 kernel messages to the console. Useful for debugging.
1783 We also add it as printk module parameter, so users
1784 could change it dynamically, usually by
1785 /sys/module/printk/parameters/ignore_loglevel.
1788 Ignore RLIMIT_DATA setting for data mappings,
1789 print warning at first misuse. Can be changed via
1790 /sys/module/kernel/parameters/ignore_rlimit_data.
1792 ihash_entries= [KNL]
1793 Set number of hash buckets for inode cache.
1795 ima_appraise= [IMA] appraise integrity measurements
1796 Format: { "off" | "enforce" | "fix" | "log" }
1799 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
1800 The builtin appraise policy appraises all files
1803 ima_canonical_fmt [IMA]
1804 Use the canonical format for the binary runtime
1805 measurements, instead of host native format.
1808 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1812 The list of supported hash algorithms is defined
1813 in crypto/hash_info.h.
1816 The builtin policies to load during IMA setup.
1817 Format: "tcb | appraise_tcb | secure_boot |
1818 fail_securely | critical_data"
1820 The "tcb" policy measures all programs exec'd, files
1821 mmap'd for exec, and all files opened with the read
1822 mode bit set by either the effective uid (euid=0) or
1825 The "appraise_tcb" policy appraises the integrity of
1826 all files owned by root.
1828 The "secure_boot" policy appraises the integrity
1829 of files (eg. kexec kernel image, kernel modules,
1830 firmware, policy, etc) based on file signatures.
1832 The "fail_securely" policy forces file signature
1833 verification failure also on privileged mounted
1834 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
1837 The "critical_data" policy measures kernel integrity
1840 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
1841 Load a policy which meets the needs of the Trusted
1842 Computing Base. This means IMA will measure all
1843 programs exec'd, files mmap'd for exec, and all files
1844 opened for read by uid=0.
1847 Select one of defined IMA measurements template formats.
1848 Formats: { "ima" | "ima-ng" | "ima-sig" }
1852 [IMA] Define a custom template format.
1853 Format: { "field1|...|fieldN" }
1855 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
1856 Format: <min_file_size>
1857 Set the minimal file size for using asynchronous hash.
1858 If left unspecified, ahash usage is disabled.
1860 ahash performance varies for different data sizes on
1861 different crypto accelerators. This option can be used
1862 to achieve the best performance for a particular HW.
1864 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
1866 Set hashing buffer size. Default: 4k.
1868 ahash performance varies for different chunk sizes on
1869 different crypto accelerators. This option can be used
1870 to achieve best performance for particular HW.
1874 Run specified binary instead of /sbin/init as init
1877 initcall_debug [KNL] Trace initcalls as they are executed. Useful
1878 for working out where the kernel is dying during
1881 initcall_blacklist= [KNL] Do not execute a comma-separated list of
1882 initcall functions. Useful for debugging built-in
1883 modules and initcalls.
1885 initramfs_async= [KNL]
1888 This parameter controls whether the initramfs
1889 image is unpacked asynchronously, concurrently
1890 with devices being probed and
1891 initialized. This should normally just work,
1892 but as a debugging aid, one can get the
1893 historical behaviour of the initramfs
1894 unpacking being completed before device_ and
1897 initrd= [BOOT] Specify the location of the initial ramdisk
1899 initrdmem= [KNL] Specify a physical address and size from which to
1900 load the initrd. If an initrd is compiled in or
1901 specified in the bootparams, it takes priority over this
1903 Format: ss[KMG],nn[KMG]
1906 init_on_alloc= [MM] Fill newly allocated pages and heap objects with
1909 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
1911 init_on_free= [MM] Fill freed pages and heap objects with zeroes.
1913 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
1915 init_pkru= [X86] Specify the default memory protection keys rights
1916 register contents for all processes. 0x55555554 by
1917 default (disallow access to all but pkey 0). Can
1918 override in debugfs after boot.
1920 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
1923 int_pln_enable [X86] Enable power limit notification interrupt
1925 integrity_audit=[IMA]
1926 Format: { "0" | "1" }
1927 0 -- basic integrity auditing messages. (Default)
1928 1 -- additional integrity auditing messages.
1930 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
1932 Enable intel iommu driver.
1934 Disable intel iommu driver.
1935 igfx_off [Default Off]
1936 By default, gfx is mapped as normal device. If a gfx
1937 device has a dedicated DMAR unit, the DMAR unit is
1938 bypassed by not enabling DMAR with this option. In
1939 this case, gfx device will use physical address for
1941 strict [Default Off]
1942 With this option on every unmap_single operation will
1943 result in a hardware IOTLB flush operation as opposed
1944 to batching them for performance.
1945 sp_off [Default Off]
1946 By default, super page will be supported if Intel IOMMU
1947 has the capability. With this option, super page will
1950 By default, scalable mode will be disabled even if the
1951 hardware advertises that it has support for the scalable
1952 mode translation. With this option set, scalable mode
1953 will be used on hardware which claims to support it.
1954 tboot_noforce [Default Off]
1955 Do not force the Intel IOMMU enabled under tboot.
1956 By default, tboot will force Intel IOMMU on, which
1957 could harm performance of some high-throughput
1958 devices like 40GBit network cards, even if identity
1960 Note that using this option lowers the security
1961 provided by tboot because it makes the system
1962 vulnerable to DMA attacks.
1964 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
1965 0 disables intel_idle and fall back on acpi_idle.
1966 1 to 9 specify maximum depth of C-state.
1970 Do not enable intel_pstate as the default
1971 scaling driver for the supported processors
1973 Use intel_pstate as a scaling driver, but configure it
1974 to work with generic cpufreq governors (instead of
1975 enabling its internal governor). This mode cannot be
1976 used along with the hardware-managed P-states (HWP)
1979 Enable intel_pstate on systems that prohibit it by default
1980 in favor of acpi-cpufreq. Forcing the intel_pstate driver
1981 instead of acpi-cpufreq may disable platform features, such
1982 as thermal controls and power capping, that rely on ACPI
1983 P-States information being indicated to OSPM and therefore
1984 should be used with caution. This option does not work with
1985 processors that aren't supported by the intel_pstate driver
1986 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
1988 Do not enable hardware P state control (HWP)
1991 Only load intel_pstate on systems which support
1992 hardware P state control (HWP) if available.
1994 Enforce ACPI _PPC performance limits. If the Fixed ACPI
1995 Description Table, specifies preferred power management
1996 profile as "Enterprise Server" or "Performance Server",
1997 then this feature is turned on by default.
1999 Allow per-logical-CPU P-State performance control limits using
2000 cpufreq sysfs interface
2002 intremap= [X86-64, Intel-IOMMU]
2003 on enable Interrupt Remapping (default)
2004 off disable Interrupt Remapping
2005 nosid disable Source ID checking
2007 BIOS x2APIC opt-out request will be ignored
2008 nopost disable Interrupt Posting
2010 iomem= Disable strict checking of access to MMIO memory
2011 strict regions from userspace.
2026 nobypass [PPC/POWERNV]
2027 Disable IOMMU bypass, using IOMMU for PCI devices.
2029 iommu.forcedac= [ARM64, X86] Control IOVA allocation for PCI devices.
2030 Format: { "0" | "1" }
2031 0 - Try to allocate a 32-bit DMA address first, before
2032 falling back to the full range if needed.
2033 1 - Allocate directly from the full usable range,
2034 forcing Dual Address Cycle for PCI cards supporting
2035 greater than 32-bit addressing.
2037 iommu.strict= [ARM64, X86] Configure TLB invalidation behaviour
2038 Format: { "0" | "1" }
2040 Request that DMA unmap operations use deferred
2041 invalidation of hardware TLBs, for increased
2042 throughput at the cost of reduced device isolation.
2043 Will fall back to strict mode if not supported by
2044 the relevant IOMMU driver.
2045 1 - Strict mode (default).
2046 DMA unmap operations invalidate IOMMU hardware TLBs
2048 Note: on x86, the default behaviour depends on the
2049 equivalent driver-specific parameters, but a strict
2050 mode explicitly specified by either method takes
2054 [ARM64, X86] Configure DMA to bypass the IOMMU by default.
2055 Format: { "0" | "1" }
2056 0 - Use IOMMU translation for DMA.
2057 1 - Bypass the IOMMU for DMA.
2058 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
2060 io7= [HW] IO7 for Marvel-based Alpha systems
2061 See comment before marvel_specify_io7 in
2062 arch/alpha/kernel/core_marvel.c.
2064 io_delay= [X86] I/O delay method
2066 Standard port 0x80 based delay
2068 Alternate port 0xed based delay (needed on some systems)
2070 Simple two microseconds delay
2075 See Documentation/admin-guide/nfs/nfsroot.rst.
2077 ipcmni_extend [KNL] Extend the maximum number of unique System V
2078 IPC identifiers from 32,768 to 16,777,216.
2080 irqaffinity= [SMP] Set the default irq affinity mask
2081 The argument is a cpu list, as described above.
2083 irqchip.gicv2_force_probe=
2086 Force the kernel to look for the second 4kB page
2087 of a GICv2 controller even if the memory range
2088 exposed by the device tree is too small.
2090 irqchip.gicv3_nolpi=
2092 Force the kernel to ignore the availability of
2093 LPIs (and by consequence ITSs). Intended for system
2094 that use the kernel as a bootloader, and thus want
2095 to let secondary kernels in charge of setting up
2098 irqchip.gicv3_pseudo_nmi= [ARM64]
2099 Enables support for pseudo-NMIs in the kernel. This
2100 requires the kernel to be built with
2101 CONFIG_ARM64_PSEUDO_NMI.
2104 When an interrupt is not handled search all handlers
2105 for it. Intended to get systems with badly broken
2109 When an interrupt is not handled search all handlers
2110 for it. Also check all handlers each timer
2111 interrupt. Intended to get systems with badly broken
2115 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
2117 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
2118 [Deprecated - use cpusets instead]
2119 Format: [flag-list,]<cpu-list>
2121 Specify one or more CPUs to isolate from disturbances
2122 specified in the flag list (default: domain):
2125 Disable the tick when a single task runs.
2127 A residual 1Hz tick is offloaded to workqueues, which you
2128 need to affine to housekeeping through the global
2129 workqueue's affinity configured via the
2130 /sys/devices/virtual/workqueue/cpumask sysfs file, or
2131 by using the 'domain' flag described below.
2133 NOTE: by default the global workqueue runs on all CPUs,
2134 so to protect individual CPUs the 'cpumask' file has to
2135 be configured manually after bootup.
2138 Isolate from the general SMP balancing and scheduling
2139 algorithms. Note that performing domain isolation this way
2140 is irreversible: it's not possible to bring back a CPU to
2141 the domains once isolated through isolcpus. It's strongly
2142 advised to use cpusets instead to disable scheduler load
2143 balancing through the "cpuset.sched_load_balance" file.
2144 It offers a much more flexible interface where CPUs can
2145 move in and out of an isolated set anytime.
2147 You can move a process onto or off an "isolated" CPU via
2148 the CPU affinity syscalls or cpuset.
2149 <cpu number> begins at 0 and the maximum value is
2150 "number of CPUs in system - 1".
2154 Isolate from being targeted by managed interrupts
2155 which have an interrupt mask containing isolated
2156 CPUs. The affinity of managed interrupts is
2157 handled by the kernel and cannot be changed via
2158 the /proc/irq/* interfaces.
2160 This isolation is best effort and only effective
2161 if the automatically assigned interrupt mask of a
2162 device queue contains isolated and housekeeping
2163 CPUs. If housekeeping CPUs are online then such
2164 interrupts are directed to the housekeeping CPU
2165 so that IO submitted on the housekeeping CPU
2166 cannot disturb the isolated CPU.
2168 If a queue's affinity mask contains only isolated
2169 CPUs then this parameter has no effect on the
2170 interrupt routing decision, though interrupts are
2171 only delivered when tasks running on those
2172 isolated CPUs submit IO. IO submitted on
2173 housekeeping CPUs has no influence on those
2176 The format of <cpu-list> is described above.
2180 ivrs_ioapic [HW,X86-64]
2181 Provide an override to the IOAPIC-ID<->DEVICE-ID
2182 mapping provided in the IVRS ACPI table. For
2183 example, to map IOAPIC-ID decimal 10 to
2184 PCI device 00:14.0 write the parameter as:
2185 ivrs_ioapic[10]=00:14.0
2187 ivrs_hpet [HW,X86-64]
2188 Provide an override to the HPET-ID<->DEVICE-ID
2189 mapping provided in the IVRS ACPI table. For
2190 example, to map HPET-ID decimal 0 to
2191 PCI device 00:14.0 write the parameter as:
2192 ivrs_hpet[0]=00:14.0
2194 ivrs_acpihid [HW,X86-64]
2195 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2196 mapping provided in the IVRS ACPI table. For
2197 example, to map UART-HID:UID AMD0020:0 to
2198 PCI device 00:14.5 write the parameter as:
2199 ivrs_acpihid[00:14.5]=AMD0020:0
2201 js= [HW,JOY] Analog joystick
2202 See Documentation/input/joydev/joystick.rst.
2205 When CONFIG_RANDOMIZE_BASE is set, this disables
2206 kernel and module base offset ASLR (Address Space
2207 Layout Randomization).
2210 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2211 report on every invalid memory access. Without this
2212 parameter KASAN will print report only for the first
2217 kernelcore= [KNL,X86,IA-64,PPC]
2218 Format: nn[KMGTPE] | nn% | "mirror"
2219 This parameter specifies the amount of memory usable by
2220 the kernel for non-movable allocations. The requested
2221 amount is spread evenly throughout all nodes in the
2222 system as ZONE_NORMAL. The remaining memory is used for
2223 movable memory in its own zone, ZONE_MOVABLE. In the
2224 event, a node is too small to have both ZONE_NORMAL and
2225 ZONE_MOVABLE, kernelcore memory will take priority and
2226 other nodes will have a larger ZONE_MOVABLE.
2228 ZONE_MOVABLE is used for the allocation of pages that
2229 may be reclaimed or moved by the page migration
2230 subsystem. Note that allocations like PTEs-from-HighMem
2231 still use the HighMem zone if it exists, and the Normal
2232 zone if it does not.
2234 It is possible to specify the exact amount of memory in
2235 the form of "nn[KMGTPE]", a percentage of total system
2236 memory in the form of "nn%", or "mirror". If "mirror"
2237 option is specified, mirrored (reliable) memory is used
2238 for non-movable allocations and remaining memory is used
2239 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2240 are exclusive, so you cannot specify multiple forms.
2242 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
2243 Format: <Controller#>[,poll interval]
2244 The controller # is the number of the ehci usb debug
2245 port as it is probed via PCI. The poll interval is
2246 optional and is the number seconds in between
2247 each poll cycle to the debug port in case you need
2248 the functionality for interrupting the kernel with
2249 gdb or control-c on the dbgp connection. When
2250 not using this parameter you use sysrq-g to break into
2251 the kernel debugger.
2253 kgdboc= [KGDB,HW] kgdb over consoles.
2254 Requires a tty driver that supports console polling,
2255 or a supported polling keyboard driver (non-usb).
2256 Serial only format: <serial_device>[,baud]
2257 keyboard only format: kbd
2258 keyboard and serial format: kbd,<serial_device>[,baud]
2259 Optional Kernel mode setting:
2260 kms, kbd format: kms,kbd
2261 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2263 kgdboc_earlycon= [KGDB,HW]
2264 If the boot console provides the ability to read
2265 characters and can work in polling mode, you can use
2266 this parameter to tell kgdb to use it as a backend
2267 until the normal console is registered. Intended to
2268 be used together with the kgdboc parameter which
2269 specifies the normal console to transition to.
2271 The name of the early console should be specified
2272 as the value of this parameter. Note that the name of
2273 the early console might be different than the tty
2274 name passed to kgdboc. It's OK to leave the value
2275 blank and the first boot console that implements
2276 read() will be picked.
2278 kgdbwait [KGDB] Stop kernel execution and enter the
2279 kernel debugger at the earliest opportunity.
2281 kmac= [MIPS] Korina ethernet MAC address.
2282 Configure the RouterBoard 532 series on-chip
2283 Ethernet adapter MAC address.
2285 kmemleak= [KNL] Boot-time kmemleak enable/disable
2286 Valid arguments: on, off
2288 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2291 kprobe_event=[probe-list]
2292 [FTRACE] Add kprobe events and enable at boot time.
2293 The probe-list is a semicolon delimited list of probe
2294 definitions. Each definition is same as kprobe_events
2295 interface, but the parameters are comma delimited.
2296 For example, to add a kprobe event on vfs_read with
2297 arg1 and arg2, add to the command line;
2299 kprobe_event=p,vfs_read,$arg1,$arg2
2301 See also Documentation/trace/kprobetrace.rst "Kernel
2302 Boot Parameter" section.
2304 kpti= [ARM64] Control page table isolation of user
2305 and kernel address spaces.
2306 Default: enabled on cores which need mitigation.
2310 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2311 Default is 0 (don't ignore, but inject #GP)
2313 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2314 Default is false (don't support).
2316 kvm.mmu_audit= [KVM] This is a R/W parameter which allows audit
2321 [KVM] Controls the software workaround for the
2322 X86_BUG_ITLB_MULTIHIT bug.
2323 force : Always deploy workaround.
2324 off : Never deploy workaround.
2325 auto : Deploy workaround based on the presence of
2326 X86_BUG_ITLB_MULTIHIT.
2330 If the software workaround is enabled for the host,
2331 guests do need not to enable it for nested guests.
2333 kvm.nx_huge_pages_recovery_ratio=
2334 [KVM] Controls how many 4KiB pages are periodically zapped
2335 back to huge pages. 0 disables the recovery, otherwise if
2336 the value is N KVM will zap 1/Nth of the 4KiB pages every
2337 minute. The default is 60.
2339 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM.
2340 Default is 1 (enabled)
2342 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU)
2344 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode.
2347 [KVM,ARM] Select one of KVM/arm64's modes of operation.
2349 nvhe: Standard nVHE-based mode, without support for
2352 protected: nVHE-based mode with support for guests whose
2353 state is kept private from the host.
2354 Not valid if the kernel is running in EL2.
2356 Defaults to VHE/nVHE based on hardware support.
2358 kvm-arm.vgic_v3_group0_trap=
2359 [KVM,ARM] Trap guest accesses to GICv3 group-0
2362 kvm-arm.vgic_v3_group1_trap=
2363 [KVM,ARM] Trap guest accesses to GICv3 group-1
2366 kvm-arm.vgic_v3_common_trap=
2367 [KVM,ARM] Trap guest accesses to GICv3 common
2370 kvm-arm.vgic_v4_enable=
2371 [KVM,ARM] Allow use of GICv4 for direct injection of
2374 kvm_cma_resv_ratio=n [PPC]
2375 Reserves given percentage from system memory area for
2376 contiguous memory allocation for KVM hash pagetable
2378 By default it reserves 5% of total system memory.
2382 kvm-intel.ept= [KVM,Intel] Disable extended page tables
2383 (virtualized MMU) support on capable Intel chips.
2384 Default is 1 (enabled)
2386 kvm-intel.emulate_invalid_guest_state=
2387 [KVM,Intel] Enable emulation of invalid guest states
2388 Default is 0 (disabled)
2390 kvm-intel.flexpriority=
2391 [KVM,Intel] Disable FlexPriority feature (TPR shadow).
2392 Default is 1 (enabled)
2395 [KVM,Intel] Enable VMX nesting (nVMX).
2396 Default is 0 (disabled)
2398 kvm-intel.unrestricted_guest=
2399 [KVM,Intel] Disable unrestricted guest feature
2400 (virtualized real and unpaged mode) on capable
2401 Intel chips. Default is 1 (enabled)
2403 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2406 Valid arguments: never, cond, always
2408 always: L1D cache flush on every VMENTER.
2409 cond: Flush L1D on VMENTER only when the code between
2410 VMEXIT and VMENTER can leak host memory.
2411 never: Disables the mitigation
2413 Default is cond (do L1 cache flush in specific instances)
2415 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification
2416 feature (tagged TLBs) on capable Intel chips.
2417 Default is 1 (enabled)
2419 l1tf= [X86] Control mitigation of the L1TF vulnerability on
2422 The kernel PTE inversion protection is unconditionally
2423 enabled and cannot be disabled.
2426 Provides all available mitigations for the
2427 L1TF vulnerability. Disables SMT and
2428 enables all mitigations in the
2429 hypervisors, i.e. unconditional L1D flush.
2431 SMT control and L1D flush control via the
2432 sysfs interface is still possible after
2433 boot. Hypervisors will issue a warning
2434 when the first VM is started in a
2435 potentially insecure configuration,
2436 i.e. SMT enabled or L1D flush disabled.
2439 Same as 'full', but disables SMT and L1D
2440 flush runtime control. Implies the
2441 'nosmt=force' command line option.
2442 (i.e. sysfs control of SMT is disabled.)
2445 Leaves SMT enabled and enables the default
2446 hypervisor mitigation, i.e. conditional
2449 SMT control and L1D flush control via the
2450 sysfs interface is still possible after
2451 boot. Hypervisors will issue a warning
2452 when the first VM is started in a
2453 potentially insecure configuration,
2454 i.e. SMT enabled or L1D flush disabled.
2458 Disables SMT and enables the default
2459 hypervisor mitigation.
2461 SMT control and L1D flush control via the
2462 sysfs interface is still possible after
2463 boot. Hypervisors will issue a warning
2464 when the first VM is started in a
2465 potentially insecure configuration,
2466 i.e. SMT enabled or L1D flush disabled.
2469 Same as 'flush', but hypervisors will not
2470 warn when a VM is started in a potentially
2471 insecure configuration.
2474 Disables hypervisor mitigations and doesn't
2476 It also drops the swap size and available
2477 RAM limit restriction on both hypervisor and
2482 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2488 lapic [X86-32,APIC] Enable the local APIC even if BIOS
2491 lapic= [X86,APIC] Do not use TSC deadline
2492 value for LAPIC timer one-shot implementation. Default
2493 back to the programmable timer unit in the LAPIC.
2494 Format: notscdeadline
2496 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
2499 libata.dma= [LIBATA] DMA control
2500 libata.dma=0 Disable all PATA and SATA DMA
2501 libata.dma=1 PATA and SATA Disk DMA only
2502 libata.dma=2 ATAPI (CDROM) DMA only
2503 libata.dma=4 Compact Flash DMA only
2504 Combinations also work, so libata.dma=3 enables DMA
2505 for disks and CDROMs, but not CFs.
2507 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2508 libata.ignore_hpa=0 keep BIOS limits (default)
2509 libata.ignore_hpa=1 ignore limits, using full disk
2511 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2515 libata.force= [LIBATA] Force configurations. The format is comma-
2516 separated list of "[ID:]VAL" where ID is
2517 PORT[.DEVICE]. PORT and DEVICE are decimal numbers
2518 matching port, link or device. Basically, it matches
2519 the ATA ID string printed on console by libata. If
2520 the whole ID part is omitted, the last PORT and DEVICE
2521 values are used. If ID hasn't been specified yet, the
2522 configuration applies to all ports, links and devices.
2524 If only DEVICE is omitted, the parameter applies to
2525 the port and all links and devices behind it. DEVICE
2526 number of 0 either selects the first device or the
2527 first fan-out link behind PMP device. It does not
2528 select the host link. DEVICE number of 15 selects the
2529 host link and device attached to it.
2531 The VAL specifies the configuration to force. As long
2532 as there's no ambiguity shortcut notation is allowed.
2533 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2534 The following configurations can be forced.
2536 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2537 Any ID with matching PORT is used.
2539 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2541 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2542 udma[/][16,25,33,44,66,100,133] notation is also
2545 * [no]ncq: Turn on or off NCQ.
2547 * [no]ncqtrim: Turn off queued DSM TRIM.
2549 * nohrst, nosrst, norst: suppress hard, soft
2552 * rstonce: only attempt one reset during
2553 hot-unplug link recovery
2555 * dump_id: dump IDENTIFY data.
2557 * atapi_dmadir: Enable ATAPI DMADIR bridge support
2559 * disable: Disable this device.
2561 If there are multiple matching configurations changing
2562 the same attribute, the last one is used.
2564 memblock=debug [KNL] Enable memblock debug messages.
2566 load_ramdisk= [RAM] [Deprecated]
2568 lockd.nlm_grace_period=P [NFS] Assign grace period.
2571 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2574 lockd.nlm_timeout=T [NFS] Assign timeout value.
2577 lockd.nlm_udpport=M [NFS] Assign UDP port.
2580 lockdown= [SECURITY]
2581 { integrity | confidentiality }
2582 Enable the kernel lockdown feature. If set to
2583 integrity, kernel features that allow userland to
2584 modify the running kernel are disabled. If set to
2585 confidentiality, kernel features that allow userland
2586 to extract confidential information from the kernel
2589 locktorture.nreaders_stress= [KNL]
2590 Set the number of locking read-acquisition kthreads.
2591 Defaults to being automatically set based on the
2592 number of online CPUs.
2594 locktorture.nwriters_stress= [KNL]
2595 Set the number of locking write-acquisition kthreads.
2597 locktorture.onoff_holdoff= [KNL]
2598 Set time (s) after boot for CPU-hotplug testing.
2600 locktorture.onoff_interval= [KNL]
2601 Set time (s) between CPU-hotplug operations, or
2602 zero to disable CPU-hotplug testing.
2604 locktorture.shuffle_interval= [KNL]
2605 Set task-shuffle interval (jiffies). Shuffling
2606 tasks allows some CPUs to go into dyntick-idle
2607 mode during the locktorture test.
2609 locktorture.shutdown_secs= [KNL]
2610 Set time (s) after boot system shutdown. This
2611 is useful for hands-off automated testing.
2613 locktorture.stat_interval= [KNL]
2614 Time (s) between statistics printk()s.
2616 locktorture.stutter= [KNL]
2617 Time (s) to stutter testing, for example,
2618 specifying five seconds causes the test to run for
2619 five seconds, wait for five seconds, and so on.
2620 This tests the locking primitive's ability to
2621 transition abruptly to and from idle.
2623 locktorture.torture_type= [KNL]
2624 Specify the locking implementation to test.
2626 locktorture.verbose= [KNL]
2627 Enable additional printk() statements.
2629 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
2632 loglevel= All Kernel Messages with a loglevel smaller than the
2633 console loglevel will be printed to the console. It can
2634 also be changed with klogd or other programs. The
2635 loglevels are defined as follows:
2637 0 (KERN_EMERG) system is unusable
2638 1 (KERN_ALERT) action must be taken immediately
2639 2 (KERN_CRIT) critical conditions
2640 3 (KERN_ERR) error conditions
2641 4 (KERN_WARNING) warning conditions
2642 5 (KERN_NOTICE) normal but significant condition
2643 6 (KERN_INFO) informational
2644 7 (KERN_DEBUG) debug-level messages
2646 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
2647 in bytes. n must be a power of two and greater
2648 than the minimal size. The minimal size is defined
2649 by LOG_BUF_SHIFT kernel config parameter. There is
2650 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2651 that allows to increase the default size depending on
2652 the number of CPUs. See init/Kconfig for more details.
2654 logo.nologo [FB] Disables display of the built-in Linux logo.
2655 This may be used to provide more screen space for
2656 kernel log messages and is useful when debugging
2657 kernel boot problems.
2659 lp=0 [LP] Specify parallel ports to use, e.g,
2660 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2661 lp=reset first parallel port). 'lp=0' disables the
2662 lp=auto printer driver. 'lp=reset' (which can be
2663 specified in addition to the ports) causes
2664 attached printers to be reset. Using
2665 lp=port1,port2,... specifies the parallel ports
2666 to associate lp devices with, starting with
2667 lp0. A port specification may be 'none' to skip
2668 that lp device, or a parport name such as
2669 'parport0'. Specifying 'lp=auto' instead of a
2670 port specification list means that device IDs
2671 from each port should be examined, to see if
2672 an IEEE 1284-compliant printer is attached; if
2673 so, the driver will manage that printer.
2674 See also header of drivers/char/lp.c.
2677 Sets loops_per_jiffy to given constant, thus avoiding
2678 time-consuming boot-time autodetection (up to 250 ms per
2679 CPU). 0 enables autodetection (default). To determine
2680 the correct value for your kernel, boot with normal
2681 autodetection and see what value is printed. Note that
2682 on SMP systems the preset will be applied to all CPUs,
2683 which is likely to cause problems if your CPUs need
2684 significantly divergent settings. An incorrect value
2685 will cause delays in the kernel to be wrong, leading to
2686 unpredictable I/O errors and other breakage. Although
2687 unlikely, in the extreme case this might damage your
2691 Format: <io>,<irq>,<dma>
2693 lsm.debug [SECURITY] Enable LSM initialization debugging output.
2696 [SECURITY] Choose order of LSM initialization. This
2697 overrides CONFIG_LSM, and the "security=" parameter.
2699 machvec= [IA-64] Force the use of a particular machine-vector
2700 (machvec) in a generic kernel.
2701 Example: machvec=hpzx1
2703 machtype= [Loongson] Share the same kernel image file between
2704 different yeeloong laptops.
2705 Example: machtype=lemote-yeeloong-2f-7inch
2707 max_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory greater
2708 than or equal to this physical address is ignored.
2710 maxcpus= [SMP] Maximum number of processors that an SMP kernel
2711 will bring up during bootup. maxcpus=n : n >= 0 limits
2712 the kernel to bring up 'n' processors. Surely after
2713 bootup you can bring up the other plugged cpu by executing
2714 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
2715 only takes effect during system bootup.
2716 While n=0 is a special case, it is equivalent to "nosmp",
2717 which also disables the IO APIC.
2719 max_loop= [LOOP] The number of loop block devices that get
2720 (loop.max_loop) unconditionally pre-created at init time. The default
2721 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
2722 of statically allocating a predefined number, loop
2723 devices can be requested on-demand with the
2724 /dev/loop-control interface.
2726 mce [X86-32] Machine Check Exception
2728 mce=option [X86-64] See Documentation/x86/x86_64/boot-options.rst
2730 md= [HW] RAID subsystems devices and level
2731 See Documentation/admin-guide/md.rst.
2734 Format: <first>,<last>
2735 Specifies range of consoles to be captured by the MDA.
2738 Control mitigation for the Micro-architectural Data
2739 Sampling (MDS) vulnerability.
2741 Certain CPUs are vulnerable to an exploit against CPU
2742 internal buffers which can forward information to a
2743 disclosure gadget under certain conditions.
2745 In vulnerable processors, the speculatively
2746 forwarded data can be used in a cache side channel
2747 attack, to access data to which the attacker does
2748 not have direct access.
2750 This parameter controls the MDS mitigation. The
2753 full - Enable MDS mitigation on vulnerable CPUs
2754 full,nosmt - Enable MDS mitigation and disable
2755 SMT on vulnerable CPUs
2756 off - Unconditionally disable MDS mitigation
2758 On TAA-affected machines, mds=off can be prevented by
2759 an active TAA mitigation as both vulnerabilities are
2760 mitigated with the same mechanism so in order to disable
2761 this mitigation, you need to specify tsx_async_abort=off
2764 Not specifying this option is equivalent to
2767 For details see: Documentation/admin-guide/hw-vuln/mds.rst
2769 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
2770 Amount of memory to be used in cases as follows:
2773 2 when the kernel is not able to see the whole system memory;
2774 3 memory that lies after 'mem=' boundary is excluded from
2775 the hypervisor, then assigned to KVM guests.
2777 [X86] Work as limiting max address. Use together
2778 with memmap= to avoid physical address space collisions.
2779 Without memmap= PCI devices could be placed at addresses
2780 belonging to unused RAM.
2782 Note that this only takes effects during boot time since
2783 in above case 3, memory may need be hot added after boot
2784 if system memory of hypervisor is not sufficient.
2786 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
2790 [KNL,SH] Allow user to override the default size for
2791 per-device physically contiguous DMA buffers.
2793 memhp_default_state=online/offline
2794 [KNL] Set the initial state for the memory hotplug
2795 onlining policy. If not specified, the default value is
2796 set according to the
2797 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
2799 See Documentation/admin-guide/mm/memory-hotplug.rst.
2801 memmap=exactmap [KNL,X86] Enable setting of an exact
2802 E820 memory map, as specified by the user.
2803 Such memmap=exactmap lines can be constructed based on
2804 BIOS output or other requirements. See the memmap=nn@ss
2807 memmap=nn[KMG]@ss[KMG]
2808 [KNL, X86, MIPS, XTENSA] Force usage of a specific region of memory.
2809 Region of memory to be used is from ss to ss+nn.
2810 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
2811 which limits max address to nn[KMG].
2812 Multiple different regions can be specified,
2815 memmap=100M@2G,100M#3G,1G!1024G
2817 memmap=nn[KMG]#ss[KMG]
2818 [KNL,ACPI] Mark specific memory as ACPI data.
2819 Region of memory to be marked is from ss to ss+nn.
2821 memmap=nn[KMG]$ss[KMG]
2822 [KNL,ACPI] Mark specific memory as reserved.
2823 Region of memory to be reserved is from ss to ss+nn.
2824 Example: Exclude memory from 0x18690000-0x1869ffff
2825 memmap=64K$0x18690000
2827 memmap=0x10000$0x18690000
2828 Some bootloaders may need an escape character before '$',
2829 like Grub2, otherwise '$' and the following number
2832 memmap=nn[KMG]!ss[KMG]
2833 [KNL,X86] Mark specific memory as protected.
2834 Region of memory to be used, from ss to ss+nn.
2835 The memory region may be marked as e820 type 12 (0xc)
2836 and is NVDIMM or ADR memory.
2838 memmap=<size>%<offset>-<oldtype>+<newtype>
2839 [KNL,ACPI] Convert memory within the specified region
2840 from <oldtype> to <newtype>. If "-<oldtype>" is left
2841 out, the whole region will be marked as <newtype>,
2842 even if previously unavailable. If "+<newtype>" is left
2843 out, matching memory will be removed. Types are
2844 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
2845 3 = ACPI, 12 = PRAM.
2847 memory_corruption_check=0/1 [X86]
2848 Some BIOSes seem to corrupt the first 64k of
2849 memory when doing things like suspend/resume.
2850 Setting this option will scan the memory
2851 looking for corruption. Enabling this will
2852 both detect corruption and prevent the kernel
2853 from using the memory being corrupted.
2854 However, its intended as a diagnostic tool; if
2855 repeatable BIOS-originated corruption always
2856 affects the same memory, you can use memmap=
2857 to prevent the kernel from using that memory.
2859 memory_corruption_check_size=size [X86]
2860 By default it checks for corruption in the low
2861 64k, making this memory unavailable for normal
2862 use. Use this parameter to scan for
2863 corruption in more or less memory.
2865 memory_corruption_check_period=seconds [X86]
2866 By default it checks for corruption every 60
2867 seconds. Use this parameter to check at some
2868 other rate. 0 disables periodic checking.
2870 memory_hotplug.memmap_on_memory
2871 [KNL,X86,ARM] Boolean flag to enable this feature.
2872 Format: {on | off (default)}
2873 When enabled, runtime hotplugged memory will
2874 allocate its internal metadata (struct pages)
2875 from the hotadded memory which will allow to
2876 hotadd a lot of memory without requiring
2877 additional memory to do so.
2878 This feature is disabled by default because it
2879 has some implication on large (e.g. GB)
2880 allocations in some configurations (e.g. small
2882 The state of the flag can be read in
2883 /sys/module/memory_hotplug/parameters/memmap_on_memory.
2884 Note that even when enabled, there are a few cases where
2885 the feature is not effective.
2887 This is not compatible with hugetlb_free_vmemmap. If
2888 both parameters are enabled, hugetlb_free_vmemmap takes
2889 precedence over memory_hotplug.memmap_on_memory.
2891 memtest= [KNL,X86,ARM,PPC,RISCV] Enable memtest
2893 default : 0 <disable>
2894 Specifies the number of memtest passes to be
2895 performed. Each pass selects another test
2896 pattern from a given set of patterns. Memtest
2897 fills the memory with this pattern, validates
2898 memory contents and reserves bad memory
2899 regions that are detected.
2901 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
2902 Valid arguments: on, off
2903 Default (depends on kernel configuration option):
2904 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
2905 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
2906 mem_encrypt=on: Activate SME
2907 mem_encrypt=off: Do not activate SME
2909 Refer to Documentation/virt/kvm/amd-memory-encryption.rst
2910 for details on when memory encryption can be activated.
2912 mem_sleep_default= [SUSPEND] Default system suspend mode:
2913 s2idle - Suspend-To-Idle
2914 shallow - Power-On Suspend or equivalent (if supported)
2915 deep - Suspend-To-RAM or equivalent (if supported)
2916 See Documentation/admin-guide/pm/sleep-states.rst.
2918 meye.*= [HW] Set MotionEye Camera parameters
2919 See Documentation/admin-guide/media/meye.rst.
2921 mfgpt_irq= [IA-32] Specify the IRQ to use for the
2922 Multi-Function General Purpose Timers on AMD Geode
2925 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
2926 the BIOS has incorrectly applied a workaround. TinyBIOS
2927 version 0.98 is known to be affected, 0.99 fixes the
2928 problem by letting the user disable the workaround.
2932 min_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory below this
2933 physical address is ignored.
2935 mini2440= [ARM,HW,KNL]
2936 Format:[0..2][b][c][t]
2938 MINI2440 configuration specification:
2939 0 - The attached screen is the 3.5" TFT
2940 1 - The attached screen is the 7" TFT
2941 2 - The VGA Shield is attached (1024x768)
2942 Leaving out the screen size parameter will not load
2943 the TFT driver, and the framebuffer will be left
2945 b - Enable backlight. The TFT backlight pin will be
2946 linked to the kernel VESA blanking code and a GPIO
2947 LED. This parameter is not necessary when using the
2949 c - Enable the s3c camera interface.
2950 t - Reserved for enabling touchscreen support. The
2951 touchscreen support is not enabled in the mainstream
2952 kernel as of 2.6.30, a preliminary port can be found
2953 in the "bleeding edge" mini2440 support kernel at
2954 https://repo.or.cz/w/linux-2.6/mini2440.git
2957 [X86,PPC,S390,ARM64] Control optional mitigations for
2958 CPU vulnerabilities. This is a set of curated,
2959 arch-independent options, each of which is an
2960 aggregation of existing arch-specific options.
2963 Disable all optional CPU mitigations. This
2964 improves system performance, but it may also
2965 expose users to several CPU vulnerabilities.
2966 Equivalent to: nopti [X86,PPC]
2968 nospectre_v1 [X86,PPC]
2970 nospectre_v2 [X86,PPC,S390,ARM64]
2971 spectre_v2_user=off [X86]
2972 spec_store_bypass_disable=off [X86,PPC]
2973 ssbd=force-off [ARM64]
2976 tsx_async_abort=off [X86]
2977 kvm.nx_huge_pages=off [X86]
2978 no_entry_flush [PPC]
2979 no_uaccess_flush [PPC]
2982 This does not have any effect on
2983 kvm.nx_huge_pages when
2984 kvm.nx_huge_pages=force.
2987 Mitigate all CPU vulnerabilities, but leave SMT
2988 enabled, even if it's vulnerable. This is for
2989 users who don't want to be surprised by SMT
2990 getting disabled across kernel upgrades, or who
2991 have other ways of avoiding SMT-based attacks.
2992 Equivalent to: (default behavior)
2995 Mitigate all CPU vulnerabilities, disabling SMT
2996 if needed. This is for users who always want to
2997 be fully mitigated, even if it means losing SMT.
2998 Equivalent to: l1tf=flush,nosmt [X86]
2999 mds=full,nosmt [X86]
3000 tsx_async_abort=full,nosmt [X86]
3003 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
3004 parameter allows control of the logging verbosity for
3005 the additional memory initialisation checks. A value
3006 of 0 disables mminit logging and a level of 4 will
3007 log everything. Information is printed at KERN_DEBUG
3008 so loglevel=8 may also need to be specified.
3011 [KNL] When CONFIG_MODULE_SIG is set, this means that
3012 modules without (valid) signatures will fail to load.
3013 Note that if CONFIG_MODULE_SIG_FORCE is set, that
3014 is always true, so this option does nothing.
3016 module_blacklist= [KNL] Do not load a comma-separated list of
3017 modules. Useful for debugging problem modules.
3020 [MOUSE] Maximum time between finger touching and
3021 leaving touchpad surface for touch to be considered
3022 a tap and be reported as a left button click (for
3023 touchpads working in absolute mode only).
3025 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
3026 reporting absolute coordinates, such as tablets
3027 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
3028 reporting absolute coordinates, such as tablets
3030 movablecore= [KNL,X86,IA-64,PPC]
3031 Format: nn[KMGTPE] | nn%
3032 This parameter is the complement to kernelcore=, it
3033 specifies the amount of memory used for migratable
3034 allocations. If both kernelcore and movablecore is
3035 specified, then kernelcore will be at *least* the
3036 specified value but may be more. If movablecore on its
3037 own is specified, the administrator must be careful
3038 that the amount of memory usable for all allocations
3041 movable_node [KNL] Boot-time switch to make hotplugable memory
3042 NUMA nodes to be movable. This means that the memory
3043 of such nodes will be usable only for movable
3044 allocations which rules out almost all kernel
3045 allocations. Use with caution!
3047 MTD_Partition= [MTD]
3048 Format: <name>,<region-number>,<size>,<offset>
3050 MTD_Region= [MTD] Format:
3051 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
3054 See drivers/mtd/parsers/cmdlinepart.c
3056 multitce=off [PPC] This parameter disables the use of the pSeries
3057 firmware feature for updating multiple TCE entries
3060 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
3062 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
3064 boundary - index of last SLC block on Flex-OneNAND.
3065 The remaining blocks are configured as MLC blocks.
3066 lock - Configure if Flex-OneNAND boundary should be locked.
3067 Once locked, the boundary cannot be changed.
3068 1 indicates lock status, 0 indicates unlock status.
3071 ARM/S3C2412 JIVE boot control
3073 See arch/arm/mach-s3c/mach-jive.c
3075 mtouchusb.raw_coordinates=
3076 [HW] Make the MicroTouch USB driver use raw coordinates
3077 ('y', default) or cooked coordinates ('n')
3079 mtrr_chunk_size=nn[KMG] [X86]
3080 used for mtrr cleanup. It is largest continuous chunk
3081 that could hold holes aka. UC entries.
3083 mtrr_gran_size=nn[KMG] [X86]
3084 Used for mtrr cleanup. It is granularity of mtrr block.
3086 Large value could prevent small alignment from
3089 mtrr_spare_reg_nr=n [X86]
3091 Range: 0,7 : spare reg number
3093 Used for mtrr cleanup. It is spare mtrr entries number.
3094 Set to 2 or more if your graphical card needs more.
3096 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
3098 netdev= [NET] Network devices parameters
3099 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
3100 Note that mem_start is often overloaded to mean
3101 something different and driver-specific.
3102 This usage is only documented in each driver source
3106 [NETFILTER] Enable connection tracking flow accounting
3107 0 to disable accounting
3108 1 to enable accounting
3111 nfsaddrs= [NFS] Deprecated. Use ip= instead.
3112 See Documentation/admin-guide/nfs/nfsroot.rst.
3114 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
3115 See Documentation/admin-guide/nfs/nfsroot.rst.
3117 nfsrootdebug [NFS] enable nfsroot debugging messages.
3118 See Documentation/admin-guide/nfs/nfsroot.rst.
3120 nfs.callback_nr_threads=
3121 [NFSv4] set the total number of threads that the
3122 NFS client will assign to service NFSv4 callback
3125 nfs.callback_tcpport=
3126 [NFS] set the TCP port on which the NFSv4 callback
3127 channel should listen.
3130 [NFS] sets the pathname to the program which is used
3131 to update the NFS client cache entries.
3133 nfs.cache_getent_timeout=
3134 [NFS] sets the timeout after which an attempt to
3135 update a cache entry is deemed to have failed.
3137 nfs.idmap_cache_timeout=
3138 [NFS] set the maximum lifetime for idmapper cache
3142 [NFS] enable 64-bit inode numbers.
3143 If zero, the NFS client will fake up a 32-bit inode
3144 number for the readdir() and stat() syscalls instead
3145 of returning the full 64-bit number.
3146 The default is to return 64-bit inode numbers.
3148 nfs.max_session_cb_slots=
3149 [NFSv4.1] Sets the maximum number of session
3150 slots the client will assign to the callback
3151 channel. This determines the maximum number of
3152 callbacks the client will process in parallel for
3153 a particular server.
3155 nfs.max_session_slots=
3156 [NFSv4.1] Sets the maximum number of session slots
3157 the client will attempt to negotiate with the server.
3158 This limits the number of simultaneous RPC requests
3159 that the client can send to the NFSv4.1 server.
3160 Note that there is little point in setting this
3161 value higher than the max_tcp_slot_table_limit.
3163 nfs.nfs4_disable_idmapping=
3164 [NFSv4] When set to the default of '1', this option
3165 ensures that both the RPC level authentication
3166 scheme and the NFS level operations agree to use
3167 numeric uids/gids if the mount is using the
3168 'sec=sys' security flavour. In effect it is
3169 disabling idmapping, which can make migration from
3170 legacy NFSv2/v3 systems to NFSv4 easier.
3171 Servers that do not support this mode of operation
3172 will be autodetected by the client, and it will fall
3173 back to using the idmapper.
3174 To turn off this behaviour, set the value to '0'.
3176 [NFS4] Specify an additional fixed unique ident-
3177 ification string that NFSv4 clients can insert into
3178 their nfs_client_id4 string. This is typically a
3179 UUID that is generated at system install time.
3181 nfs.send_implementation_id =
3182 [NFSv4.1] Send client implementation identification
3183 information in exchange_id requests.
3184 If zero, no implementation identification information
3186 The default is to send the implementation identification
3189 nfs.recover_lost_locks =
3190 [NFSv4] Attempt to recover locks that were lost due
3191 to a lease timeout on the server. Please note that
3192 doing this risks data corruption, since there are
3193 no guarantees that the file will remain unchanged
3194 after the locks are lost.
3195 If you want to enable the kernel legacy behaviour of
3196 attempting to recover these locks, then set this
3198 The default parameter value of '0' causes the kernel
3199 not to attempt recovery of lost locks.
3201 nfs4.layoutstats_timer =
3202 [NFSv4.2] Change the rate at which the kernel sends
3203 layoutstats to the pNFS metadata server.
3205 Setting this to value to 0 causes the kernel to use
3206 whatever value is the default set by the layout
3207 driver. A non-zero value sets the minimum interval
3208 in seconds between layoutstats transmissions.
3210 nfsd.nfs4_disable_idmapping=
3211 [NFSv4] When set to the default of '1', the NFSv4
3212 server will return only numeric uids and gids to
3213 clients using auth_sys, and will accept numeric uids
3214 and gids from such clients. This is intended to ease
3215 migration from NFSv2/v3.
3217 nmi_backtrace.backtrace_idle [KNL]
3218 Dump stacks even of idle CPUs in response to an
3219 NMI stack-backtrace request.
3221 nmi_debug= [KNL,SH] Specify one or more actions to take
3222 when a NMI is triggered.
3223 Format: [state][,regs][,debounce][,die]
3225 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
3226 Format: [panic,][nopanic,][num]
3228 0 - turn hardlockup detector in nmi_watchdog off
3229 1 - turn hardlockup detector in nmi_watchdog on
3230 When panic is specified, panic when an NMI watchdog
3231 timeout occurs (or 'nopanic' to not panic on an NMI
3232 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
3233 To disable both hard and soft lockup detectors,
3234 please see 'nowatchdog'.
3235 This is useful when you use a panic=... timeout and
3236 need the box quickly up again.
3238 These settings can be accessed at runtime via
3239 the nmi_watchdog and hardlockup_panic sysctls.
3241 netpoll.carrier_timeout=
3242 [NET] Specifies amount of time (in seconds) that
3243 netpoll should wait for a carrier. By default netpoll
3246 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3247 emulation library even if a 387 maths coprocessor
3250 no5lvl [X86-64] Disable 5-level paging mode. Forces
3251 kernel to use 4-level paging instead.
3253 nofsgsbase [X86] Disables FSGSBASE instructions.
3256 [HW] Never suspend the console
3257 Disable suspending of consoles during suspend and
3258 hibernate operations. Once disabled, debugging
3259 messages can reach various consoles while the rest
3260 of the system is being put to sleep (ie, while
3261 debugging driver suspend/resume hooks). This may
3262 not work reliably with all consoles, but is known
3263 to work with serial and VGA consoles.
3264 To facilitate more flexible debugging, we also add
3265 console_suspend, a printk module parameter to control
3266 it. Users could use console_suspend (usually
3267 /sys/module/printk/parameters/console_suspend) to
3268 turn on/off it dynamically.
3270 novmcoredd [KNL,KDUMP]
3271 Disable device dump. Device dump allows drivers to
3272 append dump data to vmcore so you can collect driver
3273 specified debug info. Drivers can append the data
3274 without any limit and this data is stored in memory,
3275 so this may cause significant memory stress. Disabling
3276 device dump can help save memory but the driver debug
3277 data will be no longer available. This parameter
3278 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
3281 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
3282 caches in the slab allocator. Saves per-node memory,
3283 but will impact performance.
3287 noaltinstr [S390] Disables alternative instructions patching
3288 (CPU alternatives feature).
3290 noapic [SMP,APIC] Tells the kernel to not make use of any
3291 IOAPICs that may be present in the system.
3293 noautogroup Disable scheduler automatic task group creation.
3295 nobats [PPC] Do not use BATs for mapping kernel lowmem
3296 on "Classic" PPC cores.
3300 noclflush [BUGS=X86] Don't use the CLFLUSH instruction
3302 delayacct [KNL] Enable per-task delay accounting
3304 nodsp [SH] Disable hardware DSP at boot time.
3306 noefi Disable EFI runtime services support.
3308 no_entry_flush [PPC] Don't flush the L1-D cache when entering the kernel.
3313 On X86-32 available only on PAE configured kernels.
3314 noexec=on: enable non-executable mappings (default)
3315 noexec=off: disable non-executable mappings
3318 Disable SMAP (Supervisor Mode Access Prevention)
3319 even if it is supported by processor.
3322 Disable SMEP (Supervisor Mode Execution Prevention)
3323 even if it is supported by processor.
3326 This affects only 32-bit executables.
3327 noexec32=on: enable non-executable mappings (default)
3328 read doesn't imply executable mappings
3329 noexec32=off: disable non-executable mappings
3330 read implies executable mappings
3332 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3334 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3335 register save and restore. The kernel will only save
3336 legacy floating-point registers on task switch.
3338 nohugeiomap [KNL,X86,PPC,ARM64] Disable kernel huge I/O mappings.
3340 nohugevmalloc [PPC] Disable kernel huge vmalloc mappings.
3342 nosmt [KNL,S390] Disable symmetric multithreading (SMT).
3343 Equivalent to smt=1.
3345 [KNL,X86] Disable symmetric multithreading (SMT).
3346 nosmt=force: Force disable SMT, cannot be undone
3347 via the sysfs control file.
3349 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1
3350 (bounds check bypass). With this option data leaks are
3351 possible in the system.
3353 nospectre_v2 [X86,PPC_FSL_BOOK3E,ARM64] Disable all mitigations for
3354 the Spectre variant 2 (indirect branch prediction)
3355 vulnerability. System may allow data leaks with this
3358 nospec_store_bypass_disable
3359 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
3362 [PPC] Don't flush the L1-D cache after accessing user data.
3364 noxsave [BUGS=X86] Disables x86 extended register state save
3365 and restore using xsave. The kernel will fallback to
3366 enabling legacy floating-point and sse state.
3368 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
3369 register states. The kernel will fall back to use
3370 xsave to save the states. By using this parameter,
3371 performance of saving the states is degraded because
3372 xsave doesn't support modified optimization while
3373 xsaveopt supports it on xsaveopt enabled systems.
3375 noxsaves [X86] Disables xsaves and xrstors used in saving and
3376 restoring x86 extended register state in compacted
3377 form of xsave area. The kernel will fall back to use
3378 xsaveopt and xrstor to save and restore the states
3379 in standard form of xsave area. By using this
3380 parameter, xsave area per process might occupy more
3381 memory on xsaves enabled systems.
3383 nohlt [ARM,ARM64,MICROBLAZE,SH] Forces the kernel to busy wait
3384 in do_idle() and not use the arch_cpu_idle()
3385 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
3386 to be effective. This is useful on platforms where the
3387 sleep(SH) or wfi(ARM,ARM64) instructions do not work
3388 correctly or when doing power measurements to evalute
3389 the impact of the sleep instructions. This is also
3390 useful when using JTAG debugger.
3392 no_file_caps Tells the kernel not to honor file capabilities. The
3393 only way then for a file to be executed with privilege
3394 is to be setuid root or executed by root.
3396 nohalt [IA-64] Tells the kernel not to use the power saving
3397 function PAL_HALT_LIGHT when idle. This increases
3398 power-consumption. On the positive side, it reduces
3399 interrupt wake-up latency, which may improve performance
3400 in certain environments such as networked servers or
3404 Force pointers printed to the console or buffers to be
3405 unhashed. By default, when a pointer is printed via %p
3406 format string, that pointer is "hashed", i.e. obscured
3407 by hashing the pointer value. This is a security feature
3408 that hides actual kernel addresses from unprivileged
3409 users, but it also makes debugging the kernel more
3410 difficult since unequal pointers can no longer be
3411 compared. However, if this command-line option is
3412 specified, then all normal pointers will have their true
3413 value printed. Pointers printed via %pK may still be
3414 hashed. This option should only be specified when
3415 debugging the kernel. Please do not use on production
3418 nohibernate [HIBERNATION] Disable hibernation and resume.
3420 nohz= [KNL] Boottime enable/disable dynamic ticks
3421 Valid arguments: on, off
3424 nohz_full= [KNL,BOOT,SMP,ISOL]
3425 The argument is a cpu list, as described above.
3426 In kernels built with CONFIG_NO_HZ_FULL=y, set
3427 the specified list of CPUs whose tick will be stopped
3428 whenever possible. The boot CPU will be forced outside
3429 the range to maintain the timekeeping. Any CPUs
3430 in this list will have their RCU callbacks offloaded,
3431 just as if they had also been called out in the
3432 rcu_nocbs= boot parameter.
3434 noiotrap [SH] Disables trapped I/O port accesses.
3436 noirqdebug [X86-32] Disables the code which attempts to detect and
3437 disable unhandled interrupt sources.
3439 no_timer_check [X86,APIC] Disables the code which tests for
3440 broken timer IRQ sources.
3442 noisapnp [ISAPNP] Disables ISA PnP code.
3444 noinitrd [RAM] Tells the kernel not to load any configured
3447 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
3449 [Deprecated - use intremap=off]
3453 noinvpcid [X86] Disable the INVPCID cpu feature.
3455 nojitter [IA-64] Disables jitter checking for ITC timers.
3457 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
3459 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
3463 [X86,PV_OPS] Disable paravirtualized VMware scheduler
3464 clock and use the default one.
3466 no-steal-acc [X86,PV_OPS,ARM64] Disable paravirtualized steal time
3467 accounting. steal time is computed, but won't
3468 influence scheduler behaviour
3470 nolapic [X86-32,APIC] Do not enable or use the local APIC.
3472 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
3474 noltlbs [PPC] Do not use large page/tlb entries for kernel
3475 lowmem mapping on PPC40x and PPC8xx
3477 nomca [IA-64] Disable machine check abort handling
3479 nomce [X86-32] Disable Machine Check Exception
3481 nomfgpt [X86-32] Disable Multi-Function General Purpose
3482 Timer usage (for AMD Geode machines).
3484 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3485 shutdown the other cpus. Instead use the REBOOT_VECTOR
3488 nomodule Disable module load
3490 nopat [X86] Disable PAT (page attribute table extension of
3491 pagetables) support.
3493 nopcid [X86-64] Disable the PCID cpu feature.
3495 norandmaps Don't use address space randomization. Equivalent to
3496 echo 0 > /proc/sys/kernel/randomize_va_space
3498 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
3499 with UP alternatives
3501 nordrand [X86] Disable kernel use of the RDRAND and
3502 RDSEED instructions even if they are supported
3503 by the processor. RDRAND and RDSEED are still
3504 available to user space applications.
3506 noresume [SWSUSP] Disables resume and restores original swap
3509 no-scroll [VGA] Disables scrollback.
3510 This is required for the Braillex ib80-piezo Braille
3511 reader made by F.H. Papenmeier (Germany).
3515 nosep [BUGS=X86-32] Disables x86 SYSENTER/SYSEXIT support.
3517 nosgx [X86-64,SGX] Disables Intel SGX kernel support.
3519 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
3520 and disable the IO APIC. legacy for "maxcpus=0".
3522 nosoftlockup [KNL] Disable the soft-lockup detector.
3524 nosync [HW,M68K] Disables sync negotiation for all devices.
3526 nowatchdog [KNL] Disable both lockup detectors, i.e.
3527 soft-lockup and NMI watchdog (hard-lockup).
3531 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
3533 cpu0_hotplug [X86] Turn on CPU0 hotplug feature when
3534 CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off.
3535 Some features depend on CPU0. Known dependencies are:
3536 1. Resume from suspend/hibernate depends on CPU0.
3537 Suspend/hibernate will fail if CPU0 is offline and you
3538 need to online CPU0 before suspend/hibernate.
3539 2. PIC interrupts also depend on CPU0. CPU0 can't be
3540 removed if a PIC interrupt is detected.
3541 It's said poweroff/reboot may depend on CPU0 on some
3542 machines although I haven't seen such issues so far
3543 after CPU0 is offline on a few tested machines.
3544 If the dependencies are under your control, you can
3545 turn on cpu0_hotplug.
3547 nps_mtm_hs_ctr= [KNL,ARC]
3548 This parameter sets the maximum duration, in
3549 cycles, each HW thread of the CTOP can run
3550 without interruptions, before HW switches it.
3551 The actual maximum duration is 16 times this
3553 Format: integer between 1 and 255
3556 nptcg= [IA-64] Override max number of concurrent global TLB
3557 purges which is reported from either PAL_VM_SUMMARY or
3560 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
3561 could support. nr_cpus=n : n >= 1 limits the kernel to
3562 support 'n' processors. It could be larger than the
3563 number of already plugged CPU during bootup, later in
3564 runtime you can physically add extra cpu until it reaches
3565 n. So during boot up some boot time memory for per-cpu
3566 variables need be pre-allocated for later physical cpu
3569 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
3571 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86] Disable NUMA, Only
3572 set up a single NUMA node spanning all memory.
3574 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic
3576 Allowed values are enable and disable
3578 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
3579 'node', 'default' can be specified
3580 This can be set from sysctl after boot.
3581 See Documentation/admin-guide/sysctl/vm.rst for details.
3583 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
3584 See Documentation/core-api/debugging-via-ohci1394.rst for more
3587 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
3588 Rather than timing out after 20 ms if an EC
3589 command is not properly ACKed, override the length
3590 of the timeout. We have interrupts disabled while
3591 waiting for the ACK, so if this is set too high
3592 interrupts *may* be lost!
3594 omap_mux= [OMAP] Override bootloader pin multiplexing.
3595 Format: <mux_mode0.mode_name=value>...
3596 For example, to override I2C bus2:
3597 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
3599 oops=panic Always panic on oopses. Default is to just kill the
3600 process, but there is a small probability of
3601 deadlocking the machine.
3602 This will also cause panics on machine check exceptions.
3603 Useful together with panic=30 to trigger a reboot.
3606 [KNL] Boolean flag to control whether the page allocator
3607 should randomize its free lists. The randomization may
3608 be automatically enabled if the kernel detects it is
3609 running on a platform with a direct-mapped memory-side
3610 cache, and this parameter can be used to
3611 override/disable that behavior. The state of the flag
3612 can be read from sysfs at:
3613 /sys/module/page_alloc/parameters/shuffle.
3615 page_owner= [KNL] Boot-time page_owner enabling option.
3616 Storage of the information about who allocated
3617 each page is disabled in default. With this switch,
3619 on: enable the feature
3621 page_poison= [KNL] Boot-time parameter changing the state of
3622 poisoning on the buddy allocator, available with
3623 CONFIG_PAGE_POISONING=y.
3624 off: turn off poisoning (default)
3625 on: turn on poisoning
3627 page_reporting.page_reporting_order=
3628 [KNL] Minimal page reporting order
3630 Adjust the minimal page reporting order. The page
3631 reporting is disabled when it exceeds (MAX_ORDER-1).
3633 panic= [KNL] Kernel behaviour on panic: delay <timeout>
3634 timeout > 0: seconds before rebooting
3635 timeout = 0: wait forever
3636 timeout < 0: reboot immediately
3639 panic_print= Bitmask for printing system info when panic happens.
3640 User can chose combination of the following bits:
3641 bit 0: print all tasks info
3642 bit 1: print system memory info
3643 bit 2: print timer info
3644 bit 3: print locks info if CONFIG_LOCKDEP is on
3645 bit 4: print ftrace buffer
3646 bit 5: print all printk messages in buffer
3648 panic_on_taint= Bitmask for conditionally calling panic() in add_taint()
3649 Format: <hex>[,nousertaint]
3650 Hexadecimal bitmask representing the set of TAINT flags
3651 that will cause the kernel to panic when add_taint() is
3652 called with any of the flags in this set.
3653 The optional switch "nousertaint" can be utilized to
3654 prevent userspace forced crashes by writing to sysctl
3655 /proc/sys/kernel/tainted any flagset matching with the
3656 bitmask set on panic_on_taint.
3657 See Documentation/admin-guide/tainted-kernels.rst for
3658 extra details on the taint flags that users can pick
3659 to compose the bitmask to assign to panic_on_taint.
3661 panic_on_warn panic() instead of WARN(). Useful to cause kdump
3664 crash_kexec_post_notifiers
3665 Run kdump after running panic-notifiers and dumping
3666 kmsg. This only for the users who doubt kdump always
3667 succeeds in any situation.
3668 Note that this also increases risks of kdump failure,
3669 because some panic notifiers can make the crashed
3670 kernel more unstable.
3672 parkbd.port= [HW] Parallel port number the keyboard adapter is
3673 connected to, default is 0.
3675 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
3676 0 for XT, 1 for AT (default is AT).
3679 parport= [HW,PPT] Specify parallel ports. 0 disables.
3680 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
3681 Use 'auto' to force the driver to use any
3682 IRQ/DMA settings detected (the default is to
3683 ignore detected IRQ/DMA settings because of
3684 possible conflicts). You can specify the base
3685 address, IRQ, and DMA settings; IRQ and DMA
3686 should be numbers, or 'auto' (for using detected
3687 settings on that particular port), or 'nofifo'
3688 (to avoid using a FIFO even if it is detected).
3689 Parallel ports are assigned in the order they
3690 are specified on the command line, starting
3693 parport_init_mode= [HW,PPT]
3694 Configure VIA parallel port to operate in
3695 a specific mode. This is necessary on Pegasos
3696 computer where firmware has no options for setting
3697 up parallel port mode and sets it to spp.
3698 Currently this function knows 686a and 8231 chips.
3699 Format: [spp|ps2|epp|ecp|ecpepp]
3701 pata_legacy.all= [HW,LIBATA]
3703 Set to non-zero to probe primary and secondary ISA
3704 port ranges on PCI systems where no PCI PATA device
3705 has been found at either range. Disabled by default.
3707 pata_legacy.autospeed= [HW,LIBATA]
3709 Set to non-zero if a chip is present that snoops speed
3710 changes. Disabled by default.
3712 pata_legacy.ht6560a= [HW,LIBATA]
3714 Set to 1, 2, or 3 for HT 6560A on the primary channel,
3715 the secondary channel, or both channels respectively.
3716 Disabled by default.
3718 pata_legacy.ht6560b= [HW,LIBATA]
3720 Set to 1, 2, or 3 for HT 6560B on the primary channel,
3721 the secondary channel, or both channels respectively.
3722 Disabled by default.
3724 pata_legacy.iordy_mask= [HW,LIBATA]
3726 IORDY enable mask. Set individual bits to allow IORDY
3727 for the respective channel. Bit 0 is for the first
3728 legacy channel handled by this driver, bit 1 is for
3729 the second channel, and so on. The sequence will often
3730 correspond to the primary legacy channel, the secondary
3731 legacy channel, and so on, but the handling of a PCI
3732 bus and the use of other driver options may interfere
3733 with the sequence. By default IORDY is allowed across
3736 pata_legacy.opti82c46x= [HW,LIBATA]
3738 Set to 1, 2, or 3 for Opti 82c611A on the primary
3739 channel, the secondary channel, or both channels
3740 respectively. Disabled by default.
3742 pata_legacy.opti82c611a= [HW,LIBATA]
3744 Set to 1, 2, or 3 for Opti 82c465MV on the primary
3745 channel, the secondary channel, or both channels
3746 respectively. Disabled by default.
3748 pata_legacy.pio_mask= [HW,LIBATA]
3750 PIO mode mask for autospeed devices. Set individual
3751 bits to allow the use of the respective PIO modes.
3752 Bit 0 is for mode 0, bit 1 is for mode 1, and so on.
3753 All modes allowed by default.
3755 pata_legacy.probe_all= [HW,LIBATA]
3757 Set to non-zero to probe tertiary and further ISA
3758 port ranges on PCI systems. Disabled by default.
3760 pata_legacy.probe_mask= [HW,LIBATA]
3762 Probe mask for legacy ISA PATA ports. Depending on
3763 platform configuration and the use of other driver
3764 options up to 6 legacy ports are supported: 0x1f0,
3765 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing
3766 of individual ports can be disabled by setting the
3767 corresponding bits in the mask to 1. Bit 0 is for
3768 the first port in the list above (0x1f0), and so on.
3769 By default all supported ports are probed.
3771 pata_legacy.qdi= [HW,LIBATA]
3773 Set to non-zero to probe QDI controllers. By default
3774 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise.
3776 pata_legacy.winbond= [HW,LIBATA]
3778 Set to non-zero to probe Winbond controllers. Use
3779 the standard I/O port (0x130) if 1, otherwise the
3780 value given is the I/O port to use (typically 0x1b0).
3781 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE,
3784 pata_platform.pio_mask= [HW,LIBATA]
3786 Supported PIO mode mask. Set individual bits to allow
3787 the use of the respective PIO modes. Bit 0 is for
3788 mode 0, bit 1 is for mode 1, and so on. Mode 0 only
3792 Halt all CPUs after the first oops has been printed for
3793 the specified number of seconds. This is to be used if
3794 your oopses keep scrolling off the screen.
3799 See header of drivers/block/paride/pcd.c.
3800 See also Documentation/admin-guide/blockdev/paride.rst.
3802 pci=option[,option...] [PCI] various PCI subsystem options.
3804 Some options herein operate on a specific device
3805 or a set of devices (<pci_dev>). These are
3806 specified in one of the following formats:
3808 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
3809 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
3811 Note: the first format specifies a PCI
3812 bus/device/function address which may change
3813 if new hardware is inserted, if motherboard
3814 firmware changes, or due to changes caused
3815 by other kernel parameters. If the
3816 domain is left unspecified, it is
3817 taken to be zero. Optionally, a path
3818 to a device through multiple device/function
3819 addresses can be specified after the base
3820 address (this is more robust against
3821 renumbering issues). The second format
3822 selects devices using IDs from the
3823 configuration space which may match multiple
3824 devices in the system.
3826 earlydump dump PCI config space before the kernel
3828 off [X86] don't probe for the PCI bus
3829 bios [X86-32] force use of PCI BIOS, don't access
3830 the hardware directly. Use this if your machine
3831 has a non-standard PCI host bridge.
3832 nobios [X86-32] disallow use of PCI BIOS, only direct
3833 hardware access methods are allowed. Use this
3834 if you experience crashes upon bootup and you
3835 suspect they are caused by the BIOS.
3836 conf1 [X86] Force use of PCI Configuration Access
3837 Mechanism 1 (config address in IO port 0xCF8,
3838 data in IO port 0xCFC, both 32-bit).
3839 conf2 [X86] Force use of PCI Configuration Access
3840 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
3841 the function, IO port 0xCFA, also 8-bit, sets
3842 bus number. The config space is then accessed
3843 through ports 0xC000-0xCFFF).
3844 See http://wiki.osdev.org/PCI for more info
3845 on the configuration access mechanisms.
3846 noaer [PCIE] If the PCIEAER kernel config parameter is
3847 enabled, this kernel boot option can be used to
3848 disable the use of PCIE advanced error reporting.
3849 nodomains [PCI] Disable support for multiple PCI
3850 root domains (aka PCI segments, in ACPI-speak).
3851 nommconf [X86] Disable use of MMCONFIG for PCI
3853 check_enable_amd_mmconf [X86] check for and enable
3854 properly configured MMIO access to PCI
3855 config space on AMD family 10h CPU
3856 nomsi [MSI] If the PCI_MSI kernel config parameter is
3857 enabled, this kernel boot option can be used to
3858 disable the use of MSI interrupts system-wide.
3859 noioapicquirk [APIC] Disable all boot interrupt quirks.
3860 Safety option to keep boot IRQs enabled. This
3861 should never be necessary.
3862 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
3863 primary IO-APIC for bridges that cannot disable
3864 boot IRQs. This fixes a source of spurious IRQs
3865 when the system masks IRQs.
3866 noioapicreroute [APIC] Disable workaround that uses the
3867 boot IRQ equivalent of an IRQ that connects to
3868 a chipset where boot IRQs cannot be disabled.
3869 The opposite of ioapicreroute.
3870 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
3871 routing table. These calls are known to be buggy
3872 on several machines and they hang the machine
3873 when used, but on other computers it's the only
3874 way to get the interrupt routing table. Try
3875 this option if the kernel is unable to allocate
3876 IRQs or discover secondary PCI buses on your
3878 rom [X86] Assign address space to expansion ROMs.
3879 Use with caution as certain devices share
3880 address decoders between ROMs and other
3882 norom [X86] Do not assign address space to
3883 expansion ROMs that do not already have
3884 BIOS assigned address ranges.
3885 nobar [X86] Do not assign address space to the
3886 BARs that weren't assigned by the BIOS.
3887 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
3888 assigned automatically to PCI devices. You can
3889 make the kernel exclude IRQs of your ISA cards
3891 pirqaddr=0xAAAAA [X86] Specify the physical address
3892 of the PIRQ table (normally generated
3893 by the BIOS) if it is outside the
3894 F0000h-100000h range.
3895 lastbus=N [X86] Scan all buses thru bus #N. Can be
3896 useful if the kernel is unable to find your
3897 secondary buses and you want to tell it
3898 explicitly which ones they are.
3899 assign-busses [X86] Always assign all PCI bus
3900 numbers ourselves, overriding
3901 whatever the firmware may have done.
3902 usepirqmask [X86] Honor the possible IRQ mask stored
3903 in the BIOS $PIR table. This is needed on
3904 some systems with broken BIOSes, notably
3905 some HP Pavilion N5400 and Omnibook XE3
3906 notebooks. This will have no effect if ACPI
3907 IRQ routing is enabled.
3908 noacpi [X86] Do not use ACPI for IRQ routing
3909 or for PCI scanning.
3910 use_crs [X86] Use PCI host bridge window information
3911 from ACPI. On BIOSes from 2008 or later, this
3912 is enabled by default. If you need to use this,
3913 please report a bug.
3914 nocrs [X86] Ignore PCI host bridge windows from ACPI.
3915 If you need to use this, please report a bug.
3916 routeirq Do IRQ routing for all PCI devices.
3917 This is normally done in pci_enable_device(),
3918 so this option is a temporary workaround
3919 for broken drivers that don't call it.
3920 skip_isa_align [X86] do not align io start addr, so can
3921 handle more pci cards
3922 noearly [X86] Don't do any early type 1 scanning.
3923 This might help on some broken boards which
3924 machine check when some devices' config space
3925 is read. But various workarounds are disabled
3926 and some IOMMU drivers will not work.
3927 bfsort Sort PCI devices into breadth-first order.
3928 This sorting is done to get a device
3929 order compatible with older (<= 2.4) kernels.
3930 nobfsort Don't sort PCI devices into breadth-first order.
3931 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
3932 tuning and use the BIOS-configured MPS defaults.
3933 pcie_bus_safe Set every device's MPS to the largest value
3934 supported by all devices below the root complex.
3935 pcie_bus_perf Set device MPS to the largest allowable MPS
3936 based on its parent bus. Also set MRRS (Max
3937 Read Request Size) to the largest supported
3938 value (no larger than the MPS that the device
3939 or bus can support) for best performance.
3940 pcie_bus_peer2peer Set every device's MPS to 128B, which
3941 every device is guaranteed to support. This
3942 configuration allows peer-to-peer DMA between
3943 any pair of devices, possibly at the cost of
3944 reduced performance. This also guarantees
3945 that hot-added devices will work.
3946 cbiosize=nn[KMG] The fixed amount of bus space which is
3947 reserved for the CardBus bridge's IO window.
3948 The default value is 256 bytes.
3949 cbmemsize=nn[KMG] The fixed amount of bus space which is
3950 reserved for the CardBus bridge's memory
3951 window. The default value is 64 megabytes.
3954 [<order of align>@]<pci_dev>[; ...]
3955 Specifies alignment and device to reassign
3956 aligned memory resources. How to
3957 specify the device is described above.
3958 If <order of align> is not specified,
3959 PAGE_SIZE is used as alignment.
3960 A PCI-PCI bridge can be specified if resource
3961 windows need to be expanded.
3962 To specify the alignment for several
3963 instances of a device, the PCI vendor,
3964 device, subvendor, and subdevice may be
3965 specified, e.g., 12@pci:8086:9c22:103c:198f
3966 for 4096-byte alignment.
3967 ecrc= Enable/disable PCIe ECRC (transaction layer
3968 end-to-end CRC checking).
3969 bios: Use BIOS/firmware settings. This is the
3973 hpiosize=nn[KMG] The fixed amount of bus space which is
3974 reserved for hotplug bridge's IO window.
3975 Default size is 256 bytes.
3976 hpmmiosize=nn[KMG] The fixed amount of bus space which is
3977 reserved for hotplug bridge's MMIO window.
3978 Default size is 2 megabytes.
3979 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
3980 reserved for hotplug bridge's MMIO_PREF window.
3981 Default size is 2 megabytes.
3982 hpmemsize=nn[KMG] The fixed amount of bus space which is
3983 reserved for hotplug bridge's MMIO and
3985 Default size is 2 megabytes.
3986 hpbussize=nn The minimum amount of additional bus numbers
3987 reserved for buses below a hotplug bridge.
3989 realloc= Enable/disable reallocating PCI bridge resources
3990 if allocations done by BIOS are too small to
3991 accommodate resources required by all child
3993 off: Turn realloc off
3995 realloc same as realloc=on
3996 noari do not use PCIe ARI.
3997 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
3998 do not use PCIe ATS (and IOMMU device IOTLB).
3999 pcie_scan_all Scan all possible PCIe devices. Otherwise we
4000 only look for one device below a PCIe downstream
4002 big_root_window Try to add a big 64bit memory window to the PCIe
4003 root complex on AMD CPUs. Some GFX hardware
4004 can resize a BAR to allow access to all VRAM.
4005 Adding the window is slightly risky (it may
4006 conflict with unreported devices), so this
4008 disable_acs_redir=<pci_dev>[; ...]
4009 Specify one or more PCI devices (in the format
4010 specified above) separated by semicolons.
4011 Each device specified will have the PCI ACS
4012 redirect capabilities forced off which will
4013 allow P2P traffic between devices through
4014 bridges without forcing it upstream. Note:
4015 this removes isolation between devices and
4016 may put more devices in an IOMMU group.
4017 force_floating [S390] Force usage of floating interrupts.
4018 nomio [S390] Do not use MIO instructions.
4019 norid [S390] ignore the RID field and force use of
4020 one PCI domain per PCI function
4022 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
4025 force Enable ASPM even on devices that claim not to support it.
4026 WARNING: Forcing ASPM on may cause system lockups.
4028 pcie_ports= [PCIE] PCIe port services handling:
4029 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
4030 even if the platform doesn't give the OS permission to
4031 use them. This may cause conflicts if the platform
4032 also tries to use these services.
4033 dpc-native Use native PCIe service for DPC only. May
4034 cause conflicts if firmware uses AER or DPC.
4035 compat Disable native PCIe services (PME, AER, DPC, PCIe
4038 pcie_port_pm= [PCIE] PCIe port power management handling:
4039 off Disable power management of all PCIe ports
4040 force Forcibly enable power management of all PCIe ports
4042 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
4043 nomsi Do not use MSI for native PCIe PME signaling (this makes
4044 all PCIe root ports use INTx for all services).
4046 pcmv= [HW,PCMCIA] BadgePAD 4
4050 Keep all power-domains already enabled by bootloader on,
4051 even if no driver has claimed them. This is useful
4052 for debug and development, but should not be
4053 needed on a platform with proper driver support.
4056 See Documentation/admin-guide/blockdev/paride.rst.
4058 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
4061 See arch/parisc/kernel/pdc_chassis.c
4063 percpu_alloc= Select which percpu first chunk allocator to use.
4064 Currently supported values are "embed" and "page".
4065 Archs may support subset or none of the selections.
4066 See comments in mm/percpu.c for details on each
4067 allocator. This parameter is primarily for debugging
4068 and performance comparison.
4071 See Documentation/admin-guide/blockdev/paride.rst.
4074 See Documentation/admin-guide/blockdev/paride.rst.
4076 pirq= [SMP,APIC] Manual mp-table setup
4077 See Documentation/x86/i386/IO-APIC.rst.
4079 plip= [PPT,NET] Parallel port network link
4080 Format: { parport<nr> | timid | 0 }
4081 See also Documentation/admin-guide/parport.rst.
4083 pmtmr= [X86] Manual setup of pmtmr I/O Port.
4084 Override pmtimer IOPort with a hex value.
4087 pm_debug_messages [SUSPEND,KNL]
4088 Enable suspend/resume debug messages during boot up.
4091 Enable PNP debug messages (depends on the
4092 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
4093 via /sys/module/pnp/parameters/debug. We always show
4094 current resource usage; turning this on also shows
4095 possible settings and some assignment information.
4101 { on | off | curr | res | no-curr | no-res }
4104 [ISAPNP] Exclude IRQs for the autoconfiguration
4107 [ISAPNP] Exclude DMAs for the autoconfiguration
4109 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
4110 Ranges are in pairs (I/O port base and size).
4113 [ISAPNP] Exclude memory regions for the
4115 Ranges are in pairs (memory base and size).
4117 ports= [IP_VS_FTP] IPVS ftp helper module
4119 Up to 8 (IP_VS_APP_MAX_PORTS) ports
4121 Format: <port>,<port>....
4123 powersave=off [PPC] This option disables power saving features.
4124 It specifically disables cpuidle and sets the
4125 platform machine description specific power_save
4126 function to NULL. On Idle the CPU just reduces
4129 ppc_strict_facility_enable
4130 [PPC] This option catches any kernel floating point,
4131 Altivec, VSX and SPE outside of regions specifically
4132 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
4133 There is some performance impact when enabling this.
4137 Disable Hardware Transactional Memory
4140 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC
4141 none - Limited to cond_resched() calls
4142 voluntary - Limited to cond_resched() and might_sleep() calls
4143 full - Any section that isn't explicitly preempt disabled
4144 can be preempted anytime.
4146 print-fatal-signals=
4147 [KNL] debug: print fatal signals
4149 If enabled, warn about various signal handling
4150 related application anomalies: too many signals,
4151 too many POSIX.1 timers, fatal signals causing a
4154 If you hit the warning due to signal overflow,
4155 you might want to try "ulimit -i unlimited".
4159 printk.always_kmsg_dump=
4160 Trigger kmsg_dump for cases other than kernel oops or
4162 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4165 printk.devkmsg={on,off,ratelimit}
4166 Control writing to /dev/kmsg.
4167 on - unlimited logging to /dev/kmsg from userspace
4168 off - logging to /dev/kmsg disabled
4169 ratelimit - ratelimit the logging
4172 printk.time= Show timing data prefixed to each printk message line
4173 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4175 processor.max_cstate= [HW,ACPI]
4176 Limit processor to maximum C-state
4177 max_cstate=9 overrides any DMI blacklist limit.
4179 processor.nocst [HW,ACPI]
4180 Ignore the _CST method to determine C-states,
4181 instead using the legacy FADT method
4183 profile= [KNL] Enable kernel profiling via /proc/profile
4184 Format: [<profiletype>,]<number>
4185 Param: <profiletype>: "schedule", "sleep", or "kvm"
4186 [defaults to kernel profiling]
4187 Param: "schedule" - profile schedule points.
4188 Param: "sleep" - profile D-state sleeping (millisecs).
4189 Requires CONFIG_SCHEDSTATS
4190 Param: "kvm" - profile VM exits.
4191 Param: <number> - step/bucket size as a power of 2 for
4192 statistical time based profiling.
4194 prompt_ramdisk= [RAM] [Deprecated]
4196 prot_virt= [S390] enable hosting protected virtual machines
4197 isolated from the hypervisor (if hardware supports
4201 psi= [KNL] Enable or disable pressure stall information
4205 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
4206 probe for; one of (bare|imps|exps|lifebook|any).
4207 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
4209 psmouse.resetafter= [HW,MOUSE]
4210 Try to reset the device after so many bad packets
4213 [HW,MOUSE] Set desired mouse resolution, in dpi.
4214 psmouse.smartscroll=
4215 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
4216 0 = disabled, 1 = enabled (default).
4218 pstore.backend= Specify the name of the pstore backend to use
4221 See Documentation/admin-guide/blockdev/paride.rst.
4223 pti= [X86-64] Control Page Table Isolation of user and
4224 kernel address spaces. Disabling this feature
4225 removes hardening, but improves performance of
4226 system calls and interrupts.
4228 on - unconditionally enable
4229 off - unconditionally disable
4230 auto - kernel detects whether your CPU model is
4231 vulnerable to issues that PTI mitigates
4233 Not specifying this option is equivalent to pti=auto.
4236 Equivalent to pti=off
4239 [KNL] Number of legacy pty's. Overwrites compiled-in
4242 quiet [KNL] Disable most log messages
4247 See Documentation/admin-guide/md.rst.
4249 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
4250 See Documentation/admin-guide/blockdev/ramdisk.rst.
4252 ramdisk_start= [RAM] RAM disk image start address
4254 random.trust_cpu={on,off}
4255 [KNL] Enable or disable trusting the use of the
4256 CPU's random number generator (if available) to
4257 fully seed the kernel's CRNG. Default is controlled
4258 by CONFIG_RANDOM_TRUST_CPU.
4260 randomize_kstack_offset=
4261 [KNL] Enable or disable kernel stack offset
4262 randomization, which provides roughly 5 bits of
4263 entropy, frustrating memory corruption attacks
4264 that depend on stack address determinism or
4265 cross-syscall address exposures. This is only
4266 available on architectures that have defined
4267 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET.
4268 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4269 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT.
4271 ras=option[,option,...] [KNL] RAS-specific options
4274 Disable the Correctable Errors Collector,
4275 see CONFIG_RAS_CEC help text.
4278 The argument is a cpu list, as described above.
4280 In kernels built with CONFIG_RCU_NOCB_CPU=y, set
4281 the specified list of CPUs to be no-callback CPUs.
4282 Invocation of these CPUs' RCU callbacks will be
4283 offloaded to "rcuox/N" kthreads created for that
4284 purpose, where "x" is "p" for RCU-preempt, and
4285 "s" for RCU-sched, and "N" is the CPU number.
4286 This reduces OS jitter on the offloaded CPUs,
4287 which can be useful for HPC and real-time
4288 workloads. It can also improve energy efficiency
4289 for asymmetric multiprocessors.
4292 Rather than requiring that offloaded CPUs
4293 (specified by rcu_nocbs= above) explicitly
4294 awaken the corresponding "rcuoN" kthreads,
4295 make these kthreads poll for callbacks.
4296 This improves the real-time response for the
4297 offloaded CPUs by relieving them of the need to
4298 wake up the corresponding kthread, but degrades
4299 energy efficiency by requiring that the kthreads
4300 periodically wake up to do the polling.
4302 rcutree.blimit= [KNL]
4303 Set maximum number of finished RCU callbacks to
4304 process in one batch.
4306 rcutree.dump_tree= [KNL]
4307 Dump the structure of the rcu_node combining tree
4308 out at early boot. This is used for diagnostic
4309 purposes, to verify correct tree setup.
4311 rcutree.gp_cleanup_delay= [KNL]
4312 Set the number of jiffies to delay each step of
4313 RCU grace-period cleanup.
4315 rcutree.gp_init_delay= [KNL]
4316 Set the number of jiffies to delay each step of
4317 RCU grace-period initialization.
4319 rcutree.gp_preinit_delay= [KNL]
4320 Set the number of jiffies to delay each step of
4321 RCU grace-period pre-initialization, that is,
4322 the propagation of recent CPU-hotplug changes up
4323 the rcu_node combining tree.
4325 rcutree.use_softirq= [KNL]
4326 If set to zero, move all RCU_SOFTIRQ processing to
4327 per-CPU rcuc kthreads. Defaults to a non-zero
4328 value, meaning that RCU_SOFTIRQ is used by default.
4329 Specify rcutree.use_softirq=0 to use rcuc kthreads.
4331 But note that CONFIG_PREEMPT_RT=y kernels disable
4332 this kernel boot parameter, forcibly setting it
4335 rcutree.rcu_fanout_exact= [KNL]
4336 Disable autobalancing of the rcu_node combining
4337 tree. This is used by rcutorture, and might
4338 possibly be useful for architectures having high
4339 cache-to-cache transfer latencies.
4341 rcutree.rcu_fanout_leaf= [KNL]
4342 Change the number of CPUs assigned to each
4343 leaf rcu_node structure. Useful for very
4344 large systems, which will choose the value 64,
4345 and for NUMA systems with large remote-access
4346 latencies, which will choose a value aligned
4347 with the appropriate hardware boundaries.
4349 rcutree.rcu_min_cached_objs= [KNL]
4350 Minimum number of objects which are cached and
4351 maintained per one CPU. Object size is equal
4352 to PAGE_SIZE. The cache allows to reduce the
4353 pressure to page allocator, also it makes the
4354 whole algorithm to behave better in low memory
4357 rcutree.rcu_delay_page_cache_fill_msec= [KNL]
4358 Set the page-cache refill delay (in milliseconds)
4359 in response to low-memory conditions. The range
4360 of permitted values is in the range 0:100000.
4362 rcutree.jiffies_till_first_fqs= [KNL]
4363 Set delay from grace-period initialization to
4364 first attempt to force quiescent states.
4365 Units are jiffies, minimum value is zero,
4366 and maximum value is HZ.
4368 rcutree.jiffies_till_next_fqs= [KNL]
4369 Set delay between subsequent attempts to force
4370 quiescent states. Units are jiffies, minimum
4371 value is one, and maximum value is HZ.
4373 rcutree.jiffies_till_sched_qs= [KNL]
4374 Set required age in jiffies for a
4375 given grace period before RCU starts
4376 soliciting quiescent-state help from
4377 rcu_note_context_switch() and cond_resched().
4378 If not specified, the kernel will calculate
4379 a value based on the most recent settings
4380 of rcutree.jiffies_till_first_fqs
4381 and rcutree.jiffies_till_next_fqs.
4382 This calculated value may be viewed in
4383 rcutree.jiffies_to_sched_qs. Any attempt to set
4384 rcutree.jiffies_to_sched_qs will be cheerfully
4387 rcutree.kthread_prio= [KNL,BOOT]
4388 Set the SCHED_FIFO priority of the RCU per-CPU
4389 kthreads (rcuc/N). This value is also used for
4390 the priority of the RCU boost threads (rcub/N)
4391 and for the RCU grace-period kthreads (rcu_bh,
4392 rcu_preempt, and rcu_sched). If RCU_BOOST is
4393 set, valid values are 1-99 and the default is 1
4394 (the least-favored priority). Otherwise, when
4395 RCU_BOOST is not set, valid values are 0-99 and
4396 the default is zero (non-realtime operation).
4398 rcutree.rcu_nocb_gp_stride= [KNL]
4399 Set the number of NOCB callback kthreads in
4400 each group, which defaults to the square root
4401 of the number of CPUs. Larger numbers reduce
4402 the wakeup overhead on the global grace-period
4403 kthread, but increases that same overhead on
4404 each group's NOCB grace-period kthread.
4406 rcutree.qhimark= [KNL]
4407 Set threshold of queued RCU callbacks beyond which
4408 batch limiting is disabled.
4410 rcutree.qlowmark= [KNL]
4411 Set threshold of queued RCU callbacks below which
4412 batch limiting is re-enabled.
4414 rcutree.qovld= [KNL]
4415 Set threshold of queued RCU callbacks beyond which
4416 RCU's force-quiescent-state scan will aggressively
4417 enlist help from cond_resched() and sched IPIs to
4418 help CPUs more quickly reach quiescent states.
4419 Set to less than zero to make this be set based
4420 on rcutree.qhimark at boot time and to zero to
4421 disable more aggressive help enlistment.
4423 rcutree.rcu_idle_gp_delay= [KNL]
4424 Set wakeup interval for idle CPUs that have
4425 RCU callbacks (RCU_FAST_NO_HZ=y).
4427 rcutree.rcu_kick_kthreads= [KNL]
4428 Cause the grace-period kthread to get an extra
4429 wake_up() if it sleeps three times longer than
4430 it should at force-quiescent-state time.
4431 This wake_up() will be accompanied by a
4432 WARN_ONCE() splat and an ftrace_dump().
4434 rcutree.rcu_unlock_delay= [KNL]
4435 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels,
4436 this specifies an rcu_read_unlock()-time delay
4437 in microseconds. This defaults to zero.
4438 Larger delays increase the probability of
4439 catching RCU pointer leaks, that is, buggy use
4440 of RCU-protected pointers after the relevant
4441 rcu_read_unlock() has completed.
4443 rcutree.sysrq_rcu= [KNL]
4444 Commandeer a sysrq key to dump out Tree RCU's
4445 rcu_node tree with an eye towards determining
4446 why a new grace period has not yet started.
4448 rcuscale.gp_async= [KNL]
4449 Measure performance of asynchronous
4450 grace-period primitives such as call_rcu().
4452 rcuscale.gp_async_max= [KNL]
4453 Specify the maximum number of outstanding
4454 callbacks per writer thread. When a writer
4455 thread exceeds this limit, it invokes the
4456 corresponding flavor of rcu_barrier() to allow
4457 previously posted callbacks to drain.
4459 rcuscale.gp_exp= [KNL]
4460 Measure performance of expedited synchronous
4461 grace-period primitives.
4463 rcuscale.holdoff= [KNL]
4464 Set test-start holdoff period. The purpose of
4465 this parameter is to delay the start of the
4466 test until boot completes in order to avoid
4469 rcuscale.kfree_rcu_test= [KNL]
4470 Set to measure performance of kfree_rcu() flooding.
4472 rcuscale.kfree_rcu_test_double= [KNL]
4473 Test the double-argument variant of kfree_rcu().
4474 If this parameter has the same value as
4475 rcuscale.kfree_rcu_test_single, both the single-
4476 and double-argument variants are tested.
4478 rcuscale.kfree_rcu_test_single= [KNL]
4479 Test the single-argument variant of kfree_rcu().
4480 If this parameter has the same value as
4481 rcuscale.kfree_rcu_test_double, both the single-
4482 and double-argument variants are tested.
4484 rcuscale.kfree_nthreads= [KNL]
4485 The number of threads running loops of kfree_rcu().
4487 rcuscale.kfree_alloc_num= [KNL]
4488 Number of allocations and frees done in an iteration.
4490 rcuscale.kfree_loops= [KNL]
4491 Number of loops doing rcuscale.kfree_alloc_num number
4492 of allocations and frees.
4494 rcuscale.nreaders= [KNL]
4495 Set number of RCU readers. The value -1 selects
4496 N, where N is the number of CPUs. A value
4497 "n" less than -1 selects N-n+1, where N is again
4498 the number of CPUs. For example, -2 selects N
4499 (the number of CPUs), -3 selects N+1, and so on.
4500 A value of "n" less than or equal to -N selects
4503 rcuscale.nwriters= [KNL]
4504 Set number of RCU writers. The values operate
4505 the same as for rcuscale.nreaders.
4506 N, where N is the number of CPUs
4508 rcuscale.perf_type= [KNL]
4509 Specify the RCU implementation to test.
4511 rcuscale.shutdown= [KNL]
4512 Shut the system down after performance tests
4513 complete. This is useful for hands-off automated
4516 rcuscale.verbose= [KNL]
4517 Enable additional printk() statements.
4519 rcuscale.writer_holdoff= [KNL]
4520 Write-side holdoff between grace periods,
4521 in microseconds. The default of zero says
4524 rcutorture.fqs_duration= [KNL]
4525 Set duration of force_quiescent_state bursts
4528 rcutorture.fqs_holdoff= [KNL]
4529 Set holdoff time within force_quiescent_state bursts
4532 rcutorture.fqs_stutter= [KNL]
4533 Set wait time between force_quiescent_state bursts
4536 rcutorture.fwd_progress= [KNL]
4537 Enable RCU grace-period forward-progress testing
4538 for the types of RCU supporting this notion.
4540 rcutorture.fwd_progress_div= [KNL]
4541 Specify the fraction of a CPU-stall-warning
4542 period to do tight-loop forward-progress testing.
4544 rcutorture.fwd_progress_holdoff= [KNL]
4545 Number of seconds to wait between successive
4546 forward-progress tests.
4548 rcutorture.fwd_progress_need_resched= [KNL]
4549 Enclose cond_resched() calls within checks for
4550 need_resched() during tight-loop forward-progress
4553 rcutorture.gp_cond= [KNL]
4554 Use conditional/asynchronous update-side
4555 primitives, if available.
4557 rcutorture.gp_exp= [KNL]
4558 Use expedited update-side primitives, if available.
4560 rcutorture.gp_normal= [KNL]
4561 Use normal (non-expedited) asynchronous
4562 update-side primitives, if available.
4564 rcutorture.gp_sync= [KNL]
4565 Use normal (non-expedited) synchronous
4566 update-side primitives, if available. If all
4567 of rcutorture.gp_cond=, rcutorture.gp_exp=,
4568 rcutorture.gp_normal=, and rcutorture.gp_sync=
4569 are zero, rcutorture acts as if is interpreted
4570 they are all non-zero.
4572 rcutorture.irqreader= [KNL]
4573 Run RCU readers from irq handlers, or, more
4574 accurately, from a timer handler. Not all RCU
4575 flavors take kindly to this sort of thing.
4577 rcutorture.leakpointer= [KNL]
4578 Leak an RCU-protected pointer out of the reader.
4579 This can of course result in splats, and is
4580 intended to test the ability of things like
4581 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect
4584 rcutorture.n_barrier_cbs= [KNL]
4585 Set callbacks/threads for rcu_barrier() testing.
4587 rcutorture.nfakewriters= [KNL]
4588 Set number of concurrent RCU writers. These just
4589 stress RCU, they don't participate in the actual
4590 test, hence the "fake".
4592 rcutorture.nocbs_nthreads= [KNL]
4593 Set number of RCU callback-offload togglers.
4594 Zero (the default) disables toggling.
4596 rcutorture.nocbs_toggle= [KNL]
4597 Set the delay in milliseconds between successive
4598 callback-offload toggling attempts.
4600 rcutorture.nreaders= [KNL]
4601 Set number of RCU readers. The value -1 selects
4602 N-1, where N is the number of CPUs. A value
4603 "n" less than -1 selects N-n-2, 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.
4607 rcutorture.object_debug= [KNL]
4608 Enable debug-object double-call_rcu() testing.
4610 rcutorture.onoff_holdoff= [KNL]
4611 Set time (s) after boot for CPU-hotplug testing.
4613 rcutorture.onoff_interval= [KNL]
4614 Set time (jiffies) between CPU-hotplug operations,
4615 or zero to disable CPU-hotplug testing.
4617 rcutorture.read_exit= [KNL]
4618 Set the number of read-then-exit kthreads used
4619 to test the interaction of RCU updaters and
4620 task-exit processing.
4622 rcutorture.read_exit_burst= [KNL]
4623 The number of times in a given read-then-exit
4624 episode that a set of read-then-exit kthreads
4627 rcutorture.read_exit_delay= [KNL]
4628 The delay, in seconds, between successive
4629 read-then-exit testing episodes.
4631 rcutorture.shuffle_interval= [KNL]
4632 Set task-shuffle interval (s). Shuffling tasks
4633 allows some CPUs to go into dyntick-idle mode
4634 during the rcutorture test.
4636 rcutorture.shutdown_secs= [KNL]
4637 Set time (s) after boot system shutdown. This
4638 is useful for hands-off automated testing.
4640 rcutorture.stall_cpu= [KNL]
4641 Duration of CPU stall (s) to test RCU CPU stall
4642 warnings, zero to disable.
4644 rcutorture.stall_cpu_block= [KNL]
4645 Sleep while stalling if set. This will result
4646 in warnings from preemptible RCU in addition
4647 to any other stall-related activity.
4649 rcutorture.stall_cpu_holdoff= [KNL]
4650 Time to wait (s) after boot before inducing stall.
4652 rcutorture.stall_cpu_irqsoff= [KNL]
4653 Disable interrupts while stalling if set.
4655 rcutorture.stall_gp_kthread= [KNL]
4656 Duration (s) of forced sleep within RCU
4657 grace-period kthread to test RCU CPU stall
4658 warnings, zero to disable. If both stall_cpu
4659 and stall_gp_kthread are specified, the
4660 kthread is starved first, then the CPU.
4662 rcutorture.stat_interval= [KNL]
4663 Time (s) between statistics printk()s.
4665 rcutorture.stutter= [KNL]
4666 Time (s) to stutter testing, for example, specifying
4667 five seconds causes the test to run for five seconds,
4668 wait for five seconds, and so on. This tests RCU's
4669 ability to transition abruptly to and from idle.
4671 rcutorture.test_boost= [KNL]
4672 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
4673 "Maybe" means test if the RCU implementation
4674 under test support RCU priority boosting.
4676 rcutorture.test_boost_duration= [KNL]
4677 Duration (s) of each individual boost test.
4679 rcutorture.test_boost_interval= [KNL]
4680 Interval (s) between each boost test.
4682 rcutorture.test_no_idle_hz= [KNL]
4683 Test RCU's dyntick-idle handling. See also the
4684 rcutorture.shuffle_interval parameter.
4686 rcutorture.torture_type= [KNL]
4687 Specify the RCU implementation to test.
4689 rcutorture.verbose= [KNL]
4690 Enable additional printk() statements.
4692 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
4693 Dump ftrace buffer after reporting RCU CPU
4696 rcupdate.rcu_cpu_stall_suppress= [KNL]
4697 Suppress RCU CPU stall warning messages.
4699 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL]
4700 Suppress RCU CPU stall warning messages and
4701 rcutorture writer stall warnings that occur
4702 during early boot, that is, during the time
4703 before the init task is spawned.
4705 rcupdate.rcu_cpu_stall_timeout= [KNL]
4706 Set timeout for RCU CPU stall warning messages.
4708 rcupdate.rcu_expedited= [KNL]
4709 Use expedited grace-period primitives, for
4710 example, synchronize_rcu_expedited() instead
4711 of synchronize_rcu(). This reduces latency,
4712 but can increase CPU utilization, degrade
4713 real-time latency, and degrade energy efficiency.
4714 No effect on CONFIG_TINY_RCU kernels.
4716 rcupdate.rcu_normal= [KNL]
4717 Use only normal grace-period primitives,
4718 for example, synchronize_rcu() instead of
4719 synchronize_rcu_expedited(). This improves
4720 real-time latency, CPU utilization, and
4721 energy efficiency, but can expose users to
4722 increased grace-period latency. This parameter
4723 overrides rcupdate.rcu_expedited. No effect on
4724 CONFIG_TINY_RCU kernels.
4726 rcupdate.rcu_normal_after_boot= [KNL]
4727 Once boot has completed (that is, after
4728 rcu_end_inkernel_boot() has been invoked), use
4729 only normal grace-period primitives. No effect
4730 on CONFIG_TINY_RCU kernels.
4732 But note that CONFIG_PREEMPT_RT=y kernels enables
4733 this kernel boot parameter, forcibly setting
4734 it to the value one, that is, converting any
4735 post-boot attempt at an expedited RCU grace
4736 period to instead use normal non-expedited
4737 grace-period processing.
4739 rcupdate.rcu_task_ipi_delay= [KNL]
4740 Set time in jiffies during which RCU tasks will
4741 avoid sending IPIs, starting with the beginning
4742 of a given grace period. Setting a large
4743 number avoids disturbing real-time workloads,
4744 but lengthens grace periods.
4746 rcupdate.rcu_task_stall_timeout= [KNL]
4747 Set timeout in jiffies for RCU task stall warning
4748 messages. Disable with a value less than or equal
4751 rcupdate.rcu_self_test= [KNL]
4752 Run the RCU early boot self tests
4756 Run specified binary instead of /init from the ramdisk,
4757 used for early userspace startup. See initrd.
4760 force - Override the decision by the kernel to hide the
4761 advertisement of RDRAND support (this affects
4762 certain AMD processors because of buggy BIOS
4763 support, specifically around the suspend/resume
4767 Turn on/off individual RDT features. List is:
4768 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
4770 E.g. to turn on cmt and turn off mba use:
4774 Format (x86 or x86_64):
4775 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] \
4777 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
4779 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
4780 (prefix with 'panic_' to set mode for panic
4782 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
4783 reboot_force is either force or not specified,
4784 reboot_cpu is s[mp]#### with #### being the processor
4785 to be used for rebooting.
4787 refscale.holdoff= [KNL]
4788 Set test-start holdoff period. The purpose of
4789 this parameter is to delay the start of the
4790 test until boot completes in order to avoid
4793 refscale.loops= [KNL]
4794 Set the number of loops over the synchronization
4795 primitive under test. Increasing this number
4796 reduces noise due to loop start/end overhead,
4797 but the default has already reduced the per-pass
4798 noise to a handful of picoseconds on ca. 2020
4801 refscale.nreaders= [KNL]
4802 Set number of readers. The default value of -1
4803 selects N, where N is roughly 75% of the number
4804 of CPUs. A value of zero is an interesting choice.
4806 refscale.nruns= [KNL]
4807 Set number of runs, each of which is dumped onto
4810 refscale.readdelay= [KNL]
4811 Set the read-side critical-section duration,
4812 measured in microseconds.
4814 refscale.scale_type= [KNL]
4815 Specify the read-protection implementation to test.
4817 refscale.shutdown= [KNL]
4818 Shut down the system at the end of the performance
4819 test. This defaults to 1 (shut it down) when
4820 refscale is built into the kernel and to 0 (leave
4821 it running) when refscale is built as a module.
4823 refscale.verbose= [KNL]
4824 Enable additional printk() statements.
4826 refscale.verbose_batched= [KNL]
4827 Batch the additional printk() statements. If zero
4828 (the default) or negative, print everything. Otherwise,
4829 print every Nth verbose statement, where N is the value
4833 [KNL, SMP] Set scheduler's default relax_domain_level.
4834 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
4836 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
4837 Format: <base1>,<size1>[,<base2>,<size2>,...]
4838 Reserve I/O ports or memory so the kernel won't use
4839 them. If <base> is less than 0x10000, the region
4840 is assumed to be I/O ports; otherwise it is memory.
4842 reservetop= [X86-32]
4844 Reserves a hole at the top of the kernel virtual
4847 reset_devices [KNL] Force drivers to reset the underlying device
4848 during initialization.
4851 Specify the partition device for software suspend
4853 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
4855 resume_offset= [SWSUSP]
4856 Specify the offset from the beginning of the partition
4857 given by "resume=" at which the swap header is located,
4858 in <PAGE_SIZE> units (needed only for swap files).
4859 See Documentation/power/swsusp-and-swap-files.rst
4861 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
4862 read the resume files
4864 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
4865 Useful for devices that are detected asynchronously
4866 (e.g. USB and MMC devices).
4868 hibernate= [HIBERNATION]
4869 noresume Don't check if there's a hibernation image
4870 present during boot.
4871 nocompress Don't compress/decompress hibernation images.
4872 no Disable hibernation and resume.
4873 protect_image Turn on image protection during restoration
4874 (that will set all pages holding image data
4875 during restoration read-only).
4877 retain_initrd [RAM] Keep initrd memory after extraction
4879 rfkill.default_state=
4880 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
4881 etc. communication is blocked by default.
4884 rfkill.master_switch_mode=
4885 0 The "airplane mode" button does nothing.
4886 1 The "airplane mode" button toggles between everything
4887 blocked and the previous configuration.
4888 2 The "airplane mode" button toggles between everything
4889 blocked and everything unblocked.
4891 rhash_entries= [KNL,NET]
4892 Set number of hash buckets for route cache
4895 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
4898 ro [KNL] Mount root device read-only on boot
4901 on Mark read-only kernel memory as read-only (default).
4902 off Leave read-only kernel memory writable for debugging.
4905 Enable the uart passthrough on the designated usb port
4906 on Rockchip SoCs. When active, the signals of the
4907 debug-uart get routed to the D+ and D- pins of the usb
4908 port and the regular usb controller gets disabled.
4910 root= [KNL] Root filesystem
4911 See name_to_dev_t comment in init/do_mounts.c.
4913 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
4914 mount the root filesystem
4916 rootflags= [KNL] Set root filesystem mount option string
4918 rootfstype= [KNL] Set root filesystem type
4920 rootwait [KNL] Wait (indefinitely) for root device to show up.
4921 Useful for devices that are detected asynchronously
4922 (e.g. USB and MMC devices).
4924 rproc_mem=nn[KMG][@address]
4925 [KNL,ARM,CMA] Remoteproc physical memory block.
4926 Memory area to be used by remote processor image,
4929 rw [KNL] Mount root device read-write on boot
4931 S [KNL] Run init in single mode
4933 s390_iommu= [HW,S390]
4934 Set s390 IOTLB flushing mode
4936 With strict flushing every unmap operation will result in
4937 an IOTLB flush. Default is lazy flushing before reuse,
4941 See drivers/net/irda/sa1100_ir.c.
4943 sbni= [NET] Granch SBNI12 leased line adapter
4945 sched_verbose [KNL] Enables verbose scheduler debug messages.
4947 schedstats= [KNL,X86] Enable or disable scheduled statistics.
4948 Allowed values are enable and disable. This feature
4949 incurs a small amount of overhead in the scheduler
4950 but is useful for debugging and performance tuning.
4952 sched_thermal_decay_shift=
4953 [KNL, SMP] Set a decay shift for scheduler thermal
4954 pressure signal. Thermal pressure signal follows the
4955 default decay period of other scheduler pelt
4956 signals(usually 32 ms but configurable). Setting
4957 sched_thermal_decay_shift will left shift the decay
4958 period for the thermal pressure signal by the shift
4960 i.e. with the default pelt decay period of 32 ms
4961 sched_thermal_decay_shift thermal pressure decay pr
4965 Format: integer between 0 and 10
4968 scftorture.holdoff= [KNL]
4969 Number of seconds to hold off before starting
4970 test. Defaults to zero for module insertion and
4971 to 10 seconds for built-in smp_call_function()
4974 scftorture.longwait= [KNL]
4975 Request ridiculously long waits randomly selected
4976 up to the chosen limit in seconds. Zero (the
4977 default) disables this feature. Please note
4978 that requesting even small non-zero numbers of
4979 seconds can result in RCU CPU stall warnings,
4980 softlockup complaints, and so on.
4982 scftorture.nthreads= [KNL]
4983 Number of kthreads to spawn to invoke the
4984 smp_call_function() family of functions.
4985 The default of -1 specifies a number of kthreads
4986 equal to the number of CPUs.
4988 scftorture.onoff_holdoff= [KNL]
4989 Number seconds to wait after the start of the
4990 test before initiating CPU-hotplug operations.
4992 scftorture.onoff_interval= [KNL]
4993 Number seconds to wait between successive
4994 CPU-hotplug operations. Specifying zero (which
4995 is the default) disables CPU-hotplug operations.
4997 scftorture.shutdown_secs= [KNL]
4998 The number of seconds following the start of the
4999 test after which to shut down the system. The
5000 default of zero avoids shutting down the system.
5001 Non-zero values are useful for automated tests.
5003 scftorture.stat_interval= [KNL]
5004 The number of seconds between outputting the
5005 current test statistics to the console. A value
5006 of zero disables statistics output.
5008 scftorture.stutter_cpus= [KNL]
5009 The number of jiffies to wait between each change
5010 to the set of CPUs under test.
5012 scftorture.use_cpus_read_lock= [KNL]
5013 Use use_cpus_read_lock() instead of the default
5014 preempt_disable() to disable CPU hotplug
5015 while invoking one of the smp_call_function*()
5018 scftorture.verbose= [KNL]
5019 Enable additional printk() statements.
5021 scftorture.weight_single= [KNL]
5022 The probability weighting to use for the
5023 smp_call_function_single() function with a zero
5024 "wait" parameter. A value of -1 selects the
5025 default if all other weights are -1. However,
5026 if at least one weight has some other value, a
5027 value of -1 will instead select a weight of zero.
5029 scftorture.weight_single_wait= [KNL]
5030 The probability weighting to use for the
5031 smp_call_function_single() function with a
5032 non-zero "wait" parameter. See weight_single.
5034 scftorture.weight_many= [KNL]
5035 The probability weighting to use for the
5036 smp_call_function_many() function with a zero
5037 "wait" parameter. See weight_single.
5038 Note well that setting a high probability for
5039 this weighting can place serious IPI load
5042 scftorture.weight_many_wait= [KNL]
5043 The probability weighting to use for the
5044 smp_call_function_many() function with a
5045 non-zero "wait" parameter. See weight_single
5048 scftorture.weight_all= [KNL]
5049 The probability weighting to use for the
5050 smp_call_function_all() function with a zero
5051 "wait" parameter. See weight_single and
5054 scftorture.weight_all_wait= [KNL]
5055 The probability weighting to use for the
5056 smp_call_function_all() function with a
5057 non-zero "wait" parameter. See weight_single
5060 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
5061 xtime_lock contention on larger systems, and/or RCU lock
5062 contention on all systems with CONFIG_MAXSMP set.
5063 Format: { "0" | "1" }
5064 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
5066 Note: increases power consumption, thus should only be
5067 enabled if running jitter sensitive (HPC/RT) workloads.
5069 security= [SECURITY] Choose a legacy "major" security module to
5070 enable at boot. This has been deprecated by the
5073 selinux= [SELINUX] Disable or enable SELinux at boot time.
5074 Format: { "0" | "1" }
5075 See security/selinux/Kconfig help text.
5080 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
5081 Format: { "0" | "1" }
5082 See security/apparmor/Kconfig help text
5085 Default value is set via kernel config option.
5087 serialnumber [BUGS=X86-32]
5090 Maximal number of shapers.
5098 Enable merging of slabs with similar size when the
5099 kernel is built without CONFIG_SLAB_MERGE_DEFAULT.
5102 Disable merging of slabs with similar size. May be
5103 necessary if there is some reason to distinguish
5104 allocs to different slabs, especially in hardened
5105 environments where the risk of heap overflows and
5106 layout control by attackers can usually be
5107 frustrated by disabling merging. This will reduce
5108 most of the exposure of a heap attack to a single
5109 cache (risks via metadata attacks are mostly
5110 unchanged). Debug options disable merging on their
5112 For more information see Documentation/vm/slub.rst.
5114 slab_max_order= [MM, SLAB]
5115 Determines the maximum allowed order for slabs.
5116 A high setting may cause OOMs due to memory
5117 fragmentation. Defaults to 1 for systems with
5118 more than 32MB of RAM, 0 otherwise.
5120 slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB]
5121 Enabling slub_debug allows one to determine the
5122 culprit if slab objects become corrupted. Enabling
5123 slub_debug can create guard zones around objects and
5124 may poison objects when not in use. Also tracks the
5125 last alloc / free. For more information see
5126 Documentation/vm/slub.rst.
5128 slub_max_order= [MM, SLUB]
5129 Determines the maximum allowed order for slabs.
5130 A high setting may cause OOMs due to memory
5131 fragmentation. For more information see
5132 Documentation/vm/slub.rst.
5134 slub_min_objects= [MM, SLUB]
5135 The minimum number of objects per slab. SLUB will
5136 increase the slab order up to slub_max_order to
5137 generate a sufficiently large slab able to contain
5138 the number of objects indicated. The higher the number
5139 of objects the smaller the overhead of tracking slabs
5140 and the less frequently locks need to be acquired.
5141 For more information see Documentation/vm/slub.rst.
5143 slub_min_order= [MM, SLUB]
5144 Determines the minimum page order for slabs. Must be
5145 lower than slub_max_order.
5146 For more information see Documentation/vm/slub.rst.
5148 slub_merge [MM, SLUB]
5149 Same with slab_merge.
5151 slub_nomerge [MM, SLUB]
5152 Same with slab_nomerge. This is supported for legacy.
5153 See slab_nomerge for more information.
5156 Format: <io1>[,<io2>[,...,<io8>]]
5158 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
5159 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
5160 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
5161 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
5162 smsc-ircc2.ircc_irq= [HW] IRQ line
5163 smsc-ircc2.ircc_dma= [HW] DMA channel
5164 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
5165 0: Toshiba Satellite 1800 (GP data pin select)
5166 1: Fast pin select (default)
5169 smt [KNL,S390] Set the maximum number of threads (logical
5170 CPUs) to use per physical CPU on systems capable of
5171 symmetric multithreading (SMT). Will be capped to the
5172 actual hardware limit.
5174 Default: -1 (no limit)
5177 [KNL] Should the soft-lockup detector generate panics.
5180 A value of 1 instructs the soft-lockup detector
5181 to panic the machine when a soft-lockup occurs. It is
5182 also controlled by the kernel.softlockup_panic sysctl
5183 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
5184 respective build-time switch to that functionality.
5186 softlockup_all_cpu_backtrace=
5187 [KNL] Should the soft-lockup detector generate
5188 backtraces on all cpus.
5191 sonypi.*= [HW] Sony Programmable I/O Control Device driver
5192 See Documentation/admin-guide/laptops/sonypi.rst
5194 spectre_v2= [X86] Control mitigation of Spectre variant 2
5195 (indirect branch speculation) vulnerability.
5196 The default operation protects the kernel from
5199 on - unconditionally enable, implies
5201 off - unconditionally disable, implies
5203 auto - kernel detects whether your CPU model is
5206 Selecting 'on' will, and 'auto' may, choose a
5207 mitigation method at run time according to the
5208 CPU, the available microcode, the setting of the
5209 CONFIG_RETPOLINE configuration option, and the
5210 compiler with which the kernel was built.
5212 Selecting 'on' will also enable the mitigation
5213 against user space to user space task attacks.
5215 Selecting 'off' will disable both the kernel and
5216 the user space protections.
5218 Specific mitigations can also be selected manually:
5220 retpoline - replace indirect branches
5221 retpoline,generic - google's original retpoline
5222 retpoline,amd - AMD-specific minimal thunk
5224 Not specifying this option is equivalent to
5228 [X86] Control mitigation of Spectre variant 2
5229 (indirect branch speculation) vulnerability between
5232 on - Unconditionally enable mitigations. Is
5233 enforced by spectre_v2=on
5235 off - Unconditionally disable mitigations. Is
5236 enforced by spectre_v2=off
5238 prctl - Indirect branch speculation is enabled,
5239 but mitigation can be enabled via prctl
5240 per thread. The mitigation control state
5241 is inherited on fork.
5244 - Like "prctl" above, but only STIBP is
5245 controlled per thread. IBPB is issued
5246 always when switching between different user
5250 - Same as "prctl" above, but all seccomp
5251 threads will enable the mitigation unless
5252 they explicitly opt out.
5255 - Like "seccomp" above, but only STIBP is
5256 controlled per thread. IBPB is issued
5257 always when switching between different
5258 user space processes.
5260 auto - Kernel selects the mitigation depending on
5261 the available CPU features and vulnerability.
5264 If CONFIG_SECCOMP=y then "seccomp", otherwise "prctl"
5266 Not specifying this option is equivalent to
5267 spectre_v2_user=auto.
5269 spec_store_bypass_disable=
5270 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
5271 (Speculative Store Bypass vulnerability)
5273 Certain CPUs are vulnerable to an exploit against a
5274 a common industry wide performance optimization known
5275 as "Speculative Store Bypass" in which recent stores
5276 to the same memory location may not be observed by
5277 later loads during speculative execution. The idea
5278 is that such stores are unlikely and that they can
5279 be detected prior to instruction retirement at the
5280 end of a particular speculation execution window.
5282 In vulnerable processors, the speculatively forwarded
5283 store can be used in a cache side channel attack, for
5284 example to read memory to which the attacker does not
5285 directly have access (e.g. inside sandboxed code).
5287 This parameter controls whether the Speculative Store
5288 Bypass optimization is used.
5290 On x86 the options are:
5292 on - Unconditionally disable Speculative Store Bypass
5293 off - Unconditionally enable Speculative Store Bypass
5294 auto - Kernel detects whether the CPU model contains an
5295 implementation of Speculative Store Bypass and
5296 picks the most appropriate mitigation. If the
5297 CPU is not vulnerable, "off" is selected. If the
5298 CPU is vulnerable the default mitigation is
5299 architecture and Kconfig dependent. See below.
5300 prctl - Control Speculative Store Bypass per thread
5301 via prctl. Speculative Store Bypass is enabled
5302 for a process by default. The state of the control
5303 is inherited on fork.
5304 seccomp - Same as "prctl" above, but all seccomp threads
5305 will disable SSB unless they explicitly opt out.
5307 Default mitigations:
5308 X86: If CONFIG_SECCOMP=y "seccomp", otherwise "prctl"
5310 On powerpc the options are:
5312 on,auto - On Power8 and Power9 insert a store-forwarding
5313 barrier on kernel entry and exit. On Power7
5314 perform a software flush on kernel entry and
5318 Not specifying this option is equivalent to
5319 spec_store_bypass_disable=auto.
5321 spia_io_base= [HW,MTD]
5327 [X86] Enable split lock detection or bus lock detection
5329 When enabled (and if hardware support is present), atomic
5330 instructions that access data across cache line
5331 boundaries will result in an alignment check exception
5332 for split lock detection or a debug exception for
5337 warn - the kernel will emit rate-limited warnings
5338 about applications triggering the #AC
5339 exception or the #DB exception. This mode is
5340 the default on CPUs that support split lock
5341 detection or bus lock detection. Default
5342 behavior is by #AC if both features are
5343 enabled in hardware.
5345 fatal - the kernel will send SIGBUS to applications
5346 that trigger the #AC exception or the #DB
5347 exception. Default behavior is by #AC if
5348 both features are enabled in hardware.
5351 Set system wide rate limit to N bus locks
5352 per second for bus lock detection.
5355 N/A for split lock detection.
5358 If an #AC exception is hit in the kernel or in
5359 firmware (i.e. not while executing in user mode)
5360 the kernel will oops in either "warn" or "fatal"
5363 #DB exception for bus lock is triggered only when
5367 Control the Special Register Buffer Data Sampling
5370 Certain CPUs are vulnerable to an MDS-like
5371 exploit which can leak bits from the random
5374 By default, this issue is mitigated by
5375 microcode. However, the microcode fix can cause
5376 the RDRAND and RDSEED instructions to become
5377 much slower. Among other effects, this will
5378 result in reduced throughput from /dev/urandom.
5380 The microcode mitigation can be disabled with
5381 the following option:
5383 off: Disable mitigation and remove
5384 performance impact to RDRAND and RDSEED
5386 srcutree.counter_wrap_check [KNL]
5387 Specifies how frequently to check for
5388 grace-period sequence counter wrap for the
5389 srcu_data structure's ->srcu_gp_seq_needed field.
5390 The greater the number of bits set in this kernel
5391 parameter, the less frequently counter wrap will
5392 be checked for. Note that the bottom two bits
5395 srcutree.exp_holdoff [KNL]
5396 Specifies how many nanoseconds must elapse
5397 since the end of the last SRCU grace period for
5398 a given srcu_struct until the next normal SRCU
5399 grace period will be considered for automatic
5400 expediting. Set to zero to disable automatic
5404 Speculative Store Bypass Disable control
5406 On CPUs that are vulnerable to the Speculative
5407 Store Bypass vulnerability and offer a
5408 firmware based mitigation, this parameter
5409 indicates how the mitigation should be used:
5411 force-on: Unconditionally enable mitigation for
5412 for both kernel and userspace
5413 force-off: Unconditionally disable mitigation for
5414 for both kernel and userspace
5415 kernel: Always enable mitigation in the
5416 kernel, and offer a prctl interface
5417 to allow userspace to register its
5418 interest in being mitigated too.
5420 stack_guard_gap= [MM]
5421 override the default stack gap protection. The value
5422 is in page units and it defines how many pages prior
5423 to (for stacks growing down) resp. after (for stacks
5424 growing up) the main stack are reserved for no other
5425 mapping. Default value is 256 pages.
5427 stack_depot_disable= [KNL]
5428 Setting this to true through kernel command line will
5429 disable the stack depot thereby saving the static memory
5430 consumed by the stack hash table. By default this is set
5434 Enabled the stack tracer on boot up.
5436 stacktrace_filter=[function-list]
5437 [FTRACE] Limit the functions that the stack tracer
5438 will trace at boot up. function-list is a comma-separated
5439 list of functions. This list can be changed at run
5440 time by the stack_trace_filter file in the debugfs
5441 tracing directory. Note, this enables stack tracing
5442 and the stacktrace above is not needed.
5446 Set the STI (builtin display/keyboard on the HP-PARISC
5447 machines) console (graphic card) which should be used
5448 as the initial boot-console.
5449 See also comment in drivers/video/console/sticore.c.
5452 See comment in drivers/video/console/sticore.c.
5455 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
5457 sunrpc.min_resvport=
5458 sunrpc.max_resvport=
5460 SunRPC servers often require that client requests
5461 originate from a privileged port (i.e. a port in the
5462 range 0 < portnr < 1024).
5463 An administrator who wishes to reserve some of these
5464 ports for other uses may adjust the range that the
5465 kernel's sunrpc client considers to be privileged
5466 using these two parameters to set the minimum and
5467 maximum port values.
5469 sunrpc.svc_rpc_per_connection_limit=
5471 Limit the number of requests that the server will
5472 process in parallel from a single connection.
5473 The default value is 0 (no limit).
5477 Control how the NFS server code allocates CPUs to
5478 service thread pools. Depending on how many NICs
5479 you have and where their interrupts are bound, this
5480 option will affect which CPUs will do NFS serving.
5481 Note: this parameter cannot be changed while the
5482 NFS server is running.
5484 auto the server chooses an appropriate mode
5485 automatically using heuristics
5486 global a single global pool contains all CPUs
5487 percpu one pool for each CPU
5488 pernode one pool for each NUMA node (equivalent
5489 to global on non-NUMA machines)
5491 sunrpc.tcp_slot_table_entries=
5492 sunrpc.udp_slot_table_entries=
5494 Sets the upper limit on the number of simultaneous
5495 RPC calls that can be sent from the client to a
5496 server. Increasing these values may allow you to
5497 improve throughput, but will also increase the
5498 amount of memory reserved for use by the client.
5500 suspend.pm_test_delay=
5502 Sets the number of seconds to remain in a suspend test
5503 mode before resuming the system (see
5504 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
5505 is set. Default value is 5.
5508 Format: { on | off | y | n | 1 | 0 }
5509 This parameter controls use of the Protected
5510 Execution Facility on pSeries.
5513 [KNL] Enable accounting of swap in memory resource
5514 controller if no parameter or 1 is given or disable
5515 it if 0 is given (See Documentation/admin-guide/cgroup-v1/memory.rst)
5517 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
5518 Format: { <int> | force | noforce }
5519 <int> -- Number of I/O TLB slabs
5520 force -- force using of bounce buffers even if they
5521 wouldn't be automatically used by the kernel
5522 noforce -- Never use bounce buffers (for debugging)
5527 Set a sysctl parameter, right before loading the init
5528 process, as if the value was written to the respective
5529 /proc/sys/... file. Both '.' and '/' are recognized as
5530 separators. Unrecognized parameters and invalid values
5531 are reported in the kernel log. Sysctls registered
5532 later by a loaded module cannot be set this way.
5533 Example: sysctl.vm.swappiness=40
5535 sysfs.deprecated=0|1 [KNL]
5536 Enable/disable old style sysfs layout for old udev
5537 on older distributions. When this option is enabled
5538 very new udev will not work anymore. When this option
5539 is disabled (or CONFIG_SYSFS_DEPRECATED not compiled)
5540 in older udev will not work anymore.
5541 Default depends on CONFIG_SYSFS_DEPRECATED_V2 set in
5542 the kernel configuration.
5544 sysrq_always_enabled
5546 Ignore sysrq setting - this boot parameter will
5547 neutralize any effect of /proc/sys/kernel/sysrq.
5548 Useful for debugging.
5550 tcpmhash_entries= [KNL,NET]
5551 Set the number of tcp_metrics_hash slots.
5552 Default value is 8192 or 16384 depending on total
5553 ram pages. This is used to specify the TCP metrics
5554 cache size. See Documentation/networking/ip-sysctl.rst
5555 "tcp_no_metrics_save" section for more details.
5559 test_suspend= [SUSPEND][,N]
5560 Specify "mem" (for Suspend-to-RAM) or "standby" (for
5561 standby suspend) or "freeze" (for suspend type freeze)
5562 as the system sleep state during system startup with
5563 the optional capability to repeat N number of times.
5564 The system is woken from this state using a
5565 wakeup-capable RTC alarm.
5567 thash_entries= [KNL,NET]
5568 Set number of hash buckets for TCP connection
5570 thermal.act= [HW,ACPI]
5571 -1: disable all active trip points in all thermal zones
5572 <degrees C>: override all lowest active trip points
5574 thermal.crt= [HW,ACPI]
5575 -1: disable all critical trip points in all thermal zones
5576 <degrees C>: override all critical trip points
5578 thermal.nocrt= [HW,ACPI]
5579 Set to disable actions on ACPI thermal zone
5580 critical and hot trip points.
5582 thermal.off= [HW,ACPI]
5583 1: disable ACPI thermal control
5585 thermal.psv= [HW,ACPI]
5586 -1: disable all passive trip points
5587 <degrees C>: override all passive trip points to this
5590 thermal.tzp= [HW,ACPI]
5591 Specify global default ACPI thermal zone polling rate
5592 <deci-seconds>: poll all this frequency
5593 0: no polling (default)
5596 Force threading of all interrupt handlers except those
5597 marked explicitly IRQF_NO_THREAD.
5601 Specify if the kernel should make use of the cpu
5602 topology information if the hardware supports this.
5603 The scheduler will make use of this information and
5604 e.g. base its process migration decisions on it.
5607 topology_updates= [KNL, PPC, NUMA]
5609 Specify if the kernel should ignore (off)
5610 topology updates sent by the hypervisor to this
5613 torture.disable_onoff_at_boot= [KNL]
5614 Prevent the CPU-hotplug component of torturing
5615 until after init has spawned.
5617 torture.ftrace_dump_at_shutdown= [KNL]
5618 Dump the ftrace buffer at torture-test shutdown,
5619 even if there were no errors. This can be a
5620 very costly operation when many torture tests
5621 are running concurrently, especially on systems
5622 with rotating-rust storage.
5624 torture.verbose_sleep_frequency= [KNL]
5625 Specifies how many verbose printk()s should be
5626 emitted between each sleep. The default of zero
5627 disables verbose-printk() sleeping.
5629 torture.verbose_sleep_duration= [KNL]
5630 Duration of each verbose-printk() sleep in jiffies.
5634 tpm_suspend_pcr=[HW,TPM]
5635 Format: integer pcr id
5636 Specify that at suspend time, the tpm driver
5637 should extend the specified pcr with zeros,
5638 as a workaround for some chips which fail to
5639 flush the last written pcr on TPM_SaveState.
5640 This will guarantee that all the other pcrs
5643 trace_buf_size=nn[KMG]
5644 [FTRACE] will set tracing buffer size on each cpu.
5646 trace_event=[event-list]
5647 [FTRACE] Set and start specified trace events in order
5648 to facilitate early boot debugging. The event-list is a
5649 comma-separated list of trace events to enable. See
5650 also Documentation/trace/events.rst
5652 trace_options=[option-list]
5653 [FTRACE] Enable or disable tracer options at boot.
5654 The option-list is a comma delimited list of options
5655 that can be enabled or disabled just as if you were
5656 to echo the option name into
5658 /sys/kernel/debug/tracing/trace_options
5660 For example, to enable stacktrace option (to dump the
5661 stack trace of each event), add to the command line:
5663 trace_options=stacktrace
5665 See also Documentation/trace/ftrace.rst "trace options"
5669 Have the tracepoints sent to printk as well as the
5670 tracing ring buffer. This is useful for early boot up
5671 where the system hangs or reboots and does not give the
5672 option for reading the tracing buffer or performing a
5673 ftrace_dump_on_oops.
5675 To turn off having tracepoints sent to printk,
5676 echo 0 > /proc/sys/kernel/tracepoint_printk
5677 Note, echoing 1 into this file without the
5678 tracepoint_printk kernel cmdline option has no effect.
5680 The tp_printk_stop_on_boot (see below) can also be used
5681 to stop the printing of events to console at
5686 Having tracepoints sent to printk() and activating high
5687 frequency tracepoints such as irq or sched, can cause
5688 the system to live lock.
5690 tp_printk_stop_on_boot[FTRACE]
5691 When tp_printk (above) is set, it can cause a lot of noise
5692 on the console. It may be useful to only include the
5693 printing of events during boot up, as user space may
5694 make the system inoperable.
5696 This command line option will stop the printing of events
5697 to console at the late_initcall_sync() time frame.
5700 [FTRACE] enable this option to disable tracing when a
5701 warning is hit. This turns off "tracing_on". Tracing can
5702 be enabled again by echoing '1' into the "tracing_on"
5703 file located in /sys/kernel/debug/tracing/
5705 This option is useful, as it disables the trace before
5706 the WARNING dump is called, which prevents the trace to
5707 be filled with content caused by the warning output.
5709 This option can also be set at run time via the sysctl
5710 option: kernel/traceoff_on_warning
5712 transparent_hugepage=
5714 Format: [always|madvise|never]
5715 Can be used to control the default behavior of the system
5716 with respect to transparent hugepages.
5717 See Documentation/admin-guide/mm/transhuge.rst
5720 trusted.source= [KEYS]
5722 This parameter identifies the trust source as a backend
5723 for trusted keys implementation. Supported trust
5727 If not specified then it defaults to iterating through
5728 the trust source list starting with TPM and assigns the
5729 first trust source as a backend which is initialized
5730 successfully during iteration.
5732 tsc= Disable clocksource stability checks for TSC.
5734 [x86] reliable: mark tsc clocksource as reliable, this
5735 disables clocksource verification at runtime, as well
5736 as the stability checks done at bootup. Used to enable
5737 high-resolution timer mode on older hardware, and in
5738 virtualized environment.
5739 [x86] noirqtime: Do not use TSC to do irq accounting.
5740 Used to run time disable IRQ_TIME_ACCOUNTING on any
5741 platforms where RDTSC is slow and this accounting
5743 [x86] unstable: mark the TSC clocksource as unstable, this
5744 marks the TSC unconditionally unstable at bootup and
5745 avoids any further wobbles once the TSC watchdog notices.
5746 [x86] nowatchdog: disable clocksource watchdog. Used
5747 in situations with strict latency requirements (where
5748 interruptions from clocksource watchdog are not
5751 tsc_early_khz= [X86] Skip early TSC calibration and use the given
5752 value instead. Useful when the early TSC frequency discovery
5753 procedure is not reliable, such as on overclocked systems
5754 with CPUID.16h support and partial CPUID.15h support.
5755 Format: <unsigned int>
5757 tsx= [X86] Control Transactional Synchronization
5758 Extensions (TSX) feature in Intel processors that
5759 support TSX control.
5761 This parameter controls the TSX feature. The options are:
5763 on - Enable TSX on the system. Although there are
5764 mitigations for all known security vulnerabilities,
5765 TSX has been known to be an accelerator for
5766 several previous speculation-related CVEs, and
5767 so there may be unknown security risks associated
5768 with leaving it enabled.
5770 off - Disable TSX on the system. (Note that this
5771 option takes effect only on newer CPUs which are
5772 not vulnerable to MDS, i.e., have
5773 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
5774 the new IA32_TSX_CTRL MSR through a microcode
5775 update. This new MSR allows for the reliable
5776 deactivation of the TSX functionality.)
5778 auto - Disable TSX if X86_BUG_TAA is present,
5779 otherwise enable TSX on the system.
5781 Not specifying this option is equivalent to tsx=off.
5783 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5786 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
5787 Abort (TAA) vulnerability.
5789 Similar to Micro-architectural Data Sampling (MDS)
5790 certain CPUs that support Transactional
5791 Synchronization Extensions (TSX) are vulnerable to an
5792 exploit against CPU internal buffers which can forward
5793 information to a disclosure gadget under certain
5796 In vulnerable processors, the speculatively forwarded
5797 data can be used in a cache side channel attack, to
5798 access data to which the attacker does not have direct
5801 This parameter controls the TAA mitigation. The
5804 full - Enable TAA mitigation on vulnerable CPUs
5807 full,nosmt - Enable TAA mitigation and disable SMT on
5808 vulnerable CPUs. If TSX is disabled, SMT
5809 is not disabled because CPU is not
5810 vulnerable to cross-thread TAA attacks.
5811 off - Unconditionally disable TAA mitigation
5813 On MDS-affected machines, tsx_async_abort=off can be
5814 prevented by an active MDS mitigation as both vulnerabilities
5815 are mitigated with the same mechanism so in order to disable
5816 this mitigation, you need to specify mds=off too.
5818 Not specifying this option is equivalent to
5819 tsx_async_abort=full. On CPUs which are MDS affected
5820 and deploy MDS mitigation, TAA mitigation is not
5821 required and doesn't provide any additional
5825 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5827 turbografx.map[2|3]= [HW,JOY]
5828 TurboGraFX parallel port interface
5830 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
5831 See also Documentation/input/devices/joystick-parport.rst
5833 udbg-immortal [PPC] When debugging early kernel crashes that
5834 happen after console_init() and before a proper
5835 console driver takes over, this boot options might
5836 help "seeing" what's going on.
5838 uhash_entries= [KNL,NET]
5839 Set number of hash buckets for UDP/UDP-Lite connections
5842 [USB] Ignore overcurrent events (default N).
5843 Some badly-designed motherboards generate lots of
5844 bogus events, for ports that aren't wired to
5845 anything. Set this parameter to avoid log spamming.
5846 Note that genuine overcurrent events won't be
5850 [X86] Cause panic on unknown NMI.
5852 usbcore.authorized_default=
5853 [USB] Default USB device authorization:
5854 (default -1 = authorized except for wireless USB,
5855 0 = not authorized, 1 = authorized, 2 = authorized
5856 if device connected to internal port)
5858 usbcore.autosuspend=
5859 [USB] The autosuspend time delay (in seconds) used
5860 for newly-detected USB devices (default 2). This
5861 is the time required before an idle device will be
5862 autosuspended. Devices for which the delay is set
5863 to a negative value won't be autosuspended at all.
5865 usbcore.usbfs_snoop=
5866 [USB] Set to log all usbfs traffic (default 0 = off).
5868 usbcore.usbfs_snoop_max=
5869 [USB] Maximum number of bytes to snoop in each URB
5872 usbcore.blinkenlights=
5873 [USB] Set to cycle leds on hubs (default 0 = off).
5875 usbcore.old_scheme_first=
5876 [USB] Start with the old device initialization
5877 scheme (default 0 = off).
5879 usbcore.usbfs_memory_mb=
5880 [USB] Memory limit (in MB) for buffers allocated by
5881 usbfs (default = 16, 0 = max = 2047).
5883 usbcore.use_both_schemes=
5884 [USB] Try the other device initialization scheme
5885 if the first one fails (default 1 = enabled).
5887 usbcore.initial_descriptor_timeout=
5888 [USB] Specifies timeout for the initial 64-byte
5889 USB_REQ_GET_DESCRIPTOR request in milliseconds
5890 (default 5000 = 5.0 seconds).
5892 usbcore.nousb [USB] Disable the USB subsystem
5895 [USB] A list of quirk entries to augment the built-in
5896 usb core quirk list. List entries are separated by
5897 commas. Each entry has the form
5898 VendorID:ProductID:Flags. The IDs are 4-digit hex
5899 numbers and Flags is a set of letters. Each letter
5900 will change the built-in quirk; setting it if it is
5901 clear and clearing it if it is set. The letters have
5902 the following meanings:
5903 a = USB_QUIRK_STRING_FETCH_255 (string
5904 descriptors must not be fetched using
5906 b = USB_QUIRK_RESET_RESUME (device can't resume
5907 correctly so reset it instead);
5908 c = USB_QUIRK_NO_SET_INTF (device can't handle
5909 Set-Interface requests);
5910 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
5911 handle its Configuration or Interface
5913 e = USB_QUIRK_RESET (device can't be reset
5914 (e.g morph devices), don't use reset);
5915 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
5916 more interface descriptions than the
5917 bNumInterfaces count, and can't handle
5918 talking to these interfaces);
5919 g = USB_QUIRK_DELAY_INIT (device needs a pause
5920 during initialization, after we read
5921 the device descriptor);
5922 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
5923 high speed and super speed interrupt
5924 endpoints, the USB 2.0 and USB 3.0 spec
5925 require the interval in microframes (1
5926 microframe = 125 microseconds) to be
5927 calculated as interval = 2 ^
5929 Devices with this quirk report their
5930 bInterval as the result of this
5931 calculation instead of the exponent
5932 variable used in the calculation);
5933 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
5934 handle device_qualifier descriptor
5936 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
5937 generates spurious wakeup, ignore
5938 remote wakeup capability);
5939 k = USB_QUIRK_NO_LPM (device can't handle Link
5941 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
5942 (Device reports its bInterval as linear
5943 frames instead of the USB 2.0
5945 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
5946 to be disconnected before suspend to
5947 prevent spurious wakeup);
5948 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
5949 pause after every control message);
5950 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
5951 delay after resetting its port);
5952 Example: quirks=0781:5580:bk,0a5c:5834:gij
5955 [USBHID] The interval which mice are to be polled at.
5958 [USBHID] The interval which joysticks are to be polled at.
5961 [USBHID] The interval which keyboards are to be polled at.
5963 usb-storage.delay_use=
5964 [UMS] The delay in seconds before a new device is
5965 scanned for Logical Units (default 1).
5968 [UMS] A list of quirks entries to supplement or
5969 override the built-in unusual_devs list. List
5970 entries are separated by commas. Each entry has
5971 the form VID:PID:Flags where VID and PID are Vendor
5972 and Product ID values (4-digit hex numbers) and
5973 Flags is a set of characters, each corresponding
5974 to a common usb-storage quirk flag as follows:
5975 a = SANE_SENSE (collect more than 18 bytes
5976 of sense data, not on uas);
5977 b = BAD_SENSE (don't collect more than 18
5978 bytes of sense data, not on uas);
5979 c = FIX_CAPACITY (decrease the reported
5980 device capacity by one sector);
5981 d = NO_READ_DISC_INFO (don't use
5982 READ_DISC_INFO command, not on uas);
5983 e = NO_READ_CAPACITY_16 (don't use
5984 READ_CAPACITY_16 command);
5985 f = NO_REPORT_OPCODES (don't use report opcodes
5987 g = MAX_SECTORS_240 (don't transfer more than
5988 240 sectors at a time, uas only);
5989 h = CAPACITY_HEURISTICS (decrease the
5990 reported device capacity by one
5991 sector if the number is odd);
5992 i = IGNORE_DEVICE (don't bind to this
5994 j = NO_REPORT_LUNS (don't use report luns
5996 k = NO_SAME (do not use WRITE_SAME, uas only)
5997 l = NOT_LOCKABLE (don't try to lock and
5998 unlock ejectable media, not on uas);
5999 m = MAX_SECTORS_64 (don't transfer more
6000 than 64 sectors = 32 KB at a time,
6002 n = INITIAL_READ10 (force a retry of the
6003 initial READ(10) command, not on uas);
6004 o = CAPACITY_OK (accept the capacity
6005 reported by the device, not on uas);
6006 p = WRITE_CACHE (the device cache is ON
6007 by default, not on uas);
6008 r = IGNORE_RESIDUE (the device reports
6009 bogus residue values, not on uas);
6010 s = SINGLE_LUN (the device has only one
6012 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
6013 commands, uas only);
6014 u = IGNORE_UAS (don't bind to the uas driver);
6015 w = NO_WP_DETECT (don't test whether the
6016 medium is write-protected).
6017 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
6018 even if the device claims no cache,
6020 Example: quirks=0419:aaf5:rl,0421:0433:rc
6022 user_debug= [KNL,ARM]
6024 See arch/arm/Kconfig.debug help text.
6025 1 - undefined instruction events
6027 4 - invalid data aborts
6030 Example: user_debug=31
6033 [X86] Flags controlling user PTE allocations.
6035 nohigh = do not allocate PTE pages in
6036 HIGHMEM regardless of setting
6040 On X86_32, this is an alias for vdso32=. Otherwise:
6042 vdso=1: enable VDSO (the default)
6043 vdso=0: disable VDSO mapping
6045 vdso32= [X86] Control the 32-bit vDSO
6046 vdso32=1: enable 32-bit VDSO
6047 vdso32=0 or vdso32=2: disable 32-bit VDSO
6049 See the help text for CONFIG_COMPAT_VDSO for more
6050 details. If CONFIG_COMPAT_VDSO is set, the default is
6051 vdso32=0; otherwise, the default is vdso32=1.
6053 For compatibility with older kernels, vdso32=2 is an
6056 Try vdso32=0 if you encounter an error that says:
6057 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
6060 vector=percpu: enable percpu vector domain
6062 video= [FB] Frame buffer configuration
6063 See Documentation/fb/modedb.rst.
6065 video.brightness_switch_enabled= [0,1]
6066 If set to 1, on receiving an ACPI notify event
6067 generated by hotkey, video driver will adjust brightness
6068 level and then send out the event to user space through
6069 the allocated input device; If set to 0, video driver
6070 will only send out the event without touching backlight
6075 [VMMIO] Memory mapped virtio (platform) device.
6077 <size>@<baseaddr>:<irq>[:<id>]
6079 <size> := size (can use standard suffixes
6081 <baseaddr> := physical base address
6082 <irq> := interrupt number (as passed to
6084 <id> := (optional) platform device id
6086 virtio_mmio.device=1K@0x100b0000:48:7
6088 Can be used multiple times for multiple devices.
6090 vga= [BOOT,X86-32] Select a particular video mode
6091 See Documentation/x86/boot.rst and
6092 Documentation/admin-guide/svga.rst.
6093 Use vga=ask for menu.
6094 This is actually a boot loader parameter; the value is
6095 passed to the kernel using a special protocol.
6097 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
6098 May slow down system boot speed, especially when
6099 enabled on systems with a large amount of memory.
6100 All options are enabled by default, and this
6101 interface is meant to allow for selectively
6102 enabling or disabling specific virtual memory
6105 Available options are:
6106 P Enable page structure init time poisoning
6107 - Disable all of the above options
6109 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
6110 size of <nn>. This can be used to increase the
6111 minimum size (128MB on x86). It can also be used to
6112 decrease the size and leave more room for directly
6115 vmcp_cma=nn[MG] [KNL,S390]
6116 Sets the memory size reserved for contiguous memory
6117 allocations for the vmcp device driver.
6119 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
6122 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
6125 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
6129 Controls the behavior of vsyscalls (i.e. calls to
6130 fixed addresses of 0xffffffffff600x00 from legacy
6131 code). Most statically-linked binaries and older
6132 versions of glibc use these calls. Because these
6133 functions are at fixed addresses, they make nice
6134 targets for exploits that can control RIP.
6136 emulate [default] Vsyscalls turn into traps and are
6137 emulated reasonably safely. The vsyscall
6140 xonly Vsyscalls turn into traps and are
6141 emulated reasonably safely. The vsyscall
6142 page is not readable.
6144 none Vsyscalls don't work at all. This makes
6145 them quite hard to use for exploits but
6146 might break your system.
6148 vt.color= [VT] Default text color.
6149 Format: 0xYX, X = foreground, Y = background.
6150 Default: 0x07 = light gray on black.
6152 vt.cur_default= [VT] Default cursor shape.
6153 Format: 0xCCBBAA, where AA, BB, and CC are the same as
6154 the parameters of the <Esc>[?A;B;Cc escape sequence;
6155 see VGA-softcursor.txt. Default: 2 = underline.
6157 vt.default_blu= [VT]
6158 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
6159 Change the default blue palette of the console.
6160 This is a 16-member array composed of values
6163 vt.default_grn= [VT]
6164 Format: <green0>,<green1>,<green2>,...,<green15>
6165 Change the default green palette of the console.
6166 This is a 16-member array composed of values
6169 vt.default_red= [VT]
6170 Format: <red0>,<red1>,<red2>,...,<red15>
6171 Change the default red palette of the console.
6172 This is a 16-member array composed of values
6178 Set system-wide default UTF-8 mode for all tty's.
6179 Default is 1, i.e. UTF-8 mode is enabled for all
6180 newly opened terminals.
6182 vt.global_cursor_default=
6185 Set system-wide default for whether a cursor
6186 is shown on new VTs. Default is -1,
6187 i.e. cursors will be created by default unless
6188 overridden by individual drivers. 0 will hide
6189 cursors, 1 will display them.
6191 vt.italic= [VT] Default color for italic text; 0-15.
6194 vt.underline= [VT] Default color for underlined text; 0-15.
6197 watchdog timers [HW,WDT] For information on watchdog timers,
6198 see Documentation/watchdog/watchdog-parameters.rst
6199 or other driver-specific files in the
6200 Documentation/watchdog/ directory.
6204 Set the hard lockup detector stall duration
6205 threshold in seconds. The soft lockup detector
6206 threshold is set to twice the value. A value of 0
6207 disables both lockup detectors. Default is 10
6210 workqueue.watchdog_thresh=
6211 If CONFIG_WQ_WATCHDOG is configured, workqueue can
6212 warn stall conditions and dump internal state to
6213 help debugging. 0 disables workqueue stall
6214 detection; otherwise, it's the stall threshold
6215 duration in seconds. The default value is 30 and
6216 it can be updated at runtime by writing to the
6217 corresponding sysfs file.
6219 workqueue.disable_numa
6220 By default, all work items queued to unbound
6221 workqueues are affine to the NUMA nodes they're
6222 issued on, which results in better behavior in
6223 general. If NUMA affinity needs to be disabled for
6224 whatever reason, this option can be used. Note
6225 that this also can be controlled per-workqueue for
6226 workqueues visible under /sys/bus/workqueue/.
6228 workqueue.power_efficient
6229 Per-cpu workqueues are generally preferred because
6230 they show better performance thanks to cache
6231 locality; unfortunately, per-cpu workqueues tend to
6232 be more power hungry than unbound workqueues.
6234 Enabling this makes the per-cpu workqueues which
6235 were observed to contribute significantly to power
6236 consumption unbound, leading to measurably lower
6237 power usage at the cost of small performance
6240 The default value of this parameter is determined by
6241 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
6243 workqueue.debug_force_rr_cpu
6244 Workqueue used to implicitly guarantee that work
6245 items queued without explicit CPU specified are put
6246 on the local CPU. This guarantee is no longer true
6247 and while local CPU is still preferred work items
6248 may be put on foreign CPUs. This debug option
6249 forces round-robin CPU selection to flush out
6250 usages which depend on the now broken guarantee.
6251 When enabled, memory and cache locality will be
6254 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
6255 default x2apic cluster mode on platforms
6258 xen_512gb_limit [KNL,X86-64,XEN]
6259 Restricts the kernel running paravirtualized under Xen
6260 to use only up to 512 GB of RAM. The reason to do so is
6261 crash analysis tools and Xen tools for doing domain
6262 save/restore/migration must be enabled to handle larger
6265 xen_emul_unplug= [HW,X86,XEN]
6266 Unplug Xen emulated devices
6267 Format: [unplug0,][unplug1]
6268 ide-disks -- unplug primary master IDE devices
6269 aux-ide-disks -- unplug non-primary-master IDE devices
6270 nics -- unplug network devices
6271 all -- unplug all emulated devices (NICs and IDE disks)
6272 unnecessary -- unplugging emulated devices is
6273 unnecessary even if the host did not respond to
6275 never -- do not unplug even if version check succeeds
6277 xen_legacy_crash [X86,XEN]
6278 Crash from Xen panic notifier, without executing late
6279 panic() code such as dumping handler.
6281 xen_nopvspin [X86,XEN]
6282 Disables the qspinlock slowpath using Xen PV optimizations.
6283 This parameter is obsoleted by "nopvspin" parameter, which
6284 has equivalent effect for XEN platform.
6287 Disables the PV optimizations forcing the HVM guest to
6288 run as generic HVM guest with no PV drivers.
6289 This option is obsoleted by the "nopv" option, which
6290 has equivalent effect for XEN platform.
6292 xen_no_vector_callback
6293 [KNL,X86,XEN] Disable the vector callback for Xen
6294 event channel interrupts.
6296 xen_scrub_pages= [XEN]
6297 Boolean option to control scrubbing pages before giving them back
6298 to Xen, for use by other domains. Can be also changed at runtime
6299 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
6300 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
6302 xen_timer_slop= [X86-64,XEN]
6303 Set the timer slop (in nanoseconds) for the virtual Xen
6304 timers (default is 100000). This adjusts the minimum
6305 delta of virtualized Xen timers, where lower values
6306 improve timer resolution at the expense of processing
6307 more timer interrupts.
6309 xen.event_eoi_delay= [XEN]
6310 How long to delay EOI handling in case of event
6311 storms (jiffies). Default is 10.
6313 xen.event_loop_timeout= [XEN]
6314 After which time (jiffies) the event handling loop
6315 should start to delay EOI handling. Default is 2.
6317 xen.fifo_events= [XEN]
6318 Boolean parameter to disable using fifo event handling
6319 even if available. Normally fifo event handling is
6320 preferred over the 2-level event handling, as it is
6321 fairer and the number of possible event channels is
6322 much higher. Default is on (use fifo events).
6324 nopv= [X86,XEN,KVM,HYPER_V,VMWARE]
6325 Disables the PV optimizations forcing the guest to run
6326 as generic guest with no PV drivers. Currently support
6327 XEN HVM, KVM, HYPER_V and VMWARE guest.
6329 nopvspin [X86,XEN,KVM]
6330 Disables the qspinlock slow path using PV optimizations
6331 which allow the hypervisor to 'idle' the guest on lock
6334 xirc2ps_cs= [NET,PCMCIA]
6336 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
6339 By default on POWER9 and above, the kernel will
6340 natively use the XIVE interrupt controller. This option
6341 allows the fallback firmware mode to be used:
6343 off Fallback to firmware control of XIVE interrupt
6344 controller on both pseries and powernv
6345 platforms. Only useful on POWER9 and above.
6347 xhci-hcd.quirks [USB,KNL]
6348 A hex value specifying bitmask with supplemental xhci
6349 host controller quirks. Meaning of each bit can be
6350 consulted in header drivers/usb/host/xhci.h.
6353 Format: { early | on | rw | ro | off }
6354 Controls if xmon debugger is enabled. Default is off.
6355 Passing only "xmon" is equivalent to "xmon=early".
6356 early Call xmon as early as possible on boot; xmon
6357 debugger is called from setup_arch().
6358 on xmon debugger hooks will be installed so xmon
6359 is only called on a kernel crash. Default mode,
6360 i.e. either "ro" or "rw" mode, is controlled
6361 with CONFIG_XMON_DEFAULT_RO_MODE.
6362 rw xmon debugger hooks will be installed so xmon
6363 is called only on a kernel crash, mode is write,
6364 meaning SPR registers, memory and, other data
6365 can be written using xmon commands.
6366 ro same as "rw" option above but SPR registers,
6367 memory, and other data can't be written using
6369 off xmon is disabled.