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_PCI_COMPONENT
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 PCI/PCI interrupt routing info messages:
64 acpi.debug_layer=0x400000
65 Enable AML "Debug" output, i.e., stores to the Debug
66 object while interpreting AML:
67 acpi.debug_layer=0xffffffff acpi.debug_level=0x2
68 Enable all messages related to ACPI hardware:
69 acpi.debug_layer=0x2 acpi.debug_level=0xffffffff
71 Some values produce so much output that the system is
72 unusable. The "log_buf_len" parameter may be useful
73 if you need to capture more output.
75 acpi_enforce_resources= [ACPI]
77 Check for resource conflicts between native drivers
78 and ACPI OperationRegions (SystemIO and SystemMemory
79 only). IO ports and memory declared in ACPI might be
80 used by the ACPI subsystem in arbitrary AML code and
81 can interfere with legacy drivers.
82 strict (default): access to resources claimed by ACPI
83 is denied; legacy drivers trying to access reserved
84 resources will fail to bind to device using them.
85 lax: access to resources claimed by ACPI is allowed;
86 legacy drivers trying to access reserved resources
87 will bind successfully but a warning message is logged.
88 no: ACPI OperationRegions are not marked as reserved,
89 no further checks are performed.
91 acpi_force_table_verification [HW,ACPI]
92 Enable table checksum verification during early stage.
93 By default, this is disabled due to x86 early mapping
96 acpi_irq_balance [HW,ACPI]
97 ACPI will balance active IRQs
100 acpi_irq_nobalance [HW,ACPI]
101 ACPI will not move active IRQs (default)
104 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA
105 Format: <irq>,<irq>...
107 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for
109 Format: <irq>,<irq>...
111 acpi_mask_gpe= [HW,ACPI]
112 Due to the existence of _Lxx/_Exx, some GPEs triggered
113 by unsupported hardware/firmware features can result in
114 GPE floodings that cannot be automatically disabled by
116 This facility can be used to prevent such uncontrolled
120 acpi_no_auto_serialize [HW,ACPI]
121 Disable auto-serialization of AML methods
122 AML control methods that contain the opcodes to create
123 named objects will be marked as "Serialized" by the
124 auto-serialization feature.
125 This feature is enabled by default.
126 This option allows to turn off the feature.
128 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump
131 acpi_no_static_ssdt [HW,ACPI]
132 Disable installation of static SSDTs at early boot time
133 By default, SSDTs contained in the RSDT/XSDT will be
134 installed automatically and they will appear under
135 /sys/firmware/acpi/tables.
136 This option turns off this feature.
137 Note that specifying this option does not affect
138 dynamic table installation which will install SSDT
139 tables to /sys/firmware/acpi/tables/dynamic.
141 acpi_no_watchdog [HW,ACPI,WDT]
142 Ignore the ACPI-based watchdog interface (WDAT) and let
143 a native driver control the watchdog device instead.
145 acpi_rsdp= [ACPI,EFI,KEXEC]
146 Pass the RSDP address to the kernel, mostly used
147 on machines running EFI runtime service to boot the
148 second kernel for kdump.
150 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS
151 Format: To spoof as Windows 98: ="Microsoft Windows"
153 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead
154 of 2 which is mandated by ACPI 6) as the supported ACPI
155 specification revision (when using this switch, it may
156 be necessary to carry out a cold reboot _twice_ in a
157 row to make it take effect on the platform firmware).
159 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings
160 acpi_osi="string1" # add string1
161 acpi_osi="!string2" # remove string2
162 acpi_osi=!* # remove all strings
163 acpi_osi=! # disable all built-in OS vendor
165 acpi_osi=!! # enable all built-in OS vendor
167 acpi_osi= # disable all strings
169 'acpi_osi=!' can be used in combination with single or
170 multiple 'acpi_osi="string1"' to support specific OS
171 vendor string(s). Note that such command can only
172 affect the default state of the OS vendor strings, thus
173 it cannot affect the default state of the feature group
174 strings and the current state of the OS vendor strings,
175 specifying it multiple times through kernel command line
176 is meaningless. This command is useful when one do not
177 care about the state of the feature group strings which
178 should be controlled by the OSPM.
180 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent
181 to 'acpi_osi="Windows 2000" acpi_osi=!', they all
182 can make '_OSI("Windows 2000")' TRUE.
184 'acpi_osi=' cannot be used in combination with other
185 'acpi_osi=' command lines, the _OSI method will not
186 exist in the ACPI namespace. NOTE that such command can
187 only affect the _OSI support state, thus specifying it
188 multiple times through kernel command line is also
191 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)'
194 'acpi_osi=!*' can be used in combination with single or
195 multiple 'acpi_osi="string1"' to support specific
196 string(s). Note that such command can affect the
197 current state of both the OS vendor strings and the
198 feature group strings, thus specifying it multiple times
199 through kernel command line is meaningful. But it may
200 still not able to affect the final state of a string if
201 there are quirks related to this string. This command
202 is useful when one want to control the state of the
203 feature group strings to debug BIOS issues related to
206 1. 'acpi_osi="Module Device" acpi_osi=!*' can make
207 '_OSI("Module Device")' FALSE.
208 2. 'acpi_osi=!* acpi_osi="Module Device"' can make
209 '_OSI("Module Device")' TRUE.
210 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is
212 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"'
214 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!',
215 they all will make '_OSI("Windows 2000")' TRUE.
218 Override the pmtimer bug detection: force the kernel
219 to assume that this machine's pmtimer latches its value
220 and always returns good values.
222 acpi_sci= [HW,ACPI] ACPI System Control Interrupt trigger mode
223 Format: { level | edge | high | low }
225 acpi_skip_timer_override [HW,ACPI]
226 Recognize and ignore IRQ0/pin2 Interrupt Override.
227 For broken nForce2 BIOS resulting in XT-PIC timer.
229 acpi_sleep= [HW,ACPI] Sleep options
230 Format: { s3_bios, s3_mode, s3_beep, s4_nohwsig,
231 old_ordering, nonvs, sci_force_enable, nobl }
232 See Documentation/power/video.rst for information on
234 s3_beep is for debugging; it makes the PC's speaker beep
235 as soon as the kernel's real-mode entry point is called.
236 s4_nohwsig prevents ACPI hardware signature from being
237 used during resume from hibernation.
238 old_ordering causes the ACPI 1.0 ordering of the _PTS
239 control method, with respect to putting devices into
240 low power states, to be enforced (the ACPI 2.0 ordering
241 of _PTS is used by default).
242 nonvs prevents the kernel from saving/restoring the
243 ACPI NVS memory during suspend/hibernation and resume.
244 sci_force_enable causes the kernel to set SCI_EN directly
245 on resume from S1/S3 (which is against the ACPI spec,
246 but some broken systems don't work without it).
247 nobl causes the internal blacklist of systems known to
248 behave incorrectly in some ways with respect to system
249 suspend and resume to be ignored (use wisely).
251 acpi_use_timer_override [HW,ACPI]
252 Use timer override. For some broken Nvidia NF5 boards
253 that require a timer override, but don't have HPET
255 add_efi_memmap [EFI; X86] Include EFI memory map in
256 kernel's map of available physical RAM.
259 { off | try_unsupported }
260 off: disable AGP support
261 try_unsupported: try to drive unsupported chipsets
262 (may crash computer or cause data corruption)
265 See Documentation/sound/alsa-configuration.rst
268 Allow the default userspace alignment fault handler
269 behaviour to be specified. Bit 0 enables warnings,
270 bit 1 enables fixups, and bit 2 sends a segfault.
272 align_va_addr= [X86-64]
273 Align virtual addresses by clearing slice [14:12] when
274 allocating a VMA at process creation time. This option
275 gives you up to 3% performance improvement on AMD F15h
276 machines (where it is enabled by default) for a
277 CPU-intensive style benchmark, and it can vary highly in
278 a microbenchmark depending on workload and compiler.
280 32: only for 32-bit processes
281 64: only for 64-bit processes
282 on: enable for both 32- and 64-bit processes
283 off: disable for both 32- and 64-bit processes
285 alloc_snapshot [FTRACE]
286 Allocate the ftrace snapshot buffer on boot up when the
287 main buffer is allocated. This is handy if debugging
288 and you need to use tracing_snapshot() on boot up, and
289 do not want to use tracing_snapshot_alloc() as it needs
290 to be done where GFP_KERNEL allocations are allowed.
292 amd_iommu= [HW,X86-64]
293 Pass parameters to the AMD IOMMU driver in the system.
295 fullflush - enable flushing of IO/TLB entries when
296 they are unmapped. Otherwise they are
297 flushed before they will be reused, which
299 off - do not initialize any AMD IOMMU found in
301 force_isolation - Force device isolation for all
302 devices. The IOMMU driver is not
303 allowed anymore to lift isolation
304 requirements as needed. This option
305 does not override iommu=pt
307 amd_iommu_dump= [HW,X86-64]
308 Enable AMD IOMMU driver option to dump the ACPI table
309 for AMD IOMMU. With this option enabled, AMD IOMMU
310 driver will print ACPI tables for AMD IOMMU during
311 IOMMU initialization.
313 amd_iommu_intr= [HW,X86-64]
314 Specifies one of the following AMD IOMMU interrupt
316 legacy - Use legacy interrupt remapping mode.
317 vapic - Use virtual APIC mode, which allows IOMMU
318 to inject interrupts directly into guest.
319 This mode requires kvm-amd.avic=1.
320 (Default when IOMMU HW support is present.)
322 amijoy.map= [HW,JOY] Amiga joystick support
323 Map of devices attached to JOY0DAT and JOY1DAT
325 See also Documentation/input/joydev/joystick.rst
327 analog.map= [HW,JOY] Analog joystick and gamepad support
328 Specifies type or capabilities of an analog joystick
329 connected to one of 16 gameports
330 Format: <type1>,<type2>,..<type16>
333 Power management functions (SPARCstation-4/5 + deriv.)
335 Disable APC CPU standby support. SPARCstation-Fox does
336 not play well with APC CPU idle - disable it if you have
337 APC and your system crashes randomly.
339 apic= [APIC,X86] Advanced Programmable Interrupt Controller
340 Change the output verbosity while booting
341 Format: { quiet (default) | verbose | debug }
342 Change the amount of debugging information output
343 when initialising the APIC and IO-APIC components.
344 For X86-32, this can also be used to specify an APIC
346 Format: apic=driver_name
347 Examples: apic=bigsmp
349 apic_extnmi= [APIC,X86] External NMI delivery setting
350 Format: { bsp (default) | all | none }
351 bsp: External NMI is delivered only to CPU 0
352 all: External NMIs are broadcast to all CPUs as a
354 none: External NMI is masked for all CPUs. This is
355 useful so that a dump capture kernel won't be
359 See Documentation/networking/ipv6.rst.
361 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
362 Limit apic dumping. The parameter defines the maximal
363 number of local apics being dumped. Also it is possible
364 to set it to "all" by meaning -- no limit here.
365 Format: { 1 (default) | 2 | ... | all }.
366 The parameter valid if only apic=debug or
367 apic=verbose is specified.
368 Example: apic=debug show_lapic=all
370 apm= [APM] Advanced Power Management
371 See header of arch/x86/kernel/apm_32.c.
373 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
374 Format: <io>,<irq>,<nodeID>
376 arm64.nobti [ARM64] Unconditionally disable Branch Target
377 Identification support
379 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication
384 atarimouse= [HW,MOUSE] Atari Mouse
386 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
387 EzKey and similar keyboards
389 atkbd.reset= [HW] Reset keyboard during initialization
391 atkbd.set= [HW] Select keyboard code set
392 Format: <int> (2 = AT (default), 3 = PS/2)
394 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
397 atkbd.softraw= [HW] Choose between synthetic and real raw mode
398 Format: <bool> (0 = real, 1 = synthetic (default))
400 atkbd.softrepeat= [HW]
401 Use software keyboard repeat
403 audit= [KNL] Enable the audit sub-system
404 Format: { "0" | "1" | "off" | "on" }
405 0 | off - kernel audit is disabled and can not be
406 enabled until the next reboot
407 unset - kernel audit is initialized but disabled and
408 will be fully enabled by the userspace auditd.
409 1 | on - kernel audit is initialized and partially
410 enabled, storing at most audit_backlog_limit
411 messages in RAM until it is fully enabled by the
415 audit_backlog_limit= [KNL] Set the audit queue size limit.
416 Format: <int> (must be >=0)
419 bau= [X86_UV] Enable the BAU on SGI UV. The default
420 behavior is to disable the BAU (i.e. bau=0).
421 Format: { "0" | "1" }
424 unset - Disable the BAU.
426 baycom_epp= [HW,AX25]
429 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
431 See header of drivers/net/hamradio/baycom_par.c.
433 baycom_ser_fdx= [HW,AX25]
434 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
435 Format: <io>,<irq>,<mode>[,<baud>]
436 See header of drivers/net/hamradio/baycom_ser_fdx.c.
438 baycom_ser_hdx= [HW,AX25]
439 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
440 Format: <io>,<irq>,<mode>
441 See header of drivers/net/hamradio/baycom_ser_hdx.c.
443 blkdevparts= Manual partition parsing of block device(s) for
444 embedded devices based on command line input.
445 See Documentation/block/cmdline-partition.rst
447 boot_delay= Milliseconds to delay each printk during boot.
448 Values larger than 10 seconds (10000) are changed to
453 Extended command line options can be added to an initrd
454 and this will cause the kernel to look for it.
456 See Documentation/admin-guide/bootconfig.rst
459 Disable BERT OS support on buggy BIOSes.
461 bgrt_disable [ACPI][X86]
462 Disable BGRT to avoid flickering OEM logo.
464 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
465 bttv.radio= Most important insmod options are available as
467 bttv.pll= See Documentation/admin-guide/media/bttv.rst
470 bulk_remove=off [PPC] This parameter disables the use of the pSeries
471 firmware feature for flushing multiple hpte entries
474 c101= [NET] Moxa C101 synchronous serial card
476 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
477 Sometimes CPU hardware bugs make them report the cache
478 size incorrectly. The kernel will attempt work arounds
479 to fix known problems, but for some CPUs it is not
480 possible to determine what the correct size should be.
481 This option provides an override for these situations.
484 [NET] Specifies amount of time (in seconds) that
485 the kernel should wait for a network carrier. By default
486 it waits 120 seconds.
488 ca_keys= [KEYS] This parameter identifies a specific key(s) on
489 the system trusted keyring to be used for certificate
491 format: { id:<keyid> | builtin }
493 cca= [MIPS] Override the kernel pages' cache coherency
494 algorithm. Accepted values range from 0 to 7
495 inclusive. See arch/mips/include/asm/pgtable-bits.h
496 for platform specific values (SB1, Loongson3 and
499 ccw_timeout_log [S390]
500 See Documentation/s390/common_io.rst for details.
502 cgroup_disable= [KNL] Disable a particular controller
503 Format: {name of the controller(s) to disable}
504 The effects of cgroup_disable=foo are:
505 - foo isn't auto-mounted if you mount all cgroups in
507 - foo isn't visible as an individually mountable
509 {Currently only "memory" controller deal with this and
510 cut the overhead, others just disable the usage. So
511 only cgroup_disable=memory is actually worthy}
513 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
514 Format: { { controller | "all" | "named" }
515 [,{ controller | "all" | "named" }...] }
516 Like cgroup_disable, but only applies to cgroup v1;
517 the blacklisted controllers remain available in cgroup2.
518 "all" blacklists all controllers and "named" disables
519 named mounts. Specifying both "all" and "named" disables
522 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
524 nosocket -- Disable socket memory accounting.
525 nokmem -- Disable kernel memory accounting.
527 checkreqprot [SELINUX] Set initial checkreqprot flag value.
528 Format: { "0" | "1" }
529 See security/selinux/Kconfig help text.
530 0 -- check protection applied by kernel (includes
531 any implied execute protection).
532 1 -- check protection requested by application.
533 Default value is set via a kernel config option.
534 Value can be changed at runtime via
535 /sys/fs/selinux/checkreqprot.
536 Setting checkreqprot to 1 is deprecated.
539 See Documentation/s390/common_io.rst for details.
542 Prevents the clock framework from automatically gating
543 clocks that have not been explicitly enabled by a Linux
544 device driver but are enabled in hardware at reset or
545 by the bootloader/firmware. Note that this does not
546 force such clocks to be always-on nor does it reserve
547 those clocks in any way. This parameter is useful for
548 debug and development, but should not be needed on a
549 platform with proper driver support. For more
550 information, see Documentation/driver-api/clk.rst.
552 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
554 Forces specified clocksource (if available) to be used
555 when calculating gettimeofday(). If specified
556 clocksource is not available, it defaults to PIT.
557 Format: { pit | tsc | cyclone | pmtmr }
559 clocksource= Override the default clocksource
561 Override the default clocksource and use the clocksource
562 with the name specified.
563 Some clocksource names to choose from, depending on
565 [all] jiffies (this is the base, fallback clocksource)
567 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
568 pxa_timer,timer3,32k_counter,timer0_1
569 [X86-32] pit,hpet,tsc;
570 scx200_hrt on Geode; cyclone on IBM x440
578 clocksource.arm_arch_timer.evtstrm=
581 Enable/disable the eventstream feature of the ARM
582 architected timer so that code using WFE-based polling
583 loops can be debugged more effectively on production
586 clearcpuid=BITNUM[,BITNUM...] [X86]
587 Disable CPUID feature X for the kernel. See
588 arch/x86/include/asm/cpufeatures.h for the valid bit
589 numbers. Note the Linux specific bits are not necessarily
590 stable over kernel options, but the vendor specific
592 Also note that user programs calling CPUID directly
593 or using the feature without checking anything
594 will still see it. This just prevents it from
595 being used by the kernel or shown in /proc/cpuinfo.
596 Also note the kernel might malfunction if you disable
599 cma=nn[MG]@[start[MG][-end[MG]]]
601 Sets the size of kernel global memory area for
602 contiguous memory allocations and optionally the
603 placement constraint by the physical address range of
604 memory allocations. A value of 0 disables CMA
605 altogether. For more information, see
606 kernel/dma/contiguous.c
610 Sets the size of kernel per-numa memory area for
611 contiguous memory allocations. A value of 0 disables
612 per-numa CMA altogether. And If this option is not
613 specificed, the default value is 0.
614 With per-numa CMA enabled, DMA users on node nid will
615 first try to allocate buffer from the pernuma area
616 which is located in node nid, if the allocation fails,
617 they will fallback to the global default memory area.
619 cmo_free_hint= [PPC] Format: { yes | no }
620 Specify whether pages are marked as being inactive
621 when they are freed. This is used in CMO environments
622 to determine OS memory pressure for page stealing by
626 coherent_pool=nn[KMG] [ARM,KNL]
627 Sets the size of memory pool for coherent, atomic dma
628 allocations, by default set to 256K.
630 com20020= [HW,NET] ARCnet - COM20020 chipset
632 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
634 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
638 ARCnet - COM90xx chipset (memory-mapped buffers)
639 Format: <io>[,<irq>[,<memstart>]]
641 condev= [HW,S390] console device
644 console= [KNL] Output console device and options.
646 tty<n> Use the virtual console device <n>.
650 Use the specified serial port. The options are of
651 the form "bbbbpnf", where "bbbb" is the baud rate,
652 "p" is parity ("n", "o", or "e"), "n" is number of
653 bits, and "f" is flow control ("r" for RTS or
654 omit it). Default is "9600n8".
656 See Documentation/admin-guide/serial-console.rst for more
658 Documentation/networking/netconsole.rst for an
661 uart[8250],io,<addr>[,options]
662 uart[8250],mmio,<addr>[,options]
663 uart[8250],mmio16,<addr>[,options]
664 uart[8250],mmio32,<addr>[,options]
665 uart[8250],0x<addr>[,options]
666 Start an early, polled-mode console on the 8250/16550
667 UART at the specified I/O port or MMIO address,
668 switching to the matching ttyS device later.
669 MMIO inter-register address stride is either 8-bit
670 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
671 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
672 to be equivalent to 'mmio'. 'options' are specified in
673 the same format described for ttyS above; if unspecified,
674 the h/w is not re-initialized.
676 hvc<n> Use the hypervisor console device <n>. This is for
677 both Xen and PowerPC hypervisors.
679 If the device connected to the port is not a TTY but a braille
680 device, prepend "brl," before the device type, for instance
682 For now, only VisioBraille is supported.
685 [KNL] Change console messages format
687 By default we print messages on consoles in
688 "[time stamp] text\n" format (time stamp may not be
689 printed, depending on CONFIG_PRINTK_TIME or
690 `printk_time' param).
692 Switch to syslog format: "<%u>[time stamp] text\n"
693 IOW, each message will have a facility and loglevel
694 prefix. The format is similar to one used by syslog()
695 syscall, or to executing "dmesg -S --raw" or to reading
698 consoleblank= [KNL] The console blank (screen saver) timeout in
699 seconds. A value of 0 disables the blank timer.
703 [KNL] Change the default value for
704 /proc/<pid>/coredump_filter.
705 See also Documentation/filesystems/proc.rst.
707 coresight_cpu_debug.enable
710 Enable/disable the CPU sampling based debugging.
711 0: default value, disable debugging
712 1: enable debugging at boot time
714 cpuidle.off=1 [CPU_IDLE]
715 disable the cpuidle sub-system
718 [CPU_IDLE] Name of the cpuidle governor to use.
720 cpufreq.off=1 [CPU_FREQ]
721 disable the cpufreq sub-system
723 cpufreq.default_governor=
724 [CPU_FREQ] Name of the default cpufreq governor or
725 policy to use. This governor must be registered in the
726 kernel before the cpufreq driver probes.
729 [X86] Delay for N microsec between assert and de-assert
730 of APIC INIT to start processors. This delay occurs
731 on every CPU online, such as boot, and resume from suspend.
734 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
736 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
738 crashkernel=size[KMG][@offset[KMG]]
739 [KNL] Using kexec, Linux can switch to a 'crash kernel'
740 upon panic. This parameter reserves the physical
741 memory region [offset, offset + size] for that kernel
742 image. If '@offset' is omitted, then a suitable offset
743 is selected automatically.
744 [KNL, X86-64] Select a region under 4G first, and
745 fall back to reserve region above 4G when '@offset'
746 hasn't been specified.
747 See Documentation/admin-guide/kdump/kdump.rst for further details.
749 crashkernel=range1:size1[,range2:size2,...][@offset]
750 [KNL] Same as above, but depends on the memory
751 in the running system. The syntax of range is
752 start-[end] where start and end are both
753 a memory unit (amount[KMG]). See also
754 Documentation/admin-guide/kdump/kdump.rst for an example.
756 crashkernel=size[KMG],high
757 [KNL, X86-64] range could be above 4G. Allow kernel
758 to allocate physical memory region from top, so could
759 be above 4G if system have more than 4G ram installed.
760 Otherwise memory region will be allocated below 4G, if
762 It will be ignored if crashkernel=X is specified.
763 crashkernel=size[KMG],low
764 [KNL, X86-64] range under 4G. When crashkernel=X,high
765 is passed, kernel could allocate physical memory region
766 above 4G, that cause second kernel crash on system
767 that require some amount of low memory, e.g. swiotlb
768 requires at least 64M+32K low memory, also enough extra
769 low memory is needed to make sure DMA buffers for 32-bit
770 devices won't run out. Kernel would try to allocate at
771 at least 256M below 4G automatically.
772 This one let user to specify own low range under 4G
773 for second kernel instead.
774 0: to disable low allocation.
775 It will be ignored when crashkernel=X,high is not used
776 or memory reserved is below 4G.
779 [KNL] Disable crypto self-tests
784 cs89x0_media= [HW,NET]
785 Format: { rj45 | aui | bnc }
788 See header of drivers/s390/block/dasd_devmap.c.
790 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
791 (one device per port)
792 Format: <port#>,<type>
793 See also Documentation/input/devices/joystick-parport.rst
795 ddebug_query= [KNL,DYNAMIC_DEBUG] Enable debug messages at early boot
797 Documentation/admin-guide/dynamic-debug-howto.rst for
798 details. Deprecated, see dyndbg.
800 debug [KNL] Enable kernel debugging (events log level).
803 [KNL] Enable printing [hashed] pointers early in the
804 boot sequence. If enabled, we use a weak hash instead
805 of siphash to hash pointers. Use this option if you are
806 seeing instances of '(___ptrval___)') and need to see a
807 value (hashed pointer) instead. Cryptographically
808 insecure, please do not use on production kernels.
811 [KNL] verbose locking self-tests
813 Print debugging info while doing the locking API
815 Bitmask for the various LOCKTYPE_ tests. Defaults to 0
816 (no extra messages), setting it to -1 (all bits set)
817 will print _a_lot_ more information - normally only
818 useful to lockdep developers.
820 debug_objects [KNL] Enable object debugging
823 [KNL] Disable object debugging
825 debug_guardpage_minorder=
826 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
827 parameter allows control of the order of pages that will
828 be intentionally kept free (and hence protected) by the
829 buddy allocator. Bigger value increase the probability
830 of catching random memory corruption, but reduce the
831 amount of memory for normal system use. The maximum
832 possible value is MAX_ORDER/2. Setting this parameter
833 to 1 or 2 should be enough to identify most random
834 memory corruption problems caused by bugs in kernel or
835 driver code when a CPU writes to (or reads from) a
836 random memory location. Note that there exists a class
837 of memory corruptions problems caused by buggy H/W or
838 F/W or by drivers badly programing DMA (basically when
839 memory is written at bus level and the CPU MMU is
840 bypassed) which are not detectable by
841 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
842 tracking down these problems.
845 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
846 enables the feature at boot time. By default, it is
847 disabled and the system will work mostly the same as a
848 kernel built without CONFIG_DEBUG_PAGEALLOC.
849 Note: to get most of debug_pagealloc error reports, it's
850 useful to also enable the page_owner functionality.
851 on: enable the feature
853 debugfs= [KNL] This parameter enables what is exposed to userspace
854 and debugfs internal clients.
855 Format: { on, no-mount, off }
856 on: All functions are enabled.
858 Filesystem is not registered but kernel clients can
859 access APIs and a crashkernel can be used to read
860 its content. There is nothing to mount.
861 off: Filesystem is not registered and clients
862 get a -EPERM as result when trying to register files
863 or directories within debugfs.
864 This is equivalent of the runtime functionality if
865 debugfs was not enabled in the kernel at all.
866 Default value is set in build-time with a kernel configuration.
868 debugpat [X86] Enable PAT debugging
870 decnet.addr= [HW,NET]
871 Format: <area>[,<node>]
872 See also Documentation/networking/decnet.rst.
875 [HW] The size of the default HugeTLB page. This is
876 the size represented by the legacy /proc/ hugepages
877 APIs. In addition, this is the default hugetlb size
878 used for shmget(), mmap() and mounting hugetlbfs
879 filesystems. If not specified, defaults to the
880 architecture's default huge page size. Huge page
881 sizes are architecture dependent. See also
882 Documentation/admin-guide/mm/hugetlbpage.rst.
885 deferred_probe_timeout=
886 [KNL] Debugging option to set a timeout in seconds for
887 deferred probe to give up waiting on dependencies to
888 probe. Only specific dependencies (subsystems or
889 drivers) that have opted in will be ignored. A timeout of 0
890 will timeout at the end of initcalls. This option will also
891 dump out devices still on the deferred probe list after
895 Format: { on | off | def_only | inf_only | always }
896 on: s390 zlib hardware support for compression on
897 level 1 and decompression (default)
898 off: No s390 zlib hardware support
899 def_only: s390 zlib hardware support for deflate
900 only (compression on level 1)
901 inf_only: s390 zlib hardware support for inflate
903 always: Same as 'on' but ignores the selected compression
904 level always using hardware support (used for debugging)
907 Set number of hash buckets for dentry cache.
909 disable_1tb_segments [PPC]
910 Disables the use of 1TB hash page table segments. This
911 causes the kernel to fall back to 256MB segments which
912 can be useful when debugging issues that require an SLB
916 Limits the number of kernel SLB entries, and flushes
917 them frequently to increase the rate of SLB faults
921 See Documentation/networking/ipv6.rst.
924 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
925 hardening is enabled for this boot. Hardened
926 usercopy checking is used to protect the kernel
927 from reading or writing beyond known memory
928 allocation boundaries as a proactive defense
929 against bounds-checking flaws in the kernel's
930 copy_to_user()/copy_from_user() interface.
931 on Perform hardened usercopy checks (default).
932 off Disable hardened usercopy checks.
935 Disable RADIX MMU mode on POWER9
937 radix_hcall_invalidate=on [PPC/PSERIES]
938 Disable RADIX GTSE feature and use hcall for TLB
942 Disable TLBIE instruction. Currently does not work
943 with KVM, with HASH MMU, or with coherent accelerators.
945 disable_cpu_apicid= [X86,APIC,SMP]
947 The number of initial APIC ID for the
948 corresponding CPU to be disabled at boot,
949 mostly used for the kdump 2nd kernel to
950 disable BSP to wake up multiple CPUs without
951 causing system reset or hang due to sending
954 disable_ddw [PPC/PSERIES]
955 Disable Dynamic DMA Window support. Use this
956 to workaround buggy firmware.
959 See Documentation/networking/ipv6.rst.
961 disable_mtrr_cleanup [X86]
962 The kernel tries to adjust MTRR layout from continuous
963 to discrete, to make X server driver able to add WB
964 entry later. This parameter disables that.
966 disable_mtrr_trim [X86, Intel and AMD only]
967 By default the kernel will trim any uncacheable
968 memory out of your available memory pool based on
969 MTRR settings. This parameter disables that behavior,
970 possibly causing your machine to run very slowly.
972 disable_timer_pin_1 [X86]
973 Disable PIN 1 of APIC timer
974 Can be useful to work around chipset bugs.
976 dis_ucode_ldr [X86] Disable the microcode loader.
978 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
979 this option disables the debugging code at boot.
981 dma_debug_entries=<number>
982 This option allows to tune the number of preallocated
983 entries for DMA-API debugging code. One entry is
984 required per DMA-API allocation. Use this if the
985 DMA-API debugging code disables itself because the
986 architectural default is too low.
988 dma_debug_driver=<driver_name>
989 With this option the DMA-API debugging driver
990 filter feature can be enabled at boot time. Just
991 pass the driver to filter for as the parameter.
992 The filter can be disabled or changed to another
993 driver later using sysfs.
995 driver_async_probe= [KNL]
996 List of driver names to be probed asynchronously.
997 Format: <driver_name1>,<driver_name2>...
999 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
1000 Broken monitors, graphic adapters, KVMs and EDIDless
1001 panels may send no or incorrect EDID data sets.
1002 This parameter allows to specify an EDID data sets
1003 in the /lib/firmware directory that are used instead.
1004 Generic built-in EDID data sets are used, if one of
1005 edid/1024x768.bin, edid/1280x1024.bin,
1006 edid/1680x1050.bin, or edid/1920x1080.bin is given
1007 and no file with the same name exists. Details and
1008 instructions how to build your own EDID data are
1009 available in Documentation/admin-guide/edid.rst. An EDID
1010 data set will only be used for a particular connector,
1011 if its name and a colon are prepended to the EDID
1012 name. Each connector may use a unique EDID data
1013 set by separating the files with a comma. An EDID
1014 data set with no connector name will be used for
1015 any connectors not explicitly specified.
1020 Format: {"off" | "known"}
1021 Control how the dt_cpu_ftrs device-tree binding is
1022 used for CPU feature discovery and setup (if it
1024 off: Do not use it, fall back to legacy cpu table.
1025 known: Do not pass through unknown features to guests
1026 or userspace, only those that the kernel is aware of.
1028 dump_apple_properties [X86]
1029 Dump name and content of EFI device properties on
1030 x86 Macs. Useful for driver authors to determine
1031 what data is available or for reverse-engineering.
1033 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
1034 <module>.dyndbg[="val"]
1035 Enable debug messages at boot time. See
1036 Documentation/admin-guide/dynamic-debug-howto.rst
1039 nopku [X86] Disable Memory Protection Keys CPU feature found
1042 <module>.async_probe [KNL]
1043 Enable asynchronous probe on this module.
1045 early_ioremap_debug [KNL]
1046 Enable debug messages in early_ioremap support. This
1047 is useful for tracking down temporary early mappings
1048 which are not unmapped.
1050 earlycon= [KNL] Output early console device and options.
1052 When used with no options, the early console is
1053 determined by stdout-path property in device tree's
1054 chosen node or the ACPI SPCR table if supported by
1057 cdns,<addr>[,options]
1058 Start an early, polled-mode console on a Cadence
1059 (xuartps) serial port at the specified address. Only
1060 supported option is baud rate. If baud rate is not
1061 specified, the serial port must already be setup and
1064 uart[8250],io,<addr>[,options]
1065 uart[8250],mmio,<addr>[,options]
1066 uart[8250],mmio32,<addr>[,options]
1067 uart[8250],mmio32be,<addr>[,options]
1068 uart[8250],0x<addr>[,options]
1069 Start an early, polled-mode console on the 8250/16550
1070 UART at the specified I/O port or MMIO address.
1071 MMIO inter-register address stride is either 8-bit
1072 (mmio) or 32-bit (mmio32 or mmio32be).
1073 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1074 to be equivalent to 'mmio'. 'options' are specified
1075 in the same format described for "console=ttyS<n>"; if
1076 unspecified, the h/w is not initialized.
1080 Start an early, polled-mode console on a pl011 serial
1081 port at the specified address. The pl011 serial port
1082 must already be setup and configured. Options are not
1083 yet supported. If 'mmio32' is specified, then only
1084 the driver will use only 32-bit accessors to read/write
1085 the device registers.
1088 Start an early, polled-mode console on a meson serial
1089 port at the specified address. The serial port must
1090 already be setup and configured. Options are not yet
1094 Start an early, polled-mode console on an msm serial
1095 port at the specified address. The serial port
1096 must already be setup and configured. Options are not
1099 msm_serial_dm,<addr>
1100 Start an early, polled-mode console on an msm serial
1101 dm port at the specified address. The serial port
1102 must already be setup and configured. Options are not
1106 Start an early, polled-mode console on a serial port
1107 of an Actions Semi SoC, such as S500 or S900, at the
1108 specified address. The serial port must already be
1109 setup and configured. Options are not yet supported.
1112 Start an early, polled-mode console on a serial port
1113 of an RDA Micro SoC, such as RDA8810PL, at the
1114 specified address. The serial port must already be
1115 setup and configured. Options are not yet supported.
1118 Use RISC-V SBI (Supervisor Binary Interface) for early
1121 smh Use ARM semihosting calls for early console.
1129 Use early console provided by serial driver available
1130 on Samsung SoCs, requires selecting proper type and
1131 a correct base address of the selected UART port. The
1132 serial port must already be setup and configured.
1133 Options are not yet supported.
1136 Start an early, polled-mode console on a lantiq serial
1137 (lqasc) port at the specified address. The serial port
1138 must already be setup and configured. Options are not
1143 Use early console provided by Freescale LP UART driver
1144 found on Freescale Vybrid and QorIQ LS1021A processors.
1145 A valid base address must be provided, and the serial
1146 port must already be setup and configured.
1150 Start an early, polled-mode, output-only console on the
1151 Freescale i.MX UART at the specified address. The UART
1152 must already be setup and configured.
1155 Start an early, polled-mode console on the
1156 Armada 3700 serial port at the specified
1157 address. The serial port must already be setup
1158 and configured. Options are not yet supported.
1161 Start an early, polled-mode console on a Qualcomm
1162 Generic Interface (GENI) based serial port at the
1163 specified address. The serial port must already be
1164 setup and configured. Options are not yet supported.
1167 Start an early, unaccelerated console on the EFI
1168 memory mapped framebuffer (if available). On cache
1169 coherent non-x86 systems that use system memory for
1170 the framebuffer, pass the 'ram' option so that it is
1171 mapped with the correct attributes.
1174 Use early console provided by Freescale LINFlexD UART
1175 serial driver for NXP S32V234 SoCs. A valid base
1176 address must be provided, and the serial port must
1177 already be setup and configured.
1179 earlyprintk= [X86,SH,ARM,M68k,S390]
1183 earlyprintk=serial[,ttySn[,baudrate]]
1184 earlyprintk=serial[,0x...[,baudrate]]
1185 earlyprintk=ttySn[,baudrate]
1186 earlyprintk=dbgp[debugController#]
1187 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1188 earlyprintk=xdbc[xhciController#]
1190 earlyprintk is useful when the kernel crashes before
1191 the normal console is initialized. It is not enabled by
1192 default because it has some cosmetic problems.
1194 Append ",keep" to not disable it when the real console
1197 Only one of vga, efi, serial, or usb debug port can
1200 Currently only ttyS0 and ttyS1 may be specified by
1201 name. Other I/O ports may be explicitly specified
1202 on some architectures (x86 and arm at least) by
1203 replacing ttySn with an I/O port address, like this:
1204 earlyprintk=serial,0x1008,115200
1205 You can find the port for a given device in
1206 /proc/tty/driver/serial:
1207 2: uart:ST16650V2 port:00001008 irq:18 ...
1209 Interaction with the standard serial driver is not
1212 The VGA and EFI output is eventually overwritten by
1215 The xen output can only be used by Xen PV guests.
1217 The sclp output can only be used on s390.
1219 The optional "force" to "pciserial" enables use of a
1220 PCI device even when its classcode is not of the
1223 edac_report= [HW,EDAC] Control how to report EDAC event
1224 Format: {"on" | "off" | "force"}
1225 on: enable EDAC to report H/W event. May be overridden
1226 by other higher priority error reporting module.
1227 off: disable H/W event reporting through EDAC.
1228 force: enforce the use of EDAC to report H/W event.
1231 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1234 This is designed to be used in conjunction with
1235 the boot argument: earlyprintk=vga
1237 This parameter works in place of the kgdboc parameter
1238 but can only be used if the backing tty is available
1239 very early in the boot process. For early debugging
1240 via a serial port see kgdboc_earlycon instead.
1243 Format: {"off" | "on" | "skip[mbr]"}
1246 Format: { "debug", "disable_early_pci_dma",
1247 "nochunk", "noruntime", "nosoftreserve",
1248 "novamap", "no_disable_early_pci_dma" }
1249 debug: enable misc debug output.
1250 disable_early_pci_dma: disable the busmaster bit on all
1251 PCI bridges while in the EFI boot stub.
1252 nochunk: disable reading files in "chunks" in the EFI
1253 boot stub, as chunking can cause problems with some
1254 firmware implementations.
1255 noruntime : disable EFI runtime services support
1256 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1257 attribute may cause the kernel to reserve the
1258 memory range for a memory mapping driver to
1259 claim. Specify efi=nosoftreserve to disable this
1260 reservation and treat the memory by its base type
1261 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1262 novamap: do not call SetVirtualAddressMap().
1263 no_disable_early_pci_dma: Leave the busmaster bit set
1264 on all PCI bridges while in the EFI boot stub
1266 efi_no_storage_paranoia [EFI; X86]
1267 Using this parameter you can use more than 50% of
1268 your efi variable storage. Use this parameter only if
1269 you are really sure that your UEFI does sane gc and
1270 fulfills the spec otherwise your board may brick.
1272 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1273 Add arbitrary attribute to specific memory range by
1274 updating original EFI memory map.
1275 Region of memory which aa attribute is added to is
1278 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1279 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1280 attribute is added to range 0x100000000-0x180000000 and
1281 0x10a0000000-0x1120000000.
1283 If efi_fake_mem=8G@9G:0x40000 is specified, the
1284 EFI_MEMORY_SP(0x40000) attribute is added to
1285 range 0x240000000-0x43fffffff.
1287 Using this parameter you can do debugging of EFI memmap
1288 related features. For example, you can do debugging of
1289 Address Range Mirroring feature even if your box
1290 doesn't support it, or mark specific memory as
1293 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1294 that is to be dynamically loaded by Linux. If there are
1295 multiple variables with the same name but with different
1296 vendor GUIDs, all of them will be loaded. See
1297 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1300 eisa_irq_edge= [PARISC,HW]
1301 See header of drivers/parisc/eisa.c.
1304 See comment before function elanfreq_setup() in
1305 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1307 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1308 Specifies physical address of start of kernel core
1309 image elf header and optionally the size. Generally
1310 kexec loader will pass this option to capture kernel.
1311 See Documentation/admin-guide/kdump/kdump.rst for details.
1313 enable_mtrr_cleanup [X86]
1314 The kernel tries to adjust MTRR layout from continuous
1315 to discrete, to make X server driver able to add WB
1316 entry later. This parameter enables that.
1318 enable_timer_pin_1 [X86]
1319 Enable PIN 1 of APIC timer
1320 Can be useful to work around chipset bugs
1321 (in particular on some ATI chipsets).
1322 The kernel tries to set a reasonable default.
1324 enforcing [SELINUX] Set initial enforcing status.
1326 See security/selinux/Kconfig help text.
1327 0 -- permissive (log only, no denials).
1328 1 -- enforcing (deny and log).
1330 Value can be changed at runtime via
1331 /sys/fs/selinux/enforce.
1334 Disable Error Record Serialization Table (ERST)
1337 ether= [HW,NET] Ethernet cards parameters
1338 This option is obsoleted by the "netdev=" option, which
1339 has equivalent usage. See its documentation for details.
1343 Permit 'security.evm' to be updated regardless of
1344 current integrity status.
1349 fail_make_request=[KNL]
1350 General fault injection mechanism.
1351 Format: <interval>,<probability>,<space>,<times>
1352 See also Documentation/fault-injection/.
1355 Format: { initns | none }
1356 See Documentation/admin-guide/sysctl/net.rst for
1357 fb_tunnels_only_for_init_ns
1360 See Documentation/admin-guide/blockdev/floppy.rst.
1362 force_pal_cache_flush
1363 [IA-64] Avoid check_sal_cache_flush which may hang on
1364 buggy SAL_CACHE_FLUSH implementations. Using this
1365 parameter will force ia64_sal_cache_flush to call
1366 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1369 Forcefully enable Physical Address Extension (PAE).
1370 Many Pentium M systems disable PAE but may have a
1371 functionally usable PAE implementation.
1372 Warning: use of this parameter will taint the kernel
1373 and may cause unknown problems.
1376 [FTRACE] will set and start the specified tracer
1377 as early as possible in order to facilitate early
1380 ftrace_dump_on_oops[=orig_cpu]
1381 [FTRACE] will dump the trace buffers on oops.
1382 If no parameter is passed, ftrace will dump
1383 buffers of all CPUs, but if you pass orig_cpu, it will
1384 dump only the buffer of the CPU that triggered the
1387 ftrace_filter=[function-list]
1388 [FTRACE] Limit the functions traced by the function
1389 tracer at boot up. function-list is a comma-separated
1390 list of functions. This list can be changed at run
1391 time by the set_ftrace_filter file in the debugfs
1394 ftrace_notrace=[function-list]
1395 [FTRACE] Do not trace the functions specified in
1396 function-list. This list can be changed at run time
1397 by the set_ftrace_notrace file in the debugfs
1400 ftrace_graph_filter=[function-list]
1401 [FTRACE] Limit the top level callers functions traced
1402 by the function graph tracer at boot up.
1403 function-list is a comma-separated list of functions
1404 that can be changed at run time by the
1405 set_graph_function file in the debugfs tracing directory.
1407 ftrace_graph_notrace=[function-list]
1408 [FTRACE] Do not trace from the functions specified in
1409 function-list. This list is a comma-separated list of
1410 functions that can be changed at run time by the
1411 set_graph_notrace file in the debugfs tracing directory.
1413 ftrace_graph_max_depth=<uint>
1414 [FTRACE] Used with the function graph tracer. This is
1415 the max depth it will trace into a function. This value
1416 can be changed at run time by the max_graph_depth file
1417 in the tracefs tracing directory. default: 0 (no limit)
1419 fw_devlink= [KNL] Create device links between consumer and supplier
1420 devices by scanning the firmware to infer the
1421 consumer/supplier relationships. This feature is
1422 especially useful when drivers are loaded as modules as
1423 it ensures proper ordering of tasks like device probing
1424 (suppliers first, then consumers), supplier boot state
1425 clean up (only after all consumers have probed),
1426 suspend/resume & runtime PM (consumers first, then
1428 Format: { off | permissive | on | rpm }
1429 off -- Don't create device links from firmware info.
1430 permissive -- Create device links from firmware info
1431 but use it only for ordering boot state clean
1432 up (sync_state() calls).
1433 on -- Create device links from firmware info and use it
1434 to enforce probe and suspend/resume ordering.
1435 rpm -- Like "on", but also use to order runtime PM.
1437 fw_devlink.strict=<bool>
1438 [KNL] Treat all inferred dependencies as mandatory
1439 dependencies. This only applies for fw_devlink=on|rpm.
1443 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1444 support via parallel port (up to 5 devices per port)
1445 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1446 See also Documentation/input/devices/joystick-parport.rst
1450 gart_fix_e820= [X86-64] disable the fix e820 for K8 GART
1454 gcov_persist= [GCOV] When non-zero (default), profiling data for
1455 kernel modules is saved and remains accessible via
1456 debugfs, even when the module is unloaded/reloaded.
1457 When zero, profiling data is discarded and associated
1458 debugfs files are removed at module unload time.
1460 goldfish [X86] Enable the goldfish android emulator platform.
1461 Don't use this when you are not running on the
1464 gpt [EFI] Forces disk with valid GPT signature but
1465 invalid Protective MBR to be treated as GPT. If the
1466 primary GPT is corrupted, it enables the backup/alternate
1467 GPT to be used instead.
1469 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1470 the "Enable 0" bit of the configuration register.
1473 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1474 the "Enable 0" bit of the configuration register.
1477 grcan.select= [HW] Select which physical interface to use.
1480 grcan.txsize= [HW] Sets the size of the tx buffer.
1481 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1483 grcan.rxsize= [HW] Sets the size of the rx buffer.
1484 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1487 gpio-mockup.gpio_mockup_ranges
1488 [HW] Sets the ranges of gpiochip of for this device.
1489 Format: <start1>,<end1>,<start2>,<end2>...
1491 hardlockup_all_cpu_backtrace=
1492 [KNL] Should the hard-lockup detector generate
1493 backtraces on all cpus.
1496 hashdist= [KNL,NUMA] Large hashes allocated during boot
1497 are distributed across NUMA nodes. Defaults on
1498 for 64-bit NUMA, off otherwise.
1499 Format: 0 | 1 (for off | on)
1501 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1503 hd= [EIDE] (E)IDE hard drive subsystem geometry
1504 Format: <cyl>,<head>,<sect>
1507 Disable Hardware Error Source Table (HEST) support;
1508 corresponding firmware-first mode error processing
1509 logic will be disabled.
1511 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1512 size of <nn>. This works even on boxes that have no
1513 highmem otherwise. This also works to reduce highmem
1514 size on bigger boxes.
1516 highres= [KNL] Enable/disable high resolution timer mode.
1517 Valid parameters: "on", "off"
1522 hpet= [X86-32,HPET] option to control HPET usage
1523 Format: { enable (default) | disable | force |
1525 disable: disable HPET and use PIT instead
1526 force: allow force enabled of undocumented chips (ICH4,
1528 verbose: show contents of HPET registers during setup
1530 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1531 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1533 hugetlb_cma= [HW,CMA] The size of a CMA area used for allocation
1534 of gigantic hugepages.
1537 Reserve a CMA area of given size and allocate gigantic
1538 hugepages using the CMA allocator. If enabled, the
1539 boot-time allocation of gigantic hugepages is skipped.
1541 hugepages= [HW] Number of HugeTLB pages to allocate at boot.
1542 If this follows hugepagesz (below), it specifies
1543 the number of pages of hugepagesz to be allocated.
1544 If this is the first HugeTLB parameter on the command
1545 line, it specifies the number of pages to allocate for
1546 the default huge page size. See also
1547 Documentation/admin-guide/mm/hugetlbpage.rst.
1551 [HW] The size of the HugeTLB pages. This is used in
1552 conjunction with hugepages (above) to allocate huge
1553 pages of a specific size at boot. The pair
1554 hugepagesz=X hugepages=Y can be specified once for
1555 each supported huge page size. Huge page sizes are
1556 architecture dependent. See also
1557 Documentation/admin-guide/mm/hugetlbpage.rst.
1561 [KNL] Should the hung task detector generate panics.
1564 A value of 1 instructs the kernel to panic when a
1565 hung task is detected. The default value is controlled
1566 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1567 option. The value selected by this boot parameter can
1568 be changed later by the kernel.hung_task_panic sysctl.
1570 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1571 terminal devices. Valid values: 0..8
1572 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1573 If specified, z/VM IUCV HVC accepts connections
1574 from listed z/VM user IDs only.
1576 hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations
1577 which allow the hypervisor to 'idle' the
1578 guest on lock contention.
1581 Do not unregister boot console at start. This is only
1582 useful for debugging when something happens in the window
1583 between unregistering the boot console and initializing
1586 i2c_bus= [HW] Override the default board specific I2C bus speed
1587 or register an additional I2C bus that is not
1588 registered from board initialization code.
1592 i8042.debug [HW] Toggle i8042 debug mode
1593 i8042.unmask_kbd_data
1594 [HW] Enable printing of interrupt data from the KBD port
1595 (disabled by default, and as a pre-condition
1596 requires that i8042.debug=1 be enabled)
1597 i8042.direct [HW] Put keyboard port into non-translated mode
1598 i8042.dumbkbd [HW] Pretend that controller can only read data from
1599 keyboard and cannot control its state
1600 (Don't attempt to blink the leds)
1601 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1602 i8042.nokbd [HW] Don't check/create keyboard port
1603 i8042.noloop [HW] Disable the AUX Loopback command while probing
1605 i8042.nomux [HW] Don't check presence of an active multiplexing
1607 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1609 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1610 i8042.reset [HW] Reset the controller during init, cleanup and
1611 suspend-to-ram transitions, only during s2r
1612 transitions, or never reset
1613 Format: { 1 | Y | y | 0 | N | n }
1614 1, Y, y: always reset controller
1615 0, N, n: don't ever reset controller
1616 Default: only on s2r transitions on x86; most other
1617 architectures force reset to be always executed
1618 i8042.unlock [HW] Unlock (ignore) the keylock
1619 i8042.kbdreset [HW] Reset device connected to KBD port
1623 i8k.ignore_dmi [HW] Continue probing hardware even if DMI data
1624 indicates that the driver is running on unsupported
1626 i8k.force [HW] Activate i8k driver even if SMM BIOS signature
1627 does not match list of supported models.
1629 [HW] Report power status in /proc/i8k
1630 (disabled by default)
1631 i8k.restricted [HW] Allow controlling fans only if SYS_ADMIN
1634 i915.invert_brightness=
1635 [DRM] Invert the sense of the variable that is used to
1636 set the brightness of the panel backlight. Normally a
1637 brightness value of 0 indicates backlight switched off,
1638 and the maximum of the brightness value sets the backlight
1639 to maximum brightness. If this parameter is set to 0
1640 (default) and the machine requires it, or this parameter
1641 is set to 1, a brightness value of 0 sets the backlight
1642 to maximum brightness, and the maximum of the brightness
1643 value switches the backlight off.
1644 -1 -- never invert brightness
1645 0 -- machine default
1646 1 -- force brightness inversion
1649 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1651 ide-core.nodma= [HW] (E)IDE subsystem
1652 Format: =0.0 to prevent dma on hda, =0.1 hdb =1.0 hdc
1653 .vlb_clock .pci_clock .noflush .nohpa .noprobe .nowerr
1654 .cdrom .chs .ignore_cable are additional options
1655 See Documentation/ide/ide.rst.
1657 ide-generic.probe-mask= [HW] (E)IDE subsystem
1659 Probe mask for legacy ISA IDE ports. Depending on
1660 platform up to 6 ports are supported, enabled by
1661 setting corresponding bits in the mask to 1. The
1662 default value is 0x0, which has a special meaning.
1663 On systems that have PCI, it triggers scanning the
1664 PCI bus for the first and the second port, which
1665 are then probed. On systems without PCI the value
1666 of 0x0 enables probing the two first ports as if it
1669 ide-pci-generic.all-generic-ide [HW] (E)IDE subsystem
1670 Claim all unknown PCI IDE storage controllers.
1673 Format: idle=poll, idle=halt, idle=nomwait
1674 Poll forces a polling idle loop that can slightly
1675 improve the performance of waking up a idle CPU, but
1676 will use a lot of power and make the system run hot.
1678 idle=halt: Halt is forced to be used for CPU idle.
1679 In such case C2/C3 won't be used again.
1680 idle=nomwait: Disable mwait for CPU C-states
1684 Allow force disabling of Shared Virtual Memory (SVA)
1685 support for the idxd driver. By default it is set to
1688 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1689 Format: { strict | legacy | 2008 | relaxed }
1692 Choose which programs will be accepted for execution
1693 based on the IEEE 754 NaN encoding(s) supported by
1694 the FPU and the NaN encoding requested with the value
1695 of an ELF file header flag individually set by each
1696 binary. Hardware implementations are permitted to
1697 support either or both of the legacy and the 2008 NaN
1700 Available settings are as follows:
1701 strict accept binaries that request a NaN encoding
1702 supported by the FPU
1703 legacy only accept legacy-NaN binaries, if supported
1705 2008 only accept 2008-NaN binaries, if supported
1707 relaxed accept any binaries regardless of whether
1708 supported by the FPU
1710 The FPU emulator is always able to support both NaN
1711 encodings, so if no FPU hardware is present or it has
1712 been disabled with 'nofpu', then the settings of
1713 'legacy' and '2008' strap the emulator accordingly,
1714 'relaxed' straps the emulator for both legacy-NaN and
1715 2008-NaN, whereas 'strict' enables legacy-NaN only on
1716 legacy processors and both NaN encodings on MIPS32 or
1719 The setting for ABS.fmt/NEG.fmt instruction execution
1720 mode generally follows that for the NaN encoding,
1721 except where unsupported by hardware.
1723 ignore_loglevel [KNL]
1724 Ignore loglevel setting - this will print /all/
1725 kernel messages to the console. Useful for debugging.
1726 We also add it as printk module parameter, so users
1727 could change it dynamically, usually by
1728 /sys/module/printk/parameters/ignore_loglevel.
1731 Ignore RLIMIT_DATA setting for data mappings,
1732 print warning at first misuse. Can be changed via
1733 /sys/module/kernel/parameters/ignore_rlimit_data.
1735 ihash_entries= [KNL]
1736 Set number of hash buckets for inode cache.
1738 ima_appraise= [IMA] appraise integrity measurements
1739 Format: { "off" | "enforce" | "fix" | "log" }
1742 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
1743 The builtin appraise policy appraises all files
1746 ima_canonical_fmt [IMA]
1747 Use the canonical format for the binary runtime
1748 measurements, instead of host native format.
1751 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1755 The list of supported hash algorithms is defined
1756 in crypto/hash_info.h.
1759 The builtin policies to load during IMA setup.
1760 Format: "tcb | appraise_tcb | secure_boot |
1761 fail_securely | critical_data"
1763 The "tcb" policy measures all programs exec'd, files
1764 mmap'd for exec, and all files opened with the read
1765 mode bit set by either the effective uid (euid=0) or
1768 The "appraise_tcb" policy appraises the integrity of
1769 all files owned by root.
1771 The "secure_boot" policy appraises the integrity
1772 of files (eg. kexec kernel image, kernel modules,
1773 firmware, policy, etc) based on file signatures.
1775 The "fail_securely" policy forces file signature
1776 verification failure also on privileged mounted
1777 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
1780 The "critical_data" policy measures kernel integrity
1783 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
1784 Load a policy which meets the needs of the Trusted
1785 Computing Base. This means IMA will measure all
1786 programs exec'd, files mmap'd for exec, and all files
1787 opened for read by uid=0.
1790 Select one of defined IMA measurements template formats.
1791 Formats: { "ima" | "ima-ng" | "ima-sig" }
1795 [IMA] Define a custom template format.
1796 Format: { "field1|...|fieldN" }
1798 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
1799 Format: <min_file_size>
1800 Set the minimal file size for using asynchronous hash.
1801 If left unspecified, ahash usage is disabled.
1803 ahash performance varies for different data sizes on
1804 different crypto accelerators. This option can be used
1805 to achieve the best performance for a particular HW.
1807 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
1809 Set hashing buffer size. Default: 4k.
1811 ahash performance varies for different chunk sizes on
1812 different crypto accelerators. This option can be used
1813 to achieve best performance for particular HW.
1817 Run specified binary instead of /sbin/init as init
1820 initcall_debug [KNL] Trace initcalls as they are executed. Useful
1821 for working out where the kernel is dying during
1824 initcall_blacklist= [KNL] Do not execute a comma-separated list of
1825 initcall functions. Useful for debugging built-in
1826 modules and initcalls.
1828 initrd= [BOOT] Specify the location of the initial ramdisk
1830 initrdmem= [KNL] Specify a physical address and size from which to
1831 load the initrd. If an initrd is compiled in or
1832 specified in the bootparams, it takes priority over this
1834 Format: ss[KMG],nn[KMG]
1837 init_on_alloc= [MM] Fill newly allocated pages and heap objects with
1840 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
1842 init_on_free= [MM] Fill freed pages and heap objects with zeroes.
1844 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
1846 init_pkru= [X86] Specify the default memory protection keys rights
1847 register contents for all processes. 0x55555554 by
1848 default (disallow access to all but pkey 0). Can
1849 override in debugfs after boot.
1851 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
1854 int_pln_enable [X86] Enable power limit notification interrupt
1856 integrity_audit=[IMA]
1857 Format: { "0" | "1" }
1858 0 -- basic integrity auditing messages. (Default)
1859 1 -- additional integrity auditing messages.
1861 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
1863 Enable intel iommu driver.
1865 Disable intel iommu driver.
1866 igfx_off [Default Off]
1867 By default, gfx is mapped as normal device. If a gfx
1868 device has a dedicated DMAR unit, the DMAR unit is
1869 bypassed by not enabling DMAR with this option. In
1870 this case, gfx device will use physical address for
1873 With this option iommu will not optimize to look
1874 for io virtual address below 32-bit forcing dual
1875 address cycle on pci bus for cards supporting greater
1876 than 32-bit addressing. The default is to look
1877 for translation below 32-bit and if not available
1878 then look in the higher range.
1879 strict [Default Off]
1880 With this option on every unmap_single operation will
1881 result in a hardware IOTLB flush operation as opposed
1882 to batching them for performance.
1883 sp_off [Default Off]
1884 By default, super page will be supported if Intel IOMMU
1885 has the capability. With this option, super page will
1888 By default, scalable mode will be disabled even if the
1889 hardware advertises that it has support for the scalable
1890 mode translation. With this option set, scalable mode
1891 will be used on hardware which claims to support it.
1892 tboot_noforce [Default Off]
1893 Do not force the Intel IOMMU enabled under tboot.
1894 By default, tboot will force Intel IOMMU on, which
1895 could harm performance of some high-throughput
1896 devices like 40GBit network cards, even if identity
1898 Note that using this option lowers the security
1899 provided by tboot because it makes the system
1900 vulnerable to DMA attacks.
1902 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
1903 0 disables intel_idle and fall back on acpi_idle.
1904 1 to 9 specify maximum depth of C-state.
1908 Do not enable intel_pstate as the default
1909 scaling driver for the supported processors
1911 Use intel_pstate as a scaling driver, but configure it
1912 to work with generic cpufreq governors (instead of
1913 enabling its internal governor). This mode cannot be
1914 used along with the hardware-managed P-states (HWP)
1917 Enable intel_pstate on systems that prohibit it by default
1918 in favor of acpi-cpufreq. Forcing the intel_pstate driver
1919 instead of acpi-cpufreq may disable platform features, such
1920 as thermal controls and power capping, that rely on ACPI
1921 P-States information being indicated to OSPM and therefore
1922 should be used with caution. This option does not work with
1923 processors that aren't supported by the intel_pstate driver
1924 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
1926 Do not enable hardware P state control (HWP)
1929 Only load intel_pstate on systems which support
1930 hardware P state control (HWP) if available.
1932 Enforce ACPI _PPC performance limits. If the Fixed ACPI
1933 Description Table, specifies preferred power management
1934 profile as "Enterprise Server" or "Performance Server",
1935 then this feature is turned on by default.
1937 Allow per-logical-CPU P-State performance control limits using
1938 cpufreq sysfs interface
1940 intremap= [X86-64, Intel-IOMMU]
1941 on enable Interrupt Remapping (default)
1942 off disable Interrupt Remapping
1943 nosid disable Source ID checking
1945 BIOS x2APIC opt-out request will be ignored
1946 nopost disable Interrupt Posting
1948 iomem= Disable strict checking of access to MMIO memory
1949 strict regions from userspace.
1964 nobypass [PPC/POWERNV]
1965 Disable IOMMU bypass, using IOMMU for PCI devices.
1967 iommu.strict= [ARM64] Configure TLB invalidation behaviour
1968 Format: { "0" | "1" }
1970 Request that DMA unmap operations use deferred
1971 invalidation of hardware TLBs, for increased
1972 throughput at the cost of reduced device isolation.
1973 Will fall back to strict mode if not supported by
1974 the relevant IOMMU driver.
1975 1 - Strict mode (default).
1976 DMA unmap operations invalidate IOMMU hardware TLBs
1980 [ARM64, X86] Configure DMA to bypass the IOMMU by default.
1981 Format: { "0" | "1" }
1982 0 - Use IOMMU translation for DMA.
1983 1 - Bypass the IOMMU for DMA.
1984 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
1986 io7= [HW] IO7 for Marvel-based Alpha systems
1987 See comment before marvel_specify_io7 in
1988 arch/alpha/kernel/core_marvel.c.
1990 io_delay= [X86] I/O delay method
1992 Standard port 0x80 based delay
1994 Alternate port 0xed based delay (needed on some systems)
1996 Simple two microseconds delay
2001 See Documentation/admin-guide/nfs/nfsroot.rst.
2003 ipcmni_extend [KNL] Extend the maximum number of unique System V
2004 IPC identifiers from 32,768 to 16,777,216.
2006 irqaffinity= [SMP] Set the default irq affinity mask
2007 The argument is a cpu list, as described above.
2009 irqchip.gicv2_force_probe=
2012 Force the kernel to look for the second 4kB page
2013 of a GICv2 controller even if the memory range
2014 exposed by the device tree is too small.
2016 irqchip.gicv3_nolpi=
2018 Force the kernel to ignore the availability of
2019 LPIs (and by consequence ITSs). Intended for system
2020 that use the kernel as a bootloader, and thus want
2021 to let secondary kernels in charge of setting up
2024 irqchip.gicv3_pseudo_nmi= [ARM64]
2025 Enables support for pseudo-NMIs in the kernel. This
2026 requires the kernel to be built with
2027 CONFIG_ARM64_PSEUDO_NMI.
2030 When an interrupt is not handled search all handlers
2031 for it. Intended to get systems with badly broken
2035 When an interrupt is not handled search all handlers
2036 for it. Also check all handlers each timer
2037 interrupt. Intended to get systems with badly broken
2041 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
2043 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
2044 [Deprecated - use cpusets instead]
2045 Format: [flag-list,]<cpu-list>
2047 Specify one or more CPUs to isolate from disturbances
2048 specified in the flag list (default: domain):
2051 Disable the tick when a single task runs.
2053 A residual 1Hz tick is offloaded to workqueues, which you
2054 need to affine to housekeeping through the global
2055 workqueue's affinity configured via the
2056 /sys/devices/virtual/workqueue/cpumask sysfs file, or
2057 by using the 'domain' flag described below.
2059 NOTE: by default the global workqueue runs on all CPUs,
2060 so to protect individual CPUs the 'cpumask' file has to
2061 be configured manually after bootup.
2064 Isolate from the general SMP balancing and scheduling
2065 algorithms. Note that performing domain isolation this way
2066 is irreversible: it's not possible to bring back a CPU to
2067 the domains once isolated through isolcpus. It's strongly
2068 advised to use cpusets instead to disable scheduler load
2069 balancing through the "cpuset.sched_load_balance" file.
2070 It offers a much more flexible interface where CPUs can
2071 move in and out of an isolated set anytime.
2073 You can move a process onto or off an "isolated" CPU via
2074 the CPU affinity syscalls or cpuset.
2075 <cpu number> begins at 0 and the maximum value is
2076 "number of CPUs in system - 1".
2080 Isolate from being targeted by managed interrupts
2081 which have an interrupt mask containing isolated
2082 CPUs. The affinity of managed interrupts is
2083 handled by the kernel and cannot be changed via
2084 the /proc/irq/* interfaces.
2086 This isolation is best effort and only effective
2087 if the automatically assigned interrupt mask of a
2088 device queue contains isolated and housekeeping
2089 CPUs. If housekeeping CPUs are online then such
2090 interrupts are directed to the housekeeping CPU
2091 so that IO submitted on the housekeeping CPU
2092 cannot disturb the isolated CPU.
2094 If a queue's affinity mask contains only isolated
2095 CPUs then this parameter has no effect on the
2096 interrupt routing decision, though interrupts are
2097 only delivered when tasks running on those
2098 isolated CPUs submit IO. IO submitted on
2099 housekeeping CPUs has no influence on those
2102 The format of <cpu-list> is described above.
2106 ivrs_ioapic [HW,X86-64]
2107 Provide an override to the IOAPIC-ID<->DEVICE-ID
2108 mapping provided in the IVRS ACPI table. For
2109 example, to map IOAPIC-ID decimal 10 to
2110 PCI device 00:14.0 write the parameter as:
2111 ivrs_ioapic[10]=00:14.0
2113 ivrs_hpet [HW,X86-64]
2114 Provide an override to the HPET-ID<->DEVICE-ID
2115 mapping provided in the IVRS ACPI table. For
2116 example, to map HPET-ID decimal 0 to
2117 PCI device 00:14.0 write the parameter as:
2118 ivrs_hpet[0]=00:14.0
2120 ivrs_acpihid [HW,X86-64]
2121 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2122 mapping provided in the IVRS ACPI table. For
2123 example, to map UART-HID:UID AMD0020:0 to
2124 PCI device 00:14.5 write the parameter as:
2125 ivrs_acpihid[00:14.5]=AMD0020:0
2127 js= [HW,JOY] Analog joystick
2128 See Documentation/input/joydev/joystick.rst.
2131 When CONFIG_RANDOMIZE_BASE is set, this disables
2132 kernel and module base offset ASLR (Address Space
2133 Layout Randomization).
2136 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2137 report on every invalid memory access. Without this
2138 parameter KASAN will print report only for the first
2143 kernelcore= [KNL,X86,IA-64,PPC]
2144 Format: nn[KMGTPE] | nn% | "mirror"
2145 This parameter specifies the amount of memory usable by
2146 the kernel for non-movable allocations. The requested
2147 amount is spread evenly throughout all nodes in the
2148 system as ZONE_NORMAL. The remaining memory is used for
2149 movable memory in its own zone, ZONE_MOVABLE. In the
2150 event, a node is too small to have both ZONE_NORMAL and
2151 ZONE_MOVABLE, kernelcore memory will take priority and
2152 other nodes will have a larger ZONE_MOVABLE.
2154 ZONE_MOVABLE is used for the allocation of pages that
2155 may be reclaimed or moved by the page migration
2156 subsystem. Note that allocations like PTEs-from-HighMem
2157 still use the HighMem zone if it exists, and the Normal
2158 zone if it does not.
2160 It is possible to specify the exact amount of memory in
2161 the form of "nn[KMGTPE]", a percentage of total system
2162 memory in the form of "nn%", or "mirror". If "mirror"
2163 option is specified, mirrored (reliable) memory is used
2164 for non-movable allocations and remaining memory is used
2165 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2166 are exclusive, so you cannot specify multiple forms.
2168 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
2169 Format: <Controller#>[,poll interval]
2170 The controller # is the number of the ehci usb debug
2171 port as it is probed via PCI. The poll interval is
2172 optional and is the number seconds in between
2173 each poll cycle to the debug port in case you need
2174 the functionality for interrupting the kernel with
2175 gdb or control-c on the dbgp connection. When
2176 not using this parameter you use sysrq-g to break into
2177 the kernel debugger.
2179 kgdboc= [KGDB,HW] kgdb over consoles.
2180 Requires a tty driver that supports console polling,
2181 or a supported polling keyboard driver (non-usb).
2182 Serial only format: <serial_device>[,baud]
2183 keyboard only format: kbd
2184 keyboard and serial format: kbd,<serial_device>[,baud]
2185 Optional Kernel mode setting:
2186 kms, kbd format: kms,kbd
2187 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2189 kgdboc_earlycon= [KGDB,HW]
2190 If the boot console provides the ability to read
2191 characters and can work in polling mode, you can use
2192 this parameter to tell kgdb to use it as a backend
2193 until the normal console is registered. Intended to
2194 be used together with the kgdboc parameter which
2195 specifies the normal console to transition to.
2197 The name of the early console should be specified
2198 as the value of this parameter. Note that the name of
2199 the early console might be different than the tty
2200 name passed to kgdboc. It's OK to leave the value
2201 blank and the first boot console that implements
2202 read() will be picked.
2204 kgdbwait [KGDB] Stop kernel execution and enter the
2205 kernel debugger at the earliest opportunity.
2207 kmac= [MIPS] Korina ethernet MAC address.
2208 Configure the RouterBoard 532 series on-chip
2209 Ethernet adapter MAC address.
2211 kmemleak= [KNL] Boot-time kmemleak enable/disable
2212 Valid arguments: on, off
2214 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2217 kprobe_event=[probe-list]
2218 [FTRACE] Add kprobe events and enable at boot time.
2219 The probe-list is a semicolon delimited list of probe
2220 definitions. Each definition is same as kprobe_events
2221 interface, but the parameters are comma delimited.
2222 For example, to add a kprobe event on vfs_read with
2223 arg1 and arg2, add to the command line;
2225 kprobe_event=p,vfs_read,$arg1,$arg2
2227 See also Documentation/trace/kprobetrace.rst "Kernel
2228 Boot Parameter" section.
2230 kpti= [ARM64] Control page table isolation of user
2231 and kernel address spaces.
2232 Default: enabled on cores which need mitigation.
2236 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2237 Default is 0 (don't ignore, but inject #GP)
2239 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2240 Default is false (don't support).
2242 kvm.mmu_audit= [KVM] This is a R/W parameter which allows audit
2247 [KVM] Controls the software workaround for the
2248 X86_BUG_ITLB_MULTIHIT bug.
2249 force : Always deploy workaround.
2250 off : Never deploy workaround.
2251 auto : Deploy workaround based on the presence of
2252 X86_BUG_ITLB_MULTIHIT.
2256 If the software workaround is enabled for the host,
2257 guests do need not to enable it for nested guests.
2259 kvm.nx_huge_pages_recovery_ratio=
2260 [KVM] Controls how many 4KiB pages are periodically zapped
2261 back to huge pages. 0 disables the recovery, otherwise if
2262 the value is N KVM will zap 1/Nth of the 4KiB pages every
2263 minute. The default is 60.
2265 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM.
2266 Default is 1 (enabled)
2268 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU)
2270 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode.
2273 [KVM,ARM] Select one of KVM/arm64's modes of operation.
2275 nvhe: Standard nVHE-based mode, without support for
2278 protected: nVHE-based mode with support for guests whose
2279 state is kept private from the host.
2280 Not valid if the kernel is running in EL2.
2282 Defaults to VHE/nVHE based on hardware support and
2283 the value of CONFIG_ARM64_VHE.
2285 kvm-arm.vgic_v3_group0_trap=
2286 [KVM,ARM] Trap guest accesses to GICv3 group-0
2289 kvm-arm.vgic_v3_group1_trap=
2290 [KVM,ARM] Trap guest accesses to GICv3 group-1
2293 kvm-arm.vgic_v3_common_trap=
2294 [KVM,ARM] Trap guest accesses to GICv3 common
2297 kvm-arm.vgic_v4_enable=
2298 [KVM,ARM] Allow use of GICv4 for direct injection of
2301 kvm_cma_resv_ratio=n [PPC]
2302 Reserves given percentage from system memory area for
2303 contiguous memory allocation for KVM hash pagetable
2305 By default it reserves 5% of total system memory.
2309 kvm-intel.ept= [KVM,Intel] Disable extended page tables
2310 (virtualized MMU) support on capable Intel chips.
2311 Default is 1 (enabled)
2313 kvm-intel.emulate_invalid_guest_state=
2314 [KVM,Intel] Enable emulation of invalid guest states
2315 Default is 0 (disabled)
2317 kvm-intel.flexpriority=
2318 [KVM,Intel] Disable FlexPriority feature (TPR shadow).
2319 Default is 1 (enabled)
2322 [KVM,Intel] Enable VMX nesting (nVMX).
2323 Default is 0 (disabled)
2325 kvm-intel.unrestricted_guest=
2326 [KVM,Intel] Disable unrestricted guest feature
2327 (virtualized real and unpaged mode) on capable
2328 Intel chips. Default is 1 (enabled)
2330 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2333 Valid arguments: never, cond, always
2335 always: L1D cache flush on every VMENTER.
2336 cond: Flush L1D on VMENTER only when the code between
2337 VMEXIT and VMENTER can leak host memory.
2338 never: Disables the mitigation
2340 Default is cond (do L1 cache flush in specific instances)
2342 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification
2343 feature (tagged TLBs) on capable Intel chips.
2344 Default is 1 (enabled)
2346 l1tf= [X86] Control mitigation of the L1TF vulnerability on
2349 The kernel PTE inversion protection is unconditionally
2350 enabled and cannot be disabled.
2353 Provides all available mitigations for the
2354 L1TF vulnerability. Disables SMT and
2355 enables all mitigations in the
2356 hypervisors, i.e. unconditional L1D flush.
2358 SMT control and L1D flush control via the
2359 sysfs interface is still possible after
2360 boot. Hypervisors will issue a warning
2361 when the first VM is started in a
2362 potentially insecure configuration,
2363 i.e. SMT enabled or L1D flush disabled.
2366 Same as 'full', but disables SMT and L1D
2367 flush runtime control. Implies the
2368 'nosmt=force' command line option.
2369 (i.e. sysfs control of SMT is disabled.)
2372 Leaves SMT enabled and enables the default
2373 hypervisor mitigation, i.e. conditional
2376 SMT control and L1D flush control via the
2377 sysfs interface is still possible after
2378 boot. Hypervisors will issue a warning
2379 when the first VM is started in a
2380 potentially insecure configuration,
2381 i.e. SMT enabled or L1D flush disabled.
2385 Disables SMT and enables the default
2386 hypervisor mitigation.
2388 SMT control and L1D flush control via the
2389 sysfs interface is still possible after
2390 boot. Hypervisors will issue a warning
2391 when the first VM is started in a
2392 potentially insecure configuration,
2393 i.e. SMT enabled or L1D flush disabled.
2396 Same as 'flush', but hypervisors will not
2397 warn when a VM is started in a potentially
2398 insecure configuration.
2401 Disables hypervisor mitigations and doesn't
2403 It also drops the swap size and available
2404 RAM limit restriction on both hypervisor and
2409 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2415 lapic [X86-32,APIC] Enable the local APIC even if BIOS
2418 lapic= [X86,APIC] Do not use TSC deadline
2419 value for LAPIC timer one-shot implementation. Default
2420 back to the programmable timer unit in the LAPIC.
2421 Format: notscdeadline
2423 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
2426 libata.dma= [LIBATA] DMA control
2427 libata.dma=0 Disable all PATA and SATA DMA
2428 libata.dma=1 PATA and SATA Disk DMA only
2429 libata.dma=2 ATAPI (CDROM) DMA only
2430 libata.dma=4 Compact Flash DMA only
2431 Combinations also work, so libata.dma=3 enables DMA
2432 for disks and CDROMs, but not CFs.
2434 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2435 libata.ignore_hpa=0 keep BIOS limits (default)
2436 libata.ignore_hpa=1 ignore limits, using full disk
2438 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2442 libata.force= [LIBATA] Force configurations. The format is comma-
2443 separated list of "[ID:]VAL" where ID is
2444 PORT[.DEVICE]. PORT and DEVICE are decimal numbers
2445 matching port, link or device. Basically, it matches
2446 the ATA ID string printed on console by libata. If
2447 the whole ID part is omitted, the last PORT and DEVICE
2448 values are used. If ID hasn't been specified yet, the
2449 configuration applies to all ports, links and devices.
2451 If only DEVICE is omitted, the parameter applies to
2452 the port and all links and devices behind it. DEVICE
2453 number of 0 either selects the first device or the
2454 first fan-out link behind PMP device. It does not
2455 select the host link. DEVICE number of 15 selects the
2456 host link and device attached to it.
2458 The VAL specifies the configuration to force. As long
2459 as there's no ambiguity shortcut notation is allowed.
2460 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2461 The following configurations can be forced.
2463 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2464 Any ID with matching PORT is used.
2466 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2468 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2469 udma[/][16,25,33,44,66,100,133] notation is also
2472 * [no]ncq: Turn on or off NCQ.
2474 * [no]ncqtrim: Turn off queued DSM TRIM.
2476 * nohrst, nosrst, norst: suppress hard, soft
2479 * rstonce: only attempt one reset during
2480 hot-unplug link recovery
2482 * dump_id: dump IDENTIFY data.
2484 * atapi_dmadir: Enable ATAPI DMADIR bridge support
2486 * disable: Disable this device.
2488 If there are multiple matching configurations changing
2489 the same attribute, the last one is used.
2491 memblock=debug [KNL] Enable memblock debug messages.
2493 load_ramdisk= [RAM] [Deprecated]
2495 lockd.nlm_grace_period=P [NFS] Assign grace period.
2498 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2501 lockd.nlm_timeout=T [NFS] Assign timeout value.
2504 lockd.nlm_udpport=M [NFS] Assign UDP port.
2507 lockdown= [SECURITY]
2508 { integrity | confidentiality }
2509 Enable the kernel lockdown feature. If set to
2510 integrity, kernel features that allow userland to
2511 modify the running kernel are disabled. If set to
2512 confidentiality, kernel features that allow userland
2513 to extract confidential information from the kernel
2516 locktorture.nreaders_stress= [KNL]
2517 Set the number of locking read-acquisition kthreads.
2518 Defaults to being automatically set based on the
2519 number of online CPUs.
2521 locktorture.nwriters_stress= [KNL]
2522 Set the number of locking write-acquisition kthreads.
2524 locktorture.onoff_holdoff= [KNL]
2525 Set time (s) after boot for CPU-hotplug testing.
2527 locktorture.onoff_interval= [KNL]
2528 Set time (s) between CPU-hotplug operations, or
2529 zero to disable CPU-hotplug testing.
2531 locktorture.shuffle_interval= [KNL]
2532 Set task-shuffle interval (jiffies). Shuffling
2533 tasks allows some CPUs to go into dyntick-idle
2534 mode during the locktorture test.
2536 locktorture.shutdown_secs= [KNL]
2537 Set time (s) after boot system shutdown. This
2538 is useful for hands-off automated testing.
2540 locktorture.stat_interval= [KNL]
2541 Time (s) between statistics printk()s.
2543 locktorture.stutter= [KNL]
2544 Time (s) to stutter testing, for example,
2545 specifying five seconds causes the test to run for
2546 five seconds, wait for five seconds, and so on.
2547 This tests the locking primitive's ability to
2548 transition abruptly to and from idle.
2550 locktorture.torture_type= [KNL]
2551 Specify the locking implementation to test.
2553 locktorture.verbose= [KNL]
2554 Enable additional printk() statements.
2556 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
2559 loglevel= All Kernel Messages with a loglevel smaller than the
2560 console loglevel will be printed to the console. It can
2561 also be changed with klogd or other programs. The
2562 loglevels are defined as follows:
2564 0 (KERN_EMERG) system is unusable
2565 1 (KERN_ALERT) action must be taken immediately
2566 2 (KERN_CRIT) critical conditions
2567 3 (KERN_ERR) error conditions
2568 4 (KERN_WARNING) warning conditions
2569 5 (KERN_NOTICE) normal but significant condition
2570 6 (KERN_INFO) informational
2571 7 (KERN_DEBUG) debug-level messages
2573 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
2574 in bytes. n must be a power of two and greater
2575 than the minimal size. The minimal size is defined
2576 by LOG_BUF_SHIFT kernel config parameter. There is
2577 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2578 that allows to increase the default size depending on
2579 the number of CPUs. See init/Kconfig for more details.
2581 logo.nologo [FB] Disables display of the built-in Linux logo.
2582 This may be used to provide more screen space for
2583 kernel log messages and is useful when debugging
2584 kernel boot problems.
2586 lp=0 [LP] Specify parallel ports to use, e.g,
2587 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2588 lp=reset first parallel port). 'lp=0' disables the
2589 lp=auto printer driver. 'lp=reset' (which can be
2590 specified in addition to the ports) causes
2591 attached printers to be reset. Using
2592 lp=port1,port2,... specifies the parallel ports
2593 to associate lp devices with, starting with
2594 lp0. A port specification may be 'none' to skip
2595 that lp device, or a parport name such as
2596 'parport0'. Specifying 'lp=auto' instead of a
2597 port specification list means that device IDs
2598 from each port should be examined, to see if
2599 an IEEE 1284-compliant printer is attached; if
2600 so, the driver will manage that printer.
2601 See also header of drivers/char/lp.c.
2604 Sets loops_per_jiffy to given constant, thus avoiding
2605 time-consuming boot-time autodetection (up to 250 ms per
2606 CPU). 0 enables autodetection (default). To determine
2607 the correct value for your kernel, boot with normal
2608 autodetection and see what value is printed. Note that
2609 on SMP systems the preset will be applied to all CPUs,
2610 which is likely to cause problems if your CPUs need
2611 significantly divergent settings. An incorrect value
2612 will cause delays in the kernel to be wrong, leading to
2613 unpredictable I/O errors and other breakage. Although
2614 unlikely, in the extreme case this might damage your
2618 Format: <io>,<irq>,<dma>
2620 lsm.debug [SECURITY] Enable LSM initialization debugging output.
2623 [SECURITY] Choose order of LSM initialization. This
2624 overrides CONFIG_LSM, and the "security=" parameter.
2626 machvec= [IA-64] Force the use of a particular machine-vector
2627 (machvec) in a generic kernel.
2628 Example: machvec=hpzx1
2630 machtype= [Loongson] Share the same kernel image file between
2631 different yeeloong laptops.
2632 Example: machtype=lemote-yeeloong-2f-7inch
2634 max_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory greater
2635 than or equal to this physical address is ignored.
2637 maxcpus= [SMP] Maximum number of processors that an SMP kernel
2638 will bring up during bootup. maxcpus=n : n >= 0 limits
2639 the kernel to bring up 'n' processors. Surely after
2640 bootup you can bring up the other plugged cpu by executing
2641 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
2642 only takes effect during system bootup.
2643 While n=0 is a special case, it is equivalent to "nosmp",
2644 which also disables the IO APIC.
2646 max_loop= [LOOP] The number of loop block devices that get
2647 (loop.max_loop) unconditionally pre-created at init time. The default
2648 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
2649 of statically allocating a predefined number, loop
2650 devices can be requested on-demand with the
2651 /dev/loop-control interface.
2653 mce [X86-32] Machine Check Exception
2655 mce=option [X86-64] See Documentation/x86/x86_64/boot-options.rst
2657 md= [HW] RAID subsystems devices and level
2658 See Documentation/admin-guide/md.rst.
2661 Format: <first>,<last>
2662 Specifies range of consoles to be captured by the MDA.
2665 Control mitigation for the Micro-architectural Data
2666 Sampling (MDS) vulnerability.
2668 Certain CPUs are vulnerable to an exploit against CPU
2669 internal buffers which can forward information to a
2670 disclosure gadget under certain conditions.
2672 In vulnerable processors, the speculatively
2673 forwarded data can be used in a cache side channel
2674 attack, to access data to which the attacker does
2675 not have direct access.
2677 This parameter controls the MDS mitigation. The
2680 full - Enable MDS mitigation on vulnerable CPUs
2681 full,nosmt - Enable MDS mitigation and disable
2682 SMT on vulnerable CPUs
2683 off - Unconditionally disable MDS mitigation
2685 On TAA-affected machines, mds=off can be prevented by
2686 an active TAA mitigation as both vulnerabilities are
2687 mitigated with the same mechanism so in order to disable
2688 this mitigation, you need to specify tsx_async_abort=off
2691 Not specifying this option is equivalent to
2694 For details see: Documentation/admin-guide/hw-vuln/mds.rst
2696 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
2697 Amount of memory to be used in cases as follows:
2700 2 when the kernel is not able to see the whole system memory;
2701 3 memory that lies after 'mem=' boundary is excluded from
2702 the hypervisor, then assigned to KVM guests.
2704 [X86] Work as limiting max address. Use together
2705 with memmap= to avoid physical address space collisions.
2706 Without memmap= PCI devices could be placed at addresses
2707 belonging to unused RAM.
2709 Note that this only takes effects during boot time since
2710 in above case 3, memory may need be hot added after boot
2711 if system memory of hypervisor is not sufficient.
2713 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
2717 [KNL,SH] Allow user to override the default size for
2718 per-device physically contiguous DMA buffers.
2720 memhp_default_state=online/offline
2721 [KNL] Set the initial state for the memory hotplug
2722 onlining policy. If not specified, the default value is
2723 set according to the
2724 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
2726 See Documentation/admin-guide/mm/memory-hotplug.rst.
2728 memmap=exactmap [KNL,X86] Enable setting of an exact
2729 E820 memory map, as specified by the user.
2730 Such memmap=exactmap lines can be constructed based on
2731 BIOS output or other requirements. See the memmap=nn@ss
2734 memmap=nn[KMG]@ss[KMG]
2735 [KNL, X86, MIPS, XTENSA] Force usage of a specific region of memory.
2736 Region of memory to be used is from ss to ss+nn.
2737 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
2738 which limits max address to nn[KMG].
2739 Multiple different regions can be specified,
2742 memmap=100M@2G,100M#3G,1G!1024G
2744 memmap=nn[KMG]#ss[KMG]
2745 [KNL,ACPI] Mark specific memory as ACPI data.
2746 Region of memory to be marked is from ss to ss+nn.
2748 memmap=nn[KMG]$ss[KMG]
2749 [KNL,ACPI] Mark specific memory as reserved.
2750 Region of memory to be reserved is from ss to ss+nn.
2751 Example: Exclude memory from 0x18690000-0x1869ffff
2752 memmap=64K$0x18690000
2754 memmap=0x10000$0x18690000
2755 Some bootloaders may need an escape character before '$',
2756 like Grub2, otherwise '$' and the following number
2759 memmap=nn[KMG]!ss[KMG]
2760 [KNL,X86] Mark specific memory as protected.
2761 Region of memory to be used, from ss to ss+nn.
2762 The memory region may be marked as e820 type 12 (0xc)
2763 and is NVDIMM or ADR memory.
2765 memmap=<size>%<offset>-<oldtype>+<newtype>
2766 [KNL,ACPI] Convert memory within the specified region
2767 from <oldtype> to <newtype>. If "-<oldtype>" is left
2768 out, the whole region will be marked as <newtype>,
2769 even if previously unavailable. If "+<newtype>" is left
2770 out, matching memory will be removed. Types are
2771 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
2772 3 = ACPI, 12 = PRAM.
2774 memory_corruption_check=0/1 [X86]
2775 Some BIOSes seem to corrupt the first 64k of
2776 memory when doing things like suspend/resume.
2777 Setting this option will scan the memory
2778 looking for corruption. Enabling this will
2779 both detect corruption and prevent the kernel
2780 from using the memory being corrupted.
2781 However, its intended as a diagnostic tool; if
2782 repeatable BIOS-originated corruption always
2783 affects the same memory, you can use memmap=
2784 to prevent the kernel from using that memory.
2786 memory_corruption_check_size=size [X86]
2787 By default it checks for corruption in the low
2788 64k, making this memory unavailable for normal
2789 use. Use this parameter to scan for
2790 corruption in more or less memory.
2792 memory_corruption_check_period=seconds [X86]
2793 By default it checks for corruption every 60
2794 seconds. Use this parameter to check at some
2795 other rate. 0 disables periodic checking.
2797 memtest= [KNL,X86,ARM,PPC] Enable memtest
2799 default : 0 <disable>
2800 Specifies the number of memtest passes to be
2801 performed. Each pass selects another test
2802 pattern from a given set of patterns. Memtest
2803 fills the memory with this pattern, validates
2804 memory contents and reserves bad memory
2805 regions that are detected.
2807 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
2808 Valid arguments: on, off
2809 Default (depends on kernel configuration option):
2810 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
2811 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
2812 mem_encrypt=on: Activate SME
2813 mem_encrypt=off: Do not activate SME
2815 Refer to Documentation/virt/kvm/amd-memory-encryption.rst
2816 for details on when memory encryption can be activated.
2818 mem_sleep_default= [SUSPEND] Default system suspend mode:
2819 s2idle - Suspend-To-Idle
2820 shallow - Power-On Suspend or equivalent (if supported)
2821 deep - Suspend-To-RAM or equivalent (if supported)
2822 See Documentation/admin-guide/pm/sleep-states.rst.
2824 meye.*= [HW] Set MotionEye Camera parameters
2825 See Documentation/admin-guide/media/meye.rst.
2827 mfgpt_irq= [IA-32] Specify the IRQ to use for the
2828 Multi-Function General Purpose Timers on AMD Geode
2831 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
2832 the BIOS has incorrectly applied a workaround. TinyBIOS
2833 version 0.98 is known to be affected, 0.99 fixes the
2834 problem by letting the user disable the workaround.
2838 min_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory below this
2839 physical address is ignored.
2841 mini2440= [ARM,HW,KNL]
2842 Format:[0..2][b][c][t]
2844 MINI2440 configuration specification:
2845 0 - The attached screen is the 3.5" TFT
2846 1 - The attached screen is the 7" TFT
2847 2 - The VGA Shield is attached (1024x768)
2848 Leaving out the screen size parameter will not load
2849 the TFT driver, and the framebuffer will be left
2851 b - Enable backlight. The TFT backlight pin will be
2852 linked to the kernel VESA blanking code and a GPIO
2853 LED. This parameter is not necessary when using the
2855 c - Enable the s3c camera interface.
2856 t - Reserved for enabling touchscreen support. The
2857 touchscreen support is not enabled in the mainstream
2858 kernel as of 2.6.30, a preliminary port can be found
2859 in the "bleeding edge" mini2440 support kernel at
2860 https://repo.or.cz/w/linux-2.6/mini2440.git
2863 [X86,PPC,S390,ARM64] Control optional mitigations for
2864 CPU vulnerabilities. This is a set of curated,
2865 arch-independent options, each of which is an
2866 aggregation of existing arch-specific options.
2869 Disable all optional CPU mitigations. This
2870 improves system performance, but it may also
2871 expose users to several CPU vulnerabilities.
2872 Equivalent to: nopti [X86,PPC]
2874 nospectre_v1 [X86,PPC]
2876 nospectre_v2 [X86,PPC,S390,ARM64]
2877 spectre_v2_user=off [X86]
2878 spec_store_bypass_disable=off [X86,PPC]
2879 ssbd=force-off [ARM64]
2882 tsx_async_abort=off [X86]
2883 kvm.nx_huge_pages=off [X86]
2884 no_entry_flush [PPC]
2885 no_uaccess_flush [PPC]
2888 This does not have any effect on
2889 kvm.nx_huge_pages when
2890 kvm.nx_huge_pages=force.
2893 Mitigate all CPU vulnerabilities, but leave SMT
2894 enabled, even if it's vulnerable. This is for
2895 users who don't want to be surprised by SMT
2896 getting disabled across kernel upgrades, or who
2897 have other ways of avoiding SMT-based attacks.
2898 Equivalent to: (default behavior)
2901 Mitigate all CPU vulnerabilities, disabling SMT
2902 if needed. This is for users who always want to
2903 be fully mitigated, even if it means losing SMT.
2904 Equivalent to: l1tf=flush,nosmt [X86]
2905 mds=full,nosmt [X86]
2906 tsx_async_abort=full,nosmt [X86]
2909 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
2910 parameter allows control of the logging verbosity for
2911 the additional memory initialisation checks. A value
2912 of 0 disables mminit logging and a level of 4 will
2913 log everything. Information is printed at KERN_DEBUG
2914 so loglevel=8 may also need to be specified.
2917 [KNL] When CONFIG_MODULE_SIG is set, this means that
2918 modules without (valid) signatures will fail to load.
2919 Note that if CONFIG_MODULE_SIG_FORCE is set, that
2920 is always true, so this option does nothing.
2922 module_blacklist= [KNL] Do not load a comma-separated list of
2923 modules. Useful for debugging problem modules.
2926 [MOUSE] Maximum time between finger touching and
2927 leaving touchpad surface for touch to be considered
2928 a tap and be reported as a left button click (for
2929 touchpads working in absolute mode only).
2931 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
2932 reporting absolute coordinates, such as tablets
2933 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
2934 reporting absolute coordinates, such as tablets
2936 movablecore= [KNL,X86,IA-64,PPC]
2937 Format: nn[KMGTPE] | nn%
2938 This parameter is the complement to kernelcore=, it
2939 specifies the amount of memory used for migratable
2940 allocations. If both kernelcore and movablecore is
2941 specified, then kernelcore will be at *least* the
2942 specified value but may be more. If movablecore on its
2943 own is specified, the administrator must be careful
2944 that the amount of memory usable for all allocations
2947 movable_node [KNL] Boot-time switch to make hotplugable memory
2948 NUMA nodes to be movable. This means that the memory
2949 of such nodes will be usable only for movable
2950 allocations which rules out almost all kernel
2951 allocations. Use with caution!
2953 MTD_Partition= [MTD]
2954 Format: <name>,<region-number>,<size>,<offset>
2956 MTD_Region= [MTD] Format:
2957 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
2960 See drivers/mtd/parsers/cmdlinepart.c
2962 multitce=off [PPC] This parameter disables the use of the pSeries
2963 firmware feature for updating multiple TCE entries
2966 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
2968 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
2970 boundary - index of last SLC block on Flex-OneNAND.
2971 The remaining blocks are configured as MLC blocks.
2972 lock - Configure if Flex-OneNAND boundary should be locked.
2973 Once locked, the boundary cannot be changed.
2974 1 indicates lock status, 0 indicates unlock status.
2977 ARM/S3C2412 JIVE boot control
2979 See arch/arm/mach-s3c/mach-jive.c
2981 mtouchusb.raw_coordinates=
2982 [HW] Make the MicroTouch USB driver use raw coordinates
2983 ('y', default) or cooked coordinates ('n')
2985 mtrr_chunk_size=nn[KMG] [X86]
2986 used for mtrr cleanup. It is largest continuous chunk
2987 that could hold holes aka. UC entries.
2989 mtrr_gran_size=nn[KMG] [X86]
2990 Used for mtrr cleanup. It is granularity of mtrr block.
2992 Large value could prevent small alignment from
2995 mtrr_spare_reg_nr=n [X86]
2997 Range: 0,7 : spare reg number
2999 Used for mtrr cleanup. It is spare mtrr entries number.
3000 Set to 2 or more if your graphical card needs more.
3002 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
3004 netdev= [NET] Network devices parameters
3005 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
3006 Note that mem_start is often overloaded to mean
3007 something different and driver-specific.
3008 This usage is only documented in each driver source
3012 [NETFILTER] Enable connection tracking flow accounting
3013 0 to disable accounting
3014 1 to enable accounting
3017 nfsaddrs= [NFS] Deprecated. Use ip= instead.
3018 See Documentation/admin-guide/nfs/nfsroot.rst.
3020 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
3021 See Documentation/admin-guide/nfs/nfsroot.rst.
3023 nfsrootdebug [NFS] enable nfsroot debugging messages.
3024 See Documentation/admin-guide/nfs/nfsroot.rst.
3026 nfs.callback_nr_threads=
3027 [NFSv4] set the total number of threads that the
3028 NFS client will assign to service NFSv4 callback
3031 nfs.callback_tcpport=
3032 [NFS] set the TCP port on which the NFSv4 callback
3033 channel should listen.
3036 [NFS] sets the pathname to the program which is used
3037 to update the NFS client cache entries.
3039 nfs.cache_getent_timeout=
3040 [NFS] sets the timeout after which an attempt to
3041 update a cache entry is deemed to have failed.
3043 nfs.idmap_cache_timeout=
3044 [NFS] set the maximum lifetime for idmapper cache
3048 [NFS] enable 64-bit inode numbers.
3049 If zero, the NFS client will fake up a 32-bit inode
3050 number for the readdir() and stat() syscalls instead
3051 of returning the full 64-bit number.
3052 The default is to return 64-bit inode numbers.
3054 nfs.max_session_cb_slots=
3055 [NFSv4.1] Sets the maximum number of session
3056 slots the client will assign to the callback
3057 channel. This determines the maximum number of
3058 callbacks the client will process in parallel for
3059 a particular server.
3061 nfs.max_session_slots=
3062 [NFSv4.1] Sets the maximum number of session slots
3063 the client will attempt to negotiate with the server.
3064 This limits the number of simultaneous RPC requests
3065 that the client can send to the NFSv4.1 server.
3066 Note that there is little point in setting this
3067 value higher than the max_tcp_slot_table_limit.
3069 nfs.nfs4_disable_idmapping=
3070 [NFSv4] When set to the default of '1', this option
3071 ensures that both the RPC level authentication
3072 scheme and the NFS level operations agree to use
3073 numeric uids/gids if the mount is using the
3074 'sec=sys' security flavour. In effect it is
3075 disabling idmapping, which can make migration from
3076 legacy NFSv2/v3 systems to NFSv4 easier.
3077 Servers that do not support this mode of operation
3078 will be autodetected by the client, and it will fall
3079 back to using the idmapper.
3080 To turn off this behaviour, set the value to '0'.
3082 [NFS4] Specify an additional fixed unique ident-
3083 ification string that NFSv4 clients can insert into
3084 their nfs_client_id4 string. This is typically a
3085 UUID that is generated at system install time.
3087 nfs.send_implementation_id =
3088 [NFSv4.1] Send client implementation identification
3089 information in exchange_id requests.
3090 If zero, no implementation identification information
3092 The default is to send the implementation identification
3095 nfs.recover_lost_locks =
3096 [NFSv4] Attempt to recover locks that were lost due
3097 to a lease timeout on the server. Please note that
3098 doing this risks data corruption, since there are
3099 no guarantees that the file will remain unchanged
3100 after the locks are lost.
3101 If you want to enable the kernel legacy behaviour of
3102 attempting to recover these locks, then set this
3104 The default parameter value of '0' causes the kernel
3105 not to attempt recovery of lost locks.
3107 nfs4.layoutstats_timer =
3108 [NFSv4.2] Change the rate at which the kernel sends
3109 layoutstats to the pNFS metadata server.
3111 Setting this to value to 0 causes the kernel to use
3112 whatever value is the default set by the layout
3113 driver. A non-zero value sets the minimum interval
3114 in seconds between layoutstats transmissions.
3116 nfsd.nfs4_disable_idmapping=
3117 [NFSv4] When set to the default of '1', the NFSv4
3118 server will return only numeric uids and gids to
3119 clients using auth_sys, and will accept numeric uids
3120 and gids from such clients. This is intended to ease
3121 migration from NFSv2/v3.
3123 nmi_backtrace.backtrace_idle [KNL]
3124 Dump stacks even of idle CPUs in response to an
3125 NMI stack-backtrace request.
3127 nmi_debug= [KNL,SH] Specify one or more actions to take
3128 when a NMI is triggered.
3129 Format: [state][,regs][,debounce][,die]
3131 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
3132 Format: [panic,][nopanic,][num]
3134 0 - turn hardlockup detector in nmi_watchdog off
3135 1 - turn hardlockup detector in nmi_watchdog on
3136 When panic is specified, panic when an NMI watchdog
3137 timeout occurs (or 'nopanic' to not panic on an NMI
3138 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
3139 To disable both hard and soft lockup detectors,
3140 please see 'nowatchdog'.
3141 This is useful when you use a panic=... timeout and
3142 need the box quickly up again.
3144 These settings can be accessed at runtime via
3145 the nmi_watchdog and hardlockup_panic sysctls.
3147 netpoll.carrier_timeout=
3148 [NET] Specifies amount of time (in seconds) that
3149 netpoll should wait for a carrier. By default netpoll
3152 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3153 emulation library even if a 387 maths coprocessor
3156 no5lvl [X86-64] Disable 5-level paging mode. Forces
3157 kernel to use 4-level paging instead.
3159 nofsgsbase [X86] Disables FSGSBASE instructions.
3162 [HW] Never suspend the console
3163 Disable suspending of consoles during suspend and
3164 hibernate operations. Once disabled, debugging
3165 messages can reach various consoles while the rest
3166 of the system is being put to sleep (ie, while
3167 debugging driver suspend/resume hooks). This may
3168 not work reliably with all consoles, but is known
3169 to work with serial and VGA consoles.
3170 To facilitate more flexible debugging, we also add
3171 console_suspend, a printk module parameter to control
3172 it. Users could use console_suspend (usually
3173 /sys/module/printk/parameters/console_suspend) to
3174 turn on/off it dynamically.
3176 novmcoredd [KNL,KDUMP]
3177 Disable device dump. Device dump allows drivers to
3178 append dump data to vmcore so you can collect driver
3179 specified debug info. Drivers can append the data
3180 without any limit and this data is stored in memory,
3181 so this may cause significant memory stress. Disabling
3182 device dump can help save memory but the driver debug
3183 data will be no longer available. This parameter
3184 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
3187 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
3188 caches in the slab allocator. Saves per-node memory,
3189 but will impact performance.
3193 noaltinstr [S390] Disables alternative instructions patching
3194 (CPU alternatives feature).
3196 noapic [SMP,APIC] Tells the kernel to not make use of any
3197 IOAPICs that may be present in the system.
3199 noautogroup Disable scheduler automatic task group creation.
3201 nobats [PPC] Do not use BATs for mapping kernel lowmem
3202 on "Classic" PPC cores.
3206 noclflush [BUGS=X86] Don't use the CLFLUSH instruction
3208 nodelayacct [KNL] Disable per-task delay accounting
3210 nodsp [SH] Disable hardware DSP at boot time.
3212 noefi Disable EFI runtime services support.
3214 no_entry_flush [PPC] Don't flush the L1-D cache when entering the kernel.
3219 On X86-32 available only on PAE configured kernels.
3220 noexec=on: enable non-executable mappings (default)
3221 noexec=off: disable non-executable mappings
3224 Disable SMAP (Supervisor Mode Access Prevention)
3225 even if it is supported by processor.
3228 Disable SMEP (Supervisor Mode Execution Prevention)
3229 even if it is supported by processor.
3232 This affects only 32-bit executables.
3233 noexec32=on: enable non-executable mappings (default)
3234 read doesn't imply executable mappings
3235 noexec32=off: disable non-executable mappings
3236 read implies executable mappings
3238 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3240 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3241 register save and restore. The kernel will only save
3242 legacy floating-point registers on task switch.
3244 nohugeiomap [KNL,X86,PPC,ARM64] Disable kernel huge I/O mappings.
3246 nosmt [KNL,S390] Disable symmetric multithreading (SMT).
3247 Equivalent to smt=1.
3249 [KNL,X86] Disable symmetric multithreading (SMT).
3250 nosmt=force: Force disable SMT, cannot be undone
3251 via the sysfs control file.
3253 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1
3254 (bounds check bypass). With this option data leaks are
3255 possible in the system.
3257 nospectre_v2 [X86,PPC_FSL_BOOK3E,ARM64] Disable all mitigations for
3258 the Spectre variant 2 (indirect branch prediction)
3259 vulnerability. System may allow data leaks with this
3262 nospec_store_bypass_disable
3263 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
3266 [PPC] Don't flush the L1-D cache after accessing user data.
3268 noxsave [BUGS=X86] Disables x86 extended register state save
3269 and restore using xsave. The kernel will fallback to
3270 enabling legacy floating-point and sse state.
3272 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
3273 register states. The kernel will fall back to use
3274 xsave to save the states. By using this parameter,
3275 performance of saving the states is degraded because
3276 xsave doesn't support modified optimization while
3277 xsaveopt supports it on xsaveopt enabled systems.
3279 noxsaves [X86] Disables xsaves and xrstors used in saving and
3280 restoring x86 extended register state in compacted
3281 form of xsave area. The kernel will fall back to use
3282 xsaveopt and xrstor to save and restore the states
3283 in standard form of xsave area. By using this
3284 parameter, xsave area per process might occupy more
3285 memory on xsaves enabled systems.
3287 nohlt [ARM,ARM64,MICROBLAZE,SH] Forces the kernel to busy wait
3288 in do_idle() and not use the arch_cpu_idle()
3289 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
3290 to be effective. This is useful on platforms where the
3291 sleep(SH) or wfi(ARM,ARM64) instructions do not work
3292 correctly or when doing power measurements to evalute
3293 the impact of the sleep instructions. This is also
3294 useful when using JTAG debugger.
3296 no_file_caps Tells the kernel not to honor file capabilities. The
3297 only way then for a file to be executed with privilege
3298 is to be setuid root or executed by root.
3300 nohalt [IA-64] Tells the kernel not to use the power saving
3301 function PAL_HALT_LIGHT when idle. This increases
3302 power-consumption. On the positive side, it reduces
3303 interrupt wake-up latency, which may improve performance
3304 in certain environments such as networked servers or
3308 Force pointers printed to the console or buffers to be
3309 unhashed. By default, when a pointer is printed via %p
3310 format string, that pointer is "hashed", i.e. obscured
3311 by hashing the pointer value. This is a security feature
3312 that hides actual kernel addresses from unprivileged
3313 users, but it also makes debugging the kernel more
3314 difficult since unequal pointers can no longer be
3315 compared. However, if this command-line option is
3316 specified, then all normal pointers will have their true
3317 value printed. Pointers printed via %pK may still be
3318 hashed. This option should only be specified when
3319 debugging the kernel. Please do not use on production
3322 nohibernate [HIBERNATION] Disable hibernation and resume.
3324 nohz= [KNL] Boottime enable/disable dynamic ticks
3325 Valid arguments: on, off
3328 nohz_full= [KNL,BOOT,SMP,ISOL]
3329 The argument is a cpu list, as described above.
3330 In kernels built with CONFIG_NO_HZ_FULL=y, set
3331 the specified list of CPUs whose tick will be stopped
3332 whenever possible. The boot CPU will be forced outside
3333 the range to maintain the timekeeping. Any CPUs
3334 in this list will have their RCU callbacks offloaded,
3335 just as if they had also been called out in the
3336 rcu_nocbs= boot parameter.
3338 noiotrap [SH] Disables trapped I/O port accesses.
3340 noirqdebug [X86-32] Disables the code which attempts to detect and
3341 disable unhandled interrupt sources.
3343 no_timer_check [X86,APIC] Disables the code which tests for
3344 broken timer IRQ sources.
3346 noisapnp [ISAPNP] Disables ISA PnP code.
3348 noinitrd [RAM] Tells the kernel not to load any configured
3351 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
3353 [Deprecated - use intremap=off]
3357 noinvpcid [X86] Disable the INVPCID cpu feature.
3359 nojitter [IA-64] Disables jitter checking for ITC timers.
3361 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
3363 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
3367 [X86,PV_OPS] Disable paravirtualized VMware scheduler
3368 clock and use the default one.
3370 no-steal-acc [X86,PV_OPS,ARM64] Disable paravirtualized steal time
3371 accounting. steal time is computed, but won't
3372 influence scheduler behaviour
3374 nolapic [X86-32,APIC] Do not enable or use the local APIC.
3376 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
3378 noltlbs [PPC] Do not use large page/tlb entries for kernel
3379 lowmem mapping on PPC40x and PPC8xx
3381 nomca [IA-64] Disable machine check abort handling
3383 nomce [X86-32] Disable Machine Check Exception
3385 nomfgpt [X86-32] Disable Multi-Function General Purpose
3386 Timer usage (for AMD Geode machines).
3388 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3389 shutdown the other cpus. Instead use the REBOOT_VECTOR
3392 nomodule Disable module load
3394 nopat [X86] Disable PAT (page attribute table extension of
3395 pagetables) support.
3397 nopcid [X86-64] Disable the PCID cpu feature.
3399 norandmaps Don't use address space randomization. Equivalent to
3400 echo 0 > /proc/sys/kernel/randomize_va_space
3402 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
3403 with UP alternatives
3405 nordrand [X86] Disable kernel use of the RDRAND and
3406 RDSEED instructions even if they are supported
3407 by the processor. RDRAND and RDSEED are still
3408 available to user space applications.
3410 noresume [SWSUSP] Disables resume and restores original swap
3413 no-scroll [VGA] Disables scrollback.
3414 This is required for the Braillex ib80-piezo Braille
3415 reader made by F.H. Papenmeier (Germany).
3419 nosep [BUGS=X86-32] Disables x86 SYSENTER/SYSEXIT support.
3421 nosgx [X86-64,SGX] Disables Intel SGX kernel support.
3423 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
3424 and disable the IO APIC. legacy for "maxcpus=0".
3426 nosoftlockup [KNL] Disable the soft-lockup detector.
3428 nosync [HW,M68K] Disables sync negotiation for all devices.
3430 nowatchdog [KNL] Disable both lockup detectors, i.e.
3431 soft-lockup and NMI watchdog (hard-lockup).
3435 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
3437 cpu0_hotplug [X86] Turn on CPU0 hotplug feature when
3438 CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off.
3439 Some features depend on CPU0. Known dependencies are:
3440 1. Resume from suspend/hibernate depends on CPU0.
3441 Suspend/hibernate will fail if CPU0 is offline and you
3442 need to online CPU0 before suspend/hibernate.
3443 2. PIC interrupts also depend on CPU0. CPU0 can't be
3444 removed if a PIC interrupt is detected.
3445 It's said poweroff/reboot may depend on CPU0 on some
3446 machines although I haven't seen such issues so far
3447 after CPU0 is offline on a few tested machines.
3448 If the dependencies are under your control, you can
3449 turn on cpu0_hotplug.
3451 nps_mtm_hs_ctr= [KNL,ARC]
3452 This parameter sets the maximum duration, in
3453 cycles, each HW thread of the CTOP can run
3454 without interruptions, before HW switches it.
3455 The actual maximum duration is 16 times this
3457 Format: integer between 1 and 255
3460 nptcg= [IA-64] Override max number of concurrent global TLB
3461 purges which is reported from either PAL_VM_SUMMARY or
3464 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
3465 could support. nr_cpus=n : n >= 1 limits the kernel to
3466 support 'n' processors. It could be larger than the
3467 number of already plugged CPU during bootup, later in
3468 runtime you can physically add extra cpu until it reaches
3469 n. So during boot up some boot time memory for per-cpu
3470 variables need be pre-allocated for later physical cpu
3473 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
3475 numa_balancing= [KNL,X86] Enable or disable automatic NUMA balancing.
3476 Allowed values are enable and disable
3478 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
3479 'node', 'default' can be specified
3480 This can be set from sysctl after boot.
3481 See Documentation/admin-guide/sysctl/vm.rst for details.
3483 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
3484 See Documentation/core-api/debugging-via-ohci1394.rst for more
3487 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
3488 Rather than timing out after 20 ms if an EC
3489 command is not properly ACKed, override the length
3490 of the timeout. We have interrupts disabled while
3491 waiting for the ACK, so if this is set too high
3492 interrupts *may* be lost!
3494 omap_mux= [OMAP] Override bootloader pin multiplexing.
3495 Format: <mux_mode0.mode_name=value>...
3496 For example, to override I2C bus2:
3497 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
3499 oops=panic Always panic on oopses. Default is to just kill the
3500 process, but there is a small probability of
3501 deadlocking the machine.
3502 This will also cause panics on machine check exceptions.
3503 Useful together with panic=30 to trigger a reboot.
3506 [KNL] Boolean flag to control whether the page allocator
3507 should randomize its free lists. The randomization may
3508 be automatically enabled if the kernel detects it is
3509 running on a platform with a direct-mapped memory-side
3510 cache, and this parameter can be used to
3511 override/disable that behavior. The state of the flag
3512 can be read from sysfs at:
3513 /sys/module/page_alloc/parameters/shuffle.
3515 page_owner= [KNL] Boot-time page_owner enabling option.
3516 Storage of the information about who allocated
3517 each page is disabled in default. With this switch,
3519 on: enable the feature
3521 page_poison= [KNL] Boot-time parameter changing the state of
3522 poisoning on the buddy allocator, available with
3523 CONFIG_PAGE_POISONING=y.
3524 off: turn off poisoning (default)
3525 on: turn on poisoning
3527 panic= [KNL] Kernel behaviour on panic: delay <timeout>
3528 timeout > 0: seconds before rebooting
3529 timeout = 0: wait forever
3530 timeout < 0: reboot immediately
3533 panic_print= Bitmask for printing system info when panic happens.
3534 User can chose combination of the following bits:
3535 bit 0: print all tasks info
3536 bit 1: print system memory info
3537 bit 2: print timer info
3538 bit 3: print locks info if CONFIG_LOCKDEP is on
3539 bit 4: print ftrace buffer
3540 bit 5: print all printk messages in buffer
3542 panic_on_taint= Bitmask for conditionally calling panic() in add_taint()
3543 Format: <hex>[,nousertaint]
3544 Hexadecimal bitmask representing the set of TAINT flags
3545 that will cause the kernel to panic when add_taint() is
3546 called with any of the flags in this set.
3547 The optional switch "nousertaint" can be utilized to
3548 prevent userspace forced crashes by writing to sysctl
3549 /proc/sys/kernel/tainted any flagset matching with the
3550 bitmask set on panic_on_taint.
3551 See Documentation/admin-guide/tainted-kernels.rst for
3552 extra details on the taint flags that users can pick
3553 to compose the bitmask to assign to panic_on_taint.
3555 panic_on_warn panic() instead of WARN(). Useful to cause kdump
3558 crash_kexec_post_notifiers
3559 Run kdump after running panic-notifiers and dumping
3560 kmsg. This only for the users who doubt kdump always
3561 succeeds in any situation.
3562 Note that this also increases risks of kdump failure,
3563 because some panic notifiers can make the crashed
3564 kernel more unstable.
3566 parkbd.port= [HW] Parallel port number the keyboard adapter is
3567 connected to, default is 0.
3569 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
3570 0 for XT, 1 for AT (default is AT).
3573 parport= [HW,PPT] Specify parallel ports. 0 disables.
3574 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
3575 Use 'auto' to force the driver to use any
3576 IRQ/DMA settings detected (the default is to
3577 ignore detected IRQ/DMA settings because of
3578 possible conflicts). You can specify the base
3579 address, IRQ, and DMA settings; IRQ and DMA
3580 should be numbers, or 'auto' (for using detected
3581 settings on that particular port), or 'nofifo'
3582 (to avoid using a FIFO even if it is detected).
3583 Parallel ports are assigned in the order they
3584 are specified on the command line, starting
3587 parport_init_mode= [HW,PPT]
3588 Configure VIA parallel port to operate in
3589 a specific mode. This is necessary on Pegasos
3590 computer where firmware has no options for setting
3591 up parallel port mode and sets it to spp.
3592 Currently this function knows 686a and 8231 chips.
3593 Format: [spp|ps2|epp|ecp|ecpepp]
3596 Halt all CPUs after the first oops has been printed for
3597 the specified number of seconds. This is to be used if
3598 your oopses keep scrolling off the screen.
3603 See header of drivers/block/paride/pcd.c.
3604 See also Documentation/admin-guide/blockdev/paride.rst.
3606 pci=option[,option...] [PCI] various PCI subsystem options.
3608 Some options herein operate on a specific device
3609 or a set of devices (<pci_dev>). These are
3610 specified in one of the following formats:
3612 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
3613 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
3615 Note: the first format specifies a PCI
3616 bus/device/function address which may change
3617 if new hardware is inserted, if motherboard
3618 firmware changes, or due to changes caused
3619 by other kernel parameters. If the
3620 domain is left unspecified, it is
3621 taken to be zero. Optionally, a path
3622 to a device through multiple device/function
3623 addresses can be specified after the base
3624 address (this is more robust against
3625 renumbering issues). The second format
3626 selects devices using IDs from the
3627 configuration space which may match multiple
3628 devices in the system.
3630 earlydump dump PCI config space before the kernel
3632 off [X86] don't probe for the PCI bus
3633 bios [X86-32] force use of PCI BIOS, don't access
3634 the hardware directly. Use this if your machine
3635 has a non-standard PCI host bridge.
3636 nobios [X86-32] disallow use of PCI BIOS, only direct
3637 hardware access methods are allowed. Use this
3638 if you experience crashes upon bootup and you
3639 suspect they are caused by the BIOS.
3640 conf1 [X86] Force use of PCI Configuration Access
3641 Mechanism 1 (config address in IO port 0xCF8,
3642 data in IO port 0xCFC, both 32-bit).
3643 conf2 [X86] Force use of PCI Configuration Access
3644 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
3645 the function, IO port 0xCFA, also 8-bit, sets
3646 bus number. The config space is then accessed
3647 through ports 0xC000-0xCFFF).
3648 See http://wiki.osdev.org/PCI for more info
3649 on the configuration access mechanisms.
3650 noaer [PCIE] If the PCIEAER kernel config parameter is
3651 enabled, this kernel boot option can be used to
3652 disable the use of PCIE advanced error reporting.
3653 nodomains [PCI] Disable support for multiple PCI
3654 root domains (aka PCI segments, in ACPI-speak).
3655 nommconf [X86] Disable use of MMCONFIG for PCI
3657 check_enable_amd_mmconf [X86] check for and enable
3658 properly configured MMIO access to PCI
3659 config space on AMD family 10h CPU
3660 nomsi [MSI] If the PCI_MSI kernel config parameter is
3661 enabled, this kernel boot option can be used to
3662 disable the use of MSI interrupts system-wide.
3663 noioapicquirk [APIC] Disable all boot interrupt quirks.
3664 Safety option to keep boot IRQs enabled. This
3665 should never be necessary.
3666 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
3667 primary IO-APIC for bridges that cannot disable
3668 boot IRQs. This fixes a source of spurious IRQs
3669 when the system masks IRQs.
3670 noioapicreroute [APIC] Disable workaround that uses the
3671 boot IRQ equivalent of an IRQ that connects to
3672 a chipset where boot IRQs cannot be disabled.
3673 The opposite of ioapicreroute.
3674 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
3675 routing table. These calls are known to be buggy
3676 on several machines and they hang the machine
3677 when used, but on other computers it's the only
3678 way to get the interrupt routing table. Try
3679 this option if the kernel is unable to allocate
3680 IRQs or discover secondary PCI buses on your
3682 rom [X86] Assign address space to expansion ROMs.
3683 Use with caution as certain devices share
3684 address decoders between ROMs and other
3686 norom [X86] Do not assign address space to
3687 expansion ROMs that do not already have
3688 BIOS assigned address ranges.
3689 nobar [X86] Do not assign address space to the
3690 BARs that weren't assigned by the BIOS.
3691 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
3692 assigned automatically to PCI devices. You can
3693 make the kernel exclude IRQs of your ISA cards
3695 pirqaddr=0xAAAAA [X86] Specify the physical address
3696 of the PIRQ table (normally generated
3697 by the BIOS) if it is outside the
3698 F0000h-100000h range.
3699 lastbus=N [X86] Scan all buses thru bus #N. Can be
3700 useful if the kernel is unable to find your
3701 secondary buses and you want to tell it
3702 explicitly which ones they are.
3703 assign-busses [X86] Always assign all PCI bus
3704 numbers ourselves, overriding
3705 whatever the firmware may have done.
3706 usepirqmask [X86] Honor the possible IRQ mask stored
3707 in the BIOS $PIR table. This is needed on
3708 some systems with broken BIOSes, notably
3709 some HP Pavilion N5400 and Omnibook XE3
3710 notebooks. This will have no effect if ACPI
3711 IRQ routing is enabled.
3712 noacpi [X86] Do not use ACPI for IRQ routing
3713 or for PCI scanning.
3714 use_crs [X86] Use PCI host bridge window information
3715 from ACPI. On BIOSes from 2008 or later, this
3716 is enabled by default. If you need to use this,
3717 please report a bug.
3718 nocrs [X86] Ignore PCI host bridge windows from ACPI.
3719 If you need to use this, please report a bug.
3720 routeirq Do IRQ routing for all PCI devices.
3721 This is normally done in pci_enable_device(),
3722 so this option is a temporary workaround
3723 for broken drivers that don't call it.
3724 skip_isa_align [X86] do not align io start addr, so can
3725 handle more pci cards
3726 noearly [X86] Don't do any early type 1 scanning.
3727 This might help on some broken boards which
3728 machine check when some devices' config space
3729 is read. But various workarounds are disabled
3730 and some IOMMU drivers will not work.
3731 bfsort Sort PCI devices into breadth-first order.
3732 This sorting is done to get a device
3733 order compatible with older (<= 2.4) kernels.
3734 nobfsort Don't sort PCI devices into breadth-first order.
3735 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
3736 tuning and use the BIOS-configured MPS defaults.
3737 pcie_bus_safe Set every device's MPS to the largest value
3738 supported by all devices below the root complex.
3739 pcie_bus_perf Set device MPS to the largest allowable MPS
3740 based on its parent bus. Also set MRRS (Max
3741 Read Request Size) to the largest supported
3742 value (no larger than the MPS that the device
3743 or bus can support) for best performance.
3744 pcie_bus_peer2peer Set every device's MPS to 128B, which
3745 every device is guaranteed to support. This
3746 configuration allows peer-to-peer DMA between
3747 any pair of devices, possibly at the cost of
3748 reduced performance. This also guarantees
3749 that hot-added devices will work.
3750 cbiosize=nn[KMG] The fixed amount of bus space which is
3751 reserved for the CardBus bridge's IO window.
3752 The default value is 256 bytes.
3753 cbmemsize=nn[KMG] The fixed amount of bus space which is
3754 reserved for the CardBus bridge's memory
3755 window. The default value is 64 megabytes.
3758 [<order of align>@]<pci_dev>[; ...]
3759 Specifies alignment and device to reassign
3760 aligned memory resources. How to
3761 specify the device is described above.
3762 If <order of align> is not specified,
3763 PAGE_SIZE is used as alignment.
3764 A PCI-PCI bridge can be specified if resource
3765 windows need to be expanded.
3766 To specify the alignment for several
3767 instances of a device, the PCI vendor,
3768 device, subvendor, and subdevice may be
3769 specified, e.g., 12@pci:8086:9c22:103c:198f
3770 for 4096-byte alignment.
3771 ecrc= Enable/disable PCIe ECRC (transaction layer
3772 end-to-end CRC checking).
3773 bios: Use BIOS/firmware settings. This is the
3777 hpiosize=nn[KMG] The fixed amount of bus space which is
3778 reserved for hotplug bridge's IO window.
3779 Default size is 256 bytes.
3780 hpmmiosize=nn[KMG] The fixed amount of bus space which is
3781 reserved for hotplug bridge's MMIO window.
3782 Default size is 2 megabytes.
3783 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
3784 reserved for hotplug bridge's MMIO_PREF window.
3785 Default size is 2 megabytes.
3786 hpmemsize=nn[KMG] The fixed amount of bus space which is
3787 reserved for hotplug bridge's MMIO and
3789 Default size is 2 megabytes.
3790 hpbussize=nn The minimum amount of additional bus numbers
3791 reserved for buses below a hotplug bridge.
3793 realloc= Enable/disable reallocating PCI bridge resources
3794 if allocations done by BIOS are too small to
3795 accommodate resources required by all child
3797 off: Turn realloc off
3799 realloc same as realloc=on
3800 noari do not use PCIe ARI.
3801 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
3802 do not use PCIe ATS (and IOMMU device IOTLB).
3803 pcie_scan_all Scan all possible PCIe devices. Otherwise we
3804 only look for one device below a PCIe downstream
3806 big_root_window Try to add a big 64bit memory window to the PCIe
3807 root complex on AMD CPUs. Some GFX hardware
3808 can resize a BAR to allow access to all VRAM.
3809 Adding the window is slightly risky (it may
3810 conflict with unreported devices), so this
3812 disable_acs_redir=<pci_dev>[; ...]
3813 Specify one or more PCI devices (in the format
3814 specified above) separated by semicolons.
3815 Each device specified will have the PCI ACS
3816 redirect capabilities forced off which will
3817 allow P2P traffic between devices through
3818 bridges without forcing it upstream. Note:
3819 this removes isolation between devices and
3820 may put more devices in an IOMMU group.
3821 force_floating [S390] Force usage of floating interrupts.
3822 nomio [S390] Do not use MIO instructions.
3823 norid [S390] ignore the RID field and force use of
3824 one PCI domain per PCI function
3826 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
3829 force Enable ASPM even on devices that claim not to support it.
3830 WARNING: Forcing ASPM on may cause system lockups.
3832 pcie_ports= [PCIE] PCIe port services handling:
3833 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
3834 even if the platform doesn't give the OS permission to
3835 use them. This may cause conflicts if the platform
3836 also tries to use these services.
3837 dpc-native Use native PCIe service for DPC only. May
3838 cause conflicts if firmware uses AER or DPC.
3839 compat Disable native PCIe services (PME, AER, DPC, PCIe
3842 pcie_port_pm= [PCIE] PCIe port power management handling:
3843 off Disable power management of all PCIe ports
3844 force Forcibly enable power management of all PCIe ports
3846 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
3847 nomsi Do not use MSI for native PCIe PME signaling (this makes
3848 all PCIe root ports use INTx for all services).
3850 pcmv= [HW,PCMCIA] BadgePAD 4
3854 Keep all power-domains already enabled by bootloader on,
3855 even if no driver has claimed them. This is useful
3856 for debug and development, but should not be
3857 needed on a platform with proper driver support.
3860 See Documentation/admin-guide/blockdev/paride.rst.
3862 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
3865 See arch/parisc/kernel/pdc_chassis.c
3867 percpu_alloc= Select which percpu first chunk allocator to use.
3868 Currently supported values are "embed" and "page".
3869 Archs may support subset or none of the selections.
3870 See comments in mm/percpu.c for details on each
3871 allocator. This parameter is primarily for debugging
3872 and performance comparison.
3875 See Documentation/admin-guide/blockdev/paride.rst.
3878 See Documentation/admin-guide/blockdev/paride.rst.
3880 pirq= [SMP,APIC] Manual mp-table setup
3881 See Documentation/x86/i386/IO-APIC.rst.
3883 plip= [PPT,NET] Parallel port network link
3884 Format: { parport<nr> | timid | 0 }
3885 See also Documentation/admin-guide/parport.rst.
3887 pmtmr= [X86] Manual setup of pmtmr I/O Port.
3888 Override pmtimer IOPort with a hex value.
3891 pm_debug_messages [SUSPEND,KNL]
3892 Enable suspend/resume debug messages during boot up.
3895 Enable PNP debug messages (depends on the
3896 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
3897 via /sys/module/pnp/parameters/debug. We always show
3898 current resource usage; turning this on also shows
3899 possible settings and some assignment information.
3905 { on | off | curr | res | no-curr | no-res }
3908 [ISAPNP] Exclude IRQs for the autoconfiguration
3911 [ISAPNP] Exclude DMAs for the autoconfiguration
3913 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
3914 Ranges are in pairs (I/O port base and size).
3917 [ISAPNP] Exclude memory regions for the
3919 Ranges are in pairs (memory base and size).
3921 ports= [IP_VS_FTP] IPVS ftp helper module
3923 Up to 8 (IP_VS_APP_MAX_PORTS) ports
3925 Format: <port>,<port>....
3927 powersave=off [PPC] This option disables power saving features.
3928 It specifically disables cpuidle and sets the
3929 platform machine description specific power_save
3930 function to NULL. On Idle the CPU just reduces
3933 ppc_strict_facility_enable
3934 [PPC] This option catches any kernel floating point,
3935 Altivec, VSX and SPE outside of regions specifically
3936 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
3937 There is some performance impact when enabling this.
3941 Disable Hardware Transactional Memory
3944 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC
3945 none - Limited to cond_resched() calls
3946 voluntary - Limited to cond_resched() and might_sleep() calls
3947 full - Any section that isn't explicitly preempt disabled
3948 can be preempted anytime.
3950 print-fatal-signals=
3951 [KNL] debug: print fatal signals
3953 If enabled, warn about various signal handling
3954 related application anomalies: too many signals,
3955 too many POSIX.1 timers, fatal signals causing a
3958 If you hit the warning due to signal overflow,
3959 you might want to try "ulimit -i unlimited".
3963 printk.always_kmsg_dump=
3964 Trigger kmsg_dump for cases other than kernel oops or
3966 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
3969 printk.devkmsg={on,off,ratelimit}
3970 Control writing to /dev/kmsg.
3971 on - unlimited logging to /dev/kmsg from userspace
3972 off - logging to /dev/kmsg disabled
3973 ratelimit - ratelimit the logging
3976 printk.time= Show timing data prefixed to each printk message line
3977 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
3979 processor.max_cstate= [HW,ACPI]
3980 Limit processor to maximum C-state
3981 max_cstate=9 overrides any DMI blacklist limit.
3983 processor.nocst [HW,ACPI]
3984 Ignore the _CST method to determine C-states,
3985 instead using the legacy FADT method
3987 profile= [KNL] Enable kernel profiling via /proc/profile
3988 Format: [<profiletype>,]<number>
3989 Param: <profiletype>: "schedule", "sleep", or "kvm"
3990 [defaults to kernel profiling]
3991 Param: "schedule" - profile schedule points.
3992 Param: "sleep" - profile D-state sleeping (millisecs).
3993 Requires CONFIG_SCHEDSTATS
3994 Param: "kvm" - profile VM exits.
3995 Param: <number> - step/bucket size as a power of 2 for
3996 statistical time based profiling.
3998 prompt_ramdisk= [RAM] [Deprecated]
4000 prot_virt= [S390] enable hosting protected virtual machines
4001 isolated from the hypervisor (if hardware supports
4005 psi= [KNL] Enable or disable pressure stall information
4009 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
4010 probe for; one of (bare|imps|exps|lifebook|any).
4011 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
4013 psmouse.resetafter= [HW,MOUSE]
4014 Try to reset the device after so many bad packets
4017 [HW,MOUSE] Set desired mouse resolution, in dpi.
4018 psmouse.smartscroll=
4019 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
4020 0 = disabled, 1 = enabled (default).
4022 pstore.backend= Specify the name of the pstore backend to use
4025 See Documentation/admin-guide/blockdev/paride.rst.
4027 pti= [X86-64] Control Page Table Isolation of user and
4028 kernel address spaces. Disabling this feature
4029 removes hardening, but improves performance of
4030 system calls and interrupts.
4032 on - unconditionally enable
4033 off - unconditionally disable
4034 auto - kernel detects whether your CPU model is
4035 vulnerable to issues that PTI mitigates
4037 Not specifying this option is equivalent to pti=auto.
4040 Equivalent to pti=off
4043 [KNL] Number of legacy pty's. Overwrites compiled-in
4046 quiet [KNL] Disable most log messages
4051 See Documentation/admin-guide/md.rst.
4053 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
4054 See Documentation/admin-guide/blockdev/ramdisk.rst.
4056 ramdisk_start= [RAM] RAM disk image start address
4058 random.trust_cpu={on,off}
4059 [KNL] Enable or disable trusting the use of the
4060 CPU's random number generator (if available) to
4061 fully seed the kernel's CRNG. Default is controlled
4062 by CONFIG_RANDOM_TRUST_CPU.
4064 ras=option[,option,...] [KNL] RAS-specific options
4067 Disable the Correctable Errors Collector,
4068 see CONFIG_RAS_CEC help text.
4071 The argument is a cpu list, as described above,
4072 except that the string "all" can be used to
4073 specify every CPU on the system.
4075 In kernels built with CONFIG_RCU_NOCB_CPU=y, set
4076 the specified list of CPUs to be no-callback CPUs.
4077 Invocation of these CPUs' RCU callbacks will be
4078 offloaded to "rcuox/N" kthreads created for that
4079 purpose, where "x" is "p" for RCU-preempt, and
4080 "s" for RCU-sched, and "N" is the CPU number.
4081 This reduces OS jitter on the offloaded CPUs,
4082 which can be useful for HPC and real-time
4083 workloads. It can also improve energy efficiency
4084 for asymmetric multiprocessors.
4087 Rather than requiring that offloaded CPUs
4088 (specified by rcu_nocbs= above) explicitly
4089 awaken the corresponding "rcuoN" kthreads,
4090 make these kthreads poll for callbacks.
4091 This improves the real-time response for the
4092 offloaded CPUs by relieving them of the need to
4093 wake up the corresponding kthread, but degrades
4094 energy efficiency by requiring that the kthreads
4095 periodically wake up to do the polling.
4097 rcutree.blimit= [KNL]
4098 Set maximum number of finished RCU callbacks to
4099 process in one batch.
4101 rcutree.dump_tree= [KNL]
4102 Dump the structure of the rcu_node combining tree
4103 out at early boot. This is used for diagnostic
4104 purposes, to verify correct tree setup.
4106 rcutree.gp_cleanup_delay= [KNL]
4107 Set the number of jiffies to delay each step of
4108 RCU grace-period cleanup.
4110 rcutree.gp_init_delay= [KNL]
4111 Set the number of jiffies to delay each step of
4112 RCU grace-period initialization.
4114 rcutree.gp_preinit_delay= [KNL]
4115 Set the number of jiffies to delay each step of
4116 RCU grace-period pre-initialization, that is,
4117 the propagation of recent CPU-hotplug changes up
4118 the rcu_node combining tree.
4120 rcutree.use_softirq= [KNL]
4121 If set to zero, move all RCU_SOFTIRQ processing to
4122 per-CPU rcuc kthreads. Defaults to a non-zero
4123 value, meaning that RCU_SOFTIRQ is used by default.
4124 Specify rcutree.use_softirq=0 to use rcuc kthreads.
4126 But note that CONFIG_PREEMPT_RT=y kernels disable
4127 this kernel boot parameter, forcibly setting it
4130 rcutree.rcu_fanout_exact= [KNL]
4131 Disable autobalancing of the rcu_node combining
4132 tree. This is used by rcutorture, and might
4133 possibly be useful for architectures having high
4134 cache-to-cache transfer latencies.
4136 rcutree.rcu_fanout_leaf= [KNL]
4137 Change the number of CPUs assigned to each
4138 leaf rcu_node structure. Useful for very
4139 large systems, which will choose the value 64,
4140 and for NUMA systems with large remote-access
4141 latencies, which will choose a value aligned
4142 with the appropriate hardware boundaries.
4144 rcutree.rcu_min_cached_objs= [KNL]
4145 Minimum number of objects which are cached and
4146 maintained per one CPU. Object size is equal
4147 to PAGE_SIZE. The cache allows to reduce the
4148 pressure to page allocator, also it makes the
4149 whole algorithm to behave better in low memory
4152 rcutree.jiffies_till_first_fqs= [KNL]
4153 Set delay from grace-period initialization to
4154 first attempt to force quiescent states.
4155 Units are jiffies, minimum value is zero,
4156 and maximum value is HZ.
4158 rcutree.jiffies_till_next_fqs= [KNL]
4159 Set delay between subsequent attempts to force
4160 quiescent states. Units are jiffies, minimum
4161 value is one, and maximum value is HZ.
4163 rcutree.jiffies_till_sched_qs= [KNL]
4164 Set required age in jiffies for a
4165 given grace period before RCU starts
4166 soliciting quiescent-state help from
4167 rcu_note_context_switch() and cond_resched().
4168 If not specified, the kernel will calculate
4169 a value based on the most recent settings
4170 of rcutree.jiffies_till_first_fqs
4171 and rcutree.jiffies_till_next_fqs.
4172 This calculated value may be viewed in
4173 rcutree.jiffies_to_sched_qs. Any attempt to set
4174 rcutree.jiffies_to_sched_qs will be cheerfully
4177 rcutree.kthread_prio= [KNL,BOOT]
4178 Set the SCHED_FIFO priority of the RCU per-CPU
4179 kthreads (rcuc/N). This value is also used for
4180 the priority of the RCU boost threads (rcub/N)
4181 and for the RCU grace-period kthreads (rcu_bh,
4182 rcu_preempt, and rcu_sched). If RCU_BOOST is
4183 set, valid values are 1-99 and the default is 1
4184 (the least-favored priority). Otherwise, when
4185 RCU_BOOST is not set, valid values are 0-99 and
4186 the default is zero (non-realtime operation).
4188 rcutree.rcu_nocb_gp_stride= [KNL]
4189 Set the number of NOCB callback kthreads in
4190 each group, which defaults to the square root
4191 of the number of CPUs. Larger numbers reduce
4192 the wakeup overhead on the global grace-period
4193 kthread, but increases that same overhead on
4194 each group's NOCB grace-period kthread.
4196 rcutree.qhimark= [KNL]
4197 Set threshold of queued RCU callbacks beyond which
4198 batch limiting is disabled.
4200 rcutree.qlowmark= [KNL]
4201 Set threshold of queued RCU callbacks below which
4202 batch limiting is re-enabled.
4204 rcutree.qovld= [KNL]
4205 Set threshold of queued RCU callbacks beyond which
4206 RCU's force-quiescent-state scan will aggressively
4207 enlist help from cond_resched() and sched IPIs to
4208 help CPUs more quickly reach quiescent states.
4209 Set to less than zero to make this be set based
4210 on rcutree.qhimark at boot time and to zero to
4211 disable more aggressive help enlistment.
4213 rcutree.rcu_idle_gp_delay= [KNL]
4214 Set wakeup interval for idle CPUs that have
4215 RCU callbacks (RCU_FAST_NO_HZ=y).
4217 rcutree.rcu_kick_kthreads= [KNL]
4218 Cause the grace-period kthread to get an extra
4219 wake_up() if it sleeps three times longer than
4220 it should at force-quiescent-state time.
4221 This wake_up() will be accompanied by a
4222 WARN_ONCE() splat and an ftrace_dump().
4224 rcutree.rcu_unlock_delay= [KNL]
4225 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels,
4226 this specifies an rcu_read_unlock()-time delay
4227 in microseconds. This defaults to zero.
4228 Larger delays increase the probability of
4229 catching RCU pointer leaks, that is, buggy use
4230 of RCU-protected pointers after the relevant
4231 rcu_read_unlock() has completed.
4233 rcutree.sysrq_rcu= [KNL]
4234 Commandeer a sysrq key to dump out Tree RCU's
4235 rcu_node tree with an eye towards determining
4236 why a new grace period has not yet started.
4238 rcuscale.gp_async= [KNL]
4239 Measure performance of asynchronous
4240 grace-period primitives such as call_rcu().
4242 rcuscale.gp_async_max= [KNL]
4243 Specify the maximum number of outstanding
4244 callbacks per writer thread. When a writer
4245 thread exceeds this limit, it invokes the
4246 corresponding flavor of rcu_barrier() to allow
4247 previously posted callbacks to drain.
4249 rcuscale.gp_exp= [KNL]
4250 Measure performance of expedited synchronous
4251 grace-period primitives.
4253 rcuscale.holdoff= [KNL]
4254 Set test-start holdoff period. The purpose of
4255 this parameter is to delay the start of the
4256 test until boot completes in order to avoid
4259 rcuscale.kfree_rcu_test= [KNL]
4260 Set to measure performance of kfree_rcu() flooding.
4262 rcuscale.kfree_nthreads= [KNL]
4263 The number of threads running loops of kfree_rcu().
4265 rcuscale.kfree_alloc_num= [KNL]
4266 Number of allocations and frees done in an iteration.
4268 rcuscale.kfree_loops= [KNL]
4269 Number of loops doing rcuscale.kfree_alloc_num number
4270 of allocations and frees.
4272 rcuscale.nreaders= [KNL]
4273 Set number of RCU readers. The value -1 selects
4274 N, where N is the number of CPUs. A value
4275 "n" less than -1 selects N-n+1, where N is again
4276 the number of CPUs. For example, -2 selects N
4277 (the number of CPUs), -3 selects N+1, and so on.
4278 A value of "n" less than or equal to -N selects
4281 rcuscale.nwriters= [KNL]
4282 Set number of RCU writers. The values operate
4283 the same as for rcuscale.nreaders.
4284 N, where N is the number of CPUs
4286 rcuscale.perf_type= [KNL]
4287 Specify the RCU implementation to test.
4289 rcuscale.shutdown= [KNL]
4290 Shut the system down after performance tests
4291 complete. This is useful for hands-off automated
4294 rcuscale.verbose= [KNL]
4295 Enable additional printk() statements.
4297 rcuscale.writer_holdoff= [KNL]
4298 Write-side holdoff between grace periods,
4299 in microseconds. The default of zero says
4302 rcutorture.fqs_duration= [KNL]
4303 Set duration of force_quiescent_state bursts
4306 rcutorture.fqs_holdoff= [KNL]
4307 Set holdoff time within force_quiescent_state bursts
4310 rcutorture.fqs_stutter= [KNL]
4311 Set wait time between force_quiescent_state bursts
4314 rcutorture.fwd_progress= [KNL]
4315 Enable RCU grace-period forward-progress testing
4316 for the types of RCU supporting this notion.
4318 rcutorture.fwd_progress_div= [KNL]
4319 Specify the fraction of a CPU-stall-warning
4320 period to do tight-loop forward-progress testing.
4322 rcutorture.fwd_progress_holdoff= [KNL]
4323 Number of seconds to wait between successive
4324 forward-progress tests.
4326 rcutorture.fwd_progress_need_resched= [KNL]
4327 Enclose cond_resched() calls within checks for
4328 need_resched() during tight-loop forward-progress
4331 rcutorture.gp_cond= [KNL]
4332 Use conditional/asynchronous update-side
4333 primitives, if available.
4335 rcutorture.gp_exp= [KNL]
4336 Use expedited update-side primitives, if available.
4338 rcutorture.gp_normal= [KNL]
4339 Use normal (non-expedited) asynchronous
4340 update-side primitives, if available.
4342 rcutorture.gp_sync= [KNL]
4343 Use normal (non-expedited) synchronous
4344 update-side primitives, if available. If all
4345 of rcutorture.gp_cond=, rcutorture.gp_exp=,
4346 rcutorture.gp_normal=, and rcutorture.gp_sync=
4347 are zero, rcutorture acts as if is interpreted
4348 they are all non-zero.
4350 rcutorture.irqreader= [KNL]
4351 Run RCU readers from irq handlers, or, more
4352 accurately, from a timer handler. Not all RCU
4353 flavors take kindly to this sort of thing.
4355 rcutorture.leakpointer= [KNL]
4356 Leak an RCU-protected pointer out of the reader.
4357 This can of course result in splats, and is
4358 intended to test the ability of things like
4359 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect
4362 rcutorture.n_barrier_cbs= [KNL]
4363 Set callbacks/threads for rcu_barrier() testing.
4365 rcutorture.nfakewriters= [KNL]
4366 Set number of concurrent RCU writers. These just
4367 stress RCU, they don't participate in the actual
4368 test, hence the "fake".
4370 rcutorture.nocbs_nthreads= [KNL]
4371 Set number of RCU callback-offload togglers.
4372 Zero (the default) disables toggling.
4374 rcutorture.nocbs_toggle= [KNL]
4375 Set the delay in milliseconds between successive
4376 callback-offload toggling attempts.
4378 rcutorture.nreaders= [KNL]
4379 Set number of RCU readers. The value -1 selects
4380 N-1, where N is the number of CPUs. A value
4381 "n" less than -1 selects N-n-2, where N is again
4382 the number of CPUs. For example, -2 selects N
4383 (the number of CPUs), -3 selects N+1, and so on.
4385 rcutorture.object_debug= [KNL]
4386 Enable debug-object double-call_rcu() testing.
4388 rcutorture.onoff_holdoff= [KNL]
4389 Set time (s) after boot for CPU-hotplug testing.
4391 rcutorture.onoff_interval= [KNL]
4392 Set time (jiffies) between CPU-hotplug operations,
4393 or zero to disable CPU-hotplug testing.
4395 rcutorture.read_exit= [KNL]
4396 Set the number of read-then-exit kthreads used
4397 to test the interaction of RCU updaters and
4398 task-exit processing.
4400 rcutorture.read_exit_burst= [KNL]
4401 The number of times in a given read-then-exit
4402 episode that a set of read-then-exit kthreads
4405 rcutorture.read_exit_delay= [KNL]
4406 The delay, in seconds, between successive
4407 read-then-exit testing episodes.
4409 rcutorture.shuffle_interval= [KNL]
4410 Set task-shuffle interval (s). Shuffling tasks
4411 allows some CPUs to go into dyntick-idle mode
4412 during the rcutorture test.
4414 rcutorture.shutdown_secs= [KNL]
4415 Set time (s) after boot system shutdown. This
4416 is useful for hands-off automated testing.
4418 rcutorture.stall_cpu= [KNL]
4419 Duration of CPU stall (s) to test RCU CPU stall
4420 warnings, zero to disable.
4422 rcutorture.stall_cpu_block= [KNL]
4423 Sleep while stalling if set. This will result
4424 in warnings from preemptible RCU in addition
4425 to any other stall-related activity.
4427 rcutorture.stall_cpu_holdoff= [KNL]
4428 Time to wait (s) after boot before inducing stall.
4430 rcutorture.stall_cpu_irqsoff= [KNL]
4431 Disable interrupts while stalling if set.
4433 rcutorture.stall_gp_kthread= [KNL]
4434 Duration (s) of forced sleep within RCU
4435 grace-period kthread to test RCU CPU stall
4436 warnings, zero to disable. If both stall_cpu
4437 and stall_gp_kthread are specified, the
4438 kthread is starved first, then the CPU.
4440 rcutorture.stat_interval= [KNL]
4441 Time (s) between statistics printk()s.
4443 rcutorture.stutter= [KNL]
4444 Time (s) to stutter testing, for example, specifying
4445 five seconds causes the test to run for five seconds,
4446 wait for five seconds, and so on. This tests RCU's
4447 ability to transition abruptly to and from idle.
4449 rcutorture.test_boost= [KNL]
4450 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
4451 "Maybe" means test if the RCU implementation
4452 under test support RCU priority boosting.
4454 rcutorture.test_boost_duration= [KNL]
4455 Duration (s) of each individual boost test.
4457 rcutorture.test_boost_interval= [KNL]
4458 Interval (s) between each boost test.
4460 rcutorture.test_no_idle_hz= [KNL]
4461 Test RCU's dyntick-idle handling. See also the
4462 rcutorture.shuffle_interval parameter.
4464 rcutorture.torture_type= [KNL]
4465 Specify the RCU implementation to test.
4467 rcutorture.verbose= [KNL]
4468 Enable additional printk() statements.
4470 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
4471 Dump ftrace buffer after reporting RCU CPU
4474 rcupdate.rcu_cpu_stall_suppress= [KNL]
4475 Suppress RCU CPU stall warning messages.
4477 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL]
4478 Suppress RCU CPU stall warning messages and
4479 rcutorture writer stall warnings that occur
4480 during early boot, that is, during the time
4481 before the init task is spawned.
4483 rcupdate.rcu_cpu_stall_timeout= [KNL]
4484 Set timeout for RCU CPU stall warning messages.
4486 rcupdate.rcu_expedited= [KNL]
4487 Use expedited grace-period primitives, for
4488 example, synchronize_rcu_expedited() instead
4489 of synchronize_rcu(). This reduces latency,
4490 but can increase CPU utilization, degrade
4491 real-time latency, and degrade energy efficiency.
4492 No effect on CONFIG_TINY_RCU kernels.
4494 rcupdate.rcu_normal= [KNL]
4495 Use only normal grace-period primitives,
4496 for example, synchronize_rcu() instead of
4497 synchronize_rcu_expedited(). This improves
4498 real-time latency, CPU utilization, and
4499 energy efficiency, but can expose users to
4500 increased grace-period latency. This parameter
4501 overrides rcupdate.rcu_expedited. No effect on
4502 CONFIG_TINY_RCU kernels.
4504 rcupdate.rcu_normal_after_boot= [KNL]
4505 Once boot has completed (that is, after
4506 rcu_end_inkernel_boot() has been invoked), use
4507 only normal grace-period primitives. No effect
4508 on CONFIG_TINY_RCU kernels.
4510 But note that CONFIG_PREEMPT_RT=y kernels enables
4511 this kernel boot parameter, forcibly setting
4512 it to the value one, that is, converting any
4513 post-boot attempt at an expedited RCU grace
4514 period to instead use normal non-expedited
4515 grace-period processing.
4517 rcupdate.rcu_task_ipi_delay= [KNL]
4518 Set time in jiffies during which RCU tasks will
4519 avoid sending IPIs, starting with the beginning
4520 of a given grace period. Setting a large
4521 number avoids disturbing real-time workloads,
4522 but lengthens grace periods.
4524 rcupdate.rcu_task_stall_timeout= [KNL]
4525 Set timeout in jiffies for RCU task stall warning
4526 messages. Disable with a value less than or equal
4529 rcupdate.rcu_self_test= [KNL]
4530 Run the RCU early boot self tests
4534 Run specified binary instead of /init from the ramdisk,
4535 used for early userspace startup. See initrd.
4538 force - Override the decision by the kernel to hide the
4539 advertisement of RDRAND support (this affects
4540 certain AMD processors because of buggy BIOS
4541 support, specifically around the suspend/resume
4545 Turn on/off individual RDT features. List is:
4546 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
4548 E.g. to turn on cmt and turn off mba use:
4552 Format (x86 or x86_64):
4553 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] \
4555 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
4557 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
4558 (prefix with 'panic_' to set mode for panic
4560 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
4561 reboot_force is either force or not specified,
4562 reboot_cpu is s[mp]#### with #### being the processor
4563 to be used for rebooting.
4565 refscale.holdoff= [KNL]
4566 Set test-start holdoff period. The purpose of
4567 this parameter is to delay the start of the
4568 test until boot completes in order to avoid
4571 refscale.loops= [KNL]
4572 Set the number of loops over the synchronization
4573 primitive under test. Increasing this number
4574 reduces noise due to loop start/end overhead,
4575 but the default has already reduced the per-pass
4576 noise to a handful of picoseconds on ca. 2020
4579 refscale.nreaders= [KNL]
4580 Set number of readers. The default value of -1
4581 selects N, where N is roughly 75% of the number
4582 of CPUs. A value of zero is an interesting choice.
4584 refscale.nruns= [KNL]
4585 Set number of runs, each of which is dumped onto
4588 refscale.readdelay= [KNL]
4589 Set the read-side critical-section duration,
4590 measured in microseconds.
4592 refscale.scale_type= [KNL]
4593 Specify the read-protection implementation to test.
4595 refscale.shutdown= [KNL]
4596 Shut down the system at the end of the performance
4597 test. This defaults to 1 (shut it down) when
4598 refscale is built into the kernel and to 0 (leave
4599 it running) when refscale is built as a module.
4601 refscale.verbose= [KNL]
4602 Enable additional printk() statements.
4604 refscale.verbose_batched= [KNL]
4605 Batch the additional printk() statements. If zero
4606 (the default) or negative, print everything. Otherwise,
4607 print every Nth verbose statement, where N is the value
4611 [KNL, SMP] Set scheduler's default relax_domain_level.
4612 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
4614 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
4615 Format: <base1>,<size1>[,<base2>,<size2>,...]
4616 Reserve I/O ports or memory so the kernel won't use
4617 them. If <base> is less than 0x10000, the region
4618 is assumed to be I/O ports; otherwise it is memory.
4620 reservetop= [X86-32]
4622 Reserves a hole at the top of the kernel virtual
4627 Set the amount of memory to reserve for BIOS at
4628 the bottom of the address space.
4630 reset_devices [KNL] Force drivers to reset the underlying device
4631 during initialization.
4634 Specify the partition device for software suspend
4636 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
4638 resume_offset= [SWSUSP]
4639 Specify the offset from the beginning of the partition
4640 given by "resume=" at which the swap header is located,
4641 in <PAGE_SIZE> units (needed only for swap files).
4642 See Documentation/power/swsusp-and-swap-files.rst
4644 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
4645 read the resume files
4647 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
4648 Useful for devices that are detected asynchronously
4649 (e.g. USB and MMC devices).
4651 hibernate= [HIBERNATION]
4652 noresume Don't check if there's a hibernation image
4653 present during boot.
4654 nocompress Don't compress/decompress hibernation images.
4655 no Disable hibernation and resume.
4656 protect_image Turn on image protection during restoration
4657 (that will set all pages holding image data
4658 during restoration read-only).
4660 retain_initrd [RAM] Keep initrd memory after extraction
4662 rfkill.default_state=
4663 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
4664 etc. communication is blocked by default.
4667 rfkill.master_switch_mode=
4668 0 The "airplane mode" button does nothing.
4669 1 The "airplane mode" button toggles between everything
4670 blocked and the previous configuration.
4671 2 The "airplane mode" button toggles between everything
4672 blocked and everything unblocked.
4674 rhash_entries= [KNL,NET]
4675 Set number of hash buckets for route cache
4678 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
4681 ro [KNL] Mount root device read-only on boot
4684 on Mark read-only kernel memory as read-only (default).
4685 off Leave read-only kernel memory writable for debugging.
4688 Enable the uart passthrough on the designated usb port
4689 on Rockchip SoCs. When active, the signals of the
4690 debug-uart get routed to the D+ and D- pins of the usb
4691 port and the regular usb controller gets disabled.
4693 root= [KNL] Root filesystem
4694 See name_to_dev_t comment in init/do_mounts.c.
4696 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
4697 mount the root filesystem
4699 rootflags= [KNL] Set root filesystem mount option string
4701 rootfstype= [KNL] Set root filesystem type
4703 rootwait [KNL] Wait (indefinitely) for root device to show up.
4704 Useful for devices that are detected asynchronously
4705 (e.g. USB and MMC devices).
4707 rproc_mem=nn[KMG][@address]
4708 [KNL,ARM,CMA] Remoteproc physical memory block.
4709 Memory area to be used by remote processor image,
4712 rw [KNL] Mount root device read-write on boot
4714 S [KNL] Run init in single mode
4716 s390_iommu= [HW,S390]
4717 Set s390 IOTLB flushing mode
4719 With strict flushing every unmap operation will result in
4720 an IOTLB flush. Default is lazy flushing before reuse,
4724 See drivers/net/irda/sa1100_ir.c.
4726 sbni= [NET] Granch SBNI12 leased line adapter
4728 sched_debug [KNL] Enables verbose scheduler debug messages.
4730 schedstats= [KNL,X86] Enable or disable scheduled statistics.
4731 Allowed values are enable and disable. This feature
4732 incurs a small amount of overhead in the scheduler
4733 but is useful for debugging and performance tuning.
4735 sched_thermal_decay_shift=
4736 [KNL, SMP] Set a decay shift for scheduler thermal
4737 pressure signal. Thermal pressure signal follows the
4738 default decay period of other scheduler pelt
4739 signals(usually 32 ms but configurable). Setting
4740 sched_thermal_decay_shift will left shift the decay
4741 period for the thermal pressure signal by the shift
4743 i.e. with the default pelt decay period of 32 ms
4744 sched_thermal_decay_shift thermal pressure decay pr
4748 Format: integer between 0 and 10
4751 scftorture.holdoff= [KNL]
4752 Number of seconds to hold off before starting
4753 test. Defaults to zero for module insertion and
4754 to 10 seconds for built-in smp_call_function()
4757 scftorture.longwait= [KNL]
4758 Request ridiculously long waits randomly selected
4759 up to the chosen limit in seconds. Zero (the
4760 default) disables this feature. Please note
4761 that requesting even small non-zero numbers of
4762 seconds can result in RCU CPU stall warnings,
4763 softlockup complaints, and so on.
4765 scftorture.nthreads= [KNL]
4766 Number of kthreads to spawn to invoke the
4767 smp_call_function() family of functions.
4768 The default of -1 specifies a number of kthreads
4769 equal to the number of CPUs.
4771 scftorture.onoff_holdoff= [KNL]
4772 Number seconds to wait after the start of the
4773 test before initiating CPU-hotplug operations.
4775 scftorture.onoff_interval= [KNL]
4776 Number seconds to wait between successive
4777 CPU-hotplug operations. Specifying zero (which
4778 is the default) disables CPU-hotplug operations.
4780 scftorture.shutdown_secs= [KNL]
4781 The number of seconds following the start of the
4782 test after which to shut down the system. The
4783 default of zero avoids shutting down the system.
4784 Non-zero values are useful for automated tests.
4786 scftorture.stat_interval= [KNL]
4787 The number of seconds between outputting the
4788 current test statistics to the console. A value
4789 of zero disables statistics output.
4791 scftorture.stutter_cpus= [KNL]
4792 The number of jiffies to wait between each change
4793 to the set of CPUs under test.
4795 scftorture.use_cpus_read_lock= [KNL]
4796 Use use_cpus_read_lock() instead of the default
4797 preempt_disable() to disable CPU hotplug
4798 while invoking one of the smp_call_function*()
4801 scftorture.verbose= [KNL]
4802 Enable additional printk() statements.
4804 scftorture.weight_single= [KNL]
4805 The probability weighting to use for the
4806 smp_call_function_single() function with a zero
4807 "wait" parameter. A value of -1 selects the
4808 default if all other weights are -1. However,
4809 if at least one weight has some other value, a
4810 value of -1 will instead select a weight of zero.
4812 scftorture.weight_single_wait= [KNL]
4813 The probability weighting to use for the
4814 smp_call_function_single() function with a
4815 non-zero "wait" parameter. See weight_single.
4817 scftorture.weight_many= [KNL]
4818 The probability weighting to use for the
4819 smp_call_function_many() function with a zero
4820 "wait" parameter. See weight_single.
4821 Note well that setting a high probability for
4822 this weighting can place serious IPI load
4825 scftorture.weight_many_wait= [KNL]
4826 The probability weighting to use for the
4827 smp_call_function_many() function with a
4828 non-zero "wait" parameter. See weight_single
4831 scftorture.weight_all= [KNL]
4832 The probability weighting to use for the
4833 smp_call_function_all() function with a zero
4834 "wait" parameter. See weight_single and
4837 scftorture.weight_all_wait= [KNL]
4838 The probability weighting to use for the
4839 smp_call_function_all() function with a
4840 non-zero "wait" parameter. See weight_single
4843 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
4844 xtime_lock contention on larger systems, and/or RCU lock
4845 contention on all systems with CONFIG_MAXSMP set.
4846 Format: { "0" | "1" }
4847 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
4849 Note: increases power consumption, thus should only be
4850 enabled if running jitter sensitive (HPC/RT) workloads.
4852 security= [SECURITY] Choose a legacy "major" security module to
4853 enable at boot. This has been deprecated by the
4856 selinux= [SELINUX] Disable or enable SELinux at boot time.
4857 Format: { "0" | "1" }
4858 See security/selinux/Kconfig help text.
4863 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
4864 Format: { "0" | "1" }
4865 See security/apparmor/Kconfig help text
4868 Default value is set via kernel config option.
4870 serialnumber [BUGS=X86-32]
4873 Maximal number of shapers.
4881 Disable merging of slabs with similar size. May be
4882 necessary if there is some reason to distinguish
4883 allocs to different slabs, especially in hardened
4884 environments where the risk of heap overflows and
4885 layout control by attackers can usually be
4886 frustrated by disabling merging. This will reduce
4887 most of the exposure of a heap attack to a single
4888 cache (risks via metadata attacks are mostly
4889 unchanged). Debug options disable merging on their
4891 For more information see Documentation/vm/slub.rst.
4893 slab_max_order= [MM, SLAB]
4894 Determines the maximum allowed order for slabs.
4895 A high setting may cause OOMs due to memory
4896 fragmentation. Defaults to 1 for systems with
4897 more than 32MB of RAM, 0 otherwise.
4899 slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB]
4900 Enabling slub_debug allows one to determine the
4901 culprit if slab objects become corrupted. Enabling
4902 slub_debug can create guard zones around objects and
4903 may poison objects when not in use. Also tracks the
4904 last alloc / free. For more information see
4905 Documentation/vm/slub.rst.
4907 slub_memcg_sysfs= [MM, SLUB]
4908 Determines whether to enable sysfs directories for
4909 memory cgroup sub-caches. 1 to enable, 0 to disable.
4910 The default is determined by CONFIG_SLUB_MEMCG_SYSFS_ON.
4911 Enabling this can lead to a very high number of debug
4912 directories and files being created under
4915 slub_max_order= [MM, SLUB]
4916 Determines the maximum allowed order for slabs.
4917 A high setting may cause OOMs due to memory
4918 fragmentation. For more information see
4919 Documentation/vm/slub.rst.
4921 slub_min_objects= [MM, SLUB]
4922 The minimum number of objects per slab. SLUB will
4923 increase the slab order up to slub_max_order to
4924 generate a sufficiently large slab able to contain
4925 the number of objects indicated. The higher the number
4926 of objects the smaller the overhead of tracking slabs
4927 and the less frequently locks need to be acquired.
4928 For more information see Documentation/vm/slub.rst.
4930 slub_min_order= [MM, SLUB]
4931 Determines the minimum page order for slabs. Must be
4932 lower than slub_max_order.
4933 For more information see Documentation/vm/slub.rst.
4935 slub_nomerge [MM, SLUB]
4936 Same with slab_nomerge. This is supported for legacy.
4937 See slab_nomerge for more information.
4940 Format: <io1>[,<io2>[,...,<io8>]]
4942 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
4943 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
4944 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
4945 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
4946 smsc-ircc2.ircc_irq= [HW] IRQ line
4947 smsc-ircc2.ircc_dma= [HW] DMA channel
4948 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
4949 0: Toshiba Satellite 1800 (GP data pin select)
4950 1: Fast pin select (default)
4953 smt [KNL,S390] Set the maximum number of threads (logical
4954 CPUs) to use per physical CPU on systems capable of
4955 symmetric multithreading (SMT). Will be capped to the
4956 actual hardware limit.
4958 Default: -1 (no limit)
4961 [KNL] Should the soft-lockup detector generate panics.
4964 A value of 1 instructs the soft-lockup detector
4965 to panic the machine when a soft-lockup occurs. It is
4966 also controlled by the kernel.softlockup_panic sysctl
4967 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
4968 respective build-time switch to that functionality.
4970 softlockup_all_cpu_backtrace=
4971 [KNL] Should the soft-lockup detector generate
4972 backtraces on all cpus.
4975 sonypi.*= [HW] Sony Programmable I/O Control Device driver
4976 See Documentation/admin-guide/laptops/sonypi.rst
4978 spectre_v2= [X86] Control mitigation of Spectre variant 2
4979 (indirect branch speculation) vulnerability.
4980 The default operation protects the kernel from
4983 on - unconditionally enable, implies
4985 off - unconditionally disable, implies
4987 auto - kernel detects whether your CPU model is
4990 Selecting 'on' will, and 'auto' may, choose a
4991 mitigation method at run time according to the
4992 CPU, the available microcode, the setting of the
4993 CONFIG_RETPOLINE configuration option, and the
4994 compiler with which the kernel was built.
4996 Selecting 'on' will also enable the mitigation
4997 against user space to user space task attacks.
4999 Selecting 'off' will disable both the kernel and
5000 the user space protections.
5002 Specific mitigations can also be selected manually:
5004 retpoline - replace indirect branches
5005 retpoline,generic - google's original retpoline
5006 retpoline,amd - AMD-specific minimal thunk
5008 Not specifying this option is equivalent to
5012 [X86] Control mitigation of Spectre variant 2
5013 (indirect branch speculation) vulnerability between
5016 on - Unconditionally enable mitigations. Is
5017 enforced by spectre_v2=on
5019 off - Unconditionally disable mitigations. Is
5020 enforced by spectre_v2=off
5022 prctl - Indirect branch speculation is enabled,
5023 but mitigation can be enabled via prctl
5024 per thread. The mitigation control state
5025 is inherited on fork.
5028 - Like "prctl" above, but only STIBP is
5029 controlled per thread. IBPB is issued
5030 always when switching between different user
5034 - Same as "prctl" above, but all seccomp
5035 threads will enable the mitigation unless
5036 they explicitly opt out.
5039 - Like "seccomp" above, but only STIBP is
5040 controlled per thread. IBPB is issued
5041 always when switching between different
5042 user space processes.
5044 auto - Kernel selects the mitigation depending on
5045 the available CPU features and vulnerability.
5048 If CONFIG_SECCOMP=y then "seccomp", otherwise "prctl"
5050 Not specifying this option is equivalent to
5051 spectre_v2_user=auto.
5053 spec_store_bypass_disable=
5054 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
5055 (Speculative Store Bypass vulnerability)
5057 Certain CPUs are vulnerable to an exploit against a
5058 a common industry wide performance optimization known
5059 as "Speculative Store Bypass" in which recent stores
5060 to the same memory location may not be observed by
5061 later loads during speculative execution. The idea
5062 is that such stores are unlikely and that they can
5063 be detected prior to instruction retirement at the
5064 end of a particular speculation execution window.
5066 In vulnerable processors, the speculatively forwarded
5067 store can be used in a cache side channel attack, for
5068 example to read memory to which the attacker does not
5069 directly have access (e.g. inside sandboxed code).
5071 This parameter controls whether the Speculative Store
5072 Bypass optimization is used.
5074 On x86 the options are:
5076 on - Unconditionally disable Speculative Store Bypass
5077 off - Unconditionally enable Speculative Store Bypass
5078 auto - Kernel detects whether the CPU model contains an
5079 implementation of Speculative Store Bypass and
5080 picks the most appropriate mitigation. If the
5081 CPU is not vulnerable, "off" is selected. If the
5082 CPU is vulnerable the default mitigation is
5083 architecture and Kconfig dependent. See below.
5084 prctl - Control Speculative Store Bypass per thread
5085 via prctl. Speculative Store Bypass is enabled
5086 for a process by default. The state of the control
5087 is inherited on fork.
5088 seccomp - Same as "prctl" above, but all seccomp threads
5089 will disable SSB unless they explicitly opt out.
5091 Default mitigations:
5092 X86: If CONFIG_SECCOMP=y "seccomp", otherwise "prctl"
5094 On powerpc the options are:
5096 on,auto - On Power8 and Power9 insert a store-forwarding
5097 barrier on kernel entry and exit. On Power7
5098 perform a software flush on kernel entry and
5102 Not specifying this option is equivalent to
5103 spec_store_bypass_disable=auto.
5105 spia_io_base= [HW,MTD]
5111 [X86] Enable split lock detection
5113 When enabled (and if hardware support is present), atomic
5114 instructions that access data across cache line
5115 boundaries will result in an alignment check exception.
5119 warn - the kernel will emit rate limited warnings
5120 about applications triggering the #AC
5121 exception. This mode is the default on CPUs
5122 that supports split lock detection.
5124 fatal - the kernel will send SIGBUS to applications
5125 that trigger the #AC exception.
5127 If an #AC exception is hit in the kernel or in
5128 firmware (i.e. not while executing in user mode)
5129 the kernel will oops in either "warn" or "fatal"
5133 Control the Special Register Buffer Data Sampling
5136 Certain CPUs are vulnerable to an MDS-like
5137 exploit which can leak bits from the random
5140 By default, this issue is mitigated by
5141 microcode. However, the microcode fix can cause
5142 the RDRAND and RDSEED instructions to become
5143 much slower. Among other effects, this will
5144 result in reduced throughput from /dev/urandom.
5146 The microcode mitigation can be disabled with
5147 the following option:
5149 off: Disable mitigation and remove
5150 performance impact to RDRAND and RDSEED
5152 srcutree.counter_wrap_check [KNL]
5153 Specifies how frequently to check for
5154 grace-period sequence counter wrap for the
5155 srcu_data structure's ->srcu_gp_seq_needed field.
5156 The greater the number of bits set in this kernel
5157 parameter, the less frequently counter wrap will
5158 be checked for. Note that the bottom two bits
5161 srcutree.exp_holdoff [KNL]
5162 Specifies how many nanoseconds must elapse
5163 since the end of the last SRCU grace period for
5164 a given srcu_struct until the next normal SRCU
5165 grace period will be considered for automatic
5166 expediting. Set to zero to disable automatic
5170 Speculative Store Bypass Disable control
5172 On CPUs that are vulnerable to the Speculative
5173 Store Bypass vulnerability and offer a
5174 firmware based mitigation, this parameter
5175 indicates how the mitigation should be used:
5177 force-on: Unconditionally enable mitigation for
5178 for both kernel and userspace
5179 force-off: Unconditionally disable mitigation for
5180 for both kernel and userspace
5181 kernel: Always enable mitigation in the
5182 kernel, and offer a prctl interface
5183 to allow userspace to register its
5184 interest in being mitigated too.
5186 stack_guard_gap= [MM]
5187 override the default stack gap protection. The value
5188 is in page units and it defines how many pages prior
5189 to (for stacks growing down) resp. after (for stacks
5190 growing up) the main stack are reserved for no other
5191 mapping. Default value is 256 pages.
5194 Enabled the stack tracer on boot up.
5196 stacktrace_filter=[function-list]
5197 [FTRACE] Limit the functions that the stack tracer
5198 will trace at boot up. function-list is a comma-separated
5199 list of functions. This list can be changed at run
5200 time by the stack_trace_filter file in the debugfs
5201 tracing directory. Note, this enables stack tracing
5202 and the stacktrace above is not needed.
5206 Set the STI (builtin display/keyboard on the HP-PARISC
5207 machines) console (graphic card) which should be used
5208 as the initial boot-console.
5209 See also comment in drivers/video/console/sticore.c.
5212 See comment in drivers/video/console/sticore.c.
5215 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
5217 sunrpc.min_resvport=
5218 sunrpc.max_resvport=
5220 SunRPC servers often require that client requests
5221 originate from a privileged port (i.e. a port in the
5222 range 0 < portnr < 1024).
5223 An administrator who wishes to reserve some of these
5224 ports for other uses may adjust the range that the
5225 kernel's sunrpc client considers to be privileged
5226 using these two parameters to set the minimum and
5227 maximum port values.
5229 sunrpc.svc_rpc_per_connection_limit=
5231 Limit the number of requests that the server will
5232 process in parallel from a single connection.
5233 The default value is 0 (no limit).
5237 Control how the NFS server code allocates CPUs to
5238 service thread pools. Depending on how many NICs
5239 you have and where their interrupts are bound, this
5240 option will affect which CPUs will do NFS serving.
5241 Note: this parameter cannot be changed while the
5242 NFS server is running.
5244 auto the server chooses an appropriate mode
5245 automatically using heuristics
5246 global a single global pool contains all CPUs
5247 percpu one pool for each CPU
5248 pernode one pool for each NUMA node (equivalent
5249 to global on non-NUMA machines)
5251 sunrpc.tcp_slot_table_entries=
5252 sunrpc.udp_slot_table_entries=
5254 Sets the upper limit on the number of simultaneous
5255 RPC calls that can be sent from the client to a
5256 server. Increasing these values may allow you to
5257 improve throughput, but will also increase the
5258 amount of memory reserved for use by the client.
5260 suspend.pm_test_delay=
5262 Sets the number of seconds to remain in a suspend test
5263 mode before resuming the system (see
5264 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
5265 is set. Default value is 5.
5268 Format: { on | off | y | n | 1 | 0 }
5269 This parameter controls use of the Protected
5270 Execution Facility on pSeries.
5273 [KNL] Enable accounting of swap in memory resource
5274 controller if no parameter or 1 is given or disable
5275 it if 0 is given (See Documentation/admin-guide/cgroup-v1/memory.rst)
5277 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
5278 Format: { <int> | force | noforce }
5279 <int> -- Number of I/O TLB slabs
5280 force -- force using of bounce buffers even if they
5281 wouldn't be automatically used by the kernel
5282 noforce -- Never use bounce buffers (for debugging)
5287 Set a sysctl parameter, right before loading the init
5288 process, as if the value was written to the respective
5289 /proc/sys/... file. Both '.' and '/' are recognized as
5290 separators. Unrecognized parameters and invalid values
5291 are reported in the kernel log. Sysctls registered
5292 later by a loaded module cannot be set this way.
5293 Example: sysctl.vm.swappiness=40
5295 sysfs.deprecated=0|1 [KNL]
5296 Enable/disable old style sysfs layout for old udev
5297 on older distributions. When this option is enabled
5298 very new udev will not work anymore. When this option
5299 is disabled (or CONFIG_SYSFS_DEPRECATED not compiled)
5300 in older udev will not work anymore.
5301 Default depends on CONFIG_SYSFS_DEPRECATED_V2 set in
5302 the kernel configuration.
5304 sysrq_always_enabled
5306 Ignore sysrq setting - this boot parameter will
5307 neutralize any effect of /proc/sys/kernel/sysrq.
5308 Useful for debugging.
5310 tcpmhash_entries= [KNL,NET]
5311 Set the number of tcp_metrics_hash slots.
5312 Default value is 8192 or 16384 depending on total
5313 ram pages. This is used to specify the TCP metrics
5314 cache size. See Documentation/networking/ip-sysctl.rst
5315 "tcp_no_metrics_save" section for more details.
5319 test_suspend= [SUSPEND][,N]
5320 Specify "mem" (for Suspend-to-RAM) or "standby" (for
5321 standby suspend) or "freeze" (for suspend type freeze)
5322 as the system sleep state during system startup with
5323 the optional capability to repeat N number of times.
5324 The system is woken from this state using a
5325 wakeup-capable RTC alarm.
5327 thash_entries= [KNL,NET]
5328 Set number of hash buckets for TCP connection
5330 thermal.act= [HW,ACPI]
5331 -1: disable all active trip points in all thermal zones
5332 <degrees C>: override all lowest active trip points
5334 thermal.crt= [HW,ACPI]
5335 -1: disable all critical trip points in all thermal zones
5336 <degrees C>: override all critical trip points
5338 thermal.nocrt= [HW,ACPI]
5339 Set to disable actions on ACPI thermal zone
5340 critical and hot trip points.
5342 thermal.off= [HW,ACPI]
5343 1: disable ACPI thermal control
5345 thermal.psv= [HW,ACPI]
5346 -1: disable all passive trip points
5347 <degrees C>: override all passive trip points to this
5350 thermal.tzp= [HW,ACPI]
5351 Specify global default ACPI thermal zone polling rate
5352 <deci-seconds>: poll all this frequency
5353 0: no polling (default)
5356 Force threading of all interrupt handlers except those
5357 marked explicitly IRQF_NO_THREAD.
5361 Specify if the kernel should make use of the cpu
5362 topology information if the hardware supports this.
5363 The scheduler will make use of this information and
5364 e.g. base its process migration decisions on it.
5367 topology_updates= [KNL, PPC, NUMA]
5369 Specify if the kernel should ignore (off)
5370 topology updates sent by the hypervisor to this
5373 torture.disable_onoff_at_boot= [KNL]
5374 Prevent the CPU-hotplug component of torturing
5375 until after init has spawned.
5377 torture.ftrace_dump_at_shutdown= [KNL]
5378 Dump the ftrace buffer at torture-test shutdown,
5379 even if there were no errors. This can be a
5380 very costly operation when many torture tests
5381 are running concurrently, especially on systems
5382 with rotating-rust storage.
5384 torture.verbose_sleep_frequency= [KNL]
5385 Specifies how many verbose printk()s should be
5386 emitted between each sleep. The default of zero
5387 disables verbose-printk() sleeping.
5389 torture.verbose_sleep_duration= [KNL]
5390 Duration of each verbose-printk() sleep in jiffies.
5394 tpm_suspend_pcr=[HW,TPM]
5395 Format: integer pcr id
5396 Specify that at suspend time, the tpm driver
5397 should extend the specified pcr with zeros,
5398 as a workaround for some chips which fail to
5399 flush the last written pcr on TPM_SaveState.
5400 This will guarantee that all the other pcrs
5403 trace_buf_size=nn[KMG]
5404 [FTRACE] will set tracing buffer size on each cpu.
5406 trace_event=[event-list]
5407 [FTRACE] Set and start specified trace events in order
5408 to facilitate early boot debugging. The event-list is a
5409 comma-separated list of trace events to enable. See
5410 also Documentation/trace/events.rst
5412 trace_options=[option-list]
5413 [FTRACE] Enable or disable tracer options at boot.
5414 The option-list is a comma delimited list of options
5415 that can be enabled or disabled just as if you were
5416 to echo the option name into
5418 /sys/kernel/debug/tracing/trace_options
5420 For example, to enable stacktrace option (to dump the
5421 stack trace of each event), add to the command line:
5423 trace_options=stacktrace
5425 See also Documentation/trace/ftrace.rst "trace options"
5429 Have the tracepoints sent to printk as well as the
5430 tracing ring buffer. This is useful for early boot up
5431 where the system hangs or reboots and does not give the
5432 option for reading the tracing buffer or performing a
5433 ftrace_dump_on_oops.
5435 To turn off having tracepoints sent to printk,
5436 echo 0 > /proc/sys/kernel/tracepoint_printk
5437 Note, echoing 1 into this file without the
5438 tracepoint_printk kernel cmdline option has no effect.
5442 Having tracepoints sent to printk() and activating high
5443 frequency tracepoints such as irq or sched, can cause
5444 the system to live lock.
5447 [FTRACE] enable this option to disable tracing when a
5448 warning is hit. This turns off "tracing_on". Tracing can
5449 be enabled again by echoing '1' into the "tracing_on"
5450 file located in /sys/kernel/debug/tracing/
5452 This option is useful, as it disables the trace before
5453 the WARNING dump is called, which prevents the trace to
5454 be filled with content caused by the warning output.
5456 This option can also be set at run time via the sysctl
5457 option: kernel/traceoff_on_warning
5459 transparent_hugepage=
5461 Format: [always|madvise|never]
5462 Can be used to control the default behavior of the system
5463 with respect to transparent hugepages.
5464 See Documentation/admin-guide/mm/transhuge.rst
5467 tsc= Disable clocksource stability checks for TSC.
5469 [x86] reliable: mark tsc clocksource as reliable, this
5470 disables clocksource verification at runtime, as well
5471 as the stability checks done at bootup. Used to enable
5472 high-resolution timer mode on older hardware, and in
5473 virtualized environment.
5474 [x86] noirqtime: Do not use TSC to do irq accounting.
5475 Used to run time disable IRQ_TIME_ACCOUNTING on any
5476 platforms where RDTSC is slow and this accounting
5478 [x86] unstable: mark the TSC clocksource as unstable, this
5479 marks the TSC unconditionally unstable at bootup and
5480 avoids any further wobbles once the TSC watchdog notices.
5481 [x86] nowatchdog: disable clocksource watchdog. Used
5482 in situations with strict latency requirements (where
5483 interruptions from clocksource watchdog are not
5486 tsc_early_khz= [X86] Skip early TSC calibration and use the given
5487 value instead. Useful when the early TSC frequency discovery
5488 procedure is not reliable, such as on overclocked systems
5489 with CPUID.16h support and partial CPUID.15h support.
5490 Format: <unsigned int>
5492 tsx= [X86] Control Transactional Synchronization
5493 Extensions (TSX) feature in Intel processors that
5494 support TSX control.
5496 This parameter controls the TSX feature. The options are:
5498 on - Enable TSX on the system. Although there are
5499 mitigations for all known security vulnerabilities,
5500 TSX has been known to be an accelerator for
5501 several previous speculation-related CVEs, and
5502 so there may be unknown security risks associated
5503 with leaving it enabled.
5505 off - Disable TSX on the system. (Note that this
5506 option takes effect only on newer CPUs which are
5507 not vulnerable to MDS, i.e., have
5508 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
5509 the new IA32_TSX_CTRL MSR through a microcode
5510 update. This new MSR allows for the reliable
5511 deactivation of the TSX functionality.)
5513 auto - Disable TSX if X86_BUG_TAA is present,
5514 otherwise enable TSX on the system.
5516 Not specifying this option is equivalent to tsx=off.
5518 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5521 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
5522 Abort (TAA) vulnerability.
5524 Similar to Micro-architectural Data Sampling (MDS)
5525 certain CPUs that support Transactional
5526 Synchronization Extensions (TSX) are vulnerable to an
5527 exploit against CPU internal buffers which can forward
5528 information to a disclosure gadget under certain
5531 In vulnerable processors, the speculatively forwarded
5532 data can be used in a cache side channel attack, to
5533 access data to which the attacker does not have direct
5536 This parameter controls the TAA mitigation. The
5539 full - Enable TAA mitigation on vulnerable CPUs
5542 full,nosmt - Enable TAA mitigation and disable SMT on
5543 vulnerable CPUs. If TSX is disabled, SMT
5544 is not disabled because CPU is not
5545 vulnerable to cross-thread TAA attacks.
5546 off - Unconditionally disable TAA mitigation
5548 On MDS-affected machines, tsx_async_abort=off can be
5549 prevented by an active MDS mitigation as both vulnerabilities
5550 are mitigated with the same mechanism so in order to disable
5551 this mitigation, you need to specify mds=off too.
5553 Not specifying this option is equivalent to
5554 tsx_async_abort=full. On CPUs which are MDS affected
5555 and deploy MDS mitigation, TAA mitigation is not
5556 required and doesn't provide any additional
5560 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5562 turbografx.map[2|3]= [HW,JOY]
5563 TurboGraFX parallel port interface
5565 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
5566 See also Documentation/input/devices/joystick-parport.rst
5568 udbg-immortal [PPC] When debugging early kernel crashes that
5569 happen after console_init() and before a proper
5570 console driver takes over, this boot options might
5571 help "seeing" what's going on.
5573 uhash_entries= [KNL,NET]
5574 Set number of hash buckets for UDP/UDP-Lite connections
5577 [USB] Ignore overcurrent events (default N).
5578 Some badly-designed motherboards generate lots of
5579 bogus events, for ports that aren't wired to
5580 anything. Set this parameter to avoid log spamming.
5581 Note that genuine overcurrent events won't be
5585 [X86] Cause panic on unknown NMI.
5587 usbcore.authorized_default=
5588 [USB] Default USB device authorization:
5589 (default -1 = authorized except for wireless USB,
5590 0 = not authorized, 1 = authorized, 2 = authorized
5591 if device connected to internal port)
5593 usbcore.autosuspend=
5594 [USB] The autosuspend time delay (in seconds) used
5595 for newly-detected USB devices (default 2). This
5596 is the time required before an idle device will be
5597 autosuspended. Devices for which the delay is set
5598 to a negative value won't be autosuspended at all.
5600 usbcore.usbfs_snoop=
5601 [USB] Set to log all usbfs traffic (default 0 = off).
5603 usbcore.usbfs_snoop_max=
5604 [USB] Maximum number of bytes to snoop in each URB
5607 usbcore.blinkenlights=
5608 [USB] Set to cycle leds on hubs (default 0 = off).
5610 usbcore.old_scheme_first=
5611 [USB] Start with the old device initialization
5612 scheme (default 0 = off).
5614 usbcore.usbfs_memory_mb=
5615 [USB] Memory limit (in MB) for buffers allocated by
5616 usbfs (default = 16, 0 = max = 2047).
5618 usbcore.use_both_schemes=
5619 [USB] Try the other device initialization scheme
5620 if the first one fails (default 1 = enabled).
5622 usbcore.initial_descriptor_timeout=
5623 [USB] Specifies timeout for the initial 64-byte
5624 USB_REQ_GET_DESCRIPTOR request in milliseconds
5625 (default 5000 = 5.0 seconds).
5627 usbcore.nousb [USB] Disable the USB subsystem
5630 [USB] A list of quirk entries to augment the built-in
5631 usb core quirk list. List entries are separated by
5632 commas. Each entry has the form
5633 VendorID:ProductID:Flags. The IDs are 4-digit hex
5634 numbers and Flags is a set of letters. Each letter
5635 will change the built-in quirk; setting it if it is
5636 clear and clearing it if it is set. The letters have
5637 the following meanings:
5638 a = USB_QUIRK_STRING_FETCH_255 (string
5639 descriptors must not be fetched using
5641 b = USB_QUIRK_RESET_RESUME (device can't resume
5642 correctly so reset it instead);
5643 c = USB_QUIRK_NO_SET_INTF (device can't handle
5644 Set-Interface requests);
5645 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
5646 handle its Configuration or Interface
5648 e = USB_QUIRK_RESET (device can't be reset
5649 (e.g morph devices), don't use reset);
5650 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
5651 more interface descriptions than the
5652 bNumInterfaces count, and can't handle
5653 talking to these interfaces);
5654 g = USB_QUIRK_DELAY_INIT (device needs a pause
5655 during initialization, after we read
5656 the device descriptor);
5657 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
5658 high speed and super speed interrupt
5659 endpoints, the USB 2.0 and USB 3.0 spec
5660 require the interval in microframes (1
5661 microframe = 125 microseconds) to be
5662 calculated as interval = 2 ^
5664 Devices with this quirk report their
5665 bInterval as the result of this
5666 calculation instead of the exponent
5667 variable used in the calculation);
5668 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
5669 handle device_qualifier descriptor
5671 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
5672 generates spurious wakeup, ignore
5673 remote wakeup capability);
5674 k = USB_QUIRK_NO_LPM (device can't handle Link
5676 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
5677 (Device reports its bInterval as linear
5678 frames instead of the USB 2.0
5680 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
5681 to be disconnected before suspend to
5682 prevent spurious wakeup);
5683 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
5684 pause after every control message);
5685 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
5686 delay after resetting its port);
5687 Example: quirks=0781:5580:bk,0a5c:5834:gij
5690 [USBHID] The interval which mice are to be polled at.
5693 [USBHID] The interval which joysticks are to be polled at.
5696 [USBHID] The interval which keyboards are to be polled at.
5698 usb-storage.delay_use=
5699 [UMS] The delay in seconds before a new device is
5700 scanned for Logical Units (default 1).
5703 [UMS] A list of quirks entries to supplement or
5704 override the built-in unusual_devs list. List
5705 entries are separated by commas. Each entry has
5706 the form VID:PID:Flags where VID and PID are Vendor
5707 and Product ID values (4-digit hex numbers) and
5708 Flags is a set of characters, each corresponding
5709 to a common usb-storage quirk flag as follows:
5710 a = SANE_SENSE (collect more than 18 bytes
5711 of sense data, not on uas);
5712 b = BAD_SENSE (don't collect more than 18
5713 bytes of sense data, not on uas);
5714 c = FIX_CAPACITY (decrease the reported
5715 device capacity by one sector);
5716 d = NO_READ_DISC_INFO (don't use
5717 READ_DISC_INFO command, not on uas);
5718 e = NO_READ_CAPACITY_16 (don't use
5719 READ_CAPACITY_16 command);
5720 f = NO_REPORT_OPCODES (don't use report opcodes
5722 g = MAX_SECTORS_240 (don't transfer more than
5723 240 sectors at a time, uas only);
5724 h = CAPACITY_HEURISTICS (decrease the
5725 reported device capacity by one
5726 sector if the number is odd);
5727 i = IGNORE_DEVICE (don't bind to this
5729 j = NO_REPORT_LUNS (don't use report luns
5731 k = NO_SAME (do not use WRITE_SAME, uas only)
5732 l = NOT_LOCKABLE (don't try to lock and
5733 unlock ejectable media, not on uas);
5734 m = MAX_SECTORS_64 (don't transfer more
5735 than 64 sectors = 32 KB at a time,
5737 n = INITIAL_READ10 (force a retry of the
5738 initial READ(10) command, not on uas);
5739 o = CAPACITY_OK (accept the capacity
5740 reported by the device, not on uas);
5741 p = WRITE_CACHE (the device cache is ON
5742 by default, not on uas);
5743 r = IGNORE_RESIDUE (the device reports
5744 bogus residue values, not on uas);
5745 s = SINGLE_LUN (the device has only one
5747 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
5748 commands, uas only);
5749 u = IGNORE_UAS (don't bind to the uas driver);
5750 w = NO_WP_DETECT (don't test whether the
5751 medium is write-protected).
5752 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
5753 even if the device claims no cache,
5755 Example: quirks=0419:aaf5:rl,0421:0433:rc
5757 user_debug= [KNL,ARM]
5759 See arch/arm/Kconfig.debug help text.
5760 1 - undefined instruction events
5762 4 - invalid data aborts
5765 Example: user_debug=31
5768 [X86] Flags controlling user PTE allocations.
5770 nohigh = do not allocate PTE pages in
5771 HIGHMEM regardless of setting
5775 On X86_32, this is an alias for vdso32=. Otherwise:
5777 vdso=1: enable VDSO (the default)
5778 vdso=0: disable VDSO mapping
5780 vdso32= [X86] Control the 32-bit vDSO
5781 vdso32=1: enable 32-bit VDSO
5782 vdso32=0 or vdso32=2: disable 32-bit VDSO
5784 See the help text for CONFIG_COMPAT_VDSO for more
5785 details. If CONFIG_COMPAT_VDSO is set, the default is
5786 vdso32=0; otherwise, the default is vdso32=1.
5788 For compatibility with older kernels, vdso32=2 is an
5791 Try vdso32=0 if you encounter an error that says:
5792 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
5795 vector=percpu: enable percpu vector domain
5797 video= [FB] Frame buffer configuration
5798 See Documentation/fb/modedb.rst.
5800 video.brightness_switch_enabled= [0,1]
5801 If set to 1, on receiving an ACPI notify event
5802 generated by hotkey, video driver will adjust brightness
5803 level and then send out the event to user space through
5804 the allocated input device; If set to 0, video driver
5805 will only send out the event without touching backlight
5810 [VMMIO] Memory mapped virtio (platform) device.
5812 <size>@<baseaddr>:<irq>[:<id>]
5814 <size> := size (can use standard suffixes
5816 <baseaddr> := physical base address
5817 <irq> := interrupt number (as passed to
5819 <id> := (optional) platform device id
5821 virtio_mmio.device=1K@0x100b0000:48:7
5823 Can be used multiple times for multiple devices.
5825 vga= [BOOT,X86-32] Select a particular video mode
5826 See Documentation/x86/boot.rst and
5827 Documentation/admin-guide/svga.rst.
5828 Use vga=ask for menu.
5829 This is actually a boot loader parameter; the value is
5830 passed to the kernel using a special protocol.
5832 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
5833 May slow down system boot speed, especially when
5834 enabled on systems with a large amount of memory.
5835 All options are enabled by default, and this
5836 interface is meant to allow for selectively
5837 enabling or disabling specific virtual memory
5840 Available options are:
5841 P Enable page structure init time poisoning
5842 - Disable all of the above options
5844 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
5845 size of <nn>. This can be used to increase the
5846 minimum size (128MB on x86). It can also be used to
5847 decrease the size and leave more room for directly
5850 vmcp_cma=nn[MG] [KNL,S390]
5851 Sets the memory size reserved for contiguous memory
5852 allocations for the vmcp device driver.
5854 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
5857 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
5860 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
5864 Controls the behavior of vsyscalls (i.e. calls to
5865 fixed addresses of 0xffffffffff600x00 from legacy
5866 code). Most statically-linked binaries and older
5867 versions of glibc use these calls. Because these
5868 functions are at fixed addresses, they make nice
5869 targets for exploits that can control RIP.
5871 emulate [default] Vsyscalls turn into traps and are
5872 emulated reasonably safely. The vsyscall
5875 xonly Vsyscalls turn into traps and are
5876 emulated reasonably safely. The vsyscall
5877 page is not readable.
5879 none Vsyscalls don't work at all. This makes
5880 them quite hard to use for exploits but
5881 might break your system.
5883 vt.color= [VT] Default text color.
5884 Format: 0xYX, X = foreground, Y = background.
5885 Default: 0x07 = light gray on black.
5887 vt.cur_default= [VT] Default cursor shape.
5888 Format: 0xCCBBAA, where AA, BB, and CC are the same as
5889 the parameters of the <Esc>[?A;B;Cc escape sequence;
5890 see VGA-softcursor.txt. Default: 2 = underline.
5892 vt.default_blu= [VT]
5893 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
5894 Change the default blue palette of the console.
5895 This is a 16-member array composed of values
5898 vt.default_grn= [VT]
5899 Format: <green0>,<green1>,<green2>,...,<green15>
5900 Change the default green palette of the console.
5901 This is a 16-member array composed of values
5904 vt.default_red= [VT]
5905 Format: <red0>,<red1>,<red2>,...,<red15>
5906 Change the default red palette of the console.
5907 This is a 16-member array composed of values
5913 Set system-wide default UTF-8 mode for all tty's.
5914 Default is 1, i.e. UTF-8 mode is enabled for all
5915 newly opened terminals.
5917 vt.global_cursor_default=
5920 Set system-wide default for whether a cursor
5921 is shown on new VTs. Default is -1,
5922 i.e. cursors will be created by default unless
5923 overridden by individual drivers. 0 will hide
5924 cursors, 1 will display them.
5926 vt.italic= [VT] Default color for italic text; 0-15.
5929 vt.underline= [VT] Default color for underlined text; 0-15.
5932 watchdog timers [HW,WDT] For information on watchdog timers,
5933 see Documentation/watchdog/watchdog-parameters.rst
5934 or other driver-specific files in the
5935 Documentation/watchdog/ directory.
5939 Set the hard lockup detector stall duration
5940 threshold in seconds. The soft lockup detector
5941 threshold is set to twice the value. A value of 0
5942 disables both lockup detectors. Default is 10
5945 workqueue.watchdog_thresh=
5946 If CONFIG_WQ_WATCHDOG is configured, workqueue can
5947 warn stall conditions and dump internal state to
5948 help debugging. 0 disables workqueue stall
5949 detection; otherwise, it's the stall threshold
5950 duration in seconds. The default value is 30 and
5951 it can be updated at runtime by writing to the
5952 corresponding sysfs file.
5954 workqueue.disable_numa
5955 By default, all work items queued to unbound
5956 workqueues are affine to the NUMA nodes they're
5957 issued on, which results in better behavior in
5958 general. If NUMA affinity needs to be disabled for
5959 whatever reason, this option can be used. Note
5960 that this also can be controlled per-workqueue for
5961 workqueues visible under /sys/bus/workqueue/.
5963 workqueue.power_efficient
5964 Per-cpu workqueues are generally preferred because
5965 they show better performance thanks to cache
5966 locality; unfortunately, per-cpu workqueues tend to
5967 be more power hungry than unbound workqueues.
5969 Enabling this makes the per-cpu workqueues which
5970 were observed to contribute significantly to power
5971 consumption unbound, leading to measurably lower
5972 power usage at the cost of small performance
5975 The default value of this parameter is determined by
5976 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
5978 workqueue.debug_force_rr_cpu
5979 Workqueue used to implicitly guarantee that work
5980 items queued without explicit CPU specified are put
5981 on the local CPU. This guarantee is no longer true
5982 and while local CPU is still preferred work items
5983 may be put on foreign CPUs. This debug option
5984 forces round-robin CPU selection to flush out
5985 usages which depend on the now broken guarantee.
5986 When enabled, memory and cache locality will be
5989 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
5990 default x2apic cluster mode on platforms
5993 xen_512gb_limit [KNL,X86-64,XEN]
5994 Restricts the kernel running paravirtualized under Xen
5995 to use only up to 512 GB of RAM. The reason to do so is
5996 crash analysis tools and Xen tools for doing domain
5997 save/restore/migration must be enabled to handle larger
6000 xen_emul_unplug= [HW,X86,XEN]
6001 Unplug Xen emulated devices
6002 Format: [unplug0,][unplug1]
6003 ide-disks -- unplug primary master IDE devices
6004 aux-ide-disks -- unplug non-primary-master IDE devices
6005 nics -- unplug network devices
6006 all -- unplug all emulated devices (NICs and IDE disks)
6007 unnecessary -- unplugging emulated devices is
6008 unnecessary even if the host did not respond to
6010 never -- do not unplug even if version check succeeds
6012 xen_legacy_crash [X86,XEN]
6013 Crash from Xen panic notifier, without executing late
6014 panic() code such as dumping handler.
6016 xen_nopvspin [X86,XEN]
6017 Disables the qspinlock slowpath using Xen PV optimizations.
6018 This parameter is obsoleted by "nopvspin" parameter, which
6019 has equivalent effect for XEN platform.
6022 Disables the PV optimizations forcing the HVM guest to
6023 run as generic HVM guest with no PV drivers.
6024 This option is obsoleted by the "nopv" option, which
6025 has equivalent effect for XEN platform.
6027 xen_no_vector_callback
6028 [KNL,X86,XEN] Disable the vector callback for Xen
6029 event channel interrupts.
6031 xen_scrub_pages= [XEN]
6032 Boolean option to control scrubbing pages before giving them back
6033 to Xen, for use by other domains. Can be also changed at runtime
6034 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
6035 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
6037 xen_timer_slop= [X86-64,XEN]
6038 Set the timer slop (in nanoseconds) for the virtual Xen
6039 timers (default is 100000). This adjusts the minimum
6040 delta of virtualized Xen timers, where lower values
6041 improve timer resolution at the expense of processing
6042 more timer interrupts.
6044 xen.event_eoi_delay= [XEN]
6045 How long to delay EOI handling in case of event
6046 storms (jiffies). Default is 10.
6048 xen.event_loop_timeout= [XEN]
6049 After which time (jiffies) the event handling loop
6050 should start to delay EOI handling. Default is 2.
6052 xen.fifo_events= [XEN]
6053 Boolean parameter to disable using fifo event handling
6054 even if available. Normally fifo event handling is
6055 preferred over the 2-level event handling, as it is
6056 fairer and the number of possible event channels is
6057 much higher. Default is on (use fifo events).
6059 nopv= [X86,XEN,KVM,HYPER_V,VMWARE]
6060 Disables the PV optimizations forcing the guest to run
6061 as generic guest with no PV drivers. Currently support
6062 XEN HVM, KVM, HYPER_V and VMWARE guest.
6064 nopvspin [X86,XEN,KVM]
6065 Disables the qspinlock slow path using PV optimizations
6066 which allow the hypervisor to 'idle' the guest on lock
6069 xirc2ps_cs= [NET,PCMCIA]
6071 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
6074 By default on POWER9 and above, the kernel will
6075 natively use the XIVE interrupt controller. This option
6076 allows the fallback firmware mode to be used:
6078 off Fallback to firmware control of XIVE interrupt
6079 controller on both pseries and powernv
6080 platforms. Only useful on POWER9 and above.
6082 xhci-hcd.quirks [USB,KNL]
6083 A hex value specifying bitmask with supplemental xhci
6084 host controller quirks. Meaning of each bit can be
6085 consulted in header drivers/usb/host/xhci.h.
6088 Format: { early | on | rw | ro | off }
6089 Controls if xmon debugger is enabled. Default is off.
6090 Passing only "xmon" is equivalent to "xmon=early".
6091 early Call xmon as early as possible on boot; xmon
6092 debugger is called from setup_arch().
6093 on xmon debugger hooks will be installed so xmon
6094 is only called on a kernel crash. Default mode,
6095 i.e. either "ro" or "rw" mode, is controlled
6096 with CONFIG_XMON_DEFAULT_RO_MODE.
6097 rw xmon debugger hooks will be installed so xmon
6098 is called only on a kernel crash, mode is write,
6099 meaning SPR registers, memory and, other data
6100 can be written using xmon commands.
6101 ro same as "rw" option above but SPR registers,
6102 memory, and other data can't be written using
6104 off xmon is disabled.
projects / linux-2.6-microblaze.git / blob