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]
27 If set to vendor, prefer vendor specific driver
28 (e.g. thinkpad_acpi, sony_acpi, etc.) instead
29 of the ACPI video.ko driver.
31 acpi_force_32bit_fadt_addr
32 force FADT to use 32 bit addresses rather than the
33 64 bit X_* addresses. Some firmware have broken 64
34 bit addresses for force ACPI ignore these and use
35 the older legacy 32 bit addresses.
37 acpica_no_return_repair [HW, ACPI]
38 Disable AML predefined validation mechanism
39 This mechanism can repair the evaluation result to make
40 the return objects more ACPI specification compliant.
41 This option is useful for developers to identify the
42 root cause of an AML interpreter issue when the issue
43 has something to do with the repair mechanism.
45 acpi.debug_layer= [HW,ACPI,ACPI_DEBUG]
46 acpi.debug_level= [HW,ACPI,ACPI_DEBUG]
48 CONFIG_ACPI_DEBUG must be enabled to produce any ACPI
49 debug output. Bits in debug_layer correspond to a
50 _COMPONENT in an ACPI source file, e.g.,
51 #define _COMPONENT ACPI_PCI_COMPONENT
52 Bits in debug_level correspond to a level in
53 ACPI_DEBUG_PRINT statements, e.g.,
54 ACPI_DEBUG_PRINT((ACPI_DB_INFO, ...
55 The debug_level mask defaults to "info". See
56 Documentation/firmware-guide/acpi/debug.rst for more information about
57 debug layers and levels.
59 Enable processor driver info messages:
60 acpi.debug_layer=0x20000000
61 Enable PCI/PCI interrupt routing info messages:
62 acpi.debug_layer=0x400000
63 Enable AML "Debug" output, i.e., stores to the Debug
64 object while interpreting AML:
65 acpi.debug_layer=0xffffffff acpi.debug_level=0x2
66 Enable all messages related to ACPI hardware:
67 acpi.debug_layer=0x2 acpi.debug_level=0xffffffff
69 Some values produce so much output that the system is
70 unusable. The "log_buf_len" parameter may be useful
71 if you need to capture more output.
73 acpi_enforce_resources= [ACPI]
75 Check for resource conflicts between native drivers
76 and ACPI OperationRegions (SystemIO and SystemMemory
77 only). IO ports and memory declared in ACPI might be
78 used by the ACPI subsystem in arbitrary AML code and
79 can interfere with legacy drivers.
80 strict (default): access to resources claimed by ACPI
81 is denied; legacy drivers trying to access reserved
82 resources will fail to bind to device using them.
83 lax: access to resources claimed by ACPI is allowed;
84 legacy drivers trying to access reserved resources
85 will bind successfully but a warning message is logged.
86 no: ACPI OperationRegions are not marked as reserved,
87 no further checks are performed.
89 acpi_force_table_verification [HW,ACPI]
90 Enable table checksum verification during early stage.
91 By default, this is disabled due to x86 early mapping
94 acpi_irq_balance [HW,ACPI]
95 ACPI will balance active IRQs
98 acpi_irq_nobalance [HW,ACPI]
99 ACPI will not move active IRQs (default)
102 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA
103 Format: <irq>,<irq>...
105 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for
107 Format: <irq>,<irq>...
109 acpi_mask_gpe= [HW,ACPI]
110 Due to the existence of _Lxx/_Exx, some GPEs triggered
111 by unsupported hardware/firmware features can result in
112 GPE floodings that cannot be automatically disabled by
114 This facility can be used to prevent such uncontrolled
118 acpi_no_auto_serialize [HW,ACPI]
119 Disable auto-serialization of AML methods
120 AML control methods that contain the opcodes to create
121 named objects will be marked as "Serialized" by the
122 auto-serialization feature.
123 This feature is enabled by default.
124 This option allows to turn off the feature.
126 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump
129 acpi_no_static_ssdt [HW,ACPI]
130 Disable installation of static SSDTs at early boot time
131 By default, SSDTs contained in the RSDT/XSDT will be
132 installed automatically and they will appear under
133 /sys/firmware/acpi/tables.
134 This option turns off this feature.
135 Note that specifying this option does not affect
136 dynamic table installation which will install SSDT
137 tables to /sys/firmware/acpi/tables/dynamic.
139 acpi_no_watchdog [HW,ACPI,WDT]
140 Ignore the ACPI-based watchdog interface (WDAT) and let
141 a native driver control the watchdog device instead.
143 acpi_rsdp= [ACPI,EFI,KEXEC]
144 Pass the RSDP address to the kernel, mostly used
145 on machines running EFI runtime service to boot the
146 second kernel for kdump.
148 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS
149 Format: To spoof as Windows 98: ="Microsoft Windows"
151 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead
152 of 2 which is mandated by ACPI 6) as the supported ACPI
153 specification revision (when using this switch, it may
154 be necessary to carry out a cold reboot _twice_ in a
155 row to make it take effect on the platform firmware).
157 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings
158 acpi_osi="string1" # add string1
159 acpi_osi="!string2" # remove string2
160 acpi_osi=!* # remove all strings
161 acpi_osi=! # disable all built-in OS vendor
163 acpi_osi=!! # enable all built-in OS vendor
165 acpi_osi= # disable all strings
167 'acpi_osi=!' can be used in combination with single or
168 multiple 'acpi_osi="string1"' to support specific OS
169 vendor string(s). Note that such command can only
170 affect the default state of the OS vendor strings, thus
171 it cannot affect the default state of the feature group
172 strings and the current state of the OS vendor strings,
173 specifying it multiple times through kernel command line
174 is meaningless. This command is useful when one do not
175 care about the state of the feature group strings which
176 should be controlled by the OSPM.
178 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent
179 to 'acpi_osi="Windows 2000" acpi_osi=!', they all
180 can make '_OSI("Windows 2000")' TRUE.
182 'acpi_osi=' cannot be used in combination with other
183 'acpi_osi=' command lines, the _OSI method will not
184 exist in the ACPI namespace. NOTE that such command can
185 only affect the _OSI support state, thus specifying it
186 multiple times through kernel command line is also
189 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)'
192 'acpi_osi=!*' can be used in combination with single or
193 multiple 'acpi_osi="string1"' to support specific
194 string(s). Note that such command can affect the
195 current state of both the OS vendor strings and the
196 feature group strings, thus specifying it multiple times
197 through kernel command line is meaningful. But it may
198 still not able to affect the final state of a string if
199 there are quirks related to this string. This command
200 is useful when one want to control the state of the
201 feature group strings to debug BIOS issues related to
204 1. 'acpi_osi="Module Device" acpi_osi=!*' can make
205 '_OSI("Module Device")' FALSE.
206 2. 'acpi_osi=!* acpi_osi="Module Device"' can make
207 '_OSI("Module Device")' TRUE.
208 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is
210 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"'
212 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!',
213 they all will make '_OSI("Windows 2000")' TRUE.
216 Override the pmtimer bug detection: force the kernel
217 to assume that this machine's pmtimer latches its value
218 and always returns good values.
220 acpi_sci= [HW,ACPI] ACPI System Control Interrupt trigger mode
221 Format: { level | edge | high | low }
223 acpi_skip_timer_override [HW,ACPI]
224 Recognize and ignore IRQ0/pin2 Interrupt Override.
225 For broken nForce2 BIOS resulting in XT-PIC timer.
227 acpi_sleep= [HW,ACPI] Sleep options
228 Format: { s3_bios, s3_mode, s3_beep, s4_nohwsig,
229 old_ordering, nonvs, sci_force_enable, nobl }
230 See Documentation/power/video.rst for information on
232 s3_beep is for debugging; it makes the PC's speaker beep
233 as soon as the kernel's real-mode entry point is called.
234 s4_nohwsig prevents ACPI hardware signature from being
235 used during resume from hibernation.
236 old_ordering causes the ACPI 1.0 ordering of the _PTS
237 control method, with respect to putting devices into
238 low power states, to be enforced (the ACPI 2.0 ordering
239 of _PTS is used by default).
240 nonvs prevents the kernel from saving/restoring the
241 ACPI NVS memory during suspend/hibernation and resume.
242 sci_force_enable causes the kernel to set SCI_EN directly
243 on resume from S1/S3 (which is against the ACPI spec,
244 but some broken systems don't work without it).
245 nobl causes the internal blacklist of systems known to
246 behave incorrectly in some ways with respect to system
247 suspend and resume to be ignored (use wisely).
249 acpi_use_timer_override [HW,ACPI]
250 Use timer override. For some broken Nvidia NF5 boards
251 that require a timer override, but don't have HPET
253 add_efi_memmap [EFI; X86] Include EFI memory map in
254 kernel's map of available physical RAM.
257 { off | try_unsupported }
258 off: disable AGP support
259 try_unsupported: try to drive unsupported chipsets
260 (may crash computer or cause data corruption)
263 See Documentation/sound/alsa-configuration.rst
266 Allow the default userspace alignment fault handler
267 behaviour to be specified. Bit 0 enables warnings,
268 bit 1 enables fixups, and bit 2 sends a segfault.
270 align_va_addr= [X86-64]
271 Align virtual addresses by clearing slice [14:12] when
272 allocating a VMA at process creation time. This option
273 gives you up to 3% performance improvement on AMD F15h
274 machines (where it is enabled by default) for a
275 CPU-intensive style benchmark, and it can vary highly in
276 a microbenchmark depending on workload and compiler.
278 32: only for 32-bit processes
279 64: only for 64-bit processes
280 on: enable for both 32- and 64-bit processes
281 off: disable for both 32- and 64-bit processes
283 alloc_snapshot [FTRACE]
284 Allocate the ftrace snapshot buffer on boot up when the
285 main buffer is allocated. This is handy if debugging
286 and you need to use tracing_snapshot() on boot up, and
287 do not want to use tracing_snapshot_alloc() as it needs
288 to be done where GFP_KERNEL allocations are allowed.
290 amd_iommu= [HW,X86-64]
291 Pass parameters to the AMD IOMMU driver in the system.
293 fullflush - enable flushing of IO/TLB entries when
294 they are unmapped. Otherwise they are
295 flushed before they will be reused, which
297 off - do not initialize any AMD IOMMU found in
299 force_isolation - Force device isolation for all
300 devices. The IOMMU driver is not
301 allowed anymore to lift isolation
302 requirements as needed. This option
303 does not override iommu=pt
305 amd_iommu_dump= [HW,X86-64]
306 Enable AMD IOMMU driver option to dump the ACPI table
307 for AMD IOMMU. With this option enabled, AMD IOMMU
308 driver will print ACPI tables for AMD IOMMU during
309 IOMMU initialization.
311 amd_iommu_intr= [HW,X86-64]
312 Specifies one of the following AMD IOMMU interrupt
314 legacy - Use legacy interrupt remapping mode.
315 vapic - Use virtual APIC mode, which allows IOMMU
316 to inject interrupts directly into guest.
317 This mode requires kvm-amd.avic=1.
318 (Default when IOMMU HW support is present.)
320 amijoy.map= [HW,JOY] Amiga joystick support
321 Map of devices attached to JOY0DAT and JOY1DAT
323 See also Documentation/input/joydev/joystick.rst
325 analog.map= [HW,JOY] Analog joystick and gamepad support
326 Specifies type or capabilities of an analog joystick
327 connected to one of 16 gameports
328 Format: <type1>,<type2>,..<type16>
331 Power management functions (SPARCstation-4/5 + deriv.)
333 Disable APC CPU standby support. SPARCstation-Fox does
334 not play well with APC CPU idle - disable it if you have
335 APC and your system crashes randomly.
337 apic= [APIC,X86] Advanced Programmable Interrupt Controller
338 Change the output verbosity while booting
339 Format: { quiet (default) | verbose | debug }
340 Change the amount of debugging information output
341 when initialising the APIC and IO-APIC components.
342 For X86-32, this can also be used to specify an APIC
344 Format: apic=driver_name
345 Examples: apic=bigsmp
347 apic_extnmi= [APIC,X86] External NMI delivery setting
348 Format: { bsp (default) | all | none }
349 bsp: External NMI is delivered only to CPU 0
350 all: External NMIs are broadcast to all CPUs as a
352 none: External NMI is masked for all CPUs. This is
353 useful so that a dump capture kernel won't be
357 See Documentation/networking/ipv6.txt.
359 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
360 Limit apic dumping. The parameter defines the maximal
361 number of local apics being dumped. Also it is possible
362 to set it to "all" by meaning -- no limit here.
363 Format: { 1 (default) | 2 | ... | all }.
364 The parameter valid if only apic=debug or
365 apic=verbose is specified.
366 Example: apic=debug show_lapic=all
368 apm= [APM] Advanced Power Management
369 See header of arch/x86/kernel/apm_32.c.
371 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
372 Format: <io>,<irq>,<nodeID>
376 atarimouse= [HW,MOUSE] Atari Mouse
378 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
379 EzKey and similar keyboards
381 atkbd.reset= [HW] Reset keyboard during initialization
383 atkbd.set= [HW] Select keyboard code set
384 Format: <int> (2 = AT (default), 3 = PS/2)
386 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
389 atkbd.softraw= [HW] Choose between synthetic and real raw mode
390 Format: <bool> (0 = real, 1 = synthetic (default))
392 atkbd.softrepeat= [HW]
393 Use software keyboard repeat
395 audit= [KNL] Enable the audit sub-system
396 Format: { "0" | "1" | "off" | "on" }
397 0 | off - kernel audit is disabled and can not be
398 enabled until the next reboot
399 unset - kernel audit is initialized but disabled and
400 will be fully enabled by the userspace auditd.
401 1 | on - kernel audit is initialized and partially
402 enabled, storing at most audit_backlog_limit
403 messages in RAM until it is fully enabled by the
407 audit_backlog_limit= [KNL] Set the audit queue size limit.
408 Format: <int> (must be >=0)
411 bau= [X86_UV] Enable the BAU on SGI UV. The default
412 behavior is to disable the BAU (i.e. bau=0).
413 Format: { "0" | "1" }
416 unset - Disable the BAU.
418 baycom_epp= [HW,AX25]
421 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
423 See header of drivers/net/hamradio/baycom_par.c.
425 baycom_ser_fdx= [HW,AX25]
426 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
427 Format: <io>,<irq>,<mode>[,<baud>]
428 See header of drivers/net/hamradio/baycom_ser_fdx.c.
430 baycom_ser_hdx= [HW,AX25]
431 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
432 Format: <io>,<irq>,<mode>
433 See header of drivers/net/hamradio/baycom_ser_hdx.c.
435 blkdevparts= Manual partition parsing of block device(s) for
436 embedded devices based on command line input.
437 See Documentation/block/cmdline-partition.rst
439 boot_delay= Milliseconds to delay each printk during boot.
440 Values larger than 10 seconds (10000) are changed to
445 Extended command line options can be added to an initrd
446 and this will cause the kernel to look for it.
448 See Documentation/admin-guide/bootconfig.rst
451 Disable BERT OS support on buggy BIOSes.
453 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
454 bttv.radio= Most important insmod options are available as
456 bttv.pll= See Documentation/media/v4l-drivers/bttv.rst
459 bulk_remove=off [PPC] This parameter disables the use of the pSeries
460 firmware feature for flushing multiple hpte entries
463 c101= [NET] Moxa C101 synchronous serial card
465 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
466 Sometimes CPU hardware bugs make them report the cache
467 size incorrectly. The kernel will attempt work arounds
468 to fix known problems, but for some CPUs it is not
469 possible to determine what the correct size should be.
470 This option provides an override for these situations.
473 [NET] Specifies amount of time (in seconds) that
474 the kernel should wait for a network carrier. By default
475 it waits 120 seconds.
477 ca_keys= [KEYS] This parameter identifies a specific key(s) on
478 the system trusted keyring to be used for certificate
480 format: { id:<keyid> | builtin }
482 cca= [MIPS] Override the kernel pages' cache coherency
483 algorithm. Accepted values range from 0 to 7
484 inclusive. See arch/mips/include/asm/pgtable-bits.h
485 for platform specific values (SB1, Loongson3 and
488 ccw_timeout_log [S390]
489 See Documentation/s390/common_io.rst for details.
491 cgroup_disable= [KNL] Disable a particular controller
492 Format: {name of the controller(s) to disable}
493 The effects of cgroup_disable=foo are:
494 - foo isn't auto-mounted if you mount all cgroups in
496 - foo isn't visible as an individually mountable
498 {Currently only "memory" controller deal with this and
499 cut the overhead, others just disable the usage. So
500 only cgroup_disable=memory is actually worthy}
502 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
503 Format: { { controller | "all" | "named" }
504 [,{ controller | "all" | "named" }...] }
505 Like cgroup_disable, but only applies to cgroup v1;
506 the blacklisted controllers remain available in cgroup2.
507 "all" blacklists all controllers and "named" disables
508 named mounts. Specifying both "all" and "named" disables
511 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
513 nosocket -- Disable socket memory accounting.
514 nokmem -- Disable kernel memory accounting.
516 checkreqprot [SELINUX] Set initial checkreqprot flag value.
517 Format: { "0" | "1" }
518 See security/selinux/Kconfig help text.
519 0 -- check protection applied by kernel (includes
520 any implied execute protection).
521 1 -- check protection requested by application.
522 Default value is set via a kernel config option.
523 Value can be changed at runtime via
524 /sys/fs/selinux/checkreqprot.
527 See Documentation/s390/common_io.rst for details.
530 Prevents the clock framework from automatically gating
531 clocks that have not been explicitly enabled by a Linux
532 device driver but are enabled in hardware at reset or
533 by the bootloader/firmware. Note that this does not
534 force such clocks to be always-on nor does it reserve
535 those clocks in any way. This parameter is useful for
536 debug and development, but should not be needed on a
537 platform with proper driver support. For more
538 information, see Documentation/driver-api/clk.rst.
540 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
542 Forces specified clocksource (if available) to be used
543 when calculating gettimeofday(). If specified
544 clocksource is not available, it defaults to PIT.
545 Format: { pit | tsc | cyclone | pmtmr }
547 clocksource= Override the default clocksource
549 Override the default clocksource and use the clocksource
550 with the name specified.
551 Some clocksource names to choose from, depending on
553 [all] jiffies (this is the base, fallback clocksource)
555 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
556 pxa_timer,timer3,32k_counter,timer0_1
557 [X86-32] pit,hpet,tsc;
558 scx200_hrt on Geode; cyclone on IBM x440
566 clocksource.arm_arch_timer.evtstrm=
569 Enable/disable the eventstream feature of the ARM
570 architected timer so that code using WFE-based polling
571 loops can be debugged more effectively on production
574 clearcpuid=BITNUM [X86]
575 Disable CPUID feature X for the kernel. See
576 arch/x86/include/asm/cpufeatures.h for the valid bit
577 numbers. Note the Linux specific bits are not necessarily
578 stable over kernel options, but the vendor specific
580 Also note that user programs calling CPUID directly
581 or using the feature without checking anything
582 will still see it. This just prevents it from
583 being used by the kernel or shown in /proc/cpuinfo.
584 Also note the kernel might malfunction if you disable
587 cma=nn[MG]@[start[MG][-end[MG]]]
589 Sets the size of kernel global memory area for
590 contiguous memory allocations and optionally the
591 placement constraint by the physical address range of
592 memory allocations. A value of 0 disables CMA
593 altogether. For more information, see
594 include/linux/dma-contiguous.h
596 cmo_free_hint= [PPC] Format: { yes | no }
597 Specify whether pages are marked as being inactive
598 when they are freed. This is used in CMO environments
599 to determine OS memory pressure for page stealing by
603 coherent_pool=nn[KMG] [ARM,KNL]
604 Sets the size of memory pool for coherent, atomic dma
605 allocations, by default set to 256K.
607 com20020= [HW,NET] ARCnet - COM20020 chipset
609 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
611 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
615 ARCnet - COM90xx chipset (memory-mapped buffers)
616 Format: <io>[,<irq>[,<memstart>]]
618 condev= [HW,S390] console device
621 console= [KNL] Output console device and options.
623 tty<n> Use the virtual console device <n>.
627 Use the specified serial port. The options are of
628 the form "bbbbpnf", where "bbbb" is the baud rate,
629 "p" is parity ("n", "o", or "e"), "n" is number of
630 bits, and "f" is flow control ("r" for RTS or
631 omit it). Default is "9600n8".
633 See Documentation/admin-guide/serial-console.rst for more
635 Documentation/networking/netconsole.txt for an
638 uart[8250],io,<addr>[,options]
639 uart[8250],mmio,<addr>[,options]
640 uart[8250],mmio16,<addr>[,options]
641 uart[8250],mmio32,<addr>[,options]
642 uart[8250],0x<addr>[,options]
643 Start an early, polled-mode console on the 8250/16550
644 UART at the specified I/O port or MMIO address,
645 switching to the matching ttyS device later.
646 MMIO inter-register address stride is either 8-bit
647 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
648 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
649 to be equivalent to 'mmio'. 'options' are specified in
650 the same format described for ttyS above; if unspecified,
651 the h/w is not re-initialized.
653 hvc<n> Use the hypervisor console device <n>. This is for
654 both Xen and PowerPC hypervisors.
656 If the device connected to the port is not a TTY but a braille
657 device, prepend "brl," before the device type, for instance
659 For now, only VisioBraille is supported.
662 [KNL] Change console messages format
664 By default we print messages on consoles in
665 "[time stamp] text\n" format (time stamp may not be
666 printed, depending on CONFIG_PRINTK_TIME or
667 `printk_time' param).
669 Switch to syslog format: "<%u>[time stamp] text\n"
670 IOW, each message will have a facility and loglevel
671 prefix. The format is similar to one used by syslog()
672 syscall, or to executing "dmesg -S --raw" or to reading
675 consoleblank= [KNL] The console blank (screen saver) timeout in
676 seconds. A value of 0 disables the blank timer.
680 [KNL] Change the default value for
681 /proc/<pid>/coredump_filter.
682 See also Documentation/filesystems/proc.txt.
684 coresight_cpu_debug.enable
687 Enable/disable the CPU sampling based debugging.
688 0: default value, disable debugging
689 1: enable debugging at boot time
691 cpuidle.off=1 [CPU_IDLE]
692 disable the cpuidle sub-system
695 [CPU_IDLE] Name of the cpuidle governor to use.
697 cpufreq.off=1 [CPU_FREQ]
698 disable the cpufreq sub-system
701 [X86] Delay for N microsec between assert and de-assert
702 of APIC INIT to start processors. This delay occurs
703 on every CPU online, such as boot, and resume from suspend.
706 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
708 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
710 crashkernel=size[KMG][@offset[KMG]]
711 [KNL] Using kexec, Linux can switch to a 'crash kernel'
712 upon panic. This parameter reserves the physical
713 memory region [offset, offset + size] for that kernel
714 image. If '@offset' is omitted, then a suitable offset
715 is selected automatically.
716 [KNL, x86_64] select a region under 4G first, and
717 fall back to reserve region above 4G when '@offset'
718 hasn't been specified.
719 See Documentation/admin-guide/kdump/kdump.rst for further details.
721 crashkernel=range1:size1[,range2:size2,...][@offset]
722 [KNL] Same as above, but depends on the memory
723 in the running system. The syntax of range is
724 start-[end] where start and end are both
725 a memory unit (amount[KMG]). See also
726 Documentation/admin-guide/kdump/kdump.rst for an example.
728 crashkernel=size[KMG],high
729 [KNL, x86_64] range could be above 4G. Allow kernel
730 to allocate physical memory region from top, so could
731 be above 4G if system have more than 4G ram installed.
732 Otherwise memory region will be allocated below 4G, if
734 It will be ignored if crashkernel=X is specified.
735 crashkernel=size[KMG],low
736 [KNL, x86_64] range under 4G. When crashkernel=X,high
737 is passed, kernel could allocate physical memory region
738 above 4G, that cause second kernel crash on system
739 that require some amount of low memory, e.g. swiotlb
740 requires at least 64M+32K low memory, also enough extra
741 low memory is needed to make sure DMA buffers for 32-bit
742 devices won't run out. Kernel would try to allocate at
743 at least 256M below 4G automatically.
744 This one let user to specify own low range under 4G
745 for second kernel instead.
746 0: to disable low allocation.
747 It will be ignored when crashkernel=X,high is not used
748 or memory reserved is below 4G.
751 [KNL] Disable crypto self-tests
756 cs89x0_media= [HW,NET]
757 Format: { rj45 | aui | bnc }
760 See header of drivers/s390/block/dasd_devmap.c.
762 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
763 (one device per port)
764 Format: <port#>,<type>
765 See also Documentation/input/devices/joystick-parport.rst
767 ddebug_query= [KNL,DYNAMIC_DEBUG] Enable debug messages at early boot
769 Documentation/admin-guide/dynamic-debug-howto.rst for
770 details. Deprecated, see dyndbg.
772 debug [KNL] Enable kernel debugging (events log level).
775 [KNL] Enable printing [hashed] pointers early in the
776 boot sequence. If enabled, we use a weak hash instead
777 of siphash to hash pointers. Use this option if you are
778 seeing instances of '(___ptrval___)') and need to see a
779 value (hashed pointer) instead. Cryptographically
780 insecure, please do not use on production kernels.
783 [KNL] verbose self-tests
785 Print debugging info while doing the locking API
787 We default to 0 (no extra messages), setting it to
788 1 will print _a lot_ more information - normally
789 only useful to kernel developers.
791 debug_objects [KNL] Enable object debugging
794 [KNL] Disable object debugging
796 debug_guardpage_minorder=
797 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
798 parameter allows control of the order of pages that will
799 be intentionally kept free (and hence protected) by the
800 buddy allocator. Bigger value increase the probability
801 of catching random memory corruption, but reduce the
802 amount of memory for normal system use. The maximum
803 possible value is MAX_ORDER/2. Setting this parameter
804 to 1 or 2 should be enough to identify most random
805 memory corruption problems caused by bugs in kernel or
806 driver code when a CPU writes to (or reads from) a
807 random memory location. Note that there exists a class
808 of memory corruptions problems caused by buggy H/W or
809 F/W or by drivers badly programing DMA (basically when
810 memory is written at bus level and the CPU MMU is
811 bypassed) which are not detectable by
812 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
813 tracking down these problems.
816 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
817 enables the feature at boot time. By default, it is
818 disabled and the system will work mostly the same as a
819 kernel built without CONFIG_DEBUG_PAGEALLOC.
820 Note: to get most of debug_pagealloc error reports, it's
821 useful to also enable the page_owner functionality.
822 on: enable the feature
824 debugpat [X86] Enable PAT debugging
826 decnet.addr= [HW,NET]
827 Format: <area>[,<node>]
828 See also Documentation/networking/decnet.txt.
831 [same as hugepagesz=] The size of the default
832 HugeTLB page size. This is the size represented by
833 the legacy /proc/ hugepages APIs, used for SHM, and
834 default size when mounting hugetlbfs filesystems.
835 Defaults to the default architecture's huge page size
838 deferred_probe_timeout=
839 [KNL] Debugging option to set a timeout in seconds for
840 deferred probe to give up waiting on dependencies to
841 probe. Only specific dependencies (subsystems or
842 drivers) that have opted in will be ignored. A timeout of 0
843 will timeout at the end of initcalls. This option will also
844 dump out devices still on the deferred probe list after
848 Format: { on | off | def_only | inf_only | always }
849 on: s390 zlib hardware support for compression on
850 level 1 and decompression (default)
851 off: No s390 zlib hardware support
852 def_only: s390 zlib hardware support for deflate
853 only (compression on level 1)
854 inf_only: s390 zlib hardware support for inflate
856 always: Same as 'on' but ignores the selected compression
857 level always using hardware support (used for debugging)
860 Set number of hash buckets for dentry cache.
862 disable_1tb_segments [PPC]
863 Disables the use of 1TB hash page table segments. This
864 causes the kernel to fall back to 256MB segments which
865 can be useful when debugging issues that require an SLB
869 See Documentation/networking/ipv6.txt.
872 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
873 hardening is enabled for this boot. Hardened
874 usercopy checking is used to protect the kernel
875 from reading or writing beyond known memory
876 allocation boundaries as a proactive defense
877 against bounds-checking flaws in the kernel's
878 copy_to_user()/copy_from_user() interface.
879 on Perform hardened usercopy checks (default).
880 off Disable hardened usercopy checks.
883 Disable RADIX MMU mode on POWER9
886 Disable TLBIE instruction. Currently does not work
887 with KVM, with HASH MMU, or with coherent accelerators.
889 disable_cpu_apicid= [X86,APIC,SMP]
891 The number of initial APIC ID for the
892 corresponding CPU to be disabled at boot,
893 mostly used for the kdump 2nd kernel to
894 disable BSP to wake up multiple CPUs without
895 causing system reset or hang due to sending
898 perf_v4_pmi= [X86,INTEL]
900 Disable Intel PMU counter freezing feature.
901 The feature only exists starting from
902 Arch Perfmon v4 (Skylake and newer).
904 disable_ddw [PPC/PSERIES]
905 Disable Dynamic DMA Window support. Use this if
906 to workaround buggy firmware.
909 See Documentation/networking/ipv6.txt.
911 disable_mtrr_cleanup [X86]
912 The kernel tries to adjust MTRR layout from continuous
913 to discrete, to make X server driver able to add WB
914 entry later. This parameter disables that.
916 disable_mtrr_trim [X86, Intel and AMD only]
917 By default the kernel will trim any uncacheable
918 memory out of your available memory pool based on
919 MTRR settings. This parameter disables that behavior,
920 possibly causing your machine to run very slowly.
922 disable_timer_pin_1 [X86]
923 Disable PIN 1 of APIC timer
924 Can be useful to work around chipset bugs.
926 dis_ucode_ldr [X86] Disable the microcode loader.
928 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
929 this option disables the debugging code at boot.
931 dma_debug_entries=<number>
932 This option allows to tune the number of preallocated
933 entries for DMA-API debugging code. One entry is
934 required per DMA-API allocation. Use this if the
935 DMA-API debugging code disables itself because the
936 architectural default is too low.
938 dma_debug_driver=<driver_name>
939 With this option the DMA-API debugging driver
940 filter feature can be enabled at boot time. Just
941 pass the driver to filter for as the parameter.
942 The filter can be disabled or changed to another
943 driver later using sysfs.
945 driver_async_probe= [KNL]
946 List of driver names to be probed asynchronously.
947 Format: <driver_name1>,<driver_name2>...
949 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
950 Broken monitors, graphic adapters, KVMs and EDIDless
951 panels may send no or incorrect EDID data sets.
952 This parameter allows to specify an EDID data sets
953 in the /lib/firmware directory that are used instead.
954 Generic built-in EDID data sets are used, if one of
955 edid/1024x768.bin, edid/1280x1024.bin,
956 edid/1680x1050.bin, or edid/1920x1080.bin is given
957 and no file with the same name exists. Details and
958 instructions how to build your own EDID data are
959 available in Documentation/driver-api/edid.rst. An EDID
960 data set will only be used for a particular connector,
961 if its name and a colon are prepended to the EDID
962 name. Each connector may use a unique EDID data
963 set by separating the files with a comma. An EDID
964 data set with no connector name will be used for
965 any connectors not explicitly specified.
970 Format: {"off" | "known"}
971 Control how the dt_cpu_ftrs device-tree binding is
972 used for CPU feature discovery and setup (if it
974 off: Do not use it, fall back to legacy cpu table.
975 known: Do not pass through unknown features to guests
976 or userspace, only those that the kernel is aware of.
978 dump_apple_properties [X86]
979 Dump name and content of EFI device properties on
980 x86 Macs. Useful for driver authors to determine
981 what data is available or for reverse-engineering.
983 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
984 module.dyndbg[="val"]
985 Enable debug messages at boot time. See
986 Documentation/admin-guide/dynamic-debug-howto.rst
989 nompx [X86] Disables Intel Memory Protection Extensions.
990 See Documentation/x86/intel_mpx.rst for more
991 information about the feature.
993 nopku [X86] Disable Memory Protection Keys CPU feature found
996 module.async_probe [KNL]
997 Enable asynchronous probe on this module.
999 early_ioremap_debug [KNL]
1000 Enable debug messages in early_ioremap support. This
1001 is useful for tracking down temporary early mappings
1002 which are not unmapped.
1004 earlycon= [KNL] Output early console device and options.
1006 When used with no options, the early console is
1007 determined by stdout-path property in device tree's
1008 chosen node or the ACPI SPCR table if supported by
1011 cdns,<addr>[,options]
1012 Start an early, polled-mode console on a Cadence
1013 (xuartps) serial port at the specified address. Only
1014 supported option is baud rate. If baud rate is not
1015 specified, the serial port must already be setup and
1018 uart[8250],io,<addr>[,options]
1019 uart[8250],mmio,<addr>[,options]
1020 uart[8250],mmio32,<addr>[,options]
1021 uart[8250],mmio32be,<addr>[,options]
1022 uart[8250],0x<addr>[,options]
1023 Start an early, polled-mode console on the 8250/16550
1024 UART at the specified I/O port or MMIO address.
1025 MMIO inter-register address stride is either 8-bit
1026 (mmio) or 32-bit (mmio32 or mmio32be).
1027 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1028 to be equivalent to 'mmio'. 'options' are specified
1029 in the same format described for "console=ttyS<n>"; if
1030 unspecified, the h/w is not initialized.
1034 Start an early, polled-mode console on a pl011 serial
1035 port at the specified address. The pl011 serial port
1036 must already be setup and configured. Options are not
1037 yet supported. If 'mmio32' is specified, then only
1038 the driver will use only 32-bit accessors to read/write
1039 the device registers.
1042 Start an early, polled-mode console on a meson serial
1043 port at the specified address. The serial port must
1044 already be setup and configured. Options are not yet
1048 Start an early, polled-mode console on an msm serial
1049 port at the specified address. The serial port
1050 must already be setup and configured. Options are not
1053 msm_serial_dm,<addr>
1054 Start an early, polled-mode console on an msm serial
1055 dm port at the specified address. The serial port
1056 must already be setup and configured. Options are not
1060 Start an early, polled-mode console on a serial port
1061 of an Actions Semi SoC, such as S500 or S900, at the
1062 specified address. The serial port must already be
1063 setup and configured. Options are not yet supported.
1066 Start an early, polled-mode console on a serial port
1067 of an RDA Micro SoC, such as RDA8810PL, at the
1068 specified address. The serial port must already be
1069 setup and configured. Options are not yet supported.
1072 Use RISC-V SBI (Supervisor Binary Interface) for early
1075 smh Use ARM semihosting calls for early console.
1083 Use early console provided by serial driver available
1084 on Samsung SoCs, requires selecting proper type and
1085 a correct base address of the selected UART port. The
1086 serial port must already be setup and configured.
1087 Options are not yet supported.
1090 Start an early, polled-mode console on a lantiq serial
1091 (lqasc) port at the specified address. The serial port
1092 must already be setup and configured. Options are not
1097 Use early console provided by Freescale LP UART driver
1098 found on Freescale Vybrid and QorIQ LS1021A processors.
1099 A valid base address must be provided, and the serial
1100 port must already be setup and configured.
1104 Start an early, polled-mode, output-only console on the
1105 Freescale i.MX UART at the specified address. The UART
1106 must already be setup and configured.
1109 Start an early, polled-mode console on the
1110 Armada 3700 serial port at the specified
1111 address. The serial port must already be setup
1112 and configured. Options are not yet supported.
1115 Start an early, polled-mode console on a Qualcomm
1116 Generic Interface (GENI) based serial port at the
1117 specified address. The serial port must already be
1118 setup and configured. Options are not yet supported.
1121 Start an early, unaccelerated console on the EFI
1122 memory mapped framebuffer (if available). On cache
1123 coherent non-x86 systems that use system memory for
1124 the framebuffer, pass the 'ram' option so that it is
1125 mapped with the correct attributes.
1128 Use early console provided by Freescale LINFlexD UART
1129 serial driver for NXP S32V234 SoCs. A valid base
1130 address must be provided, and the serial port must
1131 already be setup and configured.
1133 earlyprintk= [X86,SH,ARM,M68k,S390]
1137 earlyprintk=serial[,ttySn[,baudrate]]
1138 earlyprintk=serial[,0x...[,baudrate]]
1139 earlyprintk=ttySn[,baudrate]
1140 earlyprintk=dbgp[debugController#]
1141 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1142 earlyprintk=xdbc[xhciController#]
1144 earlyprintk is useful when the kernel crashes before
1145 the normal console is initialized. It is not enabled by
1146 default because it has some cosmetic problems.
1148 Append ",keep" to not disable it when the real console
1151 Only one of vga, efi, serial, or usb debug port can
1154 Currently only ttyS0 and ttyS1 may be specified by
1155 name. Other I/O ports may be explicitly specified
1156 on some architectures (x86 and arm at least) by
1157 replacing ttySn with an I/O port address, like this:
1158 earlyprintk=serial,0x1008,115200
1159 You can find the port for a given device in
1160 /proc/tty/driver/serial:
1161 2: uart:ST16650V2 port:00001008 irq:18 ...
1163 Interaction with the standard serial driver is not
1166 The VGA and EFI output is eventually overwritten by
1169 The xen output can only be used by Xen PV guests.
1171 The sclp output can only be used on s390.
1173 The optional "force" to "pciserial" enables use of a
1174 PCI device even when its classcode is not of the
1177 edac_report= [HW,EDAC] Control how to report EDAC event
1178 Format: {"on" | "off" | "force"}
1179 on: enable EDAC to report H/W event. May be overridden
1180 by other higher priority error reporting module.
1181 off: disable H/W event reporting through EDAC.
1182 force: enforce the use of EDAC to report H/W event.
1185 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1188 This is designed to be used in conjunction with
1189 the boot argument: earlyprintk=vga
1192 Format: {"off" | "on" | "skip[mbr]"}
1195 Format: { "old_map", "nochunk", "noruntime", "debug",
1196 "nosoftreserve", "disable_early_pci_dma",
1197 "no_disable_early_pci_dma" }
1198 old_map [X86-64]: switch to the old ioremap-based EFI
1199 runtime services mapping. [Needs CONFIG_X86_UV=y]
1200 nochunk: disable reading files in "chunks" in the EFI
1201 boot stub, as chunking can cause problems with some
1202 firmware implementations.
1203 noruntime : disable EFI runtime services support
1204 debug: enable misc debug output
1205 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1206 attribute may cause the kernel to reserve the
1207 memory range for a memory mapping driver to
1208 claim. Specify efi=nosoftreserve to disable this
1209 reservation and treat the memory by its base type
1210 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1211 disable_early_pci_dma: Disable the busmaster bit on all
1212 PCI bridges while in the EFI boot stub
1213 no_disable_early_pci_dma: Leave the busmaster bit set
1214 on all PCI bridges while in the EFI boot stub
1216 efi_no_storage_paranoia [EFI; X86]
1217 Using this parameter you can use more than 50% of
1218 your efi variable storage. Use this parameter only if
1219 you are really sure that your UEFI does sane gc and
1220 fulfills the spec otherwise your board may brick.
1222 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1223 Add arbitrary attribute to specific memory range by
1224 updating original EFI memory map.
1225 Region of memory which aa attribute is added to is
1228 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1229 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1230 attribute is added to range 0x100000000-0x180000000 and
1231 0x10a0000000-0x1120000000.
1233 If efi_fake_mem=8G@9G:0x40000 is specified, the
1234 EFI_MEMORY_SP(0x40000) attribute is added to
1235 range 0x240000000-0x43fffffff.
1237 Using this parameter you can do debugging of EFI memmap
1238 related features. For example, you can do debugging of
1239 Address Range Mirroring feature even if your box
1240 doesn't support it, or mark specific memory as
1243 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1244 that is to be dynamically loaded by Linux. If there are
1245 multiple variables with the same name but with different
1246 vendor GUIDs, all of them will be loaded. See
1247 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1250 eisa_irq_edge= [PARISC,HW]
1251 See header of drivers/parisc/eisa.c.
1254 See comment before function elanfreq_setup() in
1255 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1257 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1258 Specifies physical address of start of kernel core
1259 image elf header and optionally the size. Generally
1260 kexec loader will pass this option to capture kernel.
1261 See Documentation/admin-guide/kdump/kdump.rst for details.
1263 enable_mtrr_cleanup [X86]
1264 The kernel tries to adjust MTRR layout from continuous
1265 to discrete, to make X server driver able to add WB
1266 entry later. This parameter enables that.
1268 enable_timer_pin_1 [X86]
1269 Enable PIN 1 of APIC timer
1270 Can be useful to work around chipset bugs
1271 (in particular on some ATI chipsets).
1272 The kernel tries to set a reasonable default.
1274 enforcing [SELINUX] Set initial enforcing status.
1276 See security/selinux/Kconfig help text.
1277 0 -- permissive (log only, no denials).
1278 1 -- enforcing (deny and log).
1280 Value can be changed at runtime via
1281 /sys/fs/selinux/enforce.
1284 Disable Error Record Serialization Table (ERST)
1287 ether= [HW,NET] Ethernet cards parameters
1288 This option is obsoleted by the "netdev=" option, which
1289 has equivalent usage. See its documentation for details.
1293 Permit 'security.evm' to be updated regardless of
1294 current integrity status.
1298 fail_make_request=[KNL]
1299 General fault injection mechanism.
1300 Format: <interval>,<probability>,<space>,<times>
1301 See also Documentation/fault-injection/.
1304 See Documentation/admin-guide/blockdev/floppy.rst.
1306 force_pal_cache_flush
1307 [IA-64] Avoid check_sal_cache_flush which may hang on
1308 buggy SAL_CACHE_FLUSH implementations. Using this
1309 parameter will force ia64_sal_cache_flush to call
1310 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1313 Forcefully enable Physical Address Extension (PAE).
1314 Many Pentium M systems disable PAE but may have a
1315 functionally usable PAE implementation.
1316 Warning: use of this parameter will taint the kernel
1317 and may cause unknown problems.
1320 [FTRACE] will set and start the specified tracer
1321 as early as possible in order to facilitate early
1324 ftrace_dump_on_oops[=orig_cpu]
1325 [FTRACE] will dump the trace buffers on oops.
1326 If no parameter is passed, ftrace will dump
1327 buffers of all CPUs, but if you pass orig_cpu, it will
1328 dump only the buffer of the CPU that triggered the
1331 ftrace_filter=[function-list]
1332 [FTRACE] Limit the functions traced by the function
1333 tracer at boot up. function-list is a comma separated
1334 list of functions. This list can be changed at run
1335 time by the set_ftrace_filter file in the debugfs
1338 ftrace_notrace=[function-list]
1339 [FTRACE] Do not trace the functions specified in
1340 function-list. This list can be changed at run time
1341 by the set_ftrace_notrace file in the debugfs
1344 ftrace_graph_filter=[function-list]
1345 [FTRACE] Limit the top level callers functions traced
1346 by the function graph tracer at boot up.
1347 function-list is a comma separated list of functions
1348 that can be changed at run time by the
1349 set_graph_function file in the debugfs tracing directory.
1351 ftrace_graph_notrace=[function-list]
1352 [FTRACE] Do not trace from the functions specified in
1353 function-list. This list is a comma separated list of
1354 functions that can be changed at run time by the
1355 set_graph_notrace file in the debugfs tracing directory.
1357 ftrace_graph_max_depth=<uint>
1358 [FTRACE] Used with the function graph tracer. This is
1359 the max depth it will trace into a function. This value
1360 can be changed at run time by the max_graph_depth file
1361 in the tracefs tracing directory. default: 0 (no limit)
1363 fw_devlink= [KNL] Create device links between consumer and supplier
1364 devices by scanning the firmware to infer the
1365 consumer/supplier relationships. This feature is
1366 especially useful when drivers are loaded as modules as
1367 it ensures proper ordering of tasks like device probing
1368 (suppliers first, then consumers), supplier boot state
1369 clean up (only after all consumers have probed),
1370 suspend/resume & runtime PM (consumers first, then
1372 Format: { off | permissive | on | rpm }
1373 off -- Don't create device links from firmware info.
1374 permissive -- Create device links from firmware info
1375 but use it only for ordering boot state clean
1376 up (sync_state() calls).
1377 on -- Create device links from firmware info and use it
1378 to enforce probe and suspend/resume ordering.
1379 rpm -- Like "on", but also use to order runtime PM.
1382 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1383 support via parallel port (up to 5 devices per port)
1384 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1385 See also Documentation/input/devices/joystick-parport.rst
1389 gart_fix_e820= [X86_64] disable the fix e820 for K8 GART
1393 gcov_persist= [GCOV] When non-zero (default), profiling data for
1394 kernel modules is saved and remains accessible via
1395 debugfs, even when the module is unloaded/reloaded.
1396 When zero, profiling data is discarded and associated
1397 debugfs files are removed at module unload time.
1399 goldfish [X86] Enable the goldfish android emulator platform.
1400 Don't use this when you are not running on the
1403 gpt [EFI] Forces disk with valid GPT signature but
1404 invalid Protective MBR to be treated as GPT. If the
1405 primary GPT is corrupted, it enables the backup/alternate
1406 GPT to be used instead.
1408 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1409 the "Enable 0" bit of the configuration register.
1412 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1413 the "Enable 0" bit of the configuration register.
1416 grcan.select= [HW] Select which physical interface to use.
1419 grcan.txsize= [HW] Sets the size of the tx buffer.
1420 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1422 grcan.rxsize= [HW] Sets the size of the rx buffer.
1423 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1426 gpio-mockup.gpio_mockup_ranges
1427 [HW] Sets the ranges of gpiochip of for this device.
1428 Format: <start1>,<end1>,<start2>,<end2>...
1430 hardlockup_all_cpu_backtrace=
1431 [KNL] Should the hard-lockup detector generate
1432 backtraces on all cpus.
1435 hashdist= [KNL,NUMA] Large hashes allocated during boot
1436 are distributed across NUMA nodes. Defaults on
1437 for 64-bit NUMA, off otherwise.
1438 Format: 0 | 1 (for off | on)
1440 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1442 hd= [EIDE] (E)IDE hard drive subsystem geometry
1443 Format: <cyl>,<head>,<sect>
1446 Disable Hardware Error Source Table (HEST) support;
1447 corresponding firmware-first mode error processing
1448 logic will be disabled.
1450 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1451 size of <nn>. This works even on boxes that have no
1452 highmem otherwise. This also works to reduce highmem
1453 size on bigger boxes.
1455 highres= [KNL] Enable/disable high resolution timer mode.
1456 Valid parameters: "on", "off"
1461 hpet= [X86-32,HPET] option to control HPET usage
1462 Format: { enable (default) | disable | force |
1464 disable: disable HPET and use PIT instead
1465 force: allow force enabled of undocumented chips (ICH4,
1467 verbose: show contents of HPET registers during setup
1469 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1470 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1472 hugepages= [HW,X86-32,IA-64] HugeTLB pages to allocate at boot.
1473 hugepagesz= [HW,IA-64,PPC,X86-64] The size of the HugeTLB pages.
1474 On x86-64 and powerpc, this option can be specified
1475 multiple times interleaved with hugepages= to reserve
1476 huge pages of different sizes. Valid pages sizes on
1477 x86-64 are 2M (when the CPU supports "pse") and 1G
1478 (when the CPU supports the "pdpe1gb" cpuinfo flag).
1481 [KNL] Should the hung task detector generate panics.
1484 A nonzero value instructs the kernel to panic when a
1485 hung task is detected. The default value is controlled
1486 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1487 option. The value selected by this boot parameter can
1488 be changed later by the kernel.hung_task_panic sysctl.
1490 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1491 terminal devices. Valid values: 0..8
1492 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1493 If specified, z/VM IUCV HVC accepts connections
1494 from listed z/VM user IDs only.
1496 hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations
1497 which allow the hypervisor to 'idle' the
1498 guest on lock contention.
1501 Do not unregister boot console at start. This is only
1502 useful for debugging when something happens in the window
1503 between unregistering the boot console and initializing
1506 i2c_bus= [HW] Override the default board specific I2C bus speed
1507 or register an additional I2C bus that is not
1508 registered from board initialization code.
1512 i8042.debug [HW] Toggle i8042 debug mode
1513 i8042.unmask_kbd_data
1514 [HW] Enable printing of interrupt data from the KBD port
1515 (disabled by default, and as a pre-condition
1516 requires that i8042.debug=1 be enabled)
1517 i8042.direct [HW] Put keyboard port into non-translated mode
1518 i8042.dumbkbd [HW] Pretend that controller can only read data from
1519 keyboard and cannot control its state
1520 (Don't attempt to blink the leds)
1521 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1522 i8042.nokbd [HW] Don't check/create keyboard port
1523 i8042.noloop [HW] Disable the AUX Loopback command while probing
1525 i8042.nomux [HW] Don't check presence of an active multiplexing
1527 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1529 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1530 i8042.reset [HW] Reset the controller during init, cleanup and
1531 suspend-to-ram transitions, only during s2r
1532 transitions, or never reset
1533 Format: { 1 | Y | y | 0 | N | n }
1534 1, Y, y: always reset controller
1535 0, N, n: don't ever reset controller
1536 Default: only on s2r transitions on x86; most other
1537 architectures force reset to be always executed
1538 i8042.unlock [HW] Unlock (ignore) the keylock
1539 i8042.kbdreset [HW] Reset device connected to KBD port
1543 i8k.ignore_dmi [HW] Continue probing hardware even if DMI data
1544 indicates that the driver is running on unsupported
1546 i8k.force [HW] Activate i8k driver even if SMM BIOS signature
1547 does not match list of supported models.
1549 [HW] Report power status in /proc/i8k
1550 (disabled by default)
1551 i8k.restricted [HW] Allow controlling fans only if SYS_ADMIN
1554 i915.invert_brightness=
1555 [DRM] Invert the sense of the variable that is used to
1556 set the brightness of the panel backlight. Normally a
1557 brightness value of 0 indicates backlight switched off,
1558 and the maximum of the brightness value sets the backlight
1559 to maximum brightness. If this parameter is set to 0
1560 (default) and the machine requires it, or this parameter
1561 is set to 1, a brightness value of 0 sets the backlight
1562 to maximum brightness, and the maximum of the brightness
1563 value switches the backlight off.
1564 -1 -- never invert brightness
1565 0 -- machine default
1566 1 -- force brightness inversion
1569 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1571 ide-core.nodma= [HW] (E)IDE subsystem
1572 Format: =0.0 to prevent dma on hda, =0.1 hdb =1.0 hdc
1573 .vlb_clock .pci_clock .noflush .nohpa .noprobe .nowerr
1574 .cdrom .chs .ignore_cable are additional options
1575 See Documentation/ide/ide.rst.
1577 ide-generic.probe-mask= [HW] (E)IDE subsystem
1579 Probe mask for legacy ISA IDE ports. Depending on
1580 platform up to 6 ports are supported, enabled by
1581 setting corresponding bits in the mask to 1. The
1582 default value is 0x0, which has a special meaning.
1583 On systems that have PCI, it triggers scanning the
1584 PCI bus for the first and the second port, which
1585 are then probed. On systems without PCI the value
1586 of 0x0 enables probing the two first ports as if it
1589 ide-pci-generic.all-generic-ide [HW] (E)IDE subsystem
1590 Claim all unknown PCI IDE storage controllers.
1593 Format: idle=poll, idle=halt, idle=nomwait
1594 Poll forces a polling idle loop that can slightly
1595 improve the performance of waking up a idle CPU, but
1596 will use a lot of power and make the system run hot.
1598 idle=halt: Halt is forced to be used for CPU idle.
1599 In such case C2/C3 won't be used again.
1600 idle=nomwait: Disable mwait for CPU C-states
1602 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1603 Format: { strict | legacy | 2008 | relaxed }
1606 Choose which programs will be accepted for execution
1607 based on the IEEE 754 NaN encoding(s) supported by
1608 the FPU and the NaN encoding requested with the value
1609 of an ELF file header flag individually set by each
1610 binary. Hardware implementations are permitted to
1611 support either or both of the legacy and the 2008 NaN
1614 Available settings are as follows:
1615 strict accept binaries that request a NaN encoding
1616 supported by the FPU
1617 legacy only accept legacy-NaN binaries, if supported
1619 2008 only accept 2008-NaN binaries, if supported
1621 relaxed accept any binaries regardless of whether
1622 supported by the FPU
1624 The FPU emulator is always able to support both NaN
1625 encodings, so if no FPU hardware is present or it has
1626 been disabled with 'nofpu', then the settings of
1627 'legacy' and '2008' strap the emulator accordingly,
1628 'relaxed' straps the emulator for both legacy-NaN and
1629 2008-NaN, whereas 'strict' enables legacy-NaN only on
1630 legacy processors and both NaN encodings on MIPS32 or
1633 The setting for ABS.fmt/NEG.fmt instruction execution
1634 mode generally follows that for the NaN encoding,
1635 except where unsupported by hardware.
1637 ignore_loglevel [KNL]
1638 Ignore loglevel setting - this will print /all/
1639 kernel messages to the console. Useful for debugging.
1640 We also add it as printk module parameter, so users
1641 could change it dynamically, usually by
1642 /sys/module/printk/parameters/ignore_loglevel.
1645 Ignore RLIMIT_DATA setting for data mappings,
1646 print warning at first misuse. Can be changed via
1647 /sys/module/kernel/parameters/ignore_rlimit_data.
1649 ihash_entries= [KNL]
1650 Set number of hash buckets for inode cache.
1652 ima_appraise= [IMA] appraise integrity measurements
1653 Format: { "off" | "enforce" | "fix" | "log" }
1656 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
1657 The builtin appraise policy appraises all files
1660 ima_canonical_fmt [IMA]
1661 Use the canonical format for the binary runtime
1662 measurements, instead of host native format.
1665 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1669 The list of supported hash algorithms is defined
1670 in crypto/hash_info.h.
1673 The builtin policies to load during IMA setup.
1674 Format: "tcb | appraise_tcb | secure_boot |
1677 The "tcb" policy measures all programs exec'd, files
1678 mmap'd for exec, and all files opened with the read
1679 mode bit set by either the effective uid (euid=0) or
1682 The "appraise_tcb" policy appraises the integrity of
1683 all files owned by root.
1685 The "secure_boot" policy appraises the integrity
1686 of files (eg. kexec kernel image, kernel modules,
1687 firmware, policy, etc) based on file signatures.
1689 The "fail_securely" policy forces file signature
1690 verification failure also on privileged mounted
1691 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
1694 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
1695 Load a policy which meets the needs of the Trusted
1696 Computing Base. This means IMA will measure all
1697 programs exec'd, files mmap'd for exec, and all files
1698 opened for read by uid=0.
1701 Select one of defined IMA measurements template formats.
1702 Formats: { "ima" | "ima-ng" | "ima-sig" }
1706 [IMA] Define a custom template format.
1707 Format: { "field1|...|fieldN" }
1709 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
1710 Format: <min_file_size>
1711 Set the minimal file size for using asynchronous hash.
1712 If left unspecified, ahash usage is disabled.
1714 ahash performance varies for different data sizes on
1715 different crypto accelerators. This option can be used
1716 to achieve the best performance for a particular HW.
1718 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
1720 Set hashing buffer size. Default: 4k.
1722 ahash performance varies for different chunk sizes on
1723 different crypto accelerators. This option can be used
1724 to achieve best performance for particular HW.
1728 Run specified binary instead of /sbin/init as init
1731 initcall_debug [KNL] Trace initcalls as they are executed. Useful
1732 for working out where the kernel is dying during
1735 initcall_blacklist= [KNL] Do not execute a comma-separated list of
1736 initcall functions. Useful for debugging built-in
1737 modules and initcalls.
1739 initrd= [BOOT] Specify the location of the initial ramdisk
1741 init_on_alloc= [MM] Fill newly allocated pages and heap objects with
1744 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
1746 init_on_free= [MM] Fill freed pages and heap objects with zeroes.
1748 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
1750 init_pkru= [x86] Specify the default memory protection keys rights
1751 register contents for all processes. 0x55555554 by
1752 default (disallow access to all but pkey 0). Can
1753 override in debugfs after boot.
1755 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
1758 int_pln_enable [x86] Enable power limit notification interrupt
1760 integrity_audit=[IMA]
1761 Format: { "0" | "1" }
1762 0 -- basic integrity auditing messages. (Default)
1763 1 -- additional integrity auditing messages.
1765 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
1767 Enable intel iommu driver.
1769 Disable intel iommu driver.
1770 igfx_off [Default Off]
1771 By default, gfx is mapped as normal device. If a gfx
1772 device has a dedicated DMAR unit, the DMAR unit is
1773 bypassed by not enabling DMAR with this option. In
1774 this case, gfx device will use physical address for
1777 With this option iommu will not optimize to look
1778 for io virtual address below 32-bit forcing dual
1779 address cycle on pci bus for cards supporting greater
1780 than 32-bit addressing. The default is to look
1781 for translation below 32-bit and if not available
1782 then look in the higher range.
1783 strict [Default Off]
1784 With this option on every unmap_single operation will
1785 result in a hardware IOTLB flush operation as opposed
1786 to batching them for performance.
1787 sp_off [Default Off]
1788 By default, super page will be supported if Intel IOMMU
1789 has the capability. With this option, super page will
1792 By default, scalable mode will be disabled even if the
1793 hardware advertises that it has support for the scalable
1794 mode translation. With this option set, scalable mode
1795 will be used on hardware which claims to support it.
1796 tboot_noforce [Default Off]
1797 Do not force the Intel IOMMU enabled under tboot.
1798 By default, tboot will force Intel IOMMU on, which
1799 could harm performance of some high-throughput
1800 devices like 40GBit network cards, even if identity
1802 Note that using this option lowers the security
1803 provided by tboot because it makes the system
1804 vulnerable to DMA attacks.
1805 nobounce [Default off]
1806 Disable bounce buffer for untrusted devices such as
1807 the Thunderbolt devices. This will treat the untrusted
1808 devices as the trusted ones, hence might expose security
1809 risks of DMA attacks.
1811 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
1812 0 disables intel_idle and fall back on acpi_idle.
1813 1 to 9 specify maximum depth of C-state.
1817 Do not enable intel_pstate as the default
1818 scaling driver for the supported processors
1820 Use intel_pstate as a scaling driver, but configure it
1821 to work with generic cpufreq governors (instead of
1822 enabling its internal governor). This mode cannot be
1823 used along with the hardware-managed P-states (HWP)
1826 Enable intel_pstate on systems that prohibit it by default
1827 in favor of acpi-cpufreq. Forcing the intel_pstate driver
1828 instead of acpi-cpufreq may disable platform features, such
1829 as thermal controls and power capping, that rely on ACPI
1830 P-States information being indicated to OSPM and therefore
1831 should be used with caution. This option does not work with
1832 processors that aren't supported by the intel_pstate driver
1833 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
1835 Do not enable hardware P state control (HWP)
1838 Only load intel_pstate on systems which support
1839 hardware P state control (HWP) if available.
1841 Enforce ACPI _PPC performance limits. If the Fixed ACPI
1842 Description Table, specifies preferred power management
1843 profile as "Enterprise Server" or "Performance Server",
1844 then this feature is turned on by default.
1846 Allow per-logical-CPU P-State performance control limits using
1847 cpufreq sysfs interface
1849 intremap= [X86-64, Intel-IOMMU]
1850 on enable Interrupt Remapping (default)
1851 off disable Interrupt Remapping
1852 nosid disable Source ID checking
1854 BIOS x2APIC opt-out request will be ignored
1855 nopost disable Interrupt Posting
1857 iomem= Disable strict checking of access to MMIO memory
1858 strict regions from userspace.
1873 nobypass [PPC/POWERNV]
1874 Disable IOMMU bypass, using IOMMU for PCI devices.
1876 iommu.strict= [ARM64] Configure TLB invalidation behaviour
1877 Format: { "0" | "1" }
1879 Request that DMA unmap operations use deferred
1880 invalidation of hardware TLBs, for increased
1881 throughput at the cost of reduced device isolation.
1882 Will fall back to strict mode if not supported by
1883 the relevant IOMMU driver.
1884 1 - Strict mode (default).
1885 DMA unmap operations invalidate IOMMU hardware TLBs
1889 [ARM64, X86] Configure DMA to bypass the IOMMU by default.
1890 Format: { "0" | "1" }
1891 0 - Use IOMMU translation for DMA.
1892 1 - Bypass the IOMMU for DMA.
1893 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
1895 io7= [HW] IO7 for Marvel based alpha systems
1896 See comment before marvel_specify_io7 in
1897 arch/alpha/kernel/core_marvel.c.
1899 io_delay= [X86] I/O delay method
1901 Standard port 0x80 based delay
1903 Alternate port 0xed based delay (needed on some systems)
1905 Simple two microseconds delay
1910 See Documentation/admin-guide/nfs/nfsroot.rst.
1912 ipcmni_extend [KNL] Extend the maximum number of unique System V
1913 IPC identifiers from 32,768 to 16,777,216.
1915 irqaffinity= [SMP] Set the default irq affinity mask
1916 The argument is a cpu list, as described above.
1918 irqchip.gicv2_force_probe=
1921 Force the kernel to look for the second 4kB page
1922 of a GICv2 controller even if the memory range
1923 exposed by the device tree is too small.
1925 irqchip.gicv3_nolpi=
1927 Force the kernel to ignore the availability of
1928 LPIs (and by consequence ITSs). Intended for system
1929 that use the kernel as a bootloader, and thus want
1930 to let secondary kernels in charge of setting up
1933 irqchip.gicv3_pseudo_nmi= [ARM64]
1934 Enables support for pseudo-NMIs in the kernel. This
1935 requires the kernel to be built with
1936 CONFIG_ARM64_PSEUDO_NMI.
1939 When an interrupt is not handled search all handlers
1940 for it. Intended to get systems with badly broken
1944 When an interrupt is not handled search all handlers
1945 for it. Also check all handlers each timer
1946 interrupt. Intended to get systems with badly broken
1950 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
1952 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
1953 [Deprecated - use cpusets instead]
1954 Format: [flag-list,]<cpu-list>
1956 Specify one or more CPUs to isolate from disturbances
1957 specified in the flag list (default: domain):
1960 Disable the tick when a single task runs.
1962 A residual 1Hz tick is offloaded to workqueues, which you
1963 need to affine to housekeeping through the global
1964 workqueue's affinity configured via the
1965 /sys/devices/virtual/workqueue/cpumask sysfs file, or
1966 by using the 'domain' flag described below.
1968 NOTE: by default the global workqueue runs on all CPUs,
1969 so to protect individual CPUs the 'cpumask' file has to
1970 be configured manually after bootup.
1973 Isolate from the general SMP balancing and scheduling
1974 algorithms. Note that performing domain isolation this way
1975 is irreversible: it's not possible to bring back a CPU to
1976 the domains once isolated through isolcpus. It's strongly
1977 advised to use cpusets instead to disable scheduler load
1978 balancing through the "cpuset.sched_load_balance" file.
1979 It offers a much more flexible interface where CPUs can
1980 move in and out of an isolated set anytime.
1982 You can move a process onto or off an "isolated" CPU via
1983 the CPU affinity syscalls or cpuset.
1984 <cpu number> begins at 0 and the maximum value is
1985 "number of CPUs in system - 1".
1989 Isolate from being targeted by managed interrupts
1990 which have an interrupt mask containing isolated
1991 CPUs. The affinity of managed interrupts is
1992 handled by the kernel and cannot be changed via
1993 the /proc/irq/* interfaces.
1995 This isolation is best effort and only effective
1996 if the automatically assigned interrupt mask of a
1997 device queue contains isolated and housekeeping
1998 CPUs. If housekeeping CPUs are online then such
1999 interrupts are directed to the housekeeping CPU
2000 so that IO submitted on the housekeeping CPU
2001 cannot disturb the isolated CPU.
2003 If a queue's affinity mask contains only isolated
2004 CPUs then this parameter has no effect on the
2005 interrupt routing decision, though interrupts are
2006 only delivered when tasks running on those
2007 isolated CPUs submit IO. IO submitted on
2008 housekeeping CPUs has no influence on those
2011 The format of <cpu-list> is described above.
2015 ivrs_ioapic [HW,X86_64]
2016 Provide an override to the IOAPIC-ID<->DEVICE-ID
2017 mapping provided in the IVRS ACPI table. For
2018 example, to map IOAPIC-ID decimal 10 to
2019 PCI device 00:14.0 write the parameter as:
2020 ivrs_ioapic[10]=00:14.0
2022 ivrs_hpet [HW,X86_64]
2023 Provide an override to the HPET-ID<->DEVICE-ID
2024 mapping provided in the IVRS ACPI table. For
2025 example, to map HPET-ID decimal 0 to
2026 PCI device 00:14.0 write the parameter as:
2027 ivrs_hpet[0]=00:14.0
2029 ivrs_acpihid [HW,X86_64]
2030 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2031 mapping provided in the IVRS ACPI table. For
2032 example, to map UART-HID:UID AMD0020:0 to
2033 PCI device 00:14.5 write the parameter as:
2034 ivrs_acpihid[00:14.5]=AMD0020:0
2036 js= [HW,JOY] Analog joystick
2037 See Documentation/input/joydev/joystick.rst.
2040 When CONFIG_RANDOMIZE_BASE is set, this disables
2041 kernel and module base offset ASLR (Address Space
2042 Layout Randomization).
2045 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2046 report on every invalid memory access. Without this
2047 parameter KASAN will print report only for the first
2052 kernelcore= [KNL,X86,IA-64,PPC]
2053 Format: nn[KMGTPE] | nn% | "mirror"
2054 This parameter specifies the amount of memory usable by
2055 the kernel for non-movable allocations. The requested
2056 amount is spread evenly throughout all nodes in the
2057 system as ZONE_NORMAL. The remaining memory is used for
2058 movable memory in its own zone, ZONE_MOVABLE. In the
2059 event, a node is too small to have both ZONE_NORMAL and
2060 ZONE_MOVABLE, kernelcore memory will take priority and
2061 other nodes will have a larger ZONE_MOVABLE.
2063 ZONE_MOVABLE is used for the allocation of pages that
2064 may be reclaimed or moved by the page migration
2065 subsystem. Note that allocations like PTEs-from-HighMem
2066 still use the HighMem zone if it exists, and the Normal
2067 zone if it does not.
2069 It is possible to specify the exact amount of memory in
2070 the form of "nn[KMGTPE]", a percentage of total system
2071 memory in the form of "nn%", or "mirror". If "mirror"
2072 option is specified, mirrored (reliable) memory is used
2073 for non-movable allocations and remaining memory is used
2074 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2075 are exclusive, so you cannot specify multiple forms.
2077 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
2078 Format: <Controller#>[,poll interval]
2079 The controller # is the number of the ehci usb debug
2080 port as it is probed via PCI. The poll interval is
2081 optional and is the number seconds in between
2082 each poll cycle to the debug port in case you need
2083 the functionality for interrupting the kernel with
2084 gdb or control-c on the dbgp connection. When
2085 not using this parameter you use sysrq-g to break into
2086 the kernel debugger.
2088 kgdboc= [KGDB,HW] kgdb over consoles.
2089 Requires a tty driver that supports console polling,
2090 or a supported polling keyboard driver (non-usb).
2091 Serial only format: <serial_device>[,baud]
2092 keyboard only format: kbd
2093 keyboard and serial format: kbd,<serial_device>[,baud]
2094 Optional Kernel mode setting:
2095 kms, kbd format: kms,kbd
2096 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2098 kgdbwait [KGDB] Stop kernel execution and enter the
2099 kernel debugger at the earliest opportunity.
2101 kmac= [MIPS] korina ethernet MAC address.
2102 Configure the RouterBoard 532 series on-chip
2103 Ethernet adapter MAC address.
2105 kmemleak= [KNL] Boot-time kmemleak enable/disable
2106 Valid arguments: on, off
2108 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2111 kprobe_event=[probe-list]
2112 [FTRACE] Add kprobe events and enable at boot time.
2113 The probe-list is a semicolon delimited list of probe
2114 definitions. Each definition is same as kprobe_events
2115 interface, but the parameters are comma delimited.
2116 For example, to add a kprobe event on vfs_read with
2117 arg1 and arg2, add to the command line;
2119 kprobe_event=p,vfs_read,$arg1,$arg2
2121 See also Documentation/trace/kprobetrace.rst "Kernel
2122 Boot Parameter" section.
2124 kpti= [ARM64] Control page table isolation of user
2125 and kernel address spaces.
2126 Default: enabled on cores which need mitigation.
2130 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2131 Default is 0 (don't ignore, but inject #GP)
2133 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2134 Default is false (don't support).
2136 kvm.mmu_audit= [KVM] This is a R/W parameter which allows audit
2141 [KVM] Controls the software workaround for the
2142 X86_BUG_ITLB_MULTIHIT bug.
2143 force : Always deploy workaround.
2144 off : Never deploy workaround.
2145 auto : Deploy workaround based on the presence of
2146 X86_BUG_ITLB_MULTIHIT.
2150 If the software workaround is enabled for the host,
2151 guests do need not to enable it for nested guests.
2153 kvm.nx_huge_pages_recovery_ratio=
2154 [KVM] Controls how many 4KiB pages are periodically zapped
2155 back to huge pages. 0 disables the recovery, otherwise if
2156 the value is N KVM will zap 1/Nth of the 4KiB pages every
2157 minute. The default is 60.
2159 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM.
2160 Default is 1 (enabled)
2162 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU)
2164 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode.
2166 kvm-arm.vgic_v3_group0_trap=
2167 [KVM,ARM] Trap guest accesses to GICv3 group-0
2170 kvm-arm.vgic_v3_group1_trap=
2171 [KVM,ARM] Trap guest accesses to GICv3 group-1
2174 kvm-arm.vgic_v3_common_trap=
2175 [KVM,ARM] Trap guest accesses to GICv3 common
2178 kvm-arm.vgic_v4_enable=
2179 [KVM,ARM] Allow use of GICv4 for direct injection of
2182 kvm-intel.ept= [KVM,Intel] Disable extended page tables
2183 (virtualized MMU) support on capable Intel chips.
2184 Default is 1 (enabled)
2186 kvm-intel.emulate_invalid_guest_state=
2187 [KVM,Intel] Enable emulation of invalid guest states
2188 Default is 0 (disabled)
2190 kvm-intel.flexpriority=
2191 [KVM,Intel] Disable FlexPriority feature (TPR shadow).
2192 Default is 1 (enabled)
2195 [KVM,Intel] Enable VMX nesting (nVMX).
2196 Default is 0 (disabled)
2198 kvm-intel.unrestricted_guest=
2199 [KVM,Intel] Disable unrestricted guest feature
2200 (virtualized real and unpaged mode) on capable
2201 Intel chips. Default is 1 (enabled)
2203 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2206 Valid arguments: never, cond, always
2208 always: L1D cache flush on every VMENTER.
2209 cond: Flush L1D on VMENTER only when the code between
2210 VMEXIT and VMENTER can leak host memory.
2211 never: Disables the mitigation
2213 Default is cond (do L1 cache flush in specific instances)
2215 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification
2216 feature (tagged TLBs) on capable Intel chips.
2217 Default is 1 (enabled)
2219 l1tf= [X86] Control mitigation of the L1TF vulnerability on
2222 The kernel PTE inversion protection is unconditionally
2223 enabled and cannot be disabled.
2226 Provides all available mitigations for the
2227 L1TF vulnerability. Disables SMT and
2228 enables all mitigations in the
2229 hypervisors, i.e. unconditional L1D flush.
2231 SMT control and L1D flush control via the
2232 sysfs interface is still possible after
2233 boot. Hypervisors will issue a warning
2234 when the first VM is started in a
2235 potentially insecure configuration,
2236 i.e. SMT enabled or L1D flush disabled.
2239 Same as 'full', but disables SMT and L1D
2240 flush runtime control. Implies the
2241 'nosmt=force' command line option.
2242 (i.e. sysfs control of SMT is disabled.)
2245 Leaves SMT enabled and enables the default
2246 hypervisor mitigation, i.e. conditional
2249 SMT control and L1D flush control via the
2250 sysfs interface is still possible after
2251 boot. Hypervisors will issue a warning
2252 when the first VM is started in a
2253 potentially insecure configuration,
2254 i.e. SMT enabled or L1D flush disabled.
2258 Disables SMT and enables the default
2259 hypervisor mitigation.
2261 SMT control and L1D flush control via the
2262 sysfs interface is still possible after
2263 boot. Hypervisors will issue a warning
2264 when the first VM is started in a
2265 potentially insecure configuration,
2266 i.e. SMT enabled or L1D flush disabled.
2269 Same as 'flush', but hypervisors will not
2270 warn when a VM is started in a potentially
2271 insecure configuration.
2274 Disables hypervisor mitigations and doesn't
2276 It also drops the swap size and available
2277 RAM limit restriction on both hypervisor and
2282 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2288 lapic [X86-32,APIC] Enable the local APIC even if BIOS
2291 lapic= [x86,APIC] "notscdeadline" Do not use TSC deadline
2292 value for LAPIC timer one-shot implementation. Default
2293 back to the programmable timer unit in the LAPIC.
2295 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
2298 libata.dma= [LIBATA] DMA control
2299 libata.dma=0 Disable all PATA and SATA DMA
2300 libata.dma=1 PATA and SATA Disk DMA only
2301 libata.dma=2 ATAPI (CDROM) DMA only
2302 libata.dma=4 Compact Flash DMA only
2303 Combinations also work, so libata.dma=3 enables DMA
2304 for disks and CDROMs, but not CFs.
2306 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2307 libata.ignore_hpa=0 keep BIOS limits (default)
2308 libata.ignore_hpa=1 ignore limits, using full disk
2310 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2314 libata.force= [LIBATA] Force configurations. The format is comma
2315 separated list of "[ID:]VAL" where ID is
2316 PORT[.DEVICE]. PORT and DEVICE are decimal numbers
2317 matching port, link or device. Basically, it matches
2318 the ATA ID string printed on console by libata. If
2319 the whole ID part is omitted, the last PORT and DEVICE
2320 values are used. If ID hasn't been specified yet, the
2321 configuration applies to all ports, links and devices.
2323 If only DEVICE is omitted, the parameter applies to
2324 the port and all links and devices behind it. DEVICE
2325 number of 0 either selects the first device or the
2326 first fan-out link behind PMP device. It does not
2327 select the host link. DEVICE number of 15 selects the
2328 host link and device attached to it.
2330 The VAL specifies the configuration to force. As long
2331 as there's no ambiguity shortcut notation is allowed.
2332 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2333 The following configurations can be forced.
2335 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2336 Any ID with matching PORT is used.
2338 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2340 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2341 udma[/][16,25,33,44,66,100,133] notation is also
2344 * [no]ncq: Turn on or off NCQ.
2346 * [no]ncqtrim: Turn off queued DSM TRIM.
2348 * nohrst, nosrst, norst: suppress hard, soft
2351 * rstonce: only attempt one reset during
2352 hot-unplug link recovery
2354 * dump_id: dump IDENTIFY data.
2356 * atapi_dmadir: Enable ATAPI DMADIR bridge support
2358 * disable: Disable this device.
2360 If there are multiple matching configurations changing
2361 the same attribute, the last one is used.
2363 memblock=debug [KNL] Enable memblock debug messages.
2365 load_ramdisk= [RAM] List of ramdisks to load from floppy
2366 See Documentation/admin-guide/blockdev/ramdisk.rst.
2368 lockd.nlm_grace_period=P [NFS] Assign grace period.
2371 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2374 lockd.nlm_timeout=T [NFS] Assign timeout value.
2377 lockd.nlm_udpport=M [NFS] Assign UDP port.
2380 lockdown= [SECURITY]
2381 { integrity | confidentiality }
2382 Enable the kernel lockdown feature. If set to
2383 integrity, kernel features that allow userland to
2384 modify the running kernel are disabled. If set to
2385 confidentiality, kernel features that allow userland
2386 to extract confidential information from the kernel
2389 locktorture.nreaders_stress= [KNL]
2390 Set the number of locking read-acquisition kthreads.
2391 Defaults to being automatically set based on the
2392 number of online CPUs.
2394 locktorture.nwriters_stress= [KNL]
2395 Set the number of locking write-acquisition kthreads.
2397 locktorture.onoff_holdoff= [KNL]
2398 Set time (s) after boot for CPU-hotplug testing.
2400 locktorture.onoff_interval= [KNL]
2401 Set time (s) between CPU-hotplug operations, or
2402 zero to disable CPU-hotplug testing.
2404 locktorture.shuffle_interval= [KNL]
2405 Set task-shuffle interval (jiffies). Shuffling
2406 tasks allows some CPUs to go into dyntick-idle
2407 mode during the locktorture test.
2409 locktorture.shutdown_secs= [KNL]
2410 Set time (s) after boot system shutdown. This
2411 is useful for hands-off automated testing.
2413 locktorture.stat_interval= [KNL]
2414 Time (s) between statistics printk()s.
2416 locktorture.stutter= [KNL]
2417 Time (s) to stutter testing, for example,
2418 specifying five seconds causes the test to run for
2419 five seconds, wait for five seconds, and so on.
2420 This tests the locking primitive's ability to
2421 transition abruptly to and from idle.
2423 locktorture.torture_type= [KNL]
2424 Specify the locking implementation to test.
2426 locktorture.verbose= [KNL]
2427 Enable additional printk() statements.
2429 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
2432 loglevel= All Kernel Messages with a loglevel smaller than the
2433 console loglevel will be printed to the console. It can
2434 also be changed with klogd or other programs. The
2435 loglevels are defined as follows:
2437 0 (KERN_EMERG) system is unusable
2438 1 (KERN_ALERT) action must be taken immediately
2439 2 (KERN_CRIT) critical conditions
2440 3 (KERN_ERR) error conditions
2441 4 (KERN_WARNING) warning conditions
2442 5 (KERN_NOTICE) normal but significant condition
2443 6 (KERN_INFO) informational
2444 7 (KERN_DEBUG) debug-level messages
2446 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
2447 in bytes. n must be a power of two and greater
2448 than the minimal size. The minimal size is defined
2449 by LOG_BUF_SHIFT kernel config parameter. There is
2450 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2451 that allows to increase the default size depending on
2452 the number of CPUs. See init/Kconfig for more details.
2454 logo.nologo [FB] Disables display of the built-in Linux logo.
2455 This may be used to provide more screen space for
2456 kernel log messages and is useful when debugging
2457 kernel boot problems.
2459 lp=0 [LP] Specify parallel ports to use, e.g,
2460 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2461 lp=reset first parallel port). 'lp=0' disables the
2462 lp=auto printer driver. 'lp=reset' (which can be
2463 specified in addition to the ports) causes
2464 attached printers to be reset. Using
2465 lp=port1,port2,... specifies the parallel ports
2466 to associate lp devices with, starting with
2467 lp0. A port specification may be 'none' to skip
2468 that lp device, or a parport name such as
2469 'parport0'. Specifying 'lp=auto' instead of a
2470 port specification list means that device IDs
2471 from each port should be examined, to see if
2472 an IEEE 1284-compliant printer is attached; if
2473 so, the driver will manage that printer.
2474 See also header of drivers/char/lp.c.
2477 Sets loops_per_jiffy to given constant, thus avoiding
2478 time-consuming boot-time autodetection (up to 250 ms per
2479 CPU). 0 enables autodetection (default). To determine
2480 the correct value for your kernel, boot with normal
2481 autodetection and see what value is printed. Note that
2482 on SMP systems the preset will be applied to all CPUs,
2483 which is likely to cause problems if your CPUs need
2484 significantly divergent settings. An incorrect value
2485 will cause delays in the kernel to be wrong, leading to
2486 unpredictable I/O errors and other breakage. Although
2487 unlikely, in the extreme case this might damage your
2491 Format: <io>,<irq>,<dma>
2493 lsm.debug [SECURITY] Enable LSM initialization debugging output.
2496 [SECURITY] Choose order of LSM initialization. This
2497 overrides CONFIG_LSM, and the "security=" parameter.
2499 machvec= [IA-64] Force the use of a particular machine-vector
2500 (machvec) in a generic kernel.
2501 Example: machvec=hpzx1
2503 machtype= [Loongson] Share the same kernel image file between different
2505 Example: machtype=lemote-yeeloong-2f-7inch
2507 max_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory greater
2508 than or equal to this physical address is ignored.
2510 maxcpus= [SMP] Maximum number of processors that an SMP kernel
2511 will bring up during bootup. maxcpus=n : n >= 0 limits
2512 the kernel to bring up 'n' processors. Surely after
2513 bootup you can bring up the other plugged cpu by executing
2514 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
2515 only takes effect during system bootup.
2516 While n=0 is a special case, it is equivalent to "nosmp",
2517 which also disables the IO APIC.
2519 max_loop= [LOOP] The number of loop block devices that get
2520 (loop.max_loop) unconditionally pre-created at init time. The default
2521 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
2522 of statically allocating a predefined number, loop
2523 devices can be requested on-demand with the
2524 /dev/loop-control interface.
2526 mce [X86-32] Machine Check Exception
2528 mce=option [X86-64] See Documentation/x86/x86_64/boot-options.rst
2530 md= [HW] RAID subsystems devices and level
2531 See Documentation/admin-guide/md.rst.
2534 Format: <first>,<last>
2535 Specifies range of consoles to be captured by the MDA.
2538 Control mitigation for the Micro-architectural Data
2539 Sampling (MDS) vulnerability.
2541 Certain CPUs are vulnerable to an exploit against CPU
2542 internal buffers which can forward information to a
2543 disclosure gadget under certain conditions.
2545 In vulnerable processors, the speculatively
2546 forwarded data can be used in a cache side channel
2547 attack, to access data to which the attacker does
2548 not have direct access.
2550 This parameter controls the MDS mitigation. The
2553 full - Enable MDS mitigation on vulnerable CPUs
2554 full,nosmt - Enable MDS mitigation and disable
2555 SMT on vulnerable CPUs
2556 off - Unconditionally disable MDS mitigation
2558 On TAA-affected machines, mds=off can be prevented by
2559 an active TAA mitigation as both vulnerabilities are
2560 mitigated with the same mechanism so in order to disable
2561 this mitigation, you need to specify tsx_async_abort=off
2564 Not specifying this option is equivalent to
2567 For details see: Documentation/admin-guide/hw-vuln/mds.rst
2569 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
2570 Amount of memory to be used when the kernel is not able
2571 to see the whole system memory or for test.
2572 [X86] Work as limiting max address. Use together
2573 with memmap= to avoid physical address space collisions.
2574 Without memmap= PCI devices could be placed at addresses
2575 belonging to unused RAM.
2577 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
2581 [KNL,SH] Allow user to override the default size for
2582 per-device physically contiguous DMA buffers.
2584 memhp_default_state=online/offline
2585 [KNL] Set the initial state for the memory hotplug
2586 onlining policy. If not specified, the default value is
2587 set according to the
2588 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
2590 See Documentation/admin-guide/mm/memory-hotplug.rst.
2592 memmap=exactmap [KNL,X86] Enable setting of an exact
2593 E820 memory map, as specified by the user.
2594 Such memmap=exactmap lines can be constructed based on
2595 BIOS output or other requirements. See the memmap=nn@ss
2598 memmap=nn[KMG]@ss[KMG]
2599 [KNL] Force usage of a specific region of memory.
2600 Region of memory to be used is from ss to ss+nn.
2601 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
2602 which limits max address to nn[KMG].
2603 Multiple different regions can be specified,
2606 memmap=100M@2G,100M#3G,1G!1024G
2608 memmap=nn[KMG]#ss[KMG]
2609 [KNL,ACPI] Mark specific memory as ACPI data.
2610 Region of memory to be marked is from ss to ss+nn.
2612 memmap=nn[KMG]$ss[KMG]
2613 [KNL,ACPI] Mark specific memory as reserved.
2614 Region of memory to be reserved is from ss to ss+nn.
2615 Example: Exclude memory from 0x18690000-0x1869ffff
2616 memmap=64K$0x18690000
2618 memmap=0x10000$0x18690000
2619 Some bootloaders may need an escape character before '$',
2620 like Grub2, otherwise '$' and the following number
2623 memmap=nn[KMG]!ss[KMG]
2624 [KNL,X86] Mark specific memory as protected.
2625 Region of memory to be used, from ss to ss+nn.
2626 The memory region may be marked as e820 type 12 (0xc)
2627 and is NVDIMM or ADR memory.
2629 memmap=<size>%<offset>-<oldtype>+<newtype>
2630 [KNL,ACPI] Convert memory within the specified region
2631 from <oldtype> to <newtype>. If "-<oldtype>" is left
2632 out, the whole region will be marked as <newtype>,
2633 even if previously unavailable. If "+<newtype>" is left
2634 out, matching memory will be removed. Types are
2635 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
2636 3 = ACPI, 12 = PRAM.
2638 memory_corruption_check=0/1 [X86]
2639 Some BIOSes seem to corrupt the first 64k of
2640 memory when doing things like suspend/resume.
2641 Setting this option will scan the memory
2642 looking for corruption. Enabling this will
2643 both detect corruption and prevent the kernel
2644 from using the memory being corrupted.
2645 However, its intended as a diagnostic tool; if
2646 repeatable BIOS-originated corruption always
2647 affects the same memory, you can use memmap=
2648 to prevent the kernel from using that memory.
2650 memory_corruption_check_size=size [X86]
2651 By default it checks for corruption in the low
2652 64k, making this memory unavailable for normal
2653 use. Use this parameter to scan for
2654 corruption in more or less memory.
2656 memory_corruption_check_period=seconds [X86]
2657 By default it checks for corruption every 60
2658 seconds. Use this parameter to check at some
2659 other rate. 0 disables periodic checking.
2661 memtest= [KNL,X86,ARM,PPC] Enable memtest
2663 default : 0 <disable>
2664 Specifies the number of memtest passes to be
2665 performed. Each pass selects another test
2666 pattern from a given set of patterns. Memtest
2667 fills the memory with this pattern, validates
2668 memory contents and reserves bad memory
2669 regions that are detected.
2671 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
2672 Valid arguments: on, off
2673 Default (depends on kernel configuration option):
2674 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
2675 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
2676 mem_encrypt=on: Activate SME
2677 mem_encrypt=off: Do not activate SME
2679 Refer to Documentation/virt/kvm/amd-memory-encryption.rst
2680 for details on when memory encryption can be activated.
2682 mem_sleep_default= [SUSPEND] Default system suspend mode:
2683 s2idle - Suspend-To-Idle
2684 shallow - Power-On Suspend or equivalent (if supported)
2685 deep - Suspend-To-RAM or equivalent (if supported)
2686 See Documentation/admin-guide/pm/sleep-states.rst.
2688 meye.*= [HW] Set MotionEye Camera parameters
2689 See Documentation/media/v4l-drivers/meye.rst.
2691 mfgpt_irq= [IA-32] Specify the IRQ to use for the
2692 Multi-Function General Purpose Timers on AMD Geode
2695 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
2696 the BIOS has incorrectly applied a workaround. TinyBIOS
2697 version 0.98 is known to be affected, 0.99 fixes the
2698 problem by letting the user disable the workaround.
2702 min_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory below this
2703 physical address is ignored.
2705 mini2440= [ARM,HW,KNL]
2706 Format:[0..2][b][c][t]
2708 MINI2440 configuration specification:
2709 0 - The attached screen is the 3.5" TFT
2710 1 - The attached screen is the 7" TFT
2711 2 - The VGA Shield is attached (1024x768)
2712 Leaving out the screen size parameter will not load
2713 the TFT driver, and the framebuffer will be left
2715 b - Enable backlight. The TFT backlight pin will be
2716 linked to the kernel VESA blanking code and a GPIO
2717 LED. This parameter is not necessary when using the
2719 c - Enable the s3c camera interface.
2720 t - Reserved for enabling touchscreen support. The
2721 touchscreen support is not enabled in the mainstream
2722 kernel as of 2.6.30, a preliminary port can be found
2723 in the "bleeding edge" mini2440 support kernel at
2724 http://repo.or.cz/w/linux-2.6/mini2440.git
2727 [X86,PPC,S390,ARM64] Control optional mitigations for
2728 CPU vulnerabilities. This is a set of curated,
2729 arch-independent options, each of which is an
2730 aggregation of existing arch-specific options.
2733 Disable all optional CPU mitigations. This
2734 improves system performance, but it may also
2735 expose users to several CPU vulnerabilities.
2736 Equivalent to: nopti [X86,PPC]
2738 nospectre_v1 [X86,PPC]
2740 nospectre_v2 [X86,PPC,S390,ARM64]
2741 spectre_v2_user=off [X86]
2742 spec_store_bypass_disable=off [X86,PPC]
2743 ssbd=force-off [ARM64]
2746 tsx_async_abort=off [X86]
2747 kvm.nx_huge_pages=off [X86]
2750 This does not have any effect on
2751 kvm.nx_huge_pages when
2752 kvm.nx_huge_pages=force.
2755 Mitigate all CPU vulnerabilities, but leave SMT
2756 enabled, even if it's vulnerable. This is for
2757 users who don't want to be surprised by SMT
2758 getting disabled across kernel upgrades, or who
2759 have other ways of avoiding SMT-based attacks.
2760 Equivalent to: (default behavior)
2763 Mitigate all CPU vulnerabilities, disabling SMT
2764 if needed. This is for users who always want to
2765 be fully mitigated, even if it means losing SMT.
2766 Equivalent to: l1tf=flush,nosmt [X86]
2767 mds=full,nosmt [X86]
2768 tsx_async_abort=full,nosmt [X86]
2771 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
2772 parameter allows control of the logging verbosity for
2773 the additional memory initialisation checks. A value
2774 of 0 disables mminit logging and a level of 4 will
2775 log everything. Information is printed at KERN_DEBUG
2776 so loglevel=8 may also need to be specified.
2779 [KNL] When CONFIG_MODULE_SIG is set, this means that
2780 modules without (valid) signatures will fail to load.
2781 Note that if CONFIG_MODULE_SIG_FORCE is set, that
2782 is always true, so this option does nothing.
2784 module_blacklist= [KNL] Do not load a comma-separated list of
2785 modules. Useful for debugging problem modules.
2788 [MOUSE] Maximum time between finger touching and
2789 leaving touchpad surface for touch to be considered
2790 a tap and be reported as a left button click (for
2791 touchpads working in absolute mode only).
2793 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
2794 reporting absolute coordinates, such as tablets
2795 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
2796 reporting absolute coordinates, such as tablets
2798 movablecore= [KNL,X86,IA-64,PPC]
2799 Format: nn[KMGTPE] | nn%
2800 This parameter is the complement to kernelcore=, it
2801 specifies the amount of memory used for migratable
2802 allocations. If both kernelcore and movablecore is
2803 specified, then kernelcore will be at *least* the
2804 specified value but may be more. If movablecore on its
2805 own is specified, the administrator must be careful
2806 that the amount of memory usable for all allocations
2809 movable_node [KNL] Boot-time switch to make hotplugable memory
2810 NUMA nodes to be movable. This means that the memory
2811 of such nodes will be usable only for movable
2812 allocations which rules out almost all kernel
2813 allocations. Use with caution!
2815 MTD_Partition= [MTD]
2816 Format: <name>,<region-number>,<size>,<offset>
2818 MTD_Region= [MTD] Format:
2819 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
2822 See drivers/mtd/parsers/cmdlinepart.c
2824 multitce=off [PPC] This parameter disables the use of the pSeries
2825 firmware feature for updating multiple TCE entries
2828 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
2830 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
2832 boundary - index of last SLC block on Flex-OneNAND.
2833 The remaining blocks are configured as MLC blocks.
2834 lock - Configure if Flex-OneNAND boundary should be locked.
2835 Once locked, the boundary cannot be changed.
2836 1 indicates lock status, 0 indicates unlock status.
2839 ARM/S3C2412 JIVE boot control
2841 See arch/arm/mach-s3c2412/mach-jive.c
2843 mtouchusb.raw_coordinates=
2844 [HW] Make the MicroTouch USB driver use raw coordinates
2845 ('y', default) or cooked coordinates ('n')
2847 mtrr_chunk_size=nn[KMG] [X86]
2848 used for mtrr cleanup. It is largest continuous chunk
2849 that could hold holes aka. UC entries.
2851 mtrr_gran_size=nn[KMG] [X86]
2852 Used for mtrr cleanup. It is granularity of mtrr block.
2854 Large value could prevent small alignment from
2857 mtrr_spare_reg_nr=n [X86]
2859 Range: 0,7 : spare reg number
2861 Used for mtrr cleanup. It is spare mtrr entries number.
2862 Set to 2 or more if your graphical card needs more.
2864 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
2866 netdev= [NET] Network devices parameters
2867 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
2868 Note that mem_start is often overloaded to mean
2869 something different and driver-specific.
2870 This usage is only documented in each driver source
2874 [NETFILTER] Enable connection tracking flow accounting
2875 0 to disable accounting
2876 1 to enable accounting
2879 nfsaddrs= [NFS] Deprecated. Use ip= instead.
2880 See Documentation/admin-guide/nfs/nfsroot.rst.
2882 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
2883 See Documentation/admin-guide/nfs/nfsroot.rst.
2885 nfsrootdebug [NFS] enable nfsroot debugging messages.
2886 See Documentation/admin-guide/nfs/nfsroot.rst.
2888 nfs.callback_nr_threads=
2889 [NFSv4] set the total number of threads that the
2890 NFS client will assign to service NFSv4 callback
2893 nfs.callback_tcpport=
2894 [NFS] set the TCP port on which the NFSv4 callback
2895 channel should listen.
2898 [NFS] sets the pathname to the program which is used
2899 to update the NFS client cache entries.
2901 nfs.cache_getent_timeout=
2902 [NFS] sets the timeout after which an attempt to
2903 update a cache entry is deemed to have failed.
2905 nfs.idmap_cache_timeout=
2906 [NFS] set the maximum lifetime for idmapper cache
2910 [NFS] enable 64-bit inode numbers.
2911 If zero, the NFS client will fake up a 32-bit inode
2912 number for the readdir() and stat() syscalls instead
2913 of returning the full 64-bit number.
2914 The default is to return 64-bit inode numbers.
2916 nfs.max_session_cb_slots=
2917 [NFSv4.1] Sets the maximum number of session
2918 slots the client will assign to the callback
2919 channel. This determines the maximum number of
2920 callbacks the client will process in parallel for
2921 a particular server.
2923 nfs.max_session_slots=
2924 [NFSv4.1] Sets the maximum number of session slots
2925 the client will attempt to negotiate with the server.
2926 This limits the number of simultaneous RPC requests
2927 that the client can send to the NFSv4.1 server.
2928 Note that there is little point in setting this
2929 value higher than the max_tcp_slot_table_limit.
2931 nfs.nfs4_disable_idmapping=
2932 [NFSv4] When set to the default of '1', this option
2933 ensures that both the RPC level authentication
2934 scheme and the NFS level operations agree to use
2935 numeric uids/gids if the mount is using the
2936 'sec=sys' security flavour. In effect it is
2937 disabling idmapping, which can make migration from
2938 legacy NFSv2/v3 systems to NFSv4 easier.
2939 Servers that do not support this mode of operation
2940 will be autodetected by the client, and it will fall
2941 back to using the idmapper.
2942 To turn off this behaviour, set the value to '0'.
2944 [NFS4] Specify an additional fixed unique ident-
2945 ification string that NFSv4 clients can insert into
2946 their nfs_client_id4 string. This is typically a
2947 UUID that is generated at system install time.
2949 nfs.send_implementation_id =
2950 [NFSv4.1] Send client implementation identification
2951 information in exchange_id requests.
2952 If zero, no implementation identification information
2954 The default is to send the implementation identification
2957 nfs.recover_lost_locks =
2958 [NFSv4] Attempt to recover locks that were lost due
2959 to a lease timeout on the server. Please note that
2960 doing this risks data corruption, since there are
2961 no guarantees that the file will remain unchanged
2962 after the locks are lost.
2963 If you want to enable the kernel legacy behaviour of
2964 attempting to recover these locks, then set this
2966 The default parameter value of '0' causes the kernel
2967 not to attempt recovery of lost locks.
2969 nfs4.layoutstats_timer =
2970 [NFSv4.2] Change the rate at which the kernel sends
2971 layoutstats to the pNFS metadata server.
2973 Setting this to value to 0 causes the kernel to use
2974 whatever value is the default set by the layout
2975 driver. A non-zero value sets the minimum interval
2976 in seconds between layoutstats transmissions.
2978 nfsd.nfs4_disable_idmapping=
2979 [NFSv4] When set to the default of '1', the NFSv4
2980 server will return only numeric uids and gids to
2981 clients using auth_sys, and will accept numeric uids
2982 and gids from such clients. This is intended to ease
2983 migration from NFSv2/v3.
2985 nmi_debug= [KNL,SH] Specify one or more actions to take
2986 when a NMI is triggered.
2987 Format: [state][,regs][,debounce][,die]
2989 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
2990 Format: [panic,][nopanic,][num]
2992 0 - turn hardlockup detector in nmi_watchdog off
2993 1 - turn hardlockup detector in nmi_watchdog on
2994 When panic is specified, panic when an NMI watchdog
2995 timeout occurs (or 'nopanic' to not panic on an NMI
2996 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
2997 To disable both hard and soft lockup detectors,
2998 please see 'nowatchdog'.
2999 This is useful when you use a panic=... timeout and
3000 need the box quickly up again.
3002 These settings can be accessed at runtime via
3003 the nmi_watchdog and hardlockup_panic sysctls.
3005 netpoll.carrier_timeout=
3006 [NET] Specifies amount of time (in seconds) that
3007 netpoll should wait for a carrier. By default netpoll
3010 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3011 emulation library even if a 387 maths coprocessor
3014 no5lvl [X86-64] Disable 5-level paging mode. Forces
3015 kernel to use 4-level paging instead.
3018 [HW] Never suspend the console
3019 Disable suspending of consoles during suspend and
3020 hibernate operations. Once disabled, debugging
3021 messages can reach various consoles while the rest
3022 of the system is being put to sleep (ie, while
3023 debugging driver suspend/resume hooks). This may
3024 not work reliably with all consoles, but is known
3025 to work with serial and VGA consoles.
3026 To facilitate more flexible debugging, we also add
3027 console_suspend, a printk module parameter to control
3028 it. Users could use console_suspend (usually
3029 /sys/module/printk/parameters/console_suspend) to
3030 turn on/off it dynamically.
3032 novmcoredd [KNL,KDUMP]
3033 Disable device dump. Device dump allows drivers to
3034 append dump data to vmcore so you can collect driver
3035 specified debug info. Drivers can append the data
3036 without any limit and this data is stored in memory,
3037 so this may cause significant memory stress. Disabling
3038 device dump can help save memory but the driver debug
3039 data will be no longer available. This parameter
3040 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
3043 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
3044 caches in the slab allocator. Saves per-node memory,
3045 but will impact performance.
3049 noaltinstr [S390] Disables alternative instructions patching
3050 (CPU alternatives feature).
3052 noapic [SMP,APIC] Tells the kernel to not make use of any
3053 IOAPICs that may be present in the system.
3055 noautogroup Disable scheduler automatic task group creation.
3057 nobats [PPC] Do not use BATs for mapping kernel lowmem
3058 on "Classic" PPC cores.
3062 noclflush [BUGS=X86] Don't use the CLFLUSH instruction
3064 nodelayacct [KNL] Disable per-task delay accounting
3066 nodsp [SH] Disable hardware DSP at boot time.
3068 noefi Disable EFI runtime services support.
3073 On X86-32 available only on PAE configured kernels.
3074 noexec=on: enable non-executable mappings (default)
3075 noexec=off: disable non-executable mappings
3078 Disable SMAP (Supervisor Mode Access Prevention)
3079 even if it is supported by processor.
3082 Disable SMEP (Supervisor Mode Execution Prevention)
3083 even if it is supported by processor.
3086 This affects only 32-bit executables.
3087 noexec32=on: enable non-executable mappings (default)
3088 read doesn't imply executable mappings
3089 noexec32=off: disable non-executable mappings
3090 read implies executable mappings
3092 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3094 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3095 register save and restore. The kernel will only save
3096 legacy floating-point registers on task switch.
3098 nohugeiomap [KNL,x86,PPC] Disable kernel huge I/O mappings.
3100 nosmt [KNL,S390] Disable symmetric multithreading (SMT).
3101 Equivalent to smt=1.
3103 [KNL,x86] Disable symmetric multithreading (SMT).
3104 nosmt=force: Force disable SMT, cannot be undone
3105 via the sysfs control file.
3107 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1
3108 (bounds check bypass). With this option data leaks are
3109 possible in the system.
3111 nospectre_v2 [X86,PPC_FSL_BOOK3E,ARM64] Disable all mitigations for
3112 the Spectre variant 2 (indirect branch prediction)
3113 vulnerability. System may allow data leaks with this
3116 nospec_store_bypass_disable
3117 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
3119 noxsave [BUGS=X86] Disables x86 extended register state save
3120 and restore using xsave. The kernel will fallback to
3121 enabling legacy floating-point and sse state.
3123 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
3124 register states. The kernel will fall back to use
3125 xsave to save the states. By using this parameter,
3126 performance of saving the states is degraded because
3127 xsave doesn't support modified optimization while
3128 xsaveopt supports it on xsaveopt enabled systems.
3130 noxsaves [X86] Disables xsaves and xrstors used in saving and
3131 restoring x86 extended register state in compacted
3132 form of xsave area. The kernel will fall back to use
3133 xsaveopt and xrstor to save and restore the states
3134 in standard form of xsave area. By using this
3135 parameter, xsave area per process might occupy more
3136 memory on xsaves enabled systems.
3138 nohlt [BUGS=ARM,SH] Tells the kernel that the sleep(SH) or
3139 wfi(ARM) instruction doesn't work correctly and not to
3140 use it. This is also useful when using JTAG debugger.
3142 no_file_caps Tells the kernel not to honor file capabilities. The
3143 only way then for a file to be executed with privilege
3144 is to be setuid root or executed by root.
3146 nohalt [IA-64] Tells the kernel not to use the power saving
3147 function PAL_HALT_LIGHT when idle. This increases
3148 power-consumption. On the positive side, it reduces
3149 interrupt wake-up latency, which may improve performance
3150 in certain environments such as networked servers or
3153 nohibernate [HIBERNATION] Disable hibernation and resume.
3155 nohz= [KNL] Boottime enable/disable dynamic ticks
3156 Valid arguments: on, off
3159 nohz_full= [KNL,BOOT,SMP,ISOL]
3160 The argument is a cpu list, as described above.
3161 In kernels built with CONFIG_NO_HZ_FULL=y, set
3162 the specified list of CPUs whose tick will be stopped
3163 whenever possible. The boot CPU will be forced outside
3164 the range to maintain the timekeeping. Any CPUs
3165 in this list will have their RCU callbacks offloaded,
3166 just as if they had also been called out in the
3167 rcu_nocbs= boot parameter.
3169 noiotrap [SH] Disables trapped I/O port accesses.
3171 noirqdebug [X86-32] Disables the code which attempts to detect and
3172 disable unhandled interrupt sources.
3174 no_timer_check [X86,APIC] Disables the code which tests for
3175 broken timer IRQ sources.
3177 noisapnp [ISAPNP] Disables ISA PnP code.
3179 noinitrd [RAM] Tells the kernel not to load any configured
3182 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
3184 [Deprecated - use intremap=off]
3188 noinvpcid [X86] Disable the INVPCID cpu feature.
3190 nojitter [IA-64] Disables jitter checking for ITC timers.
3192 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
3194 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
3198 [X86,PV_OPS] Disable paravirtualized VMware scheduler
3199 clock and use the default one.
3201 no-steal-acc [X86,KVM,ARM64] Disable paravirtualized steal time
3202 accounting. steal time is computed, but won't
3203 influence scheduler behaviour
3205 nolapic [X86-32,APIC] Do not enable or use the local APIC.
3207 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
3209 noltlbs [PPC] Do not use large page/tlb entries for kernel
3210 lowmem mapping on PPC40x and PPC8xx
3212 nomca [IA-64] Disable machine check abort handling
3214 nomce [X86-32] Disable Machine Check Exception
3216 nomfgpt [X86-32] Disable Multi-Function General Purpose
3217 Timer usage (for AMD Geode machines).
3219 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3220 shutdown the other cpus. Instead use the REBOOT_VECTOR
3223 nomodule Disable module load
3225 nopat [X86] Disable PAT (page attribute table extension of
3226 pagetables) support.
3228 nopcid [X86-64] Disable the PCID cpu feature.
3230 norandmaps Don't use address space randomization. Equivalent to
3231 echo 0 > /proc/sys/kernel/randomize_va_space
3233 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
3234 with UP alternatives
3236 nordrand [X86] Disable kernel use of the RDRAND and
3237 RDSEED instructions even if they are supported
3238 by the processor. RDRAND and RDSEED are still
3239 available to user space applications.
3241 noresume [SWSUSP] Disables resume and restores original swap
3244 no-scroll [VGA] Disables scrollback.
3245 This is required for the Braillex ib80-piezo Braille
3246 reader made by F.H. Papenmeier (Germany).
3250 nosep [BUGS=X86-32] Disables x86 SYSENTER/SYSEXIT support.
3252 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
3253 and disable the IO APIC. legacy for "maxcpus=0".
3255 nosoftlockup [KNL] Disable the soft-lockup detector.
3257 nosync [HW,M68K] Disables sync negotiation for all devices.
3259 nowatchdog [KNL] Disable both lockup detectors, i.e.
3260 soft-lockup and NMI watchdog (hard-lockup).
3264 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
3266 cpu0_hotplug [X86] Turn on CPU0 hotplug feature when
3267 CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off.
3268 Some features depend on CPU0. Known dependencies are:
3269 1. Resume from suspend/hibernate depends on CPU0.
3270 Suspend/hibernate will fail if CPU0 is offline and you
3271 need to online CPU0 before suspend/hibernate.
3272 2. PIC interrupts also depend on CPU0. CPU0 can't be
3273 removed if a PIC interrupt is detected.
3274 It's said poweroff/reboot may depend on CPU0 on some
3275 machines although I haven't seen such issues so far
3276 after CPU0 is offline on a few tested machines.
3277 If the dependencies are under your control, you can
3278 turn on cpu0_hotplug.
3280 nps_mtm_hs_ctr= [KNL,ARC]
3281 This parameter sets the maximum duration, in
3282 cycles, each HW thread of the CTOP can run
3283 without interruptions, before HW switches it.
3284 The actual maximum duration is 16 times this
3286 Format: integer between 1 and 255
3289 nptcg= [IA-64] Override max number of concurrent global TLB
3290 purges which is reported from either PAL_VM_SUMMARY or
3293 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
3294 could support. nr_cpus=n : n >= 1 limits the kernel to
3295 support 'n' processors. It could be larger than the
3296 number of already plugged CPU during bootup, later in
3297 runtime you can physically add extra cpu until it reaches
3298 n. So during boot up some boot time memory for per-cpu
3299 variables need be pre-allocated for later physical cpu
3302 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
3304 numa_balancing= [KNL,X86] Enable or disable automatic NUMA balancing.
3305 Allowed values are enable and disable
3307 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
3308 'node', 'default' can be specified
3309 This can be set from sysctl after boot.
3310 See Documentation/admin-guide/sysctl/vm.rst for details.
3312 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
3313 See Documentation/debugging-via-ohci1394.txt for more
3316 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
3317 Rather than timing out after 20 ms if an EC
3318 command is not properly ACKed, override the length
3319 of the timeout. We have interrupts disabled while
3320 waiting for the ACK, so if this is set too high
3321 interrupts *may* be lost!
3323 omap_mux= [OMAP] Override bootloader pin multiplexing.
3324 Format: <mux_mode0.mode_name=value>...
3325 For example, to override I2C bus2:
3326 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
3328 oprofile.timer= [HW]
3329 Use timer interrupt instead of performance counters
3331 oprofile.cpu_type= Force an oprofile cpu type
3332 This might be useful if you have an older oprofile
3333 userland or if you want common events.
3334 Format: { arch_perfmon }
3335 arch_perfmon: [X86] Force use of architectural
3336 perfmon on Intel CPUs instead of the
3337 CPU specific event set.
3338 timer: [X86] Force use of architectural NMI
3339 timer mode (see also oprofile.timer
3340 for generic hr timer mode)
3342 oops=panic Always panic on oopses. Default is to just kill the
3343 process, but there is a small probability of
3344 deadlocking the machine.
3345 This will also cause panics on machine check exceptions.
3346 Useful together with panic=30 to trigger a reboot.
3349 [KNL] Boolean flag to control whether the page allocator
3350 should randomize its free lists. The randomization may
3351 be automatically enabled if the kernel detects it is
3352 running on a platform with a direct-mapped memory-side
3353 cache, and this parameter can be used to
3354 override/disable that behavior. The state of the flag
3355 can be read from sysfs at:
3356 /sys/module/page_alloc/parameters/shuffle.
3358 page_owner= [KNL] Boot-time page_owner enabling option.
3359 Storage of the information about who allocated
3360 each page is disabled in default. With this switch,
3362 on: enable the feature
3364 page_poison= [KNL] Boot-time parameter changing the state of
3365 poisoning on the buddy allocator, available with
3366 CONFIG_PAGE_POISONING=y.
3367 off: turn off poisoning (default)
3368 on: turn on poisoning
3370 panic= [KNL] Kernel behaviour on panic: delay <timeout>
3371 timeout > 0: seconds before rebooting
3372 timeout = 0: wait forever
3373 timeout < 0: reboot immediately
3376 panic_print= Bitmask for printing system info when panic happens.
3377 User can chose combination of the following bits:
3378 bit 0: print all tasks info
3379 bit 1: print system memory info
3380 bit 2: print timer info
3381 bit 3: print locks info if CONFIG_LOCKDEP is on
3382 bit 4: print ftrace buffer
3383 bit 5: print all printk messages in buffer
3385 panic_on_warn panic() instead of WARN(). Useful to cause kdump
3388 crash_kexec_post_notifiers
3389 Run kdump after running panic-notifiers and dumping
3390 kmsg. This only for the users who doubt kdump always
3391 succeeds in any situation.
3392 Note that this also increases risks of kdump failure,
3393 because some panic notifiers can make the crashed
3394 kernel more unstable.
3396 parkbd.port= [HW] Parallel port number the keyboard adapter is
3397 connected to, default is 0.
3399 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
3400 0 for XT, 1 for AT (default is AT).
3403 parport= [HW,PPT] Specify parallel ports. 0 disables.
3404 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
3405 Use 'auto' to force the driver to use any
3406 IRQ/DMA settings detected (the default is to
3407 ignore detected IRQ/DMA settings because of
3408 possible conflicts). You can specify the base
3409 address, IRQ, and DMA settings; IRQ and DMA
3410 should be numbers, or 'auto' (for using detected
3411 settings on that particular port), or 'nofifo'
3412 (to avoid using a FIFO even if it is detected).
3413 Parallel ports are assigned in the order they
3414 are specified on the command line, starting
3417 parport_init_mode= [HW,PPT]
3418 Configure VIA parallel port to operate in
3419 a specific mode. This is necessary on Pegasos
3420 computer where firmware has no options for setting
3421 up parallel port mode and sets it to spp.
3422 Currently this function knows 686a and 8231 chips.
3423 Format: [spp|ps2|epp|ecp|ecpepp]
3426 Halt all CPUs after the first oops has been printed for
3427 the specified number of seconds. This is to be used if
3428 your oopses keep scrolling off the screen.
3433 See header of drivers/block/paride/pcd.c.
3434 See also Documentation/admin-guide/blockdev/paride.rst.
3436 pci=option[,option...] [PCI] various PCI subsystem options.
3438 Some options herein operate on a specific device
3439 or a set of devices (<pci_dev>). These are
3440 specified in one of the following formats:
3442 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
3443 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
3445 Note: the first format specifies a PCI
3446 bus/device/function address which may change
3447 if new hardware is inserted, if motherboard
3448 firmware changes, or due to changes caused
3449 by other kernel parameters. If the
3450 domain is left unspecified, it is
3451 taken to be zero. Optionally, a path
3452 to a device through multiple device/function
3453 addresses can be specified after the base
3454 address (this is more robust against
3455 renumbering issues). The second format
3456 selects devices using IDs from the
3457 configuration space which may match multiple
3458 devices in the system.
3460 earlydump dump PCI config space before the kernel
3462 off [X86] don't probe for the PCI bus
3463 bios [X86-32] force use of PCI BIOS, don't access
3464 the hardware directly. Use this if your machine
3465 has a non-standard PCI host bridge.
3466 nobios [X86-32] disallow use of PCI BIOS, only direct
3467 hardware access methods are allowed. Use this
3468 if you experience crashes upon bootup and you
3469 suspect they are caused by the BIOS.
3470 conf1 [X86] Force use of PCI Configuration Access
3471 Mechanism 1 (config address in IO port 0xCF8,
3472 data in IO port 0xCFC, both 32-bit).
3473 conf2 [X86] Force use of PCI Configuration Access
3474 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
3475 the function, IO port 0xCFA, also 8-bit, sets
3476 bus number. The config space is then accessed
3477 through ports 0xC000-0xCFFF).
3478 See http://wiki.osdev.org/PCI for more info
3479 on the configuration access mechanisms.
3480 noaer [PCIE] If the PCIEAER kernel config parameter is
3481 enabled, this kernel boot option can be used to
3482 disable the use of PCIE advanced error reporting.
3483 nodomains [PCI] Disable support for multiple PCI
3484 root domains (aka PCI segments, in ACPI-speak).
3485 nommconf [X86] Disable use of MMCONFIG for PCI
3487 check_enable_amd_mmconf [X86] check for and enable
3488 properly configured MMIO access to PCI
3489 config space on AMD family 10h CPU
3490 nomsi [MSI] If the PCI_MSI kernel config parameter is
3491 enabled, this kernel boot option can be used to
3492 disable the use of MSI interrupts system-wide.
3493 noioapicquirk [APIC] Disable all boot interrupt quirks.
3494 Safety option to keep boot IRQs enabled. This
3495 should never be necessary.
3496 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
3497 primary IO-APIC for bridges that cannot disable
3498 boot IRQs. This fixes a source of spurious IRQs
3499 when the system masks IRQs.
3500 noioapicreroute [APIC] Disable workaround that uses the
3501 boot IRQ equivalent of an IRQ that connects to
3502 a chipset where boot IRQs cannot be disabled.
3503 The opposite of ioapicreroute.
3504 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
3505 routing table. These calls are known to be buggy
3506 on several machines and they hang the machine
3507 when used, but on other computers it's the only
3508 way to get the interrupt routing table. Try
3509 this option if the kernel is unable to allocate
3510 IRQs or discover secondary PCI buses on your
3512 rom [X86] Assign address space to expansion ROMs.
3513 Use with caution as certain devices share
3514 address decoders between ROMs and other
3516 norom [X86] Do not assign address space to
3517 expansion ROMs that do not already have
3518 BIOS assigned address ranges.
3519 nobar [X86] Do not assign address space to the
3520 BARs that weren't assigned by the BIOS.
3521 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
3522 assigned automatically to PCI devices. You can
3523 make the kernel exclude IRQs of your ISA cards
3525 pirqaddr=0xAAAAA [X86] Specify the physical address
3526 of the PIRQ table (normally generated
3527 by the BIOS) if it is outside the
3528 F0000h-100000h range.
3529 lastbus=N [X86] Scan all buses thru bus #N. Can be
3530 useful if the kernel is unable to find your
3531 secondary buses and you want to tell it
3532 explicitly which ones they are.
3533 assign-busses [X86] Always assign all PCI bus
3534 numbers ourselves, overriding
3535 whatever the firmware may have done.
3536 usepirqmask [X86] Honor the possible IRQ mask stored
3537 in the BIOS $PIR table. This is needed on
3538 some systems with broken BIOSes, notably
3539 some HP Pavilion N5400 and Omnibook XE3
3540 notebooks. This will have no effect if ACPI
3541 IRQ routing is enabled.
3542 noacpi [X86] Do not use ACPI for IRQ routing
3543 or for PCI scanning.
3544 use_crs [X86] Use PCI host bridge window information
3545 from ACPI. On BIOSes from 2008 or later, this
3546 is enabled by default. If you need to use this,
3547 please report a bug.
3548 nocrs [X86] Ignore PCI host bridge windows from ACPI.
3549 If you need to use this, please report a bug.
3550 routeirq Do IRQ routing for all PCI devices.
3551 This is normally done in pci_enable_device(),
3552 so this option is a temporary workaround
3553 for broken drivers that don't call it.
3554 skip_isa_align [X86] do not align io start addr, so can
3555 handle more pci cards
3556 noearly [X86] Don't do any early type 1 scanning.
3557 This might help on some broken boards which
3558 machine check when some devices' config space
3559 is read. But various workarounds are disabled
3560 and some IOMMU drivers will not work.
3561 bfsort Sort PCI devices into breadth-first order.
3562 This sorting is done to get a device
3563 order compatible with older (<= 2.4) kernels.
3564 nobfsort Don't sort PCI devices into breadth-first order.
3565 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
3566 tuning and use the BIOS-configured MPS defaults.
3567 pcie_bus_safe Set every device's MPS to the largest value
3568 supported by all devices below the root complex.
3569 pcie_bus_perf Set device MPS to the largest allowable MPS
3570 based on its parent bus. Also set MRRS (Max
3571 Read Request Size) to the largest supported
3572 value (no larger than the MPS that the device
3573 or bus can support) for best performance.
3574 pcie_bus_peer2peer Set every device's MPS to 128B, which
3575 every device is guaranteed to support. This
3576 configuration allows peer-to-peer DMA between
3577 any pair of devices, possibly at the cost of
3578 reduced performance. This also guarantees
3579 that hot-added devices will work.
3580 cbiosize=nn[KMG] The fixed amount of bus space which is
3581 reserved for the CardBus bridge's IO window.
3582 The default value is 256 bytes.
3583 cbmemsize=nn[KMG] The fixed amount of bus space which is
3584 reserved for the CardBus bridge's memory
3585 window. The default value is 64 megabytes.
3588 [<order of align>@]<pci_dev>[; ...]
3589 Specifies alignment and device to reassign
3590 aligned memory resources. How to
3591 specify the device is described above.
3592 If <order of align> is not specified,
3593 PAGE_SIZE is used as alignment.
3594 A PCI-PCI bridge can be specified if resource
3595 windows need to be expanded.
3596 To specify the alignment for several
3597 instances of a device, the PCI vendor,
3598 device, subvendor, and subdevice may be
3599 specified, e.g., 12@pci:8086:9c22:103c:198f
3600 for 4096-byte alignment.
3601 ecrc= Enable/disable PCIe ECRC (transaction layer
3602 end-to-end CRC checking).
3603 bios: Use BIOS/firmware settings. This is the
3607 hpiosize=nn[KMG] The fixed amount of bus space which is
3608 reserved for hotplug bridge's IO window.
3609 Default size is 256 bytes.
3610 hpmmiosize=nn[KMG] The fixed amount of bus space which is
3611 reserved for hotplug bridge's MMIO window.
3612 Default size is 2 megabytes.
3613 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
3614 reserved for hotplug bridge's MMIO_PREF window.
3615 Default size is 2 megabytes.
3616 hpmemsize=nn[KMG] The fixed amount of bus space which is
3617 reserved for hotplug bridge's MMIO and
3619 Default size is 2 megabytes.
3620 hpbussize=nn The minimum amount of additional bus numbers
3621 reserved for buses below a hotplug bridge.
3623 realloc= Enable/disable reallocating PCI bridge resources
3624 if allocations done by BIOS are too small to
3625 accommodate resources required by all child
3627 off: Turn realloc off
3629 realloc same as realloc=on
3630 noari do not use PCIe ARI.
3631 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
3632 do not use PCIe ATS (and IOMMU device IOTLB).
3633 pcie_scan_all Scan all possible PCIe devices. Otherwise we
3634 only look for one device below a PCIe downstream
3636 big_root_window Try to add a big 64bit memory window to the PCIe
3637 root complex on AMD CPUs. Some GFX hardware
3638 can resize a BAR to allow access to all VRAM.
3639 Adding the window is slightly risky (it may
3640 conflict with unreported devices), so this
3642 disable_acs_redir=<pci_dev>[; ...]
3643 Specify one or more PCI devices (in the format
3644 specified above) separated by semicolons.
3645 Each device specified will have the PCI ACS
3646 redirect capabilities forced off which will
3647 allow P2P traffic between devices through
3648 bridges without forcing it upstream. Note:
3649 this removes isolation between devices and
3650 may put more devices in an IOMMU group.
3651 force_floating [S390] Force usage of floating interrupts.
3652 nomio [S390] Do not use MIO instructions.
3654 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
3657 force Enable ASPM even on devices that claim not to support it.
3658 WARNING: Forcing ASPM on may cause system lockups.
3660 pcie_ports= [PCIE] PCIe port services handling:
3661 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
3662 even if the platform doesn't give the OS permission to
3663 use them. This may cause conflicts if the platform
3664 also tries to use these services.
3665 dpc-native Use native PCIe service for DPC only. May
3666 cause conflicts if firmware uses AER or DPC.
3667 compat Disable native PCIe services (PME, AER, DPC, PCIe
3670 pcie_port_pm= [PCIE] PCIe port power management handling:
3671 off Disable power management of all PCIe ports
3672 force Forcibly enable power management of all PCIe ports
3674 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
3675 nomsi Do not use MSI for native PCIe PME signaling (this makes
3676 all PCIe root ports use INTx for all services).
3678 pcmv= [HW,PCMCIA] BadgePAD 4
3682 Keep all power-domains already enabled by bootloader on,
3683 even if no driver has claimed them. This is useful
3684 for debug and development, but should not be
3685 needed on a platform with proper driver support.
3688 See Documentation/admin-guide/blockdev/paride.rst.
3690 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
3693 See arch/parisc/kernel/pdc_chassis.c
3695 percpu_alloc= Select which percpu first chunk allocator to use.
3696 Currently supported values are "embed" and "page".
3697 Archs may support subset or none of the selections.
3698 See comments in mm/percpu.c for details on each
3699 allocator. This parameter is primarily for debugging
3700 and performance comparison.
3703 See Documentation/admin-guide/blockdev/paride.rst.
3706 See Documentation/admin-guide/blockdev/paride.rst.
3708 pirq= [SMP,APIC] Manual mp-table setup
3709 See Documentation/x86/i386/IO-APIC.rst.
3711 plip= [PPT,NET] Parallel port network link
3712 Format: { parport<nr> | timid | 0 }
3713 See also Documentation/admin-guide/parport.rst.
3715 pmtmr= [X86] Manual setup of pmtmr I/O Port.
3716 Override pmtimer IOPort with a hex value.
3720 Enable PNP debug messages (depends on the
3721 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
3722 via /sys/module/pnp/parameters/debug. We always show
3723 current resource usage; turning this on also shows
3724 possible settings and some assignment information.
3730 { on | off | curr | res | no-curr | no-res }
3733 [ISAPNP] Exclude IRQs for the autoconfiguration
3736 [ISAPNP] Exclude DMAs for the autoconfiguration
3738 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
3739 Ranges are in pairs (I/O port base and size).
3742 [ISAPNP] Exclude memory regions for the
3744 Ranges are in pairs (memory base and size).
3746 ports= [IP_VS_FTP] IPVS ftp helper module
3748 Up to 8 (IP_VS_APP_MAX_PORTS) ports
3750 Format: <port>,<port>....
3752 powersave=off [PPC] This option disables power saving features.
3753 It specifically disables cpuidle and sets the
3754 platform machine description specific power_save
3755 function to NULL. On Idle the CPU just reduces
3758 ppc_strict_facility_enable
3759 [PPC] This option catches any kernel floating point,
3760 Altivec, VSX and SPE outside of regions specifically
3761 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
3762 There is some performance impact when enabling this.
3766 Disable Hardware Transactional Memory
3768 print-fatal-signals=
3769 [KNL] debug: print fatal signals
3771 If enabled, warn about various signal handling
3772 related application anomalies: too many signals,
3773 too many POSIX.1 timers, fatal signals causing a
3776 If you hit the warning due to signal overflow,
3777 you might want to try "ulimit -i unlimited".
3781 printk.always_kmsg_dump=
3782 Trigger kmsg_dump for cases other than kernel oops or
3784 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
3787 printk.devkmsg={on,off,ratelimit}
3788 Control writing to /dev/kmsg.
3789 on - unlimited logging to /dev/kmsg from userspace
3790 off - logging to /dev/kmsg disabled
3791 ratelimit - ratelimit the logging
3794 printk.time= Show timing data prefixed to each printk message line
3795 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
3797 processor.max_cstate= [HW,ACPI]
3798 Limit processor to maximum C-state
3799 max_cstate=9 overrides any DMI blacklist limit.
3801 processor.nocst [HW,ACPI]
3802 Ignore the _CST method to determine C-states,
3803 instead using the legacy FADT method
3805 profile= [KNL] Enable kernel profiling via /proc/profile
3806 Format: [<profiletype>,]<number>
3807 Param: <profiletype>: "schedule", "sleep", or "kvm"
3808 [defaults to kernel profiling]
3809 Param: "schedule" - profile schedule points.
3810 Param: "sleep" - profile D-state sleeping (millisecs).
3811 Requires CONFIG_SCHEDSTATS
3812 Param: "kvm" - profile VM exits.
3813 Param: <number> - step/bucket size as a power of 2 for
3814 statistical time based profiling.
3816 prompt_ramdisk= [RAM] List of RAM disks to prompt for floppy disk
3818 See Documentation/admin-guide/blockdev/ramdisk.rst.
3820 psi= [KNL] Enable or disable pressure stall information
3824 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
3825 probe for; one of (bare|imps|exps|lifebook|any).
3826 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
3828 psmouse.resetafter= [HW,MOUSE]
3829 Try to reset the device after so many bad packets
3832 [HW,MOUSE] Set desired mouse resolution, in dpi.
3833 psmouse.smartscroll=
3834 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
3835 0 = disabled, 1 = enabled (default).
3837 pstore.backend= Specify the name of the pstore backend to use
3840 See Documentation/admin-guide/blockdev/paride.rst.
3842 pti= [X86_64] Control Page Table Isolation of user and
3843 kernel address spaces. Disabling this feature
3844 removes hardening, but improves performance of
3845 system calls and interrupts.
3847 on - unconditionally enable
3848 off - unconditionally disable
3849 auto - kernel detects whether your CPU model is
3850 vulnerable to issues that PTI mitigates
3852 Not specifying this option is equivalent to pti=auto.
3855 Equivalent to pti=off
3858 [KNL] Number of legacy pty's. Overwrites compiled-in
3861 quiet [KNL] Disable most log messages
3866 See Documentation/admin-guide/md.rst.
3868 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
3869 See Documentation/admin-guide/blockdev/ramdisk.rst.
3871 random.trust_cpu={on,off}
3872 [KNL] Enable or disable trusting the use of the
3873 CPU's random number generator (if available) to
3874 fully seed the kernel's CRNG. Default is controlled
3875 by CONFIG_RANDOM_TRUST_CPU.
3877 ras=option[,option,...] [KNL] RAS-specific options
3880 Disable the Correctable Errors Collector,
3881 see CONFIG_RAS_CEC help text.
3884 The argument is a cpu list, as described above,
3885 except that the string "all" can be used to
3886 specify every CPU on the system.
3888 In kernels built with CONFIG_RCU_NOCB_CPU=y, set
3889 the specified list of CPUs to be no-callback CPUs.
3890 Invocation of these CPUs' RCU callbacks will be
3891 offloaded to "rcuox/N" kthreads created for that
3892 purpose, where "x" is "p" for RCU-preempt, and
3893 "s" for RCU-sched, and "N" is the CPU number.
3894 This reduces OS jitter on the offloaded CPUs,
3895 which can be useful for HPC and real-time
3896 workloads. It can also improve energy efficiency
3897 for asymmetric multiprocessors.
3900 Rather than requiring that offloaded CPUs
3901 (specified by rcu_nocbs= above) explicitly
3902 awaken the corresponding "rcuoN" kthreads,
3903 make these kthreads poll for callbacks.
3904 This improves the real-time response for the
3905 offloaded CPUs by relieving them of the need to
3906 wake up the corresponding kthread, but degrades
3907 energy efficiency by requiring that the kthreads
3908 periodically wake up to do the polling.
3910 rcutree.blimit= [KNL]
3911 Set maximum number of finished RCU callbacks to
3912 process in one batch.
3914 rcutree.dump_tree= [KNL]
3915 Dump the structure of the rcu_node combining tree
3916 out at early boot. This is used for diagnostic
3917 purposes, to verify correct tree setup.
3919 rcutree.gp_cleanup_delay= [KNL]
3920 Set the number of jiffies to delay each step of
3921 RCU grace-period cleanup.
3923 rcutree.gp_init_delay= [KNL]
3924 Set the number of jiffies to delay each step of
3925 RCU grace-period initialization.
3927 rcutree.gp_preinit_delay= [KNL]
3928 Set the number of jiffies to delay each step of
3929 RCU grace-period pre-initialization, that is,
3930 the propagation of recent CPU-hotplug changes up
3931 the rcu_node combining tree.
3933 rcutree.use_softirq= [KNL]
3934 If set to zero, move all RCU_SOFTIRQ processing to
3935 per-CPU rcuc kthreads. Defaults to a non-zero
3936 value, meaning that RCU_SOFTIRQ is used by default.
3937 Specify rcutree.use_softirq=0 to use rcuc kthreads.
3939 rcutree.rcu_fanout_exact= [KNL]
3940 Disable autobalancing of the rcu_node combining
3941 tree. This is used by rcutorture, and might
3942 possibly be useful for architectures having high
3943 cache-to-cache transfer latencies.
3945 rcutree.rcu_fanout_leaf= [KNL]
3946 Change the number of CPUs assigned to each
3947 leaf rcu_node structure. Useful for very
3948 large systems, which will choose the value 64,
3949 and for NUMA systems with large remote-access
3950 latencies, which will choose a value aligned
3951 with the appropriate hardware boundaries.
3953 rcutree.jiffies_till_first_fqs= [KNL]
3954 Set delay from grace-period initialization to
3955 first attempt to force quiescent states.
3956 Units are jiffies, minimum value is zero,
3957 and maximum value is HZ.
3959 rcutree.jiffies_till_next_fqs= [KNL]
3960 Set delay between subsequent attempts to force
3961 quiescent states. Units are jiffies, minimum
3962 value is one, and maximum value is HZ.
3964 rcutree.jiffies_till_sched_qs= [KNL]
3965 Set required age in jiffies for a
3966 given grace period before RCU starts
3967 soliciting quiescent-state help from
3968 rcu_note_context_switch() and cond_resched().
3969 If not specified, the kernel will calculate
3970 a value based on the most recent settings
3971 of rcutree.jiffies_till_first_fqs
3972 and rcutree.jiffies_till_next_fqs.
3973 This calculated value may be viewed in
3974 rcutree.jiffies_to_sched_qs. Any attempt to set
3975 rcutree.jiffies_to_sched_qs will be cheerfully
3978 rcutree.kthread_prio= [KNL,BOOT]
3979 Set the SCHED_FIFO priority of the RCU per-CPU
3980 kthreads (rcuc/N). This value is also used for
3981 the priority of the RCU boost threads (rcub/N)
3982 and for the RCU grace-period kthreads (rcu_bh,
3983 rcu_preempt, and rcu_sched). If RCU_BOOST is
3984 set, valid values are 1-99 and the default is 1
3985 (the least-favored priority). Otherwise, when
3986 RCU_BOOST is not set, valid values are 0-99 and
3987 the default is zero (non-realtime operation).
3989 rcutree.rcu_nocb_gp_stride= [KNL]
3990 Set the number of NOCB callback kthreads in
3991 each group, which defaults to the square root
3992 of the number of CPUs. Larger numbers reduce
3993 the wakeup overhead on the global grace-period
3994 kthread, but increases that same overhead on
3995 each group's NOCB grace-period kthread.
3997 rcutree.qhimark= [KNL]
3998 Set threshold of queued RCU callbacks beyond which
3999 batch limiting is disabled.
4001 rcutree.qlowmark= [KNL]
4002 Set threshold of queued RCU callbacks below which
4003 batch limiting is re-enabled.
4005 rcutree.rcu_idle_gp_delay= [KNL]
4006 Set wakeup interval for idle CPUs that have
4007 RCU callbacks (RCU_FAST_NO_HZ=y).
4009 rcutree.rcu_idle_lazy_gp_delay= [KNL]
4010 Set wakeup interval for idle CPUs that have
4011 only "lazy" RCU callbacks (RCU_FAST_NO_HZ=y).
4012 Lazy RCU callbacks are those which RCU can
4013 prove do nothing more than free memory.
4015 rcutree.rcu_kick_kthreads= [KNL]
4016 Cause the grace-period kthread to get an extra
4017 wake_up() if it sleeps three times longer than
4018 it should at force-quiescent-state time.
4019 This wake_up() will be accompanied by a
4020 WARN_ONCE() splat and an ftrace_dump().
4022 rcutree.sysrq_rcu= [KNL]
4023 Commandeer a sysrq key to dump out Tree RCU's
4024 rcu_node tree with an eye towards determining
4025 why a new grace period has not yet started.
4027 rcuperf.gp_async= [KNL]
4028 Measure performance of asynchronous
4029 grace-period primitives such as call_rcu().
4031 rcuperf.gp_async_max= [KNL]
4032 Specify the maximum number of outstanding
4033 callbacks per writer thread. When a writer
4034 thread exceeds this limit, it invokes the
4035 corresponding flavor of rcu_barrier() to allow
4036 previously posted callbacks to drain.
4038 rcuperf.gp_exp= [KNL]
4039 Measure performance of expedited synchronous
4040 grace-period primitives.
4042 rcuperf.holdoff= [KNL]
4043 Set test-start holdoff period. The purpose of
4044 this parameter is to delay the start of the
4045 test until boot completes in order to avoid
4048 rcuperf.kfree_rcu_test= [KNL]
4049 Set to measure performance of kfree_rcu() flooding.
4051 rcuperf.kfree_nthreads= [KNL]
4052 The number of threads running loops of kfree_rcu().
4054 rcuperf.kfree_alloc_num= [KNL]
4055 Number of allocations and frees done in an iteration.
4057 rcuperf.kfree_loops= [KNL]
4058 Number of loops doing rcuperf.kfree_alloc_num number
4059 of allocations and frees.
4061 rcuperf.nreaders= [KNL]
4062 Set number of RCU readers. The value -1 selects
4063 N, where N is the number of CPUs. A value
4064 "n" less than -1 selects N-n+1, where N is again
4065 the number of CPUs. For example, -2 selects N
4066 (the number of CPUs), -3 selects N+1, and so on.
4067 A value of "n" less than or equal to -N selects
4070 rcuperf.nwriters= [KNL]
4071 Set number of RCU writers. The values operate
4072 the same as for rcuperf.nreaders.
4073 N, where N is the number of CPUs
4075 rcuperf.perf_type= [KNL]
4076 Specify the RCU implementation to test.
4078 rcuperf.shutdown= [KNL]
4079 Shut the system down after performance tests
4080 complete. This is useful for hands-off automated
4083 rcuperf.verbose= [KNL]
4084 Enable additional printk() statements.
4086 rcuperf.writer_holdoff= [KNL]
4087 Write-side holdoff between grace periods,
4088 in microseconds. The default of zero says
4091 rcutorture.fqs_duration= [KNL]
4092 Set duration of force_quiescent_state bursts
4095 rcutorture.fqs_holdoff= [KNL]
4096 Set holdoff time within force_quiescent_state bursts
4099 rcutorture.fqs_stutter= [KNL]
4100 Set wait time between force_quiescent_state bursts
4103 rcutorture.fwd_progress= [KNL]
4104 Enable RCU grace-period forward-progress testing
4105 for the types of RCU supporting this notion.
4107 rcutorture.fwd_progress_div= [KNL]
4108 Specify the fraction of a CPU-stall-warning
4109 period to do tight-loop forward-progress testing.
4111 rcutorture.fwd_progress_holdoff= [KNL]
4112 Number of seconds to wait between successive
4113 forward-progress tests.
4115 rcutorture.fwd_progress_need_resched= [KNL]
4116 Enclose cond_resched() calls within checks for
4117 need_resched() during tight-loop forward-progress
4120 rcutorture.gp_cond= [KNL]
4121 Use conditional/asynchronous update-side
4122 primitives, if available.
4124 rcutorture.gp_exp= [KNL]
4125 Use expedited update-side primitives, if available.
4127 rcutorture.gp_normal= [KNL]
4128 Use normal (non-expedited) asynchronous
4129 update-side primitives, if available.
4131 rcutorture.gp_sync= [KNL]
4132 Use normal (non-expedited) synchronous
4133 update-side primitives, if available. If all
4134 of rcutorture.gp_cond=, rcutorture.gp_exp=,
4135 rcutorture.gp_normal=, and rcutorture.gp_sync=
4136 are zero, rcutorture acts as if is interpreted
4137 they are all non-zero.
4139 rcutorture.n_barrier_cbs= [KNL]
4140 Set callbacks/threads for rcu_barrier() testing.
4142 rcutorture.nfakewriters= [KNL]
4143 Set number of concurrent RCU writers. These just
4144 stress RCU, they don't participate in the actual
4145 test, hence the "fake".
4147 rcutorture.nreaders= [KNL]
4148 Set number of RCU readers. The value -1 selects
4149 N-1, where N is the number of CPUs. A value
4150 "n" less than -1 selects N-n-2, where N is again
4151 the number of CPUs. For example, -2 selects N
4152 (the number of CPUs), -3 selects N+1, and so on.
4154 rcutorture.object_debug= [KNL]
4155 Enable debug-object double-call_rcu() testing.
4157 rcutorture.onoff_holdoff= [KNL]
4158 Set time (s) after boot for CPU-hotplug testing.
4160 rcutorture.onoff_interval= [KNL]
4161 Set time (jiffies) between CPU-hotplug operations,
4162 or zero to disable CPU-hotplug testing.
4164 rcutorture.shuffle_interval= [KNL]
4165 Set task-shuffle interval (s). Shuffling tasks
4166 allows some CPUs to go into dyntick-idle mode
4167 during the rcutorture test.
4169 rcutorture.shutdown_secs= [KNL]
4170 Set time (s) after boot system shutdown. This
4171 is useful for hands-off automated testing.
4173 rcutorture.stall_cpu= [KNL]
4174 Duration of CPU stall (s) to test RCU CPU stall
4175 warnings, zero to disable.
4177 rcutorture.stall_cpu_holdoff= [KNL]
4178 Time to wait (s) after boot before inducing stall.
4180 rcutorture.stall_cpu_irqsoff= [KNL]
4181 Disable interrupts while stalling if set.
4183 rcutorture.stat_interval= [KNL]
4184 Time (s) between statistics printk()s.
4186 rcutorture.stutter= [KNL]
4187 Time (s) to stutter testing, for example, specifying
4188 five seconds causes the test to run for five seconds,
4189 wait for five seconds, and so on. This tests RCU's
4190 ability to transition abruptly to and from idle.
4192 rcutorture.test_boost= [KNL]
4193 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
4194 "Maybe" means test if the RCU implementation
4195 under test support RCU priority boosting.
4197 rcutorture.test_boost_duration= [KNL]
4198 Duration (s) of each individual boost test.
4200 rcutorture.test_boost_interval= [KNL]
4201 Interval (s) between each boost test.
4203 rcutorture.test_no_idle_hz= [KNL]
4204 Test RCU's dyntick-idle handling. See also the
4205 rcutorture.shuffle_interval parameter.
4207 rcutorture.torture_type= [KNL]
4208 Specify the RCU implementation to test.
4210 rcutorture.verbose= [KNL]
4211 Enable additional printk() statements.
4213 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
4214 Dump ftrace buffer after reporting RCU CPU
4217 rcupdate.rcu_cpu_stall_suppress= [KNL]
4218 Suppress RCU CPU stall warning messages.
4220 rcupdate.rcu_cpu_stall_timeout= [KNL]
4221 Set timeout for RCU CPU stall warning messages.
4223 rcupdate.rcu_expedited= [KNL]
4224 Use expedited grace-period primitives, for
4225 example, synchronize_rcu_expedited() instead
4226 of synchronize_rcu(). This reduces latency,
4227 but can increase CPU utilization, degrade
4228 real-time latency, and degrade energy efficiency.
4229 No effect on CONFIG_TINY_RCU kernels.
4231 rcupdate.rcu_normal= [KNL]
4232 Use only normal grace-period primitives,
4233 for example, synchronize_rcu() instead of
4234 synchronize_rcu_expedited(). This improves
4235 real-time latency, CPU utilization, and
4236 energy efficiency, but can expose users to
4237 increased grace-period latency. This parameter
4238 overrides rcupdate.rcu_expedited. No effect on
4239 CONFIG_TINY_RCU kernels.
4241 rcupdate.rcu_normal_after_boot= [KNL]
4242 Once boot has completed (that is, after
4243 rcu_end_inkernel_boot() has been invoked), use
4244 only normal grace-period primitives. No effect
4245 on CONFIG_TINY_RCU kernels.
4247 rcupdate.rcu_task_stall_timeout= [KNL]
4248 Set timeout in jiffies for RCU task stall warning
4249 messages. Disable with a value less than or equal
4252 rcupdate.rcu_self_test= [KNL]
4253 Run the RCU early boot self tests
4257 Run specified binary instead of /init from the ramdisk,
4258 used for early userspace startup. See initrd.
4261 force - Override the decision by the kernel to hide the
4262 advertisement of RDRAND support (this affects
4263 certain AMD processors because of buggy BIOS
4264 support, specifically around the suspend/resume
4268 Turn on/off individual RDT features. List is:
4269 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
4271 E.g. to turn on cmt and turn off mba use:
4275 Format (x86 or x86_64):
4276 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] \
4278 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
4280 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
4281 (prefix with 'panic_' to set mode for panic
4283 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
4284 reboot_force is either force or not specified,
4285 reboot_cpu is s[mp]#### with #### being the processor
4286 to be used for rebooting.
4289 [KNL, SMP] Set scheduler's default relax_domain_level.
4290 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
4292 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
4293 Format: <base1>,<size1>[,<base2>,<size2>,...]
4294 Reserve I/O ports or memory so the kernel won't use
4295 them. If <base> is less than 0x10000, the region
4296 is assumed to be I/O ports; otherwise it is memory.
4298 reservetop= [X86-32]
4300 Reserves a hole at the top of the kernel virtual
4305 Set the amount of memory to reserve for BIOS at
4306 the bottom of the address space.
4308 reset_devices [KNL] Force drivers to reset the underlying device
4309 during initialization.
4312 Specify the partition device for software suspend
4314 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
4316 resume_offset= [SWSUSP]
4317 Specify the offset from the beginning of the partition
4318 given by "resume=" at which the swap header is located,
4319 in <PAGE_SIZE> units (needed only for swap files).
4320 See Documentation/power/swsusp-and-swap-files.rst
4322 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
4323 read the resume files
4325 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
4326 Useful for devices that are detected asynchronously
4327 (e.g. USB and MMC devices).
4329 hibernate= [HIBERNATION]
4330 noresume Don't check if there's a hibernation image
4331 present during boot.
4332 nocompress Don't compress/decompress hibernation images.
4333 no Disable hibernation and resume.
4334 protect_image Turn on image protection during restoration
4335 (that will set all pages holding image data
4336 during restoration read-only).
4338 retain_initrd [RAM] Keep initrd memory after extraction
4340 rfkill.default_state=
4341 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
4342 etc. communication is blocked by default.
4345 rfkill.master_switch_mode=
4346 0 The "airplane mode" button does nothing.
4347 1 The "airplane mode" button toggles between everything
4348 blocked and the previous configuration.
4349 2 The "airplane mode" button toggles between everything
4350 blocked and everything unblocked.
4352 rhash_entries= [KNL,NET]
4353 Set number of hash buckets for route cache
4356 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
4359 ro [KNL] Mount root device read-only on boot
4362 on Mark read-only kernel memory as read-only (default).
4363 off Leave read-only kernel memory writable for debugging.
4366 Enable the uart passthrough on the designated usb port
4367 on Rockchip SoCs. When active, the signals of the
4368 debug-uart get routed to the D+ and D- pins of the usb
4369 port and the regular usb controller gets disabled.
4371 root= [KNL] Root filesystem
4372 See name_to_dev_t comment in init/do_mounts.c.
4374 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
4375 mount the root filesystem
4377 rootflags= [KNL] Set root filesystem mount option string
4379 rootfstype= [KNL] Set root filesystem type
4381 rootwait [KNL] Wait (indefinitely) for root device to show up.
4382 Useful for devices that are detected asynchronously
4383 (e.g. USB and MMC devices).
4385 rproc_mem=nn[KMG][@address]
4386 [KNL,ARM,CMA] Remoteproc physical memory block.
4387 Memory area to be used by remote processor image,
4390 rw [KNL] Mount root device read-write on boot
4392 S [KNL] Run init in single mode
4394 s390_iommu= [HW,S390]
4395 Set s390 IOTLB flushing mode
4397 With strict flushing every unmap operation will result in
4398 an IOTLB flush. Default is lazy flushing before reuse,
4402 See drivers/net/irda/sa1100_ir.c.
4404 sbni= [NET] Granch SBNI12 leased line adapter
4406 sched_debug [KNL] Enables verbose scheduler debug messages.
4408 schedstats= [KNL,X86] Enable or disable scheduled statistics.
4409 Allowed values are enable and disable. This feature
4410 incurs a small amount of overhead in the scheduler
4411 but is useful for debugging and performance tuning.
4413 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
4414 xtime_lock contention on larger systems, and/or RCU lock
4415 contention on all systems with CONFIG_MAXSMP set.
4416 Format: { "0" | "1" }
4417 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
4419 Note: increases power consumption, thus should only be
4420 enabled if running jitter sensitive (HPC/RT) workloads.
4422 security= [SECURITY] Choose a legacy "major" security module to
4423 enable at boot. This has been deprecated by the
4426 selinux= [SELINUX] Disable or enable SELinux at boot time.
4427 Format: { "0" | "1" }
4428 See security/selinux/Kconfig help text.
4433 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
4434 Format: { "0" | "1" }
4435 See security/apparmor/Kconfig help text
4438 Default value is set via kernel config option.
4440 serialnumber [BUGS=X86-32]
4443 Maximal number of shapers.
4451 Disable merging of slabs with similar size. May be
4452 necessary if there is some reason to distinguish
4453 allocs to different slabs, especially in hardened
4454 environments where the risk of heap overflows and
4455 layout control by attackers can usually be
4456 frustrated by disabling merging. This will reduce
4457 most of the exposure of a heap attack to a single
4458 cache (risks via metadata attacks are mostly
4459 unchanged). Debug options disable merging on their
4461 For more information see Documentation/vm/slub.rst.
4463 slab_max_order= [MM, SLAB]
4464 Determines the maximum allowed order for slabs.
4465 A high setting may cause OOMs due to memory
4466 fragmentation. Defaults to 1 for systems with
4467 more than 32MB of RAM, 0 otherwise.
4469 slub_debug[=options[,slabs]] [MM, SLUB]
4470 Enabling slub_debug allows one to determine the
4471 culprit if slab objects become corrupted. Enabling
4472 slub_debug can create guard zones around objects and
4473 may poison objects when not in use. Also tracks the
4474 last alloc / free. For more information see
4475 Documentation/vm/slub.rst.
4477 slub_memcg_sysfs= [MM, SLUB]
4478 Determines whether to enable sysfs directories for
4479 memory cgroup sub-caches. 1 to enable, 0 to disable.
4480 The default is determined by CONFIG_SLUB_MEMCG_SYSFS_ON.
4481 Enabling this can lead to a very high number of debug
4482 directories and files being created under
4485 slub_max_order= [MM, SLUB]
4486 Determines the maximum allowed order for slabs.
4487 A high setting may cause OOMs due to memory
4488 fragmentation. For more information see
4489 Documentation/vm/slub.rst.
4491 slub_min_objects= [MM, SLUB]
4492 The minimum number of objects per slab. SLUB will
4493 increase the slab order up to slub_max_order to
4494 generate a sufficiently large slab able to contain
4495 the number of objects indicated. The higher the number
4496 of objects the smaller the overhead of tracking slabs
4497 and the less frequently locks need to be acquired.
4498 For more information see Documentation/vm/slub.rst.
4500 slub_min_order= [MM, SLUB]
4501 Determines the minimum page order for slabs. Must be
4502 lower than slub_max_order.
4503 For more information see Documentation/vm/slub.rst.
4505 slub_nomerge [MM, SLUB]
4506 Same with slab_nomerge. This is supported for legacy.
4507 See slab_nomerge for more information.
4510 Format: <io1>[,<io2>[,...,<io8>]]
4512 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
4513 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
4514 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
4515 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
4516 smsc-ircc2.ircc_irq= [HW] IRQ line
4517 smsc-ircc2.ircc_dma= [HW] DMA channel
4518 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
4519 0: Toshiba Satellite 1800 (GP data pin select)
4520 1: Fast pin select (default)
4523 smt [KNL,S390] Set the maximum number of threads (logical
4524 CPUs) to use per physical CPU on systems capable of
4525 symmetric multithreading (SMT). Will be capped to the
4526 actual hardware limit.
4528 Default: -1 (no limit)
4531 [KNL] Should the soft-lockup detector generate panics.
4534 A nonzero value instructs the soft-lockup detector
4535 to panic the machine when a soft-lockup occurs. It is
4536 also controlled by the kernel.softlockup_panic sysctl
4537 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
4538 respective build-time switch to that functionality.
4540 softlockup_all_cpu_backtrace=
4541 [KNL] Should the soft-lockup detector generate
4542 backtraces on all cpus.
4545 sonypi.*= [HW] Sony Programmable I/O Control Device driver
4546 See Documentation/admin-guide/laptops/sonypi.rst
4548 spectre_v2= [X86] Control mitigation of Spectre variant 2
4549 (indirect branch speculation) vulnerability.
4550 The default operation protects the kernel from
4553 on - unconditionally enable, implies
4555 off - unconditionally disable, implies
4557 auto - kernel detects whether your CPU model is
4560 Selecting 'on' will, and 'auto' may, choose a
4561 mitigation method at run time according to the
4562 CPU, the available microcode, the setting of the
4563 CONFIG_RETPOLINE configuration option, and the
4564 compiler with which the kernel was built.
4566 Selecting 'on' will also enable the mitigation
4567 against user space to user space task attacks.
4569 Selecting 'off' will disable both the kernel and
4570 the user space protections.
4572 Specific mitigations can also be selected manually:
4574 retpoline - replace indirect branches
4575 retpoline,generic - google's original retpoline
4576 retpoline,amd - AMD-specific minimal thunk
4578 Not specifying this option is equivalent to
4582 [X86] Control mitigation of Spectre variant 2
4583 (indirect branch speculation) vulnerability between
4586 on - Unconditionally enable mitigations. Is
4587 enforced by spectre_v2=on
4589 off - Unconditionally disable mitigations. Is
4590 enforced by spectre_v2=off
4592 prctl - Indirect branch speculation is enabled,
4593 but mitigation can be enabled via prctl
4594 per thread. The mitigation control state
4595 is inherited on fork.
4598 - Like "prctl" above, but only STIBP is
4599 controlled per thread. IBPB is issued
4600 always when switching between different user
4604 - Same as "prctl" above, but all seccomp
4605 threads will enable the mitigation unless
4606 they explicitly opt out.
4609 - Like "seccomp" above, but only STIBP is
4610 controlled per thread. IBPB is issued
4611 always when switching between different
4612 user space processes.
4614 auto - Kernel selects the mitigation depending on
4615 the available CPU features and vulnerability.
4618 If CONFIG_SECCOMP=y then "seccomp", otherwise "prctl"
4620 Not specifying this option is equivalent to
4621 spectre_v2_user=auto.
4623 spec_store_bypass_disable=
4624 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
4625 (Speculative Store Bypass vulnerability)
4627 Certain CPUs are vulnerable to an exploit against a
4628 a common industry wide performance optimization known
4629 as "Speculative Store Bypass" in which recent stores
4630 to the same memory location may not be observed by
4631 later loads during speculative execution. The idea
4632 is that such stores are unlikely and that they can
4633 be detected prior to instruction retirement at the
4634 end of a particular speculation execution window.
4636 In vulnerable processors, the speculatively forwarded
4637 store can be used in a cache side channel attack, for
4638 example to read memory to which the attacker does not
4639 directly have access (e.g. inside sandboxed code).
4641 This parameter controls whether the Speculative Store
4642 Bypass optimization is used.
4644 On x86 the options are:
4646 on - Unconditionally disable Speculative Store Bypass
4647 off - Unconditionally enable Speculative Store Bypass
4648 auto - Kernel detects whether the CPU model contains an
4649 implementation of Speculative Store Bypass and
4650 picks the most appropriate mitigation. If the
4651 CPU is not vulnerable, "off" is selected. If the
4652 CPU is vulnerable the default mitigation is
4653 architecture and Kconfig dependent. See below.
4654 prctl - Control Speculative Store Bypass per thread
4655 via prctl. Speculative Store Bypass is enabled
4656 for a process by default. The state of the control
4657 is inherited on fork.
4658 seccomp - Same as "prctl" above, but all seccomp threads
4659 will disable SSB unless they explicitly opt out.
4661 Default mitigations:
4662 X86: If CONFIG_SECCOMP=y "seccomp", otherwise "prctl"
4664 On powerpc the options are:
4666 on,auto - On Power8 and Power9 insert a store-forwarding
4667 barrier on kernel entry and exit. On Power7
4668 perform a software flush on kernel entry and
4672 Not specifying this option is equivalent to
4673 spec_store_bypass_disable=auto.
4675 spia_io_base= [HW,MTD]
4680 srcutree.counter_wrap_check [KNL]
4681 Specifies how frequently to check for
4682 grace-period sequence counter wrap for the
4683 srcu_data structure's ->srcu_gp_seq_needed field.
4684 The greater the number of bits set in this kernel
4685 parameter, the less frequently counter wrap will
4686 be checked for. Note that the bottom two bits
4689 srcutree.exp_holdoff [KNL]
4690 Specifies how many nanoseconds must elapse
4691 since the end of the last SRCU grace period for
4692 a given srcu_struct until the next normal SRCU
4693 grace period will be considered for automatic
4694 expediting. Set to zero to disable automatic
4698 Speculative Store Bypass Disable control
4700 On CPUs that are vulnerable to the Speculative
4701 Store Bypass vulnerability and offer a
4702 firmware based mitigation, this parameter
4703 indicates how the mitigation should be used:
4705 force-on: Unconditionally enable mitigation for
4706 for both kernel and userspace
4707 force-off: Unconditionally disable mitigation for
4708 for both kernel and userspace
4709 kernel: Always enable mitigation in the
4710 kernel, and offer a prctl interface
4711 to allow userspace to register its
4712 interest in being mitigated too.
4714 stack_guard_gap= [MM]
4715 override the default stack gap protection. The value
4716 is in page units and it defines how many pages prior
4717 to (for stacks growing down) resp. after (for stacks
4718 growing up) the main stack are reserved for no other
4719 mapping. Default value is 256 pages.
4722 Enabled the stack tracer on boot up.
4724 stacktrace_filter=[function-list]
4725 [FTRACE] Limit the functions that the stack tracer
4726 will trace at boot up. function-list is a comma separated
4727 list of functions. This list can be changed at run
4728 time by the stack_trace_filter file in the debugfs
4729 tracing directory. Note, this enables stack tracing
4730 and the stacktrace above is not needed.
4734 Set the STI (builtin display/keyboard on the HP-PARISC
4735 machines) console (graphic card) which should be used
4736 as the initial boot-console.
4737 See also comment in drivers/video/console/sticore.c.
4740 See comment in drivers/video/console/sticore.c.
4743 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
4745 sunrpc.min_resvport=
4746 sunrpc.max_resvport=
4748 SunRPC servers often require that client requests
4749 originate from a privileged port (i.e. a port in the
4750 range 0 < portnr < 1024).
4751 An administrator who wishes to reserve some of these
4752 ports for other uses may adjust the range that the
4753 kernel's sunrpc client considers to be privileged
4754 using these two parameters to set the minimum and
4755 maximum port values.
4757 sunrpc.svc_rpc_per_connection_limit=
4759 Limit the number of requests that the server will
4760 process in parallel from a single connection.
4761 The default value is 0 (no limit).
4765 Control how the NFS server code allocates CPUs to
4766 service thread pools. Depending on how many NICs
4767 you have and where their interrupts are bound, this
4768 option will affect which CPUs will do NFS serving.
4769 Note: this parameter cannot be changed while the
4770 NFS server is running.
4772 auto the server chooses an appropriate mode
4773 automatically using heuristics
4774 global a single global pool contains all CPUs
4775 percpu one pool for each CPU
4776 pernode one pool for each NUMA node (equivalent
4777 to global on non-NUMA machines)
4779 sunrpc.tcp_slot_table_entries=
4780 sunrpc.udp_slot_table_entries=
4782 Sets the upper limit on the number of simultaneous
4783 RPC calls that can be sent from the client to a
4784 server. Increasing these values may allow you to
4785 improve throughput, but will also increase the
4786 amount of memory reserved for use by the client.
4788 suspend.pm_test_delay=
4790 Sets the number of seconds to remain in a suspend test
4791 mode before resuming the system (see
4792 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
4793 is set. Default value is 5.
4796 Format: { on | off | y | n | 1 | 0 }
4797 This parameter controls use of the Protected
4798 Execution Facility on pSeries.
4801 [KNL] Enable accounting of swap in memory resource
4802 controller if no parameter or 1 is given or disable
4803 it if 0 is given (See Documentation/admin-guide/cgroup-v1/memory.rst)
4805 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
4806 Format: { <int> | force | noforce }
4807 <int> -- Number of I/O TLB slabs
4808 force -- force using of bounce buffers even if they
4809 wouldn't be automatically used by the kernel
4810 noforce -- Never use bounce buffers (for debugging)
4814 sysfs.deprecated=0|1 [KNL]
4815 Enable/disable old style sysfs layout for old udev
4816 on older distributions. When this option is enabled
4817 very new udev will not work anymore. When this option
4818 is disabled (or CONFIG_SYSFS_DEPRECATED not compiled)
4819 in older udev will not work anymore.
4820 Default depends on CONFIG_SYSFS_DEPRECATED_V2 set in
4821 the kernel configuration.
4823 sysrq_always_enabled
4825 Ignore sysrq setting - this boot parameter will
4826 neutralize any effect of /proc/sys/kernel/sysrq.
4827 Useful for debugging.
4829 tcpmhash_entries= [KNL,NET]
4830 Set the number of tcp_metrics_hash slots.
4831 Default value is 8192 or 16384 depending on total
4832 ram pages. This is used to specify the TCP metrics
4833 cache size. See Documentation/networking/ip-sysctl.txt
4834 "tcp_no_metrics_save" section for more details.
4838 test_suspend= [SUSPEND][,N]
4839 Specify "mem" (for Suspend-to-RAM) or "standby" (for
4840 standby suspend) or "freeze" (for suspend type freeze)
4841 as the system sleep state during system startup with
4842 the optional capability to repeat N number of times.
4843 The system is woken from this state using a
4844 wakeup-capable RTC alarm.
4846 thash_entries= [KNL,NET]
4847 Set number of hash buckets for TCP connection
4849 thermal.act= [HW,ACPI]
4850 -1: disable all active trip points in all thermal zones
4851 <degrees C>: override all lowest active trip points
4853 thermal.crt= [HW,ACPI]
4854 -1: disable all critical trip points in all thermal zones
4855 <degrees C>: override all critical trip points
4857 thermal.nocrt= [HW,ACPI]
4858 Set to disable actions on ACPI thermal zone
4859 critical and hot trip points.
4861 thermal.off= [HW,ACPI]
4862 1: disable ACPI thermal control
4864 thermal.psv= [HW,ACPI]
4865 -1: disable all passive trip points
4866 <degrees C>: override all passive trip points to this
4869 thermal.tzp= [HW,ACPI]
4870 Specify global default ACPI thermal zone polling rate
4871 <deci-seconds>: poll all this frequency
4872 0: no polling (default)
4875 Force threading of all interrupt handlers except those
4876 marked explicitly IRQF_NO_THREAD.
4880 Specify if the kernel should make use of the cpu
4881 topology information if the hardware supports this.
4882 The scheduler will make use of this information and
4883 e.g. base its process migration decisions on it.
4886 topology_updates= [KNL, PPC, NUMA]
4888 Specify if the kernel should ignore (off)
4889 topology updates sent by the hypervisor to this
4894 tpm_suspend_pcr=[HW,TPM]
4895 Format: integer pcr id
4896 Specify that at suspend time, the tpm driver
4897 should extend the specified pcr with zeros,
4898 as a workaround for some chips which fail to
4899 flush the last written pcr on TPM_SaveState.
4900 This will guarantee that all the other pcrs
4903 trace_buf_size=nn[KMG]
4904 [FTRACE] will set tracing buffer size on each cpu.
4906 trace_event=[event-list]
4907 [FTRACE] Set and start specified trace events in order
4908 to facilitate early boot debugging. The event-list is a
4909 comma separated list of trace events to enable. See
4910 also Documentation/trace/events.rst
4912 trace_options=[option-list]
4913 [FTRACE] Enable or disable tracer options at boot.
4914 The option-list is a comma delimited list of options
4915 that can be enabled or disabled just as if you were
4916 to echo the option name into
4918 /sys/kernel/debug/tracing/trace_options
4920 For example, to enable stacktrace option (to dump the
4921 stack trace of each event), add to the command line:
4923 trace_options=stacktrace
4925 See also Documentation/trace/ftrace.rst "trace options"
4929 Have the tracepoints sent to printk as well as the
4930 tracing ring buffer. This is useful for early boot up
4931 where the system hangs or reboots and does not give the
4932 option for reading the tracing buffer or performing a
4933 ftrace_dump_on_oops.
4935 To turn off having tracepoints sent to printk,
4936 echo 0 > /proc/sys/kernel/tracepoint_printk
4937 Note, echoing 1 into this file without the
4938 tracepoint_printk kernel cmdline option has no effect.
4942 Having tracepoints sent to printk() and activating high
4943 frequency tracepoints such as irq or sched, can cause
4944 the system to live lock.
4947 [FTRACE] enable this option to disable tracing when a
4948 warning is hit. This turns off "tracing_on". Tracing can
4949 be enabled again by echoing '1' into the "tracing_on"
4950 file located in /sys/kernel/debug/tracing/
4952 This option is useful, as it disables the trace before
4953 the WARNING dump is called, which prevents the trace to
4954 be filled with content caused by the warning output.
4956 This option can also be set at run time via the sysctl
4957 option: kernel/traceoff_on_warning
4959 transparent_hugepage=
4961 Format: [always|madvise|never]
4962 Can be used to control the default behavior of the system
4963 with respect to transparent hugepages.
4964 See Documentation/admin-guide/mm/transhuge.rst
4967 tsc= Disable clocksource stability checks for TSC.
4969 [x86] reliable: mark tsc clocksource as reliable, this
4970 disables clocksource verification at runtime, as well
4971 as the stability checks done at bootup. Used to enable
4972 high-resolution timer mode on older hardware, and in
4973 virtualized environment.
4974 [x86] noirqtime: Do not use TSC to do irq accounting.
4975 Used to run time disable IRQ_TIME_ACCOUNTING on any
4976 platforms where RDTSC is slow and this accounting
4978 [x86] unstable: mark the TSC clocksource as unstable, this
4979 marks the TSC unconditionally unstable at bootup and
4980 avoids any further wobbles once the TSC watchdog notices.
4981 [x86] nowatchdog: disable clocksource watchdog. Used
4982 in situations with strict latency requirements (where
4983 interruptions from clocksource watchdog are not
4986 tsx= [X86] Control Transactional Synchronization
4987 Extensions (TSX) feature in Intel processors that
4988 support TSX control.
4990 This parameter controls the TSX feature. The options are:
4992 on - Enable TSX on the system. Although there are
4993 mitigations for all known security vulnerabilities,
4994 TSX has been known to be an accelerator for
4995 several previous speculation-related CVEs, and
4996 so there may be unknown security risks associated
4997 with leaving it enabled.
4999 off - Disable TSX on the system. (Note that this
5000 option takes effect only on newer CPUs which are
5001 not vulnerable to MDS, i.e., have
5002 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
5003 the new IA32_TSX_CTRL MSR through a microcode
5004 update. This new MSR allows for the reliable
5005 deactivation of the TSX functionality.)
5007 auto - Disable TSX if X86_BUG_TAA is present,
5008 otherwise enable TSX on the system.
5010 Not specifying this option is equivalent to tsx=off.
5012 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5015 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
5016 Abort (TAA) vulnerability.
5018 Similar to Micro-architectural Data Sampling (MDS)
5019 certain CPUs that support Transactional
5020 Synchronization Extensions (TSX) are vulnerable to an
5021 exploit against CPU internal buffers which can forward
5022 information to a disclosure gadget under certain
5025 In vulnerable processors, the speculatively forwarded
5026 data can be used in a cache side channel attack, to
5027 access data to which the attacker does not have direct
5030 This parameter controls the TAA mitigation. The
5033 full - Enable TAA mitigation on vulnerable CPUs
5036 full,nosmt - Enable TAA mitigation and disable SMT on
5037 vulnerable CPUs. If TSX is disabled, SMT
5038 is not disabled because CPU is not
5039 vulnerable to cross-thread TAA attacks.
5040 off - Unconditionally disable TAA mitigation
5042 On MDS-affected machines, tsx_async_abort=off can be
5043 prevented by an active MDS mitigation as both vulnerabilities
5044 are mitigated with the same mechanism so in order to disable
5045 this mitigation, you need to specify mds=off too.
5047 Not specifying this option is equivalent to
5048 tsx_async_abort=full. On CPUs which are MDS affected
5049 and deploy MDS mitigation, TAA mitigation is not
5050 required and doesn't provide any additional
5054 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5056 turbografx.map[2|3]= [HW,JOY]
5057 TurboGraFX parallel port interface
5059 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
5060 See also Documentation/input/devices/joystick-parport.rst
5062 udbg-immortal [PPC] When debugging early kernel crashes that
5063 happen after console_init() and before a proper
5064 console driver takes over, this boot options might
5065 help "seeing" what's going on.
5067 uhash_entries= [KNL,NET]
5068 Set number of hash buckets for UDP/UDP-Lite connections
5071 [USB] Ignore overcurrent events (default N).
5072 Some badly-designed motherboards generate lots of
5073 bogus events, for ports that aren't wired to
5074 anything. Set this parameter to avoid log spamming.
5075 Note that genuine overcurrent events won't be
5079 [X86] Cause panic on unknown NMI.
5081 usbcore.authorized_default=
5082 [USB] Default USB device authorization:
5083 (default -1 = authorized except for wireless USB,
5084 0 = not authorized, 1 = authorized, 2 = authorized
5085 if device connected to internal port)
5087 usbcore.autosuspend=
5088 [USB] The autosuspend time delay (in seconds) used
5089 for newly-detected USB devices (default 2). This
5090 is the time required before an idle device will be
5091 autosuspended. Devices for which the delay is set
5092 to a negative value won't be autosuspended at all.
5094 usbcore.usbfs_snoop=
5095 [USB] Set to log all usbfs traffic (default 0 = off).
5097 usbcore.usbfs_snoop_max=
5098 [USB] Maximum number of bytes to snoop in each URB
5101 usbcore.blinkenlights=
5102 [USB] Set to cycle leds on hubs (default 0 = off).
5104 usbcore.old_scheme_first=
5105 [USB] Start with the old device initialization
5106 scheme, applies only to low and full-speed devices
5109 usbcore.usbfs_memory_mb=
5110 [USB] Memory limit (in MB) for buffers allocated by
5111 usbfs (default = 16, 0 = max = 2047).
5113 usbcore.use_both_schemes=
5114 [USB] Try the other device initialization scheme
5115 if the first one fails (default 1 = enabled).
5117 usbcore.initial_descriptor_timeout=
5118 [USB] Specifies timeout for the initial 64-byte
5119 USB_REQ_GET_DESCRIPTOR request in milliseconds
5120 (default 5000 = 5.0 seconds).
5122 usbcore.nousb [USB] Disable the USB subsystem
5125 [USB] A list of quirk entries to augment the built-in
5126 usb core quirk list. List entries are separated by
5127 commas. Each entry has the form
5128 VendorID:ProductID:Flags. The IDs are 4-digit hex
5129 numbers and Flags is a set of letters. Each letter
5130 will change the built-in quirk; setting it if it is
5131 clear and clearing it if it is set. The letters have
5132 the following meanings:
5133 a = USB_QUIRK_STRING_FETCH_255 (string
5134 descriptors must not be fetched using
5136 b = USB_QUIRK_RESET_RESUME (device can't resume
5137 correctly so reset it instead);
5138 c = USB_QUIRK_NO_SET_INTF (device can't handle
5139 Set-Interface requests);
5140 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
5141 handle its Configuration or Interface
5143 e = USB_QUIRK_RESET (device can't be reset
5144 (e.g morph devices), don't use reset);
5145 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
5146 more interface descriptions than the
5147 bNumInterfaces count, and can't handle
5148 talking to these interfaces);
5149 g = USB_QUIRK_DELAY_INIT (device needs a pause
5150 during initialization, after we read
5151 the device descriptor);
5152 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
5153 high speed and super speed interrupt
5154 endpoints, the USB 2.0 and USB 3.0 spec
5155 require the interval in microframes (1
5156 microframe = 125 microseconds) to be
5157 calculated as interval = 2 ^
5159 Devices with this quirk report their
5160 bInterval as the result of this
5161 calculation instead of the exponent
5162 variable used in the calculation);
5163 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
5164 handle device_qualifier descriptor
5166 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
5167 generates spurious wakeup, ignore
5168 remote wakeup capability);
5169 k = USB_QUIRK_NO_LPM (device can't handle Link
5171 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
5172 (Device reports its bInterval as linear
5173 frames instead of the USB 2.0
5175 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
5176 to be disconnected before suspend to
5177 prevent spurious wakeup);
5178 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
5179 pause after every control message);
5180 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
5181 delay after resetting its port);
5182 Example: quirks=0781:5580:bk,0a5c:5834:gij
5185 [USBHID] The interval which mice are to be polled at.
5188 [USBHID] The interval which joysticks are to be polled at.
5191 [USBHID] The interval which keyboards are to be polled at.
5193 usb-storage.delay_use=
5194 [UMS] The delay in seconds before a new device is
5195 scanned for Logical Units (default 1).
5198 [UMS] A list of quirks entries to supplement or
5199 override the built-in unusual_devs list. List
5200 entries are separated by commas. Each entry has
5201 the form VID:PID:Flags where VID and PID are Vendor
5202 and Product ID values (4-digit hex numbers) and
5203 Flags is a set of characters, each corresponding
5204 to a common usb-storage quirk flag as follows:
5205 a = SANE_SENSE (collect more than 18 bytes
5206 of sense data, not on uas);
5207 b = BAD_SENSE (don't collect more than 18
5208 bytes of sense data, not on uas);
5209 c = FIX_CAPACITY (decrease the reported
5210 device capacity by one sector);
5211 d = NO_READ_DISC_INFO (don't use
5212 READ_DISC_INFO command, not on uas);
5213 e = NO_READ_CAPACITY_16 (don't use
5214 READ_CAPACITY_16 command);
5215 f = NO_REPORT_OPCODES (don't use report opcodes
5217 g = MAX_SECTORS_240 (don't transfer more than
5218 240 sectors at a time, uas only);
5219 h = CAPACITY_HEURISTICS (decrease the
5220 reported device capacity by one
5221 sector if the number is odd);
5222 i = IGNORE_DEVICE (don't bind to this
5224 j = NO_REPORT_LUNS (don't use report luns
5226 l = NOT_LOCKABLE (don't try to lock and
5227 unlock ejectable media, not on uas);
5228 m = MAX_SECTORS_64 (don't transfer more
5229 than 64 sectors = 32 KB at a time,
5231 n = INITIAL_READ10 (force a retry of the
5232 initial READ(10) command, not on uas);
5233 o = CAPACITY_OK (accept the capacity
5234 reported by the device, not on uas);
5235 p = WRITE_CACHE (the device cache is ON
5236 by default, not on uas);
5237 r = IGNORE_RESIDUE (the device reports
5238 bogus residue values, not on uas);
5239 s = SINGLE_LUN (the device has only one
5241 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
5242 commands, uas only);
5243 u = IGNORE_UAS (don't bind to the uas driver);
5244 w = NO_WP_DETECT (don't test whether the
5245 medium is write-protected).
5246 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
5247 even if the device claims no cache,
5249 Example: quirks=0419:aaf5:rl,0421:0433:rc
5251 user_debug= [KNL,ARM]
5253 See arch/arm/Kconfig.debug help text.
5254 1 - undefined instruction events
5256 4 - invalid data aborts
5259 Example: user_debug=31
5262 [X86] Flags controlling user PTE allocations.
5264 nohigh = do not allocate PTE pages in
5265 HIGHMEM regardless of setting
5269 On X86_32, this is an alias for vdso32=. Otherwise:
5271 vdso=1: enable VDSO (the default)
5272 vdso=0: disable VDSO mapping
5274 vdso32= [X86] Control the 32-bit vDSO
5275 vdso32=1: enable 32-bit VDSO
5276 vdso32=0 or vdso32=2: disable 32-bit VDSO
5278 See the help text for CONFIG_COMPAT_VDSO for more
5279 details. If CONFIG_COMPAT_VDSO is set, the default is
5280 vdso32=0; otherwise, the default is vdso32=1.
5282 For compatibility with older kernels, vdso32=2 is an
5285 Try vdso32=0 if you encounter an error that says:
5286 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
5289 vector=percpu: enable percpu vector domain
5291 video= [FB] Frame buffer configuration
5292 See Documentation/fb/modedb.rst.
5294 video.brightness_switch_enabled= [0,1]
5295 If set to 1, on receiving an ACPI notify event
5296 generated by hotkey, video driver will adjust brightness
5297 level and then send out the event to user space through
5298 the allocated input device; If set to 0, video driver
5299 will only send out the event without touching backlight
5304 [VMMIO] Memory mapped virtio (platform) device.
5306 <size>@<baseaddr>:<irq>[:<id>]
5308 <size> := size (can use standard suffixes
5310 <baseaddr> := physical base address
5311 <irq> := interrupt number (as passed to
5313 <id> := (optional) platform device id
5315 virtio_mmio.device=1K@0x100b0000:48:7
5317 Can be used multiple times for multiple devices.
5319 vga= [BOOT,X86-32] Select a particular video mode
5320 See Documentation/x86/boot.rst and
5321 Documentation/admin-guide/svga.rst.
5322 Use vga=ask for menu.
5323 This is actually a boot loader parameter; the value is
5324 passed to the kernel using a special protocol.
5326 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
5327 May slow down system boot speed, especially when
5328 enabled on systems with a large amount of memory.
5329 All options are enabled by default, and this
5330 interface is meant to allow for selectively
5331 enabling or disabling specific virtual memory
5334 Available options are:
5335 P Enable page structure init time poisoning
5336 - Disable all of the above options
5338 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
5339 size of <nn>. This can be used to increase the
5340 minimum size (128MB on x86). It can also be used to
5341 decrease the size and leave more room for directly
5344 vmcp_cma=nn[MG] [KNL,S390]
5345 Sets the memory size reserved for contiguous memory
5346 allocations for the vmcp device driver.
5348 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
5351 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
5354 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
5358 Controls the behavior of vsyscalls (i.e. calls to
5359 fixed addresses of 0xffffffffff600x00 from legacy
5360 code). Most statically-linked binaries and older
5361 versions of glibc use these calls. Because these
5362 functions are at fixed addresses, they make nice
5363 targets for exploits that can control RIP.
5365 emulate [default] Vsyscalls turn into traps and are
5366 emulated reasonably safely. The vsyscall
5369 xonly Vsyscalls turn into traps and are
5370 emulated reasonably safely. The vsyscall
5371 page is not readable.
5373 none Vsyscalls don't work at all. This makes
5374 them quite hard to use for exploits but
5375 might break your system.
5377 vt.color= [VT] Default text color.
5378 Format: 0xYX, X = foreground, Y = background.
5379 Default: 0x07 = light gray on black.
5381 vt.cur_default= [VT] Default cursor shape.
5382 Format: 0xCCBBAA, where AA, BB, and CC are the same as
5383 the parameters of the <Esc>[?A;B;Cc escape sequence;
5384 see VGA-softcursor.txt. Default: 2 = underline.
5386 vt.default_blu= [VT]
5387 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
5388 Change the default blue palette of the console.
5389 This is a 16-member array composed of values
5392 vt.default_grn= [VT]
5393 Format: <green0>,<green1>,<green2>,...,<green15>
5394 Change the default green palette of the console.
5395 This is a 16-member array composed of values
5398 vt.default_red= [VT]
5399 Format: <red0>,<red1>,<red2>,...,<red15>
5400 Change the default red palette of the console.
5401 This is a 16-member array composed of values
5407 Set system-wide default UTF-8 mode for all tty's.
5408 Default is 1, i.e. UTF-8 mode is enabled for all
5409 newly opened terminals.
5411 vt.global_cursor_default=
5414 Set system-wide default for whether a cursor
5415 is shown on new VTs. Default is -1,
5416 i.e. cursors will be created by default unless
5417 overridden by individual drivers. 0 will hide
5418 cursors, 1 will display them.
5420 vt.italic= [VT] Default color for italic text; 0-15.
5423 vt.underline= [VT] Default color for underlined text; 0-15.
5426 watchdog timers [HW,WDT] For information on watchdog timers,
5427 see Documentation/watchdog/watchdog-parameters.rst
5428 or other driver-specific files in the
5429 Documentation/watchdog/ directory.
5433 Set the hard lockup detector stall duration
5434 threshold in seconds. The soft lockup detector
5435 threshold is set to twice the value. A value of 0
5436 disables both lockup detectors. Default is 10
5439 workqueue.watchdog_thresh=
5440 If CONFIG_WQ_WATCHDOG is configured, workqueue can
5441 warn stall conditions and dump internal state to
5442 help debugging. 0 disables workqueue stall
5443 detection; otherwise, it's the stall threshold
5444 duration in seconds. The default value is 30 and
5445 it can be updated at runtime by writing to the
5446 corresponding sysfs file.
5448 workqueue.disable_numa
5449 By default, all work items queued to unbound
5450 workqueues are affine to the NUMA nodes they're
5451 issued on, which results in better behavior in
5452 general. If NUMA affinity needs to be disabled for
5453 whatever reason, this option can be used. Note
5454 that this also can be controlled per-workqueue for
5455 workqueues visible under /sys/bus/workqueue/.
5457 workqueue.power_efficient
5458 Per-cpu workqueues are generally preferred because
5459 they show better performance thanks to cache
5460 locality; unfortunately, per-cpu workqueues tend to
5461 be more power hungry than unbound workqueues.
5463 Enabling this makes the per-cpu workqueues which
5464 were observed to contribute significantly to power
5465 consumption unbound, leading to measurably lower
5466 power usage at the cost of small performance
5469 The default value of this parameter is determined by
5470 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
5472 workqueue.debug_force_rr_cpu
5473 Workqueue used to implicitly guarantee that work
5474 items queued without explicit CPU specified are put
5475 on the local CPU. This guarantee is no longer true
5476 and while local CPU is still preferred work items
5477 may be put on foreign CPUs. This debug option
5478 forces round-robin CPU selection to flush out
5479 usages which depend on the now broken guarantee.
5480 When enabled, memory and cache locality will be
5483 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
5484 default x2apic cluster mode on platforms
5487 x86_intel_mid_timer= [X86-32,APBT]
5488 Choose timer option for x86 Intel MID platform.
5489 Two valid options are apbt timer only and lapic timer
5490 plus one apbt timer for broadcast timer.
5491 x86_intel_mid_timer=apbt_only | lapic_and_apbt
5493 xen_512gb_limit [KNL,X86-64,XEN]
5494 Restricts the kernel running paravirtualized under Xen
5495 to use only up to 512 GB of RAM. The reason to do so is
5496 crash analysis tools and Xen tools for doing domain
5497 save/restore/migration must be enabled to handle larger
5500 xen_emul_unplug= [HW,X86,XEN]
5501 Unplug Xen emulated devices
5502 Format: [unplug0,][unplug1]
5503 ide-disks -- unplug primary master IDE devices
5504 aux-ide-disks -- unplug non-primary-master IDE devices
5505 nics -- unplug network devices
5506 all -- unplug all emulated devices (NICs and IDE disks)
5507 unnecessary -- unplugging emulated devices is
5508 unnecessary even if the host did not respond to
5510 never -- do not unplug even if version check succeeds
5512 xen_legacy_crash [X86,XEN]
5513 Crash from Xen panic notifier, without executing late
5514 panic() code such as dumping handler.
5516 xen_nopvspin [X86,XEN]
5517 Disables the ticketlock slowpath using Xen PV
5521 Disables the PV optimizations forcing the HVM guest to
5522 run as generic HVM guest with no PV drivers.
5523 This option is obsoleted by the "nopv" option, which
5524 has equivalent effect for XEN platform.
5526 xen_scrub_pages= [XEN]
5527 Boolean option to control scrubbing pages before giving them back
5528 to Xen, for use by other domains. Can be also changed at runtime
5529 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
5530 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
5532 xen_timer_slop= [X86-64,XEN]
5533 Set the timer slop (in nanoseconds) for the virtual Xen
5534 timers (default is 100000). This adjusts the minimum
5535 delta of virtualized Xen timers, where lower values
5536 improve timer resolution at the expense of processing
5537 more timer interrupts.
5539 nopv= [X86,XEN,KVM,HYPER_V,VMWARE]
5540 Disables the PV optimizations forcing the guest to run
5541 as generic guest with no PV drivers. Currently support
5542 XEN HVM, KVM, HYPER_V and VMWARE guest.
5544 xirc2ps_cs= [NET,PCMCIA]
5546 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
5549 By default on POWER9 and above, the kernel will
5550 natively use the XIVE interrupt controller. This option
5551 allows the fallback firmware mode to be used:
5553 off Fallback to firmware control of XIVE interrupt
5554 controller on both pseries and powernv
5555 platforms. Only useful on POWER9 and above.
5557 xhci-hcd.quirks [USB,KNL]
5558 A hex value specifying bitmask with supplemental xhci
5559 host controller quirks. Meaning of each bit can be
5560 consulted in header drivers/usb/host/xhci.h.
5563 Format: { early | on | rw | ro | off }
5564 Controls if xmon debugger is enabled. Default is off.
5565 Passing only "xmon" is equivalent to "xmon=early".
5566 early Call xmon as early as possible on boot; xmon
5567 debugger is called from setup_arch().
5568 on xmon debugger hooks will be installed so xmon
5569 is only called on a kernel crash. Default mode,
5570 i.e. either "ro" or "rw" mode, is controlled
5571 with CONFIG_XMON_DEFAULT_RO_MODE.
5572 rw xmon debugger hooks will be installed so xmon
5573 is called only on a kernel crash, mode is write,
5574 meaning SPR registers, memory and, other data
5575 can be written using xmon commands.
5576 ro same as "rw" option above but SPR registers,
5577 memory, and other data can't be written using
5579 off xmon is disabled.