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
444 bootmem_debug [KNL] Enable bootmem allocator debug messages.
447 Disable BERT OS support on buggy BIOSes.
449 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
450 bttv.radio= Most important insmod options are available as
452 bttv.pll= See Documentation/media/v4l-drivers/bttv.rst
455 bulk_remove=off [PPC] This parameter disables the use of the pSeries
456 firmware feature for flushing multiple hpte entries
459 c101= [NET] Moxa C101 synchronous serial card
461 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
462 Sometimes CPU hardware bugs make them report the cache
463 size incorrectly. The kernel will attempt work arounds
464 to fix known problems, but for some CPUs it is not
465 possible to determine what the correct size should be.
466 This option provides an override for these situations.
469 [NET] Specifies amount of time (in seconds) that
470 the kernel should wait for a network carrier. By default
471 it waits 120 seconds.
473 ca_keys= [KEYS] This parameter identifies a specific key(s) on
474 the system trusted keyring to be used for certificate
476 format: { id:<keyid> | builtin }
478 cca= [MIPS] Override the kernel pages' cache coherency
479 algorithm. Accepted values range from 0 to 7
480 inclusive. See arch/mips/include/asm/pgtable-bits.h
481 for platform specific values (SB1, Loongson3 and
484 ccw_timeout_log [S390]
485 See Documentation/s390/common_io.rst for details.
487 cgroup_disable= [KNL] Disable a particular controller
488 Format: {name of the controller(s) to disable}
489 The effects of cgroup_disable=foo are:
490 - foo isn't auto-mounted if you mount all cgroups in
492 - foo isn't visible as an individually mountable
494 {Currently only "memory" controller deal with this and
495 cut the overhead, others just disable the usage. So
496 only cgroup_disable=memory is actually worthy}
498 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
499 Format: { { controller | "all" | "named" }
500 [,{ controller | "all" | "named" }...] }
501 Like cgroup_disable, but only applies to cgroup v1;
502 the blacklisted controllers remain available in cgroup2.
503 "all" blacklists all controllers and "named" disables
504 named mounts. Specifying both "all" and "named" disables
507 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
509 nosocket -- Disable socket memory accounting.
510 nokmem -- Disable kernel memory accounting.
512 checkreqprot [SELINUX] Set initial checkreqprot flag value.
513 Format: { "0" | "1" }
514 See security/selinux/Kconfig help text.
515 0 -- check protection applied by kernel (includes
516 any implied execute protection).
517 1 -- check protection requested by application.
518 Default value is set via a kernel config option.
519 Value can be changed at runtime via
520 /selinux/checkreqprot.
523 See Documentation/s390/common_io.rst for details.
526 Prevents the clock framework from automatically gating
527 clocks that have not been explicitly enabled by a Linux
528 device driver but are enabled in hardware at reset or
529 by the bootloader/firmware. Note that this does not
530 force such clocks to be always-on nor does it reserve
531 those clocks in any way. This parameter is useful for
532 debug and development, but should not be needed on a
533 platform with proper driver support. For more
534 information, see Documentation/driver-api/clk.rst.
536 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
538 Forces specified clocksource (if available) to be used
539 when calculating gettimeofday(). If specified
540 clocksource is not available, it defaults to PIT.
541 Format: { pit | tsc | cyclone | pmtmr }
543 clocksource= Override the default clocksource
545 Override the default clocksource and use the clocksource
546 with the name specified.
547 Some clocksource names to choose from, depending on
549 [all] jiffies (this is the base, fallback clocksource)
551 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
552 pxa_timer,timer3,32k_counter,timer0_1
553 [X86-32] pit,hpet,tsc;
554 scx200_hrt on Geode; cyclone on IBM x440
562 clocksource.arm_arch_timer.evtstrm=
565 Enable/disable the eventstream feature of the ARM
566 architected timer so that code using WFE-based polling
567 loops can be debugged more effectively on production
570 clocksource.max_cswd_read_retries= [KNL]
571 Number of clocksource_watchdog() retries due to
572 external delays before the clock will be marked
573 unstable. Defaults to three retries, that is,
574 four attempts to read the clock under test.
576 clearcpuid=BITNUM[,BITNUM...] [X86]
577 Disable CPUID feature X for the kernel. See
578 arch/x86/include/asm/cpufeatures.h for the valid bit
579 numbers. Note the Linux specific bits are not necessarily
580 stable over kernel options, but the vendor specific
582 Also note that user programs calling CPUID directly
583 or using the feature without checking anything
584 will still see it. This just prevents it from
585 being used by the kernel or shown in /proc/cpuinfo.
586 Also note the kernel might malfunction if you disable
589 cma=nn[MG]@[start[MG][-end[MG]]]
591 Sets the size of kernel global memory area for
592 contiguous memory allocations and optionally the
593 placement constraint by the physical address range of
594 memory allocations. A value of 0 disables CMA
595 altogether. For more information, see
596 include/linux/dma-contiguous.h
598 cmo_free_hint= [PPC] Format: { yes | no }
599 Specify whether pages are marked as being inactive
600 when they are freed. This is used in CMO environments
601 to determine OS memory pressure for page stealing by
605 coherent_pool=nn[KMG] [ARM,KNL]
606 Sets the size of memory pool for coherent, atomic dma
607 allocations, by default set to 256K.
609 com20020= [HW,NET] ARCnet - COM20020 chipset
611 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
613 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
617 ARCnet - COM90xx chipset (memory-mapped buffers)
618 Format: <io>[,<irq>[,<memstart>]]
620 condev= [HW,S390] console device
623 console= [KNL] Output console device and options.
625 tty<n> Use the virtual console device <n>.
629 Use the specified serial port. The options are of
630 the form "bbbbpnf", where "bbbb" is the baud rate,
631 "p" is parity ("n", "o", or "e"), "n" is number of
632 bits, and "f" is flow control ("r" for RTS or
633 omit it). Default is "9600n8".
635 See Documentation/admin-guide/serial-console.rst for more
637 Documentation/networking/netconsole.txt for an
640 uart[8250],io,<addr>[,options]
641 uart[8250],mmio,<addr>[,options]
642 uart[8250],mmio16,<addr>[,options]
643 uart[8250],mmio32,<addr>[,options]
644 uart[8250],0x<addr>[,options]
645 Start an early, polled-mode console on the 8250/16550
646 UART at the specified I/O port or MMIO address,
647 switching to the matching ttyS device later.
648 MMIO inter-register address stride is either 8-bit
649 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
650 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
651 to be equivalent to 'mmio'. 'options' are specified in
652 the same format described for ttyS above; if unspecified,
653 the h/w is not re-initialized.
655 hvc<n> Use the hypervisor console device <n>. This is for
656 both Xen and PowerPC hypervisors.
658 If the device connected to the port is not a TTY but a braille
659 device, prepend "brl," before the device type, for instance
661 For now, only VisioBraille is supported.
664 [KNL] Change console messages format
666 By default we print messages on consoles in
667 "[time stamp] text\n" format (time stamp may not be
668 printed, depending on CONFIG_PRINTK_TIME or
669 `printk_time' param).
671 Switch to syslog format: "<%u>[time stamp] text\n"
672 IOW, each message will have a facility and loglevel
673 prefix. The format is similar to one used by syslog()
674 syscall, or to executing "dmesg -S --raw" or to reading
677 consoleblank= [KNL] The console blank (screen saver) timeout in
678 seconds. A value of 0 disables the blank timer.
682 [KNL] Change the default value for
683 /proc/<pid>/coredump_filter.
684 See also Documentation/filesystems/proc.txt.
686 coresight_cpu_debug.enable
689 Enable/disable the CPU sampling based debugging.
690 0: default value, disable debugging
691 1: enable debugging at boot time
693 cpuidle.off=1 [CPU_IDLE]
694 disable the cpuidle sub-system
697 [CPU_IDLE] Name of the cpuidle governor to use.
699 cpufreq.off=1 [CPU_FREQ]
700 disable the cpufreq sub-system
703 [X86] Delay for N microsec between assert and de-assert
704 of APIC INIT to start processors. This delay occurs
705 on every CPU online, such as boot, and resume from suspend.
708 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
710 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
712 crashkernel=size[KMG][@offset[KMG]]
713 [KNL] Using kexec, Linux can switch to a 'crash kernel'
714 upon panic. This parameter reserves the physical
715 memory region [offset, offset + size] for that kernel
716 image. If '@offset' is omitted, then a suitable offset
717 is selected automatically.
718 [KNL, x86_64] select a region under 4G first, and
719 fall back to reserve region above 4G when '@offset'
720 hasn't been specified.
721 See Documentation/admin-guide/kdump/kdump.rst for further details.
723 crashkernel=range1:size1[,range2:size2,...][@offset]
724 [KNL] Same as above, but depends on the memory
725 in the running system. The syntax of range is
726 start-[end] where start and end are both
727 a memory unit (amount[KMG]). See also
728 Documentation/admin-guide/kdump/kdump.rst for an example.
730 crashkernel=size[KMG],high
731 [KNL, x86_64] range could be above 4G. Allow kernel
732 to allocate physical memory region from top, so could
733 be above 4G if system have more than 4G ram installed.
734 Otherwise memory region will be allocated below 4G, if
736 It will be ignored if crashkernel=X is specified.
737 crashkernel=size[KMG],low
738 [KNL, x86_64] range under 4G. When crashkernel=X,high
739 is passed, kernel could allocate physical memory region
740 above 4G, that cause second kernel crash on system
741 that require some amount of low memory, e.g. swiotlb
742 requires at least 64M+32K low memory, also enough extra
743 low memory is needed to make sure DMA buffers for 32-bit
744 devices won't run out. Kernel would try to allocate at
745 at least 256M below 4G automatically.
746 This one let user to specify own low range under 4G
747 for second kernel instead.
748 0: to disable low allocation.
749 It will be ignored when crashkernel=X,high is not used
750 or memory reserved is below 4G.
753 [KNL] Disable crypto self-tests
758 cs89x0_media= [HW,NET]
759 Format: { rj45 | aui | bnc }
762 See header of drivers/s390/block/dasd_devmap.c.
764 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
765 (one device per port)
766 Format: <port#>,<type>
767 See also Documentation/input/devices/joystick-parport.rst
769 ddebug_query= [KNL,DYNAMIC_DEBUG] Enable debug messages at early boot
771 Documentation/admin-guide/dynamic-debug-howto.rst for
772 details. Deprecated, see dyndbg.
774 debug [KNL] Enable kernel debugging (events log level).
777 [KNL] Enable printing [hashed] pointers early in the
778 boot sequence. If enabled, we use a weak hash instead
779 of siphash to hash pointers. Use this option if you are
780 seeing instances of '(___ptrval___)') and need to see a
781 value (hashed pointer) instead. Cryptographically
782 insecure, please do not use on production kernels.
785 [KNL] verbose self-tests
787 Print debugging info while doing the locking API
789 We default to 0 (no extra messages), setting it to
790 1 will print _a lot_ more information - normally
791 only useful to kernel developers.
793 debug_objects [KNL] Enable object debugging
796 [KNL] Disable object debugging
798 debug_guardpage_minorder=
799 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
800 parameter allows control of the order of pages that will
801 be intentionally kept free (and hence protected) by the
802 buddy allocator. Bigger value increase the probability
803 of catching random memory corruption, but reduce the
804 amount of memory for normal system use. The maximum
805 possible value is MAX_ORDER/2. Setting this parameter
806 to 1 or 2 should be enough to identify most random
807 memory corruption problems caused by bugs in kernel or
808 driver code when a CPU writes to (or reads from) a
809 random memory location. Note that there exists a class
810 of memory corruptions problems caused by buggy H/W or
811 F/W or by drivers badly programing DMA (basically when
812 memory is written at bus level and the CPU MMU is
813 bypassed) which are not detectable by
814 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
815 tracking down these problems.
818 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
819 enables the feature at boot time. By default, it is
820 disabled and the system will work mostly the same as a
821 kernel built without CONFIG_DEBUG_PAGEALLOC.
822 Note: to get most of debug_pagealloc error reports, it's
823 useful to also enable the page_owner functionality.
824 on: enable the feature
826 debugpat [X86] Enable PAT debugging
829 [same as hugepagesz=] The size of the default
830 HugeTLB page size. This is the size represented by
831 the legacy /proc/ hugepages APIs, used for SHM, and
832 default size when mounting hugetlbfs filesystems.
833 Defaults to the default architecture's huge page size
836 deferred_probe_timeout=
837 [KNL] Debugging option to set a timeout in seconds for
838 deferred probe to give up waiting on dependencies to
839 probe. Only specific dependencies (subsystems or
840 drivers) that have opted in will be ignored. A timeout of 0
841 will timeout at the end of initcalls. This option will also
842 dump out devices still on the deferred probe list after
846 Set number of hash buckets for dentry cache.
848 disable_1tb_segments [PPC]
849 Disables the use of 1TB hash page table segments. This
850 causes the kernel to fall back to 256MB segments which
851 can be useful when debugging issues that require an SLB
855 See Documentation/networking/ipv6.txt.
858 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
859 hardening is enabled for this boot. Hardened
860 usercopy checking is used to protect the kernel
861 from reading or writing beyond known memory
862 allocation boundaries as a proactive defense
863 against bounds-checking flaws in the kernel's
864 copy_to_user()/copy_from_user() interface.
865 on Perform hardened usercopy checks (default).
866 off Disable hardened usercopy checks.
869 Disable RADIX MMU mode on POWER9
872 Disable TLBIE instruction. Currently does not work
873 with KVM, with HASH MMU, or with coherent accelerators.
875 disable_cpu_apicid= [X86,APIC,SMP]
877 The number of initial APIC ID for the
878 corresponding CPU to be disabled at boot,
879 mostly used for the kdump 2nd kernel to
880 disable BSP to wake up multiple CPUs without
881 causing system reset or hang due to sending
884 perf_v4_pmi= [X86,INTEL]
886 Disable Intel PMU counter freezing feature.
887 The feature only exists starting from
888 Arch Perfmon v4 (Skylake and newer).
890 disable_ddw [PPC/PSERIES]
891 Disable Dynamic DMA Window support. Use this if
892 to workaround buggy firmware.
895 See Documentation/networking/ipv6.txt.
897 disable_mtrr_cleanup [X86]
898 The kernel tries to adjust MTRR layout from continuous
899 to discrete, to make X server driver able to add WB
900 entry later. This parameter disables that.
902 disable_mtrr_trim [X86, Intel and AMD only]
903 By default the kernel will trim any uncacheable
904 memory out of your available memory pool based on
905 MTRR settings. This parameter disables that behavior,
906 possibly causing your machine to run very slowly.
908 disable_timer_pin_1 [X86]
909 Disable PIN 1 of APIC timer
910 Can be useful to work around chipset bugs.
912 dis_ucode_ldr [X86] Disable the microcode loader.
914 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
915 this option disables the debugging code at boot.
917 dma_debug_entries=<number>
918 This option allows to tune the number of preallocated
919 entries for DMA-API debugging code. One entry is
920 required per DMA-API allocation. Use this if the
921 DMA-API debugging code disables itself because the
922 architectural default is too low.
924 dma_debug_driver=<driver_name>
925 With this option the DMA-API debugging driver
926 filter feature can be enabled at boot time. Just
927 pass the driver to filter for as the parameter.
928 The filter can be disabled or changed to another
929 driver later using sysfs.
931 driver_async_probe= [KNL]
932 List of driver names to be probed asynchronously.
933 Format: <driver_name1>,<driver_name2>...
935 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
936 Broken monitors, graphic adapters, KVMs and EDIDless
937 panels may send no or incorrect EDID data sets.
938 This parameter allows to specify an EDID data sets
939 in the /lib/firmware directory that are used instead.
940 Generic built-in EDID data sets are used, if one of
941 edid/1024x768.bin, edid/1280x1024.bin,
942 edid/1680x1050.bin, or edid/1920x1080.bin is given
943 and no file with the same name exists. Details and
944 instructions how to build your own EDID data are
945 available in Documentation/driver-api/edid.rst. An EDID
946 data set will only be used for a particular connector,
947 if its name and a colon are prepended to the EDID
948 name. Each connector may use a unique EDID data
949 set by separating the files with a comma. An EDID
950 data set with no connector name will be used for
951 any connectors not explicitly specified.
956 Format: {"off" | "known"}
957 Control how the dt_cpu_ftrs device-tree binding is
958 used for CPU feature discovery and setup (if it
960 off: Do not use it, fall back to legacy cpu table.
961 known: Do not pass through unknown features to guests
962 or userspace, only those that the kernel is aware of.
964 dump_apple_properties [X86]
965 Dump name and content of EFI device properties on
966 x86 Macs. Useful for driver authors to determine
967 what data is available or for reverse-engineering.
969 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
970 module.dyndbg[="val"]
971 Enable debug messages at boot time. See
972 Documentation/admin-guide/dynamic-debug-howto.rst
975 nompx [X86] Disables Intel Memory Protection Extensions.
976 See Documentation/x86/intel_mpx.rst for more
977 information about the feature.
979 nopku [X86] Disable Memory Protection Keys CPU feature found
982 module.async_probe [KNL]
983 Enable asynchronous probe on this module.
985 early_ioremap_debug [KNL]
986 Enable debug messages in early_ioremap support. This
987 is useful for tracking down temporary early mappings
988 which are not unmapped.
990 earlycon= [KNL] Output early console device and options.
992 [ARM64] The early console is determined by the
993 stdout-path property in device tree's chosen node,
994 or determined by the ACPI SPCR table.
996 [X86] When used with no options the early console is
997 determined by the ACPI SPCR table.
999 cdns,<addr>[,options]
1000 Start an early, polled-mode console on a Cadence
1001 (xuartps) serial port at the specified address. Only
1002 supported option is baud rate. If baud rate is not
1003 specified, the serial port must already be setup and
1006 uart[8250],io,<addr>[,options]
1007 uart[8250],mmio,<addr>[,options]
1008 uart[8250],mmio32,<addr>[,options]
1009 uart[8250],mmio32be,<addr>[,options]
1010 uart[8250],0x<addr>[,options]
1011 Start an early, polled-mode console on the 8250/16550
1012 UART at the specified I/O port or MMIO address.
1013 MMIO inter-register address stride is either 8-bit
1014 (mmio) or 32-bit (mmio32 or mmio32be).
1015 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1016 to be equivalent to 'mmio'. 'options' are specified
1017 in the same format described for "console=ttyS<n>"; if
1018 unspecified, the h/w is not initialized.
1022 Start an early, polled-mode console on a pl011 serial
1023 port at the specified address. The pl011 serial port
1024 must already be setup and configured. Options are not
1025 yet supported. If 'mmio32' is specified, then only
1026 the driver will use only 32-bit accessors to read/write
1027 the device registers.
1030 Start an early, polled-mode console on a meson serial
1031 port at the specified address. The serial port must
1032 already be setup and configured. Options are not yet
1036 Start an early, polled-mode console on an msm serial
1037 port at the specified address. The serial port
1038 must already be setup and configured. Options are not
1041 msm_serial_dm,<addr>
1042 Start an early, polled-mode console on an msm serial
1043 dm port at the specified address. The serial port
1044 must already be setup and configured. Options are not
1048 Start an early, polled-mode console on a serial port
1049 of an Actions Semi SoC, such as S500 or S900, at the
1050 specified address. The serial port must already be
1051 setup and configured. Options are not yet supported.
1054 Start an early, polled-mode console on a serial port
1055 of an RDA Micro SoC, such as RDA8810PL, at the
1056 specified address. The serial port must already be
1057 setup and configured. Options are not yet supported.
1060 Use RISC-V SBI (Supervisor Binary Interface) for early
1063 smh Use ARM semihosting calls for early console.
1071 Use early console provided by serial driver available
1072 on Samsung SoCs, requires selecting proper type and
1073 a correct base address of the selected UART port. The
1074 serial port must already be setup and configured.
1075 Options are not yet supported.
1078 Start an early, polled-mode console on a lantiq serial
1079 (lqasc) port at the specified address. The serial port
1080 must already be setup and configured. Options are not
1085 Use early console provided by Freescale LP UART driver
1086 found on Freescale Vybrid and QorIQ LS1021A processors.
1087 A valid base address must be provided, and the serial
1088 port must already be setup and configured.
1091 Start an early, polled-mode console on the
1092 Armada 3700 serial port at the specified
1093 address. The serial port must already be setup
1094 and configured. Options are not yet supported.
1097 Start an early, polled-mode console on a Qualcomm
1098 Generic Interface (GENI) based serial port at the
1099 specified address. The serial port must already be
1100 setup and configured. Options are not yet supported.
1103 Start an early, unaccelerated console on the EFI
1104 memory mapped framebuffer (if available). On cache
1105 coherent non-x86 systems that use system memory for
1106 the framebuffer, pass the 'ram' option so that it is
1107 mapped with the correct attributes.
1110 Use early console provided by Freescale LinFlex UART
1111 serial driver for NXP S32V234 SoCs. A valid base
1112 address must be provided, and the serial port must
1113 already be setup and configured.
1115 earlyprintk= [X86,SH,ARM,M68k,S390]
1119 earlyprintk=serial[,ttySn[,baudrate]]
1120 earlyprintk=serial[,0x...[,baudrate]]
1121 earlyprintk=ttySn[,baudrate]
1122 earlyprintk=dbgp[debugController#]
1123 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1124 earlyprintk=xdbc[xhciController#]
1126 earlyprintk is useful when the kernel crashes before
1127 the normal console is initialized. It is not enabled by
1128 default because it has some cosmetic problems.
1130 Append ",keep" to not disable it when the real console
1133 Only one of vga, efi, serial, or usb debug port can
1136 Currently only ttyS0 and ttyS1 may be specified by
1137 name. Other I/O ports may be explicitly specified
1138 on some architectures (x86 and arm at least) by
1139 replacing ttySn with an I/O port address, like this:
1140 earlyprintk=serial,0x1008,115200
1141 You can find the port for a given device in
1142 /proc/tty/driver/serial:
1143 2: uart:ST16650V2 port:00001008 irq:18 ...
1145 Interaction with the standard serial driver is not
1148 The VGA and EFI output is eventually overwritten by
1151 The xen output can only be used by Xen PV guests.
1153 The sclp output can only be used on s390.
1155 The optional "force" to "pciserial" enables use of a
1156 PCI device even when its classcode is not of the
1159 edac_report= [HW,EDAC] Control how to report EDAC event
1160 Format: {"on" | "off" | "force"}
1161 on: enable EDAC to report H/W event. May be overridden
1162 by other higher priority error reporting module.
1163 off: disable H/W event reporting through EDAC.
1164 force: enforce the use of EDAC to report H/W event.
1167 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1170 This is designed to be used in conjunction with
1171 the boot argument: earlyprintk=vga
1174 Format: {"off" | "on" | "skip[mbr]"}
1177 Format: { "old_map", "nochunk", "noruntime", "debug" }
1178 old_map [X86-64]: switch to the old ioremap-based EFI
1179 runtime services mapping. 32-bit still uses this one by
1181 nochunk: disable reading files in "chunks" in the EFI
1182 boot stub, as chunking can cause problems with some
1183 firmware implementations.
1184 noruntime : disable EFI runtime services support
1185 debug: enable misc debug output
1187 efi_no_storage_paranoia [EFI; X86]
1188 Using this parameter you can use more than 50% of
1189 your efi variable storage. Use this parameter only if
1190 you are really sure that your UEFI does sane gc and
1191 fulfills the spec otherwise your board may brick.
1193 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1194 Add arbitrary attribute to specific memory range by
1195 updating original EFI memory map.
1196 Region of memory which aa attribute is added to is
1198 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1199 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1200 attribute is added to range 0x100000000-0x180000000 and
1201 0x10a0000000-0x1120000000.
1203 Using this parameter you can do debugging of EFI memmap
1204 related feature. For example, you can do debugging of
1205 Address Range Mirroring feature even if your box
1208 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1209 that is to be dynamically loaded by Linux. If there are
1210 multiple variables with the same name but with different
1211 vendor GUIDs, all of them will be loaded. See
1212 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1215 eisa_irq_edge= [PARISC,HW]
1216 See header of drivers/parisc/eisa.c.
1219 See comment before function elanfreq_setup() in
1220 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1222 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1223 Specifies physical address of start of kernel core
1224 image elf header and optionally the size. Generally
1225 kexec loader will pass this option to capture kernel.
1226 See Documentation/admin-guide/kdump/kdump.rst for details.
1228 enable_mtrr_cleanup [X86]
1229 The kernel tries to adjust MTRR layout from continuous
1230 to discrete, to make X server driver able to add WB
1231 entry later. This parameter enables that.
1233 enable_timer_pin_1 [X86]
1234 Enable PIN 1 of APIC timer
1235 Can be useful to work around chipset bugs
1236 (in particular on some ATI chipsets).
1237 The kernel tries to set a reasonable default.
1239 enforcing [SELINUX] Set initial enforcing status.
1241 See security/selinux/Kconfig help text.
1242 0 -- permissive (log only, no denials).
1243 1 -- enforcing (deny and log).
1245 Value can be changed at runtime via /selinux/enforce.
1248 Disable Error Record Serialization Table (ERST)
1251 ether= [HW,NET] Ethernet cards parameters
1252 This option is obsoleted by the "netdev=" option, which
1253 has equivalent usage. See its documentation for details.
1257 Permit 'security.evm' to be updated regardless of
1258 current integrity status.
1262 fail_make_request=[KNL]
1263 General fault injection mechanism.
1264 Format: <interval>,<probability>,<space>,<times>
1265 See also Documentation/fault-injection/.
1268 See Documentation/admin-guide/blockdev/floppy.rst.
1270 force_pal_cache_flush
1271 [IA-64] Avoid check_sal_cache_flush which may hang on
1272 buggy SAL_CACHE_FLUSH implementations. Using this
1273 parameter will force ia64_sal_cache_flush to call
1274 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1277 Forcefully enable Physical Address Extension (PAE).
1278 Many Pentium M systems disable PAE but may have a
1279 functionally usable PAE implementation.
1280 Warning: use of this parameter will taint the kernel
1281 and may cause unknown problems.
1284 [FTRACE] will set and start the specified tracer
1285 as early as possible in order to facilitate early
1288 ftrace_dump_on_oops[=orig_cpu]
1289 [FTRACE] will dump the trace buffers on oops.
1290 If no parameter is passed, ftrace will dump
1291 buffers of all CPUs, but if you pass orig_cpu, it will
1292 dump only the buffer of the CPU that triggered the
1295 ftrace_filter=[function-list]
1296 [FTRACE] Limit the functions traced by the function
1297 tracer at boot up. function-list is a comma separated
1298 list of functions. This list can be changed at run
1299 time by the set_ftrace_filter file in the debugfs
1302 ftrace_notrace=[function-list]
1303 [FTRACE] Do not trace the functions specified in
1304 function-list. This list can be changed at run time
1305 by the set_ftrace_notrace file in the debugfs
1308 ftrace_graph_filter=[function-list]
1309 [FTRACE] Limit the top level callers functions traced
1310 by the function graph tracer at boot up.
1311 function-list is a comma separated list of functions
1312 that can be changed at run time by the
1313 set_graph_function file in the debugfs tracing directory.
1315 ftrace_graph_notrace=[function-list]
1316 [FTRACE] Do not trace from the functions specified in
1317 function-list. This list is a comma separated list of
1318 functions that can be changed at run time by the
1319 set_graph_notrace file in the debugfs tracing directory.
1321 ftrace_graph_max_depth=<uint>
1322 [FTRACE] Used with the function graph tracer. This is
1323 the max depth it will trace into a function. This value
1324 can be changed at run time by the max_graph_depth file
1325 in the tracefs tracing directory. default: 0 (no limit)
1328 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1329 support via parallel port (up to 5 devices per port)
1330 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1331 See also Documentation/input/devices/joystick-parport.rst
1335 gart_fix_e820= [X86_64] disable the fix e820 for K8 GART
1339 gather_data_sampling=
1340 [X86,INTEL] Control the Gather Data Sampling (GDS)
1343 Gather Data Sampling is a hardware vulnerability which
1344 allows unprivileged speculative access to data which was
1345 previously stored in vector registers.
1347 This issue is mitigated by default in updated microcode.
1348 The mitigation may have a performance impact but can be
1349 disabled. On systems without the microcode mitigation
1350 disabling AVX serves as a mitigation.
1352 force: Disable AVX to mitigate systems without
1353 microcode mitigation. No effect if the microcode
1354 mitigation is present. Known to cause crashes in
1355 userspace with buggy AVX enumeration.
1357 off: Disable GDS mitigation.
1359 gcov_persist= [GCOV] When non-zero (default), profiling data for
1360 kernel modules is saved and remains accessible via
1361 debugfs, even when the module is unloaded/reloaded.
1362 When zero, profiling data is discarded and associated
1363 debugfs files are removed at module unload time.
1365 goldfish [X86] Enable the goldfish android emulator platform.
1366 Don't use this when you are not running on the
1369 gpt [EFI] Forces disk with valid GPT signature but
1370 invalid Protective MBR to be treated as GPT. If the
1371 primary GPT is corrupted, it enables the backup/alternate
1372 GPT to be used instead.
1374 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1375 the "Enable 0" bit of the configuration register.
1378 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1379 the "Enable 0" bit of the configuration register.
1382 grcan.select= [HW] Select which physical interface to use.
1385 grcan.txsize= [HW] Sets the size of the tx buffer.
1386 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1388 grcan.rxsize= [HW] Sets the size of the rx buffer.
1389 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1392 gpio-mockup.gpio_mockup_ranges
1393 [HW] Sets the ranges of gpiochip of for this device.
1394 Format: <start1>,<end1>,<start2>,<end2>...
1396 hardlockup_all_cpu_backtrace=
1397 [KNL] Should the hard-lockup detector generate
1398 backtraces on all cpus.
1401 hashdist= [KNL,NUMA] Large hashes allocated during boot
1402 are distributed across NUMA nodes. Defaults on
1403 for 64-bit NUMA, off otherwise.
1404 Format: 0 | 1 (for off | on)
1406 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1408 hd= [EIDE] (E)IDE hard drive subsystem geometry
1409 Format: <cyl>,<head>,<sect>
1412 Disable Hardware Error Source Table (HEST) support;
1413 corresponding firmware-first mode error processing
1414 logic will be disabled.
1416 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1417 size of <nn>. This works even on boxes that have no
1418 highmem otherwise. This also works to reduce highmem
1419 size on bigger boxes.
1421 highres= [KNL] Enable/disable high resolution timer mode.
1422 Valid parameters: "on", "off"
1427 hpet= [X86-32,HPET] option to control HPET usage
1428 Format: { enable (default) | disable | force |
1430 disable: disable HPET and use PIT instead
1431 force: allow force enabled of undocumented chips (ICH4,
1433 verbose: show contents of HPET registers during setup
1435 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1436 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1438 hugepages= [HW,X86-32,IA-64] HugeTLB pages to allocate at boot.
1439 hugepagesz= [HW,IA-64,PPC,X86-64] The size of the HugeTLB pages.
1440 On x86-64 and powerpc, this option can be specified
1441 multiple times interleaved with hugepages= to reserve
1442 huge pages of different sizes. Valid pages sizes on
1443 x86-64 are 2M (when the CPU supports "pse") and 1G
1444 (when the CPU supports the "pdpe1gb" cpuinfo flag).
1447 [KNL] Should the hung task detector generate panics.
1450 A nonzero value instructs the kernel to panic when a
1451 hung task is detected. The default value is controlled
1452 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1453 option. The value selected by this boot parameter can
1454 be changed later by the kernel.hung_task_panic sysctl.
1456 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1457 terminal devices. Valid values: 0..8
1458 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1459 If specified, z/VM IUCV HVC accepts connections
1460 from listed z/VM user IDs only.
1462 hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations
1463 which allow the hypervisor to 'idle' the
1464 guest on lock contention.
1467 Do not unregister boot console at start. This is only
1468 useful for debugging when something happens in the window
1469 between unregistering the boot console and initializing
1472 i2c_bus= [HW] Override the default board specific I2C bus speed
1473 or register an additional I2C bus that is not
1474 registered from board initialization code.
1478 i8042.debug [HW] Toggle i8042 debug mode
1479 i8042.unmask_kbd_data
1480 [HW] Enable printing of interrupt data from the KBD port
1481 (disabled by default, and as a pre-condition
1482 requires that i8042.debug=1 be enabled)
1483 i8042.direct [HW] Put keyboard port into non-translated mode
1484 i8042.dumbkbd [HW] Pretend that controller can only read data from
1485 keyboard and cannot control its state
1486 (Don't attempt to blink the leds)
1487 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1488 i8042.nokbd [HW] Don't check/create keyboard port
1489 i8042.noloop [HW] Disable the AUX Loopback command while probing
1491 i8042.nomux [HW] Don't check presence of an active multiplexing
1493 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1495 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1496 i8042.reset [HW] Reset the controller during init, cleanup and
1497 suspend-to-ram transitions, only during s2r
1498 transitions, or never reset
1499 Format: { 1 | Y | y | 0 | N | n }
1500 1, Y, y: always reset controller
1501 0, N, n: don't ever reset controller
1502 Default: only on s2r transitions on x86; most other
1503 architectures force reset to be always executed
1504 i8042.unlock [HW] Unlock (ignore) the keylock
1505 i8042.kbdreset [HW] Reset device connected to KBD port
1507 [HW] Allow deferred probing upon i8042 probe errors
1511 i8k.ignore_dmi [HW] Continue probing hardware even if DMI data
1512 indicates that the driver is running on unsupported
1514 i8k.force [HW] Activate i8k driver even if SMM BIOS signature
1515 does not match list of supported models.
1517 [HW] Report power status in /proc/i8k
1518 (disabled by default)
1519 i8k.restricted [HW] Allow controlling fans only if SYS_ADMIN
1522 i915.invert_brightness=
1523 [DRM] Invert the sense of the variable that is used to
1524 set the brightness of the panel backlight. Normally a
1525 brightness value of 0 indicates backlight switched off,
1526 and the maximum of the brightness value sets the backlight
1527 to maximum brightness. If this parameter is set to 0
1528 (default) and the machine requires it, or this parameter
1529 is set to 1, a brightness value of 0 sets the backlight
1530 to maximum brightness, and the maximum of the brightness
1531 value switches the backlight off.
1532 -1 -- never invert brightness
1533 0 -- machine default
1534 1 -- force brightness inversion
1537 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1539 ide-core.nodma= [HW] (E)IDE subsystem
1540 Format: =0.0 to prevent dma on hda, =0.1 hdb =1.0 hdc
1541 .vlb_clock .pci_clock .noflush .nohpa .noprobe .nowerr
1542 .cdrom .chs .ignore_cable are additional options
1543 See Documentation/ide/ide.rst.
1545 ide-generic.probe-mask= [HW] (E)IDE subsystem
1547 Probe mask for legacy ISA IDE ports. Depending on
1548 platform up to 6 ports are supported, enabled by
1549 setting corresponding bits in the mask to 1. The
1550 default value is 0x0, which has a special meaning.
1551 On systems that have PCI, it triggers scanning the
1552 PCI bus for the first and the second port, which
1553 are then probed. On systems without PCI the value
1554 of 0x0 enables probing the two first ports as if it
1557 ide-pci-generic.all-generic-ide [HW] (E)IDE subsystem
1558 Claim all unknown PCI IDE storage controllers.
1561 Format: idle=poll, idle=halt, idle=nomwait
1562 Poll forces a polling idle loop that can slightly
1563 improve the performance of waking up a idle CPU, but
1564 will use a lot of power and make the system run hot.
1566 idle=halt: Halt is forced to be used for CPU idle.
1567 In such case C2/C3 won't be used again.
1568 idle=nomwait: Disable mwait for CPU C-states
1570 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1571 Format: { strict | legacy | 2008 | relaxed }
1574 Choose which programs will be accepted for execution
1575 based on the IEEE 754 NaN encoding(s) supported by
1576 the FPU and the NaN encoding requested with the value
1577 of an ELF file header flag individually set by each
1578 binary. Hardware implementations are permitted to
1579 support either or both of the legacy and the 2008 NaN
1582 Available settings are as follows:
1583 strict accept binaries that request a NaN encoding
1584 supported by the FPU
1585 legacy only accept legacy-NaN binaries, if supported
1587 2008 only accept 2008-NaN binaries, if supported
1589 relaxed accept any binaries regardless of whether
1590 supported by the FPU
1592 The FPU emulator is always able to support both NaN
1593 encodings, so if no FPU hardware is present or it has
1594 been disabled with 'nofpu', then the settings of
1595 'legacy' and '2008' strap the emulator accordingly,
1596 'relaxed' straps the emulator for both legacy-NaN and
1597 2008-NaN, whereas 'strict' enables legacy-NaN only on
1598 legacy processors and both NaN encodings on MIPS32 or
1601 The setting for ABS.fmt/NEG.fmt instruction execution
1602 mode generally follows that for the NaN encoding,
1603 except where unsupported by hardware.
1605 ignore_loglevel [KNL]
1606 Ignore loglevel setting - this will print /all/
1607 kernel messages to the console. Useful for debugging.
1608 We also add it as printk module parameter, so users
1609 could change it dynamically, usually by
1610 /sys/module/printk/parameters/ignore_loglevel.
1613 Ignore RLIMIT_DATA setting for data mappings,
1614 print warning at first misuse. Can be changed via
1615 /sys/module/kernel/parameters/ignore_rlimit_data.
1617 ihash_entries= [KNL]
1618 Set number of hash buckets for inode cache.
1620 ima_appraise= [IMA] appraise integrity measurements
1621 Format: { "off" | "enforce" | "fix" | "log" }
1624 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
1625 The builtin appraise policy appraises all files
1628 ima_canonical_fmt [IMA]
1629 Use the canonical format for the binary runtime
1630 measurements, instead of host native format.
1633 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1637 The list of supported hash algorithms is defined
1638 in crypto/hash_info.h.
1641 The builtin policies to load during IMA setup.
1642 Format: "tcb | appraise_tcb | secure_boot |
1645 The "tcb" policy measures all programs exec'd, files
1646 mmap'd for exec, and all files opened with the read
1647 mode bit set by either the effective uid (euid=0) or
1650 The "appraise_tcb" policy appraises the integrity of
1651 all files owned by root.
1653 The "secure_boot" policy appraises the integrity
1654 of files (eg. kexec kernel image, kernel modules,
1655 firmware, policy, etc) based on file signatures.
1657 The "fail_securely" policy forces file signature
1658 verification failure also on privileged mounted
1659 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
1662 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
1663 Load a policy which meets the needs of the Trusted
1664 Computing Base. This means IMA will measure all
1665 programs exec'd, files mmap'd for exec, and all files
1666 opened for read by uid=0.
1669 Select one of defined IMA measurements template formats.
1670 Formats: { "ima" | "ima-ng" | "ima-sig" }
1674 [IMA] Define a custom template format.
1675 Format: { "field1|...|fieldN" }
1677 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
1678 Format: <min_file_size>
1679 Set the minimal file size for using asynchronous hash.
1680 If left unspecified, ahash usage is disabled.
1682 ahash performance varies for different data sizes on
1683 different crypto accelerators. This option can be used
1684 to achieve the best performance for a particular HW.
1686 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
1688 Set hashing buffer size. Default: 4k.
1690 ahash performance varies for different chunk sizes on
1691 different crypto accelerators. This option can be used
1692 to achieve best performance for particular HW.
1696 Run specified binary instead of /sbin/init as init
1699 initcall_debug [KNL] Trace initcalls as they are executed. Useful
1700 for working out where the kernel is dying during
1703 initcall_blacklist= [KNL] Do not execute a comma-separated list of
1704 initcall functions. Useful for debugging built-in
1705 modules and initcalls.
1707 initrd= [BOOT] Specify the location of the initial ramdisk
1709 init_on_alloc= [MM] Fill newly allocated pages and heap objects with
1712 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
1714 init_on_free= [MM] Fill freed pages and heap objects with zeroes.
1716 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
1718 init_pkru= [x86] Specify the default memory protection keys rights
1719 register contents for all processes. 0x55555554 by
1720 default (disallow access to all but pkey 0). Can
1721 override in debugfs after boot.
1723 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
1726 int_pln_enable [x86] Enable power limit notification interrupt
1728 integrity_audit=[IMA]
1729 Format: { "0" | "1" }
1730 0 -- basic integrity auditing messages. (Default)
1731 1 -- additional integrity auditing messages.
1733 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
1735 Enable intel iommu driver.
1737 Disable intel iommu driver.
1738 igfx_off [Default Off]
1739 By default, gfx is mapped as normal device. If a gfx
1740 device has a dedicated DMAR unit, the DMAR unit is
1741 bypassed by not enabling DMAR with this option. In
1742 this case, gfx device will use physical address for
1745 With this option iommu will not optimize to look
1746 for io virtual address below 32-bit forcing dual
1747 address cycle on pci bus for cards supporting greater
1748 than 32-bit addressing. The default is to look
1749 for translation below 32-bit and if not available
1750 then look in the higher range.
1751 strict [Default Off]
1752 With this option on every unmap_single operation will
1753 result in a hardware IOTLB flush operation as opposed
1754 to batching them for performance.
1755 sp_off [Default Off]
1756 By default, super page will be supported if Intel IOMMU
1757 has the capability. With this option, super page will
1760 By default, scalable mode will be disabled even if the
1761 hardware advertises that it has support for the scalable
1762 mode translation. With this option set, scalable mode
1763 will be used on hardware which claims to support it.
1764 tboot_noforce [Default Off]
1765 Do not force the Intel IOMMU enabled under tboot.
1766 By default, tboot will force Intel IOMMU on, which
1767 could harm performance of some high-throughput
1768 devices like 40GBit network cards, even if identity
1770 Note that using this option lowers the security
1771 provided by tboot because it makes the system
1772 vulnerable to DMA attacks.
1773 nobounce [Default off]
1774 Disable bounce buffer for unstrusted devices such as
1775 the Thunderbolt devices. This will treat the untrusted
1776 devices as the trusted ones, hence might expose security
1777 risks of DMA attacks.
1779 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
1780 0 disables intel_idle and fall back on acpi_idle.
1781 1 to 9 specify maximum depth of C-state.
1785 Do not enable intel_pstate as the default
1786 scaling driver for the supported processors
1788 Use intel_pstate as a scaling driver, but configure it
1789 to work with generic cpufreq governors (instead of
1790 enabling its internal governor). This mode cannot be
1791 used along with the hardware-managed P-states (HWP)
1794 Enable intel_pstate on systems that prohibit it by default
1795 in favor of acpi-cpufreq. Forcing the intel_pstate driver
1796 instead of acpi-cpufreq may disable platform features, such
1797 as thermal controls and power capping, that rely on ACPI
1798 P-States information being indicated to OSPM and therefore
1799 should be used with caution. This option does not work with
1800 processors that aren't supported by the intel_pstate driver
1801 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
1803 Do not enable hardware P state control (HWP)
1806 Only load intel_pstate on systems which support
1807 hardware P state control (HWP) if available.
1809 Enforce ACPI _PPC performance limits. If the Fixed ACPI
1810 Description Table, specifies preferred power management
1811 profile as "Enterprise Server" or "Performance Server",
1812 then this feature is turned on by default.
1814 Allow per-logical-CPU P-State performance control limits using
1815 cpufreq sysfs interface
1817 intremap= [X86-64, Intel-IOMMU]
1818 on enable Interrupt Remapping (default)
1819 off disable Interrupt Remapping
1820 nosid disable Source ID checking
1822 BIOS x2APIC opt-out request will be ignored
1823 nopost disable Interrupt Posting
1825 iomem= Disable strict checking of access to MMIO memory
1826 strict regions from userspace.
1841 nobypass [PPC/POWERNV]
1842 Disable IOMMU bypass, using IOMMU for PCI devices.
1844 iommu.strict= [ARM64] Configure TLB invalidation behaviour
1845 Format: { "0" | "1" }
1847 Request that DMA unmap operations use deferred
1848 invalidation of hardware TLBs, for increased
1849 throughput at the cost of reduced device isolation.
1850 Will fall back to strict mode if not supported by
1851 the relevant IOMMU driver.
1852 1 - Strict mode (default).
1853 DMA unmap operations invalidate IOMMU hardware TLBs
1857 [ARM64, X86] Configure DMA to bypass the IOMMU by default.
1858 Format: { "0" | "1" }
1859 0 - Use IOMMU translation for DMA.
1860 1 - Bypass the IOMMU for DMA.
1861 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
1863 io7= [HW] IO7 for Marvel based alpha systems
1864 See comment before marvel_specify_io7 in
1865 arch/alpha/kernel/core_marvel.c.
1867 io_delay= [X86] I/O delay method
1869 Standard port 0x80 based delay
1871 Alternate port 0xed based delay (needed on some systems)
1873 Simple two microseconds delay
1878 See Documentation/filesystems/nfs/nfsroot.txt.
1880 ipcmni_extend [KNL] Extend the maximum number of unique System V
1881 IPC identifiers from 32,768 to 16,777,216.
1883 irqaffinity= [SMP] Set the default irq affinity mask
1884 The argument is a cpu list, as described above.
1886 irqchip.gicv2_force_probe=
1889 Force the kernel to look for the second 4kB page
1890 of a GICv2 controller even if the memory range
1891 exposed by the device tree is too small.
1893 irqchip.gicv3_nolpi=
1895 Force the kernel to ignore the availability of
1896 LPIs (and by consequence ITSs). Intended for system
1897 that use the kernel as a bootloader, and thus want
1898 to let secondary kernels in charge of setting up
1901 irqchip.gicv3_pseudo_nmi= [ARM64]
1902 Enables support for pseudo-NMIs in the kernel. This
1903 requires the kernel to be built with
1904 CONFIG_ARM64_PSEUDO_NMI.
1907 When an interrupt is not handled search all handlers
1908 for it. Intended to get systems with badly broken
1912 When an interrupt is not handled search all handlers
1913 for it. Also check all handlers each timer
1914 interrupt. Intended to get systems with badly broken
1918 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
1920 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
1921 [Deprecated - use cpusets instead]
1922 Format: [flag-list,]<cpu-list>
1924 Specify one or more CPUs to isolate from disturbances
1925 specified in the flag list (default: domain):
1928 Disable the tick when a single task runs.
1930 A residual 1Hz tick is offloaded to workqueues, which you
1931 need to affine to housekeeping through the global
1932 workqueue's affinity configured via the
1933 /sys/devices/virtual/workqueue/cpumask sysfs file, or
1934 by using the 'domain' flag described below.
1936 NOTE: by default the global workqueue runs on all CPUs,
1937 so to protect individual CPUs the 'cpumask' file has to
1938 be configured manually after bootup.
1941 Isolate from the general SMP balancing and scheduling
1942 algorithms. Note that performing domain isolation this way
1943 is irreversible: it's not possible to bring back a CPU to
1944 the domains once isolated through isolcpus. It's strongly
1945 advised to use cpusets instead to disable scheduler load
1946 balancing through the "cpuset.sched_load_balance" file.
1947 It offers a much more flexible interface where CPUs can
1948 move in and out of an isolated set anytime.
1950 You can move a process onto or off an "isolated" CPU via
1951 the CPU affinity syscalls or cpuset.
1952 <cpu number> begins at 0 and the maximum value is
1953 "number of CPUs in system - 1".
1955 The format of <cpu-list> is described above.
1961 ivrs_ioapic [HW,X86_64]
1962 Provide an override to the IOAPIC-ID<->DEVICE-ID
1963 mapping provided in the IVRS ACPI table.
1964 By default, PCI segment is 0, and can be omitted.
1966 For example, to map IOAPIC-ID decimal 10 to
1967 PCI segment 0x1 and PCI device 00:14.0,
1968 write the parameter as:
1969 ivrs_ioapic=10@0001:00:14.0
1972 * To map IOAPIC-ID decimal 10 to PCI device 00:14.0
1973 write the parameter as:
1974 ivrs_ioapic[10]=00:14.0
1975 * To map IOAPIC-ID decimal 10 to PCI segment 0x1 and
1976 PCI device 00:14.0 write the parameter as:
1977 ivrs_ioapic[10]=0001:00:14.0
1979 ivrs_hpet [HW,X86_64]
1980 Provide an override to the HPET-ID<->DEVICE-ID
1981 mapping provided in the IVRS ACPI table.
1982 By default, PCI segment is 0, and can be omitted.
1984 For example, to map HPET-ID decimal 10 to
1985 PCI segment 0x1 and PCI device 00:14.0,
1986 write the parameter as:
1987 ivrs_hpet=10@0001:00:14.0
1990 * To map HPET-ID decimal 0 to PCI device 00:14.0
1991 write the parameter as:
1992 ivrs_hpet[0]=00:14.0
1993 * To map HPET-ID decimal 10 to PCI segment 0x1 and
1994 PCI device 00:14.0 write the parameter as:
1995 ivrs_ioapic[10]=0001:00:14.0
1997 ivrs_acpihid [HW,X86_64]
1998 Provide an override to the ACPI-HID:UID<->DEVICE-ID
1999 mapping provided in the IVRS ACPI table.
2000 By default, PCI segment is 0, and can be omitted.
2002 For example, to map UART-HID:UID AMD0020:0 to
2003 PCI segment 0x1 and PCI device ID 00:14.5,
2004 write the parameter as:
2005 ivrs_acpihid=AMD0020:0@0001:00:14.5
2008 * To map UART-HID:UID AMD0020:0 to PCI segment is 0,
2009 PCI device ID 00:14.5, write the parameter as:
2010 ivrs_acpihid[00:14.5]=AMD0020:0
2011 * To map UART-HID:UID AMD0020:0 to PCI segment 0x1 and
2012 PCI device ID 00:14.5, write the parameter as:
2013 ivrs_acpihid[0001:00:14.5]=AMD0020:0
2015 js= [HW,JOY] Analog joystick
2016 See Documentation/input/joydev/joystick.rst.
2019 When CONFIG_RANDOMIZE_BASE is set, this disables
2020 kernel and module base offset ASLR (Address Space
2021 Layout Randomization).
2024 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2025 report on every invalid memory access. Without this
2026 parameter KASAN will print report only for the first
2031 kernelcore= [KNL,X86,IA-64,PPC]
2032 Format: nn[KMGTPE] | nn% | "mirror"
2033 This parameter specifies the amount of memory usable by
2034 the kernel for non-movable allocations. The requested
2035 amount is spread evenly throughout all nodes in the
2036 system as ZONE_NORMAL. The remaining memory is used for
2037 movable memory in its own zone, ZONE_MOVABLE. In the
2038 event, a node is too small to have both ZONE_NORMAL and
2039 ZONE_MOVABLE, kernelcore memory will take priority and
2040 other nodes will have a larger ZONE_MOVABLE.
2042 ZONE_MOVABLE is used for the allocation of pages that
2043 may be reclaimed or moved by the page migration
2044 subsystem. Note that allocations like PTEs-from-HighMem
2045 still use the HighMem zone if it exists, and the Normal
2046 zone if it does not.
2048 It is possible to specify the exact amount of memory in
2049 the form of "nn[KMGTPE]", a percentage of total system
2050 memory in the form of "nn%", or "mirror". If "mirror"
2051 option is specified, mirrored (reliable) memory is used
2052 for non-movable allocations and remaining memory is used
2053 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2054 are exclusive, so you cannot specify multiple forms.
2056 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
2057 Format: <Controller#>[,poll interval]
2058 The controller # is the number of the ehci usb debug
2059 port as it is probed via PCI. The poll interval is
2060 optional and is the number seconds in between
2061 each poll cycle to the debug port in case you need
2062 the functionality for interrupting the kernel with
2063 gdb or control-c on the dbgp connection. When
2064 not using this parameter you use sysrq-g to break into
2065 the kernel debugger.
2067 kgdboc= [KGDB,HW] kgdb over consoles.
2068 Requires a tty driver that supports console polling,
2069 or a supported polling keyboard driver (non-usb).
2070 Serial only format: <serial_device>[,baud]
2071 keyboard only format: kbd
2072 keyboard and serial format: kbd,<serial_device>[,baud]
2073 Optional Kernel mode setting:
2074 kms, kbd format: kms,kbd
2075 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2077 kgdbwait [KGDB] Stop kernel execution and enter the
2078 kernel debugger at the earliest opportunity.
2080 kmac= [MIPS] korina ethernet MAC address.
2081 Configure the RouterBoard 532 series on-chip
2082 Ethernet adapter MAC address.
2084 kmemleak= [KNL] Boot-time kmemleak enable/disable
2085 Valid arguments: on, off
2087 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2090 kprobe_event=[probe-list]
2091 [FTRACE] Add kprobe events and enable at boot time.
2092 The probe-list is a semicolon delimited list of probe
2093 definitions. Each definition is same as kprobe_events
2094 interface, but the parameters are comma delimited.
2095 For example, to add a kprobe event on vfs_read with
2096 arg1 and arg2, add to the command line;
2098 kprobe_event=p,vfs_read,$arg1,$arg2
2100 See also Documentation/trace/kprobetrace.rst "Kernel
2101 Boot Parameter" section.
2103 kpti= [ARM64] Control page table isolation of user
2104 and kernel address spaces.
2105 Default: enabled on cores which need mitigation.
2109 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2110 Default is 0 (don't ignore, but inject #GP)
2112 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2113 Default is false (don't support).
2115 kvm.mmu_audit= [KVM] This is a R/W parameter which allows audit
2120 [KVM] Controls the software workaround for the
2121 X86_BUG_ITLB_MULTIHIT bug.
2122 force : Always deploy workaround.
2123 off : Never deploy workaround.
2124 auto : Deploy workaround based on the presence of
2125 X86_BUG_ITLB_MULTIHIT.
2129 If the software workaround is enabled for the host,
2130 guests do need not to enable it for nested guests.
2132 kvm.nx_huge_pages_recovery_ratio=
2133 [KVM] Controls how many 4KiB pages are periodically zapped
2134 back to huge pages. 0 disables the recovery, otherwise if
2135 the value is N KVM will zap 1/Nth of the 4KiB pages every
2136 minute. The default is 60.
2138 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM.
2139 Default is 1 (enabled)
2141 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU)
2143 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode.
2145 kvm-arm.vgic_v3_group0_trap=
2146 [KVM,ARM] Trap guest accesses to GICv3 group-0
2149 kvm-arm.vgic_v3_group1_trap=
2150 [KVM,ARM] Trap guest accesses to GICv3 group-1
2153 kvm-arm.vgic_v3_common_trap=
2154 [KVM,ARM] Trap guest accesses to GICv3 common
2157 kvm-arm.vgic_v4_enable=
2158 [KVM,ARM] Allow use of GICv4 for direct injection of
2161 kvm-intel.ept= [KVM,Intel] Disable extended page tables
2162 (virtualized MMU) support on capable Intel chips.
2163 Default is 1 (enabled)
2165 kvm-intel.emulate_invalid_guest_state=
2166 [KVM,Intel] Disable emulation of invalid guest state.
2167 Ignored if kvm-intel.enable_unrestricted_guest=1, as
2168 guest state is never invalid for unrestricted guests.
2169 This param doesn't apply to nested guests (L2), as KVM
2170 never emulates invalid L2 guest state.
2171 Default is 1 (enabled)
2173 kvm-intel.flexpriority=
2174 [KVM,Intel] Disable FlexPriority feature (TPR shadow).
2175 Default is 1 (enabled)
2178 [KVM,Intel] Enable VMX nesting (nVMX).
2179 Default is 0 (disabled)
2181 kvm-intel.unrestricted_guest=
2182 [KVM,Intel] Disable unrestricted guest feature
2183 (virtualized real and unpaged mode) on capable
2184 Intel chips. Default is 1 (enabled)
2186 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2189 Valid arguments: never, cond, always
2191 always: L1D cache flush on every VMENTER.
2192 cond: Flush L1D on VMENTER only when the code between
2193 VMEXIT and VMENTER can leak host memory.
2194 never: Disables the mitigation
2196 Default is cond (do L1 cache flush in specific instances)
2198 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification
2199 feature (tagged TLBs) on capable Intel chips.
2200 Default is 1 (enabled)
2202 l1tf= [X86] Control mitigation of the L1TF vulnerability on
2205 The kernel PTE inversion protection is unconditionally
2206 enabled and cannot be disabled.
2209 Provides all available mitigations for the
2210 L1TF vulnerability. Disables SMT and
2211 enables all mitigations in the
2212 hypervisors, i.e. unconditional L1D flush.
2214 SMT control and L1D flush control via the
2215 sysfs interface is still possible after
2216 boot. Hypervisors will issue a warning
2217 when the first VM is started in a
2218 potentially insecure configuration,
2219 i.e. SMT enabled or L1D flush disabled.
2222 Same as 'full', but disables SMT and L1D
2223 flush runtime control. Implies the
2224 'nosmt=force' command line option.
2225 (i.e. sysfs control of SMT is disabled.)
2228 Leaves SMT enabled and enables the default
2229 hypervisor mitigation, i.e. conditional
2232 SMT control and L1D flush control via the
2233 sysfs interface is still possible after
2234 boot. Hypervisors will issue a warning
2235 when the first VM is started in a
2236 potentially insecure configuration,
2237 i.e. SMT enabled or L1D flush disabled.
2241 Disables SMT and enables the default
2242 hypervisor mitigation.
2244 SMT control and L1D flush control via the
2245 sysfs interface is still possible after
2246 boot. Hypervisors will issue a warning
2247 when the first VM is started in a
2248 potentially insecure configuration,
2249 i.e. SMT enabled or L1D flush disabled.
2252 Same as 'flush', but hypervisors will not
2253 warn when a VM is started in a potentially
2254 insecure configuration.
2257 Disables hypervisor mitigations and doesn't
2259 It also drops the swap size and available
2260 RAM limit restriction on both hypervisor and
2265 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2271 lapic [X86-32,APIC] Enable the local APIC even if BIOS
2274 lapic= [x86,APIC] "notscdeadline" Do not use TSC deadline
2275 value for LAPIC timer one-shot implementation. Default
2276 back to the programmable timer unit in the LAPIC.
2278 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
2281 libata.dma= [LIBATA] DMA control
2282 libata.dma=0 Disable all PATA and SATA DMA
2283 libata.dma=1 PATA and SATA Disk DMA only
2284 libata.dma=2 ATAPI (CDROM) DMA only
2285 libata.dma=4 Compact Flash DMA only
2286 Combinations also work, so libata.dma=3 enables DMA
2287 for disks and CDROMs, but not CFs.
2289 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2290 libata.ignore_hpa=0 keep BIOS limits (default)
2291 libata.ignore_hpa=1 ignore limits, using full disk
2293 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2297 libata.force= [LIBATA] Force configurations. The format is comma
2298 separated list of "[ID:]VAL" where ID is
2299 PORT[.DEVICE]. PORT and DEVICE are decimal numbers
2300 matching port, link or device. Basically, it matches
2301 the ATA ID string printed on console by libata. If
2302 the whole ID part is omitted, the last PORT and DEVICE
2303 values are used. If ID hasn't been specified yet, the
2304 configuration applies to all ports, links and devices.
2306 If only DEVICE is omitted, the parameter applies to
2307 the port and all links and devices behind it. DEVICE
2308 number of 0 either selects the first device or the
2309 first fan-out link behind PMP device. It does not
2310 select the host link. DEVICE number of 15 selects the
2311 host link and device attached to it.
2313 The VAL specifies the configuration to force. As long
2314 as there's no ambiguity shortcut notation is allowed.
2315 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2316 The following configurations can be forced.
2318 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2319 Any ID with matching PORT is used.
2321 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2323 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2324 udma[/][16,25,33,44,66,100,133] notation is also
2327 * [no]ncq: Turn on or off NCQ.
2329 * [no]ncqtrim: Turn off queued DSM TRIM.
2331 * nohrst, nosrst, norst: suppress hard, soft
2334 * rstonce: only attempt one reset during
2335 hot-unplug link recovery
2337 * dump_id: dump IDENTIFY data.
2339 * atapi_dmadir: Enable ATAPI DMADIR bridge support
2341 * disable: Disable this device.
2343 If there are multiple matching configurations changing
2344 the same attribute, the last one is used.
2346 memblock=debug [KNL] Enable memblock debug messages.
2348 load_ramdisk= [RAM] List of ramdisks to load from floppy
2349 See Documentation/admin-guide/blockdev/ramdisk.rst.
2351 lockd.nlm_grace_period=P [NFS] Assign grace period.
2354 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2357 lockd.nlm_timeout=T [NFS] Assign timeout value.
2360 lockd.nlm_udpport=M [NFS] Assign UDP port.
2363 lockdown= [SECURITY]
2364 { integrity | confidentiality }
2365 Enable the kernel lockdown feature. If set to
2366 integrity, kernel features that allow userland to
2367 modify the running kernel are disabled. If set to
2368 confidentiality, kernel features that allow userland
2369 to extract confidential information from the kernel
2372 locktorture.nreaders_stress= [KNL]
2373 Set the number of locking read-acquisition kthreads.
2374 Defaults to being automatically set based on the
2375 number of online CPUs.
2377 locktorture.nwriters_stress= [KNL]
2378 Set the number of locking write-acquisition kthreads.
2380 locktorture.onoff_holdoff= [KNL]
2381 Set time (s) after boot for CPU-hotplug testing.
2383 locktorture.onoff_interval= [KNL]
2384 Set time (s) between CPU-hotplug operations, or
2385 zero to disable CPU-hotplug testing.
2387 locktorture.shuffle_interval= [KNL]
2388 Set task-shuffle interval (jiffies). Shuffling
2389 tasks allows some CPUs to go into dyntick-idle
2390 mode during the locktorture test.
2392 locktorture.shutdown_secs= [KNL]
2393 Set time (s) after boot system shutdown. This
2394 is useful for hands-off automated testing.
2396 locktorture.stat_interval= [KNL]
2397 Time (s) between statistics printk()s.
2399 locktorture.stutter= [KNL]
2400 Time (s) to stutter testing, for example,
2401 specifying five seconds causes the test to run for
2402 five seconds, wait for five seconds, and so on.
2403 This tests the locking primitive's ability to
2404 transition abruptly to and from idle.
2406 locktorture.torture_type= [KNL]
2407 Specify the locking implementation to test.
2409 locktorture.verbose= [KNL]
2410 Enable additional printk() statements.
2412 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
2415 loglevel= All Kernel Messages with a loglevel smaller than the
2416 console loglevel will be printed to the console. It can
2417 also be changed with klogd or other programs. The
2418 loglevels are defined as follows:
2420 0 (KERN_EMERG) system is unusable
2421 1 (KERN_ALERT) action must be taken immediately
2422 2 (KERN_CRIT) critical conditions
2423 3 (KERN_ERR) error conditions
2424 4 (KERN_WARNING) warning conditions
2425 5 (KERN_NOTICE) normal but significant condition
2426 6 (KERN_INFO) informational
2427 7 (KERN_DEBUG) debug-level messages
2429 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
2430 in bytes. n must be a power of two and greater
2431 than the minimal size. The minimal size is defined
2432 by LOG_BUF_SHIFT kernel config parameter. There is
2433 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2434 that allows to increase the default size depending on
2435 the number of CPUs. See init/Kconfig for more details.
2437 logo.nologo [FB] Disables display of the built-in Linux logo.
2438 This may be used to provide more screen space for
2439 kernel log messages and is useful when debugging
2440 kernel boot problems.
2442 lp=0 [LP] Specify parallel ports to use, e.g,
2443 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2444 lp=reset first parallel port). 'lp=0' disables the
2445 lp=auto printer driver. 'lp=reset' (which can be
2446 specified in addition to the ports) causes
2447 attached printers to be reset. Using
2448 lp=port1,port2,... specifies the parallel ports
2449 to associate lp devices with, starting with
2450 lp0. A port specification may be 'none' to skip
2451 that lp device, or a parport name such as
2452 'parport0'. Specifying 'lp=auto' instead of a
2453 port specification list means that device IDs
2454 from each port should be examined, to see if
2455 an IEEE 1284-compliant printer is attached; if
2456 so, the driver will manage that printer.
2457 See also header of drivers/char/lp.c.
2460 Sets loops_per_jiffy to given constant, thus avoiding
2461 time-consuming boot-time autodetection (up to 250 ms per
2462 CPU). 0 enables autodetection (default). To determine
2463 the correct value for your kernel, boot with normal
2464 autodetection and see what value is printed. Note that
2465 on SMP systems the preset will be applied to all CPUs,
2466 which is likely to cause problems if your CPUs need
2467 significantly divergent settings. An incorrect value
2468 will cause delays in the kernel to be wrong, leading to
2469 unpredictable I/O errors and other breakage. Although
2470 unlikely, in the extreme case this might damage your
2474 Format: <io>,<irq>,<dma>
2476 lsm.debug [SECURITY] Enable LSM initialization debugging output.
2479 [SECURITY] Choose order of LSM initialization. This
2480 overrides CONFIG_LSM, and the "security=" parameter.
2482 machvec= [IA-64] Force the use of a particular machine-vector
2483 (machvec) in a generic kernel.
2484 Example: machvec=hpzx1
2486 machtype= [Loongson] Share the same kernel image file between different
2488 Example: machtype=lemote-yeeloong-2f-7inch
2490 max_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory greater
2491 than or equal to this physical address is ignored.
2493 maxcpus= [SMP] Maximum number of processors that an SMP kernel
2494 will bring up during bootup. maxcpus=n : n >= 0 limits
2495 the kernel to bring up 'n' processors. Surely after
2496 bootup you can bring up the other plugged cpu by executing
2497 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
2498 only takes effect during system bootup.
2499 While n=0 is a special case, it is equivalent to "nosmp",
2500 which also disables the IO APIC.
2502 max_loop= [LOOP] The number of loop block devices that get
2503 (loop.max_loop) unconditionally pre-created at init time. The default
2504 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
2505 of statically allocating a predefined number, loop
2506 devices can be requested on-demand with the
2507 /dev/loop-control interface.
2509 mce [X86-32] Machine Check Exception
2511 mce=option [X86-64] See Documentation/x86/x86_64/boot-options.rst
2513 md= [HW] RAID subsystems devices and level
2514 See Documentation/admin-guide/md.rst.
2517 Format: <first>,<last>
2518 Specifies range of consoles to be captured by the MDA.
2521 Control mitigation for the Micro-architectural Data
2522 Sampling (MDS) vulnerability.
2524 Certain CPUs are vulnerable to an exploit against CPU
2525 internal buffers which can forward information to a
2526 disclosure gadget under certain conditions.
2528 In vulnerable processors, the speculatively
2529 forwarded data can be used in a cache side channel
2530 attack, to access data to which the attacker does
2531 not have direct access.
2533 This parameter controls the MDS mitigation. The
2536 full - Enable MDS mitigation on vulnerable CPUs
2537 full,nosmt - Enable MDS mitigation and disable
2538 SMT on vulnerable CPUs
2539 off - Unconditionally disable MDS mitigation
2541 On TAA-affected machines, mds=off can be prevented by
2542 an active TAA mitigation as both vulnerabilities are
2543 mitigated with the same mechanism so in order to disable
2544 this mitigation, you need to specify tsx_async_abort=off
2547 Not specifying this option is equivalent to
2550 For details see: Documentation/admin-guide/hw-vuln/mds.rst
2552 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
2553 Amount of memory to be used when the kernel is not able
2554 to see the whole system memory or for test.
2555 [X86] Work as limiting max address. Use together
2556 with memmap= to avoid physical address space collisions.
2557 Without memmap= PCI devices could be placed at addresses
2558 belonging to unused RAM.
2560 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
2564 [KNL,SH] Allow user to override the default size for
2565 per-device physically contiguous DMA buffers.
2567 memhp_default_state=online/offline
2568 [KNL] Set the initial state for the memory hotplug
2569 onlining policy. If not specified, the default value is
2570 set according to the
2571 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
2573 See Documentation/admin-guide/mm/memory-hotplug.rst.
2575 memmap=exactmap [KNL,X86] Enable setting of an exact
2576 E820 memory map, as specified by the user.
2577 Such memmap=exactmap lines can be constructed based on
2578 BIOS output or other requirements. See the memmap=nn@ss
2581 memmap=nn[KMG]@ss[KMG]
2582 [KNL] Force usage of a specific region of memory.
2583 Region of memory to be used is from ss to ss+nn.
2584 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
2585 which limits max address to nn[KMG].
2586 Multiple different regions can be specified,
2589 memmap=100M@2G,100M#3G,1G!1024G
2591 memmap=nn[KMG]#ss[KMG]
2592 [KNL,ACPI] Mark specific memory as ACPI data.
2593 Region of memory to be marked is from ss to ss+nn.
2595 memmap=nn[KMG]$ss[KMG]
2596 [KNL,ACPI] Mark specific memory as reserved.
2597 Region of memory to be reserved is from ss to ss+nn.
2598 Example: Exclude memory from 0x18690000-0x1869ffff
2599 memmap=64K$0x18690000
2601 memmap=0x10000$0x18690000
2602 Some bootloaders may need an escape character before '$',
2603 like Grub2, otherwise '$' and the following number
2606 memmap=nn[KMG]!ss[KMG]
2607 [KNL,X86] Mark specific memory as protected.
2608 Region of memory to be used, from ss to ss+nn.
2609 The memory region may be marked as e820 type 12 (0xc)
2610 and is NVDIMM or ADR memory.
2612 memmap=<size>%<offset>-<oldtype>+<newtype>
2613 [KNL,ACPI] Convert memory within the specified region
2614 from <oldtype> to <newtype>. If "-<oldtype>" is left
2615 out, the whole region will be marked as <newtype>,
2616 even if previously unavailable. If "+<newtype>" is left
2617 out, matching memory will be removed. Types are
2618 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
2619 3 = ACPI, 12 = PRAM.
2621 memory_corruption_check=0/1 [X86]
2622 Some BIOSes seem to corrupt the first 64k of
2623 memory when doing things like suspend/resume.
2624 Setting this option will scan the memory
2625 looking for corruption. Enabling this will
2626 both detect corruption and prevent the kernel
2627 from using the memory being corrupted.
2628 However, its intended as a diagnostic tool; if
2629 repeatable BIOS-originated corruption always
2630 affects the same memory, you can use memmap=
2631 to prevent the kernel from using that memory.
2633 memory_corruption_check_size=size [X86]
2634 By default it checks for corruption in the low
2635 64k, making this memory unavailable for normal
2636 use. Use this parameter to scan for
2637 corruption in more or less memory.
2639 memory_corruption_check_period=seconds [X86]
2640 By default it checks for corruption every 60
2641 seconds. Use this parameter to check at some
2642 other rate. 0 disables periodic checking.
2644 memtest= [KNL,X86,ARM,PPC] Enable memtest
2646 default : 0 <disable>
2647 Specifies the number of memtest passes to be
2648 performed. Each pass selects another test
2649 pattern from a given set of patterns. Memtest
2650 fills the memory with this pattern, validates
2651 memory contents and reserves bad memory
2652 regions that are detected.
2654 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
2655 Valid arguments: on, off
2656 Default (depends on kernel configuration option):
2657 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
2658 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
2659 mem_encrypt=on: Activate SME
2660 mem_encrypt=off: Do not activate SME
2662 Refer to Documentation/virt/kvm/amd-memory-encryption.rst
2663 for details on when memory encryption can be activated.
2665 mem_sleep_default= [SUSPEND] Default system suspend mode:
2666 s2idle - Suspend-To-Idle
2667 shallow - Power-On Suspend or equivalent (if supported)
2668 deep - Suspend-To-RAM or equivalent (if supported)
2669 See Documentation/admin-guide/pm/sleep-states.rst.
2671 meye.*= [HW] Set MotionEye Camera parameters
2672 See Documentation/media/v4l-drivers/meye.rst.
2674 mfgpt_irq= [IA-32] Specify the IRQ to use for the
2675 Multi-Function General Purpose Timers on AMD Geode
2678 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
2679 the BIOS has incorrectly applied a workaround. TinyBIOS
2680 version 0.98 is known to be affected, 0.99 fixes the
2681 problem by letting the user disable the workaround.
2685 min_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory below this
2686 physical address is ignored.
2688 mini2440= [ARM,HW,KNL]
2689 Format:[0..2][b][c][t]
2691 MINI2440 configuration specification:
2692 0 - The attached screen is the 3.5" TFT
2693 1 - The attached screen is the 7" TFT
2694 2 - The VGA Shield is attached (1024x768)
2695 Leaving out the screen size parameter will not load
2696 the TFT driver, and the framebuffer will be left
2698 b - Enable backlight. The TFT backlight pin will be
2699 linked to the kernel VESA blanking code and a GPIO
2700 LED. This parameter is not necessary when using the
2702 c - Enable the s3c camera interface.
2703 t - Reserved for enabling touchscreen support. The
2704 touchscreen support is not enabled in the mainstream
2705 kernel as of 2.6.30, a preliminary port can be found
2706 in the "bleeding edge" mini2440 support kernel at
2707 http://repo.or.cz/w/linux-2.6/mini2440.git
2710 [X86,PPC,S390,ARM64] Control optional mitigations for
2711 CPU vulnerabilities. This is a set of curated,
2712 arch-independent options, each of which is an
2713 aggregation of existing arch-specific options.
2716 Disable all optional CPU mitigations. This
2717 improves system performance, but it may also
2718 expose users to several CPU vulnerabilities.
2719 Equivalent to: gather_data_sampling=off [X86]
2721 kvm.nx_huge_pages=off [X86]
2724 mmio_stale_data=off [X86]
2725 no_entry_flush [PPC]
2726 no_uaccess_flush [PPC]
2729 nospectre_v1 [X86,PPC]
2730 nospectre_v2 [X86,PPC,S390,ARM64]
2731 spec_store_bypass_disable=off [X86,PPC]
2732 spectre_v2_user=off [X86]
2733 ssbd=force-off [ARM64]
2734 tsx_async_abort=off [X86]
2737 This does not have any effect on
2738 kvm.nx_huge_pages when
2739 kvm.nx_huge_pages=force.
2742 Mitigate all CPU vulnerabilities, but leave SMT
2743 enabled, even if it's vulnerable. This is for
2744 users who don't want to be surprised by SMT
2745 getting disabled across kernel upgrades, or who
2746 have other ways of avoiding SMT-based attacks.
2747 Equivalent to: (default behavior)
2750 Mitigate all CPU vulnerabilities, disabling SMT
2751 if needed. This is for users who always want to
2752 be fully mitigated, even if it means losing SMT.
2753 Equivalent to: l1tf=flush,nosmt [X86]
2754 mds=full,nosmt [X86]
2755 tsx_async_abort=full,nosmt [X86]
2756 mmio_stale_data=full,nosmt [X86]
2759 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
2760 parameter allows control of the logging verbosity for
2761 the additional memory initialisation checks. A value
2762 of 0 disables mminit logging and a level of 4 will
2763 log everything. Information is printed at KERN_DEBUG
2764 so loglevel=8 may also need to be specified.
2767 [X86,INTEL] Control mitigation for the Processor
2768 MMIO Stale Data vulnerabilities.
2770 Processor MMIO Stale Data is a class of
2771 vulnerabilities that may expose data after an MMIO
2772 operation. Exposed data could originate or end in
2773 the same CPU buffers as affected by MDS and TAA.
2774 Therefore, similar to MDS and TAA, the mitigation
2775 is to clear the affected CPU buffers.
2777 This parameter controls the mitigation. The
2780 full - Enable mitigation on vulnerable CPUs
2782 full,nosmt - Enable mitigation and disable SMT on
2785 off - Unconditionally disable mitigation
2787 On MDS or TAA affected machines,
2788 mmio_stale_data=off can be prevented by an active
2789 MDS or TAA mitigation as these vulnerabilities are
2790 mitigated with the same mechanism so in order to
2791 disable this mitigation, you need to specify
2792 mds=off and tsx_async_abort=off too.
2794 Not specifying this option is equivalent to
2795 mmio_stale_data=full.
2798 Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst
2801 [KNL] When CONFIG_MODULE_SIG is set, this means that
2802 modules without (valid) signatures will fail to load.
2803 Note that if CONFIG_MODULE_SIG_FORCE is set, that
2804 is always true, so this option does nothing.
2806 module_blacklist= [KNL] Do not load a comma-separated list of
2807 modules. Useful for debugging problem modules.
2810 [MOUSE] Maximum time between finger touching and
2811 leaving touchpad surface for touch to be considered
2812 a tap and be reported as a left button click (for
2813 touchpads working in absolute mode only).
2815 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
2816 reporting absolute coordinates, such as tablets
2817 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
2818 reporting absolute coordinates, such as tablets
2820 movablecore= [KNL,X86,IA-64,PPC]
2821 Format: nn[KMGTPE] | nn%
2822 This parameter is the complement to kernelcore=, it
2823 specifies the amount of memory used for migratable
2824 allocations. If both kernelcore and movablecore is
2825 specified, then kernelcore will be at *least* the
2826 specified value but may be more. If movablecore on its
2827 own is specified, the administrator must be careful
2828 that the amount of memory usable for all allocations
2831 movable_node [KNL] Boot-time switch to make hotplugable memory
2832 NUMA nodes to be movable. This means that the memory
2833 of such nodes will be usable only for movable
2834 allocations which rules out almost all kernel
2835 allocations. Use with caution!
2837 MTD_Partition= [MTD]
2838 Format: <name>,<region-number>,<size>,<offset>
2840 MTD_Region= [MTD] Format:
2841 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
2844 See drivers/mtd/parsers/cmdlinepart.c
2846 multitce=off [PPC] This parameter disables the use of the pSeries
2847 firmware feature for updating multiple TCE entries
2850 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
2852 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
2854 boundary - index of last SLC block on Flex-OneNAND.
2855 The remaining blocks are configured as MLC blocks.
2856 lock - Configure if Flex-OneNAND boundary should be locked.
2857 Once locked, the boundary cannot be changed.
2858 1 indicates lock status, 0 indicates unlock status.
2861 ARM/S3C2412 JIVE boot control
2863 See arch/arm/mach-s3c2412/mach-jive.c
2865 mtouchusb.raw_coordinates=
2866 [HW] Make the MicroTouch USB driver use raw coordinates
2867 ('y', default) or cooked coordinates ('n')
2869 mtrr_chunk_size=nn[KMG] [X86]
2870 used for mtrr cleanup. It is largest continuous chunk
2871 that could hold holes aka. UC entries.
2873 mtrr_gran_size=nn[KMG] [X86]
2874 Used for mtrr cleanup. It is granularity of mtrr block.
2876 Large value could prevent small alignment from
2879 mtrr_spare_reg_nr=n [X86]
2881 Range: 0,7 : spare reg number
2883 Used for mtrr cleanup. It is spare mtrr entries number.
2884 Set to 2 or more if your graphical card needs more.
2886 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
2888 netdev= [NET] Network devices parameters
2889 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
2890 Note that mem_start is often overloaded to mean
2891 something different and driver-specific.
2892 This usage is only documented in each driver source
2896 [NETFILTER] Enable connection tracking flow accounting
2897 0 to disable accounting
2898 1 to enable accounting
2901 nfsaddrs= [NFS] Deprecated. Use ip= instead.
2902 See Documentation/filesystems/nfs/nfsroot.txt.
2904 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
2905 See Documentation/filesystems/nfs/nfsroot.txt.
2907 nfsrootdebug [NFS] enable nfsroot debugging messages.
2908 See Documentation/filesystems/nfs/nfsroot.txt.
2910 nfs.callback_nr_threads=
2911 [NFSv4] set the total number of threads that the
2912 NFS client will assign to service NFSv4 callback
2915 nfs.callback_tcpport=
2916 [NFS] set the TCP port on which the NFSv4 callback
2917 channel should listen.
2920 [NFS] sets the pathname to the program which is used
2921 to update the NFS client cache entries.
2923 nfs.cache_getent_timeout=
2924 [NFS] sets the timeout after which an attempt to
2925 update a cache entry is deemed to have failed.
2927 nfs.idmap_cache_timeout=
2928 [NFS] set the maximum lifetime for idmapper cache
2932 [NFS] enable 64-bit inode numbers.
2933 If zero, the NFS client will fake up a 32-bit inode
2934 number for the readdir() and stat() syscalls instead
2935 of returning the full 64-bit number.
2936 The default is to return 64-bit inode numbers.
2938 nfs.max_session_cb_slots=
2939 [NFSv4.1] Sets the maximum number of session
2940 slots the client will assign to the callback
2941 channel. This determines the maximum number of
2942 callbacks the client will process in parallel for
2943 a particular server.
2945 nfs.max_session_slots=
2946 [NFSv4.1] Sets the maximum number of session slots
2947 the client will attempt to negotiate with the server.
2948 This limits the number of simultaneous RPC requests
2949 that the client can send to the NFSv4.1 server.
2950 Note that there is little point in setting this
2951 value higher than the max_tcp_slot_table_limit.
2953 nfs.nfs4_disable_idmapping=
2954 [NFSv4] When set to the default of '1', this option
2955 ensures that both the RPC level authentication
2956 scheme and the NFS level operations agree to use
2957 numeric uids/gids if the mount is using the
2958 'sec=sys' security flavour. In effect it is
2959 disabling idmapping, which can make migration from
2960 legacy NFSv2/v3 systems to NFSv4 easier.
2961 Servers that do not support this mode of operation
2962 will be autodetected by the client, and it will fall
2963 back to using the idmapper.
2964 To turn off this behaviour, set the value to '0'.
2966 [NFS4] Specify an additional fixed unique ident-
2967 ification string that NFSv4 clients can insert into
2968 their nfs_client_id4 string. This is typically a
2969 UUID that is generated at system install time.
2971 nfs.send_implementation_id =
2972 [NFSv4.1] Send client implementation identification
2973 information in exchange_id requests.
2974 If zero, no implementation identification information
2976 The default is to send the implementation identification
2979 nfs.recover_lost_locks =
2980 [NFSv4] Attempt to recover locks that were lost due
2981 to a lease timeout on the server. Please note that
2982 doing this risks data corruption, since there are
2983 no guarantees that the file will remain unchanged
2984 after the locks are lost.
2985 If you want to enable the kernel legacy behaviour of
2986 attempting to recover these locks, then set this
2988 The default parameter value of '0' causes the kernel
2989 not to attempt recovery of lost locks.
2991 nfs4.layoutstats_timer =
2992 [NFSv4.2] Change the rate at which the kernel sends
2993 layoutstats to the pNFS metadata server.
2995 Setting this to value to 0 causes the kernel to use
2996 whatever value is the default set by the layout
2997 driver. A non-zero value sets the minimum interval
2998 in seconds between layoutstats transmissions.
3000 nfsd.nfs4_disable_idmapping=
3001 [NFSv4] When set to the default of '1', the NFSv4
3002 server will return only numeric uids and gids to
3003 clients using auth_sys, and will accept numeric uids
3004 and gids from such clients. This is intended to ease
3005 migration from NFSv2/v3.
3007 nmi_debug= [KNL,SH] Specify one or more actions to take
3008 when a NMI is triggered.
3009 Format: [state][,regs][,debounce][,die]
3011 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
3012 Format: [panic,][nopanic,][num]
3014 0 - turn hardlockup detector in nmi_watchdog off
3015 1 - turn hardlockup detector in nmi_watchdog on
3016 When panic is specified, panic when an NMI watchdog
3017 timeout occurs (or 'nopanic' to not panic on an NMI
3018 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
3019 To disable both hard and soft lockup detectors,
3020 please see 'nowatchdog'.
3021 This is useful when you use a panic=... timeout and
3022 need the box quickly up again.
3024 These settings can be accessed at runtime via
3025 the nmi_watchdog and hardlockup_panic sysctls.
3027 netpoll.carrier_timeout=
3028 [NET] Specifies amount of time (in seconds) that
3029 netpoll should wait for a carrier. By default netpoll
3032 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3033 emulation library even if a 387 maths coprocessor
3036 no5lvl [X86-64] Disable 5-level paging mode. Forces
3037 kernel to use 4-level paging instead.
3040 [HW] Never suspend the console
3041 Disable suspending of consoles during suspend and
3042 hibernate operations. Once disabled, debugging
3043 messages can reach various consoles while the rest
3044 of the system is being put to sleep (ie, while
3045 debugging driver suspend/resume hooks). This may
3046 not work reliably with all consoles, but is known
3047 to work with serial and VGA consoles.
3048 To facilitate more flexible debugging, we also add
3049 console_suspend, a printk module parameter to control
3050 it. Users could use console_suspend (usually
3051 /sys/module/printk/parameters/console_suspend) to
3052 turn on/off it dynamically.
3054 novmcoredd [KNL,KDUMP]
3055 Disable device dump. Device dump allows drivers to
3056 append dump data to vmcore so you can collect driver
3057 specified debug info. Drivers can append the data
3058 without any limit and this data is stored in memory,
3059 so this may cause significant memory stress. Disabling
3060 device dump can help save memory but the driver debug
3061 data will be no longer available. This parameter
3062 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
3065 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
3066 caches in the slab allocator. Saves per-node memory,
3067 but will impact performance.
3071 noaltinstr [S390] Disables alternative instructions patching
3072 (CPU alternatives feature).
3074 noapic [SMP,APIC] Tells the kernel to not make use of any
3075 IOAPICs that may be present in the system.
3077 noautogroup Disable scheduler automatic task group creation.
3079 nobats [PPC] Do not use BATs for mapping kernel lowmem
3080 on "Classic" PPC cores.
3084 noclflush [BUGS=X86] Don't use the CLFLUSH instruction
3086 nodelayacct [KNL] Disable per-task delay accounting
3088 nodsp [SH] Disable hardware DSP at boot time.
3090 noefi Disable EFI runtime services support.
3092 no_entry_flush [PPC] Don't flush the L1-D cache when entering the kernel.
3097 On X86-32 available only on PAE configured kernels.
3098 noexec=on: enable non-executable mappings (default)
3099 noexec=off: disable non-executable mappings
3102 Disable SMAP (Supervisor Mode Access Prevention)
3103 even if it is supported by processor.
3106 Disable SMEP (Supervisor Mode Execution Prevention)
3107 even if it is supported by processor.
3110 This affects only 32-bit executables.
3111 noexec32=on: enable non-executable mappings (default)
3112 read doesn't imply executable mappings
3113 noexec32=off: disable non-executable mappings
3114 read implies executable mappings
3116 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3118 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3119 register save and restore. The kernel will only save
3120 legacy floating-point registers on task switch.
3122 nohugeiomap [KNL,x86,PPC] Disable kernel huge I/O mappings.
3124 nosmt [KNL,S390] Disable symmetric multithreading (SMT).
3125 Equivalent to smt=1.
3127 [KNL,x86] Disable symmetric multithreading (SMT).
3128 nosmt=force: Force disable SMT, cannot be undone
3129 via the sysfs control file.
3131 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1
3132 (bounds check bypass). With this option data leaks are
3133 possible in the system.
3135 nospectre_v2 [X86,PPC_FSL_BOOK3E,ARM64] Disable all mitigations for
3136 the Spectre variant 2 (indirect branch prediction)
3137 vulnerability. System may allow data leaks with this
3140 nospec_store_bypass_disable
3141 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
3144 [PPC] Don't flush the L1-D cache after accessing user data.
3146 noxsave [BUGS=X86] Disables x86 extended register state save
3147 and restore using xsave. The kernel will fallback to
3148 enabling legacy floating-point and sse state.
3150 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
3151 register states. The kernel will fall back to use
3152 xsave to save the states. By using this parameter,
3153 performance of saving the states is degraded because
3154 xsave doesn't support modified optimization while
3155 xsaveopt supports it on xsaveopt enabled systems.
3157 noxsaves [X86] Disables xsaves and xrstors used in saving and
3158 restoring x86 extended register state in compacted
3159 form of xsave area. The kernel will fall back to use
3160 xsaveopt and xrstor to save and restore the states
3161 in standard form of xsave area. By using this
3162 parameter, xsave area per process might occupy more
3163 memory on xsaves enabled systems.
3165 nohlt [BUGS=ARM,SH] Tells the kernel that the sleep(SH) or
3166 wfi(ARM) instruction doesn't work correctly and not to
3167 use it. This is also useful when using JTAG debugger.
3169 no_file_caps Tells the kernel not to honor file capabilities. The
3170 only way then for a file to be executed with privilege
3171 is to be setuid root or executed by root.
3173 nohalt [IA-64] Tells the kernel not to use the power saving
3174 function PAL_HALT_LIGHT when idle. This increases
3175 power-consumption. On the positive side, it reduces
3176 interrupt wake-up latency, which may improve performance
3177 in certain environments such as networked servers or
3180 nohibernate [HIBERNATION] Disable hibernation and resume.
3182 nohz= [KNL] Boottime enable/disable dynamic ticks
3183 Valid arguments: on, off
3186 nohz_full= [KNL,BOOT,SMP,ISOL]
3187 The argument is a cpu list, as described above.
3188 In kernels built with CONFIG_NO_HZ_FULL=y, set
3189 the specified list of CPUs whose tick will be stopped
3190 whenever possible. The boot CPU will be forced outside
3191 the range to maintain the timekeeping. Any CPUs
3192 in this list will have their RCU callbacks offloaded,
3193 just as if they had also been called out in the
3194 rcu_nocbs= boot parameter.
3196 noiotrap [SH] Disables trapped I/O port accesses.
3198 noirqdebug [X86-32] Disables the code which attempts to detect and
3199 disable unhandled interrupt sources.
3201 no_timer_check [X86,APIC] Disables the code which tests for
3202 broken timer IRQ sources.
3204 noisapnp [ISAPNP] Disables ISA PnP code.
3206 noinitrd [RAM] Tells the kernel not to load any configured
3209 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
3211 [Deprecated - use intremap=off]
3215 noinvpcid [X86] Disable the INVPCID cpu feature.
3217 nojitter [IA-64] Disables jitter checking for ITC timers.
3219 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
3221 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
3225 [X86,PV_OPS] Disable paravirtualized VMware scheduler
3226 clock and use the default one.
3228 no-steal-acc [X86,KVM] Disable paravirtualized steal time accounting.
3229 steal time is computed, but won't influence scheduler
3232 nolapic [X86-32,APIC] Do not enable or use the local APIC.
3234 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
3236 noltlbs [PPC] Do not use large page/tlb entries for kernel
3237 lowmem mapping on PPC40x and PPC8xx
3239 nomca [IA-64] Disable machine check abort handling
3241 nomce [X86-32] Disable Machine Check Exception
3243 nomfgpt [X86-32] Disable Multi-Function General Purpose
3244 Timer usage (for AMD Geode machines).
3246 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3247 shutdown the other cpus. Instead use the REBOOT_VECTOR
3250 nomodule Disable module load
3252 nopat [X86] Disable PAT (page attribute table extension of
3253 pagetables) support.
3255 nopcid [X86-64] Disable the PCID cpu feature.
3257 norandmaps Don't use address space randomization. Equivalent to
3258 echo 0 > /proc/sys/kernel/randomize_va_space
3260 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
3261 with UP alternatives
3263 nordrand [X86] Disable kernel use of the RDRAND and
3264 RDSEED instructions even if they are supported
3265 by the processor. RDRAND and RDSEED are still
3266 available to user space applications.
3268 noresume [SWSUSP] Disables resume and restores original swap
3271 no-scroll [VGA] Disables scrollback.
3272 This is required for the Braillex ib80-piezo Braille
3273 reader made by F.H. Papenmeier (Germany).
3277 nosep [BUGS=X86-32] Disables x86 SYSENTER/SYSEXIT support.
3279 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
3280 and disable the IO APIC. legacy for "maxcpus=0".
3282 nosoftlockup [KNL] Disable the soft-lockup detector.
3284 nosync [HW,M68K] Disables sync negotiation for all devices.
3286 nowatchdog [KNL] Disable both lockup detectors, i.e.
3287 soft-lockup and NMI watchdog (hard-lockup).
3291 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
3293 cpu0_hotplug [X86] Turn on CPU0 hotplug feature when
3294 CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off.
3295 Some features depend on CPU0. Known dependencies are:
3296 1. Resume from suspend/hibernate depends on CPU0.
3297 Suspend/hibernate will fail if CPU0 is offline and you
3298 need to online CPU0 before suspend/hibernate.
3299 2. PIC interrupts also depend on CPU0. CPU0 can't be
3300 removed if a PIC interrupt is detected.
3301 It's said poweroff/reboot may depend on CPU0 on some
3302 machines although I haven't seen such issues so far
3303 after CPU0 is offline on a few tested machines.
3304 If the dependencies are under your control, you can
3305 turn on cpu0_hotplug.
3307 nps_mtm_hs_ctr= [KNL,ARC]
3308 This parameter sets the maximum duration, in
3309 cycles, each HW thread of the CTOP can run
3310 without interruptions, before HW switches it.
3311 The actual maximum duration is 16 times this
3313 Format: integer between 1 and 255
3316 nptcg= [IA-64] Override max number of concurrent global TLB
3317 purges which is reported from either PAL_VM_SUMMARY or
3320 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
3321 could support. nr_cpus=n : n >= 1 limits the kernel to
3322 support 'n' processors. It could be larger than the
3323 number of already plugged CPU during bootup, later in
3324 runtime you can physically add extra cpu until it reaches
3325 n. So during boot up some boot time memory for per-cpu
3326 variables need be pre-allocated for later physical cpu
3329 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
3331 numa_balancing= [KNL,X86] Enable or disable automatic NUMA balancing.
3332 Allowed values are enable and disable
3334 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
3335 'node', 'default' can be specified
3336 This can be set from sysctl after boot.
3337 See Documentation/admin-guide/sysctl/vm.rst for details.
3339 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
3340 See Documentation/debugging-via-ohci1394.txt for more
3343 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
3344 Rather than timing out after 20 ms if an EC
3345 command is not properly ACKed, override the length
3346 of the timeout. We have interrupts disabled while
3347 waiting for the ACK, so if this is set too high
3348 interrupts *may* be lost!
3350 omap_mux= [OMAP] Override bootloader pin multiplexing.
3351 Format: <mux_mode0.mode_name=value>...
3352 For example, to override I2C bus2:
3353 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
3355 oprofile.timer= [HW]
3356 Use timer interrupt instead of performance counters
3358 oprofile.cpu_type= Force an oprofile cpu type
3359 This might be useful if you have an older oprofile
3360 userland or if you want common events.
3361 Format: { arch_perfmon }
3362 arch_perfmon: [X86] Force use of architectural
3363 perfmon on Intel CPUs instead of the
3364 CPU specific event set.
3365 timer: [X86] Force use of architectural NMI
3366 timer mode (see also oprofile.timer
3367 for generic hr timer mode)
3369 oops=panic Always panic on oopses. Default is to just kill the
3370 process, but there is a small probability of
3371 deadlocking the machine.
3372 This will also cause panics on machine check exceptions.
3373 Useful together with panic=30 to trigger a reboot.
3376 [KNL] Boolean flag to control whether the page allocator
3377 should randomize its free lists. The randomization may
3378 be automatically enabled if the kernel detects it is
3379 running on a platform with a direct-mapped memory-side
3380 cache, and this parameter can be used to
3381 override/disable that behavior. The state of the flag
3382 can be read from sysfs at:
3383 /sys/module/page_alloc/parameters/shuffle.
3385 page_owner= [KNL] Boot-time page_owner enabling option.
3386 Storage of the information about who allocated
3387 each page is disabled in default. With this switch,
3389 on: enable the feature
3391 page_poison= [KNL] Boot-time parameter changing the state of
3392 poisoning on the buddy allocator, available with
3393 CONFIG_PAGE_POISONING=y.
3394 off: turn off poisoning (default)
3395 on: turn on poisoning
3397 panic= [KNL] Kernel behaviour on panic: delay <timeout>
3398 timeout > 0: seconds before rebooting
3399 timeout = 0: wait forever
3400 timeout < 0: reboot immediately
3403 panic_print= Bitmask for printing system info when panic happens.
3404 User can chose combination of the following bits:
3405 bit 0: print all tasks info
3406 bit 1: print system memory info
3407 bit 2: print timer info
3408 bit 3: print locks info if CONFIG_LOCKDEP is on
3409 bit 4: print ftrace buffer
3410 bit 5: print all printk messages in buffer
3412 panic_on_warn panic() instead of WARN(). Useful to cause kdump
3415 crash_kexec_post_notifiers
3416 Run kdump after running panic-notifiers and dumping
3417 kmsg. This only for the users who doubt kdump always
3418 succeeds in any situation.
3419 Note that this also increases risks of kdump failure,
3420 because some panic notifiers can make the crashed
3421 kernel more unstable.
3423 parkbd.port= [HW] Parallel port number the keyboard adapter is
3424 connected to, default is 0.
3426 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
3427 0 for XT, 1 for AT (default is AT).
3430 parport= [HW,PPT] Specify parallel ports. 0 disables.
3431 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
3432 Use 'auto' to force the driver to use any
3433 IRQ/DMA settings detected (the default is to
3434 ignore detected IRQ/DMA settings because of
3435 possible conflicts). You can specify the base
3436 address, IRQ, and DMA settings; IRQ and DMA
3437 should be numbers, or 'auto' (for using detected
3438 settings on that particular port), or 'nofifo'
3439 (to avoid using a FIFO even if it is detected).
3440 Parallel ports are assigned in the order they
3441 are specified on the command line, starting
3444 parport_init_mode= [HW,PPT]
3445 Configure VIA parallel port to operate in
3446 a specific mode. This is necessary on Pegasos
3447 computer where firmware has no options for setting
3448 up parallel port mode and sets it to spp.
3449 Currently this function knows 686a and 8231 chips.
3450 Format: [spp|ps2|epp|ecp|ecpepp]
3453 Halt all CPUs after the first oops has been printed for
3454 the specified number of seconds. This is to be used if
3455 your oopses keep scrolling off the screen.
3460 See header of drivers/block/paride/pcd.c.
3461 See also Documentation/admin-guide/blockdev/paride.rst.
3463 pci=option[,option...] [PCI] various PCI subsystem options.
3465 Some options herein operate on a specific device
3466 or a set of devices (<pci_dev>). These are
3467 specified in one of the following formats:
3469 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
3470 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
3472 Note: the first format specifies a PCI
3473 bus/device/function address which may change
3474 if new hardware is inserted, if motherboard
3475 firmware changes, or due to changes caused
3476 by other kernel parameters. If the
3477 domain is left unspecified, it is
3478 taken to be zero. Optionally, a path
3479 to a device through multiple device/function
3480 addresses can be specified after the base
3481 address (this is more robust against
3482 renumbering issues). The second format
3483 selects devices using IDs from the
3484 configuration space which may match multiple
3485 devices in the system.
3487 earlydump dump PCI config space before the kernel
3489 off [X86] don't probe for the PCI bus
3490 bios [X86-32] force use of PCI BIOS, don't access
3491 the hardware directly. Use this if your machine
3492 has a non-standard PCI host bridge.
3493 nobios [X86-32] disallow use of PCI BIOS, only direct
3494 hardware access methods are allowed. Use this
3495 if you experience crashes upon bootup and you
3496 suspect they are caused by the BIOS.
3497 conf1 [X86] Force use of PCI Configuration Access
3498 Mechanism 1 (config address in IO port 0xCF8,
3499 data in IO port 0xCFC, both 32-bit).
3500 conf2 [X86] Force use of PCI Configuration Access
3501 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
3502 the function, IO port 0xCFA, also 8-bit, sets
3503 bus number. The config space is then accessed
3504 through ports 0xC000-0xCFFF).
3505 See http://wiki.osdev.org/PCI for more info
3506 on the configuration access mechanisms.
3507 noaer [PCIE] If the PCIEAER kernel config parameter is
3508 enabled, this kernel boot option can be used to
3509 disable the use of PCIE advanced error reporting.
3510 nodomains [PCI] Disable support for multiple PCI
3511 root domains (aka PCI segments, in ACPI-speak).
3512 nommconf [X86] Disable use of MMCONFIG for PCI
3514 check_enable_amd_mmconf [X86] check for and enable
3515 properly configured MMIO access to PCI
3516 config space on AMD family 10h CPU
3517 nomsi [MSI] If the PCI_MSI kernel config parameter is
3518 enabled, this kernel boot option can be used to
3519 disable the use of MSI interrupts system-wide.
3520 noioapicquirk [APIC] Disable all boot interrupt quirks.
3521 Safety option to keep boot IRQs enabled. This
3522 should never be necessary.
3523 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
3524 primary IO-APIC for bridges that cannot disable
3525 boot IRQs. This fixes a source of spurious IRQs
3526 when the system masks IRQs.
3527 noioapicreroute [APIC] Disable workaround that uses the
3528 boot IRQ equivalent of an IRQ that connects to
3529 a chipset where boot IRQs cannot be disabled.
3530 The opposite of ioapicreroute.
3531 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
3532 routing table. These calls are known to be buggy
3533 on several machines and they hang the machine
3534 when used, but on other computers it's the only
3535 way to get the interrupt routing table. Try
3536 this option if the kernel is unable to allocate
3537 IRQs or discover secondary PCI buses on your
3539 rom [X86] Assign address space to expansion ROMs.
3540 Use with caution as certain devices share
3541 address decoders between ROMs and other
3543 norom [X86] Do not assign address space to
3544 expansion ROMs that do not already have
3545 BIOS assigned address ranges.
3546 nobar [X86] Do not assign address space to the
3547 BARs that weren't assigned by the BIOS.
3548 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
3549 assigned automatically to PCI devices. You can
3550 make the kernel exclude IRQs of your ISA cards
3552 pirqaddr=0xAAAAA [X86] Specify the physical address
3553 of the PIRQ table (normally generated
3554 by the BIOS) if it is outside the
3555 F0000h-100000h range.
3556 lastbus=N [X86] Scan all buses thru bus #N. Can be
3557 useful if the kernel is unable to find your
3558 secondary buses and you want to tell it
3559 explicitly which ones they are.
3560 assign-busses [X86] Always assign all PCI bus
3561 numbers ourselves, overriding
3562 whatever the firmware may have done.
3563 usepirqmask [X86] Honor the possible IRQ mask stored
3564 in the BIOS $PIR table. This is needed on
3565 some systems with broken BIOSes, notably
3566 some HP Pavilion N5400 and Omnibook XE3
3567 notebooks. This will have no effect if ACPI
3568 IRQ routing is enabled.
3569 noacpi [X86] Do not use ACPI for IRQ routing
3570 or for PCI scanning.
3571 use_crs [X86] Use PCI host bridge window information
3572 from ACPI. On BIOSes from 2008 or later, this
3573 is enabled by default. If you need to use this,
3574 please report a bug.
3575 nocrs [X86] Ignore PCI host bridge windows from ACPI.
3576 If you need to use this, please report a bug.
3577 routeirq Do IRQ routing for all PCI devices.
3578 This is normally done in pci_enable_device(),
3579 so this option is a temporary workaround
3580 for broken drivers that don't call it.
3581 skip_isa_align [X86] do not align io start addr, so can
3582 handle more pci cards
3583 noearly [X86] Don't do any early type 1 scanning.
3584 This might help on some broken boards which
3585 machine check when some devices' config space
3586 is read. But various workarounds are disabled
3587 and some IOMMU drivers will not work.
3588 bfsort Sort PCI devices into breadth-first order.
3589 This sorting is done to get a device
3590 order compatible with older (<= 2.4) kernels.
3591 nobfsort Don't sort PCI devices into breadth-first order.
3592 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
3593 tuning and use the BIOS-configured MPS defaults.
3594 pcie_bus_safe Set every device's MPS to the largest value
3595 supported by all devices below the root complex.
3596 pcie_bus_perf Set device MPS to the largest allowable MPS
3597 based on its parent bus. Also set MRRS (Max
3598 Read Request Size) to the largest supported
3599 value (no larger than the MPS that the device
3600 or bus can support) for best performance.
3601 pcie_bus_peer2peer Set every device's MPS to 128B, which
3602 every device is guaranteed to support. This
3603 configuration allows peer-to-peer DMA between
3604 any pair of devices, possibly at the cost of
3605 reduced performance. This also guarantees
3606 that hot-added devices will work.
3607 cbiosize=nn[KMG] The fixed amount of bus space which is
3608 reserved for the CardBus bridge's IO window.
3609 The default value is 256 bytes.
3610 cbmemsize=nn[KMG] The fixed amount of bus space which is
3611 reserved for the CardBus bridge's memory
3612 window. The default value is 64 megabytes.
3615 [<order of align>@]<pci_dev>[; ...]
3616 Specifies alignment and device to reassign
3617 aligned memory resources. How to
3618 specify the device is described above.
3619 If <order of align> is not specified,
3620 PAGE_SIZE is used as alignment.
3621 A PCI-PCI bridge can be specified if resource
3622 windows need to be expanded.
3623 To specify the alignment for several
3624 instances of a device, the PCI vendor,
3625 device, subvendor, and subdevice may be
3626 specified, e.g., 12@pci:8086:9c22:103c:198f
3627 for 4096-byte alignment.
3628 ecrc= Enable/disable PCIe ECRC (transaction layer
3629 end-to-end CRC checking).
3630 bios: Use BIOS/firmware settings. This is the
3634 hpiosize=nn[KMG] The fixed amount of bus space which is
3635 reserved for hotplug bridge's IO window.
3636 Default size is 256 bytes.
3637 hpmemsize=nn[KMG] The fixed amount of bus space which is
3638 reserved for hotplug bridge's memory window.
3639 Default size is 2 megabytes.
3640 hpbussize=nn The minimum amount of additional bus numbers
3641 reserved for buses below a hotplug bridge.
3643 realloc= Enable/disable reallocating PCI bridge resources
3644 if allocations done by BIOS are too small to
3645 accommodate resources required by all child
3647 off: Turn realloc off
3649 realloc same as realloc=on
3650 noari do not use PCIe ARI.
3651 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
3652 do not use PCIe ATS (and IOMMU device IOTLB).
3653 pcie_scan_all Scan all possible PCIe devices. Otherwise we
3654 only look for one device below a PCIe downstream
3656 big_root_window Try to add a big 64bit memory window to the PCIe
3657 root complex on AMD CPUs. Some GFX hardware
3658 can resize a BAR to allow access to all VRAM.
3659 Adding the window is slightly risky (it may
3660 conflict with unreported devices), so this
3662 disable_acs_redir=<pci_dev>[; ...]
3663 Specify one or more PCI devices (in the format
3664 specified above) separated by semicolons.
3665 Each device specified will have the PCI ACS
3666 redirect capabilities forced off which will
3667 allow P2P traffic between devices through
3668 bridges without forcing it upstream. Note:
3669 this removes isolation between devices and
3670 may put more devices in an IOMMU group.
3671 force_floating [S390] Force usage of floating interrupts.
3672 nomio [S390] Do not use MIO instructions.
3674 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
3677 force Enable ASPM even on devices that claim not to support it.
3678 WARNING: Forcing ASPM on may cause system lockups.
3680 pcie_ports= [PCIE] PCIe port services handling:
3681 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
3682 even if the platform doesn't give the OS permission to
3683 use them. This may cause conflicts if the platform
3684 also tries to use these services.
3685 compat Disable native PCIe services (PME, AER, DPC, PCIe
3688 pcie_port_pm= [PCIE] PCIe port power management handling:
3689 off Disable power management of all PCIe ports
3690 force Forcibly enable power management of all PCIe ports
3692 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
3693 nomsi Do not use MSI for native PCIe PME signaling (this makes
3694 all PCIe root ports use INTx for all services).
3696 pcmv= [HW,PCMCIA] BadgePAD 4
3700 Keep all power-domains already enabled by bootloader on,
3701 even if no driver has claimed them. This is useful
3702 for debug and development, but should not be
3703 needed on a platform with proper driver support.
3706 See Documentation/admin-guide/blockdev/paride.rst.
3708 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
3711 See arch/parisc/kernel/pdc_chassis.c
3713 percpu_alloc= Select which percpu first chunk allocator to use.
3714 Currently supported values are "embed" and "page".
3715 Archs may support subset or none of the selections.
3716 See comments in mm/percpu.c for details on each
3717 allocator. This parameter is primarily for debugging
3718 and performance comparison.
3721 See Documentation/admin-guide/blockdev/paride.rst.
3724 See Documentation/admin-guide/blockdev/paride.rst.
3726 pirq= [SMP,APIC] Manual mp-table setup
3727 See Documentation/x86/i386/IO-APIC.rst.
3729 plip= [PPT,NET] Parallel port network link
3730 Format: { parport<nr> | timid | 0 }
3731 See also Documentation/admin-guide/parport.rst.
3733 pmtmr= [X86] Manual setup of pmtmr I/O Port.
3734 Override pmtimer IOPort with a hex value.
3738 Enable PNP debug messages (depends on the
3739 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
3740 via /sys/module/pnp/parameters/debug. We always show
3741 current resource usage; turning this on also shows
3742 possible settings and some assignment information.
3748 { on | off | curr | res | no-curr | no-res }
3751 [ISAPNP] Exclude IRQs for the autoconfiguration
3754 [ISAPNP] Exclude DMAs for the autoconfiguration
3756 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
3757 Ranges are in pairs (I/O port base and size).
3760 [ISAPNP] Exclude memory regions for the
3762 Ranges are in pairs (memory base and size).
3764 ports= [IP_VS_FTP] IPVS ftp helper module
3766 Up to 8 (IP_VS_APP_MAX_PORTS) ports
3768 Format: <port>,<port>....
3770 powersave=off [PPC] This option disables power saving features.
3771 It specifically disables cpuidle and sets the
3772 platform machine description specific power_save
3773 function to NULL. On Idle the CPU just reduces
3776 ppc_strict_facility_enable
3777 [PPC] This option catches any kernel floating point,
3778 Altivec, VSX and SPE outside of regions specifically
3779 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
3780 There is some performance impact when enabling this.
3784 Disable Hardware Transactional Memory
3786 print-fatal-signals=
3787 [KNL] debug: print fatal signals
3789 If enabled, warn about various signal handling
3790 related application anomalies: too many signals,
3791 too many POSIX.1 timers, fatal signals causing a
3794 If you hit the warning due to signal overflow,
3795 you might want to try "ulimit -i unlimited".
3799 printk.always_kmsg_dump=
3800 Trigger kmsg_dump for cases other than kernel oops or
3802 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
3805 printk.devkmsg={on,off,ratelimit}
3806 Control writing to /dev/kmsg.
3807 on - unlimited logging to /dev/kmsg from userspace
3808 off - logging to /dev/kmsg disabled
3809 ratelimit - ratelimit the logging
3812 printk.time= Show timing data prefixed to each printk message line
3813 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
3815 processor.max_cstate= [HW,ACPI]
3816 Limit processor to maximum C-state
3817 max_cstate=9 overrides any DMI blacklist limit.
3819 processor.nocst [HW,ACPI]
3820 Ignore the _CST method to determine C-states,
3821 instead using the legacy FADT method
3823 profile= [KNL] Enable kernel profiling via /proc/profile
3824 Format: [<profiletype>,]<number>
3825 Param: <profiletype>: "schedule", "sleep", or "kvm"
3826 [defaults to kernel profiling]
3827 Param: "schedule" - profile schedule points.
3828 Param: "sleep" - profile D-state sleeping (millisecs).
3829 Requires CONFIG_SCHEDSTATS
3830 Param: "kvm" - profile VM exits.
3831 Param: <number> - step/bucket size as a power of 2 for
3832 statistical time based profiling.
3834 prompt_ramdisk= [RAM] List of RAM disks to prompt for floppy disk
3836 See Documentation/admin-guide/blockdev/ramdisk.rst.
3838 psi= [KNL] Enable or disable pressure stall information
3842 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
3843 probe for; one of (bare|imps|exps|lifebook|any).
3844 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
3846 psmouse.resetafter= [HW,MOUSE]
3847 Try to reset the device after so many bad packets
3850 [HW,MOUSE] Set desired mouse resolution, in dpi.
3851 psmouse.smartscroll=
3852 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
3853 0 = disabled, 1 = enabled (default).
3855 pstore.backend= Specify the name of the pstore backend to use
3858 See Documentation/admin-guide/blockdev/paride.rst.
3860 pti= [X86_64] Control Page Table Isolation of user and
3861 kernel address spaces. Disabling this feature
3862 removes hardening, but improves performance of
3863 system calls and interrupts.
3865 on - unconditionally enable
3866 off - unconditionally disable
3867 auto - kernel detects whether your CPU model is
3868 vulnerable to issues that PTI mitigates
3870 Not specifying this option is equivalent to pti=auto.
3873 Equivalent to pti=off
3876 [KNL] Number of legacy pty's. Overwrites compiled-in
3879 quiet [KNL] Disable most log messages
3884 See Documentation/admin-guide/md.rst.
3886 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
3887 See Documentation/admin-guide/blockdev/ramdisk.rst.
3889 random.trust_cpu={on,off}
3890 [KNL] Enable or disable trusting the use of the
3891 CPU's random number generator (if available) to
3892 fully seed the kernel's CRNG. Default is controlled
3893 by CONFIG_RANDOM_TRUST_CPU.
3895 random.trust_bootloader={on,off}
3896 [KNL] Enable or disable trusting the use of a
3897 seed passed by the bootloader (if available) to
3898 fully seed the kernel's CRNG. Default is controlled
3899 by CONFIG_RANDOM_TRUST_BOOTLOADER.
3901 ras=option[,option,...] [KNL] RAS-specific options
3904 Disable the Correctable Errors Collector,
3905 see CONFIG_RAS_CEC help text.
3908 The argument is a cpu list, as described above,
3909 except that the string "all" can be used to
3910 specify every CPU on the system.
3912 In kernels built with CONFIG_RCU_NOCB_CPU=y, set
3913 the specified list of CPUs to be no-callback CPUs.
3914 Invocation of these CPUs' RCU callbacks will be
3915 offloaded to "rcuox/N" kthreads created for that
3916 purpose, where "x" is "p" for RCU-preempt, and
3917 "s" for RCU-sched, and "N" is the CPU number.
3918 This reduces OS jitter on the offloaded CPUs,
3919 which can be useful for HPC and real-time
3920 workloads. It can also improve energy efficiency
3921 for asymmetric multiprocessors.
3924 Rather than requiring that offloaded CPUs
3925 (specified by rcu_nocbs= above) explicitly
3926 awaken the corresponding "rcuoN" kthreads,
3927 make these kthreads poll for callbacks.
3928 This improves the real-time response for the
3929 offloaded CPUs by relieving them of the need to
3930 wake up the corresponding kthread, but degrades
3931 energy efficiency by requiring that the kthreads
3932 periodically wake up to do the polling.
3934 rcutree.blimit= [KNL]
3935 Set maximum number of finished RCU callbacks to
3936 process in one batch.
3938 rcutree.dump_tree= [KNL]
3939 Dump the structure of the rcu_node combining tree
3940 out at early boot. This is used for diagnostic
3941 purposes, to verify correct tree setup.
3943 rcutree.gp_cleanup_delay= [KNL]
3944 Set the number of jiffies to delay each step of
3945 RCU grace-period cleanup.
3947 rcutree.gp_init_delay= [KNL]
3948 Set the number of jiffies to delay each step of
3949 RCU grace-period initialization.
3951 rcutree.gp_preinit_delay= [KNL]
3952 Set the number of jiffies to delay each step of
3953 RCU grace-period pre-initialization, that is,
3954 the propagation of recent CPU-hotplug changes up
3955 the rcu_node combining tree.
3957 rcutree.use_softirq= [KNL]
3958 If set to zero, move all RCU_SOFTIRQ processing to
3959 per-CPU rcuc kthreads. Defaults to a non-zero
3960 value, meaning that RCU_SOFTIRQ is used by default.
3961 Specify rcutree.use_softirq=0 to use rcuc kthreads.
3963 rcutree.rcu_fanout_exact= [KNL]
3964 Disable autobalancing of the rcu_node combining
3965 tree. This is used by rcutorture, and might
3966 possibly be useful for architectures having high
3967 cache-to-cache transfer latencies.
3969 rcutree.rcu_fanout_leaf= [KNL]
3970 Change the number of CPUs assigned to each
3971 leaf rcu_node structure. Useful for very
3972 large systems, which will choose the value 64,
3973 and for NUMA systems with large remote-access
3974 latencies, which will choose a value aligned
3975 with the appropriate hardware boundaries.
3977 rcutree.jiffies_till_first_fqs= [KNL]
3978 Set delay from grace-period initialization to
3979 first attempt to force quiescent states.
3980 Units are jiffies, minimum value is zero,
3981 and maximum value is HZ.
3983 rcutree.jiffies_till_next_fqs= [KNL]
3984 Set delay between subsequent attempts to force
3985 quiescent states. Units are jiffies, minimum
3986 value is one, and maximum value is HZ.
3988 rcutree.jiffies_till_sched_qs= [KNL]
3989 Set required age in jiffies for a
3990 given grace period before RCU starts
3991 soliciting quiescent-state help from
3992 rcu_note_context_switch() and cond_resched().
3993 If not specified, the kernel will calculate
3994 a value based on the most recent settings
3995 of rcutree.jiffies_till_first_fqs
3996 and rcutree.jiffies_till_next_fqs.
3997 This calculated value may be viewed in
3998 rcutree.jiffies_to_sched_qs. Any attempt to set
3999 rcutree.jiffies_to_sched_qs will be cheerfully
4002 rcutree.kthread_prio= [KNL,BOOT]
4003 Set the SCHED_FIFO priority of the RCU per-CPU
4004 kthreads (rcuc/N). This value is also used for
4005 the priority of the RCU boost threads (rcub/N)
4006 and for the RCU grace-period kthreads (rcu_bh,
4007 rcu_preempt, and rcu_sched). If RCU_BOOST is
4008 set, valid values are 1-99 and the default is 1
4009 (the least-favored priority). Otherwise, when
4010 RCU_BOOST is not set, valid values are 0-99 and
4011 the default is zero (non-realtime operation).
4013 rcutree.rcu_nocb_gp_stride= [KNL]
4014 Set the number of NOCB callback kthreads in
4015 each group, which defaults to the square root
4016 of the number of CPUs. Larger numbers reduce
4017 the wakeup overhead on the global grace-period
4018 kthread, but increases that same overhead on
4019 each group's NOCB grace-period kthread.
4021 rcutree.qhimark= [KNL]
4022 Set threshold of queued RCU callbacks beyond which
4023 batch limiting is disabled.
4025 rcutree.qlowmark= [KNL]
4026 Set threshold of queued RCU callbacks below which
4027 batch limiting is re-enabled.
4029 rcutree.rcu_idle_gp_delay= [KNL]
4030 Set wakeup interval for idle CPUs that have
4031 RCU callbacks (RCU_FAST_NO_HZ=y).
4033 rcutree.rcu_idle_lazy_gp_delay= [KNL]
4034 Set wakeup interval for idle CPUs that have
4035 only "lazy" RCU callbacks (RCU_FAST_NO_HZ=y).
4036 Lazy RCU callbacks are those which RCU can
4037 prove do nothing more than free memory.
4039 rcutree.rcu_kick_kthreads= [KNL]
4040 Cause the grace-period kthread to get an extra
4041 wake_up() if it sleeps three times longer than
4042 it should at force-quiescent-state time.
4043 This wake_up() will be accompanied by a
4044 WARN_ONCE() splat and an ftrace_dump().
4046 rcutree.sysrq_rcu= [KNL]
4047 Commandeer a sysrq key to dump out Tree RCU's
4048 rcu_node tree with an eye towards determining
4049 why a new grace period has not yet started.
4051 rcuperf.gp_async= [KNL]
4052 Measure performance of asynchronous
4053 grace-period primitives such as call_rcu().
4055 rcuperf.gp_async_max= [KNL]
4056 Specify the maximum number of outstanding
4057 callbacks per writer thread. When a writer
4058 thread exceeds this limit, it invokes the
4059 corresponding flavor of rcu_barrier() to allow
4060 previously posted callbacks to drain.
4062 rcuperf.gp_exp= [KNL]
4063 Measure performance of expedited synchronous
4064 grace-period primitives.
4066 rcuperf.holdoff= [KNL]
4067 Set test-start holdoff period. The purpose of
4068 this parameter is to delay the start of the
4069 test until boot completes in order to avoid
4072 rcuperf.nreaders= [KNL]
4073 Set number of RCU readers. The value -1 selects
4074 N, where N is the number of CPUs. A value
4075 "n" less than -1 selects N-n+1, where N is again
4076 the number of CPUs. For example, -2 selects N
4077 (the number of CPUs), -3 selects N+1, and so on.
4078 A value of "n" less than or equal to -N selects
4081 rcuperf.nwriters= [KNL]
4082 Set number of RCU writers. The values operate
4083 the same as for rcuperf.nreaders.
4084 N, where N is the number of CPUs
4086 rcuperf.perf_type= [KNL]
4087 Specify the RCU implementation to test.
4089 rcuperf.shutdown= [KNL]
4090 Shut the system down after performance tests
4091 complete. This is useful for hands-off automated
4094 rcuperf.verbose= [KNL]
4095 Enable additional printk() statements.
4097 rcuperf.writer_holdoff= [KNL]
4098 Write-side holdoff between grace periods,
4099 in microseconds. The default of zero says
4102 rcutorture.fqs_duration= [KNL]
4103 Set duration of force_quiescent_state bursts
4106 rcutorture.fqs_holdoff= [KNL]
4107 Set holdoff time within force_quiescent_state bursts
4110 rcutorture.fqs_stutter= [KNL]
4111 Set wait time between force_quiescent_state bursts
4114 rcutorture.fwd_progress= [KNL]
4115 Enable RCU grace-period forward-progress testing
4116 for the types of RCU supporting this notion.
4118 rcutorture.fwd_progress_div= [KNL]
4119 Specify the fraction of a CPU-stall-warning
4120 period to do tight-loop forward-progress testing.
4122 rcutorture.fwd_progress_holdoff= [KNL]
4123 Number of seconds to wait between successive
4124 forward-progress tests.
4126 rcutorture.fwd_progress_need_resched= [KNL]
4127 Enclose cond_resched() calls within checks for
4128 need_resched() during tight-loop forward-progress
4131 rcutorture.gp_cond= [KNL]
4132 Use conditional/asynchronous update-side
4133 primitives, if available.
4135 rcutorture.gp_exp= [KNL]
4136 Use expedited update-side primitives, if available.
4138 rcutorture.gp_normal= [KNL]
4139 Use normal (non-expedited) asynchronous
4140 update-side primitives, if available.
4142 rcutorture.gp_sync= [KNL]
4143 Use normal (non-expedited) synchronous
4144 update-side primitives, if available. If all
4145 of rcutorture.gp_cond=, rcutorture.gp_exp=,
4146 rcutorture.gp_normal=, and rcutorture.gp_sync=
4147 are zero, rcutorture acts as if is interpreted
4148 they are all non-zero.
4150 rcutorture.n_barrier_cbs= [KNL]
4151 Set callbacks/threads for rcu_barrier() testing.
4153 rcutorture.nfakewriters= [KNL]
4154 Set number of concurrent RCU writers. These just
4155 stress RCU, they don't participate in the actual
4156 test, hence the "fake".
4158 rcutorture.nreaders= [KNL]
4159 Set number of RCU readers. The value -1 selects
4160 N-1, where N is the number of CPUs. A value
4161 "n" less than -1 selects N-n-2, where N is again
4162 the number of CPUs. For example, -2 selects N
4163 (the number of CPUs), -3 selects N+1, and so on.
4165 rcutorture.object_debug= [KNL]
4166 Enable debug-object double-call_rcu() testing.
4168 rcutorture.onoff_holdoff= [KNL]
4169 Set time (s) after boot for CPU-hotplug testing.
4171 rcutorture.onoff_interval= [KNL]
4172 Set time (jiffies) between CPU-hotplug operations,
4173 or zero to disable CPU-hotplug testing.
4175 rcutorture.shuffle_interval= [KNL]
4176 Set task-shuffle interval (s). Shuffling tasks
4177 allows some CPUs to go into dyntick-idle mode
4178 during the rcutorture test.
4180 rcutorture.shutdown_secs= [KNL]
4181 Set time (s) after boot system shutdown. This
4182 is useful for hands-off automated testing.
4184 rcutorture.stall_cpu= [KNL]
4185 Duration of CPU stall (s) to test RCU CPU stall
4186 warnings, zero to disable.
4188 rcutorture.stall_cpu_holdoff= [KNL]
4189 Time to wait (s) after boot before inducing stall.
4191 rcutorture.stall_cpu_irqsoff= [KNL]
4192 Disable interrupts while stalling if set.
4194 rcutorture.stat_interval= [KNL]
4195 Time (s) between statistics printk()s.
4197 rcutorture.stutter= [KNL]
4198 Time (s) to stutter testing, for example, specifying
4199 five seconds causes the test to run for five seconds,
4200 wait for five seconds, and so on. This tests RCU's
4201 ability to transition abruptly to and from idle.
4203 rcutorture.test_boost= [KNL]
4204 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
4205 "Maybe" means test if the RCU implementation
4206 under test support RCU priority boosting.
4208 rcutorture.test_boost_duration= [KNL]
4209 Duration (s) of each individual boost test.
4211 rcutorture.test_boost_interval= [KNL]
4212 Interval (s) between each boost test.
4214 rcutorture.test_no_idle_hz= [KNL]
4215 Test RCU's dyntick-idle handling. See also the
4216 rcutorture.shuffle_interval parameter.
4218 rcutorture.torture_type= [KNL]
4219 Specify the RCU implementation to test.
4221 rcutorture.verbose= [KNL]
4222 Enable additional printk() statements.
4224 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
4225 Dump ftrace buffer after reporting RCU CPU
4228 rcupdate.rcu_cpu_stall_suppress= [KNL]
4229 Suppress RCU CPU stall warning messages.
4231 rcupdate.rcu_cpu_stall_timeout= [KNL]
4232 Set timeout for RCU CPU stall warning messages.
4234 rcupdate.rcu_expedited= [KNL]
4235 Use expedited grace-period primitives, for
4236 example, synchronize_rcu_expedited() instead
4237 of synchronize_rcu(). This reduces latency,
4238 but can increase CPU utilization, degrade
4239 real-time latency, and degrade energy efficiency.
4240 No effect on CONFIG_TINY_RCU kernels.
4242 rcupdate.rcu_normal= [KNL]
4243 Use only normal grace-period primitives,
4244 for example, synchronize_rcu() instead of
4245 synchronize_rcu_expedited(). This improves
4246 real-time latency, CPU utilization, and
4247 energy efficiency, but can expose users to
4248 increased grace-period latency. This parameter
4249 overrides rcupdate.rcu_expedited. No effect on
4250 CONFIG_TINY_RCU kernels.
4252 rcupdate.rcu_normal_after_boot= [KNL]
4253 Once boot has completed (that is, after
4254 rcu_end_inkernel_boot() has been invoked), use
4255 only normal grace-period primitives. No effect
4256 on CONFIG_TINY_RCU kernels.
4258 rcupdate.rcu_task_stall_timeout= [KNL]
4259 Set timeout in jiffies for RCU task stall warning
4260 messages. Disable with a value less than or equal
4263 rcupdate.rcu_self_test= [KNL]
4264 Run the RCU early boot self tests
4268 Run specified binary instead of /init from the ramdisk,
4269 used for early userspace startup. See initrd.
4272 force - Override the decision by the kernel to hide the
4273 advertisement of RDRAND support (this affects
4274 certain AMD processors because of buggy BIOS
4275 support, specifically around the suspend/resume
4279 Turn on/off individual RDT features. List is:
4280 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
4282 E.g. to turn on cmt and turn off mba use:
4286 Format (x86 or x86_64):
4287 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] \
4289 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
4291 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
4292 (prefix with 'panic_' to set mode for panic
4294 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
4295 reboot_force is either force or not specified,
4296 reboot_cpu is s[mp]#### with #### being the processor
4297 to be used for rebooting.
4300 [KNL, SMP] Set scheduler's default relax_domain_level.
4301 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
4303 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
4304 Format: <base1>,<size1>[,<base2>,<size2>,...]
4305 Reserve I/O ports or memory so the kernel won't use
4306 them. If <base> is less than 0x10000, the region
4307 is assumed to be I/O ports; otherwise it is memory.
4309 reservetop= [X86-32]
4311 Reserves a hole at the top of the kernel virtual
4316 Set the amount of memory to reserve for BIOS at
4317 the bottom of the address space.
4319 reset_devices [KNL] Force drivers to reset the underlying device
4320 during initialization.
4323 Specify the partition device for software suspend
4325 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
4327 resume_offset= [SWSUSP]
4328 Specify the offset from the beginning of the partition
4329 given by "resume=" at which the swap header is located,
4330 in <PAGE_SIZE> units (needed only for swap files).
4331 See Documentation/power/swsusp-and-swap-files.rst
4333 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
4334 read the resume files
4336 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
4337 Useful for devices that are detected asynchronously
4338 (e.g. USB and MMC devices).
4340 hibernate= [HIBERNATION]
4341 noresume Don't check if there's a hibernation image
4342 present during boot.
4343 nocompress Don't compress/decompress hibernation images.
4344 no Disable hibernation and resume.
4345 protect_image Turn on image protection during restoration
4346 (that will set all pages holding image data
4347 during restoration read-only).
4349 retain_initrd [RAM] Keep initrd memory after extraction
4351 retbleed= [X86] Control mitigation of RETBleed (Arbitrary
4352 Speculative Code Execution with Return Instructions)
4355 off - unconditionally disable
4356 auto - automatically select a migitation
4358 Selecting 'auto' will choose a mitigation method at run
4359 time according to the CPU.
4361 Not specifying this option is equivalent to retbleed=auto.
4363 rfkill.default_state=
4364 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
4365 etc. communication is blocked by default.
4368 rfkill.master_switch_mode=
4369 0 The "airplane mode" button does nothing.
4370 1 The "airplane mode" button toggles between everything
4371 blocked and the previous configuration.
4372 2 The "airplane mode" button toggles between everything
4373 blocked and everything unblocked.
4375 rhash_entries= [KNL,NET]
4376 Set number of hash buckets for route cache
4379 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
4382 ro [KNL] Mount root device read-only on boot
4385 on Mark read-only kernel memory as read-only (default).
4386 off Leave read-only kernel memory writable for debugging.
4389 Enable the uart passthrough on the designated usb port
4390 on Rockchip SoCs. When active, the signals of the
4391 debug-uart get routed to the D+ and D- pins of the usb
4392 port and the regular usb controller gets disabled.
4394 root= [KNL] Root filesystem
4395 See name_to_dev_t comment in init/do_mounts.c.
4397 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
4398 mount the root filesystem
4400 rootflags= [KNL] Set root filesystem mount option string
4402 rootfstype= [KNL] Set root filesystem type
4404 rootwait [KNL] Wait (indefinitely) for root device to show up.
4405 Useful for devices that are detected asynchronously
4406 (e.g. USB and MMC devices).
4408 rproc_mem=nn[KMG][@address]
4409 [KNL,ARM,CMA] Remoteproc physical memory block.
4410 Memory area to be used by remote processor image,
4413 rw [KNL] Mount root device read-write on boot
4415 S [KNL] Run init in single mode
4417 s390_iommu= [HW,S390]
4418 Set s390 IOTLB flushing mode
4420 With strict flushing every unmap operation will result in
4421 an IOTLB flush. Default is lazy flushing before reuse,
4425 See drivers/net/irda/sa1100_ir.c.
4427 sbni= [NET] Granch SBNI12 leased line adapter
4429 sched_debug [KNL] Enables verbose scheduler debug messages.
4431 schedstats= [KNL,X86] Enable or disable scheduled statistics.
4432 Allowed values are enable and disable. This feature
4433 incurs a small amount of overhead in the scheduler
4434 but is useful for debugging and performance tuning.
4436 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
4437 xtime_lock contention on larger systems, and/or RCU lock
4438 contention on all systems with CONFIG_MAXSMP set.
4439 Format: { "0" | "1" }
4440 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
4442 Note: increases power consumption, thus should only be
4443 enabled if running jitter sensitive (HPC/RT) workloads.
4445 security= [SECURITY] Choose a legacy "major" security module to
4446 enable at boot. This has been deprecated by the
4449 selinux= [SELINUX] Disable or enable SELinux at boot time.
4450 Format: { "0" | "1" }
4451 See security/selinux/Kconfig help text.
4454 Default value is set via kernel config option.
4455 If enabled at boot time, /selinux/disable can be used
4456 later to disable prior to initial policy load.
4458 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
4459 Format: { "0" | "1" }
4460 See security/apparmor/Kconfig help text
4463 Default value is set via kernel config option.
4465 serialnumber [BUGS=X86-32]
4468 Maximal number of shapers.
4476 Disable merging of slabs with similar size. May be
4477 necessary if there is some reason to distinguish
4478 allocs to different slabs, especially in hardened
4479 environments where the risk of heap overflows and
4480 layout control by attackers can usually be
4481 frustrated by disabling merging. This will reduce
4482 most of the exposure of a heap attack to a single
4483 cache (risks via metadata attacks are mostly
4484 unchanged). Debug options disable merging on their
4486 For more information see Documentation/vm/slub.rst.
4488 slab_max_order= [MM, SLAB]
4489 Determines the maximum allowed order for slabs.
4490 A high setting may cause OOMs due to memory
4491 fragmentation. Defaults to 1 for systems with
4492 more than 32MB of RAM, 0 otherwise.
4494 slub_debug[=options[,slabs]] [MM, SLUB]
4495 Enabling slub_debug allows one to determine the
4496 culprit if slab objects become corrupted. Enabling
4497 slub_debug can create guard zones around objects and
4498 may poison objects when not in use. Also tracks the
4499 last alloc / free. For more information see
4500 Documentation/vm/slub.rst.
4502 slub_memcg_sysfs= [MM, SLUB]
4503 Determines whether to enable sysfs directories for
4504 memory cgroup sub-caches. 1 to enable, 0 to disable.
4505 The default is determined by CONFIG_SLUB_MEMCG_SYSFS_ON.
4506 Enabling this can lead to a very high number of debug
4507 directories and files being created under
4510 slub_max_order= [MM, SLUB]
4511 Determines the maximum allowed order for slabs.
4512 A high setting may cause OOMs due to memory
4513 fragmentation. For more information see
4514 Documentation/vm/slub.rst.
4516 slub_min_objects= [MM, SLUB]
4517 The minimum number of objects per slab. SLUB will
4518 increase the slab order up to slub_max_order to
4519 generate a sufficiently large slab able to contain
4520 the number of objects indicated. The higher the number
4521 of objects the smaller the overhead of tracking slabs
4522 and the less frequently locks need to be acquired.
4523 For more information see Documentation/vm/slub.rst.
4525 slub_min_order= [MM, SLUB]
4526 Determines the minimum page order for slabs. Must be
4527 lower than slub_max_order.
4528 For more information see Documentation/vm/slub.rst.
4530 slub_nomerge [MM, SLUB]
4531 Same with slab_nomerge. This is supported for legacy.
4532 See slab_nomerge for more information.
4535 Format: <io1>[,<io2>[,...,<io8>]]
4537 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
4538 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
4539 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
4540 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
4541 smsc-ircc2.ircc_irq= [HW] IRQ line
4542 smsc-ircc2.ircc_dma= [HW] DMA channel
4543 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
4544 0: Toshiba Satellite 1800 (GP data pin select)
4545 1: Fast pin select (default)
4548 smt [KNL,S390] Set the maximum number of threads (logical
4549 CPUs) to use per physical CPU on systems capable of
4550 symmetric multithreading (SMT). Will be capped to the
4551 actual hardware limit.
4553 Default: -1 (no limit)
4556 [KNL] Should the soft-lockup detector generate panics.
4559 A nonzero value instructs the soft-lockup detector
4560 to panic the machine when a soft-lockup occurs. This
4561 is also controlled by CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC
4562 which is the respective build-time switch to that
4565 softlockup_all_cpu_backtrace=
4566 [KNL] Should the soft-lockup detector generate
4567 backtraces on all cpus.
4570 sonypi.*= [HW] Sony Programmable I/O Control Device driver
4571 See Documentation/admin-guide/laptops/sonypi.rst
4573 spectre_v2= [X86] Control mitigation of Spectre variant 2
4574 (indirect branch speculation) vulnerability.
4575 The default operation protects the kernel from
4578 on - unconditionally enable, implies
4580 off - unconditionally disable, implies
4582 auto - kernel detects whether your CPU model is
4585 Selecting 'on' will, and 'auto' may, choose a
4586 mitigation method at run time according to the
4587 CPU, the available microcode, the setting of the
4588 CONFIG_RETPOLINE configuration option, and the
4589 compiler with which the kernel was built.
4591 Selecting 'on' will also enable the mitigation
4592 against user space to user space task attacks.
4594 Selecting 'off' will disable both the kernel and
4595 the user space protections.
4597 Specific mitigations can also be selected manually:
4599 retpoline - replace indirect branches
4600 retpoline,generic - Retpolines
4601 retpoline,lfence - LFENCE; indirect branch
4602 retpoline,amd - alias for retpoline,lfence
4603 eibrs - Enhanced/Auto IBRS
4604 eibrs,retpoline - Enhanced/Auto IBRS + Retpolines
4605 eibrs,lfence - Enhanced/Auto IBRS + LFENCE
4606 ibrs - use IBRS to protect kernel
4608 Not specifying this option is equivalent to
4612 [X86] Control mitigation of Spectre variant 2
4613 (indirect branch speculation) vulnerability between
4616 on - Unconditionally enable mitigations. Is
4617 enforced by spectre_v2=on
4619 off - Unconditionally disable mitigations. Is
4620 enforced by spectre_v2=off
4622 prctl - Indirect branch speculation is enabled,
4623 but mitigation can be enabled via prctl
4624 per thread. The mitigation control state
4625 is inherited on fork.
4628 - Like "prctl" above, but only STIBP is
4629 controlled per thread. IBPB is issued
4630 always when switching between different user
4634 - Same as "prctl" above, but all seccomp
4635 threads will enable the mitigation unless
4636 they explicitly opt out.
4639 - Like "seccomp" above, but only STIBP is
4640 controlled per thread. IBPB is issued
4641 always when switching between different
4642 user space processes.
4644 auto - Kernel selects the mitigation depending on
4645 the available CPU features and vulnerability.
4648 If CONFIG_SECCOMP=y then "seccomp", otherwise "prctl"
4650 Not specifying this option is equivalent to
4651 spectre_v2_user=auto.
4653 spec_store_bypass_disable=
4654 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
4655 (Speculative Store Bypass vulnerability)
4657 Certain CPUs are vulnerable to an exploit against a
4658 a common industry wide performance optimization known
4659 as "Speculative Store Bypass" in which recent stores
4660 to the same memory location may not be observed by
4661 later loads during speculative execution. The idea
4662 is that such stores are unlikely and that they can
4663 be detected prior to instruction retirement at the
4664 end of a particular speculation execution window.
4666 In vulnerable processors, the speculatively forwarded
4667 store can be used in a cache side channel attack, for
4668 example to read memory to which the attacker does not
4669 directly have access (e.g. inside sandboxed code).
4671 This parameter controls whether the Speculative Store
4672 Bypass optimization is used.
4674 On x86 the options are:
4676 on - Unconditionally disable Speculative Store Bypass
4677 off - Unconditionally enable Speculative Store Bypass
4678 auto - Kernel detects whether the CPU model contains an
4679 implementation of Speculative Store Bypass and
4680 picks the most appropriate mitigation. If the
4681 CPU is not vulnerable, "off" is selected. If the
4682 CPU is vulnerable the default mitigation is
4683 architecture and Kconfig dependent. See below.
4684 prctl - Control Speculative Store Bypass per thread
4685 via prctl. Speculative Store Bypass is enabled
4686 for a process by default. The state of the control
4687 is inherited on fork.
4688 seccomp - Same as "prctl" above, but all seccomp threads
4689 will disable SSB unless they explicitly opt out.
4691 Default mitigations:
4692 X86: If CONFIG_SECCOMP=y "seccomp", otherwise "prctl"
4694 On powerpc the options are:
4696 on,auto - On Power8 and Power9 insert a store-forwarding
4697 barrier on kernel entry and exit. On Power7
4698 perform a software flush on kernel entry and
4702 Not specifying this option is equivalent to
4703 spec_store_bypass_disable=auto.
4705 spia_io_base= [HW,MTD]
4711 Control the Special Register Buffer Data Sampling
4714 Certain CPUs are vulnerable to an MDS-like
4715 exploit which can leak bits from the random
4718 By default, this issue is mitigated by
4719 microcode. However, the microcode fix can cause
4720 the RDRAND and RDSEED instructions to become
4721 much slower. Among other effects, this will
4722 result in reduced throughput from /dev/urandom.
4724 The microcode mitigation can be disabled with
4725 the following option:
4727 off: Disable mitigation and remove
4728 performance impact to RDRAND and RDSEED
4730 srcutree.counter_wrap_check [KNL]
4731 Specifies how frequently to check for
4732 grace-period sequence counter wrap for the
4733 srcu_data structure's ->srcu_gp_seq_needed field.
4734 The greater the number of bits set in this kernel
4735 parameter, the less frequently counter wrap will
4736 be checked for. Note that the bottom two bits
4739 srcutree.exp_holdoff [KNL]
4740 Specifies how many nanoseconds must elapse
4741 since the end of the last SRCU grace period for
4742 a given srcu_struct until the next normal SRCU
4743 grace period will be considered for automatic
4744 expediting. Set to zero to disable automatic
4748 Speculative Store Bypass Disable control
4750 On CPUs that are vulnerable to the Speculative
4751 Store Bypass vulnerability and offer a
4752 firmware based mitigation, this parameter
4753 indicates how the mitigation should be used:
4755 force-on: Unconditionally enable mitigation for
4756 for both kernel and userspace
4757 force-off: Unconditionally disable mitigation for
4758 for both kernel and userspace
4759 kernel: Always enable mitigation in the
4760 kernel, and offer a prctl interface
4761 to allow userspace to register its
4762 interest in being mitigated too.
4764 stack_guard_gap= [MM]
4765 override the default stack gap protection. The value
4766 is in page units and it defines how many pages prior
4767 to (for stacks growing down) resp. after (for stacks
4768 growing up) the main stack are reserved for no other
4769 mapping. Default value is 256 pages.
4772 Enabled the stack tracer on boot up.
4774 stacktrace_filter=[function-list]
4775 [FTRACE] Limit the functions that the stack tracer
4776 will trace at boot up. function-list is a comma separated
4777 list of functions. This list can be changed at run
4778 time by the stack_trace_filter file in the debugfs
4779 tracing directory. Note, this enables stack tracing
4780 and the stacktrace above is not needed.
4784 Set the STI (builtin display/keyboard on the HP-PARISC
4785 machines) console (graphic card) which should be used
4786 as the initial boot-console.
4787 See also comment in drivers/video/console/sticore.c.
4790 See comment in drivers/video/console/sticore.c.
4793 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
4795 sunrpc.min_resvport=
4796 sunrpc.max_resvport=
4798 SunRPC servers often require that client requests
4799 originate from a privileged port (i.e. a port in the
4800 range 0 < portnr < 1024).
4801 An administrator who wishes to reserve some of these
4802 ports for other uses may adjust the range that the
4803 kernel's sunrpc client considers to be privileged
4804 using these two parameters to set the minimum and
4805 maximum port values.
4807 sunrpc.svc_rpc_per_connection_limit=
4809 Limit the number of requests that the server will
4810 process in parallel from a single connection.
4811 The default value is 0 (no limit).
4815 Control how the NFS server code allocates CPUs to
4816 service thread pools. Depending on how many NICs
4817 you have and where their interrupts are bound, this
4818 option will affect which CPUs will do NFS serving.
4819 Note: this parameter cannot be changed while the
4820 NFS server is running.
4822 auto the server chooses an appropriate mode
4823 automatically using heuristics
4824 global a single global pool contains all CPUs
4825 percpu one pool for each CPU
4826 pernode one pool for each NUMA node (equivalent
4827 to global on non-NUMA machines)
4829 sunrpc.tcp_slot_table_entries=
4830 sunrpc.udp_slot_table_entries=
4832 Sets the upper limit on the number of simultaneous
4833 RPC calls that can be sent from the client to a
4834 server. Increasing these values may allow you to
4835 improve throughput, but will also increase the
4836 amount of memory reserved for use by the client.
4838 suspend.pm_test_delay=
4840 Sets the number of seconds to remain in a suspend test
4841 mode before resuming the system (see
4842 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
4843 is set. Default value is 5.
4846 Format: { on | off | y | n | 1 | 0 }
4847 This parameter controls use of the Protected
4848 Execution Facility on pSeries.
4851 [KNL] Enable accounting of swap in memory resource
4852 controller if no parameter or 1 is given or disable
4853 it if 0 is given (See Documentation/admin-guide/cgroup-v1/memory.rst)
4855 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
4856 Format: { <int> | force | noforce }
4857 <int> -- Number of I/O TLB slabs
4858 force -- force using of bounce buffers even if they
4859 wouldn't be automatically used by the kernel
4860 noforce -- Never use bounce buffers (for debugging)
4864 sysfs.deprecated=0|1 [KNL]
4865 Enable/disable old style sysfs layout for old udev
4866 on older distributions. When this option is enabled
4867 very new udev will not work anymore. When this option
4868 is disabled (or CONFIG_SYSFS_DEPRECATED not compiled)
4869 in older udev will not work anymore.
4870 Default depends on CONFIG_SYSFS_DEPRECATED_V2 set in
4871 the kernel configuration.
4873 sysrq_always_enabled
4875 Ignore sysrq setting - this boot parameter will
4876 neutralize any effect of /proc/sys/kernel/sysrq.
4877 Useful for debugging.
4879 tcpmhash_entries= [KNL,NET]
4880 Set the number of tcp_metrics_hash slots.
4881 Default value is 8192 or 16384 depending on total
4882 ram pages. This is used to specify the TCP metrics
4883 cache size. See Documentation/networking/ip-sysctl.txt
4884 "tcp_no_metrics_save" section for more details.
4888 test_suspend= [SUSPEND][,N]
4889 Specify "mem" (for Suspend-to-RAM) or "standby" (for
4890 standby suspend) or "freeze" (for suspend type freeze)
4891 as the system sleep state during system startup with
4892 the optional capability to repeat N number of times.
4893 The system is woken from this state using a
4894 wakeup-capable RTC alarm.
4896 thash_entries= [KNL,NET]
4897 Set number of hash buckets for TCP connection
4899 thermal.act= [HW,ACPI]
4900 -1: disable all active trip points in all thermal zones
4901 <degrees C>: override all lowest active trip points
4903 thermal.crt= [HW,ACPI]
4904 -1: disable all critical trip points in all thermal zones
4905 <degrees C>: override all critical trip points
4907 thermal.nocrt= [HW,ACPI]
4908 Set to disable actions on ACPI thermal zone
4909 critical and hot trip points.
4911 thermal.off= [HW,ACPI]
4912 1: disable ACPI thermal control
4914 thermal.psv= [HW,ACPI]
4915 -1: disable all passive trip points
4916 <degrees C>: override all passive trip points to this
4919 thermal.tzp= [HW,ACPI]
4920 Specify global default ACPI thermal zone polling rate
4921 <deci-seconds>: poll all this frequency
4922 0: no polling (default)
4925 Force threading of all interrupt handlers except those
4926 marked explicitly IRQF_NO_THREAD.
4930 Specify if the kernel should make use of the cpu
4931 topology information if the hardware supports this.
4932 The scheduler will make use of this information and
4933 e.g. base its process migration decisions on it.
4936 topology_updates= [KNL, PPC, NUMA]
4938 Specify if the kernel should ignore (off)
4939 topology updates sent by the hypervisor to this
4944 tpm_suspend_pcr=[HW,TPM]
4945 Format: integer pcr id
4946 Specify that at suspend time, the tpm driver
4947 should extend the specified pcr with zeros,
4948 as a workaround for some chips which fail to
4949 flush the last written pcr on TPM_SaveState.
4950 This will guarantee that all the other pcrs
4953 trace_buf_size=nn[KMG]
4954 [FTRACE] will set tracing buffer size on each cpu.
4956 trace_event=[event-list]
4957 [FTRACE] Set and start specified trace events in order
4958 to facilitate early boot debugging. The event-list is a
4959 comma separated list of trace events to enable. See
4960 also Documentation/trace/events.rst
4962 trace_options=[option-list]
4963 [FTRACE] Enable or disable tracer options at boot.
4964 The option-list is a comma delimited list of options
4965 that can be enabled or disabled just as if you were
4966 to echo the option name into
4968 /sys/kernel/debug/tracing/trace_options
4970 For example, to enable stacktrace option (to dump the
4971 stack trace of each event), add to the command line:
4973 trace_options=stacktrace
4975 See also Documentation/trace/ftrace.rst "trace options"
4979 Have the tracepoints sent to printk as well as the
4980 tracing ring buffer. This is useful for early boot up
4981 where the system hangs or reboots and does not give the
4982 option for reading the tracing buffer or performing a
4983 ftrace_dump_on_oops.
4985 To turn off having tracepoints sent to printk,
4986 echo 0 > /proc/sys/kernel/tracepoint_printk
4987 Note, echoing 1 into this file without the
4988 tracepoint_printk kernel cmdline option has no effect.
4992 Having tracepoints sent to printk() and activating high
4993 frequency tracepoints such as irq or sched, can cause
4994 the system to live lock.
4997 [FTRACE] enable this option to disable tracing when a
4998 warning is hit. This turns off "tracing_on". Tracing can
4999 be enabled again by echoing '1' into the "tracing_on"
5000 file located in /sys/kernel/debug/tracing/
5002 This option is useful, as it disables the trace before
5003 the WARNING dump is called, which prevents the trace to
5004 be filled with content caused by the warning output.
5006 This option can also be set at run time via the sysctl
5007 option: kernel/traceoff_on_warning
5009 transparent_hugepage=
5011 Format: [always|madvise|never]
5012 Can be used to control the default behavior of the system
5013 with respect to transparent hugepages.
5014 See Documentation/admin-guide/mm/transhuge.rst
5017 tsc= Disable clocksource stability checks for TSC.
5019 [x86] reliable: mark tsc clocksource as reliable, this
5020 disables clocksource verification at runtime, as well
5021 as the stability checks done at bootup. Used to enable
5022 high-resolution timer mode on older hardware, and in
5023 virtualized environment.
5024 [x86] noirqtime: Do not use TSC to do irq accounting.
5025 Used to run time disable IRQ_TIME_ACCOUNTING on any
5026 platforms where RDTSC is slow and this accounting
5028 [x86] unstable: mark the TSC clocksource as unstable, this
5029 marks the TSC unconditionally unstable at bootup and
5030 avoids any further wobbles once the TSC watchdog notices.
5031 [x86] nowatchdog: disable clocksource watchdog. Used
5032 in situations with strict latency requirements (where
5033 interruptions from clocksource watchdog are not
5036 tsx= [X86] Control Transactional Synchronization
5037 Extensions (TSX) feature in Intel processors that
5038 support TSX control.
5040 This parameter controls the TSX feature. The options are:
5042 on - Enable TSX on the system. Although there are
5043 mitigations for all known security vulnerabilities,
5044 TSX has been known to be an accelerator for
5045 several previous speculation-related CVEs, and
5046 so there may be unknown security risks associated
5047 with leaving it enabled.
5049 off - Disable TSX on the system. (Note that this
5050 option takes effect only on newer CPUs which are
5051 not vulnerable to MDS, i.e., have
5052 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
5053 the new IA32_TSX_CTRL MSR through a microcode
5054 update. This new MSR allows for the reliable
5055 deactivation of the TSX functionality.)
5057 auto - Disable TSX if X86_BUG_TAA is present,
5058 otherwise enable TSX on the system.
5060 Not specifying this option is equivalent to tsx=off.
5062 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5065 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
5066 Abort (TAA) vulnerability.
5068 Similar to Micro-architectural Data Sampling (MDS)
5069 certain CPUs that support Transactional
5070 Synchronization Extensions (TSX) are vulnerable to an
5071 exploit against CPU internal buffers which can forward
5072 information to a disclosure gadget under certain
5075 In vulnerable processors, the speculatively forwarded
5076 data can be used in a cache side channel attack, to
5077 access data to which the attacker does not have direct
5080 This parameter controls the TAA mitigation. The
5083 full - Enable TAA mitigation on vulnerable CPUs
5086 full,nosmt - Enable TAA mitigation and disable SMT on
5087 vulnerable CPUs. If TSX is disabled, SMT
5088 is not disabled because CPU is not
5089 vulnerable to cross-thread TAA attacks.
5090 off - Unconditionally disable TAA mitigation
5092 On MDS-affected machines, tsx_async_abort=off can be
5093 prevented by an active MDS mitigation as both vulnerabilities
5094 are mitigated with the same mechanism so in order to disable
5095 this mitigation, you need to specify mds=off too.
5097 Not specifying this option is equivalent to
5098 tsx_async_abort=full. On CPUs which are MDS affected
5099 and deploy MDS mitigation, TAA mitigation is not
5100 required and doesn't provide any additional
5104 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5106 turbografx.map[2|3]= [HW,JOY]
5107 TurboGraFX parallel port interface
5109 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
5110 See also Documentation/input/devices/joystick-parport.rst
5112 udbg-immortal [PPC] When debugging early kernel crashes that
5113 happen after console_init() and before a proper
5114 console driver takes over, this boot options might
5115 help "seeing" what's going on.
5117 uhash_entries= [KNL,NET]
5118 Set number of hash buckets for UDP/UDP-Lite connections
5121 [USB] Ignore overcurrent events (default N).
5122 Some badly-designed motherboards generate lots of
5123 bogus events, for ports that aren't wired to
5124 anything. Set this parameter to avoid log spamming.
5125 Note that genuine overcurrent events won't be
5129 [X86] Cause panic on unknown NMI.
5131 usbcore.authorized_default=
5132 [USB] Default USB device authorization:
5133 (default -1 = authorized except for wireless USB,
5134 0 = not authorized, 1 = authorized, 2 = authorized
5135 if device connected to internal port)
5137 usbcore.autosuspend=
5138 [USB] The autosuspend time delay (in seconds) used
5139 for newly-detected USB devices (default 2). This
5140 is the time required before an idle device will be
5141 autosuspended. Devices for which the delay is set
5142 to a negative value won't be autosuspended at all.
5144 usbcore.usbfs_snoop=
5145 [USB] Set to log all usbfs traffic (default 0 = off).
5147 usbcore.usbfs_snoop_max=
5148 [USB] Maximum number of bytes to snoop in each URB
5151 usbcore.blinkenlights=
5152 [USB] Set to cycle leds on hubs (default 0 = off).
5154 usbcore.old_scheme_first=
5155 [USB] Start with the old device initialization
5156 scheme (default 0 = off).
5158 usbcore.usbfs_memory_mb=
5159 [USB] Memory limit (in MB) for buffers allocated by
5160 usbfs (default = 16, 0 = max = 2047).
5162 usbcore.use_both_schemes=
5163 [USB] Try the other device initialization scheme
5164 if the first one fails (default 1 = enabled).
5166 usbcore.initial_descriptor_timeout=
5167 [USB] Specifies timeout for the initial 64-byte
5168 USB_REQ_GET_DESCRIPTOR request in milliseconds
5169 (default 5000 = 5.0 seconds).
5171 usbcore.nousb [USB] Disable the USB subsystem
5174 [USB] A list of quirk entries to augment the built-in
5175 usb core quirk list. List entries are separated by
5176 commas. Each entry has the form
5177 VendorID:ProductID:Flags. The IDs are 4-digit hex
5178 numbers and Flags is a set of letters. Each letter
5179 will change the built-in quirk; setting it if it is
5180 clear and clearing it if it is set. The letters have
5181 the following meanings:
5182 a = USB_QUIRK_STRING_FETCH_255 (string
5183 descriptors must not be fetched using
5185 b = USB_QUIRK_RESET_RESUME (device can't resume
5186 correctly so reset it instead);
5187 c = USB_QUIRK_NO_SET_INTF (device can't handle
5188 Set-Interface requests);
5189 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
5190 handle its Configuration or Interface
5192 e = USB_QUIRK_RESET (device can't be reset
5193 (e.g morph devices), don't use reset);
5194 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
5195 more interface descriptions than the
5196 bNumInterfaces count, and can't handle
5197 talking to these interfaces);
5198 g = USB_QUIRK_DELAY_INIT (device needs a pause
5199 during initialization, after we read
5200 the device descriptor);
5201 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
5202 high speed and super speed interrupt
5203 endpoints, the USB 2.0 and USB 3.0 spec
5204 require the interval in microframes (1
5205 microframe = 125 microseconds) to be
5206 calculated as interval = 2 ^
5208 Devices with this quirk report their
5209 bInterval as the result of this
5210 calculation instead of the exponent
5211 variable used in the calculation);
5212 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
5213 handle device_qualifier descriptor
5215 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
5216 generates spurious wakeup, ignore
5217 remote wakeup capability);
5218 k = USB_QUIRK_NO_LPM (device can't handle Link
5220 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
5221 (Device reports its bInterval as linear
5222 frames instead of the USB 2.0
5224 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
5225 to be disconnected before suspend to
5226 prevent spurious wakeup);
5227 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
5228 pause after every control message);
5229 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
5230 delay after resetting its port);
5231 Example: quirks=0781:5580:bk,0a5c:5834:gij
5234 [USBHID] The interval which mice are to be polled at.
5237 [USBHID] The interval which joysticks are to be polled at.
5240 [USBHID] The interval which keyboards are to be polled at.
5242 usb-storage.delay_use=
5243 [UMS] The delay in seconds before a new device is
5244 scanned for Logical Units (default 1).
5247 [UMS] A list of quirks entries to supplement or
5248 override the built-in unusual_devs list. List
5249 entries are separated by commas. Each entry has
5250 the form VID:PID:Flags where VID and PID are Vendor
5251 and Product ID values (4-digit hex numbers) and
5252 Flags is a set of characters, each corresponding
5253 to a common usb-storage quirk flag as follows:
5254 a = SANE_SENSE (collect more than 18 bytes
5255 of sense data, not on uas);
5256 b = BAD_SENSE (don't collect more than 18
5257 bytes of sense data, not on uas);
5258 c = FIX_CAPACITY (decrease the reported
5259 device capacity by one sector);
5260 d = NO_READ_DISC_INFO (don't use
5261 READ_DISC_INFO command, not on uas);
5262 e = NO_READ_CAPACITY_16 (don't use
5263 READ_CAPACITY_16 command);
5264 f = NO_REPORT_OPCODES (don't use report opcodes
5266 g = MAX_SECTORS_240 (don't transfer more than
5267 240 sectors at a time, uas only);
5268 h = CAPACITY_HEURISTICS (decrease the
5269 reported device capacity by one
5270 sector if the number is odd);
5271 i = IGNORE_DEVICE (don't bind to this
5273 j = NO_REPORT_LUNS (don't use report luns
5275 k = NO_SAME (do not use WRITE_SAME, uas only)
5276 l = NOT_LOCKABLE (don't try to lock and
5277 unlock ejectable media, not on uas);
5278 m = MAX_SECTORS_64 (don't transfer more
5279 than 64 sectors = 32 KB at a time,
5281 n = INITIAL_READ10 (force a retry of the
5282 initial READ(10) command, not on uas);
5283 o = CAPACITY_OK (accept the capacity
5284 reported by the device, not on uas);
5285 p = WRITE_CACHE (the device cache is ON
5286 by default, not on uas);
5287 r = IGNORE_RESIDUE (the device reports
5288 bogus residue values, not on uas);
5289 s = SINGLE_LUN (the device has only one
5291 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
5292 commands, uas only);
5293 u = IGNORE_UAS (don't bind to the uas driver);
5294 w = NO_WP_DETECT (don't test whether the
5295 medium is write-protected).
5296 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
5297 even if the device claims no cache,
5299 Example: quirks=0419:aaf5:rl,0421:0433:rc
5301 user_debug= [KNL,ARM]
5303 See arch/arm/Kconfig.debug help text.
5304 1 - undefined instruction events
5306 4 - invalid data aborts
5309 Example: user_debug=31
5312 [X86] Flags controlling user PTE allocations.
5314 nohigh = do not allocate PTE pages in
5315 HIGHMEM regardless of setting
5319 On X86_32, this is an alias for vdso32=. Otherwise:
5321 vdso=1: enable VDSO (the default)
5322 vdso=0: disable VDSO mapping
5324 vdso32= [X86] Control the 32-bit vDSO
5325 vdso32=1: enable 32-bit VDSO
5326 vdso32=0 or vdso32=2: disable 32-bit VDSO
5328 See the help text for CONFIG_COMPAT_VDSO for more
5329 details. If CONFIG_COMPAT_VDSO is set, the default is
5330 vdso32=0; otherwise, the default is vdso32=1.
5332 For compatibility with older kernels, vdso32=2 is an
5335 Try vdso32=0 if you encounter an error that says:
5336 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
5339 vector=percpu: enable percpu vector domain
5341 video= [FB] Frame buffer configuration
5342 See Documentation/fb/modedb.rst.
5344 video.brightness_switch_enabled= [0,1]
5345 If set to 1, on receiving an ACPI notify event
5346 generated by hotkey, video driver will adjust brightness
5347 level and then send out the event to user space through
5348 the allocated input device; If set to 0, video driver
5349 will only send out the event without touching backlight
5354 [VMMIO] Memory mapped virtio (platform) device.
5356 <size>@<baseaddr>:<irq>[:<id>]
5358 <size> := size (can use standard suffixes
5360 <baseaddr> := physical base address
5361 <irq> := interrupt number (as passed to
5363 <id> := (optional) platform device id
5365 virtio_mmio.device=1K@0x100b0000:48:7
5367 Can be used multiple times for multiple devices.
5369 vga= [BOOT,X86-32] Select a particular video mode
5370 See Documentation/x86/boot.rst and
5371 Documentation/admin-guide/svga.rst.
5372 Use vga=ask for menu.
5373 This is actually a boot loader parameter; the value is
5374 passed to the kernel using a special protocol.
5376 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
5377 May slow down system boot speed, especially when
5378 enabled on systems with a large amount of memory.
5379 All options are enabled by default, and this
5380 interface is meant to allow for selectively
5381 enabling or disabling specific virtual memory
5384 Available options are:
5385 P Enable page structure init time poisoning
5386 - Disable all of the above options
5388 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
5389 size of <nn>. This can be used to increase the
5390 minimum size (128MB on x86). It can also be used to
5391 decrease the size and leave more room for directly
5394 vmcp_cma=nn[MG] [KNL,S390]
5395 Sets the memory size reserved for contiguous memory
5396 allocations for the vmcp device driver.
5398 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
5401 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
5404 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
5408 Controls the behavior of vsyscalls (i.e. calls to
5409 fixed addresses of 0xffffffffff600x00 from legacy
5410 code). Most statically-linked binaries and older
5411 versions of glibc use these calls. Because these
5412 functions are at fixed addresses, they make nice
5413 targets for exploits that can control RIP.
5415 emulate [default] Vsyscalls turn into traps and are
5416 emulated reasonably safely. The vsyscall
5419 xonly Vsyscalls turn into traps and are
5420 emulated reasonably safely. The vsyscall
5421 page is not readable.
5423 none Vsyscalls don't work at all. This makes
5424 them quite hard to use for exploits but
5425 might break your system.
5427 vt.color= [VT] Default text color.
5428 Format: 0xYX, X = foreground, Y = background.
5429 Default: 0x07 = light gray on black.
5431 vt.cur_default= [VT] Default cursor shape.
5432 Format: 0xCCBBAA, where AA, BB, and CC are the same as
5433 the parameters of the <Esc>[?A;B;Cc escape sequence;
5434 see VGA-softcursor.txt. Default: 2 = underline.
5436 vt.default_blu= [VT]
5437 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
5438 Change the default blue palette of the console.
5439 This is a 16-member array composed of values
5442 vt.default_grn= [VT]
5443 Format: <green0>,<green1>,<green2>,...,<green15>
5444 Change the default green palette of the console.
5445 This is a 16-member array composed of values
5448 vt.default_red= [VT]
5449 Format: <red0>,<red1>,<red2>,...,<red15>
5450 Change the default red palette of the console.
5451 This is a 16-member array composed of values
5457 Set system-wide default UTF-8 mode for all tty's.
5458 Default is 1, i.e. UTF-8 mode is enabled for all
5459 newly opened terminals.
5461 vt.global_cursor_default=
5464 Set system-wide default for whether a cursor
5465 is shown on new VTs. Default is -1,
5466 i.e. cursors will be created by default unless
5467 overridden by individual drivers. 0 will hide
5468 cursors, 1 will display them.
5470 vt.italic= [VT] Default color for italic text; 0-15.
5473 vt.underline= [VT] Default color for underlined text; 0-15.
5476 watchdog timers [HW,WDT] For information on watchdog timers,
5477 see Documentation/watchdog/watchdog-parameters.rst
5478 or other driver-specific files in the
5479 Documentation/watchdog/ directory.
5483 Set the hard lockup detector stall duration
5484 threshold in seconds. The soft lockup detector
5485 threshold is set to twice the value. A value of 0
5486 disables both lockup detectors. Default is 10
5489 workqueue.watchdog_thresh=
5490 If CONFIG_WQ_WATCHDOG is configured, workqueue can
5491 warn stall conditions and dump internal state to
5492 help debugging. 0 disables workqueue stall
5493 detection; otherwise, it's the stall threshold
5494 duration in seconds. The default value is 30 and
5495 it can be updated at runtime by writing to the
5496 corresponding sysfs file.
5498 workqueue.disable_numa
5499 By default, all work items queued to unbound
5500 workqueues are affine to the NUMA nodes they're
5501 issued on, which results in better behavior in
5502 general. If NUMA affinity needs to be disabled for
5503 whatever reason, this option can be used. Note
5504 that this also can be controlled per-workqueue for
5505 workqueues visible under /sys/bus/workqueue/.
5507 workqueue.power_efficient
5508 Per-cpu workqueues are generally preferred because
5509 they show better performance thanks to cache
5510 locality; unfortunately, per-cpu workqueues tend to
5511 be more power hungry than unbound workqueues.
5513 Enabling this makes the per-cpu workqueues which
5514 were observed to contribute significantly to power
5515 consumption unbound, leading to measurably lower
5516 power usage at the cost of small performance
5519 The default value of this parameter is determined by
5520 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
5522 workqueue.debug_force_rr_cpu
5523 Workqueue used to implicitly guarantee that work
5524 items queued without explicit CPU specified are put
5525 on the local CPU. This guarantee is no longer true
5526 and while local CPU is still preferred work items
5527 may be put on foreign CPUs. This debug option
5528 forces round-robin CPU selection to flush out
5529 usages which depend on the now broken guarantee.
5530 When enabled, memory and cache locality will be
5533 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
5534 default x2apic cluster mode on platforms
5537 x86_intel_mid_timer= [X86-32,APBT]
5538 Choose timer option for x86 Intel MID platform.
5539 Two valid options are apbt timer only and lapic timer
5540 plus one apbt timer for broadcast timer.
5541 x86_intel_mid_timer=apbt_only | lapic_and_apbt
5543 xen_512gb_limit [KNL,X86-64,XEN]
5544 Restricts the kernel running paravirtualized under Xen
5545 to use only up to 512 GB of RAM. The reason to do so is
5546 crash analysis tools and Xen tools for doing domain
5547 save/restore/migration must be enabled to handle larger
5550 xen_emul_unplug= [HW,X86,XEN]
5551 Unplug Xen emulated devices
5552 Format: [unplug0,][unplug1]
5553 ide-disks -- unplug primary master IDE devices
5554 aux-ide-disks -- unplug non-primary-master IDE devices
5555 nics -- unplug network devices
5556 all -- unplug all emulated devices (NICs and IDE disks)
5557 unnecessary -- unplugging emulated devices is
5558 unnecessary even if the host did not respond to
5560 never -- do not unplug even if version check succeeds
5562 xen_legacy_crash [X86,XEN]
5563 Crash from Xen panic notifier, without executing late
5564 panic() code such as dumping handler.
5566 xen_nopvspin [X86,XEN]
5567 Disables the ticketlock slowpath using Xen PV
5571 Disables the PV optimizations forcing the HVM guest to
5572 run as generic HVM guest with no PV drivers.
5573 This option is obsoleted by the "nopv" option, which
5574 has equivalent effect for XEN platform.
5576 xen_no_vector_callback
5577 [KNL,X86,XEN] Disable the vector callback for Xen
5578 event channel interrupts.
5580 xen_scrub_pages= [XEN]
5581 Boolean option to control scrubbing pages before giving them back
5582 to Xen, for use by other domains. Can be also changed at runtime
5583 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
5584 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
5586 xen_timer_slop= [X86-64,XEN]
5587 Set the timer slop (in nanoseconds) for the virtual Xen
5588 timers (default is 100000). This adjusts the minimum
5589 delta of virtualized Xen timers, where lower values
5590 improve timer resolution at the expense of processing
5591 more timer interrupts.
5593 nopv= [X86,XEN,KVM,HYPER_V,VMWARE]
5594 Disables the PV optimizations forcing the guest to run
5595 as generic guest with no PV drivers. Currently support
5596 XEN HVM, KVM, HYPER_V and VMWARE guest.
5598 xen.balloon_boot_timeout= [XEN]
5599 The time (in seconds) to wait before giving up to boot
5600 in case initial ballooning fails to free enough memory.
5601 Applies only when running as HVM or PVH guest and
5602 started with less memory configured than allowed at
5603 max. Default is 180.
5605 xen.event_eoi_delay= [XEN]
5606 How long to delay EOI handling in case of event
5607 storms (jiffies). Default is 10.
5609 xen.event_loop_timeout= [XEN]
5610 After which time (jiffies) the event handling loop
5611 should start to delay EOI handling. Default is 2.
5613 xirc2ps_cs= [NET,PCMCIA]
5615 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
5618 By default on POWER9 and above, the kernel will
5619 natively use the XIVE interrupt controller. This option
5620 allows the fallback firmware mode to be used:
5622 off Fallback to firmware control of XIVE interrupt
5623 controller on both pseries and powernv
5624 platforms. Only useful on POWER9 and above.
5626 xhci-hcd.quirks [USB,KNL]
5627 A hex value specifying bitmask with supplemental xhci
5628 host controller quirks. Meaning of each bit can be
5629 consulted in header drivers/usb/host/xhci.h.
5632 Format: { early | on | rw | ro | off }
5633 Controls if xmon debugger is enabled. Default is off.
5634 Passing only "xmon" is equivalent to "xmon=early".
5635 early Call xmon as early as possible on boot; xmon
5636 debugger is called from setup_arch().
5637 on xmon debugger hooks will be installed so xmon
5638 is only called on a kernel crash. Default mode,
5639 i.e. either "ro" or "rw" mode, is controlled
5640 with CONFIG_XMON_DEFAULT_RO_MODE.
5641 rw xmon debugger hooks will be installed so xmon
5642 is called only on a kernel crash, mode is write,
5643 meaning SPR registers, memory and, other data
5644 can be written using xmon commands.
5645 ro same as "rw" option above but SPR registers,
5646 memory, and other data can't be written using
5648 off xmon is disabled.