1 acpi= [HW,ACPI,X86,ARM64]
2 Advanced Configuration and Power Interface
3 Format: { force | on | off | strict | noirq | rsdt |
5 force -- enable ACPI if default was off
6 on -- enable ACPI but allow fallback to DT [arm64]
7 off -- disable ACPI if default was on
8 noirq -- do not use ACPI for IRQ routing
9 strict -- Be less tolerant of platforms that are not
10 strictly ACPI specification compliant.
11 rsdt -- prefer RSDT over (default) XSDT
12 copy_dsdt -- copy DSDT to memory
13 For ARM64, ONLY "acpi=off", "acpi=on" or "acpi=force"
16 See also Documentation/power/runtime_pm.rst, pci=noacpi
18 acpi_apic_instance= [ACPI, IOAPIC]
20 2: use 2nd APIC table, if available
21 1,0: use 1st APIC table
24 acpi_backlight= [HW,ACPI]
25 { vendor | video | native | none }
26 If set to vendor, prefer vendor-specific driver
27 (e.g. thinkpad_acpi, sony_acpi, etc.) instead
28 of the ACPI video.ko driver.
29 If set to video, use the ACPI video.ko driver.
30 If set to native, use the device's native backlight mode.
31 If set to none, disable the ACPI backlight interface.
33 acpi_force_32bit_fadt_addr
34 force FADT to use 32 bit addresses rather than the
35 64 bit X_* addresses. Some firmware have broken 64
36 bit addresses for force ACPI ignore these and use
37 the older legacy 32 bit addresses.
39 acpica_no_return_repair [HW, ACPI]
40 Disable AML predefined validation mechanism
41 This mechanism can repair the evaluation result to make
42 the return objects more ACPI specification compliant.
43 This option is useful for developers to identify the
44 root cause of an AML interpreter issue when the issue
45 has something to do with the repair mechanism.
47 acpi.debug_layer= [HW,ACPI,ACPI_DEBUG]
48 acpi.debug_level= [HW,ACPI,ACPI_DEBUG]
50 CONFIG_ACPI_DEBUG must be enabled to produce any ACPI
51 debug output. Bits in debug_layer correspond to a
52 _COMPONENT in an ACPI source file, e.g.,
53 #define _COMPONENT ACPI_EVENTS
54 Bits in debug_level correspond to a level in
55 ACPI_DEBUG_PRINT statements, e.g.,
56 ACPI_DEBUG_PRINT((ACPI_DB_INFO, ...
57 The debug_level mask defaults to "info". See
58 Documentation/firmware-guide/acpi/debug.rst for more information about
59 debug layers and levels.
61 Enable processor driver info messages:
62 acpi.debug_layer=0x20000000
63 Enable AML "Debug" output, i.e., stores to the Debug
64 object while interpreting AML:
65 acpi.debug_layer=0xffffffff acpi.debug_level=0x2
66 Enable all messages related to ACPI hardware:
67 acpi.debug_layer=0x2 acpi.debug_level=0xffffffff
69 Some values produce so much output that the system is
70 unusable. The "log_buf_len" parameter may be useful
71 if you need to capture more output.
73 acpi_enforce_resources= [ACPI]
75 Check for resource conflicts between native drivers
76 and ACPI OperationRegions (SystemIO and SystemMemory
77 only). IO ports and memory declared in ACPI might be
78 used by the ACPI subsystem in arbitrary AML code and
79 can interfere with legacy drivers.
80 strict (default): access to resources claimed by ACPI
81 is denied; legacy drivers trying to access reserved
82 resources will fail to bind to device using them.
83 lax: access to resources claimed by ACPI is allowed;
84 legacy drivers trying to access reserved resources
85 will bind successfully but a warning message is logged.
86 no: ACPI OperationRegions are not marked as reserved,
87 no further checks are performed.
89 acpi_force_table_verification [HW,ACPI]
90 Enable table checksum verification during early stage.
91 By default, this is disabled due to x86 early mapping
94 acpi_irq_balance [HW,ACPI]
95 ACPI will balance active IRQs
98 acpi_irq_nobalance [HW,ACPI]
99 ACPI will not move active IRQs (default)
102 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA
103 Format: <irq>,<irq>...
105 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for
107 Format: <irq>,<irq>...
109 acpi_mask_gpe= [HW,ACPI]
110 Due to the existence of _Lxx/_Exx, some GPEs triggered
111 by unsupported hardware/firmware features can result in
112 GPE floodings that cannot be automatically disabled by
114 This facility can be used to prevent such uncontrolled
116 Format: <byte> or <bitmap-list>
118 acpi_no_auto_serialize [HW,ACPI]
119 Disable auto-serialization of AML methods
120 AML control methods that contain the opcodes to create
121 named objects will be marked as "Serialized" by the
122 auto-serialization feature.
123 This feature is enabled by default.
124 This option allows to turn off the feature.
126 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump
129 acpi_no_static_ssdt [HW,ACPI]
130 Disable installation of static SSDTs at early boot time
131 By default, SSDTs contained in the RSDT/XSDT will be
132 installed automatically and they will appear under
133 /sys/firmware/acpi/tables.
134 This option turns off this feature.
135 Note that specifying this option does not affect
136 dynamic table installation which will install SSDT
137 tables to /sys/firmware/acpi/tables/dynamic.
139 acpi_no_watchdog [HW,ACPI,WDT]
140 Ignore the ACPI-based watchdog interface (WDAT) and let
141 a native driver control the watchdog device instead.
143 acpi_rsdp= [ACPI,EFI,KEXEC]
144 Pass the RSDP address to the kernel, mostly used
145 on machines running EFI runtime service to boot the
146 second kernel for kdump.
148 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS
149 Format: To spoof as Windows 98: ="Microsoft Windows"
151 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead
152 of 2 which is mandated by ACPI 6) as the supported ACPI
153 specification revision (when using this switch, it may
154 be necessary to carry out a cold reboot _twice_ in a
155 row to make it take effect on the platform firmware).
157 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings
158 acpi_osi="string1" # add string1
159 acpi_osi="!string2" # remove string2
160 acpi_osi=!* # remove all strings
161 acpi_osi=! # disable all built-in OS vendor
163 acpi_osi=!! # enable all built-in OS vendor
165 acpi_osi= # disable all strings
167 'acpi_osi=!' can be used in combination with single or
168 multiple 'acpi_osi="string1"' to support specific OS
169 vendor string(s). Note that such command can only
170 affect the default state of the OS vendor strings, thus
171 it cannot affect the default state of the feature group
172 strings and the current state of the OS vendor strings,
173 specifying it multiple times through kernel command line
174 is meaningless. This command is useful when one do not
175 care about the state of the feature group strings which
176 should be controlled by the OSPM.
178 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent
179 to 'acpi_osi="Windows 2000" acpi_osi=!', they all
180 can make '_OSI("Windows 2000")' TRUE.
182 'acpi_osi=' cannot be used in combination with other
183 'acpi_osi=' command lines, the _OSI method will not
184 exist in the ACPI namespace. NOTE that such command can
185 only affect the _OSI support state, thus specifying it
186 multiple times through kernel command line is also
189 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)'
192 'acpi_osi=!*' can be used in combination with single or
193 multiple 'acpi_osi="string1"' to support specific
194 string(s). Note that such command can affect the
195 current state of both the OS vendor strings and the
196 feature group strings, thus specifying it multiple times
197 through kernel command line is meaningful. But it may
198 still not able to affect the final state of a string if
199 there are quirks related to this string. This command
200 is useful when one want to control the state of the
201 feature group strings to debug BIOS issues related to
204 1. 'acpi_osi="Module Device" acpi_osi=!*' can make
205 '_OSI("Module Device")' FALSE.
206 2. 'acpi_osi=!* acpi_osi="Module Device"' can make
207 '_OSI("Module Device")' TRUE.
208 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is
210 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"'
212 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!',
213 they all will make '_OSI("Windows 2000")' TRUE.
216 Override the pmtimer bug detection: force the kernel
217 to assume that this machine's pmtimer latches its value
218 and always returns good values.
220 acpi_sci= [HW,ACPI] ACPI System Control Interrupt trigger mode
221 Format: { level | edge | high | low }
223 acpi_skip_timer_override [HW,ACPI]
224 Recognize and ignore IRQ0/pin2 Interrupt Override.
225 For broken nForce2 BIOS resulting in XT-PIC timer.
227 acpi_sleep= [HW,ACPI] Sleep options
228 Format: { s3_bios, s3_mode, s3_beep, s4_hwsig,
229 s4_nohwsig, old_ordering, nonvs,
230 sci_force_enable, nobl }
231 See Documentation/power/video.rst for information on
233 s3_beep is for debugging; it makes the PC's speaker beep
234 as soon as the kernel's real-mode entry point is called.
235 s4_hwsig causes the kernel to check the ACPI hardware
236 signature during resume from hibernation, and gracefully
237 refuse to resume if it has changed. This complies with
238 the ACPI specification but not with reality, since
239 Windows does not do this and many laptops do change it
240 on docking. So the default behaviour is to allow resume
241 and simply warn when the signature changes, unless the
242 s4_hwsig option is enabled.
243 s4_nohwsig prevents ACPI hardware signature from being
244 used (or even warned about) during resume.
245 old_ordering causes the ACPI 1.0 ordering of the _PTS
246 control method, with respect to putting devices into
247 low power states, to be enforced (the ACPI 2.0 ordering
248 of _PTS is used by default).
249 nonvs prevents the kernel from saving/restoring the
250 ACPI NVS memory during suspend/hibernation and resume.
251 sci_force_enable causes the kernel to set SCI_EN directly
252 on resume from S1/S3 (which is against the ACPI spec,
253 but some broken systems don't work without it).
254 nobl causes the internal blacklist of systems known to
255 behave incorrectly in some ways with respect to system
256 suspend and resume to be ignored (use wisely).
258 acpi_use_timer_override [HW,ACPI]
259 Use timer override. For some broken Nvidia NF5 boards
260 that require a timer override, but don't have HPET
262 add_efi_memmap [EFI; X86] Include EFI memory map in
263 kernel's map of available physical RAM.
266 { off | try_unsupported }
267 off: disable AGP support
268 try_unsupported: try to drive unsupported chipsets
269 (may crash computer or cause data corruption)
272 See Documentation/sound/alsa-configuration.rst
275 Allow the default userspace alignment fault handler
276 behaviour to be specified. Bit 0 enables warnings,
277 bit 1 enables fixups, and bit 2 sends a segfault.
279 align_va_addr= [X86-64]
280 Align virtual addresses by clearing slice [14:12] when
281 allocating a VMA at process creation time. This option
282 gives you up to 3% performance improvement on AMD F15h
283 machines (where it is enabled by default) for a
284 CPU-intensive style benchmark, and it can vary highly in
285 a microbenchmark depending on workload and compiler.
287 32: only for 32-bit processes
288 64: only for 64-bit processes
289 on: enable for both 32- and 64-bit processes
290 off: disable for both 32- and 64-bit processes
292 alloc_snapshot [FTRACE]
293 Allocate the ftrace snapshot buffer on boot up when the
294 main buffer is allocated. This is handy if debugging
295 and you need to use tracing_snapshot() on boot up, and
296 do not want to use tracing_snapshot_alloc() as it needs
297 to be done where GFP_KERNEL allocations are allowed.
299 allow_mismatched_32bit_el0 [ARM64]
300 Allow execve() of 32-bit applications and setting of the
301 PER_LINUX32 personality on systems where only a strict
302 subset of the CPUs support 32-bit EL0. When this
303 parameter is present, the set of CPUs supporting 32-bit
304 EL0 is indicated by /sys/devices/system/cpu/aarch32_el0
305 and hot-unplug operations may be restricted.
307 See Documentation/arm64/asymmetric-32bit.rst for more
310 amd_iommu= [HW,X86-64]
311 Pass parameters to the AMD IOMMU driver in the system.
313 fullflush - Deprecated, equivalent to iommu.strict=1
314 off - do not initialize any AMD IOMMU found in
316 force_isolation - Force device isolation for all
317 devices. The IOMMU driver is not
318 allowed anymore to lift isolation
319 requirements as needed. This option
320 does not override iommu=pt
321 force_enable - Force enable the IOMMU on platforms known
322 to be buggy with IOMMU enabled. Use this
325 amd_iommu_dump= [HW,X86-64]
326 Enable AMD IOMMU driver option to dump the ACPI table
327 for AMD IOMMU. With this option enabled, AMD IOMMU
328 driver will print ACPI tables for AMD IOMMU during
329 IOMMU initialization.
331 amd_iommu_intr= [HW,X86-64]
332 Specifies one of the following AMD IOMMU interrupt
334 legacy - Use legacy interrupt remapping mode.
335 vapic - Use virtual APIC mode, which allows IOMMU
336 to inject interrupts directly into guest.
337 This mode requires kvm-amd.avic=1.
338 (Default when IOMMU HW support is present.)
340 amijoy.map= [HW,JOY] Amiga joystick support
341 Map of devices attached to JOY0DAT and JOY1DAT
343 See also Documentation/input/joydev/joystick.rst
345 analog.map= [HW,JOY] Analog joystick and gamepad support
346 Specifies type or capabilities of an analog joystick
347 connected to one of 16 gameports
348 Format: <type1>,<type2>,..<type16>
351 Power management functions (SPARCstation-4/5 + deriv.)
353 Disable APC CPU standby support. SPARCstation-Fox does
354 not play well with APC CPU idle - disable it if you have
355 APC and your system crashes randomly.
357 apic= [APIC,X86] Advanced Programmable Interrupt Controller
358 Change the output verbosity while booting
359 Format: { quiet (default) | verbose | debug }
360 Change the amount of debugging information output
361 when initialising the APIC and IO-APIC components.
362 For X86-32, this can also be used to specify an APIC
364 Format: apic=driver_name
365 Examples: apic=bigsmp
367 apic_extnmi= [APIC,X86] External NMI delivery setting
368 Format: { bsp (default) | all | none }
369 bsp: External NMI is delivered only to CPU 0
370 all: External NMIs are broadcast to all CPUs as a
372 none: External NMI is masked for all CPUs. This is
373 useful so that a dump capture kernel won't be
377 See Documentation/networking/ipv6.rst.
379 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
380 Limit apic dumping. The parameter defines the maximal
381 number of local apics being dumped. Also it is possible
382 to set it to "all" by meaning -- no limit here.
383 Format: { 1 (default) | 2 | ... | all }.
384 The parameter valid if only apic=debug or
385 apic=verbose is specified.
386 Example: apic=debug show_lapic=all
388 apm= [APM] Advanced Power Management
389 See header of arch/x86/kernel/apm_32.c.
391 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
392 Format: <io>,<irq>,<nodeID>
394 arm64.nobti [ARM64] Unconditionally disable Branch Target
395 Identification support
397 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication
400 arm64.nomte [ARM64] Unconditionally disable Memory Tagging Extension
405 atarimouse= [HW,MOUSE] Atari Mouse
407 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
408 EzKey and similar keyboards
410 atkbd.reset= [HW] Reset keyboard during initialization
412 atkbd.set= [HW] Select keyboard code set
413 Format: <int> (2 = AT (default), 3 = PS/2)
415 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
418 atkbd.softraw= [HW] Choose between synthetic and real raw mode
419 Format: <bool> (0 = real, 1 = synthetic (default))
421 atkbd.softrepeat= [HW]
422 Use software keyboard repeat
424 audit= [KNL] Enable the audit sub-system
425 Format: { "0" | "1" | "off" | "on" }
426 0 | off - kernel audit is disabled and can not be
427 enabled until the next reboot
428 unset - kernel audit is initialized but disabled and
429 will be fully enabled by the userspace auditd.
430 1 | on - kernel audit is initialized and partially
431 enabled, storing at most audit_backlog_limit
432 messages in RAM until it is fully enabled by the
436 audit_backlog_limit= [KNL] Set the audit queue size limit.
437 Format: <int> (must be >=0)
440 bau= [X86_UV] Enable the BAU on SGI UV. The default
441 behavior is to disable the BAU (i.e. bau=0).
442 Format: { "0" | "1" }
445 unset - Disable the BAU.
447 baycom_epp= [HW,AX25]
450 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
452 See header of drivers/net/hamradio/baycom_par.c.
454 baycom_ser_fdx= [HW,AX25]
455 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
456 Format: <io>,<irq>,<mode>[,<baud>]
457 See header of drivers/net/hamradio/baycom_ser_fdx.c.
459 baycom_ser_hdx= [HW,AX25]
460 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
461 Format: <io>,<irq>,<mode>
462 See header of drivers/net/hamradio/baycom_ser_hdx.c.
464 blkdevparts= Manual partition parsing of block device(s) for
465 embedded devices based on command line input.
466 See Documentation/block/cmdline-partition.rst
468 boot_delay= Milliseconds to delay each printk during boot.
469 Values larger than 10 seconds (10000) are changed to
474 Extended command line options can be added to an initrd
475 and this will cause the kernel to look for it.
477 See Documentation/admin-guide/bootconfig.rst
480 Disable BERT OS support on buggy BIOSes.
482 bgrt_disable [ACPI][X86]
483 Disable BGRT to avoid flickering OEM logo.
485 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
486 bttv.radio= Most important insmod options are available as
488 bttv.pll= See Documentation/admin-guide/media/bttv.rst
491 bulk_remove=off [PPC] This parameter disables the use of the pSeries
492 firmware feature for flushing multiple hpte entries
495 c101= [NET] Moxa C101 synchronous serial card
497 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
498 Sometimes CPU hardware bugs make them report the cache
499 size incorrectly. The kernel will attempt work arounds
500 to fix known problems, but for some CPUs it is not
501 possible to determine what the correct size should be.
502 This option provides an override for these situations.
505 [NET] Specifies amount of time (in seconds) that
506 the kernel should wait for a network carrier. By default
507 it waits 120 seconds.
509 ca_keys= [KEYS] This parameter identifies a specific key(s) on
510 the system trusted keyring to be used for certificate
512 format: { id:<keyid> | builtin }
514 cca= [MIPS] Override the kernel pages' cache coherency
515 algorithm. Accepted values range from 0 to 7
516 inclusive. See arch/mips/include/asm/pgtable-bits.h
517 for platform specific values (SB1, Loongson3 and
520 ccw_timeout_log [S390]
521 See Documentation/s390/common_io.rst for details.
523 cgroup_disable= [KNL] Disable a particular controller or optional feature
524 Format: {name of the controller(s) or feature(s) to disable}
525 The effects of cgroup_disable=foo are:
526 - foo isn't auto-mounted if you mount all cgroups in
528 - foo isn't visible as an individually mountable
530 - if foo is an optional feature then the feature is
531 disabled and corresponding cgroup files are not
533 {Currently only "memory" controller deal with this and
534 cut the overhead, others just disable the usage. So
535 only cgroup_disable=memory is actually worthy}
536 Specifying "pressure" disables per-cgroup pressure
537 stall information accounting feature
539 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
540 Format: { { controller | "all" | "named" }
541 [,{ controller | "all" | "named" }...] }
542 Like cgroup_disable, but only applies to cgroup v1;
543 the blacklisted controllers remain available in cgroup2.
544 "all" blacklists all controllers and "named" disables
545 named mounts. Specifying both "all" and "named" disables
548 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
550 nosocket -- Disable socket memory accounting.
551 nokmem -- Disable kernel memory accounting.
553 checkreqprot [SELINUX] Set initial checkreqprot flag value.
554 Format: { "0" | "1" }
555 See security/selinux/Kconfig help text.
556 0 -- check protection applied by kernel (includes
557 any implied execute protection).
558 1 -- check protection requested by application.
559 Default value is set via a kernel config option.
560 Value can be changed at runtime via
561 /sys/fs/selinux/checkreqprot.
562 Setting checkreqprot to 1 is deprecated.
565 See Documentation/s390/common_io.rst for details.
568 Prevents the clock framework from automatically gating
569 clocks that have not been explicitly enabled by a Linux
570 device driver but are enabled in hardware at reset or
571 by the bootloader/firmware. Note that this does not
572 force such clocks to be always-on nor does it reserve
573 those clocks in any way. This parameter is useful for
574 debug and development, but should not be needed on a
575 platform with proper driver support. For more
576 information, see Documentation/driver-api/clk.rst.
578 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
580 Forces specified clocksource (if available) to be used
581 when calculating gettimeofday(). If specified
582 clocksource is not available, it defaults to PIT.
583 Format: { pit | tsc | cyclone | pmtmr }
585 clocksource= Override the default clocksource
587 Override the default clocksource and use the clocksource
588 with the name specified.
589 Some clocksource names to choose from, depending on
591 [all] jiffies (this is the base, fallback clocksource)
593 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
594 pxa_timer,timer3,32k_counter,timer0_1
595 [X86-32] pit,hpet,tsc;
596 scx200_hrt on Geode; cyclone on IBM x440
604 clocksource.arm_arch_timer.evtstrm=
607 Enable/disable the eventstream feature of the ARM
608 architected timer so that code using WFE-based polling
609 loops can be debugged more effectively on production
612 clocksource.max_cswd_read_retries= [KNL]
613 Number of clocksource_watchdog() retries due to
614 external delays before the clock will be marked
615 unstable. Defaults to two retries, that is,
616 three attempts to read the clock under test.
618 clocksource.verify_n_cpus= [KNL]
619 Limit the number of CPUs checked for clocksources
620 marked with CLOCK_SOURCE_VERIFY_PERCPU that
621 are marked unstable due to excessive skew.
622 A negative value says to check all CPUs, while
623 zero says not to check any. Values larger than
624 nr_cpu_ids are silently truncated to nr_cpu_ids.
625 The actual CPUs are chosen randomly, with
626 no replacement if the same CPU is chosen twice.
628 clocksource-wdtest.holdoff= [KNL]
629 Set the time in seconds that the clocksource
630 watchdog test waits before commencing its tests.
631 Defaults to zero when built as a module and to
632 10 seconds when built into the kernel.
634 clearcpuid=X[,X...] [X86]
635 Disable CPUID feature X for the kernel. See
636 arch/x86/include/asm/cpufeatures.h for the valid bit
637 numbers X. Note the Linux-specific bits are not necessarily
638 stable over kernel options, but the vendor-specific
640 X can also be a string as appearing in the flags: line
641 in /proc/cpuinfo which does not have the above
642 instability issue. However, not all features have names
644 Note that using this option will taint your kernel.
645 Also note that user programs calling CPUID directly
646 or using the feature without checking anything
647 will still see it. This just prevents it from
648 being used by the kernel or shown in /proc/cpuinfo.
649 Also note the kernel might malfunction if you disable
652 cma=nn[MG]@[start[MG][-end[MG]]]
654 Sets the size of kernel global memory area for
655 contiguous memory allocations and optionally the
656 placement constraint by the physical address range of
657 memory allocations. A value of 0 disables CMA
658 altogether. For more information, see
659 kernel/dma/contiguous.c
663 Sets the size of kernel per-numa memory area for
664 contiguous memory allocations. A value of 0 disables
665 per-numa CMA altogether. And If this option is not
666 specificed, the default value is 0.
667 With per-numa CMA enabled, DMA users on node nid will
668 first try to allocate buffer from the pernuma area
669 which is located in node nid, if the allocation fails,
670 they will fallback to the global default memory area.
672 cmo_free_hint= [PPC] Format: { yes | no }
673 Specify whether pages are marked as being inactive
674 when they are freed. This is used in CMO environments
675 to determine OS memory pressure for page stealing by
679 coherent_pool=nn[KMG] [ARM,KNL]
680 Sets the size of memory pool for coherent, atomic dma
681 allocations, by default set to 256K.
683 com20020= [HW,NET] ARCnet - COM20020 chipset
685 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
687 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
691 ARCnet - COM90xx chipset (memory-mapped buffers)
692 Format: <io>[,<irq>[,<memstart>]]
694 condev= [HW,S390] console device
697 console= [KNL] Output console device and options.
699 tty<n> Use the virtual console device <n>.
703 Use the specified serial port. The options are of
704 the form "bbbbpnf", where "bbbb" is the baud rate,
705 "p" is parity ("n", "o", or "e"), "n" is number of
706 bits, and "f" is flow control ("r" for RTS or
707 omit it). Default is "9600n8".
709 See Documentation/admin-guide/serial-console.rst for more
711 Documentation/networking/netconsole.rst for an
714 uart[8250],io,<addr>[,options]
715 uart[8250],mmio,<addr>[,options]
716 uart[8250],mmio16,<addr>[,options]
717 uart[8250],mmio32,<addr>[,options]
718 uart[8250],0x<addr>[,options]
719 Start an early, polled-mode console on the 8250/16550
720 UART at the specified I/O port or MMIO address,
721 switching to the matching ttyS device later.
722 MMIO inter-register address stride is either 8-bit
723 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
724 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
725 to be equivalent to 'mmio'. 'options' are specified in
726 the same format described for ttyS above; if unspecified,
727 the h/w is not re-initialized.
729 hvc<n> Use the hypervisor console device <n>. This is for
730 both Xen and PowerPC hypervisors.
733 Use to disable console output, i.e., to have kernel
734 console messages discarded.
735 This must be the only console= parameter used on the
738 If the device connected to the port is not a TTY but a braille
739 device, prepend "brl," before the device type, for instance
741 For now, only VisioBraille is supported.
744 [KNL] Change console messages format
746 By default we print messages on consoles in
747 "[time stamp] text\n" format (time stamp may not be
748 printed, depending on CONFIG_PRINTK_TIME or
749 `printk_time' param).
751 Switch to syslog format: "<%u>[time stamp] text\n"
752 IOW, each message will have a facility and loglevel
753 prefix. The format is similar to one used by syslog()
754 syscall, or to executing "dmesg -S --raw" or to reading
757 consoleblank= [KNL] The console blank (screen saver) timeout in
758 seconds. A value of 0 disables the blank timer.
762 [KNL] Change the default value for
763 /proc/<pid>/coredump_filter.
764 See also Documentation/filesystems/proc.rst.
766 coresight_cpu_debug.enable
769 Enable/disable the CPU sampling based debugging.
770 0: default value, disable debugging
771 1: enable debugging at boot time
773 cpuidle.off=1 [CPU_IDLE]
774 disable the cpuidle sub-system
777 [CPU_IDLE] Name of the cpuidle governor to use.
779 cpufreq.off=1 [CPU_FREQ]
780 disable the cpufreq sub-system
782 cpufreq.default_governor=
783 [CPU_FREQ] Name of the default cpufreq governor or
784 policy to use. This governor must be registered in the
785 kernel before the cpufreq driver probes.
788 [X86] Delay for N microsec between assert and de-assert
789 of APIC INIT to start processors. This delay occurs
790 on every CPU online, such as boot, and resume from suspend.
793 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
795 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
797 crashkernel=size[KMG][@offset[KMG]]
798 [KNL] Using kexec, Linux can switch to a 'crash kernel'
799 upon panic. This parameter reserves the physical
800 memory region [offset, offset + size] for that kernel
801 image. If '@offset' is omitted, then a suitable offset
802 is selected automatically.
803 [KNL, X86-64] Select a region under 4G first, and
804 fall back to reserve region above 4G when '@offset'
805 hasn't been specified.
806 See Documentation/admin-guide/kdump/kdump.rst for further details.
808 crashkernel=range1:size1[,range2:size2,...][@offset]
809 [KNL] Same as above, but depends on the memory
810 in the running system. The syntax of range is
811 start-[end] where start and end are both
812 a memory unit (amount[KMG]). See also
813 Documentation/admin-guide/kdump/kdump.rst for an example.
815 crashkernel=size[KMG],high
816 [KNL, X86-64, ARM64] range could be above 4G. Allow kernel
817 to allocate physical memory region from top, so could
818 be above 4G if system have more than 4G ram installed.
819 Otherwise memory region will be allocated below 4G, if
821 It will be ignored if crashkernel=X is specified.
822 crashkernel=size[KMG],low
823 [KNL, X86-64] range under 4G. When crashkernel=X,high
824 is passed, kernel could allocate physical memory region
825 above 4G, that cause second kernel crash on system
826 that require some amount of low memory, e.g. swiotlb
827 requires at least 64M+32K low memory, also enough extra
828 low memory is needed to make sure DMA buffers for 32-bit
829 devices won't run out. Kernel would try to allocate
830 at least 256M below 4G automatically.
831 This one lets the user specify own low range under 4G
832 for second kernel instead.
833 0: to disable low allocation.
834 It will be ignored when crashkernel=X,high is not used
835 or memory reserved is below 4G.
837 [KNL, ARM64] range in low memory.
838 This one lets the user specify a low range in the
839 DMA zone for the crash dump kernel.
840 It will be ignored when crashkernel=X,high is not used
841 or memory reserved is located in the DMA zones.
844 [KNL] Disable crypto self-tests
849 cs89x0_media= [HW,NET]
850 Format: { rj45 | aui | bnc }
852 csdlock_debug= [KNL] Enable debug add-ons of cross-CPU function call
853 handling. When switched on, additional debug data is
854 printed to the console in case a hanging CPU is
855 detected, and that CPU is pinged again in order to try
856 to resolve the hang situation.
857 0: disable csdlock debugging (default)
858 1: enable basic csdlock debugging (minor impact)
859 ext: enable extended csdlock debugging (more impact,
863 See header of drivers/s390/block/dasd_devmap.c.
865 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
866 (one device per port)
867 Format: <port#>,<type>
868 See also Documentation/input/devices/joystick-parport.rst
870 debug [KNL] Enable kernel debugging (events log level).
873 [KNL] Enable printing [hashed] pointers early in the
874 boot sequence. If enabled, we use a weak hash instead
875 of siphash to hash pointers. Use this option if you are
876 seeing instances of '(___ptrval___)') and need to see a
877 value (hashed pointer) instead. Cryptographically
878 insecure, please do not use on production kernels.
881 [KNL] verbose locking self-tests
883 Print debugging info while doing the locking API
885 Bitmask for the various LOCKTYPE_ tests. Defaults to 0
886 (no extra messages), setting it to -1 (all bits set)
887 will print _a_lot_ more information - normally only
888 useful to lockdep developers.
890 debug_objects [KNL] Enable object debugging
893 [KNL] Disable object debugging
895 debug_guardpage_minorder=
896 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
897 parameter allows control of the order of pages that will
898 be intentionally kept free (and hence protected) by the
899 buddy allocator. Bigger value increase the probability
900 of catching random memory corruption, but reduce the
901 amount of memory for normal system use. The maximum
902 possible value is MAX_ORDER/2. Setting this parameter
903 to 1 or 2 should be enough to identify most random
904 memory corruption problems caused by bugs in kernel or
905 driver code when a CPU writes to (or reads from) a
906 random memory location. Note that there exists a class
907 of memory corruptions problems caused by buggy H/W or
908 F/W or by drivers badly programing DMA (basically when
909 memory is written at bus level and the CPU MMU is
910 bypassed) which are not detectable by
911 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
912 tracking down these problems.
915 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
916 enables the feature at boot time. By default, it is
917 disabled and the system will work mostly the same as a
918 kernel built without CONFIG_DEBUG_PAGEALLOC.
919 Note: to get most of debug_pagealloc error reports, it's
920 useful to also enable the page_owner functionality.
921 on: enable the feature
923 debugfs= [KNL] This parameter enables what is exposed to userspace
924 and debugfs internal clients.
925 Format: { on, no-mount, off }
926 on: All functions are enabled.
928 Filesystem is not registered but kernel clients can
929 access APIs and a crashkernel can be used to read
930 its content. There is nothing to mount.
931 off: Filesystem is not registered and clients
932 get a -EPERM as result when trying to register files
933 or directories within debugfs.
934 This is equivalent of the runtime functionality if
935 debugfs was not enabled in the kernel at all.
936 Default value is set in build-time with a kernel configuration.
938 debugpat [X86] Enable PAT debugging
940 decnet.addr= [HW,NET]
941 Format: <area>[,<node>]
942 See also Documentation/networking/decnet.rst.
945 [HW] The size of the default HugeTLB page. This is
946 the size represented by the legacy /proc/ hugepages
947 APIs. In addition, this is the default hugetlb size
948 used for shmget(), mmap() and mounting hugetlbfs
949 filesystems. If not specified, defaults to the
950 architecture's default huge page size. Huge page
951 sizes are architecture dependent. See also
952 Documentation/admin-guide/mm/hugetlbpage.rst.
955 deferred_probe_timeout=
956 [KNL] Debugging option to set a timeout in seconds for
957 deferred probe to give up waiting on dependencies to
958 probe. Only specific dependencies (subsystems or
959 drivers) that have opted in will be ignored. A timeout of 0
960 will timeout at the end of initcalls. This option will also
961 dump out devices still on the deferred probe list after
964 dell_smm_hwmon.ignore_dmi=
965 [HW] Continue probing hardware even if DMI data
966 indicates that the driver is running on unsupported
969 dell_smm_hwmon.force=
970 [HW] Activate driver even if SMM BIOS signature does
971 not match list of supported models and enable otherwise
972 blacklisted features.
974 dell_smm_hwmon.power_status=
975 [HW] Report power status in /proc/i8k
976 (disabled by default).
978 dell_smm_hwmon.restricted=
979 [HW] Allow controlling fans only if SYS_ADMIN
982 dell_smm_hwmon.fan_mult=
983 [HW] Factor to multiply fan speed with.
985 dell_smm_hwmon.fan_max=
986 [HW] Maximum configurable fan speed.
989 Format: { on | off | def_only | inf_only | always }
990 on: s390 zlib hardware support for compression on
991 level 1 and decompression (default)
992 off: No s390 zlib hardware support
993 def_only: s390 zlib hardware support for deflate
994 only (compression on level 1)
995 inf_only: s390 zlib hardware support for inflate
997 always: Same as 'on' but ignores the selected compression
998 level always using hardware support (used for debugging)
1000 dhash_entries= [KNL]
1001 Set number of hash buckets for dentry cache.
1003 disable_1tb_segments [PPC]
1004 Disables the use of 1TB hash page table segments. This
1005 causes the kernel to fall back to 256MB segments which
1006 can be useful when debugging issues that require an SLB
1010 Limits the number of kernel SLB entries, and flushes
1011 them frequently to increase the rate of SLB faults
1012 on kernel addresses.
1015 See Documentation/networking/ipv6.rst.
1018 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
1019 hardening is enabled for this boot. Hardened
1020 usercopy checking is used to protect the kernel
1021 from reading or writing beyond known memory
1022 allocation boundaries as a proactive defense
1023 against bounds-checking flaws in the kernel's
1024 copy_to_user()/copy_from_user() interface.
1025 on Perform hardened usercopy checks (default).
1026 off Disable hardened usercopy checks.
1029 Disable RADIX MMU mode on POWER9
1031 radix_hcall_invalidate=on [PPC/PSERIES]
1032 Disable RADIX GTSE feature and use hcall for TLB
1036 Disable TLBIE instruction. Currently does not work
1037 with KVM, with HASH MMU, or with coherent accelerators.
1039 disable_cpu_apicid= [X86,APIC,SMP]
1041 The number of initial APIC ID for the
1042 corresponding CPU to be disabled at boot,
1043 mostly used for the kdump 2nd kernel to
1044 disable BSP to wake up multiple CPUs without
1045 causing system reset or hang due to sending
1046 INIT from AP to BSP.
1048 disable_ddw [PPC/PSERIES]
1049 Disable Dynamic DMA Window support. Use this
1050 to workaround buggy firmware.
1052 disable_ipv6= [IPV6]
1053 See Documentation/networking/ipv6.rst.
1055 disable_mtrr_cleanup [X86]
1056 The kernel tries to adjust MTRR layout from continuous
1057 to discrete, to make X server driver able to add WB
1058 entry later. This parameter disables that.
1060 disable_mtrr_trim [X86, Intel and AMD only]
1061 By default the kernel will trim any uncacheable
1062 memory out of your available memory pool based on
1063 MTRR settings. This parameter disables that behavior,
1064 possibly causing your machine to run very slowly.
1066 disable_timer_pin_1 [X86]
1067 Disable PIN 1 of APIC timer
1068 Can be useful to work around chipset bugs.
1070 dis_ucode_ldr [X86] Disable the microcode loader.
1072 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
1073 this option disables the debugging code at boot.
1075 dma_debug_entries=<number>
1076 This option allows to tune the number of preallocated
1077 entries for DMA-API debugging code. One entry is
1078 required per DMA-API allocation. Use this if the
1079 DMA-API debugging code disables itself because the
1080 architectural default is too low.
1082 dma_debug_driver=<driver_name>
1083 With this option the DMA-API debugging driver
1084 filter feature can be enabled at boot time. Just
1085 pass the driver to filter for as the parameter.
1086 The filter can be disabled or changed to another
1087 driver later using sysfs.
1089 driver_async_probe= [KNL]
1090 List of driver names to be probed asynchronously.
1091 Format: <driver_name1>,<driver_name2>...
1093 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
1094 Broken monitors, graphic adapters, KVMs and EDIDless
1095 panels may send no or incorrect EDID data sets.
1096 This parameter allows to specify an EDID data sets
1097 in the /lib/firmware directory that are used instead.
1098 Generic built-in EDID data sets are used, if one of
1099 edid/1024x768.bin, edid/1280x1024.bin,
1100 edid/1680x1050.bin, or edid/1920x1080.bin is given
1101 and no file with the same name exists. Details and
1102 instructions how to build your own EDID data are
1103 available in Documentation/admin-guide/edid.rst. An EDID
1104 data set will only be used for a particular connector,
1105 if its name and a colon are prepended to the EDID
1106 name. Each connector may use a unique EDID data
1107 set by separating the files with a comma. An EDID
1108 data set with no connector name will be used for
1109 any connectors not explicitly specified.
1114 Format: {"off" | "known"}
1115 Control how the dt_cpu_ftrs device-tree binding is
1116 used for CPU feature discovery and setup (if it
1118 off: Do not use it, fall back to legacy cpu table.
1119 known: Do not pass through unknown features to guests
1120 or userspace, only those that the kernel is aware of.
1122 dump_apple_properties [X86]
1123 Dump name and content of EFI device properties on
1124 x86 Macs. Useful for driver authors to determine
1125 what data is available or for reverse-engineering.
1127 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
1128 <module>.dyndbg[="val"]
1129 Enable debug messages at boot time. See
1130 Documentation/admin-guide/dynamic-debug-howto.rst
1133 nopku [X86] Disable Memory Protection Keys CPU feature found
1136 <module>.async_probe [KNL]
1137 Enable asynchronous probe on this module.
1139 early_ioremap_debug [KNL]
1140 Enable debug messages in early_ioremap support. This
1141 is useful for tracking down temporary early mappings
1142 which are not unmapped.
1144 earlycon= [KNL] Output early console device and options.
1146 When used with no options, the early console is
1147 determined by stdout-path property in device tree's
1148 chosen node or the ACPI SPCR table if supported by
1151 cdns,<addr>[,options]
1152 Start an early, polled-mode console on a Cadence
1153 (xuartps) serial port at the specified address. Only
1154 supported option is baud rate. If baud rate is not
1155 specified, the serial port must already be setup and
1158 uart[8250],io,<addr>[,options]
1159 uart[8250],mmio,<addr>[,options]
1160 uart[8250],mmio32,<addr>[,options]
1161 uart[8250],mmio32be,<addr>[,options]
1162 uart[8250],0x<addr>[,options]
1163 Start an early, polled-mode console on the 8250/16550
1164 UART at the specified I/O port or MMIO address.
1165 MMIO inter-register address stride is either 8-bit
1166 (mmio) or 32-bit (mmio32 or mmio32be).
1167 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1168 to be equivalent to 'mmio'. 'options' are specified
1169 in the same format described for "console=ttyS<n>"; if
1170 unspecified, the h/w is not initialized.
1174 Start an early, polled-mode console on a pl011 serial
1175 port at the specified address. The pl011 serial port
1176 must already be setup and configured. Options are not
1177 yet supported. If 'mmio32' is specified, then only
1178 the driver will use only 32-bit accessors to read/write
1179 the device registers.
1182 Start an early console on a litex serial port at the
1183 specified address. The serial port must already be
1184 setup and configured. Options are not yet supported.
1187 Start an early, polled-mode console on a meson serial
1188 port at the specified address. The serial port must
1189 already be setup and configured. Options are not yet
1193 Start an early, polled-mode console on an msm serial
1194 port at the specified address. The serial port
1195 must already be setup and configured. Options are not
1198 msm_serial_dm,<addr>
1199 Start an early, polled-mode console on an msm serial
1200 dm port at the specified address. The serial port
1201 must already be setup and configured. Options are not
1205 Start an early, polled-mode console on a serial port
1206 of an Actions Semi SoC, such as S500 or S900, at the
1207 specified address. The serial port must already be
1208 setup and configured. Options are not yet supported.
1211 Start an early, polled-mode console on a serial port
1212 of an RDA Micro SoC, such as RDA8810PL, at the
1213 specified address. The serial port must already be
1214 setup and configured. Options are not yet supported.
1217 Use RISC-V SBI (Supervisor Binary Interface) for early
1220 smh Use ARM semihosting calls for early console.
1228 Use early console provided by serial driver available
1229 on Samsung SoCs, requires selecting proper type and
1230 a correct base address of the selected UART port. The
1231 serial port must already be setup and configured.
1232 Options are not yet supported.
1235 Start an early, polled-mode console on a lantiq serial
1236 (lqasc) port at the specified address. The serial port
1237 must already be setup and configured. Options are not
1242 Use early console provided by Freescale LP UART driver
1243 found on Freescale Vybrid and QorIQ LS1021A processors.
1244 A valid base address must be provided, and the serial
1245 port must already be setup and configured.
1249 Start an early, polled-mode, output-only console on the
1250 Freescale i.MX UART at the specified address. The UART
1251 must already be setup and configured.
1254 Start an early, polled-mode console on the
1255 Armada 3700 serial port at the specified
1256 address. The serial port must already be setup
1257 and configured. Options are not yet supported.
1260 Start an early, polled-mode console on a Qualcomm
1261 Generic Interface (GENI) based serial port at the
1262 specified address. The serial port must already be
1263 setup and configured. Options are not yet supported.
1266 Start an early, unaccelerated console on the EFI
1267 memory mapped framebuffer (if available). On cache
1268 coherent non-x86 systems that use system memory for
1269 the framebuffer, pass the 'ram' option so that it is
1270 mapped with the correct attributes.
1273 Use early console provided by Freescale LINFlexD UART
1274 serial driver for NXP S32V234 SoCs. A valid base
1275 address must be provided, and the serial port must
1276 already be setup and configured.
1278 earlyprintk= [X86,SH,ARM,M68k,S390]
1282 earlyprintk=serial[,ttySn[,baudrate]]
1283 earlyprintk=serial[,0x...[,baudrate]]
1284 earlyprintk=ttySn[,baudrate]
1285 earlyprintk=dbgp[debugController#]
1286 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1287 earlyprintk=xdbc[xhciController#]
1289 earlyprintk is useful when the kernel crashes before
1290 the normal console is initialized. It is not enabled by
1291 default because it has some cosmetic problems.
1293 Append ",keep" to not disable it when the real console
1296 Only one of vga, efi, serial, or usb debug port can
1299 Currently only ttyS0 and ttyS1 may be specified by
1300 name. Other I/O ports may be explicitly specified
1301 on some architectures (x86 and arm at least) by
1302 replacing ttySn with an I/O port address, like this:
1303 earlyprintk=serial,0x1008,115200
1304 You can find the port for a given device in
1305 /proc/tty/driver/serial:
1306 2: uart:ST16650V2 port:00001008 irq:18 ...
1308 Interaction with the standard serial driver is not
1311 The VGA and EFI output is eventually overwritten by
1314 The xen option can only be used in Xen domains.
1316 The sclp output can only be used on s390.
1318 The optional "force" to "pciserial" enables use of a
1319 PCI device even when its classcode is not of the
1322 edac_report= [HW,EDAC] Control how to report EDAC event
1323 Format: {"on" | "off" | "force"}
1324 on: enable EDAC to report H/W event. May be overridden
1325 by other higher priority error reporting module.
1326 off: disable H/W event reporting through EDAC.
1327 force: enforce the use of EDAC to report H/W event.
1330 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1333 This is designed to be used in conjunction with
1334 the boot argument: earlyprintk=vga
1336 This parameter works in place of the kgdboc parameter
1337 but can only be used if the backing tty is available
1338 very early in the boot process. For early debugging
1339 via a serial port see kgdboc_earlycon instead.
1342 Format: {"off" | "on" | "skip[mbr]"}
1345 Format: { "debug", "disable_early_pci_dma",
1346 "nochunk", "noruntime", "nosoftreserve",
1347 "novamap", "no_disable_early_pci_dma" }
1348 debug: enable misc debug output.
1349 disable_early_pci_dma: disable the busmaster bit on all
1350 PCI bridges while in the EFI boot stub.
1351 nochunk: disable reading files in "chunks" in the EFI
1352 boot stub, as chunking can cause problems with some
1353 firmware implementations.
1354 noruntime : disable EFI runtime services support
1355 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1356 attribute may cause the kernel to reserve the
1357 memory range for a memory mapping driver to
1358 claim. Specify efi=nosoftreserve to disable this
1359 reservation and treat the memory by its base type
1360 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1361 novamap: do not call SetVirtualAddressMap().
1362 no_disable_early_pci_dma: Leave the busmaster bit set
1363 on all PCI bridges while in the EFI boot stub
1365 efi_no_storage_paranoia [EFI; X86]
1366 Using this parameter you can use more than 50% of
1367 your efi variable storage. Use this parameter only if
1368 you are really sure that your UEFI does sane gc and
1369 fulfills the spec otherwise your board may brick.
1371 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1372 Add arbitrary attribute to specific memory range by
1373 updating original EFI memory map.
1374 Region of memory which aa attribute is added to is
1377 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1378 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1379 attribute is added to range 0x100000000-0x180000000 and
1380 0x10a0000000-0x1120000000.
1382 If efi_fake_mem=8G@9G:0x40000 is specified, the
1383 EFI_MEMORY_SP(0x40000) attribute is added to
1384 range 0x240000000-0x43fffffff.
1386 Using this parameter you can do debugging of EFI memmap
1387 related features. For example, you can do debugging of
1388 Address Range Mirroring feature even if your box
1389 doesn't support it, or mark specific memory as
1392 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1393 that is to be dynamically loaded by Linux. If there are
1394 multiple variables with the same name but with different
1395 vendor GUIDs, all of them will be loaded. See
1396 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1399 eisa_irq_edge= [PARISC,HW]
1400 See header of drivers/parisc/eisa.c.
1403 See comment before function elanfreq_setup() in
1404 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1406 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1407 Specifies physical address of start of kernel core
1408 image elf header and optionally the size. Generally
1409 kexec loader will pass this option to capture kernel.
1410 See Documentation/admin-guide/kdump/kdump.rst for details.
1412 enable_mtrr_cleanup [X86]
1413 The kernel tries to adjust MTRR layout from continuous
1414 to discrete, to make X server driver able to add WB
1415 entry later. This parameter enables that.
1417 enable_timer_pin_1 [X86]
1418 Enable PIN 1 of APIC timer
1419 Can be useful to work around chipset bugs
1420 (in particular on some ATI chipsets).
1421 The kernel tries to set a reasonable default.
1423 enforcing [SELINUX] Set initial enforcing status.
1425 See security/selinux/Kconfig help text.
1426 0 -- permissive (log only, no denials).
1427 1 -- enforcing (deny and log).
1429 Value can be changed at runtime via
1430 /sys/fs/selinux/enforce.
1433 Disable Error Record Serialization Table (ERST)
1436 ether= [HW,NET] Ethernet cards parameters
1437 This option is obsoleted by the "netdev=" option, which
1438 has equivalent usage. See its documentation for details.
1442 Permit 'security.evm' to be updated regardless of
1443 current integrity status.
1448 fail_make_request=[KNL]
1449 General fault injection mechanism.
1450 Format: <interval>,<probability>,<space>,<times>
1451 See also Documentation/fault-injection/.
1454 Format: { initns | none }
1455 See Documentation/admin-guide/sysctl/net.rst for
1456 fb_tunnels_only_for_init_ns
1459 See Documentation/admin-guide/blockdev/floppy.rst.
1461 force_pal_cache_flush
1462 [IA-64] Avoid check_sal_cache_flush which may hang on
1463 buggy SAL_CACHE_FLUSH implementations. Using this
1464 parameter will force ia64_sal_cache_flush to call
1465 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1468 Forcefully enable Physical Address Extension (PAE).
1469 Many Pentium M systems disable PAE but may have a
1470 functionally usable PAE implementation.
1471 Warning: use of this parameter will taint the kernel
1472 and may cause unknown problems.
1475 [FTRACE] will set and start the specified tracer
1476 as early as possible in order to facilitate early
1479 ftrace_boot_snapshot
1480 [FTRACE] On boot up, a snapshot will be taken of the
1481 ftrace ring buffer that can be read at:
1482 /sys/kernel/tracing/snapshot.
1483 This is useful if you need tracing information from kernel
1484 boot up that is likely to be overridden by user space
1485 start up functionality.
1487 ftrace_dump_on_oops[=orig_cpu]
1488 [FTRACE] will dump the trace buffers on oops.
1489 If no parameter is passed, ftrace will dump
1490 buffers of all CPUs, but if you pass orig_cpu, it will
1491 dump only the buffer of the CPU that triggered the
1494 ftrace_filter=[function-list]
1495 [FTRACE] Limit the functions traced by the function
1496 tracer at boot up. function-list is a comma-separated
1497 list of functions. This list can be changed at run
1498 time by the set_ftrace_filter file in the debugfs
1501 ftrace_notrace=[function-list]
1502 [FTRACE] Do not trace the functions specified in
1503 function-list. This list can be changed at run time
1504 by the set_ftrace_notrace file in the debugfs
1507 ftrace_graph_filter=[function-list]
1508 [FTRACE] Limit the top level callers functions traced
1509 by the function graph tracer at boot up.
1510 function-list is a comma-separated list of functions
1511 that can be changed at run time by the
1512 set_graph_function file in the debugfs tracing directory.
1514 ftrace_graph_notrace=[function-list]
1515 [FTRACE] Do not trace from the functions specified in
1516 function-list. This list is a comma-separated list of
1517 functions that can be changed at run time by the
1518 set_graph_notrace file in the debugfs tracing directory.
1520 ftrace_graph_max_depth=<uint>
1521 [FTRACE] Used with the function graph tracer. This is
1522 the max depth it will trace into a function. This value
1523 can be changed at run time by the max_graph_depth file
1524 in the tracefs tracing directory. default: 0 (no limit)
1526 fw_devlink= [KNL] Create device links between consumer and supplier
1527 devices by scanning the firmware to infer the
1528 consumer/supplier relationships. This feature is
1529 especially useful when drivers are loaded as modules as
1530 it ensures proper ordering of tasks like device probing
1531 (suppliers first, then consumers), supplier boot state
1532 clean up (only after all consumers have probed),
1533 suspend/resume & runtime PM (consumers first, then
1535 Format: { off | permissive | on | rpm }
1536 off -- Don't create device links from firmware info.
1537 permissive -- Create device links from firmware info
1538 but use it only for ordering boot state clean
1539 up (sync_state() calls).
1540 on -- Create device links from firmware info and use it
1541 to enforce probe and suspend/resume ordering.
1542 rpm -- Like "on", but also use to order runtime PM.
1544 fw_devlink.strict=<bool>
1545 [KNL] Treat all inferred dependencies as mandatory
1546 dependencies. This only applies for fw_devlink=on|rpm.
1550 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1551 support via parallel port (up to 5 devices per port)
1552 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1553 See also Documentation/input/devices/joystick-parport.rst
1557 gart_fix_e820= [X86-64] disable the fix e820 for K8 GART
1561 gcov_persist= [GCOV] When non-zero (default), profiling data for
1562 kernel modules is saved and remains accessible via
1563 debugfs, even when the module is unloaded/reloaded.
1564 When zero, profiling data is discarded and associated
1565 debugfs files are removed at module unload time.
1567 goldfish [X86] Enable the goldfish android emulator platform.
1568 Don't use this when you are not running on the
1571 gpio-mockup.gpio_mockup_ranges
1572 [HW] Sets the ranges of gpiochip of for this device.
1573 Format: <start1>,<end1>,<start2>,<end2>...
1574 gpio-mockup.gpio_mockup_named_lines
1575 [HW] Let the driver know GPIO lines should be named.
1577 gpt [EFI] Forces disk with valid GPT signature but
1578 invalid Protective MBR to be treated as GPT. If the
1579 primary GPT is corrupted, it enables the backup/alternate
1580 GPT to be used instead.
1582 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1583 the "Enable 0" bit of the configuration register.
1586 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1587 the "Enable 0" bit of the configuration register.
1590 grcan.select= [HW] Select which physical interface to use.
1593 grcan.txsize= [HW] Sets the size of the tx buffer.
1594 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1596 grcan.rxsize= [HW] Sets the size of the rx buffer.
1597 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1600 hardlockup_all_cpu_backtrace=
1601 [KNL] Should the hard-lockup detector generate
1602 backtraces on all cpus.
1605 hashdist= [KNL,NUMA] Large hashes allocated during boot
1606 are distributed across NUMA nodes. Defaults on
1607 for 64-bit NUMA, off otherwise.
1608 Format: 0 | 1 (for off | on)
1610 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1612 hd= [EIDE] (E)IDE hard drive subsystem geometry
1613 Format: <cyl>,<head>,<sect>
1616 Disable Hardware Error Source Table (HEST) support;
1617 corresponding firmware-first mode error processing
1618 logic will be disabled.
1620 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1621 size of <nn>. This works even on boxes that have no
1622 highmem otherwise. This also works to reduce highmem
1623 size on bigger boxes.
1625 highres= [KNL] Enable/disable high resolution timer mode.
1626 Valid parameters: "on", "off"
1631 hpet= [X86-32,HPET] option to control HPET usage
1632 Format: { enable (default) | disable | force |
1634 disable: disable HPET and use PIT instead
1635 force: allow force enabled of undocumented chips (ICH4,
1637 verbose: show contents of HPET registers during setup
1639 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1640 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1642 hugetlb_cma= [HW,CMA] The size of a CMA area used for allocation
1643 of gigantic hugepages. Or using node format, the size
1644 of a CMA area per node can be specified.
1645 Format: nn[KMGTPE] or (node format)
1646 <node>:nn[KMGTPE][,<node>:nn[KMGTPE]]
1648 Reserve a CMA area of given size and allocate gigantic
1649 hugepages using the CMA allocator. If enabled, the
1650 boot-time allocation of gigantic hugepages is skipped.
1652 hugepages= [HW] Number of HugeTLB pages to allocate at boot.
1653 If this follows hugepagesz (below), it specifies
1654 the number of pages of hugepagesz to be allocated.
1655 If this is the first HugeTLB parameter on the command
1656 line, it specifies the number of pages to allocate for
1657 the default huge page size. If using node format, the
1658 number of pages to allocate per-node can be specified.
1659 See also Documentation/admin-guide/mm/hugetlbpage.rst.
1660 Format: <integer> or (node format)
1661 <node>:<integer>[,<node>:<integer>]
1664 [HW] The size of the HugeTLB pages. This is used in
1665 conjunction with hugepages (above) to allocate huge
1666 pages of a specific size at boot. The pair
1667 hugepagesz=X hugepages=Y can be specified once for
1668 each supported huge page size. Huge page sizes are
1669 architecture dependent. See also
1670 Documentation/admin-guide/mm/hugetlbpage.rst.
1673 hugetlb_free_vmemmap=
1674 [KNL] Reguires CONFIG_HUGETLB_PAGE_FREE_VMEMMAP
1676 Allows heavy hugetlb users to free up some more
1677 memory (7 * PAGE_SIZE for each 2MB hugetlb page).
1678 Format: { on | off (default) }
1680 on: enable the feature
1681 off: disable the feature
1683 Built with CONFIG_HUGETLB_PAGE_FREE_VMEMMAP_DEFAULT_ON=y,
1686 This is not compatible with memory_hotplug.memmap_on_memory.
1687 If both parameters are enabled, hugetlb_free_vmemmap takes
1688 precedence over memory_hotplug.memmap_on_memory.
1691 [KNL] Should the hung task detector generate panics.
1694 A value of 1 instructs the kernel to panic when a
1695 hung task is detected. The default value is controlled
1696 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1697 option. The value selected by this boot parameter can
1698 be changed later by the kernel.hung_task_panic sysctl.
1700 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1701 terminal devices. Valid values: 0..8
1702 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1703 If specified, z/VM IUCV HVC accepts connections
1704 from listed z/VM user IDs only.
1706 hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations
1707 which allow the hypervisor to 'idle' the
1708 guest on lock contention.
1711 Do not unregister boot console at start. This is only
1712 useful for debugging when something happens in the window
1713 between unregistering the boot console and initializing
1716 i2c_bus= [HW] Override the default board specific I2C bus speed
1717 or register an additional I2C bus that is not
1718 registered from board initialization code.
1722 i8042.debug [HW] Toggle i8042 debug mode
1723 i8042.unmask_kbd_data
1724 [HW] Enable printing of interrupt data from the KBD port
1725 (disabled by default, and as a pre-condition
1726 requires that i8042.debug=1 be enabled)
1727 i8042.direct [HW] Put keyboard port into non-translated mode
1728 i8042.dumbkbd [HW] Pretend that controller can only read data from
1729 keyboard and cannot control its state
1730 (Don't attempt to blink the leds)
1731 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1732 i8042.nokbd [HW] Don't check/create keyboard port
1733 i8042.noloop [HW] Disable the AUX Loopback command while probing
1735 i8042.nomux [HW] Don't check presence of an active multiplexing
1737 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1739 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1740 i8042.reset [HW] Reset the controller during init, cleanup and
1741 suspend-to-ram transitions, only during s2r
1742 transitions, or never reset
1743 Format: { 1 | Y | y | 0 | N | n }
1744 1, Y, y: always reset controller
1745 0, N, n: don't ever reset controller
1746 Default: only on s2r transitions on x86; most other
1747 architectures force reset to be always executed
1748 i8042.unlock [HW] Unlock (ignore) the keylock
1749 i8042.kbdreset [HW] Reset device connected to KBD port
1751 [HW] Allow deferred probing upon i8042 probe errors
1755 i915.invert_brightness=
1756 [DRM] Invert the sense of the variable that is used to
1757 set the brightness of the panel backlight. Normally a
1758 brightness value of 0 indicates backlight switched off,
1759 and the maximum of the brightness value sets the backlight
1760 to maximum brightness. If this parameter is set to 0
1761 (default) and the machine requires it, or this parameter
1762 is set to 1, a brightness value of 0 sets the backlight
1763 to maximum brightness, and the maximum of the brightness
1764 value switches the backlight off.
1765 -1 -- never invert brightness
1766 0 -- machine default
1767 1 -- force brightness inversion
1770 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1772 ide-core.nodma= [HW] (E)IDE subsystem
1773 Format: =0.0 to prevent dma on hda, =0.1 hdb =1.0 hdc
1774 .vlb_clock .pci_clock .noflush .nohpa .noprobe .nowerr
1775 .cdrom .chs .ignore_cable are additional options
1776 See Documentation/ide/ide.rst.
1778 ide-generic.probe-mask= [HW] (E)IDE subsystem
1780 Probe mask for legacy ISA IDE ports. Depending on
1781 platform up to 6 ports are supported, enabled by
1782 setting corresponding bits in the mask to 1. The
1783 default value is 0x0, which has a special meaning.
1784 On systems that have PCI, it triggers scanning the
1785 PCI bus for the first and the second port, which
1786 are then probed. On systems without PCI the value
1787 of 0x0 enables probing the two first ports as if it
1790 ide-pci-generic.all-generic-ide [HW] (E)IDE subsystem
1791 Claim all unknown PCI IDE storage controllers.
1794 Format: idle=poll, idle=halt, idle=nomwait
1795 Poll forces a polling idle loop that can slightly
1796 improve the performance of waking up a idle CPU, but
1797 will use a lot of power and make the system run hot.
1799 idle=halt: Halt is forced to be used for CPU idle.
1800 In such case C2/C3 won't be used again.
1801 idle=nomwait: Disable mwait for CPU C-states
1805 Allow force disabling of Shared Virtual Memory (SVA)
1806 support for the idxd driver. By default it is set to
1809 idxd.tc_override= [HW]
1811 Allow override of default traffic class configuration
1812 for the device. By default it is set to false (0).
1814 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1815 Format: { strict | legacy | 2008 | relaxed }
1818 Choose which programs will be accepted for execution
1819 based on the IEEE 754 NaN encoding(s) supported by
1820 the FPU and the NaN encoding requested with the value
1821 of an ELF file header flag individually set by each
1822 binary. Hardware implementations are permitted to
1823 support either or both of the legacy and the 2008 NaN
1826 Available settings are as follows:
1827 strict accept binaries that request a NaN encoding
1828 supported by the FPU
1829 legacy only accept legacy-NaN binaries, if supported
1831 2008 only accept 2008-NaN binaries, if supported
1833 relaxed accept any binaries regardless of whether
1834 supported by the FPU
1836 The FPU emulator is always able to support both NaN
1837 encodings, so if no FPU hardware is present or it has
1838 been disabled with 'nofpu', then the settings of
1839 'legacy' and '2008' strap the emulator accordingly,
1840 'relaxed' straps the emulator for both legacy-NaN and
1841 2008-NaN, whereas 'strict' enables legacy-NaN only on
1842 legacy processors and both NaN encodings on MIPS32 or
1845 The setting for ABS.fmt/NEG.fmt instruction execution
1846 mode generally follows that for the NaN encoding,
1847 except where unsupported by hardware.
1849 ignore_loglevel [KNL]
1850 Ignore loglevel setting - this will print /all/
1851 kernel messages to the console. Useful for debugging.
1852 We also add it as printk module parameter, so users
1853 could change it dynamically, usually by
1854 /sys/module/printk/parameters/ignore_loglevel.
1857 Ignore RLIMIT_DATA setting for data mappings,
1858 print warning at first misuse. Can be changed via
1859 /sys/module/kernel/parameters/ignore_rlimit_data.
1861 ihash_entries= [KNL]
1862 Set number of hash buckets for inode cache.
1864 ima_appraise= [IMA] appraise integrity measurements
1865 Format: { "off" | "enforce" | "fix" | "log" }
1868 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
1869 The builtin appraise policy appraises all files
1872 ima_canonical_fmt [IMA]
1873 Use the canonical format for the binary runtime
1874 measurements, instead of host native format.
1877 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1881 The list of supported hash algorithms is defined
1882 in crypto/hash_info.h.
1885 The builtin policies to load during IMA setup.
1886 Format: "tcb | appraise_tcb | secure_boot |
1887 fail_securely | critical_data"
1889 The "tcb" policy measures all programs exec'd, files
1890 mmap'd for exec, and all files opened with the read
1891 mode bit set by either the effective uid (euid=0) or
1894 The "appraise_tcb" policy appraises the integrity of
1895 all files owned by root.
1897 The "secure_boot" policy appraises the integrity
1898 of files (eg. kexec kernel image, kernel modules,
1899 firmware, policy, etc) based on file signatures.
1901 The "fail_securely" policy forces file signature
1902 verification failure also on privileged mounted
1903 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
1906 The "critical_data" policy measures kernel integrity
1909 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
1910 Load a policy which meets the needs of the Trusted
1911 Computing Base. This means IMA will measure all
1912 programs exec'd, files mmap'd for exec, and all files
1913 opened for read by uid=0.
1916 Select one of defined IMA measurements template formats.
1917 Formats: { "ima" | "ima-ng" | "ima-ngv2" | "ima-sig" |
1922 [IMA] Define a custom template format.
1923 Format: { "field1|...|fieldN" }
1925 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
1926 Format: <min_file_size>
1927 Set the minimal file size for using asynchronous hash.
1928 If left unspecified, ahash usage is disabled.
1930 ahash performance varies for different data sizes on
1931 different crypto accelerators. This option can be used
1932 to achieve the best performance for a particular HW.
1934 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
1936 Set hashing buffer size. Default: 4k.
1938 ahash performance varies for different chunk sizes on
1939 different crypto accelerators. This option can be used
1940 to achieve best performance for particular HW.
1944 Run specified binary instead of /sbin/init as init
1947 initcall_debug [KNL] Trace initcalls as they are executed. Useful
1948 for working out where the kernel is dying during
1951 initcall_blacklist= [KNL] Do not execute a comma-separated list of
1952 initcall functions. Useful for debugging built-in
1953 modules and initcalls.
1955 initramfs_async= [KNL]
1958 This parameter controls whether the initramfs
1959 image is unpacked asynchronously, concurrently
1960 with devices being probed and
1961 initialized. This should normally just work,
1962 but as a debugging aid, one can get the
1963 historical behaviour of the initramfs
1964 unpacking being completed before device_ and
1967 initrd= [BOOT] Specify the location of the initial ramdisk
1969 initrdmem= [KNL] Specify a physical address and size from which to
1970 load the initrd. If an initrd is compiled in or
1971 specified in the bootparams, it takes priority over this
1973 Format: ss[KMG],nn[KMG]
1976 init_on_alloc= [MM] Fill newly allocated pages and heap objects with
1979 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
1981 init_on_free= [MM] Fill freed pages and heap objects with zeroes.
1983 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
1985 init_pkru= [X86] Specify the default memory protection keys rights
1986 register contents for all processes. 0x55555554 by
1987 default (disallow access to all but pkey 0). Can
1988 override in debugfs after boot.
1990 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
1993 int_pln_enable [X86] Enable power limit notification interrupt
1995 integrity_audit=[IMA]
1996 Format: { "0" | "1" }
1997 0 -- basic integrity auditing messages. (Default)
1998 1 -- additional integrity auditing messages.
2000 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
2002 Enable intel iommu driver.
2004 Disable intel iommu driver.
2005 igfx_off [Default Off]
2006 By default, gfx is mapped as normal device. If a gfx
2007 device has a dedicated DMAR unit, the DMAR unit is
2008 bypassed by not enabling DMAR with this option. In
2009 this case, gfx device will use physical address for
2011 strict [Default Off]
2012 Deprecated, equivalent to iommu.strict=1.
2013 sp_off [Default Off]
2014 By default, super page will be supported if Intel IOMMU
2015 has the capability. With this option, super page will
2018 Enable the Intel IOMMU scalable mode if the hardware
2019 advertises that it has support for the scalable mode
2022 Disallow use of the Intel IOMMU scalable mode.
2023 tboot_noforce [Default Off]
2024 Do not force the Intel IOMMU enabled under tboot.
2025 By default, tboot will force Intel IOMMU on, which
2026 could harm performance of some high-throughput
2027 devices like 40GBit network cards, even if identity
2029 Note that using this option lowers the security
2030 provided by tboot because it makes the system
2031 vulnerable to DMA attacks.
2033 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
2034 0 disables intel_idle and fall back on acpi_idle.
2035 1 to 9 specify maximum depth of C-state.
2039 Do not enable intel_pstate as the default
2040 scaling driver for the supported processors
2042 Use intel_pstate as a scaling driver, but configure it
2043 to work with generic cpufreq governors (instead of
2044 enabling its internal governor). This mode cannot be
2045 used along with the hardware-managed P-states (HWP)
2048 Enable intel_pstate on systems that prohibit it by default
2049 in favor of acpi-cpufreq. Forcing the intel_pstate driver
2050 instead of acpi-cpufreq may disable platform features, such
2051 as thermal controls and power capping, that rely on ACPI
2052 P-States information being indicated to OSPM and therefore
2053 should be used with caution. This option does not work with
2054 processors that aren't supported by the intel_pstate driver
2055 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
2057 Do not enable hardware P state control (HWP)
2060 Only load intel_pstate on systems which support
2061 hardware P state control (HWP) if available.
2063 Enforce ACPI _PPC performance limits. If the Fixed ACPI
2064 Description Table, specifies preferred power management
2065 profile as "Enterprise Server" or "Performance Server",
2066 then this feature is turned on by default.
2068 Allow per-logical-CPU P-State performance control limits using
2069 cpufreq sysfs interface
2071 intremap= [X86-64, Intel-IOMMU]
2072 on enable Interrupt Remapping (default)
2073 off disable Interrupt Remapping
2074 nosid disable Source ID checking
2076 BIOS x2APIC opt-out request will be ignored
2077 nopost disable Interrupt Posting
2079 iomem= Disable strict checking of access to MMIO memory
2080 strict regions from userspace.
2095 nobypass [PPC/POWERNV]
2096 Disable IOMMU bypass, using IOMMU for PCI devices.
2098 iommu.forcedac= [ARM64, X86] Control IOVA allocation for PCI devices.
2099 Format: { "0" | "1" }
2100 0 - Try to allocate a 32-bit DMA address first, before
2101 falling back to the full range if needed.
2102 1 - Allocate directly from the full usable range,
2103 forcing Dual Address Cycle for PCI cards supporting
2104 greater than 32-bit addressing.
2106 iommu.strict= [ARM64, X86] Configure TLB invalidation behaviour
2107 Format: { "0" | "1" }
2109 Request that DMA unmap operations use deferred
2110 invalidation of hardware TLBs, for increased
2111 throughput at the cost of reduced device isolation.
2112 Will fall back to strict mode if not supported by
2113 the relevant IOMMU driver.
2115 DMA unmap operations invalidate IOMMU hardware TLBs
2117 unset - Use value of CONFIG_IOMMU_DEFAULT_DMA_{LAZY,STRICT}.
2118 Note: on x86, strict mode specified via one of the
2119 legacy driver-specific options takes precedence.
2122 [ARM64, X86] Configure DMA to bypass the IOMMU by default.
2123 Format: { "0" | "1" }
2124 0 - Use IOMMU translation for DMA.
2125 1 - Bypass the IOMMU for DMA.
2126 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
2128 io7= [HW] IO7 for Marvel-based Alpha systems
2129 See comment before marvel_specify_io7 in
2130 arch/alpha/kernel/core_marvel.c.
2132 io_delay= [X86] I/O delay method
2134 Standard port 0x80 based delay
2136 Alternate port 0xed based delay (needed on some systems)
2138 Simple two microseconds delay
2143 See Documentation/admin-guide/nfs/nfsroot.rst.
2145 ipcmni_extend [KNL] Extend the maximum number of unique System V
2146 IPC identifiers from 32,768 to 16,777,216.
2148 irqaffinity= [SMP] Set the default irq affinity mask
2149 The argument is a cpu list, as described above.
2151 irqchip.gicv2_force_probe=
2154 Force the kernel to look for the second 4kB page
2155 of a GICv2 controller even if the memory range
2156 exposed by the device tree is too small.
2158 irqchip.gicv3_nolpi=
2160 Force the kernel to ignore the availability of
2161 LPIs (and by consequence ITSs). Intended for system
2162 that use the kernel as a bootloader, and thus want
2163 to let secondary kernels in charge of setting up
2166 irqchip.gicv3_pseudo_nmi= [ARM64]
2167 Enables support for pseudo-NMIs in the kernel. This
2168 requires the kernel to be built with
2169 CONFIG_ARM64_PSEUDO_NMI.
2172 When an interrupt is not handled search all handlers
2173 for it. Intended to get systems with badly broken
2177 When an interrupt is not handled search all handlers
2178 for it. Also check all handlers each timer
2179 interrupt. Intended to get systems with badly broken
2183 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
2185 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
2186 [Deprecated - use cpusets instead]
2187 Format: [flag-list,]<cpu-list>
2189 Specify one or more CPUs to isolate from disturbances
2190 specified in the flag list (default: domain):
2193 Disable the tick when a single task runs.
2195 A residual 1Hz tick is offloaded to workqueues, which you
2196 need to affine to housekeeping through the global
2197 workqueue's affinity configured via the
2198 /sys/devices/virtual/workqueue/cpumask sysfs file, or
2199 by using the 'domain' flag described below.
2201 NOTE: by default the global workqueue runs on all CPUs,
2202 so to protect individual CPUs the 'cpumask' file has to
2203 be configured manually after bootup.
2206 Isolate from the general SMP balancing and scheduling
2207 algorithms. Note that performing domain isolation this way
2208 is irreversible: it's not possible to bring back a CPU to
2209 the domains once isolated through isolcpus. It's strongly
2210 advised to use cpusets instead to disable scheduler load
2211 balancing through the "cpuset.sched_load_balance" file.
2212 It offers a much more flexible interface where CPUs can
2213 move in and out of an isolated set anytime.
2215 You can move a process onto or off an "isolated" CPU via
2216 the CPU affinity syscalls or cpuset.
2217 <cpu number> begins at 0 and the maximum value is
2218 "number of CPUs in system - 1".
2222 Isolate from being targeted by managed interrupts
2223 which have an interrupt mask containing isolated
2224 CPUs. The affinity of managed interrupts is
2225 handled by the kernel and cannot be changed via
2226 the /proc/irq/* interfaces.
2228 This isolation is best effort and only effective
2229 if the automatically assigned interrupt mask of a
2230 device queue contains isolated and housekeeping
2231 CPUs. If housekeeping CPUs are online then such
2232 interrupts are directed to the housekeeping CPU
2233 so that IO submitted on the housekeeping CPU
2234 cannot disturb the isolated CPU.
2236 If a queue's affinity mask contains only isolated
2237 CPUs then this parameter has no effect on the
2238 interrupt routing decision, though interrupts are
2239 only delivered when tasks running on those
2240 isolated CPUs submit IO. IO submitted on
2241 housekeeping CPUs has no influence on those
2244 The format of <cpu-list> is described above.
2248 ivrs_ioapic [HW,X86-64]
2249 Provide an override to the IOAPIC-ID<->DEVICE-ID
2250 mapping provided in the IVRS ACPI table. For
2251 example, to map IOAPIC-ID decimal 10 to
2252 PCI device 00:14.0 write the parameter as:
2253 ivrs_ioapic[10]=00:14.0
2255 ivrs_hpet [HW,X86-64]
2256 Provide an override to the HPET-ID<->DEVICE-ID
2257 mapping provided in the IVRS ACPI table. For
2258 example, to map HPET-ID decimal 0 to
2259 PCI device 00:14.0 write the parameter as:
2260 ivrs_hpet[0]=00:14.0
2262 ivrs_acpihid [HW,X86-64]
2263 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2264 mapping provided in the IVRS ACPI table. For
2265 example, to map UART-HID:UID AMD0020:0 to
2266 PCI device 00:14.5 write the parameter as:
2267 ivrs_acpihid[00:14.5]=AMD0020:0
2269 js= [HW,JOY] Analog joystick
2270 See Documentation/input/joydev/joystick.rst.
2273 When CONFIG_RANDOMIZE_BASE is set, this disables
2274 kernel and module base offset ASLR (Address Space
2275 Layout Randomization).
2278 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2279 report on every invalid memory access. Without this
2280 parameter KASAN will print report only for the first
2285 kernelcore= [KNL,X86,IA-64,PPC]
2286 Format: nn[KMGTPE] | nn% | "mirror"
2287 This parameter specifies the amount of memory usable by
2288 the kernel for non-movable allocations. The requested
2289 amount is spread evenly throughout all nodes in the
2290 system as ZONE_NORMAL. The remaining memory is used for
2291 movable memory in its own zone, ZONE_MOVABLE. In the
2292 event, a node is too small to have both ZONE_NORMAL and
2293 ZONE_MOVABLE, kernelcore memory will take priority and
2294 other nodes will have a larger ZONE_MOVABLE.
2296 ZONE_MOVABLE is used for the allocation of pages that
2297 may be reclaimed or moved by the page migration
2298 subsystem. Note that allocations like PTEs-from-HighMem
2299 still use the HighMem zone if it exists, and the Normal
2300 zone if it does not.
2302 It is possible to specify the exact amount of memory in
2303 the form of "nn[KMGTPE]", a percentage of total system
2304 memory in the form of "nn%", or "mirror". If "mirror"
2305 option is specified, mirrored (reliable) memory is used
2306 for non-movable allocations and remaining memory is used
2307 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2308 are exclusive, so you cannot specify multiple forms.
2310 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
2311 Format: <Controller#>[,poll interval]
2312 The controller # is the number of the ehci usb debug
2313 port as it is probed via PCI. The poll interval is
2314 optional and is the number seconds in between
2315 each poll cycle to the debug port in case you need
2316 the functionality for interrupting the kernel with
2317 gdb or control-c on the dbgp connection. When
2318 not using this parameter you use sysrq-g to break into
2319 the kernel debugger.
2321 kgdboc= [KGDB,HW] kgdb over consoles.
2322 Requires a tty driver that supports console polling,
2323 or a supported polling keyboard driver (non-usb).
2324 Serial only format: <serial_device>[,baud]
2325 keyboard only format: kbd
2326 keyboard and serial format: kbd,<serial_device>[,baud]
2327 Optional Kernel mode setting:
2328 kms, kbd format: kms,kbd
2329 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2331 kgdboc_earlycon= [KGDB,HW]
2332 If the boot console provides the ability to read
2333 characters and can work in polling mode, you can use
2334 this parameter to tell kgdb to use it as a backend
2335 until the normal console is registered. Intended to
2336 be used together with the kgdboc parameter which
2337 specifies the normal console to transition to.
2339 The name of the early console should be specified
2340 as the value of this parameter. Note that the name of
2341 the early console might be different than the tty
2342 name passed to kgdboc. It's OK to leave the value
2343 blank and the first boot console that implements
2344 read() will be picked.
2346 kgdbwait [KGDB] Stop kernel execution and enter the
2347 kernel debugger at the earliest opportunity.
2349 kmac= [MIPS] Korina ethernet MAC address.
2350 Configure the RouterBoard 532 series on-chip
2351 Ethernet adapter MAC address.
2353 kmemleak= [KNL] Boot-time kmemleak enable/disable
2354 Valid arguments: on, off
2356 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2359 kprobe_event=[probe-list]
2360 [FTRACE] Add kprobe events and enable at boot time.
2361 The probe-list is a semicolon delimited list of probe
2362 definitions. Each definition is same as kprobe_events
2363 interface, but the parameters are comma delimited.
2364 For example, to add a kprobe event on vfs_read with
2365 arg1 and arg2, add to the command line;
2367 kprobe_event=p,vfs_read,$arg1,$arg2
2369 See also Documentation/trace/kprobetrace.rst "Kernel
2370 Boot Parameter" section.
2372 kpti= [ARM64] Control page table isolation of user
2373 and kernel address spaces.
2374 Default: enabled on cores which need mitigation.
2378 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2379 Default is 0 (don't ignore, but inject #GP)
2381 kvm.eager_page_split=
2382 [KVM,X86] Controls whether or not KVM will try to
2383 proactively split all huge pages during dirty logging.
2384 Eager page splitting reduces interruptions to vCPU
2385 execution by eliminating the write-protection faults
2386 and MMU lock contention that would otherwise be
2387 required to split huge pages lazily.
2389 VM workloads that rarely perform writes or that write
2390 only to a small region of VM memory may benefit from
2391 disabling eager page splitting to allow huge pages to
2392 still be used for reads.
2394 The behavior of eager page splitting depends on whether
2395 KVM_DIRTY_LOG_INITIALLY_SET is enabled or disabled. If
2396 disabled, all huge pages in a memslot will be eagerly
2397 split when dirty logging is enabled on that memslot. If
2398 enabled, eager page splitting will be performed during
2399 the KVM_CLEAR_DIRTY ioctl, and only for the pages being
2402 Eager page splitting currently only supports splitting
2403 huge pages mapped by the TDP MMU.
2407 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2408 Default is false (don't support).
2411 [KVM] Controls the software workaround for the
2412 X86_BUG_ITLB_MULTIHIT bug.
2413 force : Always deploy workaround.
2414 off : Never deploy workaround.
2415 auto : Deploy workaround based on the presence of
2416 X86_BUG_ITLB_MULTIHIT.
2420 If the software workaround is enabled for the host,
2421 guests do need not to enable it for nested guests.
2423 kvm.nx_huge_pages_recovery_ratio=
2424 [KVM] Controls how many 4KiB pages are periodically zapped
2425 back to huge pages. 0 disables the recovery, otherwise if
2426 the value is N KVM will zap 1/Nth of the 4KiB pages every
2427 period (see below). The default is 60.
2429 kvm.nx_huge_pages_recovery_period_ms=
2430 [KVM] Controls the time period at which KVM zaps 4KiB pages
2431 back to huge pages. If the value is a non-zero N, KVM will
2432 zap a portion (see ratio above) of the pages every N msecs.
2433 If the value is 0 (the default), KVM will pick a period based
2434 on the ratio, such that a page is zapped after 1 hour on average.
2436 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM.
2437 Default is 1 (enabled)
2439 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU)
2441 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode.
2444 [KVM,ARM] Select one of KVM/arm64's modes of operation.
2446 none: Forcefully disable KVM.
2448 nvhe: Standard nVHE-based mode, without support for
2451 protected: nVHE-based mode with support for guests whose
2452 state is kept private from the host.
2453 Not valid if the kernel is running in EL2.
2455 Defaults to VHE/nVHE based on hardware support. Setting
2456 mode to "protected" will disable kexec and hibernation
2459 kvm-arm.vgic_v3_group0_trap=
2460 [KVM,ARM] Trap guest accesses to GICv3 group-0
2463 kvm-arm.vgic_v3_group1_trap=
2464 [KVM,ARM] Trap guest accesses to GICv3 group-1
2467 kvm-arm.vgic_v3_common_trap=
2468 [KVM,ARM] Trap guest accesses to GICv3 common
2471 kvm-arm.vgic_v4_enable=
2472 [KVM,ARM] Allow use of GICv4 for direct injection of
2475 kvm_cma_resv_ratio=n [PPC]
2476 Reserves given percentage from system memory area for
2477 contiguous memory allocation for KVM hash pagetable
2479 By default it reserves 5% of total system memory.
2483 kvm-intel.ept= [KVM,Intel] Disable extended page tables
2484 (virtualized MMU) support on capable Intel chips.
2485 Default is 1 (enabled)
2487 kvm-intel.emulate_invalid_guest_state=
2488 [KVM,Intel] Disable emulation of invalid guest state.
2489 Ignored if kvm-intel.enable_unrestricted_guest=1, as
2490 guest state is never invalid for unrestricted guests.
2491 This param doesn't apply to nested guests (L2), as KVM
2492 never emulates invalid L2 guest state.
2493 Default is 1 (enabled)
2495 kvm-intel.flexpriority=
2496 [KVM,Intel] Disable FlexPriority feature (TPR shadow).
2497 Default is 1 (enabled)
2500 [KVM,Intel] Enable VMX nesting (nVMX).
2501 Default is 0 (disabled)
2503 kvm-intel.unrestricted_guest=
2504 [KVM,Intel] Disable unrestricted guest feature
2505 (virtualized real and unpaged mode) on capable
2506 Intel chips. Default is 1 (enabled)
2508 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2511 Valid arguments: never, cond, always
2513 always: L1D cache flush on every VMENTER.
2514 cond: Flush L1D on VMENTER only when the code between
2515 VMEXIT and VMENTER can leak host memory.
2516 never: Disables the mitigation
2518 Default is cond (do L1 cache flush in specific instances)
2520 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification
2521 feature (tagged TLBs) on capable Intel chips.
2522 Default is 1 (enabled)
2524 l1d_flush= [X86,INTEL]
2525 Control mitigation for L1D based snooping vulnerability.
2527 Certain CPUs are vulnerable to an exploit against CPU
2528 internal buffers which can forward information to a
2529 disclosure gadget under certain conditions.
2531 In vulnerable processors, the speculatively
2532 forwarded data can be used in a cache side channel
2533 attack, to access data to which the attacker does
2534 not have direct access.
2536 This parameter controls the mitigation. The
2539 on - enable the interface for the mitigation
2541 l1tf= [X86] Control mitigation of the L1TF vulnerability on
2544 The kernel PTE inversion protection is unconditionally
2545 enabled and cannot be disabled.
2548 Provides all available mitigations for the
2549 L1TF vulnerability. Disables SMT and
2550 enables all mitigations in the
2551 hypervisors, i.e. unconditional L1D flush.
2553 SMT control and L1D flush control via the
2554 sysfs interface is still possible after
2555 boot. Hypervisors will issue a warning
2556 when the first VM is started in a
2557 potentially insecure configuration,
2558 i.e. SMT enabled or L1D flush disabled.
2561 Same as 'full', but disables SMT and L1D
2562 flush runtime control. Implies the
2563 'nosmt=force' command line option.
2564 (i.e. sysfs control of SMT is disabled.)
2567 Leaves SMT enabled and enables the default
2568 hypervisor mitigation, i.e. conditional
2571 SMT control and L1D flush control via the
2572 sysfs interface is still possible after
2573 boot. Hypervisors will issue a warning
2574 when the first VM is started in a
2575 potentially insecure configuration,
2576 i.e. SMT enabled or L1D flush disabled.
2580 Disables SMT and enables the default
2581 hypervisor mitigation.
2583 SMT control and L1D flush control via the
2584 sysfs interface is still possible after
2585 boot. Hypervisors will issue a warning
2586 when the first VM is started in a
2587 potentially insecure configuration,
2588 i.e. SMT enabled or L1D flush disabled.
2591 Same as 'flush', but hypervisors will not
2592 warn when a VM is started in a potentially
2593 insecure configuration.
2596 Disables hypervisor mitigations and doesn't
2598 It also drops the swap size and available
2599 RAM limit restriction on both hypervisor and
2604 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2610 lapic [X86-32,APIC] Enable the local APIC even if BIOS
2613 lapic= [X86,APIC] Do not use TSC deadline
2614 value for LAPIC timer one-shot implementation. Default
2615 back to the programmable timer unit in the LAPIC.
2616 Format: notscdeadline
2618 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
2621 libata.dma= [LIBATA] DMA control
2622 libata.dma=0 Disable all PATA and SATA DMA
2623 libata.dma=1 PATA and SATA Disk DMA only
2624 libata.dma=2 ATAPI (CDROM) DMA only
2625 libata.dma=4 Compact Flash DMA only
2626 Combinations also work, so libata.dma=3 enables DMA
2627 for disks and CDROMs, but not CFs.
2629 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2630 libata.ignore_hpa=0 keep BIOS limits (default)
2631 libata.ignore_hpa=1 ignore limits, using full disk
2633 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2637 libata.force= [LIBATA] Force configurations. The format is a comma-
2638 separated list of "[ID:]VAL" where ID is PORT[.DEVICE].
2639 PORT and DEVICE are decimal numbers matching port, link
2640 or device. Basically, it matches the ATA ID string
2641 printed on console by libata. If the whole ID part is
2642 omitted, the last PORT and DEVICE values are used. If
2643 ID hasn't been specified yet, the configuration applies
2644 to all ports, links and devices.
2646 If only DEVICE is omitted, the parameter applies to
2647 the port and all links and devices behind it. DEVICE
2648 number of 0 either selects the first device or the
2649 first fan-out link behind PMP device. It does not
2650 select the host link. DEVICE number of 15 selects the
2651 host link and device attached to it.
2653 The VAL specifies the configuration to force. As long
2654 as there is no ambiguity, shortcut notation is allowed.
2655 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2656 The following configurations can be forced.
2658 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2659 Any ID with matching PORT is used.
2661 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2663 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2664 udma[/][16,25,33,44,66,100,133] notation is also
2667 * nohrst, nosrst, norst: suppress hard, soft and both
2670 * rstonce: only attempt one reset during hot-unplug
2673 * [no]dbdelay: Enable or disable the extra 200ms delay
2674 before debouncing a link PHY and device presence
2677 * [no]ncq: Turn on or off NCQ.
2679 * [no]ncqtrim: Enable or disable queued DSM TRIM.
2681 * [no]ncqati: Enable or disable NCQ trim on ATI chipset.
2683 * [no]trim: Enable or disable (unqueued) TRIM.
2685 * trim_zero: Indicate that TRIM command zeroes data.
2687 * max_trim_128m: Set 128M maximum trim size limit.
2689 * [no]dma: Turn on or off DMA transfers.
2691 * atapi_dmadir: Enable ATAPI DMADIR bridge support.
2693 * atapi_mod16_dma: Enable the use of ATAPI DMA for
2694 commands that are not a multiple of 16 bytes.
2696 * [no]dmalog: Enable or disable the use of the
2697 READ LOG DMA EXT command to access logs.
2699 * [no]iddevlog: Enable or disable access to the
2700 identify device data log.
2702 * [no]logdir: Enable or disable access to the general
2703 purpose log directory.
2705 * max_sec_128: Set transfer size limit to 128 sectors.
2707 * max_sec_1024: Set or clear transfer size limit to
2710 * max_sec_lba48: Set or clear transfer size limit to
2713 * [no]lpm: Enable or disable link power management.
2715 * [no]setxfer: Indicate if transfer speed mode setting
2718 * dump_id: Dump IDENTIFY data.
2720 * disable: Disable this device.
2722 If there are multiple matching configurations changing
2723 the same attribute, the last one is used.
2725 memblock=debug [KNL] Enable memblock debug messages.
2727 load_ramdisk= [RAM] [Deprecated]
2729 lockd.nlm_grace_period=P [NFS] Assign grace period.
2732 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2735 lockd.nlm_timeout=T [NFS] Assign timeout value.
2738 lockd.nlm_udpport=M [NFS] Assign UDP port.
2741 lockdown= [SECURITY]
2742 { integrity | confidentiality }
2743 Enable the kernel lockdown feature. If set to
2744 integrity, kernel features that allow userland to
2745 modify the running kernel are disabled. If set to
2746 confidentiality, kernel features that allow userland
2747 to extract confidential information from the kernel
2750 locktorture.nreaders_stress= [KNL]
2751 Set the number of locking read-acquisition kthreads.
2752 Defaults to being automatically set based on the
2753 number of online CPUs.
2755 locktorture.nwriters_stress= [KNL]
2756 Set the number of locking write-acquisition kthreads.
2758 locktorture.onoff_holdoff= [KNL]
2759 Set time (s) after boot for CPU-hotplug testing.
2761 locktorture.onoff_interval= [KNL]
2762 Set time (s) between CPU-hotplug operations, or
2763 zero to disable CPU-hotplug testing.
2765 locktorture.shuffle_interval= [KNL]
2766 Set task-shuffle interval (jiffies). Shuffling
2767 tasks allows some CPUs to go into dyntick-idle
2768 mode during the locktorture test.
2770 locktorture.shutdown_secs= [KNL]
2771 Set time (s) after boot system shutdown. This
2772 is useful for hands-off automated testing.
2774 locktorture.stat_interval= [KNL]
2775 Time (s) between statistics printk()s.
2777 locktorture.stutter= [KNL]
2778 Time (s) to stutter testing, for example,
2779 specifying five seconds causes the test to run for
2780 five seconds, wait for five seconds, and so on.
2781 This tests the locking primitive's ability to
2782 transition abruptly to and from idle.
2784 locktorture.torture_type= [KNL]
2785 Specify the locking implementation to test.
2787 locktorture.verbose= [KNL]
2788 Enable additional printk() statements.
2790 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
2793 loglevel= All Kernel Messages with a loglevel smaller than the
2794 console loglevel will be printed to the console. It can
2795 also be changed with klogd or other programs. The
2796 loglevels are defined as follows:
2798 0 (KERN_EMERG) system is unusable
2799 1 (KERN_ALERT) action must be taken immediately
2800 2 (KERN_CRIT) critical conditions
2801 3 (KERN_ERR) error conditions
2802 4 (KERN_WARNING) warning conditions
2803 5 (KERN_NOTICE) normal but significant condition
2804 6 (KERN_INFO) informational
2805 7 (KERN_DEBUG) debug-level messages
2807 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
2808 in bytes. n must be a power of two and greater
2809 than the minimal size. The minimal size is defined
2810 by LOG_BUF_SHIFT kernel config parameter. There is
2811 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2812 that allows to increase the default size depending on
2813 the number of CPUs. See init/Kconfig for more details.
2815 logo.nologo [FB] Disables display of the built-in Linux logo.
2816 This may be used to provide more screen space for
2817 kernel log messages and is useful when debugging
2818 kernel boot problems.
2820 lp=0 [LP] Specify parallel ports to use, e.g,
2821 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2822 lp=reset first parallel port). 'lp=0' disables the
2823 lp=auto printer driver. 'lp=reset' (which can be
2824 specified in addition to the ports) causes
2825 attached printers to be reset. Using
2826 lp=port1,port2,... specifies the parallel ports
2827 to associate lp devices with, starting with
2828 lp0. A port specification may be 'none' to skip
2829 that lp device, or a parport name such as
2830 'parport0'. Specifying 'lp=auto' instead of a
2831 port specification list means that device IDs
2832 from each port should be examined, to see if
2833 an IEEE 1284-compliant printer is attached; if
2834 so, the driver will manage that printer.
2835 See also header of drivers/char/lp.c.
2838 Sets loops_per_jiffy to given constant, thus avoiding
2839 time-consuming boot-time autodetection (up to 250 ms per
2840 CPU). 0 enables autodetection (default). To determine
2841 the correct value for your kernel, boot with normal
2842 autodetection and see what value is printed. Note that
2843 on SMP systems the preset will be applied to all CPUs,
2844 which is likely to cause problems if your CPUs need
2845 significantly divergent settings. An incorrect value
2846 will cause delays in the kernel to be wrong, leading to
2847 unpredictable I/O errors and other breakage. Although
2848 unlikely, in the extreme case this might damage your
2852 Format: <io>,<irq>,<dma>
2854 lsm.debug [SECURITY] Enable LSM initialization debugging output.
2857 [SECURITY] Choose order of LSM initialization. This
2858 overrides CONFIG_LSM, and the "security=" parameter.
2860 machvec= [IA-64] Force the use of a particular machine-vector
2861 (machvec) in a generic kernel.
2862 Example: machvec=hpzx1
2864 machtype= [Loongson] Share the same kernel image file between
2865 different yeeloong laptops.
2866 Example: machtype=lemote-yeeloong-2f-7inch
2868 max_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory greater
2869 than or equal to this physical address is ignored.
2871 maxcpus= [SMP] Maximum number of processors that an SMP kernel
2872 will bring up during bootup. maxcpus=n : n >= 0 limits
2873 the kernel to bring up 'n' processors. Surely after
2874 bootup you can bring up the other plugged cpu by executing
2875 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
2876 only takes effect during system bootup.
2877 While n=0 is a special case, it is equivalent to "nosmp",
2878 which also disables the IO APIC.
2880 max_loop= [LOOP] The number of loop block devices that get
2881 (loop.max_loop) unconditionally pre-created at init time. The default
2882 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
2883 of statically allocating a predefined number, loop
2884 devices can be requested on-demand with the
2885 /dev/loop-control interface.
2887 mce [X86-32] Machine Check Exception
2889 mce=option [X86-64] See Documentation/x86/x86_64/boot-options.rst
2891 md= [HW] RAID subsystems devices and level
2892 See Documentation/admin-guide/md.rst.
2895 Format: <first>,<last>
2896 Specifies range of consoles to be captured by the MDA.
2899 Control mitigation for the Micro-architectural Data
2900 Sampling (MDS) vulnerability.
2902 Certain CPUs are vulnerable to an exploit against CPU
2903 internal buffers which can forward information to a
2904 disclosure gadget under certain conditions.
2906 In vulnerable processors, the speculatively
2907 forwarded data can be used in a cache side channel
2908 attack, to access data to which the attacker does
2909 not have direct access.
2911 This parameter controls the MDS mitigation. The
2914 full - Enable MDS mitigation on vulnerable CPUs
2915 full,nosmt - Enable MDS mitigation and disable
2916 SMT on vulnerable CPUs
2917 off - Unconditionally disable MDS mitigation
2919 On TAA-affected machines, mds=off can be prevented by
2920 an active TAA mitigation as both vulnerabilities are
2921 mitigated with the same mechanism so in order to disable
2922 this mitigation, you need to specify tsx_async_abort=off
2925 Not specifying this option is equivalent to
2928 For details see: Documentation/admin-guide/hw-vuln/mds.rst
2930 mem=nn[KMG] [HEXAGON] Set the memory size.
2931 Must be specified, otherwise memory size will be 0.
2933 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
2934 Amount of memory to be used in cases as follows:
2937 2 when the kernel is not able to see the whole system memory;
2938 3 memory that lies after 'mem=' boundary is excluded from
2939 the hypervisor, then assigned to KVM guests.
2940 4 to limit the memory available for kdump kernel.
2942 [ARC,MICROBLAZE] - the limit applies only to low memory,
2943 high memory is not affected.
2945 [ARM64] - only limits memory covered by the linear
2946 mapping. The NOMAP regions are not affected.
2948 [X86] Work as limiting max address. Use together
2949 with memmap= to avoid physical address space collisions.
2950 Without memmap= PCI devices could be placed at addresses
2951 belonging to unused RAM.
2953 Note that this only takes effects during boot time since
2954 in above case 3, memory may need be hot added after boot
2955 if system memory of hypervisor is not sufficient.
2958 [ARM,MIPS] - override the memory layout reported by
2960 Define a memory region of size nn[KMG] starting at
2962 Multiple different regions can be specified with
2963 multiple mem= parameters on the command line.
2965 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
2969 [KNL,SH] Allow user to override the default size for
2970 per-device physically contiguous DMA buffers.
2972 memhp_default_state=online/offline
2973 [KNL] Set the initial state for the memory hotplug
2974 onlining policy. If not specified, the default value is
2975 set according to the
2976 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
2978 See Documentation/admin-guide/mm/memory-hotplug.rst.
2980 memmap=exactmap [KNL,X86] Enable setting of an exact
2981 E820 memory map, as specified by the user.
2982 Such memmap=exactmap lines can be constructed based on
2983 BIOS output or other requirements. See the memmap=nn@ss
2986 memmap=nn[KMG]@ss[KMG]
2987 [KNL, X86, MIPS, XTENSA] Force usage of a specific region of memory.
2988 Region of memory to be used is from ss to ss+nn.
2989 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
2990 which limits max address to nn[KMG].
2991 Multiple different regions can be specified,
2994 memmap=100M@2G,100M#3G,1G!1024G
2996 memmap=nn[KMG]#ss[KMG]
2997 [KNL,ACPI] Mark specific memory as ACPI data.
2998 Region of memory to be marked is from ss to ss+nn.
3000 memmap=nn[KMG]$ss[KMG]
3001 [KNL,ACPI] Mark specific memory as reserved.
3002 Region of memory to be reserved is from ss to ss+nn.
3003 Example: Exclude memory from 0x18690000-0x1869ffff
3004 memmap=64K$0x18690000
3006 memmap=0x10000$0x18690000
3007 Some bootloaders may need an escape character before '$',
3008 like Grub2, otherwise '$' and the following number
3011 memmap=nn[KMG]!ss[KMG]
3012 [KNL,X86] Mark specific memory as protected.
3013 Region of memory to be used, from ss to ss+nn.
3014 The memory region may be marked as e820 type 12 (0xc)
3015 and is NVDIMM or ADR memory.
3017 memmap=<size>%<offset>-<oldtype>+<newtype>
3018 [KNL,ACPI] Convert memory within the specified region
3019 from <oldtype> to <newtype>. If "-<oldtype>" is left
3020 out, the whole region will be marked as <newtype>,
3021 even if previously unavailable. If "+<newtype>" is left
3022 out, matching memory will be removed. Types are
3023 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
3024 3 = ACPI, 12 = PRAM.
3026 memory_corruption_check=0/1 [X86]
3027 Some BIOSes seem to corrupt the first 64k of
3028 memory when doing things like suspend/resume.
3029 Setting this option will scan the memory
3030 looking for corruption. Enabling this will
3031 both detect corruption and prevent the kernel
3032 from using the memory being corrupted.
3033 However, its intended as a diagnostic tool; if
3034 repeatable BIOS-originated corruption always
3035 affects the same memory, you can use memmap=
3036 to prevent the kernel from using that memory.
3038 memory_corruption_check_size=size [X86]
3039 By default it checks for corruption in the low
3040 64k, making this memory unavailable for normal
3041 use. Use this parameter to scan for
3042 corruption in more or less memory.
3044 memory_corruption_check_period=seconds [X86]
3045 By default it checks for corruption every 60
3046 seconds. Use this parameter to check at some
3047 other rate. 0 disables periodic checking.
3049 memory_hotplug.memmap_on_memory
3050 [KNL,X86,ARM] Boolean flag to enable this feature.
3051 Format: {on | off (default)}
3052 When enabled, runtime hotplugged memory will
3053 allocate its internal metadata (struct pages)
3054 from the hotadded memory which will allow to
3055 hotadd a lot of memory without requiring
3056 additional memory to do so.
3057 This feature is disabled by default because it
3058 has some implication on large (e.g. GB)
3059 allocations in some configurations (e.g. small
3061 The state of the flag can be read in
3062 /sys/module/memory_hotplug/parameters/memmap_on_memory.
3063 Note that even when enabled, there are a few cases where
3064 the feature is not effective.
3066 This is not compatible with hugetlb_free_vmemmap. If
3067 both parameters are enabled, hugetlb_free_vmemmap takes
3068 precedence over memory_hotplug.memmap_on_memory.
3070 memtest= [KNL,X86,ARM,M68K,PPC,RISCV] Enable memtest
3072 default : 0 <disable>
3073 Specifies the number of memtest passes to be
3074 performed. Each pass selects another test
3075 pattern from a given set of patterns. Memtest
3076 fills the memory with this pattern, validates
3077 memory contents and reserves bad memory
3078 regions that are detected.
3080 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
3081 Valid arguments: on, off
3082 Default (depends on kernel configuration option):
3083 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
3084 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
3085 mem_encrypt=on: Activate SME
3086 mem_encrypt=off: Do not activate SME
3088 Refer to Documentation/virt/kvm/amd-memory-encryption.rst
3089 for details on when memory encryption can be activated.
3091 mem_sleep_default= [SUSPEND] Default system suspend mode:
3092 s2idle - Suspend-To-Idle
3093 shallow - Power-On Suspend or equivalent (if supported)
3094 deep - Suspend-To-RAM or equivalent (if supported)
3095 See Documentation/admin-guide/pm/sleep-states.rst.
3097 meye.*= [HW] Set MotionEye Camera parameters
3098 See Documentation/admin-guide/media/meye.rst.
3100 mfgpt_irq= [IA-32] Specify the IRQ to use for the
3101 Multi-Function General Purpose Timers on AMD Geode
3104 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
3105 the BIOS has incorrectly applied a workaround. TinyBIOS
3106 version 0.98 is known to be affected, 0.99 fixes the
3107 problem by letting the user disable the workaround.
3111 min_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory below this
3112 physical address is ignored.
3114 mini2440= [ARM,HW,KNL]
3115 Format:[0..2][b][c][t]
3117 MINI2440 configuration specification:
3118 0 - The attached screen is the 3.5" TFT
3119 1 - The attached screen is the 7" TFT
3120 2 - The VGA Shield is attached (1024x768)
3121 Leaving out the screen size parameter will not load
3122 the TFT driver, and the framebuffer will be left
3124 b - Enable backlight. The TFT backlight pin will be
3125 linked to the kernel VESA blanking code and a GPIO
3126 LED. This parameter is not necessary when using the
3128 c - Enable the s3c camera interface.
3129 t - Reserved for enabling touchscreen support. The
3130 touchscreen support is not enabled in the mainstream
3131 kernel as of 2.6.30, a preliminary port can be found
3132 in the "bleeding edge" mini2440 support kernel at
3133 https://repo.or.cz/w/linux-2.6/mini2440.git
3136 [X86,PPC,S390,ARM64] Control optional mitigations for
3137 CPU vulnerabilities. This is a set of curated,
3138 arch-independent options, each of which is an
3139 aggregation of existing arch-specific options.
3142 Disable all optional CPU mitigations. This
3143 improves system performance, but it may also
3144 expose users to several CPU vulnerabilities.
3145 Equivalent to: nopti [X86,PPC]
3147 nospectre_v1 [X86,PPC]
3149 nospectre_v2 [X86,PPC,S390,ARM64]
3150 spectre_v2_user=off [X86]
3151 spec_store_bypass_disable=off [X86,PPC]
3152 ssbd=force-off [ARM64]
3155 tsx_async_abort=off [X86]
3156 kvm.nx_huge_pages=off [X86]
3157 srbds=off [X86,INTEL]
3158 no_entry_flush [PPC]
3159 no_uaccess_flush [PPC]
3162 This does not have any effect on
3163 kvm.nx_huge_pages when
3164 kvm.nx_huge_pages=force.
3167 Mitigate all CPU vulnerabilities, but leave SMT
3168 enabled, even if it's vulnerable. This is for
3169 users who don't want to be surprised by SMT
3170 getting disabled across kernel upgrades, or who
3171 have other ways of avoiding SMT-based attacks.
3172 Equivalent to: (default behavior)
3175 Mitigate all CPU vulnerabilities, disabling SMT
3176 if needed. This is for users who always want to
3177 be fully mitigated, even if it means losing SMT.
3178 Equivalent to: l1tf=flush,nosmt [X86]
3179 mds=full,nosmt [X86]
3180 tsx_async_abort=full,nosmt [X86]
3183 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
3184 parameter allows control of the logging verbosity for
3185 the additional memory initialisation checks. A value
3186 of 0 disables mminit logging and a level of 4 will
3187 log everything. Information is printed at KERN_DEBUG
3188 so loglevel=8 may also need to be specified.
3191 [KNL] When CONFIG_MODULE_SIG is set, this means that
3192 modules without (valid) signatures will fail to load.
3193 Note that if CONFIG_MODULE_SIG_FORCE is set, that
3194 is always true, so this option does nothing.
3196 module_blacklist= [KNL] Do not load a comma-separated list of
3197 modules. Useful for debugging problem modules.
3200 [MOUSE] Maximum time between finger touching and
3201 leaving touchpad surface for touch to be considered
3202 a tap and be reported as a left button click (for
3203 touchpads working in absolute mode only).
3205 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
3206 reporting absolute coordinates, such as tablets
3207 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
3208 reporting absolute coordinates, such as tablets
3210 movablecore= [KNL,X86,IA-64,PPC]
3211 Format: nn[KMGTPE] | nn%
3212 This parameter is the complement to kernelcore=, it
3213 specifies the amount of memory used for migratable
3214 allocations. If both kernelcore and movablecore is
3215 specified, then kernelcore will be at *least* the
3216 specified value but may be more. If movablecore on its
3217 own is specified, the administrator must be careful
3218 that the amount of memory usable for all allocations
3221 movable_node [KNL] Boot-time switch to make hotplugable memory
3222 NUMA nodes to be movable. This means that the memory
3223 of such nodes will be usable only for movable
3224 allocations which rules out almost all kernel
3225 allocations. Use with caution!
3227 MTD_Partition= [MTD]
3228 Format: <name>,<region-number>,<size>,<offset>
3230 MTD_Region= [MTD] Format:
3231 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
3234 See drivers/mtd/parsers/cmdlinepart.c
3236 multitce=off [PPC] This parameter disables the use of the pSeries
3237 firmware feature for updating multiple TCE entries
3240 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
3242 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
3244 boundary - index of last SLC block on Flex-OneNAND.
3245 The remaining blocks are configured as MLC blocks.
3246 lock - Configure if Flex-OneNAND boundary should be locked.
3247 Once locked, the boundary cannot be changed.
3248 1 indicates lock status, 0 indicates unlock status.
3251 ARM/S3C2412 JIVE boot control
3253 See arch/arm/mach-s3c/mach-jive.c
3255 mtouchusb.raw_coordinates=
3256 [HW] Make the MicroTouch USB driver use raw coordinates
3257 ('y', default) or cooked coordinates ('n')
3259 mtrr_chunk_size=nn[KMG] [X86]
3260 used for mtrr cleanup. It is largest continuous chunk
3261 that could hold holes aka. UC entries.
3263 mtrr_gran_size=nn[KMG] [X86]
3264 Used for mtrr cleanup. It is granularity of mtrr block.
3266 Large value could prevent small alignment from
3269 mtrr_spare_reg_nr=n [X86]
3271 Range: 0,7 : spare reg number
3273 Used for mtrr cleanup. It is spare mtrr entries number.
3274 Set to 2 or more if your graphical card needs more.
3276 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
3278 netdev= [NET] Network devices parameters
3279 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
3280 Note that mem_start is often overloaded to mean
3281 something different and driver-specific.
3282 This usage is only documented in each driver source
3286 [NETFILTER] Enable connection tracking flow accounting
3287 0 to disable accounting
3288 1 to enable accounting
3291 nfsaddrs= [NFS] Deprecated. Use ip= instead.
3292 See Documentation/admin-guide/nfs/nfsroot.rst.
3294 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
3295 See Documentation/admin-guide/nfs/nfsroot.rst.
3297 nfsrootdebug [NFS] enable nfsroot debugging messages.
3298 See Documentation/admin-guide/nfs/nfsroot.rst.
3300 nfs.callback_nr_threads=
3301 [NFSv4] set the total number of threads that the
3302 NFS client will assign to service NFSv4 callback
3305 nfs.callback_tcpport=
3306 [NFS] set the TCP port on which the NFSv4 callback
3307 channel should listen.
3310 [NFS] sets the pathname to the program which is used
3311 to update the NFS client cache entries.
3313 nfs.cache_getent_timeout=
3314 [NFS] sets the timeout after which an attempt to
3315 update a cache entry is deemed to have failed.
3317 nfs.idmap_cache_timeout=
3318 [NFS] set the maximum lifetime for idmapper cache
3322 [NFS] enable 64-bit inode numbers.
3323 If zero, the NFS client will fake up a 32-bit inode
3324 number for the readdir() and stat() syscalls instead
3325 of returning the full 64-bit number.
3326 The default is to return 64-bit inode numbers.
3328 nfs.max_session_cb_slots=
3329 [NFSv4.1] Sets the maximum number of session
3330 slots the client will assign to the callback
3331 channel. This determines the maximum number of
3332 callbacks the client will process in parallel for
3333 a particular server.
3335 nfs.max_session_slots=
3336 [NFSv4.1] Sets the maximum number of session slots
3337 the client will attempt to negotiate with the server.
3338 This limits the number of simultaneous RPC requests
3339 that the client can send to the NFSv4.1 server.
3340 Note that there is little point in setting this
3341 value higher than the max_tcp_slot_table_limit.
3343 nfs.nfs4_disable_idmapping=
3344 [NFSv4] When set to the default of '1', this option
3345 ensures that both the RPC level authentication
3346 scheme and the NFS level operations agree to use
3347 numeric uids/gids if the mount is using the
3348 'sec=sys' security flavour. In effect it is
3349 disabling idmapping, which can make migration from
3350 legacy NFSv2/v3 systems to NFSv4 easier.
3351 Servers that do not support this mode of operation
3352 will be autodetected by the client, and it will fall
3353 back to using the idmapper.
3354 To turn off this behaviour, set the value to '0'.
3356 [NFS4] Specify an additional fixed unique ident-
3357 ification string that NFSv4 clients can insert into
3358 their nfs_client_id4 string. This is typically a
3359 UUID that is generated at system install time.
3361 nfs.send_implementation_id =
3362 [NFSv4.1] Send client implementation identification
3363 information in exchange_id requests.
3364 If zero, no implementation identification information
3366 The default is to send the implementation identification
3369 nfs.recover_lost_locks =
3370 [NFSv4] Attempt to recover locks that were lost due
3371 to a lease timeout on the server. Please note that
3372 doing this risks data corruption, since there are
3373 no guarantees that the file will remain unchanged
3374 after the locks are lost.
3375 If you want to enable the kernel legacy behaviour of
3376 attempting to recover these locks, then set this
3378 The default parameter value of '0' causes the kernel
3379 not to attempt recovery of lost locks.
3381 nfs4.layoutstats_timer =
3382 [NFSv4.2] Change the rate at which the kernel sends
3383 layoutstats to the pNFS metadata server.
3385 Setting this to value to 0 causes the kernel to use
3386 whatever value is the default set by the layout
3387 driver. A non-zero value sets the minimum interval
3388 in seconds between layoutstats transmissions.
3390 nfsd.inter_copy_offload_enable =
3391 [NFSv4.2] When set to 1, the server will support
3392 server-to-server copies for which this server is
3393 the destination of the copy.
3395 nfsd.nfsd4_ssc_umount_timeout =
3396 [NFSv4.2] When used as the destination of a
3397 server-to-server copy, knfsd temporarily mounts
3398 the source server. It caches the mount in case
3399 it will be needed again, and discards it if not
3400 used for the number of milliseconds specified by
3403 nfsd.nfs4_disable_idmapping=
3404 [NFSv4] When set to the default of '1', the NFSv4
3405 server will return only numeric uids and gids to
3406 clients using auth_sys, and will accept numeric uids
3407 and gids from such clients. This is intended to ease
3408 migration from NFSv2/v3.
3411 nmi_backtrace.backtrace_idle [KNL]
3412 Dump stacks even of idle CPUs in response to an
3413 NMI stack-backtrace request.
3415 nmi_debug= [KNL,SH] Specify one or more actions to take
3416 when a NMI is triggered.
3417 Format: [state][,regs][,debounce][,die]
3419 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
3420 Format: [panic,][nopanic,][num]
3422 0 - turn hardlockup detector in nmi_watchdog off
3423 1 - turn hardlockup detector in nmi_watchdog on
3424 When panic is specified, panic when an NMI watchdog
3425 timeout occurs (or 'nopanic' to not panic on an NMI
3426 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
3427 To disable both hard and soft lockup detectors,
3428 please see 'nowatchdog'.
3429 This is useful when you use a panic=... timeout and
3430 need the box quickly up again.
3432 These settings can be accessed at runtime via
3433 the nmi_watchdog and hardlockup_panic sysctls.
3435 netpoll.carrier_timeout=
3436 [NET] Specifies amount of time (in seconds) that
3437 netpoll should wait for a carrier. By default netpoll
3440 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3441 emulation library even if a 387 maths coprocessor
3444 no5lvl [X86-64] Disable 5-level paging mode. Forces
3445 kernel to use 4-level paging instead.
3447 nofsgsbase [X86] Disables FSGSBASE instructions.
3450 [HW] Never suspend the console
3451 Disable suspending of consoles during suspend and
3452 hibernate operations. Once disabled, debugging
3453 messages can reach various consoles while the rest
3454 of the system is being put to sleep (ie, while
3455 debugging driver suspend/resume hooks). This may
3456 not work reliably with all consoles, but is known
3457 to work with serial and VGA consoles.
3458 To facilitate more flexible debugging, we also add
3459 console_suspend, a printk module parameter to control
3460 it. Users could use console_suspend (usually
3461 /sys/module/printk/parameters/console_suspend) to
3462 turn on/off it dynamically.
3464 novmcoredd [KNL,KDUMP]
3465 Disable device dump. Device dump allows drivers to
3466 append dump data to vmcore so you can collect driver
3467 specified debug info. Drivers can append the data
3468 without any limit and this data is stored in memory,
3469 so this may cause significant memory stress. Disabling
3470 device dump can help save memory but the driver debug
3471 data will be no longer available. This parameter
3472 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
3475 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
3476 caches in the slab allocator. Saves per-node memory,
3477 but will impact performance.
3481 noaltinstr [S390] Disables alternative instructions patching
3482 (CPU alternatives feature).
3484 noapic [SMP,APIC] Tells the kernel to not make use of any
3485 IOAPICs that may be present in the system.
3487 noautogroup Disable scheduler automatic task group creation.
3489 nobats [PPC] Do not use BATs for mapping kernel lowmem
3490 on "Classic" PPC cores.
3494 delayacct [KNL] Enable per-task delay accounting
3496 nodsp [SH] Disable hardware DSP at boot time.
3498 noefi Disable EFI runtime services support.
3500 no_entry_flush [PPC] Don't flush the L1-D cache when entering the kernel.
3505 Disable SMAP (Supervisor Mode Access Prevention)
3506 even if it is supported by processor.
3509 Disable SMEP (Supervisor Mode Execution Prevention)
3510 even if it is supported by processor.
3513 This affects only 32-bit executables.
3514 noexec32=on: enable non-executable mappings (default)
3515 read doesn't imply executable mappings
3516 noexec32=off: disable non-executable mappings
3517 read implies executable mappings
3519 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3521 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3522 register save and restore. The kernel will only save
3523 legacy floating-point registers on task switch.
3525 nohugeiomap [KNL,X86,PPC,ARM64] Disable kernel huge I/O mappings.
3527 nohugevmalloc [PPC] Disable kernel huge vmalloc mappings.
3529 nosmt [KNL,S390] Disable symmetric multithreading (SMT).
3530 Equivalent to smt=1.
3532 [KNL,X86] Disable symmetric multithreading (SMT).
3533 nosmt=force: Force disable SMT, cannot be undone
3534 via the sysfs control file.
3536 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1
3537 (bounds check bypass). With this option data leaks are
3538 possible in the system.
3540 nospectre_v2 [X86,PPC_FSL_BOOK3E,ARM64] Disable all mitigations for
3541 the Spectre variant 2 (indirect branch prediction)
3542 vulnerability. System may allow data leaks with this
3545 nospec_store_bypass_disable
3546 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
3549 [PPC] Don't flush the L1-D cache after accessing user data.
3551 noxsave [BUGS=X86] Disables x86 extended register state save
3552 and restore using xsave. The kernel will fallback to
3553 enabling legacy floating-point and sse state.
3555 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
3556 register states. The kernel will fall back to use
3557 xsave to save the states. By using this parameter,
3558 performance of saving the states is degraded because
3559 xsave doesn't support modified optimization while
3560 xsaveopt supports it on xsaveopt enabled systems.
3562 noxsaves [X86] Disables xsaves and xrstors used in saving and
3563 restoring x86 extended register state in compacted
3564 form of xsave area. The kernel will fall back to use
3565 xsaveopt and xrstor to save and restore the states
3566 in standard form of xsave area. By using this
3567 parameter, xsave area per process might occupy more
3568 memory on xsaves enabled systems.
3570 nohlt [ARM,ARM64,MICROBLAZE,SH] Forces the kernel to busy wait
3571 in do_idle() and not use the arch_cpu_idle()
3572 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
3573 to be effective. This is useful on platforms where the
3574 sleep(SH) or wfi(ARM,ARM64) instructions do not work
3575 correctly or when doing power measurements to evalute
3576 the impact of the sleep instructions. This is also
3577 useful when using JTAG debugger.
3579 no_file_caps Tells the kernel not to honor file capabilities. The
3580 only way then for a file to be executed with privilege
3581 is to be setuid root or executed by root.
3583 nohalt [IA-64] Tells the kernel not to use the power saving
3584 function PAL_HALT_LIGHT when idle. This increases
3585 power-consumption. On the positive side, it reduces
3586 interrupt wake-up latency, which may improve performance
3587 in certain environments such as networked servers or
3591 Force pointers printed to the console or buffers to be
3592 unhashed. By default, when a pointer is printed via %p
3593 format string, that pointer is "hashed", i.e. obscured
3594 by hashing the pointer value. This is a security feature
3595 that hides actual kernel addresses from unprivileged
3596 users, but it also makes debugging the kernel more
3597 difficult since unequal pointers can no longer be
3598 compared. However, if this command-line option is
3599 specified, then all normal pointers will have their true
3600 value printed. This option should only be specified when
3601 debugging the kernel. Please do not use on production
3604 nohibernate [HIBERNATION] Disable hibernation and resume.
3606 nohz= [KNL] Boottime enable/disable dynamic ticks
3607 Valid arguments: on, off
3610 nohz_full= [KNL,BOOT,SMP,ISOL]
3611 The argument is a cpu list, as described above.
3612 In kernels built with CONFIG_NO_HZ_FULL=y, set
3613 the specified list of CPUs whose tick will be stopped
3614 whenever possible. The boot CPU will be forced outside
3615 the range to maintain the timekeeping. Any CPUs
3616 in this list will have their RCU callbacks offloaded,
3617 just as if they had also been called out in the
3618 rcu_nocbs= boot parameter.
3620 noiotrap [SH] Disables trapped I/O port accesses.
3622 noirqdebug [X86-32] Disables the code which attempts to detect and
3623 disable unhandled interrupt sources.
3625 no_timer_check [X86,APIC] Disables the code which tests for
3626 broken timer IRQ sources.
3628 noisapnp [ISAPNP] Disables ISA PnP code.
3630 noinitrd [RAM] Tells the kernel not to load any configured
3633 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
3635 [Deprecated - use intremap=off]
3639 noinvpcid [X86] Disable the INVPCID cpu feature.
3641 nojitter [IA-64] Disables jitter checking for ITC timers.
3643 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
3645 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
3649 [X86,PV_OPS] Disable paravirtualized VMware scheduler
3650 clock and use the default one.
3652 no-steal-acc [X86,PV_OPS,ARM64] Disable paravirtualized steal time
3653 accounting. steal time is computed, but won't
3654 influence scheduler behaviour
3656 nolapic [X86-32,APIC] Do not enable or use the local APIC.
3658 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
3660 noltlbs [PPC] Do not use large page/tlb entries for kernel
3661 lowmem mapping on PPC40x and PPC8xx
3663 nomca [IA-64] Disable machine check abort handling
3665 nomce [X86-32] Disable Machine Check Exception
3667 nomfgpt [X86-32] Disable Multi-Function General Purpose
3668 Timer usage (for AMD Geode machines).
3670 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3671 shutdown the other cpus. Instead use the REBOOT_VECTOR
3674 nomodeset Disable kernel modesetting. DRM drivers will not perform
3675 display-mode changes or accelerated rendering. Only the
3676 system framebuffer will be available for use if this was
3677 set-up by the firmware or boot loader.
3679 Useful as fallback, or for testing and debugging.
3681 nomodule Disable module load
3683 nopat [X86] Disable PAT (page attribute table extension of
3684 pagetables) support.
3686 nopcid [X86-64] Disable the PCID cpu feature.
3688 norandmaps Don't use address space randomization. Equivalent to
3689 echo 0 > /proc/sys/kernel/randomize_va_space
3691 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
3692 with UP alternatives
3694 nordrand [X86] Disable kernel use of the RDRAND and
3695 RDSEED instructions even if they are supported
3696 by the processor. RDRAND and RDSEED are still
3697 available to user space applications.
3699 noresume [SWSUSP] Disables resume and restores original swap
3702 no-scroll [VGA] Disables scrollback.
3703 This is required for the Braillex ib80-piezo Braille
3704 reader made by F.H. Papenmeier (Germany).
3708 nosgx [X86-64,SGX] Disables Intel SGX kernel support.
3710 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
3711 and disable the IO APIC. legacy for "maxcpus=0".
3713 nosoftlockup [KNL] Disable the soft-lockup detector.
3715 nosync [HW,M68K] Disables sync negotiation for all devices.
3717 nowatchdog [KNL] Disable both lockup detectors, i.e.
3718 soft-lockup and NMI watchdog (hard-lockup).
3722 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
3724 cpu0_hotplug [X86] Turn on CPU0 hotplug feature when
3725 CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off.
3726 Some features depend on CPU0. Known dependencies are:
3727 1. Resume from suspend/hibernate depends on CPU0.
3728 Suspend/hibernate will fail if CPU0 is offline and you
3729 need to online CPU0 before suspend/hibernate.
3730 2. PIC interrupts also depend on CPU0. CPU0 can't be
3731 removed if a PIC interrupt is detected.
3732 It's said poweroff/reboot may depend on CPU0 on some
3733 machines although I haven't seen such issues so far
3734 after CPU0 is offline on a few tested machines.
3735 If the dependencies are under your control, you can
3736 turn on cpu0_hotplug.
3738 nps_mtm_hs_ctr= [KNL,ARC]
3739 This parameter sets the maximum duration, in
3740 cycles, each HW thread of the CTOP can run
3741 without interruptions, before HW switches it.
3742 The actual maximum duration is 16 times this
3744 Format: integer between 1 and 255
3747 nptcg= [IA-64] Override max number of concurrent global TLB
3748 purges which is reported from either PAL_VM_SUMMARY or
3751 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
3752 could support. nr_cpus=n : n >= 1 limits the kernel to
3753 support 'n' processors. It could be larger than the
3754 number of already plugged CPU during bootup, later in
3755 runtime you can physically add extra cpu until it reaches
3756 n. So during boot up some boot time memory for per-cpu
3757 variables need be pre-allocated for later physical cpu
3760 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
3762 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86] Disable NUMA, Only
3763 set up a single NUMA node spanning all memory.
3765 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic
3767 Allowed values are enable and disable
3769 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
3770 'node', 'default' can be specified
3771 This can be set from sysctl after boot.
3772 See Documentation/admin-guide/sysctl/vm.rst for details.
3774 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
3775 See Documentation/core-api/debugging-via-ohci1394.rst for more
3778 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
3779 Rather than timing out after 20 ms if an EC
3780 command is not properly ACKed, override the length
3781 of the timeout. We have interrupts disabled while
3782 waiting for the ACK, so if this is set too high
3783 interrupts *may* be lost!
3785 omap_mux= [OMAP] Override bootloader pin multiplexing.
3786 Format: <mux_mode0.mode_name=value>...
3787 For example, to override I2C bus2:
3788 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
3790 oops=panic Always panic on oopses. Default is to just kill the
3791 process, but there is a small probability of
3792 deadlocking the machine.
3793 This will also cause panics on machine check exceptions.
3794 Useful together with panic=30 to trigger a reboot.
3797 [KNL] Boolean flag to control whether the page allocator
3798 should randomize its free lists. The randomization may
3799 be automatically enabled if the kernel detects it is
3800 running on a platform with a direct-mapped memory-side
3801 cache, and this parameter can be used to
3802 override/disable that behavior. The state of the flag
3803 can be read from sysfs at:
3804 /sys/module/page_alloc/parameters/shuffle.
3806 page_owner= [KNL] Boot-time page_owner enabling option.
3807 Storage of the information about who allocated
3808 each page is disabled in default. With this switch,
3810 on: enable the feature
3812 page_poison= [KNL] Boot-time parameter changing the state of
3813 poisoning on the buddy allocator, available with
3814 CONFIG_PAGE_POISONING=y.
3815 off: turn off poisoning (default)
3816 on: turn on poisoning
3818 page_reporting.page_reporting_order=
3819 [KNL] Minimal page reporting order
3821 Adjust the minimal page reporting order. The page
3822 reporting is disabled when it exceeds (MAX_ORDER-1).
3824 panic= [KNL] Kernel behaviour on panic: delay <timeout>
3825 timeout > 0: seconds before rebooting
3826 timeout = 0: wait forever
3827 timeout < 0: reboot immediately
3830 panic_print= Bitmask for printing system info when panic happens.
3831 User can chose combination of the following bits:
3832 bit 0: print all tasks info
3833 bit 1: print system memory info
3834 bit 2: print timer info
3835 bit 3: print locks info if CONFIG_LOCKDEP is on
3836 bit 4: print ftrace buffer
3837 bit 5: print all printk messages in buffer
3838 bit 6: print all CPUs backtrace (if available in the arch)
3839 *Be aware* that this option may print a _lot_ of lines,
3840 so there are risks of losing older messages in the log.
3841 Use this option carefully, maybe worth to setup a
3842 bigger log buffer with "log_buf_len" along with this.
3844 panic_on_taint= Bitmask for conditionally calling panic() in add_taint()
3845 Format: <hex>[,nousertaint]
3846 Hexadecimal bitmask representing the set of TAINT flags
3847 that will cause the kernel to panic when add_taint() is
3848 called with any of the flags in this set.
3849 The optional switch "nousertaint" can be utilized to
3850 prevent userspace forced crashes by writing to sysctl
3851 /proc/sys/kernel/tainted any flagset matching with the
3852 bitmask set on panic_on_taint.
3853 See Documentation/admin-guide/tainted-kernels.rst for
3854 extra details on the taint flags that users can pick
3855 to compose the bitmask to assign to panic_on_taint.
3857 panic_on_warn panic() instead of WARN(). Useful to cause kdump
3860 crash_kexec_post_notifiers
3861 Run kdump after running panic-notifiers and dumping
3862 kmsg. This only for the users who doubt kdump always
3863 succeeds in any situation.
3864 Note that this also increases risks of kdump failure,
3865 because some panic notifiers can make the crashed
3866 kernel more unstable.
3868 parkbd.port= [HW] Parallel port number the keyboard adapter is
3869 connected to, default is 0.
3871 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
3872 0 for XT, 1 for AT (default is AT).
3875 parport= [HW,PPT] Specify parallel ports. 0 disables.
3876 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
3877 Use 'auto' to force the driver to use any
3878 IRQ/DMA settings detected (the default is to
3879 ignore detected IRQ/DMA settings because of
3880 possible conflicts). You can specify the base
3881 address, IRQ, and DMA settings; IRQ and DMA
3882 should be numbers, or 'auto' (for using detected
3883 settings on that particular port), or 'nofifo'
3884 (to avoid using a FIFO even if it is detected).
3885 Parallel ports are assigned in the order they
3886 are specified on the command line, starting
3889 parport_init_mode= [HW,PPT]
3890 Configure VIA parallel port to operate in
3891 a specific mode. This is necessary on Pegasos
3892 computer where firmware has no options for setting
3893 up parallel port mode and sets it to spp.
3894 Currently this function knows 686a and 8231 chips.
3895 Format: [spp|ps2|epp|ecp|ecpepp]
3897 pata_legacy.all= [HW,LIBATA]
3899 Set to non-zero to probe primary and secondary ISA
3900 port ranges on PCI systems where no PCI PATA device
3901 has been found at either range. Disabled by default.
3903 pata_legacy.autospeed= [HW,LIBATA]
3905 Set to non-zero if a chip is present that snoops speed
3906 changes. Disabled by default.
3908 pata_legacy.ht6560a= [HW,LIBATA]
3910 Set to 1, 2, or 3 for HT 6560A on the primary channel,
3911 the secondary channel, or both channels respectively.
3912 Disabled by default.
3914 pata_legacy.ht6560b= [HW,LIBATA]
3916 Set to 1, 2, or 3 for HT 6560B on the primary channel,
3917 the secondary channel, or both channels respectively.
3918 Disabled by default.
3920 pata_legacy.iordy_mask= [HW,LIBATA]
3922 IORDY enable mask. Set individual bits to allow IORDY
3923 for the respective channel. Bit 0 is for the first
3924 legacy channel handled by this driver, bit 1 is for
3925 the second channel, and so on. The sequence will often
3926 correspond to the primary legacy channel, the secondary
3927 legacy channel, and so on, but the handling of a PCI
3928 bus and the use of other driver options may interfere
3929 with the sequence. By default IORDY is allowed across
3932 pata_legacy.opti82c46x= [HW,LIBATA]
3934 Set to 1, 2, or 3 for Opti 82c611A on the primary
3935 channel, the secondary channel, or both channels
3936 respectively. Disabled by default.
3938 pata_legacy.opti82c611a= [HW,LIBATA]
3940 Set to 1, 2, or 3 for Opti 82c465MV on the primary
3941 channel, the secondary channel, or both channels
3942 respectively. Disabled by default.
3944 pata_legacy.pio_mask= [HW,LIBATA]
3946 PIO mode mask for autospeed devices. Set individual
3947 bits to allow the use of the respective PIO modes.
3948 Bit 0 is for mode 0, bit 1 is for mode 1, and so on.
3949 All modes allowed by default.
3951 pata_legacy.probe_all= [HW,LIBATA]
3953 Set to non-zero to probe tertiary and further ISA
3954 port ranges on PCI systems. Disabled by default.
3956 pata_legacy.probe_mask= [HW,LIBATA]
3958 Probe mask for legacy ISA PATA ports. Depending on
3959 platform configuration and the use of other driver
3960 options up to 6 legacy ports are supported: 0x1f0,
3961 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing
3962 of individual ports can be disabled by setting the
3963 corresponding bits in the mask to 1. Bit 0 is for
3964 the first port in the list above (0x1f0), and so on.
3965 By default all supported ports are probed.
3967 pata_legacy.qdi= [HW,LIBATA]
3969 Set to non-zero to probe QDI controllers. By default
3970 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise.
3972 pata_legacy.winbond= [HW,LIBATA]
3974 Set to non-zero to probe Winbond controllers. Use
3975 the standard I/O port (0x130) if 1, otherwise the
3976 value given is the I/O port to use (typically 0x1b0).
3977 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE,
3980 pata_platform.pio_mask= [HW,LIBATA]
3982 Supported PIO mode mask. Set individual bits to allow
3983 the use of the respective PIO modes. Bit 0 is for
3984 mode 0, bit 1 is for mode 1, and so on. Mode 0 only
3988 Halt all CPUs after the first oops has been printed for
3989 the specified number of seconds. This is to be used if
3990 your oopses keep scrolling off the screen.
3995 See header of drivers/block/paride/pcd.c.
3996 See also Documentation/admin-guide/blockdev/paride.rst.
3998 pci=option[,option...] [PCI] various PCI subsystem options.
4000 Some options herein operate on a specific device
4001 or a set of devices (<pci_dev>). These are
4002 specified in one of the following formats:
4004 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
4005 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
4007 Note: the first format specifies a PCI
4008 bus/device/function address which may change
4009 if new hardware is inserted, if motherboard
4010 firmware changes, or due to changes caused
4011 by other kernel parameters. If the
4012 domain is left unspecified, it is
4013 taken to be zero. Optionally, a path
4014 to a device through multiple device/function
4015 addresses can be specified after the base
4016 address (this is more robust against
4017 renumbering issues). The second format
4018 selects devices using IDs from the
4019 configuration space which may match multiple
4020 devices in the system.
4022 earlydump dump PCI config space before the kernel
4024 off [X86] don't probe for the PCI bus
4025 bios [X86-32] force use of PCI BIOS, don't access
4026 the hardware directly. Use this if your machine
4027 has a non-standard PCI host bridge.
4028 nobios [X86-32] disallow use of PCI BIOS, only direct
4029 hardware access methods are allowed. Use this
4030 if you experience crashes upon bootup and you
4031 suspect they are caused by the BIOS.
4032 conf1 [X86] Force use of PCI Configuration Access
4033 Mechanism 1 (config address in IO port 0xCF8,
4034 data in IO port 0xCFC, both 32-bit).
4035 conf2 [X86] Force use of PCI Configuration Access
4036 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
4037 the function, IO port 0xCFA, also 8-bit, sets
4038 bus number. The config space is then accessed
4039 through ports 0xC000-0xCFFF).
4040 See http://wiki.osdev.org/PCI for more info
4041 on the configuration access mechanisms.
4042 noaer [PCIE] If the PCIEAER kernel config parameter is
4043 enabled, this kernel boot option can be used to
4044 disable the use of PCIE advanced error reporting.
4045 nodomains [PCI] Disable support for multiple PCI
4046 root domains (aka PCI segments, in ACPI-speak).
4047 nommconf [X86] Disable use of MMCONFIG for PCI
4049 check_enable_amd_mmconf [X86] check for and enable
4050 properly configured MMIO access to PCI
4051 config space on AMD family 10h CPU
4052 nomsi [MSI] If the PCI_MSI kernel config parameter is
4053 enabled, this kernel boot option can be used to
4054 disable the use of MSI interrupts system-wide.
4055 noioapicquirk [APIC] Disable all boot interrupt quirks.
4056 Safety option to keep boot IRQs enabled. This
4057 should never be necessary.
4058 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
4059 primary IO-APIC for bridges that cannot disable
4060 boot IRQs. This fixes a source of spurious IRQs
4061 when the system masks IRQs.
4062 noioapicreroute [APIC] Disable workaround that uses the
4063 boot IRQ equivalent of an IRQ that connects to
4064 a chipset where boot IRQs cannot be disabled.
4065 The opposite of ioapicreroute.
4066 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
4067 routing table. These calls are known to be buggy
4068 on several machines and they hang the machine
4069 when used, but on other computers it's the only
4070 way to get the interrupt routing table. Try
4071 this option if the kernel is unable to allocate
4072 IRQs or discover secondary PCI buses on your
4074 rom [X86] Assign address space to expansion ROMs.
4075 Use with caution as certain devices share
4076 address decoders between ROMs and other
4078 norom [X86] Do not assign address space to
4079 expansion ROMs that do not already have
4080 BIOS assigned address ranges.
4081 nobar [X86] Do not assign address space to the
4082 BARs that weren't assigned by the BIOS.
4083 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
4084 assigned automatically to PCI devices. You can
4085 make the kernel exclude IRQs of your ISA cards
4087 pirqaddr=0xAAAAA [X86] Specify the physical address
4088 of the PIRQ table (normally generated
4089 by the BIOS) if it is outside the
4090 F0000h-100000h range.
4091 lastbus=N [X86] Scan all buses thru bus #N. Can be
4092 useful if the kernel is unable to find your
4093 secondary buses and you want to tell it
4094 explicitly which ones they are.
4095 assign-busses [X86] Always assign all PCI bus
4096 numbers ourselves, overriding
4097 whatever the firmware may have done.
4098 usepirqmask [X86] Honor the possible IRQ mask stored
4099 in the BIOS $PIR table. This is needed on
4100 some systems with broken BIOSes, notably
4101 some HP Pavilion N5400 and Omnibook XE3
4102 notebooks. This will have no effect if ACPI
4103 IRQ routing is enabled.
4104 noacpi [X86] Do not use ACPI for IRQ routing
4105 or for PCI scanning.
4106 use_crs [X86] Use PCI host bridge window information
4107 from ACPI. On BIOSes from 2008 or later, this
4108 is enabled by default. If you need to use this,
4109 please report a bug.
4110 nocrs [X86] Ignore PCI host bridge windows from ACPI.
4111 If you need to use this, please report a bug.
4112 routeirq Do IRQ routing for all PCI devices.
4113 This is normally done in pci_enable_device(),
4114 so this option is a temporary workaround
4115 for broken drivers that don't call it.
4116 skip_isa_align [X86] do not align io start addr, so can
4117 handle more pci cards
4118 noearly [X86] Don't do any early type 1 scanning.
4119 This might help on some broken boards which
4120 machine check when some devices' config space
4121 is read. But various workarounds are disabled
4122 and some IOMMU drivers will not work.
4123 bfsort Sort PCI devices into breadth-first order.
4124 This sorting is done to get a device
4125 order compatible with older (<= 2.4) kernels.
4126 nobfsort Don't sort PCI devices into breadth-first order.
4127 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
4128 tuning and use the BIOS-configured MPS defaults.
4129 pcie_bus_safe Set every device's MPS to the largest value
4130 supported by all devices below the root complex.
4131 pcie_bus_perf Set device MPS to the largest allowable MPS
4132 based on its parent bus. Also set MRRS (Max
4133 Read Request Size) to the largest supported
4134 value (no larger than the MPS that the device
4135 or bus can support) for best performance.
4136 pcie_bus_peer2peer Set every device's MPS to 128B, which
4137 every device is guaranteed to support. This
4138 configuration allows peer-to-peer DMA between
4139 any pair of devices, possibly at the cost of
4140 reduced performance. This also guarantees
4141 that hot-added devices will work.
4142 cbiosize=nn[KMG] The fixed amount of bus space which is
4143 reserved for the CardBus bridge's IO window.
4144 The default value is 256 bytes.
4145 cbmemsize=nn[KMG] The fixed amount of bus space which is
4146 reserved for the CardBus bridge's memory
4147 window. The default value is 64 megabytes.
4150 [<order of align>@]<pci_dev>[; ...]
4151 Specifies alignment and device to reassign
4152 aligned memory resources. How to
4153 specify the device is described above.
4154 If <order of align> is not specified,
4155 PAGE_SIZE is used as alignment.
4156 A PCI-PCI bridge can be specified if resource
4157 windows need to be expanded.
4158 To specify the alignment for several
4159 instances of a device, the PCI vendor,
4160 device, subvendor, and subdevice may be
4161 specified, e.g., 12@pci:8086:9c22:103c:198f
4162 for 4096-byte alignment.
4163 ecrc= Enable/disable PCIe ECRC (transaction layer
4164 end-to-end CRC checking).
4165 bios: Use BIOS/firmware settings. This is the
4169 hpiosize=nn[KMG] The fixed amount of bus space which is
4170 reserved for hotplug bridge's IO window.
4171 Default size is 256 bytes.
4172 hpmmiosize=nn[KMG] The fixed amount of bus space which is
4173 reserved for hotplug bridge's MMIO window.
4174 Default size is 2 megabytes.
4175 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
4176 reserved for hotplug bridge's MMIO_PREF window.
4177 Default size is 2 megabytes.
4178 hpmemsize=nn[KMG] The fixed amount of bus space which is
4179 reserved for hotplug bridge's MMIO and
4181 Default size is 2 megabytes.
4182 hpbussize=nn The minimum amount of additional bus numbers
4183 reserved for buses below a hotplug bridge.
4185 realloc= Enable/disable reallocating PCI bridge resources
4186 if allocations done by BIOS are too small to
4187 accommodate resources required by all child
4189 off: Turn realloc off
4191 realloc same as realloc=on
4192 noari do not use PCIe ARI.
4193 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
4194 do not use PCIe ATS (and IOMMU device IOTLB).
4195 pcie_scan_all Scan all possible PCIe devices. Otherwise we
4196 only look for one device below a PCIe downstream
4198 big_root_window Try to add a big 64bit memory window to the PCIe
4199 root complex on AMD CPUs. Some GFX hardware
4200 can resize a BAR to allow access to all VRAM.
4201 Adding the window is slightly risky (it may
4202 conflict with unreported devices), so this
4204 disable_acs_redir=<pci_dev>[; ...]
4205 Specify one or more PCI devices (in the format
4206 specified above) separated by semicolons.
4207 Each device specified will have the PCI ACS
4208 redirect capabilities forced off which will
4209 allow P2P traffic between devices through
4210 bridges without forcing it upstream. Note:
4211 this removes isolation between devices and
4212 may put more devices in an IOMMU group.
4213 force_floating [S390] Force usage of floating interrupts.
4214 nomio [S390] Do not use MIO instructions.
4215 norid [S390] ignore the RID field and force use of
4216 one PCI domain per PCI function
4218 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
4221 force Enable ASPM even on devices that claim not to support it.
4222 WARNING: Forcing ASPM on may cause system lockups.
4224 pcie_ports= [PCIE] PCIe port services handling:
4225 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
4226 even if the platform doesn't give the OS permission to
4227 use them. This may cause conflicts if the platform
4228 also tries to use these services.
4229 dpc-native Use native PCIe service for DPC only. May
4230 cause conflicts if firmware uses AER or DPC.
4231 compat Disable native PCIe services (PME, AER, DPC, PCIe
4234 pcie_port_pm= [PCIE] PCIe port power management handling:
4235 off Disable power management of all PCIe ports
4236 force Forcibly enable power management of all PCIe ports
4238 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
4239 nomsi Do not use MSI for native PCIe PME signaling (this makes
4240 all PCIe root ports use INTx for all services).
4242 pcmv= [HW,PCMCIA] BadgePAD 4
4246 Keep all power-domains already enabled by bootloader on,
4247 even if no driver has claimed them. This is useful
4248 for debug and development, but should not be
4249 needed on a platform with proper driver support.
4252 See Documentation/admin-guide/blockdev/paride.rst.
4254 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
4257 See arch/parisc/kernel/pdc_chassis.c
4259 percpu_alloc= Select which percpu first chunk allocator to use.
4260 Currently supported values are "embed" and "page".
4261 Archs may support subset or none of the selections.
4262 See comments in mm/percpu.c for details on each
4263 allocator. This parameter is primarily for debugging
4264 and performance comparison.
4267 See Documentation/admin-guide/blockdev/paride.rst.
4270 See Documentation/admin-guide/blockdev/paride.rst.
4272 pirq= [SMP,APIC] Manual mp-table setup
4273 See Documentation/x86/i386/IO-APIC.rst.
4275 plip= [PPT,NET] Parallel port network link
4276 Format: { parport<nr> | timid | 0 }
4277 See also Documentation/admin-guide/parport.rst.
4279 pmtmr= [X86] Manual setup of pmtmr I/O Port.
4280 Override pmtimer IOPort with a hex value.
4283 pmu_override= [PPC] Override the PMU.
4284 This option takes over the PMU facility, so it is no
4285 longer usable by perf. Setting this option starts the
4286 PMU counters by setting MMCR0 to 0 (the FC bit is
4287 cleared). If a number is given, then MMCR1 is set to
4288 that number, otherwise (e.g., 'pmu_override=on'), MMCR1
4291 pm_debug_messages [SUSPEND,KNL]
4292 Enable suspend/resume debug messages during boot up.
4295 Enable PNP debug messages (depends on the
4296 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
4297 via /sys/module/pnp/parameters/debug. We always show
4298 current resource usage; turning this on also shows
4299 possible settings and some assignment information.
4305 { on | off | curr | res | no-curr | no-res }
4308 [ISAPNP] Exclude IRQs for the autoconfiguration
4311 [ISAPNP] Exclude DMAs for the autoconfiguration
4313 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
4314 Ranges are in pairs (I/O port base and size).
4317 [ISAPNP] Exclude memory regions for the
4319 Ranges are in pairs (memory base and size).
4321 ports= [IP_VS_FTP] IPVS ftp helper module
4323 Up to 8 (IP_VS_APP_MAX_PORTS) ports
4325 Format: <port>,<port>....
4327 powersave=off [PPC] This option disables power saving features.
4328 It specifically disables cpuidle and sets the
4329 platform machine description specific power_save
4330 function to NULL. On Idle the CPU just reduces
4333 ppc_strict_facility_enable
4334 [PPC] This option catches any kernel floating point,
4335 Altivec, VSX and SPE outside of regions specifically
4336 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
4337 There is some performance impact when enabling this.
4341 Disable Hardware Transactional Memory
4344 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC
4345 none - Limited to cond_resched() calls
4346 voluntary - Limited to cond_resched() and might_sleep() calls
4347 full - Any section that isn't explicitly preempt disabled
4348 can be preempted anytime.
4350 print-fatal-signals=
4351 [KNL] debug: print fatal signals
4353 If enabled, warn about various signal handling
4354 related application anomalies: too many signals,
4355 too many POSIX.1 timers, fatal signals causing a
4358 If you hit the warning due to signal overflow,
4359 you might want to try "ulimit -i unlimited".
4363 printk.always_kmsg_dump=
4364 Trigger kmsg_dump for cases other than kernel oops or
4366 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4369 printk.console_no_auto_verbose=
4370 Disable console loglevel raise on oops, panic
4371 or lockdep-detected issues (only if lock debug is on).
4372 With an exception to setups with low baudrate on
4373 serial console, keeping this 0 is a good choice
4374 in order to provide more debug information.
4376 default: 0 (auto_verbose is enabled)
4378 printk.devkmsg={on,off,ratelimit}
4379 Control writing to /dev/kmsg.
4380 on - unlimited logging to /dev/kmsg from userspace
4381 off - logging to /dev/kmsg disabled
4382 ratelimit - ratelimit the logging
4385 printk.time= Show timing data prefixed to each printk message line
4386 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4388 processor.max_cstate= [HW,ACPI]
4389 Limit processor to maximum C-state
4390 max_cstate=9 overrides any DMI blacklist limit.
4392 processor.nocst [HW,ACPI]
4393 Ignore the _CST method to determine C-states,
4394 instead using the legacy FADT method
4396 profile= [KNL] Enable kernel profiling via /proc/profile
4397 Format: [<profiletype>,]<number>
4398 Param: <profiletype>: "schedule", "sleep", or "kvm"
4399 [defaults to kernel profiling]
4400 Param: "schedule" - profile schedule points.
4401 Param: "sleep" - profile D-state sleeping (millisecs).
4402 Requires CONFIG_SCHEDSTATS
4403 Param: "kvm" - profile VM exits.
4404 Param: <number> - step/bucket size as a power of 2 for
4405 statistical time based profiling.
4407 prompt_ramdisk= [RAM] [Deprecated]
4409 prot_virt= [S390] enable hosting protected virtual machines
4410 isolated from the hypervisor (if hardware supports
4414 psi= [KNL] Enable or disable pressure stall information
4418 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
4419 probe for; one of (bare|imps|exps|lifebook|any).
4420 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
4422 psmouse.resetafter= [HW,MOUSE]
4423 Try to reset the device after so many bad packets
4426 [HW,MOUSE] Set desired mouse resolution, in dpi.
4427 psmouse.smartscroll=
4428 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
4429 0 = disabled, 1 = enabled (default).
4431 pstore.backend= Specify the name of the pstore backend to use
4434 See Documentation/admin-guide/blockdev/paride.rst.
4436 pti= [X86-64] Control Page Table Isolation of user and
4437 kernel address spaces. Disabling this feature
4438 removes hardening, but improves performance of
4439 system calls and interrupts.
4441 on - unconditionally enable
4442 off - unconditionally disable
4443 auto - kernel detects whether your CPU model is
4444 vulnerable to issues that PTI mitigates
4446 Not specifying this option is equivalent to pti=auto.
4449 Equivalent to pti=off
4452 [KNL] Number of legacy pty's. Overwrites compiled-in
4455 quiet [KNL] Disable most log messages
4460 See Documentation/admin-guide/md.rst.
4462 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
4463 See Documentation/admin-guide/blockdev/ramdisk.rst.
4465 ramdisk_start= [RAM] RAM disk image start address
4467 random.trust_cpu={on,off}
4468 [KNL] Enable or disable trusting the use of the
4469 CPU's random number generator (if available) to
4470 fully seed the kernel's CRNG. Default is controlled
4471 by CONFIG_RANDOM_TRUST_CPU.
4473 random.trust_bootloader={on,off}
4474 [KNL] Enable or disable trusting the use of a
4475 seed passed by the bootloader (if available) to
4476 fully seed the kernel's CRNG. Default is controlled
4477 by CONFIG_RANDOM_TRUST_BOOTLOADER.
4479 randomize_kstack_offset=
4480 [KNL] Enable or disable kernel stack offset
4481 randomization, which provides roughly 5 bits of
4482 entropy, frustrating memory corruption attacks
4483 that depend on stack address determinism or
4484 cross-syscall address exposures. This is only
4485 available on architectures that have defined
4486 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET.
4487 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4488 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT.
4490 ras=option[,option,...] [KNL] RAS-specific options
4493 Disable the Correctable Errors Collector,
4494 see CONFIG_RAS_CEC help text.
4496 rcu_nocbs[=cpu-list]
4497 [KNL] The optional argument is a cpu list,
4500 In kernels built with CONFIG_RCU_NOCB_CPU=y,
4501 enable the no-callback CPU mode, which prevents
4502 such CPUs' callbacks from being invoked in
4503 softirq context. Invocation of such CPUs' RCU
4504 callbacks will instead be offloaded to "rcuox/N"
4505 kthreads created for that purpose, where "x" is
4506 "p" for RCU-preempt, "s" for RCU-sched, and "g"
4507 for the kthreads that mediate grace periods; and
4508 "N" is the CPU number. This reduces OS jitter on
4509 the offloaded CPUs, which can be useful for HPC
4510 and real-time workloads. It can also improve
4511 energy efficiency for asymmetric multiprocessors.
4513 If a cpulist is passed as an argument, the specified
4514 list of CPUs is set to no-callback mode from boot.
4516 Otherwise, if the '=' sign and the cpulist
4517 arguments are omitted, no CPU will be set to
4518 no-callback mode from boot but the mode may be
4519 toggled at runtime via cpusets.
4522 Rather than requiring that offloaded CPUs
4523 (specified by rcu_nocbs= above) explicitly
4524 awaken the corresponding "rcuoN" kthreads,
4525 make these kthreads poll for callbacks.
4526 This improves the real-time response for the
4527 offloaded CPUs by relieving them of the need to
4528 wake up the corresponding kthread, but degrades
4529 energy efficiency by requiring that the kthreads
4530 periodically wake up to do the polling.
4532 rcutree.blimit= [KNL]
4533 Set maximum number of finished RCU callbacks to
4534 process in one batch.
4536 rcutree.dump_tree= [KNL]
4537 Dump the structure of the rcu_node combining tree
4538 out at early boot. This is used for diagnostic
4539 purposes, to verify correct tree setup.
4541 rcutree.gp_cleanup_delay= [KNL]
4542 Set the number of jiffies to delay each step of
4543 RCU grace-period cleanup.
4545 rcutree.gp_init_delay= [KNL]
4546 Set the number of jiffies to delay each step of
4547 RCU grace-period initialization.
4549 rcutree.gp_preinit_delay= [KNL]
4550 Set the number of jiffies to delay each step of
4551 RCU grace-period pre-initialization, that is,
4552 the propagation of recent CPU-hotplug changes up
4553 the rcu_node combining tree.
4555 rcutree.use_softirq= [KNL]
4556 If set to zero, move all RCU_SOFTIRQ processing to
4557 per-CPU rcuc kthreads. Defaults to a non-zero
4558 value, meaning that RCU_SOFTIRQ is used by default.
4559 Specify rcutree.use_softirq=0 to use rcuc kthreads.
4561 But note that CONFIG_PREEMPT_RT=y kernels disable
4562 this kernel boot parameter, forcibly setting it
4565 rcutree.rcu_fanout_exact= [KNL]
4566 Disable autobalancing of the rcu_node combining
4567 tree. This is used by rcutorture, and might
4568 possibly be useful for architectures having high
4569 cache-to-cache transfer latencies.
4571 rcutree.rcu_fanout_leaf= [KNL]
4572 Change the number of CPUs assigned to each
4573 leaf rcu_node structure. Useful for very
4574 large systems, which will choose the value 64,
4575 and for NUMA systems with large remote-access
4576 latencies, which will choose a value aligned
4577 with the appropriate hardware boundaries.
4579 rcutree.rcu_min_cached_objs= [KNL]
4580 Minimum number of objects which are cached and
4581 maintained per one CPU. Object size is equal
4582 to PAGE_SIZE. The cache allows to reduce the
4583 pressure to page allocator, also it makes the
4584 whole algorithm to behave better in low memory
4587 rcutree.rcu_delay_page_cache_fill_msec= [KNL]
4588 Set the page-cache refill delay (in milliseconds)
4589 in response to low-memory conditions. The range
4590 of permitted values is in the range 0:100000.
4592 rcutree.jiffies_till_first_fqs= [KNL]
4593 Set delay from grace-period initialization to
4594 first attempt to force quiescent states.
4595 Units are jiffies, minimum value is zero,
4596 and maximum value is HZ.
4598 rcutree.jiffies_till_next_fqs= [KNL]
4599 Set delay between subsequent attempts to force
4600 quiescent states. Units are jiffies, minimum
4601 value is one, and maximum value is HZ.
4603 rcutree.jiffies_till_sched_qs= [KNL]
4604 Set required age in jiffies for a
4605 given grace period before RCU starts
4606 soliciting quiescent-state help from
4607 rcu_note_context_switch() and cond_resched().
4608 If not specified, the kernel will calculate
4609 a value based on the most recent settings
4610 of rcutree.jiffies_till_first_fqs
4611 and rcutree.jiffies_till_next_fqs.
4612 This calculated value may be viewed in
4613 rcutree.jiffies_to_sched_qs. Any attempt to set
4614 rcutree.jiffies_to_sched_qs will be cheerfully
4617 rcutree.kthread_prio= [KNL,BOOT]
4618 Set the SCHED_FIFO priority of the RCU per-CPU
4619 kthreads (rcuc/N). This value is also used for
4620 the priority of the RCU boost threads (rcub/N)
4621 and for the RCU grace-period kthreads (rcu_bh,
4622 rcu_preempt, and rcu_sched). If RCU_BOOST is
4623 set, valid values are 1-99 and the default is 1
4624 (the least-favored priority). Otherwise, when
4625 RCU_BOOST is not set, valid values are 0-99 and
4626 the default is zero (non-realtime operation).
4627 When RCU_NOCB_CPU is set, also adjust the
4628 priority of NOCB callback kthreads.
4630 rcutree.rcu_nocb_gp_stride= [KNL]
4631 Set the number of NOCB callback kthreads in
4632 each group, which defaults to the square root
4633 of the number of CPUs. Larger numbers reduce
4634 the wakeup overhead on the global grace-period
4635 kthread, but increases that same overhead on
4636 each group's NOCB grace-period kthread.
4638 rcutree.qhimark= [KNL]
4639 Set threshold of queued RCU callbacks beyond which
4640 batch limiting is disabled.
4642 rcutree.qlowmark= [KNL]
4643 Set threshold of queued RCU callbacks below which
4644 batch limiting is re-enabled.
4646 rcutree.qovld= [KNL]
4647 Set threshold of queued RCU callbacks beyond which
4648 RCU's force-quiescent-state scan will aggressively
4649 enlist help from cond_resched() and sched IPIs to
4650 help CPUs more quickly reach quiescent states.
4651 Set to less than zero to make this be set based
4652 on rcutree.qhimark at boot time and to zero to
4653 disable more aggressive help enlistment.
4655 rcutree.rcu_kick_kthreads= [KNL]
4656 Cause the grace-period kthread to get an extra
4657 wake_up() if it sleeps three times longer than
4658 it should at force-quiescent-state time.
4659 This wake_up() will be accompanied by a
4660 WARN_ONCE() splat and an ftrace_dump().
4662 rcutree.rcu_unlock_delay= [KNL]
4663 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels,
4664 this specifies an rcu_read_unlock()-time delay
4665 in microseconds. This defaults to zero.
4666 Larger delays increase the probability of
4667 catching RCU pointer leaks, that is, buggy use
4668 of RCU-protected pointers after the relevant
4669 rcu_read_unlock() has completed.
4671 rcutree.sysrq_rcu= [KNL]
4672 Commandeer a sysrq key to dump out Tree RCU's
4673 rcu_node tree with an eye towards determining
4674 why a new grace period has not yet started.
4676 rcuscale.gp_async= [KNL]
4677 Measure performance of asynchronous
4678 grace-period primitives such as call_rcu().
4680 rcuscale.gp_async_max= [KNL]
4681 Specify the maximum number of outstanding
4682 callbacks per writer thread. When a writer
4683 thread exceeds this limit, it invokes the
4684 corresponding flavor of rcu_barrier() to allow
4685 previously posted callbacks to drain.
4687 rcuscale.gp_exp= [KNL]
4688 Measure performance of expedited synchronous
4689 grace-period primitives.
4691 rcuscale.holdoff= [KNL]
4692 Set test-start holdoff period. The purpose of
4693 this parameter is to delay the start of the
4694 test until boot completes in order to avoid
4697 rcuscale.kfree_rcu_test= [KNL]
4698 Set to measure performance of kfree_rcu() flooding.
4700 rcuscale.kfree_rcu_test_double= [KNL]
4701 Test the double-argument variant of kfree_rcu().
4702 If this parameter has the same value as
4703 rcuscale.kfree_rcu_test_single, both the single-
4704 and double-argument variants are tested.
4706 rcuscale.kfree_rcu_test_single= [KNL]
4707 Test the single-argument variant of kfree_rcu().
4708 If this parameter has the same value as
4709 rcuscale.kfree_rcu_test_double, both the single-
4710 and double-argument variants are tested.
4712 rcuscale.kfree_nthreads= [KNL]
4713 The number of threads running loops of kfree_rcu().
4715 rcuscale.kfree_alloc_num= [KNL]
4716 Number of allocations and frees done in an iteration.
4718 rcuscale.kfree_loops= [KNL]
4719 Number of loops doing rcuscale.kfree_alloc_num number
4720 of allocations and frees.
4722 rcuscale.nreaders= [KNL]
4723 Set number of RCU readers. The value -1 selects
4724 N, where N is the number of CPUs. A value
4725 "n" less than -1 selects N-n+1, where N is again
4726 the number of CPUs. For example, -2 selects N
4727 (the number of CPUs), -3 selects N+1, and so on.
4728 A value of "n" less than or equal to -N selects
4731 rcuscale.nwriters= [KNL]
4732 Set number of RCU writers. The values operate
4733 the same as for rcuscale.nreaders.
4734 N, where N is the number of CPUs
4736 rcuscale.perf_type= [KNL]
4737 Specify the RCU implementation to test.
4739 rcuscale.shutdown= [KNL]
4740 Shut the system down after performance tests
4741 complete. This is useful for hands-off automated
4744 rcuscale.verbose= [KNL]
4745 Enable additional printk() statements.
4747 rcuscale.writer_holdoff= [KNL]
4748 Write-side holdoff between grace periods,
4749 in microseconds. The default of zero says
4752 rcutorture.fqs_duration= [KNL]
4753 Set duration of force_quiescent_state bursts
4756 rcutorture.fqs_holdoff= [KNL]
4757 Set holdoff time within force_quiescent_state bursts
4760 rcutorture.fqs_stutter= [KNL]
4761 Set wait time between force_quiescent_state bursts
4764 rcutorture.fwd_progress= [KNL]
4765 Specifies the number of kthreads to be used
4766 for RCU grace-period forward-progress testing
4767 for the types of RCU supporting this notion.
4768 Defaults to 1 kthread, values less than zero or
4769 greater than the number of CPUs cause the number
4772 rcutorture.fwd_progress_div= [KNL]
4773 Specify the fraction of a CPU-stall-warning
4774 period to do tight-loop forward-progress testing.
4776 rcutorture.fwd_progress_holdoff= [KNL]
4777 Number of seconds to wait between successive
4778 forward-progress tests.
4780 rcutorture.fwd_progress_need_resched= [KNL]
4781 Enclose cond_resched() calls within checks for
4782 need_resched() during tight-loop forward-progress
4785 rcutorture.gp_cond= [KNL]
4786 Use conditional/asynchronous update-side
4787 primitives, if available.
4789 rcutorture.gp_exp= [KNL]
4790 Use expedited update-side primitives, if available.
4792 rcutorture.gp_normal= [KNL]
4793 Use normal (non-expedited) asynchronous
4794 update-side primitives, if available.
4796 rcutorture.gp_sync= [KNL]
4797 Use normal (non-expedited) synchronous
4798 update-side primitives, if available. If all
4799 of rcutorture.gp_cond=, rcutorture.gp_exp=,
4800 rcutorture.gp_normal=, and rcutorture.gp_sync=
4801 are zero, rcutorture acts as if is interpreted
4802 they are all non-zero.
4804 rcutorture.irqreader= [KNL]
4805 Run RCU readers from irq handlers, or, more
4806 accurately, from a timer handler. Not all RCU
4807 flavors take kindly to this sort of thing.
4809 rcutorture.leakpointer= [KNL]
4810 Leak an RCU-protected pointer out of the reader.
4811 This can of course result in splats, and is
4812 intended to test the ability of things like
4813 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect
4816 rcutorture.n_barrier_cbs= [KNL]
4817 Set callbacks/threads for rcu_barrier() testing.
4819 rcutorture.nfakewriters= [KNL]
4820 Set number of concurrent RCU writers. These just
4821 stress RCU, they don't participate in the actual
4822 test, hence the "fake".
4824 rcutorture.nocbs_nthreads= [KNL]
4825 Set number of RCU callback-offload togglers.
4826 Zero (the default) disables toggling.
4828 rcutorture.nocbs_toggle= [KNL]
4829 Set the delay in milliseconds between successive
4830 callback-offload toggling attempts.
4832 rcutorture.nreaders= [KNL]
4833 Set number of RCU readers. The value -1 selects
4834 N-1, where N is the number of CPUs. A value
4835 "n" less than -1 selects N-n-2, where N is again
4836 the number of CPUs. For example, -2 selects N
4837 (the number of CPUs), -3 selects N+1, and so on.
4839 rcutorture.object_debug= [KNL]
4840 Enable debug-object double-call_rcu() testing.
4842 rcutorture.onoff_holdoff= [KNL]
4843 Set time (s) after boot for CPU-hotplug testing.
4845 rcutorture.onoff_interval= [KNL]
4846 Set time (jiffies) between CPU-hotplug operations,
4847 or zero to disable CPU-hotplug testing.
4849 rcutorture.read_exit= [KNL]
4850 Set the number of read-then-exit kthreads used
4851 to test the interaction of RCU updaters and
4852 task-exit processing.
4854 rcutorture.read_exit_burst= [KNL]
4855 The number of times in a given read-then-exit
4856 episode that a set of read-then-exit kthreads
4859 rcutorture.read_exit_delay= [KNL]
4860 The delay, in seconds, between successive
4861 read-then-exit testing episodes.
4863 rcutorture.shuffle_interval= [KNL]
4864 Set task-shuffle interval (s). Shuffling tasks
4865 allows some CPUs to go into dyntick-idle mode
4866 during the rcutorture test.
4868 rcutorture.shutdown_secs= [KNL]
4869 Set time (s) after boot system shutdown. This
4870 is useful for hands-off automated testing.
4872 rcutorture.stall_cpu= [KNL]
4873 Duration of CPU stall (s) to test RCU CPU stall
4874 warnings, zero to disable.
4876 rcutorture.stall_cpu_block= [KNL]
4877 Sleep while stalling if set. This will result
4878 in warnings from preemptible RCU in addition
4879 to any other stall-related activity.
4881 rcutorture.stall_cpu_holdoff= [KNL]
4882 Time to wait (s) after boot before inducing stall.
4884 rcutorture.stall_cpu_irqsoff= [KNL]
4885 Disable interrupts while stalling if set.
4887 rcutorture.stall_gp_kthread= [KNL]
4888 Duration (s) of forced sleep within RCU
4889 grace-period kthread to test RCU CPU stall
4890 warnings, zero to disable. If both stall_cpu
4891 and stall_gp_kthread are specified, the
4892 kthread is starved first, then the CPU.
4894 rcutorture.stat_interval= [KNL]
4895 Time (s) between statistics printk()s.
4897 rcutorture.stutter= [KNL]
4898 Time (s) to stutter testing, for example, specifying
4899 five seconds causes the test to run for five seconds,
4900 wait for five seconds, and so on. This tests RCU's
4901 ability to transition abruptly to and from idle.
4903 rcutorture.test_boost= [KNL]
4904 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
4905 "Maybe" means test if the RCU implementation
4906 under test support RCU priority boosting.
4908 rcutorture.test_boost_duration= [KNL]
4909 Duration (s) of each individual boost test.
4911 rcutorture.test_boost_interval= [KNL]
4912 Interval (s) between each boost test.
4914 rcutorture.test_no_idle_hz= [KNL]
4915 Test RCU's dyntick-idle handling. See also the
4916 rcutorture.shuffle_interval parameter.
4918 rcutorture.torture_type= [KNL]
4919 Specify the RCU implementation to test.
4921 rcutorture.verbose= [KNL]
4922 Enable additional printk() statements.
4924 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
4925 Dump ftrace buffer after reporting RCU CPU
4928 rcupdate.rcu_cpu_stall_suppress= [KNL]
4929 Suppress RCU CPU stall warning messages.
4931 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL]
4932 Suppress RCU CPU stall warning messages and
4933 rcutorture writer stall warnings that occur
4934 during early boot, that is, during the time
4935 before the init task is spawned.
4937 rcupdate.rcu_cpu_stall_timeout= [KNL]
4938 Set timeout for RCU CPU stall warning messages.
4939 The value is in seconds and the maximum allowed
4940 value is 300 seconds.
4942 rcupdate.rcu_exp_cpu_stall_timeout= [KNL]
4943 Set timeout for expedited RCU CPU stall warning
4944 messages. The value is in milliseconds
4945 and the maximum allowed value is 21000
4946 milliseconds. Please note that this value is
4947 adjusted to an arch timer tick resolution.
4948 Setting this to zero causes the value from
4949 rcupdate.rcu_cpu_stall_timeout to be used (after
4950 conversion from seconds to milliseconds).
4952 rcupdate.rcu_expedited= [KNL]
4953 Use expedited grace-period primitives, for
4954 example, synchronize_rcu_expedited() instead
4955 of synchronize_rcu(). This reduces latency,
4956 but can increase CPU utilization, degrade
4957 real-time latency, and degrade energy efficiency.
4958 No effect on CONFIG_TINY_RCU kernels.
4960 rcupdate.rcu_normal= [KNL]
4961 Use only normal grace-period primitives,
4962 for example, synchronize_rcu() instead of
4963 synchronize_rcu_expedited(). This improves
4964 real-time latency, CPU utilization, and
4965 energy efficiency, but can expose users to
4966 increased grace-period latency. This parameter
4967 overrides rcupdate.rcu_expedited. No effect on
4968 CONFIG_TINY_RCU kernels.
4970 rcupdate.rcu_normal_after_boot= [KNL]
4971 Once boot has completed (that is, after
4972 rcu_end_inkernel_boot() has been invoked), use
4973 only normal grace-period primitives. No effect
4974 on CONFIG_TINY_RCU kernels.
4976 But note that CONFIG_PREEMPT_RT=y kernels enables
4977 this kernel boot parameter, forcibly setting
4978 it to the value one, that is, converting any
4979 post-boot attempt at an expedited RCU grace
4980 period to instead use normal non-expedited
4981 grace-period processing.
4983 rcupdate.rcu_task_collapse_lim= [KNL]
4984 Set the maximum number of callbacks present
4985 at the beginning of a grace period that allows
4986 the RCU Tasks flavors to collapse back to using
4987 a single callback queue. This switching only
4988 occurs when rcupdate.rcu_task_enqueue_lim is
4989 set to the default value of -1.
4991 rcupdate.rcu_task_contend_lim= [KNL]
4992 Set the minimum number of callback-queuing-time
4993 lock-contention events per jiffy required to
4994 cause the RCU Tasks flavors to switch to per-CPU
4995 callback queuing. This switching only occurs
4996 when rcupdate.rcu_task_enqueue_lim is set to
4997 the default value of -1.
4999 rcupdate.rcu_task_enqueue_lim= [KNL]
5000 Set the number of callback queues to use for the
5001 RCU Tasks family of RCU flavors. The default
5002 of -1 allows this to be automatically (and
5003 dynamically) adjusted. This parameter is intended
5006 rcupdate.rcu_task_ipi_delay= [KNL]
5007 Set time in jiffies during which RCU tasks will
5008 avoid sending IPIs, starting with the beginning
5009 of a given grace period. Setting a large
5010 number avoids disturbing real-time workloads,
5011 but lengthens grace periods.
5013 rcupdate.rcu_task_stall_info= [KNL]
5014 Set initial timeout in jiffies for RCU task stall
5015 informational messages, which give some indication
5016 of the problem for those not patient enough to
5017 wait for ten minutes. Informational messages are
5018 only printed prior to the stall-warning message
5019 for a given grace period. Disable with a value
5020 less than or equal to zero. Defaults to ten
5021 seconds. A change in value does not take effect
5022 until the beginning of the next grace period.
5024 rcupdate.rcu_task_stall_info_mult= [KNL]
5025 Multiplier for time interval between successive
5026 RCU task stall informational messages for a given
5027 RCU tasks grace period. This value is clamped
5028 to one through ten, inclusive. It defaults to
5029 the value three, so that the first informational
5030 message is printed 10 seconds into the grace
5031 period, the second at 40 seconds, the third at
5032 160 seconds, and then the stall warning at 600
5033 seconds would prevent a fourth at 640 seconds.
5035 rcupdate.rcu_task_stall_timeout= [KNL]
5036 Set timeout in jiffies for RCU task stall
5037 warning messages. Disable with a value less
5038 than or equal to zero. Defaults to ten minutes.
5039 A change in value does not take effect until
5040 the beginning of the next grace period.
5042 rcupdate.rcu_self_test= [KNL]
5043 Run the RCU early boot self tests
5047 Run specified binary instead of /init from the ramdisk,
5048 used for early userspace startup. See initrd.
5051 force - Override the decision by the kernel to hide the
5052 advertisement of RDRAND support (this affects
5053 certain AMD processors because of buggy BIOS
5054 support, specifically around the suspend/resume
5058 Turn on/off individual RDT features. List is:
5059 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
5061 E.g. to turn on cmt and turn off mba use:
5065 Format (x86 or x86_64):
5066 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] | d[efault] \
5068 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
5070 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
5071 (prefix with 'panic_' to set mode for panic
5073 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
5074 reboot_force is either force or not specified,
5075 reboot_cpu is s[mp]#### with #### being the processor
5076 to be used for rebooting.
5078 refscale.holdoff= [KNL]
5079 Set test-start holdoff period. The purpose of
5080 this parameter is to delay the start of the
5081 test until boot completes in order to avoid
5084 refscale.loops= [KNL]
5085 Set the number of loops over the synchronization
5086 primitive under test. Increasing this number
5087 reduces noise due to loop start/end overhead,
5088 but the default has already reduced the per-pass
5089 noise to a handful of picoseconds on ca. 2020
5092 refscale.nreaders= [KNL]
5093 Set number of readers. The default value of -1
5094 selects N, where N is roughly 75% of the number
5095 of CPUs. A value of zero is an interesting choice.
5097 refscale.nruns= [KNL]
5098 Set number of runs, each of which is dumped onto
5101 refscale.readdelay= [KNL]
5102 Set the read-side critical-section duration,
5103 measured in microseconds.
5105 refscale.scale_type= [KNL]
5106 Specify the read-protection implementation to test.
5108 refscale.shutdown= [KNL]
5109 Shut down the system at the end of the performance
5110 test. This defaults to 1 (shut it down) when
5111 refscale is built into the kernel and to 0 (leave
5112 it running) when refscale is built as a module.
5114 refscale.verbose= [KNL]
5115 Enable additional printk() statements.
5117 refscale.verbose_batched= [KNL]
5118 Batch the additional printk() statements. If zero
5119 (the default) or negative, print everything. Otherwise,
5120 print every Nth verbose statement, where N is the value
5124 [KNL, SMP] Set scheduler's default relax_domain_level.
5125 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
5127 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
5128 Format: <base1>,<size1>[,<base2>,<size2>,...]
5129 Reserve I/O ports or memory so the kernel won't use
5130 them. If <base> is less than 0x10000, the region
5131 is assumed to be I/O ports; otherwise it is memory.
5133 reservetop= [X86-32]
5135 Reserves a hole at the top of the kernel virtual
5138 reset_devices [KNL] Force drivers to reset the underlying device
5139 during initialization.
5142 Specify the partition device for software suspend
5144 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
5146 resume_offset= [SWSUSP]
5147 Specify the offset from the beginning of the partition
5148 given by "resume=" at which the swap header is located,
5149 in <PAGE_SIZE> units (needed only for swap files).
5150 See Documentation/power/swsusp-and-swap-files.rst
5152 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
5153 read the resume files
5155 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
5156 Useful for devices that are detected asynchronously
5157 (e.g. USB and MMC devices).
5159 hibernate= [HIBERNATION]
5160 noresume Don't check if there's a hibernation image
5161 present during boot.
5162 nocompress Don't compress/decompress hibernation images.
5163 no Disable hibernation and resume.
5164 protect_image Turn on image protection during restoration
5165 (that will set all pages holding image data
5166 during restoration read-only).
5168 retain_initrd [RAM] Keep initrd memory after extraction
5170 rfkill.default_state=
5171 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
5172 etc. communication is blocked by default.
5175 rfkill.master_switch_mode=
5176 0 The "airplane mode" button does nothing.
5177 1 The "airplane mode" button toggles between everything
5178 blocked and the previous configuration.
5179 2 The "airplane mode" button toggles between everything
5180 blocked and everything unblocked.
5182 rhash_entries= [KNL,NET]
5183 Set number of hash buckets for route cache
5186 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
5189 ro [KNL] Mount root device read-only on boot
5192 on Mark read-only kernel memory as read-only (default).
5193 off Leave read-only kernel memory writable for debugging.
5196 Enable the uart passthrough on the designated usb port
5197 on Rockchip SoCs. When active, the signals of the
5198 debug-uart get routed to the D+ and D- pins of the usb
5199 port and the regular usb controller gets disabled.
5201 root= [KNL] Root filesystem
5202 See name_to_dev_t comment in init/do_mounts.c.
5204 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
5205 mount the root filesystem
5207 rootflags= [KNL] Set root filesystem mount option string
5209 rootfstype= [KNL] Set root filesystem type
5211 rootwait [KNL] Wait (indefinitely) for root device to show up.
5212 Useful for devices that are detected asynchronously
5213 (e.g. USB and MMC devices).
5215 rproc_mem=nn[KMG][@address]
5216 [KNL,ARM,CMA] Remoteproc physical memory block.
5217 Memory area to be used by remote processor image,
5220 rw [KNL] Mount root device read-write on boot
5222 S [KNL] Run init in single mode
5224 s390_iommu= [HW,S390]
5225 Set s390 IOTLB flushing mode
5227 With strict flushing every unmap operation will result in
5228 an IOTLB flush. Default is lazy flushing before reuse,
5231 s390_iommu_aperture= [KNL,S390]
5232 Specifies the size of the per device DMA address space
5233 accessible through the DMA and IOMMU APIs as a decimal
5234 factor of the size of main memory.
5235 The default is 1 meaning that one can concurrently use
5236 as many DMA addresses as physical memory is installed,
5237 if supported by hardware, and thus map all of memory
5238 once. With a value of 2 one can map all of memory twice
5239 and so on. As a special case a factor of 0 imposes no
5240 restrictions other than those given by hardware at the
5241 cost of significant additional memory use for tables.
5244 See drivers/net/irda/sa1100_ir.c.
5246 sched_verbose [KNL] Enables verbose scheduler debug messages.
5248 schedstats= [KNL,X86] Enable or disable scheduled statistics.
5249 Allowed values are enable and disable. This feature
5250 incurs a small amount of overhead in the scheduler
5251 but is useful for debugging and performance tuning.
5253 sched_thermal_decay_shift=
5254 [KNL, SMP] Set a decay shift for scheduler thermal
5255 pressure signal. Thermal pressure signal follows the
5256 default decay period of other scheduler pelt
5257 signals(usually 32 ms but configurable). Setting
5258 sched_thermal_decay_shift will left shift the decay
5259 period for the thermal pressure signal by the shift
5261 i.e. with the default pelt decay period of 32 ms
5262 sched_thermal_decay_shift thermal pressure decay pr
5266 Format: integer between 0 and 10
5269 scftorture.holdoff= [KNL]
5270 Number of seconds to hold off before starting
5271 test. Defaults to zero for module insertion and
5272 to 10 seconds for built-in smp_call_function()
5275 scftorture.longwait= [KNL]
5276 Request ridiculously long waits randomly selected
5277 up to the chosen limit in seconds. Zero (the
5278 default) disables this feature. Please note
5279 that requesting even small non-zero numbers of
5280 seconds can result in RCU CPU stall warnings,
5281 softlockup complaints, and so on.
5283 scftorture.nthreads= [KNL]
5284 Number of kthreads to spawn to invoke the
5285 smp_call_function() family of functions.
5286 The default of -1 specifies a number of kthreads
5287 equal to the number of CPUs.
5289 scftorture.onoff_holdoff= [KNL]
5290 Number seconds to wait after the start of the
5291 test before initiating CPU-hotplug operations.
5293 scftorture.onoff_interval= [KNL]
5294 Number seconds to wait between successive
5295 CPU-hotplug operations. Specifying zero (which
5296 is the default) disables CPU-hotplug operations.
5298 scftorture.shutdown_secs= [KNL]
5299 The number of seconds following the start of the
5300 test after which to shut down the system. The
5301 default of zero avoids shutting down the system.
5302 Non-zero values are useful for automated tests.
5304 scftorture.stat_interval= [KNL]
5305 The number of seconds between outputting the
5306 current test statistics to the console. A value
5307 of zero disables statistics output.
5309 scftorture.stutter_cpus= [KNL]
5310 The number of jiffies to wait between each change
5311 to the set of CPUs under test.
5313 scftorture.use_cpus_read_lock= [KNL]
5314 Use use_cpus_read_lock() instead of the default
5315 preempt_disable() to disable CPU hotplug
5316 while invoking one of the smp_call_function*()
5319 scftorture.verbose= [KNL]
5320 Enable additional printk() statements.
5322 scftorture.weight_single= [KNL]
5323 The probability weighting to use for the
5324 smp_call_function_single() function with a zero
5325 "wait" parameter. A value of -1 selects the
5326 default if all other weights are -1. However,
5327 if at least one weight has some other value, a
5328 value of -1 will instead select a weight of zero.
5330 scftorture.weight_single_wait= [KNL]
5331 The probability weighting to use for the
5332 smp_call_function_single() function with a
5333 non-zero "wait" parameter. See weight_single.
5335 scftorture.weight_many= [KNL]
5336 The probability weighting to use for the
5337 smp_call_function_many() function with a zero
5338 "wait" parameter. See weight_single.
5339 Note well that setting a high probability for
5340 this weighting can place serious IPI load
5343 scftorture.weight_many_wait= [KNL]
5344 The probability weighting to use for the
5345 smp_call_function_many() function with a
5346 non-zero "wait" parameter. See weight_single
5349 scftorture.weight_all= [KNL]
5350 The probability weighting to use for the
5351 smp_call_function_all() function with a zero
5352 "wait" parameter. See weight_single and
5355 scftorture.weight_all_wait= [KNL]
5356 The probability weighting to use for the
5357 smp_call_function_all() function with a
5358 non-zero "wait" parameter. See weight_single
5361 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
5362 xtime_lock contention on larger systems, and/or RCU lock
5363 contention on all systems with CONFIG_MAXSMP set.
5364 Format: { "0" | "1" }
5365 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
5367 Note: increases power consumption, thus should only be
5368 enabled if running jitter sensitive (HPC/RT) workloads.
5370 security= [SECURITY] Choose a legacy "major" security module to
5371 enable at boot. This has been deprecated by the
5374 selinux= [SELINUX] Disable or enable SELinux at boot time.
5375 Format: { "0" | "1" }
5376 See security/selinux/Kconfig help text.
5381 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
5382 Format: { "0" | "1" }
5383 See security/apparmor/Kconfig help text
5386 Default value is set via kernel config option.
5388 serialnumber [BUGS=X86-32]
5390 sev=option[,option...] [X86-64] See Documentation/x86/x86_64/boot-options.rst
5393 Maximal number of shapers.
5401 Enable merging of slabs with similar size when the
5402 kernel is built without CONFIG_SLAB_MERGE_DEFAULT.
5405 Disable merging of slabs with similar size. May be
5406 necessary if there is some reason to distinguish
5407 allocs to different slabs, especially in hardened
5408 environments where the risk of heap overflows and
5409 layout control by attackers can usually be
5410 frustrated by disabling merging. This will reduce
5411 most of the exposure of a heap attack to a single
5412 cache (risks via metadata attacks are mostly
5413 unchanged). Debug options disable merging on their
5415 For more information see Documentation/vm/slub.rst.
5417 slab_max_order= [MM, SLAB]
5418 Determines the maximum allowed order for slabs.
5419 A high setting may cause OOMs due to memory
5420 fragmentation. Defaults to 1 for systems with
5421 more than 32MB of RAM, 0 otherwise.
5423 slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB]
5424 Enabling slub_debug allows one to determine the
5425 culprit if slab objects become corrupted. Enabling
5426 slub_debug can create guard zones around objects and
5427 may poison objects when not in use. Also tracks the
5428 last alloc / free. For more information see
5429 Documentation/vm/slub.rst.
5431 slub_max_order= [MM, SLUB]
5432 Determines the maximum allowed order for slabs.
5433 A high setting may cause OOMs due to memory
5434 fragmentation. For more information see
5435 Documentation/vm/slub.rst.
5437 slub_min_objects= [MM, SLUB]
5438 The minimum number of objects per slab. SLUB will
5439 increase the slab order up to slub_max_order to
5440 generate a sufficiently large slab able to contain
5441 the number of objects indicated. The higher the number
5442 of objects the smaller the overhead of tracking slabs
5443 and the less frequently locks need to be acquired.
5444 For more information see Documentation/vm/slub.rst.
5446 slub_min_order= [MM, SLUB]
5447 Determines the minimum page order for slabs. Must be
5448 lower than slub_max_order.
5449 For more information see Documentation/vm/slub.rst.
5451 slub_merge [MM, SLUB]
5452 Same with slab_merge.
5454 slub_nomerge [MM, SLUB]
5455 Same with slab_nomerge. This is supported for legacy.
5456 See slab_nomerge for more information.
5459 Format: <io1>[,<io2>[,...,<io8>]]
5461 smp.csd_lock_timeout= [KNL]
5462 Specify the period of time in milliseconds
5463 that smp_call_function() and friends will wait
5464 for a CPU to release the CSD lock. This is
5465 useful when diagnosing bugs involving CPUs
5466 disabling interrupts for extended periods
5467 of time. Defaults to 5,000 milliseconds, and
5468 setting a value of zero disables this feature.
5469 This feature may be more efficiently disabled
5470 using the csdlock_debug- kernel parameter.
5472 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
5473 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
5474 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
5475 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
5476 smsc-ircc2.ircc_irq= [HW] IRQ line
5477 smsc-ircc2.ircc_dma= [HW] DMA channel
5478 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
5479 0: Toshiba Satellite 1800 (GP data pin select)
5480 1: Fast pin select (default)
5483 smt [KNL,S390] Set the maximum number of threads (logical
5484 CPUs) to use per physical CPU on systems capable of
5485 symmetric multithreading (SMT). Will be capped to the
5486 actual hardware limit.
5488 Default: -1 (no limit)
5491 [KNL] Should the soft-lockup detector generate panics.
5494 A value of 1 instructs the soft-lockup detector
5495 to panic the machine when a soft-lockup occurs. It is
5496 also controlled by the kernel.softlockup_panic sysctl
5497 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
5498 respective build-time switch to that functionality.
5500 softlockup_all_cpu_backtrace=
5501 [KNL] Should the soft-lockup detector generate
5502 backtraces on all cpus.
5505 sonypi.*= [HW] Sony Programmable I/O Control Device driver
5506 See Documentation/admin-guide/laptops/sonypi.rst
5508 spectre_v2= [X86] Control mitigation of Spectre variant 2
5509 (indirect branch speculation) vulnerability.
5510 The default operation protects the kernel from
5513 on - unconditionally enable, implies
5515 off - unconditionally disable, implies
5517 auto - kernel detects whether your CPU model is
5520 Selecting 'on' will, and 'auto' may, choose a
5521 mitigation method at run time according to the
5522 CPU, the available microcode, the setting of the
5523 CONFIG_RETPOLINE configuration option, and the
5524 compiler with which the kernel was built.
5526 Selecting 'on' will also enable the mitigation
5527 against user space to user space task attacks.
5529 Selecting 'off' will disable both the kernel and
5530 the user space protections.
5532 Specific mitigations can also be selected manually:
5534 retpoline - replace indirect branches
5535 retpoline,generic - Retpolines
5536 retpoline,lfence - LFENCE; indirect branch
5537 retpoline,amd - alias for retpoline,lfence
5538 eibrs - enhanced IBRS
5539 eibrs,retpoline - enhanced IBRS + Retpolines
5540 eibrs,lfence - enhanced IBRS + LFENCE
5542 Not specifying this option is equivalent to
5546 [X86] Control mitigation of Spectre variant 2
5547 (indirect branch speculation) vulnerability between
5550 on - Unconditionally enable mitigations. Is
5551 enforced by spectre_v2=on
5553 off - Unconditionally disable mitigations. Is
5554 enforced by spectre_v2=off
5556 prctl - Indirect branch speculation is enabled,
5557 but mitigation can be enabled via prctl
5558 per thread. The mitigation control state
5559 is inherited on fork.
5562 - Like "prctl" above, but only STIBP is
5563 controlled per thread. IBPB is issued
5564 always when switching between different user
5568 - Same as "prctl" above, but all seccomp
5569 threads will enable the mitigation unless
5570 they explicitly opt out.
5573 - Like "seccomp" above, but only STIBP is
5574 controlled per thread. IBPB is issued
5575 always when switching between different
5576 user space processes.
5578 auto - Kernel selects the mitigation depending on
5579 the available CPU features and vulnerability.
5581 Default mitigation: "prctl"
5583 Not specifying this option is equivalent to
5584 spectre_v2_user=auto.
5586 spec_store_bypass_disable=
5587 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
5588 (Speculative Store Bypass vulnerability)
5590 Certain CPUs are vulnerable to an exploit against a
5591 a common industry wide performance optimization known
5592 as "Speculative Store Bypass" in which recent stores
5593 to the same memory location may not be observed by
5594 later loads during speculative execution. The idea
5595 is that such stores are unlikely and that they can
5596 be detected prior to instruction retirement at the
5597 end of a particular speculation execution window.
5599 In vulnerable processors, the speculatively forwarded
5600 store can be used in a cache side channel attack, for
5601 example to read memory to which the attacker does not
5602 directly have access (e.g. inside sandboxed code).
5604 This parameter controls whether the Speculative Store
5605 Bypass optimization is used.
5607 On x86 the options are:
5609 on - Unconditionally disable Speculative Store Bypass
5610 off - Unconditionally enable Speculative Store Bypass
5611 auto - Kernel detects whether the CPU model contains an
5612 implementation of Speculative Store Bypass and
5613 picks the most appropriate mitigation. If the
5614 CPU is not vulnerable, "off" is selected. If the
5615 CPU is vulnerable the default mitigation is
5616 architecture and Kconfig dependent. See below.
5617 prctl - Control Speculative Store Bypass per thread
5618 via prctl. Speculative Store Bypass is enabled
5619 for a process by default. The state of the control
5620 is inherited on fork.
5621 seccomp - Same as "prctl" above, but all seccomp threads
5622 will disable SSB unless they explicitly opt out.
5624 Default mitigations:
5627 On powerpc the options are:
5629 on,auto - On Power8 and Power9 insert a store-forwarding
5630 barrier on kernel entry and exit. On Power7
5631 perform a software flush on kernel entry and
5635 Not specifying this option is equivalent to
5636 spec_store_bypass_disable=auto.
5638 spia_io_base= [HW,MTD]
5644 [X86] Enable split lock detection or bus lock detection
5646 When enabled (and if hardware support is present), atomic
5647 instructions that access data across cache line
5648 boundaries will result in an alignment check exception
5649 for split lock detection or a debug exception for
5654 warn - the kernel will emit rate-limited warnings
5655 about applications triggering the #AC
5656 exception or the #DB exception. This mode is
5657 the default on CPUs that support split lock
5658 detection or bus lock detection. Default
5659 behavior is by #AC if both features are
5660 enabled in hardware.
5662 fatal - the kernel will send SIGBUS to applications
5663 that trigger the #AC exception or the #DB
5664 exception. Default behavior is by #AC if
5665 both features are enabled in hardware.
5668 Set system wide rate limit to N bus locks
5669 per second for bus lock detection.
5672 N/A for split lock detection.
5675 If an #AC exception is hit in the kernel or in
5676 firmware (i.e. not while executing in user mode)
5677 the kernel will oops in either "warn" or "fatal"
5680 #DB exception for bus lock is triggered only when
5684 Control the Special Register Buffer Data Sampling
5687 Certain CPUs are vulnerable to an MDS-like
5688 exploit which can leak bits from the random
5691 By default, this issue is mitigated by
5692 microcode. However, the microcode fix can cause
5693 the RDRAND and RDSEED instructions to become
5694 much slower. Among other effects, this will
5695 result in reduced throughput from /dev/urandom.
5697 The microcode mitigation can be disabled with
5698 the following option:
5700 off: Disable mitigation and remove
5701 performance impact to RDRAND and RDSEED
5703 srcutree.big_cpu_lim [KNL]
5704 Specifies the number of CPUs constituting a
5705 large system, such that srcu_struct structures
5706 should immediately allocate an srcu_node array.
5707 This kernel-boot parameter defaults to 128,
5708 but takes effect only when the low-order four
5709 bits of srcutree.convert_to_big is equal to 3
5712 srcutree.convert_to_big [KNL]
5713 Specifies under what conditions an SRCU tree
5714 srcu_struct structure will be converted to big
5715 form, that is, with an rcu_node tree:
5718 1: At init_srcu_struct() time.
5719 2: When rcutorture decides to.
5720 3: Decide at boot time (default).
5721 0x1X: Above plus if high contention.
5723 Either way, the srcu_node tree will be sized based
5724 on the actual runtime number of CPUs (nr_cpu_ids)
5725 instead of the compile-time CONFIG_NR_CPUS.
5727 srcutree.counter_wrap_check [KNL]
5728 Specifies how frequently to check for
5729 grace-period sequence counter wrap for the
5730 srcu_data structure's ->srcu_gp_seq_needed field.
5731 The greater the number of bits set in this kernel
5732 parameter, the less frequently counter wrap will
5733 be checked for. Note that the bottom two bits
5736 srcutree.exp_holdoff [KNL]
5737 Specifies how many nanoseconds must elapse
5738 since the end of the last SRCU grace period for
5739 a given srcu_struct until the next normal SRCU
5740 grace period will be considered for automatic
5741 expediting. Set to zero to disable automatic
5744 srcutree.small_contention_lim [KNL]
5745 Specifies the number of update-side contention
5746 events per jiffy will be tolerated before
5747 initiating a conversion of an srcu_struct
5748 structure to big form. Note that the value of
5749 srcutree.convert_to_big must have the 0x10 bit
5750 set for contention-based conversions to occur.
5753 Speculative Store Bypass Disable control
5755 On CPUs that are vulnerable to the Speculative
5756 Store Bypass vulnerability and offer a
5757 firmware based mitigation, this parameter
5758 indicates how the mitigation should be used:
5760 force-on: Unconditionally enable mitigation for
5761 for both kernel and userspace
5762 force-off: Unconditionally disable mitigation for
5763 for both kernel and userspace
5764 kernel: Always enable mitigation in the
5765 kernel, and offer a prctl interface
5766 to allow userspace to register its
5767 interest in being mitigated too.
5769 stack_guard_gap= [MM]
5770 override the default stack gap protection. The value
5771 is in page units and it defines how many pages prior
5772 to (for stacks growing down) resp. after (for stacks
5773 growing up) the main stack are reserved for no other
5774 mapping. Default value is 256 pages.
5776 stack_depot_disable= [KNL]
5777 Setting this to true through kernel command line will
5778 disable the stack depot thereby saving the static memory
5779 consumed by the stack hash table. By default this is set
5783 Enabled the stack tracer on boot up.
5785 stacktrace_filter=[function-list]
5786 [FTRACE] Limit the functions that the stack tracer
5787 will trace at boot up. function-list is a comma-separated
5788 list of functions. This list can be changed at run
5789 time by the stack_trace_filter file in the debugfs
5790 tracing directory. Note, this enables stack tracing
5791 and the stacktrace above is not needed.
5795 Set the STI (builtin display/keyboard on the HP-PARISC
5796 machines) console (graphic card) which should be used
5797 as the initial boot-console.
5798 See also comment in drivers/video/console/sticore.c.
5801 See comment in drivers/video/console/sticore.c.
5804 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
5809 Enable or disable strict sigaltstack size checks
5810 against the required signal frame size which
5811 depends on the supported FPU features. This can
5812 be used to filter out binaries which have
5813 not yet been made aware of AT_MINSIGSTKSZ.
5815 sunrpc.min_resvport=
5816 sunrpc.max_resvport=
5818 SunRPC servers often require that client requests
5819 originate from a privileged port (i.e. a port in the
5820 range 0 < portnr < 1024).
5821 An administrator who wishes to reserve some of these
5822 ports for other uses may adjust the range that the
5823 kernel's sunrpc client considers to be privileged
5824 using these two parameters to set the minimum and
5825 maximum port values.
5827 sunrpc.svc_rpc_per_connection_limit=
5829 Limit the number of requests that the server will
5830 process in parallel from a single connection.
5831 The default value is 0 (no limit).
5835 Control how the NFS server code allocates CPUs to
5836 service thread pools. Depending on how many NICs
5837 you have and where their interrupts are bound, this
5838 option will affect which CPUs will do NFS serving.
5839 Note: this parameter cannot be changed while the
5840 NFS server is running.
5842 auto the server chooses an appropriate mode
5843 automatically using heuristics
5844 global a single global pool contains all CPUs
5845 percpu one pool for each CPU
5846 pernode one pool for each NUMA node (equivalent
5847 to global on non-NUMA machines)
5849 sunrpc.tcp_slot_table_entries=
5850 sunrpc.udp_slot_table_entries=
5852 Sets the upper limit on the number of simultaneous
5853 RPC calls that can be sent from the client to a
5854 server. Increasing these values may allow you to
5855 improve throughput, but will also increase the
5856 amount of memory reserved for use by the client.
5858 suspend.pm_test_delay=
5860 Sets the number of seconds to remain in a suspend test
5861 mode before resuming the system (see
5862 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
5863 is set. Default value is 5.
5866 Format: { on | off | y | n | 1 | 0 }
5867 This parameter controls use of the Protected
5868 Execution Facility on pSeries.
5871 [KNL] Enable accounting of swap in memory resource
5872 controller if no parameter or 1 is given or disable
5873 it if 0 is given (See Documentation/admin-guide/cgroup-v1/memory.rst)
5875 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
5876 Format: { <int> | force | noforce }
5877 <int> -- Number of I/O TLB slabs
5878 force -- force using of bounce buffers even if they
5879 wouldn't be automatically used by the kernel
5880 noforce -- Never use bounce buffers (for debugging)
5885 Set a sysctl parameter, right before loading the init
5886 process, as if the value was written to the respective
5887 /proc/sys/... file. Both '.' and '/' are recognized as
5888 separators. Unrecognized parameters and invalid values
5889 are reported in the kernel log. Sysctls registered
5890 later by a loaded module cannot be set this way.
5891 Example: sysctl.vm.swappiness=40
5893 sysfs.deprecated=0|1 [KNL]
5894 Enable/disable old style sysfs layout for old udev
5895 on older distributions. When this option is enabled
5896 very new udev will not work anymore. When this option
5897 is disabled (or CONFIG_SYSFS_DEPRECATED not compiled)
5898 in older udev will not work anymore.
5899 Default depends on CONFIG_SYSFS_DEPRECATED_V2 set in
5900 the kernel configuration.
5902 sysrq_always_enabled
5904 Ignore sysrq setting - this boot parameter will
5905 neutralize any effect of /proc/sys/kernel/sysrq.
5906 Useful for debugging.
5908 tcpmhash_entries= [KNL,NET]
5909 Set the number of tcp_metrics_hash slots.
5910 Default value is 8192 or 16384 depending on total
5911 ram pages. This is used to specify the TCP metrics
5912 cache size. See Documentation/networking/ip-sysctl.rst
5913 "tcp_no_metrics_save" section for more details.
5917 test_suspend= [SUSPEND][,N]
5918 Specify "mem" (for Suspend-to-RAM) or "standby" (for
5919 standby suspend) or "freeze" (for suspend type freeze)
5920 as the system sleep state during system startup with
5921 the optional capability to repeat N number of times.
5922 The system is woken from this state using a
5923 wakeup-capable RTC alarm.
5925 thash_entries= [KNL,NET]
5926 Set number of hash buckets for TCP connection
5928 thermal.act= [HW,ACPI]
5929 -1: disable all active trip points in all thermal zones
5930 <degrees C>: override all lowest active trip points
5932 thermal.crt= [HW,ACPI]
5933 -1: disable all critical trip points in all thermal zones
5934 <degrees C>: override all critical trip points
5936 thermal.nocrt= [HW,ACPI]
5937 Set to disable actions on ACPI thermal zone
5938 critical and hot trip points.
5940 thermal.off= [HW,ACPI]
5941 1: disable ACPI thermal control
5943 thermal.psv= [HW,ACPI]
5944 -1: disable all passive trip points
5945 <degrees C>: override all passive trip points to this
5948 thermal.tzp= [HW,ACPI]
5949 Specify global default ACPI thermal zone polling rate
5950 <deci-seconds>: poll all this frequency
5951 0: no polling (default)
5954 Force threading of all interrupt handlers except those
5955 marked explicitly IRQF_NO_THREAD.
5959 Specify if the kernel should make use of the cpu
5960 topology information if the hardware supports this.
5961 The scheduler will make use of this information and
5962 e.g. base its process migration decisions on it.
5965 topology_updates= [KNL, PPC, NUMA]
5967 Specify if the kernel should ignore (off)
5968 topology updates sent by the hypervisor to this
5971 torture.disable_onoff_at_boot= [KNL]
5972 Prevent the CPU-hotplug component of torturing
5973 until after init has spawned.
5975 torture.ftrace_dump_at_shutdown= [KNL]
5976 Dump the ftrace buffer at torture-test shutdown,
5977 even if there were no errors. This can be a
5978 very costly operation when many torture tests
5979 are running concurrently, especially on systems
5980 with rotating-rust storage.
5982 torture.verbose_sleep_frequency= [KNL]
5983 Specifies how many verbose printk()s should be
5984 emitted between each sleep. The default of zero
5985 disables verbose-printk() sleeping.
5987 torture.verbose_sleep_duration= [KNL]
5988 Duration of each verbose-printk() sleep in jiffies.
5992 tpm_suspend_pcr=[HW,TPM]
5993 Format: integer pcr id
5994 Specify that at suspend time, the tpm driver
5995 should extend the specified pcr with zeros,
5996 as a workaround for some chips which fail to
5997 flush the last written pcr on TPM_SaveState.
5998 This will guarantee that all the other pcrs
6001 trace_buf_size=nn[KMG]
6002 [FTRACE] will set tracing buffer size on each cpu.
6004 trace_event=[event-list]
6005 [FTRACE] Set and start specified trace events in order
6006 to facilitate early boot debugging. The event-list is a
6007 comma-separated list of trace events to enable. See
6008 also Documentation/trace/events.rst
6010 trace_options=[option-list]
6011 [FTRACE] Enable or disable tracer options at boot.
6012 The option-list is a comma delimited list of options
6013 that can be enabled or disabled just as if you were
6014 to echo the option name into
6016 /sys/kernel/debug/tracing/trace_options
6018 For example, to enable stacktrace option (to dump the
6019 stack trace of each event), add to the command line:
6021 trace_options=stacktrace
6023 See also Documentation/trace/ftrace.rst "trace options"
6027 Have the tracepoints sent to printk as well as the
6028 tracing ring buffer. This is useful for early boot up
6029 where the system hangs or reboots and does not give the
6030 option for reading the tracing buffer or performing a
6031 ftrace_dump_on_oops.
6033 To turn off having tracepoints sent to printk,
6034 echo 0 > /proc/sys/kernel/tracepoint_printk
6035 Note, echoing 1 into this file without the
6036 tracepoint_printk kernel cmdline option has no effect.
6038 The tp_printk_stop_on_boot (see below) can also be used
6039 to stop the printing of events to console at
6044 Having tracepoints sent to printk() and activating high
6045 frequency tracepoints such as irq or sched, can cause
6046 the system to live lock.
6048 tp_printk_stop_on_boot[FTRACE]
6049 When tp_printk (above) is set, it can cause a lot of noise
6050 on the console. It may be useful to only include the
6051 printing of events during boot up, as user space may
6052 make the system inoperable.
6054 This command line option will stop the printing of events
6055 to console at the late_initcall_sync() time frame.
6058 [FTRACE] enable this option to disable tracing when a
6059 warning is hit. This turns off "tracing_on". Tracing can
6060 be enabled again by echoing '1' into the "tracing_on"
6061 file located in /sys/kernel/debug/tracing/
6063 This option is useful, as it disables the trace before
6064 the WARNING dump is called, which prevents the trace to
6065 be filled with content caused by the warning output.
6067 This option can also be set at run time via the sysctl
6068 option: kernel/traceoff_on_warning
6070 transparent_hugepage=
6072 Format: [always|madvise|never]
6073 Can be used to control the default behavior of the system
6074 with respect to transparent hugepages.
6075 See Documentation/admin-guide/mm/transhuge.rst
6078 trusted.source= [KEYS]
6080 This parameter identifies the trust source as a backend
6081 for trusted keys implementation. Supported trust
6086 If not specified then it defaults to iterating through
6087 the trust source list starting with TPM and assigns the
6088 first trust source as a backend which is initialized
6089 successfully during iteration.
6093 The RNG used to generate key material for trusted keys.
6096 - the same value as trusted.source: "tpm" or "tee"
6098 If not specified, "default" is used. In this case,
6099 the RNG's choice is left to each individual trust source.
6101 tsc= Disable clocksource stability checks for TSC.
6103 [x86] reliable: mark tsc clocksource as reliable, this
6104 disables clocksource verification at runtime, as well
6105 as the stability checks done at bootup. Used to enable
6106 high-resolution timer mode on older hardware, and in
6107 virtualized environment.
6108 [x86] noirqtime: Do not use TSC to do irq accounting.
6109 Used to run time disable IRQ_TIME_ACCOUNTING on any
6110 platforms where RDTSC is slow and this accounting
6112 [x86] unstable: mark the TSC clocksource as unstable, this
6113 marks the TSC unconditionally unstable at bootup and
6114 avoids any further wobbles once the TSC watchdog notices.
6115 [x86] nowatchdog: disable clocksource watchdog. Used
6116 in situations with strict latency requirements (where
6117 interruptions from clocksource watchdog are not
6120 tsc_early_khz= [X86] Skip early TSC calibration and use the given
6121 value instead. Useful when the early TSC frequency discovery
6122 procedure is not reliable, such as on overclocked systems
6123 with CPUID.16h support and partial CPUID.15h support.
6124 Format: <unsigned int>
6126 tsx= [X86] Control Transactional Synchronization
6127 Extensions (TSX) feature in Intel processors that
6128 support TSX control.
6130 This parameter controls the TSX feature. The options are:
6132 on - Enable TSX on the system. Although there are
6133 mitigations for all known security vulnerabilities,
6134 TSX has been known to be an accelerator for
6135 several previous speculation-related CVEs, and
6136 so there may be unknown security risks associated
6137 with leaving it enabled.
6139 off - Disable TSX on the system. (Note that this
6140 option takes effect only on newer CPUs which are
6141 not vulnerable to MDS, i.e., have
6142 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
6143 the new IA32_TSX_CTRL MSR through a microcode
6144 update. This new MSR allows for the reliable
6145 deactivation of the TSX functionality.)
6147 auto - Disable TSX if X86_BUG_TAA is present,
6148 otherwise enable TSX on the system.
6150 Not specifying this option is equivalent to tsx=off.
6152 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6155 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
6156 Abort (TAA) vulnerability.
6158 Similar to Micro-architectural Data Sampling (MDS)
6159 certain CPUs that support Transactional
6160 Synchronization Extensions (TSX) are vulnerable to an
6161 exploit against CPU internal buffers which can forward
6162 information to a disclosure gadget under certain
6165 In vulnerable processors, the speculatively forwarded
6166 data can be used in a cache side channel attack, to
6167 access data to which the attacker does not have direct
6170 This parameter controls the TAA mitigation. The
6173 full - Enable TAA mitigation on vulnerable CPUs
6176 full,nosmt - Enable TAA mitigation and disable SMT on
6177 vulnerable CPUs. If TSX is disabled, SMT
6178 is not disabled because CPU is not
6179 vulnerable to cross-thread TAA attacks.
6180 off - Unconditionally disable TAA mitigation
6182 On MDS-affected machines, tsx_async_abort=off can be
6183 prevented by an active MDS mitigation as both vulnerabilities
6184 are mitigated with the same mechanism so in order to disable
6185 this mitigation, you need to specify mds=off too.
6187 Not specifying this option is equivalent to
6188 tsx_async_abort=full. On CPUs which are MDS affected
6189 and deploy MDS mitigation, TAA mitigation is not
6190 required and doesn't provide any additional
6194 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6196 turbografx.map[2|3]= [HW,JOY]
6197 TurboGraFX parallel port interface
6199 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
6200 See also Documentation/input/devices/joystick-parport.rst
6202 udbg-immortal [PPC] When debugging early kernel crashes that
6203 happen after console_init() and before a proper
6204 console driver takes over, this boot options might
6205 help "seeing" what's going on.
6207 uhash_entries= [KNL,NET]
6208 Set number of hash buckets for UDP/UDP-Lite connections
6211 [USB] Ignore overcurrent events (default N).
6212 Some badly-designed motherboards generate lots of
6213 bogus events, for ports that aren't wired to
6214 anything. Set this parameter to avoid log spamming.
6215 Note that genuine overcurrent events won't be
6219 [X86] Cause panic on unknown NMI.
6221 usbcore.authorized_default=
6222 [USB] Default USB device authorization:
6223 (default -1 = authorized except for wireless USB,
6224 0 = not authorized, 1 = authorized, 2 = authorized
6225 if device connected to internal port)
6227 usbcore.autosuspend=
6228 [USB] The autosuspend time delay (in seconds) used
6229 for newly-detected USB devices (default 2). This
6230 is the time required before an idle device will be
6231 autosuspended. Devices for which the delay is set
6232 to a negative value won't be autosuspended at all.
6234 usbcore.usbfs_snoop=
6235 [USB] Set to log all usbfs traffic (default 0 = off).
6237 usbcore.usbfs_snoop_max=
6238 [USB] Maximum number of bytes to snoop in each URB
6241 usbcore.blinkenlights=
6242 [USB] Set to cycle leds on hubs (default 0 = off).
6244 usbcore.old_scheme_first=
6245 [USB] Start with the old device initialization
6246 scheme (default 0 = off).
6248 usbcore.usbfs_memory_mb=
6249 [USB] Memory limit (in MB) for buffers allocated by
6250 usbfs (default = 16, 0 = max = 2047).
6252 usbcore.use_both_schemes=
6253 [USB] Try the other device initialization scheme
6254 if the first one fails (default 1 = enabled).
6256 usbcore.initial_descriptor_timeout=
6257 [USB] Specifies timeout for the initial 64-byte
6258 USB_REQ_GET_DESCRIPTOR request in milliseconds
6259 (default 5000 = 5.0 seconds).
6261 usbcore.nousb [USB] Disable the USB subsystem
6264 [USB] A list of quirk entries to augment the built-in
6265 usb core quirk list. List entries are separated by
6266 commas. Each entry has the form
6267 VendorID:ProductID:Flags. The IDs are 4-digit hex
6268 numbers and Flags is a set of letters. Each letter
6269 will change the built-in quirk; setting it if it is
6270 clear and clearing it if it is set. The letters have
6271 the following meanings:
6272 a = USB_QUIRK_STRING_FETCH_255 (string
6273 descriptors must not be fetched using
6275 b = USB_QUIRK_RESET_RESUME (device can't resume
6276 correctly so reset it instead);
6277 c = USB_QUIRK_NO_SET_INTF (device can't handle
6278 Set-Interface requests);
6279 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
6280 handle its Configuration or Interface
6282 e = USB_QUIRK_RESET (device can't be reset
6283 (e.g morph devices), don't use reset);
6284 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
6285 more interface descriptions than the
6286 bNumInterfaces count, and can't handle
6287 talking to these interfaces);
6288 g = USB_QUIRK_DELAY_INIT (device needs a pause
6289 during initialization, after we read
6290 the device descriptor);
6291 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
6292 high speed and super speed interrupt
6293 endpoints, the USB 2.0 and USB 3.0 spec
6294 require the interval in microframes (1
6295 microframe = 125 microseconds) to be
6296 calculated as interval = 2 ^
6298 Devices with this quirk report their
6299 bInterval as the result of this
6300 calculation instead of the exponent
6301 variable used in the calculation);
6302 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
6303 handle device_qualifier descriptor
6305 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
6306 generates spurious wakeup, ignore
6307 remote wakeup capability);
6308 k = USB_QUIRK_NO_LPM (device can't handle Link
6310 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
6311 (Device reports its bInterval as linear
6312 frames instead of the USB 2.0
6314 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
6315 to be disconnected before suspend to
6316 prevent spurious wakeup);
6317 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
6318 pause after every control message);
6319 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
6320 delay after resetting its port);
6321 Example: quirks=0781:5580:bk,0a5c:5834:gij
6324 [USBHID] The interval which mice are to be polled at.
6327 [USBHID] The interval which joysticks are to be polled at.
6330 [USBHID] The interval which keyboards are to be polled at.
6332 usb-storage.delay_use=
6333 [UMS] The delay in seconds before a new device is
6334 scanned for Logical Units (default 1).
6337 [UMS] A list of quirks entries to supplement or
6338 override the built-in unusual_devs list. List
6339 entries are separated by commas. Each entry has
6340 the form VID:PID:Flags where VID and PID are Vendor
6341 and Product ID values (4-digit hex numbers) and
6342 Flags is a set of characters, each corresponding
6343 to a common usb-storage quirk flag as follows:
6344 a = SANE_SENSE (collect more than 18 bytes
6345 of sense data, not on uas);
6346 b = BAD_SENSE (don't collect more than 18
6347 bytes of sense data, not on uas);
6348 c = FIX_CAPACITY (decrease the reported
6349 device capacity by one sector);
6350 d = NO_READ_DISC_INFO (don't use
6351 READ_DISC_INFO command, not on uas);
6352 e = NO_READ_CAPACITY_16 (don't use
6353 READ_CAPACITY_16 command);
6354 f = NO_REPORT_OPCODES (don't use report opcodes
6356 g = MAX_SECTORS_240 (don't transfer more than
6357 240 sectors at a time, uas only);
6358 h = CAPACITY_HEURISTICS (decrease the
6359 reported device capacity by one
6360 sector if the number is odd);
6361 i = IGNORE_DEVICE (don't bind to this
6363 j = NO_REPORT_LUNS (don't use report luns
6365 k = NO_SAME (do not use WRITE_SAME, uas only)
6366 l = NOT_LOCKABLE (don't try to lock and
6367 unlock ejectable media, not on uas);
6368 m = MAX_SECTORS_64 (don't transfer more
6369 than 64 sectors = 32 KB at a time,
6371 n = INITIAL_READ10 (force a retry of the
6372 initial READ(10) command, not on uas);
6373 o = CAPACITY_OK (accept the capacity
6374 reported by the device, not on uas);
6375 p = WRITE_CACHE (the device cache is ON
6376 by default, not on uas);
6377 r = IGNORE_RESIDUE (the device reports
6378 bogus residue values, not on uas);
6379 s = SINGLE_LUN (the device has only one
6381 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
6382 commands, uas only);
6383 u = IGNORE_UAS (don't bind to the uas driver);
6384 w = NO_WP_DETECT (don't test whether the
6385 medium is write-protected).
6386 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
6387 even if the device claims no cache,
6389 Example: quirks=0419:aaf5:rl,0421:0433:rc
6391 user_debug= [KNL,ARM]
6393 See arch/arm/Kconfig.debug help text.
6394 1 - undefined instruction events
6396 4 - invalid data aborts
6399 Example: user_debug=31
6402 [X86] Flags controlling user PTE allocations.
6404 nohigh = do not allocate PTE pages in
6405 HIGHMEM regardless of setting
6409 On X86_32, this is an alias for vdso32=. Otherwise:
6411 vdso=1: enable VDSO (the default)
6412 vdso=0: disable VDSO mapping
6414 vdso32= [X86] Control the 32-bit vDSO
6415 vdso32=1: enable 32-bit VDSO
6416 vdso32=0 or vdso32=2: disable 32-bit VDSO
6418 See the help text for CONFIG_COMPAT_VDSO for more
6419 details. If CONFIG_COMPAT_VDSO is set, the default is
6420 vdso32=0; otherwise, the default is vdso32=1.
6422 For compatibility with older kernels, vdso32=2 is an
6425 Try vdso32=0 if you encounter an error that says:
6426 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
6429 vector=percpu: enable percpu vector domain
6431 video= [FB] Frame buffer configuration
6432 See Documentation/fb/modedb.rst.
6434 video.brightness_switch_enabled= [0,1]
6435 If set to 1, on receiving an ACPI notify event
6436 generated by hotkey, video driver will adjust brightness
6437 level and then send out the event to user space through
6438 the allocated input device; If set to 0, video driver
6439 will only send out the event without touching backlight
6444 [VMMIO] Memory mapped virtio (platform) device.
6446 <size>@<baseaddr>:<irq>[:<id>]
6448 <size> := size (can use standard suffixes
6450 <baseaddr> := physical base address
6451 <irq> := interrupt number (as passed to
6453 <id> := (optional) platform device id
6455 virtio_mmio.device=1K@0x100b0000:48:7
6457 Can be used multiple times for multiple devices.
6459 vga= [BOOT,X86-32] Select a particular video mode
6460 See Documentation/x86/boot.rst and
6461 Documentation/admin-guide/svga.rst.
6462 Use vga=ask for menu.
6463 This is actually a boot loader parameter; the value is
6464 passed to the kernel using a special protocol.
6466 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
6467 May slow down system boot speed, especially when
6468 enabled on systems with a large amount of memory.
6469 All options are enabled by default, and this
6470 interface is meant to allow for selectively
6471 enabling or disabling specific virtual memory
6474 Available options are:
6475 P Enable page structure init time poisoning
6476 - Disable all of the above options
6478 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
6479 size of <nn>. This can be used to increase the
6480 minimum size (128MB on x86). It can also be used to
6481 decrease the size and leave more room for directly
6484 vmcp_cma=nn[MG] [KNL,S390]
6485 Sets the memory size reserved for contiguous memory
6486 allocations for the vmcp device driver.
6488 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
6491 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
6494 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
6498 Controls the behavior of vsyscalls (i.e. calls to
6499 fixed addresses of 0xffffffffff600x00 from legacy
6500 code). Most statically-linked binaries and older
6501 versions of glibc use these calls. Because these
6502 functions are at fixed addresses, they make nice
6503 targets for exploits that can control RIP.
6505 emulate [default] Vsyscalls turn into traps and are
6506 emulated reasonably safely. The vsyscall
6509 xonly Vsyscalls turn into traps and are
6510 emulated reasonably safely. The vsyscall
6511 page is not readable.
6513 none Vsyscalls don't work at all. This makes
6514 them quite hard to use for exploits but
6515 might break your system.
6517 vt.color= [VT] Default text color.
6518 Format: 0xYX, X = foreground, Y = background.
6519 Default: 0x07 = light gray on black.
6521 vt.cur_default= [VT] Default cursor shape.
6522 Format: 0xCCBBAA, where AA, BB, and CC are the same as
6523 the parameters of the <Esc>[?A;B;Cc escape sequence;
6524 see VGA-softcursor.txt. Default: 2 = underline.
6526 vt.default_blu= [VT]
6527 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
6528 Change the default blue palette of the console.
6529 This is a 16-member array composed of values
6532 vt.default_grn= [VT]
6533 Format: <green0>,<green1>,<green2>,...,<green15>
6534 Change the default green palette of the console.
6535 This is a 16-member array composed of values
6538 vt.default_red= [VT]
6539 Format: <red0>,<red1>,<red2>,...,<red15>
6540 Change the default red palette of the console.
6541 This is a 16-member array composed of values
6547 Set system-wide default UTF-8 mode for all tty's.
6548 Default is 1, i.e. UTF-8 mode is enabled for all
6549 newly opened terminals.
6551 vt.global_cursor_default=
6554 Set system-wide default for whether a cursor
6555 is shown on new VTs. Default is -1,
6556 i.e. cursors will be created by default unless
6557 overridden by individual drivers. 0 will hide
6558 cursors, 1 will display them.
6560 vt.italic= [VT] Default color for italic text; 0-15.
6563 vt.underline= [VT] Default color for underlined text; 0-15.
6566 watchdog timers [HW,WDT] For information on watchdog timers,
6567 see Documentation/watchdog/watchdog-parameters.rst
6568 or other driver-specific files in the
6569 Documentation/watchdog/ directory.
6573 Set the hard lockup detector stall duration
6574 threshold in seconds. The soft lockup detector
6575 threshold is set to twice the value. A value of 0
6576 disables both lockup detectors. Default is 10
6579 workqueue.watchdog_thresh=
6580 If CONFIG_WQ_WATCHDOG is configured, workqueue can
6581 warn stall conditions and dump internal state to
6582 help debugging. 0 disables workqueue stall
6583 detection; otherwise, it's the stall threshold
6584 duration in seconds. The default value is 30 and
6585 it can be updated at runtime by writing to the
6586 corresponding sysfs file.
6588 workqueue.disable_numa
6589 By default, all work items queued to unbound
6590 workqueues are affine to the NUMA nodes they're
6591 issued on, which results in better behavior in
6592 general. If NUMA affinity needs to be disabled for
6593 whatever reason, this option can be used. Note
6594 that this also can be controlled per-workqueue for
6595 workqueues visible under /sys/bus/workqueue/.
6597 workqueue.power_efficient
6598 Per-cpu workqueues are generally preferred because
6599 they show better performance thanks to cache
6600 locality; unfortunately, per-cpu workqueues tend to
6601 be more power hungry than unbound workqueues.
6603 Enabling this makes the per-cpu workqueues which
6604 were observed to contribute significantly to power
6605 consumption unbound, leading to measurably lower
6606 power usage at the cost of small performance
6609 The default value of this parameter is determined by
6610 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
6612 workqueue.debug_force_rr_cpu
6613 Workqueue used to implicitly guarantee that work
6614 items queued without explicit CPU specified are put
6615 on the local CPU. This guarantee is no longer true
6616 and while local CPU is still preferred work items
6617 may be put on foreign CPUs. This debug option
6618 forces round-robin CPU selection to flush out
6619 usages which depend on the now broken guarantee.
6620 When enabled, memory and cache locality will be
6623 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
6624 default x2apic cluster mode on platforms
6627 xen_512gb_limit [KNL,X86-64,XEN]
6628 Restricts the kernel running paravirtualized under Xen
6629 to use only up to 512 GB of RAM. The reason to do so is
6630 crash analysis tools and Xen tools for doing domain
6631 save/restore/migration must be enabled to handle larger
6634 xen_emul_unplug= [HW,X86,XEN]
6635 Unplug Xen emulated devices
6636 Format: [unplug0,][unplug1]
6637 ide-disks -- unplug primary master IDE devices
6638 aux-ide-disks -- unplug non-primary-master IDE devices
6639 nics -- unplug network devices
6640 all -- unplug all emulated devices (NICs and IDE disks)
6641 unnecessary -- unplugging emulated devices is
6642 unnecessary even if the host did not respond to
6644 never -- do not unplug even if version check succeeds
6646 xen_legacy_crash [X86,XEN]
6647 Crash from Xen panic notifier, without executing late
6648 panic() code such as dumping handler.
6650 xen_nopvspin [X86,XEN]
6651 Disables the qspinlock slowpath using Xen PV optimizations.
6652 This parameter is obsoleted by "nopvspin" parameter, which
6653 has equivalent effect for XEN platform.
6656 Disables the PV optimizations forcing the HVM guest to
6657 run as generic HVM guest with no PV drivers.
6658 This option is obsoleted by the "nopv" option, which
6659 has equivalent effect for XEN platform.
6661 xen_no_vector_callback
6662 [KNL,X86,XEN] Disable the vector callback for Xen
6663 event channel interrupts.
6665 xen_scrub_pages= [XEN]
6666 Boolean option to control scrubbing pages before giving them back
6667 to Xen, for use by other domains. Can be also changed at runtime
6668 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
6669 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
6671 xen_timer_slop= [X86-64,XEN]
6672 Set the timer slop (in nanoseconds) for the virtual Xen
6673 timers (default is 100000). This adjusts the minimum
6674 delta of virtualized Xen timers, where lower values
6675 improve timer resolution at the expense of processing
6676 more timer interrupts.
6678 xen.balloon_boot_timeout= [XEN]
6679 The time (in seconds) to wait before giving up to boot
6680 in case initial ballooning fails to free enough memory.
6681 Applies only when running as HVM or PVH guest and
6682 started with less memory configured than allowed at
6683 max. Default is 180.
6685 xen.event_eoi_delay= [XEN]
6686 How long to delay EOI handling in case of event
6687 storms (jiffies). Default is 10.
6689 xen.event_loop_timeout= [XEN]
6690 After which time (jiffies) the event handling loop
6691 should start to delay EOI handling. Default is 2.
6693 xen.fifo_events= [XEN]
6694 Boolean parameter to disable using fifo event handling
6695 even if available. Normally fifo event handling is
6696 preferred over the 2-level event handling, as it is
6697 fairer and the number of possible event channels is
6698 much higher. Default is on (use fifo events).
6700 nopv= [X86,XEN,KVM,HYPER_V,VMWARE]
6701 Disables the PV optimizations forcing the guest to run
6702 as generic guest with no PV drivers. Currently support
6703 XEN HVM, KVM, HYPER_V and VMWARE guest.
6705 nopvspin [X86,XEN,KVM]
6706 Disables the qspinlock slow path using PV optimizations
6707 which allow the hypervisor to 'idle' the guest on lock
6710 xirc2ps_cs= [NET,PCMCIA]
6712 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
6715 By default on POWER9 and above, the kernel will
6716 natively use the XIVE interrupt controller. This option
6717 allows the fallback firmware mode to be used:
6719 off Fallback to firmware control of XIVE interrupt
6720 controller on both pseries and powernv
6721 platforms. Only useful on POWER9 and above.
6723 xive.store-eoi=off [PPC]
6724 By default on POWER10 and above, the kernel will use
6725 stores for EOI handling when the XIVE interrupt mode
6726 is active. This option allows the XIVE driver to use
6727 loads instead, as on POWER9.
6729 xhci-hcd.quirks [USB,KNL]
6730 A hex value specifying bitmask with supplemental xhci
6731 host controller quirks. Meaning of each bit can be
6732 consulted in header drivers/usb/host/xhci.h.
6735 Format: { early | on | rw | ro | off }
6736 Controls if xmon debugger is enabled. Default is off.
6737 Passing only "xmon" is equivalent to "xmon=early".
6738 early Call xmon as early as possible on boot; xmon
6739 debugger is called from setup_arch().
6740 on xmon debugger hooks will be installed so xmon
6741 is only called on a kernel crash. Default mode,
6742 i.e. either "ro" or "rw" mode, is controlled
6743 with CONFIG_XMON_DEFAULT_RO_MODE.
6744 rw xmon debugger hooks will be installed so xmon
6745 is called only on a kernel crash, mode is write,
6746 meaning SPR registers, memory and, other data
6747 can be written using xmon commands.
6748 ro same as "rw" option above but SPR registers,
6749 memory, and other data can't be written using
6751 off xmon is disabled.